TSC22D1 promotes liver sinusoidal endothelial cell dysfunction and induces macrophage M1 polarization in non-alcoholic fatty liver disease

Abstract

BACKGROUND

The progression of non-alcoholic fatty liver disease (NAFLD) to non-alcoholic steatohepatitis (NASH) and liver fibrosis remains poorly understood, though liver sinusoidal endothelial cells (LSECs) are thought to play a central role in disease pathogenesis.

AIM

To investigate the role of TSC22D1 in NAFLD fibrosis through its regulation of LSEC dysfunction and macrophage polarization.

METHODS

We analysed single-cell transcriptomic data (GSE129516) from NASH and normal mouse models and identified TSC22D1 as a key regulator in LSECs. In vitro and in vivo experiments were conducted to validate the functional role of TSC22D1. Human LSECs were cultured and transfected to overexpress TSC22D1, and evaluated using flow cytometry, enzyme-linked immunosorbent assay, and quantitative polymerase chain reaction. NAFLD mice were used to assess TSC22D1 expression and its effects on LSEC dysfunction, endothelial-mesenchymal transition (EndMT), and microvascularization.

RESULTS

Single-cell analysis revealed that TSC22D1 mediates intercellular communication between LSECs and macrophages via the tumor necrosis factor-like weak inducer of apoptosis (TWEAK)/fibroblast growth factor-inducible 14 (FN14) signalling pathway, promoting M1 macrophage polarization and exacerbating liver fibrosis. In vitro studies revealed that TSC22D1 overexpression in LSECs exacerbated endothelial dysfunction and M1 polarization, whereas TWEAK inhibition attenuated these effects. Mechanistically, TSC22D1 drives LSEC microvascularization and EndMT through the TWEAK/FN14 pathway, leading to increased secretion of pro-inflammatory cytokines and M1 macrophage polarization. In vivo, experiments demonstrated that TSC22D1 inhibition via adeno-associated virus serotype 8-short hairpin RNA reduced NAFLD progression and liver fibrosis.

CONCLUSION

Our findings indicate a pivotal role of TSC22D1 in NAFLD fibrosis, demonstrating its dual function in regulating LSEC dysfunction and inflammatory responses. TSC22D1 may be a promising target for the treatment and the prevention and management of NAFLD progression to fibrosis.

Key Words: Liver sinusoidal endothelial cells; TSC22D1; Non-alcoholic fatty liver disease; Macrophage polarization; Tumor necrosis factor-like weak inducer of apoptosis

Core Tip: This study reveals that TSC22D1 plays a pivotal role in the progression of non-alcoholic fatty liver disease (NAFLD) fibrosis by promoting liver sinusoidal endothelial cell (LSEC) dysfunction and macrophage M1 polarization via the tumor necrosis factor-like weak inducer of apoptosis/fibroblast growth factor-inducible 14 signaling pathway. Through single-cell transcriptomic analysis and experimental validation, the authors demonstrate that TSC22D1 drives endothelial-mesenchymal transition and microvascularization in LSECs, while also enhancing pro-inflammatory responses in macrophages. Targeting TSC22D1 with adeno-associated virus serotype 8-short hairpin RNA alleviates liver fibrosis and inflammation in vivo, suggesting TSC22D1 as a promising therapeutic target for NAFLD. These findings provide novel insights into the mechanisms linking endothelial dysfunction, inflammation, and fibrosis in NAFLD.

INTRODUCTION

Non-alcoholic fatty liver disease (NAFLD) is a global health issue associated with increasing rates of obesity and metabolic syndrome. NAFLD encompasses a spectrum of conditions, ranging from simple steatosis to more severe manifestations such as non-alcoholic steatohepatitis (NASH), fibrosis, cirrhosis, and hepatocellular carcinoma. Liver fibrosis represents a critical stage in NAFLD progression because of its strong association with impaired liver function, progression to end-stage liver disease, and increased disease-related mortality[1]. The pathogenesis of NAFLD is multifactorial and involves complex interactions between genetic predispositions, insulin resistance, dietary factors, and chronic inflammation[2].

Liver sinusoidal endothelial cells (LSECs), which are highly specialized endothelial cells lining the hepatic sinusoids, critically contribute to both the pathogenesis and progression of NAFLD[3,4]. In NAFLD, LSECs undergo structural alterations such as reduced fenestrations, which impair hepatic microcirculation and hinder the exchange of lipids and other substances, thereby promoting lipid accumulation in hepatocytes[5]. Furthermore, dysfunctional LSECs exacerbate hepatic inflammation and fibrogenesis by releasing pro-inflammatory cytokines and fibrogenic mediators, such as transforming growth factor-β (TGF-β). These factors activate hepatic stellate cells (HSCs), resulting in the pathological accumulation of extracellular matrix components[6]. LSECs are also highly susceptible to oxidative stress, further aggravating hepatic injury[7,8]. Importantly, LSECs influence macrophage polarization by producing chemotactic and immunomodulatory factors, thereby promoting the recruitment and activation of M1-type pro-inflammatory macrophages, which amplify liver inflammation and fibrogenesis[9,10]. Collectively, these findings underscore the pivotal role of LSECs in modulating both inflammatory and fibrotic cascades in NAFLD/NASH pathogenesis, suggesting that therapeutic interventions targeting LSEC homeostasis could represent a viable treatment approach.

Single-cell RNA sequencing (scRNA-seq) has emerged as a powerful technology for decoding cellular heterogeneity and revealing the molecular pathogenesis of NAFLD. By providing high-resolution gene expression profiling at the single-cell level, this technology enables the precise identification of pathogenic cell subpopulations and dysregulated signalling pathways associated with NAFLD progression[11-13]. In the present study, we used scRNA-seq analysis to identify TSC22D1 as a key gene influencing LSECs in NAFLD. Subsequent experimental validation revealed that TSC22D1 mediates LSEC dysfunction through the tumor necrosis factor-like weak inducer of apoptosis (TWEAK)/fibroblast growth factor-inducible 14 (FN14) signalling pathway, leading to increased M1 macrophage polarization and aggravated liver fibrosis. These findings suggest that TSC22D1 in LSECs contributes to NAFLD progression and may represent a potential therapeutic target.

