Inhibition of TPL2 by interferon-α suppresses bladder cancer through activation of PDE4D

T24 and HEK293A cells were obtained from Huaxi Hospital (Chengdu, China) and were authenticated using Short Tandem Repeat (STR) analysis. 5637 cells were purchased from Procell Life Science & Technology Co., Ltd. (Wuhan, China). T24 and 5637 cells were cultured in Roswell Park Memorial Institute Medium (RPMI) 1640 (HyClone) containing 10% (V/V) fetal bovine serum (FBS, HyClone). HEK293A cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% (V/V) FBS. All the cell lines were maintained in an incubator with 5% CO₂ at 37 °C. Forskolin (S2449), PD98059 (S1177), and roflumilast (S2131) were purchased from Selleck Chemicals (Shanghai, China). TPL2 kinase inhibitor (#19710) was purchased from Cayman Chemical (Shanghai, China). Human IFN-α-2a (Z03003) and human EGF (Z02691) were purchased from Genscript., Ltd. (Nanjing, China). Antibodies used in this study as follow: STAT1 (#41462); STAT3 (#41465); p-STAT3 (Tyr705), (#11045); JAK1 (#35530); Tyk2 (#38374); COX-2 (#21679); p- IκBα (Ser32/36), (#11152); IκBα (#29054); CREB (#21052); ERK1/2 (#40901); p-ERK1/2 (Thr202/Tyr204), (#12082); TPL2 (#33235); PDE4D (#23049); IKKα/β (#1057); IFNAR2 (#32426); IFNAR1 (#32400) were purchased from Signalway Antibody, LLC. (Nanjing, China). p-STAT1 (Tyr701), (#9167); p-JAK1 (Tyr1034/1035), (#3331); p-TYK2 (Tyr1054/1055), (#68790); p-TPL2 (Ser400), (#4491); p-CREB (Ser133), (#9198); RACK1 (#5432); P-IKKα/β (Ser176/180), (#2697) were purchased from Cell Signaling Technology, lnc. (Shanghai, China). β-tubulin (#341002) was purchased from Zen Bio Science, Ltd. (Chengdu, China).

Whole cell lysates were extracted using RIPA buffer (Beyotime Biotechnology, China) supplemented with a protease inhibitor cocktail (Sigma, Shanghai, China). The protein concentration was measured using a BCA protein assay kit (Bestbio, Shanghai, China). Cell lysates were performed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and protein bands were electrophoretically transferred onto nitrocellulose membranes. After incubation with the primary and secondary antibodies, protein bands were visualized by enhanced-chemiluminescence reaction (Amersham Biosciences, Piscataway, NJ, USA).

PDE4D overexpression was performed in T24 and 5637 bladder cancer cells using the PDE4D-pReceiver-M11 (or the control) vectors according to the manufacturer’s instructions (Genecopoeia, Rockville, MD, USA). PDE4D knockdown in T24 and 5637 cells was performed using the siRNA sequences targeting PDE4D (stQ0007397) or the non-targeting control siRNA (stQ0007397–1) according to the manufacturer’s instructions (Ribobio, Guangdong, China).

T24 cells were seeded (5 × 10³ cells/well) in 96-well plates using 100 μL medium and incubated overnight. The specific drugs used in the particular experiments were diluted in culture medium and added to cells and the plates were incubated for an additional 72 h. The cells of the control group were treated using 0.1% DMSO. Cell proliferation was measured as the absorbance at 450 nm using a cell counting kit-8 (CCK-8) according to the manufacturer’s instructions (Solarbio, China). Experiments were performed in triplicates.

The migration of T24 and 5637 bladder cancer cells were measured by trans-well assay according to the manufacturer’s instructions (Thermo Fisher, USA). Briefly, T24 and 5637 cells (1 × 10⁵ cells/ml) were added into the trans-wells (100 μL/well) and allowed to migrate under different treatments for 6 h at 37 °C. Cotton swabs were used to remove cells from the upper surface of the trans-wells, and migratory cells attached to the undersurface were stained with crystal violet (0.5%). The numbers of migrated cells (5 different fields per well) were counted using an inverted microscope.

