Loss of BNIP3 expression is a late event in pancreatic cancer contributing to chemoresistance and worsened prognosis


Altered expression of apoptosis-regulating genes plays an important role in the aggressive growth behavior and chemoresistance of pancreatic ductal adenocarcinoma. In the present study, the hypoxia-inducible proapoptotic gene, BNIP3, was analysed in terms of expression, effect on patient survival and chemo-responsiveness in pancreatic cancer cell lines. cDNA microarray, real-time light cycler® quantitative polymerase chain reaction, laser-capture microdissection and immunohistochemistry analyses were used to evaluate BNIP3 expression in normal and diseased pancreatic specimens. Modulation of BNIP3 expression was achieved using specific siRNA molecules. The effect of chemotherapeutic agents on pancreatic cancer cells was assessed utilizing 3-(4,5-methylthiazol-2-yl)-2,5-diphenyl-tetrazolium-bromide assays. BNIP3 mRNA levels were 3.0- and 6.3-fold lower in chronic pancreatitis and pancreatic cancer compared to the normal pancreas, respectively. Microdissection analysis confirmed the reduction of BNIP3 expression in pancreatic cancer cells compared to normal duct cells. By immunohistochemistry, BNIP3 was predominantly expressed in the acinar cells of the normal and diseased pancreas. Interestingly, while BNIP3 was undetectable in the cancer cells of 59% of the cases, 75–100% of PanIN2/3 lesions displayed BNIP3 immunoreactivity. Loss of BNIP3 expression correlated with poorer survival of patients (8 vs 14 months for BNIP3 negative vs positive tumors). Hypoxia induced BNIP3 expression in four out of eight pancreatic cancer cell lines, while it was absent under normoxic and hypoxic conditions in the remaining four. Downregulation of BNIP3 resulted in increased resistance to 5-fluoro-uracil and gemcitabine. In conclusion, loss of BNIP3 expression occurs late in pancreatic cancer, contributes to resistance to chemotherapy and correlates with a worsened prognosis.


Pancreatic ductal adenocarcinoma (PDAC) has a dismal prognosis, with an incidence rate that almost equals its mortality rate. Most of the cases are unresectable at the time of diagnosis and the high incidence of local recurrence and metastasis following surgical resection requires effective new palliative and adjuvant therapies. 5-Fluoro-uracil (5-FU) and gemcitabine, currently the most commonly used agents, both induce endogenous apoptotic mechanisms to exert their cytotoxic effects (Korsmeyer, 1992; Kerr et al., 1994; White, 1996; Bold et al., 1999). However, intrinsic or acquired resistance to apoptosis is a major cause of treatment failure in PDAC with either drug (Bold et al., 1999; Shi et al., 2002). In order to improve the effectiveness of chemotherapy, it is necessary to understand the mechanisms exploited by PDAC to evade apoptosis.

Deregulated expression of apoptosis-regulating molecules – including members of the Bcl-2 family, such as Bcl-2, Bcl-XL, Bax and Bak – is a common feature in PDAC (Friess et al., 1998a, 1998b; Graber et al., 1999). Pancreatic cancer may gain resistance to apoptosis either by upregulating the expression of antiapoptotic genes or by downregulating the expression of proapoptotic genes. We have previously shown that acquired resistance of pancreatic cancer cell lines to 5-FU and gemcitabine is associated with an alteration in the Bax/Bcl-XL ratio in favor of the activation of antiapoptotic genes (Shi et al., 2002). On the other hand, expression of proapoptotic Bax gene in tumor samples from patients with PDAC is a strong indicator of a longer survival in a multivariate analysis (Friess et al., 1998a, 1998b). This observation was further consolidated by showing that overexpression of Bax sensitizes pancreatic cancer cells to either drug (Xu et al., 2002).

BNIP3 is a novel hypoxia-inducible proapoptotic mitochondrial member of the Bcl-2 family which has recently been shown by gene profiling to be differentially expressed in several malignancies (Bruick, 2000; Kennedy et al., 2000; Sowter et al., 2001, 2003; de Angelis et al., 2004; Okami et al., 2004; Tenedini et al., 2004; Tracey et al., 2004). Several studies have indicated that BNIP3 plays an important role in the regulation of apoptosis in normal and malignant cells (Kothari et al., 2003; Bacon and Harris, 2004; Bani et al., 2004; de Angelis et al., 2004; Yook et al., 2004). Under physiologic conditions, BNIP3 is involved in the necrotic-apoptotic death of cardiac myocytes and brain cells following hypoxic episodes (Graham et al., 2004; Schmidt-Kastner et al., 2004). In hematologic (cutaneous T-cell lymphoma, essential thrombocytemia) (Tenedini et al., 2004; Tracey et al., 2004), neural (malignant glioma) (Daido et al., 2004) and gastrointestinal (colorectal, pancreatic) malignancies (Kennedy et al., 2000; Okami et al., 2004) BNIP3 expression is downregulated. This downregulation was assumed to render malignant cells more resistant to therapeutic agents (IFN-α, 5-FU) (Kennedy et al., 2000; Tracey et al., 2004). Paradoxically, higher BNIP3 expression in tumor samples of non-small-cell lung carcinoma and ductal carcinoma in situ of the breast correlates with a more aggressive tumor phenotype (Sowter et al., 2003; Giatromanolaki et al., 2004). Still, it has been proposed that hypoxic areas in solid tumors may select more aggressive tumor clones which can survive under stress conditions (Sowter et al., 2003; Giatromanolaki et al., 2004). The rationale behind this hypothesis rests on the facts that: (a) hypoxia increases the mutational rate in the cells and (b) cells which gained a survival advantage by resisting apoptosis under hypoxic conditions also became more resistant to cancer therapy (Graeber et al., 1996; Reynolds et al., 1996; Schmaltz et al., 1998).

The link between hypoxic death of normal cells and the selection of BNIP3-negative cancer clones is the induction of BNIP3 during hypoxia via hypoxia-inducible factor 1-α (HIF1-α). The BNIP3 promoter contains a functional HIF1-α responsive element and is strongly induced upon exposure to hypoxia (Bruick, 2000; Sowter et al., 2001; Okami et al., 2004). Once induced, BNIP3 both antagonizes the antiapoptotic proteins (such as Bcl-2 and Bcl-XL) by heterodimerization and associates with the outer mitochondrial membrane, where it opens mitochondrial permeability transition pores (Adams and Cory, 1998; Chen et al., 1997, 1999; Yasuda et al., 1998; Ray et al., 2000). BNIP3-induced cell death proceeds via the mitochondrial pathway, independent of cytochrome c release and caspase activation, as shown in several normal and cancer cell lines (Chen et al., 1997, 1999; Bruick, 2000; Ray et al., 2000; Vande Velde et al., 2000; Lamy et al., 2003).

