CCl₄ (C128126), deferoxamine (DFO) (D105648), and erastin (E126853) used in this study were purchased from Aladdin (Shanghai, China).

The gene expression data were obtained from the Gene Expression Omnibus (GEO) dataset (Access ID: GSE25097)¹.

The study protocol, conformed to the ethical guidelines of the 1975 Declaration of Helsinki, was approved by the Institutional Ethical Review Committee of Shanghai Ninth People’s Hospital (Shanghai, China). Written informed consents were obtained from all the participants. A total of 20 liver specimens of mild fibrosis, 20 of severe fibrosis, and 20 of cirrhosis were collected.

Human HSCs line LX-2 was obtained from the cell bank of the Shanghai Biology Institute (Chinese Academy of Sciences, Shanghai, China). The cells were cultured in RPMI-1640 medium (Hyclone) supplemented with 10% fetal bovine serum (Invitrogen) and 1% streptomycin/penicillin (Invitrogen).

Total RNA was extracted using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. The mRNA levels of indicated genes were determined by qRT-PCR using SYBR^® Green (Thermo Fisher Scientific) on ABI 7300 instrument (Applied Biosystems), with GAPDH as an internal control. All the reactions were conducted based on the following cycling parameters, 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 45 s. Specific amplification was verified by dissociation curve analysis. Comparative C_t method was used for transcript quantification. The fold-changes of target genes, normalized by the internal control, were determined by the formula 2^–△△CT. All data represent the average of triplicates. The primers are listed in Supplementary Table S1.

Total cell lysates were prepared with radioimmunoprecipitation assay (RIPA) buffer containing proteinase inhibitor (Beyotime), following the manufacturer’s instructions. Proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and electroblotted onto nitrocellulose membranes (Millipore). After blocking with 5% skimmed milk, the membranes were incubated with primary antibodies (Supplementary Table S2) at 4°C overnight in accordance with the manufacturer’s instructions. After the unbound antibody was washed away, the membranes were further incubated with HRP-conjugated rabbit secondary antibody (Beyotime) at room temperature for 1 h. The enhanced chemiluminescence system (Millipore) was employed for signal detection.

Full-length TRIM26 and TRIM26 with substitute mutation of C31S were cloned into pCMV-Tag2 (Stratagene) to express Flag-tagged TRIM26 and TRIM26 E3 catalytic mutant (C31S) (Ran et al., 2016), respectively. The constructed plasmids were verified by double enzyme digestion and DNA sequencing.

Mouse TRIM26 was cloned into pShuttle-CMV (GenScript). The SalI/XhoI-linearized shuttle plasmid was recombined with backbone pAdEasy-1 (GenScript) in Escherichia coli BJ5183 to construct pAd-TRIM26. Then, pAd-TRIM26 and pAdEasy-1 were transfected into 293 cells using Lipofectamine^TM 2000 reagent (Invitrogen) according to the protocol provided by the manufacturer. Recombinant adenovirus was grown and purified based on cesium chloride gradients. Viral titers were determined by plaque assay, and the virus was aliquoted for storage at −80°C.

Short hairpin RNA (shRNA) oligos targeting TRIM26 (Supplementary Table S3) were annealed and cloned into AgeI/EcoRI-digested pLKO.1 (Addgene). Full-length human TRIM26 was cloned into pLVX-puro (Clontech). The lentivirus was produced in 293T cells along with packaging plasmids psPAX2 and pMD2.G.

Cell proliferation was determined using the Cell Counting Kit-8 (CCK-8) Assay Kit (Jiancheng Bioengineering Institute, Nanjing, China). The LX-2 cells were cultured and treated in 96-well plates, and 10 μL of the CCK-8 reagent was added to each well for 3 h incubation in darkness. The absorbance was measured at 450 nm using a Microplate Reader. The averaged optical density (OD) of each group was used to calculate the percent of relative cell proliferation by the following formula: (OD_treatment/OD_control) × 100%.

Hydroxyproline and GSH levels were measured using commercial kits (Jiancheng Bioengineering Institute) according to the manufacturer’s protocols.

Lipid peroxidation was assessed using the C11-BODIPY assay kit (D3861, Thermo Fisher Scientific) according to the manufacturer’s instructions. Briefly, the cells were treated as indicated. The cultural medium was replaced with 10 mM C11-BODIPY-containing medium for 1 h. Then, the cells were harvested by trypsinization and resuspended in 1% BSA-containing PBS. The lipid ROS level was examined by flow cytometry analysis (FACSCanto^TM II, BD Biosciences).

Total ROS level was assessed using the Reactive Oxygen Species Assay Kit (Beyotime, S0030) and analyzed by flow cytometry.

