Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Oct;20(10):1341-1352.
doi: 10.1038/s41589-024-01612-6. Epub 2024 May 8.

Tumor-repopulating cells evade ferroptosis via PCK2-dependent phospholipid remodeling

Zhe Li #  1   2   3 , Zhi-Min Xu #  1   2   3 , Wei-Peng Chen #  1   2 , Xiao-Jing Du #  2   3 , Chun-Xian Ou  1   2   3 , Zi-Kang Luo  1   2   3 , Rong Wang  4 , Chu-Qing Zhang  1   2   3 , Chao-Dong Ge  5   6 , Meng Han  7 , Fudi Wang  5   6 , Rong-Rong He  4 , Wan-Yang Sun  8   9 , Jun Ma  10   11   12 , Xiao-Yu Liang  13   14   15 , Zhuo-Wei Liu  16   17
Affiliations

Tumor-repopulating cells evade ferroptosis via PCK2-dependent phospholipid remodeling

Zhe Li et al. Nat Chem Biol. 2024 Oct.

Abstract

Whether stem-cell-like cancer cells avert ferroptosis to mediate therapy resistance remains unclear. In this study, using a soft fibrin gel culture system, we found that tumor-repopulating cells (TRCs) with stem-cell-like cancer cell characteristics resist chemotherapy and radiotherapy by decreasing ferroptosis sensitivity. Mechanistically, through quantitative mass spectrometry and lipidomic analysis, we determined that mitochondria metabolic kinase PCK2 phosphorylates and activates ACSL4 to drive ferroptosis-associated phospholipid remodeling. TRCs downregulate the PCK2 expression to confer themselves on a structural ferroptosis-resistant state. Notably, in addition to confirming the role of PCK2-pACSL4(T679) in multiple preclinical models, we discovered that higher PCK2 and pACSL4(T679) levels are correlated with better response to chemotherapy and radiotherapy as well as lower distant metastasis in nasopharyngeal carcinoma cohorts.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Fig. 1
Fig. 1. TRCs resist radiotherapy and chemotherapy by evading ferroptosis.
a,b, HONE1 TRCs and bulk cells were treated with radiation (IR, 8 Gy) with or without Fer-1 (1 µM) for 60 h. Fer-1 was added 6 h before IR treatment. The content of ferroptotic cell death signal PE (38:4)-OOH (a) and PE (40:4)-OOH (b) was measured by LC–MS/MS. P = 0.000028, 0.9999996, 0.000344 and 0.474506. c, Percentage of dead cells in TRCs and bulk tumor cells from HONE1 cells treated with IR in the absence or presence of Fer-1 (1 μM) and DFO (20 μM). P = 0.000000000000101. d, Percentage of dead cells in TRCs and bulk tumor cells from HK1 cells treated with IR in the absence or presence of Fer-1 (2 μM) and DFO (20 μM). P = 0.000000000000019. e–g, Dose-dependent toxicity of ferroptosis inducers RSL3 (e), FIN56 (f) or FINO2 (g) induced cell death of TRCs and bulk tumor cells (HONE1). n = 3 replicates from one representative of three independent experiments. h–j, Percentage of dead cells in TRCs and bulk tumor cells from HONE1 cells treated with indicated inhibitors and RSL3 (15 μM) (h), FIN56 (10 μM) (i) or FINO2 (10 μM) (j). Fer-1, 1 μM; Z-VAD (Z-VAD-FMK), 10 μM and Nec-1 (necrostatin-1), 2 μM. P = 0.000000000042792, 0.000000000075223 and 0.000000002261459. k, Accumulation of RSL3-induced oxygenated PE that contains AA (C20:4) or AdA (C22:4) in TRCs and bulk tumor cells from HONE1 cells. P = 0.00000879. n = 3 replicates from one representative of three independent experiments. Data are shown as mean ± s.d.; one-way ANOVA (a,b,k) or two-way ANOVA (c,d,h–j). n = 3 independent experiments. NS, not significant. Source data
Fig. 2
Fig. 2. ACSL4-dependent PL remodeling is required for ferroptosis resistance of TRCs.
a, The radar chart illustrates changes of SFA-PLs, MUFA-PLs and PUFA-PLs in HONE1 TRCs and bulk tumor cells (left). The bar chart (right) represents the relative increment of these PLs. b, The content of esterified AA (C20:4) or AdA (C22:4) PE or PC molecular species in HONE1 TRCs and bulk cells. c, The formation rate of AA-d8-CoA in reactions catalyzed by ACSL4 purified from HONE1 TRCs and bulk cells at different timepoints. AA-d8 and coenzyme A were used as substrates. AA-d8-CoA was undetectable in the samples with no enzyme added. d, The content of esterified AA (C20:4) or AdA (C22:4) PE molecular species in TRCs and bulk tumor cells from ACSL4-knockout HONE1 cells and parental cells. e, The content of esterified AA (C20:4) or AdA (C22:4) PC molecular species in TRCs and bulk tumor cells from ACSL4-knockout HONE1 cells and parental cells. f,g, TRCs and bulk cells from ACSL4-knockout HONE1 cells (sg ACSL4) or parental cells (sg GFP) were treated with RSL3 (15 µM) for 13 h. The content of ferroptotic cell death signal PE (38:4)-OOH (f) and PE (40:4)-OOH (g) was measured by LC–MS/MS. P = 0.0000596 and 0.5509789. h, TRCs and bulk tumor cells from ACSL4-knockout HONE1 cells or parental cells were treated with RSL3 (15 µM) or FIN56 (10 µM). The percentage of dead cells was measured after indicated treatment for 16 h. Data are shown as mean ± s.d.; unpaired two-tailed t-test (b,c,h) or one-way ANOVA (d–g). n = 3 independent experiments. NS, not significant. Source data
Fig. 3
Fig. 3. T679 phosphorylation is the key event for ACSL4 to function.
