LM3 liver cancer cell line was obtained from BeNa culture collection (BNCC102270), HuH-7, SK-HEP-1, HepG2, Hepa1-6 liver cancer cell lines, and the HEK-293T cell line were obtained from the American type culture collection (ATCC, United States) and have been proven to be negative for mycoplasma contamination. All cells were incubated in a 37°C incubator with 5% CO2 and cultured in Dulbecoo’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS). YAP^S127A-LM3, YAP^S127A-Hepa1-6, shNT-LM3/SK-HEP-1, shYAP-LM3/SK-HEP-1 cells were generated by retroviral or lentiviral transfection.

The design of siRNA interference target and the preparation of double-stranded siRNA targeting ALOXE3 were completed by GenePharma (Shanghai, China). siALOXE3#1: sense, CCUGCUAUCAGUGGAUUGATT, antisense, UCAAUCCACUGAUAGCAGGTT. siALOXE3#2: sense, GCCCAAUGUUCGAUACUCATT, antisense, UGAGUAUCGAACAUUGGGCTT. siALOXE3#3: sense, GUACCUGAAUGGUGUCAAUTT, antisense, AUUGACACCAUUCAGGUACTT.Parental LM3 and YAP^S127A - LM3 cells in the logarithmic growth phase were routinely digested for cell counting. Cells prepared at 3.0 × 10⁵ cells/well were seeded in a six-well plate. They were cultured to cell confluence of 60–70% prior to transfection. After 24 h of transfection, 200 µL serum-free Opti-MEM plus 7.5 µL PEI, and 200 µL serum-free Opti-MEM plus 5 μg siRNA were mixed and incubated at room temperature for 5 min, and then the mixture was incubated together for 20 min. Transfection mixture was applied to each well with 2 ml of serum-free medium replaced. After transfection at 37°C for 6 h, DMEM containing 10% FBS was replaced. Transfected cells at 24 and 72 h were respectively used for qRT-PCR and Western blotting. Transfected cells at 24 h were re-seeded and treated with RSL3 to measure cell death.

The 8xGTIIC-luciferase reporter was from the Piccolo laboratory (Addgene plasmid 34615). pQCXIH-Flag-YAP-S127A was from the Guan laboratory (Addgene plasmid 33092). SFB-ALOXE3 was a gift from the Chu laboratory (Shandong University). Lentiviral vectors encoding shRNAs targeting human YAP were gifts from the Jiang laboratory (Memorial Sloan Kettering Cancer Center). For retroviral particle assembly, retroviral vector pQCXIH-Flag-YAP-S127A was co-transfected with gag/pol (Addgene) and VSV-G (Addgene) into 293T cells using PEI. Lentiviruses were produced by the co-transfection of the lentiviral vector with the Delta-VPR envelope and CMV VSV-G packaging plasmids into HEK-293T cells using PEI (Sigma-Aldrich). The progeny viruses released from 293T cells were filtered and collected 48 h after transfection, and used to infect HCC cells.

Briefly, cells were scraped into ice-cold radioimmunoprecipitation assay (RIPA) buffer and 1% Halt protease and phosphatase inhibitor cocktail. Briefly, cells were scraped into ice-cold radioimmunoprecipitation assay (RIPA) buffer and 1% Halt protease and phosphatase inhibitor cocktail. Cell lysates were centrifuged at 14,000 rpm for 15 min at 4°C, and supernatants were collected. Protein concentrations were determined using the Bradford method. Cellular proteins were separated by 10% SDS-PAGE, transferred onto PVDF membranes (0.45 μM, IPVH00010, Millipore), and blocked with 5% skim milk at room temperature for 1 h. Membranes were incubated with the following primary antibodies: anti-YAP1 Polyclonal Antibody (PA5-87568, Invitrogen, 1:2000), anti-ALOXE3 (ab118470, Abcam, 1:1000), anti-β-actin (ab8226, Abcam, 1:3000) and anti-GAPDH (60004-1-Ig, Proteintech, 1:5000) overnight at 4°C. They were washed and incubated with the horseradish peroxidase conjugated anti-rabbit or mouse IgG (H + L) (1:3000; Proteintech). The enhanced chemiluminescence solution (GE Healthcare) was used for visualizing protein expression.

