Nucleus-translocated GCLM promotes chemoresistance in colorectal cancer through a moonlighting function

Abstract

Metabolic enzymes perform moonlighting functions during tumor progression, including the modulation of chemoresistance. However, the underlying mechanisms of these functions remain elusive. Here, utilizing a metabolic clustered regularly interspaced short palindromic repeats (CRISPR)-Cas9 knockout library screen, we observe that the loss of glutamate-cysteine ligase modifier subunit (GCLM), a rate-limiting enzyme in glutathione biosynthesis, noticeably increases the sensitivity of colorectal cancer (CRC) cells to platinum-based chemotherapy. Mechanistically, we unveil a noncanonical mechanism through which nuclear GCLM competitively interacts with NF-kappa-B (NF-κB)-repressing factor (NKRF), to promote NF-κB activity and facilitate chemoresistance. In response to platinum drug treatment, GCLM is phosphorylated by P38 MAPK at T17, resulting in its recognition by importin a5 and subsequent nuclear translocation. Furthermore, elevated expression of nuclear GCLM and phospho-GCLM correlate with an unfavorable prognosis and poor benefit from standard chemotherapy. Overall, our work highlights the essential nonmetabolic role and posttranslational regulatory mechanism of GCLM in enhancing NF-κB activity and subsequent chemoresistance.

Fig. 1. GCLM depletion enhances the chemosensitivity of CRC cells to oxaliplatin.

a Diagram showing the strategy for the CRISPR-Cas9 screen in HCT116 cells under the treatment with PBS or oxaliplatin (10 µM, 7 days). The genomic DNA of control and treated cells was isolated and amplified for comparison of sgRNAs via deep sequencing. b Venn diagram showing the top 100 downregulated genes and the overlapping genes in the three oxaliplatin-treated groups compared with the control group (Supplementary Data 1). c Cell viability of HCT116 cells with or without oxaliplatin treatment (40 µM, 24 h) after each candidate gene (CYP3A5, APEH, ASS1, ALG1, GGH, GFPT1, GSTO2, AGMAT, SHMT2 or GCLM) was knocked out. d IC₅₀ Analysis of oxaliplatin in HCT116 and DLD1 cells treated with different concentrations of oxaliplatin for 48 h. e Annexin V/PI staining analysis was used to evaluate the percentages of apoptotic cells among control and GCLM-knockdown CRC cells treated with PBS or oxaliplatin (40 μM for HCT116 cells and 80 μM for DLD1 cells, 24 h). f Statistical analysis of CDX tumor volumes and weights in nude mice after the implantation of GCLM-knockdown or control HCT116 cells (2 × 10⁶), followed by intraperitoneal injections of PBS or oxaliplatin (5 mg/kg). Statistical analysis of the tumor volumes and weights in nude mice in the PDX #1 (g) and PDX #2 (h) models, followed by intratumoral injections of in vivo-optimized GCLM inhibitor (siGCLM) or the control siRNA (5 nmol per injection), and intraperitoneal injections of PBS or oxaliplatin (5 mg/kg). i Representative images of H&E, IHC staining for Ki67, Tunel staining and Masson’s trichrome staining in PDX #1-based paraffin-embedded subcutaneous tumor sections. The red arrowheads indicate the positive cell of Tunel staining. Scale bar = 50 μm. j Quantification of the proliferation index (Ki67 staining) and apoptotic index (Tunel assay) in the PDX #1 models. n = 3 biologically independent experiments in (c, d) and n = 6 mice in (f–h, j). All the data are presented as the mean ± S.D. The P values were calculated by one-way ANOVA (f–h right and j), and two-way ANOVA (f–h left). Oxa oxaliplatin, CDX cell-derived xenograft, PDX patient-derived xenograft, CRC colorectal cancer, ANOVA analysis of variance.

