Abstract
Non-alcoholic fatty liver disease (NAFLD) and steatohepatitis (NASH) have become a worldwide health concern because of lifestyle changes, but it is still lack of specific therapeutic strategies as the underlying molecular mechanisms remain poorly understood. Our previous study indicated that deficiency of Arid1a, a key component of SWI/SNF chromatin remodeling complex, initiated mouse steatohepatitis, implying that Arid1a might be essentially required for the integrity of hepatic lipid metabolism. However, the exact mechanisms of the pathological process due to Arid1a loss are unclear. In the present work, we show that hepatocyte-specific deletion of Arid1a significantly increases susceptibility to develop hepatic steatosis and insulin resistance in mice fed with high-fat diet (HFD), along with the aggravated inflammatory responses marked by increment of serum alanine amino transferase (AST), aspartate amino transferase (AST) and TNFα. Mechanistically, Arid1a deficiency leads to the reduction of chromatin modification characteristic of transcriptional activation on multiple metabolic genes, especially Cpt1a and Acox1, two rate-limiting enzyme genes for fatty acid oxidation. Furthermore, our data indicated that Arid1a loss promotes hepatic steatosis by downregulating PPARα, thereby impairing fatty acid oxidation which leads to lipid accumulation and insulin resistance. These findings reveal that targeting ARID1a might be a promising therapeutic strategy for NAFLD, NASH and insulin resistance.
- Abbreviations
- NAFLD
- non-alcoholic fatty liver disease
- NASH
- non-alcoholic steatohepatitis
- FAO
- fatty acid oxidation
- FAs
- fatty acids
- FD
- high fat diet
- CD
- chow diet
- GTT
- glucose tolerance test
- ITT
- insulin tolerance test
- LKO
- liver-specific knockout
- HE
- hematoxylin and eosin
- TG
- triglyceride
- TCHO
- total cholesterol
- NEFA
- non-esterified fatty acid
- ALT
- alanine amino transferase
- AST
- aspartate amino transferase
- ALP
- alkaline phosphatase
- KEGG
- Kyoto Encyclopedia of Genes and Genomes
- GO
- gene ontology
- OA
- oleic acid
- PA
- palmitic acid.
Introduction
Non-alcoholic fatty liver disease (NAFLD), as characterized by hepatic steatosis, insulin resistance and chronic low-grade inflammation, is the leading cause of chronic liver disease, and may process to non-alcoholic steatohepatitis (NASH), cirrhosis and hepatocellular carcinoma (1,2). Ectopic lipid accumulation is proposed to be the root cause of NAFLD, leading to hepatic steatosis and insulin resistance (3). Fatty acid oxidation (FAO) is a multistep process that involves the degradation of fatty acids (FAs) by the sequential removal of two-carbon units from the acyl chain to produce acetyl-CoA, and there are compelling evidences show that undermined hepatic FAO is one of the central processes in hepatic lipid disorders (4). Inadequate oxidation of fatty acid would result in aberrant lipid accumulation and massive steatosis (2). However, the underlying mechanisms for this process are complex, multifactorial and heterogeneous, which hinder the exploration of exact pathological processes and therapeutic methods.
SWI/SNF chromatin remodeling complex manipulates the accessibility of regulatory transcriptional factors and thereby controls the transcriptional activation and repression of some genes related to physiological functions (5,6). So far, the effects of SWI/SNF complex members on nutrient metabolism remains less clear. It was reported that hepatocyte-specific deletion of SNF5, another subunit of SWI/SNF complex, caused glycogen storage deficiency and energetic metabolism impairment (7). Baf60a and Baf60c regulated metabolic gene programs in the liver and skeletal muscle (8,9), and even, Baf60a controlled the transcriptional programs of some genes responsible for hepatic bile acid synthesis and intestinal cholesterol absorption (10). Moreover, expression of BAF60a could stimulate fatty acid β-oxidation in cultured hepatocytes and ameliorated hepatic steatosis in vivo (11).
