The pro-apoptotic paradox: the BH3-only protein Bcl-2 interacting killer (Bik) is prognostic for unfavorable outcomes in breast cancer.

INTRODUCTION

Breast cancer is the most prevalent cancer in women worldwide accounting for the highest number of cancer-associated deaths [ 1 – 3 ]. Cancer is a heterogeneous disease with many factors affecting accurate estimation of prognosis, treatment decisions and quality of life. Identification of clinical biomarkers and elucidation of their roles has paved the way for a better understanding of the disease and improved patient survival [ 4 , 5 ]. Since many unknown factors still determine variable clinical outcomes, we decided to search for additional prognostic markers. We reasoned that due to inherent genomic instability, compensatory mutations would uncouple upstream molecular drivers from clinical outcome. We thus decided to analyze apoptotic markers and in particular focussed on the Bcl-2 family of proteins that are downstream effectors of cell-death or cell-survival decisions [ 6 , 7 ]. The BH3-only proteins are a subgroup of the Bcl-2 family of proteins [ 6 , 8 ]. BH3-only proteins have both distinct and overlapping developmental and tissue-specific expression patterns, highlighting both unique and redundant roles in cellular processes [ 6 , 8 ]. In addition to regulating apoptosis, members of this family also interact with such diverse cellular pathways as autophagy, checkpoint regulation and metabolism [ 8 ]. Therefore we examined whether specific BH3-only proteins were prognostic for breast cancer patient outcome and correlated expression with specific biological pathways.

Analysis of gene expression datasets linked to clinical outcomes can identify biomarkers that are significantly regulated at the transcriptional level. Indeed many of the BH3-only proteins are transcriptionally regulated with subsequent effects on apoptotic and autophagic pathways [ 8 , 9 ]. Genomic stress mediated upregulation of multiple BH3-only genes is dependent on the p53 tumor suppressor [ 6 ]. Depending on the nature of stress-stimuli Bid, Bim and Puma are also regulated by FOXO3a while Bim and Puma are subject to E2F-1 dependent regulation [ 6 , 10 – 12 ]. Additionally, NF-κB modulates transcriptional activity of Bad and Noxa in squamous cell carcinoma and neutrophils respectively [ 13 ]. Although Bik transcription is dependent on p53, various other regulators including TGF-β, E2F, TEF and PARbZIP transcription factors regulate gene expression [ 14 , 15 ]. Thus multiple pathways that determine cancer cell fate, regulate the gene expression of multiple BH3-only genes.

Importantly, BH3- only protein levels are also regulated at the post-transcriptional level [ 8 , 14 ]. For instance, Bik is degraded by both proteasomal- and autophagic pathways. [ 16 , 17 ]. In addition, post-translational modifications of BH3-proteins alter protein function and can act as a signalling switch from apoptosis to cell survival. It is thus important to compare gene expression studies with protein analysis to generate mechanistic models for clinical outcome.

Thus, we investigated the prognostic potential of BH3-only gene expression followed by validation at the protein level. Of five BH3-only genes analyzed, Bik was the only independent prognostic indicator for breast cancer patient recurrence. Bik prognostic value was not dependent on anti-apoptotic gene expression, suggesting that Bik-high tumors were not addicted to anti-apoptotic proteins. However we identified an association between autophagy marker ATG5 and Bik. Our work suggests that Bik may be an indicator of enhanced autophagy and cell survival, contributing to disease recurrence.

RESULTS

BH3-only gene expression and breast cancer patient outcome

The BH3-only proteins are comprised of multiple family members with distinct and overlapping roles depending on tissue-type and cellular signals [ 8 ]. Whether unique BH3-only proteins regulate breast cancer pathophysiology is not yet clear, and this understanding would aid in efforts to dissect relevant molecular pathways in disease. Therefore, we analyzed a gene expression microarray to identify which, if any, BH3-only genes were associated with clinical outcome. As described in previous studies, this patient cohort contained all pathological subtypes of breast cancer (64% Luminal, 5% HER2 amplified and 32% Triple negative) of variable grade and stage with 49% recurrence and 27% death with complete long-term follow-up after diagnosis (Figure 1 ) [ 18 , 19 ]. To identify markers of early recurrence, we assessed five-year disease-free and overall survival. The hazard ratio (HR) indicates the likelihood of relapse or death of the marker-high group versus the marker-low group. The HR values and statistical significance of BH3-only gene expression alongside clinicopathological factors of age, tumor size, mitotic grade, overall grade, vascular invasion, menopausal status, hormone receptor status and Her2 status are shown in Table 1 . Of the clinicopathological features, mitotic grade (HR 2.05, p=0.002), overall grade (HR 1.95, p=0.009) and progesterone receptor (PR) status (HR 0.66, p=0.05) were significantly associated with disease-free survival. Gene expression data was retrieved for the BH3-members Bad, Bid, Bik, Noxa and Puma. Univariate Cox regression analysis identified relatively high mRNA levels of only Bid (HR 1.75, p=0.011) and Bik (HR 1.79, p=0.021) to be significantly associated with poor survival outcomes.