Fig. 2. GCLM translocates to the nucleus up platinum drug treatment.

a IHC staining and scores of GCLM expression in primary CRC tumor (T) and adjacent normal tissues (N) (n = 406, CRC tissue specimens). Scale bar = 50 μm. The data are presented as a box-and-whisker graph (minimum–maximum), and the horizontal line across the box indicates the median (a right). b Overall survival and disease-free survival assays of CRC patients based on the GCLM protein level in (a). c IHC staining of GCLM expression in CRC patients receiving FOLFOX or XELOX chemotherapy. The blue arrowheads indicate the GCLM expression in nucleus. The correlation between GCLM expression and the response of patients to the standard chemotherapy (right), which is presented in the form of the percentage of total samples. (n = 58, CRC tissue specimens). Scale bar = 50 μm. d Cell viability of CRC cells overexpressing control or rGCLM WT, which were treated with PBS, oxaliplatin (40 μM for HCT116, 80 μM for DLD1, 24 h) or BSO (150 μM, 24 h). e IB detection of total, cytoplasmic and nuclear GCLM expression in HCT116 cells treated with 40 μM oxaliplatin (Oxa), cisplatin (CDDP) and carboplatin (CBP), and 20 μM irinotecan (CPT-11) and 5-fluorouracil (5-FU) for 24 h. f IF staining showing the localization of GCLM in HCT116 cells with the treatment of PBS, oxaliplatin or cisplatin (40 μM, 24 h). Scale bar = 10 μm. Cell viability (g) and apoptotic cells (h) among HCT116 cells overexpressing control, nuclear GCLM (NLS) or C193/194 A mutant (NLS-CA) with oxaliplatin or cisplatin treatment (40 μM, 24 h). i Photographs and tumor volumes, weights analysis of CDX model after the implantation of HCT116 cells overexpressing control or nuclear GCLM, followed by intraperitoneal injections of PBS or oxaliplatin (5 mg/kg). IB experiments were repeated three times and n = 3 biologically independent experiments in (d, g, h) and n = 5 mice in (i). The data are presented as the mean ± S.D (d, g, h, i). The P values were calculated by two-tailed paired Student’s t test (a), Kaplan–Meier analysis (log-rank test) (b), two-sided chi-square test (c), one-way ANOVA (h, i right), and two-way ANOVA (d, g, i middle). PD progressive disease, SD stable disease, PR partial response, CR complete response, NLS nuclear localization signals.

Fig. 3. Nuclear GCLM interacts with NKRF to orchestrate NF-κB activity and chemoresistance.

a Coomassie blue staining showing the bound proteins of cytoplasmic and nuclear GCLM in 293 T cells overexpressing Flag-tagged GCLM under oxaliplatin treatment (40 μM, 24 h). Red box indicates the main proteins that differed between the cytoplasmic and nuclear fractions. b Co-IP analysis showing the level of NKRF bound by GCLM using anti-GCLM or IgG antibody. c GST pull-down analysis detecting the level of purified His-NKRF interacting with purified GST-GCLM. d Co-IP analysis demonstrating the level of NKRF bound by total, cytoplasmic and nuclear GCLM in 293 T cells overexpressing Flag-tagged GCLM. Co-IP (e) and Duolink (f) analyses showing the interaction (red dots in f) of GCLM and NKRF in 293 T cells overexpressing Flag-tagged GCLM WT or nuclear GCLM ( + NLS). Scale bar = 10 μm. g Co-IP analysis showing the interaction of GCLM and NKRF in HCT116 cells treated with different concentrations of oxaliplatin or cisplatin for 24 h. h Co-IP analysis demonstrating the level of GCLM bound by WT or truncated NKRF mutants (M1-M5) in 293 T cells. Control or GCLM-knockdown HCT116 or 293 T cells overexpressed HA-tagged NKRF. Co-IP analysis showing the interaction of NKRF and p65/p50 (i) and the interaction of p65 and p50 (j) with oxaliplatin treatment (40 μM, 24 h). k HCT116 or 293 T cells were depleted GCLM and re-expressed nuclear GCLM (NLS) with control or NKRF inhibition in the presence of oxaliplatin (40 μM, 24 h). The transcriptional activity of NF-κB/p65 was detected. l Q-PCR analysis of the expression of NF-κB/p65-targeted genes (CCND1, XIAP, BCL2, BCL-xl, and iNOS) in HCT116 cells when depleting GCLM and re-expressiong nuclear GCLM in the presence of oxaliplatin (40 μM, 24 h). m Cell viability and apoptotic cells analysis in HCT116 cells when depleting GCLM and re-expressiong nuclear GCLM with control or p65 silencing with or without oxaliplatin treatment (40 μM, 24 h). IB experiments were repeated three times and n = 3 biologically independent experiments in (k–m). All the data are presented as the mean ± S.D. The P values were calculated by one-way ANOVA (k, l) and two-way ANOVA (m).

