The Potential of the FSP1cre-Pparb/d−/− Mouse Model for Studying Juvenile NAFLD

Non-alcoholic fatty liver disease (NAFLD) can progress from steatosis to non-alcoholic steatohepatitis (NASH) characterized by liver inflammation, possibly leading to cirrhosis and hepatocellular carcinoma (HCC). Mice with impaired macrophage activation, when fed a high-fat diet, develop severe NASH. Evidence is mounting that Kupffer cells are implicated. However, it is unknown whether the resident CD68+ or bone marrow-derived CD11b+ Kupffer cells are involved. Characterization of the FSP1cre-Pparb/d−/− mouse liver revealed that FSP1 is expressed in CD11b+ Kupffer cells. Although these cells only constitute a minute fraction of the liver cell population, Pparb/d deletion in these cells led to remarkable hepatic phenotypic changes. We report that a higher lipid content was present in postnatal day 2 (P2) FSP1cre-Pparb/d−/− livers, which diminished after weaning. Quantification of total lipids and triglycerides revealed that P2 and week 4 of age FSP1cre-Pparb/d−/− livers have higher levels of both. qPCR analysis also showed upregulation of genes involved in fatty acid β-oxidation, and fatty acid and triglyceride synthesis pathways. This result is further supported by western blot analysis of proteins in these pathways. Hence, we propose that FSP1cre-Pparb/d−/− mice, which accumulate lipids in their liver in early life, may represent a useful animal model to study juvenile NAFLD.


Introduction
Non-alcoholic fatty liver disease (NAFLD) is an emerging epidemic disease posing a major health issue worldwide. It is the most common cause of liver disease in humans. NAFLD has high prevalence ranging from 11% to 46% in developed countries [1][2][3]. NAFLD is an inclusive term that is used to cover a range of liver pathologies. It begins with steatosis, where lipid accumulation in the hepatocytes is caused by an impaired triglyceride synthesis, a reduced fatty acid β-oxidation, or both. Patients with steatosis can progress to non-alcoholic steatohepatitis (NASH), a more severe form of NAFLD. NASH involves hepatocellular injury and liver inflammation [4]. The majority of individuals with NAFLD are asymptomatic. In some patients, NAFLD is diagnosed after a blood test, which shows elevated levels of liver enzymes (alanine transaminase, aspartate aminotransferase, and alkaline phosphatase) or through an ultrasound scan. Individuals who are either diabetic, obese, or suffer from metabolic syndrome will be suspected to have NAFLD [5].
co-staining of FSP1 and CD11b identifying CD11b + cells as FSP1 expressing cells (Figure 1a). There was no FSP1 staining in CD68 + cells (Figure 1b). Hence, this observation suggests that FSP1 is expressed in resident liver macrophages, particularly in CD11b + CD68 − resident liver macrophages. 1b). Hence, this observation suggests that FSP1 is expressed in resident liver macrophages, particularly in CD11b + CD68 − resident liver macrophages.
Other non-parenchymal hepatic cells were also investigated for co-expression of specific markers and FSP1. Glial fibrillary acidic protein (GFAP) is an intermediate filament protein expressed in numerous central nervous system cells, including astrocytes. GFAP is also found to be a specific biomarker of a hepatic stellate cell (HSC). We observed no co-staining of FSP1 and GFAP in HSCs (Figure 1c), suggesting that HSCs do not express FSP1.
The LSECs were identified using a specific biomarker, CD146. CD146 is a transmembrane glycoprotein which is a member of the IgG superfamily of cell adhesion molecules. We found that there was no FSP1 co-staining with CD146-positive LSECs (Figure 1d), indicating that LSECs do not express FSP1.  Other non-parenchymal hepatic cells were also investigated for co-expression of specific markers and FSP1. Glial fibrillary acidic protein (GFAP) is an intermediate filament protein expressed in numerous central nervous system cells, including astrocytes. GFAP is also found to be a specific biomarker of a hepatic stellate cell (HSC). We observed no co-staining of FSP1 and GFAP in HSCs (Figure 1c), suggesting that HSCs do not express FSP1.
The LSECs were identified using a specific biomarker, CD146. CD146 is a transmembrane glycoprotein which is a member of the IgG superfamily of cell adhesion molecules. We found that there was no FSP1 co-staining with CD146-positive LSECs (Figure 1d), indicating that LSECs do not express FSP1.
