For CCl₄-induced liver fibrosis, 8-week-old C57 BL/6J (WT) and 5-lipoxygenase knockout (5-LO^−/−) mice received an intraperitoneal (i.p.) injection of CCl₄ (1 ml/kg body weight) twice a week. For MCD diet-induced liver fibrosis, WT and 5-LO^−/− mice were fed a methionine-choline-supplied (MCS) or MCD diet (TROPHIC Animal Feed High-Tech, China) for 6 weeks. For therapeutic experiments, zileuton loaded in cRGDyK (Cyclo [Arg-Gly-Asp-_DTyr-Lys])-guided liposome (RGD-Lip; 10 mg/kg) or vehicle was given by tail vein injection once every 3 days during the last 4 weeks of CCl₄ or MCD diet treatment. All mice were housed at West China Hospital, Sichuan University in accordance with the guidelines of the animal care utilization committee of the institute. Food and water were freely available to mice, except otherwise stated.

Liposome (Lip) were prepared by the thin-film hydration method as described (Li et al., 2019; Zhang et al., 2020). Zileuton-loaded RGD-Lips (RGD-Lip/zileuton) were prepared by adding zileuton to the lipid organic solution before the solvent evaporation. The mean particle size and zeta potential of Lip were measured by dynamic light scattering with the Zetasizer Nano ZS90 instrument (Malvern, United Kingdom). The morphology of Lip was examined by transmission electron microscopy (H-600, Hitachi, Japan) with 2% phosphotungstic acid staining.

Serum ALT and AST levels were detected by using commercial kits (BioSino Bio-Technology and Science).

An amount of 50 mg liver tissue was dissolved in acid hydrolysate in glass tube and heated in boiling water bath for 20 min. Hydroxyproline was extracted according to the manufacturer's instructions and measured by using kits (Nanjing Jiancheng Bioengineering Institute).

The left lobe of the mouse liver was removed and immediately fixed in 10% neutral-buffered formalin, embedded in paraffin, and sectioned at 4 μm. Liver sections were stained with hematoxylin and eosin (H&E). For picrosirius red staining, liver sections were incubated with 0.1% sirius red in saturated picric acid for 60 min at room temperature. The Sirus Red positive area were detected by Image J. Fibrosis was assessed by picrosirius red staining according to the Ishak fibrosis criteria (Ishak et al., 1995).

HSC were isolated from livers of WT mice and 5-LO^−/− mice via in situ collagenase perfusion and underwent differential centrifugation on Optiprep (Sigma) density gradients, as described (Kwon et al., 2014). Isolated HSC were cultured in collagen-coated dishes with DMEM supplemented with 10% fetal bovine serum and antibiotics at 37°C. The purity of HSC was >95% as determined by their typical star-like shape and abundant lipid droplets.

Mice were injected with CCl₄ to cause liver fibrosis and treated with RGD-Lip/zileuton (10 mg/kg) for 4 h. Primary hepatocytes were isolated as described (Chen et al., 2019). The remaining cells were divided into three groups, fixed, perforated and stained with anti-α-SMA (HSC), anti-F4/80 (Kupffer cells) and anti-CD31 antibodies. HSC, Kupffer cells and LSECs were sorted by flow cytometry. Biliary epithelial cells were isolated as previously described (Li et al., 2019). Before extraction, cells were added to 30 μl ddH₂O and underwent repeated freezing and thawing 5 times. A 25-μl aliquot of samples was added to a 1.5-ml polypropylene tube followed by 200 μl methanol. The mixture was vortex-mixed for 5 min and centrifuged at 14,000 rpm for 10 min at room temperature. The top layer was transferred to a new 1.5 ml polypropylene tube and evaporated to vacuum dryness at 37°C. Samples were re-dissolved with 50 μl 80% methanol. The mixture was vortex-mixed for 30 s and centrifuged at 14,000 rpm for 5 min at 4 °C. The 40-μl supernatant was transferred to a 250 μl polypropylene autosampler vial and sealed with a Teflon crimp cap. Partially purified cell samples were analyzed by using LC-MS/MS.

Serum-free supernatant from primary mouse HSC were collected in an ice bath and extracted by solid-phase extraction. The metabolomics of arachidonic acid were detected and analyzed as described (Zhang et al., 2015).

