Anti-cancer drug discovery and development

Abstract

Deregulated apoptosis is a hallmark of cancer, and the B-cell lymphoma-2 (Bcl-2) family of proteins is pivotal to mediating the intrinsic pathway of this process. Recent advances have yielded both pan-Bcl-2 small molecule inhibitors (SMIs) that inhibit both the Bcl-2 and the Mcl-1 arm of the Bcl-2 family anti-apoptotic proteins, as well as selective SMIs to differentially target the two arms. Of these SMIs, ABT-263 (navitoclax), AT-101 [(-)-gossypol], and obatoclax (GX15-070) are currently in clinical trials for multiple cancers. While pan-Bcl-2 inhibitors such as AT-101 and obatoclax can be more toxic for inhibiting all members of the anti-apoptotic Bcl-2 family of proteins, resistance can quickly develop for ABT-263, a selective Bcl-2 inhibitor. In this article, we discuss the current status of Bcl-2 family SMIs in preclinical and clinical development. As Mcl-1 upregulation is a major mechanism of ABT-263 resistance, Mcl-1-specific inhibitors are expected to be efficacious both in combination/sequential treatments and as a single agent against cancers resistant to ABT-263. 

Keywords: :

• Drug discovery
• Bcl-2
• Mcl-1
• apoptosis
• cancer

Introduction

Apoptosis is an orchestrated cellular process important for maintaining the homeostasis between cell proliferation and cell death, and is also pivotal for the removal of diseased, damaged, and inflamed cells.¹ Apoptosis can be induced in two distinct pathways, both of which lead to the activation of effector caspases. The extrinsic pathway of apoptosis is initiated at the plasma membrane upon ligation of death receptors, leading to the subsequent activation of effector caspases. These caspases then execute the apoptotic pathway to trigger DNA fragmentation and membrane blebbing, two hallmarks of apoptosis.² The intrinsic pathway is initiated in response to intracellular stress, eventually leading to mitochondrial outer membrane permeabilization (MOMP), releasing factors including cytochrome c and Smac to activate downstream caspases.³

The B-cell lymphoma-2 (Bcl-2) family of proteins is central to the intrinsic pathway of apoptosis. The Bcl-2 proteins share one of four Bcl-2 homology (BH) domains, BH1–4, and of those, the BH3 domain is critical for mediating interactions among the family members.^4,5 The Bcl-2 family of proteins can be grouped as either pro-apoptotic or anti-apoptotic. The pro-apoptotic group can be further divided as either multi-BH-domain proteins, including Bax and Bak, or as BH3-only proteins, such as Bid, Bad, Bim, Puma, and Noxa. The anti-apoptotic group of Bcl-2 proteins consists of Bcl-2, Bcl-X_L, Bcl-w, Mcl-1, and Bfl-1/A1.⁶ Two models account for how the various proteins interact to induce apoptosis (Fig. 1). In the indirect activation model, the anti-apoptotic Bcl-2 proteins physically interact and sequester Bak or Bax. Upon apoptotic stimuli, BH3-only proteins release Bak or Bax by binding and neutralizing the anti-apoptotic Bcl-2 proteins. In the direct activation model, anti-apoptotic Bcl-2 proteins function by sequestering particular BH3-only proteins called BH3 activators, including Bim, Bid, and Puma. Upon apoptotic stimuli, BH3 sensitizers such as Noxa or Bad liberate the activators, allowing them to activate Bak and Bax.^4,7 The two models converge when activated Bak and Bax oligomerize, leading to MOMP and subsequent apoptosis. Therefore, the carefully regulated balance between pro-apoptotic and anti-apoptotic members of the Bcl-2 family determine the survival of the cell. It has repeatedly been shown that the overexpression or loss of pro- or anti-apoptotic members of the Bcl-2 proteins render cells correspondingly susceptible or resistant to apoptosis.^3,8-11 Therefore, deregulated apoptosis can be detrimental to the cell, tissue, or organ.

