GINS4 suppresses ferroptosis by antagonizing p53 acetylation with Snail

Significance

GINS4 is a promoter of eukaryotic G₁/S_-cell cycle as a regulator of initiation and elongation of DNA replication, but little is known about its impact on ferroptosis. Here, we found that GINS4 was involved in the suppression of ferroptosis in LUAD. Depletion of GINS4 could effectively induce G1, G₁/S, S, and G₂/M cells to ferroptosis, especially for G2/M cells. Mechanistically, GINS4 could upregulate Snail expression to antagonize p53 acetylation, and p53 lysine residue 351 (K351 for human p53) was the key site for GINS4-suppressed p53-mediated ferroptosis, indicating that GINS4 might be a critical regulator of ferroptosis in LUAD.

Abstract

Ferroptosis is an iron-dependent oxidative, nonapoptotic form of regulated cell death caused by the destruction of redox homeostasis. Recent studies have uncovered complex cellular networks that regulate ferroptosis. GINS4 is a promoter of eukaryotic G₁/S_-cell cycle as a regulator of initiation and elongation of DNA replication, but little is known about its impact on ferroptosis. Here, we found that GINS4 was involved in the regulation of ferroptosis in lung adenocarcinoma (LUAD). CRISPR/Cas9-mediated GINS4 KO facilitated ferroptosis. Interestingly, depletion of GINS4 could effectively induce G1, G₁/S, S, and G₂/M cells to ferroptosis, especially for G2/M cells. Mechanistically, GINS4 suppressed p53 stability through activating Snail that antagonized the acetylation of p53, and p53 lysine residue 351 (K351 for human p53) was the key site for GINS4-suppressed p53-mediated ferroptosis. Together, our data demonstrate that GINS4 is a potential oncogene in LUAD that functions to destabilize p53 and then inhibits ferroptosis, providing a potential therapeutic target for LUAD.

A key precondition for DNA replication is that it must take place exactly once in each cell cycle with high fidelity and efficiency, to prevent the accumulation of genetic changes, which has potentially harmful consequences on cell survival and organism. Erroneous DNA replication events bring about diseases (1). The GINS complex, a structure of heterotetramer, consists of four different subunits (Sld5, Psf1, Psf2, and Psf3 representing go, ichi, ni, and san, respectively, from Japanese, also named GINS4, GINS1, GINS2, and GINS3 in the human genome). The GINS complex has no significant enzyme activity but is important to initiate and prolong chromosome replication by binding to Mcm helicase and enhances enzyme function (2). GINS4 is an important component of the GINS complex, which plays an indispensable role in the initiation and prolongation of DNA replication for promoting eukaryotic G₁/S cell cycle (3). Our previous studies have confirmed that GINS4 is overexpressed in nonsmall cell lung cancer (NSCLC) tissues and cell lines (4). Overexpression of GINS4 in other human cancer tissue and cell lines has also been reported (5–8). In addition, survival analysis has showed that the overall survival (OS) and disease-free survival (DFS) of cancer patients (lung, gastric, colorectal, and pancreatic cancers) with high expression of GINS4 are significantly lower than those with low expression of GINS4 (4, 5, 8). Therefore, GINS4 may participate in the malignant progression of cancer and become a potential target for cancer therapy and diagnosis. However, the molecular mechanism of GINS4-promoting tumor progression remains unclear. In the current study, we aimed to determine the function of GINS4 in suppressing ferroptosis and to identify GINS4 as a potential target for LUAD.

Ferroptosis, proposed in 2012, is a unique form of programmed cell death with iron dependency (9). It contributes to the maintenance of living/death balance in normal cells and tissues (10) and associates with the development of disease, especially cancers (11, 12). With the increase of membrane density, mitochondria become smaller, and cristae usually decrease or disappear in ferroptotic cells (9, 13). Inhibition of glutathione peroxidase 4 (GPX4) (e.g., RSL3, FIN56, and FINO2) or system Xc^- (e.g., sorafenib and Erastin), depletion of glutathione (GSH) (e.g., BSO), and some physiological conditions such as high extracellular glutamate, amino acid, or cystine starvation have been reported capable of inducing ferroptosis. On the contrary, lipophilic antioxidants (such as liprostatin-1, vitamin E), iron chelators (such as deferoxamine), lipid peroxidation inhibitors (such as eugenol), and consumption phospholipids of polyunsaturated fatty acids (such as tetraenoic acid) could inhibit ferroptosis (14). In the molecular mechanism, p53 (15), autophagy (16), neurofibromatosis 2-YAP (NF2-YAP) (17), p62-Keap1-NRF2 (18), and glutamine catabolism (19) signaling pathways are mainly engaged in the management of ferroptosis. p53 induces spermidine/spermine N1-acetyltransferase 1 (SAT1) (20) or glutaminases 2 (GSL2) (21) at transcriptional level in human osteosarcoma cells or mouse embryonic fibroblasts (MEFs) and promotes ferroptosis by regulating lipid peroxidation. It also suppresses the transcription of cystine/glutamate antiporter solute carrier family 7 member 11 (SLC7A11; also known as xCT), promotes GSH depletion, and induces ferroptosis, which leads to tumor inhibition in human osteosarcoma, breast cancer, or lung cancer cells (22). Mouse p53 3KR mutations at K117, K161, and K162 (here in after referred to as p53 3KR/ 3KR) diminished its ability to suppress cancer. However, 3KR p53 could still activate ferroptosis and inhibit tumor growth, but the additional mutation of p53 at K98 (K101 for human) would abolish this ferroptosis-promoting process (23). By identifying a new p53 acetylation site at lysine K136, mutations at all five acetylation sites (p53-5KR) totally/greatly reduced the remaining tumor suppressive function of p53 (24), which indicates that acetylation of p53 plays an important role in regulating ferroptosis and inhibition of tumor development. The p53 tetramerization domain includes a nuclear output signal (NES) (residues 340 to 351) which is masked during p53 tetramerization (25). It has been reported that lysine residues 351 for human could be ubiquitinated by MSL2, an E3 ligase, which promotes ubiquitin-dependent cytoplasmic p53 localization while not affecting its stability (26), and the lysines in the tetramerization domain of p53 are mainly involved in G₁ arrest modulation (27). Whether the acetylation site K351 associated with ferriptosis remains to be explored. A deep understanding of the role for p53 in ferroptosis may provide new strategies for cancer prevention.

Cancer cells with epithelial mesenchymal transition (EMT) characteristics are sensitive to ferroptosis (28–30). However, the molecular mechanisms underlying this phenomenon remain to be elucidated. Snail, a key EMT molecule, promotes the invasion and metastasis of tumor cells, bestows tumor cells with cancer stem cell-like trait, and promotes drug and radiation resistance (31). Snail is also a crucial molecule in regulating the acetylation of p53, for it causes the deacetylation of p53 K382 by binding to p53 and histone deacetylases (HDAC1). Therefore, Snail decreases p53 activity and plays a role in p53-dependent anti-apoptosis and cell fate in breast cancer (32, 33). On the other hand, p53 activates miRNA-34 that targets Snail and thus suppresses the activity of Snail (34). Whether this mutual regulation between Snail and p53 affects other types of cell death remains to be unknown.

