Dexamethasone mediates pancreatic cancer progression by glucocorticoid receptor, TGFβ and JNK/AP-1.

abstract CC BY

Glucocorticoids such as dexamethasone are widely co-prescribed with cytotoxic therapy because of their proapoptotic effects in lymphoid cancer, reduction of inflammation and edema and additional benefits. Concerns about glucocorticoid-induced therapy resistance, enhanced metastasis and reduced survival of patients are largely not considered. We analyzed dexamethasone-induced tumor progression in three established and one primary human pancreatic ductal adenocarcinoma (PDA) cell lines and in PDA tissue from patients and xenografts by FACS and western blot analysis, immunohistochemistry, MTT and wound assay, colony and spheroid formation, EMSA and in vivo tumor growth and metastasis of tumor xenografts on chicken eggs and mice. Dexamethasone in concentrations observed in plasma of patients favored epithelial–mesenchymal transition, self-renewal potential and cancer progression. Ras/JNK signaling, enhanced expression of TGFβ, vimentin, Notch-1 and SOX-2 and the inhibition of E-cadherin occurred. This was confirmed in patient and xenograft tissue, where dexamethasone induced tumor proliferation, gemcitabine resistance and metastasis. Inhibition of each TGFβ receptor-I, glucocorticoid receptor or JNK signaling partially reversed the dexamethasone-mediated effects, suggesting a complex signaling network. These data reveal that dexamethasone mediates progression by membrane effects and binding to glucocorticoid receptor.

dexamethasone favors emt-related tumor progression CC BY

Cancer spheres were isolated from the primary PDA xenograft line T30 and let either untreated or were treated in vitro with 1 μM (0.4 μg/ml) dexamethasone or solvent alone as control. This concentration resembles peak levels of 0.3 to 3 μg/ml that can be measured in patient plasma 2 to 16 h after intake of 2.49 mg/kg dexamethasone.28 At 48 h after treatment, the T30 cells were transplanted into the left and right flanks of immunodeficient mice. After 3 weeks, the tumors were resected and examined. The morphology resembled that of the primary patient tumor with the typical ductal structures (Figure 1a). Dexamethasone enhanced the tumor volume and proliferation, as detected by calipers (Figure 1b) and Ki67 staining (Figure 1c), whereas the expression of vimentin increased and the expression of E-cadherin decreased (Figure 1d). The observed expression pattern of vimentin and E-cadherin was confirmed by immunohistochemistry in tissue of PDA patients who had taken glucocorticoids before surgery (n=17) or not (n=17) (Figure 1e and Table 1). To confirm these results and to examine the effect of dexamethasone to standard chemotherapy, we treated AsPC-1, BxPc-3 and MIA-PaCa2 cells with dexamethasone, gemcitabine or both together, or let the cells untreated, before they were transplanted to the CAM of fertilized chicken eggs. After 9 days, large tumor xenografts developed in all dexamethasone-treated groups, even in the presence of gemcitabine, whereas the tumor sizes in groups that have not received dexamethasone were smaller and almost undetectable after gemcitabine treatment (Figure 2a and Supplementary Figure S1a). Dexamethasone usage was associated with enhanced proliferation and the combination with gemcitabine prevented it, as found by Ki-67 immunofluorescence staining and counting of positive cells (Figure 2b and Supplementary Figure S1b). In addition, the expression of E-cadherin was inhibited and the expression of vimentin was increased upon dexamethasone treatment (Figure 2c and Supplementary Figure S1c), together with an enhanced invasive potential, as concluded from PCR detection of human Alu sequences in genomic DNA from chick CAM and liver. Whereas 31% of the CAM and 0% of the liver tissues were Alu positive in eggs with untreated AsPC-1 cells, dexamethasone increased the percentage to 85% and 60%, respectively (Figure 2d and Supplementary Figure S1d). Similarly, dexamethasone increased the invasion of BxPc-3 and MIA-PaCa2 tumor cells to the CAM from 20% to 40% and from 30% to 70%, respectively (Figure 2d and Supplementary Figure S1e). Internal chicken GAPDH controls were performed and the results confirmed equal conditions (Supplementary Figure S1f). No significant differences in the mean weights of the embryos occurred between the groups (Supplementary Figure S1g), suggesting that no toxic side effects occurred. To further highlight the influence of dexamethasone to tumor progression, the cells were treated with dexamethasone, followed by western blot analysis. Each cell line exhibited an already basal expression of the glucocorticoid receptor and dexamethasone further induced (MIA-PaCa2, BxPc-3) or repressed (AsPC-1) it (Supplementary Figure S2a). This implies a functional autoregulation of the glucocorticoid receptor by its ligand, as previously described.30 Dexamethasone switched the epithelial cell morphology to a more spindle-shaped phenotype, as shown by microscopy 240 h after dexamethasone treatment (Supplementary Figure S2b). The migratory activity increased significantly upon dexamethasone treatment, as measured by scratch and transwell assays (Supplementary Figures S2c and d). This was accompanied by the inhibition of E-cadherin and the induction of vimentin in AsPC-1 and BxPc-3 cells, as detected by double immunofluorescence staining (Supplementary Figure S2e). Dexamethasone did not alter the expression of E-cadherin and vimentin in MIA-PaCa2 cells because of already high basal expression levels. These results suggest that dexamethasone induces EMT-related tumor progression.