Fig. 4. Platinum drugs promote GCLM nuclear localization via GCLM binding to importin a5.

a Schematic illustration of the potential NLS sequence of the GCLM protein predicted by the cNLS Mapper tool. The strategy for mutating the NLS of the GCLM is also shown. b HCT116 cells overexpressed Flag-tagged GCLM WT or GCLM-NLS mutant (Mut) with or without importazole treatment (40 μM, 24 h). The localization of GCLM was detected by IF staining. Scale bar = 10 μm. c, d HCT116 cells overexpressed Flag-tagged GCLM WT or GCLM-NLS Mut with oxaliplatin or IPZ treatment (40 μM, 24 h). IB detected the nuclear and total GCLM expression (c) and Co-IP analysis showed the level of interaction between NKRF and GCLM (d). e–g GCLM-knockdown HCT116 or DLD1 cells overexpressed rGCLM WT or rGCLM-NLS Mut with or without oxaliplatin treatment (40 μM, 24 h). The transcriptional activity of NF-κB/p65 (e), cell viability (f) and apoptotic cells (g) were detected. h Co-localization of endogenous GCLM and importin α5 in HCT116 cells was detected by IF. Red arrowheads showed the co-localization. Scale bar = 10 μm. i HCT116 cells overexpressed Flag-tagged GCLM WT or GCLM-NLS Mut with or without oxaliplatin treatment (40 μM, 24 h). Co-IP analysis showing the GCLM-importin α5 interaction. IF staining of the localization of GCLM (j) and IB detection of nuclear and total GCLM expression (k) in HCT116 cells with control or importin α5 silencing in presence of oxaliplatin treatment (40 μM, 24 h). Scale bar = 10 μm. l, m GCLM-knockdown HCT116 cells overexpressed vector or rGCLM WT with control or importin α5 silencing under oxaliplatin treatment (40 μM, 24 h). The transcriptional activity of NF-κB/p65 (l), cell viability and apoptotic cells (m) were detected. IB experiments were repeated three times and n = 3 biologically independent experiments in (e–g, l, m). All the data are presented as the mean ± S.D. The P values were calculated by one-way ANOVA (e, l, m) and two-way ANOVA (f, g). IPZ importazole, Imp α5 importin α5.

Fig. 5. P38 MAPK-mediated phosphorylation of GCLM participates in platinum drug-induced nuclear localization of GCLM.

a IP analysis showing the level of GCLM phosphorylated at threonine (pThr), serine (pSer) and tyrosine (pTyr) residues in 293 T cells treated with oxaliplatin or cisplatin (40 μM, 24 h). b Co-IP analysis showing the level of NKRF and importin α5 bound by GCLM in HCT116 cells treated with calf intestinal phosphatase (CIP) that resulted in GCLM dephosphorylation. c IP analysis showing the level of GCLM phosphorylated at Thr17, Thr106 and Thr164 in HCT116 cells treated with oxaliplatin or cisplatin (40 μM, 24 h). d IF staining showing the localization of Flag-GCLM in 293 T cells overexpressing Flag-tagged GCLM WT or T17A mutant with oxaliplatin treatment (40 μM, 24 h). Scale bar = 10 μm. e, f 293 T and HCT116 cells overexpressed the Flag-tagged GCLM WT or T17A, T17E mutant with or without oxaliplatin treatment (40 μM, 24 h). IB detection of nuclear and total GCLM expression (e) and Co-IP analysis showing the levels of importin α5 and NKRF interacting with GCLM (f). g 293 T cells were treated with the inhibitors of AKT (AKTi, MK-2206, 5 μM), ERK (ERKi, PD98059, 10 μM), JNK (JNKi, SP600125, 20 μM) or P38 MAPK (P38i, SB203580, 10 μM) for 24 h. Co-IP analysis showing the GCLM-importin α5 interaction and the level of GCLM phosphorylated at Thr17. h, i CRC cells were treated with control or P38 MAPK inhibitor (P38i, SB203580, 10 μM) in the presence or absence of oxaliplatin treatment (40 μM for HCT116, 80 μM for DLD1, 24 h). Co-IP analysis showing the level of importin α5 interacting with GCLM (h) and IB detection of nuclear and total GCLM expression (i). The transcriptional activity of NF-κB/p65 (j), cell viability (k) and apoptotic cells (l) in GCLM-knockdown HCT116 or DLD1 cells overexpressing rGCLM WT or T17A, T17E mutants with or without oxaliplatin treatment (40 μM, 24 h). IB experiments were repeated three times and n = 3 biologically independent experiments in (j–l). All the data are presented as the mean ± S.D. The P values were calculated by one-way ANOVA (j) and two-way ANOVA (k, l).