Bik gene expression is a significant and independent predictor of breast cancer survival

In order to assess whether Bik or Bid were independently prognostic, we performed multivariate Cox analysis. We carried out stepwise multivariate analysis on all variables associated with recurrence (Bid, Bik, PR, overall grade and mitotic grade). Only Bik and mitotic grade were retained as independent variables. We then performed pairwise multivariate analysis to assess the relationships between variables ( Supplementary Table S1 ). Bid was independent of Bik (HR 1.78, p=0.009) but was dependent on mitotic-grade and overall grade. On the other hand, Bik was independent of mitotic grade, overall grade, PR status and Bid. ROC curve analysis was performed with recurrence as the classification variable to identify an appropriate cut-point (≤1.71) that dichotomized patients into Bik-high and Bik-low groups. Kaplan-Meier analysis of those groups for disease-free survival revealed worse prognosis for Bik high patients with a HR of 1.78 (p=0.019) (Figure 2A ). Similar analysis was performed for overall survival. Kaplan-Meier analysis revealed a higher risk of death for Bik high patients with a hazard ratio of 2.05 (p=0.021) (Figure 2B ). Thus, elevated Bik mRNA level was an independent prognostic indicator of recurrence and death.

Elevated Bik protein levels are associated with poor prognosis of breast cancer patients

We were curious as to whether Bik protein was also prognostic of outcome. To rigorously test the prognostic value of Bik, we queried a different cohort of patients. We examined a tissue microarray (TMA) from 152 treatment naïve patients (Dataset-2) composed of luminal (76%), HER2 amplified (37%) and triple negative (13.5%) pathological subtypes of variable grade and stage with 11 % recurrence and 16.5 % death (Figure 1 ). The statistical significance of clinicopathological factors of age, tumor size, mitotic grade, overall grade, vascular invasion, estrogen receptor (ER)/PR and HER2 protein levels are shown in Table 2 . Mitotic grade (HR 3.91, p=0.006), overall grade (HR 3.29, p=0.037) and ER status (HR 0.37, p=0.05) were significantly associated with disease-free survival. The TMA was immunostained with an anti-Bik antibody and tumor cells within each tissue core were analyzed for Bik staining. Bik was localized to the cytoplasm and staining intensity for each core was scored on a relative scale between 0-3 by personnel blinded to outcomes (Figure 3A ). A Bik score generated from the average value of each associated core was assigned for each patient. ROC analysis identified a cutpoint of ≥1.5. Kaplan-Meier analysis confirmed that high levels of Bik protein were also associated with worse disease-free survival (Figure 3B ). Patients with high Bik protein levels had poor disease-free survival with a hazard ratio of 3.59 (p=0.007) compared to Bik low patients. Similarly Bik-high patients had a worse overall survival with a hazard ratio of 3.40 (p=0.04) compared with Bik-low patients (Figure 3C ). Multivariate analysis showed that mitotic grade and Bik protein expression were the only two variables that retained significance (Table 2 ). We then carried out pairwise multivariate analysis to identify dependence of Bik prognostic signature on pathophysiological factors ( Supplementary Table S2 ). Similar to the gene expression analysis, Bik prognostic signature was independent of mitotic grade, overall grade and ER status. Altogether, mRNA and protein analysis identify Bik as a novel independent marker for breast cancer clinical outcomes.

Bik association with poor outcomes is not explained by compensatory increase in anti-apoptotic gene or protein expression

Bik has a well-documented role as a pro-apoptotic protein, so we were initially surprised that higher levels of Bik mRNA and protein were associated with poor clinical outcomes. A possible explanation was that anti-apoptotic protein levels were upregulated and compensated for Bik activity. For example, Bcl-2 and Bik levels are significantly correlated in multiple myeloma cell lines where Bcl-2 inhibits Bik-induced apoptosis [ 15 ]. Therefore, we investigated whether a similar Bik:Bcl-2 association was recapitulated in the clinical samples in our study. We found no correlation with Bik and Bcl-2 gene expression levels (r=−0.02, P=0.79), which did not support a Bcl-2-mediated tumor adaptation model for Bik high tumors. We next examined levels of anti-apoptotic genes in relation to disease-free survival. Data was available for four anti-apoptotic genes (Bcl-2, Bcl-xL, Mcl-1 and Bcl-w). Univariate Cox analysis identified that elevated levels of Bcl-2, Bcl-xL and Mcl-1transcripts were associated with favourable disease-free survival (Table 3 ). Stepwise multivariate Cox analysis demonstrated that all variables remained significant and were independent of each other. In KM survival analyses (Figure 4 A-C ) elevated levels of Bcl-2, Bcl-xL and Mcl-1 were associated with significantly improved disease-free survival with HRs of 0.41, 0.65 and 0.55, respectively. This is opposite to what would be expected if elevated Bcl-2, Bcl-xL and Mcl-1 were indeed inhibiting apoptosis as this would be associated with higher likelihood of recurrence with a HR greater than 1. Finally, using the Bik/Bcl-2 relationship as a model, we stratified patients with respect to their individual Bik and Bcl-2 gene expression levels and looked for correlations between subgroup expression levels and recurrence ( Figure-S1 ). If the tumor adaption model were dominant, we would expect a lower recurrence rate in the Bik-high patient subgroup with low Bcl-2 and presumably unencumbered apoptotic signalling, whereas the Bik-high patients with high Bcl-2 would show significantly higher recurrence rates. However, we see no significant difference between these two subgroups, and in fact the majority of patients show the opposite effect. Thus elevated Bik levels and diminished Bcl-2 levels are both prognostic for poor outcomes, but potentially function in separate pathways.