Serum-free primary mouse HSC were collected in an ice bath and protected from light. Samples were centrifuged (600 × g, 5  min, 4°C), and with the resulting supernatant, LTB₄ and LTC₄ levels were determined by using ELISA kits (Cayman Chemical).

Primary HSC were fixed in 4% paraformaldehyde for 15 min, incubated in 0.2% Triton X-100 1×PBS for 15 min for permeabilization of cytomembrane. Antigens in paraffin sections were repaired by microwaving in 0.01 M citrate buffer (pH = 6.0) for 15 min. Both cells and tissue samples were incubated with antibodies in 4°C for 12 h. Immunoreactive compounds were incubated in room temperature for 1 h for conjugating with fluorescence-labeling secondary antibodies. All antibodies used in these experiments are in Supplementary Supplementary Table S1.

Western blot analysis was performed as described (Chen et al., 2019). The antibodies were listed in supplementary Supplementary Table S1. The expression of β-tubulin was a loading control. Immunoreactive bands were visualized on nitrocellulose membranes by using fluorescence-conjugated secondary antibodies (LI-COR, United States). The relative density was calculated by the ratio of the density of the protein of interest to β-tubulin.

Total RNA isolation and PT-PCR was performed as described (Chen et al., 2019). The primers for detected genes are in Supplementary Table S2.

Experiments were repeated at least 3 times with similar results. Quantitative results are expressed as the mean ± SEM. Statistical significance was determined by Student’s unpaired two-tailed t test or one-way ANOVA multiple comparison test as indicated in legends. p < 0.05 was considered statistically significant.

To explore the potential role of lipoxygenase pathway in arachidonic acid during HSC activation, we collected cell supernatants from quiescent HSC (q-HSC) and activated HSC (a-HSC, culture activated for 7 days) (Thi Thanh Hai et al., 2018). Among metabolites identified in the lipoxygenase pathway, target metabolomics revealed that LTB₄ level was selectively increased in a-HSC (Supplementary Figure S1). ELISA further confirmed the increased LTB₄ level in the supernatant of a-HSC (Figure 1A). Consistently, the level of LTC₄, another metabolite of 5-LO, was also increased (Figure 1B).

The high level of LTB₄ and LTC₄ released by a-HSC prompted us to explore their function in activating HSC. To exclude the influence of endogenous LTB₄ and LTC₄, we isolated primary HSC from 5-LO^−/− mice and added exogenous LTB₄ or LTC₄ to these cells for 4 days. Treatment with LTB₄ or LTC₄ significantly increased the expression of fibrotic genes including α-SMA, collagen 1a1 (Col1a1), Col3a1, Col5a1, tissue inhibitor of metalloproteinase 1 (Timp-1) and plasminogen activator inhibitor 1 (PAI-1) (Figures 1C,D). On immunofluorescence staining, LTB₄ and LTC₄ increased α-SMA accumulation in HSC as compared with controls (Figure 1E). Western blot analysis further confirmed that LTB₄ and LTC₄ elevated the protein levels of α-SMA and Col1a1 (Figures 1F,G). Therefore, LTB₄ or LTC₄ could promote HSC activation. We next explored the signaling pathway conveying the effect of LTB₄ or LTC₄. LTB₄ or LTC₄ seemed not to induce the phosphorylation of Smad2/3, a key pathway molecule for TGF-β-induced fibrosis (Supplementary Figure S2A). Instead, we found that LTB₄ and LTC₄ induced phosphorylation of the ERK1 signaling pathway (Figures 2A,B). The phosphorylation of ERK is necessary for LTB₄- and LTC₄-induced fibrosis because PD98059, a mitogen-activated protein kinase inhibitor, abolished their effects (Figures 2C,D; Supplementary Figure S2B). Other MAP kinases, according to p-p38 and p-JNK, were no change after LTB₄ or LTC₄ administration (Supplementary Figure S2C). Therefore, the effect of LTB4 and LTC4 on HSC may be mediated by ERK1 signaling.

