We have shown that one of the principle mechanisms of chemotherapy resistance involves the activation of nuclear factor kappa-B (NF-κB). In an effort to identify NF-κB-regulated chemotherapy response genes, we performed a microarray assay and observed that heparin-binding EGF-like growth factor (HB-EGF) was significantly upregulated by SN38 (a strong inducer of NF-κB activity) in colon cancer cells. Further studies revealed that HB-EGF was rapidly induced following a variety of chemotherapy treatments. Using RNA interference, we demonstrated that the chemotherapy-induced HB-EGF was largely dependent on activator protein-1 (AP-1) and NF-κB activation. Constitutive HB-EGF expression rescued AP-1/NF-κB small interfering RNA (siRNA) cells from chemotherapy-induced apoptosis. Meanwhile, we found that the enzymatic shedding of HB-EGF was also regulated by chemotherapy treatment, resulting in the elevated release of soluble HB-EGF from the cellular membrane. Induction of HB-EGF expression and ectodomain shedding synergistically led to robust epidermal growth factor receptor (EGFR) phosphorylation, whereas inhibition of HB-EGF expression by use of the HB-EGF inhibitor (CRM197) or siRNA resulted in the suppression of chemotherapy-induced EGFR phosphorylation. These results suggest that the chemotherapy-induced EGFR activation is regulated by HB-EGF. Finally, we demonstrated that overexpression of HB-EGF led to apoptotic resistance to chemotherapy, whereas suppression of HB-EGF expression by siRNA resulted in a dramatic increase in cell death. In summary, our study suggests that chemotherapy-induced HB-EGF activation represents a critical mechanism of inducible chemotherapy resistance. Therefore, therapeutic intervention aimed at inhibiting HB-EGF activity may be useful in cancer prevention and treatments.
Drug resistance is one of the principal reasons for chemotherapy failure. Currently, resistance to first-line chemotherapy affects 50–70% of patients with metastatic colorectal cancer. Similarly, resistance to conventional anticancer treatments in patients with pancreatic cancers have led to only minimal survival improvements. Independent of chemotherapy agents' mechanism of function (e.g. antimetabolite, DNA damaging, etc.), the cellular response to drug treatment ultimately determines whether cancer cells undergo apoptosis or not. Previously, we have shown that one principal mechanism of chemotherapy resistance involves the activation of the transcription factor nuclear factor kappa-B (NF-κB) (Cusack et al., 1999, 2001; Wang et al., 1999; Russo et al., 2001; Nakanishi and Toi, 2005). This activation leads to the expression of a variety of genes that are involved in cell proliferation and antiapoptosis. However, the specific NF-κB-regulated genes that are involved in the chemotherapy resistance have not been fully elucidated. Using a microarray-based differential gene expression assay, we identified that heparin-binding EGF-like growth factor (HB-EGF) may be involved in NF-κB-regulated chemotherapy resistance.
HB-EGF is a member of the EGF growth factor family (Nishi and Klagsbrun, 2004). It is synthesized as a transmembrane protein (pro-HB-EGF) that can undergo proteolytic cleavage at the cell surface to release a mature soluble 14–22-kDa N-terminal ectodomain (s-HB-EGF), a process often referred to as ‘ectodomain shedding’. s-HB-EGF subsequently binds to and activates its receptors: EGF receptor/erbB1/HER1 (epidermal growth factor receptor (EGFR)) or erbB4/HER4. Heterodimerization or homodimerization of these receptors following s-HB-EGF's binding drives signal-transduction cascades, which have critical roles in diverse cell fates, including development, proliferation, differentiation and migration. Meanwhile, the C-terminal domain of HB-EGF following shedding translocates to the nucleus and modulates cell cycle and cell proliferation via regulating gene transcription, including cyclin A and D (Nishi and Klagsbrun, 2004).
Ectodomain shedding of pro-HB-EGF is a critical step for HB-EGF activity. A number of pharmacological and physiological stimuli induce ectodomain shedding, such as G-protein-coupled receptor ligands (Prenzel et al., 1999), ultraviolet light (Takenobu et al., 2003) and osmotic pressure (Takenobu et al., 2003), etc., indicating that the shedding is a regulated process. Although the role of HB-EGF shedding in physiological processes has been increasingly reported (Asakura et al., 2002; Xu et al., 2004; Higashiyama and Nanba, 2005), the regulation of HB-EGF shedding in relation to chemotherapy resistance has not been fully elucidated.
