|Home | About | Journals | Submit | Contact Us | Français|
Constitutive classical NFκB activation has been implicated in the development of pancreatic cancer, and inhibition of classical NFκB signaling sensitizes pancreatic cancer cells to apoptosis. However, the role of the more recently described non-canonical NFκB pathway has not been specifically addressed in pancreatic cancer. The non-canonical pathway requires stabilization of NIK and IKKα-dependent phosphorylation and processing of NFκB2/p100 to p52. This leads to the activation of p52-RelB heterodimers that regulate genes encoding lymphoid-specific chemokines and cytokines. We performed qRT-PCR to detect gene expression in a panel of pancreatic ductal adenocarcinoma cell lines (BxPC-3, PCA-2, PANC-1, Capan-1, Hs-766T, AsPC-1, MiaPACA-2) and found only modest elevation of classical NFκB-dependent genes. In contrast, each of the tumor cell lines displayed dramatically elevated levels of subsets of the non-canonical NFκB target genes CCL19, CCL21, CXCL12, CXCL13 and BAFF. Consistent with activation of the non-canonical pathway, p52 and RelB co-localized in adenocarcinoma cells in sections of pancreatic tumor tissue, and each of the tumor cell lines displayed elevated p52 levels. Furthermore, p52 and RelB co-immunoprecipitated from pancreatic cancer cells and immunoblotting revealed that NIK was stabilized and p100 was constitutively phosphorylated in a subset of the cell lines. Finally, stable overexpression of dominant negative IKKα significantly inhibited non-canonical target gene expression in BxPC-3 cells. These findings therefore demonstrate that the non-canonical NFκB pathway is constitutively active and functional in pancreatic cancer cells.
Dysregulated activation of NFκB plays a major role in tumor development through maintained expression of target genes that regulate cell growth, proliferation and survival.1,2 Constitutive NFκB activity has been described in many distinct tumor types including leukemias, lymphomas and solid tumors.1,2 Consequently, selective inhibition of aberrant NFκB signaling is being actively explored for its potential therapeutic efficacy.3-5 The mechanisms underlying NFκB activation in tumor cells are poorly understood but those that have been described include mutations that activate upstream signaling components and secretion of autocrine factors.1,2 Understanding the precise mechanisms regulating aberrant NFκB activation in specific tumor cells and the functional consequences of this activation will assist the development of novel tumor-specific pharmacological strategies.
Constitutive NFκB activity regulating tumor growth, development and survival has been extensively studied in the context of dysregulated activation of the classical NFκB pathway.1,2 The classical pathway is activated by most NFκB-inducing stimuli including innate and adaptive immune receptors and pro-inflammatory cytokines.6,7 Upon activation by these stimuli, the inhibitor of NFκB (IκB) kinase (IKK) complex consisting of IKKα, IKKβ and NEMO/IKKγ phosphorylates the IκB proteins leading to their subsequent ubiquitination and proteasomal degradation.7,8 IκBs normally sequester NFκB homo- or heterodimeric complexes in the cytoplasm and IκB degradation releases these complexes to migrate to the nucleus and regulate gene expression. The prototypical NFκB complex activated by the classical pathway is heterodimer of the p65/RelA and p50 NFκB proteins and both of these NFκB subunits have been extensively described in the nucleus of tumor cells in vivo and in many distinct tumor cell lines.1,2 The classical pathway regulates expression of the majority of NFκB-dependent genes, many of which have been strongly implicated in tumorigenesis. These include genes encoding anti-apoptotic and survival proteins, pro-inflammatory cytokines, chemokines and growth factors, and inhibition of classical NFκB activity in tumor cells has been extensively demonstrated to block expression of these genes.1-5
In contrast to the classical mechanism, the more recently described non-canonical (NC) NFκB pathway has not been extensively studied in the context of cancer biology.6 The NC pathway is normally activated by ligation of only a subset of receptors that includes the lymphotoxin-β receptor (LTβR), CD40 and BAFF-R.9-13 Activation of these receptor complexes leads to the degradation of the inhibitory protein TRAF3, followed by stabilization of NFκB-inducing kinase (NIK) and specific activation of IKKα.14-16 In this regard, the NC pathway is activated in the absence of IKKβ and NEMO, which are critical for induction of the classical pathway.9-13 NIK-dependent IKKα activation mediates specific phosphorylation, ubiquitination and processing of the NFκB family member p100/NFκB2 to generate p52.13,14,16 The resulting p52 forms a heterodimer with RelB, and these RelB-p52 NFκB complexes translocate to the nucleus to regulate transcription of specific NC NFκB target genes.9-13
Although little is known about the activation status of the NC pathway in cancer, recent elegant genetic studies have demonstrated frequent mutations in key upstream signaling proteins in multiple myeloma cells.17,18 These include mutations of TRAF3 and NIK that result in upregulated p100 processing to p52 and nuclear translocation of both classical and NC NFκB heterodimers.17,18 Furthermore, in studies broadly examining the expression of NFκB family proteins, elevated levels of nuclear p52 and RelB have been reported in primary tumor tissues including prostate cancer and oral cancers, as well as in tumor cell lines including prostate and pancreatic.19-25 Additionally, Xu et al.25 recently found that blocking p52 and RelB import in prostate cancer cells sensitized them to destruction by radiation. These findings therefore suggest that in addition to the classical pathway, aberrant activation of NC in cancer cells may regulate pro-tumorigenic gene expression.