5-LO is the key enzyme in the synthesis of LTB₄ and LTC₄ (Alexander et al., 2011). We further investigated whether the high level of LTB₄ and LTC₄ released by a-HSC resulted from the increased expression of 5-LO. The mRNA and protein levels of 5-LO were significantly upregulated in a-HSC as compared with q-HSC (Figures 3A–C). In contrast, the expression of Flap was not changed by HSC activation (Figure 3B). In CCl₄-induced liver fibrosis, 5-LO was also significantly increased along with other fibrotic genes (Figures 3D–F). Again, Flap expression was not significantly changed (Figure 3E). Liver fibrosis occurs during the progression of non-alcoholic steatohepatitis (NASH) (Nasr et al., 2018). Both the mRNA and protein levels of 5-LO were increased in an independent model of MCD diet-induced NASH (Anstee and Goldin, 2006) (Supplementary Figures S3A–C). These results suggest that increased 5-LO level may be responsible for the high level of LTB₄ and LTC₄ in a-HSC.

To further investigate the effect of 5-LO in HSC activation, we isolated primary HSC from wild-type and 5-LO^−/− mice, then compared the expression of fibrotic genes after culture activation. Ablation of 5-LO significantly inhibited the expression of culture-induced fibrotic genes, such as α-SMA, Col1a1, Col3a1, Col5a1, Timp1 and PAI-1 (Figures 4A,B), indicating that 5-LO was involved in HSC activation. Inhibition of HSC activation was confirmed by immunofluorescent staining, which showed the expression of α-SMA less detectable in 5-LO^−/− HSC (Figure 4C). Western blot analysis revealed that ablation of 5-LO suppressed α-SMA and Col1a1 expression (Figures 4D,E). The incubation of supernatant from WT a-HSC (WT-CM) was more effective to activate HSC than that from 5-LO^−/− a-HSC (5-LO^−/−-CM) (Supplementary Figures S4A,B).

To explore the in vivo function of 5-LO in liver fibrosis, we first exposed WT and 5-LO^−/− mice to CCl₄ twice a week for 8 weeks by intraperitoneal injection. As expected, CCl₄ caused significant hepatic fibrosis as compared with olive oil, as assessed by picrosirius red staining (Figures 5A,B). In contrast, 5-LO^−/− mice showed improved liver fibrosis and decreased hepatic fibrosis scores (Figure 5C). Consistently, the level of hydroxyproline was significantly lower in 5-LO^−/− than WT mice (Figure 5D), which suggests that 5-LO deletion conferred resistance to CCl₄-induced hepatic fibrosis. Among the fibrotic markers, α-SMA, collagens, TGF-β1, Timp-1/2 and PAI-1 were greatly suppressed in 5-LO^−/− mice after CCl₄ treatment (Figures 5E,F). The protein levels of α-SMA and Col1a1 were also greatly reduced in 5-LO^−/− mice with chronic CCl₄ injection (Figure 5G). CCl₄ administration showed an increasing serum levels of ALT and AST (Supplementary Figure S5A). However, this liver injury was significantly decreased in 5-LO^−/− mice. Chronic liver injury accelerated the accumulation of inflammatory cells around the vessels (Supplementary Figure S5B). Ablation of 5-LO greatly suppressed CCl₄-induced inflammatory cell infiltration (Supplementary Figures S5B,C). These results were further supported by the decreased expression of inflammatory genes seen in the liver of 5-LO^−/− mice (Supplementary Figure S5D).

In another independent model of MCD diet-induced NASH, collagen deposition was reduced in livers of 5-LO^−/− mice along with decreased fibrosis scores and hydroxyproline levels (Supplementary Figures S6A–D). Improved liver fibrosis was further confirmed by gene expression and protein analysis. Indeed, 5-LO ablation decreased the mRNA levels of fibrotic genes (α-SMA, Col1a1, Col3a1, Col5a1, TGF-β1, Timp-1/2 and PAI-1) (Supplementary Figures S6E,F) and the protein levels of α-SMA and Col1a1 (Supplementary Figures S6G,H). 5-LO ablation was protective in MCD-diet induced liver injury, as indicated by reduced serum levels of ALT and AST and inflammation scores (Supplementary Figures S7A–C). Inflammation plays an important role in MCD diet-induced NASH (Locatelli et al., 2014). Consistently, the mRNA levels of inflammatory genes including tumor necrosis factor α (TNF-α), interleukin 1β (IL-1β) and monocyte chemoattractant protein 1 (Mcp-1) were decreased in 5-LO^/− mice (Supplementary Figure S7D). The hepatic expression of CD68, a marker of macrophages, was also reduced (Supplementary Figure S7D). Therefore, 5-LO ablation ameliorated CCl₄- and MCD diet-induced hepatic fibrosis, inflammation and liver injury.

