Fax: (813) 632-1708
Down-regulation of Bax-interacting factor-1 in colorectal adenocarcinoma
Article first published online: 2 OCT 2008
Copyright © 2008 American Cancer Society
Volume 113, Issue 10, pages 2665–2670, 15 November 2008
How to Cite
Coppola, D., Khalil, F., Eschrich, S. A., Boulware, D., Yeatman, T. and Wang, H.-G. (2008), Down-regulation of Bax-interacting factor-1 in colorectal adenocarcinoma. Cancer, 113: 2665–2670. doi: 10.1002/cncr.23892
- Issue published online: 3 NOV 2008
- Article first published online: 2 OCT 2008
- Manuscript Accepted: 25 JUN 2008
- Manuscript Revised: 11 JUN 2008
- Manuscript Received: 20 MAR 2008
- American Cancer Society. Grant Number: RSG-05-244-01-CCG
- National Institutes of Health. Grant Number: CA82197
- colon adenocarcinoma;
Bax-interacting factor-1 (Bif-1) protein is a member of the endophilin B family that plays a critical role in apoptosis, autophagy, and mitochondrial morphology. Loss of Bif-1 suppresses programmed cell death and promotes tumorigenesis. The connection of Bif-1 to colorectal cancer remains to be evaluated.
To determine Bif-1 expression in human colorectal adenocarcinoma (CRC), the authors performed immunohistochemistry using stage-oriented cancer tissue microarrays containing 102 CRC samples of different stages and 38 samples of normal colorectal mucosa (NR). Formalin-fixed, paraffin-embedded core sections on the tissue array were immunostained using the avidin-biotin-peroxidase method and the anti-Bif-1 murine monoclonal antibody. Bif-1 staining was scored by 2 independent observers. To examine Bif-1 mRNA levels, the authors performed DNA microarray analysis of 205 CRC and 10 NR samples.
Bif-1 expression was negative in 22.5% (23 of 102) of CRCs. Moderate to strong Bif-1 staining was identified in 36.3% (37 of 102) of the tumors, and weak staining was noted in 41.2% (42 of 102). Twenty-six of 38 (68.4%) NR samples exhibited moderate to strong Bif-1 immunoreactivity, and none of them was negative. In 12 (31.6%) cases NR demonstrated weak Bif-1 stain. The mean (median) scores for CRCs and NR differed significantly: 3.2 (3.0) and 5.2 (6.0), respectively (P = .0003). The percentage of cases with negative expression also differed significantly between NR and CRC (P = .002). Decreased Bif-1 expression in CRCs was confirmed at the mRNA level by microarray analysis.
The authors report the down-regulation of Bif-1 during the transition from NR to CRC, a novel finding in agreement with the tumor suppressor function of Bif-1. Cancer 2008. © 2008 American Cancer Society.
Colorectal adenocarcinoma (CRC) is 1 of the most common malignancies, accounting for approximately 15% of all cancer-related deaths in the US.1 The prevalence of CRC increases with age, the largest number of tumors occurring during the sixth decade of life. The expected annual incidence of this tumor has risen over the last decade, and 153,760 new cases are estimated in 2007.1, 2 If not diagnosed and treated early, this tumor spreads through the entire bowel wall, extends to adjacent organs, and eventually metastasizes to regional lymph nodes and distant sites. The majority of deaths from CRC occur in patients with late-stage tumors, which are usually incurable.3
It has been shown that inhibition of apoptosis is critical to colorectal tumorigenesis.4 For example, it has been proposed that overexpression of Bcl-XL in cancer may suppress the activity of the proapoptotic molecules Bax and Bak, contributing to cancer progression.5, 6 It appears that, also in CRC, the dissociation of Bax and Bcl-XL promotes Bax multimerization and mitochondrial translocation, triggering apoptosis.7 Similarly, dysregulation of autophagy has also been proposed to play a role in the pathogenesis of cancer. As an example, the autophagy activator Beclin 1 is found to be monoallelically deleted in a high percentage of ovarian, breast, and prostate cancers, and overexpression of Beclin 1 in MCF7 cells promotes autophagy and inhibits tumor formation in nude mice.8, 9 Moreover, the Beclin 1 binding protein UVRAG has been shown to promote autophagy and suppress the tumorigenesis of colon cancer cells in nude mice.10
Bax-interacting factor-1 (Bif-1) has been shown to interact with Bax and induce its conformational change in mammalian cells during apoptosis.11 Knockout of Bif-1 suppresses Bax/Bak conformational change, cytochrome c release, caspase activation, and cell death.12 Interestingly, we have recently discovered that Bif-1 also regulates autophagy by forming a multiprotein complex with PI3KC3-Beclin1 through UVRAG, and loss of Bif-1 suppresses autophagic cell death and promotes tumorigenesis.13 Along these lines, a recent study has reported the decrease of Bif-1 expression in malignant gastric epithelial cells as compared with the normal gastric mucosal cells.14 To our knowledge, to date, the expression of Bif-1 in CRC has not been reported.
