Research ArticleCancer

Exploiting selective BCL-2 family inhibitors to dissect cell survival dependencies and define improved strategies for cancer therapy

See allHide authors and affiliations

Science Translational Medicine  18 Mar 2015:
Vol. 7, Issue 279, pp. 279ra40
DOI: 10.1126/scitranslmed.aaa4642

A more refined antitumor strategy

The BCL-2 family is a group of related proteins that regulate apoptosis in a variety of ways. The success of anticancer treatments often hinges on the ability to induce cancer cell death by apoptosis. As a result, there has been a great deal of interest in developing drugs that can inhibit the antiapoptotic members of the BCL-2 pathway. Unfortunately, some of these drugs are also associated with dose-limiting hematologic toxicities, such as neutropenia. Now, Leverson et al. have used a toolkit of BCL-2 family inhibitors with different specificities to show that specifically inhibiting BCL-XL (one member of this protein family) is effective for killing tumors, but without the common side effects seen with less selective drugs.

Abstract

The BCL-2/BCL-XL/BCL-W inhibitor ABT-263 (navitoclax) has shown promising clinical activity in lymphoid malignancies such as chronic lymphocytic leukemia. However, its efficacy in these settings is limited by thrombocytopenia caused by BCL-XL inhibition. This prompted the generation of the BCL-2–selective inhibitor venetoclax (ABT-199/GDC-0199), which demonstrates robust activity in these cancers but spares platelets. Navitoclax has also been shown to enhance the efficacy of docetaxel in preclinical models of solid tumors, but clinical use of this combination has been limited by neutropenia. We used venetoclax and the BCL-XL–selective inhibitors A-1155463 and A-1331852 to assess the relative contributions of inhibiting BCL-2 or BCL-XL to the efficacy and toxicity of the navitoclax-docetaxel combination. Selective BCL-2 inhibition suppressed granulopoiesis in vitro and in vivo, potentially accounting for the exacerbated neutropenia observed when navitoclax was combined with docetaxel clinically. By contrast, selectively inhibiting BCL-XL did not suppress granulopoiesis but was highly efficacious in combination with docetaxel when tested against a range of solid tumors. Therefore, BCL-XL–selective inhibitors have the potential to enhance the efficacy of docetaxel in solid tumors and avoid the exacerbation of neutropenia observed with navitoclax. These studies demonstrate the translational utility of this toolkit of selective BCL-2 family inhibitors and highlight their potential as improved cancer therapeutics.

INTRODUCTION

Multicellular organisms have evolved molecular mechanisms to eliminate cells that are no longer required or that have become compromised through environmental insults. This active process of programmed cell death, or apoptosis, is especially important for eliminating cells with the potential for malignant transformation. The ability to suppress apoptosis under conditions of cell stress has been defined as one of the hallmarks of a cancer cell (1), and thus, triggering apoptosis in cancer cells represents a powerful therapeutic approach.

Apoptosis is regulated by a family of closely related proteins exemplified by BCL-2 (B cell lymphoma protein 2), the first family member discovered. BCL-2 family proteins are defined by one to four BCL-2 homology motifs (BH1 to BH4) and can be subdivided into pro- and antiapoptotic subsets. Proapoptotic proteins include the BH3-only proteins such as BIM, BAD, BID, and NOXA, and the BH1 to BH4 proteins BAK and BAX, which serve as the ultimate effectors of apoptosis by oligomerizing to form pores in the mitochondrial outer membrane. BCL-2 and the closely related antiapoptotic proteins BCL-XL and MCL-1 can sequester their proapoptotic counterparts by binding to their BH3 motifs and thereby inhibit the initiating steps of programmed cell death (24).

Small molecules that mimic the BH3 motif have been developed with the aim of binding antiapoptotic proteins, liberating proapoptotic proteins, and triggering apoptosis in cancer cells, many of which are thought to be “primed for death” due to the expression of high levels of pro- and antiapoptotic protein complexes (5). ABT-737 and the related orally bioavailable compound ABT-263 (navitoclax) are BH3 mimetics that bind with high affinity to BCL-2, BCL-XL, and BCL-W, but not to MCL-1 (6, 7). Preclinically, navitoclax inhibits tumor growth as a single agent (8) and in combination with standard-of-care therapeutics such as gemcitabine, vincristine, and docetaxel (9, 10). In clinical studies, navitoclax exhibited objective antitumor activity in lymphoid malignancies but also induced rapid, reversible, and dose-dependent thrombocytopenia that was dose-limiting in this setting (11, 12). Navitoclax-induced thrombocytopenia was found to be a consequence of inhibiting BCL-XL (13, 14), and this prompted the development of ABT-199/GDC-0199 (venetoclax), a BCL-2–selective inhibitor that maintains efficacy in hematologic tumor models but spares platelets (15). Venetoclax has shown promising signs of clinical activity in hematologic malignancies, demonstrating objective responses in chronic lymphocytic leukemia (CLL) (16) and non-Hodgkin’s lymphoma (17).