HB-EGF expression has been implicated in tumor progression due to its overexpression in many tumors, including hepatocellular carcinoma (Ito et al., 2001b), colon cancer (Itoh et al., 2005), pancreatic cancer (Ito et al., 2001a) and bladder malignancy (Ongusaha et al., 2004). Recently, Miyamoto et al. (2004) reported that tumor formation of ovarian cancer cells in nude mice was enhanced by exogenous expression of HB-EGF and completely blocked by HB-EGF gene RNA interference or by CRM197, a specific HB-EGF inhibitor. Moreover, the lysophosphatidic acid-induced ectodomain shedding of HB-EGF is critical to tumor formation in ovarian cancer. Fang et al. (2001) also demonstrated that HB-EGF is induced in response to tumor suppressor p53 as well as methyl methanesulfonate-induced DNA damage, and the induction of HB-EGF antagonizes apoptosis mediated by genotoxic stress through the activation of the Ras/Raf cascade and the AKT pathway. Suganuma et al. (2003) recently identified HB-EGF as a chemoresistance-related gene in gastric cancer by use of cDNA microarray. They observed that HB-EGF was highly expressed in 5-fluorouracil (5-FU) and cisplatin (DDP)–chemoresistance groups, but they suggested that further work was needed to investigate the mechanism of chemotherapy resistance with the overexpression of HB-EGF. These findings demonstrated that HB-EGF is not only a potent inducer of tumor growth but also a survival factor in response to cellular stress. Therefore, the therapeutic intervention aimed at inhibiting HB-EGF activity may be useful in cancer prevention and treatments.
In this study, we further investigated the potential role of HB-EGF in chemotherapy resistance. We demonstrated that the upregulation of HB-EGF following chemotherapy is regulated by activator protein-1 (AP-1) and NF-κB. Chemotherapy also triggers ectodomain shedding of HB-EGF, resulting in the activation of EGFR and resistance to chemotherapy.
HB-EGF expression is elevated in response to chemotherapy treatments
Using microarray-based differential gene expression assay, we identified approximately 80 genes that might be involved in NF-κB-regulated chemotherapy resistance. These genes were classified by two conditions in LOVO cells: induction of expression by exposure to SN38 (the active metabolite of CPT-11 and a potent inducer of NF-κB) and suppression of expression by PS1145 (a selective I kappa B kinase (IKK)β inhibitor). We observed a sevenfold increase of HB-EGF expression in SN38 treatment and a twofold increase in PS1145/SN38 co-treatments, indicating that HB-EGF might be involved in NF-κB-regulated chemotherapy response (data not shown).
To validate the microarray data, we performed real-time reverse transcriptase–polymerase chain reaction (RT–PCR) analysis to quantify the level of HB-EGF induction. As shown in Figure 1a, LOVO, pancreatic cancer cell (PANC1) and HeLa cells were treated with SN38, doxorubicin and cisplatin for 2 h, respectively. We observed the chemotherapy-induced HB-EGF expression in all three cell lines. We also detected that the induction of HB-EGF following the SN38 treatment was dose- and time-dependent as shown in Figure 1b and c. Higher dosages and longer exposure time of chemotherapy both augmented HB-EGF induction in LOVO and PANC1 cells.
Elevation of HB-EGF expression following chemotherapy is regulated by NF-κB
Consistent with the microarray data that PS1145 pretreatment suppressed HB-EGF induction in LOVO cells, we observed the similar effects of PS1145 in PANC1 and HeLa cells. Using real-time PCR, we observed a 20–30% suppression of HB-EGF induction in response to PS1145/SN38 treatments compared to SN38 treatment alone as shown in Figure 2a. Because PS1145 is a potent inhibitor of NF-κB activity (Yemelyanov et al., 2006), these data suggest that HB-EGF induction by SN38 treatment may be regulated by NF-κB.
To test this possibility, we transiently transfected PANC1 and HeLa cells with small interfering RNA (siRNA) against p65 or scramble siRNA, and the effect of suppression was assessed by Western blot analysis (Figure 2b). p65 is v-rel reticuloendotheliosis viral oncogene homolog A (Rel A), which is one of the major components to form the NF-κB complex. The p50 (NF-κB1)/p65 (RelA) heterodimer is the most abundant form of NF-κB. siRNA cells were then treated with vehicle or SN38, and the induction of HB-EGF was quantitated by real-time RT–PCR. As shown in Figure 2c, inhibition of p65 suppresses HB-EGF induction by 20–30% in response to SN38 compared to control siRNA cells receiving the same treatments.