The major functions of NC NFκB (i.e., p52-RelB) are in regulating lymphoid organogenesis and B-cell maturation, and mice lacking components of the NC pathway (i.e., NIK, IKKα, p100 or RelB) exhibit defects in these developmental processes.6,7,9,13 Consistent with the phenotype of these animals, the small number of bona fide NC NFκB-dependent genes includes the B-cell regulatory cytokine BAFF and the homeostatic and lymphoid chemokines CXCL12, CXCL13, CCL19 and CCL21, all of which function during lymphoid organogenesis.6,7,9,11,13 Intriguingly, several of these genes have been implicated in tumor survival. BAFF and CXCL13 are frequently overexpressed in B-cell lymphomas, where they promote survival of malignant cells, and CXCL12 overexpression has been described in numerous solid tumors, where it may affect angiogenesis or antitumor lymphocyte trafficking.26-32 Understanding the activation status of the NC NFκB pathway in solid tumors will therefore provide valuable insight into the mechanisms that regulate tumor promoting gene expression.
To address this question, we examined NC NFκB-dependent gene expression in a panel of pancreatic adenocarcinoma cell lines. Pancreatic cancer has the lowest survival rate of any tumor, and constitutive classical NFκB activation has been widely reported in pancreatic cancer (PC) cell lines and tumor tissue.33-42 We show here that NC NFκB target gene expression was significantly enhanced in all seven of the cell lines we examine. Furthermore, p52 and RelB co-localized in the nuclei of pancreatic adenocarcinoma cells in sections of resected tumors and these same NC NFκB complexes also co-immunoprecipitated from PC cell lines. The PC cells showed distinct hallmarks of constitutive NC NFκB activation including enhanced levels of p52 and specific phosphorylation of p100. Remarkably, a subset of the PC cells also displayed stabilized NIK protein expression. Finally, we found that selectively targeting the NC pathway using retrovirally transduced dominant negative IKKα significantly reduced NC NFκB dependent gene expression in one of the PC lines. These findings therefore demonstrate that the NC NFκB pathway is constitutively active in pancreatic adenocarcinoma cells providing a potentially novel target for the development of drugs aimed at disrupting tumor growth and survival.
To determine the levels of NC NFκB-dependent gene expression in pancreatic cancer (PC) cells we performed qRT-PCR analysis using RNA isolated from BxPC-3 adenocarcinoma cells. We compared expression of both classical and NC NFκB target genes in BxPC-3 cells with levels observed in HeLa epithelial tumor cells that exhibit low basal NFκB activity.46 Expression of the classical NFκB-dependent genes CXCL2 and IL-6 was not significantly different from the levels observed in HeLa cells, and COX-2 expression was only elevated two-fold (Fig. 1A). In contrast, BxPC-3 cells exhibited significantly elevated levels of the NC NFκB-dependent genes CCL19, CCL21 and CXCL12 together with a modest increase (three-fold) in BAFF expression compared with HeLa cells (Fig. 1A).
We next measured expression of the same panel of genes in PCA-2 cells, which are a recently derived low-passage pancreatic adenocarcinoma cell line. Similar to BxPC-3 cells, expression of CXCL2, COX-2 and IL-6 in PCA-2 cells was not significantly elevated compared with untreated HeLa (Fig. 1B). However, expression of four NC NFκB-dependent genes was dramatically elevated with the levels of CCL19, CXCL12, CXCL13 and BAFF being approximately forty-fold higher than levels in HeLa cells (Fig. 1B).