We next examined Bcl-2 protein levels in the previously described TMA from Dataset-2. Tumor cells within each core were assessed for Bcl-2 protein levels (Figure 5A ). ROC analysis identified a cutpoint of >0. Kaplan-Meier analysis confirmed that high levels of Bcl-2 protein were associated with improved disease-free survival (Figure 5B ). Patients with high Bcl-2 protein levels had a higher recurrence-free survival with a hazard ratio of 0.13 (p<0.002) compared to Bcl-2 low patients. Similarly Bcl-2-high patients had a higher overall survival with a hazard ratio of 0.18 (p=0.01) compared with Bcl-2-low patients (Figure 5C ). Additionally, we performed a multivariate Cox analysis of Bik and Bcl-2 protein levels; Bik and Bcl-2 were found to be independent variables (Table 4 ). Altogether, our analyses indicate that Bik prognostic value is independent of anti-apoptotic protein levels, suggesting that a non-apoptotic activity of Bik contributes to poor clinical outcomes.

High Bik and ATG5 levels are associated with worse patient outcome

We sought to identify a tumor-promoting pathway that accounts for the poor prognosis in Bik high patients. In addition to stimulating apoptotic pathways, Bik facilitates autophagy [ 20 – 22 ] and enhanced autophagy is associated with poor clinical outcomes [ 23 – 25 ]. Thus, using the gene expression microarray as a screening tool, we looked for a correlation between high Bik expression levels and markers of autophagy. Four autophagy gene sets (ATG5, ATG7, Beclin-1 and p62) were captured in the gene expression microarray ( Supplementary Table S3 ). ATG5, ATG7 and p62 showed clear association of high autophagy flux with poor patient outcomes. ATG5 showed the best ROC values so we analyzed ATG5 further. The ATG5 gene product is an E3 ubiquitin ligase involved in elongation of autophagosomal membranes that facilitates tumor survival in nutrient deprivation [ 26 , 27 ]. Overexpression of ATG5 is associated with aggressive tumors in squamous cell carcinoma, colorectal and prostate cancers and chemoresistant tumors in gastric cancer [ 26 – 29 ]. To examine dependence of Bik and ATG5, we performed ROC curve analysis and identified a cut-point (>0.939) that dichotomized patients into ATG5-high and -low groups. Kaplan-Meier survival analysis of patients with respect to ATG5 gene expression levels demonstrated that patients with high levels of ATG5 mRNA had a disease-free survival HR of 1.69 (p=0.018) compared to ATG5-low patients (Figure 6A ). As well, ATG5-high patients had a lower overall survival with a HR of 1.90 (p=0.04) compared to ATG5-low patients (Figure 6B ). In order to dissect the relationship between Bik and ATG5, we performed multivariate Cox analysis and found that Bik and ATG5 were independent variables (p=0.02 and p=0.02 respectively). We performed a bivariable KM analysis in order to further examine potential relationships between Bik and ATG5 (Figure 6C ). We found that while ATG5-high was a stand-alone significant prognostic indicator (Figure 6A ), Bik-high was significant only when ATG5 was also high (Figure 6C ), suggesting that the mechanism whereby high Bik contributes to poor outcomes may be through an ATG5-dependent autophagy pathway. To interrogate this further, we generated a scatter plot to correlate patients with respect to their individual Bik and ATG-5 gene levels and examined recurrence rates for each subgroup (Figure 6D ). We observed that Bik-high patients with high levels of ATG5 mRNA had 88% recurrence (p=0.009) compared to a 44% recurrence rate in the Bik high/ATG5-low patient subgroup. Of the Bik-low patient subgroups, there was no significant difference in recurrence rates relative to ATG-5 levels. Collectively, this data supports a mutual association of Bik and ATG5 in determining patient outcome.

DISCUSSION

We identified that the BH3-only protein Bik was a novel prognostic marker for breast cancer. Bik levels were prognostic for disease-free and overall survival in two independent cohorts of primary breast cancer patients. Importantly, elevated Bik levels were associated with poor outcomes (average HR= 2.75). This result suggested that Bik may have tumor-promoter activity. However, there is substantial evidence that Bik is a tumor-suppressor. For example, the Bik gene was mutated in peripheral B-cell lymphomas, deleted in gliomas, head and neck and colorectal cancers or silenced in renal cell carcinomas [ 14 , 30 – 33 ]. Additionally, a genetically-engineered mutant of Bik (BikDD) induced apoptosis in pancreatic, breast and colon cancer models resulting in tumor clearance [ 34 – 37 ]. On the other hand, Bik expression was elevated in breast, pancreas, multiple myeloma and colon cancers, suggestive of a tumor-promoting role for Bik [ 38 – 41 ]. These studies, however, did not correlate Bik levels to outcome. As far as we know, our study is the first to report that high Bik transcript and protein levels correlated with poor outcomes.