MDA (malondialdehyde) content, Fe²⁺ release level, and NADPH (reduced form of nicotinamide-adenine dinucleotide phosphate) was determined using the lipid peroxidation (MDA) assay kit (Abcam, #ab118970), iron assay kit (Abcam, #ab83366), and fluorometric NADP/NADPH (Abcam, #ab65349) assay kit, respectively, following standard instructions.

The cell lysates were reacted with anti-TRIM26 (27013-1-AP, Invitrogen), anti-SLC7A11 (PA1-16893, Invitrogen), or control IgG (Santa Cruz Biotechnology) for 1 h at 4°C. Then, they were incubated with protein A/G-agarose for 3 h at 4°C. The precipitates were washed three times with the lysis buffer and detected by western blot analysis.

Animal experiments were approved by the institutional and local animal care and use committees of the Shanghai Ninth People’s Hospital (Shanghai, China). A total of 24 C57BL/6 mice were obtained from the Sippr-BK laboratory animal Co., Ltd. (Shanghai, China). They were randomly divided into four groups: Group I, Vehicle (control); Group II, CCl₄; Group III, CCL₄ + Vector; Group IV, CCL₄ + oeTRIM26. Liver fibrosis was induced in Groups II − IV, by intraperitoneally injecting 50% carbon tetrachloride (CCl₄) in corn oil (0.1 mL/100 g body weight) over 8 weeks (3 times/week); control mice were injected with corn oil only. Then, recombinant adenovirus Vector or oeTRIM26 (5 × 10⁹ pfu/mouse, 0.5 mL) was injected into the mice of Group III or IV, respectively, through the tail vein. Four weeks after the injection, all the mice were sacrificed, and blood samples were obtained for subsequent measurements of aspartate transaminase (AST), alanine transaminase (ALT), and hydroxyproline. The livers were immediately excised; half were fixed with 4% paraformaldehyde for histological examination by hematoxylin and eosin (HE), Masson’s trichrome, and Sirius red staining; the other half were frozen in liquid nitrogen for further western blot analysis. All procedures were performed in accordance with the guidelines for animal care. The qualification of Masson’s trichrome and Sirius Red staining was performed with the ImageJ software² (Bethesda, MD, United States).

Statistical analysis was carried out using the Graphpad Prism software (version 6.0). Student’s t-test and analysis of variance (ANOVA) were performed to compare the data. P-values less than 0.05 were considered as statistically significant.

Bioinformatic analysis based on the GSE25097 dataset revealed that the mRNA level of TRIM26 in human cirrhotic liver tissues (n = 40) was significantly lower than that in normal liver samples (n = 6) (Figure 1A). Further, we collected liver specimens from fibrotic or cirrhotic liver patients in our hospital. Through qRT-PCR anaylsis, we observed that the mRNA level of TRIM26 were significantly different among patient groups; the highest TRIM26 expression was observed in those with mild liver fibrosis, followed by those with severe liver fibrosis and with liver cirrhosis (Figure 1B). Moreover, the protein expression of TRIM26 in CCl₄-induced fibrotic liver of mice (Figure 1C) was significantly lower than that of the control (Figure 1D). Combining these findings, we demonstrate that TRIM26 is downregulated in cirrhotic or fibrotic liver.

To investigate the effect of TRIM26 on HSCs, oeTRIM26 plasmid was transduced into LX-2 cells, which upregulated the protein expression of TRIM26 (Figure 2A and Supplementary Figure S1A). Western blot analysis was then performed to detected the expression of α-smooth muscle actin (α-SMA), a marker of HSCs activation (Zhao et al., 2020), and collagen I, a marker of fibrosis (Schuppan, 2015). The results revealed that TRIM26 overexpression led to the downregulation of α-SMA and collagen I at the protein level (Figure 2A). Through biochemical tests, we detected that the cellular levels of hydroxyproline, a marker of collagen production (Schuppan, 2015), in TRIM26-overexpressed LX-2 cells were substantially lower than those of the controls (Figure 2B). Furthermore, the overexpressed TRIM26 substantially inhibited LX-2 cell proliferation by approximately 15% (Figure 2C), while increased the lipid ROS level by nearly threefold (Figure 2D) and the total ROS level (Figure 2E) by more than threefold. Additionally, we also examined Fe²⁺ release, the cellular level of MDA, GSH, and NADPH to verify the occurrence of ferroptosis. We found that Fe²⁺ release and MDA content was elevated, while GSH and NAPDH was substantially downregulated in TRIM26-overexpressed LX-2 (Figure 2F).