Figure 1. Regulation of apoptosis by the Bcl-2 family. The anti-apoptotic Bcl-2 protein family members include Bcl-2, Bcl-X_L, Bcl-w, Mcl-1, and Bfl-1/A1. In the indirect activation model, the anti-apoptotic proteins sequester Bax and Bak through physical interactions. The upregulation of BH3-only proteins will disrupt this interaction, releasing Bax and Bak to oligomerize, allowing mitochondrial outer membrane permeabilization (MOMP), the release of key factors such as cytochrome c, and the induction of apoptosis. In the direct activation model, anti-apoptotic Bcl-2 protein members sequester BH3 activators including Puma, Bid, and Bim. BH3 sensitizers are able to disrupt this interaction, and the released activators directly activate Bax and Bak to allow MOMP and apoptosis. It is most likely that these two models happen in tandem.

Deregulated apoptosis is a hallmark of cancer.¹² The anti-apoptotic group of Bcl-2 family proteins is frequently found to be overexpressed in numerous cancers, causing both evasion of apoptosis and resistance to treatment.^3,13-18 Furthermore, since most anti-cancer drugs induce apoptosis through the intrinsic pathway, targeting the Bcl-2 family of proteins with small molecule inhibitors (SMIs) is especially attractive clinically either in single agent therapy or combination treatment.² While numerous compounds with anti-cancer activity have been shown to antagonize Bcl-2 family of proteins, such as epigallocatenin-3-gallate (EGCG) and chelerythrine, SMIs selective for the anti-apoptotic Bcl-2 proteins have only recently been identified.^19-24 The first Bcl-2 selective SMI, HA14–1, was identified through a virtual screen in 2000. Due to the recent advances in crystallizing Bcl-X_L and its complex with the Bak BH3 domain peptide, a hydrophobic pocket that interacts with other BH3 domains was identified and taken advantage of to aid drug design.^25,26 The Bcl-2 protein structure was then modeled and HA14–1 was discovered through a computer-aided screen.²⁷ Following this, the fluorescence polarization assay (FPA) was soon developed to allow experimental high-throughput screening for SMIs of anti-apoptotic family proteins. This assay uses fluorescein-conjugated BH3 peptides and recombinant anti-apoptotic Bcl-2 proteins.²⁸ Using this method, several pan-Bcl-2 inhibitors with moderate affinity were discovered, including BH3–1, gossypol, and obatoclax.^23,29,30 Recently, structure-based design has also emerged to synthesize numerous high-affinity Bcl-2 SMIs. Using the NMR-guided fragment-based technology, low-affinity molecules, which bind to distinct sites on the hydrophobic groove of the Bcl-X_L protein, were linked together to generate the high-affinity antagonist ABT-737.²² Further rational design approaches have been successful at generating ABT-737 based high-affinity Bcl-2 SMIs, including compounds B-11 and compound 21.^31,32 Moreover, the rational design of terphenyl-based Bak-BH3 α-helical proteomimetics resulted in the discovery of BH3-M6, a pan-Bcl-2 inhibitor that can inhibit both Bcl-X_L and Mcl-1 to induce caspase dependent apoptosis.³³ Other developmental efforts have also been fruitful, resulting in the discovery of the selective Mcl-1 antagonist maritoclax, and the high-affinity pan-Bcl-2 inhibitor sabutoclax.^24,34 Currently, three Bcl-2 inhibitors are undergoing clinical trials: ABT-737’s orally bioavailable analog ABT-263 (navitoclax), AT-101 [(-)-gossypol], and obatoclax (GX15–070).

Selectivity of Bcl-2 Inhibitors

Numerous compounds have been specifically developed or identified as Bcl-2 inhibitors. These compounds include ABT-737 and ABT-263, obatoclax, gossypol derivatives such as apogossypol, apogossypolone, and sabutoclax, other rational design-based compounds such as B-12 and BH3-M6, and the recently identified maritoclax.^{22-24,31,34-37} However, the selectivity and affinities of these compounds for the anti-apoptotic Bcl-2 family members differ greatly among the different compounds as well as between different assays (Fig. 2). The difficulty in obtaining precise values for the affinity of Bcl-2 SMIs can be attributed to the affinity differences among the peptides used for the FPA. For example, ABT-737’s apparent affinity was 4-fold lower for Bcl-2, while 50-fold higher for Bcl-w, when using a BID-peptide compared with using a BIM-peptide.^21,38 Therefore interpretations of affinity data for Bcl-2 SMIs should be taken with a grain of salt.

