Up-regulation of Nrf2 in human PDAC and predicts poor prognosis
To see the clinical significance of Nrf2, we determined Nrf2 expression in PDAC patients and found that it was markedly higher in PDAC as compared to the adjacent noncancerous tissue (n=4) (Fig. 1a), H&E analysis of the tumors also showed a reduction in cell density in cancer tissue than in the adjacent tissue (Supplementary Fig. S1a). A large-scale dataset analysis by using GEPIA (http://gepia.cancer-pku.cn) and Oncomine database confirmed that Nrf2 was overexpressed in PDAC tissues than that in normal tissues (Fig.1b & Supplementary Fig. S1b). It also found that the patients whose tumors were clinically staged later had higher expression levels of Nrf2 (Fig.1c). In addition, the protein levels of Nrf2 in human PDAC cells (Miapaca-2, Capan-2, PANC-1) and normal human gastric epithelial cells (GSE-1) and human fibroblast cells (Hs27) were detected using western blot, and it was found that Nrf2 protein levels in the cancer cells were significantly higher than those in the normal cells (Fig.1d). Moreover, Kaplan-Meier survival analysis indicated that high expression of Nrf2 in PDAC tissues was associated with a shorter overall survival of PDAC patients (Fig. 1e-f). We also detected the influence of GEM on the expression of Nrf2 and found that GEM treatment significantly upregulated the Nrf2 expression in all PDAC cell lines in protein and mRNA levels (Fig.1g-i & Supplementary Fig. S1c). Taken together, these data suggest that there is a positive role of Nrf2 in PDAC progression, and unfortunately, GEM treatment resulted in the activation of Nrf2 signaling pathway.
Knockdown of Nrf2 enhances the chemosensitivity of GEM in PDAC
Since Nrf2 is over-expressed in cancer tissues, and Nrf2 over-expression is known to be associated with chemoresistance. A reduction in Nrf2 expression should help sensitize cancer cells to chemotherapeutic drugs. We hypothesized that deletion of Nrf2 could lead to the chemosensitivity of GEM in PDAC. To verify this hypothesis, we ablated Nrf2 in Miapaca-2 and PANC-1 cells using transfection with a Nrf2-lentivirus construct. As shown in Fig. 2a-b, a pool of Nrf2 shRNA exhibited strong Nrf2 elimination efficacies in both the mRNA expression and protein levels. We then tested the sensitivity of GEM in Miapaca-2 LV-shNrf2 and PANC-1 LV-shNrf2 cells. As predicated, the Miapaca-2 and PANC-1 cells with Nrf2 knockdown were more sensitive to GEM than their LV-shCtrl cells (Fig. 2c). In addition, upon GEM treatment, the protein levels of Nrf2, Keap1, HO-1, NQO1, γGCSm, AKR1B10, MRP1 and MRP5 were markedly decreased in Miapaca-2 and PANC-1 LV-shNrf2 cells when compared with cells transfected with empty vector (Fig. 2d-e). Similar results were also obtained in Capan-2 cells, which are shown in Figure S2A-E. These results strongly indicate that Nrf2 contributes substantially to PDAC progress, and deletion of Nrf2 successfully sensitizes PDAC cells to GEM.
