Female 5-week-old age C57BL/6 mice, weighing 18–25 g, were purchased from SPF (Beijing) BIOTECHNOLOGY Co. Ltd. One group (N = 3) was fed a standard chow diet (soybean meal, fish meal, Brewer’s yeast, vegetable oil, Bran, corn, wheat, vitamins, pantothenic acid, minerals). A second group (N = 3) was fed a high fat diet (standard feed 70% + lard 20% + cholesterol 5% + yolk 5%). All mice were provided with normal water (Soret et al., 2020). All mice were housed in a 12 h light–12 h dark cycle in a standard facility. At 6 weeks, livers were harvested, washed in saline solution, and formalin-fixed. All animal experimental procedures followed the guidelines of the Ethical Committee of the Medical College of Qingdao University (Animal Ethics Approval Number: QDU-AEC-2022097).

Clinical information and surgical specimens were taken patients with NAFLD-related HCC from the Department of Hepatobiliary Surgery of Qingdao Central (Oncology) Hospital, Medical College of Qingdao University (Qingdao, China). All patients were diagnosed and treated for the first time and were not exposed to any prior treatment. The human sample collection procedure was approved by the Ethics Committee of the Medical College of Qingdao University. The informed consent was obtained before the collection of human specimens (Human Ethics Approval Number: QDU-HEC-2022055).

Protease Inhibitor cocktail (Targetmol, cat#C0001), Phosphatase Inhibitor cocktail (Targetmol, cat#C0002), Phosphatase Inhibitor Cocktail II (Targetmol, cat#C0003), Bradford Assay Reagent (Thermo, Prod#1863028), Transfer Membrane (Merck Millipore Ltd., LOT:IPVH07850), Tris (Solarbio, Cat#T8060), Ammonium persulfate (Sinopharm Chemical Reagent Co. Ltd.), 30% Acrylamide/Bissoln (MDBio, lnc, Lot#F1110419), 1,2 Bis (dimethylamino)ethane-TEMED (Solaibio, Cas#110-18-9), Sorafenib (Meilunbio, Cat# MB1666), Erastin (MCE, Cat No.:HY-15763), Fer-1 (MCE, Cat No.: HY-100579), EntransterTM-R4000 (Engreen Biosystem Cat#4000-3), TriZol (SparkJade, Cat#AC0101-B), cDNA synthesis Mix (TRANS, Cat# AE311-04), HE Staining Kit (BASO, Cat#BA4025), Rhamsan gum (Sinopharm, Cat#10004160), OPTI-MEM medium 1X (Gibco, Cat#31985), Polymeric anti-rabbit IgG-HRP kit (BOSTER, Cat#SV0002), Polymeric anti-mouse IgG-HRP kit (BOSTER, Cat#SV0001), Endogenous peroxidase blockers (BOS-TER, Cat# AR1108), EDTA Antigen repair solution (10X) (BOSTER, Cat#AR0023), Anti-body dilutions (BOSTER, Cat#AR1016), PBS Bleach (BOSTER, Cat#AR0030), DAB Color Development kit (Yellow) (BOSTER, Cat#AR1022), Mayor hematoxylin (BOSTER, Cat#AR0005), Neutral gum (BOSTER, Cat#AR0038), Protein Marker 10-180kd (ABclonal, Cat#RM19001). Formalin (Biosharp, Lot 70081800), ACSL3 antibody (Santa cruz, 1:100, Cat# sc-166374), ACSL4 Rabbit pAb (ABclonal, 1:1,000, Cat#A6826), MEF2D antibody (Santa cruz, 1:500, Cat# sc-271153), GPX4 Monoclonal antibody (proteintech, 1:1,000, Cat No.: 67763-1-Ig), Ferritin heavy chain Polyclonal antibody (proteintech, 1:1,000, Cat No.: 11682-1-AP), Vinculin Monoclonal antibody (proteintech, 1:10,000, Cat No.: 66305-1-Ig), Mouse monoclonal [JDC-10] Anti-Human IgG Fc (Abcam, Cat#ab99757).