Navitoclax has also been evaluated for the treatment of solid tumors in combination with chemotherapeutics. When combined with docetaxel, febrile neutropenia was the most commonly observed dose-limiting toxicity for navitoclax (18). However, it has remained unclear whether this effect is driven by the inhibition of BCL-2, BCL-XL, or both proteins. We reasoned that, if the contributions of BCL-XL and BCL-2 inhibition could be defined, an improved strategy to minimize toxicity and thus maximize antitumor activity might be identified. To this end, we generated the BCL-XL–selective inhibitors A-1155463 (19) and A-1331852 to complement the BCL-2–selective agent venetoclax and its close analog A-1211212. Equipped with this toolkit, we now had the opportunity to chemically dissect the biological activities of navitoclax and attribute them to the inhibition of BCL-2, BCL-XL, or both targets.

In the studies described here, we use these inhibitors to define the roles of BCL-2 and BCL-XL in maintaining the survival of multiple hematologic and solid tumor cell lines. Using a variety of in vitro, ex vivo, and in vivo model systems, we also parse the contributions of BCL-2 and BCL-XL inhibition to the antitumor efficacy and the myelotoxicity observed when navitoclax is combined with docetaxel. Our data indicate that BCL-XL inhibition is sufficient to recapitulate the efficacy of navitoclax in combination with docetaxel, whereas BCL-2 inhibition likely accounts for navitoclax-related neutropenia. Just as venetoclax represented an improvement over the efficacy/toxicity profile of navitoclax for hematologic malignancies, so BCL-XL–selective inhibitors may be better tolerated, and thus enable greater efficacy, when combined with chemotherapeutic agents for the treatment of solid tumors.

RESULTS

Selective BCL-2 family inhibitors parse the activity of navitoclax into BCL-2– and BCL-XL–dependent components

Navitoclax most closely mimics the BH3-only protein BAD, binding to BCL-2 (Ki = 0.044 nM) and BCL-XL (Ki = 0.055 nM) with high affinity and more weakly to BCL-W (Ki = 21 nM; Table 1). Because its affinity for BCL-W is >300-fold weaker, the biological effects of navitoclax are most likely due to the inhibition of BCL-2, BCL-XL, or both of these targets (Fig. 1A). Whereas venetoclax enables one to probe specifically for BCL-2–dependent phenomena, a BCL-XL–selective counterpart was needed to parse the effects of navitoclax more completely. We used the recently described BCL-XL–selective inhibitor WEHI-539 (20) as an effective lead structure to generate a more potent and chemically stable molecule, A-1155463 (19) (Fig. 1A). A-1155463 binds to BCL-XL with high affinity (Ki < 0.010 nM) but has much weaker affinity for BCL-2 (Ki = 74 nM), BCL-W (Ki = 8 nM), and MCL-1 (Ki > 444 nM) (Table 1). As predicted by this binding profile, A-1155463 disrupts BCL-XL–BIM but not BCL-2–BIM complexes in cells (Fig. 1B). A-1155463 kills BCL-XL–dependent Molt-4 cells (EC50 = 70 nM) but has no measurable cytotoxicity against BCL-2–dependent RS4;11 cells (EC50 > 5 μM) (Table 1). A-1155463 induces the hallmarks of apoptosis, as evidenced by the release of cytochrome c from mitochondria, caspase activation, and the accumulation of caspase-dependent sub–G0-G1 DNA content in BCL-XL–dependent H146 cells (8, 15) (Fig. 1, C to E). In the absence of the essential apoptosis effector proteins BAK and BAX, no toxicity was observed with A-1155463 (fig. S1), thus excluding any major off-target cytotoxicity. As anticipated (2022), potent killing of murine embryonic fibroblasts (MEFs) was observed only when Mcl-1 was deleted (fig. S1), demonstrating that this compound acted in a highly specific manner. Together, venetoclax and A-1155463 provide the ability to separate the activity of navitoclax into its respective BCL-2– and BCL-XL–dependent components.

Table 1. Biochemical and cellular activity of small-molecule BCL-2 family inhibitors.

For each compound listed, binding affinities (Ki) for BCL-2 family proteins were calculated in time-resolved fluorescence resonance energy transfer (TR-FRET) assays, and the effects on cell viability [median effective concentration (EC50)] were assessed in BCL-2–dependent (RS4;11) or BCL-XL–dependent (Molt-4) cancer cell lines. All values are in nanomolar units.

View this table:
Fig. 1. Selective BCL-2 family inhibitors enable the functional dissection of the effects of navitoclax.