a, Immunoprecipitation and immunoblot analysis of the level of phospho-Ser/Thr ACSL4 in TRCs and bulk tumor cells (HONE1 and A375). b, Mass spectrometry analysis of the phosphorylation site in ACSL4 protein purified from HONE1 TRCs or bulk cells (left). The percentage of ACSL4 Thr679 phosphorylation in HONE1 TRCs and bulk tumor cells (right). c, The formation rate of AA-d8-CoA in reactions catalyzed by ACSL4 protein purified from Dox-inducible ACSL4-WT HONE1 cells or Dox-inducible ACSL4-T679A HONE1 cells. AA-d8 and coenzyme A were used as substrates. AA-d8-CoA was undetectable in the samples with no enzyme added. d, Dose-dependent toxicity of RSL3-induced cell death of ACSL4-knockout HONE1 cells stably expressing Dox-inducible ACSL4-WT (ACSL4 WTTet) or Dox-inducible ACSL4-T679A (ACSL4 T679ATet) with or without Dox (0.3 μg ml−1). e, The radar chart indicates the changes of SFA-PLs, MUFA-PLs and PUFA-PLs in Dox-inducible ACSL4-WT HONE1 cells or Dox-inducible ACSL4-T679A HONE1 cells (left). The relative increment of these PLs is represented by a bar chart (right). f,g, Dox-inducible ACSL4-WT HONE1 cells or Dox-inducible ACSL4-T679A HONE1 cells were treated with Dox (0.3 μg ml−1) followed by AA-d8 (10 µM) for 36 h. The relative changes of PC (f) or PE (g) that contain AA (20:4)-d8 are shown. n = 4. *P < 0.05; **P < 0.01; ***P < 0.001. f, P = 0.0017, 0.0008, 0.0236, 0.0017, 0.0045, 0.0035, 0.0373, 0.0312, 0.0013 and 0.0030. g, P = 0.0481, 0.0039, 0.0263, 0.0045, 0.0446, 0.0164, 0.0073, 0.0035 and 0.0002. h, Accumulation of RSL3-induced oxygenated PE that contains AA or AdA in ACSL4-knockout HONE1 cells (sg ACSL4), Dox-inducible ACSL4-WT expression HONE1 cells (ACSL4 WTTet) or Dox-inducible ACSL4-T679A expression HONE1 cells (ACSL4 T679ATet) treated with Dox (0.3 μg ml−1). P = 0.000000182, 0.0004 and 0.0000171. Data are shown as mean ± s.d.; unpaired two-tailed t-test (c,f,g) or one-way ANOVA (h). n = 3 independent experiments. One of three experiments is shown (a). Source data
Fig. 4
Fig. 4. PCK2 directly phosphorylates ACSL4 at T679 to enhance ferroptosis sensitivity.
a, Immunoblots showing the expression of ACSL4 and p-ACSL4(T679) in PCK2-knockout cells and parental cells. b, Immunoblots showing the expression of ACSL4 and p-ACSL4(T679) in TRCs and bulk tumor cells. c, Representative immunofluorescence staining image showing the co-localization of ACSL4 and mitochondria in cells stably transfected with mito-GFP. Scale bar, 10 μm. d, Immunoblots showing the localization of ACSL4 in mitochondria. Cyto, cytosol; Mito, mitochondria; WCL, whole-cell lysate. Tom20 and α-tubulin are markers for mitochondria and cytosol, respectively. e, Immunoelectron microscopic localization of ACSL4 and PCK2 in the mitochondria by using a gold-labeled anti-ACSL4 antibody (15 nm gold) and a gold-labeled anti-PCK2 antibody (6 nm gold). The co-localization of ACSL4 and PCK2 in sections of HONE1 cells is shown. Bars, 0.2 μm. f, Immunoprecipitation and immunoblotting showing the interaction of ACSL4 and PCK2 in HONE1 TRCs and bulk cells. g, PCK2-knockout HONE1 cells and parental cells were treated with AA-d8 (10 µM). The relative changes of PE that contain AA (20:4)-d8 are shown. n = 4. *P < 0.05; **P < 0.01; ***P < 0.001. P = 0.0254, 0.0294, 0.0038, 0.0024, 0.0027, 0.0267, 0.0225 and 0.004. h, Accumulation of RSL3-induced oxPE that contains AA or AdA in PCK2-knockout HONE1 cells and parental cells. P = 0.0000417, 0.0039 and 0.0046. i, Dose-dependent toxicity of RSL3-induced cell death of PCK2-knockout HONE1 cells (sg PCK2) and parental cells (sg GFP). j, Parental HONE1 cells and TRCs from PCK2-knockout HONE1 cells stably expressing Dox-inducible PCK2 (PCK2Tet) were treated without or with Dox. Immunoblots showing the expression of p-ACSL4(T679) and ACSL4 in cells. k, The cells depicted in j were treated with or without Dox before RSL3 or FINO2 treatment (15 μM). The percentage of dead cells was measured by 7-AAD staining and flow cytometry. Data are shown as mean ± s.d. (g,h,k); unpaired two-tailed t-test (g) or one-way ANOVA (h,k). n = 3 independent experiments. One of three experiments is shown (a,b,d,f,j). Source data
Fig. 5
Fig. 5. Activation of STAT3 in TRCs promotes PCK2 degradation.
