Whole body S100a11 knockout alleviates FFC diet-induced murine MASH
We previously identified S100A11, an alarmin, as a ceramide-dependent extracellular vesicle (EV) cargo secreted from lipotoxic hepatocytes. Our assessment of S100A11 by immunohistochemistry demonstrated a significant increase in FFC-fed wild-type mouse livers (Supplementary Fig. 1A, B), in a hepatocellular pattern compared to CD-fed controls, at 24 weeks of dietary feeding. Correspondingly, the abundance of S100a11 mRNA was significantly increased in FFC-fed WT mouse livers compared to CD-fed mice (Supplementary Fig. 1C). We next examined S100A11 expression in human MASH livers by qPCR. Similar to the FFC-fed mouse model, which is known to have a high fidelity to human MASH, we observed a significant upregulation in the expression of S100A11 compared to healthy controls (Supplementary Fig. 1D). To examine the role S100A11 plays in MASH pathophysiology, WT and S100A11 whole body knockout mice (S100a11) were maintained on the FFC diet or CD for 24 weeks (Fig. 1A). Quantitative PCR analysis of liver tissues confirmed all the S100a11 mice to be knockouts, compared to the WT controls, with undetectable S100a11 transcripts (Supplementary Fig. 2A). The FFC-fed WT mice demonstrated a significant upregulation of S100a11 confirming our previous observations (Supplementary Fig. 2A). We confirmed the expression of S100A11 by immunohistochemistry in FFC-fed WT mice (Supplementary Fig. 2B). We did not detect S100A11 in S100a11 mice; CD-fed mice were not compared due to low basal levels observed in CD-fed WT livers (Supplementary Fig. 1A). At the start of the dietary studies, at 12 weeks of age, body mass was comparable across the groups (Supplementary Fig. 2C). Upon study completion, at 36 weeks of age, we observed a slightly lower body weight in S100a11 mice compared to WT on both diets (Supplementary Fig. 2D), though there was a small but significant increase in food intake in the FFC-fed S100a11 mice (Supplementary Fig. 2E). Food intake was similar between CD-fed WT and S100a11 mice. FFC-fed mice consumed fewer grams of food than CD-fed mice, consistent with the established literature that mice consume fewer grams of high-fat diets due to the higher caloric content of such diets. Liver-to-body weight ratio demonstrated no basal changes in CD-fed mice cohorts (Fig. 1B). FFC diet significantly increased liver-to-body weight ratio, though this response was blunted in S100a11 mice (Fig. 1B). Plasma ALT levels were normal in CD-fed S100a11 and WT controls (Fig. 1C). FFC-fed WT mice had significantly higher ALT levels indicating liver injury, while FFC-fed S100a11 mice displayed comparatively reduced levels (Fig. 1C). No significant differences were observed in bilirubin levels (Supplementary Fig. 3A) among all cohorts. As anticipated, serum bile acid and cholesterol levels were elevated in FFC-fed mice compared to those on CD (Supplementary Fig. 3B, C). Notably, FFC-fed S100a11 mice exhibited significantly lower levels of serum bile acid and cholesterol levels relative to FFC-fed WT controls (Supplementary Fig. 3B, C). Histological analysis of the liver tissues using H&E stained liver sections demonstrated significant steatosis and lobular inflammation in FFC-fed WT mouse livers (Fig. 1D). FFC-fed S100a11 mouse livers displayed reduced steatosis and fewer inflammatory foci (Fig. 1D). The CD-fed mouse livers were histologically normal (Supplementary Fig. 3D). Steatosis and inflammatory foci were graded according to the respective components of the NAFLD activity score, confirming reduced inflammatory foci in FFC-fed S100a11 mouse livers with a trend toward decreased steatosis (Fig. 1E, F). The reduction in steatosis was further assessed by lipid droplet morphometry. Total lipid droplet area and average lipid droplet size were significantly lower in FFC-fed S100a11 mice compared to FFC-fed WT mice (Fig. 1G, H). There was an increase in the number of lipid droplets in FFC-fed S100a11 mice consistent with the H&E observation of greater microvesicular steatosis (Fig. 1I).
