As stated in the Ishak scoring system [22], liver tissue samples were collected from four subjects. Patients with liver fibrosis at stages S2, S3, S3/S4, and S4 at the Second People’s Hospital of Guangdong Province were included. We strictly adhered to ethical guidelines during the acquisition process to ensure informed consent from the patients. Record clinical information, including patient age, gender, aetiology(including viral hepatitis type, alcohol consumption history, and metabolic syndrome-related indicators, etc.), disease course, liver function indicators (alanine aminotransferase (ALT), aspartate aminotransferase (AST), bilirubin, albumin, and globulin, etc.), imaging examination results (liver stiffness values, ultrasound findings, etc.), and treatment history. All experiments involving human tissue were conducted in accordance with the ethical policies and procedures approved by the Medical Ethics Committee of The Affiliated Guangdong Second Provincial General Hospital of Jinan University. The ethical approval number for this study is 2023-KY-SB-37.

24 SPF-grade C57BL/6 J mice, aged 6–8 weeks with a body mass of 19–23 g, were procured from Zhuhai Baishitong Biotechnology Co., Ltd., Guangdong Province, China. A Total of 24 mice were housed in a constant temperature environment (22–23 °C) and were subjected to a standard 12-hour light-dark cycle, with sufficient food and water provided. All animal experiments were conducted in accordance with the ethical guidelines and policies approved by the Laboratory Animal Ethics Committee of The Affiliated Guangdong Second Provincial General Hospital of Jinan University, under approval number 2023-DW-SB-48.

The mice were randomly assigned to four groups with six mice in each group (n = 6): control group (injected with adeno-associated virus 9 (AAV9)-GFP virus), CCl₄-treated model group (injected with AAV9-GFP virus), CCl₄-treated model group (injected with AAV9-LXN shRNA), and control group (injected with AAV9-LXN shRNA). The model group used a CCl₄-induced liver fibrosis model, mixing CCl₄ with olive oil at a ratio of 1:3 to 1:5. The mixture was administered via intraperitoneal injection based on mouse body weight (0.1 mL per 5 g of body weight), the control group received 5 µL/g of corn oil, three times per week for a period of eight weeks.

shRNA sequences targeting LXN were designed and synthesized based on the mouse LXN gene sequence. After annealing to form double strands, the strands were connected to the AAV9 vector. The constructed vector contained promoters, shRNA sequences, and poly(A) tails. This section was constructed to Cyagen Biosciences. The constructed AAV9-LXN shRNA vector and the auxiliary plasmid were co-transfected into HEK293 cells. Following cell culture and virus packaging processes, cell culture supernatants were collected and purified by ultrafiltration and ultracentrifugation to obtain the AAV9-LXN shRNA virus. After determining the virus titer, the virus was stored at –80 °C. Starting from week 3, AAV9-GFP virus and AAV9-LXN shRNA virus (at a dose of 1 $\times$ 10¹¹–1 $\times$ 10¹² vg/mouse) were injected into the corresponding groups of mice via the tail vein. Following the conclusion of the experimental study, mice were fasted overnight. We administered 50 mg/kg sodium pentobarbital intraperitoneally to anesthetize the mice, measured their body weight, and collected their blood and liver tissue, after which the mice were euthanized by cervical dislocation.

The blood collected from mice was centrifuged at 3000 rpm at room temperature for 15 minutes, the supernatant was collected and stored at –80 °C for subsequent experimental analysis. Serum ALT, AST, alkaline phosphatase (ALP), albumin (ALB), total bilirubin (TBIL), total bile acids (TBA), cholesterol (CHO), and triglycerides (TG) levels were measured using an automatic biochemical analyzer (Hitachi 7020, Japan).

After removing the liver tissue, a portion was treated with 4% formaldehyde for 36 hours, then embedded in paraffin and sectioned (4–6 µm thick). The morphology of hepatocytes and the presence of inflammatory cell infiltration were observed using HE staining, and the severity of liver fibrosis was assessed using Sirius red staining.