Nrf2 deletion suppresses the orthotopic pancreatic tumor growth and improves the chemosensitivity of GEM in vivo
Moreover, similar results that Nrf2 knockdown enhanced the sensitivity of GEM were also obtained in vivo. Miapaca-2 LV-shCtrl cells and Miapaca-2 LV-shNrf2 cells were directly injected into the tail of the pancreas of the nude mice to establish an orthotopic model of pancreatic cancer, followed by treatment with or without GEM (Fig. 2f-j). As shown in Fig. 2g, Miapaca-2 LV-shNrf2 group showed an inhibitory tumor effect as compared with LV-shCtrl group, while treatment with GEM presented better inhibitory effects on tumor growth than Miapaca-2 LV-shCtrl group. No obvious changes in the body weight were found in the animals (Fig. 2h). In addition, LV-shNrf2 group had lower protein levels of Nrf2, HO-1, NQO1, AKR1B10, γGCSm, MRP1 and MRP5 when compared with LV-shCtrl (Supplementary Fig. S2f-g). The hematoxylin & eosin (H&E) staining of excised tumors showed a significantly reduction in cell density in GEM treated LV-shNrf2 group compared with LV-shCtrl group (Fig. 2k). Furthermore, IHC analysis revealed that knockdown of Nrf2 markedly reduced the positive cells of the proliferation marker of Ki-67, and NQO1 and Nrf2 (Fig. 2l & Supplementary Fig. S2j). These results were indicative that knockdown of Nrf2 could inhibit the orthotopic pancreatic tumor growth by improving the sensitivity of GEM to PDAC in vivo.
Nrf2 activation augments the chemoresistance of GEM
To further verify the effect of Nrf2 on the chemosensitivity of GEM, we pharmacologically activated Nrf2 using Nrf2 inducer tert-butylhydroquinone (tBHQ) in vitro and in vivo. As shown in Fig. 3a, treatment with tBHQ increased the cell viability and enhanced the chemoresistance of GEM in all PDAC cell lines. Immunoblotting analysis unraveled that overexpression of Nrf2 conspicuously activated Nrf2 and Nrf2-target gene expression (Fig. 3b). Additionally, we also established an orthotopic mouse model using Capan-2 cells (Fig. 3c). In this model, intraperitoneal injection of tBHQ (2.5 mg/kg) significantly increased the tumor volume and tumor weight, as compared with the vehicle control group and the GEM treatment group (Fig. 3f-g). Moreover, tBHQ treatment also suppressed the anti-tumor effect of GEM, as compared with the GEM alone, leading to larger volume and weight of the xenograft tumor (Fig. 3d). The increases in the body weight seen in the tBHQ and control mice might be the result of more rapid tumor growth in the tBHQ and control group (Fig. 3e). In addition, immunohistochemical analysis revealed that tBHQ-induced activation of Nrf2 significantly accentuated the numbers of Ki-67, Nrf2 and NQO1 positive cells (Fig. 3h-i). These results amply demonstrated that activation of Nrf2 promoted the chemoresistance on GEM, thereby augmenting the tumor growth in the Capan-2 cells-derived orthotopic xenograft.
BD potentiates the sensitivity of PDAC cells to GEM
We then performed a mobility assay to evaluate the effect of BD in the PDAC cell lines (Fig. 4a for BD structure). BD significantly inhibited the proliferation of Miapaca-2, Capan-2 and PANC-1 cells with similar IC50 values (Fig. 4b). Since BD possesses potent inhibitory effect on PDAC cell proliferation, we speculated that BD could have a synergetic effect with GEM in PDAC in vitro. The MTT assay showed that BD effectively sensitized the Miapaca-2, Capan-2 and PANC-1 cells to GEM treatment (Fig. 4c). In addition, the combination of BD and GEM exerted synergetic inhibitory effect on the colony formation of PDAC cells (Fig. 4d-e), and increased the number of cells undergoing apoptosis, when compared with GEM treatment alone through activated the caspase-3, caspase-9 and PARP protein levels in PDAC cells (Fig. 4f-h). BD also augmented the GEM effect on the ROS accumulation following 24 h treatment (Fig. 4i). These results unequivocally indicated that BD could enhance the chemosensitivity of GEM and accentuate the effect of GEM on the ROS accumulation in Miapaca-2, Capan-2 and PANC-1 cells, thereby leading to more cellular apoptosis.