We validated the co-expression of CD11b and FSP1 in non-parenchymal liver cells using flow cytometry ( Figure 2a). All leucocytes express CD45, which is a receptor-linked protein tyrosine  as mean ± s.e.m; n = 6 biological replicates per group). Two-tailed Mann-Whitney test of the data did not find significant differences between the FSP1cre-Pparb/d −/− and Pparb/d fl/fl genotypes. (b) Schematic diagram showing the relative position of PCR genotyping primers for the Pparb/d lox/lox (400 bp band) and the deleted Pparb/d exon 4 alleles (490 bp band). FSP1-cre genotyping: 100 bp band indicates FSP1-cre. 300 bp band is from the endogenous Fsp1 gene. In both gels, the rightmost lane depicts the molecular weight 100 bp ladder. (c) Fsp1 and Pparb/d gene expression in isolated FSP1 + CD11b + cells from P2 Pparb/d fl/fl and FSP1cre-Pparb/d −/− livers. BD, below detection. Two-tailed Mann-Whitney test of the data shown as mean ± s.e.m, n = 5 biological replicates per group. ** P < 0.01; FSP1cre-Pparb/d −/− vs. Pparb/d fl/fl controls. (d) Immunoblot analysis of PPARβ/δ in P2 Pparb/d fl/fl and FSP1cre-Pparb/d −/− whole livers. Representative blots from 2 livers for each genotype showing lower PPARβ/δ expression in FSP1cre-Pparb/d −/− livers are shown. The mean PPARβ/δ expression level in Pparb/d fl/fl livers was arbitrarily given a value of 1. This level is more than 3 times lower in FSP1cre-Pparb/d −/− livers. The full-size western blot is shown in Figure SM2a. U2AF65 is used as loading and transfer control.
Remarkably, the level of PPARβ/δ mRNA was below detection in isolated FSP1 + CD11b + cells from P2 FSP1cre-Pparb/d −/− livers ( Figure 2c). As a consequence of the invalidation of Pparb/d, the level of PPARβ/δ protein in whole liver extracts was lower in FSP1cre-Pparb/d −/− livers compared to Pparb/d fl/fl livers ( Figure 2d). Due to the very low abundance of FSP1 + CD11b + cells, it was technically not possible to determine the PPARβ/δ protein level in the lysates from these cells by using western blot. To define the potential role of PPARβ/δ in FSP1-expressing non-parenchymal hepatic cells on liver structure and function, we analyzed the livers of pups and young animals at time points of important metabolic changes (suckling-P2; weaning-week 4; transition to adulthood-week 8). We examined the liver from FSP1cre-Pparb/d −/− mice at postnatal day 2 (P2), week 4, and week 8 of age after hematoxylin and eosin staining. Livers from Pparb/d fl/fl mice at the same age served as controls for comparison. FSP1cre-Pparb/d −/− livers appeared less compacted when compared with their Pparb/d fl/fl counterparts because of extended sinusoids. Furthermore, the hepatocytes in FSP1cre-Pparb/d −/− mice presented a more pronounced microvesicular structure, likely indicative of more intracellular lipid droplets (Figure 3a,b). At week 8 of age, the compactness of the hepatic cells in Pparb/d fl/fl and FSP1cre-Pparb/d −/− livers was similar and some microvesicles remained in the livers of both FSP1cre-Pparb/d −/− and Pparb/d fl/fl mice (Figure 3c). There was no significant difference in the weight of the livers of both genotypes (Figure 3d).
Because the hematoxylin and eosin staining of liver sections in FSP1cre-Pparb/d −/− mice revealed a more pronounced microvesicular like structure than in the liver of Pparb/d fl/fl mice, we performed Oil red O staining to detect intracellular lipid droplets. As expected under the milk diet, which is high in fat content, lipid droplets were present in P2 hepatocytes of both Pparb/d fl/fl and FSP1cre-Pparb/d −/− mice. However, the droplets were more abundant and larger in the FSP1cre-Pparb/d −/− livers (Figure 4a). At week 4 in both Pparb/d fl/fl and FSP1cre-Pparb/d −/− livers, the lipid droplets were more abundant in hepatocytes adjacent to the hepatic sinusoids, and the number of droplets was higher in hepatocytes of the FSP1cre-Pparb/d −/− mice. Furthermore, at week 4, the droplets are much smaller than at P2 (Figure 4b).   hepatocytes of the FSP1cre-Pparb/d −/− mice. Furthermore, at week 4, the droplets are much smaller 1 than at P2 (Figure 4b).
Remarkably, at week 8 of age, the number of the lipid droplets appeared similar in the livers of     Remarkably, at week 8 of age, the number of the lipid droplets appeared similar in the livers of both Pparb/d fl/fl and FSP1cre-Pparb/d −/− mice ( Figure 4c). Again, the lipid droplets were preferentially localized within hepatocytes that were adjacent to the hepatic sinusoids. Together, our results indicate that the observed phenotype (higher number and larger size of lipid droplets) only occurs in livers of the FSP1cre-Pparb/d −/− during early life (at P2 and to a lesser extent at week 4). The observed lipid droplet phenotype prompted us to quantify it. To this end, we measured lipid levels in both genotypes at the 3 age points. We hypothesized that the FSP1cre-Pparb/d −/− livers would have higher levels of total lipids, especially at P2 and week 4. The liver lipids were extracted using the Cell Biolabs lipid extraction kit and measured using the Cell Biolabs total lipids quantification kit. The readings were normalized to the protein concentration in the respective samples (µg lipids per µg protein).

