Activation of the PERK-ATF4 pathway promotes chemo-resistance in colon cancer cells

Colon cancer is a major health problem worldwide. While chemotherapy remains a main approach for treating late-stage colon cancer patients, most, if not all, of them will develop drug resistance and die of uncontrollable disease progression eventually. Therefore, identification of mechanism of drug resistance and development of overcoming strategy hold great significance in management of colon cancer. In this study, we discovered that activation of the PERK branch of the unfolded protein response (UPR) pathways is required for colon cancer cells to survive treatment of 5-Fluorouracil (5-FU), one of the first-line chemotherapeutics for late-stage colon cancer patients. Genetic and pharmacological inhibition of PERK or its downstream factors greatly sensitize colon cancer cells to 5-FU. Most importantly, in vivo use of PERK inhibitor synergizes with 5-FU in suppressing the growth of colon cancer cells in mouse models. In summary, our findings established a promising way to overcome resistance to chemotherapy in colon cancer.

Western blot. Cultured cells were lysed on ice with cold RIPA buffer plus complete protease inhibitor cocktail (Roche Applied Science). Cell lysates were clarified by centrifugation at 12000 g for 10 min, and protein concentration was determined by the BCA Reagent. Lysates were separated on NuPAGE 4-12% Bis-Tris gel electrophoresis, proteins were then transferred to nitrocellulose membrane and immunoblotted with the fol- Flow cytometry analysis. Flow cytometry analysis was performed according to the manufacturer's protocol (BD Biosciences), with at least 10000 live events captured per analysis. Cells after treatment as indicated were treated with RNase and stained with propidium iodide (5 μg/ml) (BD Bioscience). Percentage of sub-G0/G1 cells was measured to indicate apoptotic cell death.
In vivo tumor growth. 2 × 10 6 SW620 cells were injected subcutaneously into 6-8-wk-old female NOD/ SCID mice. After tumor reaching 60-80 mm 3 , animals were treated with PBS by IP injection, 5-FU (20 mg/kg) by IP injection, PERK inhibitor (10 mg/kg) by oral administration, or a combination of 5-FU and PERK inhibitor three times per week for 3 weeks 20 . Tumor volume over time and tumor weight at sacrifice were measured and presented as the average ± standard error of mean for 5 tumors per treatment group. All animals were randomized by weight. statistical analysis. All data are presented as mean ± standard error of mean unless otherwise specified. Student t test (two-tailed) was used to compare two groups of data, and two way ANOVA was used to analyze drug treatment. p < 0.05 was considered significant.

Results
evaluating the activity of the unfolded protein response pathways in human colon cancer cell lines. We first examined the activation of three branches of UPR pathways in six colon cancer cell lines. To assess the activity of PERK pathway, we gauged the expression level of two downstream factors of the PERK kinase, phospho-eIF2α and ATF4 (Fig. 1A,D), as well as a direct target of ATF4, GADD34 (Fig. S1A). To assess the activity of the IRE1 pathway, we measured the expression of spliced XBP1, a downstream factor of IRE1 upon its activation (Fig. 1B,D). For the ATF6 pathway, we applied a reporter-based assay to measure the transcriptional activity of ATF6, since currently there are no reliable antibodies to detect the cleavage of endogenous ATF6 protein (Figs 1C and S1A). BiP is a critical chaperon protein involved in ER stress. We found that BiP expression was also differently expressed among CRC cells (Fig. 1D). In addition to its basal level, we also measured the response of each arm of the UPR to Thapsigargin, a classical ER stressor. Similarly, the UPR pathways of CRC cells were induced differently by Thapsigargin (Fig. S1B,C). Based on the above results, we found that the UPR pathways are differentially activated in CRC cells.

