2.1 Identification of MenSCs and HBV transgenic mice

The expression of CD29, CD73, CD90, and CD105 was positive, whereas that of CD34, CD45, CD117, and human leukocyte antigen-DR (HLA-DR) was negative (Figure 1B). MenSCs displayed a spindle- and fibroblast-like shape and successfully differentiated into osteogenic, adipogenic, and chondrogenic cells under the appropriate conditions (Figure 1C).

The average of hepatitis B surface antigen (HBsAg) remained almost the same in each group of enrolled mice on day 1 (Figure S1A). C57BL/6 mice and HBV-Tg mice have no phenotypic variations (Figure S1B). According to the polymerase chain reaction (PCR) result, both the HBV-Tg mice and C57BL/6 mice were positive for the expression of the internal positive control (Figure 1D). The HBV-Tg mice expressed the HBV transgene target (size 1535 bp); however, the C57BL/6 mice did not express HBV DNA (Figure 1E), which further confirmed the establishment of HBV-Tg mice.

2.2 MenSCs migrated into injured livers and improved serological indicators in HBV-Tg mouse liver fibrosis

In vivo imaging results showed that within 14 days of MenSC transplantation, fluorescence existed at a high intensity (Figure 2A), and on day 5 post MenSC transplantation, the relative fluorescence intensity (n = 5) showed the strongest value (Figure 2B). By detecting the fluorescence intensity of the dissected liver, lung, spleen, kidneys, and heart, we found that MenSCs were mainly concentrated in the liver (Figure 2C), and the average radiation efficiency of the mouse liver imaging on 3 days after transplantation was higher than that 14 days (Figure 2D), which was consistent with the in vivo imaging data (Figure 2B). Thus, MenSCs mainly migrated into the injured liver and lasted for at least 2 weeks.

Spleen injections (Figure S1C) and surgical sutures (Figure S1D) for MenSC transplantation were administered. The HBV-Tg mouse livers in the control group were glossy, but those in the CCl₄ group sustained injury and produced granular fibrosis (Figure S1E–H). An analysis of body weight (Figure S1I) and normalized body weight (Figure S1J) revealed no significant differences among the CCl₄, MenSC-low, and MenSC-high groups. Most serological indicators in the MenSC-low group showed remission trends compared with in the CCl₄ model group. High doses of MenSCs significantly improved alanine aminotransferase (ALT) (Figure 2E), aspartate aminotransferase (AST) (Figure 2F), alkaline phosphatase (ALP) (Figure 2G), and albumin (ALB) (Figure 2H) levels on days 7 and 14 post MenSC transplantation. We studied hepatitis B-related indicators in the sera of HBV-Tg mice, including hepatitis B e antigen (HBeAg) (Figure 2I), HBsAg (Figure 2J), and HBV DNA (Figure 2K). The serum HBeAg level in the MenSC-high group was significantly lower than that in the CCl₄ model group in MenSC 2 W (Figure 2I). Compared with those in the control group, HBsAg levels in the CCl₄, MenSC-low, and MenSC-high groups were reduced (Figure 2J). This result indicated that mouse hepatocytes were damaged, causing a decrease in the expression of HBsAg. The damaged mouse hepatocytes were unable to express HBsAg, and the ability to express HBV serological indexes was reduced. This indirectly proves that the CCl₄-induced liver fibrosis models indeed affected the function of hepatocytes. After MenSC transplantation, HBsAg levels in low-dose and CCl₄ model groups for both MenSC 1 W and MenSC 2 W did not differ significantly. The HBsAg levels in the high-dose group were significantly lower than those in the CCl₄ model group for both MenSC 1 W and MenSC 2 W (Figure 2J). The HBV DNA level in the CCl₄ group was significantly lower than that in the control group (Figure 2K), indicating that damaged hepatocytes reduced the capacity for expressing HBV DNA after CCl₄ sustained injury. Compared with that in the CCl₄ group, the expression of HBV DNA in the MenSC-high group was significantly reduced in 2 weeks after MenSC transplantation (Figure 2K).