(A) Chemical structures and selectivity profiles of BH3 mimetics used for in vitro studies. ABT-263 (navitoclax) inhibits both BCL-2 and BCL-XL, whereas ABT-199 (venetoclax) selectively inhibits BCL-2 and A-1155463 selectively inhibits BCL-XL. (B) Quantitative measurement of BCL-XL–BIM and BCL-2–BIM complexes in Molt-4 cells after 4-hour treatments with increasing concentrations of A-1155463. Data represent the average of triplicate experiments, with error bars indicating the SD. (C) Cytochrome c levels present in NCI-H146 mitochondrial and cytosolic fractions as determined by immunoblotting after treatments with increasing concentrations of A-1155463, venetoclax, or navitoclax for 4 hours. (D) Caspase-3/7 activation in NCI-H146 cells after incubation with increasing concentrations of A-1155463 for 4 hours. Data represent the average of triplicate experiments, with error bars indicating the SD. (E) Apoptosis (sub–G0-G1 accumulation) as assessed by fluorescence-activated cell sorting in NCI-H146 cells after 1-hour preincubation plus or minus the caspase inhibitor Z-VAD-fmk (75 μM) and an additional 24 hours of incubation with increasing concentrations of A-1155463. Data represent the average of duplicate experiments, with error bars indicating the SD.

As a first implementation of this chemical toolkit, we investigated the roles of BCL-2 and BCL-XL in mediating the survival of cancer cell lines with known sensitivity to navitoclax. Small cell lung cancer (SCLC) cell lines have shown sensitivity to navitoclax that is inversely correlated with the expression of MCL-1 (23). Because BCL2 copy number gains were observed in many of these cell lines (24), it was reasonable to infer that this activity was due to the inhibition of BCL-2. Indeed, among a panel of 11 SCLC cell lines, two with BCL2 amplification, NCI-H889 and NCI-H211, were sensitive (EC50 ≤ 1.0 μM) to venetoclax but not to A-1155463 (Fig. 2A and Table 2), indicating that these cell lines were BCL-2–dependent. However, most navitoclax-sensitive SCLC lines either were more sensitive to the BCL-XL–selective inhibitor alone (NCI-H446, NCI-H847, NCI-H1417, and NCI-H1836) or required inhibition of both BCL-2 and BCL-XL. For example, navitoclax killed NCI-H345, NCI-H69, DMS79, and NCI-H1048 more potently than did either selective inhibitor alone. Notably, the combination of venetoclax and A-1155463 was able to recapitulate the effect of navitoclax in this setting, exhibiting synergistic killing of NCI-H69 (Bliss sum = 730 ± 26) and NCI-H345 (Bliss sum = 603 ± 89) cells (Fig. 2B). NCI-H187 cells were susceptible to all three inhibitors, indicating that inhibition of either BCL-2 or BCL-XL alone was sufficient to trigger their killing. These data demonstrate the utility of venetoclax and A-1155463 for determining whether BCL-2 or BCL-XL is necessary and sufficient for the survival of a given cell line. Furthermore, they indicate that BCL-2 and BCL-XL do not function redundantly in all cellular contexts.

Fig. 2. Venetoclax and A-1155463 define the BCL-2 family dependence profile of cancer cell lines.

(A) SCLC cell lines were incubated with increasing concentrations of navitoclax (BCL-2/BCL-XL), venetoclax (BCL-2–selective), or A-1155463 (BCL-XL–selective) for 48 hours before assessing cell viability. Cell killing EC50 values are plotted for each compound against the cell lines examined. (B) NCI-H69 and NCI-H345 cells were incubated with increasing concentrations of A-1155463 in the presence or absence of venetoclax for 48 hours before assessing cell viability. (C) AML cell lines were treated as in (A) before assessing cell viability. Cell killing EC50 values are plotted for each compound against the cell lines examined. (D) KG-1 and SKM-1 cells were incubated with increasing concentrations of A-1155463 in the presence or absence of venetoclax for 48 hours before assessing cell viability.

Table 2. Cell killing activity of small-molecule BCL-2 family inhibitors.

SCLC and AML cell lines were incubated with increasing concentrations of navitoclax, venetoclax, or A-1155463 for 48 hours before assessing cell viability. Cell killing EC50 values are listed for each compound against the cell lines examined.

View this table:

We performed similar parsing studies in a panel of 24 acute myeloid leukemia (AML) cell lines, 9 of which were sensitive to venetoclax (EC50 ≤ 0.1 μM) (Fig. 2C and Table 2), consistent with a recent report (25). Most AML cell lines (14 of 24) were more sensitive to venetoclax than to a BCL-XL–selective inhibitor. Cell lines bearing the JAK2 V617F mutation (UKE-1, SET-2, and HEL) were a notable exception, showing sensitivity to the BCL-XL–selective inhibitor but not to venetoclax. Again, the inhibition of both BCL-2 and BCL-XL was required for the effective killing of some cell lines, including KG-1 and SKM-1. Like BCL-2/BCL-XL co-dependent SCLC cell lines, KG-1 (Bliss sum = 892 ± 22) and SKM-1 (Bliss sum = 997 ± 154) showed synergistic sensitivity to the combination of venetoclax and A-1155463 (Fig. 2D). OCI-AML3, NOMO-1, and ME-1 cells were resistant to all three inhibitors, potentially implicating other BCL-2 family members in mediating their survival. Indeed, MCL-1 was shown to maintain the survival of OCI-AML3 when treated with ABT-737 or venetoclax (25).