Upregulation of Genes Involved in Fatty Acid β-Oxidation, Fatty Acid Synthesis and Triglyceride Synthesis in FSP1cre-Pparb/d −/− Liver at P2 and Week 4
Based on the important differences in triglyceride levels between the livers of FSP1cre-Pparb/d −/− and Pparb/d fl/fl mice, we examined the expression of genes involved in fatty acid β-oxidation and triglyceride synthesis. As PPARs (PPARα, PPARβ/δ, and PPARγ) are lipid-activated transcription factors that play an important role in energy metabolism, we measured their expression levels in the Next, we measured the levels of triglycerides in these livers. Stored triglycerides serve as an energy source and play a key role in metabolism. Excess carbohydrates and fats that are not immediately used by the body are converted to triglycerides. Hence, we hypothesized that the level of triglycerides might be different in the livers of FSP1cre-Pparb/d −/− and Pparb/d fl/fl mice. The quantification of triglycerides was done with the Cell Biolabs triglyceride quantification kit. The readings were normalized to the total protein in the respective samples (mg triglycerides to g protein).
In addition to changes in triglyceride levels, we hypothesized that cholesterol levels might be changed as well. Hence, we quantified total cholesterol, free cholesterol, and cholesteryl ester using the Cell Biolabs total cholesterol kit. The readings were normalized to the weight of the respective tissue samples. Surprisingly, there was no difference in total cholesterol content between livers of FSP1cre-Pparb/d Based on the important differences in triglyceride levels between the livers of FSP1cre-Pparb/d −/− and Pparb/d fl/fl mice, we examined the expression of genes involved in fatty acid β-oxidation and triglyceride synthesis. As PPARs (PPARα, PPARβ/δ, and PPARγ) are lipid-activated transcription factors that play an important role in energy metabolism, we measured their expression levels in the liver of both genotypes. We were interested in determining whether changes in PPAR isotype expression could be leading to the increased lipid accumulation and elevated triglyceride levels in the livers of P2 and week 4 FSP1cre-Pparb/d −/− mice.
Remarkably, the expression of PPARα was upregulated by 31. As mentioned above, there was no difference in total cholesterol, free cholesterol, and cholesteryl ester in the livers of Pparb/d fl/fl and FSP1cre-Pparb/d −/− mice at P2, week 4, and week 8. Hence, we conducted a gene expression analysis on cholesterol synthesis genes to consolidate (or not consolidate) these results. Srebp2, Hmgcs1, cytochrome P450 family 51 subfamily A member 1 (Cyp51), and Dhcr7 expressions were found to be similar in both liver genotypes at all 3-time points ( Figure  6e). Hence, these results further suggest that the accumulation of lipid droplets and a higher level of total lipids in P2 and week 4 FSP1cre-Pparb/d −/− livers was not the result of increased cholesterol synthesis. We reported above that there was a higher number of lipid droplets and total lipid levels in the liver of P2 and week 4 FSP1cre-Pparb/d −/− mice. First, this prompted us to examine the expression level of genes involved in fatty acid β-oxidation. Although there was higher lipid accumulation in P2 FSP1cre-Pparb/d −/− livers compared to Pparb/d fl/fl livers, fatty acid β-oxidation genes such as Cd36, acyl-CoA oxidase 1 (Acox1), Hadha, and Acaa2 were all upregulated in P2 FSP1cre-Pparb/d −/− livers ( Figure 6b). CD36, which functions as a fatty acid transporter, had its gene expression upregulated by 57.2%. Acox1 is the first enzyme in the fatty acid β-oxidation pathway, which catalyzes the desaturation of acyl-CoAs to 2-trans-enoyl-CoAs. Acox1 expression was increased by 22.5%. Hadha, which catalyzes the mitochondrial beta-oxidation of long-chain fatty acids to β-ketoacyl-CoA, had its gene expression upregulated by 31.7%. The expression of the gene coding for Acaa2, which catalyzes the final step of the mitochondrial fatty acid beta-oxidation, was increased by 95.9%. No significant change in the expression of these fatty acid β-oxidation genes was observed in week 4 and week 8 livers of Pparb/d fl/fl and FSP1cre-Pparb/d −/− mice.