Figure 2. The selectivity of Bcl-2 SMIs. For each drug, green indicates relatively high affinity for the protein, orange indicates relatively low affinity for the protein, and red indicates no affinity for the protein was detected. Black indicates a lack of published data. The colors assigned were based on the drug’s affinities relative to the affinity of ABT-737 against Bcl-2, as well as on the relative affinities between the drug’s targets.

The selectivity of the Bcl-2 SMIs for the anti-apoptotic members differ as well. While agents such as obatoclax and AT-101 are classified as pan-Bcl-2 inhibitors due to its moderate affinity for all anti-apoptotic members, ABT-737 and its analog ABT-263 are selective for Bcl-2, Bcl-X_L, and Bcl-w as they are modeled after the BH3-only protein Bad.²² It is also the most potent of the inhibitors so far possessing affinities in the sub-nanomolar range. Other selective inhibitors have also been recently developed. A-385358 was rationally designed to be almost 100-fold more selective for Bcl-X_L over Bcl-2, and is the most selective inhibitor for Bcl-X_L to date. A-385358 was able to more potently kill Bcl-X_L-dependent compared with Bcl-2-dependent lymphoma cells.³⁹

Maritoclax, also known as marinopyrrole A, was recently discovered as a Mcl-1-specific inhibitor through the screening of small compound libraries. Maritoclax is a natural product from a marine-derived species of Streptomyces⁴⁰ which binds to Mcl-1 with comparable affinity as obatoclax; however, unlike obatoclax, maritoclax does not bind to Bcl-X_L. Maritoclax induces Mcl-1 proteasomal degradation and shows similar or greater in vitro efficacy toward Mcl-1 dependent cells as obatoclax. However, in cells dependent on Bcl-2 or Bcl-X_L for survival, maritoclax demonstrates only marginal activity. Importantly, unlike obatoclax, which induces cell death indiscriminately in healthy peripheral blood mononuclear cells (PBMCs) and large granular lymphocyte leukemia (LGLL), maritoclax is significantly less effective against healthy PBMCs compared with LGLL cells. This suggests that maritoclax may be safer for treatment compared with pan-Bcl-2 inhibitors.²⁴ Interestingly, maritoclax has been shown to bind actin in addition to Mcl-1.⁴¹ However, maritoclax induces cell death selectively in Mcl-1 dependent cells in a Bax/Bak dependent manner, indicating that maritoclax acts through the intrinsic pathway of apoptosis. Moreover, there is a positive correlation between the sensitivity to maritoclax and Mcl-1 expression in primary LGLL cells.²⁴ Lastly, actin is known to bind to and sequester the pro-apoptotic BH3-only protein Bmf, which is released to sensitize the cell to apoptosis upon actin depolymerization.^42,43 Since Bmf interacts with not only Mcl-1 but also Bcl-2 and Bcl-X_L, liberation of Bmf from actin should induce cell death in both Mcl-1 and Bcl-2/Bcl-X_L dependent cells.⁴² However, given that maritoclax only kills Mcl-1 dependent cells, it is unlikely that maritoclax induces apoptosis in a mechanism mediated by actin.

Since it is difficult for SMIs with micromolar affinities to reach physiologically relevant concentrations in vivo, such compounds are often deemed to be useful only as experimental tools. However, some compounds with moderate affinity, such as obatoclax and AT-101, have already demonstrated efficacy against cancer in clinical trials.^44,45 Therefore, it is possible that these agents may act as non-selective Bcl-2 inhibitors, killing cancer cells through multiple pathways. In an important study by Vogler et al., it was shown that between obatoclax, gossypol, and ABT-737, only ABT-737 was unable to kill Bax/Bak double knockout cells.⁴⁶ Since Bax/Bak oligomerization is necessary for Bcl-2 mediated apoptosis, the ability of obatoclax and gossypol to induce cell death in these cells signifies its actions on other cellular pathways. It was similarly shown that obatoclax does not induce cytochrome c release from isolated mitochondria at physiologically relevant levels.⁴⁷ Given the inhibitory effect of Bcl-2 on Beclin-1, the mammalian homolog of the autophagy related gene 6 (Atg6), Bcl-2 SMIs often induce autophagy by disrupting Bcl-2 and Beclin-1 interactions.^48-52 However, numerous reports have demonstrated Beclin-1 independent induction of autophagy by gossypol and obatoclax, possibly causing autophagic cell death independent of Bcl-2 interactions.^50,53 Other reports have also shown obatoclax and gossypol to possess activities unrelated to Bcl-2 which may contribute to cell killing, such as reactive oxygen species generation or anion transport.^54-57 Therefore, ABT-263 is possibly the only compound in clinical trials to selectively induce Bcl-2 dependent cell death.