dexamethasone induces tgf CC BY

Because TGFβ is a key player in EMT, we measured its expression by western blot analysis.31 Dexamethasone induced the expression of the 25/26 kDa form of mature TGFβ protein in AsPC-1, MIA-PaCa2 and BxPc-3 cells within 0.5 h and further induced it during the subsequent 24 h, followed by downregulation at 48 h (Figure 3a). This was associated with a decrease of soluble TGFβ1 protein in the cell supernatant after dexamethasone that may be explained by receptor binding and as detected by ELISA assay (Figure 3b and Supplementary Figure S3a). In contrast, coincubation of the glucocorticoid receptor antagonist mifepristone (RU486) counteracted the dexamethasone-induced dropdown of soluble TGFβ1. These results are underlined by coincubation experiments with the TGFβ receptor-I and -II inhibitor LY2109761 that significantly revised the dexamethasone-induced gemcitabine resistance, as measured in AsPC-1 cells by assays for viability, clonogenicity and spheroid formation (Figures 3c–e and Supplementary Figures S3b–d). Remarkably, dexamethasone did not induce gemcitabine resistance upon cotreatment of cells shortly after detachment of the cells from the culture dish with trypsin (data not shown), suggesting that trypsin removed membrane-associated dexamethasone-binding partners, for example, membrane-bound glucocorticoid receptor and other cell surface receptors. To obtain knowledge whether cancer progression was induced, we treated the cells 48 h after trypsinization with dexamethasone for 2, 6, 12 and 24 h, followed by western blot analysis of the pluripotency markers SOX2 and Notch1 and saw that dexamethasone rapidly induced the expression of both proteins (Figure 3f and Supplementary Figure S3e). To verify these data in tissues from patients, we performed immunofluorescence staining. Fluorescence microscopy revealed a higher expression of TGFβ, SOX2 and Notch1 in tissue of patients who received a glucocorticoid before surgery (n=17) compared with tissue of patients who received no glucocorticoid ((n=17, Figure 3g and Table 1). Because diverse glucocorticoids were administered to patients, these results suggest that not only dexamethasone but also other glucocorticoids induce TGFβ1-dependent EMT and tumor progression.