MATERIALS AND METHODS

Acquisition and preprocessing of single-cell transcriptome data

Single-cell transcriptome data were obtained from the GEO database (GSE129516), comprising three NASH mouse samples and three normal mouse samples. We used the CreateSeuratObject function in the Seurat package (version 4.3.0 and R version 4.2.0) to read and create Seurat Object variables[14]. We implemented stringent quality control criteria to exclude low-quality cells from our analysis[15]. Genes expressed in fewer than five cells were removed, and cells expressing fewer than 300 genes were excluded. We used the PercentageFeatureSet function to calculate the percentage of mitochondrial and erythrocyte genome expression in individual cells. Cells whose mitochondrial genome expression exceeding 20% and erythrocyte genome expression exceeded 5% were excluded. Single-cell expression data were normalized via the Seurat pipeline using the normalize data function (scale factor = 10000) to account for library size differences. Highly variable genes were then identified using Seurat’s find variable features function, and the top 3000 genes exhibiting the most significant dispersion (variance-to-mean ratio) were selected. Integration anchors were identified using the find integration anchors function with anchor features set to 2000 and datasets were merged using the integrate data function. Gene expression values were scaled and centred using Seurat’s scale data function to normalize variance across features. Dimensionality reduction was performed via principal component analysis (PCA) using the RunPCA function, with the first 30 principal components retained for downstream analyses based on the elbow method. Cell clustering was performed through a two-step process: (1) Constructing a shared nearest neighbour graph using find neighbors (k = 20); and (2) Applying the Louvain algorithm via FindClusters (resolution = 1). Dimensionality reduction and visualization were performed using the RunUMAP and RunTSNE functions to generate uniform manifold approximation and projection and t-distributed stochastic neighbour embedding plots. Cell annotation was conducted using the SingleR package, with the mouse RNAseq data reference dataset for matching.

Identification of differentially expressed genes and functional enrichment

Cell type-specific differentially expressed genes were identified using Seurat’s find markers function and applying stringent thresholds (adjusted P value < 0.05 and absolute log₂-fold change > 0.5). Functional enrichment analysis was conducted using cluster Profiler (v4.4.4) to identify significantly overrepresented Gene Ontology (GO) biological processes and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways (P value < 0.05)[16].

Cell-cell communication analysis

Intercellular communication networks were reconstructed using CellChat (v1.6.1)[17], a computational framework that systematically infers and analyses cell-cell signalling interactions. A CellChat object was initialized with create CellChat, incorporating the species-specific ligand-receptor interaction database (CellChatDB.mouse database) to ensure biological relevance. The following functions were used to preprocess the expression data for cell communication analysis in the two groups of mouse samples: Identify over expressed genes, identify over expressed interactions, computeCommunProb, computeCommunProb pathway, aggregate Net, and net analysis compute centrality. To compare and visualize signalling pathways between the two groups, we used the compare interactions, netVisual diff-interaction, netVisual heatmap, netVisual circle, netAnalysis signalingRole heatmap, and netVisual aggregate functions.

Animals

Male C57BL/6 mice, aged 6-8 weeks, were obtained from Aniphe Biolaboratory Company (Nanjing, Jiangsu Province, China) and housed in a controlled environment (22 °C, 12-hour light/dark cycle) with free access to water and standard chow. The bedding materials were sterilized corncob and replaced twice weekly. To induce NAFLD, the mice were fed a high-fat diet (HFD) comprising 60% kcal from fat, 20% kcal from carbohydrate, and 20% kcal from protein (D12492, Research Diets, Shanghai, China) for either 12 (HFD 12W group) or 20 weeks (HFD 20W group). A control group (HFD 0W group) was maintained on a standard chow diet containing 10% kcal from fat, 70% kcal from carbohydrates, and 20% kcal from protein.

The recombinant adeno-associated virus serotype 8 (AAV8) carrying TSC22D1-targeting short hairpin RNA (shRNA) (AAV8-shRNA-TSC22D1) and its corresponding negative control (NC) vector (AAV8-shRNA-NC) were commercially obtained from GeneChem Biotechnology Co., Ltd (Shanghai, China). The mice were randomly assigned to either the AAV8-shRNA-TSC22D1 (sh-TSC22D1) group (n = 6) or the AAV8-shRNA-NC (sh-NC) group (n = 6) and were fed a HFD for 20 weeks. The sh-TSC22D1 group received tail vein injections of AAV8 virus carrying shRNA-TSC22D1 (1 × 10¹¹ GC/mouse) in a total volume of 100 μL of sterile phosphate-buffered saline (PBS) at both 12 and 16 weeks. The sh-NC group received an equal volume of sterile PBS containing the AAV8 virus carrying shRNA-NC. Following the 20-week experimental period, the mice were humanely euthanized according to approved protocols, after which the liver, spleen, and brain tissues were immediately harvested, snap-frozen in liquid nitrogen, and stored at -80 °C for subsequent molecular analyses.

Cell culture and transfection

Human LSECs (HLSECs) (Fenghui, Hunan Province, China) were cultured in specialized medium at 37 °C with 5% carbon dioxide. In the experimental group, 100 ng/mL oxidized low-density lipoprotein (oxLDL) was administered. Following a 24-hour treatment period, samples were collected for subsequent analyses, including western blot (WB), immunofluorescence, and tube formation assays.

To overexpress TSC22D1 in HLSECs, human TSC22D1 overexpression plasmids (Genecefe Biotechnology, Jiangsu Province, China) were synthesized and transfected into HLSECs using a transfection reagent (TransGen Biotech, Beijing, China). As a control, cells were transfected with empty vector plasmids. WB was used to confirm TSC22D1 overexpression. After a 24-hour transfection period, HLSECs were treated with either 100 ng/mL oxLDL for 24 hours or 100 ng/mL oxLDL combined with 10 μM TWEAK inhibitor (L524-0366, TargetMol Chemicals, Shanghai, China) for 24 hours. All the cell supernatants (approximately 10 mL each for the control group, TSC22D1 group, and TSC22D1 + L524-0366 group) were collected. A portion of the supernatant (1 mL) was used for enzyme-linked immunosorbent assay (ELISA). HLSECs were harvested for WB, immunofluorescence, and tube formation assays.