a, Immunoblots showing the expression of PCK2 in HONE1 TRCs and bulk tumor cells transfected with indicated siRNAs. b, RT–qPCR analysis of PCK2 expression in HONE1 or A375 TRCs and bulk cells with STAT3 knockdown. c, Inhibition STAT3 with Stattic (5 µM) stabilized PCK2 protein in HONE1 TRCs with CHX treatment for indicated times. The immunoblot analysis of PCK2 protein level (left) and corresponding grayscale analysis (right) are shown. P = 0.00000042 and 0.000002. d, Immunoblot analysis showing PCK2 protein level in HONE1 TRCs and bulk cells treated with MG132 (10 μM), carfilzomib (100 nM), bafilomycin A1 (Baf A1, 0.2 μM) or chloroquine (CQ, 50 μM) for 6 h. e, HONE1 TRCs and bulk cells treated with or without Stattic (5 µM), after which the cell lysates were subjected to immunoprecipitation with anti-PCK2 antibody and immunoblotting with anti-ubiquitin (Ub) antibody. f, RT–qPCR analysis of HERC6 expression in HONE1 or A375 TRCs and bulk cells with STAT3 knockdown. P = 0.0000001, 0.0000003, 0.00000037, 0.000000003, 0.0000000067 and 0.0000000077. g, Immunoblots showing the expression of PCK2 in TRCs and bulk tumor cells with or without HERC6 knockdown. h,i, TRCs from STAT3-knockout HONE1 cells or parental cells and bulk HONE1 cells were subjected to indicated treatments (h). TRCs from HERC6-knockout HONE1 cells or parental cells and bulk HONE1 cells were subjected to indicated treatments (i). RSL3, 15 µM; radiation (IR), 8 Gy; cisplatin (Cis), 20 µM. The percentage of dead cells was measured. j, HONE1 cells were implanted subcutaneously into female BALB/c nude mice and were locally exposed to radiotherapy (8 Gy, arrow) or not. Mice were provided with Stattic (3 mg kg−1, every 2 d, starting from ‘arrowhead’ until the endpoint) or not. Tumor volumes for each group are shown (n = 5). P = 0.00002 and 0.0075. Data are shown as mean ± s.d. (b,f,h,i) or mean ± s.e.m. (j); one-way ANOVA (b,f,h,i) or two-way ANOVA (j). n = 3 independent experiments. One of three experiments is shown (a,c–e,g) CHX, cycloheximide. Source data
Fig. 6
Fig. 6. PCK2-ACSL4(T679) phosphorylation promotes ferroptosis in vivo.
a, HONE1 cells with or without fibrin gel were implanted subcutaneously into female BALB/c nude mice to construct xenograft growth models and were exposed to radiotherapy (8 Gy, arrow). Tumor volumes for each group are shown (n = 5). b, Hematoxylin and eosin (H&E) staining and IHC of fibrin, PCK2 and ACSL4 pT679 showing the presence of fibrin gel and the level of PCK2 and ACSL4 pT679 in TRC tumors and bulk tumors treated with radiotherapy. Images are representative of n = 5 images. Scale bars, 100 µm. c, Flow cytometric analysis of the proportion of EpCAM+pSTAT3+ in TRC tumors and bulk tumors treated with radiotherapy. n = 5. d, Representative immunofluorescence images of PCK2 and ACSL4 pT679 in EpCAM+integrin β3+ and EpCAM+integrin β3− cells. Scale bars, 10 µm. Images are representative of n = 10 images. e, The content of esterified AA (C20:4)-PE molecular species in integrin β3+ and integrin β3− tumor cells isolated from radiation-treated HONE1 tumors. n = 5. f,g, The content of PE (38:4)-OOH (f) and PE (40:4)-OOH (g) in integrin β3+ and integrin β3− tumor cells isolated from radiation-treated HONE1 tumors. n = 5. h, HONE1 cells stably transfected with indicated plasmids were implanted subcutaneously into female BALB/c nude mice to construct xenograft growth models. Indicated groups were exposed to radiotherapy (8 Gy, arrow). All groups were provided with a Dox drink (100 mg kg−1, intragastrical administration, every 2 d, starting from ‘arrowhead’ until the endpoint). Tumor volumes for each group are shown (n = 5). i, PCK2Tet HONE1 cells with or without fibrin gel were implanted subcutaneously into female BALB/c nude mice to construct xenograft growth models. Indicated groups were exposed to radiotherapy (8 Gy, arrow). Mice were provided with a Dox drink (100 mg kg−1, intragastrical administration, every 2 d, starting from ‘arrowhead’ until the endpoint) or a normal drink. Tumor volumes for each group are shown (n = 5). Data are shown as mean ± s.e.m. (a,h,i) or mean ± s.d. (c,e–g); paired two-tailed t-test (e), unpaired two-tailed t-test (c,f,g) or two-way ANOVA (a,h,i). Source data
Extended Data Fig. 1
Extended Data Fig. 1. TRCs resist radiotherapy and chemotherapy via evading ferroptosis.
