Aging triggers steatohepatitis by impeding fatty acid oxidation and degradation.
The models of aging were established using 24-month-old mice. No significant changes of liver function and histology were detected in aged liver, except for a reduction of the relative liver weight (Supplementary Fig. 1A-1C). SA-β-gal, a hallmark of aging, was used to identify the cellular senescence in aged liver. Compared to the younger control (8-week-old mice), aged liver exhibited more senescent cells in liver staining (Fig. 1A). Staining of Oil red, Sirius red and F4/80 showed deteriorated fat deposition, fibrosis and inflammation in aged liver (Fig. 1B), indicating aging triggers steatohepatitis. The TG content was found to be higher in the aged liver than in the control (Fig. 1C). Next, pathways-associated with fibrosis and inflammation were confirmed to be activated in aged LSECs by analyses of heatmaps and GSEA according to the RNA sequencing data (Fig. 1D and 1E). Besides, genes-related with fatty acid oxidation or degradation were inactivated in aging LSEC by analyses of the heatmap (Fig. 1F) and GSEA (Fig. 1G), which further proved that aging increased abnormal fat accumulation in liver.
Aging disrupts sinusoid homeostasis and reshapes liver endothelial zonation by downregulating pericentral endothelial markers.
According to the enlarged image of Fig. 1A, SA-β-gal+ cells seemed to distribute along the sinusoids, providing the possibility that senescent cells may probably be liver nonparenchymal cells (NPCs). Thus, hepatocytes and liver NPCs were isolated and measured by qRT-PCR, which confirmed that LSECs enriched the expressions of several senescent markers, such as IL-1β, IL-1α, Gata4, P16, P53, CCl2, P21 and P27 (Fig. 2A). Since LSECs were found to be the origin of senescent cells of aged liver, we then investigated the change of liver sinusoid during the process of aging. Interestingly, the aged liver showed unique filiform alteration under scanning electronic microscopy (SEM) (Fig. 1B). Besides, VE-cadherin and Lyve1, which represent the markers of endothelial cells and LSECs respectively, were downregulated in aged liver, according to the immunofluorescent (IF) staining (Fig. 1C). Fenestrae, usually reflecting the differentiation of LSECs, were found to be reduced in aged liver under SEM investigation (Fig. 1C), which suggested aging accelerating capillarization of liver sinusoid. Aiming to investigate the changes of endothelial gene profiles in aged liver, LSECs were isolated from aged and young control mice respectively. As previously demonstrated32, liver endothelial cells (ECs) can be classified into distinct populations. According to GSEA analyses based on the RNA-sequencing data, the top 50 marker genes for liver ECs, arterial ECs, capillary ECs and venous ECs were significantly downregulated in aged mice (Fig. 2D), implying the homeostasis of liver vasculature was disrupted by aging. Next, we compared the expressions of pericentral and periportal liver endothelial markers between aged and control LSECs. The landmark genes concerning pericentral and periportal liver ECs were identified using the sequencing data provided by Halpern KB et al33. Interestingly, pericentral but not periportal EC markers were downregulated with significance by aging (Fig. 2E). Previously, we proved that C-kit, as the same as Wnt9b, Wnt2a and Rspo320,31,34, was a marker gene of pericentral liver ECs. The mRNA level of C-kit was found to be lower in aged LSECs than in control (Fig. 2F). IF staining of C-kit again proved that aging downregulated hepatic C-kit expression (Fig. 2G). These data collectively indicated that irregular pericentral liver EC-derived C-kit may be responsible for aging-associated liver metabolic disorders.
NASH inactivates pericentral liver endothelial markers and C-kit.
