Long‐term hypercaloric diet exacerbates metabolic liver disease in PNPLA3 I148M animals

Nonalcoholic fatty liver disease (NAFLD) is a major health burden associated with the metabolic syndrome leading to liver fibrosis, cirrhosis and ultimately liver cancer. In humans, the PNPLA3 I148M polymorphism of the phospholipase patatin‐like phospholipid domain containing protein 3 (PNPLA3) has a well‐documented impact on metabolic liver disease. In this study, we used a mouse model mimicking the human PNPLA3 I148M polymorphism in a long‐term high fat diet (HFD) experiment to better define its role for NAFLD progression.


| INTRODUC TI ON
The metabolic syndrome linked to non-alcoholic fatty liver disease (NAFLD) is becoming increasingly relevant due to its widespread occurrence in all populations, particularly among those with a sedentary lifestyle. In addition to lifestyle and nutrition, genetic risk factors play a major role for the progression of NAFLD, of which the patatinlike phospholipid domain containing protein 3 (PNPLA3) I148M polymorphism is of major relevance. In Caucasian populations, this polymorphism has an allele frequency of 0.28. Homozygote allele carriers have a 3.3-fold increased incidence of nonalcoholic steatohepatitis (NASH) and liver fibrosis in a European cohort. 1,2 PNPLA3 I148M mutation has been shown to be associated with increased hepatic lipids accumulation. 3 Li et al. found increased triglycerides (TGs) synthesis in mice expressing human PNPLA3 I148M. 4 Conversely, the hepatic de novo lipogenesis (DNL) was lower in a human PNPLA3 I148M cohort. 5 Moreover, the sterol regulatory element binding protein 1c (SREBP-1c) which is the key enzyme for DNL was decreased. 5 Additionally, it was reported that the PNPLA3 I148M variant protein showed less lipase activity especially the lysophosphatidic acid acyltransferase (LPAAT) effect. 6,7 Wang et al. also reported that PNPLA3 I148M together with comparative gene identification-58 (CGI-58) would inhibit adipose triglyceride lipase on lipid droplet. 8 Therefore, it is more reasonable to assume that the PNPLA3 I148M mutation leads to a loss of function, thereby exacerbating disease progression.
Fibrogenesis, as a hall marker for NAFLD, could be accelerated by the PNPLA3 I148M mutation directly or indirectly. In wild type mice, PNPLA3 protein has a protective role, can promote retinoids release from hepatic stellate cells (HSCs), which may decrease extracellular matrix formation and remodelling. However, when the PNPLA3 I148M mutation is present, this function was lost. 9 On the other hand, under liver injury, HSCs can release inflammatory cytokines and trigger stronger inflammation, which in turn promotes fibrosis. Primary human HSCs with the PNPLA3 I148M polymorphism were shown to release higher levels of proinflammatory cytokines such as chemokine (C-C motif) ligand 5 (CCL5). 10 During this process, monocyte derived macrophages and liver resident Kupffer cells play a key role in releasing proinflammatory mediators and activate myofibroblasts. 11 However, it is poorly understood how PNPLA3 I148M influences the function of macrophages and Kupffer cells. Furthermore, related genes or pathways activated specifically by the PNPLA3 I148M genotype are not fully understood.
Nevertheless, obesity is essential to amplify the impact of the PNPLA3 I148M genotype on hepatic triglyceride content (HTGC) and liver damage. In a lean population with a body mass index (BMI) of lower than 25 kg/m 2 , wild-type(I148I) or I148M homozygous individuals have similar levels of HTGC and AST. In contrast, PNPLA3 I148M homozygous carriers show significantly higher HTGC and AST levels compared to the wild type, when an increased BMI is present. 12 In addition, I148M homozygous individuals are more frequent among obese individuals who also have a higher likelihood to develop the metabolic syndrome. 13 The current study aims to define the impact of the PNPLA3 I148M genotype on NAFLD especially in obese and older animals after long-term hypercaloric feeding. We here define that PNPLA3 I148M-specific changes in gene signature and microbiota composition is associated with stronger monocyte-derived inflammation and liver disease progression in NAFLD.