These findings suggest that NC NFκB-dependent genes are constitutively elevated in pancreatic adenocarcinoma cells. To test this further, we extended our gene expression analysis to a wider panel of PC lines consisting of PANC-1, MiaPACA-2, AsPC-1, Hs766T and Capan-1 in addition to BxPC-3 and PCA-2. To provide a positive control for classical NFκB-dependent gene expression, we treated HeLa cells with TNF for 4 h; as expected, this led to increased expression of all three classical genes (Fig. 1C). We also detected a modest increase in the level of CCL19 in TNF-stimulated HeLa cells suggesting possible cross-talk between the classical and NC NFκB pathways; however, none of the other NC genes were upregulated by TNF (Fig. 1C). Recent studies have demonstrated that mutations or deletion of critical upstream signaling components leads to constitutive NC pathway activity in multiple myeloma (MM) cells although NC NFκB-dependent gene expression in these cells has not yet been determined.17,18 We therefore included the human MM cell line RPMI-8226 in which the NC signaling pathway is active,18 as a positive control in our analysis. As shown in Figure 1C, RPMI-8226 exhibited elevated expression of CCL19, CXCL12 and CXCL13 whereas none of the classical NFκB-dependent genes we tested were increased compared with untreated HeLa.
Consistent with our findings with BxPC-3 and PCA-2 cells, the PC cell panel exhibited only minor changes the classical NFκB dependent genes with the only significant increases observed in CXCL2 expression by PANC-1, MiaPACA-2 and Capan-1 cells (Fig. 1C). None of the classical genes in the PC cell lines were elevated to the levels observed in TNF-stimulated HeLa cells. In contrast, like the RPMI-8226 cells, all of the PC cell lines displayed significantly enhanced expression of NC NFκB target genes compared with HeLa cells (Fig. 1C). CCL19 and CXCL12 were elevated in all PC cell lines tested and CXCL13 and BAFF were each upregulated over basal HeLa levels in five out of the seven lines. Only BxPC-3 cells displayed increased levels of CCL21 suggesting that this gene is differentially regulated in BxPC-3 cells compared with the other cell lines. Taken together, the findings presented in Figure 1 demonstrate that NC NFκB-dependent gene expression is significantly elevated in a panel of seven pancreatic adenocarcinoma cell lines.
To determine whether the NC NFκB complex consisting of p52 and RelB was present in human tumors, we stained paraffin-embedded tissues representing malignant pancreas for the presence of p52 and RelB. We immunohistochemically (IHC) stained 15 specimens using anti-p100/p52 and anti-RelB and found positive staining for p100/p52 and RelB in the malignant tissues. The results in Figure 2 show the staining profiles of three separate specimens that were determined histologically to be adenocarcinoma. As shown in Figure 2A, strong positive staining representing the presence of both p100/p52 and RelB was clearly identified in two of the adenocarcinoma samples by one-antibody IHC. We observed strong cytoplasmic staining for both proteins and in each field of view, numerous nuclei also stained brown, demonstrating that p100/p52 and RelB were localized in the nucleus in these malignant cells.
To determine whether p100/p52 and RelB co-localize within the same cell, we performed two-antibody immunofluorescence on sections of the adenocarcinoma specimens. As shown in Figure 2B, adenocarcinoma cells stained positively for both p100/p52 and RelB. Merging of the fluorescent dyes revealed that p100/p52 (green) and RelB (red) are frequently co-localized as indicated by a yellow merge color. Furthermore, merging with a DAPI-stain (blue), demonstrated that both proteins co-localize in the nucleus in some of the cells as indicated by a white merge color. The findings in Figure 2 therefore demonstrate that RelB and p100/p52 are present and co-localize in specimens of human pancreatic adenocarcinoma. Moreover, both proteins are frequently detected in the nuclei of pancreatic adenocarcinoma cells suggesting that the NC NFκB pathway is active in these cells.
Signal induced processing of p100 to generate p52 is a hallmark of the NC NFκB pathway.6,7,9-13 In normal resting cells, the levels of p100 exceed the amount of p52 and it is only upon activation of the NC pathway that p52 levels are increased.45 To determine levels of p100 processing in non-malignant cells we stimulated HUVEC with LIGHT for 24 h and as we previously described,45 this induced the appearance of a p52 band (Fig. 3A; compare lanes 1 and 2). Notably, we found that HeLa cells displayed levels of p52 similar to that observed in stimulated HUVEC (lane 3) suggesting that basal p100 processing is upregulated in these cells. Consistent with a previous study,18 p52 levels were also elevated in the RPMI-8226 cells in which NC signaling is constitutively activated (lane 4). In light of our findings that NC NFκB target genes are upregulated in the panel of pancreatic cancer cells (Fig. 1), and that p52 and RelB co-localize in pancreatic tumor tissue sections (Fig. 2), we questioned whether p52 levels were elevated in the PC lines and whether these cells contained detectable p52-RelB NFκB complexes.
