The BH-3 only group of pro-apoptotic Bcl-2 proteins consists of Bim (BCL2L11), Puma/BBC3, Bad (Bcl-2/Bcl-x-associated death promoter), Bid (BH-3 interacting-domain death agonist), Bik (Bcl-2-interacting killer), Noxa/PMAIP1, Bmf (Bcl-2-modifying factor), and Hrk (Harakiri) [Figure 1A, (70)] and are essential for initiating the apoptotic cascade. While the BH-3 only proteins can initiate apoptotic signaling by binding directly to the anti-apoptotic Bcl-2 proteins, thereby freeing up Bax and Bak to undergo homo-dimerization, some have been reported to bind directly to and activate Bax and Bak (71). Years of studies using mouse models have revealed certain BH-3 only proteins are preferentially solicited in response to different apoptotic stimuli (72–76).

BH-3 only proteins serve as the first responders to cellular insults, including from dysregulation of oncogenes, such as Myc, and serve as a blockade against the development of B cell lymphoma. For example, loss of Bim or Puma accelerated Myc-driven B cell lymphoma development in a mouse model engineered to overexpress Myc in B cells (Eμ-myc transgenic) (11, 15, 16). Loss of BIM may also contribute to human lymphomas, as ~20% of mantle cell lymphomas (MCL) have deleted both alleles of BIM (12). In addition, single nucleotide polymorphisms in the BIM gene have been associated with risk of developing follicular lymphoma (FL), diffuse large B cell lymphoma (DLBCL), and chronic lymphocytic leukemia (CLL) (13). Furthermore, ~40% of human Burkitt lymphomas express very low levels of BIM or PUMA mRNA, which may be the result of epigenetic silencing (14, 15). In contrast to Bim and Puma, loss of Noxa had no effect on Myc-induced B cell lymphomagenesis in mice, but did increase the number of B lineage cells (16). It is unknown whether NOXA loss contributes to human B cell lymphoma. A quarter of mice with deletion of Bad develop DLBCL at old age, suggesting that it may have a tumor suppressive function in mature B cells (18). Deletion of Bad also accelerated Myc-induced B cell lymphoma (17). Despite the findings in mice, BAD loss has not been linked to DLBCL in humans. Deletion of Bid did not result in B cell lymphoma development in mice. Instead, chronic myelomonocytic leukemia emerged in Bid-null mice after a long latency period, indicating Bid function is critical for the myeloid lineage (19). It is unknown if alterations in BID contribute to human lymphoma. In mice, loss of Bik alone had no effect on hematopoietic cells and did not accelerate Myc-induced B cell lymphoma development, suggesting that it has no role in B cells (20, 21). However, somatic missense mutations have been observed in BIK in B cell lymphomas in humans, including FL, marginal zone (MZL), and DLBCL (22), suggesting that its loss may contribute to these lymphomas. Loss of Bmf in mice increased B cell numbers and cooperated with Myc overexpression to accelerate lymphomagenesis; preferentially developing an IgM+ B cell lymphoma (17). Reduced levels of BMF were observed in Burkitt lymphoma patient samples and cell lines (17). Together, the data indicate that each pro-apoptotic BH-3 only family member may facilitate tumor suppression in specific hematopoietic cells, but not all may have a role in B cells.

Activation of Bax and Bak, and possibly the lesser-studied Bok, involves homo-dimerization followed by oligomerization within the outer mitochondrial membrane (71). This conformational change induces mitochondrial outer membrane permeabilization (MOMP) by creating a pore and triggering the release of apoptosis-inducing proteins, such as cytochrome c and second mitochondria-derived activator of caspase (Smac)/direct IAP-binding protein with low pI (DIABLO) from the mitochondria [Figure 1A, (77)]. Following this release, the caspase cascade is activated, ultimately resulting in the proteolysis of intra-cellular proteins and cellular destruction.