Synergistic killing of cancer cell lines by the navitoclax-docetaxel combination is driven by BCL-XL inhibition

ABT-737 and navitoclax have been shown to synergize with taxanes in killing a variety of cancer cell lines (10, 2628). To dissect the contributions of BCL-2 and BCL-XL inhibition to the activity of this combination, we tested panels of breast cancer, non–small cell lung cancer (NSCLC), and ovarian cancer cell lines with combinations of docetaxel and navitoclax, venetoclax, or A-1155463. To assess potential combination activity, we used the Bliss independence model (2830), according to which negative integers indicate antagonism, a value of zero indicates additive activity, and positive integers indicate synergy. Bliss scores were calculated for each combination in the dose matrix and then totaled to give a “Bliss sum” (Table 3). For the purposes of these studies, Bliss sum values >150 were considered indicative of synergy. The navitoclax-docetaxel combination demonstrated synergistic killing of several breast cancer cell lines (fig. S2) and enhanced the induction of programmed cell death, as indicated by elevated caspase cleavage, poly(ADP-ribose) polymerase cleavage, and the accumulation of cells with sub–G0-G1 DNA content (fig. S3). The Bliss sum values for the A-1155463–docetaxel combinations showed a significant correlation with those of the navitoclax-docetaxel combinations when compared among breast cancer (r = 0.75, P < 0.0001, n = 28) (Fig. 3A) or NSCLC (r = 0.94, P < 0.0001, n = 15) (Fig. 3B) cell lines. However, venetoclax-docetaxel Bliss values showed no correlation to those calculated for navitoclax-docetaxel. In addition, the navitoclax-docetaxel and A-1155463–docetaxel combinations exhibited synergistic killing of four of six ovarian cancer cell lines tested (Bliss sum > 150), but no synergy was observed with the venetoclax-docetaxel combination (fig. S4). Collectively, these data indicate that BCL-XL is the key target of navitoclax for inducing the synergy observed with docetaxel in these solid tumor models.

Table 3. Bliss synergy assessment of cell killing by BCL-2 family inhibitors combined with docetaxel.

Breast cancer, NSCLC, and ovarian cancer cell lines were cultured in the presence of navitoclax, A-1155463, or venetoclax plus or minus docetaxel for 72 hours before assessing cell viability. The sum of Bliss scores across the combination dose matrix is listed for each cell line.

View this table:
Fig. 3. BCL-XL inhibition drives synergistic killing of solid tumor cell lines in combination with docetaxel.

(A and B) Panels of (A) breast cancer (n = 28) and (B) NSCLC (n = 15) cell lines were cultured in the presence of navitoclax, venetoclax, or A-1155463 plus or minus docetaxel for 72 hours before assessing cell viability. The sum of Bliss scores across the combination dose matrix was calculated for each cell line. Bliss sums were plotted for navitoclax-docetaxel combinations versus the venetoclax-docetaxel combinations or the A-1155463–docetaxel combinations. A significant correlation was observed between navitoclax-docetaxel and A-1155463–docetaxel Bliss sums for both the breast cancer (Spearman: 0.75, P < 0.0001) and NSCLC (Spearman: 0.94, P < 0.0001) cell line panels.

Orally bioavailable BCL-XL–selective inhibitor A-1331852 enhances the efficacy of docetaxel in vivo

We next assessed the ability of a selective BCL-XL inhibitor to enhance the efficacy of docetaxel in vivo. To this end, we used structure-based design to generate A-1331852 (Fig. 4A), a BCL-XL–selective inhibitor with oral bioavailability. A-1331852 is a potent BCL-XL inhibitor, binding BCL-XL with a Ki value of <0.010 nM and demonstrating cellular activity 10- to 50-fold more potent than A-1155463 and navitoclax, respectively (Table 1). This molecule selectively disrupts BCL-XL–BIM complexes and induces the hallmarks of apoptosis in BCL-XL–dependent Molt-4 cells with median inhibitory concentration (IC50) values in the low nanomolar range (Fig. 4, B to E, and Table 1) but does not affect MEF cells lacking BAK or BAX (fig. S5). Moreover, A-1331852 demonstrates antitumor efficacy in the Molt-4 xenograft model, inducing tumor regressions as a single agent (Fig. 4F). Additionally, A-1331852 combines with venetoclax to recapitulate the efficacy of navitoclax in the NCI-H1963.FP5 xenograft model of SCLC (Fig. 4G), thus providing in vivo confirmation of the combination studies shown in Fig. 2 (B and D).

Fig. 4. Orally bioavailable inhibitor A-1331852 enables the functional characterization of BCL-XL in vivo.

