Purpose: B-cell lymphomas must acquire resistance to apoptosis during their development. We recently discovered BCLW, an antiapoptotic BCL2 family member thought only to contribute to spermatogenesis, was overexpressed in diffuse large B-cell lymphoma (DLBCL) and Burkitt lymphoma. To gain insight into the contribution of BCLW to B-cell lymphomas and its potential to confer resistance to BCL2 inhibitors, we investigated the expression of BCLW and the other antiapoptotic BCL2 family members in six different B-cell lymphomas.

Experimental Design: We performed a large-scale gene expression analysis of datasets comprising approximately 2,300 lymphoma patient samples, including non-Hodgkin and Hodgkin lymphomas as well as indolent and aggressive lymphomas. Data were validated experimentally with qRT-PCR and IHC.

Results: We report BCLW is significantly overexpressed in aggressive and indolent lymphomas, including DLBCL, Burkitt, follicular, mantle cell, marginal zone, and Hodgkin lymphomas. Notably, BCLW was preferentially overexpressed over that of BCL2 and negatively correlated with BCL2 in specific lymphomas. Unexpectedly, BCLW was overexpressed as frequently as BCL2 in follicular lymphoma. Evaluation of all five antiapoptotic BCL2 family members in six types of B-cell lymphoma revealed that BCL2, BCLW, and BCLX were consistently overexpressed, whereas MCL1 and A1 were not. In addition, individual lymphomas frequently overexpressed more than one antiapoptotic BCL2 family member.

Conclusions: Our comprehensive analysis indicates B-cell lymphomas commonly select for BCLW overexpression in combination with or instead of other antiapoptotic BCL2 family members. Our results suggest BCLW may be equally as important in lymphomagenesis as BCL2 and that targeting BCLW in lymphomas should be considered. Clin Cancer Res; 23(22); 7119–29. ©2017 AACR.

Translational Relevance

With clinical trials on-going for BCL2-specific inhibitors for the treatment of lymphomas and BCLX- and MCL1-specific inhibitors being developed, there is a critical need for increased knowledge of the expression of antiapoptotic BCL2 family members in lymphomas to identify those most likely to respond to specific inhibitors. Our study, the first large-scale analysis of all antiapoptotic BCL2 family members in six types of B-cell lymphomas (non-Hodgkin and Hodgkin), reveals B-cell lymphomas typically overexpress more than one. Moreover, BCLW, a previously uncharacterized antiapoptotic BCL2 family member, was frequently overexpressed. Specific lymphomas preferentially selected for BCLW overexpression over that of BCL2. Notably, BCLW was overexpressed in follicular lymphoma, indicating it likely has a role in this lymphoma. Our results suggest targeting BCLW in B-cell lymphomas may be needed alone or in combination with other specific BCL2 family inhibitors, as targeting just one may lead to dependence on another resulting in therapy resistance.

Lymphomas, as with all human cancers, select for alterations that inhibit apoptosis, allowing them to survive during transformation (1). The BCL2 family of proteins consists of antiapoptotic (BCL2, BCLX/BCL2L1, BCLW/BCL2L2, MCL1, and A1/BFL1) and proapoptotic proteins that regulate cell death and have been linked to many human cancers, including lymphomas (2).

BCL2 translocation to the immunoglobulin locus, leading to BCL2 overexpression, is a defining feature of the non-Hodgkin indolent lymphoma follicular lymphoma (3). Diffuse large B-cell lymphoma (DLBCL), an aggressive lymphoma, can also overexpress BCL2 (4). Thirty percent of de novo DLBCL have BCL2 translocations and/or increased protein expression (5), correlating with reduced survival (4). In contrast with follicular lymphoma and DLBCL, Burkitt lymphoma, an aggressive lymphoma, express low/undetectable levels of BCL2, which is part of its diagnostics (6). We recently discovered that Burkitt lymphomas frequently overexpress BCLW, an antiapoptotic BCL2 family member that was initially reported to only function in spermatogenesis (7–9). Targeting BCLW with shRNA or pharmacologically with BH3-mimetics induced apoptosis in Burkitt lymphoma cell lines, indicating BCLW was required for its survival (9). When overexpressed, BCLW conferred resistance to BH3-mimetics (9). We also showed that DLBCL frequently overexpresses BCLW alone or in combination with BCL2, and that increased levels of BCLW in patient samples with low BCL2 expression correlated with reduced patient survival (9). Therefore, BCLW is a previously unappreciated, significant contributor to Burkitt lymphoma and DLBCL, but its involvement in other B-cell lymphomas is unknown.

There are multiple different aggressive and indolent B-cell lymphoma subtypes, but the contribution of BCLW and other antiapoptotic BCL2 family members to them is unclear. It is reported that there is increased expression of BCL2 and BCLX in a small subset of mantle cell lymphoma (an aggressive lymphoma), whereas MCL1 expression is typically low (10). BCL2 levels can be elevated in marginal zone lymphomas, which are indolent (4). Increased levels of BCLX were identified in the majority of Hodgkin lymphomas, but are not part of its diagnostics (11). Therefore, unlike follicular lymphoma and DLBCL, other B-cell lymphomas have not been linked to alterations in specific antiapoptotic BCL2 family members, which may be due to the lack of a comprehensive analysis of these genes/proteins in these lymphomas. Here, we evaluated gene expression profiling datasets of six B-cell lymphomas for the expression of BCLW and the other antiapoptotic BCL2 family members and validated the results with quantitative real-time PCR (qRT-PCR) and immunohistochemistry of patient samples. Our data show BCLW is frequently overexpressed in all six B-cell lymphomas, suggesting BCLW has an equally important role in lymphomagenesis as BCL2. In addition, most of the lymphomas analyzed overexpressed more than one antiapoptotic BCL2 family member. Our results suggest that B-cell lymphomas rely on more than one antiapoptotic BCL2 family member for their survival. With targeted therapies against antiapoptotic BCL2 family members currently being tested or developed and therapeutic resistance emerging, our data provide critical information that should have significant clinical implications.