The PGE₂ levels in cell culture supernatants and serum of mice were estimated according to the manufacturer’s instructions using human PGE₂ ELISA kit (Invitrogen, USA) and mouse PGE₂ ELISA kit (Cusabio Technology, USA), respectively.

The cAMP levels in cells and the xenograft tumor tissues were quantified according to the manufacturer’s instructions using a cAMP-Glo™ assay kit (Promega, USA). The enzyme activities of PDE4D isoforms that were immunoprecipitated from cells and xenograft tumor tissues were quantified according to the manufacturer’s instructions using a PDE-Glo™ phosphodiesterase assay kit (Promega).

The extracts of T24 cells and xenograft tumor tissues were precleared with protein A/G agarose beads (Santa Cruz Biotechnology) and incubated with primary antibodies overnight at 4 °C. Later, these samples further incubated with protein A/G agarose beads for 2 h at 4 °C. The immunoprecipitate were suspended in sample buffer and detected by performing western blotting or activity analysis.

Female BALB/c (nu/nu) nude mice (aged 5-weeks) were purchased from Dashuo Laboratory Animal Technology, Ltd. (Chengdu, China) and were kept on a 12-h day/night cycle with free access to food and water. All the experiments and procedures were performed in accordance with the National Institutes of Health guide for the care and use of Laboratory animals (NIH Publications No. 8023, revised 1978). T24 or 5637 cells (5 × 10⁶ cells/mouse) were subcutaneously injected into the flanks of mice in serum-free RPMI 1640 (100 μL). The tumor volume was calculated using the formula: volume = 1/2 (length × width²). When the tumor volume was approximately 100 to 150 mm³, mice were randomly segregated into six groups (seven mice per group) and treated using specific drugs or inhibitors. The tumor size was measured every 3rd day with a caliper. After 28 days (T24 cells) or 24 days (5637 cells), mice were sacrificed to surgically remove tumors and measure the tumor volume and weight. The serum of each mouse was collected to perform PGE₂ analysis. To determine cAMP level and for PDE4D activity analyses, the lysates of xenograft tumor tissues were extracted using SDS lysis buffer (Beyotime Biotechnology, China). The expression levels of specific proteins in the xenograft tumor tissues were analyzed using tissue microarrays by Outdo Biotech, Ltd. (Shanghai, China).

TMA chips that consisted of the bladder tumor tissue specimens (n = 126) and adjacent normal bladder tissue specimens (n = 40) were purchased from Outdo Biotech, Ltd. (Shanghai, China). Hematoxylin-eosin (H&E) staining was performed by following the routine method and IHC of TMA chips was performed using primary antibodies against PDE4D (1:150) and pTPL2 (1:150). The assignment of positive-staining score was based on the percentage of positive-staining (0% positive: 0, 1–25% positive: 1, 26–50% positive: 2, 51–75% positive: 3, and 76–100% positive: 4) and staining intensity score was based on the staining intensity (no intensity: 0, weak intensity: 1+, moderate intensity: 2+, and strong intensity: 3+). The final staining index was calculated using the formula: positive-staining score × staining intensity score. These scores were independently determined by two pathologists who were blinded to the clinical and pathological information. IHC staining of TMA using each antibody was performed in a single experiment by including the negative staining control.

The statistical significance of variations among the experimental groups was evaluated using Student t-test and one-way and two-way analyses of variance (ANOVA) tests in the analyses of cAMP levels, PGE₂ production, cell viability, and PDE4D activity. Data are represented as mean ± standard deviation (SD). Survival analysis was performed using Kaplan-Meier method and compared with the log-rank test. Wilcoxon signed rank test (unpaired comparisons) was used to determine the significant variations among the expressions of particular proteins in the bladder tumor tissues and adjacent bladder tissues. Spearman’s rank correlation coefficient was used to analyze the correlation among the expressions of specific proteins and various clinicopathological features in patients. Data are represented as mean ± standard error of mean (SEM). P < 0.05 was considered as statistically significant value. SPSS 13.0 software (SPSS, Chicago, IL, USA) was used to perform survival and correlation analyses, while Prism version 6.07 (GraphPad Software) to perform other analyses.