We further treated the LX-2 cells with iron inhibitor DFO, to examine the function of TRIM26 on HSCs ferroptosis. Compared with the untreated LX-2 cells, we observed that the proliferation of LX-2 cells treated with 100 μM DFO was substantially accelerated (Figure 3A), while their lipid ROS levels (Figure 3B) were substantially lower and cellular levels of GSH were significantly increased (Figure 3C). Moreover, DFO treatment also increased the levels of hydroxyproline (Figure 3D), α-SMA, and collagen I (Figure 3E) in the LX-2 cells. However, TRIM26 overexpression could contract the inhibitory effect of DFO on HSCs ferroptosis. At one hand, TRIM26 overexpression substantially reversed the proliferation rate (Figure 3A) and lipid ROS level (Figure 3B) of the LX-2 cells treated with DFO; at the other hand, under DFO treatment, TRIM26-overexpressed LX-2 cells exhibited lower levels of hydroxyproline (Figure 3D), α-SMA, and collagen I (Figure 3E) in comparison with the controls. Such findings demonstrate the function of TRIM26 in facilitating HSCs ferroptosis.

To identify the TRIM26-interacting proteins, we performed mass spectrometry-based proteomics analysis on the co-immunoprecipitation (co-IP) sample, in which SLC7A11 ranked high among the candidate interacting proteins of TRIM26 (Supplementary Table S4 and Supplementary Figure S2). Further, co-IP experiments with the LX-2 cell lysates confirmed that TRIM26 and SLC7A11 could physically interact with each other (Figure 4A). When TRIM26 was overexpressed in LX-2 cells, the protein expression of SLC7A11 was downregulated, while its mRNA level remained unchanged (Figure 4B). Therefore, we speculated that TRIM26 could regulate SLC7A11 merely at the protein level. Through further investigation, we identified that TRIM26 overexpression in LX-2 cells could enhance the ubiquitination of SLC7A11, while no such effect was observed in the cells overexpressed with E3 catalytic mutant TRIM26 (C31S) (Figure 4C). These findings indicate that TRIM26 can facilitate SLC7A11 ubiquitination, which was dependent on the E3 ligase activity of TRIM26.

To further investigate the role of SLC7A11 in TRIM26-induced ferroptosis, oeSLC7A11 and oeTRIM26 plasmids were co-transfected into LX-2 cells. We identified that the protein expression of SLC7A11 was downregulated in TRIM26-overexpressed LX-2 cells, while it was reversed by SLC7A11 overexpression (Figure 5A and Supplementary Figure S2). In comparison with the LX-2 cells with TRIM26 overexpression alone, oeSLC7A11 and oeTRIM26 co-transfection significantly restored lipid ROS back to the normal level of the control (Figure 5B). Meanwhile, SLC7A11 overexpression also significantly counteracted the inhibitory effect of TRIM26 overexpression on ferroptosis indicator molecules in LX-2 cells; the cellular levels of GSH (Figure 5C) and hydroxyproline (Figure 5D) in the LX-2 cells with oeSLC7A11 and oeTRIM26 co-transfection were significantly higher than those with oeTRIM26 transfection only; similarly, the protein expressions of α-SMA and collagen I were upregulated (Figure 5E). Therefore, we speculate that the ferroptosis-promotive effect of TRIM26 is through regulating SLC7A11.

After ferroptosis-promotive effect of TRIM26 was revealed in vitro, we next designated to examine its protective effect on liver fibrosis in vivo. Through intraperitoneally injecting CCl₄ into mice, we successfully established a liver fibrosis animal model. Based on morphological examinations with H&E, Masson’s trichrome, and Sirius red staining, we observed that CCl₄ injection induced extensive fibrotic deposition in liver tissues of the mice. However, the mice with oeTRIM26 adenovirus administration exhibited mild liver fibrosis (Figure 6A). At the meantime, CCl₄ considerably increased the serum levels of ALT, AST, and hydroxyproline (Figures 6B,C); while in contrast, the overexpression of TRIM26 effectively counteracted CCl₄’s effect by reducing ALT, AST (Figure 6B), and hydroxyproline (Figure 6C). Following a similar trend, CCl₄ injection downregulated the protein level of TRIM26 (Figure 6D), while upregulated SLC7A11 (Figure 6D), α-SMA, and collagen I (Figure 6E); in contrast, TRIM26 overexpression restored the expressions of these indicator proteins, i.e., a higher protein level of TRIM26 (Figure 6D) and lower protein levels of SLC7A11 (Figure 6D), α-SMA, and collagen I (Figure 6E). Overall, these observations reveal that TRIM26 overexpression can mitigate liver fibrosis triggered by CCl₄ in mice.