Microarray gene expression analysis

Thirty-seven microarray gene expression profiling datasets for six different B-cell lymphomas (Burkitt, DLBCL, follicular, mantle cell, marginal zone, and Hodgkin) and normal B cells (germinal center centroblasts and centrocytes, memory B cells, and CD19+ B cells) were downloaded from the Gene Expression Omnibus (GEO) database (12) or the authors' website (ref. 6; summarized in Supplementary Table S1). The data were generated on the Affymetrix Human Genome U133 Plus 2.0 platform following standard Affymetrix protocol. The data were separated into batches, normalized using the robust multi-array average (RMA) algorithm (13) in the Affy package for R (14). To limit/eliminate confounding, we corrected batch effects using ComBat considering two covariates (lymphoma and normal; ref. 15). Probe sets were averaged to obtain a single expression intensity measure per gene per array if multiple probe sets corresponded to the same gene ID. Expression values from replicate samples were averaged. Differential expression was measured using unpaired, two-tailed t tests as indicated. All analyses were carried out in R, version 3.2.3.

Survival analysis

Gene expression profiles and patient clinicopathologic information for DLBCL were extracted from the GEO database (GSE10846 and GSE31312). Data were normalized and probe sets averaged as explained above. Patient samples were simultaneously stratified on the basis of the median expression of BCL2 and BCLW into high or low expression groups for each gene. Then, patients with low BCL2 expression were identified and Kaplan–Meier overall survival curves plotted for the high and low BCLW expression groups and compared using the log-rank test.

Patient sample acquisition

Patient samples of formalin-fixed, paraffin-embedded (FFPE) sections were obtained from archival pathologic specimens at Thomas Jefferson University with approval from the Institutional Review Board. Thirty follicular lymphomas, 10 marginal zone lymphomas, 14 mantle cell lymphomas, 26 Hodgkin lymphomas, and 12 normal lymph nodes were identified by a board-certified hematopathologist (J.Z. Gong) and processed as described below for evaluation.

RNA isolation and quantitative real-time PCR

Total RNA was isolated from eight 10-μm sections of FFPE tissue using the RecoverAll Total Nucleic Acid Isolation Kit as per the manufacturer's instructions (Thermo Fisher). cDNA was generated using SuperScript III First-Strand Synthesis (Thermo Fisher Scientific) and SybrGreen (SABiosciences) assays were performed in triplicate to measure mRNA expression as previously described (16). mRNA expression was normalized to β-ACTIN and presented as 2−ΔΔCt comparing the mean expression of specific mRNA in lymphoma with the mean of this mRNA in the controls to determine fold-change. We previously reported the primer sequences for BCLW, BCL2, BCLX, and MCL1 (17). Primers for BCL2A1 are forward-5′-CCCGGATGTGGATACCTATAAGGAGA and reverse-5′-GTCATCCAGCCAGATTTAGGTTCA.

IHC

IHC for BCL2 and BCLW was performed on 4-μm FFPE sections from lymphoma and control tissues following standard procedures. BCL2 (SP66) and BCLW (ab38629, Abcam, 1:125 dilution) staining was performed using automated IHC stainers (Ventana Medical Systems and DAKO Autostainer Plus, respectively) according to the manufacturer's specifications. Positive and negative controls were included. Protein levels were scored by a practicing, board-certified hematopathologist (J.Z. Gong) on a 0–3 scale in a blinded fashion (representative images in Supplementary Fig. S1).

When comparing two groups, two-tailed t tests were used. For gene expression, qRT-PCR, and IHC graphs, lines represent the mean ± SEM and symbols represent individual samples. For box and whisker plots, boxes represent the 25th and 75th percentiles, lines indicate the median, circles indicate the mean, and whiskers are the maximum and minimum. Survival analyses were compared by log-rank tests. Pearson correlation coefficient and Spearman rank correlation coefficient determined the expression correlation between two genes. P <0.05 was considered statistically significant unless otherwise noted.

BCLW is overexpressed in activated B-cell (ABC) and germinal center B-cell (GCB) subtypes of DLBCL

Recently, after analyzing microarray datasets containing 57 Burkitt and 319 DLBCL patient samples, we reported that BCLW is frequently overexpressed in both of these lymphomas (9). Because RNA-sequencing datasets were limited, we obtained additional microarray datasets of DLBCL and normal human B cells and combined them with those we previously analyzed. Assessing 1,490 DLBCL samples showed that BCLW and BCL2 were significantly overexpressed compared with normal B cells (Fig. 1A). These results independently validate our previous study and further indicate BCLW is overexpressed in aggressive lymphomas, such as DLBCL.

Figure 1.

BCLW is overexpressed in GCB and ABC DLBCL. A and B, Gene-expression datasets for BCLW and BCL2 mRNA in all DLBCL patient samples evaluated (n = 1,490; A, or stratified on the basis of GCB (n = 591) and ABC (n = 519) DLBCL subtypes (B) compared with normal human B cells (n = 117; A and B). C, Association between BCLW and BCL2 mRNA in ABC DLBCL stratified on the basis of the lowest (n = 130) and highest (n = 130) quartiles of BCL2 expression. For A, *, P = 4.03 × 10−11 and **, P = 3.00 × 10−21; B, *, P = 1.66 × 10−3 and **, P < 4.78 × 10−18; C, *, P = 1.34 × 10−5; two-tailed t tests. D, Patient samples of GCB DLBCL with low BCL2 from GSE10846 and GSE31312 datasets were each stratified on the basis of the median expression of BCLW into two groups, BCLW low and BCLW high. Kaplan–Meier overall survival analyses were performed and compared using log-rank tests.

Figure 1.

BCLW is overexpressed in GCB and ABC DLBCL. A and B, Gene-expression datasets for BCLW and BCL2 mRNA in all DLBCL patient samples evaluated (n = 1,490; A, or stratified on the basis of GCB (n = 591) and ABC (n = 519) DLBCL subtypes (B) compared with normal human B cells (n = 117; A and B). C, Association between BCLW and BCL2 mRNA in ABC DLBCL stratified on the basis of the lowest (n = 130) and highest (n = 130) quartiles of BCL2 expression. For A, *, P = 4.03 × 10−11 and **, P = 3.00 × 10−21; B, *, P = 1.66 × 10−3 and **, P < 4.78 × 10−18; C, *, P = 1.34 × 10−5; two-tailed t tests. D, Patient samples of GCB DLBCL with low BCL2 from GSE10846 and GSE31312 datasets were each stratified on the basis of the median expression of BCLW into two groups, BCLW low and BCLW high. Kaplan–Meier overall survival analyses were performed and compared using log-rank tests.