Assessment of hepatic mRNA expression levels of inflammatory markers revealed a significant upregulation of pro-inflammatory genes, Ly6c and Mac2, in FFC-fed WT mice compared to CD-fed controls. Importantly, this inflammatory response was markedly attenuated in FFC-fed S100a11 mice (Supplementary Fig. 3E, F). Assessment of liver fibrosis using Picrosirius staining demonstrated no significant collagen fiber accumulation in CD-fed mice across the genotypes (Supplementary Fig. 3G). FFC-fed WT mouse livers displayed significant pericellular and bridging fibrosis while FFC-fed S100a11 mice had lower abundance of collagen fibers, verified by polarized light microscopy-based collagen quantification (Fig. 1J). mRNA analysis of fibrogenic genes in the liver demonstrated a significant increase in the expression of Col1a1 and Timp1 in FFC-fed WT mice compared to those on CD (Supplementary Fig. 3H, I). Notably, the upregulation of Col1a1 was significantly lower in FFC-fed S100a11 mice and Timp1 showed a similar pattern but did not reach significance. Taken together, these data demonstrate that whole-body S100a11 deletion reduces FFC-diet-induced MASH.
Due to the difference in body mass and relative liver mass in FFC-fed WT compared to S100a11 mice, to avoid confounding due to differences in body mass and extrahepatic effects of S100A11, we asked if hepatocyte-specific repression of S100a11 would exhibit MASH attenuation. To address this question, mice were fed CD or FFC diet for 20 weeks and then divided into four groups (Fig. 2A). Hepatotropic AAV8 viral particles encoding S100a11 shRNA or scramble shRNA were administered following 20 weeks of diet by tail vein injections. Mice were maintained on their respective diets for an additional 4 weeks. Assessment of shRNA-mediated knockdown efficacy was performed by qPCR and immunohistochemistry. We observed that FFC-shA11 (AAV8-shA11 injected FFC) mice had significantly lower S100a11 transcripts compared to FFC-scr (AAV8-scramble injected FFC) mice controls (Fig. 2B), and lower expression by immunohistochemistry (Supplementary Fig. 4A). In contrast to the whole-body knockout, both groups of FFC-fed mice gained comparable weight on the diet (Supplementary Fig. 4B). However, the relative liver mass was lower in FFC-shA11 group compared to FFC-scr group with no change in CD-fed mouse groups (Supplementary Fig. 4C). Correspondingly, there was a reduction in ALT elevation, inflammation and hepatic steatosis (Fig. 2C-F) in FFC-shA11 mice compared to FFC-scr mice while repression of S100a11 had no effect on CD fed mice (Supplementary Fig. 3D). Lipid droplet morphometry confirmed a reduction in total lipid droplet area, average droplet size, and an increase in the number of lipid droplets in the FFC-shA11 mouse livers (Fig. 2G-I) as observed in the whole-body knockout mice. Picrosirius red stained liver sections indicated no notable collagen fiber accumulation in the CD-fed mice (Supplementary Fig. 3E). FFC-scr mice cohort displayed significant fibrosis which was remarkably reduced with 4-weeks of S100a11 silencing, as noted in the FFC-shA11 group (Fig. 2J). Polarized light microscopy-based collagen quantification confirmed the histological impression of a reduction in fibrosis. Thus, hepatocyte-specific silencing of S100A11 was sufficient to attenuate MASH, while maintaining comparable body mass in mice.
