Loss of PERK function promotes ferroptosis by downregulating SLC7A11 (System Xc⁻) in colorectal cancer

Abstract

Ferroptosis, a genetically and biochemically distinct form of programmed cell death, is characterised by an iron-dependent accumulation of lipid peroxides. Therapy-resistant tumor cells display vulnerability toward ferroptosis. Endoplasmic Reticulum (ER) stress and Unfolded Protein Response (UPR) play a critical role in cancer cells to become therapy resistant. Tweaking the balance of UPR to make cancer cells susceptible to ferroptotic cell death could be an attractive therapeutic strategy. To decipher the emerging contribution of ER stress in the ferroptotic process, we observe that ferroptosis inducer RSL3 promotes UPR (PERK, ATF6, and IRE1α), along with overexpression of cystine-glutamate transporter SLC7A11 (System Xc^-). Exploring the role of a particular UPR arm in modulating SLC7A11 expression and subsequent ferroptosis, we notice that PERK is selectively critical in inducing ferroptosis in colorectal carcinoma. PERK inhibition reduces ATF4 expression and recruitment to the promoter of SLC7A11 and results in its downregulation. Loss of PERK function not only primes cancer cells for increased lipid peroxidation but also limits in vivo colorectal tumor growth, demonstrating active signs of ferroptotic cell death in situ. Further, by performing TCGA data mining and using colorectal cancer patient samples, we demonstrate that the expression of PERK and SLC7A11 is positively correlated. Overall, our experimental data indicate that PERK is a negative regulator of ferroptosis and loss of PERK function sensitizes colorectal cancer cells to ferroptosis. Therefore, small molecule PERK inhibitors hold huge promise as novel therapeutics and their potential can be harnessed against the apoptosis-resistant condition.

Keywords: Cancer; ER stress; Ferroptosis; PERK; SLC7A11; UPR.

Fig. 1

Ferroptosis inducer RSL3 causes UPR and the PERK arm of UPR regulates SLC7A11 expression (A-D) HT29, SW620, DLD1 and HCT116 cells were either treated with 1 μM RSL3 or vehicle control (VC) for 24 h and protein lysates were prepared for Western blot analysis. Immunoblot shows the expression for classical UPR marker proteins, i.e., PERK, ATF6, and IRE1α. Full-length ATF6 (ATF6-FL), membrane-associated ATF6 (ATF6-P), and nuclear-translocated S2P-cleaved ATF6 (ATF6-N) all are indicated in the immunoblot. (E) Major ferroptosis regulator proteins GPx4, SLC7A11, and ACSL4 in HT29 cells. (F–H) Immunoblot analysis of PERK, ATF6, IRE1α, SLC7A11, GPx4, and ACSL4 in (F) PERK knockdown (KD), (G) ATF6 KD and (H) IRE1α KD in HT29 cells with respective empty vector (EV). (I) Immunoblot showing the expression of PERK and SLC7A11 in EV and PERK KD HT29 cells that were either treated with vehicle or 1 μM RSL3 for 6 h. (J–K) Immunoblot analysis of PERK and SLC7A11 in (J) VC and 5 μM PERK inhibitor (GSK2656157) treated and (K) VC, PERK inhibitor and RSL3 treated HT29 cells. (L–M) EV and PERK KD, (N–O) VC and PERK inhibitor treated HT29 (L, N) and SW620 (M, O) cells were cultured in RPMI 1640 in the presence (cystine+) or absence (cystine-) of cystine for 72 h and subjected to Western blot analysis (left) for the expression of PERK and SLC7A11, quantitative analysis (right) of percent area covered by the cultured cells are shown in the graph (Original photomicrographs are shown in Supplementary Fig. 5). *p < 0.05; compared to control. β-actin was used as a loading control in all immunoblot studies. Respective molecular weight marker (left of each immunoblot) and densitometric quantifications (bottom of each immunoblot) are shown.