The combined function of Bax and Bak is critical for development and mediating apoptosis. Mice lacking both Bak and Bax have severe developmental defects and have hematopoietic cells that are resistant to diverse stimuli that activate the intrinsic apoptotic pathway (23, 78). Mice lacking either Bak or Bax are phenotypically normal (23) or have mild lymphoid hyperplasia, respectively (24). Using the Eμ-myc mouse model of Myc-driven B cell lymphomagenesis, loss of Bax accelerated lymphoma development (25), but the effects of inactivation of Bak in Myc-induced B cell lymphoma have not been reported. Loss of Bok did not alter Myc-induced B cell lymphoma development (26). In humans, mutations in BAX that caused a frameshift were detected in cell lines derived from hematologic malignancies that did not include lymphoma and were associated with resistance to cell death and microsatellite instability (79–81). Alterations of BAX, BAK, or BOK have not been reported in human lymphoma; thus, loss/inactivation of BAX, BAK, or BOK either does not occur or is a rare event in human B cell lymphoma.

BCL-2 is translocated t(14;18)(q32;q21) to the immunoglobulin heavy chains, resulting in its constitutive expression in 90% of FL (29, 86–90). Bcl-2 knockout mice showed that Bcl-2 is required for normal B cells to survive (27), providing evidence for why B cell lymphomas would select for its overexpression. Somatic mutations in BCL-2 in FL are often associated with transformation of this indolent disease to more aggressive diffuse large B cell lymphoma (DLBCL) and decreased patient survival (31). Some of these mutations increased the affinity of BCL-2 to pro-apoptotic BH-3 only proteins and have also been detected in lymphoid cell lines (91, 92).

Approximately 20% of de novo DLBCL also harbor BCL-2 translocations (30). Increased BCL-2 expression has been linked to reduced survival of patients with DLBCL (93, 94). The first large-scale gene expression profiling studies classified DLBCL into two major molecular subgroups, germinal center B cell (GCB) and activated B cell (ABC) subtypes (95). GCB DLBCL that have increased levels of BCL-2 is most often due to a t(14;18) translocation (96), whereas in ABC DLBCL, amplification of the BCL-2 gene is more often observed (93, 97). Using gene expression profiling, Iqbal and colleagues reported a significant correlation between elevated BCL-2 mRNA expression and poor overall survival within the ABC subtype (93). We also observed elevated levels of BCL-2 in both subtypes of DLBCL, but BCL-2 was more highly expressed in the ABC subtype than the GCB subtype (9). More recently, new classifications of DLBCL subtypes have been reported by two groups (98, 99). Shipp and colleagues described a new classification of DLBCL subtypes based on genetic signatures of low-frequency alterations, recurrent mutations, somatic copy number alterations, and structural variants. One of their two distinct subtypes of GCB-DLBCL (cluster 3) has structural variants of BCL-2 and correlate to poor risk (98). Their ABC-DLBCL subtype (cluster 5) had amplification of BCL-2 (98), as previously described (93). Staudt and colleagues also reported new classifications of DLBCL that include four new genetic subtypes, one of which is EZB, which has EZH2 mutations and BCL-2 translocations (99).

DLBCL lymphomas that contain both rearrangements in BCL-2 and translocation of MYC are classified as “double hit lymphomas” (DHL) and represent ~10% of DLBCL cases (100, 101). DLBCLs that co-express high levels of MYC and BCL-2 proteins due to mechanisms other than chromosomal translocations are referred to as “dual expresser lymphomas” (DEL) and represent ~30% of DLBCL (102, 103). Both DHL and DEL tend to be clinically more aggressive and have a higher frequency of treatment failure than those that are non-DHL or non-DEL (102, 103). Because of this, these subtypes of DLBCL have become a new biomarker-defined subset, which illustrates the importance of knowing the status of MYC and BCL-2 to help guide treatment and monitoring of patients. These results also provide impetus to investigate the expression of other anti-apoptotic BCL-2 family members in lymphomas, as others may also alter prognosis.