(A) The chemical structures and selectivity profiles of BH3 mimetics used for in vivo studies are depicted. A-874009 inhibits both BCL-2 and BCL-XL, whereas A-1211212 selectively inhibits BCL-2 and A-1331852 selectively inhibits BCL-XL. (B) Quantitative measurement of BCL-XL–BIM and BCL-2–BIM complexes in Molt-4 cells after 4-hour treatments with increasing concentrations of A-1331852. Data represent the average of triplicate experiments, with error bars indicating the SD. (C) Cytochrome c levels present in Molt-4 mitochondrial and cytosolic fractions as determined by immunoblotting after 4-hour treatments with increasing concentrations of A-1331852. (D) Caspase-3/7 activation in Molt-4 cells after incubation with increasing concentrations of A-1331852 for 4 hours. Data represent the average of triplicate experiments, with error bars indicating the SD. (E) Exposure of phosphatidylserine as determined by annexin V staining of Molt-4 cells after 1-hour preincubation plus or minus the caspase inhibitor Z-VAD-fmk (75 μM) and an additional 24 hours of incubation with increasing concentrations of A-1331852. Data represent the average of triplicate experiments, with error bars indicating the SD. (F) Mice bearing Molt-4 T cell acute lymphocytic leukemia–xenografted tumors were treated with a vehicle control (solid gray circles), venetoclax (daily for 14 days) at 100 mg/kg (open blue circles), or A-1331852 (twice a day for 14 days) at 25 mg/kg (open red circles). The points of each curve reflect the average volume of 10 tumors. Error bars indicate the SD of the means. (G) Mice bearing subcutaneous xenografts of NCI-H1963.FP5 were treated with vehicle control (solid gray circles), navitoclax (daily for 14 days) at 100 mg/kg (open green circles), venetoclax (daily for 14 days) at 50 mg/kg (open blue circles), A-1331852 (twice a day for 14 days) at 25 mg/kg (open red circles), or a combination of the latter two compounds (solid purple circles). The points of each curve reflect the average volume of five tumors. Error bars indicate the SD.

Inhibition of tumor growth by A-1331852 combined with docetaxel was determined in seven subcutaneous xenograft models of solid tumors, including breast cancer, NSCLC, and ovarian cancer. Given as a single agent, A-1331852 significantly (P < 0.05) inhibited tumor growth in all seven models (Table 4). Although its single-agent activity was modest (TGImax < 60% in five of seven models), A-1331852 increased the efficacy of docetaxel in all seven models. As shown in Table 4, the maximum tumor growth inhibition (TGImax) for A-1331852 as a single agent ranged between 34% (OVCAR-5) and 67% (A549-FP3). The most durable response to A-1331852 was a tumor growth delay (TGD) of 108%, observed in the A549-FP3 model. This indicates that the median time required for the tumors to reach a volume of 1 cm3 is about twice as long when treated with A-1331852 as compared to a sham-treated control. When comparing the combination to the most effective single-agent treatment, the increase in amplitude and durability of the response was statistically significant (P < 0.05) in five of seven models. The effect was most pronounced in the MDA-MB-231 LC3 metastatic breast cancer model (Fig. 5A) and the NSCLC models NCI-H1650 (Fig. 5B) and NCI-H358 (Table 4). Overall, the single-agent and combination treatments were well tolerated by mice, without overt signs of toxicity or weight loss of >9%. These data demonstrate that BCL-XL inhibition alone can enhance the efficacy of docetaxel in a variety of solid tumor models.

Table 4. Inhibition of tumor xenograft growth by administration of A-1331852, docetaxel, or the combination.

TGImax = 100 (1 − Tv/Cv), where Tv and Cv are the mean tumor volumes of the treated and control groups, respectively. TGD is the extended period of time that a treated tumor requires to reach a volume of 1 cm3 relative to the control group. TGD = 100(T/C − 1), where T and C are the median time periods required for the treated and control groups, respectively, to reach 1 cm3. Each treatment group of NCI-H747, A549-FP3, EBC-1, and OVCAR-5 models consisted of five mice. Each treatment group of the NCI-H358 model consisted of eight mice. Each treatment group of NCI-H1650 and MDA-MB-231 LC3 consisted of ten mice.

View this table:
Fig. 5. BCL-XL– selective inhibitor A-1331852 enhances the efficacy of docetaxel in vivo.

The growth inhibition of established tumors in SCID-bg mice is illustrated. Each graph describes the change of tumor volume (mm3, ordinate) as a function of the time after initiation of treatment (days, abscissa). Each point on the curves represents the mean volume of 10 tumors. Error bars depict the SD. A-1331852 was administered orally and docetaxel was administered intravenously for all studies. (A) MDA-MB-231 LC3 metastatic breast cancer xenograft. Vehicle control (solid gray circles), docetaxel (DTX) (once) at 7.5 mg/kg (open blue circles), A-1331852 (daily for 14 days) at 25 mg/kg (open red circles), and combination of docetaxel (once) at 7.5 mg/kg and A-1331852 (daily for 14 days) at 25 mg/kg (solid purple circles). (B) NCI-H1650 NSCLC xenograft. Vehicle control (solid gray circles), docetaxel (once) at 7.5 mg/kg (open blue circles), A-1331852 (daily for 14 days) at 25 mg/kg (open red circles), and combination of docetaxel (once) at 7.5 mg/kg and A-1331852 (daily for 14 days) at 25 mg/kg (solid purple circles).