Fig. 2

Loss of PERK function promotes ferroptosis in colorectal cancer (A) HT29 EV, PERK, ATF6, and IRE1α KD cells were treated with either vehicle or 1 μM RSL3 for 48 h, and SRB assay was performed to evaluate the cytotoxic effect of the same. (B) SW620 EV and PERK KD cells were treated with vehicle or 1 μM RSL3 for 48 h, and subjected to SRB assay. For A and B, percent growth inhibition was tabulated. *p < 0.05; compared to respective EV. (C) HT29 EV (untagged) and chilli-tagged PERK KD (red) cells were mixed in equal numbers and subjected to flow cytometric analysis on day 0 and day 3 following treatment with vehicle control or 500 nM RSL3. Analysed cell populations are shown in the FACS plot. (D) Different CRC cells were treated with RSL3 (HT29 and SW620 - 1 μM), (DLD1 - 125 nM), and (HCT116 - 2 μM) alone or in combination with 5 μM of PERK inhibitor (GSK2656157) for 48 h and SRB assay was performed. Percent growth inhibition was tabulated. *p < 0.05; compared to RSL3 treated cells. (E) HT29 cells were treated with vehicle control or 10 nM Thapsigargin or 1 μg/ml Tunicamycin for 48 h and subjected to SRB assay. Percent growth inhibition was calculated and tabulated in columns. *p < 0.05; compared to respective controls. (F) HT29 EV and PERK KD cells were treated with vehicle control or different chemotherapeutic drugs i.e., paclitaxel (20 nM), 5-Fluorouracil (100 μM), Doxorubicin (10 μM), Oxaliplatin (50 μM) for 48 h and cytotoxic impact of these drugs was evaluated via SRB assay. Percent growth inhibition was tabulated in the columns. (G) HT29 EV and PERK KD cells were treated either with either 1 μM RSL3 alone or in combination with (25 μM) Z-VAD-FMK (pan-caspase inhibitor) or (10 μM) ferrostatin-1 (ferroptosis inhibitor) and percent growth inhibition in different groups was estimated by SRB assay. *p < 0.05; compared to respective RSL3 control. In A-B and D-G Columns represent an average of triplicate readings of samples; error bars ± S.D. (H–I left panel) Control (EV) and PERK KD of (H) HT29 and (I) SW620 cells were treated with 1 μM RSL3 or vehicle control for 24 h followed by BODIPY C11 staining (Lipid peroxidation sensor) and cells were analysed by FACS (detailed description provided in materials and methods section). Histogram overlays show BODIPY C11 positivity correlating with lipid peroxidation levels in respective groups. (H–I) Right Panels, respective delta mean fluorescence intensity (MFI) of the cells, stained for BODIPY C11. The delta mean was calculated by subtracting the mean fluorescence intensity of the control from the RSL3 treated cells. Columns represent an average of duplicate readings of samples; error bars ±S.D. *p < 0.05; compared to EV (control). (J–K) Immunoblots representing 4-Hydroxynonenal (4-HNE) and malondialdehyde (MDA) conjugated protein expression in (J) HT29 EV and PERK KD cells and (K) HT29 VC and PERK inhibitor-treated (24 h) cells. β-actin was used as a loading control in all immunoblot studies. Respective molecular weight marker (left of each immunoblot) and densitometric quantifications (bottom of each immunoblot) are shown. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3

PERK-ATF4 axis regulates SLC7A11 expression and colorectal cancer cell ferroptosis (A) Total RNA was isolated from control and RSL3 treated HT29 cells, reverse transcribed, and subjected to qRT-PCR analysis for PERK, ATF6 and IRE1α mRNA expression. Fold change in mRNA expression is represented in the bar graph. Data is representative of three independent experiments, resulting from different samples; Columns, the fold change of mRNA expression of PERK, ATF6 and IRE1α compared to respective control. (B) Control and RSL3 treated HT29 cells were subjected to qRT-PCR analysis as described above for expression of the SLC7A11 gene. Columns represents fold change in mRNA expression of SLC7A11; bars ±SD. *, p < 0.05, compared with control. (C) HT29 cells were pre-treated for 30 min with 10 μM Ferrostatin-1 (ferroptosis inhibitor) and further treated with 1 μM RSL3 alone or in combination with FER-1. Immunoblot analysis shows the expression of PERK, Bip/GRP78, and ATF4. (D) HT29 cells were treated with Thapsigargin (ER stress inducer) alone or with PERK inhibitor (pre-treated for 2 h) for 6 h and subjected to Western blot analysis. The expression of PERK, ATF4, and SLC7A11 is shown in the immunoblot images. (E–F) Immunoblots show the expression of PERK, ATF4, and SLC7A11 in (E) HT29 EV and PERK KD or (F) HT29 EV and ATF4 KD cells. (G) qRT-PCR analysis of HT29 EV and PERK KD cells; Columns showing the fold change in mRNA expression of PERK and SLC7A11 genes. bars ±SD. *, p < 0.05, compared with respective control. (H) Diagrammatic representation of SLC7A11 promotor showing (Top) putative ATF4 DNA binding sites and transcription start site (TSS) with RNA pol II, (Down) Human ATF4 binding motif of SLC7A11 promoter on predicted binding site (-0.3 kb upstream from TSS) that is publicly available at JASPAR database (http://www.jaspar.genereg.net). (I) ChIP assay (Details described in the Methods section) was performed in vehicle control, and PERK inhibitor-treated (24 h) HT29 cells using anti-ATF4 and IgG antibodies and then examined by real-time qPCR using primer pairs targeting predicted -0.3 kb and -0.5 kb ATF4 binding sites upstream from TSS of the SLC7A11 gene. Fold change in enrichment for ATF4 and IgG with respect to % input was shown; Data is representative of three independent experiments resulting from different samples; Columns, the average value of percentage enrichment compared to input; bars ±SD. *, p < 0.05, compared with respective control. Photomicrograph of Gel showing conventional PCR validation of ChIP experiments. Lanes are vehicle control and treatment, respectively, for each group of the ChIP sample. (J) The HEK293 cells were transfected with -0.6 kb upstream of SLC7A11 promoter luciferase construct plasmid (pGL4.12) followed by treatment with vehicle or 1 μM and 5 μM of PERK inhibitor for 24 h and cells were harvested for luciferase activity (detailed description in Methods Section). Columns, the average value of relative firefly luciferase activity compared to Renila luciferase activity derived from triplicate readings of different samples; bars ±SD. **, p < 0.01, compared with vehicle control. (K) HT29 cells were treated with vehicle control or 1 μM of RSL3, and ChIP assay was performed using anti-ATF4 and IgG antibodies and then examined by real-time qRT-PCR using primer pairs targeting the predicted ATF4 binding sites on the SLC7A11 promoter. Fold change enrichment for ATF4 and IgG with respect to % input was shown in Columns; bars ±SD. *, p < 0.05, compared with vehicle control. (L) The HEK293 cells were transfected with -0.6 kb upstream of SLC7A11 promoter luciferase construct plasmid, followed by treatment with vehicle or 500 nM and 1 μM of RSL3 for 24 h, and subjected to luciferase activity. Columns, the average value of relative firefly luciferase activity compared to Renila luciferase activity derived from triplicate readings of different samples. (M) HT29 EV and ATF4 KD cells were treated with 1 μM RSL3 for 48 h, and SRB assay was performed. Percent growth inhibition was tabulated, Columns, an average of triplicate readings of samples; error bars ±S.D. *p < 0.05; compared to EV. (N) HT29 EV and ATF4 KD cells were treated with 1 μM RSL3 for 6 h and analysed by flow cytometry after staining with BODIPY C11. Histogram overlays show lipid peroxidation levels in respective treatment groups. (O) EV and PERK KD HT29 cells were made stable for overexpression of SLC7A11 and subjected to immunoblot analysis for SLC7A11 and PERK. (P) HT29 EV and PERK KD cells with stable SLC7A11 overexpression (as shown in the immunoblot) were treated with 1 μM RSL3 for 48 h and subjected to SRB assay. Percent growth inhibition was tabulated, Columns, an average of triplicate readings of samples; error bars ±S.D. *p < 0.05; compared to respective control. β-actin was used as a loading control in all immunoblot studies. Respective molecular weight marker (left of each immunoblot) and densitometric quantifications (bottom of each immunoblot) are shown.