a-b, HONE1 cells and HK1 cells were seeded in 3D fibrin gels for 5 days of culture. The colony size of the tumor spheroid was presented (a). The colony sizes at different time points were quantified and the colony size on d1 was set to 1 (b). Bar, 50 mm. n = 6. c, Immunoblots showing the expression of cancer stem cell markers (pSTAT3 and SOX2) in TRCs and bulk cells. d, RNA-seq was performed on HONE1 TRCs and bulk cells. The gene set enrichment analysis (GSEA) shows that multiple stem cell-like transition pathways were enriched in the TRCs group. Two-sided P values were calculated with Kolmogorov-Smirnov test and multiple-comparison adjusted q values were calculated using Benjamini-Hochberg (BH) method. e, The HONE1 TRCs and bulk cells (5 cells or 100 cells) were injected into the mammary fat pads of NSG mice. Eight weeks later, the tumor formation was recorded. n = 12. f-g, HONE1 TRCs and bulk cells were treated with cisplatin (Cis) with or without Fer-1. The content of ferroptotic cell death signal, PE (38:4)-OOH (f) and PE (40:4)-OOH (g) was measured. h-j, Dose-dependent toxicity of RSL3 (h), FIN56 (i) or FINO2 (j) induced cell death of HK1 cells. k, Principal component analysis (PCA) of oxygenated phospholipids in HONE1 TRCs and bulk cells. Each point represents a sample. l, Heatmap showing the relative changes in RSL3-induced oxygenated PE and PC molecular species in HONE1 TRCs and bulk cells. Row z-scores were obtained from averaged values of the content of lipid species. Data are shown as mean ± SD (b, f-g), one-way ANOVA (f-g). n = 3 independent experiments. One of three experiments is shown (c). ns, not significant. Source data
Extended Data Fig. 2
Extended Data Fig. 2. Interrogation the involvement of previously identified key genes in mediating resistance to ferroptosis in TRCs.
a, Immunoblots showing the expression of GPX4 and SLC7A11 in TRCs and bulk cells. b, GSH/GSSG ratios in TRCs and bulk cells. c, The relative mRNA expression of AIFM2 (encoding FSP1 protein) in TRCs and bulk tumor cells. d, NADH consumption assay with CoQ1 to determine the FSP1 activity. e, Representative results of a cellular thermal shift assay (CETSA). FSP1 protein was resistant to heat denaturation in the presence of CoQ10 (upper) and α-tocopherol (lower). f, Representative results of a cellular thermal shift assay (CETSA). FSP1 protein from HONE1 bulk tumor cells and TRCs showed similar heat denaturation in the presence of CoQ10 (upper) and α- tocopherol (lower). g-h, The relative mRNA expression of ALOX15 (i) or PEBP1(j) in TRCs and bulk tumor cells. P = 0.1556, 0.0683, 0.4532, 0.1007, 0.1243 and 0.7519. i, The cellular ferrous iron content of TRCs and bulk tumor cells was measured. P = 0.5932, 0.3396 and 0.6071. j, The relative mRNA expression of ACSL1, ACSL3, ACSL4, FATP2, LPCAT3, PLA2G6, PLA2G2F and PLA2G4A in TRCs and bulk tumor cells from HONE1 cells. Data are shown as mean ± SD, unpaired two-tailed t-test (b-c, g- j), n = 3 independent experiments. Bulk, black points; TRC, red points. One of three experiments is shown (a, e-f). ns, not significant. Source data
Extended Data Fig. 3
Extended Data Fig. 3. ACSL4 dependent phospholipid remodeling is required for ferroptosis resistance of TRCs.