How Bik association with poor clinical outcome is manifested at the molecular level is unclear. Bik could either be a marker of aggressive tumors that have evolved anti-apoptotic strategies, or Bik could have a direct tumor-promoting role. To address the first point, it is possible that tumors upregulate Bcl-2 expression. This compensatory mechanism to ablate apoptotic pathways is well-documented in multiple myeloma and non-small cell lung cancer (NSCLC) tumor models [ 15 , 42 ]. Of particular interest, Lu et al reported that poorly surviving non-small cell lung cancer (NSCLC) patients had high Bik expression and compensatory increase in anti-apoptotic protein Bcl-2 led to tumor adaptation [ 14 , 42 ]. Contrary to their findings, in breast cancer we found that Bik and Bcl-2 were independent variables and the increase in Bik expression was not correlated with expression levels of 4 anti-apoptotic members of the Bcl-2 family. Furthermore, the anti-apoptotic genes Bcl-2, Bcl-xL and Mcl-1 were all independently prognostic of favourable outcome, which argues against a simple anti-apoptotic/tumor-promoting role for these genes. In addition, loss of apoptosis in the face of Bik elevation may involve dysregulation of downstream apoptotic effectors and their regulators. Additionally, recent studies report BH3-only proteins can lead to tumorigenesis by compromising mitochondrial integrity, facilitating low caspase activation and chronic genomic damage [ 43 ]. It is possible that Bik-expressing tumors have greater genomic instability driving tumor adaptation and the development of more aggressive subtypes of cancer. Taken together, our data suggests that a complex network may be at play.

Variable clinical outcomes have correlated with anti-apoptotic protein expression dependent on tumor type. Negative outcome associations are seen in Hodgkin's lymphoma, myeloid leukemia and follicular lymphoma, and positive outcome associations are seen in NSCLC, breast and liver cancers [ 44 – 47 ]. In our case, Bcl-2 levels were associated with good outcome, and this is in agreement with other studies [ 44 , 48 , 49 ]. While Bcl-2 can also inhibit cell proliferation and autophagy [ 21 , 50 – 53 ], the mechanism behind the positive prognostic signature of the anti-apoptotic genes in our study, is not clear.

It is possible that Bik stimulates tumor-promoting autophagy. Bik stimulates autophagy through Bcl-2-dependent and -independent mechanisms. In the first case, Bik indirectly stimulates autophagy by ablating Bcl-2:Beclin-1 interactions [ 21 ]. Specifically, Chang et al reported that Bcl-2 interaction with the ER-resident protein, Naf-1 was crucial for inhibition of Beclin-1-mediated autophagy and Bik suppressed this Bcl-2/Naf-1 interaction [ 21 ]. In the case of the Bcl-2-independent autophagy pathway, ectopic expression of Bik induced autophagy in Bcl2 −/ − MEFs [ 20 ]. Bik was a target of autophagy and also actively contributed to autophagy induction [ 16 ]. Finally, silencing of endogenous Bik expression in the breast cancer cell line MDA-MB-231 downregulated crucial components of autophagic machinery [ 22 ]. We found that high Bik expression in breast cancer patients was significantly correlated with increased levels of the autophagy marker, ATG5. ATG5 is a prognostic indicator of squamous cell carcinoma, prostate, and colorectal cancers relapse [ 26 , 29 , 54 ]. This is consistent with a model whereby Bik induced-autophagy promotes breast cancer relapse (Figure 7 ). Additionally, we speculate that anti-apoptotic Bcl-2-like proteins act downstream of Bik and/or directly antagonize Bik to inhibit autophagic processes. This may explain in part, why elevated Bcl-2 is associated with favourable outcomes. This multiple interaction network of Bik, Bcl-2 and the marker of autophagy ATG5 may then have subset interactions, and thus retain statistical independence. As well, we expect that other signals stimulate ATG5 and autophagy independently of Bik, again resulting in the retention of the independent prognostic value of Bik and ATG5. Although we propose that Bik promotes tumor survival by facilitating autophagy, other possible mechanisms could lead to Bik prognostic association with poor outcomes.

Altogether, multiple models of Bik-induced tumor survival signaling could be at play and lead to the observation of poor clinical outcome. Our study results give rationale to investigate non-canonical roles of Bik related to tumor growth. In addition, exploration of interactions of Bik with autophagy and clinical response in breast cancer patients could set the stage for novel therapeutic regimens in the future.

MATERIALS AND METHODS

Reagents

Bik antibody (sc-1710) used for IHC was obtained from Santa Cruz, USA. Bcl-2 antibody (M0887) used for IHC was from Dako, USA.