Close modal

DLBCL has two subtypes, ABC and GCB (18). We evaluated both subtypes for BCLW and as a control, BCL2 expression. As previously reported, ABC DLBCL had significantly increased BCL2 levels compared with normal B cells (Fig. 1B; ref. 19). BCL2 levels were also increased in GCB DLBCL, but not to the same extent. In contrast, BCLW was equally overexpressed in ABC and GCB DLBCL (Fig. 1B). Further evaluation of ABC DLBCL showed that those with higher BCL2 levels had lower BCLW levels and those with lower BCL2 levels had higher BCLW levels. This inverse association between BCLW and BCL2 in ABC DLBCL was significant, but not in GCB DLBCL (Fig. 1C; Supplementary Table S2; Supplementary Fig. S2A).

Using two independent datasets containing survival data, we previously reported DLBCL patients with high levels of BCLW and low levels of BCL2 had poor survival compared with patients whose DLBCL was low for both BCLW and BCL2 (9). To assess whether this was specific for ABC or GCB DLBCL, we evaluated overall survival of each subtype based on BCLW and BCL2 expression. Although statistical significance was not reached, both datasets showed a potential for reduced overall survival for patients with high BCLW and low BCL2 in GCB, but not in ABC (Fig. 1D; Supplementary Fig. S2B). Together, our data show that BCLW is overexpressed in both GCB and ABC DLBCL, and may affect the survival of GCB DLBCL.

Indolent lymphomas also select for BCLW overexpression

To ascertain whether selection for BCLW overexpression only occurs in aggressive lymphomas, we also evaluated indolent lymphomas. Assessment of patient samples of follicular lymphoma (n = 196) showed BCL2 was characteristically overexpressed (Fig. 2A). However, there was also significant overexpression of BCLW compared with normal B cells (Fig. 2A). Notably, there was no significant inverse association between BCLW and BCL2 expression (Supplementary Fig. S3A), suggesting that BCL2 overexpression was not preferentially selected over BCLW. In support of this, of the lymphomas that highly overexpressed BCLW (top 25%, n = 49), 53.1% also overexpressed BCL2 (Fig. 2B). Therefore, BCLW overexpression is selected for even when BCL2 is overexpressed, suggesting BCLW is important in follicular lymphoma.

Figure 2.

Overexpression of BCLW is selected for in follicular lymphoma. A, Gene-expression data for BCL2 and BCLW mRNA in follicular lymphoma (n = 196) compared with normal human B cells (n = 117). *, P = 5.64 × 10−10 and **, P = 1.11 × 10−16; two-tailed t tests. B, Association between BCLW and BCL2 mRNA in follicular lymphoma stratified on the basis of the highest quartile (n = 49) of BCLW expression. Each line represents one sample. C and D, qRT-PCR analysis for BCLW and BCL2, in triplicate, from follicular lymphoma patient samples (n = 30) compared with normal lymph nodes (n = 12) in C (box and whisker plot), or stratified on the basis of grade (low, n = 19; high, n = 11) in D. For C, *, P = 5.91 × 10−5 and **, P = 7.21 × 10−7; D, *, P = 0.044 and **, P = 0.023; two-tailed t tests. E, IHC analysis for BCLW and BCL2 protein on follicular lymphoma patient samples (n = 30). Box and whisker plot of pathologist scores and representative images with scores of BCLW and BCL2 from the same tumors are shown. F, Pearson correlation coefficients for BCLW and BCL2 comparing mRNA fold-change with IHC scores, *, P = 0.039 and **, P = 1.73 × 10−6.

Figure 2.

Overexpression of BCLW is selected for in follicular lymphoma. A, Gene-expression data for BCL2 and BCLW mRNA in follicular lymphoma (n = 196) compared with normal human B cells (n = 117). *, P = 5.64 × 10−10 and **, P = 1.11 × 10−16; two-tailed t tests. B, Association between BCLW and BCL2 mRNA in follicular lymphoma stratified on the basis of the highest quartile (n = 49) of BCLW expression. Each line represents one sample. C and D, qRT-PCR analysis for BCLW and BCL2, in triplicate, from follicular lymphoma patient samples (n = 30) compared with normal lymph nodes (n = 12) in C (box and whisker plot), or stratified on the basis of grade (low, n = 19; high, n = 11) in D. For C, *, P = 5.91 × 10−5 and **, P = 7.21 × 10−7; D, *, P = 0.044 and **, P = 0.023; two-tailed t tests. E, IHC analysis for BCLW and BCL2 protein on follicular lymphoma patient samples (n = 30). Box and whisker plot of pathologist scores and representative images with scores of BCLW and BCL2 from the same tumors are shown. F, Pearson correlation coefficients for BCLW and BCL2 comparing mRNA fold-change with IHC scores, *, P = 0.039 and **, P = 1.73 × 10−6.

Close modal

To determine whether the bioinformatic analyses accurately reflected expression in patients, we first performed qRT-PCR on patient samples of follicular lymphoma compared with normal lymph nodes. BCL2 was significantly overexpressed (Fig. 2C; Supplementary Fig. S3B). BCLW was also significantly overexpressed with 93.3% (28 of 30) of the follicular lymphoma expressing at least 4-fold more BCLW than normal lymph nodes (Fig. 2C; Supplementary Fig. S3B). Furthermore, consistent with previous reports (20), high-grade follicular lymphoma showed decreased BCL2 expression compared with low-grade (Fig. 2D; Supplementary Fig. S3C). However, a similar level of BCLW overexpression was selected for regardless of grade (Fig. 2D; Supplementary Fig. S3C).

We next performed IHC to assess whether the mRNA levels detected were reflected at the protein level. BCLW protein was significantly overexpressed compared with normal lymph node (Fig. 2E). Specifically, BCLW staining (≥1 pathologist score) was present in all 30 follicular lymphoma assessed, and two-thirds received a score of 2 or 3 (Fig. 2E). Characteristically, for follicular lymphoma (21), nearly all showed increased BCL2 levels (Fig. 2E). There was a positive correlation between our qRT-PCR and IHC results (Fig. 2F), providing additional support that BCLW and BCL2 mRNA expression is reflected at the protein level. Together, these results indicate BCLW is overexpressed in follicular lymphoma and that it is overexpressed at a similar frequency as BCL2.