2.3 MenSCs reduced pathological indicators by reducing collagen fiber in HBV-Tg mice with liver fibrosis

Referring to the steatosis activity fibrosis (SAF) scoring system,⁴⁹ the pathological scores of the livers were evaluated based on the hematoxylin and eosin (HE) and Sirius red (SR) staining (Table S1). The CCl₄, MenSC-low, and MenSC-high groups exhibited hepatic lobular inflammation, hepatocyte hyaline degeneration, ballooning degeneration, bile duct proliferation, and necrosis (Table S1 and Figure 3A,B). Compared with the CCl₄ group, there was a reduction in hepatic lobular inflammation and hyalinization in the high-dose group (Figure 3A). After transplanting MenSCs into fibrotic mice for 2 weeks, the collagen fiber area was reduced (Figure 3B). Compared with the fibrosis scores of the CCl₄ model group (3.1 ± 0.6), those of the MenSC-low (2.4 ± 0.5) and MenSC-high (2.1 ± 0.8) groups were significantly reduced (Figure 3C). A quantitative real-time PCR (qRT-PCR) analysis further revealed that HBV DNA (named X-universal) expression in the liver tissue showed a significant decreasing trend in MenSC-high group compared with CCl₄ model group (Figure 3D). The expression of Col-I (Figure 3E), as shown by immunohistochemistry (IHC), and Col1a1 (Figure 3F), as shown by qRT-PCR, was significantly higher in the CCl₄ model group than in the control group. After high-dose MenSC transplantation, the expression of Col1a1 (Figure 3F) significantly decreased in MenSC 2 W. The expression of α-smooth muscle actin (α-SMA; Acta2) (Figure 3G), as shown by IHC, and Acta2 (Figure 3H), as shown by qRT-PCR, was increased in CCl₄ model group compared with that control group. The difference in Acta2 expression among the four groups was not significant (Figure 3H); however, the observed trend was similar to that of Col1a1. The expression of TGF-β1 showed no significant difference in the IHC (Figure 3I) and qRT-PCR (Figure 3J). Thus, MenSCs reduced fibrotic accumulation independent of TGF-β1 signal, as shown by qRT-PCR analysis.

According to the RNA-bulk analysis, there were 33 common genes in “CCl₄ versus Con up” and “MenSC versus CCl₄ down” and 468 common genes in “CCl₄ versus Con down” and “MenSC versus CCl₄ up” (Table S2 and Figure S2A,B). Based on the common elements, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were classified as infectious and immune-related diseases (Figure S2C,D). Extracellular region/ECM structural constituents were in the top-30 Gene Ontology (GO) Terms (Figure 3K,L) and ECM–receptor interaction was in the top-20 KEGG result (Figure 3M,N). Cell adhesion (integrin binding or cell–cell junction) was a common top-30 GO Term (Figure 3K,L). Therefore, MenSCs may mediate ECM accumulation and cell adhesion during HBV-associated liver fibrosis.

2.4 scRNA-seq identified liver multiple cell populations in HBV-Tg mice with liver fibrosis