Accumulation of lipid can result from an increase in fatty acid synthesis. Therefore, we examined genes expressed in the fatty acid synthesis pathway, such as Acat1, Fas, and Scd1 (Figure 6c). The most significantly upregulated gene was Acat1. ACAT1 is a mitochondria-localized enzyme that catalyzes the reversible formation of acetoacetyl-CoA from two molecules of acetyl-CoA to initiate fatty acid synthesis. We observed that the expression of Acat1 was increased by 187.7%, while that of Fas was increased by 52.8%. SCD1 is an enzyme involved primarily in oleic acid synthesis. Unexpectedly, Scd1 expression was downregulated by about 35.3% in P2 FSP1cre-Pparb/d −/− livers. In contrast, there was no significant change in the expression of these fatty acid synthesis genes between the two genotypes in week 4 and week 8 livers. Expression of Srebp1c, which codes for a transcription factor that regulates genes required for glucose metabolism and fatty acid and lipid production, remained similar at the 3 age points in both genotypes.
The expression of genes involved in the triglyceride synthesis pathway or lipogenesis, such as Gpd1, 1-acylglycerol-3-phosphate O-acyltransferase 3 (Agpat3), Lpin3, and Dgat1, was also examined (Figure 6d). In parallel with the upregulation of genes in the fatty acid synthesis pathway in P2 FSP1cre-Pparb/d −/− livers, the expression of these four triglyceride synthesis genes was also increased. GPD1, along with GPD2, plays a critical role in lipid metabolism by transferring glycerol-3-phosphate from the cytosol to mitochondria. This process takes place after catalyzing the conversion of NADH to NAD+ and glycerol-3-phosphate. Gpd1 expression was increased by 35.3% in P2 FSP1cre-Pparb/d −/− compared to Pparb/d fl/fl livers. AGPAT3 is a mitochondrial acyltransferase that converts lysophosphatidic acid into phosphatidic acid in the triglyceride synthesis pathway. The expression level of Agpat3 was upregulated by 57.5% in P2 FSP1cre-Pparb/d −/− livers. LPIN3 catalyzes the dephosphorylation of phosphatidic acid to form diacylglycerol. Diacylglycerol is the precursor of both triglycerides and phospholipids. A staggering 248.3% increase in Lpin3 expression was observed in P2 FSP1cre-Pparb/d −/− livers. In the final step of triglyceride synthesis, DGAT1 catalyzes the conversion of diacylglycerol and fatty acyl CoA to triacylglycerol. Dgat1 expression was elevated by 84.9% in P2 FSP1cre-Pparb/d −/− livers. Lastly, Plin5, which encodes an intracellular lipid droplet coating protein promoting triglyceride stores and preventing them from lipolytic degradation, was increased by 52.1% in P2 FSP1cre-Pparb/d −/− livers. In contrast to P2 livers, there were no significant changes in the expression of all these genes between both genotypes in week 4 and 8 livers. Collectively, these qPCR results demonstrated that both fatty acid and triglyceride synthesis pathways were upregulated in the livers of FSP1cre-Pparb/d −/− mice, which is consistent with the lipid droplets accumulation and higher total lipid levels in these livers. It is noteworthy that in parallel fatty acid β-oxidation genes were upregulated as well.
As mentioned above, there was no difference in total cholesterol, free cholesterol, and cholesteryl ester in the livers of Pparb/d fl/fl and FSP1cre-Pparb/d −/− mice at P2, week 4, and week 8. Hence, we conducted a gene expression analysis on cholesterol synthesis genes to consolidate (or not consolidate) these results. Srebp2, Hmgcs1, cytochrome P450 family 51 subfamily A member 1 (Cyp51), and Dhcr7 expressions were found to be similar in both liver genotypes at all 3-time points (Figure 6e). Hence, these results further suggest that the accumulation of lipid droplets and a higher level of total lipids in P2 and week 4 FSP1cre-Pparb/d −/− livers was not the result of increased cholesterol synthesis.
2.5. Increase in Hadha, Phospho-ACLY, Phospho-ACC and GPD2 in the Liver of P2 FSP1cre-Pparb/d −/− Mice P2 livers of FSP1cre-Pparb/d −/− mice exhibited more lipid droplets and higher levels of total lipids and triglycerides. In parallel, the expression of genes involved in lipid and triglyceride synthesis was also increased, as well as genes involved in fatty acid oxidation. We complemented these results by analyzing the expression and phosphorylation level of some lipid metabolism proteins in P2 livers of FSP1cre-Pparb/d −/− compared to control Pparb/d fl/fl mice.
First, we observed that the trifunctional HADHA protein involved in fatty acid β-oxidation was upregulated by 30.0% in the liver of FSP1cre-Pparb/d −/− mice (Figure 7a). Hence, this result is in line with the upregulated HADHA mRNA level in the liver of FSP1cre-Pparb/d −/− mice (Figure 6b).

Discussion
We demonstrated through double immunofluorescence staining that CD11b + CD68 − liver macrophages, but not in other non-parenchymal hepatic cells, express FSP1. We further showed, using flow cytometry, that 14.5% of the bone marrow-derived cells express FSP1, of which 57% of these cells are CD11b + cells. Hence, our results support an earlier observation of FSP1-positive cells in a subpopulation of inflammatory Kupffer cells/macrophages in injured livers [31].