THP-1 cells (Fenghui, Hunan Province, China) were treated with 150 ng/mL phorbol 12-myristate 13-acetate (MCE, Shanghai, China) for 24 hours to induce differentiation into M0 macrophages. The cells were treated with the supernatants from the three groups (the control group, TSC22D1 group, and TSC22D1 + L524-0366 group) described above for 48 hours. Following this treatment, cells were collected for quantitative polymerase chain reaction and flow cytometry analysis. The supernatants were also collected for ELISA.

Histological and immunostaining analyses

For histopathological examination, liver tissue sections were stained with hematoxylin and eosin (HE). Oil red O was used to stain fat deposits, while Sirius red was used to stain collagen. Immunohistochemistry and immunofluorescence are used to detect and localize the expression levels of specific antigens in tissues or cells. The antibodies used are listed in Supplementary Table 1. Images were analysed using Fiji (ImageJ2, version 2.14.0/1.54f). The colocalization finder plugin was used for positional analysis. Representative images from at least three independent animal or cell experiments were analysed for each staining method.

Liver tissue sections (5 μm thick) were processed for standard HE staining to evaluate histopathological changes. Lipid accumulation was assessed by Oil red O staining, and collagen deposition was visualized using picrosirius red staining according to standard protocols. Antigen localization and expression levels were determined by immunohistochemistry and immunofluorescence staining using established protocols. Primary and secondary antibodies (detailed in Supplementary Table 1) were optimized for appropriate dilutions and incubation conditions. Digital image analysis was performed using Fiji/ImageJ2 (v2.14.0) with standardized threshold settings across samples. Colocalization analysis was conducted using the colocalization finder plugin (v1.8) with Pearson’s correlation coefficient calculation. For each experimental condition, representative images from ≥ 3 biologically independent replicates were quantitatively analysed.

Scanning electron microscopy

Live tissue and HLSECs were fixed in electron microscopy fixation fluid (Servicebio, Wuhan, Hubei Province, China) for 2 hours. After additional fixation, dehydration, and drying, the samples were observed using scanning electron microscopy.

Tube formation assay

A total of 100 μL of matrix gel and 100 μL of medium were mixed and added to a 48-well plate. Subsequently, 300 μL of HLSECs were seeded into each well and incubated under standard conditions for 6 hours. Endothelial tube formation was monitored and digitally captured at 4-6 hours intervals using an inverted phase-contrast microscope (MS2000, Mshot, China) equipped with a CCD camera. Quantitative analysis was performed using Fiji/ImageJ2 (v2.14.0) with the angiogenesis analyzer plugin, and the total tube length, branch points, and mesh area were measured. All experiments were conducted in technical triplicates and repeated in three biologically independent experiments, with investigators blinded to the experimental conditions during image acquisition and analysis.

Quantitative real-time polymerase chain reaction

Total RNA was extracted from LSECs using TRIzol Reagent (Yeasen, Shanghai, China). Quantitative real-time polymerase chain reaction (qRT-PCR) was performed with a SYBR Green PCR Kit (Yeasen, Shanghai, China). The expression levels were normalized to those of the glyceraldehyde-3-phosphate dehydrogenase gene (Gapdh). The relative expression was calculated as 2^-ΔΔCt. The primers used are listed in Supplementary Table 2. Each qRT-PCR experiment was conducted in triplicate, and the results were derived from three independent biological replicates.

WB

Total protein was extracted using 1 mL of radio immunoprecipitation assay buffer containing 10 μL phenylmethanesulfonyl fluoride. Polyacrylamide gel electrophoresis gels were prepared according to the manufacturer’s instructions (Yeasen, Shanghai, China). After electrophoresis and transfer, the membrane was placed in tris buffered saline tween (TBST) containing 5% non-fat dry milk and blocked at room temperature for 1 hour. The primary and secondary antibodies (goat anti-mouse and goat anti-rabbit antibodies) were both diluted at 1:20000 in TBST (Yeasen, Shanghai, China) and incubated at room temperature for 1 hour each. The antibodies used are listed in Supplementary Table 1. WB experiments were independently repeated three times using samples from separate biological replicates.

Flow cytometry analysis

Macrophages were harvested and resuspended in PBS. The cells were then incubated with flow cytometry antibodies diluted 1:100 in PBS at 4 °C for 30 minutes. Following incubation, flow cytometry analysis was conducted using a NovoCyte Penteon flow cytometer (Agilent, China). Details of the antibodies used can be found in Supplementary Table1. Flow cytometry was performed on three independent biological replicates per group.

Statistical analysis

All the statistical analyses were performed using GraphPad Prism software (version 10.1.0, GraphPad Software, San Diego, CA, United States). Continuous data are presented as the mean ± SEM unless otherwise specified. Between-group comparisons were analysed using two-tailed unpaired Student’s t-tests. Multiple group comparisons were evaluated by one-way analysis of variance with either Bonferroni correction or least significant difference post-hoc test, following the verification of normality and homogeneity of variance. Statistical significance thresholds were defined as ^aP < 0.05, ^bP < 0.01, and ^cP < 0.001, with nonsignificant results indicated for P ≥ 0.05. All data visualizations were generated using GraphPad Prism.