As shown in Figure 3B, we detected elevated p52 in PANC-1, BxPC-3, Hs766T and PCA-2 cells (lanes 3–6), which were the PC lines displaying the highest levels of NC target gene expression (Fig. 1). Similar elevated p52 levels were also detected in AsPC-1, Capan-1 and MiaPACA-2 cells (data not shown). To determine whether these NFκB proteins were localized in the nuclei we immunoblotted nuclear extracts from HeLa, BxPC-3 and Hs766T cells. To ensure the integrity of the nuclear extracts, we immunoblotted using anti-PARP and this consistently revealed two bands in HeLa but not BxPC-3 or Hs766T nuclei. Although this suggests that PARP is cleaved in HeLa cells, we did not observe any enhanced apoptosis or growth defects in these cultures. Immunoblotting nuclear extracts using anti-p52 and anti-RelB demonstrated enhanced nuclear localization of both NFκB proteins in BxPC-3 and Hs766T cells compared with HeLa cells, which also displayed detectable levels of nuclear RelB and p52 (Fig. 3C).
To determine the nature of the nuclear NFκB complexes in the cell lines, we performed immunoprecipitations using anti-p52. Immunoprecipitation of NFκB complexes from BxPC-3 and Hs766T resulted in co-immunoprecipitation of RelB, but this was not the case in HeLa lysates (Fig. 3D). Similar co-immunoprecipitation of p52 and RelB was observed in the other PC cells in our panel (data not shown). Hence, although HeLa cells exhibit high basal p52 levels, and nuclear p52 and RelB are present in these cells, the non-canonical p52-RelB NFκB complex is constitutively present only in the nuclei of the PC cells.
We next questioned whether key signaling events in the non-canonical pathway are activated in the pancreatic cancer cell lines. One of the earliest steps in signal-induced NC pathway activation is degradation of TRAF3 that in turn leads to NIK stabilization.14-16 However, immunoblotting revealed similar levels of TRAF3 in all of the pancreatic cancer cells to that observed in HeLa cells (Fig. 4A). Surprisingly, we also detected abundant TRAF3 in RPMI-8226 cells that have been shown previously to lack TRAF3.18 Despite this apparent lack of TRAF3 degradation, we detected elevated levels of NIK protein expression in BxPC-3, Hs766T and PCA-2 cells but not in PANC-1, RPMI-8226 or HeLa cells (Fig. 4A). Densitometric analysis of immunoblots from multiple experiments encompassing the entire PC cell panel demonstrated that NIK expression was elevated at least four-fold over the basal level in HeLa cells in four of the seven PC cell lines tested (Fig. 4B).
Signal-induced p100 processing in the NC NFκB pathway is triggered by IKKα-dependent site-specific phosphorylation of p100.13 To determine whether p100 phosphorylation occurs in pancreatic cancer cells we immunoblotted lysates of BxPC-3 and Hs766T cells using anti-phospho-p100. This antibody detects phosphorylation at Serines 866 and 870 in p100, which are required for its subsequent ubiquitination and processing to p52. As shown in Figure 4C, phosphorylated p100 was present in both of the PC cell lines but not in HeLa cells. Taken together, the findings in Figure 4 demonstrate that critical upstream signaling events of non-canonical NFκB activation can be detected in pancreatic cancer cells.
We previously demonstrated that retroviral transduction of dominant negative IKKα into human endothelial cells selectively inhibits the non-canonical NFκB pathway.45 We therefore employed a similar approach to determine the effects of disrupting the NC pathway on gene expression in pancreatic cancer cells. BxPC-3 cells were retrovirally transduced with dominant negative IKKα harboring two serine to alanine mutations within the activation loop (IKKαSSAA). Parallel transduction with a GFP-expressing control vector demonstrated that approximately 60% of the cells were transduced (Fig. 5A) and immunoblotting confirmed elevated levels of IKKα expression in the IKKαSSAA-transduced cells (Fig. 5B). As shown in Figure 5C, IKKαSSAA did not affect expression CCL21 but significantly reduced the constitutive expression of CXCL12, CXCL13 and BAFF by BxPC-3 cells.
Extensive evidence supports a key role for constitutive classical NFκB activity in the development of pancreatic cancer.19,33-42,47,48 In this regard, both p50 and p65 have been shown to be present in the nucleus of many of the PC cell lines we have used in this study and we have confirmed these findings (data not shown). Furthermore, nuclear localization of these classical NFκB subunits occurs in adenocarcinoma cells in human pancreatic tumor tissue and also in mouse models of pancreatic cancer.33-36 Selective inhibition of classical NFκB in PC cells blocks NFκB-dependent gene expression, sensitizes the cells to apoptosis, reduces their growth and proliferation and inhibits their in vitro angiogenic potential.34-36,39,42,47 Classical pathway inhibition also prevents tumor growth and angiogenesis in mouse orthotopic transplant models of pancreatic cancer.34-36,42 These accumulated findings clearly demonstrate that the classical NFκB pathway is constitutively active and contributes to tumorigenesis in pancreatic cancer. In contrast, the role of the more recently described non-canonical NFκB pathway has not been specifically addressed in PC and is poorly defined in tumor cells in general.