BCL-2 inhibition suppresses granulocyte colony formation and decreases circulating neutrophil counts

Although the navitoclax-docetaxel combination is efficacious in preclinical studies, neutropenia has limited the dosing of this combination in the clinic (18). Using the toolkit of dual and selective inhibitors, we next asked if BCL-2 or BCL-XL inhibition alone might be sufficient to suppress granulopoiesis in colony-forming assays. Each compound was incubated with isolated human bone marrow cells cultured in semisolid medium over a period of 2 weeks. Venetoclax replicated the dose-dependent reduction in granulocyte colony formation caused by navitoclax, but the BCL-XL–selective inhibitor A-1155463 had little or no effect (Fig. 6A). Likewise, navitoclax and venetoclax showed greater inhibition of granulocyte colony formation than did A-1155463 when combined with docetaxel (Fig. 6B).

Fig. 6. BCL-2– selective inhibition suppresses granulopoiesis ex vivo and reduces circulating neutrophil counts in vivo.

(A) Isolated human bone marrow cells were used to perform granulocyte colony-forming assays in the presence of increasing concentrations of navitoclax (BCL-2/BCL-XL), venetoclax (BCL-2–selective), or A-1155463 (BCL-XL–selective). Neutrophil colonies were enumerated by light microscopy after 15 to 16 days of culture. (B) Neutrophil colony formation was assessed as in (A) after treatment of human bone marrow samples with increasing concentrations of navitoclax, venetoclax, or A-1155463 ± 5 nM docetaxel. Colonies were enumerated after 14 days of growth in culture. All data in (A) and (B) represent the means of triplicate experiments, with error bars indicating the SD. Student’s t tests compared test compound single-agent effects to solvent control samples. For combination experiments, test compounds in combination with docetaxel were compared to samples treated with docetaxel alone. (C) Groups (n = 10 per group) of male Sprague-Dawley rats were dosed with docetaxel (5 mg/kg, intravenously, once), the BCL-2/BCL-XL inhibitor A-874009 (30 mg/kg, orally, daily for 5 days), the BCL-2–selective inhibitor A-1211212 (50 mg/kg, orally, daily for 5 days), the BCL-XL–selective inhibitor A-1331852 (7 mg/kg, orally, daily for 5 days), or their respective vehicles. Groups were also dosed with combinations of docetaxel and the various BCL-2 family inhibitors. Box and whisker plots depict the means (n = 10) and ranges for total neutrophil and platelet counts assessed in blood collected on day 6. Tukey-Kramer tests compared vehicle- and compound-treated groups at a 5% significance level. Significant reductions in neutrophils or platelets relative to vehicle controls (P < 0.01) are denoted by asterisks.

To examine these effects in vivo, we evaluated orally bioavailable BCL-2 family inhibitors, including A-874009 (a close analog of navitoclax), A-1211212 (a close analog of venetoclax), and A-1331852 (Fig. 4A). Male Sprague-Dawley rats were dosed daily with each compound for 5 days, either alone or in combination with a single dose of docetaxel (5 mg/kg), before assessing their circulating platelet and neutrophil counts. Docetaxel decreased neutrophil counts significantly, with reductions between 61% (P < 0.01) and 71% (P < 0.01) relative to the mean of the vehicle control groups (Fig. 6C). Rats treated with the dual BCL-2/BCL-XL inhibitor A-874009 or the BCL-2–selective inhibitor A-1211212 had reduced neutrophils as well, in both cases showing a reduction of 41% relative to the respective vehicle-treated groups (P < 0.01). However, rats dosed with the BCL-XL–selective inhibitor A-1331852 showed no inhibition of granulopoiesis and exhibited increased neutrophil counts. When combinations of docetaxel with A-874009, A-1211212, or A-1331852 were compared to docetaxel alone, the differences in neutrophil inhibition did not reach statistical significance (P = 0.383, 0.981, and 0.773, respectively). The dual BCL-2/BCL-XL and BCL-XL–selective inhibitors both induced significant reductions in circulating platelets (P < 0.01), though the BCL-2–selective compound A-1211212 did not (Fig. 6C). This is consistent with BCL-XL maintaining platelet survival and suggests that the lack of neutrophil inhibition was not due to insufficient exposure to A-1331852.