Induction of HB-EGF expression following chemotherapy is also regulated by AP-1
Although we demonstrated here that NF-κB regulates HB-EGF induction, neither the pretreatment of PS1145 nor the siRNA against p65 showed complete suppression of HB-EGF induction, which led us to suspect that other factors may contribute to chemotherapy-induced HB-EGF expression.
We have known that SN38 treatment not only activates NF-κB but also potently induces AP-1 activity in gastric cancer AGS cells (Kishida et al., 2005). AP-1 is a heterodimer formed by c-jun and c-fos and has been shown to have consensus binding sequence on HB-EGF promoter (Park et al., 1999; Kanda and Watanabe, 2005). Strong evidence also indicates that AP-1 plays a key role in cancer development and is upregulated during tumor progression (Bode and Dong, 2004). Therefore, we hypothesized that HB-EGF induction may also be AP-1 regulated.
To test this, we first demonstrated that AP-1 activity is indeed induced by SN38 in our system. PANC1 and HeLa cells were treated with SN38 and the induction of c-fos gene was assessed by use of real-time PCR. As shown in Figure 3a, the transcription activity of c-fos gene was induced by SN38, suggesting SN38 can upregulate AP-1 activity. We then utilized the siRNA to suppress c-fos expression and the efficacy of suppression was assessed by Western blot analysis (Figure 3b). siRNA cells were then treated with vehicle or SN38, and the induction of HB-EGF was quantitated by real-time RT–PCR. As shown in Figure 3c, inhibition of c-fos markedly suppresses the induction of HB-EGF by 50–60% following SN38 treatment.
HB-EGF expression rescues c-fos/p65 siRNA cells from SN38-induced cell death
We performed combined siRNA against c-fos and p65 at the same time and observed even stronger inhibition of HB-EGF induction following SN38 treatment (70–80% inhibition) compared to single siRNA alone, indicating the additive roles of c-fos and p65 in the regulation of HB-EGF (Figure 4a).
Meanwhile, we observed that the combined c-fos/p65 siRNA had dramatic negative effects on cell viability in response to SN38 treatment, which raised the question of whether HB-EGF expression could rescue c-fos/p65 siRNA cells from SN38-induced cell death. To test this, we generated HB-EGF stable expressing PANC1 cells and vector only cells. Cells were transfected with scramble RNA or combined siRNA against c-fos/p65, followed by SN38 treatments. Apoptosis was measured by poly(ADP-ribose)polymerase (PARP) cleavage (Figure 4b) and Cell Death Elisa (Figure 4c). Here, we demonstrated that combined knockdown endogenous c-fos/p65 had significant negative impact on cell survival when challenged with SN38, and exogenous HB-EGF expression markedly rescued siRNA cells from apoptosis.
Chemotherapy treatments induce the ectodomain shedding of HB-EGF
To examine the effect of chemotherapy agents on the induction of pro-HB-EGF ectodomain shedding, we used HB-EGF stable expressing PANC1 and HeLa cells, PANC1/HB-EGF-vesicular stomatitis virus (VSV), HeLa/HB-EGF-VSV, control PANC1/pcDNA3.1 and HeLa/pcDNA3.1 cells. The cleavage of pro-HB-EGF was detected by Western blotting analysis. In cell lysate, bands ranging from 20 to 30 kDa correspond to pro-HB-EGF, whereas bands between 10 and 15 kDa represent the proteolytic C-terminal fragments, referred to as the ‘tail fragment’ (Takenobu et al., 2003). To minimize the ectodomain shedding induced by serum-containing factors, cells were cultured in serum-free medium for at least 48 h before treated with SN38, 5-FU and doxorubicin, respectively. As shown in Figure 5a and b, SN38 and doxorubicin treatments induced pro-HB-EGF cleavage to generate more C-terminal fragments, and the appearance of tail fragments in cell lysate was inhibited in the presence of GM6001. GM6001 is a potent inhibitor of matrix metalloproteases (MMPs) and it has been shown to inhibit HB-EGF shedding activity (Yang et al., 2005; Armant et al., 2006). Although the mechanism by which chemotherapy agents increase MMPs activity remains unknown, the finding that inhibition of MMPs abrogated chemotherapy-induced supports the idea of utilizing MMP inhibitor to regulate HB-EGF activity.