Four HCC cell lines (SK-HEP-1, HuH-7, LM3, HepG2) were seeded in 96-well plates with high-density group (3 × 10⁴ cells/well) and low-density group (3 × 10³ cells/well). RSL3 (Sigma) was diluted to 4 μM, 2 μM, 1 μM, 0.5 μM, and 0.25 μM respectively. Ferrostatin-1 (MCE) was used as a specific ferroptosis inhibitor. Cells were collected for SYTOX Green (Invitrogen) and Hoechst 33342 (Beyotime) staining for analyzing cell death by microscopy and the number of dead cells was counted using ImageJ software.

Cell viability was measured by CellTiter-Glo luminescent cell viability assay (Promega). Briefly, 100 μL of assay reagent was added to each well and mixed at room temperature. After 15 min, intracellular ATP content was measured by a luminescence multilabel counter.

The ability of cell proliferation was assessed using CCK-8 (Catalog No. C0005, TargetMoI) according to the manufacturer’s instructions. Parental LM3 cells, ShYAP LM3 cells, Parental SK-HEP-1 cells and ShYAP SK-HEP-1 cells were seeded in 96-well plates (3 × 10³ cells/well) and cultured overnight, followed by treatment with RSL3 for 24 h. 90 μL culture medium plus 10 μL CCK-8 solution was added to each well and then incubated about 1–4 h at 37°C until the color of medium turned orange. The absorbance of each well was measured by a microplate reader at an excitation wavelength of 450 nm and then the cell activity was calculated.

For analyzing RSL3-induced cell lipid ROS, five HCC cell lines (SK-HEP-1, HuH-7, LM3, HepG2, Hepa1-6) were seeded in 6-well plates divided into high-density group (3 × 10⁵ cells/well) and low-density group (3 × 10⁴ cells/well) and cultured overnight, and then Fer-1 and/or RSL3 was added to the well plates. After 6 h cell induction, a ratiometric lipid peroxidation sensor (lipid peroxidation assay kit, ab243377 Abcam) was added to cells and incubated for 30 min. Then, cells were collected, resuspended in 250 μL of serum-free medium and examined by flow cytometry with FITC and PE channels [Ex = 488 nm and Em = 530 nm (FITC) or 572 nm (PE)] within 2 h of staining. The sensor in the lipid peroxidation kit changes its fluorescence from PE to FITC upon peroxidation by lipid ROS in cells, thus enabling ratio metric measurement of lipid peroxidation (the decreasing PE/FITC ratio means increasing lipid peroxidation).

For immunofluorescence (IF) staining, HCC cells were seeded at 3.0 × 10⁵ cells or 3.0 × 10⁴ cells per 35 mm confocal dish. After overnight cell culture, cells were fixed in 4% PFA for 15 min, and washed with PBS. They were blocked with 5% goat serum in PBS and then incubated at 4°C overnight with the anti-YAP (rabbit, PA5-87568, Invitrogen, 1:200). On the next day, cells were washed 3 times with PBS, and incubated with the Alexa Fluor 488 anti-rabbit IgG (Invitrogen, 1:200) at room temperature for 1 h. Nuclei were stained with 4,6-diamidino-2-phenylindole (DAPI, Beyotime). IF images were acquired using a confocal microscope (Nikon).