To address this gap in our understanding we questioned whether PC cells express the known NC NFκB-dependent target genes (CCL19, CCL21, CXCL12, CXCL13 and BAFF). We analyzed gene expression in a panel of well-described PC cell lines derived from primary (Panc-1, MiaPACA-2, BxPC-3), metastasized (Hs766T, Capan-1) and ascites (AsPC1) tumors.49 As these lines have been maintained in culture for many years and exhibit diversity in their phenotypes and biological behavior,49 we also analyzed gene expression in PCA-2 cells that were derived recently from a human adenocarcinoma. PCA-2 cells remain in low passage and are more likely to retain the in situ tumor cell characteristics than the older, high-passage PC cell lines. Remarkably, we found that all of the PC lines exhibited significantly enhanced expression of at least two of the NC NFκB-dependent genes compared with HeLa cells.
Notably, the levels of the NC genes in PC cells were comparable with those in the multiple myeloma (MM) cell line RPMI-8226 in which the NC pathway has been shown to be constitutively active.18 We confirmed that p52 levels are elevated in these cells; however, we did not detect any defective TRAF3 expression or NIK stabilization in the RPMI-8226 cells as previously reported by Keats and colleagues.18 It is not clear why these differences exist, but it is possible that the experimental conditions we have employed differ from those of the previous study. Nevertheless, as NC NFκB-dependent gene expression was not previously investigated in MM cells, our results extend the understanding of the functional effects of constitutive NC NFκB activity in multiple myeloma.
Previous studies have demonstrated enhanced expression of BAFF and CXCL13 in B-cell lymphomas, and the in vitro and in vivo expression and function of CXCL12 has been extensively reported in various solid tumors.26-32,50 Consistent with our findings, elevated CXCL12 expression has been described in AsPC-1 and Hs766T PC cells49,51 although in a separate study, Koshiba and colleagues did not detect any CXCL12 expression by AsPC-1, BxPC-3 or Panc-1 cells.52 It is not clear why these disparities exist but Koshiba et al. used non-quantitative RT PCR whereas we employed qRT-PCR to detect chemokine expression. Using this technique we find that all of the cell lines tested consistently express CXCL12, suggesting that this is a common phenotype of PC cells. Supporting this notion Koshiba et al. did find that all primary pancreatic tumor tissue samples they tested expressed high levels of CXCL12.52
Pancreatic cancer cells have been shown to express high levels of the CXCL12 receptor CXCR4, and autocrine feedback via the CXCL12-CXCR4 axis may regulate tumor survival, proliferation, migration and metastasis.51,52 In addition, CXCL12 is potently pro-angiogenic and its expression by PC cells might contribute to the angiogenic potential of these tumor cells in vivo.36,42 In contrast, none of the other NC NFκB-dependent genes we detected have been described previously in PC cells. As CCL19, CCL21, CXCL13 and BAFF all have important immunoregulatory functions, we speculate that these may play a significant role in shaping the immune microenvironment in pancreatic cancer. In this regard, expression of CXCL12 in ovarian tumors promotes the recruitment of T regulatory cells that may suppress antitumor immunity.32 Clearly, further work is required to fully dissect the function of these genes in pancreatic tumorigenesis; however, our findings present the first line of evidence that this group of NC NFκB-dependent genes is constitutively expressed in any tumor cell type.
To determine whether constitutive NC NFκB activity underlies the expression of these genes, we used dominant negative IKKα to block NC signaling. We previously demonstrated that this approach selectively targets the NC pathway45 and found here that IKKαSSAA significantly reduced CXCL12, CXCL13 and BAFF expression in BxPC-3 cells. Notably however, IKKαSSAA did not affect CCL21 expression suggesting that CCL21 is regulated independently of the NC NFκB pathway in these cells. Consistent with this concept, BxPC-3 was the only PC line in which we observed upregulated CCL21. It is therefore possible that a negative regulatory mechanism exists to prevent CCL21 expression in the remaining PC cells that exhibit constitutive expression of the other NC NFκB-dependent genes. Furthermore, recent evidence indicates that some NC genes can be regulated independently of non-canonical p52-RelB dimers.50 Consequently, crosstalk between constitutive classical and NC signaling regulating gene expression may occur in tumor cells and this is certainly an area that warrants further study. Overall however, our data strongly suggest that selectively targeting the NC pathway will directly alter the gene expression profile and phenotype of pancreatic cancer cells.