BD selectively inhibits the Nrf2 pathway
As up-regulation of Nrf2 protein level was observed in PDAC cells, we then evaluated the inhibitory effect of BD on the protein level of Nrf2. The expression level of Nrf2 was significantly attenuated in Miapaca-2 and Capan-2 cells after treatment with BD (0.5, 1, 1.5 µM) (Fig. 5a-b). An inhibitory effect of BD (1.5 µM) on Nrf2 protein level in PDAC cells was also observed in the indicated time. It should be noted that the protein level of Nrf2 began to decrease from 6 h onward after BD treatment (Fig. 5c-d). In addition to Nrf2, the protein levels of Nrf2-target genes, including Keap1, HO-1, NQO1, AKR1B10 and γGCSm, were also decreased in a dose and time-dependent manners. The above experimental results amply suggested that BD could effectively inhibit Nrf2 signaling in PDAC cells.
To elucidate the possible mechanism underlying the inhibitory effect of BD on Nrf2 signaling pathway, immunofluorescence staining was applied to observe the translocation of Nrf2 in PDAC cells. As shown in Fig. 5e, the dissociation from the membrane into the cytoplasm was not found after BD treatment. Moreover, an endogenous Nrf2 immunostaining assay further corroborated the inhibitory effect of BD on the protein level of Nrf2. We also assessed the relationship between BD and Nrf2 using molecular docking. The results revealed that BD could bind to Keap-1:Nrf2 (2flu) at Keap-1 amino acid chain to form 3 hydrogen bonds at valine 369 (O-H, 2.394Å, and HN-O, 2.079 Å) and valine 420 (HN-O 2.537Å), respectively, with the binding energy of -8.18 kcal/mol (Fig. 5f). Based on the above experimental and molecular docking results, we believe that BD is a Nrf2 inhibitor. Interestingly, we also found that brusatol (also a quassinoid compound), a known Nrf2 inhibitor, also bound to Keap-1:Nrf2 (2flu) in a fashion similar to that of BD (Supplementary Fig. S3a).
BD promotes the ubiquitin-proteasome dependent degradation of Nrf2
As BD sensitized the PDAC cells to GEM through the downregulation of Nrf2 expression, we further explored the mechanism on how BD regulates Nrf2 in PDAC cells. First, the effect of BD on the transcription of Nrf2 was determined. RT-PCR results showed that Nrf2 mRNA level remained unchanged after BD treatment in the two cell lines (Fig. 6a), suggesting that BD did not affect the transcription or mRNA stability of Nrf2. In the presence of BD, the half-life of the Nrf2 protein was reduced in both Miapaca-2 and Capan-2 cells, as measured by cycloheximide (CHX) chase assays (Fig. 6b-c), indicating the accelerated degradation of the Nrf2 protein. We next evaluated the effect of BD on the Nrf2 ubiquitination level using western blotting, and the results showed that the ubiquitination level was accentuated upon BD treatment in PDAC cells (Fig. 6d). Furthermore, we employed MG-132 to inhibit 26 s proteasome activity in Miapaca-2 cells. We found that the BD-induced down-regulation of Nrf2 was significantly attenuated. Similar results were observed in Capan-2 cells (Fig. 6e-f). The above experimental results conspicuously indicate that BD inhibits Nrf2 through promoting its ubiquitin-proteasome degradation rather than decreasing the synthesis of the Nrf2 protein.
Nrf2 knockdown plus BD enhances the chemosensitivity of GEM in human PDAC cells
To evaluate whether Nrf2 was involved in the BD-mediated sensitization of GEM in PDAC cells, we used Nrf2-depleted Miapaca-2 and PANC-1 cells to verify this hypothesis. As predicated, the knockdown of endogenous expression of Nrf2 led to significant sensitization of PDAC cells to cell death upon treatment with BD or GEM. On the other hand, co-treatment with BD and GEM exerted slight synergetic effects (Fig. 7a). Similar results were also found in Capan-2 cells (Supplementary Fig. S4a). To understand the potential mechanism underlying the enhanced antitumor effect of BD after Nrf2 knockdown, we assessed the protein level of Nrf2 using western blot. As shown in Fig. 7b & Supplementary Fig. S4b, treatment with BD following silencing Nrf2 more significantly attenuated the expression of Nrf2 and its downstream protein than that in the cells without silenced Nrf2. Moreover, co-treatment with BD and GEM resulted in a more potent inhibitory effect. These results suggested that BD efficiently augmented the chemosensitivity of GEM by downregulating the Nrf2 expression, and the anti-PDAC effect was more potent when Nrf2 was depleted in the PDAC cells.