Immunohistochemistry was carried out to explore positive expressions of YAP1, ALOXE3, PTGS2, GPX4, Ki67 and 4HNE in human HCC specimens and xenografts generated from HCC cells. The high-throughput tissue microarray was purchased from Shaanxi Avila Biotechnology Co., Ltd. (Cat. No. DC-liv11014). Of the 208 cases involved in the commercial tissue microarray, 16 cases are normal liver tissue, and 180 cases are hepatocellular carcinoma or mixed hepatocellular carcinoma. Of the 180 cases of hepatocellular carcinoma, 6 sections fell off the slides or showed large area of necrosis, so 174 cases were retained for further analysis. Some cases lack the information about grades, which were excluded for analysis about grades. The thickness of tissue sections is 5 μm. Antigen retrieval was performed in citrate buffer in 100°C water bath for 2 min, aiming to eliminate endogenous peroxidase activity, and 5% BSA was added to block nonspecific epitopes. Sections were incubated with primary antibodies as follows: Anti-YAP1 (1:200; Invitrogen), anti-4HNE (1:200; Invitrogen), anti-ALOXE3 (1:400; Abcam), anti-PTGS2 (1:200; Invitrogen), anti-Ki67 (1:250; Invitrogen) and anti-GPX4 (1:250; Invitrogen) with a standard avidin-biotin HRP detection system according to manufacturer’s instructions (anti-mouse/rabbit HRP-DAB Cell & Tissue Staining Kit, R&D Systems). Tissues were counterstained with haematoxylin, dehydrated and mounted. In all cases, antigen retrieval was done with the BD Retrievagen (pH 6.0) Antigen Retrieval Systems. Slices were baked for 2 h, and the remaining steps were the same as the IHC steps. Primary antibodies of anti-YAP1 (1:200; Invitrogen) and anti-ALOXE3 (1:400; Abcam) were used. The product of the staining intensity score A and the staining positive cell ratio score B were considered as the scoring standard. Briefly, scoring standard of A: 0, negative staining; 1, weak intensity; 2, medium intensity; 3, high intensity. Scoring standard of B: 0, negative staining; 1, positive cell number ≤25%; 2, 25% < positive cell number <50%; 3, 50% < positive cell number <75%; 4, positive cell number ≥75%. The final score is obtained by multiplying A and B for analyzing.

Total RNA was extracted from YAP^S127A - LM3 cells and parental LM3 cells using TRIzol (Invitrogen). The quality of RNA was assessed with an Agilent 2,100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, United States) and checked with RNase free agarose gel electrophoresis. Eukaryotic mRNA was enriched by Oligo (dT) beads, while prokaryotic mRNA was enriched by removing rRNA by Ribo-ZeroTM Magnetic Kit (Epicentre, Madison, WI, United States). After enrichment, mRNA was fragmented using fragmentation buffer and reverse transcribed into cDNA with random primers. Second-strand cDNA were synthesized by DNA polymerase I, RNase H, dNTP and buffer, and the cDNA fragments were purified with the QiaQuick PCR extraction kit (Qiagen, Venlo, Netherlands), end repaired, poly (A) added, and ligated to Illumina sequencing adapters. The ligation products were size selected by agarose gel electrophoresis, PCR amplified, and sequenced using Illumina Novaseq6000 by Gene Denovo Biotechnology Co. (Guangzhou, China). For GSEA, the normalized enrichment score (NES) was used to quantify enrichment magnitude and statistical significance, respectively. The RNA-seq data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) database under the accession code PRJNA762931 [https://www.ncbi.nlm.nih.gov/bioproject/PRJNA762931].

Female nude mice with 4 weeks old were injected with 2 × 10⁶ Parental LM3 or YAP^S127ALM3 cells subcutaneously. Visible tumors appeared after 1 week. Tumor size was measured by an electronic caliper every 2 days and calculated using the formula: Tumor size (mm³) = 0.5 × length × width². After another 1 week, mice were randomly separated into 4 group of 10 mice per cage with roughly tumor size (100 mm³). They were treated with saline or 50 mg/kg Sorafenib (Selleck) dissolved in 5% DMSO+5% Tween-80 + 40% PEG-300 + 50% ddH₂O via gavage once a day. At 12th week, mice were sacrificed by cervical dislocation, and subcutaneous tumors were dissected, photographed and weighed. The tumors were appropriately segmented, and one part was immediately frozen in liquid nitrogen and stored at −80°C, and the remaining were soaked in 10% formalin, followed by sectioning and immunohistochemical staining.