dexamethasone mediates ras/jnk and ap-1 signaling CC BY

To measure rapid membrane and kinase effects of dexamethasone, we detected the expression of Ras/JNK signaling and AP-1 activity, known to activate TGFβ expression.32 Dexamethasone led to enhanced RAS expression and phosphorylation of cJun within 2 to 5 min in MIA-PaCa2, AsPC-1 and BxPc-3 cells, with peaks at 10 to 45 min, as measured by western blot analysis and JNK/SAP kinase assays (Figure 4a). An enhanced binding activity of the transcription factor and JNK-target cJun/AP-1 occurred within 2 h, with peaks at 12 and 24 h as detected by EMSA and oligonucleotides of the vimentin promoter AP-1 binding site (Figure 4b). Immunofluorescence staining of human PDA tissue transplanted to mice revealed that the expression of P-cJun was still increased at late time points after dexamethasone treatment (Figure 4c, compare Figures 1a and b). Similar results were obtained by staining of patient tissue that demonstrated a higher expression of P-cJun upon intake of a glucocorticoid before surgery (Figure 4d and Table 1). To evaluate the impact of each key player, we treated AsPC-1 cells with dexamethasone and gemcitabine alone or together and in the presence or absence of inhibitors of the glucocorticoid receptor (mifepristone/RU486), the TGFβ receptor (LY2109761) or JNK signaling (SP600125). Each inhibitor alone partially reversed the dexamethasone-mediated gemcitabine resistance as examined 72 and 96 h after treatment by MTT assay (Figure 5a and Supplementary Figure S4a). Western blot analysis shows that RU486 inhibited the dexamethasone-induced upregulation of TGFβ and its mediator P-Smad2, along with inhibition of Notch1, SOX2 and P-cJun expression (Figures 5b and c). LY2109761 inhibited the upregulation of TGFβ, P-Smad2, SOX2 and P-cJun, but had no effect on Notch1 expression. Similarly, SP600125 inhibited the activation of all examined proteins. These results were verified in vivo in MIA-PaCa2 cells and by the use of the TGFβ receptor inhibitor (Figure 5d). AsPC-1 cells were treated with dexamethasone and gemcitabine alone or together, ±LY2109761, followed by xenotransplantation of equal amounts of viable cells to the CAM of fertilized chicken eggs. At 10 days after transplantation, the tumors were resected and the tumor volumes were detected by calipers. In confirmation with our in vitro results, the inhibition of TGFβ signaling reversed the dexamethasone-induced gemcitabine resistance. The measurement of the invasion potential by Alu-PCR showed that 70% of CAM tissues from dexamethasone groups were Alu positive, but LY210976 reduced the percentage to 50% (Figure 5e). Accordingly, 30% of CAM tissues from the dexamethasone and gemcitabine combination groups were Alu positive, but LY210976 reduced it to 20%. Internal chicken GAPDH controls for Alu PCR reactions demonstrated equal conditions (Supplementary Figure S4b). These results suggest that a dexamethasone-induced complex signaling network mediates tumor progression with TGFβ, RAS/JNK and AP-1 as key players.