BD suppresses the orthotopic PDAC tumor growth and sensitizes the efficacy to GEM by inhibiting Nrf2 pathway
To confirm the enhanced chemotherapeutic effect of BD in PDAC was Nrf2 dependent, we established the orthotopic mouse models using Miapaca-2 LV-shCtrl cells and Miapaca-2 LV-shNrf2 cells, respectively (Fig. 7c). No significant difference in the body weight was observed among different treatment groups (Fig. 7d). As depicted in Fig. 7e-f, Miapaca-2 LV-shNrf2 cells derived-orthotopic xenografts had a smaller tumor weight, as compared with the Miapaca-2 LV-shCtrl group following BD and GEM treatment, indicating that the co-treatment elicited a greater inhibitory effect. Analysis of excised tumors showed a similar reduction in cell density (H&E) in the Miapaca-2 LV-shNrf2 group compared with Miapaca-2 LV-shCtrl group. In addition, lower Nrf2 expression in the Miapaca-2 LV shNrf2-derived xenograft after BD treatment was confirmed by IHC staining. Furthermore, BD, GEM and their combination treatment also significantly attenuated the number of the Ki-67 positive cells in the LV-shNrf2 group when compared with that in the LV-shCtrl group (Fig. 7g-h). Taken together, our experimental results unambiguously indicated that BD exerted potent inhibitory effects on pancreatic tumor growth in the orthotopic PDAC mouse model, and the underlying mechanism involves the suppression of Nrf2 expression, thereby rendering the PDAC cells more sensitive to GEM.
BD enhances the anti-tumor activity of GEM in KPC mouse model
To further corroborate the therapeutic effect of BD in PDAC, we assessed whether BD could also sensitize the anti-PDAC effect of GEM in KPC transgenic mouse model. KPC mice at 6-week-old were treated with vehicle, BD (daily), GEM (twice a week) and their combination at indicated time (Fig. 8a). As expected, treatment with the combination of BD and GEM exerted more potent anti-PDAC effect in KPC mice when compared with GEM treatment alone. Accordingly, KPC mice treated with BD had a decreased tumor weight (Fig. 8b-d) when compared to the vehicle-treated control. We further evaluated the anti-fibrotic role of BD in KPC mice using Masson’s trichrome staining. The Masson’s trichrome-stained area was significantly reduced following BD, GEM or their combination treatment (Fig. 8e-g). In addition, we also found that BD, GEM or their combination treatment markedly down-regulated the protein levels of Nrf2 and Ki67 in the tumor tissues of KPC mice (Fig. 8i).
We also evaluated the possible chronic liver and renal toxicities of BD treatment using blood biochemical analysis. At the end of BD, GEM or their combination treatment, blood samples were collected, and the serum AST and ALT levels measured to determine the liver toxicity of the KPC mice. As shown in Fig. 8h, the AST and ALT levels in BD, GEM and their combination group were similar to that of the control group. Likewise, we evaluated the renal toxicity of BD, GEM and their combination by measuring the creatinine level. No overt renal toxicity was observed among these groups. Moreover, no abnormality was seen in liver and kidney tissues as examined histopathologically (Supplementary Fig. S5a-b). Furthermore, BD induced no treatment-related abnormality concerning the gross anatomy and histological morphology (Supplementary Fig. S5c). Collectively, these data suggested that BD reduced the tumor burden and enhanced the chemotherapeutic effect of GEM in KPC mice, while it exerted no overt systemic toxicity.