Hypoxia is a common phenomenon in solid tumors and has been proven to occur in PDAC, with tissue oxygen tension being significantly lower in the cancer area than the surrounding normal pancreas (Koong et al., 2000; Harris, 2002). Recently, hypoxia-dependent expression of BNIP3 was detected in pancreatic cancer; however, this gene was silenced in the majority of tested cancer cell lines due to the hypermethylation of its promoter (Okami et al., 2004). Moreover, hypoxia treatment is unable to induce BNIP3 expression in BNIP3-negative cells. These results indicate that downregulation of proapoptotic BNIP3 may contribute to the survival and progression of pancreatic cancer in a hypoxic environment. In the present study, we extensively analysed: (a) the expression of BNIP3 in pancreatic acinar, islet and ductal cells throughout their transformation from normal through chronic pancreatitis (CP) and PanIN lesions to pancreatic cancer; (b) pathways involved in the downregulation of BNIP3 in pancreatic cancer cell lines other than promoter hypermethylation and (c) the clinico-pathological consequences of BNIP3 downregulation in terms of patient survival and chemoresistance in pancreatic cancer cell lines.


Expression analysis of BNIP3 mRNA in pancreatic tissues

To quantify BNIP3 mRNA levels, real-time light cycler® quantitative polymerase chain reaction (QRT–PCR) was utilized in normal (n=26), CP (n=20) and PDAC (n=23) samples. This analysis revealed that BNIP3 expression was threefold reduced in CP (P<0.0001) and 6.3-fold in PDAC (P<0.0001) as compared to normal tissues (Figure 1a). Furthermore, when micro-dissected ductal epithelium of CP (n=5) and PDAC (n=6) patients was compared by microarray analysis to normal pancreatic tissues (n=3) and micro-dissected normal ducts (n=3), there were significantly lower BNIP3 mRNA expression levels in the cancer cells compared to normal ducts (P=0.047) and total normal pancreatic tissues (P=0.024), as well as in CP ducts compared to total pancreatic extracts (P=0.036) (Figure 1b). Similar results were obtained using a different probe targeting another region of the BNIP3 sequence (data not shown).

Figure 1

Expression analysis of BNIP3 mRNA in pancreatic tissues. (a) QRT–PCR was carried out as described in the Materials and methods section. Individual samples are expressed on a logarithmic scale as copies/μl cDNA normalized to housekeeping genes. Horizontal bars show the median values. (b) Microarray analysis of BNIP3 expression in LCM ductal epithelium and PDAC cells. Total normal pancreatic tissue specimens (n=3), LCM-dissected epithelium of normal ducts (n=3), normal ducts in chronic pancreatitis (n=5) and cancer cells from duct-like structures in PDAC (n=6) are analysed. The results are presented as arbitrary units normalized to housekeeping genes. Large horizontal bars represent the median values

Localization of BNIP3 in pancreatic tissues and correlation with patient survival

To evaluate the pattern of BNIP3 protein distribution in the pancreas, we performed immunohistochemical analysis of normal and diseased tissues. (i) Normal tissue: The most prominent and consistent cytoplasmatic BNIP3 immunoreactivity was present in the acinar cells of the normal pancreas. Islets in the normal pancreas were very faintly stained compared to acini and ductal epithelial cells were almost always devoid of BNIP3 immunoreactivity (Figure 2a). The specificity of the observed staining was confirmed using consecutive sections with control IgG2b (Figure 2b). Only one out of 20 donor samples displayed some faint staining of the ductal epithelium. (ii) Normal tissue next to diseased foci: The staining pattern of acini, islets and ducts changed dramatically in the areas of microscopically normal tissue areas next to diseased foci (NND). In the regions where acini were insulated by the excessive stroma, the diffuse cytoplasmic staining of the acini was replaced by punctate perinuclear staining (Figure 2c). In some of the islets located in the diseased pancreas, the intensity of the punctate staining reached or exceeded that of acini (Figure 2d). The ductal epithelium became immunopositive in 50% of the samples (Figure 2e; negative control Figure 2f). (iii) PanIN lesions: BNIP3 expression in the preneoplastic lesions was assessed utilizing tissue arrays for PanIN lesions (n=96). In general, the percentage of BNIP3 staining increased as the ducts became more dysplastic, with 75–100% of PanIN 2/3 lesions displaying BNIP3 immunoreactivity (Figure 3a–d). A similar pattern of staining was also observed in incidental PanIN lesions seen in the sections of CP and PDAC tissues. Especially in PanIN-2 and 3 lesions, there was a change from homogenous cytoplasmic to punctate perinuclear staining (Figure 3c, d). This type of staining was mostly seen in the cells of the outermost layer of the PanIN 2/3 lesions. Although the number of PanIN 2 and 3 lesions evaluated was not high enough for reliable statistical calculations, this pattern was consistent. (iv) CP: In six out of the 10 sections obtained from CP patients, there was ductal BNIP3 staining, with 40% of the normal-appearing ducts showing immunoreactivity. The cytoplasmic staining of the epithelial ductal cells in CP was more intense than that observed in the ducts of the normal or NND samples (Figure 4a; negative control Figure 4b). Most of the acinar cells and islets were replaced by the excessive stromal tissue and in the remaining acini the staining pattern changed from the diffuse cytoplasmic to the punctate perinuclear in one-third of the samples evaluated. The tubular complexes seen in CP were almost always immunopositive, showing intensive staining (data not shown). (v) PDAC: Among evaluated PDAC sections (n=70), only 29 (41%) were counted as positive, mostly displaying diffuse cytoplasmic staining (Figure 4c–e). The punctate perinuclear pattern seen in PanIN 2/3 lesions was mostly lost in PDAC. The change in appearance of acinar and islet cells in the PDAC sections was similar to that in CP. Notably, there were some sections in which PanIN 3 lesions were stained, whereas the cancer cells next to these lesions displayed no immunoreactivity (Figure 4e). The studied groups, the number of counted sections and ducts, as well as results of the staining are summarized in Figure 5a and Table 1. Among 70 PDAC patients, 49 had confirmed survival data at the time of the final assessment, with 14 of these patients being alive. Patients were divided into two groups according to their BNIP3 immunopositivity. Both groups were comparable in terms of tumor stage and adjuvant therapy application. When the overall survival analysis of these patients was performed according to the BNIP3 immunoreactivity, shorter survival of PDAC patients correlated with low BNIP3 expression (P=0.013). Thus, the median survival of BNIP3-positive patients (n=23) was 14 months (s.e.±1.72; 95% CI=4.62–11.38), whereas it was 8 months (s.e.±1.51; 95% CI=11.04–16.96) in BNIP3-negative patients (n=26) (Figure 5b).