discussion CC BY

Whereas glucocorticoids have many advantages for cancer patients, our present results hint to undesirable side effects in PDA, namely the induction of tumor progression, chemotherapy resistance and metastasis by a complex sequence of signaling (Figure 6). We assume that dexamethasone-induced therapy resistance and tumor progression depends on interactions with membrane-associated factors, for example, the membrane-bound glucocorticoid receptor coupled to G proteins, and the activation of downstream signaling.33 This assumption is based on our observation that freshly trypsinized cells did not respond with tumor progression to dexamethasone treatment—most likely because trypsin removed membrane-associated binding partners from the cell surface.34 Such membrane-mediated actions of glucocorticoids were already observed in lower vertebrate species and are believed to represent the more evolutionarily ancient forms of glucocorticoid receptor signaling.33 This very old way might have become dominant in PDA cells during the dedifferentiation process of normal cancer cells into CSCs. Several recent reports support our observation of dexamethasone-induced EMT and cancer progression. For example, elevated JNK activity in breast cancer did not induce apoptosis as expected, but cell migration and EMT, along with an enhanced expression of vimentin and AP-1.35 In addition, targeting of RAS/JNK eliminated CSC features and prevented tumor formation in pancreatic cancer.36 Similarly, a pivotal role of JNK in the maintenance of self-renewal and tumorigenicity in glioma stem-like cells was demonstrated.37 Moreover, AP-1 was defined as a critical regulator of a complex program of gene expression mediating an invasive phenotype.27 However, there are reports that published the opposite effects. One study for example claimed that dexamethasone inhibited the invasiveness of a human pancreatic cancer cell line that was examined by in vitro invasion assay and expression of MMP-9.38 Another study found a dexamethasone-induced inhibition of in vitro invasion of the PDA cell line Panc-TuI and, after orthotopic transplantation to immunodeficient mice, reduced tumor growth and metastasis.39 We do not know the reason for these obvious discrepancies to our results, but suspect that it might be because of differences in cell lines, experimental conditions and data evaluation. For example, the conclusions are drawn from one cell line only in both studies. In addition, the in vitro invasion assays are highly susceptible to faults, the presented pictures are out of focus and not quantitatively evaluated. Next, the PDA cell line Panc-TuI was used that we never evaluated, and therefore it might well be that this is one of a few solid tumor cells that are sensitive to dexamethasone treatment. In the orthotopic xenograft experiments, dexamethasone was not applied to inhibit primary tumor growth as in our study, but first when the xenografts were resected with the intention to prevent a relapse.39 Moreover, orthotopic transplantation to the mouse pancreas by median laparatomy is known to result in xenograft tumors that strongly vary in size. Thus, the measurement of an increase in tumor volume over time by in vivo imaging and the presentation of the results as ‘mean increase of tumor volume over time’ may provide the most accurate results. However, Egberts et al.39 evaluated tumor growth only once at the end of the experiment and presented the results as ‘mm3’. An explanation of how the tumor volumes were evaluated was not provided and hence we are not able to judge the method. Signaling by the cytoplasmic glucocorticoid receptor was involved in our system as well, and may have mediated the enhanced expression of TGFβ by binding to the glucocorticoid-responsive element in the TGFβ promoter,19 as concluded from partial inhibition by mifepristone/RU486. Similarly, glucocorticoid receptor/AP-1 interactions might have enforced the signaling cascade by the binding of AP-1 to respective binding sites in the TGFβ and vimentin promoters,40 and to other target genes that are involved in stemness and progression. Positive feedback loops mediated by autoactivation of cJun transcription by AP-1 and the subsequent activation of TGFβ, which in turn activates Ras/JNK signaling and cJun activation by phosphorylation, have enforced the scenario.25 Our results obtained after inhibition of JNK signaling confirmed a direct link of dexamethasone-induced rapid and most likely nongenomic TGFβ and RAS/JNK action,41 with subsequent cJun/AP-1-mediated upregulation of vimentin,42 along with downregulation of E-cadherin.43 Interestingly, we observed that the amount of soluble TGFβ1 ligand in cell culture supernatant of dexamethasone treatment was diminished, hinting to the binding of TGFβ1 to its receptor. This assumption is underlined by our result that a TGFβ receptor I/II blocker partially neutralized the dexamethasone-induced tumor progression. We do not exclude that a complex network of other signaling cascades is involved, such as glucocorticoid receptor-chromatin interactions, that are described to require AP-1 as a major partner.44 In this regard, the loss of glucocorticoid receptor in transgenic mice resulted in tumorigenesis because of defective chromosome segregation.18 We observed the induction of RAS/JNK signaling by dexamethasone, whose relation to dexamethasone-induced anti-apoptotic signaling in breast cancer by interaction with protein kinases PI3-K and Akt is well known.45 This matches to our data, because PI3-K and Akt are the best-characterized RAS effectors.26 Other involved key players may be the serum and glucocorticoid-regulated kinase-1 (SGK1), as reported by Conzen and colleagues10, 46, 47, 48 in breast, ovarian and prostate cancer. In a human breast cancer xenograft model, the inhibition of glucocorticoid receptor signaling reversed the glucocorticoid-induced chemotherapy resistance,10 in line with our results, where the inhibition of the glucocorticoid receptor partially reversed the dexamethasone-induced gemcitabine resistance. In summary, we suggest a careful weighing of advantages and disadvantages of glucocorticoids in therapy of solid tumors and the use of alternatives such as nonsteroidal medicaments for treatment of inflammation, emesis, edema, allergies and immune response. Our data may be of clinical relevance with regard to the still frequent use of glucocorticoid medication, but also considering the known elevated levels of endogenous glucocorticoids in chronic pancreatitis, under severe stress conditions, in chronic depression, after protein-rich meals and acute smoking.49