Staining with the standard H&E staining technique in paraffin-embedded sections was performed according to the commercially available methods by the Boster Biological Technology center. The samples were processed by dewaxing the sections, hematoxylin sustaining for 3 min, the hydrochloride alcohol solution for 2 s dying after washing by water, staining with eosin solution for 1 s, dehydration by absolute ethyl alcohol for 2 s after washing by water. All the cell morphologic changes in liver tissue were observed under the OLYMPUS BX51 microscope and the images were collected by a Leica pathology fish slide scanner.

IHC staining was performed according to the commercially available methods by the Boster Biological Technology center. ACSL3 expression was investigated using unstained, formalin-fixed, paraffin-embedded tissue sections and was detected using anti-ACSL3 antibody. The tissues were incubated with primary antibodies overnight at 4°C and then with an HRP-conjugated secondary antibody at 37°C. Whole slide imaging magnification was performed using BX51 OLYMPUS microscope and APERIO VERSA 8 of Leica slide scanner. Images were inspected using TEKSQRAY Digital Pathology viewer software.

Huh7 and PLC/PRF/5 cells were purchased from national collection of authenticated cell cultures (Shanghai, China) and maintained in Dulbecco’s Modified Eagle Medium (DMEM) medium (Gibco, Cat#SH8122258) supplemented with 10% fetal bovine serum (FBS), 1% streptomycin-penicillin (Gibco, Cat#15140-122). All cells used in this work were maintained in culture media at 37°C in a humidified atmosphere with 5% CO₂.

Total proteins were extracted with RIPA lysis buffer containing protease and phosphatase inhibitors. The protein concentration was determined using a Bradford reagent kit. Proteins were separated by SDS-PAGE and transferred to polyvinylidene fluoride membranes. Membranes were blocked with 5% skim milk for 1 h, and then incubated with primary antibody at 4°C overnight. Membranes were then washed with Tris-buffered saline containing Tween 20, and incubated with horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG secondary antibodies for 1 h at room temperature, and finally developed with enhanced chemiluminescence kit. Vinculin served as the loading control. ACSL3 antibody (Santa cruz, 1:100, Cat# sc-166374), ACSL4 Rabbit pAb (ABclonal, 1:1,000, Cat#A6826), MEF2D antibody (Santa cruz, 1:500, Cat# sc-271153), GPX4 Monoclonal antibody (proteintech, 1:1,000, Cat No.: 67763-1-Ig), Ferritin heavy chain Polyclonal antibody (FTH1) (proteintech, 1:1,000, Cat No.: 11682-1-AP), Vinculin Monoclonal antibody (proteintech, 1:10,000, Cat No.: 66305-1-Ig).

ACSL3 siRNA sequences included sense (5′-3′) CCU​GGA​UGU​GAU​ACU​UUA​GAU and antisense (5′-3′) AUC​UAA​AGU​AUC​ACA​UCC​AGG. These sequences were purchased from Sangon (Shanghai, China). Huh7 and PLC/PRF/5 cells were seeded in 6-well plates and maintained to 50% confluence prior to transfection siRNA and EntransterTM-R4000 were mixed in OPTI-MeM medium, and transfection complexes were placed at room temperature 15 min prior to adding to DMEM. After 6 h of transfection, media was exchanged and transfected cells were incubated for 48 h mRNA and protein expression levels of ACSL3 were evaluated using qPCR.

Total RNA was extracted with Trizol reagent (Takara, Cat#9109) according to the manufacturer’s protocol. KOD -Plus- Neo (TOYOBO, Cat#KOD-401) was used to amplify the complete open reading frame (ORF) of ACSL3. Amplify ACSL3 gene was connected to pcDNA3.1 by Not I and Xba I restriction sites for expression in E. coli. In this study, Lv-GFP, Lv-MEF2D, Lv-scramble, and Lv-shMEF2D lentivirus construction details have been described in previous publications by our group (Ma et al., 2014).