RESULTS

TSC22D1 affects the intercellular communication of LSECs in NAFLD via TWEAK/FN14

To investigate intercellular communication in LSECs and key genetic changes during NAFLD fibrosis, we analysed the single-cell data described above. First, we used uniform manifold approximation and projection and t-distributed stochastic neighbour embedding methods to determine the distributions of different cell types in liver tissue (Figure 1A and B). The marker gene expression in each cell type is shown in Figure 1C. The proportions of different cell types in the six samples are shown in Figure 1D. We used CellChat to analyse differences in signalling pathways between cells in the liver tissues of the two groups of mice. Interestingly, according to our analysis, the signalling pathways of fibroblasts had more interactions than those of other cells did, while the signalling pathways of endothelial cells had greater interaction strengths, as shown in Supplementary Figure 1A. In NAFLD tissue, the midkine, growth differentiation factor, semaphorin neuropilin, and non-canonical wingless-related integration site signalling pathways predominated, whereas the kinase insert domain receptor, granulin, and angiopoietins pathways were less active (Supplementary Figure 1B). We then illustrated the intensity of various signalling pathways across different cell types in Supplementary Figure 1C. Furthermore, we performed differential expression gene (DEG) analysis among various cell types (Figure 2A). We extracted DEGs from endothelial cells for enrichment analysis. The KEGG and GO analysis results, shown in Figure 2B and C, respectively, revealed significant enrichment in migration and angiogenesis. We visualized these DEGs in an Upset plot, identifying ABCA1, INHBB, DLC1, LYVE1, TSC22D1, CSRP2, GM12840, ESM1, CCDC85B, and CTGF as specific to endothelial cells (Supplementary Figure 2A). Previous studies have shown that TSC22D1-overexpressing mice develop obesity and fatty liver, prompting us to focus on TSC22D1 for further investigation[18]. We used single-cell data from NAFLD mice to determine TSC22D1 expression levels across various cell types (Supplementary Figure 2B) and categorized endothelial cells into high and low expression groups based on the basis of the 1^st/4^th quartile cutoff (Supplementary Figure 2C). We analysed signalling pathway alterations in LSECs in different groups and other cell types between groups, as shown in Figure 2D. We discovered that hepatocytes, macrophages, endothelial cells with high TSC22D1 expression, and fibroblasts interact via the TWEAK/FN14 signalling pathway (Figure 2E). These findings indicate that TSC22D1 may promote NAFLD fibrosis through its action on HSECs.

Figure 1 Single-cell atlas of liver tissue from two groups of mice. A: T-distributed stochastic neighbor embedding plot of single cells from liver tissue of two groups of mice; B: Uniform manifold approximation plot of single cells from liver tissue of two groups of mice; C: Heatmap of characteristic genes of various cell types in liver tissue; D: Cell proportion plot in each sample. t-SNE: T-distributed stochastic neighbor embedding; UMAP: Uniform manifold approximation; NK: Nature killer.

Figure 2 Cell communication analysis in nonalcoholic fatty liver disease. A: Differential gene expression analysis showing up- and down-regulated genes across all eight clusters. An adjusted P value < 0.01 is indicated in red, while an adjusted P value ≥ 0.01 is indicated in black; B: Kyoto Encyclopedia of Genes and Genomes enrichment analysis of genetic variations in hepatic sinusoidal endothelial cells; C: Gene Ontology enrichment analysis of genetic variations in hepatic sinusoidal endothelial cells; D: Analysis of TSC22D1 expression levels in hepatic sinusoidal endothelial cells and the strength of communication with other cell types; E: Different cells interact via the tumor necrosis factor-like weak inducer of apoptosis fibroblast growth factor-inducible 14 signaling pathway. NK: Nature killer; FC: Fold change; COVID-19: Corona virus disease 2019; MAPK: Mitogen-activated protein kinase; KEGG: Kyoto Encyclopedia of Genes and Genomes; GO: Gene Ontology; BP: Biological process; CC: Cellular component; MF: Molecular function; 5’-UTR: 5’ untranslated region; mRNA: Message RNA.

A HFD upregulates TSC22D1 expression and promotes LSEC dysfunction

To further validate the increased expression of TSC22D1 in the LSECs of NAFLD mice and explore its relationship with endothelial-mesenchymal transition (EndMT) and angiogenesis, the mice were fed a HFD for 12 or 20 weeks to establish a NAFLD model and subsequently assessed. HE staining revealed that compared with the normal group, the HFD 20W group displayed marked fat vacuole degeneration and disorganized liver tissue structure, indicating successful NAFLD model establishment (Figure 3A). Additionally, Sirius red staining revealed significant collagen deposition in the portal area in the HFD 20W group compared with the normal group, suggesting progression towards fibrosis. Oil red O staining clearly revealed significantly increased lipid accumulation in the liver tissues of HFD-fed 20W mice, as shown in Figure 3A. Next, we performed biochemical examinations on the three groups of mice, and the results revealed that the serum alanine aminotransferase, aspartate aminotransferase and triglyceride values of the mice increased gradually with increasing duration of a HFD (Figure 3B). To demonstrate LSEC dysfunction in NAFLD, we performed electron microscopy, which revealed that the density of fenestrae in the livers of HFD-fed 20 weeks mice was lower than that in the livers of normal mice (Figure 3C). Hepatic sinusoidal endothelial cell dysfunction also involves the promotion of microvascularization and EndMT[19]. To elucidate microvascularization in each group, we used immunohistochemistry to detect the expression of cluster of differentiation (CD) 34, laminin, and angiotensin-2 (Ang-2) in the liver tissues of each group[20,21]. The results indicated that CD34, laminin, and Ang-2 Levels gradually increased with prolonged HFD (Figure 3D). To elucidate EndMT in each group, we used immunohistochemistry to detect the expression of α-smooth muscle actin (α-SMA), E-cadherin, and N-cadherin in the liver tissues[22]. The results indicated that with prolonged HFD, α-SMA, E-cadherin, and N-cadherin expression levels gradually increased (Figure 3E). The WB results revealed a significant increase in TSC22D1 expression levels in the liver tissues of the mice fed a prolonged HFD (Figure 3F). We used CD31 to mark endothelial cells. The immunofluorescence results revealed a significant increase in the CD31 and TSC22D1 expression levels in the liver tissues of the mice fed a prolonged HFD, with a marked increase in colocalization areas (Figure 3G). These results suggest that with the progression of NAFLD in mice, endothelial dysfunction gradually worsens, EndMT and microvascularization increase and TSC22D1 expression levels in endothelial cells also progressively increase.