Consistent with the NC gene expression and the effects of IKKαSSAA, we found that the PC lines displayed elevated levels of p52 suggesting that basal p100 processing is upregulated in these cells. We also found nuclear p52 and RelB in PC cells and we co-immunoprecipitated p52-RelB heterodimers from the nuclei of BxPC-3 and Hs766T cells. Although HeLa cells had high levels of nuclear p52 and RelB, these proteins did not co-immunoprecipitate, suggesting that they are components of other non-NC NFκB complexes. Importantly, we demonstrated that p52 and RelB are present and co-localize in human pancreatic adenocarcinoma tissue although it is not possible to distinguish between p100 and p52 immunohistochemically. Nevertheless, when combined with our IB and IP studies, these findings strongly support the presence of constitutively active NC NFκB signaling in PC cells.
Previous studies have demonstrated nuclear p52 or RelB in cells derived other solid tumors including breast, prostate, ovarian and colon cancer.20-23,53 In addition, nuclear p52 and RelB has been extensively reported in lymphoid malignancies.54 In PC cell lines, Chandler and colleagues19 reported p52 in the nuclei of Panc-1 and BxPC-3 cells and Liptay et al.36 showed that RelB was present in the nuclei of MiaPACA-2 cells. Intriguingly, Schneider et al.55 recently reported the co-localization of p52 and RelB on the promoter for the S-phase kinase associated protein 2 (skp2) in MiaPACA-2 cells and showed that this was disrupted by inhibition of IKKα. Farrow and colleagues33 reported that IKKα levels are increased in pancreatic tumor tissue samples though we did not consistently detect any overall differences in the levels of the IKK complex subunits in our panel of PC cells (data not shown). However, we found that the downstream target of IKKα in the NC pathway (i.e., p100) is constitutively phosphorylated in PC cells. Together with the effects of IKKαSSAA on NC gene expression, these accumulated lines of evidence suggest that IKKα is constitutively active in PC cells resulting in the continual phosphorylation and processing of p100 to p52 and the nuclear translocation and transcriptional activity of p52-RelB heterodimers.
The mechanisms that regulate constitutive classical NFκB activity in cancer cells have been intensely studied.1,2 However, as constitutive activity of NC signaling has not been widely reported, little is known about the potential mechanisms leading to its aberrant activation. Recently, however, compelling evidence for such a mechanism emerged from two independent studies.17,18 These studies demonstrated that an extensive array of mutations in key upstream signaling components in the NC pathway including TRAF3 and NIK, leads to constitutive activation of both the NC and classical pathways in multiple myeloma cells.17,18 A major outcome of these mutations was stable expression of NIK that in turn induced downstream activation of both NFκB pathways. Intriguingly, we found increased levels of NIK protein (suggesting increased NIK accumulation) in a subset of the PC cell lines, including the recently derived PCA-2 cells. This stabilization did not appear to involve loss of TRAF3, strongly suggesting that mutations similar to those in MM that affect upstream signaling proteins may occur in PC cells.
As we did not observe elevated NIK levels in all of the cell lines, it is likely that diverse mechanisms leading to NC NFκB activation occur in PC cells. Potential other mechanisms may include enhanced p100 expression via constitutive classical NFκB activity or autocrine feedback of factors that directly activate the NC pathway. Such autocrine activation via enhanced IL-1 expression has been suggested for the classical pathway in PC cells37 and it is notable that PC cells express CD4049, which is one of the few receptors known to activate the NC NFκB pathway.10 Finally, it is possible that aberrant activity of unrelated signaling mechanisms leads to NC NFκB activation. In this regard, the ser/thr kinases GSK-3α and β were shown recently to play a critical role in constitutive classical NFκB signaling via IKK activation in BxPC-3, Panc-1 and MiaPACA-2 cells.41,48 Determining whether these, or other, mechanisms underlie the aberrant activity of the NC pathway in PC cells will require intense future investigation.
In conclusion, our finding that PC cells express NC NFκB dependent genes represents the first comprehensive investigation of this panel of genes in any tumor cell type. We have established that the NC NFκB effector proteins p52 and RelB co-localize and associate in the nucleus of PC cells and we provide the first evidence for NIK stabilization in cells derived from a solid tumor. Given the fact that NIK stabilization occurs in multiple myeloma,17,18 we speculate that this may emerge as a major mechanism regulating constitutive NFκB activity in a broader range of tumor cell types. Collectively, these findings reveal a novel avenue of attack for pancreatic cancer and lay the foundation for further selective inhibitory strategies targeting aberrant NC NFκB activity in cancer therapy.