Besides BCL-2 translocation and amplification, additional mechanisms are reported to contribute to its increased expression, including BCL-2 gene rearrangement (104), promoter hypomethylation (105), promoter hypermutation (106), and phosphorylation (107). A small subset of mantle cell lymphoma (MCL) have increased expression of BCL-2 (33). Additionally, BCL-2 can also become overexpressed in the indolent MZL (32). For patient samples of both MCL and MZL, BCL-2 overexpression is reportedly caused by non-genomic changes to BCL-2 (33, 108). BCL-2 expression can also be regulated by non-coding RNA. In B cells, BCL-2 expression is negatively regulated by the miR-15a/miR-16-1 cluster (109, 110). This region is deleted or inactivated by mutations in ~70% of CLL (110); however, no other B cell malignancy has been associated with loss of the region (111, 112).

Although the belief in the importance of BCL-2 in human B cell lymphomas is firmly embedded, two different transgenic mice generated ~30 years ago revealed that Bcl-2 is not a driver of B cell lymphoma, but increased levels in B cells did lead to their accumulation (113, 114). Bcl-2 overexpression did cooperate with Myc overexpression to accelerate B cell lymphomagenesis (28). The requirements of BCL-2 in the continued survival of human B cell lymphomas is just now being explored with some surprising results, as described below.

Shortly following the cloning of the BCL-2 gene, the gene coding for BCL-X (BCL2L1 for Bcl-2 like 1) was discovered due to its high level of sequence similarity to BCL-2 (115, 116). Similar to Bcl-2 transgenic mice, mice engineered to overexpress Bcl-x renders lymphoid cells resistant to numerous apoptotic stimuli and causes an abnormal accumulation of mature lymphocytes, but not overt lymphoma development (37). However, mice double-transgenic for Bcl-x and Myc in B cells developed lymphoproliferative disease and plasma cell malignancies (38). Knocking out Bcl-x revealed that it is not required for lymphocyte development, but is critical for erythropoiesis and platelets (34, 35, 117, 118). Loss of Bcl-x did delay Myc-induced B cell lymphoma development, suggesting that under conditions of oncogenic stress, B cells may rely on Bcl-x for survival (36).

Burkitt lymphomas can select to overexpress BCL-X (9). Elevated BCL-X expression has been detected in other B cell non-Hodgkin lymphomas, including FL and DLBCL, as well as T cell non-Hodgkin lymphomas (39). Moreover, it has been demonstrated that in DLBCL, elevated expression of BCL-X mRNA is associated with a chemoresistant, short-lived group of patients (119). Like BCL-2, selection for BCL-X overexpression occurs in a subset of MCL (40). Moderate levels of BCL-X protein were reported in several cases of CLL, FL, and MCL, but it was lowly expressed in MZL (33). In a large-scale analysis of gene expression profiling data, we reported that BCL-X mRNA was significantly elevated compared to normal human B cells in multiple types of non-Hodgkin B cell lymphoma, including Burkitt, DLBCL, FL, MCL, and MZL (9). Unlike BCL-2, which is lowly expressed in multiple myeloma (MM), levels of BCL-X are much higher and may therefore be a critical survival factor for MM (41–44). To date, no chromosomal translocation involving BCL-X has been reported in human tumor samples, but somatic copy number amplifications have been detected in hematopoietic malignancies, including non-Hodgkin lymphomas (120). High levels of BCL-X have also been attributed to the loss or silencing of the let-7 family of miRNA that targets BCL-X (109, 121, 122). Therefore, BCL-X likely contributes to B cell lymphomas.

Upregulation of the anti-apoptotic Bcl-2 family member Mcl-1 likely also contributes to lymphomagenesis. Mcl-1 was first identified in an immortalized myeloid leukemia-derived cell line, and consequently named myeloid cell leukemia 1 (123). Consistent with Bcl-2 and Bcl-x overexpression models, Mcl-1 overexpression in transgenic mice renders hematopoietic cells largely resistant to varying apoptotic stimuli and causes the accumulation of mature B and T lymphocytes (50, 51). In addition, half of the Mcl-1 transgenic mice develop B cell lymphomas within two years (52). Mcl-1 knockout mice are embryonic lethal, but conditional knockout of Mcl-1 in mice shows a requirement for Mcl-1 in hematopoietic stem cell and neutrophil survival (45–47). Furthermore, loss of Mcl-1 inhibited Myc-induced B cell lymphomagenesis in mice (48, 49).