ARID1a, also known as Baf250a, a subunit of SWI/SNF chromatin remodeling complex, may facilitate the access of transcription factors and regulatory proteins to genomic DNA. Loss of ARID1A may result in the structural and functional alterations of SWI/SNF complex, which leads to transcriptional dysfunction through disruption of nucleosome sliding activity, assembly of variant SWI/SNF complexes, targeting to specific genomic loci, and recruitment of coactivator/corepressor activities (12-17). Our group previously demonstrated hepatocyte-specific Arid1a deficiency initiated mouse hepatic steatosis and steatohepatitis (18), suggesting that Arid1a might participate in lipid mechanism, however, the exact pathological process and the underlying molecular mechanism were unclear. In this study, we employed hepatocyte-specific Arid1a knockout mice administrated with high fat diet (HFD), and proved that wild type Arid1a protected mice against hepatic steatosis and insulin resistance. Mechanistically, Arid1a directly controlled the transcriptional activation of multiple metabolic genes by erasing H3K4me3 marks, and downregulated PPARα which thus inhibited fatty acid oxidation to regulate lipid metabolism and insulin sensitivity. In general, the finding that Arid1a may regulate lipid mechanism via directly and indirectly modulating the transcription of fatty acid oxidation-related genes provides a promising therapeutic approach for NAFLD, NASH and insulin resistance.
Results
Arid1a is required for regulating insulin sensitivity and glucose tolerance
Insulin resistance is a vital part in the development of hepatic steatosis (19). To directly address the effect of Arid1a deletion on glucose homeostasis and insulin sensitivity, liver-specific Arid1a knockout (Arid1aLKO) mice and Arid1aF/F mice were placed on a normal chow diet (CD) or high-fat diet (HFD), and then metabolic phenotypes of these mice were characterized. Interestingly, despite of similar food intake (Supplementary Fig. 1A), Arid1aLKO mice gained much more body weights than Arid1aF/F mice when placed on a HFD, whereas no difference was statistically identified between the Arid1aLKO and Arid1aF/F littermates under CD (Fig. 1A-B). In parallel, compared to Arid1aF/F littermates fed on HFD, Arid1aLKO mice showed significantly elevated levels of fasting blood glucose (Fig. 1C) and insulin (Fig. 1D), as well as the markedly decreased glucose tolerance via glucose tolerance test (GTT) (Fig. 1E) and insulin sensitivity shown by insulin tolerance test (ITT) (Fig. 1F). Furthermore, insulin sensitivity was assessed by measuring the phosphorylated Protein Kinase B (pAKT) level in livers of Arid1aLKO mice fed on HFD. As expected, after an injection of insulin, pAkt level was significantly decreased in the livers of Arid1aLKO mice (Fig. 1G), as compared to that of Arid1aF/F littermates. However, no significant difference in blood glucose and insulin levels were found between Arid1aLKO and Arid1aF/F group fed on CD during GTT and ITT tests (Supplementary Fig. 1B-E). Collectively, these data indicate that hepatocyte-specific Arid1a deletion damages insulin sensitivity, thereby leading to insulin resistance and glucose tolerance of Arid1aLKO mice under HFD challenge.
HFD aggravates Arid1a deficiency-induced hepatic steatosis, fibrosis and inflammation
To define the role of Arid1a on the development of non-alcoholic fatty liver, we carried out histological comparisons between the livers from 4 months-age Arid1aLKO and Arid1aF/F mice. Consistent with our previous data, Arid1aLKO mice fed with CD exhibited hepatic dysplasia, steatosis and fibrosis, as compared to Arid1aF/F mice (18). However, HFD significantly aggravated these phenotypes of Arid1aLKO mice, as evidenced by quantitatively analyses on histology according to the NAFLD Activity Score (NAS) (20) which measures degree of steatosis, lobular inflammation, and hepatocyte ballooning (Fig. 2A and B). In addition, more lipid droplets were accumulated in the livers of Arid1aLKO mice challenged with HFD, as visualized by Oil Red O staining (Fig. 2C). In parallel with these morphologic changes, the Arid1aLKO mice fed on HFD exhibited worse liver function, as indicated by significant elevation of serum AST, ALT and ALP levels (Fig. 2D-F). Moreover, the mRNA levels of inflammatory-related genes including Tnfα, Arg1, Mcp1 and Cd11b were also markedly elevated in HFD-fed Arid1aLKO mice compared to those in the Arid1aF/F mice (Fig. 2G). Correspondingly, blood concentration of TNF-α (Fig. 2H) was also significantly increased in obese Arid1aLKO mice, although serum level of IL-6 had no significant change (Fig. 2I). Taken together, these data indicate that HFD aggravates Arid1a deficiency-induced hepatic steatosis, fibrosis and inflammation, suggesting that the Arid1a deficiency-induced dysfunction of lipid mechanism could be crucial to the development of hepatic disease.