The liver of FSP1cre-Pparb/d −/− mice is characterized by a change in hepatic lipid metabolism in 2-day-old pups and young mice. Histological analyses revealed that the microvesicular phenotype in Second, we tested ATP citrate lyase (ACLY) which cleaves citrate into oxaloacetate and acetyl-CoA, the latter serving as a substrate for cholesterol and fatty acid synthesis. Phosphorylation of ACLY at Ser455 inhibits the homotopic allosteric regulation through citrate, which promotes the hydrolysis process to increase the yield of acetyl-CoA. Hence, ACLY has a central role in triglyceride synthesis. Both ACLY and phospho-ACLY were significantly upregulated in the liver of FSP1cre-Pparb/d −/− mice by 40.0% and 43.7%, respectively (Figure 7b). Hence, we speculate that there is more acetyl-CoA in the liver of FSP1cre-Pparb/d −/− mice available for fatty acid synthesis.
Third, we measured acetyl-CoA carboxylase (ACC), which plays a crucial role in the fatty acid synthesis pathway. In the biogenesis of long-chain fatty acids, the carboxylation process of acetyl-CoA to malonyl-CoA is catalyzed by ACC. Phosphorylation at Ser79 in ACC by AMP-activated protein kinase (AMPK) reduces the enzymatic activity of ACC. The protein expression level of ACC was similar in the livers of Pparb/d fl/fl and FSP1cre-Pparb/d −/− mice (Figure 7c). However, phospho-ACC was significantly downregulated by 27.4% in the liver of FSP1cre-Pparb/d −/− mice, which should result in a more active enzyme in these mice compared to the control mice. These results, together with those from ACLY, suggest an upregulation of fatty acid synthesis in the liver of FSP1cre-Pparb/δ −/− mice.
Last, we examined GPD2, which along with GPD1, plays a critical role in triglyceride synthesis by transferring glycerol-3-phosphate from the cytosol to mitochondria. This process takes place after catalyzing the conversion of NADH to NAD+ and glycerol-3-phosphate. We observed an upregulation of GPD2 expression by 68.6% in the liver of FSP1cre-Pparb/d −/− mice ( Figure 7d). Hence, this result further suggests an upregulation of triglyceride synthesis in the liver of FSP1cre-Pparb/d −/− mice. In contrast to ACLY, ACC, and GPD2, only the mRNA level of HADHA was not upregulated in P2 FSP1cre-Pparb/d −/− livers compared with Pparb/d fl/fl livers (Figure 6b, Figure SM1). This observation suggests a post-transcriptional regulation of these three genes in the liver of FSP1cre-Pparb/d −/− mice, which results in different protein levels and/or activities between the two genotypes, as reported above.

Discussion
We demonstrated through double immunofluorescence staining that CD11b + CD68 − liver macrophages, but not in other non-parenchymal hepatic cells, express FSP1. We further showed, using flow cytometry, that 14.5% of the bone marrow-derived cells express FSP1, of which 57% of these cells are CD11b + cells. Hence, our results support an earlier observation of FSP1-positive cells in a subpopulation of inflammatory Kupffer cells/macrophages in injured livers [31].
The liver of FSP1cre-Pparb/d −/− mice is characterized by a change in hepatic lipid metabolism in 2-day-old pups and young mice. Histological analyses revealed that the microvesicular phenotype in the liver of FSP1cre-Pparb/d −/− was due to more and larger intracellular lipid droplets. This phenotype appeared in the livers of P2 and week 4 FSP1cre-Pparb/d −/− mice compared to control livers. Hence, we propose that the alteration of lipid metabolism in the liver of FSP1cre-Pparb/d −/− mice occurs in the early days of life, especially before and around weaning. As these FSP1cre-Pparb/d −/− pups aged, the lipid accumulation phenotype disappeared. Interestingly, one of our previous studies revealed that lipid accumulation in the liver of P2 Ppara knock out mice [36] and hepatocyte-specific deletion of Ppara resulted in impairment of fatty acid catabolism which caused hepatic lipid accumulation [37]. In the present study, this lipid accumulation phenotype was seen in the P2 and week 4 FSP1cre-Pparb/d −/− livers, although the underlying cause was different than Ppara deletion. This phenotype gradually disappeared in week 8.