In the present study, we investigated a GINS4-modulated ferroptosis in LUAD. We found that GINS4 could upregulate Snail expression to antagonize p53 acetylation and thus suppress LUAD ferroptosis, indicating that GINS4 might be a critical regulator of ferroptosis in LUAD.

Results

DNA Replication-Related GINS4 Associates with Ferroptosis.

To explore the role of DNA replication-related genes in LUAD, a total of 276 genes from DNA replication-related pathway (hsa03030) in KEGG pathway database and MSigDB database (Go: 0006260, Go: 0006261) were selected. Among them, 253 genes had available data in both TCGA and GSE40419 datasets (SI Appendix, Table S1); 89 (81 upregulated and 8 downregulated) and 64 (59 upregulated and 5 downregulated) differentially expressed genes (DEGs) were identified in TCGA and GSE40419, respectively (Fig. 1 A and B and SI Appendix, Tables S2 and S3), and the 62 DEGs were overlapped in both datasets (SI Appendix, Fig. S1 A–D and Table S4) and thus were utilized to construct a protein–protein interaction (PPI) network (SI Appendix, Fig. S1E). Interestingly, we found that GINS4, which was found to promote LUAD progression in our previous study, was among the 10 hub genes in the PPI network. This indicates a critical role of GINS4 in LUAD-related DNA abnormal replication. As shown in Fig. 1 C and D, GINS4 was highly expressed in LUAD tissues, which was conformed in our previous findings (4). GSEA results showed that DNA replication and cell cycle pathways were significantly enriched in both datasets. In GSE40419, GINS4 also showed the function of regulating ferroptosis (Fig. 1E). Subsequently, we performed GSEA analysis on the expression of genes and GINS4 mRNA in TCGA dataset and made a heatmap (Fig. 1F). The genes in the heatmap were the core enrichment genes which contributed most to the enriched pathway using GSEA. The heatmap contains the core enrichment genes whose mRNA levels were related to GINS4 mRNA in ferroptosis, and they might play an important role in GINS4-modulated ferroptosis. The core enrichment genes identified by GSEA like p53 were the members in “the leading-edge subset” of more than 17,000 genes, representing relatively high association with GINS4 in TCGA dataset. Next, to analyze the effect of DEGs on the overall survival of patients with lung adenocarcinoma, Kaplan Meier analysis by log-rank test was performed based on TCGA clinical data, and the result indicated that GINS4 played a significant role in LUAD prognosis (Fig. 1G). Therefore, we constructed CRISPR/Cas9-mediated GINS4 KO cell line based on KRAS-mutant A549 cells (SI Appendix, Fig. S1 F–G). Treated cells with RAS-selective lethal (RSL) or Erastin that is selectively lethal to RAS mutant cell lines (35, 36) resulted in a time-dependent increase in cell death (Fig. 1 H and I). Taken together, these results suggested that GINS4, an important DNA replication molecule, highly expressed in lung adenocarcinoma might be involved in the regulation of ferroptosis.

Fig. 1.

DNA replication molecule GINS4 is involved in ferroptosis. (A) 81 upregulated and 8 downregulated DEGs were identified in TCGA dataset. ****P < 0.0001. (B) 59 upregulated and 5 downregulated DEGs were identified in GSE40419 dataset, ****P < 0.0001. (C) GINS4 was significantly highly expressed in LUAD tissues in TCGA dataset. ****P < 0.0001. (D) GINS4 was significantly highly expressed in LUAD tissues in GSE40419 dataset. ****P < 0.0001. (E) GSEA results showed that DNA replication, cell cycle, and ferroptosis pathways were significantly enriched GSE40419 dataset. (F) A heatmap reflecting the mRNA expression of the ferroptosis core enrichment genes according to the expression of GINS4 in TCGA dataset. (G) Kaplan Meier analysis by log-rank test was performed based on TCGA clinical data. (H) Visualization of GINS4 WT and KO A549 cells viability over time ± Erastin 5 μM or RSL3 5 μM. (n = 3 independent experiments). (I) Trypan blue staining cell death in GINS4 WT and KO A549 cells over time ± Erastin 5 μM or RSL3 5 μM. (n = 3 independent experiments).

GINS4 Depletion Promotes Ferroptosis In Vitro.

To verify the bioinformatics implication that GINS4 was involved in the regulation of ferroptosis, we constructed a LUAD cell line A549 with GINS4 stably depletion or overexpressing (SI Appendix, Fig. S2 A and B) as well. Consistent with the above bioinformatics prediction, we found that glutamate-cystine antiporter system X_c⁻ inhibitor (Erastin) could induce more cell death in GINS4 depletion-A549 cells than talabostat (PT-100, a potent pyrolytic inducer), Rapamycin (RAPA, a potent autophagy inducer), and Vincristine (VCR, a potent apoptosis inducer), while ferrostatin-1 (Ferr-1), a potent ferroptosis inhibitor, could completely reverse this kind of cell death induced by Erastin. Morphological observation and cellular reactive trypan blue staining also certified that Erastin-induced cell death could be rescued by ferroptosis inhibitor in GINS4 depletion-A549 cells (SI Appendix, Fig. S2C and Fig. 2A). Furthermore, we analyzed mitochondria changes at the ultrastructural level through transmission electron microscopy (TEM). Typical mitochondria with abundant, large, and rich cristae were present in normal A549 cells, while in GINS4 depletion-A549 cells, the mitochondria generally showed enlarged, disorganized, and thickened cristae, especially after the RSL3-induced GPX4 inhibition (Fig. 2B). Likewise, knockdown of GINS4 augmented lipid ROS (Fig. 2 C and D) and decreased GSH level in cells (Fig. 2E). Next, we overexpressed GINS4 in GINS4 depletion cells (SI Appendix, Fig. S2D). The lipid ROS level was reduced (Fig. 2 F and G), and the GSH level was increased especially in treatment with Erastin (Fig. 2H). Meanwhile, we tested ferrous iron (Fe²⁺) or total intracellular iron. Depletion of GINS4 increased total intracellular iron and ferrous iron (Fe²⁺), and they decreased when GINS4 was rescued (SI Appendix, Fig. S2 E and F). Conclusively, these data demonstrated that GINS4 depletion led to lipid peroxidation, the levels of iron, and ferroptosis. Next, we selected six genes involved in iron metabolism: ferritin heavy chain 1 (Fth1), iron-responsive element binding protein 2 (IREB2/IRP2), scavenger receptor class A, member 5 (SCARA5), six-transmembrane epithelial antigen of prostate 3 (STEAP3), and transferrin receptor 1and 2 (TfR1, TfR2). Unfortunately, there was no significant correlation between iron metabolism genes and GINS4 at mRNA and protein levels (SI Appendix, Fig. S2 G and H), suggesting that the lipid peroxidation rather than iron level was the major reason for GINS4 to affect ferroptosis.

