Tunicamycin specifically aggravates ER stress and overcomes chemoresistance in multidrug-resistant gastric cancer cells by inhibiting N-glycosylation

Adriamycin (Adr), vincristine sulfate (Vcr), hydroxychloroquine Sulfate (HCQ) and brefeldin A (BFA) were purchased from Selleck Chemicals (Houston, TX, USA). Tunicamycin (Tu, ab120296) was purchased from Abcam (Cambridge, MA, USA). Thapsigargin (Tg, #12758) was from Cell Signaling Technology (Danvers, MA, USA). Tu, Tg and BFA were dissolved in DMSO and then diluted with RPMI-1640 medium for subsequent experiments. Additionally, Adr, Vcr and HCQ were dissolved in PBS and diluted with RPMI-1640 medium for further experiments. Anti-rabbit IgG (#7074), anti-mouse IgG (#7076) and antibodies against cleaved caspase-3 (#9664), cleaved caspase-7 (#8438), cleaved PARP (#5625), β-actin (#8457), PERK (#5683), IRE1α (#3294), Bip (#3177), XBP1s (#12782), PDI (#3501), CHOP (#2895), L1CAM (#89861), and TIMP1 (#8946) were from Cell Signaling Technology. Goat anti-mouse IgG Alexa Fluor® 488 (ab150113), donkey anti-rabbit IgG Alexa Fluor® 594 (ab150076) and antibodies against p-IRE1 (ab124945), LC3B (ab192890) and P62 (ab56416) were from Abcam.

The human gastric adenocarcinoma cell line SGC7901 was obtained from the Academy of Military Medical Science (Beijing, China). The MDR derivatives SGC7901/ADR (ADR) and SGC7901/VCR (VCR) were developed and maintained as previously described [14]. Other gastric cancer cell lines (BGC823, MKN45, AGS) and the human immortalized gastric epithelial cell line (GES) were from the cell bank of our lab. All cell lines were cultured in RPMI-1640 medium with 10% FBS, 100 U/ml penicillin sodium and 100 mg/ml streptomycin sulfate at 37 °C in a humidified atmosphere containing 5% CO₂.

Cell viability was measured with Cell Counting Kit-8 (DOJINDO, Kumamoto, Japan) according to the manufacturer’s instructions. Cells were seeded in 96-well plates at a density of 3 × 10³ cells/well with five replicates and incubated overnight. We then treated the cells with the appropriate drugs. After treatment, the original medium was replaced with a mixture of 10 μl CCK-8 reagent and 100 μl fresh medium. Cells were incubated for another 4 h at 37 °C. Finally, the absorbance of each well was measured by a microplate reader (Varioskan Flash, Thermo Scientific) at 450 nm. Every experiment was performed in triplicate.

GC cells were grown in 6-well plates and treated with the appropriate chemicals for the desired period. After treatment, the cells were harvested and then resuspended in binding buffer. Afterwards, the cells were incubated with Annexin V-FITC and PI (Beyotime, China) according to the manufacturer’s instructions. Subsequently, the cells were analyzed with a flow cytometer (FACScan, BD Biosciences, USA). All experiments were performed in triplicate.

Protein expression was determined by western blot analysis. The cells were grown in 6-well plates and then lysed with RIPA buffer (Beyotime, Shanghai, China) containing a complete® protease inhibitor cocktail (Roche, Manheim, Germany). The protein concentration was determined using a BCA kit (Thermo Scientific, USA). Protein samples (30 μg) were separated on SDS-PAGE gels and then transferred to PVDF membranes (Thermo Scientific, USA). Afterwards, the membranes were blocked in 5% nonfat milk and incubated with antibodies as described in our previous work [14]. Protein signals were detected by the ChemiDoc XRS+ imaging system (BIO-RAD, CA, USA) using the ECL reagent (Millipore, MA, USA) and quantified by Image Lab software (BIO-RAD). All experiments were performed in triplicate.