These studies indicate that BCL-2 inhibition accounts for the exacerbation of docetaxel-induced neutropenia caused by navitoclax (18) and support the hypothesis that BCL-XL–selective inhibitors will avoid dose-limiting neutropenia in this setting. However, because BCL-XL–selective inhibitors will still affect platelets, a question remained whether efficacious levels of BCL-XL inhibition can be achieved in combination with docetaxel before thrombocytopenia becomes dose-limiting. We therefore analyzed data from phase 1 clinical trials of navitoclax, including single-agent studies in subjects with lymphoid malignancies (11) or solid tumors (31) and a navitoclax-docetaxel combination study in subjects with advanced solid tumors (18). No apparent pharmacokinetic (PK) interactions were observed between navitoclax and docetaxel in the latter study (18). Moreover, coadministration with docetaxel did not lead to substantial increases in navitoclax-induced thrombocytopenia relative to navitoclax treatment alone (fig. S6). However, the maximum tolerated dose (MTD) of navitoclax when given in combination with docetaxel was limited to 150 mg/day because of febrile neutropenia, which was observed at navitoclax exposures as low as 50.7 μg*hour/ml (AUC0–inf) (18) (table S1). This is in contrast to the single-agent setting, where MTDs of 315 or 325 mg/day were achieved before thrombocytopenia became dose-limiting (11, 31). These doses correspond to mean exposures of 80.5 ± 44.3 μg*hour/ml and 91.0 ± 33.5 μg*hour/ml (AUC0–inf), which overlap with highly efficacious exposures of navitoclax determined in preclinical studies (7) (Table 4). Because the PK of navitoclax is linear over the 150- to 325-mg dose range (11), these data indicate that large increases in navitoclax exposure could be obtained in combination with docetaxel if neutropenia were avoided. The discovery of BCL-XL–selective inhibitors like A-1331852 thus represents an opportunity to maximize BCL-XL inhibition and improve efficacy when dosed in combination with docetaxel.

DISCUSSION

The work that led to the discovery and functional characterization of BCL-2 family proteins relied on a variety of cell and molecular biology approaches, genetically engineered mouse models, and RNA interference–based methods. With the synthesis of the first validated BH3 mimetics, ABT-737 and navitoclax (6, 7), came the ability to inhibit BCL-2 and BCL-XL directly with cell-permeable small molecules, facilitating a host of new discoveries. Since its introduction, the BCL-2–selective inhibitor ABT-199/GDC-0199 (venetoclax) has been used to define the role of BCL-2 in specific hematopoietic lineages (32), a variety of hematologic cancers (25, 3338), and even some solid tumors, including estrogen receptor–positive breast cancer (39). In addition, certain roles for BCL-XL have been deduced using a subtractive parsing method that compares the effects of ABT-737 or navitoclax to those of venetoclax (38, 4042). For example, Bah and co-workers used ABT-737 and venetoclax to demonstrate that BCL-XL inhibition is required for the killing of breast cancer cell lines in combination with paclitaxel (41). Although this approach is informative, these inhibitors alone cannot distinguish between cell lines that are co-dependent on BCL-2 and BCL-XL for survival and those that rely solely on BCL-XL for survival. With the generation of the BCL-XL–selective inhibitors described here, it is possible to interrogate the role of BCL-XL directly in vitro and in vivo and to determine whether its selective inhibition is sufficient for a given effect. This is demonstrated by the studies in Fig. 2, which define the contributions of BCL-2 and BCL-XL in maintaining the survival of SCLC and AML cell lines. When used together, the small-molecule BH3 mimetics described here represent a powerful toolkit for dissecting the roles of BCL-2 and BCL-XL and for defining more effective therapeutic strategies. As these molecules find broader use in the research community, they should facilitate a more complete understanding of BCL-2 family biology and the roles these proteins play in normal tissues and disease states.

Antiapoptotic BCL-2 family proteins have been widely studied in cancer cells, where they are often overexpressed and maintain survival by sequestering large amounts of their proapoptotic counterparts, a state that has been referred to as being primed for death (5). An elegant peptide-based technique referred to as “BH3 profiling” has been used extensively to probe these interactions and to determine the profile of BCL-2 family dependence for a host of tumor cell types (43). One limitation of this method is the lack of highly selective BH3 peptides for targeting certain BCL-2 family members, which can make it difficult to fully define their roles. The discovery of selective BH3 mimetics like venetoclax, A-1155463, A-1331852, and cell-active MCL-1–selective inhibitors (44) will now enable a form of chemical BH3 profiling, whereby one can determine precisely which BCL-2 family members are required for the survival of any given cell population. The ability of these molecules to penetrate live cells also obviates the need for detergent-based permeabilization associated with peptide-based methods and enables one to probe BCL-2 family dependency by conducting simple cell viability assays. Furthermore, navitoclax, venetoclax, and A-1331852 are orally bioavailable and sufficiently potent to interrogate BCL-2 family member dependence in vivo, making them especially attractive translational tools.

Despite major investments into the discovery and development of small-molecule therapeutics, many compounds still fail in the clinic because of a lack of efficacy (45). In some cases, this may be due to off-target activities causing dose-limiting toxicities that preclude the achievement of efficacious exposures. This emphasizes the need to carefully dissect a molecule’s target inhibition profile and to better understand the roles each target plays in efficacy and toxicity. As demonstrated here, selective BCL-XL inhibitors can offer important advantages over their less selective predecessor, navitoclax, for the treatment of solid tumors. BCL-XL–selective inhibition enhanced the efficacy of docetaxel in breast cancer, NSCLC, and ovarian cancer models, but it did not inhibit granulopoiesis assessed ex vivo or in vivo. In contrast, BCL-2–selective inhibition suppressed granulopoiesis to a similar extent as was observed with dual BCL-2/BCL-XL inhibitors. Although Bcl-2 knockout mice show no overt signs of defective granulopoiesis before succumbing to polycystic kidney disease (46, 47), bone marrow from Bcl-2−/− reconstituted mice shows a reduced capacity for granulocyte colony formation, especially when cultured in the absence of cytokines such as interleukin-3 and stem cell factor (48). Moreover, neutropenia has been observed in a portion of CLL subjects receiving venetoclax (16).