Fig. 6. Phosphorylation of GCLM at T17 contributes to CRC chemoresistance in vivo.

Representative images (a, the black arrowheads indicate the tumor), tumor numbers and average sizes (b) of spontaneous tumors in Gclm^fl/fl and Gclm^iKO mice treated with PBS or oxaliplatin (5 mg/kg). Representative images of H&E, IHC of GCLM, Ki67 and Tunel staining (c), and quantification of Tunel staining (d) of the spontaneous mouse CRC model. The blue arrowheads indicate the adenomas and the red arrowheads indicate the positive cells of Tunel staining. Scale bar = 1 mm (low power image) and 50 μm (high power images). Statistical analysis of the CDX tumor volume and weight after the implantation of endogenous GCLM-knockdown HCT116 (e) and DLD1 (f) cells (2 × 10⁶), which overexpressed rGCLM WT or T17A, T17E mutants, followed by intraperitoneal injections of PBS or oxaliplatin (5 mg/kg). Statistical analysis of the tumor volumes and weights in the PDX #1 (g) and PDX #2 (h) models, followed by intraperitoneal injections of control or P38 inhibitor (P38i, SB203580, 5 mg/kg) and FOLFOX (oxaliplatin 5 mg/kg, 5-fluorouracil 25 mg/kg). n = 6 mice in (b, d, g, h) and n = 5 mice in (e, f). All the data are presented as the mean ± S.D. The P values were calculated by one-way ANOVA (b, d and g, h right) and two-way ANOVA (e–h left).

Fig. 7. Nuclear GCLM is highly expressed in CRC and indicates a poor prognosis.

a IHC staining and scores for nuclear GCLM expression (nGCLM, only focusing on GCLM staining in the nucleus, as shown by blue arrowheads) in paired primary CRC tumor tissues (T) and adjacent normal tissues (N) (n = 406, CRC tissue specimens). Scale bar = 50 μm. b IHC staining and scores for the level of GCLM phosphorylated at Thr17 (pT17-GCLM) in paired primary CRC tumor tissues (T) and adjacent normal tissues (N) (n = 200, CRC tissue specimens). Scale bar = 50 μm. c, d Overall survival (left) and disease-free survival (right) assays of patients with CRC based on nuclear GCLM (a) and pT17-GCLM (b) expression. e Overall survival assays of patients with GC (n = 235, GC tissue specimens) or ESCC (n = 276, ESCC tissue specimens) based on nuclear GCLM expression. f The correlation between nuclear GCLM and pT17-GCLM expression and the response of CRC patients to FOLFOX or XELOX chemotherapy (n = 58, CRC tissue specimens, PD progressive disease, SD stable disease, PR partial response, CR complete response). g IHC staining showing high and low expression of nuclear GCLM (nGCLM, blue arrowheads), pT17-GCLM, Ki67, Tunel (red arrowheads) and phospho-P38 MAPK (pP38) expression in CRC tumor tissues. Scale bar = 50 μm. h Correlations between nuclear GCLM expression and pT17-GCLM, Ki67, Tunel and pP38 MAPK expression. i Proposed working model based on this study. The model shows that P38 MAPK-mediated phosphorylated GCLM interacts with importin α5 and is translocated into the nucleus upon platinum drug treatment. Nuclear GCLM enhances NF-κB/p65 activity by interacting with NKRF to confer resistance to platinum-based chemotherapy in CRC cells. The data are presented as a box-and-whisker graph (minimum–maximum), and the horizontal line across the box indicates the median (a, b right), and the percentage of total sample (f, h). We categorized proteins levels as low or high compared with the median value of IHC score (c–e, f and h). The P values were calculated by two-tailed paired Student’s t test (a, b right), Kaplan–Meier analysis (log-rank test) (c–e), and two-sided chi-square test (f, h). GC, gastric carcinoma; ESCC, esophageal squamous cell carcinoma.