a, Immunoblots showing the expression of ACSL4 and LPCAT3 in TRCs and bulk tumor cells. b, Cell lysates of TRCs and bulk tumor cells from HONE1 cells were incubated with or without LPCAT3 inhibitors, (R)-HTS-3. LPCAT3 enzymatic activity was assessed. c, Cell lysates of TRCs and bulk tumor cells from HONE1 cells were incubated with or without ACSL4 inhibitors, Rosiglitazone. The formation of AA-d8-CoA in reactions was detected by LC-MS. d, Immunoblots showing the expression of ACSL4 in ACSL4-knockout HONE1 cells and A375 cells. e, TRCs and bulk cells from ACSL4-knockout A375 cells or parental cells were treated with RSL3 or FINO2. The percentage of dead cells was measured. f-g, Percentage of dead cells in TRCs and bulk tumor cells from HONE1 cells (f) or A375 cells (g) with indicated treatment. Rosiglitazone (ROSI, 10 µM) was added before the treatment of ferroptosis inducers. h, Immunoblots showing the expression of ACSL4 in ACSL4- knockout HONE1 cells transfected with Dox-inducible ACSL4-WT expression plasmids. Cells were treated with or without Dox. i, The parental HONE1 cells (sg GFP), TRCs and bulk tumor cells from ACSL4-knockout HONE1 cells (sg ACSL4) and ACSL4-knockout HONE1 cells stably expressing Dox-inducible ACSL4-wild type (ACSL4 WTTet) were treated with Dox before RSL3 treatment. The percentage of dead cells was measured by 7-AAD staining and flow cytometry. P = 0.8154, 0.0000173 and 0.000000000998. j, Immunoblots showing the expression of ACSL4 in the cells depicted in (i). k-l, The parental HONE1 cells (sg GFP), TRCs and bulk tumor cells from ACSL4-knockout HONE1 cells (sg ACSL4) and ACSL4- knockout HONE1 cells stably expressing Dox-inducible ACSL4-wild type (ACSL4 WTTet) were treated with or without Dox, followed by treatment with IR (k) or cisplatin (Cis) (l). The percentage of dead cells was measured. P = 0.0951, 0.0000000432, 0.6187 and 0.00000273. m, Immunoblots showing the expression of ACSL4 in the cells depicted in (k-l). Data are shown as mean ± SD, unpaired two-tailed t-test (b-c, e) or one-way ANOVA (i, k-l), n = 3 independent experiments. One of three experiments is shown (a, d, h, j, m). ns, not significant. Source data
Extended Data Fig. 4
Extended Data Fig. 4. T679 phosphorylation is the key event for ACSL4 to function.
a, Alignment of protein sequences spanning ACSL4 T679 from different species. b, Immunoblots showing the ACSL4 expression in ACSL4-knockout HONE1 cells transfected with Dox-inducible ACSL4-T679A mutation expression plasmids. Cells were treated without or with Dox. c, Dose-dependent toxicity of FINO2 induced cell death of ACSL4- knockout HONE1 cells stably expressing Doxycycline (Dox)- inducible ACSL4-wild type (ACSL4 WTTet) or Dox-inducible ACSL4-T679A (ACSL4 T679ATet) without or with Dox. d, Dox-inducible ACSL4-WT HONE1 cells or Dox-inducible ACSL4-T679A HONE1 cells pretreated with Dox, followed by IR or cisplatin. The percentage of dead cells was measured. P = 0.0000104 and 0.0005. e, Principal component analysis (PCA) of RSL3-induced oxygenated phospholipids in ACSL4-knockout HONE1 cells (sg ACSL4), Dox-inducible ACSL4-WT expression HONE1 cells (ACSL4 WTTet) or Dox-inducible ACSL4-T679A expression HONE1 cells (ACSL4 T679ATet) treated with Dox, with or without RSL3. Lipids were analyzed by LC-MS/MS. Each point represents a sample. f, Heatmap showing LC-MS analysis of the relative changes in RSL3-induced oxygenated PE molecular species in ACSL4-knockout HONE1 cells (sg ACSL4), Dox-inducible ACSL4-WT expression HONE1 cells (ACSL4 WTTet) or Dox-inducible ACSL4-T679A expression HONE1 cells (ACSL4 T679ATet) treated with Dox (0.3 μg/mL). Row z-scores were obtained from averaged values of the content of lipid species. g, Microscale thermophoresis (MST) demonstrated a direct interaction between Arachidonic acid and GFP-tagged ACSL4 in lysates from GFP-ACSL4-WT or GFP-ACSL4-T679A expressing HEK293T cells; mutation of T679A partly blocks the interaction. Data are shown as mean ± SD, one-way ANOVA (d), n = 3 independent experiments. One of three experiments is shown (b, g). aa, amino acid. Source data
Extended Data Fig. 5
Extended Data Fig. 5. PCK2 phosphorylates ACSL4 at T679.