Clinical Relevance of Bcl-2 Inhibitors

All three Bcl-2 SMIs in clinical trials, ABT-263, AT-101, and obatoclax, demonstrate remarkable preclinical efficacy in a variety of cell systems (Figure 3). Notably, these Bcl-2 SMIs kill chemoresistant cell lines as efficaciously as susceptible cell lines, demonstrating a distinct mechanism of action and clinical promise.^58-63

Figure 3. The potency of Bcl-2 SMIs currently in clinical trials against various cancer cells in culture. Green indicates high potency for the cancer type with an average EC₅₀ value below 1 µM, orange indicates a moderate potency with an average EC₅₀ value between 1 and 10 µM, and red indicates a low potency with an average EC₅₀ value above 10 µM. Black indicates that no published EC₅₀ exists for that drug and cancer type. The white numbers inside the box signify the number of experiments examined to obtain the average.

Bcl-2 inhibitors also synergize with a large number of other cancer treatments. All three Bcl-2 SMIs currently in clinical trials have shown to potentiate and synergize with chemotherapy and radiation in vitro.^22,64-72 Interestingly, mitogen-activated protein kinase (MAPK) pathway inhibitors, such as the epidermal growth factor receptor (EGFR) inhibitor gefitinib and erlotinib or the MEK inhibitor UO126 synergize with Bcl-2 SMIs.^73-76 While the MAPK pathway is frequently upregulated in cancers and is therapeutic target by itself, synergy between Bcl-2 inhibitors and MAPK pathway inhibitors may actually be due to crosstalk between the two pathways. It has been shown that EGFR activation leads to Stat3 activation, which in turn increases the expression of Bcl-2 and Bcl-X_L.^77,78 Therefore, the inhibition of the MAPK pathway reduces the expression of Bcl-2 and Bcl-X_L, while Bcl-2 inhibitors physically antagonize their anti-apoptotic actions to create a synergistic effect. Recently, ABT-737 has also been shown to possess immunosuppressive properties, possessing activity on lymphocytes and mast cells.^79-82

Due to the selectivity of ABT-737 on specific Bcl-2 family protein members, resistance can develop rapidly. As ABT-737 disrupts Bcl-2/Bim interactions more readily compared with Bcl-X_L/Bim, Bcl-X_L overexpressing cancer cells can be resistant to ABT-737 treatment.^83,84 The primary pathway for ABT-737 resistance appears to be Mcl-1 upregulation, as it is not targeted by the compound. It was repeatedly observed in a large number of cancer cell lines that Mcl-1 expression levels inversely correlate with ABT-737 susceptibility.^{38,59,65,85-90} Human B-cell lymphoma cell lines derived from long-term exposure to ABT-737 were resistant to its treatment through a significant increase in Mcl-1 transcription and protein levels. The use of flavopiridol, a cyclin-dependent kinase 9 (CDK9) inhibitor shown to downregulate Mcl-1, was able to restore ABT-737 sensitivity.⁹¹ Furthermore, melanoma, which is typically resistant to ABT-737 treatment, can be rendered susceptible to the compound through genetic or non-selective pharmacologic inhibition of Mcl-1 in multiple reports.^92-95 Obatoclax has demonstrated significant antagonistic activity against Mcl-1, and has been used experimentally to demonstrate Mcl-1 mediated ABT-737 resistance as well.^96-98 As the induction of the MAPK pathway also upregulates Mcl-1, the synergy between ABT-737 and MAPK pathway inhibitors can be due to its downregulation of Mcl-1.⁹⁹ Following the development of the first selective Mcl-1 inhibitor, maritoclax was immediately shown to be efficacious as a single agent in ABT-737 resistant, Mcl-1 dependent cell lines, and was strongly synergistic with ABT-737.²⁴ Overexpression of Noxa in multiple cancer systems have shown to restore drug susceptibility that is similar to Mcl-1 downregulation, mechanistically demonstrating that sequestration of pro-apoptotic Bcl-2 proteins by Mcl-1 can mediate drug resistance.^94,100-102 Recently it was also found that Mcl-1 can also be phosphorylated in response to ABT-737 to increase its interaction with Bim to prevent apoptosis.¹⁰¹

