Restricted Heterochromatin Formation Links NFATc2 Repressor Activity With Growth Promotion in Pancreatic Cancer

Cell Lines, Plasmids, Reverse-Transcription Polymerase Chain Reaction Analysis, and Small Interfering RNA Oligonucleotides

Detailed descriptions on cell lines, plasmid generation, small interfering RNA (siRNA) oligonucleotides, and transfection protocols used in this study are available in Supplementary Materials and Methods.

Generation of Stable Cell Lines, Microscopic Analysis, Protein Analysis, Promoter Studies, Growth Assays, and Xenograft Models

Supplementary Materials and Methods contains details of antibodies as well as protocols of standard techniques, including generation of viral vectors and stable cell lines, fluorescence microscopy, immunohistochemistry, subcellular fractionation, isolation of proteins from tumor tissues, coimmunoprecipitation, immunoblotting, chromatin immunoprecipitation (ChIP) assays, DNA pull-down assays, proliferation assays, reporter gene assays, soft agar assays, xenograft experiments, and statistical analysis, performed as described previously.^10-12

Ectopic Expression of the Calcium/Calcineurin Responsive NFATc2 Transcription Factor During Pancreatic Carcinogenesis

Immunohistochemical analysis revealed that NFATc2 was detectable neither in acinar or ductal cells of healthy pancreas nor in early PanIN-1 lesions (Figure 1A). However, we found positive NFATc2 staining in advanced PanIN-2 and PanIN-3 precursor lesions and in the majority of invasively growing cancers (19/23; Figure 1A and B). Depending on its phosphorylation level, NFATc2 can shuttle between the cytosol and the nuclear fraction.¹³ When dephosphorylated by the phosphatase calcineurin, cytoplasmic NFATc2 becomes active and translocates into the nucleus to modulate gene expression. We found NFATc2 widely distributed throughout ductal cells in preinvasive PanIN-2 and PanIN-3 lesions. More important, however, NFATc2 was exclusively or predominantly expressed in the nuclear compartment of invasively growing tumor cells in the majority of positive tumor samples (14/19), thus indicating that NFATc2 is present in an active state in advanced human exocrine pancreatic cancer (Figure 1A and B). We were not able to find a significant correlation between nuclear NFATc2 localization and clinical parameters such as tumor stages and gradings (TNM) (data not shown). Consistent with increased expression of NFATc2 in cancer cells, we found high levels of both the messenger RNA and protein in a significant portion of established human pancreatic cancer cells, as evidenced by Western blot analysis and real-time polymerase chain reaction (PCR) (Figure 1C and D). Together, these results indicate that NFATc2 expression is induced during pancreatic carcinogenesis and reaches maximum expression levels in the nucleus of invasive cancer cells, pointing to a high activity of this transcription factor state in cancer.

NFATc2 Promotes Cancer Growth and p15^INK4b Silencing In Vitro and In Vivo

Administration of the calcium ionophore ionomycin caused increased [³H]thymidine incorporation in pancreatic cancer cells (PaTu-8988t and IMIM-PC1 cells), and this effect was abolished following genetic depletion of NFATc2 (Figure 2A and B). Similarly, stable retroviral expression of NFATc2 in Suit007, Suit028, and PaTu-8988t cells increased cell proliferation and colony formation (Figure 2C and D and Supplementary Figure 1A and B) and promoted tumor growth in xenograft tumors (Figure 3A). Interestingly, NFATc2-mediated growth promotion was accompanied by silencing of the p15^INK4b tumor suppressor pathway both in vivo and in vitro (Figure 3B and C) and resulted in increased expression of D-type cyclins and their partner kinases cyclin-dependent kinases 4 and 6 (Supplementary Figure 1C and D). Conversely, genetic depletion of this transcription factor caused release of p15^INK4b expression both on messenger RNA and protein levels (Figure 3D and Supplementary Figure 2A), reduced expression levels of D-type cyclins, and prevented cells from growth promotion by ionomycin (Figure 2A and B and Supplementary Figures 1E, 1F, and 2A). Immunohistochemical analyses of serial human pancreatic cancer sections for NFATc2 and p15^INK4b expression finally showed an inverse correlation between NFATc2 and p15^INK4b expression (Figure 3E). Taken together, these results provided first evidence for NFATc2 repressor activity in cancer cells and suggested a role in silencing of the p15^INK4b tumor suppressor pathway in pancreatic cancer.