Fig. 2.

GINS4 depletion promotes ferroptosis in vitro. (A) Trypan blue staining cell death in shCon, shGINS4 #1, and shGINS4 #2 A549 cells treated with DMSO or Erastin 5 μM or Erastin 5 μM and ferr-1 5 μM or PT100 10 μM or RAPA 10 μM or VCR 2 μM for 24 h. (n = 3 independent experiments). (B) TEM showed that ultrastructure of mitochondria in shCon, shGINS4 #1, and shGINS4 #2 A549 cells, by treatment of DMSO or RSL 5 μM for 24 h. Blue arrowheads, normal mitochondria; Red arrowheads, shrunken mitochondria. (n = 3 independent experiments). (C) Lipid ROS analyzed by flow cytometry in shCon, shGINS4 #1, and shGINS4 #2 A549 cells. (n = 3 independent experiments). (D) Lipid ROS analyzed in shCon, shGINS4 #1, and shGINS4 #2 A549 cells. (n = 3 independent experiments). (E) Intracellular reduced glutathione (GSH) level was detected in shCon, shGINS4 #1, and shGINS4 #2 A549 cells, by treatment with Erastin 5 μM or not for 24 h. (n = 3 independent experiments). (F) Lipid ROS analyzed by flow cytometry in shCon, shGINS4, and shGINS4 transiently GINS4 overexpressed A549 cells. (n = 3 independent experiments). (G) Lipid ROS analyzed in shCon, shGINS4, and shGINS4 transiently GINS4 overexpressed. (n = 3 independent experiments). (H) GSH level was detected in shCon, shGINS4, and shGINS4 transiently GINS4 overexpressed, by treatment with Erastin 5 μM or not for 24 h. (n = 3 independent experiments).

GINS4 Depletion Especially Sensitizes G₂/M Cells to Ferroptosis In Vitro.

Was GINS4-controlled cell cycle involved in ferroptosis? To verify this possibility, we first synchronized A549 cells in specific cell cycle (Fig. 3A) and examined the protein levels of GINS4 in G₁/S, S, G₂/M, G₁ (Fig. 3B). The protein level of GINS4 was lowest in G₂/M and highest in G₁/S and S. To gain insight into the functional relevance of cell cycle–regulated ferroptosis, we examined the effects of GINS4 depletion on different cell cycles to ferroptosis. Combined with RSL3 treatment, an accumulation of G₁ and a reduction of the S phase population showed in four cell cycles, but more G₂/M cells were trapped in G₁ (Fig. 3 C–F). Furthermore, morphological observation and cellular reactive trypan blue staining revealed that GINS4-depleted with RSL3 treatment induced G₂/M cells ferroptosis mostly, but not with doxorubicin treatment (SI Appendix, Fig. S3 and Fig. 3G). The determination of cell counting kit 8 assay (CCK8) also proved that the activity of all stages of the cell cycle in GINS4-depleted cells treated with RSL3 was lower than that of normal cells, but the cells in G2/M phase were more sensitive to RSL3 treatment in essence, and there was no difference in the activity of cells treated with doxorubicin in GINS4-depleted and normal cells (Fig. 3H). These results suggested that all stages of the cell cycle were sensitized by GINS4 depletion, but G2/M cells are intrinsically more sensitive. Therefore, GINS4 depletion might promote ferroptosis mostly for G2/M cells.

Fig. 3.

GINS4 depletion sensitizes G2/M cells to ferroptosis in vitro. (A) Synchronized A549 cells in specific cell cycle stages analyzed by flow cytometry. (B) The protein levels of GINS4 in each stage of the cell cycle detected by Western blot in A549 cells. (C) Cell cycle distribution of shCon and shGINS4 A549 cells synchronized in G₁ phase was treated by RSL3 5 μM or not for 24 h and measured after propidium iodide staining by flow cytometry. Mean values ±SD of shCon (n = 3) and shGINS4 (n = 3) are shown in A549 cells. (D) Cell cycle distribution of shCon and shGINS4 A549 cells synchronized in G₁/S phase was treated by RSL3 5 μM or not for 24 h and measured after propidium iodide staining by flow cytometry. Mean values ±SD of shCon (n = 3) and shGINS4 (n = 3) are shown in A549 cells. (E) Cell cycle distribution of shCon and shGINS4 A549 cells synchronized in S phase was treated by RSL3 5 μM or not for 24 h and measured after propidium iodide staining by flow cytometry. Mean values ±SD of shCon (n = 3) and shGINS4 (n = 3) are shown in A549 cells. (F) Cell cycle distribution of shCon and shGINS4 A549 cells synchronized in G₂/M phase was treated by RSL3 5 μM or not for 24 h and measured after propidium iodide staining by flow cytometry. Mean values ±SD of shCon (n = 3) and shGINS4 (n = 3) are shown in A549 cells. (G) Trypan blue staining revealed cell death of shCon and shGINS4 A549 cells treated with DMSO or RSL3 5 μM or doxorubicin 1 μM for 24 h while synchronized in different cell phases. (n = 3 independent experiments). (H) CCK8 assay analyzed shCon, shGINS4 #1, and shGINS4 #2 A549 cells treated with DMSO or RSL3 5 μM or doxorubicin 1 μM for 24 h while synchronized in different cell phases. (n = 3 independent experiments).

GINS4 KO Suppresses Tumor Development Partly through Ferroptosis In Vivo.

Using a nude mice xenograft model, we noticed the phenomenon of tumor-suppress in GINS4 KO cells in vivo (Fig. 4A), where both the weight and the volume of tumors derived from GINS4 KO cells were obviously less than that of tumors derived from control cells (Fig. 4 B and C). Within GINS4 KO tumors, we observed numerous cells with lipid droplet-like structures in hematoxylin-eosin (H&E) staining (Fig. 4D). TEM analysis revealed that many tumor cells from xenograft tumors with GINS4 KO contained obviously a morphologic feature of ferroptosis, such as shrunken mitochondria, increased membrane density, and presence of abnormally large lipid droplets (Fig. 4E). Since ferroptosis is characterized by excessive lipid peroxidation, we performed IHC analysis of 4-hydroxy-2-noneal (4HNE), an indicator of ferroptosis, to characterize lipid peroxidation level in GINS4 WT or KO expressing tumor samples. These results showed that 4HNE staining increased in GINS4 KO tumor cells, compared with cells from GINS4 WT tumor, while these lesions exhibited no alterations in apoptosis marker (Fig. 4 F and G). As p53 was in the heatmap of the ferroptosis core enrichment gene which is associated with GINS4 (Fig. 1F), we detected the p53 protein level and other ferroptosis-related proteins in tumor tissues. The protein expression level of p53 increased with the decrease of GINS4 protein level. The downstream gene SLC7A11 of p53 also changed, but the change of acyl-CoA synthetase long-chain family member 4 (ACSL4) was not correlated with the change of GINS4 (Fig. 4H). Our data indicated that GINS4 promoted tumor development partly through suppressed ferroptosis and more likely via p53-mediated lipid oxidation pathway.