COX-2 plays a significant role in the bladder tumorigenesis [11]; however, the effect and mechanism of IFN-α in the regulation of COX-2 expression remains unclear. In this study, IFN-α decreased COX-2 expression in a time- and dose-dependent manner in T24 and 5637 bladder cancer cells (Fig. 1a and Additional file 1: Figure S1). COX-2 is known to be induced by activation of the NF-κB pathway that is regulated by TPL2 in addition to various other factors [9, 24]. Therefore, we examined whether TPL2 mediates the inhibitory effect of IFN-α on NF-κB activation. IFN-α inhibited the phosphorylation of TPL2, ERK, IKK α/β and IκBα and stabilized IκBα expression, indicating that IFN-α decreases COX-2 expression by inhibition of TPL2-NF-κB pathway (Fig. 1b and Additional file 2: Figure S2). To confirm the aforementioned result, bladder cancer cells were treated with TPL2 kinase inhibitor (TPL2i) and MEK inhibitor (PD98059). COX-2 expression and the phosphorylation of IKK α/β and IκBα were inhibited by TPL2i in the presence or absence of IFN-α (Fig. 1c and Additional file 3: Figure S3). Similar results were observed after PD98059 treatment (Fig. 1c). Consistent with a previous report [25], we found that the canonical JAK-STAT signaling in T24 cells was barely affected by IFN-α (Additional file 4: Figure S4A-B), indicating that IFN-α decreased the COX-2 expression through a non-canonical JAK/STAT pathway.

The cAMP/CREB pathway is another main modulator of COX-2 expression [26, 27]. The cAMP level is also regulated by IFN-α-induced MEK/ERK-mediated PDE4 activity [19]. Therefore, we further investigated whether IFN-α decreases the COX-2 expression through TPL2-mediated cAMP/CREB pathway. In T24 cells, IFN-α down-regulated the intracellular cAMP level which was further decreased by the treatment with TPL2i or PD98059 (Fig. 1d). Consistent with the aforementioned result, the CREB phosphorylation was inhibited by IFN-α in the presence or absence of TPL2i or PD98059 accompanied by the down-regulation of COX-2 expression. Conversely, forskolin (a cAMP elevator) counteracted the down-regulation of COX-2 expression induced by IFN-α and TLP2i or PD98059 (Fig. 1e). Furthermore, the reduction of COX-2 expression by IFN-α was abrogated after the treatment with epidermal growth factor (EGF) that is known to activate ERK phosphorylation [28] (Fig. 1f). To determine whether the decrease of intracellular cAMP level inhibits the bladder cancer cell growth, we used TPL2i or PD98059, and forskolin to treat the respective bladder cancer cell groups. TPL2i or PD98059 treatment reduced the bladder cancer cell viability and this reduction was attenuated by forskolin. Furthermore, the cell growth was promoted after the individual treatment of forskolin (Fig. 1g). These data suggested that IFN-α inhibited COX-2 expression through TPL2-mediated inhibition of the NF-κB activation and cAMP/CREB pathway.

In order to understand the mechanism involved in the regulation of TPL2 by IFN-α, we examined the interaction between TPL2 and IFNAR by performing co-immunoprecipitation. TPL2 was found to interact with IFNAR2 (but not IFNAR1) and this interaction was barely affected by IFN-α (Fig. 2a). IFN-α and TPL2i suppressed the level of pTPL2 that interacted with IFNAR2, whereas they did not affect the interaction between IFNAR2 and un-phosphorylated TPL2 (Fig. 2b). We previously found that RACK1 orchestrates the localization of PDE4D and protein kinase A (PKA) at IFNAR2 [29]. Therefore, we further determined the function of PDE4D in the IFN-α induced cAMP suppression. The interaction between the PDE4D and IFNAR2 was examined by performing co-immunoprecipitation. Unlike TPL2, PDE4D was recruited to IFNAR2 through RACK1 after the IFN-α treatment (Fig. 2c and Additional file 5: Figure S5A-C). Additionally, the activity of PDE4D that interacted with IFNAR2 was increased after IFN-α treatment, followed by the increase in the activity of total intracellular PDE4D. Consequently, the intracellular cAMP level was observed to be steadily reduced (Fig. 2d). These results indicated that IFN-α suppresses the cAMP level by enhancing the PDE4D activity through a dynamic interaction between PDE4D and IFNAR2. To further examine the role of TPL2-MEK/ERK pathway in the regulation of PDE4D activity, cells were treated with IFN-α and/or TPL2i or PD98059. The treatments using the individual inhibitors and combination of IFN-α and inhibitors exhibited a stronger effect on the enhancement of total intracellular PDE4D activity compared to individual IFN-α treatment (Fig. 2e). However, the variation in the PDE4D activity that leads to the interaction with IFNAR2 was very low after the treatment using individual TPL2i or PD98059 (Fig. 2f). This suggested that IFNAR2 does not recruit the reactivated PDE4D that is induced by TPL2-MEK inhibition. Collectively, these data indicated that IFN-α suppresses the cAMP level through TPL2-MEK/ERK-mediated PDE4D activity at IFNAR2.