Alteration in lipid levels can be due to changes in lipid biosynthesis or catabolism. Between P2 and week 4, there was a 70% reduction in triglyceride levels in both genotypes. However, the level of triglycerides at week 8 of age was similar to that of week 4. Therefore, we speculated that the significant difference in total lipids observed in the liver of FSP1cre-Pparb/d −/− mice at P2 and week 4 was probably associated with triglyceride uptake from milk fat. This would explain the significantly higher triglyceride level observed at P2 and week 4 of age and the similar triglyceride level in the livers of both genotypes at week 8 when the young mice were now fed on normal chow diet after weaning at P21. SREBP1c and SREBP2 are activators of the complete program of fatty acid and cholesterol synthesis in the liver [38]. We did not find any alteration in the cholesterol synthesis pathway in FSP1cre-Pparb/d −/− livers, which would impact total lipid levels at P2 and week 4. We noticed that the total cholesterol and free cholesterol levels were highest at P2, while cholesterol ester levels were highest at week 4 in both analyzed genotypes. Our qPCR results confirmed that there was no alteration in the expression of the cholesterol synthesis genes, which differed from the observation from the P2 Ppara knockout liver, which presented higher cholesterol ester levels [36]. The glucocorticoid receptor-PPARα axis that prepares neonates for milk lipid catabolism was impaired in the Ppara knockout liver. The deletion of Ppara drastically reduces fatty acid oxidation, leading to fat accumulation. Our results suggest that the mechanism of hepatic lipid accumulation in the FSP1cre-Pparb/d −/− liver is different, as already mentioned above.
The higher total lipids level in P2 and week 4 FSP1cre-Pparb/d −/− livers was attributed to a higher level of triglycerides, most likely due to an enhanced triglyceride synthesis. This alteration in triglyceride synthesis disappeared by week 8 after birth, which was in concordance with no difference in the levels of total lipids and triglycerides, as well as the number and size of lipid droplets. Remarkably, our qPCR results revealed that Pparb/d expression was lower in P2 FSP1cre-Pparb/d −/− liver. A lower Pparb/d expression in the liver is usually detected under the condition of higher Ppara expression and β-oxidation activity [39]. There was indeed an upregulation of genes involved in fatty acid β-oxidation, but also in fatty acid and triglyceride synthesis in FSP1cre-Pparb/d −/− livers. These results were supported by the protein expression analysis that confirmed the upregulation of these processes in P2 FSP1cre-Pparb/d −/− liver. However, the changes in gene expression were not seen anymore after weaning. We propose that the fat-rich milk diet of the pups promotes the lipid metabolism phenotype observed in FSP1cre-Pparb/d −/− liver. This phenotype would progressively disappear after the transition to a carbohydrate-rich diet.
The upregulation of the fatty acid β-oxidation pathway was somewhat unexpected in a situation of lipid accumulation. Fatty acid β-oxidation results in the production of acyl-CoA and then citrate via the TCA cycle ( Figure 8). In the FSP1cre-Pparb/d −/− liver, there was an increase in phospho-ACLY, which would enhance citrate hydrolysis to increase the level of acetyl-CoA. In parallel, phospho-ACC was decreased in the P2 FSP1cre-Pparb/d −/− liver, enhancing the carboxylation of acetyl-CoA to malonyl-CoA to fuel fatty acid and triglyceride synthesis. Therefore, we hypothesized that the upregulation of fatty acid β-oxidation would participate in a higher level of triglycerides at P2 in the FSP1cre-Pparb/d −/− liver. The TCA cycle activity in the P2 FSP1cre-Pparb/d −/− liver remains to be determined. Taken together, we propose that FSP1cre-Pparb/d −/− mice, whose liver accumulates fat during the early days of life, maybe a useful animal model to study juvenile NAFLD. NAFLD is currently the most prevalent form of chronic liver disease, affecting up to 20% of the general pediatric population [40].
A limitation of this study is the lack of unveiling of the mechanism by which Pparb/d deletion in FSP1-expressing non-parenchymal hepatic cells leads to fat accumulation in hepatocytes before weaning. Ongoing work is addressing this question.

Chemicals and Antibodies
Unless otherwise stated, all chemicals and antibodies were purchased from Sigma-Aldrich

Animals
Colonies of Pparb/d fl/fl and FSP1cre-Pparb/d −/− mice on a (C57BL/6J×Sv129)×BALB/c-Tg mixed background were provided by Associate Professor Tan Nguan Soon's laboratory (NTU, Singapore) and bred in our animal facility. They were generated in Tan's laboratory using Pparb/d fl/fl mice provided by Prof. Walter Wahli [41]. Mice were maintained with close monitoring in a specific pathogen-free facility at 25 • C on a 12-h/12-h light-dark cycle. They were kept in microisolator cages and fed on standard rodent chow diet and given water ad libitum. All experimental protocols were approved by the Nanyang Technological University-Institutional Animal Care and Use Committee (ARF-CSB-E0004, 10 October 2013 and ARF-CSB-E0007, 25 August2016) in Singapore and followed the NIH Guide for the Care and Use of Laboratory Animals. To avoid estrous cycle variations that might cause confounding actors in data interpretation, only male mice were used in all experiments and these mice were litter mates.