For immunohistochemical staining of Alpha-Smooth Muscle Actin ($\alpha$-SMA), collagen I (COL1), and Latexin (LXN), sodium citrate buffer was used to retrieve antigens, and paraffin-embedded sections were subjected to antibody incubation for a duration of 12 hours at 4 °C. Goat anti-rabbit horseradish peroxidase (HRP) secondary antibody (ZK0623, Bioss, China) was then added to each slide and incubated at room temperature for 50 min. The sections were then incubated with 3,3’-Diaminobenzidine (DAB) for two minutes, followed by counterstaining with hematoxylin. Finally, counterstain with blue reagent for ten seconds. The following antibodies were utilized for immunohistochemistry: anti-LXN antibody (1:300, Bioss, bs-1971R), anti-$\alpha$-SMA antibody (1:250, Bioss, bs-10196R), and anti-COL1 antibody (1:250, Bioss, bs-10423R).

An Animal Tissue Total RNA Extraction Kit (PK10021, Proteintech) was used to extract total RNA from the liver tissue. A spectrophotometer was used to detect the RNA concentration and purity.

The accepted range for the A260/A280 ratio is 1.8–2.0. The reverse transcription process was executed by means of a reverse transcription kit (B639252-0100, Sangon Biotech), and RT-qPCR was conducted using the TB Green® Premix Ex TaqTM II kit (RR420A, TAKARA) on an ABI7500 PCR instrument. We used $\beta$-actin as an internal reference gene. The 2^{-Δ⁢Δ⁢Ct} method was utilized for the statistical analysis. Table 1 presents a list of all the primer sequences.

LX-2 cells were procured from the Shanghai Institute of Cell Biology, Chinese Academy of Sciences. All cell lines used in this study were authenticated by short tandem repeat (STR) profiling and tested negative for mycoplasma contamination. After thawing from liquid nitrogen, the cells were resuspended in DMEM (Gibco, USA) supplemented with 10% fetal bovine serum (Gibco, USA) and 100 U/mL penicillin and streptomycin (Solarbio, Guangzhou, China). Cells were then maintained in a 5% CO₂ incubator for subsequent experiments. The cells were then cultured in six-well plates and grouped as follows: (1) Control group, cultured under standard conditions. (2) TGF-$\beta$ group: treated with TGF-$\beta$1 at 10 ng/mL for 12 hours. (3) TGF-$\beta$+ LXN-siRNA group: Following TGF-$\beta$1treatment, LXN-siRNA was added and transfected for 24 hours. (4) TGF-$\beta$+ siRNA-Ctrl group: Following TGF-$\beta$1 treatment, NC-siRNA was added for 24 hours of transfection. LXN-siRNA and a negative control siRNA were designed and synthesized by Shanghai Sangon Biotech Co., Ltd. The sequences are listed in Table 2.

For liver tissue paraffin sections, fixation with 10% neutral buffered formalin, boil in sodium citrate buffer for 15 minutes, and incubation at room temperature for one hour after blocking with 5% bovine serum albumin (BSA), followed by incubation with the appropriate primary antibody at 37 °C overnight. After thorough washing with PBS, sections were incubated with FITC-labelled goat anti-rabbit IgG at 20 °C for 1 h. Then, the sections were counterstained with 4,6-diamidino-2-phenylindole (DAPI), washed, and mounted using an anti-fluorescence quenching mounting medium (Prolong® Gold). The cells were observed under a fluorescent microscope (OLYMPUS IX83, Tokyo, Japan). The following antibodies were utilised for immunofluorescence staining: anti-LXN (1:3500; Bioss, bs-1971R) and anti-THBS2 (1:2000; CUSABIO, P35442).

Total protein was extracted from the liver tissue using a dedicated kit for animal tissues. Lysate proteins from LX-2 cells were prepared using 1:100 PMSF and RIPA buffer, followed by SDS-PAGE to separate 30 micrograms of protein and polyvinylidene difluoride (PVDF) membrane was used to transfer it. For 1 hour at 20 °C, the membrane was blocked with 5% bovine serum albumin (BSA), the primary antibody was incubated with the PVDF membrane at 4 °C overnight, then the PVDF membrane was thoroughly washed with TBST, and incubated with goat anti-rabbit secondary antibody (IgG) (ZK0623) at 20 °C for 70 minutes, followed by ECL chemiluminescence imaging to detect the bands. For the purpose of visual analysis of proteins, ImageJ (NIH, Bethesda, MD, USA) was employed. The primary antibodies utilized for western blotting are listed below: anti-LXN antibody (1:300, Bioss, bs-1971R), anti-$\alpha$-SMA antibody (1:250, Bioss, bs-10196R), and anti-COL1 antibody (1:250, Bioss, bs-10423R), anti-THBS2 antibody (1:1000, CUSABIO, P35442), and anti-Tubulin antibody (1:2000, Beyotime, AF0001).