The transcriptional activity of YAP with its primary binding partners, the TEAD family of transcription factors was monitored by an 8xGTIIC-luciferase reporter assay. The 8xGTIIC-luciferase reporter was from the Piccolo laboratory (Addgene plasmid 34615). Parental LM3 or YAP^S127A-LM3 cells were plated in 12-well plates at 1.5 × 10⁵ cells per well and incubated overnight. Cells were transfected the next day with 20 ng of 8xGTIIC-luciferase reporter and 1 ng of Renilla luciferase reporter plasmid (pRL-CMV; Promega). The transcriptional regulation of ALOXE3 was also investigated using the dual-luciferase reporter assay system. To analyze the regulation of ALOXE3 by TEAD4, we co-transfected pCMV6-Entry-TEAD4 with pGL3-ALOXE3 (−3000/+250) and pRL-CMV into HEK293 cells. At 24 h post-transfection, luciferase activities were assayed using the dual-luciferase reporter assay system (Promega) according to the manufacturer’s instructions. Firefly and Renilla luciferase activities were quantified using a luminometer, with firefly luciferase activity normalized for transfection efficiency based on Renilla luciferase activity. To analyze the regulation of ALOXE3 by YAP activation, we contransfected pGL3-ALOXE3 and pRL-CMV into parental cells and YAP^S127A-overexpressing cells.

A ChIP assay was carried out using a ChIP Kit (ab500, Abcam). Briefly, cell lysates were incubated with beads and anti-TEAD4 (ab58310) antibody. DNA fragments were assessed by quantitative PCR with reverse transcription (qRT–PCR) using the primer sequences. Quantitative PCR reactions generated products from the promoter region of the ALOXE3 gene. Primers were designed based on TEAD4-binding peak regions depicted in the ENCODE TEAD4 ChIP–seq datasets. The following primers were used: ALOXE3-1: forward 5′-GGC​GAA​ATC​CCG​TCT​CTA​CA-3′, and reverse 5′-CTC​TCC​TGC​CTC​AGC​CTC​CT-3'; ALOXE3-2: forward 5′-CAC​CAC​AAC​CTC​TGC​CTC​CT-3′, and reverse 5′-TGG​TGA​AAC​CCC​ATC​TCT​ACT​AAA-3'; ALOXE3-3: forward 5′-GCC​TCA​GCT​TCC​CAA​GTA​GCT-3′, and reverse 5′-TGG​GCG​ACA​AGA​GCA​ACA​CT-3'; ALOXE3-4: forward 5′-GGT​GGC​GCG​ATC​TTA​GCT​C-3′, and reverse 5′-TGG​CGG​GTG​CCT​GTA​GTC-3'.

HCC cells (SK-HEP-1, HuH-7, LM3, HepG2) and YAP^S127ALM3 cells were prepared to suspension at 2 × 10⁴ cells/mL. After centrifugation at 600 g for 5 min and mixture in 2.5% (v/v) Matrigel (Corning), cells were applied on 96-well Black/Clear Ultra-Low Attachment Microplate (22918053, Corning) and were incubated at 37°C, 5% CO₂ for 48–72 h formation of single spheroids of cells. Spheroids were then treated with RSL3 and Fer-1 in medium containing Matrigel for the indicated time. After 90% cells dead, cell viability was assessed by CellTiter-Glo® 3D Cell Viability Assay (Promega).

RNA was extracted using the TRIzol reagent (Invitrogen). cDNA was synthesized using Superscript first strand synthesis kit (Invitrogen) according to the manufacture’s protocol. RT–PCR was performed a TB Green PCR kit (TaKaRa, Otsu, Japan) on the Mxpro system to determine the expression levels of the genes. Primers used in qPCR were listed as follows: ALOXE3-Forward: 5′-CAA​GGA​CTC​TTG​GTA​CTG​TAG​C-3′, ALOXE3-Reverse: 5′-TAG​CCT​TCA​ATC​CAC​TGA​TAG​C-3’; GPX2-Forward: 5′-GGT​AGA​TTT​CAA​TAC​GTT​CCG​GG-3′, GPX2-Reverse: 5′-TGA​CAG​TTC​TCC​TGA​TGT​CCA​AA-3’; GSTk1-Forward: 5′-AGA​ACC​AGC​TCA​AGG​AGA​CC-3′, GSTk1- Reverse: 5′-CAT​CCA​CTT​CTC​TCC​CAG​CA-3’;