Fig. 4.

GINS4 KO promotes tumor development partly through ferroptosis in vivo. (A) Xenograft tumor growth of A549 cells with GINS4 WT versus GINS4 KO in nude mice. (B) The final weight of xenograft tumors based on xenograft tumor growth of A549 cells with GINS4 WT versus GINS4 KO. (C) The volume analysis of xenograft tumors based on xenograft tumor growth of A549 cell with GINS4 WT versus GINS4 KO. (D) H&E staining sections of adenocarcinoma of lung from nude mice. Red arrowheads indicate lipid droplets. Bars = 50 μm. (E) TEM of adenocarcinoma of lung from nude mice. Red arrowheads indicate mitochondria or LD; LD = lipid droplets. Bars = 2 μm. (F) IHC staining of tumor xenografts from A549 GINS4 WT versus GINS4 KO stably expressing cell lines. (Scale bar, 50 μm.) Red arrowheads indicate staining of 4HNE. (G) Percentage of 4HNE-positive stained cells per field. Error bars are mean ± SD, n = 4 randomly selected high-power fields. P value was calculated using two-tailed unpaired Student’s t test. (H) Representative immunoblots for GINS4, p53, SLC7A11, and ACSL4 in xenograft tumors.

GINS4 acts as a p53 repressor to suppress ferroptosis.

To further explore mechanism of GINS4-suppressed ferroptosis, we generated several GINS4 stably depletion LUAD cell lines (SI Appendix, Fig. S4 A–D), and we confirmed the presence of p53 in these cell lines (SI Appendix, Fig. S4E). We found that RSL3 induced more cell death in GINS4 KD cells than WT cells in wild-type p53 cells (A549, H460). However, there was no difference in the number of cell death between GINS4 KO and WT cells in p53 null cells (H358, H1299). Such RSL3-induced cell death could be rescued by Ferr-1, but not by ZVAD-FMK, indicating that the mechanism is specific in cell ferroptosis rather than apoptosis (SI Appendix, Fig. S4F, Fig. 5 A–D). Consistent with this, GINS4 depletion-A549 cells (in which wildtype p53 exists) showed higher mortality to RSL3 or tert-butyl hydroperoxideb (TBH, ROS-inducing agent) induction than control cells (Fig. 5 E and F). In contrast, GINS4 depletion-H1299 cells (in which p53 is null) were not sensitive to RSL3 or TBH treatment (Fig. 5 G and H). After reintroduction of p53 into p53 null cells (H1299 and H358), GINS4 depletion in these cells could increase cell sensibility to RSL3-induced cell death (Fig. 5 I–L).

Fig. 5.

GINS4 suppresses ferroptosis in a p53-dependent manner. (A) Trypan blue staining revealed cell death of shCon, shGINS4 #1, and shGINS4 #2 A549 cells treated with DMSO or RSL3 5 μM or RSL3 5 μM and Ferr-1 5 μM or RSL3 5 μM and Z-VAD-FMK 20 μM for 24 h. (n = 3 independent experiments). (B) Trypan blue staining revealed cell death of shCon, shGINS4 #1, and shGINS4 #2 H460 cells treated with DMSO or RSL3 5 μM or RSL3 5 μM and Ferr-1 5 μM or RSL3 5 μM and Z-VAD-FMK 20 μM for 24 h. (n = 3 independent experiments). (C) Trypan blue staining revealed cell death of shCon, shGINS4 #1, and shGINS4 #2 H358 cells treated with DMSO or RSL3 5 μM or RSL3 5 μM and Ferr-1 5 μM or RSL3 5 μM and Z-VAD-FMK 20 μM for 24 h. (n = 3 independent experiments). (D) Trypan blue staining revealed cell death of shCon, shGINS4 #1, and shGINS4 #2 H1299 cells treated with DMSO or RSL3 5 μM or RSL3 5 μM and Ferr-1 5 μM or RSL3 5 μM and Z-VAD-FMK 20 μM for 24 h. (n = 3 independent experiments). (E) CCK8 assay analyzed shCon, shGINS4 #1, and shGINS4 #2 A549 cells treated with DMSO or RSL3 in different concentration. (n = 3 independent experiments). (F) CCK8 assay analyzed shCon, shGINS4 #1, and shGINS4 #2 H1299 cells treated with DMSO or RSL3 in different concentration. (n = 3 independent experiments). (G) CCK8 assay analyzed shCon, shGINS4 #1, and shGINS4 #2 A549 cells treated with DMSO or TBH in different concentration. (n = 3 independent experiments). (H) CCK8 assay analyzed shCon, shGINS4 #1, and shGINS4 #2 H1299 cells treated with DMSO or TBH in different concentration. (n = 3 independent experiments). (I) CCK8 assay analyzed shCon, shGINS4 #1, shGINS4 #2, and shCon, shGINS4 #1, shGINS4 #2 with p53 transiently overexpressed H1299 cells treated with DMSO or RSL3 in different concentration. (n = 3 independent experiments). (J) The protein levels of indicated detected by Western blot in H1299 cells. (n = 3. independent experiments). (K) CCK8 assay analyzed shCon, shGINS4 #1, shGINS4 #2, and shCon, shGINS4 #1, shGINS4 #2 with p53 transiently overexpressed or not H358 cells treated with DMSO or RSL3 in different concentrations. (n = 3 independent experiments). (L) The protein levels of indicated detected by Western blot in H358 cells. (n = 3 independent experiments). (M and N) The volumes of subcutaneous tumors in indicated mice. Unpaired t test was used unless otherwise stated. Data represent the mean ± SD; *P < 0.05; **P < 0.01; ***P < 0.001. (O) The lipid peroxidation level of subcutaneous tumor in indicated mice. Unpaired t test was used unless otherwise stated. Data represent the mean ± SD; *P < 0.05; **P < 0.01; ***P < 0.001.

Then, using a transplanted tumor model of mouse, we found that subcutaneous tumors derived from GINS4-depletion cells with wildtype p53 expressing were smaller than tumors derived from control cells and contain higher level of lipid peroxidation as well (Fig. 5 M and N). However, this effect could not be detected in p53 KO cells. Moreover, we also observed a rescue effect of Ferr-1 on the tissue level of lipid peroxidation in the tumors derived from GINS4-depletion p53 normal cells (Fig. 5 M–O). Changes in SLC7A11 protein levels in mice tumor tissues were also consistent with this result (SI Appendix, Fig. S4J). Those results suggested that GINS4 suppressed ferroptosis in a p53-dependent manner.