Genotyping Analysis
Pparb/d fl/fl and FSP1cre-Pparb/d −/− mice used at week 4 and 8 of age were weaned and genotyped at three weeks of age as previously described [42]. Genotyping via tail nicking of P2 of age mice was carried out together with the harvesting of livers. Prior to ear biopsy or tail cut, mice were anesthetized via intraperitoneal injection of ketamine: xylazine (100 mg: 10 mg per kg body weight) dissolved in 0.9% saline solution. An ear biopsy of approximately 2 × 2 mm 2 or 2 mm of the tail cut was taken. Extraction of DNA from the biopsy was carried out with the KAPA HotStart Mouse Genotyping Kit (Kappa Biosystems, Wilmington, MA, USA), based on the suggested protocol from the kit.
Male P2 mice were determined by the sex-determining region of the Y chromosome. SRY allele was detected using primer SRY forward primer (5 -TTGTCTAGAGAGCATGGAGGGCCATGTCAA-3 ) and SRY reverse primer (5 -CCACTCCTCTGTGACACTTTAGCCCTCCGA-3 ). The primers were synthesized by Integrated DNA Technologies, Singapore. The genomic DNA was amplified by PCR using KAPA HotStart Mouse Genotyping Kit (Kappa Biosystems, Wilmington, MA, USA). The protocol is provided in Appendix A. PCR products were subsequently analyzed by electrophoresis on 1.5% agarose gels.

Histology
Specimens taken from the liver left lateral lobe were briefly rinsed in 1× phosphate buffer saline (PBS; pH 7.4), which comprises 2.7 mM KCL, 137 mM NaCl, 2 mM KH 2 PO 4 , and 10 mM Na 2 HPO 4 , before being fixed overnight with 4% paraformaldehyde/PBS (pH 7.4) at 4 • C. The tissue was subjected to dehydration procedure in 50% EtOH, 60% EtOH, 70% EtOH, 80% EtOH, 90% EtOH, twice in 100% EtOH, EtOH/xylene, twice in xylene at 1 h each. The dehydrated liver sections were infiltrated with molten paraffin at 60 • C overnight before embedding fixing into paraffin blocks. The Pparb/d fl/fl and FSP1cre-Pparb/d −/− livers were sectioned at 5 µm thickness. For hematoxylin and eosin (H&E) staining, the sections were deparaffinized, rehydrated, stained with hematoxylin for 3 min, and differentiated in 0.1% HCl for 2 s prior counterstaining with eosin for 27 s. For Oil red O staining preparation, the fresh liver samples from the left lateral lobe were collected and embedded in OCT freezing medium (Leica Microsystems, Wetzlar, Germany). A stock solution of Oil red O was prepared by dissolving 0.5 g of Oil red O powder (Sigma-Alrich, St Louis, MO, USA) in 500 mL isopropanol. 50 mL of 60% Oil red O stock solution was diluted with 20 mL of deionized water to prepare the working solution. The frozen specimens were cryosectioned at 8 µm thickness and stained with Oil Red O working solution for 10 min and followed by counterstaining with hematoxylin. Images were acquired using the ZEN 2012 LE software and Axio Scan.Z1 slide scanner microscope with 10× and 20× objective lenses (Carl Zeiss AG, Oberkochen, Germany).
Prior to double immunofluorescence staining, the sections were melted at 60 • C before the rehydrating procedure in 2 times in xylene, EtOH/xylene, 2 times in 100% EtOH, 90% EtOH, 80% EtOH, 70% EtOH, 60% EtOH, and 50% EtOH at 5 min each. The tissue sections were subjected to antigen retrieval procedure in 10 mmol/L sodium citrate buffer (pH 6.0) at 95 • C for 20 min. The slides were washed thrice with PBS before blocking with 5% BSA for 1 h. Tissue sections were incubated with either anti-GFAP (1:100), or anti-CD11b (1:100), or anti-CD68 (1:100), or anti-CD146 (1:100) in 0.5% BSA overnight at 4 • C. Except for the tissue sections incubated with anti-CD146 or anti-CD68, the rest of the tissue sections were incubated with secondary goat anti-rabbit Alexa Fluor 488 antibodies (1:200) in 0.5% BSA with incubation for 1 h at room temperature in the dark. Tissue sections incubated with CD68 were incubated with secondary goat anti-rat Alexa Fluor 488 antibodies (1:200) in 0.5% BSA with incubation for 1 h at room temperature in the dark. Tissue sections were then incubated with anti-S100A4 (1:100) in 0.5% BSA overnight at 4 • C. Secondary goat anti-rabbit Alexa Fluor 594 antibodies (1:200) in 0.5% BSA were added with incubation for 1 h at room temperature in the dark. The tissue sections were counterstained with DAPI (Vectashield, Burlingame, CA, USA). Between each step, slides were washed thrice with PBS. Immunofluorescent images were taken using a Carl Zeiss (Thornwood, NY, USA) confocal microscope LSM 710meta using a Plan-Apochromat × 63/1.4 oil DIC objective, and ZEN 2012 LE software with constant exposure and gain.