Next, we tried to clarify whether inactivation of pericentral endothelial C-kit accounted for the accelerated hepatic lipogenesis. Thus, diet-induced NASH models were introduced. The pericentral marker genes were extensively downregulated in NASH models as well. However, the periportal endothelial markers showed insignificant alterations during NASH progression (Supplementary Fig. 2A). Consistently, public databases proved that the expression of C-kit was downregulated in different NASH models (Supplementary Fig. 2B). Additionally, we evaluated the mRNA and protein levels of C-kit in LSECs isolated from MCD diet-induced NASH mice. Western-blot (WB) and qRT-PCR analyses confirmed that pericentral endothelium-derived C-kit was inactivated at different timepoint in MCD-treated LSECs (Supplementary Fig. 2C and 2D). IF staining also exhibited the downregulation of hepatic C-kit in MCD or CDAA diet-induced models (Supplementary Fig. 2E and 2F). Collectively, these findings suggested that inactivation of pericentral endothelial C-kit may be the reason for the pathogenesis of aging and diet-induced NASH.
Disruption of endothelial C-kit shifts endothelial and metabolic gene profiles in LSECs and induces LSEC senescence.
To precisely illustrate the role of pericentral liver endothelium-derived C-kit in NASH, we established endothelial C-kit knockout (KO) mice by crossing Cdh5-CreERT mice with C-kitfloxp mice as shown in Supplementary Fig. 3A. The WB and qRT-PCR analyses confirmed the successful depletion of C-kit in LSECs (Supplementary Fig. 3B and 3C). Although, the morphological analyses of quiescent liver collected from endothelial C-kit KO mice showed no difference in general look, SEM, H&E, Ki67 IHC, F4/80 IF and Laminin IF staining (Supplementary Fig. 3D), the gene profiles of liver endothelium were extensively changed following endothelial C-kit deletion. By reanalyzing previously published sequencing data32, we found that most of the top 50 liver EC marker genes were downregulated in C-kit KO LSECs (Fig. 3A), as were the top 50 metabolism marker genes (Fig. 3B), suggesting endothelium-derived C-kit play a key role in maintaining LSEC homeostasis. Next, the expression profiles of the top 50 marker genes for arterial ECs, venous ECs and capillary ECs were analyzed in LSECs with C-kit disruption. Interestingly, we found that capillary ECs, the majority of which are LSECs, lost their characteristic genotypes (Fig. 3A). However, the expression of arterial and venous EC marker genes fluctuated in C-kit KO LSECs (Fig. 3A). The GSEA analyses confirmed the above findings (Fig. 3C and 3D).
Next, we observed cellular senescence after C-kit deletion. By SA-β-gal staining, accumulated senescent cells could be found in C-kit KO liver (Fig. 3E). The heatmap and GSEA analyses showed that aging and senescence-associated pathways were activated following C-kit disruption (Fig. 3F and 3G). Collectively, these results indicated that C-kit deficiency-induced senescent LSECs underwent extensive gene profile alterations, which potentially affected liver metabolism in an EC-mediated manner.
Endothelial C-kit depletion aggravates hepatic steatosis by enhancing fatty acid biosynthesis.
Since endothelial C-kit KO affected EC metabolic gene expression, we then speculated the impact of C-kit deletion on hepatic lipid metabolism. Metabolic disordered mice were established with MCD diet for 6 weeks and CDAA diet for 10 weeks respectively (Fig. 4A). According to H&E and Oil Red staining, endothelial C-kit disruption exhibited noticeably aggravated hepatic steatosis in both MCD (Fig. 4B and 4C) and CDAA diet-treated mice (Fig. 4D and 4E). Hepatic TC and TG were then evaluated in CDAA diet-triggered NASH mice. Increased hepatic lipid content could be detected in mice with C-kit deletion (Fig. 4F). Besides, the serum glucose was found to be even higher in C-kit KO mice compared to the control (Fig. 4G). All these findings proved the hepatic lipid metabolism was disturbed following C-kit disruption. Mechanistically, molecules-associated with lipid synthesis, oxidation and lipolysis were measured by qRT-PCR in CDAA diet-triggered NASH mice. As shown in Fig. 4H, the expression of FASN was upregulated in C-kit KO LSECs, however PPARα, Acadvl and Mgll were found to be downregulated, indicating depletion of C-kit facilitated lipid synthesis, and attenuated lipid oxidation and lipolysis.
Next, we evaluated the top enrichment of the hallmark genes in C-kit-deficient LSECs, pathways-related with glycolysis, adipogenesis, oxidative phosphorylation were activated by loss of C-kit, and the homeostasis of cholesterol and fatty acid metabolism were also affected (Fig. 4I). Moreover, the sequencing data presented by the heatmap and GSEA analyses confirmed that the process of hepatic lipid biosynthesis was strikingly enhanced by loss of endothelial C-kit (Fig. 4K and 4J).