Then, we examined p53 mRNA and protein levels regulated by GINS4. GINS4 depletion remarkably increased p53 protein level and GINS4 overexpression decreased p53 protein level (SI Appendix, Fig. S5A and Fig. 6 A and B). However, the p53 mRNA level was not affected by GINS4 depletion or overexpression (Fig. 6 C and D), indicating that the impact of GINS4 on p53 was through a posttranslational way. Concurrently, we found that GINS4 could decrease the half-life of p53 and increase ubiquitylation of p53 (Fig. 6 E-G). Since it has been reported that acetylation is an indispensable modification mechanism to activate p53 both in stress response (37, 38) and ferroptosis (22, 23), we detected p53 acetylation level as well. As such, lysates prepared from control or GINS4 overexpressing cells (containing equalized levels of p53) were assessed by IP assay using antibodies directly against to pan-acetylated lysine (Pan-Ack) or p53. GINS4 overexpressing cells showed a decreased percentage of acetylated p53 compared with control cells (Fig. 6H). On the contrary, GINS4 depletion could dramatically increase p53 acetylation, to a degree equivalent to the blocking effect of TSA, an HDAC inhibitor, on p53 deacetylation (Fig. 6I).

Fig. 6.

GINS4 suppresses ferroptosis by inhibiting p53 acetylation. (A) Western blot analysis of GINS4, p53, and β-actin protein levels in A549 cells with or without GINS4 overexpression. (n = 3 independent experiments). (B) Western blot analysis of GINS4, p53, and β-actin protein levels in H460 cells with or without GINS4 overexpression. (n = 3 independent experiments). (C) qPCR analysis of GINS4, p53, and β-actin mRNA levels in A549 cells with or without GINS4 overexpression. (n = 3 independent experiments). (D) qPCR analysis of GINS4, p53, and β-actin mRNA levels in H460 cells with or without GINS4 overexpression. (n = 3 independent experiments). (E) GINS4 WT and GINS4 KO A549 cells treated with MG132 or not (20 μM) for 6 h. (n = 3 independent experiments). (F) CHX 100 μg/mL pulse-chase analysis of p53 protein levels in GINS4 WT and KO A549 cells. Quantification of p53 protein expression was shown in right panel. (n = 3 independent experiments). (G) shCon, shGINS4 #1, shGINS4 #2 A549 cells treated with MG132 (20 μM) for 6 h. Cell lysates collected for IP analysis. SEshort exposure, LE: long exposure. (n = 3 independent experiments). (H) Lysates from vector and GINS4 overexpressed A549 cells loaded at a ratio of 1:2 was subjected to IP analysis. (n = 3 independent experiments). (I) Western blot analysis of GINS4, p53, acetyl-p53, and β-actin protein levels in A549 GINS4 WT and GINS4 KO cells treated with TSA 1 μM or not for 24 h. (n = 3 independent experiments). (J) H1299 cell lines expressing wildtype, EGFP-K101R, EGFP-K164R, EGFP-K351, EGFP-K382R p53, and crude cell lysates were obtained for Western blot analysis using antibodies against p53, SLC7A11, MDM2, and β-actin. (n = 3 independent experiments). (K) The lipid peroxidation level of cells with indicated p53 mutants. Unpaired t test was used unless otherwise stated. Data represent the mean ± SD; *P < 0.05; **P < 0.01; ***P < 0.001. (L) H1299 GINS4 WT and KO cell lines transiently overexpressing wildtype, EGFP-K101R, EGFP-K382R, EGFP-K351 p53, and crude cell lysates were obtained for Western blot analysis using antibodies against GINS4, p53, SLC7A11, MDM2, and β-actin. (n = 3 independent experiments).

To screen for all p53 acetylation sites that could be affected by GINS4, we purified and analyzed p53 protein in GINS4 depletion-A549 cells and control cells by liquid chromatography-tandem mass spectrometry (LC-MS-MS). The results revealed lysine residue K351 of p53 as a new acetylation site regulated by GINS4 (SI Appendix, Fig. S5 B and C). Although this site has been previously characterized for p53 acetylation, it has not been reported to be related to cell ferroptosis (27). A study once introduced ectopic wildtype-p53 and several of its mutants (K101R, K164R, K351R, and K382R) into p53-null H1299 cells, K101 and K351 of p53, among all acetylation sites, affected SLC7A11 but not MDM2 which both were transcriptional target genes of p53. The results indicated that the acetylation state of K164 and K382 did not affect the expression of SLC7A11 and MDM2 (Fig. 6J). However, in p53 K101R and K351R mutant cells, which means incapable of p53 acetylation at these two sites, intracellular lipid peroxidation was at lower levels (Fig. 6K). These results suggested that K351, like K101, might be a key acetylation site for p53 affecting ferroptosis. Furthermore, in the above p53 and its mutants ectopically expressed H1299 cells, GINS4 depletion also did not affect ectopic p53 expression as well, including K351R. In addition, SLC7A11 expression was not affected by K351R mutation of p53 (Fig. 6L). Those results suggested that GINS4 recognized particular p53 acetylation pattern/code(s) based on K351Ac. Collectively, these data demonstrated a p53 K351Ac-dependent mechanism for GINS4 repressing cell ferroptosis.

GINS4 Inhibits Ferroptosis through Antagonizing the Acetylation of p53 by Snail.

Since Snail has been reckoned as an important molecule that regulates the acetylation of p53 and there is a mutual inhibition relationship between them (32, 39), we wondered whether Snail might be involved in such GINS4-regulated ferroptosis. Subsequently, bioinformatics analysis was conducted on the correlation of GINS4, Snail, and p53 (SI Appendix, Fig. S6A); this result further suggested the possible relationship between GINS4, Snail, p53, and ferroptosis molecule SLC7A11. Interestingly, we found that Snail mRNA and protein levels were decreased in GINS4 depletion cells while increased in GINS4 overexpressed cells (SI Appendix, Fig. S6 B–E), while for other EMT factors, such as ZEB1 and Twist, GINS4 did not alter their expression (SI Appendix, Fig. S6F).

By discovering upstream signaling pathways of Snail, p-ERK1/2 (40), HIF-1α (41), and β-catenin (42) as well, we noticed that HIF1α protein level was increased (Fig. S7 A and C). Further, we found that mRNA levels of HIF1α did not change (SI Appendix, Fig. S7 B and D). Even in the cells GINS4 knockout or reexpression, there were changes in HIF1α protein level (SI Appendix, Fig. S7E) but no changes in HIF1α mRNA level (SI Appendix, Fig. S7F). Therefore, we speculated that GINS4 upregulated Snail activity from the transcriptional level through HIF1α and activated HIF1α pathway by means of post-translational modification.