Next, we evaluated 148 patient samples of marginal zone lymphoma (MZL), an indolent lymphoma, and determined that both BCLW and BCL2 were overexpressed compared with normal human B cells (Fig. 3A). Notably, there was a significant inverse association between BCLW and BCL2 expression (Fig. 3B; Supplementary Table S2; Supplementary Fig. S4A). Measurement of BCLW and BCL2 levels by qRT-PCR in patient samples of nodal MZL also showed that both were significantly overexpressed compared with normal lymph nodes (Fig. 3C; Supplementary Fig. S4B). Although the average and median expression of BCLW was higher than that of BCL2, the negative correlation between the two was not statistically significant, likely due to our limited sample set (n = 10). Evaluation of BCLW protein by IHC showed it was consistently overexpressed, as we detected high levels in MZL (average score of 2.2 out of 3). In contrast, although BCL2 protein levels were overexpressed compared with normal lymph nodes, BCL2 protein (average score of 1.7 out of 3) was not as frequently overexpressed as BCLW (Fig. 3D), indicating that BCLW may be preferentially overexpressed in nodal MZL over BCL2. A positive correlation was observed between the qRT-PCR and IHC analyses for BCLW and BCL2, indicating protein and mRNA levels were consistent (Fig. 3E). Therefore, BCLW is overexpressed in indolent and aggressive lymphomas.

Figure 3.

Marginal zone and mantle cell lymphomas overexpress BCLW. A and F, Gene-expression data for BCLW and BCL2 mRNA in MZL (n = 148, A) and MCL (n = 196, F) compared with normal human B-cells (n = 117). For A, *, P = 7.38 × 10−10 and **, P = 1.29 × 10−13; F, *, P = 2.94 × 10−9 and **, P = 1.06 × 10−13; two-tailed t tests. B, Association between BCLW and BCL2 mRNA in MZL stratified on the basis of the lowest (n = 37) and highest (n = 37) quartiles of BCL2 expression; *, P = 5.13 × 10−5; two-tailed t tests. C and G, qRT-PCR analysis (box and whisker plot) for BCLW and BCL2, in triplicate, from nodal MZL (n = 10, C) and MCL (n = 14, G) patient samples compared with normal lymph nodes (n = 12). For C, *, P = 5.95 × 10−3 and **, P = 8.03 × 10−5; G, *, P = 9.28 × 10−9 and **, P = 2.01 × 10−12; two-tailed t tests. D and H, IHC analysis for BCLW and BCL2 protein on nodal MZL (n = 10, D) and MCL patient samples (n = 5, H). Box and whisker plots of pathologist scores and representative images of BCLW and BCL2 from the same tumors are shown. E and I, Pearson correlation coefficients for BCLW and BCL2 comparing mRNA fold-change with IHC scores (E, *, P = 0.024; I *, P = 0.027).

Figure 3.

Marginal zone and mantle cell lymphomas overexpress BCLW. A and F, Gene-expression data for BCLW and BCL2 mRNA in MZL (n = 148, A) and MCL (n = 196, F) compared with normal human B-cells (n = 117). For A, *, P = 7.38 × 10−10 and **, P = 1.29 × 10−13; F, *, P = 2.94 × 10−9 and **, P = 1.06 × 10−13; two-tailed t tests. B, Association between BCLW and BCL2 mRNA in MZL stratified on the basis of the lowest (n = 37) and highest (n = 37) quartiles of BCL2 expression; *, P = 5.13 × 10−5; two-tailed t tests. C and G, qRT-PCR analysis (box and whisker plot) for BCLW and BCL2, in triplicate, from nodal MZL (n = 10, C) and MCL (n = 14, G) patient samples compared with normal lymph nodes (n = 12). For C, *, P = 5.95 × 10−3 and **, P = 8.03 × 10−5; G, *, P = 9.28 × 10−9 and **, P = 2.01 × 10−12; two-tailed t tests. D and H, IHC analysis for BCLW and BCL2 protein on nodal MZL (n = 10, D) and MCL patient samples (n = 5, H). Box and whisker plots of pathologist scores and representative images of BCLW and BCL2 from the same tumors are shown. E and I, Pearson correlation coefficients for BCLW and BCL2 comparing mRNA fold-change with IHC scores (E, *, P = 0.024; I *, P = 0.027).

Close modal

Mantle cell lymphomas also overexpress BCLW

We evaluated BCLW expression in the aggressive lymphoma mantle cell lymphoma (MCL) to determine how widespread BCLW overexpression was. Assessment of 196 MCL samples showed significantly increased levels of BCLW and BCL2 compared with normal human B cells (Fig. 3F). There was no inverse association between BCLW and BCL2 expression (Supplementary Fig. S5A). qRT-PCR analysis of patient samples showed that BCLW and BCL2 were both significantly overexpressed in MCL compared with normal lymph nodes (Fig. 3G; Supplementary Fig. S5B). IHC for BCLW and BCL2 showed increased levels (average score of 2.2 and 2.6, respectively, out of 3; Fig. 3H). Furthermore, the mRNA expression data and IHC results were positively correlated for both BCLW and BCL2 (Fig. 3I). Therefore, our data indicate BCLW overexpression is widespread in aggressive and indolent non-Hodgkin B-cell lymphomas.

BCLW is overexpressed in Hodgkin lymphomas

To determine whether BCLW overexpression was limited to only non-Hodgkin B-cell lymphomas, we evaluated 180 Hodgkin lymphoma (HL) samples. HL showed BCLW overexpression compared with normal B cells (Fig. 4A; Supplementary Fig. S6A). Also, the HL with higher BCLW levels had lower BCL2 levels and those with higher BCL2 had lower BCLW levels. There was a statistically significant inverse association between their expression (Fig. 4B; Supplementary Table S2; Supplementary Fig. S6A). We evaluated patient samples of both nodular sclerosis and mixed cellularity HL by qRT-PCR. Although qRT-PCR verified the bioinformatic analyses, showing statistically significant increased expression of BCLW and BCL2 mRNA, BCLW was overexpressed significantly more than BCL2 in both subtypes (Fig. 4C; Supplementary Fig. S6B and 6C). The preferential selection for BCLW overexpression was also observed at the protein level, as 14 of 15 HL samples expressed high levels (pathologist score of 2 or 3) of BCLW protein, whereas only 1 sample received a high score for BCL2 protein (Fig. 4D). The results of our protein and mRNA analyses were consistent, producing statistically significant positive correlations for both BCLW and BCL2 (Fig. 4E). Therefore, HL frequently overexpresses BCLW, and preferentially selects for BCLW overexpression over that of BCL2.

Figure 4.