To understand the mechanism of how silencing of hepatic S100a11 impairs FFC diet-induced steatohepatitis compared to the respective controls, we employed bulk-RNA sequencing. We identified 13,379 protein coding genes wherein 2245 were significantly differentially expressed genes (DEGs) of which 1429 were downregulated and 816 were upregulated (Supplementary Fig. 5A). We observed more than five-fold reduction (Log FC: -2.47) of S100A11 transcripts in the FFC-shA11 livers versus FFC-scr, confirming significant S100a11 gene knockdown in this cohort (Supplementary Fig. 5A). Principal Component Analysis (PCA) confirmed significant clustering of the two independent study groups (FFC-scr and FFC-shA11) (Supplementary Fig. 5B). Ingenuity Pathways Analysis demonstrated significant dysregulation of lipid metabolism and carbohydrate metabolism in the top 10 dysregulated pathways in FFC-shA11 livers (Fig. 3A). Further interrogation of "carbohydrate metabolism" and "lipid metabolism" components, within Ingenuity Pathways Analysis, was performed by comparing the rank order list of component pathways. We noted 128 transcripts in "metabolism of carbohydrate" (p value of 6.79e and activation score of 0.008) and 101 transcripts in "synthesis of fatty acids" (p value of 4.07e-20 and activation score of -2.011) with 19 transcripts that were common to both pathways (Supplementary Table 1).
Of the 19 identified transcripts, we considered HK2 further, as glucose fixation by hexokinase-mediated phosphorylation is the crucial step in carbohydrate metabolism and links carbohydrate to lipid metabolism. We observed that the expression of Hk2 was reduced in the shA11 livers in the RNA-seq data (Fig. 2B). We confirmed this by qPCR, which showed the abundance of Hk2 transcripts was low and unchanged in CD-fed mouse livers, whereas expression was significantly decreased in FFC-shA11 mouse livers compared to FFC-scr livers (Fig. 3C). Similar to the mouse data, HK2 expression was low in normal human livers (only detected in 2 of 6 normal control samples tested) and upregulated in MASH livers, though not uniformly so (detected in 4 of 9 samples tested) (Supplementary Fig. 5C). The mRNA expression of glucokinase or hexokinase 4 (Hk4), the predominant form in hepatocytes, was upregulated with FFC diet in control mice whereas hexokinase 3 (Hk3) did not significantly change (Supplementary Fig. 5D). There were no significant differences in expression between FFC-scr and -shA11 livers. We did not detect hexokinase 1 (Hk1) and hexokinase domain-containing 1 (Hkdc1) transcripts in the liver (data not shown). Among other candidate genes, we observed a similar direction in Fasn, which encodes fatty acid synthase, and peroxisome proliferator-activated receptor alpha (Pparα), which were also attenuated in FFC-shA11 (Fig. 3C). The mRNA expression of phosphoenolpyruvate carboxykinase 1 (Pck1), diacylglycerol O-acyltransferase 2 (Dgat2), and acyl-CoA oxidase 1 (Acox1) demonstrated no significant differences in FFC-shA11 mouse livers with respect to the control groups (Fig. 3C).
Next, we asked if the upregulation of Hk2 transcripts correlated with an increase in its protein expression. Western blot of HK2 in mouse livers demonstrated low basal expression in CD-fed liver (Fig. 3D). We tested this with increasing protein lysate concentrations (5 μg, 15 μg, 30 μg, and 50 μg) from CD-fed WT livers, which confirmed expression of HK2 protein in all samples tested rather than non-specific detection (Supplementary Fig. 5E). Some variability was observed among the CD-fed samples; one mouse exhibited detectable HK2 expression even at 5 μg protein input, suggesting individual variation in basal HK2 levels (Supplementary Fig. 5E). FFC-diet feeding significantly increased HK2 protein levels and liver-specific S100a11 repression lowered FFC-diet induced HK2 protein upregulation (Fig. 3D). In vitro treatment of a murine hepatocyte cell line (IMH) with palmitate, a lipotoxic stressor significantly increased the expression of HK2 protein levels in palmitate-stimulated cells compared to the controls (Supplementary Fig. 6A). De novo lipogenesis (DNL) assay in primary mouse hepatocytes demonstrated an increase in lipid synthesis in palmitate-stimulated WT hepatocytes. (Supplementary Fig. 6B). In contrast, hepatocytes isolated from S100a11 mice did not demonstrate an increase in DNL following PA treatment (Supplementary Fig. 6C). To confirm these findings and to determine whether S100A11 regulates HK2 expression differentially in response to carbohydrate or lipid metabolism, we treated WT and S100A11 Hu1545 human hepatocyte cell line for 8 h with either vehicle or 400 µM palmitate under standard or high glucose conditions. Western blot analysis demonstrated that HK2 expression was upregulated in response to palmitate treatment in WT cells and not in response to glucose, suggesting a glucose-independent effect of palmitate on HK2 induction (Supplementary Fig. 6D). In S100A11 cells, we did not observe an increase in HK2 expression following PA treatment. Rather, the levels of HK2 were lower than standard glucose conditions. We next measured accumulation of lipid droplets in WT and S100A11 cells similarly treated with palmitate under standard or high glucose conditions (Supplementary Fig. 6E, F). We found that palmitate was able to increase lipid content, whereas high glucose did not change lipid area. Furthermore, high glucose did not modify the palmitate response. We also observed that palmitate-induced steatosis was attenuated in S100A11 cells. Altogether, our observations suggest that S100A11 mediates lipotoxicity-induced expression of HK2, which may drive de novo lipogenesis.