GSTM4-Forward: 5′-ACT​TGA​TTG​ATG​GGG​CTC​AC-3′, GSTM4- Reverse: 5′-TCT​CCA​AAA​TGT​CCA​CAC​GA-3’; SOD2-Forward: 5′-TTT​CAA​TAA​GGA​ACG​GGG​ACA​C-3′, SOD2- Reverse: 5′-GTG​CTC​CCA​CAC​ATC​AAT​CC-3’; GCLC-Forward: 5′-GGC​ACA​AGG​ACG​TTC​TCA​AGT-3′, GCLC- Reverse: 5′-CAG​ACA​GGA​CCA​ACC​GGA​C-3’; IDH1-Forward: 5′-CAC​TAC​CGC​ATG​TAC​CAG​AAA​GG-3′, IDH1-Reverse: 5′-TCT​GGT​CCA​GGC​AAA​AAT​GG-3’; GAPDH-Forward: 5′-CAG​GTG​GTC​TCC​TCT​GAC​TT-3′, GAPDH-Reverse: 5′-CCA​AAT​TCG​TTG​TCA​TAC​CA-3’.

All statistical analyses were performed using GraphPad Prism 8.0 Software. Data in Figure 8A are presented as mean ± SEM. Other data are presented as mean ± SD. p values were determined by Student’s two-tailed t-test or one-way ANOVA. p-values < 0.05 were considered to be statistically significant.

The expression of YAP was investigated by Tumor Immune Estimation Resource (TIMER). We found that YAP was highly expressed in the tumor tissues of cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), liver hepatocellular carcinoma (LIHC) and stomach adenocarcinoma (STAD) (Figure 1A). We performed immunohistochemistry to investigate the expression and localization of YAP in human HCC tissues and normal tissues using a high-throughput tissue microarray analysis. In HCC tissues, the total YAP staining was enhanced (Figures 1B,C), and YAP nuclear staining was also more prevalent than in nontumorous regions (Figures 1D,E). However, there was no significant correlation between total YAP or nuclear YAP level and Tumor-Node-Metastasis (TNM) stages (Figure 1F) or grades of HCC (Figure 1G).

Intercellular interaction of epithelial cells can signal to intracellular Hippo pathway (Kim, Koh et al., 2011). The Hippo pathway involves multiple players, such as the tumor suppressor Merlin and a kinase cascade comprising Mst1/2 and Lats1/2. Lats1/2 suppresses the function of the pro-oncogenic transcriptional cofactor YAP by inducing its nuclear exclusion through phosphorylation at its S127 residue. We examined the localization of YAP in 4 HCC cell lines. As expected, as HCC cells grew more confluent, decreased nuclear localization of YAP were observed (Figure 2A). Previous study has confirmed the role of YAP in promoting ferroptosis in mesothelioma cells, and also suggested that the transcriptional regulation mediated by YAP-TEAD interaction is responsible for this effect (Wu, Minikes et al., 2019). To test this possibility in HCC cells, we seeded different numbers of HCC cells in culture dishes. We found that higher cell confluence conferred significant resistance to ferroptosis induced by a covalent inhibitor of GPX4, RSL3, which is commonly used as an inducer for ferroptosis (Stockwell, Friedmann Angeli et al., 2017) (Figures 2B,C). Ferrostatin-1 (Fer-1), a selective inhibitor of ferroptosis, significantly blocked cell death of sparse cells (Figure 2D). Among the HCC cell lines tested, high-density LM3 cells showed the strongest resistance at high cell density to ferroptosis induction compared to other cells (Figure 2B). As cell density increased, E-cadherin expression increased and became enriched at sites of cell–cell contact in HuH-7, HepG2 and LM3 cells (Supplementary Figures S1A,B). However, in SK-HEP-1 cells, E-cadherin was located in the cytoplasm even when approaching high confluence (Supplementary Figure 1A). To better mimic an in vivo context, we cultured HCC cells into 3D multicellular tumor spheroids. Consistent with the results from 2D cell culture analysis, RSL3 triggered more prominent cell death (measured positive by SYTOX Green staining) in spheroids formed by HuH-7 cells and HepG2 cells (Figure 2E). Upon RSL3 treatment, cell viability was also most significantly reduced in HuH-7 and HepG2 spheroids (Figure 2F), whereas LM3 and SK-HEP-1 spheroids showed resistance.