Table 1 Summary of the tissue types, number of analysed sections and counted ducts, with their percentage of BNIP3 staining
Figure 2

Localization of BNIP3 in the normal pancreas and microscopically proven normal pancreas next to diseased areas. Immunohistochemistry was carried out using a monoclonal BNIP3 antibody (a, ce) or negative control antibody (b, f) on sequential sections as described in the Materials and methods section. (a, b) Normal pancreas, showing prominent homogenous acinar staining, faint islet staining and nonstained ductal epithelium (× 50). (c) Acinar cells in NND areas displaying punctate perinuclear BNIP3 staining (× 400). (d) Prominent islet staining in normal pancreas next to diseased areas (× 400). (e, f) Microscopically normal ducts next to diseased areas showing BNIP3 immunoreactivity (× 200)

Figure 3

Localization of BNIP3 in typical PanIN lesions. Immunohistochemistry was carried out using a monoclonal BNIP3 antibody as described in the Materials and methods section. (a) PanIN 1A; (b) PanIN-1B; (c) PanIN-2; (d) PanIN-3 (× 200). Note the change from homogenous cytoplasmic staining of PanIN 1A and 1B lesions to punctate perinuclear staining in PanIN 2 and 3 lesions. Also note that this punctate staining is mostly evident in the cells of the outermost layer of the papillary lesions

Figure 4

Localization of BNIP3 protein in CP and PDAC. Immunohistochemistry was carried out using a monoclonal BNIP3 antibody (a, c, e) or negative control antibody (b, d, f) in consecutive sections as described in the Materials and methods section. (a, b) Normal ductal epithelium in CP displaying BNIP3 immunoreactivity (× 50). (cf) PDAC displaying BNIP3 immunoreactivity in 41% (× 400) (c) and absent staining in the remaining 59% (× 200) (e). Note the loss of BNIP3 immunoreactivity in PDAC (e, black arrowheads) while still present in an adjacent PanIN 3 lesion (e, upper left corner)

Figure 5

The percentage of BNIP3 immunopositivity in different pancreatic tissues and correlation with survival in pancreatic cancer. (a) The comparison of percentages of BNIP3 immunostaining in pancreatic ducts, PanIN lesions and duct-like structures in PDAC. In total, 20 normal, five microscopically normal tissue areas next to disease, 10 CP and 70 PDAC samples were evaluated. PanIN lesions were analysed using tissue arrays of a total of 96 samples. The number of ducts counted are summarized in Table 1. Any duct with staining present in more than 10% of the epithelial cells was counted as positive. The black columns represent the detected BNIP3 positivity and the rest of the framed columns represent the hypothetical potential of BNIP3 inducibility under hypoxic conditions. (Since tissue hypoxia is getting more profound throughout the disease progression from normal to cancer, BNIP3 immunopositivity increases until a certain threshold is reached (PanIN2 lesions). Thereafter, although the hypoxic milieu prevails, the percentage of BNIP3 immunopositivity decreases, most likely due to BNIP3 silencing.) NND: microscopically normal tissue next to disease; CP: chronic pancreatitis; PDAC: pancreatic ductal adenocarcinoma. (b) Survival analysis of BNIP3-positive vs BNIP3-negative patients compared by immunohistochemistry. The median survival of BNIP3-positive patients (n=23) was 14 months, which was significantly longer than the median 8-month survival of BNIP3-negative patients (n=26). Both groups are comparable in terms of tumor stage and adjuvant therapy application

BNIP3 and chemoresistance

To analyse the role of BNIP3 in chemoresistance, we first selected BNIP3-positive pancreatic cell lines in which BNIP3 expression was detectable and then downregulated BNIP3 expression with siRNA transfection, prior to exposure to the chemotherapeutic drugs. BNIP3 mRNA was present in Colo-357, Panc-1, SU86.86 and T3M4, but not in ASPC-1, BxPc-3, Capan-1, or MiaPaCa-2 cell lines. When all cell lines were exposed to hypoxia for 17 h, BNIP3 expression increased significantly (P=0.007) only in the cell lines which also expressed BNIP3 mRNA under normoxic conditions: 5.3-fold in Colo-357, 11.8-fold in Panc-1, 16.2-fold in SU86.86 and 13-fold in T3M4 cells. In contrast, there was no detectable induction in ASPC-1, BxPc-3, Capan-1, or MiaPaCa-2 cell lines (Figure 6a, c). In three out of four inducible cell lines, the increase was also detected at the protein level by the appearance of a specific 30 kDa band corresponding to the known size of BNIP3, which was not detectable under normoxic conditions (Figure 6c). The additional 60 kDa band most likely represented a nonspecific band, since this band was neither inducible by hypoxia, nor did it parallel the disappearance/reduction of the 30 kDa band after siRNA transfection. In addition, it is unlikely that this band represent a BNIP3 dimer (as it has been suggested by other investigators) since it did not appear to parallel the 30 kDa band and since the protein extraction and the immunoblotting were made under reducing conditions and 1% SDS.