a, Immunoblots were performed with parental HONE1 cells (sg GFP), ACSL4-knockout HONE1 cells (sg ACSL4), ACSL4-knockout HONE1 cells stably expressing ACSL4 T679A mutation (T679A), treating cell lysates with λ protein phosphatase (λ-phosphatase) or in the presence of blocking peptide for ACSL4 Thr679 phosphorylation (peptide). The p-ACSL4 (T679) antibody was used to detect the phosphorylation level of ACSL4 T679 site. b, IHC analyses of human NPC samples were performed with the p-ACSL4(T679) antibody in the presence or absence of a blocking peptide for p-ACSL4(T679). c-g, LC–MS/MS detection for ACSL4-interacting proteins in ACSL4-containing protein complex immunoprecipitated from HONE1 TRCs and bulk cells. Volcano plot showing the fold change and −log10 (P-value of two-sided Student’s t-test) of differential interaction proteins in HONE1 TRCs and bulk cells. Dotted line indicates factors with fold change > 2 and significance P < 0.05 (c). The peak area and PSMs of PKM2 (d), PFKP (e), PPM1G (f) and PCK2 (g) are shown. n = 4. h, The relative mRNA expressions of indicated genes were analyzed by qRT-PCR for gene knocked-down efficiency in HONE1 cells. P = 0.00005, 0.000040, 0.0000050, 0.000005, 0.00003, 0.000048, 0.000013 and 0.000012. i, Immunoblots showing the expression of ACSL4 and ACSL4 pT679 in HONE1 cells transfected with indicated siRNAs. j, The relative mRNA expression of PCK2 and PCK1 in HONE1 and HK1 cells. P = 0.0001 and 0.0000366. k, PCK2-knockout HONE1 cells and parental cells were treated with or without 3-MPA. Immunoblots showing the expression of p-ACSL4(T679) and ACSL4 in cells. l, Immunoblots showing the expression of ACSL4 and ACSL4 pT679 in HONE1 cells or A375 cells treated with phosphoenolpyruvate (PEP). m, Representative immunofluorescence staining image showing the co-localization of PCK2 (red) and mitochondria (green) in HONE1 cells and HK1 cells stably transfected with mito-GFP. Scale bar: 10 μm. n, Immunoelectron microscopy of ACSL4 localization in mitochondria in HONE1 cells by using a gold-labeled anti-ACSL4 antibody (10 nm gold). Bars: 0.2 μm. o, Subfractionation of highly purified mitochondria from HONE1 cells showing the presence of ACSL4 protein in mitochondria of HONE1 cells and HK1 cells. WCL, whole cell lysate; Cyto, cytosol; OM, mitochondria outer membrane; IM, inner membrane; Mx, mitochondria matrix; IMS, mitochondria intermembrane space. α-Tubulin, Tom20, TIMM22, MnSOD and Cyto C are markers for cytosol, mitochondria outer membrane (OM), inner membrane (IM), matrix (Mx) and inter-membrane space (IMS), respectively. p, Cells expressing ACSL4-mCherry and PCK2-GFP were seeded in 3D fibrin gels (TRCs) or not (bulk cells). Immunofluorescence analysis to detect ACSL4 (red) or PCK2 (green). Scale bar: 10 μm. Data are shown as mean ± SD (d-h, j), unpaired two-tailed t-test (d-g, j) or one-way ANOVA (h), n = 3 independent experiments. One of three experiments is shown (a, i, k-l, o). Source data
Extended Data Fig. 6
Extended Data Fig. 6. PCK2 directly phosphorylates ACSL4 and promotes its activity.
a, In vitro kinase assays were performed by mixing Escherichia coli-purified His-ACSL4 (AA552-711) and 293T-purified active Flag-PCK2 with or without λ-Phosphatase. b, Schematic of PCK2 full-length and truncated mutants. c, HEK293T cells were transfected with the indicated Myc-tagged constructs, the ACSL4-PCK2 interaction was analyzed by immunoprecipitation with anti-Myc beads and immunoblotting. d, Immunoblots showing the expression of PCK2, ACSL4 and ACSL4 pT679 in HONE1 cells that expressing PCK2 wild type (FL) or catalytic domain-deletion mutation (ΔMD). e, PCK2-knockout HONE1 cells and parental cells were treated with AA-d8 (10 µM) for 36 h. The relative changes of PC that contain AA (20:4)-d8 are shown. n = 4. * P < 0.05; ** P < 0.01; *** P < 0.001. P = 0.0157, 0.0004, 0.0047, 0.0002, 0.0395, 0.0009, 0.0124, 0.0103, 0.036, 0.0085, 0.0076 and 0.0048. f, Radar chart indicates the changes of SFA-PLs, MUFA-PLs, and PUFA-PLs in PCK2-knockout cells and parental cells (upper). The relative increment of these phospholipids was represented by a bar chart (lower). g, Principal component analysis (PCA) of oxygenated phospholipids in PCK2-knockout cells and parental cells in the present or in the absence of RSL3. h, LC-MS-based heatmap showing relative changes in RSL3-induced oxygenated PE molecular species in PCK2-knockout HONE1 cells and parental cells. Row z-scores were obtained from averaged values of the content of lipid species. i, Dose-dependent toxicity of FINO2 induced cell death of PCK2-knockout HONE1 cells and parental cells. j, Percentage of dead cells in PCK2-knockout HONE1 cells and parental cells treated with IR or Cis. Data are shown as mean ± SD (e, j), unpaired two-tailed t-test (e) or one-way ANOVA (j), n = 3 independent experiments. One of three experiments is shown (a, c-d). Source data
Extended Data Fig. 7
Extended Data Fig. 7. PCK2 increases cellular ferroptosis sensitivity via mitochondria located ACSL4.