Figure 3 High-fat diet upregulates TSC22D1 expression and promotes liver sinusoidal endothelial cell dysfunction. A: Hematoxylin eosin, Sirius red, and Oil red O staining of liver tissues from three groups of mice; B: Biochemical indices of the three groups of mice; C: Electron microscopy showing fenestrae in liver tissues with orange arrows; D: Immunohistochemical staining for cluster of differentiation (CD) 34, laminin, and angiotensin-2; E: Immunohistochemical staining for α-smooth muscle actin, E-cadherin, and N-cadherin; F: Western blot analysis for TSC22D1 expression; G: Immunofluorescence staining for CD31 and TSC22D1. Scale bar 100 μm, 100 × magnification. Scale bar 50 μm, 200 × magnification. Scale bar 2 μm, 5000 × magnification. n = 6 mice each group. ^aP < 0.05; ^bP < 0.01; ^cP < 0.001. One-way analysis of variance. HE: Hematoxylin eosin; HFD: High-fat diet; CD: Cluster of differentiation; Ang-2: Angiotensin-2; NS: No significant; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; TG: Triglyceride; SEM: Scanning electron microscopy; SMA: Smooth muscle actin; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; DAPI: 4’,6-diamidino-2-phenylindole.

OxLDL promotes microvascularization and EndMT in LSECs and upregulates TSC22D1 expression

We cocultured oxLDL with LSECs to simulate the microenvironment in NAFLD and validated the model’s effectiveness using Oil red O staining (Figure 4A). The results showed that lipid deposition in LSECs was significantly increased after oxLDL intervention. We subsequently applied this model to analyse the effects and mechanisms of lipotoxicity on LSEC dysfunction. Immunofluorescence analysis revealed that the protein expression levels of microvascularization-related proteins CD34, laminin, and Ang-2, as well as the EndMT-related proteins α-SMA, E-cadherin, and N-cadherin, were elevated in oxLDL-treated LSECs compared with normal LSECs (Figure 4B). The tube formation assay demonstrated that, in comparison with those in normal LSECs, the number of node points and junction points, as well as the total length of the tubes, were greater in oxLDL-treated LSECs (Figure 4C). WB analysis revealed that, compared with that in normal LSECs, the expression of TSC22D1 was increased in oxLDL-treated LSECs (Figure 4D). Moreover, furthermore, fluorescence double staining revealed that the protein expression levels of TSC22D1 and TWEAK were both increased (Figure 4E), and colocalization analysis revealed an average Pearson’s R value of 0.80 and an overlap R value of 0.93. These results indicate that TSC22D1/TWEAK may be involved in the microvascularization and EndMT of LSECs.

Figure 4 Oxidized low-density lipoprotein upregulates TSC22D1 expression and promotes microvascular and endothelial-mesenchymal transition of liver sinusoidal endothelial cells. A: Oil red O staining for the verification of the oxidized low-density lipoprotein (oxLDL) intervention model; B: Western blot analysis of TSC22D1 expression in normal and oxLDL-treated liver sinusoidal endothelial cells; C: Tube formation assay; D: Immunofluorescence staining for cluster of differentiation 34, laminin, and angiotensin-2, α-smooth muscle actin, E-cadherin, and N-cadherin; E: Fluorescence double staining for TSC22D1 and tumor necrosis factor-like weak inducer of apoptosis. ^aP < 0.05; ^bP < 0.01; ^cP < 0.001. Student’s t test. Scale bar 50 μm, 200 × magnification. NC: Negative control; oxLDL: Oxidized low-density lipoprotein; CD: Cluster of differentiation; DAPI: 4’,6-diamidino-2-phenylindole; Ang-2: Angiotensin-2; SMA: Smooth muscle actin; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; TWEAK: Tumor necrosis factor-like weak inducer of apoptosis.

TSC22D1 promotes the microvasculature and EndMT of LSECs via the TWEAK/FN14 pathway

To further validate the role of TSC22D1 in hepatic sinusoidal endothelial cell microvascularization and EndMT, we constructed a TSC22D1 overexpression plasmid and confirmed its efficacy using western blotting (Figure 5A). Additionally, we used L524-0366 to inhibit the TWEAK/FN14 signalling pathway[23]. We confirmed through cell counting kit 8 analysis that treatment with 10 μM L524-0366 did not result in a statistically significant change in cell viability (Supplementary Figure 2D). Immunofluorescence analysis revealed that in TSC22D1-overexpressing LSECs treated with oxLDL, the protein expression levels of the microvascularization-related proteins CD34, laminin, and Ang-2, as well as the EndMT-related proteins α-SMA, E-cadherin, and N-cadherin, were elevated, an effect that was reversed by L524-0366 (Figure 5B). The tube formation assay indicated that in TSC22D1-overexpressing LSECs treated with oxLDL, the number of node points and junction points, along with the total tube length, increased, and this increase was reversed by L524-0366 (Figure 5C). Electron microscopy revealed that in TSC22D1-overexpressing LSECs, the density of fenestrae significantly decreased after oxLDL treatment, and this reduction was reversed by L524-0366 (Figure 5D). Fluorescence double staining revealed that, in TSC22D1-overexpressing LSECs, the expression levels of both the TSC22D1 and TWEAK proteins were increased following oxLDL treatment, and this increased in TWEAK protein expression was reversed by L524-0366 (Figure 5E). Subsequent WB analysis revealed that, following oxLDL treatment, the expression levels of TSC22D1 and TWEAK/FN14 pathway proteins were increased, and this increase was also reversed by L524-0366 (Figure 5F). Thus, overexpression of TSC22D1 induces microvascularization and EndMT of LSECs through the TWEAK/FN14 signalling pathway, thereby promoting the dysfunction of LSECs.

Figure 5 TSC22D1 promotes the microvascular and endothelial-mesenchymal transition of liver sinusoidal endothelial cells via the tumor necrosis factor-like weak inducer of apoptosis/fibroblast growth factor-inducible 14 pathway. A: Western blot analysis for the verification of TSC22D1 overexpression plasmid; B: Immunofluorescence staining for cluster of differentiation 34, laminin, angiotensin-2, α-smooth muscle actin, E-cadherin, and N-cadherin; C: Tube formation assay; D: Electron microscopy showing fenestrae density in liver sinusoidal endothelial cells; E: Fluorescence double staining for TSC22D1 and tumor necrosis factor-like weak inducer of apoptosis (TWEAK); F: Western blot analysis for TSC22D1 and TWEAK/fibroblast growth factor-inducible 14 pathway. Scale bar 100 μm, 100 × magnification. Scale bar 50 μm, 200 × magnification. Scale bar 10 μm, 1000 × magnification. ^aP < 0.05; ^bP < 0.01; ^cP < 0.001. One-way analysis of variance. NS: No significant; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; CD: Cluster of differentiation; DAPI: 4’,6-diamidino-2-phenylindole; Ang-2: Angiotensin-2; SMA: Smooth muscle actin; TWEAK: Tumor necrosis factor-like weak inducer of apoptosis; FN14: Fibroblast growth factor-inducible 14.