Figure 6

Expression analysis of BNIP3 in pancreatic cancer cell lines under normoxic and hypoxic conditions. Induction of hypoxia, Western blot and QRT–PCR analysis were carried out as described in the Methods section. (a) Western blot analysis of BNIP3 after 17 h of hypoxia in eight pancreatic cancer cell lines. (Lower panel) γ-Tubulin was used as a control for equal loading and the prostatic cancer cell line DU 145 was used as positive control. (ac) BNIP3 mRNA (b) and protein (c) expression under normoxic (N) and 17-h hypoxic (H) conditions in the indicated cell lines. BNIP3 mRNA expression is shown in a logarithmic scale as copies/μl cDNA normalized to housekeeping genes. Error bars represent the s.e.m. of three different experiments

When the three BNIP3-positive cell lines were treated with specific BNIP3 siRNA molecules, a significant reduction of the mRNA expression was detected in Panc-1 (2.6-fold), SU86.86 (3.4-fold) and T3M4 (1.8-fold) cells compared to negative control siRNA-transfected cells, as determined by QRT–PCR analysis (Figure 7a). Since these levels of basal and reduced protein expressions were under the detection limit of the Western blot analysis, control and BNIP3 siRNA-transfected cells were further induced by hypoxia and levels of the BNIP3 expression were examined by QRT–PCR and Western blot analysis. As seen in Figure 7b, there was a proportionate reduction of BNIP3 mRNA expression compared to normoxic conditions (Panc-1: 2.9-fold; SU86.86: 5.7-fold; T3M4: 2.5-fold), which was paralleled by a reduction/disappearance of BNIP3 protein levels as determined by immunoblotting (Figure 7c). To analyse whether reduction of the BNIP3 expression would affect sensitivity of the pancreatic cancer cells to cytotoxic drugs, control (negative control siRNA transfected) and BNIP3-silenced Panc-1, SU86.86 and T3M4 cell lines were exposed to the graded concentrations of the 5-FU and gemcitabine for 48 h under normoxic conditions. BNIP3-silenced cells showed increased chemoresistance, as illustrated in Figure 8. Panc-1 cells became significantly more resistant to gemcitabine (P=0.016) and SU86.86 cells to 5-FU (P=0.028). Although it did not reach statistical significance, the EC50 of gemcitabine increased from 8.5 ng/ml to more than 100 ng/ml in SU86.86 cells. Small shifts in the same direction were also observed in T3M4 cells; however, there were no significant changes in the response to either drug.

Figure 7

Evaluation of the efficacy of BNIP3 siRNA silencing in pancreatic cancer cells. siRNA transfection, QRT–PCR and Western blot analysis were carried out as described in the Materials and methods section. (a, b) BNIP3 mRNA expression in control and siRNA-transfected cells under normoxic (a) and hypoxic (b) conditions. (c) BNIP3 protein levels in control and siRNA-transfected cells under hypoxic condition

Figure 8

Effects of BNIP3 levels on the response to 5-FU and gemcitabine in pancreatic cancer cells. Negative control transfected cells (dashed line) and BNIP3-silenced (straight line) Panc-1, SU86.86 and T3M4 cell lines were continuously exposed to increasing 5-FU and gemcitabine concentrations for 48 h as indicated. siRNA transfection and MTT assay were carried out as described in the Materials and methods section. Results are presented as the mean±s.e.m. of three independent experiments


Pancreatic cancer is the fourth to fifth leading cause of cancer-related death in industrialized Western countries (Warshaw and Fernandez-del Castillo, 1992). Clinically, PDAC is often diagnosed at a late stage, when local invasion of vital structures and distant metastasis has already precluded curative resection. Consequently, only 10–20% of the patients are candidates for surgery at the time of diagnosis, of which only approximately 25% will survive for more than 5 years (Yeo et al., 1997; Richter et al., 2003; Wagner et al., 2004). Moreover, PDAC is often highly resistant to the currently available chemotherapeutics as well as radiotherapy (Klinkenbijl et al., 1999; Neoptolemos et al., 2004). Therefore, elucidation of the mechanisms exploited by PDAC is urgently required to develop more effective therapeutic strategies. Molecular research has contributed significantly to a better understanding of the aggressive nature of this disease and multiple genetic and epigenetic alterations have been identified over the past decade (Hanahan and Weinberg, 2000; Friess et al., 2003).

5-FU and gemcitabine are used as first-line drugs for patients with advanced PDAC. The response rate of PDAC to 5-FU is around 20% (Cullinan et al., 1990; Crown et al., 1991). Gemcitabine – a novel pyrimidine analogue – provides a slightly improved clinical results and fewer side effects (Burris et al., 1997). Both drugs induce endogenous apoptotic mechanisms to exert their cytotoxic effects (Korsmeyer, 1992; Kerr et al., 1994; White, 1996; Bold et al., 1999). However, evading apoptosis is one of the major mechanisms exploited by PDAC and imbalanced expression of the apoptosis-regulating molecules – including members of the Bcl-2 family, such as Bcl-2, Bcl-XL, Bax and Bak – is a common feature (Friess et al., 1998a, 1998b; Graber et al., 1999; Hanahan and Weinberg, 2000; Shi et al., 2002; Xu et al., 2002).

We and others have previously identified BNIP3 as a differentially expressed gene in PDAC in comparison to the normal pancreas (Crnogorac-Jurcevic et al., 2002; Friess et al., 2003; Brandt et al., 2004). By microarray analysis, BNIP3 was found to be 4.1- and 5.3-fold reduced in CP and PDAC, respectively. Recently, it has also been shown that BNIP3 expression is silenced in PDAC by hypermethylation of its promoter (Okami et al., 2004). Moreover, it has also been proposed that early hypoxic induction of BNIP3 in ductal carcinoma in situ of the breast and in non-small-cell lung carcinoma may lead to the selection of the apoptosis-resistant clones within the hypoxic–necrotic environment, resulting in a more agressive tumor phenotype (Sowter et al., 2003; Giatromanolaki et al., 2004).

Comparison of BNIP3 mRNA levels in whole tissue extracts of CP, PDAC and normal pancreas in the present study revealed a significant reduction of BNIP3 expression in both pathologies. PDAC is characterized by scarce neoplastic cells embedded in a densely desmoplastic stroma and a similar abundance of fibrotic stroma is also a characteristic of CP (Crnogorac-Jurcevic et al., 2002). Since BNIP3 was abundant in normal pancreatic acini, reduced expression of BNIP3 in CP and PDAC could result from the mere reduction of the acinar compartment in both pathologies.

However, as seen by immunohistochemistry, there was a dramatic change between normal tissues and microscopically proven normal tissue next to disease in CP and PDAC. The major differences were that (1) the diffuse cytoplasmic staining of the acini and islet cells was replaced by a sharply punctate staining in the perinuclear area which was not observed in normal tissues and (2) half of the islets, which showed very weak BNIP3 immunoreaction in normal tissues, displayed strong BNIP3 immunostaining in NND.