A schematic of scRNA-seq using 10× Genomics is presented in Figure 4A. In total, 81,902 cells were collected. After passing quality control and filtering, 79,019 cells were included in the analysis: 25,714 in the control group, 25,726 in the CCl₄ group, and 27,579 in the MenSC group. Initial visual diagrams of t-distributed Stochastic Neighborhood Embedding (t-SNE) (Figure S3A) and Uniform Mobility Approximation and Projection (UMAP) (Figure S3B) were generated, and 32 cell clusters were identified. The Gene_Counts_RNA profile was presented for 32 cell clusters (Figure S3C). During the analysis of the cell samples, cells that had a mitochondrial gene content exceeding 25% were removed from the dataset (Figure S3D). An clustering analysis of UMAP (Figure 4B) was performed based on the expression of known markers referred to related studies.^{29, 30, 50-54} Twelve major cell types were grouped along with the corresponding gene markers: LSECs (Cdh5, Kdr, Bmp2, Flt1), fibroblasts/HSCs (Wt1, Pdgfra, Dcn, Col1a1), cholangiocytes (Epcam, Sox9, Sorbs2, Spp1), hepatocytes (Alb, Ttr, Mup20, Apoc1), plasma cells (Jchain), B cells (Cd79a, Cd79b), DCs (Runx2, Ccr9, Siglech), Kupffer cells (Csf1r, Adgre1, Lyz2, C1qb), T cells (Cd3d, Cd3e, Cd3g), NKs (Il2rb, Nkg7, Klrb1c), neutrophils (S100a8, S100a9), and mast cells (Cpa3, Fcer1a, Ms4a2) (Table S3 and Figures S3E and 4C). Feature plots of representative markers for each cluster are presented in Figure S3F–Q. A heatmap with the top-10 markers of differentially expressed genes (DEGs) in 12 cell types is presented (Figure S4A), which may act as potential markers for future studies. The cluster analysis showed that the main cell populations in the normal control and liver fibrosis groups (CCl₄ and MenSC groups) were similar, but their proportions differed after CCl₄-induced liver fibrosis and following MenSC transplantation (Figures S4B and 4D–G). Compared with those in the control group, there were significant changes in LSECs, Kupffer cells, T cells, and neutrophils in the fibrosis groups. LSECs were drastically reduced in CCl₄ and MenSC groups compared with control group (Figure 4G). Conversely, the proportions of Kupffer cells, T cells, and neutrophils increased significantly, suggesting a proinflammatory state induced by CCl₄. The heatmap revealed that well-known collagen-producing genes (Col1a1, Col1a2, and Col3a1) were specifically expressed in fibroblasts/HSCs subpopulations (Figure S4A). As shown in Figure 4H, the expression of classic fibrosis-inducing genes (Col1a1, Col3a1, Acta2, and Timp1) with a cell type-specific distribution was mainly concentrated in the population of fibroblasts/HSCs. Feature plots for Col1a1 (Figure 4I), Col3a1 (Figure 4J), Acta2 (Figure 4K), and Timp1 (Figure 4L) further indicate that collagenous fibers are predominant in the populations of fibroblasts/HSCs. The relative expression of Col1a1 (Figure 4M), Col3a1 (Figure 4N), Acta2 (Figure 4O), and Timp1 (Figure 4P) was lower in the MenSC group than in the CCl₄ group, which was anastomotic with the IHC and qRT-PCR results of Col-I and α-SMA in vivo. Therefore, these results further verified the therapeutic effects of MenSCs in reducing liver fibrosis in HBV-Tg mice at the scRNA-seq level.

2.5 ScRNA-seq identified the heterogeneity of fibroblasts/HSCs and key molecules, from aHSCs to iHSCs

Unsupervised clustering analyses using UMAP (Figure S5A) identified nine cell populations in fibroblasts/HSCs. It was difficult to define clustered single cells by classic markers including Col1a1, Acta2, and Timp1 (Figure S5B) in the scRNA-seq database. Drawing on other related studies,^{13, 25, 29, 33, 50} we identified five subpopulations: aHSCs (Pdgfrb, Des, Lum), iHSCs (Tcf21, Smoc2, Fbln7), PFs (Msln, Muc16, Igfbp5), neutrophil-like fibroblasts (S100a8, S100a9, Fcgr3), and LSEC-like fibroblasts (Cdh5, Kdr, Bmp2) (Table S4 and Figure 5A,B). Representative markers (Lum, Smoc2, Msln, S100a8, Bmp2) for each cluster with a Feature Plot are presented in Figure S5C–G. The relative proportions of each cell type in the three groups are shown (Figures 5C and S5H). The heatmap of the top-10 markers of DEGs in fibroblasts/HSCs subpopulation is shown (Figure S5I). Some classic markers of qHSCs, including lecithin retinol acyltransferase (Lrat), reelin (Reln), and glial fibrillary acid protein (Gfap), were not expressed in the fibroblast/HSC subpopulation (Figure 5B). These results prove that HBV has a huge impact on the heterogeneity of fibroblasts/HSCs. The expression of Col1a1 (Figure 5D), Acta2 (Figure 5E), and Timp1 (Figure 5F) in fibroblasts/HSCs was reduced in the MenSC group compared with that in the CCl₄ group. Monocle2 pseudotime results showed that aHSCs were predominant in stage 1 and stage 3, iHSCs were predominant in stage 3 (Figures 5G and S5J–L). Among these genes, with the time course, the most upregulated genes in the fibroblast/HSC proportions were Dpt and Gsn, and the most downregulated genes were Crip, Dmkn, Igfbp5, and Saa3 (Figure S5M). Additionally, relative gene expression of six sub-subpopulations with respect to time is shown (Figure 5H).