Total RNA was extracted using the TriZol reagent and was measured using a Thermo Scientific NanoDrop instrument. Complementary DNA was synthesized using EasyScript^® One-Step gDNA Removal and cDNA Synthesis SuperMix. qRT-PCR was measured using Universal SYBR Green Fast qPCR Mix. All genes were normalized to GAPDH. mRNA expression levels of ACSL3, ACSL4, GPX4 and FTH1 were quantified using qPCR.

Viable cells were measured using a Cell Counting Kit-8 (CCK-8). In short, Specific treatment of cells were dispersed by Trypsin and seeded into 96-well plates at 4 × 103 cells/well. After 24 h of incubation, cells were treated with additional drugs (Sorafenib) at concentrations of 0 μM, 2 μM, 4 μM, 8 μM, 16 μM, 32 μM. After 24 h of treatment, the culture medium exchanged with 90 μL fresh complete medium and 10 μL CCK8 reagent (TargetMol, Boston, United States) were added to each well, followed by incubation for 1 h. The absorbance of each well was measured at 450 nm with a VICTOR^® Nivo™ microplate reader (PerkinElmer, Waltham, United States).

LPO levels were detected by Bodipy 581/591 C11 (MKBio, MX5211-1 MG, China). After the cells were pretreated with the drugs (sorafenib, erastin, fer-1), Bodipy 581/591 C11 was added and incubated for 20 min. Fluorescence was then measured at 595 and 510 nm using the SpectraMax i3x, Molecular Devices, United States. LPO levels were analyzed by calculating the ratio of 510/595 fluorescence intensity. The levels of ROS were detected using fluorescent probe DCFH-DA (Beyotime, S0033S, China). After the cells were pretreated with the drugs (sorafenib, erastin, fer-1), appropriate volume of diluted DCFH-DA was added and incubated for 20 min. Intracellular ROS can oxidize non-fluorescent DCFH to produce fluorescent DCF, and the level of intracellular ROS can be evaluated by detecting the fluorescence of DCF.

ChIP assay was performed using the ChIP assay kit (Beyotime, Cat No P2078). In brief, treated cells were cross-linked by incubation with 1% formaldehyde at 37° for 10 min and sonicated to break the DNA to a length ranging from 200 to 1,000 base pairs. Cell lysates were pre-clarified with protein A/G beads and then incubated overnight at 4°C with protein A/G beads coated with anti-MEF2D antibody (2 μg, Santa cruz). Anti-rabbit immunoglobulin G (IgG) was also used as a negative control. After extensive washing, the bead-bound immunocomplexes were eluted using elution buffer. To remove protein and genomic DNA cross-links, 5 M NaCl samples were added and heated at 65°C for 4 h, followed by proteinase K treatment and further incubation at 45°C for 1 h. The bound DNA fragments were subsequently purified and subjected to real-time PCR using specific primers.

GSEA datasets GSE63067, GSE185051 and GSE180882 were used to investigate several KEGG metabolic pathways (https://www.ncbi.nlm.nih.gov/geo/). Expression profiles of ACSL3 in the murine livers were generated using the Tabula Muris database (https://tabula-muris.ds.czbiohub.org). The mRNA and protein expression of ACSL3 in liver cancer compared to normal tissues in the UALCAN databases (http://ualcan.path.uab.edu). The GEPIA2 database (http://gepia2.cancer-pku.cn/) was used to study the correlation between ACSL3 and hub genes expression. The prognostic value of ACSL3 and related genes was tested using the Kaplan-Meier Plotter (https://kmplot.com/analysis/). Gene association study and enrichment pathways were obtained using the Linked Omics database (http://linkedomics.org/). Transcription factor binding sites in the promoter regions of ACSL3 under study were carried out using AliBaba2.1 software (http://generegulation.com/pub/programs/alibaba2/index.html).

All data were analyzed using appropriate statistical analysis methods with GraphPad (Prism V.8.3.0) and data was presented as mean ± SEM.