TWEAK secreted by LSECs overexpressing TSC22D1 promotes M1 macrophage polarization via the TWEAK/FN14 pathway

On the basis of our previous single-cell analysis, we hypothesized that LSECs mediate intercellular communication with macrophages via TWEAK. To test this hypothesis, we conducted an ELISA on the supernatants from the three LSEC groups (control group, TSC22D1 group, and TSC22D1 + L524-0366 group), which revealed a significantly greater TWEAK expression level in the TSC22D1 group than in the control group (Figure 6A). The supernatants of the three groups were subsequently extracted and cultured in macrophages. Flow cytometry analysis revealed that after 24 hours of treatment, both M1 and M2 macrophage expression levels were modestly elevated in the TSC22D1 group. Notably, after 48 hours of treatment, the number of M1 macrophages increased significantly, whereas the number of M2 macrophages decreased (Figure 6B). Subsequent ELISA revealed that after 48 hours of treatment, the expression levels of proinflammatory cytokines [interleukin (IL)-6, tumor necrosis factor (TNF)-α] increased, and those of anti-inflammatory cytokines (IL-10, TGF-β) decreased in the TSC22D1 group, with these changes being reversed by L524-0366 (Figure 6C). Additionally, PCR analysis revealed that in the TSC22D1 group, the expression of M1 polarization markers [inducible nitric oxide synthase (iNOS) and monocyte chemoattractant protein-1] increased, whereas the expression of M2 polarization markers (arginase-1 and CD206) decreased, an effect similarly reversed by L524-0366 (Figure 6D). These findings indicate that TWEAK secreted by LSECs overexpressing TSC22D1 promotes M1 macrophage polarization via the TWEAK/FN14 pathway.

Figure 6 Tumor necrosis factor-like weak inducer of apoptosis promotes macrophage M1 polarization secreted by liver sinusoidal endothelial cells overexpressing TSC22D1 via the tumor necrosis factor-like weak inducer of apoptosis/fibroblast growth factor-inducible 14 pathway. A: Enzyme-linked immunosorbent assay (ELISA) analysis of tumor necrosis factor-like weak inducer of apoptosis in the supernatants of liver sinusoidal endothelial cells (LSECs); B: Flow cytometry analysis detecting the expression levels of M1 and M2 macrophages; C: ELISA analysis of interleukin (IL)-6, tumor necrosis factor-α, IL-10, and transforming growth factor-β in the supernatants of macrophages cultured in conditioned media from LSECs; D: Quantitative real-time polymerase chain reaction analysis of inducible nitric oxide synthase, monocyte chemoattractant protein-1, arginase-1 and cluster of differentiation 206 in macrophages. ^aP < 0.05; ^bP < 0.01; ^cP < 0.001. One-way analysis of variance. TWEAK: Tumor necrosis factor-like weak inducer of apoptosis; CD: Cluster of differentiation; FITC: Fluorescein Isothiocyanate; APC: Antigen-presenting cell; NS: No significant; IL: Interleukin; TNF: Tumor necrosis factor; TGF: Transforming growth factor; iNOS: Inducible nitric oxide synthase; MCP: Monocyte chemoattractant protein; Arg-1: Arginase-1.

Targeting TSC22D1 alleviates LSEC dysfunction and inhibits macrophage M1 polarization via the TWEAK/FN14 pathway in vivo

Finally, we performed an in vivo experiment to analyse the effect of TSC22D1 on endothelial cell dysfunction and its regulation of macrophage polarization in a NAFLD murine model. We used adenovirus to inhibit mouse liver TSC22D1 expression. The sequences of the shRNA for TSC22D1 are shown in Supplementary Table 3. The interference efficiency of sh-TSC22D1-2 was the best, as verified by PCR (Figure 7A). Therefore, we encapsulated sh-TSC22D1-2 in adenovirus and injected it into mice to knock down TSC22D1. Compared with the sh-NC group, the sh-TSC22D1 group presented fewer fat vacuoles, lipid accumulation, and collagen deposition in liver tissue, indicating a milder degree of fatty liver (Figure 7B). Immunofluorescence results showed that the expression levels of CD31 and TSC22D1 in the liver tissues of the sh-TSC22D1 group were significantly decreased, and the colocalization area was significantly reduced (Figure 7C). The immunohistochemistry results revealed that the expression levels of CD34, laminin, Ang-2, α-SMA, E-cadherin, and N-cadherin in the liver tissues of the sh-TSC22D1 group were lower than those in the sh-NC group (Figure 7D). Additionally, decreased levels of mouse epidermal growth factor-like module-containing mucin-like hormone receptor-like 1 (F4/80) and iNOS were detected by immunohistochemistry, indicating that the M1 polarization of intrahepatic macrophages was attenuated after TSC22D1 knockdown (Figure 7E). As shown in Figure 7F, TSC22D1 knockdown significantly reduced the expression levels of the TWEAK/FN14 signalling pathway. These results collectively suggest that targeting TSC22D1 may improve endothelial cell dysfunction, reduce macrophage M1 polarization, and alleviate NAFLD-associated fibrosis through inhibition of the TWEAK/FN14 signalling pathway.