Gene expression and tissue microarray analyses

Pre-treatment tumor samples were accessed from the Alberta Cancer Research Biobank/CBCF Tumor Bank. Patient information was collected under research ethics board approval (HREB Biomedical). All patients received guideline based standard therapy. Gene expression data from 175 primary breast cancer patients was determined from analysis of Agilent Whole Human Genome Arrays and has been described previously [ 18 ]. The messenger RNA (mRNA) expression data sets were deposited (GEO accession ID GSE22820 ) in a publicly accessible database. This cohort is referred to as Dataset-1 in this study. For Dataset-2, tissue microarrays (TMA) were generated from formalin fixed tumor samples obtained for 152 primary breast cancer patients as described previously [ 55 ]. Briefly, the TMA was prepared by punching at least two representative tumor cores from pre-treatment formalin fixed paraffin embedded (FFPE) blocks and embedding them in recipient paraffin blocks. Glass TMA slides were generated. TMAs were deparaffinized in xylene and serially rehydrated in 95% to 0% ethanol solution. Epitope retrieval was performed by microwaving in citrate buffer (10mM citrate, 0.05% Tween-20; pH 6.0). Immunostaining on these TMAs was performed using anti-Bik antibody (1:300; Santa Cruz Biotechnology, Inc. USA) and anti-Bcl-2 antibody (1:300; Dako, USA). Dako EnVision+System-HRP (Dako, USA) system was used for signal amplification with DAB substrate.

Scoring and quantification of immunohistochemical staining

The scoring of the immunostained TMA was performed in an outcome blinded fashion according to training and guidelines from the study's breast pathologist. Images were acquired at 40X magnification using a Carl Zeiss Axio Observer microscope (Carl Zeiss, Germany)

Statistical analysis

Statistical analysis was performed using MedCalc version 15 (Ostend, Belgium). Receiver operator characteristic (ROC) curve analysis was used as a measure of sensitivity versus specificity for determining the optimum cutpoint value for gene expression and TMA scores. Disease-free and overall survival were calculated by Kaplan-Meier analysis. Significant differences between KM curves were measured by log-rank test. Uni- and multivariate analyses for recurrence or death was carried out using Cox regression methods.

SUPPLEMENTARY FIGURE AND TABLES

Operating grants from the Alberta Institute for Health Research (AIHS)/Pfizer program (ISG, JRM), Alberta Cancer Foundation (ISG, JRM) and Canadian Breast Cancer Foundation (CBCF) Prairies/NWT Chapter (ISG) supported this research. SD received operating grants from Canadian Breast Cancer Foundation (CBCF) Prairies/NWT Chapter. JRM received operating grants from Alberta Breast Cancer Initiatives program. We thank Jennifer Dufour for technical assistance. We thank Devon Germain and Krista Vincent for help with statistical analysis. We thank Dr. Alan Underhill and all members of the Goping lab for helpful discussions.

CONFLICTS OF INTEREST

The authors declare no conflict of interest

Author contributions VP and ISG designed the study; VP, DG and LV performed experiments and collected data; VP and Jo Ha performed statistical analysis; SD, JRM and JHugh provided expert opinions on oncology and pathology; VP and ISG wrote the manuscript and all authors provided input.

Author contributions

VP and ISG designed the study; VP, DG and LV performed experiments and collected data; VP and Jo Ha performed statistical analysis; SD, JRM and JHugh provided expert opinions on oncology and pathology; VP and ISG wrote the manuscript and all authors provided input.

REFERENCES

1 Jemal A Bray F Center MM Ferlay J Ward E Forman D Global cancer statistics CA Cancer J Clin 2011 61 69 90 21296855 10.3322/caac.20107

2 Ferlay J Soerjomataram I Dikshit R Eser S Mathers C Rebelo M Parkin DM Forman D Bray F Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012 International journal of cancer 2015 136 E359 386 25220842 10.1002/ijc.29210

3 Ferlay J SI Ervik M Dikshit R Eser S Mathers C Rebelo M Parkin DM Forman D Bray F 2013 Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 11 GLOBOCAN 2012 v10 10.1002/ijc.29210 25220842

4 Henry NL Hayes DF Cancer biomarkers Mol Oncol 2012 6 140 146 22356776 10.1016/j.molonc.2012.01.010 PMC5528374

5 Van Poznak C Somerfield MR Bast RC Cristofanilli M Goetz MP Gonzalez-Angulo AM Hicks DG Hill EG Liu MC Lucas W Mayer IA Mennel RG Symmans WF Hayes DF Harris LN Use of Biomarkers to Guide Decisions on Systemic Therapy for Women With Metastatic Breast Cancer: American Society of Clinical Oncology Clinical Practice Guideline Journal of clinical oncology 2015 33 2695 2704 26195705 10.1200/JCO.2015.61.1459 PMC5478102

6 Puthalakath H Strasser A Keeping killers on a tight leash: transcriptional and post-translational control of the pro-apoptotic activity of BH3-only proteins Cell death and differentiation 2002 9 505 512 11973609 10.1038/sj.cdd.4400998