The cell immunofluorescence assay was performed with a CST immunofluorescence application solutions kit (#12727) according to the standard protocol. After incubation with antibodies, the cells were counterstained with ProLong® Gold Antifade Reagent with DAPI (#8961). Then, the cells were visualized and photographed using an Olympus confocal laser scanning microscope FV1200.

Changes in UPR-related genes were determined by the RT² Profiler™ PCR Array-Human Unfolded Protein Response (PAHS-089Z, Qiagen). Cells were grown in 6-well plates and subjected to treatments for 48 h before being harvested for further analysis. Total RNA was isolated using the Qiagen RNeasy Mini Kit, and cDNA was synthesized using the Qiagen RT² First Strand Kit. The subsequent procedures followed the instructions provided by Qiagen. The reactions were performed with Bio-Rad CFX96. The web-based analysis tool of Qiagen was used to interpret the results. In addition, GO analysis of the identified genes was conducted with the PANTHER classification system.

The data are expressed as the mean ± SD. Statistical analysis was performed using one-way ANOVA in GraphPad Prism (version 7.0). Differences were considered statistically significant when P was < 0.05.

To explore the effects of Tu on GC cells, we identified the dose-response curves of GES and 6 gastric cancer cell lines, including two derivative MDR cell lines. In general, the cytotoxicity assay indicated that GES was more sensitive to Tu than any other cell line, demonstrating the more toxic effects of Tu on normal tissues. Moreover, as for the MDR cell models, MDR cells (SGC7901/ADR and SGC7901/VCR) showed higher sensitivity to Tu than the parental cells (SGC7901) (Fig. 1a).

Given that Tu can activate ER stress and that excess ER stress leads cells to death, we determined the expressions of ER stress-associated proteins in cells without treatment by western blot (WB) analysis. The results showed that PERK, p-IRE1, IRE1 and XBP1s were all upregulated in MDR cells compared to those in the parental cells, except for Bip. Bip was slightly downregulated in SGC7901/VCR compared to SGC7901. When the expressions of all the proteins were considered comprehensively, we believed that compared to the parental cells, MDR GC cells displayed higher levels of basal ER stress (Fig. 1b), which could be the determinant factor driving their greater susceptibility to Tu. That deserves further exploration.

We treated the two MDR cell lines (SGC7901/ADR and SGC7901/VCR) and their parental cell line SGC7901 with Tu (0–1 μg/ml) for 24, 48 and 72 h to obtain more detailed information about their responses to Tu. Generally, we found that Tu preferentially reduced the viability of MDR cells in a time/dose-dependent manner (Fig. 2a). Furthermore, after exposure to Tu for 48 h, MDR cells exhibited more significant enhancements of ER stress-associated proteins (PERK, IRE1, Bip, CHOP) with increasing doses, as demonstrated by WB analysis (Fig. 2b) and immunocytochemical staining (Fig. 2c). Based on the above results, we believed that Tu could specifically exacerbate basal ER stress in MDR cells, which made them more vulnerable to Tu treatment.

Because MDR GC cells were more susceptible to Tu, we next explored whether Tu could overcome chemoresistance in GC cells. Cell viability assays indicated that cotreatment of Tu (0.2/0.4/0.8 μg/ml) with Adr could significantly reduce cell survival in MDR cells at any dose of Tu, whereas only the large concentration of Tu (0.8 μg/ml) could markedly enhance cell death in the parental cell line (Fig. 3a). To further identify the interaction between Tu and Adr, we calculated the Combination index (CI) values (Additional file 1: Table S1) for GC cells using CompuSyn software. Disappointingly, despite the fact that combined therapy with Tu and Adr could induce more cell death than monotherapy with Adr, the combination did not indicate synergistic effects for MDR cells. Despite this, dual therapy with Tu and Adr triggered significantly more apoptosis of GC cells than monotherapy with Adr (Fig. 3b). Taken together, these results showed that Tu could decrease the chemoresistance of MDR cells to some extent.