PCA-2 are adenocarcinoma cells that have been in continuous culture since removal from a patient on 07/13/06. The cells used in this study were at passage 24 and were grown in E medium supplemented with 0.5% HuSA on collagen I.43,44 The cells grow in culture as both adherent and non-adherent cells with frequent sphere formation. PCA2 cells form colonies in methyl-cellulose, and tumors in mice, with histology suggesting both an undifferentiated and differentiated population. All other cell lines were obtained from either the American Type Culture Collection (Manassas, VA) or as a kind gift from Dr. Anil K. Rustgi at the Gastroenterology Division of the University of Pennsylvania. HeLa, PANC-1 and MiaPACA-2 cells were maintained in Dulbecco's Modified Eagle's Medium (Invitrogen, Carlsbad, CA) supplemented with 10% fetal calf serum, 2 mM L-glutamine, penicillin (50 units/ml) and streptomycin (50 μg/ml). RPMI-8226, AsPC-1 and BxPC-3 cells were maintained in RPMI (Invitrogen) supplemented with 10% fetal calf serum, 2 mM L-glutamine, penicillin (50 units/ml) and streptomycin (50 μg/ml). Hs766T cells were maintained in a 1:1 mixture of Ham's F12 medium with 2.5 mM L-glutamine and Dulbecco's Modified Eagle's Medium (Invitrogen) supplemented with 10% fetal calf serum, 2 mM L-glutamine, penicillin (50 units/ml) and streptomycin (50 μg/ml). Capan-1 cells were maintained in a 1:1 mixture of Ham's F12 medium with 2.5 mM L-glutamine and Dulbecco's Modified Eagle's Medium (Invitrogen) supplemented with 20% fetal calf serum, 2 mM L-glutamine, penicillin (50 units/ml) and streptomycin (50 μg/ml). Primary human umbilical vein endothelial cells (HUVEC) were isolated, cultured and treated with LIGHT (R&D Systems; Minneapolis, MN) as previously described.45
Monoclonal anti-tubulin was from Sigma (St. Louis, MI). Polyclonal anti-RelB, anti-TRAF3 and anti-NIK antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Polyclonal anti-phospho-p100 and polyclonal anti-p52 antibody was from Cell Signaling Technology (Boston, MA). Monoclonal anti-p52 and monoclonal anti-IKKα were from Upstate (Millipore, Billerica, MA). Horseradish peroxidase-conjugated, biotinylated and fluorescent-conjugated secondary antibodies against either rabbit or mouse IgG were from Jackson ImmunoResearch Labs, (West Grove, PA). Mouse IgG1 isotype control for immunoprecipitation was from R&D Systems.
RNA was isolated from cultured cells using the RNeasy Mini Kit (Qiagen, Valencia, CA) according to manufacturer's instructions for the spin protocol. For HeLa plus TNF controls, cells were incubated with 10 ng/mL TNF (R&D Systems) for 4 h prior to RNA isolation. Complimentary DNA was prepared using the High Capacity cDNA Reverse Transcription Kit from Applied Biosystems (Foster City, CA) according to the manufacturer's instructions. All quantitative real-time reverse transcription (qRT)-PCR was performed using a 7500 Fast Real-Time PCR System (Applied Biosystems) and all qRT-PCR reagents and consumables were purchased from Applied Biosystems. For each reaction, 100 ng of cDNA was added to each well of an optical 96-well fast thermal cycling plate along with 2× TaqMan Fast Universal PCR Master Mix and 1 μL of the specific commercially designed TaqMan Gene Expression Assay primer set. Each sample was analyzed in quadruplicate. Relative quantification (RQ) was derived from the difference in cycle threshold (Ct) between the gene of interest and β-actin (ΔCt) as compared to HeLa baseline controls using the equation RQ = 2-ΔΔCt. Error bars represent standard error of the mean (SEM), sample standard deviation divided by the square root of the sample size. Statistical significance was calculated using a one-tailed, unpaired t-test.
Paraffin-embedded tissues were obtained from the Cooperative Human Tissue Network at the University of Pennsylvania. The tissues were de-waxed and rehydrated by progressive transfer through xylene, ethanol and water solutions, followed by boiling in citric acid buffer (pH 6.0). Endogenous peroxidases were quenched by incubation in hydrogen peroxide, and the tissues were blocked with avidin, biotin and protein blocking solutions. The tissues were incubated in either monoclonal anti-p52 or polyclonal anti-RelB at a dilution of 1:500 in PBT buffer (phosphate-buffered saline, 0.1% BSA, 0.2% Triton X-100) overnight at 4°C. Tissues were then rinsed and incubated in biotinylated secondary antibody (1:200 for rabbit, 1:100 for mouse) for 30 minutes at 37°C, followed by 30 min incubation at 37°C in HRP-ABC reagent and developing in DAB solution with hematoxylin counterstain. The stained tissues were then dehydrated and mounted with coverslips.