Activation of the PERK-ATF4 pathway is required for resistance of CRC cells to 5-Fluorouracil treatment.
To evaluate if the activity of UPR pathways is associated with the sensitivity or resistance of CRC cells to chemotherapeutic regimens, we tested the response of CRC cell lines to treatment of 5-FU. With a 3 day-treatment, the EC50 of 5-FU on CRC cells was ranged from ~200 nM to 20 µM ( Fig. 2A,B). We next examined if the activation level of the UPR pathways correlates with the EC50 of 5-FU in these cells. We found that the level of phospho-eIF2α and ATF4 positively correlates the EC50, while the level of sXBP-1 (spliced XBP-1), activity of ATF6 or the expression of BiP does not show significant correlation (Figs 2C,D and S2A-C). In addition, the PERK-ATF4 pathway is induced upon 5-FU treatment (Fig. 2E), and the induction of ATF4 is not due www.nature.com/scientificreports www.nature.com/scientificreports/ to alteration of oxidative stress, since treatment of NAC, a reducing agent, does not reverse the increase of ATF4 upon 5-FU treatment ( Fig. 2F) while IRE1-sXBP1 and ATF6 were not significantly changed (Figs 2E and S2D). These results suggested that the PERK-ATF4 branch of the UPR pathways may regulate the sensitivity of CRC cells to 5-FU treatment.
To test this hypothesis, we knocked down PERK in SW620 cells and conducted 5-FU treatment (Fig. S2E). Decrease of PERK expression significantly reduces the level of p-eIF2α, confirming the role of PERK in regulating eIF2α in CRC cells. As a control, loss of other two eIF2α kinases, GCN2 and PKR, does not change the level of p-eIF2α (Fig. S2F). We found that reduction of PERK expression can significantly sensitize SW620 cells to 5-FU treatment (Fig. 2G). Consistently, reduction of ATF4, a downstream factor of PERK, also increases sensitivity of SW620 cells to 5-FU treatment ( Fig. 2H and S2G). To confirm the specific role of the PERK pathway in mediating CRC cells' resistance to 5-FU treatment, we went on to knock down IRE1 and ATF6, respectively. We found that reduction of the IRE1 and ATF6 pathways does not alter the sensitivities of CRC cells to 5-FU treatment ( Fig. S2H-K). In addition, treatment of NAC does not change the sensitivity of SW620 cells to 5-FU treatment (Fig. 2I). Collectively, activation of the PERK-ATF4 pathway is functionally required for CRC cells to resist chemotherapy.

Chemical inhibition of PERK sensitizes CRC cells to 5-Fluorouracil treatment.
While shRNA-mediated gene expression reduction of the PERK pathway components was effective in sensitizing CRC cells to 5-FU, we next tested if chemical inhibition of the PERK pathway can produce the same effect. By use of a GSK-developed second generation small molecule inhibitor of PERK 20 , we can effectively suppress the activity of PERK in SW620 cells lines by measuring the level of phosphorylated PERK (Fig. 3A). The reduction of activity of the PERK pathway by this inhibitor can also be gauged by the level of its downstream factors (Figs 3B and S3A). Consistent with the findings from the shRNAs-based assays, treatment of the PERK inhibitor can effectively increase the sensitivity of CRC cells to 5-FU treatment (Figs 3C,D and S3B). To further confirm this finding, we measured apoptosis caused by 5-FU and PERK inhibitor in CRC cells. We found that treatment of PERK inhibitor significantly increases apoptotic cell death caused by 5-FU (Fig. 3E,F). As a control, inhibition of the IRE1 pathway does not sensitize CRC cells to 5-FU treatment (Fig. S3C). By use of a murine colon cancer cell line, CT26, we found that inhibition of the PERK activity also sensitizes CT26 cells to 5-FU treatment (Fig. S3D,E). Taking together, chemical inhibition of the PERK pathway sensitizes CRC cells to chemotherapy.