To elucidate the underlying mechanism, we performed the PCR array for SGC7901 and SGC7901/ADR after treatment with Adr alone or Adr and Tu. The results showed that in SGC7901/ADR, the differentially expressed genes between the CT (combined therapy with Tu and Adr) group and the Adr group were mainly involved in the UPR and apoptosis signaling pathways (Fig. 3c), which was further evidenced by GO analysis of these identified genes (Additional file 2: Figure S1). However, we found few changes in such genes in SGC7901 between the CT group and Adr group (Additional file 3: Figure S2).

Moreover, at the protein level, we carried out WB and immunofluorescence (IF) experiments to explore the changes in UPR- and apoptosis-related proteins in SGC7901 and SGC7901/ADR after monotherapy (Adr) or CT (Adr and Tu). Our results indicated that Bip (GRP78), an important modulator of ER stress, was significantly increased in the CT group of SGC7901/ADR compared to the Adr group. We also identified the notable up-regulation of CHOP (C/EBP homologous protein), a key player in UPR-mediated apoptotic pathway, in the CT group of SGC7901/ADR. In addition, the effects of CT in SGC7901/ADR were further evidenced by increases in cleaved caspases and PARP (Fig. 3d). Furthermore, in SGC7901/ADR, IF assays of CHOP, Cl-caspase3 and Cl-PARP also demonstrated marked enhancements in the CT group compared to the Adr group (Fig. 3e). However, the increases in these proteins were not as significant in the CT group of SGC7901 (Fig. 3d and Additional file 4: Figure S3). Generally, our results demonstrated that CT could induce more apoptosis by aggravating ER stress in MDR GC cells, thus reversing chemoresistance.

As already reported, Tu potently induces ER stress by inhibiting the N-linked glycosylation of glycoproteins. However, thapsigargin (Tg) functions as another ER stress inducer by disrupting Ca²⁺ homeostasis in the ER. Hence, to determine whether the specific effects of Tu on MDR GC cells were mainly due to its inhibition of N-glycosylation or its role as an ER stress inducer, we attempted to mimic its particular influences on MDR cells using Tg.

First, we compared the responses of SGC7901, SGC7901/ADR and SGC7901/VCR to Tg and Tu, respectively, across large dosage ranges (0.125–16 μg/ml for Tg; 0.25–16 μg/ml for Tu). The results of CCK-8 assays showed that the two MDR cell lines SGC7901/ADR and SGC7901/VCR were less sensitive to Tg compared to the parental cell line, which was opposite to the effects of Tu on these 3 cell lines (Fig. 4a).

In addition, to verify its function of inducing ER stress, we incubated the 3 cell lines with Tg (2 μg/ml) for 48 h and subsequently determined the expressions of UPR-related proteins by WB. The results demonstrated that Tg could markedly activate ER stress, especially in SGC7901 (Fig. 4b).

Furthermore, we treated the 3 cell lines with dual therapy of Adr and Tg (0.5/2 μg/ml) to explore whether Tg could potentiate the chemosensitivity of MDR cells to Adr. However, the cell viability assays indicated that Tg notably increased chemotherapy-induced cell death of SGC7901 at any dose but had little influence on the survival of MDR cells undergoing Adr treatment (Fig. 4c).