7 Czabotar PE Lessene G Strasser A Adams JM Control of apoptosis by the BCL-2 protein family: implications for physiology and therapy Nature reviews Molecular cell biology 2014 15 49 63 24355989 10.1038/nrm3722

8 Lomonosova E Chinnadurai G BH3-only proteins in apoptosis and beyond: an overview Oncogene 2008 27 S2 19 10.1038/onc.2009.39 PMC2928556 19641503

9 Maiuri MC Criollo A Tasdemir E Vicencio JM Tajeddine N Hickman JA Geneste O Kroemer G BH3-only proteins and BH3 mimetics induce autophagy by competitively disrupting the interaction between Beclin 1 and Bcl-2/Bcl-X(L) Autophagy 2007 3 374 376 17438366 10.4161/auto.4237

10 Dey P Strom A Gustafsson JA Estrogen receptor beta upregulates FOXO3a and causes induction of apoptosis through PUMA in prostate cancer Oncogene 2014 33 4213 4225 24077289 10.1038/onc.2013.384

11 Sunters A Fernandez de Mattos S Stahl M Brosens JJ Zoumpoulidou G Saunders CA Coffer PJ Medema RH Coombes RC Lam EW FoxO3a transcriptional regulation of Bim controls apoptosis in paclitaxel-treated breast cancer cell lines The Journal of biological chemistry 2003 278 49795 49805 14527951 10.1074/jbc.M309523200

12 Cao Q Xia Y Azadniv M Crispe IN The E2F-1 transcription factor promotes caspase-8 and bid expression, and enhances Fas signaling in T cells Journal of immunology 2004 173 1111 1117 10.4049/jimmunol.173.2.1111 15240700

13 Co NN Tsang WP Tsang TY Yeung CL Yau PL Kong SK Kwok TT AF1q enhancement of gamma irradiation-induced apoptosis by up-regulation of BAD expression via NF-kappaB in human squamous carcinoma A431 cells Oncology reports 2010 24 547 554 20596645

14 Chinnadurai G Vijayalingam S Rashmi R BIK, the founding member of the BH3-only family proteins: mechanisms of cell death and role in cancer and pathogenic processes Oncogene 2008 27 S20 29 19641504 10.1038/onc.2009.40 PMC2928562

15 Bodet L Menoret E Descamps G Pellat-Deceunynck C Bataille R Le Gouill S Moreau P Amiot M Gomez-Bougie P BH3-only protein Bik is involved in both apoptosis induction and sensitivity to oxidative stress in multiple myeloma British journal of cancer 2010 103 1808 1814 21063407 10.1038/sj.bjc.6605981 PMC3008608

16 Chen S Zhou L Zhang Y Leng Y Pei XY Lin H Jones R Orlowski RZ Dai Y Grant S Targeting SQSTM1/p62 Induces Cargo-Loading Failure and Converts Autophagy to Apoptosis via NBK/Bik Molecular and cellular biology 2014 34 3435 49 10.1128/MCB.01383-13 25002530 PMC4135623

17 Lopez J Hesling C Prudent J Popgeorgiev N Gadet R Mikaelian I Rimokh R Gillet G Gonzalo P Src tyrosine kinase inhibits apoptosis through the Erk1/2- dependent degradation of the death accelerator Bik Cell death and differentiation 2012 19 1459 1469 22388352 10.1038/cdd.2012.21 PMC3422470

18 Germain DR Graham K Glubrecht DD Hugh JC Mackey JR Godbout R DEAD box 1: a novel and independent prognostic marker for early recurrence in breast cancer Breast cancer research and treatment 2011 127 53 63 20499159 10.1007/s10549-010-0943-7

19 Liu RZ Graham K Glubrecht DD Germain DR Mackey JR Godbout R Association of FABP5 expression with poor survival in triple-negative breast cancer: implication for retinoic acid therapy The American journal of pathology 2011 178 997 1008 21356353 10.1016/j.ajpath.2010.11.075 PMC3070589

20 Rashmi R Pillai SG Vijayalingam S Ryerse J Chinnadurai G BH3-only protein BIK induces caspase-independent cell death with autophagic features in Bcl-2 null cells Oncogene 2008 27 1366 1375 17873911 10.1038/sj.onc.1210783 PMC2928058

21 Chang NC Nguyen M Germain M Shore GC Antagonism of Beclin 1-dependent autophagy by BCL-2 at the endoplasmic reticulum requires NAF-1 The EMBO journal 2010 29 606 618 20010695 10.1038/emboj.2009.369 PMC2830692

22 Ruiz Esparza-Garrido R Torres-Marquez ME Viedma-Rodriguez R Velazquez-Wong AC Salamanca-Gomez F Rosas-Vargas H Velazquez-Flores MA Breast cancer cell line MDA-MB-231 miRNA profile expression after BIK interference: BIK involvement in autophagy Tumour Biol 2015 10.1007/s13277-015-4494-8 26662110

23 Guo JY Xia B White E Autophagy-mediated tumor promotion Cell 2013 155 1216 1219 24315093 10.1016/j.cell.2013.11.019 PMC3987898