Endothelial C-kit blockage provokes NASH-associated fibrosis and liver inflammatory response.
After finding C-kit KO aggravated hepatic steatosis in diet-induced NASH mice, we inspected whether NASH-associated liver fibrosis and inflammation were affected. In MCD diet-induced NASH mice, endothelial C-kit KO increased hepatic macrophages, basement membrane, liver fibrosis and impeded hepatocyte proliferation, which were identified by F4/80, Laminin, Sirius Red and Ki67 staining respectively (Fig. 5A and 5C). In consistent with the findings observed above, mice fed with CDAA-diet also exhibited aggravated fibrosis, inflammation, and incapable hepatocyte proliferation (Fig. 5B and 5D). Molecules-associated with fibrosis and inflammation were then evaluated by qRT-PCR. The mRNA levels of SM22, αSMA, PDGFβ, TGFβ, IL-1β, TNFα and IL-6 were all upregulated in mice fed with CDAA-diet after C-kit KO (Fig. 5E and 5G). Likewise, in MCD diet-treated group, TIMP-1, Collagen1, PDGFβ, IL-6 and IL-1β were also found to be increased in LSECs with C-kit deletion (Fig. 5F and 5H), indicating loss of endothelium-derived C-kit exacerbated liver fibrosis and inflammation. The sequencing data proved that hepatic inflammatory responses were largely promoted by C-kit disruption, as shown in Fig. 5I and 5J.
Endothelium-derived C-kit inhibits CXCR4-mediated macrophage and neutrophil response.
Since chemokine-mediated infiltration of inflammatory cells usually takes part in the process of hepatic pathologic challenges including NASH, we then investigated the alterations of chemokine pathways in the isolated LSECs with different genotypes. Expectedly, the heatmap combining with GSEA analyses proved the extensive activation of chemokine signaling following C-kit disruption (Fig. 6A and 6B). Strikingly, CXCR4, one of the canonical chemokine receptors, was confirmed to be upregulated significantly in the mRNA level in both quiescent and MCD-treated mice following C-kit blockage (Fig. 6C). In the protein level, abrogation of endothelial C-kit increased CXCR4 expression as well (Fig. 6D). We then identified if the increasing expression of hepatic CXCR4 was due to enhanced secretion of its ligand-SDF-1 in C-kit-deficient LSECs. As expected, IF staining confirmed the increased production of SDF-1 in C-kit KO LSECs (Fig. 6E). The ELISA assays performed in the supernatants of cultured LSECs proved that, the secreted levels of SDF-1 were significantly higher in C-kit-deficient group than in the control (Fig. 6F).
As chemokine signaling modulates the recruitment of immune cells such as macrophages and neutrophils following injury, we thus explored pathways-associated with macrophage and neutrophil biology. After C-kit depletion, GSEA analyses illustrated that the activation, migration, and chemotaxis of macrophages and neutrophils were greatly enhanced (Fig. 6G and Supplementary 4A), proving that endothelial C-kit inhibited chemokine-mediated recruitment of macrophages and neutrophils.
Blocking CXCR4 abrogates macrophage recruitment in C-kit-deficient NASH mice.
To clarify whether deletion of C-kit promoted CXCR4-mediated macrophage recruitment, AMD3100, an inhibitor of CXCR4 (Supplementary Fig. 4B), was applicated for seven doses in total before analysis, during the process of diet induction (Fig. 6H). FACS investigation proved that the augment of CD11b+, F4/80+, CD11b+F4/80+ and CD11b−F4/80+ liver NPCs caused by C-kit disruption were largely neutralized by AMD3100 in mice fed with CDAA diet (Fig. 6I), which suggested that the macrophage recruitment was modulated by C-kit-CXCR4 axis. To determine the crosstalk between LSECs and macrophages, primary LSECs and hepatic macrophages were isolated and co-cultured in vitro (Fig. 6J). The transwell assay demonstrated that wildtype macrophages migrated to C-kit KO LSECs even more than to the control, the change of which was abolished while AMD3100 was additionally administrated (Fig. 6K and 6L). Collectively, these data indicated that endothelium-derived C-kit impeded LSEC-macrophage crosstalk by inhibiting CXCR4/SDF-1 signaling.