| Statistics
Statistical analysis was performed in GraphPad Prism 9. Unpaired two-tailed Student's t-test was employed for comparisons of two groups. Two-way ANOVA and Tukey's multi-comparisons test was used in grouped diagrams. Data were considered significant between experimental groups as: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Analysis procedure for microbiome and RNA-Seq can be found in the supplementary information.

| Long term hypercaloric diet exacerbates features of NAFLD in Pnpla3 148M/M mice
To study the influence of the PNPLA3 I148M genotype on NALFD disease progression Pnpla3 148M/M knock-in mice, in which hypercaloric feeding is known to cause the enrichment of PNPLA3 on the membrane of lipid droplets, 15 were fed with HFD for 24 and 52 weeks.
When compared to Pnpla3 WT mice of the same genetic background, Pnpla3 148M/M animals gained significantly more body weight over time, which was first evident after 17 weeks of HFD ( Figure 1A).
The highest amount of weight gain was accounted for liver weight ( Figure 1B) and subcutaneous adipose tissue mass, according to body composition study upon sacrifice after 52 weeks HFD ( Figure 1C).
When compared to WT controls, Pnpla3 148M/M livers were substantially heavier and macroscopically showed steatosis after 52 weeks ( Figure 1D). Liver histology revealed abundant hepatic lipid deposition. After 24 weeks HFD, the pattern of ballooning degeneration was comparable between both genotypes. However, Pnpla3 148M/M livers showed more parenchymal and non-parenchymal cells including immune cells at 52 weeks ( Figures 1E and S1A). In Pnpla3 148M/M mice, serum liver injury markers (AST, ALT) were significantly higher compared to Pnpla3 WT mice after 24 and 52 weeks ( Figure 1F,G).
With a selected range of body weight, Pnpla3 148M/M mice have higher mean AST values compared to WT mice but did not reach statistically significant differences after 16 and 24 weeks of HFD ( Figure S3F). Pnpla3 148M/M mice fed with HFD showed significantly more liver lipids than WT mice ( Figure 1H,I).
In summary, the data demonstrate that long term HFD feeding resulted in a significantly more pronounced NASH-related phenotype in Pnpla3 148M/M compared to Pnpla3 WT mice.

| HFD triggers a stronger inflammatory response in Pnpla3 148M/M livers
Inflammation is a driver of liver injury and consecutively triggers fibrosis progression. As a result, we investigated the presence of relevant immune cell populations in the liver tissue of our animals.
In comparison to Pnpla3 WT controls, Pnpla3 148M/M mice had more CD45 + inflammatory cells in their livers after HFD feeding for 52 weeks. In contrast, no significant differences were found after 24 weeks of HFD (Figure 2A,B). Thus, we concentrated on the 52 weeks timepoint for more detailed analysis. Flow cytometry of immune cells from liver tissue showed increased numbers of CD45 + cells in Pnpla3 148M/M mice ( Figure 2C). After 52 weeks, HFD Pnpla3 148M/M livers compared to Pnpla3 WT controls had more CD8 + and NK T-lymphocytes and NK cells among those cells ( Figure 2D-F). Furthermore, monocyte-derived macrophages were significantly higher in Pnpla3 148M/M compared to WT livers ( Figure 2G).
CD11b + Ly6G + neutrophils were increased, in line with these findings ( Figure 2H). The PNPLA3 I148M mutation triggers STAT3 activation both in vivo and in vitro. 17,18 Western Blot analysis revealed that P-STAT3 expression was increased only after 52 weeks of HFD feeding in Pnpla3 148M/M livers compared to Pnpla3 WT livers ( Figure 2I). Livers of NEMO Δhepa /JNK Δhepa mice 19 were used as a positive control.
Hence, these results showed that HFD in Pnpla3 148M/M livers leads to a stronger inflammatory response compared to WT controls.