BCLW is preferentially overexpressed in Hodgkin lymphoma. A, Gene-expression data for BCL2 and BCLW mRNA in HL (n = 180) compared with normal human B cells (n = 117). B, Association between BCLW and BCL2 mRNA in HL stratified on the basis of the lowest (n = 45) and highest (n = 45) quartiles of BCL2 expression. For A, *, P = 2.27 × 10−8 and **, P = 5.25 × 10−14; B, *, P = 4.18 × 10−5; two-tailed t tests. C, Box and whisker plot of qRT-PCR analysis for BCLW and BCL2, in triplicate, from HL patient samples (n = 26) compared with normal lymph nodes (n = 12). *, P = 0.013; **, P = 2.54 × 10−4; and ***, P = 6.26 × 10−5; two-tailed t tests. D, IHC analysis for BCLW and BCL2 protein on HL patient samples (n = 15). Box and whisker plots of pathologist scores and representative images of BCLW and BCL2 from the same tumors are shown; arrows indicate HL cells. *, P = 1.43 × 10−5; two-tailed t test. E, Pearson correlation coefficients for BCLW and BCL2 comparing mRNA fold-change with IHC scores, *, P = 8.49 × 10−3 and **, P = 1.10 × 10−5.

Figure 4.

BCLW is preferentially overexpressed in Hodgkin lymphoma. A, Gene-expression data for BCL2 and BCLW mRNA in HL (n = 180) compared with normal human B cells (n = 117). B, Association between BCLW and BCL2 mRNA in HL stratified on the basis of the lowest (n = 45) and highest (n = 45) quartiles of BCL2 expression. For A, *, P = 2.27 × 10−8 and **, P = 5.25 × 10−14; B, *, P = 4.18 × 10−5; two-tailed t tests. C, Box and whisker plot of qRT-PCR analysis for BCLW and BCL2, in triplicate, from HL patient samples (n = 26) compared with normal lymph nodes (n = 12). *, P = 0.013; **, P = 2.54 × 10−4; and ***, P = 6.26 × 10−5; two-tailed t tests. D, IHC analysis for BCLW and BCL2 protein on HL patient samples (n = 15). Box and whisker plots of pathologist scores and representative images of BCLW and BCL2 from the same tumors are shown; arrows indicate HL cells. *, P = 1.43 × 10−5; two-tailed t test. E, Pearson correlation coefficients for BCLW and BCL2 comparing mRNA fold-change with IHC scores, *, P = 8.49 × 10−3 and **, P = 1.10 × 10−5.

Close modal

Combined, the data show that BCLW is not only overexpressed in Burkitt (9) and DLBCL, but it is also overexpressed in follicular, mantle cell, marginal zone, and Hodgkin lymphomas. Therefore, BCLW appears to contribute to both indolent and aggressive and non-Hodgkin and Hodgkin B-cell lymphomas. Importantly, the data show some lymphomas preferentially select for overexpression of BCLW instead of BCL2.

Lymphomas select for the overexpression of multiple antiapoptotic BCL2 family members

Because we detected increased BCLW expression across all B-cell lymphoma subtypes analyzed (Fig. 5A) and there appeared to be preferential BCLW expression in specific subtypes, we questioned the expression of the other antiapoptotic BCL2 family members and whether one or more would preferentially be expressed in particular lymphoma subtypes. To address this, we evaluated BCLX, MCL1, and A1 expression in the six-major human B-cell lymphomas (Burkitt, DLBCL, follicular lymphoma, MCL, MZL, and HL). BCLX was significantly overexpressed in all six lymphomas compared with normal human B cells with the mean expression highest in DLBCL and lowest in follicular (Fig. 5B). The mean expression of MCL1 and A1 was significantly downregulated in Burkitt lymphoma, but showed no difference in DLBCL, follicular lymphoma, MZL, MCL, and HL compared with normal B cells (Fig. 5C and D).

Figure 5.

Expression of antiapoptotic BCL2 family members in different B-cell lymphomas. A–D, Gene-expression data for BCLW and BCL2 (A), BCLX (B), MCL1(C), and A1 (D) mRNA in the six different lymphomas indicated (BL, Burkitt lymphoma). The data in A are summarized from A in Figs. 14 and 3F. *, P < 1.0 × 10−3 is significant except for BL for which significance is *, P < 0.01 due to small sample size, two-tailed t tests. E, Box and whisker plot of qRT-PCR analysis for BCLW, BCL2, BCLX, MCL1, and A1 mRNA, in triplicate, from follicular lymphoma patient samples (n = 30) compared with normal lymph nodes (N; n = 12). The data in E for BCLW and BCL2 is the same as Fig. 2C, but included here for comparison. *, P = 3.77 × 10−4; **, P = 5.91 × 10−5; and ***, P = 7.21 × 10−7; two-tailed t tests.

Figure 5.

Expression of antiapoptotic BCL2 family members in different B-cell lymphomas. A–D, Gene-expression data for BCLW and BCL2 (A), BCLX (B), MCL1(C), and A1 (D) mRNA in the six different lymphomas indicated (BL, Burkitt lymphoma). The data in A are summarized from A in Figs. 14 and 3F. *, P < 1.0 × 10−3 is significant except for BL for which significance is *, P < 0.01 due to small sample size, two-tailed t tests. E, Box and whisker plot of qRT-PCR analysis for BCLW, BCL2, BCLX, MCL1, and A1 mRNA, in triplicate, from follicular lymphoma patient samples (n = 30) compared with normal lymph nodes (N; n = 12). The data in E for BCLW and BCL2 is the same as Fig. 2C, but included here for comparison. *, P = 3.77 × 10−4; **, P = 5.91 × 10−5; and ***, P = 7.21 × 10−7; two-tailed t tests.

Close modal

To confirm our bioinformatic results, we performed qRT-PCR on our follicular lymphoma patient samples for BCLX, MCL1, and A1. BCLX was significantly overexpressed compared with normal B cells, but MCL1 and A1 were not (Fig. 5E; Supplementary Fig. S7A–S7C). However, some individual follicular lymphoma overexpressed MCL1 or A1 (Supplementary Fig. S7B and S7C). Overall, BCLW, BCL2 and BCLX, but not MCL1 or A1, were consistently overexpressed in the lymphomas analyzed.