A somatically acquired increase in MCL-1 copy number has been documented in a variety of non-hematopoietic malignancies, but only in a limited number of non-Hodgkin lymphomas (120). Gene amplification or chromosomal gains of MCL-1 occur in 20–25% of the activated B cell (ABC) subtype of DLBCL (53). Although individual lymphomas may select for overexpression of MCL-1, we determined MCL-1 expression was not elevated in a cohort of Burkitt lymphomas compared to normal human B cells, and that it was also not significantly different in patient samples of DLBCL, FL, MZL, and MCL compared to normal B cells (9). However, increased levels of MCL-1 mRNA are suggested to be essential for sustained growth, survival, and resistance to chemotherapeutics in multiple types of lymphoma as well as CLL (54, 55). Unlike BCL-2 and BCL-X, which are overexpressed in a subset of MCL, MCL-1 expression is typically low in this lymphoma (33). Increased levels of MCL-1 have been observed in multiple myeloma (56) and shown to correlate with disease progression (57).

The expression of Mcl-1 is tightly regulated at both the transcriptional and post-transcriptional level. In contrast to Bcl-2 and Bcl-x, Mcl-1 is a short-lived protein with a half-life of ~30 min compared to the >6 h half-lives observed for Bcl-2 and Bcl-x (124–126). Mcl-1 levels are also regulated by miRNA. Loss of miR-29 and decreased levels of miR-125b and miR-133b, miRNA that bind and negatively regulate MCL-1 expression (127–130), have been reported in many lymphomas, including Burkitt, anaplastic large cell, and DLBCL, which may also contribute to increased MCL-1 expression (131–133). Elevated expression of MCL-1 can also be the result of aberrant post-translational mechanisms. For instance, increased expression of the deubiquitinase USP9X, which is responsible for removing polyubiquitin chains that target MCL-1 protein for degradation, correlates with increased MCL-1 protein in FL, DLBCL, and multiple myeloma (134). In addition, increased protein stability of MCL-1 can also lead to protein overexpression as a result of genetic inactivation of FBW7, a ubiquitin ligase (135, 136).

Protein-based structural analyses indicate, A1 (BCL2A1), the mouse homolog of human BFL-1, is most highly related to the anti-apoptotic Mcl-1 (137). Because A1/Bfl-1-null mice are embryonic lethal, recently, an A1/Bfl-1-conditional knockout mouse was generated that was viable, fertile, and only showed minor defects in the hematopoietic system (58, 59). A1/Bfl-1 transgenic mice demonstrated that overexpression of A1/Bfl-1 did not cooperate with Myc to drive B cell lymphomagenesis (60). However, overexpression of A1/BFL-1 in lymphoma cell lines provided protection from apoptosis induced by a number of stimuli, including ligation of the B cell receptor (138), cytokine deprivation (139), and treatment with staurosporine or etoposide (139, 140). In contrast, knocking down A1/BFL-1 resulted in increased sensitivity of B cell lymphoma cells to cell death caused by CD20 cross-linking and DNA-damaging drugs (141).

Increased levels of A1/BFL-1 mRNA have been reported in DLBCL and CLL (61–63). However, our analysis of public gene expression data showed that compared to normal B cells, A1/BFL-1 was not overexpressed in DLBCL, FL, MZL, and MCL (9). We also observed reduced A1/BFL-1 expression in Burkitt lymphoma patient samples compared to normal human B cells (9). A1/BFL-1 protein can be post-translationally regulated by ubiquitin-mediated proteasomal degradation (139, 140). While regulatory mechanisms, including ubiquitination/proteasomal degradation (139, 140) and direct transcriptional activation by NF-κB (nuclear factor-kappaB) (142) have been reported, the contributions of these mechanisms to increased levels of A1/BFL-1 in B cell lymphomas have not been evaluated. Based on the available data, the contribution of A1/BFL-1 to B cell lymphomas does not appear to be significant.