Loss of endothelial C-kit exacerbates diet-induced NASH in a CXCR4-dependent manner.
Subsequently, we determined the role of CXCR4 in C-kit-regulated NASH progression. The strategy of AMD3100 administration was shown in Fig. 6H as well. H&E, Oil Red, Sirius Red, MPO and F4/80 staining showed that blocking CXCR4/SDF-1 signaling successfully recovered the aggravated hepatic steatosis, fibrosis and inflammation caused by C-kit deficiency in NASH mice (Fig. 7A-7D). Besides, elevated hepatic TC and TG content levels were counteracted by AMD3100 in NASH mice with C-kit deletion (Fig. 7E). Next, we identified the changes of genes-associated with fatty acid biosynthesis, the expressions of ACC1 and FASN were both decreased while AMD3100 was applicated (Fig. 7F). Additionally, senescence was observed in liver of CDAA diet-induced NASH mice. The enhancement of cellular senescence caused by C-kit KO was abolished while AMD3100 was administrated (Supplementary Fig. 4C). Taken together, these data proved that CXCR4 was required for C-kit-mediated senescence and NASH development.
Splenic implantation of C-kit + LSECs alleviates diet or aging-induced NASH by restoring the metabolic homeostasis of pericentral LSECs.
Since loss of endothelial C-kit accelerated NASH progression, whether gain of C-kit-dominated endothelium could improve NASH was then explored. C-kit+ LSECs were sorted by anti-CD117 magnetic beads. As shown by the FACS, the purity of C-kit+ (CD117) cells after isolation reached 97.77% (Supplementary Fig. 5A). C-kit+ LSECs were infused into the spleen of MCD diet-treated mice, one week before analysis (Fig. 8B), while C-kit− LSECs were used as the control. Few minutes after the infusion, fluorescence-labeled C-kit+ LSECs could only be detected in liver by observation of spectrum, but not in spleen, kidney, heart, and lung (Supplementary Fig. 5B), implying that the implanted cells predominantly recruited to the liver. Microscopic analysis of the liver also confirmed the existence of spleen-originated cells which were labeled by GFP (Supplementary Fig. 5C). Interestingly, implantation of C-kit+ LSECs not only increased the expression of hepatic C-kit, but also attenuated diet-induced hepatic steatosis, fibrogenesis, macrophage accumulation, and incapable hepatocyte proliferation (Fig. 8A). Mechanistically, expressions of some chemokine receptors, including CXCR4, were found to be lower in C-kit+ LSEC group than in C-kit− one (Fig. 8C), which was determined by the RNA sequencing. Besides, pathways-associated with macrophage and neutrophil recruitment were seemed to be inactivated in C-kit+ LSEC group (Fig. 8D). Moreover, GSEA analyses revealed that LSECs of C-kit+ group enriched pericentral marker genes, but not periportal ones (Fig. 8E). The top 50 liver-specific EC marker genes and metabolic marker genes which downregulated in C-kit deficient LSECs (Fig. 3A-3D) were upregulated in C-kit+ LSEC group (Fig. 8E), Besides, inflammation-associated genes were found to be inactivated in LSECs of C-kit+ group determined by the heatmap (Fig. 8E and Supplementary Fig. 5D).
Finally, C-kit+ LSECs were implanted into aged mice as well. According to the RNA sequencing, senescence and aging-associated genes were extensively inactivated in C-kit+ LSEC group (Fig. 8F and 8G), implying that C-kit+ LSECs possessed the potential to treat aging. Expectedly, C-kit+ LSEC infusion successfully diminished cellular senescence, increased liver ECs (Fig. 8H), and reduced hepatic fat deposition (Fig. 8I). These data collectively suggested that gain of C-kit-dominated liver ECs rectified diet or aging-induced metabolic disorders by restoring the homeostasis of pericentral liver endothelium.