We next assessed whether negative correlations existed between any two antiapoptotic BCL2 family members in the six types of lymphomas. BCLX, MCL1, and A1 showed no significant negative correlations with other antiapoptotic BCL2 family members with three exceptions. BCL2 was significantly negatively correlated with MCL1 and BCLX in MZL, and BCLW was significantly negatively correlated with A1 in HL (Supplementary Table S2). These data suggest preferential section of one antiapoptotic BCL2 family member over another in MZL and HL.

To determine whether individual lymphomas that overexpress BCLW simultaneously overexpress other antiapoptotic BCL2 family members, we selected the top 25% of BCLW-overexpressing DLBCL samples, which was validated to highly overexpress BCLW (Supplementary Fig. S8), and aligned them with the expression of the other four antiapoptotic BCL2 family members. We observed 56.8% of the DLBCL samples that overexpressed BCLW expressed low (below median) levels of BCL2 (Fig. 6A; Supplementary Fig. S8). Notably, a third of the DLBCL samples that overexpressed BCLW only overexpressed BCLW and no other antiapoptotic BCL2 family members (Fig. 6B). However, 36.5% of the BCLW-overexpressing DLBCL samples also concurrently overexpressed BCL2, BCLX, MCL1, or A1 (Fig. 6B). Thirty percent of the lymphomas simultaneously overexpressed BCLW and two or three other antiapoptotic BCL2 family members, whereas only one BCLW-overexpressing DLBCL also overexpressed the other four family members (Fig. 6B).

Figure 6.

Non-Hodgkin and Hodgkin lymphomas preferentially overexpress select antiapoptotic BCL2 family members. A, Gene-expression data showing association between BCLW and BCL2, BCLX, MCL1, and A1 mRNA in DLBCL stratified on the basis of the top 25% of BCLW expression. Each line represents one sample. B, Summarized results of A showing the percentage of samples with overexpression of the indicated mRNA within the top 25% of BCLW expression in DLBCL. The +2, +3, and +4 indicate the number of additional antiapoptotic BCL2 family members that were also overexpressed. C–F (left), Gene-expression data showing association between BCLW and BCL2, BCLX, MCL1, and A1 mRNA in follicular lymphoma (FL; C), MZL (D), MCL (E), and HL (F) stratified on the basis of the top 25% of BCLW expression. Each line represents one sample. C–F (right), Summarized results of left showing the percentage of samples with overexpression of the indicated mRNA within the top 25% of BCLW expression.

Figure 6.

Non-Hodgkin and Hodgkin lymphomas preferentially overexpress select antiapoptotic BCL2 family members. A, Gene-expression data showing association between BCLW and BCL2, BCLX, MCL1, and A1 mRNA in DLBCL stratified on the basis of the top 25% of BCLW expression. Each line represents one sample. B, Summarized results of A showing the percentage of samples with overexpression of the indicated mRNA within the top 25% of BCLW expression in DLBCL. The +2, +3, and +4 indicate the number of additional antiapoptotic BCL2 family members that were also overexpressed. C–F (left), Gene-expression data showing association between BCLW and BCL2, BCLX, MCL1, and A1 mRNA in follicular lymphoma (FL; C), MZL (D), MCL (E), and HL (F) stratified on the basis of the top 25% of BCLW expression. Each line represents one sample. C–F (right), Summarized results of left showing the percentage of samples with overexpression of the indicated mRNA within the top 25% of BCLW expression.

Close modal

We expanded our analysis to evaluate the expression patterns of the other antiapoptotic BCL2 family members in follicular lymphoma, MZL, MCL, and HL in the top 25% of BCLW-overexpressing lymphomas (Fig. 6C–F). In MZL, MCL, and HL, of the lymphomas that overexpressed BCLW, 24.3% to 33.3% only overexpressed BCLW, whereas only 12.2% of the follicular lymphoma that overexpressed BCLW solely overexpressed BCLW. Instead, BCLW-overexpressing follicular lymphoma preferentially overexpressed (42.9%) two other antiapoptotic BCL2 family members (Fig. 6C). We show that two or more antiapoptotic BCL2 family members are frequently overexpressed together, suggesting that these lymphomas likely rely on more than one for survival. Collectively, our data provide evidence that BCLW is a previously unappreciated contributor to 6 different B-cell lymphomas and that B-cell lymphomas frequently select for the overexpression of BCLW as well as other antiapoptotic BCL2 family members.

Until our recent investigation of Burkitt lymphoma and DLBCL showing that BCLW significantly contributes to these lymphomas (9), BCLW was considered only necessary for spermatogenesis and dispensable for hematopoietic cells (7, 8, 22, 23). Here, we significantly expand upon the lymphomas that select for BCLW overexpression, revealing that both aggressive and indolent lymphomas and non-Hodgkin and Hodgkin lymphomas commonly overexpress BCLW. Our data show that BCLW is overexpressed at a frequency that is similar to or greater than, depending on the lymphoma subtype, the quintessential BCL2. Unexpectedly, follicular lymphoma, in which BCL2 overexpression is diagnostic, also overexpressed BCLW at the same frequency as BCL2 overexpression. We also determined a significant inverse association existed between BCLW and BCL2 levels in DLBCL, MZL, and HL, indicating these lymphomas preferentially select for overexpression of one over the other. For HL, BCLW was preferentially selected for overexpression over BCL2, with samples containing high levels of BCLW and low levels of BCL2. A fraction of each lymphoma subtype evaluated preferentially selected for BCLW overexpression alone. Together, our results indicate that BCLW has a significant role in B-cell lymphomas and is likely as important as BCL2.

Our results also provide an unprecedented evaluation of all five antiapoptotic BCL2 family members in the six major B-cell lymphoma subtypes. Analyses of the antiapoptotic BCL2 family members revealed overexpression of more than one family member is a frequent event in all six lymphoma subtypes evaluated. These data suggest a threshold level of expression of BCL2 antiapoptotic proteins may be needed and once it is reached, overexpression of others is not selected for. However, overall, when mean expression levels in the six lymphomas were compared to normal B cells, there appeared to be preferential selection for BCLW, BCL2, and/or BCLX overexpression, but not MCL1 or A1. Interestingly, in Burkitt lymphoma, the mean expression of MCL1 and A1 was actually lower than their levels in normal human B cells, suggesting MCL1 and A1 may not significantly contribute to this lymphoma. However, in individual lymphomas of all subtypes, of those that overexpressed BCLW, a similar low frequency of cooverexpression of BCL2, BCLX, MCL1, or A1 was detected. Our data provide evidence that each B-cell lymphoma subtype is heterogeneous at the antiapoptotic BCL2 family level, but most have a preference for which family member to overexpress.