In addition, we employed another glycosylation inhibitor to further demonstrate the significance of glycosylation inhibition in the specific effects of Tu on MDR cells. First, we further illustrated the inhibitory effects of Tu on N-glycosylation by investigating example glycoproteins (Additional file 5: Figure S4a). L1CAM and TIMP1 are two identified glycoproteins modified with N-glycans, as recorded in the Uniprot database. Our previous study discovered that the N-glycosylation status of L1CAM and TIMP1 dramatically changed after the acquisition of chemoresistance in GC cells [14]. Thus, we explored whether Tu could inhibit the glycosylation of these two glycoproteins in GC cells. The WB results indicated that the expressions of L1CAM and TIMP1 decreased after Tu treatment (0.8 μg/ml) for 48 h in GC cells, particularly MDR cells (Additional file 5: Figure S4a). In addition, as indicated by Uniprot and our work [14], L1CAM possesses more glycosites and is modified with more N-glycans. As a result, the expression of L1CAM is more vulnerable to glycosylation inhibition. Accordingly, the WB bands of L1CAM showed an obvious downward shift in MDR GC cells after Tu treatment (Additional file 5: Figure S4a). These results demonstrated that Tu could potently inhibit the N-glycosylation of glycoproteins in GC cells. Brefeldin A (BFA) can block protein transport from the ER to the Golgi apparatus and thus inhibit the glycosylation process. Therefore, we explored the effects of BFA on GC cells. The results demonstrated that MDR cells were more vulnerable to BFA treatment (Additional file 5: Figure S4b), and BFA could inhibit the glycosylation of L1CAM and activate ER stress in GC cells (Additional file 5: Figure S4c). More importantly, BFA significantly increased the chemosensitivity of GC cells to Adr (Additional file 5: Figure S4d). In summary, our results indicated that BFA could mimic the effects of Tu on MDR cells to some extent.

In conclusion, our results demonstrated that compared to Tg, BFA could imitate the effects of Tu on MDR GC cells, which indicated that the inhibition of N-glycosylation by Tu determined its specific impacts on MDR cells rather than its role as an ER stress inducer.

At present, it is widely accepted that ER stress is a potent inducer of autophagy [29‐31]. As Tu could significantly activate ER stress in GC cells (as evidenced above), we sought to determine whether Tu-induced ER stress triggered autophagy and the exact role of autophagy in the effects of Tu on GC cells.

We incubated SGC7901 and SGC7901/ADR with Tu (0.8 μg/ml) for 48 h and then subjected their cell lysates to WB for detection of LC3II/LC3I and P62, which are common markers of autophagy. The results indicated a marked increase in the ratio of LC3II/LC3I and a corresponding decrease in P62 in SGC7901/ADR after Tu treatment but no obvious changes in SGC7901 (Fig. 5a). In addition, the fluorescence of LC3II was also markedly enhanced in SGC7901/ADR after 48 h treatment of Tu, as shown in Fig. 5b.

We used hydroxychloroquine (HCQ), a lysosome inhibitor, to block autophagic flux, and then identified whether autophagy inhibition impacted the effects of Tu on GC cells. Firstly, we chose 25 μM HCQ for further experiments because HCQ effectively inhibited Tu-induced autophagy and exerted little effect on the viability of GC cells at 25 μM (Additional file 6: Figure S5). Fig. 5c showed that HCQ significantly decreased the cell viability of GC cells when combined with the dual therapy of Adr and Tu. Moreover, we discovered that HCQ markedly increased the apoptosis induced by dual therapy of Adr and Tu (Fig. 5d).

Furthermore, WB analysis showed that after the addition of HCQ, the expressions of cleaved caspases and PARP increased compared to those in the CT (Adr and Tu) group in GC cells. Besides, we also found a substantial increase in the LC3II/LC3I ratio in GC cells with triple therapy (Adr, Tu and HCQ) (Fig. 5e), which might have resulted from the simultaneous autophagy induction by drugs and autophagic flux blockade by HCQ.

After comprehensive analysis, we found that Tu could significantly trigger autophagy in MDR cells and that HCQ enhanced the combined effects of Adr and Tu by inhibiting Tu-induced autophagy. However, in SGC7901, Adr activated autophagy, and HCQ promoted the inhibitory effects of Adr and Tu by blocking Adr-induced autophagy.

In summary, Tu-induced autophagy did not contribute to the effects of Tu on MDR GC cells but rather antagonized its chemosensitizing impacts. Autophagy inhibition could further potentiate the cell death of GC cells induced by CT with Adr and Tu.