The in vivo activities of these Bcl-2 SMIs were similarly promising. ABT-737 was shown to be extremely efficacious as a single agent in SCLC mouse xenograft models, as the compound was able to induce complete regression of tumors and prevent its regrowth for at least 2 months after the cessation of therapy.²² ABT-737 by itself has also been shown to be effective in xenograft models for glioblastoma, multiple myeloma, acute myeloid leukemia, chronic myeloid leukemia, acute lymphoblastic leukemia, among others.^{38,65,103-105} Furthermore, the pan-Bcl-2 inhibitor AT-101 was shown to be efficacious in prostate cancer xenografts in different cell line models, resulting in tumor growth reductions.^106,107 In another study, however, AT-101 was ineffective in a similar prostate cancer xenograft model, but was active in potentiating the effects of docetaxel.¹⁰⁸ Obatoclax has also been shown to potentiate other cancer treatments in xenograft models.^109,110

All Bcl-2 SMIs in clinical trials have published Phase II results. ABT-263 (navitoclax), which is being developed by Abbott Laboratories and Genentech, was shown to be the most efficacious against chronic lymphocytic leukemia (CLL) in Phase I trials, with all 7 patients in one trial and 9 out of 21 patients with relapsed or refractory disease experiencing a partial response with median progression-free survival of more than a year.^111,112 It was observed in these patients that Mcl-1 expression inversely correlated with response to drug, matching previous preclinical observations.¹¹² The recently published Phase II clinical trial data for ABT-263 indicate that it lacks efficacy in patients with relapsed small cell lung carcinoma (SCLC), as only 1 out of 39 patients experienced a partial response, and 36 patients withdrew due to disease progression or adverse events.¹¹³ Currently multiple clinical trials are still also underway to assess the efficacy of ABT-263 in combination treatments.¹¹⁴

Since Bcl-X_L is important for platelet survival, thrombocytopenia was observed across all clinical trials for ABT-263.^115,116 However, the platelet decreases were all observed to be transient and reversible upon treatment termination, and has not been associated with clinically significant bleeding.^111-113,117 Other hematological toxicities including neutropenia have also been observed, and could present significant problems when treating patients with hematological malignancies with an already compromised bone marrow reserve.¹¹²

AT-101 is in the pipeline for Ascenta Therapeutics and has only demonstrated moderate efficacy in one Phase I/II clinical trial against castration-resistant prostate cancer. In this trial, two of 24 patients experienced a partial decrease in prostate-specific antigen release, while no objective partial responses were observed. However, 4 individuals unexpectedly developed small intestinal obstructions possibly related to the drug, causing discontinuation of treatment.⁴⁴ In a following larger clinical trial in combination with docetaxel and prednisone, however, efficacy was not met against castration-resistant prostate cancer.¹¹⁸ AT-101 did not manage to meet efficacy endpoints both as a single agent against nor in combination with topotecan or docetaxel/prednisone against SCLC.^119-121 Gastrointestinal toxicities were observed across all clinical trials with single-agent trials, but were not increased in combination treatments.

Obatoclax is developed by Gemin X Pharmaceuticals. As it is not orally bioavailable, treatment must be administered through infusion. Unlike ABT-263, obatoclax demonstrated moderate to no efficacy against hematological malignancies in Phase I trials. However, in these clinical trials where obatoclax was either infused for one hour or three hours, significant neural adverse events were observed, including somnolence, euphoria, and ataxia. It was observed that a three-hour infusion was necessary to reduce the occurrence of toxic events.^45,122-124 In a phase I trial, obatoclax demonstrated significant efficacy against SCLC as a single agent, where 2 of 8 patients with SCLC had partial response and 3 experienced stable disease.¹²⁵ Moreover, another Phase I clinical trial in combination with carboplatin and etoposide against extensive-stage SCLC was conducted. Obatoclax demonstrated efficacy in combination treatments to improve outcomes.¹²⁶ However, obatoclax in combination with topotecan against relapsed SCLC did not meet efficacy endpoints.¹²⁷ Another Phase II trial also indicates obatoclax to be ineffective toward myelofibrosis, despite the disease’s dependence on the Bcl-2 family anti-apoptotic proteins.¹²⁸