GINS4 reexpressed in GINS4 depletion cells could successfully rescue the expression level of Snail and remarkably decreased p53 protein level (Fig. 7A). Further, exogenous Snail expression in GINS4 depletion A549 cells did not change the GINS4 protein level (SI Appendix, Fig. S6 G–H and 7B), but increased p53 protein level (Fig. 7B); the result suggested that GINS4 caused the inhibition of the p53 protein level by Snail. Subsequently, we verified that GINS4, Snail, and p53 have endogenous and exogenous interactions (SI Appendix, Fig. S8 A and B), suggesting that a complex could be formed between them.

Fig. 7.

GINS4 suppresses ferroptosis by antagonizing p53 acetylation by Snail. (A) Western blot analysis of GINS4, p53, Snail, and β-actin protein levels in shCon, shGINS4 with vector transiently expressed, and shGINS4 with GINS4 transiently expressed A549 cells. (n = 3 independent experiments). (B) Western blot analysis of GINS4, p53, Snail, and β-actin protein levels in shCon, shGINS4 with vector transiently expressed, and shCon, shGINS4 with Flag-Snail transiently expressed A549 cells. (n = 3 independent experiments). (C) CCK8 assay revealed cell viability of shCon, shGINS4, and shGINS4 with Flag-Snail transiently expressed A549 cells treated with DMSO or RSL3 5 μM or RSL3 5 μM and Ferr-1 5 μM or RSL3 5 μM and Z-VAD-FMK 20 μM for 24 h. (n = 3 independent experiments). (D) GSH detected assay performed on shCon, shGINS4, and shGINS4 with Flag-Snail transiently expressed A549 cells treated with DMSO or Erastin 5 μM or Erastin 5 μM and Ferr-1 5 μM or Erastin 5 μM and Z-VAD-FMK 20 μM for 24 h. (n = 3 independent experiments). (E) CCK8 assay revealed cell viability of shCon, shGINS4, and shGINS4 with Flag-Snail transiently expressed H1299 cells treated with DMSO or RSL3 5 μM or RSL3 5 μM and Ferr-1 5 μM or RSL3 5 μM and Z-VAD-FMK 20 μM for 24 h. (n = 3 independent experiments). (F) GSH detected assay performed on shCon, shGINS4, and shGINS4 with Flag-Snail transiently expressed H1299 cells treated with DMSO or Erastin 5 μM or Erastin 5 μM and Ferr-1 5 μM or Erastin 5 μM and Z-VAD-FMK 20 μM for 24 h. (n = 3 independent experiments). (G) Western blot analysis of GINS4, Snail, p53, acetyl-p53 and β-actin protein levels in Snail WT and Snail KO A549 cells overexpressed with vector or GINS4. (n = 3 independent experiments). (n = 3 independent experiments). (H) Western blot analysis of GINS4, Snail, p53, and β-actin protein levels in Snail WT and Snail KO A549 cells. (n = 3 independent experiments). (I) Western blot analysis of GINS4, Snail, p53, and β-actin protein levels in Snail WT and Snail KO A549 cells with MG132 (20 μM) for 6 h. Lysates from Snail WT and Snail KO A549 cells were subjected to IP assay using anti-ubiquitin antibody. (n = 3 independent experiments). (J) Lysates from shCon, shGINS4, and shGINS4 with Flag-Snail transiently overexpressed A549 cells loaded at a ratio of 2:1:2 was subjected to IP assay. (n = 3 independent experiments). (K) Snail WT and KO H1299 cells transiently overexpressed wildtype and vector, EGFP-K101R and vector, EGFP-K351R p53 and vector, wildtype and GINS4, EGFP-K101R and GINS4, and EGFP-K351R p53 4 and GINS4, crude cell lysates were obtained for Western blot analysis using antibodies against GINS4, p53, SLC7A11, and β-actin. (n = 3 independent experiments). (L) A model of GINS4-mediated ferroptosis.

Snail acts as a bridge between p53 and HDAC1 (32), indicating that GINS4 was also involved in the formation of this complex and indirectly binds to Snail and p53. Next, cell viability and the intracellular GSH level were enhanced by exogenous Snail expression in GINS4 depletion A549 cells (Fig. 7 C and D) but not in H1299 cells (Fig. 7 E and F). As a positive control, Snail induced cell death that could be antagonized by the apoptosis inhibitor Z-VAD-FMK in both A549 and H1299 cells (Fig. 7 C and E). By CRISPR-Cas9 deletion of endogenous Snail and overexpression of GINS4 in A549 cells, we found that GINS4 could decrease total p53 protein and acetylated p53 only in control cells with normal Snail expression but not in Snail-deleted cells (Fig. 7G). Then, we further depleted GINS4 in Snail depletion cells and found that the p53 protein level no longer increases in Snail-deleted cells (Fig. 7H). Consistent with the previous results in this study, Snail decreased p53 protein degradation and GINS4 promoted p53 ubiquitination in Snail wild-type cells (Fig. 7I). By overexpressing Snail into GINS4 depletion cells and detecting p53 acetylation level, we found that Snail notably antagonized p53 protein and acetylation levels which was induced by GINS4 depletion (Fig. 7J). Lastly, by introducing wild-type-p53 and its mutants, including K101R, and K351R, into Snail depletion-GINS4 overexpressed-H1299 cells, we also found that GINS4 overexpression only reduced p53 protein level in Snail existing cells, and such regulation could be abolished by K351R mutation of p53 (Fig. 7K). Consistently, the SLC7A11 level was not affected by K351R mutation of p53 (Fig. 7K). Collectively, our study demonstrated a suppressive function of GINS4 for ferroptosis through specific recognition and suppression of p53 acetylation at K351 by Snail.

Discussion

Ferroptosis is a newly named programmed cell death process, signed by iron-dependent lipid peroxides accumulation (43, 44). Ferroptosis differs from other established cell death procedures and showed unique morphological and biological characteristics (45–47). The physiological and pathological role of ferroptosis in development has not been well-described (48). It is conventionally reckoned that GINS4 is a key component of the eukaryotic replisome (49). However, the emerging studies in recent years found that GINS4 is overexpressed in many kinds of cancer cells, which may exert an oncogenic role on cancer cell behaviors (4, 7, 8). Noteworthily, we recently discovered that activation of GINS4 is required for malignant progression of lung adenocarcinoma (4). According to bioinformatics analysis, we noticed that GINS4 was involved in the regulation of ferroptosis and shown a time effect of ferroptosis after ferroptotic inducer treatment (Fig. 1). Upon induction of ferroptosis, GINS4 depletion conferred promotion of ferroptosis characterized by mitochondrial shrinkage, lipid peroxidation redox-active, and GSH depletion (Fig. 2 B–H). Since GINS4 promoted G₁/S, we used double thymidine or nocodazole block and release assay to block the cells in G₁/S, S, G₂/M, and G₁ (50). We found that GINS4 protein levels were inconsistent in different cell cycles, especially less in G₂/M. Interestingly, upon ferroptosis activation, knockdown of GINS4 increased the sensitivity of cells in G₁/S, S, G₂/M, and G₁ to ferroptosis, particularly for G₂/M cells (Fig. 3 G and H). These results suggested that different cell cycles had different sensitivity to ferroptosis and might be related to the different expression levels of proteins in different cell cycles. Further, if we block cells in G₂/M and knock down or knock out GINS4, it will greatly increase ferroptosis in G₂/M cells, indicating that tumor cells might be better killed by trapping them in the G₂/M phase with chemotherapy drugs and then inducing ferroptosis. Next, CRISPR/Cas9-mediated GINS4 KO also induces nude mice tumor ferroptosis in vivo (Fig. 4). Although more experiments are required to determine the exact role of GINS4 in ferroptosis, our results indicate a function of GINS4 in LUAD development.