| Distinct proliferative pattern in Pnpla3 148M/M livers after long term hypercaloric diet
Increased liver cell proliferation is characteristic of NAFLD and is required for the initiation and progression of hepatic fibrosis in both human and animal models of the disease. 20,21 Since Pnpla3 148M/M livers had increased cell numbers per view field in H&E staining after 52 weeks HFD feeding, we stained liver tissue for Ki-67 to study changes in hepatic cell proliferation. In the livers of Pnpla3 148M/M mice, we found a significant increase in Ki-67 positive cells after 24 and 52 weeks of HFD ( Figure 3A  F I G U R E 1 Long term hypercaloric diet exacerbates features of NAFLD in Pnpla3 148M/M mice. (A) Pnpla3 WT and Pnpla3 148M/M mice were treated with diet experiments at around 8-10 weeks old and were fed with HFD or chow diet for up to 52 weeks. Pnpla3 148M/M mice receiving HFD showed significant differences in body weight from 17 weeks. Each value represents the mean value of one group and shown as every 2 weeks. Two-way ANOVA and Tukey's multiple comparisons test were performed. (n = 5-10/group) *p ≤ 0.05, **p ≤ 0.01. (B) The liver weight is shown before feeding (0 weeks) and after 24 and 52 weeks of feeding. Pnpla3 148M/M mice fed with HFD had more liver weight at 52 weeks. Data is represented as mean ± SD. Two-way ANOVA and Tukey's multiple comparisons test were performed. *p ≤ 0.05, **p ≤ 0.01 (n = 5-10/ group). (C) Adipose tissue weight from both PNPLA3 genotypes fed with chow diet and HFD for 24 and 52 weeks. Pnpla3 148M/M mice with HFD gained more weight of sWAT at 52 weeks. (p ≤ 0.0001) Values are presented as mean ± SD. Two-way ANOVA and Tukey's multiple comparisons test were performed.  When the animals were separated by body weight, we found that CK19 positive area in the livers was linked to increased body weight in Pnpla3 148M/M mice, but not in WT mice fed an HFD for 52 weeks ( Figure 3H). However, it should be considered that on average Pnpla3 148M/M mice gained more weight under HFD than WT animals.
We next investigated whether enhanced ductular prolifera-

| Susceptibility to liver fibrosis is increased in Pnpla3 148M/M mice after long term hypercaloric diet
In humans, the PNPLA3 I148M polymorphism has been linked to more severe liver fibrosis. 2 Here we studied the impact of the PNPLA3 I148M genotype on fibrosis in a long term HFD mouse model. Representative Western Blot analysis for alpha SMA of entire liver lysates from 52 weeks fed animals further strengthened these findings ( Figure 4E,F). Since PNPLA3 protein expression was associated with HSCs activation, we analysed PNPLA3 protein expression.
There was a trend that Pnpla3 148M/M mouse livers expressed more PNPLA3 protein compared to Pnpla3 WT mice after 52 weeks HFD.
However, these differences did not reach statistical significance ( Figure S3G,H). As we fasted the animals for 3 h before sacrificing, and PNPLA3 protein expression is regulated via food intake, 23 this might influence PNPLA3 protein expression.
In conclusion, HFD triggered a more pronounced fibrotic reaction in Pnpla3 148M/M compared to Pnpla3 WT control animals.

| Exacerbated NAFLD in Pnpla3 148M/M mice show a distinct intestinal microbial pattern, associated with obesity and high fat diet
The intestine microbiome is becoming increasingly recognized as a driver in various diseases, and changes have been connected to disease progression in obesity and NAFLD. 24 To explore the effect of host genotype on microbiome composition, faecal samples col-   showed enriched Gracilibacteraceae ( Figure 5D) and from this fam-  (Figures 5F and S1D). In addition, in the comparison of diet in each genetic group, the families Erysipelotrichaceae, Enterobacteriaceae, Deferribacteraceae and Coriobacteriaceae were altered ( Figure S1E).