In the following analyses, we characterized the changes in aHSCs and iHSCs proportions and their transcriptional profiles (Figure 5I,J). The heatmap showing the top-10 markers of DEGs in aHSCs and iHSCs is presented (Figure 5K). Monocle2 results showed that aHSCs can progress towards iHSCs according to the pseudotime (Figure 5L,M and S5N). With the time course, the most upregulated genes for the aHSCs into iHSCs were Dpt, Gsn, and Sfrp1, and the most downregulated genes were Fn1, Serpinb2, and Saa3 (Figure 5N). Additionally, their relative gene expression with respect to time in the clusters is shown in Figure S5O. According to the relative expression trend and top-10 DEGs of aHSCs/iHSCs (Figure 5K), liver fibrosis may be reduced by increasing the expression of Dpt and Gsn and reducing the expression of Saa3. Therefore, these genes may be potential therapeutic targets.

2.6 Reduced ICAM crosstalk of aHSCs and neutrophils was responsible for ameliorating liver fibrosis after MenSC transplant

The Icam1 expression for aHSC (LX-2 cells stimulated with recombinant TGF-β1) by coculturing with MenSC in vitro (Figure S5P,Q) was assessed. Although the relative expression of Icam1 was reduced by qRT-PCR (Figure S5R, 1.02 ± 0.23 in LX-2 group and 0.78 ± 0.28 in coculture group), however, there are no statistical discrepancy. To further examine the interactions among aHSCs, iHSCs, and other cells, we used CellChat to investigate the relationship between aHSCs, iHSCs, T cells, B cells, Kupffer cells, neutrophils, and LSECs in the CCl₄ versus control groups (Figure 6A) and MenSC versus CCl₄ groups (Figure 6B). The strength of the aHSC-exported signals was significantly increased in neutrophils, Kupffer cells, and B cells in the CCl₄ group compared with that in the control group (Figure 6A), while the signal strength of the MenSC group significantly decreased after the MenSC therapy (Figure 6B). Further research found that 15 signaling pathways were stronger in the CCl₄ group than in the control group (Figures S6A and 6C), while 23 signals were weaker in the MenSC group than in the CCl₄ group (Figures S6B and 6D). Although some common pathways (marked with pink font in Figure 6C,D) were found to be consistent, but there are either no expression or no signals targeting neutrophils/Kupffer cells/B cells. By further contrast, five signal channels were selected (marked with red font in Figure 6C,D), including ICAM, junctional adhesion molecule (JAM), C–C motif chemokine ligand (CCL), CHEMERIN, and growth arrest-specific (GAS) and they were further confirmed by a chord diagram, respectively (Figure 6E–I). Through stratigraphy of these chord maps, we identified ICAM (Figure S6C), JAM (Figure S6D), CCL (Figure S6E), CHEMERIN (Figure S6F), and GAS (Figure S6G) pathways with relatively contributing ligands in aHSCs and receptors in neutrophils, Kupffer cells, and B cells in the control, CCl₄, and MenSC groups (Table S5). The aHSCs expressed ligands in five signaling pathways in the control, CCl₄ and MenSC groups (Figure 6J). The expression of receptors in the five signaling pathways was studied in neutrophils (Figure 6K), Kupffer cells (Figure 6L), and B cells (Figure 6M), respectively. The results showed that the ICAM signaling pathway may be regulated by a combination of the ligand Icam1 in aHSCs (Figure 6J, blue box markup) and its receptors Itgal, Itgam, and Itgb2 in neutrophils (Figure 6K, blue box markup). In the aHSCs population, the expression of Icam1 was significantly higher in the CCl₄ group than in the control and MenSCs groups (Figure 6N). Similarly, in neutrophils, the expression of Itgal (Figure 6O), Itgam (Figure 6P), and Itgb2 (Figure 6Q) was significantly higher in the CCl₄ group than in the control and MenSCs groups. Furthermore, a whole-group expression spectrum analysis revealed that Icam1 was highly expressed in fibroblasts/HSCs, and Itgal, Itgam, and Itgb2 were highly expressed in neutrophils (Figure 6R).