BNIP3 belongs to a subfamiliy of proapoptotic mitochondrial proteins (Chen et al., 1999). Upon induction by hypoxia, BNIP3 is known to co-localize with the outer mitochondrial membrane, where it initiates programmed cell death (Chen et al., 1997; Adams and Cory, 1998; Yasuda et al., 1998; Imazu et al., 1999; Ray et al., 2000). Since tissue hypoxia occurs in PDAC and CP (Koong et al., 2000; Harris, 2002), it is probable that hypoxia-induced BNIP3 overexpression (as reflected by focal accumulation of BNIP3 in the perinuclear area) in the acinar and islet cells adjacent to chronically inflamed or malignant tissue areas may cause acinar and islet cell death. However, this hypothesis requires further experimental studies.

Ductal epithelial cells are considered to be the normal counterpart of malignant PDAC cells (Crnogorac-Jurcevic et al., 2002). Therefore, homogenous populations of normal, CP and PDAC ducts obtained by laser capture microdissection were compared for differences in BNIP3 expression. The level of BNIP3 expression was highest in normal ducts and decreased gradually for CP ducts and finally for cancer cells. PanIN lesions are currently classified as the pre-neoplastic lesions reflecting different stages of malignant transformation characterized by the accumulation of genetic alterations such as mutations of the K-ras proto-oncogene, p53, p16 (INK4A) and SMAD4 tumor suppressor genes (Moskaluk et al., 1997; Wilentz et al., 2000; Takaori et al., 2004).

Using PanIN tissue arrays allowed us to follow the alteration of BNIP3 expression during the malignant transformation process. The percentage of staining increased as the ducts became more dysplastic, with 75–100% of PanIN 2/3 lesions displaying BNIP3 staining. Similar to acini and islets seen in areas next to disease, the subcellular staining pattern changed from homogenous cytoplasmic to punctate perinuclear, especially in PanIN 2 and 3 lesions. This type of staining was most prominent in the cells of the outermost layer of the PanIN 3 lesions, an area where hypoxia is expected to be more profound than at the base. In contrast, only 41% of the cancer cells were immunopositive. Apparently, the major loss of BNIP3 occurs at a later stage of the malignant transformation. Since there is a hypoxic environment in advanced stages of malignant transformation, we hypothesized that the observed late reduction of immunopositivity may be due to the apoptotic extinction of cells in which BNIP3 was initially induced by hypoxia.

It is an already known phenomenon that cells in rapidly growing tumors are intermittently, or sometimes constantly, exposed to hypoxic conditions (Sowter et al., 2001, 2003; Greijer and van der Wall, 2004). Some cells may adapt to the environmental stress and survive. These selected hypoxia-resistant cells have a more aggressive phenotype (Harris, 2002). Therefore, the hypoxic environment of pancreatic cancer may contribute to selection of BNIP3-negative cancer clones. As published previously and demonstrated by our hypoxia experiment, these BNIP3-negative cells remained resistant to the future hypoxia-mediated BNIP3 induction.

An alternative explanation of BNIP3 downregulation may come from our observation that TGF-beta treatment of Colo-357 and Panc-1 cells (which are the only BNIP3-expressing tested cell lines with an intact TGF-beta signaling pathway) reduced BNIP3 mRNA levels by half (data not shown). Together with the ability of TGF-beta to slow the growth of these cells (Kleeff et al., 1999a, 1999b, 1999c), the concomitant downregulation of proapoptotic BNIP3 may offer a survival advantage in the hostile (e.g., hypoxic) environment. As to the clinical consequences, BNIP3-negative patients had a significantly shorter median survival compared to BNIP3-positive patients.

The tumor cells with a reduced sensitivity to apoptosis were also less responsive to anticancer treatment (Schmaltz et al., 1998; Sowter et al., 2003). The susceptibility of PDAC to cytotoxic drugs might therefore be – at least in part – dependent on the balance between pro- and antiapoptotic members of the Bcl-2 family. BNIP3, in this context, is believed to be one of the important members, rather than the sole executioner. Moreover, it has been recently shown that all pancreatic cancer cell lines are at least in vitro relatively insensitive to treatment with 5-FU and showed variable degrees of susceptibility to gemcitabine (Monti et al., 2004). In order to understand the role of BNIP3 in the mechanism of chemoresistance, we used siRNA transfection to downregulate BNIP3 expression in pancreatic cancer cells. The analysis of cytotoxic effects of 5-FU and gemcitabine on BNIP3-modified cells showed that silencing of BNIP3 generally correlated with increased chemoresistance, although cell type- and drug-dependent variations were present. The ambient amount of BNIP3 protein in cancer cells may be important in this context, as well as the amount of silencing induced by siRNA transfection. The maximal mRNA reduction both under normoxic and hypoxic conditions occurred in SU86.86, followed by Panc-1 and T3M4, which is reflected by their chemoresistance in the same order. As observed in three different experiments (Figures 6a, c and 7c), T3M4 cell line displayed the lowest levels of BNIP3 protein, silencing of which did not cause significant chemoresistance.

In concordance with our results, downregulation of BNIP3 in colonic cancer cells paralleled the 5-FU resistance as well (de Angelis et al., 2004). Therefore, silencing of BNIP3 ostensibly shifts the balance between the pro- and antiapoptotic factors in favor of survival.

In summary, we hypothesize that the hypoxic environment in CP and PDAC increases BNIP3 expression, which in turn might lead to the extinction of acinar and islet cells. Loss of BNIP3 at advanced stages of pancreatic cancer may create a more aggressive tumor phenotype and contribute – at least in part – to the chemoresistance observed in this malignancy. Finally, BNIP3 status in patients should further be evaluated as a marker of responsiveness to chemotherapy in order to tailor postoperative adjuvant therapy to the individual patient.

Materials and methods

Pancreatic tissues

Tissue samples were collected from patients following pancreatic resection for PDAC or CP at the University of Bern, Switzerland and the University of Heidelberg, Germany. All patients were informed and written consent was obtained. The studies were approved by the Ethics Committees of the University of Bern (Switzerland) and the University of Heidelberg (Germany). Normal pancreatic tissue samples were obtained at the University of Bern through an organ donor procurement program whenever there was no suitable recipient for pancreas transplantation. During tissue collection, freshly removed samples were either snap frozen in liquid nitrogen and stored at −80°C for protein extraction or preserved in RNA-later solution for future RNA extraction. A portion of the samples was also fixed in 5% parafomaldehyde solution for 12–24 h and then embedded in paraffin for histological analysis.