In this study, we focused on the evaluation of Bif-1 expression and significance in CRC. Bif-1 expression levels in CRC were determined using semiquantitative immunohistochemistry and microarray analysis of archival specimens. The results of this study may help in allowing the evolution of improved therapies for CRC, based on the better understanding of the underlying biology of this disease process.
MATERIALS AND METHODS
Selection of Human Tissues
By using stage-oriented human colorectal cancer tissue microarrays (prepared in the Histology laboratory of the Moffitt Cancer Center Tissue Core Facility), 140 tissue samples (102 CRC and 38 samples of normal colonic mucosa [NR]) were analyzed for Bif-1 expression by immunohistochemistry. All the tumors used for the tissue array construction were CRC identified from the Moffitt Cancer Center Anatomic Pathology Division's database, CoPath, and representing surgical resection specimens obtained between 1990 and 2002. All the specimens were preserved in 10% buffered formalin before embedding in paraffin. The patients had a median age of 65 years (range, 24 years-92 years), 61 were men and 41 were women. The tumors ranged in size between 1.4 cm and 14.5 cm. The tumors were staged according to the TNM system, following the recommendations of the American Joint Committee on Cancer (1988). The stage of the invasive tumors was as follows: 10 patients had stage I (Dukes stage A), 33 had stage II (Dukes stage B), 38 had stage III (Dukes stage C), and 21 had stage IV (Dukes stage D). All tumors occurred in the absence of genetic cancer syndromes such as human nonpolyposis colon cancer syndrome, familial adenomatous polyposis syndrome, etc; in addition, cancers arising in the background of ulcerative colitis or Crohn disease were excluded from the study. The NR samples were taken near the resected colorectal margin, away from the tumor site, from CRC colon resection specimens included in this study.
The tissues were stained for Bif-1 using a mouse monoclonal antibody (Imgenex, San Diego, Calif). The slides were dewaxed by heating at 55°C for 30 minutes and by 3 washes of 5 minutes each with xylene. Tissues were rehydrated by a series of 5-minute washes in 100%, 95%, and 80% ethanol, and distilled water. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 20 minutes. After blocking with universal blocking serum (Ventana Medical Systems, Inc., Tucson, Ariz) for 30 minutes, the samples were incubated with anti-Bif-1 mouse monoclonal antibody (Imgenex, dilution 1:2500) at 4°C overnight. The samples were then incubated with biotin-labeled secondary antibody and streptavidin-horseradish peroxidase for 30 minutes each (Ventana Medical Systems). The slides were developed with 3,3′-diaminobenzidine tetrahydrochloride substrate (Ventana Medical Systems) and counterstained with hematoxylin (Ventana Medical Systems). The tissue samples were dehydrated and coverslipped. Standard cell conditioning (following the Ventana proprietary recommendations) was used for antigen retrieval. The specificity of the anti-Bif-1 monoclonal antibody was confirmed by immunostains of Bif-1 overexpression and knockout cell lines.12, 13 Negative control was included by using nonimmune mouse sera and omitting the monoclonal Bif-1 antibody during the primary antibody incubation step.
Immunohistochemical Data Analysis
The Bif-1-stained tissue cores were examined by 2 independent observers (F.K. and D.C.) simultaneously and a consensus score was reached for each specimen. The positive reaction of Bif-1 was scored into 4 grades, according to the intensity of the staining: 0, 1+, 2+, and 3+. The percentages of Bif-1-positive cells were also scored into 4 categories: 0 (0%), 1 (1%-33%), 2 (34%-66%), and 3 (67%-100%). The product of the intensity by percentage scores was used as the final score. The final scores were classified as: 0, negative; 1 to 3, weak; 4 to 6, moderate; and 7 to 9, strong. The specimens were also classified by the types of tissue staining positive: NR and CRC.