2.7 Potential genes in subpopulations of neutrophils for ameliorating liver fibrosis, as shown by scRNA-seq

The neutrophils were reclustered into five subgroups: Neu-C (neutrophil cluster)-1, Neu-C2, Neu-C3, Neu-C4, and Neu-C5 (Figure 7A). The heatmap with top-5 markers of DEGs in five subgroups of neutrophils was presented (Figure S7A). A cell percentage analysis revealed that the highest percentage of neutrophils was found in Neu-C1 and Neu-C2 (Figure 7B). A dotplot showed that Itgal, Itgam, and Itgb2 were most highly expressed in the Neu-C1 compared with those in Neu-C2–C5 (Figure 7C). We split Neu-C1–C5 in which Itgal (Figure 7D), Itgam (Figure 7E), and Itgb2 (Figure 7F) were expressed in the control, CCl₄, and MenSCs groups. They were not only most strongly expressed in Neu-C1 (Figure 7C) but were also expressed at higher levels in the CCl₄ group than in the control group and tended to decrease after the MenSC treatment. A two-by-two differential significance expression comparison analysis showed that the expression of Itgal was significantly higher in the CCl₄ group than in the control group, and it was significantly decreased after the MenSC treatment (Figure 7G). The expression trends of Itgam (Figure 7H) and Itgb2 (Figure 7I) were consistent with those of Itgal. These results imply that Neu-C1 is the major effector subpopulation of Icam1 in aHSCs ligand and Itgal/Itgb2 and Itgam/Itgb2 act as receptors for subsequent signal transduction.

To explore the specific expression properties of Neu-C1, the top-10 genes with higher expression in Neu-C1 compared with that in other subpopulations were assessed (Figure 7J). The pseudotime analysis found that Neu-C1 was mainly distributed in stages 2 and 3 (Figure S7B,C). Additionally, 8 of the 10 most expressed genes, including Anxa1, Retnlg, Wfdc21, Mmp8, peptidoglycan recognition protein 1 (Pglyrp1), S100a8, S100a9, and lipocalin-2 (Lcn2), clustered together (Figure S7D). Of these molecules, S100a8, S100a9, and Retnlg are well known as canonical neutrophil markers,⁵⁵ and the relative expression profiles of the three genes (Figure S7E–G) were further verified in the present study. Wfdc21 was expressed only in hepatocytes and neutrophils (Figure 7K), suggesting that neutrophil–hepatocyte crosstalk may occur through this factor. Lcn2 was mainly expressed in neutrophils, but cholangiocytes and fibroblasts/HSCs also expressed Lcn2 (Figure 7L). Further studies revealed Anxa1 (Figure S7H) to be nonspecific, whereas Pglyrp1 (Figure 7M) and Mmp8 (Figure 7N) were almost exclusively expressed in neutrophils and could be considered as genes specifically expressed in neutrophils. A two-by-two differential significance expression comparison analysis revealed that the expression of Wfdc21 was significantly lower in the CCl₄ group than in the control group and significantly increased after the MenSC treatment (Figure 7O). The expression trends of Lcn2 (Figure 7P) and Pglyrp1 (Figure 7Q) were consistent with those of Wfdc21. These results indicate that Wfdc21, Lcn2, and Pglyrp1 in Neu-C1 can serve as potential targets or effector molecules with a negative correlation trend. In the Neu-C1 cell population, the expression of Mmp8 was significantly higher in the CCl₄ group than in the control group and remained significantly elevated after the MenSC injection (Figure 7R).

A GO analysis of the subpopulation of aHSCs in the MenSC and CCl₄ groups (Figure 7S) revealed that cell migration was the major mode of action (blue boxes). In addition, a GO analysis of the Neu-C1 subpopulation in the MenSC and CCl₄ groups (Figure 7T) revealed that cell migration and adhesion were the main modes of action (pink boxes). These results suggested aHSC and Neu-C1 showed strong differences in cell migration and adhesion between the MenSC and CCl₄ groups.