Human pancreatic cancer cell lines

Aspc-1, BxPc-3, Capan-1, Colo-357, MiaPaCa-2, Panc-1, SU86.86,. and T3M4 were either purchased from ATCC (Rockville, MD, USA) or received as a kind gift of Dr RS Metzgar (Durham, NC, USA). The cells were routinely grown in complete medium (RPMI-1640 supplemented with 10% fetal bovine serum, 100 U/ml penicillin and 100 μg/ml streptomycin) at 37°C, saturated with 5% CO2 in a humid atmosphere.


The following reagents were purchased: RPMI-1640, Trypsin-EDTA and penicillin–streptomycin from Invitrogen (Karlsruhe, Germany); fetal bovine serum from PAN Biotech (Aidenbach, Germany); mouse monoclonal anti-BNIP3 antibody from Sigma-Aldrich (Taufkirchen, Germany), monoclonal mouse IgG2b negative control from Cymbus (Hampshire, UK); anti-mouse IgG HRPO-linked antibodies and ECL immunoblotting detection reagents from Amersham Biosciences (Amersham Life Science, Amersham, UK); Envision system, antibody diluent and liquid DAB+ substrate from DAKO (Hamburg, Germany); mini EDTA-Free Protease inhibitor from Roche Molecular Biochemicals (Basel, Switzerland); BCA protein assay from Pierce Chemical Co. (Rockford, IL, USA); RNA-later solution from Ambion (Huntington, UK); BNIP3 validated small interfering RNA (siRNA), negative control siRNA and RNAiFect transfection reagent from Qiagen (Hilden, Germany) and gemcitabine from Lily (Fegersheim, France). All other reagents were from Sigma-Aldrich Chemical Company (Taufkirchen, Germany).

Real-time Light Cycler® QRT–PCR

All reagents and equipment for mRNA/cDNA preparation were purchased from Roche Applied Science Diagnostics (Mannheim, Germany). mRNA extractions were prepared by automated isolation using the MagNA Pure LC instrument and isolation kits I (for cells) and II (for tissue). cDNA was prepared using the first-strand cDNA synthesis kit (AMV) according to the manufacturer's instructions. Real-time PCR was performed with the Light Cycler Fast Start DNA SYBR Green kit. All primers were obtained from Search-LC (Heidelberg, Germany). The calculated number of specific transcripts was normalized to the housekeeping genes cyclophilin B and hypoxanthine guanine phosphoribosyltransferase and expressed as number of copies per μl of input cDNA.

Laser capture microdissection (LCM)

Tissue samples were embedded in OCT (Sakura Fietek, CA, USA) by freezing the blocks with liquid nitrogen in an acetone bath. Tissue sections (6–8 μm thick) were prepared using the Reichard-Jung 1800 cryostat. LCM and RNA extraction were carried out as described previously (Ketterer et al., 2003).

cDNA array

The HG-U95Av2 array from Affymetrix (Santa Clara, CA, USA) was utilized for analysis. Poly(A)+RNA isolation, cDNA synthesis and cRNA in vitro transcription, purification and fragmentation were carried out as reported previously (Ketterer et al., 2003). Hybridization of the fragmented in vitro transcription products to oligonucleotide arrays was performed as suggested by the manufacturer (Affymetrix).


Two sequential 3 μm thick paraffin-embedded tissue sections were placed on the same slide, de-paraffinized and re-hydrated. One section was used for analysis while the other was used as the negative control. Antigen retrieval was performed by boiling the slides in 10 mM citrate buffer two times for 10 min. Peroxidase was quenched with a 3% H2O2 solution in 30% methanol. DakoCytomation antibody diluent was used to dilute both the mouse monoclonal anti-BNIP3 antibody (1 : 7500) and the negative-control IgG2b (1 : 375). After an overnight incubation at 4°C, slides were washed with Tris buffer supplemented with 0.05% Tween-20 (TBS-T) and exposed to the HRPO-linked anti-mouse secondary antibody for 45 min at room temperature. Color reaction was carried out by incubation for 4 min with liquid DAB+substrate and counterstaining by Mayer's hematoxylin solution. Two researchers (ME, IE) evaluated the results independently for lesion severity, intensity of staining and percentage of positive cells after inspection of all the fields in the tissue sections. For evaluation of the ductal structures, sections were scanned at low (× 50) magnification. Each identified duct was then evaluated at higher magnifications (× 200/ × 400). Any duct with staining present in more than 10% of the epithelial cells was counted as positive. When more than 30% of the ducts in a section were positive, that section was counted as positive. Cancer sections without the formation of duct-like structures were scored as positive if more than 10% of the cancer cells were stained. A Zeiss Axiocam 3.1 system was used throughout the process.


Cell culture monolayers were homogenized and lysed with 0.5 ml. buffer containing Tris-HCl (pH 7.5), 150 mM NaCl, 2 mM EDTA, 1% SDS and one tablet of complete mini-EDTA-free protease inhibitor cocktail (per 10 ml of the buffer). Protein concentration was determined by BCA protein assay. Samples containing 30 μg of the protein extract were size-fractioned by 12% SDS–PAGE and transferred onto nitrocellulose membranes by the application of 100 V for 90 min. Blots were blocked with TBS-T plus 5% non-fat milk for 1 h, incubated with mouse monoclonal anti-BNIP3 antibody (1 : 1000) overnight at 4°C, washed with TBS-T and incubated with HRPO-linked sheep anti-mouse antibody (1 : 3000) for 1 h at room temperature. After washing with TBS-T, antibody detection was performed using the enhanced chemoilluminescence (ECL) reaction system. Protein extract of the DU145 prostatic cancer cell line was used as a positive control. Equal loading was verified by re-probing the membranes with anti-gamma-tubulin antibodies.