Figure 7 Targeting TSC22D1 alleviates liver sinusoidal endothelial cells dysfunction and inhibits macrophage M1 polarization via the tumor necrosis factor-like weak inducer of apoptosis/fibroblast growth factor-inducible 14 pathway in vivo. A: The interference efficiency of different short hairpin RNAs against TSC22D1 via quantitative real-time polymerase chain reaction analysis; B: Hematoxylin eosin, Sirius red, and Oil red O staining of liver tissues from the adeno-associated virus serotype 8-short hairpin RNA-negative control group and the recombinant adeno-associated virus serotype 8 carrying TSC22D1-targeting short hairpin RNA vector group; C: Fluorescence double staining for cluster of differentiation (CD) 31 and TSC22D1; D: Immunohistochemical staining for CD34, laminin, angiotensin-2, α-smooth muscle actin, E-cadherin, and N-cadherin in liver tissues; E: Immunohistochemical staining for mouse epidermal growth factor-like module-containing mucin-like hormone receptor-like 1, inducible nitric oxide synthase, and CD206; F: Western blot analysis for TSC22D1 and tumor necrosis factor-like weak inducer of apoptosis/fibroblast growth factor-inducible 14 pathway. Scale bar 100 μm, 100 × magnification. Scale bar 50 μm, 200 × magnification. ^aP < 0.05; ^bP < 0.01; ^cP < 0.001. Student’s t test. n = 6 mice each group. sh-TSC22D1: The recombinant adeno-associated virus serotype 8 carrying TSC22D1-targeting short hairpin RNA vector; sh-NC: The recombinant adeno-associated virus serotype 8 carrying negative control short hairpin RNA vector; F4/80: Mouse epidermal growth factor-like module-containing mucin-like hormone receptor-like 1; HE: Hematoxylin eosin; CD: Cluster of differentiation; DAPI: 4’,6-diamidino-2-phenylindole; Ang-2: Angiotensin-2; SMA: Smooth muscle actin; iNOS: Inducible nitric oxide synthase; NS: No significant; TWEAK: Tumor necrosis factor-like weak inducer of apoptosis; FN14: Fibroblast growth factor-inducible 14; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase.

DISCUSSION

Our results demonstrate that TSC22D1 significantly influences LSEC mesenchymal transition and microvascularization by activating the TWEAK/FN14 signalling pathway, leading to LSEC dysfunction. Additionally, TSC22D1 promotes the secretion of TWEAK from LSECs, thereby promoting macrophage polarization towards the M1 phenotype, and these M1 macrophages secreting substantial quantities of inflammatory cytokines that exacerbate liver inflammation and fibrosis progression. We comprehensively revealed the crucial role of TSC22D1 in NAFLD fibrosis, highlighting its dual role in regulating endothelial cell function and the inflammatory response and highlighting to the potential of TSC22D1 as a therapeutic target[24].

The TSC22D1 gene, encoding a TSC22 domain family protein, has been the subject of recent research given its pleiotropic roles in fundamental biological pathways and established associations with pathological conditions[25]. Emerging evidence has established TSC22D1 as a critical regulatory node governing oncogenic processes, immune cell homeostasis, and systemic lipid metabolism[26,27]. Furthermore, TSC22D1 has been identified as a novel susceptibility gene for spontaneous pulmonary adenoma formation, positioning it as a potential therapeutic target for lung adenocarcinoma intervention[28]. Zheng et al[29] demonstrated that the RNA-binding protein MEX3D drives cervical carcinogenesis through TSC22D1 message RNA destabilization, revealing a crucial posttranscriptional regulatory axis in tumour progression. Pépin et al[30] demonstrated that TSC-22 promotes apoptosis in T-lymphocytes under IL-2 deprivation, emphasizing its importance in immune regulation. Vargas et al[31] identified TSC22D1 as a potential new target for the treatment of Alzheimer’s disease. Additionally, TSC22D1 stimulated by TGF-β1, controls systemic cholesterol metabolism in the liver, underscoring its importance in metabolic regulation[32]. Collectively, these findings suggest that TSC22D1 is a pivotal gene in various biological processes and could serve as a novel target for therapeutic interventions in cancer, immune disorders, and metabolic diseases.

Our findings align with and extend existing research on the role of TWEAK/FN14 in fibrotic diseases, identifying TSC22D1 as a novel regulator of NAFLD fibrosis. TWEAK is a multifunctional cytokine belonging to the TNF superfamily that is predominantly secreted by myeloid and immune cells and mediates diverse pathophysiological processes including inflammatory responses, tissue regeneration, and cellular viability through specific binding to its cognate receptor FN14[33-36]. In NASH, pathological overactivation of the TWEAK/FN14 signalling axis drives sustained proinflammatory responses and potentiates fibrogenesis[37]. Mechanistically, TWEAK stimulates HSC chemotaxis by sequentially activating the epidermal growth factor receptor (EGFR)/SRC-mediated motility machinery and phosphatidylinositol 3-kinase (PI3K)/protein kinase B (AKT)-dependent survival signalling, while concurrently enhancing nuclear factor kappa-B (NF-κB)/signal transducers and activators of transcription 3 (STAT3)-driven production of profibrogenic cytokines through transcriptional activation[38,39]. Moreover, FN14-deficient mice exhibited significantly reduced hematopoietic progenitor cell (HPC) proliferation, inflammation, and fibrosis when subjected to a choline-deficient, ethionine-supplemented diet model[40,41]. These results suggest that the inhibition of TWEAK or its receptor FN14 can effectively reduce HSC activation and endothelial cell dysfunction, thereby mitigating liver fibrosis. In addition to TWEAK, TSC22D1 may also promote fatty liver fibrosis by regulating the TGF-β or collagen 3a1 genes[42,43]. Recent studies have demonstrated that the activation of the Yes-associated protein signalling pathway by LSEC-S1pr2 enhances the transactivation of TGF-β, which subsequently acts on HSCs through a paracrine mechanism, thereby exacerbating liver fibrosis[44].