2.8 Colocalization for ICAM signaling interaction in HSCs and neutrophils by the immunofluorescence

By focusing on ICAM signaling in HSCs and neutrophils, the Icam1 expression was checked on HSCs (marked as Des) using immunofluorescence (Figure 8A). And LFA-1 (Itgal/Itgb2) expression on neutrophils (S100a8+S100a9) using immunofluorescence (Figure 8B). Mac-1 (Itgam/Itgb2) expression on neutrophils (S100a8+S100a9) using immunofluorescence (Figure 8C). Although the positive rate of Des/Icam1 was reduced by immunofluorescence (Figure 8D, average 28.99% in CCl₄ group and average 24.13% in MenSC group), however, there are no statistical discrepancy. The coexpression of S100a8+S100a9/Itgal/Itgb2 (Figure 8E) and S100a8+S100a9/Itgam/Itgb2 (Figure 8F) were almost 0 in the Control group. The positive rate of S100a8+S100a9/Itgal/Itgb2 in MenSC group (average 0.74%) were significantly decreased in contrast to the CCl₄ group (average 5.78%) (Figure 8E). Similarly, the positive rate of S100a8+S100a9/Itgam/Itgb2 in MenSC group (average 0.62%) were significantly decreased in contrast to the CCl₄ group (average 1.49%) (Figure 8F). MenSCs secrete cytokines that act on aHSCs to reduce the expression of the aHSC ligand, Icam1, which specifically binds to LFA-1 (Itgal/Itgb2) and Mac-1 (Itgam/Itgb2) in neutrophils. Furthermore, the mechanistic diagram is shown (Figure 8G).

4.1 HBV transgenic mice

Male HBV-Tg mice (GenBank No. AF305422.1) aged 7−8 weeks were purchased from Beijing Vitalstar Biotechnology Co., Ltd. (Beijing, China).⁸⁰ The experimental mice, excrements, litter, and related consumables were processed in accordance with the requirements of ABSL-2 (BSL-2). The animal experiments were approved jointly by the Animal Care and Use Committee of the First Affiliated Hospital, Zhejiang University School of Medicine (No. 2022-1082) and the Beijing Vitalstar Biotechnology (No. VST-SY-20221018).

4.2 Source and preparation of MenSCs

Human MenSCs (no. 00612−210415P), provided by the Innovative Precision Medicine (IPM) Group (Hangzhou, China), were acquired from a healthy female donor who provided informed consent before donation, as described in previous studies.^{47, 65, 81} In brief, menstrual blood samples were collected and cultured in α-MEM (Invitrogen, CA, USA) culture media supplemented with 15% fetal bovine serum (Gibco, Australia). After reaching 80−90% confluence, MenSCs were washed, harvested, resuspended, and cryopreserved. To identify MenSC surface molecular markers, fifth-generation MenSCs were digested with trypsin-EDTA (Gibco). The cells were then precipitated and resuspended by adding an appropriate amount of Stain Buffer (BD Biosciences, USA). CD29, CD34, CD45, CD73, CD90, CD105, CD117, and HLA-DR were added to the Eppendorf tubes, according to the dosage recommended by the antibody manufacturers. The detail procedure has been previously described in our previous studies.^{47, 81} IgG1 or IgG2a served as isotype controls; detailed information is presented in Table S6. To identify the three-line differentiation of MenSCs, we used the OriCell^® human bone marrow MSC osteogenic, adipogenic, and chondrogenic induction differentiation kit (Cyagen Biosciences, Guangzhou, China) according to the instructions. The detail three-line differentiation procedure has been previously described.^{81, 82} After approximately four weeks, the MenSCs differentiated into osteoblasts, adipocytes, and chondrocytes, after which they were stained and photographed.

4.3 CCl₄-induced liver fibrosis and MenSC transplantation

HBV-Tg mice do not spontaneously develop liver fibrosis. The control group (N = 8) was injected with corn oil (Macklin Biochemical Technology Co., Ltd., Shanghai, China). To induce liver fibrosis in the mice, the experimental groups were intraperitoneally injected with CCl₄ (Macklin Biochemical Technology Co., Ltd.; dissolved in corn oil at 20% solution, 2 mL/kg body weight) twice weekly for 6 weeks. MenSCs were administered in different doses (MenSC-low group, 1 × 10⁷ cells/kg, N = 8; MenSC-high group, 4 × 10⁷ cells/kg, N = 8) in week 7, and an equal amount of solution (CCl₄ model group, N = 8) was used for the control. Serological indicators were detected at the time point after 1 week of MenSC administration (MenSC 1 W), and all samples were collected at the time point of MenSC 2 W (Figure 1A).