The protective effect of 5-LO ablation prompted us to explore whether pharmacological inhibition of 5-LO would have a similar effect in restraining the activation of primary HSC. We used an HSC-specific drug delivery system by modifying sterically stable liposome (Lip) with cRGDyK, a pentapeptide that binds to integrin αvβ3 on the surface of a-HSC (Li et al., 2019). cRGDyK-guided Lip showed high selectivity toward activated but not quiescent HSC, and preferentially accumulated in the fibrotic liver (Li et al., 2019; Zhang et al., 2020). We loaded with zileuton, an inhibitor of 5-LO, into this delivery system (RGD-Lip/zileuton). The schematic illustration, particle size, morphology and entrapment efficiency were comparable to regular liposome (Supplementary Figures S8A,B, Supplementary Table S3). As expected, RGD-Lip/zileuton significantly suppressed the secretion of LTB₄ by a-HSC (Supplementary Figure S9), which indicates the inhibition of 5-LO. In a-HSC, zileuton delivered by RGD-Lip greatly inhibited the expression of fibrotic genes including α-SMA, Col1a1, Col3a1, Col5a1, Timp-1 and PAI-1 (Figures 6A,B). These effects were further confirmed by immunofluorescent staining and western blot analysis (Figures 6C–E). Therefore, pharmacological inhibition of 5-LO suppressed HSC activation.

We then evaluated the in vivo therapeutic effect of RGD-Lip/zileuton in CCl₄-and MCD diet-induced liver fibrosis. In this experiment, mice were injected with CCl₄ for 4 weeks, then treated with RGD-Lip/zileuton (10 mg/kg) every 3 days (Supplementary Figure S10A). RGD-Lip could specifically deliver zileuton to HSC because the concentration of zileuton in HSC was about 28.99-, 4.71-, 4.67- and 34.01-times higher than that in hepatocytes, Kupffer cells, endothelial cells and biliary epithelial cells, respectively (Supplementary Figure S10B). Specific inactivation of 5-LO by RGD-Lip/zileuton in HSC greatly decreased liver fibrosis as shown by picrosirius red staining and liver hydroxyproline quantification (Figures 7A–D). The expression of fibrotic genes such as α-SMA, Col1a1, Col3a1, Col5a1, Timp1 and PAI-1 was also mitigated (Figures 6E,F). Western blot analysis confirmed the reduced protein levels of Col1a1 and α-SMA (Figures 6G,H). Targeted inhibition of 5-LO in HSC also alleviated CCl₄-induced liver injury and hepatic inflammation (Supplementary Figures S11A–C), but did not reduce the accumulation of F4/80 positive cells (Supplementary Figure S11D).

The therapeutic effects of RGD-Lip/zileuton were further demonstrated in an MCD diet-induced NASH model (Supplementary Figure S12). Specific inactivation of 5-LO in HSC explicitly alleviated MCD diet-induced liver fibrosis (Supplementary Figures S13A–D). Expression of fibrotic genes was suppressed in RGD-Lip/zileuton-treated mice (Supplementary Figures S13E,F) and α-SMA and collagen accumulation was reduced (Supplementary Figures S13G,H). Targeted delivery of zileuton also relieved liver injury in MCD-induced NASH (Supplementary Figure S14A). H&E staining and the expression of proinflammatory genes such as TNF-α and Mcp-1 indicated decreased inflammation in livers of RGD-Lip/zileuton-treated mice (Supplementary Figures S14B–D).

We next determined whether 5-LO expression was changed in patients with NASH or liver fibrosis. The diagnosis of NASH and fibrosis was confirmed by H&E and picrosirius red staining (Supplementary Figures S15A–C). As compared with healthy individuals, the expression of 5-LO was increased in liver sections from patients with NASH and fibrosis (Figure 8). The expression of 5-LO was largely co-localized in α-SMA-positive cells, which were HSC (Figure 8). These results were consistent with 5-LO possibly having a positive role in activation of HSC in rodents.