Induction of hypoxia

Pancreatic cancer cell lines were incubated in the modular chamber saturated with the hypoxic air mixture (89.25% N2+10% CO2+0.75%O2) for 17 h at 37°C.

siRNA transfection

Panc-1, SU86.86 and T3M4 cells were seeded in six-well plates in duplicates at the density of 2.5 × 105 cells/well in 2.5 ml of complete medium. At 24 h after seeding, either specific BNIP3 siRNA or negative control siRNA was added to give a final concentration of 100 nM for 72 h. For 3-(4,5-methylthiazol-2-yl)-2,5-diphenyl-tetrazolium-bromide (MTT) assays, cells were trypsinized and transferred into 96-well plates. For mRNA and protein extraction, cells were kept in six-well plates. At 55 h after transfection, six-well plates were put into the hypoxia chamber for 17 h, while the controls were kept under normoxic conditions. All experiments were repeated three times.

MTT assay

To assess cell proliferation, the MTT test was used as described previously (Guo et al., 2004). siRNA-transfected cells were trypsinized after 12 h and seeded in triplicate in 96-well plates at densities of 5000 cells/well for SU86.86 and 6000 cells/well for Panc-1 and T3M4. After 12 h, 5-FU or gemcitabine was added at the indicated concentration. Dimethyl sulfoxide (0.01%) and PBS were used as controls. After 48 h of incubation, 20 μl MTT (5 mg/ml) was added for 4 h. Formazan products were solubilized with acidic isopropanol and the optical density was measured at 570 nm. All experiments were repeated three times.


Statistical analysis and graph presentation were made using GraphPad Prism 2 Software (GraphPad, San Diego, CA, USA) and SPSS 12 for Windows (SPSS Inc.,. Chicago, IL, USA). QRT–PCR measurements in pancreatic tissues are presented as median and range and in cell lines as mean±s.e.m. MTT results are presented as mean±s.e.m. The Mann–Whitney U-test was used to test differences between groups. The Kaplan–Meier method and log-rank test were used for survival analysis. P<0.05 was taken as the level of significance.


  1. Adams JM and Cory S . (1998). Science, 281, 1322–1326.

  2. Bacon AL and Harris AL . (2004). Ann. Med., 36, 530–539.

  3. Bani MR, Nicoletti MI, Alkharouf NW, Ghilardi C, Petersen D, Erba E, Sausville EA, Liu ET and Giavazzi R . (2004). Mol. Cancer Ther., 3, 111–121.

  4. Bold RJ, Chandra J and McConkey DJ . (1999). Ann. Surg. Oncol., 6, 279–285.

  5. Brandt R, Grutzmann R, Bauer A, Jesnowski R, Ringel J, Lohr M, Pilarsky C and Hoheisel JD . (2004). Pancreatology, 4, 587–597.

  6. Bruick RK . (2000). Proc. Natl. Acad. Sci. USA, 97, 9082–9087.

  7. Burris III HA, Moore MJ, Andersen J, Green MR, Rothenberg ML, Modiano MR, Cripps MC, Portenoy RK, Storniolo AM, Tarassoff P, Nelson R, Dorr FA, Stephens CD and Von Hoff DD . (1997). J. Clin. Oncol., 15, 2403–2413.

  8. Chen G, Cizeau J, Vande Velde C, Park JH, Bozek G, Bolton J, Shi L, Dubik D and Greenberg A . (1999). J. Biol. Chem., 274, 7–10.

  9. Chen G, Ray R, Dubik D, Shi L, Cizeau J, Bleackley RC, Saxena S, Gietz RD and Greenberg AH . (1997). J. Exp. Med., 186, 1975–1983.

  10. Crnogorac-Jurcevic T, Efthimiou E, Nielsen T, Loader J, Terris B, Stamp G, Baron A, Scarpa A and Lemoine NR . (2002). Oncogene, 21, 4587–4594.

  11. Crown J, Casper ES, Botet J, Murray P and Kelsen DP . (1991). J. Clin. Oncol., 9, 1682–1686.

  12. Cullinan S, Moertel CG, Wieand HS, Schutt AJ, Krook JE, Foley JF, Norris BD, Kardinal CG, Tschetter LK and Barlow JF . (1990). Cancer, 65, 2207–2212.

  13. Daido S, Kanzawa T, Yamamoto A, Takeuchi H, Kondo Y and Kondo S . (2004). Cancer Res., 64, 4286–4293.

  14. de Angelis PM, Fjell B, Kravik KL, Haug T, Tunheim SH, Reichelt W, Beigi M, Clausen OP, Galteland E and Stokke T . (2004). Int. J. Oncol., 24, 1279–1288.

  15. Friess H, Buchler MW, Mueller C and Malfertheiner P . (1998a). Gastroenterology, 115, 1018–1022.

  16. Friess H, Ding J, Kleeff J, Fenkell L, Rosinski JA, Guweidhi A, Reidhaar-Olson JF, Korc M, Hammer J and Buchler MW . (2003). Cell Mol. Life Sci., 60, 1180–1199.

  17. Friess H, Lu Z, Graber HU, Zimmermann A, Adler G, Korc M, Schmid RM and Buchler MW . (1998b). Gut, 43, 414–421.

  18. Giatromanolaki A, Koukourakis MI, Sowter HM, Sivridis E, Gibson S, Gatter KC and Harris AL . (2004). Clin. Cancer Res., 10, 5566–5571.

  19. Graber HU, Friess H, Zimmermann A, Korc M, Adler G, Schmid R and Buchler MW . (1999). J. Gastrointest. Surg., 3, 74–80; discussion 81.

  20. Graeber TG, Osmanian C, Jacks T, Housman DE, Koch CJ, Lowe SW and Giaccia AJ . (1996). Nature, 379, 88–91.

  21. Graham RM, Frazier DP, Thompson JW, Haliko S, Li H, Wasserlauf BJ, Spiga MG, Bishopric NH and Webster KA . (2004). J. Exp. Biol., 207, 3189–3200.

  22. Greijer AE and van der Wall E . (2004). J. Clin. Pathol., 57, 1009–1014.

  23. Guo J, Kleeff J, Li J, Ding J, Hammer J, Zhao Y, Giese T, Korc M, Buchler MW and Friess H . (2004). Oncogene, 23, 71–81.

  24. Hanahan D and Weinberg RA . (2000). Cell, 100, 57–70.

  25. Harris AL . (2002). Nat. Rev. Cancer, 2, 38–47.

  26. Imazu T, Shimizu S, Tagami S, Matsushima M, Nakamura Y, Miki T, Okuyama A and Tsujimoto Y . (1999). Oncogene, 18, 4523–4529.