4.4 Cell tracking in CCl₄-induced liver fibrosis

For cell tracking experiments in vivo, DiR (DiIC18(7), Thermo Fisher) was used to label the MenSCs in CCl₄-induced liver fibrosis HBV-Tg mice. Follow the kit instructions, DiR was diluted to the final concentration of 10 µg/mL. MenSCs were diluted to the concentration of 5 × 10⁶ cells/mL. DiR-labeled MenSCs (1 × 10⁶ cells, 200 µL) were then injected into the mouse spleens, with 10 mice in the group being treated. An IVIS small animal optical imaging system (IVIS Lumia series III; PerkinElmer, MA, USA) was used to detect the distribution and intensity of the fluorescence signals before the MenSC injections and 16 h, 24 h, 3 days, 5 days, 7 days, 11 days, and 14 days after the MenSC injections (Figure 1A). On days 3 and 14 after the DiR-MenSC transplantation, the livers, lungs, spleens, kidneys, and hearts were removed for tissue photography.

4.5 Serum liver function and qRT-PCR

The mice were weighed once a week (D1-D56). The sera were sent for liver function testing by the Dian Diagnostics Medical Laboratory (Beijing, China), including testing for the liver function indices ALT, AST, ALP, and ALB and serum HBV-related virological indicators, including HBV DNA, HBsAg, and HBeAg.

Total RNA from the mouse livers was extracted by the manufacturer's instructions for the TRIzol reagent (Invitrogen, MA, USA), using TransScript® One-Step gDNA Removal and cDNA Synthesis SuperMix (TransGen Biotech, Beijing, China). A qRT-PCR analysis was performed using UltraSYBR Mixture (CoWin Biosciences, Beijing, China) with the Applied Biosystems 7500. The detection indicators were HBV product (X-universal), TGF-β1 (Tgfb1), α-SMA (Acta2), and Col-I (Col1a1). GAPDH was used as an internal control. The primer sequences are presented in Table S7.

For coculturing experiment in vitro, recombinant human TGF-β1 (7745-BH; R&D System, USA) stimulated LX-2 cells (human HSC cell line; Wuhan Procell Life Science & Technology Co., Ltd, China) for 24 h to activate HSCs. MenSC and LX-2 were cocultured in six-well transwell (Corning, USA) for three biological duplication. Individual LX-2 cells named as LX-2 group, MenSC and LX-2 in the transwell named as coculture group. The Icam1 expression of LX-2 cells in LX-2 group and coculture group were assessed by qRT-PCR. The β-actin was used as an internal control. The primer sequences are presented in Table S7.

4.6 Histological staining, IHC, and immunofluorescence

Histopathological indicators were assessed at the experimental endpoint (D56/MenSC 2 W). HE staining and SR staining were performed after routine paraffin section preparation (Jkgreen Technology, Beijing, China). The histological characteristics of the liver lesions were used to evaluate the degrees of the liver lesions. NASH activity and liver fibrosis scores were evaluated using the SAF scoring system.⁴⁹ IHC was used to identify α-SMA, Col-I, and TGF-β1, using a Dako EnVision IHC system (Dako, Denmark) as previously described.^{47, 82} Immunofluorescence was performed to examine Des, Icam1, S100a8+S100a9, Itgal, Itgam, and Itgb2 after paraffin section preparation with the instruction by HaoKe Biotechnology Co. Ltd., Hangzhou, China. DAPI (HKI0005; HaoKe Biotechnology) was used for staining nucleus, and the coexpression of Des/Icam1; S100a8+S100a9/Itgal/Itgb2; and S100a8+S100a9/Itgam/Itgb2 were statistically analyzed in control group, CCl₄ group, and MenSC group. Three independent samples were used for each group with five horizons. DAB solution (HaoKe Biotechnology) was used for color development, the duration of which was controlled under a microscope (NIKON-ECLIPSE C1). The detailed antibody information and usage are shown in Table S6.