The expression of PDE4 isoforms (PDE4A-D) is induced by PDE4 inhibitors [30]. Recently, roflumilast was reported to induce the PDE4B/4D expression in human respiratory tract epithelial cells [31]. We attempted to determine whether roflumilast induces the PDE4D expression in bladder cancer cells. The PDE4D expression was up-regulated by roflumilast treatment in a dose-dependent manner (Fig. 3a). Additionally, the PDE4D expression was significantly increased during 12–24 h after roflumilast treatment (Fig. 3b). However, the concomitant intracellular PDE4D activity and cAMP level reverted to their normal levels after 12 h and subsequently remained unchanged (Fig. 3c). These results indicated that the increased PDE4D by roflumilast did not reduce the intracellular cAMP level continuously. These observations prompted us to speculate whether IFN-α enhances the PDE4D activity that is induced by roflumilast and further reduces the intracellular cAMP level. The result showed that more PDE4D were recruited at IFNAR2 after the combination treatment of IFN-α and roflumilast (Fig. 3d). Moreover, the treatment using the combination of IFN-α with roflumilast caused a stronger effect on the enhancement of total PDE4D activity and reduction of intracellular cAMP level compared to individual IFN-α or roflumilast treatment (Fig. 3e). cAMP was reported to stimulate the proliferation in renal epithelial cells [17, 18]. Therefore, we further investigated whether the synergistic reduction of cAMP level by the combinations of IFN-α and inhibitors of TPL2-MEK-PDE4D pathway have effects on the proliferation of bladder cancer cells. The treatment using the combinations of IFN-α and inhibitors (TPL2i, PD98059 or roflumilast) exhibited a stronger inhibitory effect on the viability of bladder cancer cells than the individual IFN-α treatment (Fig. 3f and Additional file 6: Figure S6A-B). PDE4D overexpression and knockdown were then performed to further examine the role of PDE4D in the regulation of proliferation and migration in bladder cancer cells. The results showed that the overexpression of PDE4D inhibited the proliferation of bladder cancer cells and the knockdown of PDE4D, conversely, promoted the cell growth (Additional file 6: Figure S6C-E). The knockdown of PDE4D also increased the migration of bladder cancer cells and IFN-α (or TPL2 inhibitor) reduced the number of migrated cells only when the PDE4D protein was overexpressed (Additional file 6: Figure S6F-G). We then observed the morphological alterations of cells after the knockdown of PDE4D and found that only the 5637 bladder cancer cells became irregular in shape and extended tentacles (Additional file 6: Figure S6H). Furthermore, roflumilast was found to enhance the inhibitory effect of IFN-α on PGE₂ production, which plays an important role in tumorigenesis of bladder cancer (Fig. 3g). Taken together, these data suggested that the induction of PDE4D expression by roflumilast synergized with IFN-α activity to reduce the intracellular cAMP level and potentiates the antiproliferative effect of IFN-α on bladder cancer cells.