In this study, we discovered that TWEAK secreted by LSECs plays a pivotal role in the polarization of macrophages into the M1 phenotype, potentially mediated through the TWEAK/FN14 interaction. Consistent with these findings, TWEAK has been shown to promote M1 polarization in both liver Kupffer cells and mononuclear macrophages, an effect associated with increased expression of proinflammatory cytokines. Disrupting TWEAK signalling through the inhibition of the CD266 or NOD-like receptor thermal protein domain associated protein 3 pathways significantly attenuates this effect, underscoring the importance of these signalling pathways in TWEAK-mediated M1 polarization[45]. Furthermore, the interaction between TWEAK and its receptor FN14 is crucial for the regulation of macrophage polarization, as demonstrated by the increased M1-like phenotype following the reduction in the CD163-TWEAK complex in lipopolysaccharide-treated macrophages[46]. Additionally, the TWEAK/FN14 interaction has been implicated in the upregulation of high mobility group protein 1 expression specifically in M1 macrophages, further linking TWEAK to the enhancement of inflammatory responses within atherosclerotic plaques[47]. These findings suggest that TWEAK/FN14 signalling functions as a critical mediator of M1 polarization and may serve as a therapeutic target for modulating inflammatory responses in various pathological conditions.

In addition to inducing M1 macrophage polarization, TWEAK secreted by LSECs significantly affects hepatocytes, hepatic progenitor cells, and HSCs. The TWEAK/FN14 signalling pathway is highly active in metabolic dysfunction-associated fatty liver disease, suggesting its critical role in both liver injury and regeneration processes[48]. Specifically, Liao et al[49] demonstrated that TWEAK/FN14 promotes hepatocyte pyroptosis through the NF-κB/caspase-1/gasdermin D pathway. Furthermore, TWEAK serves as a direct proliferative factor for HPCs, and the TWEAK/FN14 axis activates the NF-κB and STAT3 signalling pathways, leading to increased production of proinflammatory cytokines during liver disease progression[38]. Additionally, Chen et al[50] revealed that TWEAK promotes HSC migration through the activation of the EGFR/SRC and PI3K/AKT pathways, thus facilitating the transition from NAFLD to liver fibrosis. TWEAK secreted by LSECs may act on multiple cell types within the liver, including hepatocytes, HSCs, and macrophages, triggering a cascade of signalling pathways that drive inflammation, fibrosis, and tissue remodelling in fatty liver disease.

Unlike classical regulators of LSECs, such as vascular endothelial growth factor (which promotes endothelial survival and microvascular integrity) or notch (which regulates angiocrine signalling and regeneration), TSC22D1 drives endothelial dysfunction via mesenchymal transition, aberrant microvascularization, and proinflammatory activation[51,52]. Mechanistically, TSC22D1 uniquely activates the TWEAK/FN14 axis, shifting LSECs towards pathological rather than homeostatic signaling-a distinction underscoring its nonredundant role in chronic liver injury and fibrosis. The TWEAK/FN14 pathway itself plays dual roles: Transient activation supports hepatocyte proliferation and progenitor cell expansion for regeneration, whereas chronic activation exacerbates inflammation and fibrogenesis by activating stellate cells and recruiting M1 macrophages[46,53]. This duality complicates therapeutic targeting, as interventions must balance timing, dosage, and disease context to suppress fibrosis without impairing repair. Thus, TSC22D1 represents a promising target for NAFLD-related fibrosis, but its downstream effects on TWEAK/FN14 necessitate precision modulation to uncouple detrimental signalling from regenerative responses.

Although this study elucidates the critical role of TSC22D1 in NAFLD fibrosis and its regulatory mechanisms through the TWEAK/FN14 signalling pathway in LSEC mesenchymal transition and macrophage polarization, several limitations should be acknowledged. First, the study relies primarily on animal models and in vitro experiments, with a relatively limited sample size, which may not fully capture the complex pathophysiology of human NAFLD. Although the HFD-induced mouse model and oxLDL-treated LSECs are widely used to mimic the pathophysiological features of NAFLD and NASH, these models have certain limitations. The HFD model primarily replicates the metabolic aspects and early inflammatory responses of NAFLD, but it may not fully reproduce the complex histopathological features of advanced NASH, such as significant hepatocellular ballooning or progressive fibrosis. Similarly, while oxLDL stimulation of LSECs can effectively simulate the endothelial stress and dysfunction observed in NAFLD, it may not capture the full spectrum of in vivo cellular interactions, immune responses, or chronic injury processes occurring within the liver microenvironment. Therefore, the findings derived from these models should be interpreted with caution. Future studies incorporating additional NASH models (e.g., methionine- and choline-deficient diets or Western diets) or coculture systems that better recapitulate the liver’s multicellular milieu may help to further validate and extend our observations. Second, inherent differences between animal models and human diseases, such as metabolic profiles, immune responses, and fibrosis progression, may restrict the direct clinical translation of the findings. Third, in the in vivo experiments, AAV8 vectors were delivered via the tail vein, and while AAV8 has high liver tropism and preferentially transduces hepatocytes and endothelial cells, off-target knockdown may occur in other hepatic cell types. Further studies using cell-type-specific knockout models are warranted to confirm these results. Additionally, while current methodologies (e.g., Western blotting and immunohistochemistry) provide molecular-level insights, their limitations in spatial resolution and dynamic monitoring may hinder a comprehensive understanding of the multifaceted functions of TSC22D1. Finally, although our study demonstrated that TSC22D1 overexpression leads to increased TWEAK expression at both the message RNA and protein levels, the precise regulatory mechanism underlying this relationship remains unclear. It is currently unknown whether TSC22D1 directly modulates TWEAK transcription through promoter binding or indirectly affects TWEAK expression via intermediate signalling pathways or RNA-binding interactions. Future studies are warranted to elucidate the molecular interactions between TSC22D1 and the TWEAK/FN14 signalling axis.

CONCLUSION

In conclusion, this study provides a comprehensive understanding of the role of TSC22D1 in the pathogenesis of NAFLD fibrosis. We elucidated the mechanisms through which TSC22D1 drives LSEC microvascularization and EndMT, as well as its role in promoting the secretion of TWEAK, which induces macrophage polarization towards the M1 phenotype. These findings offer novel insights into the pathophysiology of NAFLD, particularly the interplay between endothelial dysfunction, inflammation, and fibrosis. Importantly, our results highlight the potential of TSC22D1 as a therapeutic target for NAFLD. Future research should focus on validating these mechanisms in human clinical cohorts and developing targeted interventions, such as TSC22D1 inhibitors or modulators of the TWEAK/FN14 signalling pathway, to translate these findings into effective treatments for NAFLD.