Meanwhile, we also measured the concentration change of s-HB-EGF released into the culture medium following chemotherapy. As shown in Figure 5c, the accumulated s-HB-EGF in culture medium was increased by 20–50% after 3 h exposure to different chemotherapy agents in both PANC1 and HeLa cells. 5-FU appears to have different effects on the shedding of HB-EGF in different cell lines, as we did not observe activated shedding in PANC1 cells following 5-FU treatment.
HB-EGF mediates chemotherapy-induced EGFR activation
EGFR signaling pathway has been well established as a tumorigenetic pathway that promotes cellular proliferation, angiogenesis and inhibition of apoptosis. However, little has been done in terms of investigating the EGFR activation in response to chemotherapy. Kishida et al. (2005) have shown that CPT-11 enhanced EGF signaling and addressed the potential benefit of the combination therapy of CPT-11 and EGFR inhibitor in certain gastric cancers. Based on our findings, we hypothesized that chemotherapy-induced HB-EGF could activate EGFR signaling cascade and may contribute to chemoresistance. To test this possibility, we treated PANC1 cells with SN38 and doxorubicin and observed the increased EGFR phosphorylation in both drug treatments (Figure 6a). When PANC1 cells were pretreated with HB-EGF inhibitor, CRM197, and followed by chemotherapy, the phosphorylation of EGFR was suppressed (Figure 6b), indicating that chemotherapy-induced EGFR phosphorylation is HB-EGF dependent.
We also exposed PANC1/HB-EGF-expressing cells to SN38 and doxorubicin, and found out that chemotherapy-induced EGFR phosphorylation is dramatically increased in HB-EGF group compared to vector alone groups receiving the same treatments (Figure 6c). Suppression HB-EGF activity by transiently transfected siRNA against HB-EGF almost completely abolished SN38- and doxorubicin-induced EGFR phosphorylation (Figure 6d and e). These data suggest the critical role of HB-EGF in chemotherapy-induced EGFR activation.
Overexpression of HB-EGF protects cells from chemotherapy-induced cell death
To address the key question of whether overexpression of HB-EGF leads to chemotherapy resistance, we measured SN38-induced apoptosis using flow cytometry. As shown in Figure 7, when challenged with SN38, HB-EGF expression significantly protected cells from SN38-induced apoptosis (9% Annexin-positive cells) compared to vector alone group (18% Annexin-positive cells). Meanwhile, suppression of HB-EGF expression by use of siRNA resulted in a dramatic increase in apoptotic rate (26% Annexin-positive cells) following the same treatment. These results point to a significant protective role of HB-EGF in cell viability and in chemotherapy protection.
We and others have shown that NF-κB activation suppresses the apoptotic potential of chemotherapeutic agents and contributes to drug resistance (Wang et al., 1999; Nakanishi and Toi, 2005). However, the downstream mediators of NF-κB that are involved in the mechanism of chemotherapy resistance remain poorly understood. For this reason, we attempted to identify NF-κB-regulated chemotherapy resistance gene(s) using microarray. Somewhat surprisingly, our studies of colon cancer cells did not find previous known antiapoptotic genes to be either induced by SN38 or suppressed by selective inhibitors of NF-κB. Those genes include c inhibitor of apoptosis-1 (cIAP-1), cIAP-2, X-chromosome-linked inhibitor of apoptosis protein, tumor necrosis factor receptor-associated factor (TRAF)-1 and TRAF-2, etc. (data not shown). These results suggested that other genes might be involved in the NF-κB-regulated antiapoptotic response.
In this study, we have provided various lines of evidences that HB-EGF is a chemotherapy response gene. HB-EGF expression activates the EGFR signaling pathway and contributes to chemotherapy resistance.