Fig. 4

Loss of PERK has compromised in vivo colorectal tumor growth due to increased ferroptosis 2 x 10⁶ HT29 EV and PERK KD cells in 100 μl PBS were injected subcutaneously in the flanks of the right hind leg of 4–6 weeks old Crl: CD1-Foxn1nu mice in two different groups for each condition. Tumor volumes were measured twice a week with a caliper. (A) Tumor progression of the same is shown in the graph. Each point indicates the average tumor volumes at a particular time; error bars ± SEM (n = 4 for each group); *p < 0.05 compared to control tumors. (B) Photographs of tumor-bearing mice (top) and harvested tumors (bottom) from respective groups were shown. (C) The average tumor weight of each group is shown in the graph. error bars ± SEM (n = 4 for each group); *p < 0.05 compared to control tumors. (D) Harvested tumors were lysed and subjected to Western blot analysis to visualize the protein expression of PERK SLC7A11, ATF4, MDA (malondialdehyde) and 4-HNE (4-Hydroxynonenal). β-actin and GAPDH were used as loading control in immunoblot studies. Respective molecular weight marker (left of each immunoblot) and densitometric quantifications (bottom of each immunoblot) are shown. Respective molecular weight marker (right) and densitometric quantifications (below) are shown for respective blots of all Western blot images. (E) Immunohistochemistry was conducted to detect MDA and 4-HNE in formalin-fixed paraffin-embedded serial sections of harvested tumors with respective antibodies. Representative photomicrographs are shown at 10X and 40X magnifications. Scale bar, 50 μm (10X) or 15 μm (40X). (F–G) Quantitative ATM scores for the expression of (F) MDA and (G) 4-HNE are represented as scatter plots; Error bar, +/- SEM, *p-value, <0.05, compared to expression in control tumors.

Fig. 5

PERK (EIF2AK3) and SLC7A11 are positively correlated in human colorectal tumors (A) GDC TCGA COAD patient data were acquired from the Xena browser, and a correlation graph was plotted between EIF2AK3 (PERK) and SLC7A11. R (Pearson’s correlation coefficient) (B) RT-PCR in matched non-malignant (Normal) and malignant tumor samples of colorectal cancer patients showed the correlation between delta Ct of EIF2AK3 (PERK) and SLC7A11. (C–D) Total RNA was isolated from colorectal cancer patient tumor tissue samples along with their respective matched non-malignant counterparts, reverse transcribed, and RT-qPCR was performed for PERK and SLC7A11 expression analysis. 18s is used as an internal control. Fold change in mRNA expression in (C) PERK and (D) SLC7A11 is shown in bar diagram; Columns, the average value of fold change as compared to control; error bars ± SEM. *, p < 0.05, compared to control, n = 15. (E) The findings are illustrated in a graphical abstract showing how selectively PERK arm of ER stress regulates ferroptosis in colorectal cancer.