Arid1a deletion leads to hepatic steatosis possibly through FAO impairment
Metabolic imbalance between lipid acquisition and removal in the liver is the first step in the pathophysiology of NAFLD (21). To explore the effects of Arid1a on hepatic lipid metabolism, the mice were fed with CD or HFD for 12 weeks and then sacrificed to observe their phenotypes. Significantly, the liver weights of the Arid1aLKO mice fed on HFD were much higher than those of the Arid1aF/F mice (Fig. 3A). Additionally, both subcutaneous white adipose tissue (WAT) and inguinal WAT fat pads were larger in Arid1aLKO mice (Fig. 3B and C), indicative of decreased adipose tissue homeostasis. Afterwards, blood lipid testing revealed that plasma cholesterol (TCHO) as well as low-density lipoprotein (LDL) and high-density lipoprotein cholesterol (HDL) cholesterol levels, but not non-esterified fatty acids (NEFA), were much higher in Arid1aLKO animals (Fig. 3D-F). Surprisingly, plasma triglycerides (TG) of Arid1aLKO mice was reduced (Fig. 3D). Furthermore, we determined liver lipids. TCHO and TG content in the livers from HFD-fed Arid1aLKO mice were significantly increased, where hepatic TG in Arid1aLKO mice were even 8-fold higher than those in Arid1aF/F mice (Fig. 3G), indicating that the hyperlipidemia of Arid1aLKO mice on HFD is likely caused by hepatic Arid1a loss, although there is barrier to block the secretion of liver TG to blood.
Based on the observations that dysregulated lipid metabolism was induced by Arid1a deletion, we assessed transcriptional expression of genes related to lipid acquisition and removal, including fatty acid synthesis, oxidation and TG secretion. The mRNAs encoded by these genes involved in FAO (Pparα, Cpt1a, Acox1, Pgc1α, Hmgcs2, etc.) and secretion (Mttp) were significantly reduced in the Arid1aLKO mice with HFD administration whereas there was no obvious difference of genes expression related to fatty acid synthesis (Fas, Acc1, Pparγ, Sreb1c and Chrebp) between Arid1aLKO and Arid1aF/F mice (Fig. 3H-J), suggesting that the reduced FAO and impaired TG secretion may contribute to hepatic lipid accumulation in Arid1aLKO mice.
Arid1a deficiency augments FFA-induced lipid accumulation and insulin resistance
To directly explore the role of Arid1a in FAO process, we investigated whether Arid1a can regulate free fatty acids (FFA)-induced lipid accumulation and insulin signaling in the hepatocytes isolated from Arid1aF/F mice, where Arid1a was deleted by adenovirus containing Cre (Ad-Cre) infection (Fig. 4A). Interestingly, Arid1a deficiency significantly augmented the oleic acid (OA, the unsaturated fatty acid) and palmitic acid (PA, the saturated fatty acid) -induced lipid accumulation in these hepatocytes (Fig. 4B), indicating that Arid1a was required for FAO process to break down these saturated and unsaturated fatty acid molecules. Later on, we assessed the alterations of insulin response in the primary hepatocytes with Arid1a loss. The results showed that, regardless of the presence or absence of OA, Arid1a deficiency led to decreased phosphorylation levels of AKT and GSK-3β in hepatocytes upon insulin stimulation (Fig. 4C). Similar results were observed in hepatocytes isolated from tamoxifen-induced Arid1a knockout mice, as well as in hepatocellular carcinoma (HCC) cell lines MHCC-97H and SK-Hep1-6 with Arid1a knockout by CRISPR-Cas9 system upon insulin treatment (Supplementary Fig. 2).