The statistical analyses were conducted using the GraphPad Prism v8.00 software (La Jolla, CA, USA). The experimental data were then compared using Student’s t-test or analysis of variance. The statistical significance of the results was determined by applying a p-value cutoff of 0.05.

We obtained three sets of transcriptome data related to human and mouse liver fibrosis from the GEO database (GSE25583, GSE55747, and GSE84044), from which we screened for differentially expressed genes ($|$LogFC$|$ $\ge$1, p 0.005; $|$LogFC$|$ $\ge$2, p 0.00001; $|$LogFC$|$ $\ge$1, p 0.05). After obtaining the intersection of the three datasets, we identified a common differentially expressed gene LXN (Fig. 1A). Next, LXN expression levels were analyzed in different groups using GEO chips (GSE89377) derived from human cirrhotic tissue samples and GEO chips for mouse cirrhosis (GSE25583 and GSE55747). The results showed that LXN expression was higher in human cirrhosis than in hepatitis, and treatment of mice with CCl₄ to induce cirrhosis also led to increased LXN expression (Fig. 1B–D). Next, we performed differential gene analysis of 124 patients with different stages of liver fibrosis using gene expression profiling chips (GSE84044). Through RNA-seq analysis, we screened for differentially expressed genes (FDR 0.05) at different stages of liver fibrosis. Subsequently, a heat map was generated for the 40 genes with the highest expression levels (Fig. 1E). We performed correlation analysis on the top 20 genes that were upregulated and downregulated in patients with different stages of liver fibrosis, and LXN exhibited a significantly and positively correlated with THBS2 and multiple chemokines (CXCL6, CCL20, CXCL10, CXCL9, CXCL11, and CCL19) (Fig. 1F). THBS2 regulates ECM and collagen deposition, which is linked to liver fibrosis. Pathway and enrichment analyses showed that compared with the low LXN expression group, the high LXN expression group had differentially expressed genes linked to T lymphocyte, granulocyte, and monocyte activation, proliferation, and chemotaxis (Fig. 1G,H).

Fig. 1.

Latexin (LXN) is upregulated in liver fibrosis and highly correlated with Thrombospondin-2 (THBS2). (A) The Venn diagram illustrates the correlations among differentially expressed genes across three liver fibrosis datasets. (B) LXN expression in four distinct pathological states within the human liver tissue dataset. NLs, Normal liver tissue; LGCHs, Low-grade chronic hepatitis; HGCHs, High-grade chronic hepatitis; CSs, Cirrhosis. * p 0.05, ** p 0.01. (C) Expression of LXN in mouse liver fibrosis datasets following no carbon tetrachloride (CCl₄) treatment, and at different time points following CCl₄ treatment. * p 0.05, ** p 0.01. (D) Expression of LXN in healthy individuals and patients with liver cirrhosis. ** p 0.01. (E) The heatmap depicts the differential expression patterns of the top 40 most highly expressed genes across 124 patients at different stages of liver fibrosis. (F) The correlation heatmap illustrates the associations between the top 20 genes showing upregulation and downregulation. (G,H) Pathway and enrichment analysis of the high-expression LXN group.

After analyzing the expression profiles of 124 patients with different stages of liver fibrosis (GSE84044), it was established that there was a gradual increase in LXN expression concomitant with the progression of inflammatory activity grading (G0–G4) and fibrosis staging (S0–S4) (Fig. 2A,B). To validate this result, we collected liver tissue samples from four patients with liver fibrosis stages S2, S3, S3/S4, and S4, and performed HE staining, Sirius red staining, and LXN immunohistochemical staining. HE staining revealed that as the stage of liver fibrosis increased, the portal areas gradually expanded, forming fibrous septa, and the hepatic lobule structure became disordered with increased inflammatory cell infiltration. Sirius Red staining showed that collagen deposition gradually increased and fibrous septa formed as liver fibrosis progressed. The immunohistochemical staining results for LXN were consistent with our expectations, with LXN expression increasing as the stage of liver fibrosis progressed (Fig. 2C, Supplementary Fig. 1A,B).

Fig. 2.

Expression of LXN in human liver tissue increases with the progression of liver fibrosis. All experiments were performed in triplicate (n = 3). (A,B) In a dataset comprising 124 patients with distinct stages of hepatic fibrosis, the expression of LXN correlates with both the grading and staging of fibrosis. ** p 0.01. (C) Histological, Sirius red, and immunohistochemical staining findings in four patients with different stages of liver fibrosis. Scale is 400 µm.

