1. Spatial transcriptome reveals the spatial distribution characteristics of NASH fibrosis
Twelve samples, including four healthy controls, four NAFL patients and four NASH patients, were selected for spatial transcriptome sequencing. After quality inspection, four samples were finally selected for space transcriptome sequencing, including one from the healthy control (pathology number: 191518), one from an NAFL patient (pathology number: 20201869, SAF score: S2AOF1) and two from NASH patients (pathology number: 20200714, SAF score: S3A2F2; pathology number: 20201896, SAF score: S3A2F2). The detection of unique molecular identifiers (UMIs) corresponding to each spot of each chip is shown in Figure 1A. PCA linear dimension reduction was performed on the spatial transcriptomic data using Seurat_4.1.1, and the appropriate principal component (PC) was selected for cluster analysis. t-SNE&UMAP was used to visualize the clustering results, as shown in Figure 1B. After more than 30 principal components, the standard deviation of the data tended to be stable, and the first 30 PCs were selected for downstream data analysis. The samples were divided into 8 clusters by dimensionality reduction and clustering (Figure 2A). Combined with the pathological results, it can be inferred that Cluster 5 is a fibrotic region with high probability. NASH fibrosis was mainly distributed in lobules, and a small amount of fibrosis was also observed in the portal area (Figure 2B).
Then, functional analysis of Cluster 5 was carried out, and GO analysis showed that biological processes (BP), such as ECM receptor interaction, were mainly concentrated in the ECM and structural tissues. Molecular function (MF), such as those of growth factors and integrins, was mainly enriched in the ECM structural components and signal molecule transmission. The cell composition (CC) was mainly concentrated in the plasma membrane matrix, as shown in Figure 3A. According to KEGG analysis, differentially expressed genes were mainly enriched in the PI3K Akt signalling pathway and ECM-receptor interaction, as shown in Figure 3B. GSEA enrichment analysis focused on ECM-related pathways (Figure 3C, D), such as NABA CORE MATRISOME and NABA ECM GLYCOPROTEINS. NABA CORE MATRISOME is a collection of genes encoding the core extracellular matrix, including ECM glycoprotein, collagen and proteoglycan. NABA ECM GLYCOPROTEINS is a gene encoding the structure of ECM glycoprotein. In summary, functional clustering results preliminarily verified cluster5 as the fibrosis region.
2. Single-cell and spatial transcriptomics reveal the key role of HSCs in NASH fibrosis
SPOTLight is centred around a seeded nonnegative matrix factorization (NMF) regression, initialized using cell-type marker genes and nonnegative least squares (NNLS) to subsequently deconvolute spatial transcriptomics capture locations (spots). In integrating the spatial transcriptome data and the single-cell sequencing dataset (GSE189175), a total of 6 cell types were identified from GSE189175. Compared with the healthy control and NAFL groups, the NASH group had significantly increased proportions of HSCs and myofibroblasts, which were distributed in the lobule and the portal area around the fibrotic area. At the same time, the infiltration of Kupffer cells around the fibrotic area were also increased (Figure 4). The cell communication analysis showed that diffusive cell communication was the main type, including endocrine, paracrine and autocrine communication, followed by ECM-receptor cell communication (Figure 5A). In NASH fibrosis, hepatocytes, HSCs, myofibroblasts and bile duct cells have strong communication with each other (Figure 5B), and hepatocytes, HSCs and myofibroblasts are the main types of cells that communicate via the collagen signalling pathways (Figure 5C).
3. Key genes specifically overexpressed in the NASH fibrosis region
By differential analysis of subgroups, 303 differentially expressed genes were obtained from Cluster 5 (fibrosis region). The protein interaction network shows that there are many key fibrosis and fibrin genes in Cluster 5 (Figure 6). A literature search was performed for differentially expressed genes, and signalling pathway analysis was combined to predict the mRNAs associated with fibrosis. The distribution and expression of key fibrosis genes in tissues and subpopulations were analysed. Finally, AEBP1, DPT, CCL19 and NOTCH3 were highly expressed in the fibrotic area (Figures 7,8). We further analysed the expression of the above genes in the single-cell transcriptome; AEBP1 and DPT were relatively highly expressed specifically in HSCs and myofibroblasts (Figure 9).
To explore the potential role of AEBP1+ and DPT+ myofibroblasts in liver fibrosis, SCENIC analysis was used to divide myofibroblasts into two groups according to the expression of AEBP1 and DPT: AEBP1+ and AEBP1−, DPT+ and DPT-. The differences in cellular signalling pathways between the two groups were analysed (Tables 1, 2). We found that AEBP1+ and DPT+ myofibroblasts are involved in the activation of HSCs and the formation of fibrosis. For example, AEBP1+ myofibroblasts are involved in the positive regulation of HSC proliferation and the transforming growth factor receptor signalling pathway. DPT+ myofibroblasts participate in the positive regulation of myofibroblast proliferation and transforming TGF-β production. The specific TFs of AEBP1+/AEBP1- and DPT+/DPT− myoblasts were predicted by SCENIC, and TFs with high relative activity scores were identified (Figure 10). We found that these two groups of cells were regulated by four TFs: SOX4, IRF8, GATA6 and TBX2. At the same time, we found that these four TFs were highly expressed in cluster5, suggesting that they may be important potential TFs of AEBP1+ and DPT+ myofibroblasts.
4. Verification of the expression and function of key genes in NASH fibrosis by in vitro and in vivo experiments
First, immunohistochemical detection of AEBP1 and DPT in liver tissues of healthy controls, NAFL patients and NASH patients was carried out. The results showed that the protein expression levels of AEBP1 and DPT in liver tissues of NASH patients were significantly higher than those in the first two groups, as shown in Figure 11. Second, after LX2 cells were transfected with three AEBP1 and DPT interference fragments, siRNA-AEBP1-2 and siRNA-DPT-2 with the best interference effect were selected for subsequent experiments. The mRNA levels of collagen I in siRNA-AEBP1-2 and siRNA-DPT-2 cells were significantly lower than those in the siRNA-NC group and blank control group, as shown in Figure 12.