cell culture CC BY

The human established PDA cell lines MIA-PaCa2, AsPC-1 and BxPc-3 were obtained from the American Type Culture Collection (Manassas, VA, USA) and were cultured as previously described.50

reagents CC BY

Stock solutions of dexamethasone (25 mM, ≥98% pure, Sigma-Aldrich Chemie, Munich, Germany) and mifepristone (RU486, 50 mM, ≥98% pure, Sigma-Aldrich) were prepared in ethanol. LY2109761 (10 mM, ≥99% pure, Selleckchem, Munich, Germany) and SP600125 (20 mM, ≥99% pure, Selleckchem) were prepared in DMSO. Gemcitabine (126 mM, Lilly Deutschland, Bad Homburg, Germany) was freshly diluted in cell culture medium to prepare a 100 μM stock solution. The final concentrations of the solvents in media were 0.1% or less.

cell viability CC BY

Cell viability was measured using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) as previously described.50

facs analysis CC BY

The cells were treated with FITC-conjugated annexin-V (Life Technologies, Darmstadt, Germany), and FACS analysis was performed as previously described.50

colony assay CC BY

At 48 h after treatment, the cells were reseeded in complete medium in 6-well tissue culture plates. The colonies were evaluated 14 days later, as previously described.50

spheroid assay CC BY

Cells were cultured in human NeuroCult NS-A serum-free medium with supplements (STEMCELL Technologies, Cologne, Germany), and analysis of spheroid formation was performed as previously described.50

western blot CC BY

Western blot was performed as previously described.50 The antibodies were: mouse monoclonal against vimentin (V9, Abcam, Cambridge, UK), phosphorylated (P)-p44/42 MAPK (Cell Signaling Technology, Cambridge, UK), β-actin (Sigma-Aldrich, Munich, Germany), rabbit monoclonal against E-cadherin (24E10, Abcam), Ras (EPR3255, Abcam), Notch1 (Cell Signaling), rabbit polyclonal against TGFβ1/2/3 (Cell Signaling), SOX2 (Abcam), phosphorylated Smad2 (Life Technologies, Darmstadt, Germany), cJun and phosphorylated cJun (P-cJun) (Cell Signaling), glucocorticoid receptor (H-300, Santa Cruz Biotechnology, Heidelberg, Germany) and secondary goat anti-mouse or goat-anti rabbit IgG horseradish peroxidase-conjugated antibodies (Santa Cruz).

tgf CC BY

TGF-β1 concentrations in supernatant were determined according to the ELISA kit instructions of the manufacturer (Enzo Life Sciences, Lörrach, Germany).

sap/jnk assay CC BY

JNK/SAP kinases were immunoprecipitated from cell extract, and a kinase reaction was performed according to the instructions of the manufacturer (New England Biolabs, Frankfurt, Germany). Phosphorylated cJun was detected by western blot using a rabbit polyclonal antibody specific for cJun phosphorylated at serine 63. Mouse mAb cJun (L70B11, New England Biolabs) served as a control for equal conditions.

immunohistochemistry CC BY

Staining was performed on established cell lines growing in chambers of the Nunc Lab-Tek Chamber Slide system (Sigma-Aldrich Chemie), or on 6 μm frozen or paraffin-embedded tissue sections as previously described.50 The antibodies were mouse monoclonal against cytokeratin (Cyt)-19 (Abcam), rabbit monoclonal against E-cadherin (Cell Signaling Technology), Ki-67 (Abcam), Notch1, P-cJun (Cell Signaling), goat polyclonal against vimentin (R&D Systems, Wiesbaden-Nordenstadt, Germany), SOX2 (Santa Cruz), c-Met (Biozol, Eching, Germany) and TGFβ1 (Acris, Herford, Germany).