This study used a CCl₄-induced liver fibrosis model in which mice received intraperitoneal injections of CCl₄ for eight weeks. To determine whether LXN knockout could alleviate CCl₄-induced liver fibrosis in mice, AAV9-LXN-shRNA or control (GFP) was injected into C57BL/6 mice starting at week 3 (Fig. 3A). After the experiment, the mice were euthanised and whole blood and liver tissue were collected. The body weight and ALT, AST, ALP, ALB, TBIL, TBA, CHO, and TG levels of the mice were measured (Fig. 3B–J). The results showed that murine subjects in the CCl₄-modelling group exhibited a reduced mean body mass compared with the control group, and indicators such as TBIL, TBA, ALT, AST and ALP were all elevated, indicating that CCl₄ treatment damaged the liver function of mice. Following LXN knockdown using AAV9-LXN-shRNA, the levels of TBIL, TBA, ALT, AST, and ALP were reduced compared to the model group. TBA is a sensitive marker of liver damage and can directly participate in the fibrotic process through mechanisms such as HSC activation and oxidative stress. This finding suggests that suppression of LXN may lead to a reduction in HSC activation and oxidative stress, consequently alleviating CCl₄-induced liver injury and mitigating liver fibrosis.

Fig. 3.

Weight and blood biochemical parameters in the four groups of mice. (A) Schematic diagram of the treatment protocol for the CCl₄-induced adeno-associated virus 9 (AAV9)-LXN-shRNA mouse model. (B–J) Body weight and Serum Alanine Aminotransferase (ALT), Aspartate Aminotransferase (AST), alkaline phosphatase (ALP), albumin (ALB), total bilirubin (TBIL), total bile acids (TBA), cholesterol (CHO), and triglycerides (TG) levels in the four mouse groups. * p 0.05, ** p 0.01, *** p 0.001, ns indicates no significant difference.

Subsequently, the present study investigated the potential of AAV9-LXN shRNA to reduce liver fibrosis and protect the liver from the harmful effects of CCl₄. Photographs of the isolated livers were taken, and HE and Sirius red staining were utilised in order to ascertain the histopathological status of the mouse liver tissue and collagen deposition. Immunohistochemical staining for $\alpha$-SMA, collagen I, and LXN was performed to examine HSC activation and LXN expression. As shown in Fig. 4A, a significant decrease in liver volume was observed in mice exposed to CCl₄ in compared to the control group. HE staining revealed a disordered hepatic lobule structure, dilated sinusoids, inflammatory cell infiltration, and interlobular septum formation. Sirius red staining showed a significant increase in collagen fiber deposition, whereas the liver tissue in the reversal group spontaneously recovered. These results demonstrated the successful establishment of a liver fibrosis model. Immunohistochemical staining results showed that CCl₄-induced HSCs were activated, with a substantial increase in the number of $\alpha$-SMA+ cells in the liver; collagen I also showed a similar trend. However, treatment with AAV9-LXN shRNA alleviated these changes. As HSCs are activated in the liver, LXN expression was significantly upregulated. We also performed quantitative analysis of Sirius Red staining and immunohistochemical staining, as shown in Supplementary Fig. 1C,D. We subsequently performed immunofluorescence staining for THBS2, and the results were consistent with the changes observed in LXN. THBS2 expression was upregulated in the CCl₄-induced liver fibrosis model group, and LXN interference led to its downregulation (Fig. 4B).

Fig. 4.

AAV9-LXN shRNA reverses CCl₄-induced liver fibrosis. All experiments were performed in triplicate (n = 3). (A) Histological examination of liver tissue from oil 8w, CCl₄ 8w-induced model, CCl₄ 8w AAV9 LXN-shRNA, and oil 8w AAV9 LXN-shRNA-treated mice via hematoxylin and eosin staining, Sirius red staining, and immunohistochemical staining for $\alpha$-SMA, collagen I (COL1), and LXN. Scale bars are as indicated. The scale of HE staining and Sirius red staining is 400 µm, The immunohistochemical staining scale of $\alpha$-SMA is 625 µm, Collagen I and LXN is 400 µm. (B) Immunofluorescence staining of THBS2 in liver tissue from four groups of mice, the area indicated by the arrow is the positive zone, the original image and enlarged graphic are at magnifications of 20$\times$ (20 µm) and 40$\times$ (10 µm) respectively. (C) Western blotting and quantitative analysis of LXN, THBS2, $\alpha$-SMA, and COL1 proteins in liver tissue from four groups of mice, using tubulin as the internal reference protein. (D) Quantitative analysis by Western blot, oil 8w, CCl₄ 8w-induced model, CCl₄ 8w AAV9 LXN-shRNA, and oil 8w AAV9 LXN-shRNA are denoted as 1/2/3/4 respectively. * p 0.05, ** p 0.01, *** p 0.001, ns indicates no significant difference.