Therapies targeted at antiapoptotic BCL2 family members have entered the clinic. ABT-737 and ABT-263 (navitoclax), which inhibit BCL2, BCLX, and BCLW showed promise in chronic lymphocytic leukemia (CLL; ref. 24). The agents TW-37 and Apogossypolone also target multiple antiapoptotic BCL2 family members (25). However, ABT-263, TW-37, and Apogossypolone cause acute cytotoxicity from thrombocytopenia (25), which has led to their limited use and the development of compounds only targeting one of the family members. The more selective BCL2 inhibitor, ABT-199 (venetoclax) does not deplete platelets, but causes neutropenia when used with other drugs (26). Targeting BCL2 with a BCL2 antisense DNA oligonucleotide (PNT100) leads to lymphopenia (27). The BCLX-specific inhibitors, A1155463 and A1331852, have been evaluated in non–small cell lung cancer (NSCLC), breast cancer, and T-cell acute lymphoblastic leukemia cell lines (28). MCL1 targeted therapy has shown single-agent killing of multiple myeloma and NSCLC (29). So far, only ABT-199 has been FDA approved for treatment of relapsed or refractory CLL (30). Because we have exposed BCLW as a critical contributor to B-cell lymphomagenesis, BCLW-specific inhibitors will need to be developed and tested, alone and in combination with, other selective BCL2 family inhibitors for future lymphoma treatment.

Despite significant investments into the development of small-molecule therapeutics, many compounds still fail in the clinic due to acquired resistance. Resistance mechanisms to specific antiapoptotic BCL2 family inhibitors can occur at least, in part, by overexpression of other BCL2 family members. For example, although mantle cell lymphomas are more resistant to ABT-199 when they express increased levels of BCLX and MCL1, knockdown of MCL1 confers sensitivity to ABT-199 (31). In addition, development of ABT-199 resistance was linked to increased BCLX and MCL1 levels (BCLW not evaluated) in two DLBCL cell lines (32). In addition, we recently reported that BCLW or BCL2 overexpression conferred resistance to ABT-263 in Burkitt lymphoma cells (9). Therefore, it will be important to monitor other antiapoptotic BCL2 family members when using therapies that target one or more of them.

Our comprehensive evaluation of BCLW and four other antiapoptotic BCL2 family members in six different lymphomas reveals the contribution these genes have to B-cell lymphomas by the frequency they are selected for overexpression, individually and with each other. Importantly, BCLW overexpression was selected for in all six non-Hodgkin and Hodgkin lymphomas evaluated. Furthermore, each of the lymphomas frequently selected other antiapoptotic BCL2 family members to overexpress in addition to BCLW. These results illustrate the complexity of treating these lymphomas, as they depend on one or more antiapoptotic BCL2 family members. At a time when inhibitors are actively being developed for use in the clinic to target individual BCL2 family members, our analysis demonstrates the importance of considering the expression of all antiapoptotic BCL2 family members for improved targeted therapies and to minimize the occurrence of drug resistance.

No potential conflicts of interest were disclosed.

Conception and design: C.M. Adams, R. Mitra, J.Z. Gong, C.M. Eischen

Development of methodology: C.M. Adams, J.Z. Gong

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.M. Adams, J.Z. Gong

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.M. Adams, R. Mitra, J.Z. Gong, C.M. Eischen

Writing, review, and/or revision of the manuscript: C.M. Adams, R. Mitra, J.Z. Gong, C.M. Eischen

Study supervision: C.M. Eischen

We thank Dr. Zhijiu Zhong for performing the BCLW IHC in the Translational Pathology Shared Resource core supported by NCI Cancer Center support grant P30CA056036.