The gene encoding BCL-W (BCL2L2 for Bcl-2 like 2) was initially discovered via a PCR-based strategy while searching for additional BCL-2 related genes (143). Mice lacking Bcl-w were determined to be essentially normal, except for a profound block in male spermatogenesis (64, 65). Several observations have pointed to a potential role for Bcl-w in tumorigenesis. For example, in a mouse model of Myc-driven myeloid leukemogenesis, Bcl-w overexpression cooperated with Myc to accelerate leukemia development (66). In addition, high levels of BCL-W were present in cell lines derived from human lymphomas, leukemias, and multiple solid organ cancers (66). Similar to its anti-apoptotic relatives, overexpression of Bcl-w in mouse B and T lymphocytes imparted resistance to cytotoxic agents (143, 144). Recently, we discovered Bcl-w has a critical, previously unexplored function in B cell survival and lymphomagenesis (8, 9). We demonstrated with mouse models that loss of Bcl-w profoundly delayed Myc-induced B cell lymphoma development and sensitized B cells to Myc-induced apoptosis (8). We also evaluated the importance of BCL-W in human lymphomas known to be driven by or reliant on MYC expression, specifically Burkitt and DLBCL. The vast majority of Burkitt lymphoma patient samples examined overexpressed BCL-W at both the mRNA and protein levels (8). When BCL-W-specific shRNA was introduced into human Burkitt lymphoma cell lines, they rapidly underwent apoptosis, indicating BCL-W is essential for their continued survival. Additionally, we determined BCL-W was frequently overexpressed in DLBCL. We detected BCL-W mRNA and protein were as often and as highly expressed as BCL-2 in DLBCL, where BCL-2 has long served as the hallmark of a prognostically unfavorable subset (94). Notably, we also observed that increased BCL-W mRNA expression correlated with poor patient survival when levels of BCL-2 mRNA were lower in DLBCL (8).

More recently, we expanded our analyses to explore the additional contributions of BCL-W in other non-Hodgkin B cell lymphomas. We performed an unprecedented analysis of all anti-apoptotic BCL-2 family members across different B cell lymphomas, including Burkitt, DLBCL, FL, MZL, and MCL (9). In all five types of B cell lymphomas, BCL-W was overexpressed compared to normal B cell controls. Increased levels of BCL-W mRNA and protein mirrored those of BCL-2 in FL, which was unexpected, as FL is historically recognized to be driven by a BCL-2 translocation (9, 86–90). Of note, and consistent with previous reports, as the grade of FL increased, levels of BCL-2 decreased (9, 145), whereas BCL-W expression remained elevated in both low and high grade FL (9). Taken together, BCL-W appears to have a critically important, and previously unrecognized, anti-apoptotic role in B cell lymphoma.

Multiple avenues to regulate levels of Bcl-w have been proposed. Overexpression of BCL-W can result from increased activity at the BCL-W promoter (146). A number of studies using different tumor types have documented that elevated expression of BCL-W may be attributed to the downregulation of miRNA that target BCL-W mRNA (128, 147, 148). We recently demonstrated that tumor suppressive miRNA target BCL-W, BCL-2, and BCL-X as a novel, miRNA-mediated mechanism of apoptosis induced by the oncogenic transcription factor Myc (8, 109, 121). Specifically, in normal cells, Myc transcriptionally activates the miR-15 family and let-7a, which bind and negatively regulate the expression of BCL-W, BCL-2, and BCL-X, causing cells to undergo apoptosis (Figure 2). However, cancer cells, including lymphoma, inactivate this mechanism. By means that have not been fully elucidated, but do involve histone deacetylase (HDAC) enzymes, Myc switches from a transcriptional activator to a transcriptional repressor of the miR-15 family and let-7a in lymphoma. This, in turn, allows the expression of anti-apoptotic BCL-W, BCL-2, and BCL-X to increase, thereby facilitating tumorigenesis (Figure 2). We established a link between regulation of BCL-W expression and Myc. Identifying this novel mechanism of miRNA-mediated apoptosis also answered a question that has remained largely unexplored. Previously, it was shown that Myc suppressed the expression of Bcl-2 and Bcl-x through an unknown, indirect mechanism (149, 150). These recent studies provide direct evidence of a mechanism whereby Myc transcriptionally modulates miRNA expression leading to altered expression of Bcl-w as well as Bcl-2 and Bcl-x (8, 109, 121). Therefore, increased expression of anti-apoptotic Bcl-2 family members, through a variety of mechanisms, are essential survival factors to B cell lymphoma.