24 Yang S Wang X Contino G Liesa M Sahin E Ying H Bause A Li Y Stommel JM Dell'antonio G Mautner J Tonon G Haigis M Shirihai OS Doglioni C Bardeesy N Pancreatic cancers require autophagy for tumor growth Genes & development 2011 25 717 729 21406549 10.1101/gad.2016111 PMC3070934

25 Wei H Wei S Gan B Peng X Zou W Guan JL Suppression of autophagy by FIP200 deletion inhibits mammary tumorigenesis Genes & development 2011 25 1510 1527 21764854 10.1101/gad.2051011 PMC3143941

26 Tang JY Fang YY Hsi E Huang YC Hsu NC Yang WC Chang HW Chai CY Chu PY Immunopositivity of Beclin-1 and ATG5 as indicators of survival and disease recurrence in oral squamous cell carcinoma Anticancer research 2013 33 5611 5616 24324106

27 Kim MS Song SY Lee JY Yoo NJ Lee SH Expressional and mutational analyses of ATG5 gene in prostate cancers APMIS 2011 119 802 807 21995634 10.1111/j.1600-0463.2011.02812.x

28 Ge J Chen Z Huang J Chen J Yuan W Deng Z Chen Z Upregulation of autophagy-related gene-5 (ATG-5) is associated with chemoresistance in human gastric cancer PloS one 2014 9 e110293 25329677 10.1371/journal.pone.0110293 PMC4201506

29 Cho DH Jo YK Kim SC Park IJ Kim JC Down-regulated expression of ATG5 in colorectal cancer Anticancer research 2012 32 4091 4096 22993366

30 Arena V Martini M Luongo M Capelli A Larocca LM Mutations of the BIK gene in human peripheral B-cell lymphomas Genes, chromosomes & cancer 2003 38 91 96 12874789 10.1002/gcc.10245

31 Bredel M Bredel C Juric D Harsh GR Vogel H Recht LD Sikic BI High-resolution genome-wide mapping of genetic alterations in human glial brain tumors Cancer research 2005 65 4088 4096 15899798 10.1158/0008-5472.CAN-04-4229

32 Castells A Ino Y Louis DN Ramesh V Gusella JF Rustgi AK Mapping of a target region of allelic loss to a 0. 5-cM interval on chromosome 22q13 in human colorectal cancer Gastroenterology 1999 117 831 837 10500065 10.1016/s0016-5085(99)70341-0

33 Sturm I Stephan C Gillissen B Siebert R Janz M Radetzki S Jung K Loening S Dorken B Daniel PT Loss of the tissue-specific proapoptotic BH3-only protein Nbk/Bik is a unifying feature of renal cell carcinoma Cell death and differentiation 2006 13 619 627 16322756 10.1038/sj.cdd.4401782

34 Li C Li R Grandis JR Johnson DE Bortezomib induces apoptosis via Bim and Bik up-regulation and synergizes with cisplatin in the killing of head and neck squamous cell carcinoma cells Molecular cancer therapeutics 2008 7 1647 1655 18566236 10.1158/1535-7163.MCT-07-2444 PMC2534142

35 Day CP Rau KM Qiu L Liu CW Kuo HP Xie X Lopez-Berestein G Hortobagyi GN Hung MC Mutant Bik expression mediated by the enhanced minimal topoisomerase IIalpha promoter selectively suppressed breast tumors in an animal model Cancer gene therapy 2006 13 706 719 16514421 10.1038/sj.cgt.7700945

36 Jiao S Wu M Ye F Tang H Xie X Xie X BikDDA, a mutant of Bik with longer half-life expression protein, can be a novel therapeutic gene for triple-negative breast cancer PloS one 2014 9 e92172 24637719 10.1371/journal.pone.0092172 PMC3956915

37 Lan KL Yen SH Liu RS Shih HL Tseng FW Lan KH Mutant Bik gene transferred by cationic liposome inhibits peritoneal disseminated murine colon cancer Clinical & experimental metastasis 2007 24 461 470 17636408 10.1007/s10585-007-9083-9

38 Garcia N Salamanca F Astudillo-de la Vega H Curiel-Quesada E Alvarado I Penaloza R Arenas D A molecular analysis by gene expression profiling reveals Bik/NBK overexpression in sporadic breast tumor samples of Mexican females BMC cancer 2005 5 93 16060964 10.1186/1471-2407-5-93 PMC1198222

39 Lopez-Munoz E Hernandez-Zarco A Garcia-Hernandez N Alvarado-Cabrero I Zarco-Espinosa G Salamanca-Gomez F Arenas-Aranda D BIK/NBK gene as potential marker of prognostic and therapeutic target in breast cancer patients Clinical & translational oncology 2012 14 586 591 22855140 10.1007/s12094-012-0845-8