Paraffin-embedded tissues were obtained and rehydrated/de-waxed for IHC. Following protein blocking, the tissues were simultaneously incubated in monoclonal anti-p52 and polyclonal anti-RelB at a dilution of 1:500 in PBT buffer overnight at 4°C. The slides were then incubated in Cy3- and Cy5-labelled secondary antibodies (1:600 in PBT) for 30 min at 37°C. Tissues were counterstained with DAPI and mounted with cover slips using fluorescent mounting medium.
Whole cell lysates from adherent cells were prepared by washing with phosphate-buffered saline, incubating in TNT lysis buffer (50 mM Tris pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 mM DTT, 2 mM NaF, 2 mM β-glycerophosphate) on ice for 10 min, followed by scraping and brief vortexing. RPMI-8226 suspension cells and PCA-2 cells were lysed in the same buffer applied directly to a rinsed pellet. Nuclear lysates were prepared by cell trypsinization, followed by one wash in phosphate-buffered saline. Cells were then lysed for 15 min at 4°C in 200 μL buffer (10 mM HEPES pH 7.9, 10 mM KCl, 0.1 mM EDTA), followed by the addition of 25 μL 1% NP-40 detergent for 15 minutes at room temperature. Lysates were then vortexed strongly for 30 s, and centrifuged to pellet the nuclei. The nuclear fraction was then rinsed, followed by nuclear membrane lysis in 50 μL lysis buffer (20 mM HEPES pH 7.9, 0.4 M NaCl, 1 mM EDTA) and shaking at 4°C for 15 min. Lysates were then clarified by centrifugation and supernatants containing nuclear protein were retained.
Protein concentrations were measured using Coomassie Plus Reagent (Pierce, Rockford, IL). Due to the differing structural and metabolic properties among the separate members of the cell panel, standard loading controls such as anti-α-tubulin could not be used to normalize blots. We therefore loaded identical concentrations of protein for each sample in blots in which more than one cell line was used. For each sample, 10 μg of protein lysate was separated by SDS-PAGE then transferred electrophoretically to a polyvinylidene difluoride membrane (Immobilon P, Millipore, Milford, MA) and immunoblotted with primary and horseradish peroxidase-conjugated secondary antibodies in 5% milk. Detection of the bound antibody was performed by enhanced chemiluminescence using Lumi-Light western blotting Substrate (Roche, Palo Alto, CA) and HyBlot autoradiography film (Denville Scientific, Metuchen, NJ). Anti-NIK and anti-phospho-p100 primary antibodies were used at a dilution of 1:200 in 5% milk. Anti-a-tubulin primary antibody was used at a dilution of 1:5,000 in 5% milk. All other primary antibodies were used at a dilution of 1:1,000 in 5% milk. Densitometry was performed using a Gel-Doc EQ and the QuantityOne software package (Bio-Rad Laboratories). Pixel intensity was measured in identical rectangular volumes around each band on immunoblots and a background value in an equal rectangular volume separate from the bands was subtracted to obtain the mean pixel intensity/mm2.
Whole cell lysates were obtained as described above, and for each sample, 100 μg of protein was further diluted to a total volume of 500 μL. The lysate was then incubated with 2 μg of either isotype control or monoclonal anti-p52 overnight at 4°C, followed by a 4 h incubation with 30 μl of a 50% suspension of Protein A/G agarose beads from Pierce. The beads were washed three times with lysis buffer (0.1% Triton X-100) and boiled in sample buffer containing SDS. Protein samples were analyzed by immunoblotting using anti-RelB and anti-p52.
GFP control and dominant negative IKKαSSAA cDNA were cloned into the retroviral vector LZRS, and transfected into ΦNX packaging cells with Fugene6 reagent (Roche) as previously described.45 Cells were then selected for successful transfection with 1 μg/mL puromycin. Viral supernatant was harvested with 1 μg/mL polybrene and stored at -80°C. BxPC-3 cells were incubated with viral supernatant overnight and replaced with standard growth medium during the day. This infection protocol was repeated through six rounds, with cell passage as required. To determine the efficiency of GFP control transduction, a pellet of BxPC-3 cells transduced with the GFP construct were resuspended in PBS and analyzed by FACs (FACSort, BD Biosciences). Whole cell lysates were then prepared and immunoblotted for IKKα expression and RNA was prepared and analyzed by qRT-PCR as described. Statistical significance was calculated using a one-tailed, paired t-test.
The Cancer Research Institute (Pre-doctoral Emphasis Pathway in Tumor Immunology Grant), the National Institutes of Health (R01-HL080612 and P30-DK050306) and an award from the Abramson Cancer Center, University of Pennsylvania, funded this work. The authors thank Dr. Andres Klein-Szanto of the Fox Chase Cancer Center (Philadelphia, PA) for pathology assistance and Dr. Lisa Madge (University of Pennsylvania) for providing primary HUVEC.