PTMs are an important layer of p53 functional regulation mechanism (38), especially acetylation is very important in the regulation of its function in ferroptosis and tumor inhibition. In fact, mouse p53 3KR mutations at K117, K161, and K162 diminished its ability to suppress cancer. However, 3KR p53 could still activate ferroptosis and inhibit tumor growth, but the additional mutation of p53 at K98 would abolish this ferroptosis-promoting process (22, 23). Based on 4KR, acetylation at position 139 for human was a key locus regulating p53 to inhibit tumor and embryonic death (24). In our article, we observed that GINS4 knockdown increased ferroptosis in a p53-dependent manner both in vitro and vivo (Fig. 5). We noticed that GINS4 reduced the stability of p53 protein (Fig. 6 E and F), for the ubiquitination level of p53 upregulated by decreasing its acetylation level (Fig. 6 G–I). That was, GINS4 reduced the acetylation level of p53. Based on this, we analyzed the effect of GINS4 on p53 acetylation sites by mass spectrometry. Then, through mass spectrometry, we found that p53 K351 might be a site which GINS4 affected (SI Appendix, Fig. S5 B and C). According to Fig. 6J, the mutation of K351R increased the SLC7A11 protein level more than that of WT. The lipid ROS level was detected for such p53 mutants as well, and the result confirmed that K351R decreases more lipid ROS level than WT (Fig. 6K). Furthermore, the site affected GINS4-suppressed ferroptosis, for K351R prevented the ability of GINS4 to induce SLC7A11 (Fig. 6L). Altogether, we hypothesized that GINS4 decreased p53 promoting ferroptosis by affecting the acetylation level of p53 K351 in human. This observation brought us a new example to underline the critical role of p53 acetylation in controlling ferroptosis. Since the K351 affected p53 regulation of cell cycle arrest (51, 52), we speculated that GINS4-regulated cell cycle might be also related to p53 regulated cell cycle. Lysine residues 351 have been reported to be ubiquitinated by MSL2, facilitating p53 localization of cytoplasmic, but not its degradation (26). Thus, the effect of GINS4 on the stability of the p53 protein might also be correlated with other acetylation modification sites, or the acetylation of K351 could also affect the protein stability of p53, which remains to be further explored.

EMT is a dynamic process of epithelial cells losing connections and polarity so that they change shape easy to move. This is a basic process by which epithelial cells break free from their adhesion between cells and cells to extracellular matrix. The loss of epithelial properties makes it easier for cancer cells to invade adjacent tissues and to propagate to remote sites. In addition, EMT entrusts cancer cells with chemotherapy resistance and reduces the efficiency of chemotherapy (53, 54). However, recent studies have shown that cancer cells with mesenchymal or metastatic specificity are more likely to be induced by ferroptosis if they are drug resistant or persistent (29, 55). The key EMT protein Snail is an important molecule that regulates the acetylation of p53, and there is a mutually inhibitory relationship between the two (33, 39). Further, according to the previous research of our research group, GINS4 could promote EMT (7). Through the detection of multiple EMT molecules, we found that the expression of Snail was significantly upregulated by GINS4 (SI Appendix, Fig. S6F). By discovering upstream signaling pathways of Snail, p-ERK1/2 (40), HIF-1α (41), and β-catenin (42) as well, we noticed that the HIF1α pathway was activated (SI Appendix, Fig. S7). Therefore, we speculated that GINS4 upregulated Snail activity from the transcriptional level through HIF1α. There was evidence that hypoxia could promote ferroptosis (44, 56), and our findings further confirmed that EMT-inclined cells were sensitive to ferroptosis. Snail overexpressing could antagonize GSH which was caused by knockdown of GINS4 (Fig. 7D). Therefore, additional effects of Snail-regulated p53 activity probably involved not only apoptosis but also ferroptosis.

In Snail KO A549 cells, the overexpression of GINS4 did not affect the protein level and acetylation level of p53 (Fig. 7G), nor did the knockdown of GINS4 affect the ubiquitination level of p53 (Fig. 7I). Overexpression of Snail in GINS4 knockdown cells did indeed reduce the originally elevated acetylation level of p53 (Fig. 7J). Those results suggested that GINS4 inhibiting ferroptosis might be through Snail antagonizing p53 acetylation. Moreover, GINS4 almost lost the ability to increase SLC7A11 protein level at Snail KO situation and change it when K351 mutant existed no matter Snail KO or not (Fig. 7K), suggesting that K351 was an important site that suppresses p53 acetylation mediated by Snail and further supplemented the regulation of Snail to p53 acetylation site. In our working model (Fig. 7L), GINS4 allowed Snail-associated HDAC1 to catalyze p53 deacetylation, thereby hastening the stability of p53, and then inactivation of p53 pathway, particularly through p53 K351 to suppress ferroptosis. While most studies emphasize the role of GINS4 in cancer development, we outlined a previously undiscovered role in its controlling ferroptosis. Our studies also indicated once GINS4 was inhibited, Snail would be successively suppressed, at which condition the tumor cells were more prone to ferroptosis for p53 activated. More importantly, most LUAD patients express wild-type TP53 (57), and GINS4 suppressed could effectively induce ferroptosis in EMT-specific LUAD (especially G₂/M cells). Indepth and detailed exploration of the above-mentioned functions of GINS4 protein was expected to provide new potential targets and theoretical basis for the treatment of p53 wildtype and metastatic LUAD.

Materials and Methods

Cell Culture and Stable Lines.

293T cells were grown in DMEM (Gibco). H460, H1299, and H358 cells were grown in RPMI-1640 (Gibco). A549 cells were grown in DMEM/F12 (Gibco). Culture medium was supplemented with 10% bovine calf serum (B7446, Sigma-Aldrich) and 1% penicillin/streptomycin/gentamicin. Cells were treated with mycoplasma clearance, and only mycoplasma-negative cells were used.

Constructs.