The pro-oncogenic transcriptional co-activator YAP is one of the downstream effectors suppressed by Merlin-Hippo signaling (Li, Cooper et al., 2012; Varelas and Wrana 2012). In addition, previous studies and our experiments have shown an exquisite correlation between YAP/TAZ activity and cell-density-dependent ferroptosis in different tumor cell lines (Wu, Minikes et al., 2019; Yang, Ding et al., 2019; Yang, Huang et al., 2020). These observations prompted us to perform a series of functional experiments to determine whether YAP promotes ferroptosis.

First, we utilized the constitutively active YAP mutant, S127A (serine-127 mutated to alanine). This mutant cannot be phosphorylated by Lats1/2 at the S127 residue and thus is retained in the nucleus to exert its transcriptional co-regulatory activity. Compared with the parental control, LM3 cells ectopically expressing the YAP^S127A mutant were unable to exclude the YAP mutant from the nucleus, even when cultured at a high density (Figure 3A). YAP is functionally activated with overexpression of YAP^S127A, measured by an 8xGTIIC-luciferase reporter assay that monitors the transcriptional activity of YAP with its primary binding partners, the TEAD family of transcription factors (Figure 3B). LM3 or Hepa1-6 HCC cells that express YAP^S127A were markedly more sensitive to ferroptosis induced by RSL3 (Figures 3C–J). YAP^S127A also enhanced lipid ROS generation (Figures 3K,L). Consistently, the YAP ^S127A mutant also sensitized ferroptosis in 3D tumor spheroids derived from LM3 cells (Figure 3M). We found that YAP ^S127A mutant enhanced ferroptosis induced by imidazole ketone erastin (IKE) (Figure 4A), a potent and metabolically stable analog of erastin that has been validated for in vivo use, as well as sorafenib (Figure 4B). Treatment with the lipid peroxide-trapping agent ferrostatin-1 (Fer-1) prevented cell death triggered by IKE in LM3 cells with overexpression of YAP ^S127A (Figures 4C,D). Notably, Fer-1 didn’t completely prevent cell death triggered by sorafenib, suggesting the existence of ferroptosis-independent cell death. Furthermore, knockdown of YAP in sparse LM3 and SK-HEP-1 cells conferred resistance to ferroptosis (Figure 5). These results confirm the role of oncogenic activation of YAP signaling in sensitizing ferroptosis in HCC cells.

To investigate the mechanism of YAP activation in sensitizing ferroptosis, we performed RNA-sequencing in parental and YAP ^S127A mutant-overexpressing LM3 cells. In total, 567 differentially expressed genes were identified, including 370 upregulated and 197 downregulated genes in YAP ^S127A mutant-overexpressing LM3 cells (Figure 6A). In order to functionally annotate the expression differences between the 2 groups, we performed a gene set enrichment analysis (GSEA) and selected the signaling pathways related to ferroptosis for further analysis (Figure 6B). YAP ^S127A mutant-overexpressing LM3 cells showed a significant enrichment of genes in categories related to peroxisome proliferator activated receptor signaling pathway, while parental cells were strongly characterized by the expression of genes involved in cysteine and methionine metabolism and glutathione metabolism. Among genes comprising the leading edge of Enrichment score in the category of peroxisome proliferator activated receptor signaling pathway in YAP ^S127A-overexpressing cells compared with parental cells, ALOXE3 belongs to the mammalian lipoxygenase family (Figure 6C), which has been revealed as a ferroptosis regulator (Yang, Kim et al., 2016; Xia, Liu et al., 2020; Yang, Liu et al., 2021). Based on publicly available ENCODE TEAD4 ChIP-seq datasets (GSM1614035, GSM1010860, GSM1010868, GSM1010772 and GSM1010875), we found that ALOXE3 is a putative YAP-TEAD4 gene target in various cancer cell lines. As shown in Figures 6D,E, the expression of ALOXE3 increased upon YAP ^S127A overexpression, suggesting that ALOXE3 might be transcriptionally regulated. As cell density increased, the level of ALOXE3 decreased in HuH-7, HepG2 and SK-HEP-1 cells (Figure 6F). To validate the contribution of YAP activation on ALOXE3 expression, the transcriptional regulation of ALOXE3 was investigated using a dual-luciferase reporter assay system. pGL3-ALOXE3 (−3000/+250) was co-transfected with pCMV6-Entry-TEAD4 into LM3 cells. As shown in Figure 6G, transfection of TEAD4 plasmid significantly increase the luciferase activity compared to cells transfected with empty plasmid. In addition, co-transfection of YAP ^S127A and the ALOXE3 promoter region also led to an increase in luciferase activity compared to parental cells (Figure 6H). ChIP analysis of TEAD4 binding to the ALOXE3 promoter in LM3 cells using beads control or an anti-TEAD4 antibody. To validate the TEAD4 binding to the regulation regions of ALOXE3 in HCC cells, we did ChIP assay with anti-TEAD4 antibody. We found that TEAD4 binds to the promoter regions of ALOXE3 genes in parental LM3 cells (Figure 6I), and the binding was enhanced by overexpression of YAP^S127A (Figure 6J). Effective knockdown of ALOXE3 protected YAP ^S127A-overexpressing cells from ferroptosis upon RSL3 treatment (Figures 6K−M), while overexpression of ALOXE3 restored ferroptosis induced by RSL3 in confluent cells (Figures 6N,O). These data indicate that ALOXE3 is essential for YAP-sensitized ferroptosis.