Using Gene Set Enrichment Analysis (GSEA), we observed significant enrichment in fatty acid metabolism in tissues affected by NAFLD compared to healthy human liver tissues (Figure 1A). We also identified that ACSL3 was aberrantly expressed among genes linked to fatty acid metabolic pathways in NAFLD (Figure 1B), and ASCL3 mRNA levels were upregulated in NASH (Figure 1C). In NAFLD mice, H&E staining of liver tissue demonstrated an increase in scattered lipid vacuoles and compressed liver sinusoids compared to the control groups. Furthermore, IHC staining showed that increased ACSL3 protein expression (indicated by brown particles) in the liver tissue of NAFLD mice (Figure 1D). Single-cell sequencing of mouse tissues revealed that among the top five liver cell types, ACSL3 expression was predominantly confined to hepatocytes (Figure 1E).

To investigate the role of ACSL3 in the progression of NAFLD to HCC, we assessed cell morphology and ACSL3 antigenicity in human NAFLD-related HCC tissues, ranging from mild to severe. H&E and IHC staining showed that elevated levels of lipid vacuolization and overall ACSL3 expression in HCC tissues were positively correlated with the increased severity of NAFLD (Figure 1F). For information on patients with NAFLD-related HCC, we adopted the FLIP-SAF evaluation standard for the histological grading of NAFLD tissues (Supplementary Table S1).

By analyzing The Cancer Genome Atlas (TCGA), we found that the gene expression boxplot indicated significantly higher expression of ACSL3 in liver tumors compared to adjacent normal tissues (Figure 1G). Consistent with this, analysis using the UALCAN database also confirmed higher mRNA and protein expression of ACSL3 in liver cancer (Figure 1H). To assess the prognostic value of ACSL3 in HCC, we analyzed survival curves using TCGA database. Our analysis revealed that low expression levels of ACSL3 were associated with a higher survival rate in HCC patients (Figure 1I).

These results suggested that ACSL3 expression was elevated in NAFLD and HCC and promotes the progression from NAFLD to HCC, while high expression of ACSL3 was positively correlated with poor prognosis of HCC patients.

To investigate the expression networks and signaling pathways related to ACSL3, we examined its co-expression profiles using the LinkedOmics database. As illustrated in the volcano plot (Figure 2A), red points indicated positive correlations with ACSL3, while green points represented negative correlations in HCC. Simultaneously, the heatmaps displayed the top 50 genes positively and negatively correlated with ACSL3 in HCC (Figures 2B, C; Supplementary Tables S2, S3).

GSEA indicated that the functional genes co-expressed with ACSL3 were primarily enriched in fatty acid biosynthesis and the ferroptosis signaling pathway (Figure 2D). Since ferroptosis is induced by the peroxidation of polyunsaturated fatty acids and regulation of lipid metabolism is a key determinant of ferroptosis sensitivity, the aberrant expression of ACSL3 in HCC may be closely linked to cellular ferroptosis. We evaluated the ferroptosis role of ACSL3 in HCC by overexpressing ACSL3 in PLC/PRF/5 cells, which have low endogenous levels of ACSL3. We then assessed the mRNA and protein levels of ferroptosis biomarkers by qPCR and Western blotting (Wu et al., 2019; Fang et al., 2021). Compared to the control group, the expression of the ferroptosis-positive regulator ACSL4 was downregulated, while the expression of the ferroptosis-inhibitory genes GPX4 and FTH1 was increased (Figures 2E, F).

These results collectively demonstrated that increased expression of ACSL3 played a significant role in ferroptosis within HCC.

To determine the association of ACSL3 with sorafenib resistance in HCC, we established resistant cell lines, Huh7 (Huh7-DR3) and PLC/PRF/5 (PLC/PRF/5-DR3), by culturing them with increasing doses of sorafenib (Figure 3A). Western blotting showed that ACSL3 expression was increased in sorafenib-resistant HCC cells compared to the control sorafenib-sensitive cells (Figure 3B). Subsequently, PLC/PRF/5 and PLC/PRF/5-DR3 cells were injected subcutaneously into mice. After tumor formation, the mice were treated with sorafenib. IHC staining revealed that ACSL3 expression was significantly higher in tumors derived from drug-resistant cells compared to the control group (Figure 3C).