Navitoclax clinical data indicated that, if exacerbated neutropenia in combination with docetaxel could be avoided, higher exposures and more complete inhibition of BCL-XL might be achieved before thrombocytopenia becomes dose-limiting. However, these analyses were limited by the small number of subjects (n = 7) with both exposure and hematology data available. In addition, our in vivo studies of BCL-XL–selective inhibitors were limited to rodent models, and so the possibility of encountering dose-limiting toxicities other than thrombocytopenia in the clinic cannot be ruled out. Nevertheless, the data presented here indicate that BCL-XL–selective inhibitors have the potential for improved safety and efficacy profiles, and provide further impetus for exploring this concept in the clinic.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/7/279/279ra40/DC1

Materials and Methods

Fig. S1. BCL-XL–selective inhibitor A-1155463 kills Mcl-1−/− MEF cells but not Bak−/− Bax−/− MEFs.

Fig. S2. Navitoclax synergizes with docetaxel to kill breast cancer cell lines.

Fig. S3. Navitoclax-docetaxel combination kills breast cancer cell lines by inducing apoptosis.

Fig. S4. Selective BCL-XL inhibition suffices for synergy with docetaxel in ovarian cancer cell lines.

Fig. S5. BCL-XL–selective inhibitor A-1331852 kills Mcl-1−/− MEF cells but not Bak−/− Bax−/− MEFs.

Fig. S6. Relationship between exposure and platelet reduction is similar for navitoclax with or without docetaxel.

Table S1. Plasma exposures and platelet effects of navitoclax-docetaxel combinations.

References (4951)

REFERENCES AND NOTES

Acknowledgments: We thank J. Damen and M. Huber of StemCell Technologies for technical advice and performing granulocyte colony-forming assays. We thank D. Sheinson for performing statistical analyses. We also thank M. Mabry for helpful discussions and coining the phrase “chemical parsing.” Funding: Work in the Huang Lab is supported by grants and fellowships from the Australian National Health and Medical Research Council (research fellowships to D.C.S.H.; project grants to D.C.S.H.; program grants 461219, 461221, and 1016701; and Independent Research Institutes Infrastructure Support Scheme grant 361646), the Cancer Council Victoria (grant-in-aid to D.C.S.H.), the Leukemia and Lymphoma Society (Specialized Centers of Research grants), the Australian Cancer Research Foundation, and a Victorian State Government Operational Infrastructure Support grant. Author contributions: J.D.L. designed experiments, oversaw biology experiments, and wrote the manuscript. D.C.P. designed and performed breast cancer experiments in vitro, oversaw biology experiments, and wrote the manuscript. M.J.M., A.O., and T.J.M. performed in vivo efficacy experiments. K.S.V., E.R.B., D.H.A., and D.S. designed, oversaw, and analyzed in vivo efficacy experiments. D.D. designed and oversaw rat studies. J.M.T. and N.L. performed rat studies. L.D.B. designed and oversaw NSCLC experiments in vitro. K.N.L. performed ovarian cancer studies in vitro. P.K. performed biochemical affinity assessments of all compounds. Y.X. performed breast cancer experiments in vitro. X.M.M. performed AML experiments in vitro. P.N., S.J., J.X., J.C., and H.Z. performed biological characterizations of BCL-XL–selective inhibitors in vitro. M.S. and S.K.T. designed colony-forming experiments and performed SCLC experiments in vitro. L.W., Z.-F.T., and M.D.W. designed and synthesized compounds. C.T., D.C.S.H., and W.J.F. oversaw biology efforts. S.H.R. and S.W.E. oversaw chemistry and biology efforts. A.J.S. designed compounds, oversaw chemistry and biology efforts, and wrote the manuscript. All authors contributed to the interpretation of data and to the review and editing of the manuscript. Competing interests: J.D.L., D.C.P., M.J.M., E.R.B., J.C., P.N., S.K.T., J.X., P.K., M.S., A.O., T.J.M., K.S.V., D.H.A., Y.X., H.Z., L.W., Z.T., M.D.W., C.T., S.H.R., S.W.E., and A.J.S. are employees of AbbVie and hold company stock. D.D., L.D.B., J.M.T., N.L., D.S., and W.J.F. are employees of Genentech and hold company stock. Financial support for this research was provided by AbbVie and Genentech. AbbVie, Genentech, and Walter and Eliza Hall Institute of Medical Research participated in the design and conduct of studies, interpretation of data, and review and approval of the publication. Patents cover all of the small molecules described in this study. Data and materials availability: ABT-263 (navitoclax), ABT-199 (venetoclax), A-874009, A-1155463, A-1211212, and A-1331852 may be obtained through appropriate material transfer agreements. Please address compound requests to joel.leverson{at}abbvie.com.
View Abstract

Navigate This Article