Descriptive statistics for the scores were generated and reported for each tissue group. The initial method used to compare Bif-1 expression in CRC and NR was the Wilcoxon rank sum test. In addition, the Fisher exact test was used to compare Bif-1 negativity between CRC and NR. For CRC, the exact Cochran-Armitage trend test was used to compare Bif-1 negativity across stages. The Holm step down method was used to adjust for multiple testing. Age and sex differences between cohorts were examined using the Wilcoxon rank sum test and the chi-square test, respectively. Spearman correlation was used to examine the correlation between age and Bif-1 expression, and the Wilcoxon rank sum test was used to compare Bif-1 expression differences between genders.
mRNA Microarray Analysis
To evaluate whether the variation in Bif-1 protein expression between NR and CRC reflected a corresponding modulation of Bif-1 mRNA, we resorted to the Moffitt Cancer Center gene profiling database. Two hundred and five CRC specimens and 10 NR samples, from patients treated at the Moffitt Cancer Center under a protocol approved by the institutional review board of University of South Florida were arrayed on Affymetrix (Santa Clara, Calif) HG-U133+ GeneChip microarrays. The tumors used for the mRNA microarray analysis included mirror image samples from all of the CRCs used to construct the colon tissue microarray used in this study.
The data were processed using MAS5.0 and scaled to a mean intensity of 500. Three probe sets were identified by Affymetrix NetAffx as detecting Bif-1: 209090_s_at, 209091_s_at, and 210101_x_at. The R statistical software was used for expression analysis (freely available open source statistical package available at: www.r-project.org, accessed on January 14, 2008). The Anderson-Darling test for normality was used to verify the distribution of gene expression for each probe set across samples, and a t test was used to compare differences between groups (normal vs tumor and normal vs each stage). Expression is graphed using the mean and standard error for each probe set across the different groups.
Clinical Pathologic Findings
The patients had a median age of 65 years (range, 24 years-92 years). Sixty-one were men, and 41 were women. The tumors ranged in size between 1.4 cm and 14.5 cm, mostly polypoid and ulcerated. Twenty tumors involved the cecum, 26 involved the ascending colon, 4 involved the transverse colon, 9 involved the descending colon, 24 involved the sigmoid, 10 involved the rectosigmoid junction, and 9 involved the rectum. Fourteen tumors were well differentiated, 74 were moderately differentiated, and 14 were poorly differentiated. Ten tumors were Dukes stage A, 33 were Dukes stage B, 38 were Dukes stage C, and 21 were Dukes stage D. Only 2 patients, both with rectal cancer, received preoperative radiation to decrease the size of their tumors.
All of the positively stained cases had cytoplasmic staining, which was diffusely granular with variation in intensity observed within the same lesion of some cases. Cases with variable staining were graded based on the predominant staining intensity, and the percentage of positive stain was determined based on the amount of the lesion demonstrating the predominant intensity. In CRC specimens, approximately 41.2% (42 of 102) had weak Bif-1 staining (Fig. 1A), 36.3% (37 of 102) exhibited moderate to strong Bif-1 staining (Figs. 1B and 1C), and 22.5% (23 of 102) were Bif-1 negative (Fig. 1D). In contrast, in the NR samples, 68.4% (26 of 38) exhibited moderate to strong Bif-1 immunoreactivity (Fig. 1E and 1F), 31.6% (12 of 38) demonstrated weak Bif-1 staining, and none was Bif-1 negative. The specificity of the anti-Bif-1 antibody was confirmed by immunostaining of Bif-1 overexpressing cells (Fig. 1G) as compared with the Bif-1 knockout cell line (Fig. 1H).
There was a statistically significant difference found in Bif-1 staining score between NR and CRC, using the Wilcoxon rank sum test (P = .0003), and there was also a significant difference in Bif-1 negativity (P = .002) when comparing the percentage negative (as dichotomous groups) between the 2 tissue types. By using the exact Cochran-Armitage trend test, there was no statistically significant increasing trend noted between Bif-1 negativity score and tumor stage (P = .29). When considering the expression of Bif-1 in the NR versus the CRC, it became evident that although 22.5% of CRC were negative, none of the NR was negative. This difference was statistically significant (P = .002). There were no statistically significant differences found in age or sex between the NR and CRC cohorts (P = .39 and P = .63, respectively) (Table 1), and age and sex were not found to be significantly correlated with Bif-1 staining score (P = .68 and P = .26, respectively).
To determine whether Bif-1 changes in protein level during human colorectal cancer development reflect changes in gene expression, we compared the Bif-1 mRNA levels by DNA microarray in 10 normal human colon tissues and 205 colorectal tumors grouped by Dukes staging system. The samples included 32 stage A, 66 stage B, 65 stage C, and 42 stage D CRC samples. The expression of Bif-1 mRNA decreased significantly between NRs and stage A CRCs and remained at approximately the same levels during tumor progression (Fig. 2), suggesting that loss of Bif-1 expression may play a role at an early stage of colorectal tumorigenesis. The mRNA levels of Bif-1 in the 10 NRs were approximately 3-fold higher than the Bif-1 mRNA levels in the tumors.