40 Maxfield KE Taus PJ Corcoran K Wooten J Macion J Zhou Y Borromeo M Kollipara RK Yan J Xie Y Xie XJ Whitehurst AW Comprehensive functional characterization of cancer-testis antigens defines obligate participation in multiple hallmarks of cancer Nat Commun 2015 6 8840 26567849 10.1038/ncomms9840 PMC4660212

41 Zhan F Huang Y Colla S Stewart JP Hanamura I Gupta S Epstein J Yaccoby S Sawyer J Burington B Anaissie E Hollmig K Pineda-Roman M Tricot G van Rhee F Walker R The molecular classification of multiple myeloma Blood 2006 108 2020 2028 16728703 10.1182/blood-2005-11-013458 PMC1895543

42 Lu Y Lemon W Liu PY Yi Y Morrison C Yang P Sun Z Szoke J Gerald WL Watson M Govindan R You M A gene expression signature predicts survival of patients with stage I non-small cell lung cancer PLoS medicine 2006 3 e467 17194181 10.1371/journal.pmed.0030467 PMC1716187

43 Ichim G Lopez J Ahmed SU Muthalagu N Giampazolias E Delgado ME Haller M Riley JS Mason SM Athineos D Parsons MJ van de Kooij B Bouchier-Hayes L Chalmers AJ Rooswinkel RW Oberst A Limited mitochondrial permeabilization causes DNA damage and genomic instability in the absence of cell death Molecular cell 2015 57 860 872 25702873 10.1016/j.molcel.2015.01.018 PMC4352766

44 Abdel-Fatah TM Perry C Dickinson P Ball G Moseley P Madhusudan S Ellis IO Chan SY Bcl2 is an independent prognostic marker of triple negative breast cancer (TNBC) and predicts response to anthracycline combination (ATC) chemotherapy (CT) in adjuvant and neoadjuvant settings Annals of oncology 2013 24 2801 2807 23908177 10.1093/annonc/mdt277

45 Anagnostou VK Lowery FJ Zolota V Tzelepi V Gopinath A Liceaga C Panagopoulos N Frangia K Tanoue L Boffa D Gettinger S Detterbeck F Homer RJ Dougenis D Rimm DL Syrigos KN High expression of BCL-2 predicts favorable outcome in non-small cell lung cancer patients with non squamous histology BMC cancer 2010 10 186 20459695 10.1186/1471-2407-10-186 PMC2875218

46 Karakas T Maurer U Weidmann E Miething CC Hoelzer D Bergmann L High expression of bcl-2 mRNA as a determinant of poor prognosis in acute myeloid leukemia Annals of oncology 1998 9 159 165 9553660 10.1023/a:1008255511404

47 Sup SJ Alemany CA Pohlman B Elson P Malhi S Thakkar S Steinle R Hsi ED Expression of bcl-2 in classical Hodgkin's lymphoma: an independent predictor of poor outcome Journal of clinical oncology 2005 23 3773 3779 15809450 10.1200/JCO.2005.04.358

48 Lipponen P Pietilainen T Kosma VM Aaltomaa S Eskelinen M Syrjanen K Apoptosis suppressing protein bcl-2 is expressed in well-differentiated breast carcinomas with favourable prognosis The Journal of pathology 1995 177 49 55 7472779 10.1002/path.1711770109

49 Gasparini G Barbareschi M Doglioni C Palma PD Mauri FA Boracchi P Bevilacqua P Caffo O Morelli L Verderio P Expression of bcl-2 protein predicts efficacy of adjuvant treatments in operable node-positive breast cancer Clinical cancer research 1995 1 189 198 9815973

50 Pattingre S Tassa A Qu X Garuti R Liang XH Mizushima N Packer M Schneider MD Levine B Bcl-2 antiapoptotic proteins inhibit Beclin 1-dependent autophagy Cell 2005 122 927 939 16179260 10.1016/j.cell.2005.07.002

51 Wei Y Pattingre S Sinha S Bassik M Levine B JNK1-mediated phosphorylation of Bcl-2 regulates starvation-induced autophagy Molecular cell 2008 30 678 688 18570871 10.1016/j.molcel.2008.06.001 PMC2478643

52 Mazel S Burtrum D Petrie HT Regulation of cell division cycle progression by bcl-2 expression: a potential mechanism for inhibition of programmed cell death The Journal of experimental medicine 1996 183 2219 2226 8642331 10.1084/jem.183.5.2219 PMC2192544

53 Vairo G Innes KM Adams JM Bcl-2 has a cell cycle inhibitory function separable from its enhancement of cell survival Oncogene 1996 13 1511 1519 8875989

54 An CH Kim MS Yoo NJ Park SW Lee SH Mutational and expressional analyses of ATG5, an autophagy-related gene, in gastrointestinal cancers Pathology, research and practice 2011 207 433 437 10.1016/j.prp.2011.05.002 21664058

55 Craik AC Veldhoen RA Czernick M Buckland TW Kyselytzia K Ghosh S Lai R Damaraju S Underhill DA Mackey JR Goping IS The BH3-only protein Bad confers breast cancer taxane sensitivity through a nonapoptotic mechanism Oncogene 2010 29 5381 5391 20603619 10.1038/onc.2010.272