Support for this study was provided by NCI R01CA148950 and R21CA184352 (to C.M. Eischen), NCI P30CA056036, and the Sidney Kimmel Cancer Center.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Hanahan
D
,
Weinberg
RA
. 
Hallmarks of cancer: the next generation
.
Cell
2011
;
144
:
646
74
.
2.
Czabotar
PE
,
Lessene
G
,
Strasser
A
,
Adams
JM
. 
Control of apoptosis by the BCL-2 protein family: implications for physiology and therapy
.
Nat Rev Mol Cell Biol
2014
;
15
:
49
63
.
3.
Bakhshi
A
,
Jensen
JP
,
Goldman
P
,
Wright
JJ
,
McBride
OW
,
Epstein
AL
, et al
Cloning the chromosomal breakpoint of t(14;18) human lymphomas: clustering around JH on chromosome 14 and near a transcriptional unit on 18
.
Cell
1985
;
41
:
899
906
.
4.
Thieblemont
C
,
Briere
J
. 
MYC, BCL2, BCL6 in DLBCL: impact for clinics in the future?
Blood
2013
;
121
:
2165
6
.
5.
Lenz
G
,
Wright
GW
,
Emre
NC
,
Kohlhammer
H
,
Dave
SS
,
Davis
RE
, et al
Molecular subtypes of diffuse large B-cell lymphoma arise by distinct genetic pathways
.
Proc Natl Acad Sci U S A
2008
;
105
:
13520
5
.
6.
Dave
SS
,
Fu
K
,
Wright
GW
,
Lam
LT
,
Kluin
P
,
Boerma
EJ
, et al
Molecular diagnosis of Burkitt's lymphoma
.
N Engl J Med
2006
;
354
:
2431
42
.
7.
Print
CG
,
Loveland
KL
,
Gibson
L
,
Meehan
T
,
Stylianou
A
,
Wreford
N
, et al
Apoptosis regulator bcl-w is essential for spermatogenesis but appears otherwise redundant
.
Proc Natl Acad Sci U S A
1998
;
95
:
12424
31
.
8.
Ross
AJ
,
Waymire
KG
,
Moss
JE
,
Parlow
AF
,
Skinner
MK
,
Russell
LD
, et al
Testicular degeneration in Bclw-deficient mice
.
Nat Genet
1998
;
18
:
251
6
.
9.
Adams
CM
,
Kim
AS
,
Mitra
R
,
Choi
JK
,
Gong
JZ
,
Eischen
CM
. 
BCL-W has a fundamental role in B cell survival and lymphomagenesis
.
J Clin Invest
2017
;
127
:
635
50
.
10.
Agarwal
B
,
Naresh
KN
. 
Bcl-2 family of proteins in indolent B-cell non-Hodgkin's lymphoma: study of 116 cases
.
Am J Hematol
2002
;
70
:
278
82
.
11.
Bai
M
,
Papoudou-Bai
A
,
Horianopoulos
N
,
Grepi
C
,
Agnantis
NJ
,
Kanavaros
P
. 
Expression of bcl2 family proteins and active caspase 3 in classical Hodgkin's lymphomas
.
Hum Pathol
2007
;
38
:
103
13
.
12.
Barrett
T
,
Edgar
R
. 
Gene expression omnibus: microarray data storage, submission, retrieval, and analysis
.
Methods Enzymol
2006
;
411
:
352
69
.
13.
Irizarry
RA
,
Hobbs
B
,
Collin
F
,
Beazer-Barclay
YD
,
Antonellis
KJ
,
Scherf
U
, et al
Exploration, normalization, and summaries of high density oligonucleotide array probe level data
.
Biostatistics
2003
;
4
:
249
64
.
14.
Gautier
L
,
Cope
L
,
Bolstad
BM
,
Irizarry
RA
. 
affy–analysis of Affymetrix GeneChip data at the probe level
.
Bioinformatics
2004
;
20
:
307
15
.
15.
Johnson
WE
,
Li
C
,
Rabinovic
A
. 
Adjusting batch effects in microarray expression data using empirical Bayes methods
.
Biostatistics
2007
;
8
:
118
27
.
16.
Wang
P
,
Lushnikova
T
,
Odvody
J
,
Greiner
TC
,
Jones
SN
,
Eischen
CM
. 
Elevated Mdm2 expression induces chromosomal instability and confers a survival and growth advantage to B cells
.
Oncogene
2008
;
27
:
1590
8
.
17.
Adams
CM
,
Hiebert
SW
,
Eischen
CM
. 
Myc induces miRNA-mediated apoptosis in response to HDAC inhibition in hematologic malignancies
.
Cancer Res
2016
;
76
:
736
48
.
18.
Sehn
LH
,
Gascoyne
RD
. 
Diffuse large B-cell lymphoma: optimizing outcome in the context of clinical and biologic heterogeneity
.
Blood
2015
;
125
:
22
32
.
19.
Iqbal
J
,
Neppalli
VT
,
Wright
G
,
Dave
BJ
,
Horsman
DE
,
Rosenwald
A
, et al
BCL2 expression is a prognostic marker for the activated B-cell-like type of diffuse large B-cell lymphoma
.
J Clin Oncol
2006
;
24
:
961
8
.
20.
Ott
G
,
Rosenwald
A
. 
Molecular pathogenesis of follicular lymphoma
.
Haematologica
2008
;
93
:
1773
6
.
21.
Kahl
BS
,
Yang
DT
. 
Follicular lymphoma: evolving therapeutic strategies
.
Blood
2016
;
127
:
2055
63
.
22.
Gibson
L
,
Holmgreen
SP
,
Huang
DC
,
Bernard
O
,
Copeland
NG
,
Jenkins
NA
, et al
bcl-w, a novel member of the bcl-2 family, promotes cell survival
.
Oncogene
1996
;
13
:
665
75
.
23.
O'Reilly
LA
,
Print
C
,
Hausmann
G
,
Moriishi
K
,
Cory
S
,
Huang
DC
, et al
Tissue expression and subcellular localization of the pro-survival molecule Bcl-w
.
Cell Death Differ
2001
;
8
:
486
94
.
24.
Delbridge
AR
,
Grabow
S
,
Strasser
A
,
Vaux
DL
. 
Thirty years of BCL-2: translating cell death discoveries into novel cancer therapies
.
Nat Rev Cancer
2016
;
16
:
99
109
.
25.
Besbes
S
,
Mirshahi
M
,
Pocard
M
,
Billard
C
. 
New dimension in therapeutic targeting of BCL-2 family proteins
.
Oncotarget
2015
;
6
:
12862
71
.
26.
Souers
AJ
,
Leverson
JD
,
Boghaert
ER
,
Ackler
SL
,
Catron
ND
,
Chen
J
, et al
ABT-199, a potent and selective BCL-2 inhibitor, achieves antitumor activity while sparing platelets
.
Nat Med
2013
;
19
:
202
8
.
27.
Tolcher
AW
,
Rodrigueza
WV
,
Rasco
DW
,
Patnaik
A
,
Papadopoulos
KP
,
Amaya
A
, et al
A phase 1 study of the BCL2-targeted deoxyribonucleic acid inhibitor (DNAi) PNT2258 in patients with advanced solid tumors
.
Cancer Chemother Pharmacol
2014
;
73
:
363
71
.
28.
Leverson
JD
,
Phillips
DC
,
Mitten
MJ
,
Boghaert
ER
,
Diaz
D
,
Tahir
SK
, et al
Exploiting selective BCL-2 family inhibitors to dissect cell survival dependencies and define improved strategies for cancer therapy
.
Sci Transl Med
2015
;
7
:
279ra40
.
29.
Leverson
JD
,
Zhang
H
,
Chen
J
,
Tahir
SK
,
Phillips
DC
,
Xue
J
, et al
Potent and selective small-molecule MCL-1 inhibitors demonstrate on-target cancer cell killing activity as single agents and in combination with ABT-263 (navitoclax)
.
Cell Death Dis
2015
;
6
:
e1590
.
30.
Cang
S
,
Iragavarapu
C
,
Savooji
J
,
Song
Y
,
Liu
D
. 
ABT-199 (venetoclax) and BCL-2 inhibitors in clinical development
.
J Hematol Oncol
2015
;
8
:
129
.
31.
Chiron
D
,
Dousset
C
,
Brosseau
C
,
Touzeau
C
,
Maiga
S
,
Moreau
P
, et al
Biological rational for sequential targeting of Bruton tyrosine kinase and Bcl-2 to overcome CD40-induced ABT-199 resistance in mantle cell lymphoma
.
Oncotarget
2015
;
6
:
8750
9
.
32.
Choudhary
GS
,
Al-Harbi
S
,
Mazumder
S
,
Hill
BT
,
Smith
MR
,
Bodo
J
, et al
MCL-1 and BCL-xL-dependent resistance to the BCL-2 inhibitor ABT-199 can be overcome by preventing PI3K/AKT/mTOR activation in lymphoid malignancies
.
Cell Death Dis
2015
;
6
:
e1593
.