The construct expressing pEGFP-C3-p53 with point mutation(s) was generated by using Q5 ^® Ultra-Fidelity 2X premix (#M0492S, NEB) and QuickCut™ Dpn I (1609, TAKARA) according to the manufacturer’s protocol.

CRISPR/Cas9-Mediated GINS4 or Snail Knockout (KO).

293T cells were generated by transfection of lentiCRISPRv2 plasmid (#52961, addgene) which cloned into the single guide RNA with LipoMAX (#32012, Sudgen) according to the manufacturer’s protocol. Cell lines were selected with puromycin (1 μg/mL) for 3 d after transfection. Transfection efficiency was verified by real-time PCR. Guide RNA sequences targeting the GINS4 gene are 5′- CACCGATGACACATTCTACAATCTC-3′, 5′-AAACGAGATTGTAGAATGTGTC ATC-3′. Guide RNA sequences targeting the Snail gene are 5′- CACCGAGAAGGTCCGAGCACACGCC-3′, 5′-AAACGGCGTGTGCTCGGAC CTC-3′. Guide RNA sequences targeting the p53 gene are 5′- CACCGATGACACATTCTACAATCTC-3′, 5′-AAACTTGCTTGGGACGGCAAGGGGC-3′. Single colonies with GINS4 or Snail or p53 KO were selected and used for experiments.

CCK8 Viability Assay.

Here, 1,500 cells/well were seeded in a 96-well plate 24 h before 100 μL culture medium that contained 10% CCK8 (ZETA life, China) was incubated in each well for 2 h. OD values under 450 nm on a multiple-well plate reader (BioTek) were measured. The cell viability was expressed as A/B, where A was the absorbance value from the experimental group, and B was that from the control cells.

Iron Assay.

The relative iron concentration in cell lysates was assessed using an Iron Assay Kit (MAK125, Sigma-Aldrich) according to the manufacturer’s instructions. In brief, cells were plated in a sex-well plate (723001, NEST Biotechnology) and were homogenized with 5 vol of buffer. The insoluble material was abandoned after centrifugated at 13,000 g for 15 min at 4 °C. Sample wells were then added iron reducer. The buffer was gently mixed and then reacted for 30 min in the dark at 37 °C. Subsequently, 100 μL iron probe was put into each sample, and the mixed sample was reacted for 1 h at room temperature darkly. Finally, absorbance was measured at 593 nm by a microplate reader (Biotek).

Analysis of Lipid ROS Production.

Cells were washed once with PBS and incubated with PBS containing 10 μM C11-BODIPY (581/591) (#D3861, Thermo Fisher Scientific) at 37 °C for 30 min in a cell culture incubator. Cells then were washed, harvested by trypsinization, and resuspended in 500 μL fresh PBS. ROS levels were analyzed using a flow cytometer (Fortessa, BD Biosciences) with fluorescein isothiocyanate (FITC) green channel and Texas red channel.

GSH Assay.

The Glutathione Assay Kit (Beyotime Biotechnology) was used for these experiments. Cells in six-well plates were added with or without 5 μM of Erastin for 24 h and then washed with PBS once and centrifuged at 800 rmp 5 min to collect the supernatant. Then, 30μl of protein removal reagent S solution was mixed with cells which were three times the volume of cell precipitation. Then, after being fully vortexed, the samples were frozen and thawed twice in liquid nitrogen and 37 °C water bath. After 4 °C or ice bath for 5 min, and centrifugation at 4 °C, 10,000 g for 10 min. The supernatant was used for the determination of total glutathione. The sample should be kept at 4 °C temporarily, and the sample not determined immediately should be kept at −70 °C for 10 d. Finally, the absorbance at 405 nm was calculated (Biotek).

Transmission Electron Microscopy.

A549 cells and Xenograft tissue samples were fixed and then submitted to Laboratory of Electron Microscopy, Department of Pathology, Xiangya Hospital.

Cell Synchronization.

After being treated with 50 ng/mL nocodazole for 12 h, A549 cells were synchronized to mitosis. Mitotic cells were washed with PBS and released in a fresh medium for 4 h into the G1 phase. Double-thymidine block and release were used to collect G1/S-boundary, S-phase, and G₂-phase cells. In brief, cells were treated with 2 mM thymidine for 24 h, then washed with PBS, and cultured for 12 h. In addition, 2 mM thymidine was added for another 24 h, and then collected cells were in the G₁/S phase. The cells were collected in the S phase in 4 h and the G₂ phase in 8 h (50, 58). Cell cycle analysis was measured by using a FACSCalibur (BD Immunocytometry Systems) after staining cells at 37 °C for 30 min with 50 μg/mL propidium iodide, and analyzed results by Flow Jo software.

Immunoprecipitation (IP) Assay.

Control and GINS4-overexpressed cells were incubated with anti-Pan-AcK and precleared Protein G magnetic beads (Invitrogen). The immunocomplexes were tested by 10% (vol/vol) SDS–PAGE and immunoblot assay adopting indicated antibodies.

Mouse xenograft assay.

The xenograft tumor experiment was carried out as previously described (59). The 4-wk-old female nude mice used in this study were purchased from Hunan SJA Laboratory Animal Co., Ltd (Changsha). Then, 2 × 10⁶ GINS4 WT or KO A549 cells were injected s.c. into nude mice without matrigel. Tumor size was measured every 2 d with a caliper, and volume of tumor was calculated with the formula: L × W² × 0.5, for L represents the longest diameter and W means the shortest diameter. Then, 2 × 10⁶ p53 WT and KO A549 cells with shCon or shGINS4 overexpressing were injected s.c. into nude mice without matrigel. When tumors reached 60 to 100 mm³, p53 WT with shGINS4 mice were treated with ferrostatin-1 (S7243, Selleckchem) (20 mg/kg) in 2% DMSO, 50% PEG300, 5% Tween80, and 43% water by daily intraperitoneal injection. Tumors were measured three times a week. Mice were sacrificed and tumors were collected finally. Tumor tissue was made into a single-cell suspension for lipidROS assay using Tumor Dissociation Kit (Miltenyi Biotec).

Histology and Immunohistochemistry.

Xenograft tissue samples were obtained and immediately fixed in 10% formalin 4 °C overnight. Samples were then submitted to the Department of Pathology, Xiangya Hospital for paraffin embedding, hematoxylin and eosin (HE), and IHC staining. The antibodies used for IHC were anti-4HNE (1:50, Abcam, ab46545) and anti-Cleaved Caspase-3 (1:50, CST, 9664).

Statistical Analyses.

All the experiments were repeated at least three times apart from the nude mouse experiments. Data are analyzed as the mean ± SD or SEM. GraphPad Prism 8.0 software performed statistical analyses. Student’s t test was performed to decide the significant differences between two groups, and variance (ANOVA) was accessed to compare more groups. Differences were judged statistically significant in the case as P < 0.05 (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the iProX partner repository with the dataset identifier PXD040993 (60). All study data are included in the article and/or SI Appendix.