| Transcriptional analysis revealed genes involved in NAFLD progression and specific genes altered by PNPLA3 I148M
To compare HFD-induced expression changes in Pnpla3 WT and Compared to the influence of HFD, the number of differential genes between Pnpla3 WT and Pnpla3 148M/M in either the HFD ( Figure 6D) or the chow diet group ( Figure 6E) was much lower. Overrepresentation analysis of the differential genes in Figure 6B-E demonstrated that genes upregulated due to HFD were overrepresented in inflammation and immune response associated GO terms ( Figure S2A-D).
In contrast, the downregulated genes were dominated by metabolic processes, such as cholesterol and steroid biosynthetic processes and further differentiated liver functions ( Figure S2E-H).
Overrepresentation analysis of genes differentially expressed between Pnpla3 WT and Pnpla3 148M/M did not result in significant GO terms after adjustment for multiple testing.
To identify possible differences in global gene expression changes between mutant and wild-type mice, the fold-changes of individual genes between chow and HFD were plotted for Pnpla3 WT and Pnpla3 148M/M mice ( Figure 6F). The scatter plot of individual genes showed a strong correlation of HFD-induced expression changes in both mouse strains. Validated with qPCR, we found a similar strong induction by HFD for example for Ccr1, Ccr5 and Ccr2, genes that may trigger an inflammatory response ( Figure S3A-C) and for Timp1 and Vcan that are involved in fibrogenesis ( Figure S3D,E), corresponding to the NAFLD phenotypes in humans. 25,26 However, for several genes the log 2 -fold change differed by more than 0.5 in WT and mutant mice. The dots for these genes appeared outside the blue lines in Figure 6F. For a set of 171 genes, indicated as 'group 1' in Figure 6F, HFD caused a stronger increase in Pnpla3 148M/M than in Pnpla3 WT livers (list of genes in Tables S10 and S11). These genes were overrepresented predominantly in immune associated GO groups ( Figure 6G). Among the genes more upregulated in the livers of mutant than WT mice after 52 weeks of HFD were Cracc and Eat2 ( Figure 7A,B). It was reported that CRACC can inhibit NK-and T-cell function in the absence of EAT2. However, with the participation of EAT2, CRACC can positively regulate NK cells and T cells, which promotes inflammation. 27 Ephb2 expressed in HSCs is a crucial gene that regulates Ccl2 and Ccr2 expression. 28,29 Ephb2 was significantly increased in Pnpla3 148M/M livers both at 24 and 52 weeks ( Figure 7C).
Osteopontin encoded by Spp1 and expressed in macrophages, is a biomarker for liver damage and fibrosis. 30 Here we found that Spp1 expression was higher in Pnpla3 148M/M mice at 52 weeks ( Figure 7D).
The proinflammatory chemokine Ccl2 was increased in line with this finding ( Figure 7H). Osteopontin staining confirmed mRNA analysis ( Figure 7J,K).
Taken together, the PNPLA3 I148M genotype was associated with genes involved in inflammation and fibrogenesis and may contribute to stronger NAFLD-dependent disease progression.