Having noted a significant upregulation of HK2 in FFC-fed livers and repression of HK2 in FFC-shA11 livers, we asked if a gain-of-function approach would help understand the relationship between S100A11 and HK2. First, we confirmed that HK2 expression was lower in FFC-fed S100a11 livers compared to WT (Supplementary Fig. 7A). Next, we injected 1 × 10 genome copies of AAV8-Hk2 viral particles (Supplementary Fig. 7B) via tail vein into 12-week-old WT and S100a11 male mice. The mice were maintained on CD for 4 weeks after virus administration (Fig. 4A). During tissue collection, the livers of Hk2-overexpressing CD-fed mice appeared pale and enlarged compared to normal livers of scramble controls. In keeping with these morphological observations, the relative liver mass was increased significantly in the Hk2 overexpression group compared to scramble controls in both WT and S100a11 mice (Fig. 4B). Western blotting of liver tissues confirmed overexpression of exogenous HK2 in all the AAV8-Hk2 injected mice compared to the scramble controls (Fig. 4C). Interestingly, we also noted increased abundance of DNL-regulating master enzymes: total acetyl-CoA carboxylase (ACC) and ATP citrate lyase (ACLY) in Hk2-overexpressing WT and S100a11 mouse livers compared to the scramble mice (Fig. 4C). Steatosis was observed in H&E-stained liver sections from WT and S100a11 Hk2 overexpressing mice (Fig. 4D, E). Quantification of steatosis and of the BODIPY signal demonstrated comparable steatosis in WT and S100a11 Hk2 overexpressing mice (Fig. 4F, G). Lastly, biochemical measurement of liver triglyceride content confirmed that HK2 overexpression was sufficient to drive steatosis in WT and S100a11 mouse livers (Fig. 4H). Altogether, these observations demonstrate that even under basal (chow diet) conditions, the overexpression of HK2 is sufficient to drive steatosis in WT and S100a11 mouse livers.
In complementary experiments, we employed BNBZ, a pharmacological inhibitor of HK2. Palmitate-treated Hu1545 cells demonstrated significant lipid-droplet accumulation (Fig. 5A, B). In contrast, in cells treated with palmitate in the presence of BNBZ there was no increase in lipid droplet accumulation. Thus, HK2 mediates palmitate-induced steatosis in hepatocytes.
Though bulk liver mRNA and protein expression mostly represent hepatocytes, given the significantly greater hepatocellular mass than other cell types in the liver, we asked whether S100A11 and HK2 may be upregulated in other liver cell types. We used bone marrow-derived macrophages (BMDM) and the human hepatic stellate cell line LX-2 to address this question. Neither S100a11 nor Hk2 was upregulated in BMDM or LX2 cells (Supplementary Fig. 8A-D). We have previously demonstrated an increase in RAGE (Ager) positive recruited macrophages in MASH. We did not detect Ager in hepatocytes (data not shown). As Ager is also expressed in hepatic stellate cells, we examined its expression in this cell type; Ager was not upregulated in palmitate-treated LX-2 cells (Supplementary Fig. 8E). Altogether, our data implicate a hepatocellular role for palmitate-induced upregulation of S100A11 and consequently HK2 in promoting MASH.