  27. Kennedy AS, Harrison GH, Mansfield CM, Zhou XJ, Xu JF and Balcer-Kubiczek EK . (2000). Int. J. Cancer, 90, 175–185.

  28. Kerr JF, Winterford CM and Harmon BV . (1994). Cancer, 73, 2013–2026.

  29. Ketterer K, Rao S, Friess H, Weiss J, Buchler MW and Korc M . (2003). Clin. Cancer Res., 9, 5127–5136.

  30. Kleeff J, Friess H, Simon P, Susmallian S, Buchler P, Zimmermann A, Buchler MW and Korc M . (1999a). Digest. Dis. Sci., 44, 1793–1802.

  31. Kleeff J, Ishiwata T, Maruyama H, Friess H, Truong P, Buchler MW, Falb D and Korc M . (1999b). Oncogene, 18, 5363–5372.

  32. Kleeff J, Maruyama H, Friess H, Buchler MW, Falb D and Korc M . (1999c). Biochem. Biophys. Res. Commun., 255, 268–273.

  33. Klinkenbijl JH, Jeekel J, Sahmoud T, van Pel R, Couvreur ML, Veenhof CH, Arnaud JP, Gonzalez DG, de Wit LT, Hennipman A and Wils J . (1999). Ann. Surg., 230, 776–782; discussion 782–784.

  34. Koong AC, Mehta VK, Le QT, Fisher GA, Terris DJ, Brown JM, Bastidas AJ and Vierra M . (2000). Int. J. Radiat. Oncol. Biol. Phys., 48, 919–922.

  35. Korsmeyer SJ . (1992). Blood, 80, 879–886.

  36. Kothari S, Cizeau J, McMillan-Ward E, Israels SJ, Bailes M, Ens K, Kirshenbaum LA and Gibson SB . (2003). Oncogene, 22, 4734–4744.

  37. Lamy L, Ticchioni M, Rouquette-Jazdanian AK, Samson M, Deckert M, Greenberg AH and Bernard A . (2003). J. Biol. Chem., 278, 23915–23921.

  38. Monti P, Marchesi F, Reni M, Mercalli A, Sordi V, Zerbi A, Balzano G, Di Carlo V, Allavena P and Piemonti L . (2004). Virchows Arch., 445, 236–247.

  39. Moskaluk CA, Hruban RH and Kern SE . (1997). Cancer Res., 57, 2140–2143.

  40. Neoptolemos JP, Stocken DD, Friess H, Bassi C, Dunn JA, Hickey H, Beger H, Fernandez-Cruz L, Dervenis C, Lacaine F, Falconi M, Pederzoli P, Pap A, Spooner D, Kerr DJ and Buchler MW . (2004). N. Engl. J. Med., 350, 1200–1210.

  41. Okami J, Simeone DM and Logsdon CD . (2004). Cancer Res., 64, 5338–5346.

  42. Ray R, Chen G, Vande Velde C, Cizeau J, Park JH, Reed JC, Gietz RD and Greenberg AH . (2000). J. Biol. Chem., 275, 1439–1448.

  43. Reynolds TY, Rockwell S and Glazer PM . (1996). Cancer Res., 56, 5754–5757.

  44. Richter A, Niedergethmann M, Sturm JW, Lorenz D, Post S and Trede M . (2003). World J. Surg., 27, 324–329.

  45. Schmaltz C, Hardenbergh PH, Wells A and Fisher DE . (1998). Mol. Cell. Biol., 18, 2845–2854.

  46. Schmidt-Kastner R, Aguirre-Chen C, Kietzmann T, Saul I, Busto R and Ginsberg MD . (2004). Brain Res., 1001, 133–142.

  47. Shi X, Liu S, Kleeff J, Friess H and Buchler MW . (2002). Oncology, 62, 354–362.

  48. Sowter HM, Ferguson M, Pym C, Watson P, Fox SB, Han C and Harris AL . (2003). J. Pathol., 201, 573–580.

  49. Sowter HM, Ratcliffe PJ, Watson P, Greenberg AH and Harris AL . (2001). Cancer Res., 61, 6669–6673.

  50. Takaori K, Hruban RH, Maitra A and Tanigawa N . (2004). Pancreas, 28, 257–262.

  51. Tenedini E, Fagioli ME, Vianelli N, Tazzari PL, Ricci F, Tagliafico E, Ricci P, Gugliotta L, Martinelli G, Tura S, Baccarani M, Ferrari S and Catani L . (2004). Blood, 104, 3126–3135.

  52. Tracey L, Spiteri I, Ortiz P, Lawler M, Piris MA and Villuendas R . (2004). J. Interferon Cytokine Res., 24, 185–195.

  53. Vande Velde C, Cizeau J, Dubik D, Alimonti J, Brown T, Israels S, Hakem R and Greenberg AH . (2000). Mol. Cell. Biol., 20, 5454–5468.

  54. Wagner M, Redaelli C, Lietz M, Seiler CA, Friess H and Buchler MW . (2004). Br. J. Surg., 91, 586–594.

  55. Warshaw AL and Fernandez-del Castillo C . (1992). N. Engl. J. Med., 326, 455–465.

  56. White E . (1996). Genes Dev., 10, 1–15.

  57. Wilentz RE, Iacobuzio-Donahue CA, Argani P, McCarthy DM, Parsons JL, Yeo CJ, Kern SE and Hruban RH . (2000). Cancer Res., 60, 2002–2006.

  58. Xu ZW, Friess H, Buchler MW and Solioz M . (2002). Cancer Chemother. Pharmacol., 49, 504–510.

  59. Yasuda M, Theodorakis P, Subramanian T and Chinnadurai G . (1998). J. Biol. Chem., 273, 12415–12421.

  60. Yeo CJ, Cameron JL, Sohn TA, Lillemoe KD, Pitt HA, Talamini MA, Hruban RH, Ord SE, Sauter PK, Coleman J, Zahurak ML, Grochow LB and Abrams RA . (1997). Ann. Surg., 226, 248–257; discussion 257–260.

  61. Yook YH, Kang KH, Maeng O, Kim TR, Lee JO, Kang KI, Kim YS, Paik SG and Lee H . (2004). Biochem. Biophys. Res. Commun., 321, 298–305.

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article


  • hypoxia
  • apoptosis
  • PanIN
  • Bcl-2
  • chemoresistance
  • TGF-beta

Further reading