NFATc2 Targets the p15^INK4b Promoter for Silencing

We next questioned whether in growing pancreatic cancer cells the mechanism underlying NFATc2-mediated repression of p15^INK4b is mediated through the silencing of its promoter. ChIP assays showed NFATc2 promoter occupancy under serum-free conditions, and this binding was further enhanced on mitogenic stimulation with 10% fetal calf serum (Figure 4A). Consistent with p15^INK4b promoter regulation by NFATc2, reporter gene assays displayed dose-dependent repression of the full-length p15^INK4b promoter (Figure 4B). Further analysis showed that deletion of the p15^INK4b promoter to nucleotide −166 rendered the promoter insensitive to NFATc2-mediated silencing, indicative of the existence of an NFATc2-responsive repression element located in the distal promoter region beyond base pair −165 (Figure 4C and D). Computer-based sequence analysis of the −751/−165 promoter region confirmed this hypothesis and revealed the presence of a single NFAT (GGAAA) recognition site within the distal p15^INK4b promoter. This site, located at −593/−589 base pairs, was disrupted in the context of the full-length promoter through site-directed mutagenesis (p15^INK4b NFAT-mut) and used for further promoter analysis. The functional relevance of the −593/−589 NFAT binding element was confirmed by DNA pull-down experiments and additional reporter gene assays, showing that mutational disruption of this site (p15^INK4b NFAT-mut) rendered the p15^INK4b promoter resistant to inducible NFATc2 binding and repression (Figure 4E and F and Supplementary Figure 2B and C). Together, these experiments defined a novel repressor function for NFATc2 and identified the p15^INK4b promoter as an NFATc2 target in pancreatic cancer cells.

Heterochromatin Formation Underlies NFATc2-Mediated Silencing of p15^INK4b Pancreatic Cancer Cells

Regulation of gene transcription requires posttranslational modifications of histones, which, in concert with chromatin remodeling factors, shape the structure of chromatin.¹⁴ Euchromatin is characterized by weak DNA staining and corresponds to less compacted chromatin, whereas heterochromatin is compacted and stains intensively with DNA dyes. Consistent with previous observations, immunofluorescence microscopy indicated a cytoplasmic localization of NFATc2 in unstimulated cancer cells (Supplementary Figure 2D). Activation of NFATc2, however, caused nuclear translocation of the factor with a predominant punctual localization pattern in euchromatic regions (Supplementary Figure 2D). We then questioned whether euchromatic NFATc2 recruits heterochromatin forming complexes to the p15^INK4b promoter for silencing of the tumor suppressor. The following observations indicated that this is indeed the case. First, activation of NFATc2 binding to the p15^INK4b promoter, as shown in Figure 4A, was associated with deacetylation of histone 3 (Supplementary Figure 2E) and recruitment of the Suv39H1 histone methyltransferase and a strong induction of histone 3 lysine 9 trimethylation (H3K9-me), a mark associated with gene silencing and heterochromatin formation (Figure 5A). Second and even more important, coimmunoprecipitation experiments and DNA pull-down (Figure 5B) revealed that NFATc2 forms a complex with Suv39H1 and recruits it to the p15^INK4b promoter via the −593/−589 binding site. Additionally, disruption of the NFATc2 promoter-binding site or siRNA depletion of NFATc2 abolished Suv39H1 recruitment to the p15^INK4b promoter and subsequent trimethylation of H3K9 without affecting Suv39H1 and H3K9-me protein expression levels (Figure 5B [right panel], Figure 5C, and Supplementary Figure 3A). Most importantly, NFATc2-dependent regulation of H3K9-me could be observed neither on transcriptional start site/exon I and intron within the INK4b locus nor on the p14^ARF promoter, clearly high-lighting that the heterochromatin formation is restricted to the p15^INK4b regulatory sequence (Supplementary Figure 3B–D and data not shown). In contrast, depletion of Suv39H1 did not affect NFATc2 recruitment to the p15^INK4b promoter (Supplementary Figure 3E), suggesting that NFATc2 recruits Suv39H1 for subsequent gene repression. Consistent with these findings, pharmacologic inhibition of Suv39H1 using chaetocin or siRNA-mediated depletion abrogated H3K9 trimethylation and subsequent NFATc2-mediated repression of the p15^INK4 promoter (Figure 5D–F and Supplementary Figure 3F). Together, these data identified an NFATc2-Suv39H1–dependent transcriptional repression mechanism at the p15^INK4b promoter.