First, we observed different levels of induction by chemotherapy treatments in LOVO, PANC1 and HeLa cells, and the induction of HB-EGF is time- and dose dependent. Using siRNA and PS1145, we found out that the induction of HB-EGF following chemotherapy is regulated by NF-κB. However, neither PS1145 nor siRNA against p65 showed complete suppression of HB-EGF following chemotherapy, suggesting that other factor(s) may regulate HB-EGF expression in response to chemotherapy. Here, we demonstrated that not only NF-κB but also AP-1 was rapidly induced by SN38. Using siRNA against c-fos gene, we found that the induction of HB-EGF following chemotherapy was markedly inhibited, and even the basal level of HB-EGF was suppressed by c-fos siRNA. Combined knockdown c-fos/p65 markedly blocked the inductions of HB-EGF in response to SN38 compared to single c-fos siRNA or p65 siRNA alone, suggesting an additive effect of AP-1 and NF-κB activity. Meanwhile, we also observed that HB-EGF expression rescued c-fos/p65 siRNA cells from SN38-induced apoptosis, indicating that HB-EGF is a critical downstream effector of AP-1 and NF-κB and its protective role in response to chemotherapy.
Ectodomain shedding of HB-EGF has emerged as a critical step in the functional activation of EGFR in the inter-receptor crosstalk. Despite the extensive studies of the mechanistic aspects of HB-EGF shedding in wound healing (Xu et al., 2004; Higashiyama and Nanba, 2005) and cardiac hypertrophy (Asakura et al., 2002), the mechanism(s) underlying the chemotherapy-induced HB-EGF shedding and its relationship with chemotherapy resistance have remained elusive. Our data indicate that the ectodomain shedding of HB-EGF is regulated by chemotherapy exposure. The elevated shedding of HB-EGF was observed following 3 h treatment of SN38 or doxorubicin. Usually, serum starvation for 24 h could eliminate the shedding activity of most metalloproteases (Hirata et al., 2001), but we observed very strong shedding activity of HB-EGF in PANC1 and HeLa cells even after serum starvation for 72 h (data not shown). This may partly explain the potent mitogenic potential of HB-EGF and the high resistance to cytotoxicity following chemotherapy, as cells constantly produce the s-HB-EGF ligand to activate the survival signaling cascade.
After generating HB-EGF-expressing clones, we found out that the capacity of tumor cells to maintain active HB-EGF signaling not only functions as a mitogenic force to promote cell proliferation but also represents a potential mechanism of resistance by which tumor cells escape chemotherapy-induced genotoxic stress. Overexpression of HB-EGF dramatically decreased the susceptibility of tumor cells to chemotherapy treatments and suppression of HB-EGF by siRNA led to significantly increased sensitivity to chemotherapy.
Another important aspect of our study is that we demonstrated that chemotherapy-induced EGFR activation is mediated by HB-EGF. Induction of HB-EGF significantly enhances chemotherapy-induced EGFR activation and inhibition of HB-EGF by CRM197 or siRNA suppresses EGFR activation following chemotherapy treatments. A previous study has suggested that EGFR activation by SN38 causes the activation of NF-κB and AP-1 and the activated EGFR pathway by SN38 may also induce HB-EGF indirectly (Kishida et al., 2005). In our study, we suggest another possible chemotherapy-triggered EGFR activation pathway (Figure 8). Namely, chemotherapy agents activate NF-κB and AP-1 (the specific mechanism of activation is unknown), which leads to the transcription activation and the increased ectodomain shedding activity of HB-EGF. These two effects triggered by chemotherapy synergistically activate EGFR pathway and lead to the inhibition of apoptosis.
In summary, our findings provide evidence of a novel reaction of chemotherapy-induced HB-EGF activity. This response may contribute to the survival mechanism by which HB-EGF protects cells from chemotherapy-induced cell death. Thus, selective targeting HB-EGF may yield significant improvement in chemotherapy treatment.
Materials and methods
Polyclonal antibodies against HB-EGF, phospho-EGFR and EGFR were purchased from Santa Cruz Biotech (Santa Cruz, CA, USA). SignalSilence NF-κB p65 siRNA Kit was purchased from Cell Signaling (Beverly, MA, USA). HB-EGF inhibitor (diphtheria toxin, CRM197) was purchased from EMD Biosciences Inc. (San Diego, CA, USA). Cisplatin, 5-FU, doxorubicin and anti-VSV antibody were purchased from Sigma-Aldrich (Saint Louis, MO, USA). SN38 was provided by Pharmacia and Upjohn Company (Kalamazoo, MI, USA). PS1145 (a selective IKKβ inhibitor) was provided by Millennium Pharmaceutical Inc. (Cambridge, MA, USA). Smartpool siRNA kits against c-fos, HB-EGF and scramble siRNA control were purchased from Dharmacon Inc. (Chicago, IL, USA). HB-EGF-VSV construct (pcDNA3-HB-EGF-VSV) was generously provided by Dr Axel Ullrich (Max-Planck-Institute for Biochemistry, Germany). Cell Death Detection ELISA Plus kit was purchased from Roche (Indianapolis, IN, USA).