Additionally, RT-qPCR and Western blot analyses further confirmed these results, showing that the CCl₄-treated groups exhibited significantly increased LXN expression and markedly elevated HSC activation indices, whereas LXN-silenced mice in the reversal group demonstrated reduced HSC activation indices. Furthermore, the expression of these genes was significantly reduced following LXN knockdown (Supplementary Fig. 2 and Fig. 4C,D). In summary, liver fibrosis upregulates LXN expression, and LXN knockdown can reverse and regress liver fibrosis.

The LX-2 cells were subjected to a treatment with TGF-$\beta$ to induce their activation, and LXN-siRNA was used to silence LXN in TGF-$\beta$-treated LX-2 cells. The duration of the treatment was 10 h. After treatment, Oil red O staining and immunofluorescence analyses of LXN and THBS2 were performed. The proteins were extracted for Western blot analysis. Unactivated LX-2 cells and empty siRNA were used as controls. Immunofluorescence staining showed that after TGF-$\beta$ induction, the fluorescence intensity of LXN increased, whereas that in the LXN-siRNA group was significantly reduced, indicating that the cell model was successfully established. The results of THBS2 immunofluorescence staining were consistent with those of LXN, with TGF-$\beta$ induction increasing the fluorescence intensity, whereas knockdown of LXN reduced the expression of THBS2 (Fig. 5A). Oil red staining results showed that LX-2 cells treated with TGF-$\beta$ had larger volumes than unactivated cells, with a flat, star-shaped, or polygonal morphology, dense cell arrangement, darker nuclear staining, and a marked reduction in lipid droplets. Treatment with LXN-siRNA reversed these effects (Fig. 5B). Western blot results confirmed that, compared with untreated cells, TGF-$\beta$ induction can lead to HSC activation and upregulation of collagen I and $\alpha$-SMA expression, while LXN silencing can weaken LX-2 cell activation and downregulate collagen I and $\alpha$-SMA expression. However, the empty siRNA vector group did not show this change. THBS2 expression showed a trend consistent with that of LXN (Fig. 5C,D).

Fig. 5.

LXN siRNA interference reduces the activation of hepatic stellate cells (HSCs). All experiments were performed in triplicate (n = 3). (A) Immunofluorescence staining for LXN and THBS2 in cells from the control group, transforming growth factor-$\beta$ (TGF-$\beta$)-induced activation group, TGF-$\beta$-siLXN knockdown group, and TGF-$\beta$-siRNA control group. Scale bars are 50 µm. (B) Oil red staining of the above four cell groups. Scale bars are 50 µm and 25 µm. (C,D) Western blot analysis and quantitative assessment of four cellular proteins—LXN, THBS2, COL1, and $\alpha$-SMA—were conducted, with Tubulin employed as the internal control protein. LXN-siRNA, siRNA-Ctrl, TGF-$\beta$ and Ctrl are denoted as 1/2/3/4 respectively (D). *** p 0.001, **** p 0.0001.

Single-cell sequencing data of liver tissues during hepatic fibrosis, retrieved from the GEO database (GSE174748), was further analyzed. Unsupervised clustering of the qualified single-cell dataset identified 10 distinct cell clusters (Supplementary Fig. 3A). Subsequently, a panel of classical HSC marker genes (DCN, HGF, ACTA2, and ADAMTSL2) was utilized to validate and annotate each cell cluster (Supplementary Fig. 3B). Via this marker gene-mediated identification strategy, the HSC cluster was successfully localized and defined within the heterogeneous cellular population of liver tissues (Supplementary Fig. 3C). Further differential expression analysis demonstrated that the LXN gene was specifically highly expressed in HSCs, with negligible expression detected in other cell clusters (Supplementary Fig. 3D).