HP1γ Is Required for Sustained Silencing of the p15^INK4b Promoter

HP1 family members are critical mediators of repressor complex formation and maintenance, thus greatly contributing to gene silencing.¹⁵ Growing evidence indicates that the three known HP1 isoforms may not fulfill identical functions and instead show differences in their nuclear distribution pattern and nuclear activities. In contrast to HP1α and HP1β, which are almost exclusively found in the heterochromatin area, HP1γ is localized in both heterochromatic and euchromatic regions, where it contributes to transcription complex stabilization.^16-18 We found HP1γ widely distributed throughout the nuclei of pancreatic cancer cells before and after growth stimulation (Figure 6A). However, after serum stimulation, HP1γ partially colocalizes with NFATc2 in euchromatic foci and interacts with the repressor in an Suv39H1-dependent manner (Figure 6A and B). In fact, depletion of Suv39H1 prevented HP1γ from both interaction with NFATc2 and docking to the p15^INK4b promoter (Figure 6B and C). Similarly, genetic knockdown of NFATc2 not only disabled Suv39H1 recruitment and local H3K9 trimethylation, as shown in Figure 5C, but prevented HP1γ promoter interaction (Supplementary Figure 4A). Functionally, HP1γ binding to the repressor complex is essential for heterochromatin stabilization and sufficient promoter silencing by NFATc2. Accordingly, DNA pull-down assays showed inducible NFATc2-Suv39H1 complex formation at the p15^INK4b promoter, and this complex remained stable for at least 3 hours when performed in the presence of HP1γ (Figure 6D). However, depletion of HP1γ destabilized repressor complex formation at the p15^INK4 promoter in vitro (Figure 6D) and in vivo (Figure 6E); hence, sufficient binding of neither NFATc2 or Suv39H1 nor local H3K9 trimethylation was observed in ChIP assays performed 3 hours after stimulation (Figure 6E). Consistent with destabilization of the repressor complex, depletion of HP1γ caused a loss of NFATc2-mediated p15^INK4b promoter silencing (Figure 6F and Supplementary Figure 4B). Together, these results defined a novel NFATc2 silencing mechanism in which the binding of this transcription factor to the p15^INK4 promoter induces local heterochromatin formation involving the stepwise recruitment of Suv39H1, trimethylation of H3K9, and docking of HP1γ.

Inactivation of NFATc2 Restores Expression of p15^INK4b and Inhibits Growth of Pancreatic Cancer

Finally, to define the biological relevance of our mechanistic findings, we analyzed the effect of NFATc2 inactivation on p15^INK4b expression and pancreatic tumor growth in vivo. For this purpose, we subcutaneously injected IMIM-PC1 pancreatic cancer cells with high NFATc2 expression levels into the flanks of nude mice. After 2 weeks, mice were treated intraperitoneally with a widely used specific inhibitor of calcineurin/NFAT signaling, cyclosporin A (CsA), or dimethyl sulfoxide 5 times weekly. Xenograft tumor growth was significantly delayed on administration of CsA, which was associated with increased p15^INK4b messenger RNA levels (Figure 7A and B). We complemented this experiment using a second NFAT inhibitor and treated IMIM-PC1 tumor cell–bearing nude mice with the GSK-3β inhibitor AR-A014418 (3 times per week). The GSK-3β serine kinase stabilizes NFATc2 in cancer cells and protects the transcription factor from rapid ubiquitination and proteasomal degradation. Inhibition of GSK-3β, in turn, promotes NFATc2 inhibition and degradation in cancer cells.¹² Consistent with the CsA results, treatment of tumor-bearing mice with AR-A014418 caused loss of NFATc2 expression and a significant retardation of tumor growth over time, reaching statistical significance after 2 weeks (Supplementary Figure 4C). Moreover, loss of NFATc2 expression abrogated p15^INK4b silencing and restored expression of this tumor suppressor gene and cancer defense mechanism in all animals, as shown by the decrease in tumor growth (Supplementary Figure 4D and E).

Taken together, in congruency with our biochemical and molecular analysis, these results showed that promotion of pancreatic cancer growth by NFATc2 involves silencing of the p15^INK4b tumor suppressor pathway. In addition, our xenograft tumor models revealed that inactivation of NFATc2 rescues p15^INK4b and blocks tumor growth in vivo. Thus, our results indicate that specific targeting of oncogenic NFATc2 signaling and transcription can restore the function of a critical cellular defense mechanism against mitogenic stimuli and therefore NFAT inactivation might be an attractive approach in the treatment of pancreatic cancer.