While compounds selective for BCL-2 over BCL-X have shown anti-tumor effects in vivo with limited platelet toxicities (175, 180), not all cancers express BCL-2 or require BCL-2 for their continued survival (5, 197). The first reported BCL-X selective inhibitor with sub-nanomolar affinity and selectivity was WEHI-539, which antagonized the anti-apoptotic activity of BCL-X (198). However, WEHI-539 has been limited to in vitro studies as a tool compound due to its unfavorable chemical properties for use in vivo (198). Other efforts using high-throughput screening with nuclear magnetic resonance and structure-based medicinal chemistry led to the development of the BCL-X selective inhibitors A-1155463 and its orally bioavailable analog A-1331852 [Figure 3, (199, 200)]. Cell lines from a variety of malignancies, including AML showed sensitivity to the BCL-X inhibitors. In vivo studies in mice indicated inhibition of BCL-X with A-1331852 as a single agent reduced the volume of a T-cell leukemia, whereas venetoclax did not (180). In addition, the mice tolerated the BCL-X inhibitor. These data suggest that some cancers appear to be dependent on BCL-X and not on BCL-2 and that targeting one over the other may be beneficial.

Multiple approaches have been explored to selectively target MCL-1. Initially, stabilized alpha-helices of BH-3 only proteins [known as “stapled” peptides (201)] and alpha- or beta-peptide foldamers (202) were designed to target MCL-1. Several compounds with affinity for MCL-1 were identified by screening libraries of natural compounds and small molecules. However, many of these have reported off-target or minimal effects (203, 204). Although obatoclax (GX15-070) targets other anti-apoptotic BCL-2 family members, it was one of the first BH-3 mimetics reported to inhibit MCL-1 [Figure 3, (205)]. Obatoclax has completed phase I/II clinical trials for a number of malignancies, including CLL, but with modest efficacy and neutropenia as a common toxicity (206). At this time, no further clinical trials with obatoclax are ongoing. Similar to obatoclax, TW-37, a gossypol derivative, has shown some toxicity, but has demonstrated potency as a single agent in the treatment of DLBCL cell lines and synergized with other chemotherapies in xenograft models [Figure 3, (207)]. Phase I clinical trials for AT-101, a more refined variant of the gossypol family of pan-BCL-2 family inhibitors, show that it is well-tolerated with treatable neutropenia (208). Two clinical trials of AT-101 as a single agent in R/R B cell lymphomas (NCT00275431) and in combination with rituximab in CLL (NCT00286780) have been completed, but final results have not been published.

A-1210477 was the first selective MCL-1 inhibitor to demonstrate picomolar binding to MCL-1 while maintaining selectivity for MCL-1 over other anti-apoptotic family members [Figure 3, (209)]. A-1210477 forms complexes with MCL-1, thereby disrupting endogenous MCL-1:BIM complexes. As a single agent, this compound induced apoptosis in MCL-1-dependent multiple myeloma cell lines (210). However, A-1210477 does not demonstrate the necessary pharmacokinetics for in vivo use. AMG176 was the first putative MCL-1 inhibitor to undergo clinical evaluation (NCT02675452) (211), but no data in humans has been reported. In addition, phase I trials are actively recruiting or soon to recruit patients for testing the MCL-1 specific inhibitors S64315 (MIK665) in patients with R/R DLBCL and multiple myeloma, AML, and myelodysplastic syndrome (NCT02992483, NCT02979366, and NCT03672695) and AZD5991 in patients with R/R hematologic malignancies, CLL, T-cell lymphoma, and multiple myeloma (NCT03218683). A question that remains largely unanswered is whether normal cells will tolerate MCL-1 inhibition at the level necessary for therapeutic benefit. Potential on-target toxicities may include cardiac (212, 213), hepatic (214, 215), and hematological (45–47), which are based on those observed in Mcl-1 knockout mouse models (45–47, 212–215).