Cell culture and chemotherapy treatments
Human LOVO (colorectal cancer cell), PANC1 (pancreatic cancer cell) and HeLa (cervical cancer cell) cells were purchased from the American Type Culture Collection (ATCC) (Manassas, VA, USA) and cultured according to ATCC recommendation. For chemotherapy treatments, cells were seeded in six-well plates at a density of 5 × 104 cells/well and treated with PS-1145 at the concentration of 20 μM for 4 h. The medium was refreshed, and cells were treated with SN38 (2.5 μg/ml), doxorubicin (6 μg/ml) and cisplatin (10 μg/ml) for indicated time.
Expression of HB-EGF and c-fos genes was measured by real-time PCR. Taqman-labeled specific probes of HB-EGF and c-fos were purchased from Applied Biosystem (Foster City, CA, USA). The reaction of RT–PCR was performed using TaqMan Gene Expression Assays on Applied Biosystems 7300 Real-Time PCR Systems. Human endogenous GAPDH gene was used as control group.
p65, c-fos and HB-EGF gene siRNA treatment
Cells were grown to 80% confluence and p65 siRNA, c-fos siRNA, HB-EGF siRNA or scramble control siRNA were transfected into PANC1 or HeLa cells using the TransIT-TKO (Mirus, Madison, WI, USA) according to the manufacturer's instructions. Briefly, TransIT-TKO reagent was incubated with serum-free medium for 5 min. Subsequently, a mixture of respective siRNA was added. After incubation for 15 min at room temperature, the mixture was diluted with medium and added to each well. The final concentration of each siRNA in transfection was 25 nM. After incubation for 48 h, cells were washed, resuspended in new culture media and treated accordingly.
HB-EGF enzyme-linked immunosorbent assay (ELISA)
Cells were cultured in serum-free media for 48 h before treated with SN38 (2.5 μg/ml), doxorubicin (6 μg/ml) and 5-FU (10 μg/ml), respectively for 3 h. Growth media were collected by centrifugation at 5000 g for 10 min. Enzyme-linked immunosorbent assay (ELISA) to detect the s-HB-EGF in media was performed as described (Armant et al., 2006). Briefly, conditioned medium were serially diluted with phosphate-buffered saline (PBS)/0.1% BSA and then incubated in anti-HB-EGF- (R&D System, Minneapolis, MN, USA) coated microtiter plate for 2 h at 37°C. After washing three times with PBS, biotin-labeled anti-HB-EGF antibody was added and incubated at room temperature for 1 h. After washing three times with PBS, streptavidin-conjugated horseradish peroxidase (1:100, Santa Cruz Biotech) was added to each well and incubated for 1 h. The wells were washed and peroxidase activity was detected using 3,3′,5,5′-tetramethyl benzidine (TMB)/E solution (Chemicon, Temecula, CA, USA). Oxidation of TMB produces a blue reaction product that is immediately measured at 450 nm using a programmable multiplate spectrophotometer. Experiments were repeated three times and all samples were measured in triplicates.
PANC1/pcDNA3.1, PANC1/HB-EGF and PANC1/si-HB-EGF cells grew in serum-free media before treated with SN38 (2.5 μg/ml) for 18 h. Cells were then collected and stained with Annexin V and propidium iodide using TACS™ Annexin V-FITC kit (R&D Systems Inc., Minneapolis, MN, USA). Flow cytometry was performed and analysed on a Becton Dickison FACScalibur flow cytometer (BD, Franklin Lakes, NJ, USA).
Western blot analysis
Whole-cell lysate (50 μg) were electrophoresis through 8, 10 or 15% denaturing polyacrylamide slab gels and transferred to a polyvinylidene difluoride membrane using semi-dry electro blot (Bio-Rad, Hercules, CA, USA). The blot was probed with the primary antibodies and then antibody binding was detected using the ECL detection system (Amersham Biosciences, Piscataway, NJ, USA) according to the manufacturer's protocol.
This study was supported by NIH grant support CA77278-01A1 and CA98871-01 (J Cusack).