scratch assay CC BY

Cells (6 × 105) were seeded in 6-well plates and grown to confluence overnight. A line was scraped within the cell layer using the fine end of a 10 μl white pipette tip. Images of migrating cells were obtained with a Nikon Eclipse TS100 microscope (Nikon, Tokyo, Japan).

transwell assay CC BY

Cells (105 cells/cm2) were seeded in 24-well plates with transwell polycarbonate filters (8 mm, Corning Life Sciences, Amsterdam, The Netherlands). The number of transmigrated cells was counted under a Nikon Eclipse TS100 microscope.

emsa CC BY

Nuclear protein extracts were prepared with the NucBuster Protein Extraction Kit (Merck, Darmstadt, Germany), and the bandshift reaction was performed using the Light Shift Chemiluminescent EMSA Kit, according to the manufacturer’s instructions (Thermo Fisher Scientific, Bonn, Germany). Biotin 3' end-labeled oligonucleotides or unlabeled oligonucleotides (MWG Biotech, Ebersberg, Germany), corresponding to the vimentin AP-1 site at the promoter region -715/-676,51 were: AP-1-sense: 5'- AGGGCGCGGTGAGTCACCGCCGGTGACTAAGCGACCCCAC-3' and AP-1-antisense: 5'-GTGGGGTCGCTTAGTCACCGGCGGTGACTCACCGCGCCCT-3' (Consensus AP-1 sites are underlined).

xenotransplantation to chicken eggs CC BY

Tumor cells were transplanted to the chorioallantoic membrane52 of fertilized eggs from genetically identical hybrid LB chicken, followed by treatment and evaluation of tumor take and volume as previously described.50

alu pcr CC BY

Genomic DNA was isolated from chick tissue using the DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany), and PCR was performed with primers for human Alu sequences: Alu-sense: 5'-GTAAGAGTTCCGTAACAGGACAGC T-3' and Alu-antisense: 5'-CCCCACCCTAGGAGAACTTCTCTTT-3' (MWG Biotech, Ebersberg, Germany). Chicken GAPDH-sense: 5'-GAGGAAAGGTCGCCTGGTGGATCG-3' chicken GAPDH-antisense: 5'-GGTGAGGACAGGCAGTGAGGAACG-3' (MWG Biotech).

tumor xenotransplantation to immunodeficient mice CC BY

The primary PDA xenograft line T30 was isolated from surgical specimens, which were mechanically minced, and transplanted into the flanks of NMRI (nu/nu) mice, followed by subtransplantation as recently described.53 Cancer spheres were isolated from T30 xenografts and cultured in NeuroCult NS-A serum-free medium with supplements (STEMCELL Technologies, Cologne, Germany), followed by in vitro treatment and subcutaneous injection into the left and right flanks of 6-week-old NMRI (nu/nu) male mice, 6 animals per group. The tumor volumes (V) were determined 3 weeks after transplantation using two diameters and were calculated with the formula: V=1/2 (length × width2). The experiments were performed in the animal facilities of the University of Heidelberg after receiving approval from the authorities (Regierungspräsidium Karlsruhe, Karlsruhe, Germany).

statistical analysis CC BY

All in vitro experiments except the colony-forming assays were performed three times in triplicate (n=9), and the quantitative data are presented as the means±S.D. The colony-forming assays were performed twice in sextuplicate (n=12). For the in vivo transplantation experiments, 20 eggs (n=20) or 6 mice with tumors on both flanks per group (n=12) were used. The significance of the data was analyzed with Student’s t-test. Variances in the tumor volumes were evaluated with the Kruskal–Wallis test and the Mann–Whitney test with Bonferroni correction. *P<0.05 was considered statistically significant and **P<0.01 was considered highly statistically significant.

publisher’s note CC BY

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.