Isolation of Hepatic Cells
P2 Pparb/d fl/fl and FSP1cre-Pparb/d −/− livers were harvested and briefly rinsed with 1× PBS (pH 7.4) prior they were minced into smaller pieces. Each liver was digested in 2 mL HBSS solution with calcium and magnesium (#14025092) (Life Technologies) with 200 U/mL collagenase type IV (#17104019) (Life Technologies) for 30 min in 37 • C thermo-shaker at 120 rpm. The remaining liver tissue pieces were disaggregated with the use of the 18-gauge blunt needle syringe. Upon confirmation of the genotype and gender, the cell suspension was pooled and filtered through the 100 µm cell strainer into a 15 mL Eppendorf tube and topped up to 15 mL with Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (#16000036; Life Technologies). The cell suspension was centrifuged at 300 g for 5 min. The supernatant was discarded, and the cell pellet resuspended in HBSS solution with calcium and magnesium. The cell suspension was centrifuged at 300 g for 5 min; the supernatant was discarded prior adding 5 mL of red blood cell lysis buffer (1.5 M NH 4 Cl, 0.1 M NaHCO 3 , and 0.01 M EDTA) for 10 min on ice before filtering through the 100-µm cell strainer. The cell suspension was washed 2 times before proceeding to flow cytometric analysis.  Cholesterol Assay Kit generated a dye that absorbs light at 540 nm, whereas enzymatic reactions from the Serum Triglyceride Quantification Kit generated a dye that absorbs light at 570 nm. A standard curve was generated for each individual assay, using the serial dilutions of the standard solution provided in each kit. Using the optical density values at the specific absorbance wavelength against the respective concentrations of standards, the standard curves were generated using Microsoft ® Office Excel 2010 and best fitted to a linear function with an R2 ≥ 0.98. The calculated total lipids level was normalized with the mass of liver tissue, while the calculated triglyceride, total cholesterol, cholesterol ester, and free cholesterol levels were normalized with the total protein determined with the Bradford assay protein assay #5000006 (Bio-Rad Laboratories, Hercules, CA, USA).

Total RNA Extraction and Reverse Transcription
50-100 mg of liver tissue in 1 mL of TrIzol ® Reagent (Invitrogen, Carlsbad, CA, USA) was processed by TissueLyser II (Qiagen, Hilden, Germany). Each homogenate was subjected to centrifugation at 12,000 g for 10 min at 4 • C to remove the insoluble materials. The soluble RNA in the supernatant was transferred to a new tube. 0.2 mL of chloroform was added and mixed vigorously for 15 s, prior incubation at room temperature for 5 min. The samples were centrifuged at 12,000 g for 15 min at 4 • C. Approximately 400 µL of the colorless RNA-containing aqueous phase was transferred to a fresh tube and an equal volume of 70% ethanol was added and vortexed. These exacted RNA samples were further purified using the PureLink RNA Mini Kit (Invitrogen, Carlsbad, CA, USA) accompanied with a DNase I treatment (Roche Applied Science, Penzberg, Upper Bavaria, Germany), according to the protocol from the manufacturer. The extracted RNA was spectrophotometrically measured, and quality was determined by absorbance ratios at 260 nm/280 nm and 260 nm/ 230 nm through Nanodrop Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Every one microgram of RNA was reverse transcribed with the iScript™ cDNA Synthesis Kit (Hercules, CA, USA), according to the manufacturer's protocol. The cDNA products were diluted 5 times and stored at −20 • C.

Quantitative Real-Time PCR
Ten sets of tissue samples were used for each genotype and each age group. The sequence of the primer pairs was obtained from the NCBI Primer Blast available online at https:// www.ncbi.nlm.nih.gov/tools/primer-blast/ and summarized in Appendix B. Quantitative q(PCR) was performed on a StepOnePlus Real-Time PCR System (Thermo Fisher Scientific, Waltham, MA, USA) using KAPA FAST qPCR kit (Kappa Biosystems, Wilmington, MA, USA). Gadph (forward primer sequence: 5 -AGGTCGGTGTGAACGGATTTG-3 ; reverse primer sequence: 5 -TGTAGACCATGTAGTTGAGGTCA-3 ) was used as the normalizing housekeeping gene.

Statistical Analysis
Values are shown as mean±standard error of the mean (SEM). The two-tailed Mann-Whitney test and two-way ANOVA were performed using GraphPad Prism (version 4.00) software. Two-tailed Mann-Whitney test is a non-parametric test that was used to compare two sample means that come from the same population, and used to test whether the two sample means are equal or not, without assuming a gaussian distribution. For quantification of total lipids, total cholesterol, and total triglyceride levels, two-way ANOVA was also used to compare the mean differences between the ages (P2, 4 weeks, 8 weeks) of Pparb/d fl/fl and FSP1cre-Pparb/d −/− livers. In both statistical tests, P < 0.05 was considered statistically significant.