Future Perspectives

As clinical trial data for Bcl-2 SMIs continue to be published, we will likely gain more insight on the efficacy of these inhibitors against cancer. So far, clinical trials show promise for ABT-263 against CLL, and obatoclax in combination treatments against SCLC. However, numerous clinical trials have demonstrated a lack of efficacy where it was expected, such as AT-101 against prostate cancer. A higher affinity pan Bcl-2 inhibitor sabutoclax (BI-97C1) has recently been developed, and may yield clinical promise as it possesses nanomolar affinities for all of Mcl-1, Bcl-2, and Bcl-X_L proteins, and induces cell death in a Bax/Bak-dependent manner. Moreover, it has demonstrated both in vitro and in vivo xenograft model efficacy against Bcl-2 dependent and Mcl-1 dependent prostate cancer cell lines.³⁴ Currently, research efforts have began focusing on different dosing regimens using Bcl-2 SMIs to improve efficacy.¹²⁹ Metronomic dosing, the uninterrupted use of a low dose of compound, have been suggested for Bcl-2 inhibitors including AT-101 and TW-37.^130,131 It is hypothesized that the metronomic dosing of Bcl-2 inhibitors is able to more effectively induce growth arrest by targeting the tumor microenvironment. Sequential dosing of multiple therapies also shows preclinical promise. The sequential use of bortezomib and HA14–1 was more efficacious than both single agent and combination treatment regimens in in vitro models of multiple myeloma.¹³² Sequential dosing may be particularly useful for ABT-263, as it cannot be administered for extended periods of time due to its severe toxicity on platelets. Moreover, both preclinical and clinical data have indicated the importance of Mcl-1 mediated ABT-263 resistance. As maritoclax is strongly synergistic with ABT-737, the sequential use of ABT-263 with Mcl-1 selective inhibitors may be particularly useful both to prevent the development of resistant cell populations and to allow the recovery of platelets in patients.²⁴

High-affinity pan Bcl-2 inhibitors could cause much toxicity in the patient, as numerous healthy cells depend on the maintenance of anti-apoptotic Bcl-2 family members for survival. Thus selective Bcl-2 SMIs demonstrate a distinct advantage in drug safety, although selectivity can also lead to resistance, as was seen in the case of ABT-263 and ABT-737. As simultaneous treatment of patients with multiple Bcl-2 inhibitors is imprudent due to toxicity reasons, it may be more sensible to use sequential treatments in order to minimize undesired side effects.

Maritoclax is expected to demonstrate single-agent efficacy in addition to combination therapies. It is known that the downregulation of Mcl-1 itself is efficacious against numerous cancer cells to mediate both cell killing and drug sensitivity.^110,133-136 Importantly, experiments show that multiple cancers, including multiple myeloma and acute myeloid leukemia, are highly dependent on Mcl-1 to sustain growth.^134,137,138 As ABT-737 does not target Mcl-1, these cancer cell lines are resistant to ABT-737 killing. For example, downregulation of Mcl-1 is able to completely kill acute myeloid leukemia cells in both cell culture and in vivo mouse models.¹³⁴ Following optimization, more Mcl-1 inhibitors with higher selectivity and affinity are likely to be generated for clinical use.

The Bcl-2 family proteins play a significant role in maintaining survival for a large number of cancers. SMIs for the anti-apoptotic proteins have already demonstrated that cancer cells can be killed through this distinct mechanism of action despite their resistance status to previous therapies. These inhibitors have demonstrated efficacy both as single agents and in combination therapy to improve patient outcome. However, selectivity, toxicity, resistance, and bioavailability remain a challenge for many of these compounds. Further efforts should focus on reducing undesired toxicities from these drugs through dosing or combination therapies, as well as chemical advances in synthesizing more selective Bcl-2 inhibitors.

Acknowledgments

We thank Yoshinori Takahashi and Megan Young for critical reading of the manuscript. This work is supported by a NIH/NCI grant (CA82197).