a, Comparison between the levels of cardiolipin (CL) in the cytoplasm (Cyto), endoplasmic reticulum (ER) and mitochondria (Mito). b, The content of PUFA in the cytoplasm (Cyto), endoplasmic reticulum (ER) and mitochondria (Mito). c, Immunoblot analysis of ACSL4 expression in ACSL4-knockout (sg ACSL4) HONE1 cells stably expressing vector (Vector), human NLS-ACSL4 (Nu-ACSL4), human ACSL4 wild type (ACSL4-WT), or human mito-ACSL4 (mito-ACSL4). d, Immunoblots showing increased ACSL4 mitochondrial localization in ACSL4-knockout HONE1 cells stably expressing mitoACSL4 (mitoACSL4) compared to ACSL4 wild type overexpressing HONE1 cells (ACSL4 WT). Cyto, cytosol; Mito, mitochondrial. VDAC1 and α-Tubulin are markers for mitochondrial and cytosol, respectively. e, Representative immunofluorescence staining image showing the co-localization of ACSL4 (green) and Nucleus (blue) in the ACSL4-knockout HONE1 cells stably expressing human ACSL4 wild type (ACSL4-WT) or human NLS-ACSL4 (Nu-ACSL4). Scale bar: 10 μm. f, Dose-dependent toxicity of RSL3 of the cells depicted in ACSL4- knockout (sg ACSL4) HONE1 cells stably expressing vector (Vector), human NLS-ACSL4 (Nu-ACSL4), human ACSL4 wild type (ACSL4-WT) or human mito-ACSL4 (mito-ACSL4). g, The relative phospholipid composition changes of cells in ACSL4-knockout (sg ACSL4) HONE1 cells stably expressing vector (Vector), human NLS-ACSL4 (Nu-ACSL4), human ACSL4 wild type (ACSL4-WT) or human mito-ACSL4 (mito-ACSL4). P = 0.9910, 0.4435, 0.4162, 0.999997, 0.6996, 0.0243, 0.0005, 0.0422, 0.5183, 0.0000991, 0.00000209 and 0.00167. h, The content of esterified AA (C20:4) or AdA (C22:4) PE molecular species in cells in (i) are shown. P = 0.5369, 0.0032, 0.000073, 0.021, 0.0315, 0.0000190, 0.00000224, 0.0326, 0.0158, 0.000000079, 0.00000000851, 0.0004, 0.0991, 0.4924, 0.9986, 0.6389, 0.7626, 0.0002, 0.0024, 0.1500, 0.1951, 0.0005, 0.0005, 0.99994. i, The content of esterified AA (C20:4) or AdA (C22:4) PC molecular species in cells in (i) are shown. P = 0.0344, 0.000000134, 0.0000000163, 0.000947, 0.2556, 0.000000694, 0.0000000808, 0.0031, 0.9897, 0.0000000844, 0.00000002, 0.0097, 0.1196, 0.00000042, 0.000000063, 0.0051, 0.7878, 0.0856, 0.0256, 0.8973, 0.0966, 0.307, 0.311, 0.999999996. Data are shown as mean ± SD (a-b, g-i), one-way ANOVA (g-i), n = 3 independent experiments. One of three experiments is shown (c, d). Source data
Extended Data Fig. 8
Extended Data Fig. 8. STAT3 promotes PCK2 degradation via HERC6- dependent proteasome pathway in TRCs.
a, The indicated mRNA expression was analyzed by qRT-PCR for genes knockdown efficiency assessment in HONE1 cells. P = 0.0000005, 0.0000006, 0.0000005, 0.00000058, 0.00000028, 0.00000016, 0.000000025 and 0.000000019. b, Immunoblots showing the expression of PCK2 in HONE1 cells treated with or without STAT3 inhibitor (Stattic). c, Volcano plot showing upregulation of seven E3 ligases in HONE1 TRCs. d, Quantitative RT-PCR analysis of the relative mRNA expression of indicated genes in HONE1 TRCs and bulk cells treated with or without Stattic. P = 0.168, 0.168, 0.00000057, 0.0000016, 0.267, 0.987, 0.999, 0.818, 0.536, 0.872, 0.581, 0.913, 0.000006, 0.00002. e, STAT3-binding elements on the promoters of HERC6 and TRIM47 genes were predicted by JASPAR. f, qRT-PCR analysis of TRIM47 expression in HONE1 or A375 TRCs and bulk cells with STAT3 knockdown. P = 0.0000000047, 0.000000017, 0.000000014, 0.00000011, 0.00000034 and 0.00000029. g, Immunoprecipitation and immunoblot analysis showing the binding of PCK2 to HERC6 in HEK293T cells transfected with Flag- PCK2 and HA-HERC6. h, Immunoprecipitation and immunoblot analysis showing that PCK2 does not bind to TRIM47 in HEK293T cells that were transfected with Flag-PCK2 and HA-TRIM47. i, HERC6-knockout HONE1 cells (sg HERC6) or parental HONE1 cells (sg GFP) were cultured in 3D fibrin gels, after which the cell lysates were subjected to immunoprecipitation with anti-PCK2 antibody and immunoblotting with anti-ubiquitin (Ub) antibody. j, Immunoblots showed that knocking out HERC6 stabilized PCK2 protein in HONE1 TRCs with CHX treatment for indicated times. k, The HONE1 cells were implanted subcutaneously into female BALB/c nude mice to construct xenograft growth models. Mice were provided with Stattic (3 mg/kg, every two days, starting from ‘arrowhead’, until the endpoint) or not. Cisplatin or vehicle was intraperitoneally injected (arrow). Tumors volumes for each group were shown (n = 5). Data are shown as mean ± SD (a, d, f) or mean ± SEM (k), one- way ANOVA (a, d, f) or two-way ANOVA (k). n = 3 independent experiments. One of three experiments is shown (b, g-j). Source data
Extended Data Fig. 9
Extended Data Fig. 9. PCK2-ACSL4(T679) phosphorylation axis promotes ferroptosis in vivo.