Pharmacological inhibition of PERK potentiates treatment of 5-Fluorouracil in vivo.
We next investigated if inhibition of PERK can synergize with chemotherapy in vivo. Consistent with prior studies, mono-treatment of 5-FU or PERK inhibitor, respectively, led to partial inhibition of tumor growth of the SW620 cells in nude mice (Fig. 4A-C). Remarkably, combo-treatment of 5-FU and PERK inhibitor exhibited a striking synergistic inhibition on tumor growth (Fig. 4A-C). We then examined the expression of Ki67, a classical marker of proliferation and cell growth, in the control, mono-treated and combo-treated tumors. We found the percentage of Ki67 positive cells was markedly decreased upon combined treatment with 5-FU and the PERK inhibitor (Fig. 4D). Lastly, we found that cleaved PARP was also increased in tumors treated with both PERK inhibitor and 5-FU (Fig. 4E), further showing the exceptional anti-tumor effects delivered by the combinatory administration with chemotherapeutic agent and PERK inhibitor. www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
Resistance to chemotherapy is a major hurdle in clinical management of late-stage colon cancer, as well as other types of human cancers. In this study, we set out to investigate the molecular mechanism of drug resistance of colon cancer cells, and discovered that activation of the PERK kinase-mediated stress signaling plays an important role in promoting cell survival under chemotherapy.
Activation of the unfolded protein response pathway has been implicated in cell survival. It has been reported that BiP, a classical chaperone protein induced by the UPR pathway, is required for breast cancer and lung cancer cells to grow and survive anti-cancer drugs 15 . This might be the underlying mechanism how two branches of the www.nature.com/scientificreports www.nature.com/scientificreports/ UPR signaling, the IRE1-and ATF6-led pathways, promote chemo-resistance of advanced cancers 21 . In addition, activation of the PERK pathway is required for breast cancer cells that have undergone an EMT to survive treatment of doxorubicin and paclitaxel, primarily through the NRF2 cascade, a non-canonical downstream of PERK 22 . Interestingly, while it was recently reported that the PERK-NRF2 pathway is activated and required for CRC cells to grow in the condition of ER stress and chemotherapy treatment 23 , we did not observe that NRF2 is differentially expressed in CRC cells with different sensitivity to 5-FU (data not shown). Instead, we found that expression of ATF4, a canonical downstream factor of the PERK kinase, is up-regulated in CRC cells that hold heightened resistance to chemotherapy drugs. We further showed that both PERK and ATF4 are functionally required for cell survival under 5-FU treatment. It is also possible that different downstream factors of PERK mediate resistance to different chemo-drugs.
The PERK-ATF4 signaling contributes to many cancer-related characteristics during disease progression. It has been reported that the PERK pathway is required for cell migration and invasion in breast and cervix cancers 24,25 . In addition, ATF4 promotes metastasis by reducing anoikis of cancer cells in suspension 26 . Consistent with its role in cancer cell migration and metastasis, the PERK branch of the UPR pathways is specifically activated in cancer cells that undergo an EMT, which is a key molecular program to drive metastasis 27 . Two recent studies suggested that the PERK pathway and its downstream factors promote metastasis through EMT-driven activation of the protein secretion pathways 28,29 . In addition to its role in migration and metastasis, the EMT molecular program is also closely related to drug resistance of cancer cells to chemotherapy and other anti-cancer regimens 30 . Therefore, our findings suggested that the PERK-ATF4 pathway may mediate the drug resistance caused by cell-state transition and plasticity program, e.g., the EMT program, in CRC. Since PERK is targetable by use of pharmacological inhibitors, it could be a promising approach to overcome drug resistance to CRC through inhibition of the PERK pathways. At the meantime, we noticed that other UPR-related mechanisms that underlie the drug resistance of CRC have been reported 31 . While CHOP, the downstream protein of ATF4 is expressed in prostate cancer cells even in the absence of ER stressors, CHOP is not expressed in the CRC cells that we tested. This tissue-specific CHOP expression might differentiate the role of PERK pathway in prostate cancer and colorectal cancer, because CHOP is largely known as a pro-apoptotic protein, while ATF4 is primarily pro-cancer development and progression. We also noticed that by use of the HCT116 cell model, another group has reported that the PKR kinase is responsible for the increase of p-eIF2α in 5-FU-treated colon cancer cells 32 . It suggests that colon cancer develop multiple stress-related mechanisms to resist 5-FU treatment, and different cell lines may apply different strategy in this process.
Cancer seems to acquire multiple malignant phenotypes at its late stage: recurrent and metastatic cancers are usually multidrug resistant, and enriched for cancer stem-like cells. One possibility would be that there are some central pathways that govern multiple cancer characteristics simultaneously. While we found that the PERK-ATF4 pathway is essential for drug-resistance of CRC cells, this pathway may also promote metastatic progression of CRC cells. It will be of great interest to examine if activation of the PERK-ATF4 pathway is also functionally required for migration and metastasis of CRC cells. Further investigation on this will open a window on an entirely new avenue of targeting cancer progression.