We first analyzed the correlation between YAP mRNA expression and ALOXE3 in hepatocellular carcinoma using data from the TCGA. The mRNA expression of YAP is positively correlated with ALOXE3 (Figure 7A). We then performed immunohistochemistry to investigate the correlation between nuclear YAP and ALOXE3 in human HCC tissues and normal liver tissue using a high-throughput tissue microarray analysis (Figure 7B). As shown in Figure 7C, ALOXE3 level was significantly higher in HCC patients than in normal liver tissues (p < 0.001) but showed no significant correlation with TNM stages and grades of HCC (Figures 7D,E). Statistical analysis showed that the level of nuclear YAP in the primary HCC lesions was positively correlated with the expression of ALOXE3 ( Figure 7F) (r = 0.7046, p < 0.0001). Among nuclear YAP negative tumors samples, 53.13% tumor samples exhibited negative staining of ALOXE3. However, among nuclear YAP positive tumors, only 3.21% exhibited negative staining of ALOXE3 (Figure 7G), suggesting the potential regulation of ALOXE3 by YAP.

To explore the cancer therapeutic potential of ferroptosis induction in HCC with enhanced activation of YAP, we used parental LM3 cells and cells ectopically expressing the YAP^S127A mutant to produce subcutaneous xenograft tumors in athymic nude mice. In mice xenografted with these cells, we allowed the average volume of tumors to reach ∼100 mm³, and then started ferroptosis induction by sorafenib. We found that overexpression of YAP^S127A mutant rendered xenograft tumors significantly more sensitive to sorafenib (Figures 8A,B). H&E staining showed that sorafenib administration resulted in significantly larger zones of coagulation necrosis in tumors generated by YAP^S127A mutant LM3 cells compared with tumors generated by parental cells (Figure 8C). As confirmed by immunohistochemical analysis, levels of YAP in nucleus and ALOXE3 in cytoplasm were greatly increased in tumor tissues generated by YAP^S127A mutant LM3 cells. Immunohistochemical analysis of 4HNE and PTGS2, markers of lipid peroxidation and ferroptosis, supported the synergistic effect of sorafenib in inducing tumor ferroptosis in HCC cells with enhanced activation of YAP in vivo (Figure 8C). Ki67 expression, indicative of cell proliferation, was reduced (Figure 8C).

To further investigate the correlation of YAP expression and the survival rate of patients with HCC, we used the survival information retrieved by the GEPIA database to further validate our results analysis. We found that YAP level was not significantly associated with the overall survival of patients (Figure 8D). However, for patients who had been treated with sorafenib, patients with tumors exhibiting high level of YAP had better survival rates than those with tumors exhibiting low YAP expression level (Figure 8D). A significantly negative association between GPX4 status and the overall survival rate was also observed in HCC patients with sorafenib treatment (Figure 8E). Thus, enhancing the response of HCC cells to sorafenib-induced ferroptosis may become an effective strategy to improve the antitumor performance of sorafenib in vivo.