4.7 RNA-bulk analysis for HBV-Tg mouse livers

For mRNA deep (or RNA-bulk) sequencing, RNA samples were prepared using TruSeq RNA Sample Preparation and performed by Beijing CapitalBio Technology. In brief, mRNA molecules containing poly-A were purified from total RNA using poly-T oligonucleotide-conjugated magnetic beads. The cDNA fragments were purified and the “A” tails were ligated. The cDNA library was constructed (identified by Agilent 2100 Bioanalyzer). The resulting libraries were sequenced on the NovaSeq 6000 platform according to the confidence intervals used for the serial analysis of gene expression data analysis.⁸³

4.8 Single-cell capture and cDNA synthesis for mouse livers

A mouse liver dissociation kit (Miltenyi Biotec, Germany) was used to digest the livers by gentle shaking at 37°C for 30 min. Filtration was performed through a Falcon^® 70-µM cell filter (BD Biosciences), and the samples were centrifuged at 300 g for 6 min to collect single-cell suspensions. To comprehensively define HBV-Tg mouse livers at the single-cell level, we collected HBV-Tg mouse liver NPCs and partial liver parenchymal cells (hepatocytes/cholangiocytes) from the normal control (Con group), CCl₄ fibrosis model (CCl₄ group), and MenSC high-dose therapy (named as MenSC group) at the time point of MenSC 2 W (Figure 1A). Liver NPCs and partial liver parenchymal cells from three groups of nine HBV-Tg mice (three samples per group) were subjected to scRNA-seq using the 10× Genomics platform provided by Beijing CapitalBio Technology (Beijing, China). The scRNA-seq was performed using the Single-Cell 3′ Library and Gel Bead Kit V3.1 (10× Genomics, 1000121) and Chromium Single-Cell G Chip Kit (10× Genomics, 1000120) for cell capture and cDNA synthesis. Cell suspensions were loaded onto the Chromium Single Cell Controller (10× Genomics) to generate single-cell gel beads, according to the manufacturer's instructions. In brief, single cells were suspended in PBS containing 0.04% bovine serum albumin (Sigma, USA). RNA was barcoded by the reverse transcription of an individual GEM in turn released. Reverse transcription was performed using a S1000TM Touch Thermal Cycler (Bio-Rad, CA, USA). The cDNA was amplified, and its quality was evaluated for subsequent analysis using an Agilent 4200 (CapitalBio Technology).

4.9 ScRNA-seq library preparation and data processing

ScRNA-seq libraries were established using Single-Cell 3′ Libraries and Gel Bead kits according to the manufacturer's protocols. Sequencing was performed using an Illumina NovaSeq6000 sequencer at a sequencing depth over 100,000 reads per cell (CapitalBio Technology). Barcode and UMI counting were performed using the CellRanger counting module to generate a feature barcode matrix and identify clusters. Dimensionality reduction was performed using a principal component analysis (PCA) to generate clusters with K-means and graph-based algorithms. Another clustering method used was Seurat 3.0 (R package), where cells with gene counts less than 200 or gene counts ranked in the top 1% were considered abnormal and filtered out. During the analysis or preparation of the cell samples, cells with a mitochondrial gene content exceeding 25% were removed from the dataset. Dimensionality reduction was performed using a PCA, and visualization was achieved by t-SNE and UMAP.⁸⁴ All the genes in the cells of a sample were extracted using gene set enrichment analysis software. The cells were quasi-temporally sorted using DEGs identified with Seurat. The merging of multiple samples was based on the Harmony algorithm in Seurat.⁸⁵ The analyzed single-cell trajectory obtained with the “plot_cell_trajectory” function was constructed using Monocle2 (R package) with the introduction of pseudotimes.

4.10 Enrichment and CellChat analysis

GO enrichment terms and KEGG enrichment pathways were functionally analyzed, classified, and annotated using the ClusterProfiler package in R.⁸⁶ Fisher's exact test was used to calculate the differences in representative GO function sets. A KEGG functional analysis was performed to annotate and categorize the pathways.⁸⁷ To predict cell–cell communication between different cell types, we applied the CellChat R package to the scRNA-seq data. The quantification of signaling pathways is derived from important ligand–receptor pairs between cell types. In the computeCommunProbe function, CellChat determined the effects of the proportion of cells in each cell group on the probability calculations. This was accomplished by separately applying the dot heatmap and chord plots using the netVisual_bubble and netVisual_chord_gene functions.

4.11 Statistical analysis

Statistical analyses were performed using GraphPad Prism 9 software. The data were exhibited as the mean ± standard deviation and analyzed using Student's t-tests, χ² tests, or a one-way analysis of variance. Correlation analyses were performed using Spearman's rank-order correlations. Statistical significance was set at *p < 0.05, **p < 0.01, and ***p < 0.001.