a-b, The TRCs and bulk tumor cells from HONE1 cells were implanted subcutaneously into female BALB/c nude mice to construct xenograft growth models and exposed to radiotherapy (8 Gy, arrow) (a) or cisplatin (5 mg/kg, arrow) (b). Tumor volume for each group were shown (n = 5). c, Immunoblots showing the expression of PCK2, pSTAT3 and STAT3 in bulk tumor cells and TRCs from HONE1 cells in indicated conditions. One of three experiments is shown. d, The HONE1 cells with or without fibrin gel were implanted subcutaneously into female BALB/c nude mice to construct xenograft growth models and exposed to cisplatin (arrow). Tumor volumes for each group were shown (n = 5). e, H&E staining and Immunohistochemistry of fibrin showing the presence of fibrin gel in TRC tumors. Images are representative of n = 5 images. f, H&E staining and immunohistochemistry of fibrin, PCK2 and ACSL4 pT679 showing the presence of fibrin gel and the level of PCK2 and ACSL4 pT679 in TRC tumors and bulk tumors treated with cisplatin. Images are representative of n = 5 images. g, Flow cytometric analysis of the proportion of EpCAM+ pSTAT3+ in TRC tumors and bulk tumors treated with cisplatin. n = 5. h-i, The HONE1 cells stably transfected with indicated plasmids (h) or PCK2Tet HONE1 cells with or without fibrin gel (i) were implanted subcutaneously into female BALB/c nude mice to construct xenograft growth models. Indicated groups were exposed to cisplatin (arrow). Mice were provided with a Dox drink (100 mg/kg, every two days, starting from ‘arrowhead’, until the endpoint) or a normal drink. Tumor volumes for each group were shown (n = 5). Data are shown as mean ± SEM (a-b, d, h-i) or mean ± SD (g), unpaired two-tailed t-test (g) or two-way ANOVA (a-b, d, h-i). Source data
Extended Data Fig. 10
Extended Data Fig. 10. PCK2-ACSL4(T679) phosphorylation is related to better therapy response in NPC.
a, Correlation analysis of PCK2 and ACSL4 pT679 expression detected by IHC. The P value was determined using the two- tailed χ2 test. b-c, The PCK2 (b) and ACSL4 pT679 (c) expression in pre-treatment biopsies are positively related to clinical response of NPC patients to platinum-based induction chemotherapy. CR, complete response, n = 17; PR, partial response, n = 142; SD, stable disease, n = 23. The P value was determined using the two-tailed χ2 test. d, Correlations of locoregional recurrence status with the level of ACSL4 pT679 detected by IHC. The P value was determined using the two-tailed χ2 test. e-f, Kaplan-Meier analysis of disease-free survival (h) and overall survival (i) according to the ACSL4 pT679 expression levels. The P values in h-i were determined using the log-rank test. P = 0.0025 and 0.000049. g, Forest plots showing the results of multivariate Cox regression analysis, indicated that ACSL4 pT679 expression is a significant prognostic indicator for overall survival in NPC patients (Cox proportional-hazards model). pACSL4(T679) level (High vs. Low: HR = 0.219–0.631), gender (Male vs. Female: HR = 1.163–3.777), WHO type (III vs. II: HR = 0.206–0.887) and TNM stage (IV vs. III: HR = 1.252–2.886). h, Kaplan-Meier analysis of distant metastasis-free survival according to the ACSL4 pT679 expression levels. The P value was determined using the log-rank test. Source data
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Lei, G., Zhuang, L. & Gan, B. Targeting ferroptosis as a vulnerability in cancer. Nat. Rev. Cancer22, 381–396 (2022). - PMC - PubMed
    1. Jiang, L. et al. Ferroptosis as a p53-mediated activity during tumour suppression. Nature520, 57–62 (2015). - PMC - PubMed
    1. Wang, W. et al. CD8+ T cells regulate tumour ferroptosis during cancer immunotherapy. Nature569, 270–274 (2019). - PMC - PubMed
    1. Guo, J. et al. Ferroptosis: a novel anti-tumor action for cisplatin. Cancer Res. Treat.50, 445–460 (2018). - PMC - PubMed
    1. Lei, G. et al. The role of ferroptosis in ionizing radiation-induced cell death and tumor suppression. Cell Res.30, 146–162 (2020). - PMC - PubMed

LinkOut - more resources