Programmed cell death (PCD) is defined as a physiologic process that plays a critical role in normal development, cellular differentiation, and tissue homeostasis of multicellular organisms.15–17 Dysregulation of this physiologic cell death process contributes to the pathogenesis of human diseases, including cancer.18 In addition to apoptosis (type I cell death), which has long been used as a synonym for PCD, accumulating evidence suggests that autophagy (type II cell death) also belongs to PCD.19 Autophagy is a highly orchestrated self-digestion process that involves multiple steps from the formation of autophagic vesicles to lysosomal degradation of the vesicles and their contents.19–21 As with apoptosis, autophagy also contributes to proper morphogenesis during development and tissue homeostasis in mature organisms.21
Bif-1, also known as endophilin B1 and SH3GLB1 (SH3 domain GRB2-like endophilin B1), was originally identified as a Bax-binding protein by yeast 2-hybrid screens using Bax as the bait.11, 22 The human Bif-1 gene encodes a 365 amino-acid polypeptide that contains an N-terminal BAR (Bin/Amphiphysin/Rvs) domain, a central coiled-coil domain, and a C-terminal SH3 domain. The N-terminal part (1-27 amino acids) of Bif-1 is required for its binding to Bax.11, 22 Moreover, the interaction between Bif-1 and Bax is enhanced in mammalian cells during apoptosis, which is accompanied by a conformational change in the Bax protein.11, 12, 23
Overexpression of Bif-1 promotes Bax activation and apoptosis,11 whereas inhibition of Bif-1 expression suppresses Bax/Bak conformational activation, cytochrome c release, caspase activation, and cell death in response to intrinsic apoptosis signals.12 It has also been demonstrated that Bif-1 regulates apoptosis by mediating the mitochondrial fission process.24 This suggests that Bif-1 may represent a new type of Bax activator controlling the mitochondrial pathway of apoptosis.
In addition to its role as a Bax activator, Bif-1 has recently been found to be involved in autophagosome formation and autophagic cell death.13 Bif-1 interacts with Beclin 1 through UVRAG to positively regulate the class III PI3-kinase (PI3KC3) lipid kinase during autophagy. Although the C-terminal SH3 domain of Bif-1 is sufficient for binding to UVRAG, the N-terminal BAR domain of Bif-1 is also required for Bif-1 to activate PI3KC3 lipid kinase and induce autophagosome formation. Suppression of Bif-1 expression inhibits autophagy and prolongs cell survival under nutrient starvation. Moreover, Bif-1 ablation promotes the development of spontaneous tumors in mice, consistent with the notion that both apoptosis and autophagy play crucial roles in tumor suppression.19, 25–27
It has been shown that the Bif-1 mRNA levels are down-regulated in lung carcinomas,28 and that approximately 60% of gastric carcinomas express undetectable levels of Bif-1 protein.14 In addition, loss of heterozygosity on 1p22, in which the bif-1 gene is localized, is frequently observed in many types of tumor.29–37 In CRC, 1p22 deletions were identified in >70% of advanced stage and metastatic tumors.36 These results are in agreement with other studies showing that 1p deletion was significantly more common in metastatic as compared with primary CRC.37 Others have described a 38% incidence of 1p deletions in 34 sporadic colorectal adenomas, using a centromeric probe for chromosome 1 and a simultaneous telomeric probe mapping to 1p36. These authors concluded that 1p deletion is an early event in colorectal tumorigenesis. These data and the observation that inhibition of Bif-1 expression promotes tumor development in mice12, 13 suggest Bif-1 as a candidate tumor suppressor gene.
In this study, we found that the expression of Bif-1 was absent in 22.5% of CRC, but all of the NR samples were Bif-1 positive. This difference was statistically significant (P = .002). This finding is in agreement with the tumor suppressor function of Bif-1. Remarkably, this trend of Bif-1 protein expression down-regulation in CRC was mirrored by significant decrease in the mRNA levels of Bif-1 at an early stage of colorectal cancer development. Because loss of Bif-1 not only suppresses Bax/Bak activation and apoptosis12 but also inhibits PI3KC3 activation and autophagy,13 it remains to be determined whether the tumor suppressor activity of Bif-1 is because of its proapoptotic activity, proautophagic activity, or both.
A previous study has reported allelic loss in the 1p36 and 1p32 regions of chromosome 1 as an independent predictor of poor prognosis in patients with CRC.37 In the current study, complete follow-up information was available only for a subset of the patients studied. The correlation of Bif-1 protein expression and patient survival in CRC warrants further evaluation.
We thank the Histology Section of the Tissue Core at the Moffitt Cancer Center and Research Institute for the support in performing the immunohistochemical stains. The DNA microarray analysis was performed in the Moffitt Microarray Core facility.