Subsequently, we transiently knocked down ACSL3 using specific siRNAs. As shown in Figure 3D, the activity of ACSL3-silenced Huh7 cells decreased following sorafenib treatment. By examining the levels of redox species, we validated the role of ACSL3 in ferroptosis and found that ACSL3 silencing consistently enhanced sorafenib-induced production of LPO and ROS. Additionally, we added the ferroptosis inhibitor fer-1 as a control during treatment (Figure 3F). Moreover, overexpression of ACSL3 in PLC/PRF/5 cells conferred resistance to sorafenib treatment and enhanced cell activity compared to the control group (Figure 3E). Cells with high ACSL3 expression were resistant to sorafenib-induced increases in LPO and ROS levels. The ferroptosis inducer erastin was also included as a control in the treatment (Figure 3G).

The results consistently demonstrated that the upregulation of ACSL3 expression in HCC protects against sorafenib-induced ferroptosis.

To understand the specific reasons of abnormal expression of ACSL3 in HCC, we conducted further exploration. Previous studies in our group have shown that MEF2D is overexpressed in HCC and sorafenib resistant-HCC. Abnormal expression of MEF2D enhances resistance to sorafenib therapy and is associated with poor prognosis in patients (Ma et al., 2021). Sorafenib is not only a multi-targeted tyrosine kinase inhibitor but also an inducer of ferroptosis, triggering oxidative stress in cells. We investigated whether there is a critical link between MEF2D, ACSL3 and ferroptosis in HCC.

Previously, our research group analyzed the gene differential expression in PLC/PRF/5 cells with or without MEF2D through RNA sequencing (Ma et al., 2014). GSEA revealed enrichment of the peroxisome pathway in MEF2D-silenced cells (Figure 4A). Peroxisomes are key mediators of cellular oxidative stress responses. Based on this, we found downregulated expression of several candidate genes related to oxidative stress in MEF2D-silenced HCC cells by qPCR. Notably, there was a significant difference in the expression of ACSL3, a gene associated with ferroptosis (Figure 4B). Additionally, the results indicated that in PLC/PRF/5 cells infected with Lv-GFP or Lv-MEF2D, the mRNA and protein expression levels of ACSL3 were positively correlated with MEF2D expression (Figure 4C).

To investigate the mechanism by which MEF2D directly regulates ACSL3, we used a bioinformatics approach to analyze the upstream regions of the ACSL3 transcription sites and identified a potential MEF2 recognition element (MRE) (Figure 4D). Subsequently, we performed a luciferase reporter assay in which luciferase expression was driven by the promoter of ACSL3. Overexpression of MEF2D enhanced the promoter activity of ACSL3. Conversely, knockdown of endogenous MEF2D levels also reduced luciferase activity driven by the ACSL3 promoter (Figure 4E). To further confirm the binding of MEF2D to the putative MRE in the upstream regions of the ACSL3 promoter, we conducted a ChIP assay in Huh7 cells. The results showed that the anti-MEF2D antibody co-precipitated the DNA fragment containing the predicted MRE in the regulatory regions of the ACSL3 gene, indicating that MEF2D bound directly to the MRE in Huh7 cells (Figure 4F). We found that ACSL3 was highly expressed in HCC tissues (Figure 1H). Further analysis of RNA sequencing data from the TCGA database showed that MEF2D expression was also significantly elevated in HCC tissues compared to adjacent normal tissues (Figure 4G). To investigate the potential relationship between ACSL3 and MEF2D, we conducted a correlation analysis, which revealed a significant positive association between their expression levels in HCC (Figure 4H).

These results provide strong evidence for the ability of MEF2D to regulate the transcription of the ACSL3 gene.