| DISCUSS ION
In the present study, we used a knock-in mouse model for PNPLA3 to mimic the most common genetic risk factor for disease progression of metabolic liver injury the PNPLA3 I148M genotype. This model helps to better understand findings in humans, 32 but also adds essential knowledge to previous findings in the PNPLA3 mouse model. 15 In animals with the I148M substitution we found increased liver damage, fibrosis, alterations in intestinal microbiota and a cluster of genes associated with stronger HFD-induced expression in PNPLA3 I148M compared with WT mice. These observations were only found after long term HFD feeding.
Changes in intestinal microbiota composition has been found in different human cohorts with liver diseases including NASH. These findings reflect a dysregulation within the gut-liver axis and earlier results including our own recent data suggest that this finding is associated with increased bacterial translocation triggering a stronger inflammatory milieu in the liver. 24 In our study, we saw a decrease in microbial diversity under HFD consistent with prior findings. These effects were mainly diet-dependent rather than Altered BAs concentrations in the gut have been linked to initiation and progression of NAFLD. 36 We found an increase in total bile acids in faeces of Pnpla3 148M/M compared with mice Pnpla3 WT mice, supporting prior NAFLD studies. 36 Serum cholesterol levels were increased after 52 weeks HFD in both genotypes. Additionally, the crucial bile acid synthesis enzyme CYP7A1 was upregulated in those liver. In contrast NTCP, allowing efficient reabsorption and recycling of bile acids, was decreased. Hence BAs metabolism in the liver was not different between Pnpla3 WT and Pnpla3 148M/M mice after 52 weeks of HFD. As we found a 12% difference in gut microbiota linked to the PNPLA3 I148M genotype, it is likely that these differences are of relevance to explain the differences in faecal BAs concentration.
Increased cell proliferation following liver injury is a key event that drives progression of NAFLD. 21 We observed an increase in proliferation of liver cells at 24  Hence our finding suggest that higher liver injury was only found after long term HFD feeding of Pnpla3 148M/M animals reflecting the situation in humans where only patients with high BMI and the PNPLA3 I148M genotype suffer from more progressive metabolic liver disease. 12 In accordance with the human situation, we only observed first signs of more progressive liver disease after 24 weeks.
However, the impact on fibrosis progression was only evident after 52 weeks of HFD feeding. This cascade of events agrees with aging in humans showing fibrosis progression only after longer periods of 10-15 years in patients with NASH and not NAFLD. Therefore, our data in the Pnpla3 148M/M mouse model reflect key aspects of metabolic liver disease in humans. Earlier results using these mice were limited as they used shorter feeding intervals not leading to sufficient obesity. 4,38 In HFD-fed Pnpla3 148M/M mice, not only liver weight, but especially subcutaneous adipose tissue contributed significantly to increased body weight. Hence in Pnpla3 148M/M mice compared to WT controls with increased disease progression the liver-fat axis seems to be stronger activated potentially contributing to the overall phenotype in these animals. In our experiments we found weight gain mainly in subcutaneous adipose tissue. There is evidence in humans that visceral adipose tissue significantly contributes to metabolic syndrome-related diseases. 39 However, in a recent study in NAFLD patients, a link between liver function measured by ALT and adipose tissue mass was described, and here both visceral and subcutaneous adipose tissue contributed to the overall phenotype. 40 Therefore, there is presently a certain lack of data to which increased subcutaneous adipose tissue contributes to the overall liver phenotype. To better define the molecular mechanism leading to more advanced liver disease progression in Pnpla3 148M/M livers following long term HFD feeding we included RNA seq analysis. Significant expression differences were obtained between HFD and chow fed animals.
Upregulated genes were associated with inflammation and changes in the immune response. Downregulated genes were linked with metabolism and other differentiated functions of the mature liver.
This response pattern is known from previous studies after animals were challenged with hypercaloric diets rich in fat and calories 16,41 but also for toxic liver injury. 42 As expected, large sets of genes were regulated similarly in response to HFD in Pnpla3 148M/M and Pnpla3 WT livers. Interestingly, a specific set of genes associated with immune regulation responded stronger to HFD in Pnpla3 148M/M compared to wild-type livers.
In the natural course of NAFLD, NASH is an inflammatory state in the liver that contributes to tissue damage and fibrosis formation. Consistent with these findings, the pattern of genes specifically found in the Pnpla3 148M/M livers link increased inflammation with fibrosis after HFD. In Pnpla3 148M/M animals, higher Ephb2 was expressed by HSCs and likely promotes higher levels of Ccl2 expression. In cooperation with Ccr2, more macrophages may be recruited into the liver which in turn regulate the expression of Ephb2. 28,29 Besides, earlier studies showed that Pilrα reduces macrophage infiltration, while in Pnpla3 148M/M livers its expression was downregulated indicating that this regulation might be involved in controlling a stronger influx of macrophages into these livers. 31 Eventually, macrophages expressed more Osteopontin which is a biomarker for liver

Our study shows increased immune cell infiltration in
damage. An earlier study provided evidence that Osteopontin could induce fibrosis by triggering ductular response. 43 Taken together, the PNPLA3 I148M genotype supports expression of a gene cluster involved in monocyte-/Kupffer cell-derived inflammation, fibrosis and an increased ductular response in obese animals.

| Conclusion
Our findings provide long-term data on NAFLD progression in a Pnpla3 148M/M mouse model which mimics human liver disease.

CO N FLI C T O F I NTER E S T S TATEM ENT
There are no potential conflicts (financial, professional, or personal) that are relevant to the manuscript. All authors had access to the study data and had reviewed and approved the final manuscript.