To investigate whether roflumilast potentiates the anti-tumor effect of IFN-α in vivo, we used a tumor xenograft model by injecting human 5637 bladder cancer cells into BALB/c nude mice (see materials and methods). The combination treatment of IFN-α and roflumilast (5 mg/kg/day) drastically suppressed the tumor growth compared to individual IFN-α treatment (Fig. 4a-c). We further investigated whether roflumilast (5 mg/kg/day) potentiates the anti-tumor effect of IFN-α through the cAMP reduction. The lysates obtained from xenograft tumor tissues were used to detect the cAMP level. The cAMP level in tumor lysates was significantly reduced after the combination treatment of IFN-α and roflumilast compared to individual IFN-α or roflumilast treatment (Fig. 4d). Next, the PDE4D expression in tumor tissues was evaluated by western blotting. IFN-α did not affect the PDE4D expression; however, roflumilast treatment induced the PDE4D expression when used individually or in combination with IFN-α (Fig. 4e). The results from T24 tumor xenograft model also showed that the combination treatment of IFN-α and roflumilast (5 mg/kg/day) potentiated the anti-tumor effect of IFN-α through the cAMP reduction (Additional file 7: Figure S7A-D). Consistent with the in vitro result, the PDE4D activity was increased in the individual IFN-α treatment, and in the treatment of roflumilast combined with IFN-α further enhanced the PDE4D activity (Additional file 7: Figure S7E). Moreover, we examined the PGE₂ production in the mice serum. IFN-α as well as roflumilast individual treatment exhibited inhibitory effects on the PGE₂ production; IFN-α and roflumilast combination treatment further reduced the PGE₂ production than either of the individual treatments (Additional file 7: Figure S7F). The PDE4D expression and TPL2 phosphorylation in T24 tumor tissues were evaluated by immunohistochemistry (IHC). The PDE4D expression increased when roflumilast was used individually or in combination with IFN-α (Fig. 4f). Furthermore, high pTPL2 level was observed in tumor tissues and was inhibited by IFN-α. However, roflumilast did not affect both the TPL2 phosphorylation and IFN-α induced inhibition of TPL2 phosphorylation (Fig. 4g). These data suggested that the induction of PDE4D expression by roflumilast potentiated the anti-tumor effect of IFN-α through the elevated PDE4D expression and reduction of intracellular cAMP.

To explore the clinical importance of the TPL2 phosphorylation and PDE4D expression levels in human bladder cancer development, the tissue microarray chips that consisted of MIBC specimens (n = 126) were used to perform IHC analysis. The bladder cancer sections and the adjacent bladder normal tissues were obtained from the patients who underwent surgical resection. The H&E staining was performed using the routine method (Additional file 8: Figure S8) and the IHC staining results were analyzed using the staining index (see materials and methods). Statistically, the PDE4D expression was found to be significantly lower in the bladder tumor tissues than that in adjacent normal bladder tissues (P = 0.009) (Fig. 5a-c, Additional file 9: Figure S9, Additional file 10: Table S1), and the low PDE4D expression was positively correlated to the poor prognosis (Fig. 5d).

Unlike PDE4D expression, the level of TPL2 phosphorylation was found to be significantly higher in the bladder tumor tissues than that in the adjacent bladder normal tissues (P = 0.003) (Fig. 5e-g, Additional file 11: Figure S10, Additional file 10: Table S1), and the high level of pTPL2 was positively correlated to poor prognosis (Fig. 5h). The survival curves that correspond to each of the clinicopathologic features were also analyzed as the basic information about the tissue microarray chips (Additional file 12: Figure S11). Additionally, the low PDE4D expression and high pTPL2 levels were found to positively correlate with the age and TNM stages of bladder cancer patients (Additional file 10: Table S2). To further validate our TMA-IHC results, we analyzed the data of bladder cancer patients derived from The Cancer Genome Atlas (TCGA) database. In TCGA data, a significantly low (cancer vs. normal) PDE4D expression was observed in bladder cancer patients (Fig. 5i) and correlated with their poor prognosis (Fig. 5j). Moreover, the data also revealed that the PDE4B expression was significantly lower in the bladder cancer tissues than that in the bladder normal tissues (Additional file 13: Figure S12B); however, variations were not observed in the cases of PDE4A and PDE4C expressions (Additional file 13: Figure S12A and C). The data derived from Oncomine database indicated similar results of PDE4D expression (Additional file 14: Figure S13A-D); however, regarding the un-phosphorylated TPL2 expression, variations were not observed between the bladder mucosal and bladder tumors (Additional file 14: Figure S13E). Together, these results suggest that the low PDE4D expression and high pTPL2 levels are correlated to the MIBC development and poor prognosis in MIBC patients.