Next, we performed the rescue assay on the Arid1a-/- primary mouse hepatocytes via ectopic ARID1a expression. Remarkably, the rescued ARID1a expression resulted in the restoration of lipid clearance in Arid1a-deficient hepatocytes under both OA and PA challenge, which was evidenced by quantitative analysis of Oil Red O staining (Fig. 4D and 6F). Furthermore, the phosphorylation levels of AKT and GSK3β were also significantly recovered by ectopic ARID1a expression, and similar results were acquired in the MHCC-97H cells with Arid1a deletion (Fig. 4E and F). Altogether, these data indicate that Arid1a may disrupt the fatty acid β-oxidation as a critical process for lipid metabolism regulation.
Arid1a loss down-regulates essential genes in FAO and Ppar signaling pathways
Then we performed RNA sequencing on the mouse hepatocytes infected with Ad-Cre or Ad-GFP to figure out the Arid1a-regulated downstream genes (GEO). Significantly, pathway analysis using the GO and KEGG databases revealed that the down-regulated genes induced by Arid1a deficiency were enriched in lipid and fatty acid metabolic processes, PPAR and insulin signaling pathways, respectively (Fig. 5A and B). Additionally, we performed the gene set enrichment analysis and found that fatty acid metabolism was the Top5 ranked among the ‘Hallmark’ gene sets (Fig. 5C). Many genes related to fatty acid metabolism and PPAR signaling pathways were downregulated in the Arid1a deficient group (Figure 5D), some of which were consistent with qRT-PCR results in the mouse liver (Fig. 3H), suggesting that Arid1a loss induces the dysregulated lipid metabolism and PPAR function.
We further validated some RNA-seq revealed differentially expressed genes in hepatocytes by qRT-PCR, showing that Arid1a deficiency caused a significant reduction of FAO genes, such as Pparα, Acox1, Cpt1α, and Hmgcs2, in the primary mouse hepatocytes (Fig. 6A) and immortalized hepatocytes with SV40 expression (Supplementary Fig. 3). In contrast, the mRNA levels of lipogenic genes, including Acc1, Chrebp, Srebp1c and Fas, were not obviously changed in hepatocytes with Arid1a deletion. Moreover, the mRNA expression levels of key gluconeogenic enzymes were determined. As shown in Figure 6A, glucose 6-phosphatase (G6pase), but not phosphoenolpyruvate carboxykinase (Pepck) mRNAs were increased 2.5-fold in the hepatocytes with Arid1a deficiency (Fig. 6A). qRT-PCR and western blotting assay also revealed that insulin receptor substrate 1 (Irs1) was significantly down-regulated in mouse hepatocytes as well as in MHCC-97H cells with Arid1a knockout (Supplementary information, Fig. 4). Consistently, ectopic ARID1a expression recovered the expression of Pparα, Acox1, Cpt1a, Hmgcs2, and Pgc1α in the mouse hepatocytes with Arid1a deletion (Fig. 6B).
Among the mentioned effector genes of ARID1a, we paid particularly attention to Pparα, the major regulator of fatty acid oxidation (22), which was largely reduced in the liver and hepatocytes without Arid1a (Fig. 6A and C). Remarkably, western blotting assay confirmed that Pparα was recovered by ectopic ARID1a expression in the Arid1a-/- hepatocytes (Fig. 6D), suggesting that the down-regulated Pparα might be critical for Arid1a deletion-induced lipid accumulation and insulin resistance. Later on, we investigated whether induction of PPARα can improve FAO in Arid1a-/- hepatocytes. Interestingly, the results showed that the enforced expression of PPARα partially recovered the defective expression of its targeting downstream genes Acox1, Cpt1a, and Hmgcs2 in Arid1a-/- hepatocytes (Fig. 6E). Consequently, ectopic PPARα expression alleviated the lipid accumulations induced by OA and PA (Fig. 6F), and insulin sensitivity, indicated by pAKT and pGSK3β in mouse Arid1a-/- hepatocytes (Fig. 4E). The results testified that Arid1a deficiency-induced transcription repression of Pparα and related fatty acid oxidation genes could be responsible for hepatic lipid accumulation and steatosis.
Arid1a deficiency decreases H3K4me3 and chromatin accessibility on promoters of metabolic genes
It is noteworthy that Arid11a may facilitate access of transcription factors and regulatory proteins to the genomic DNA, and thus regulates the transcription of downstream genes. Therefore, we proposed that Arid1a may impact on the local epigenetic landscape to regulate transcription activity of key metabolic genes. Here, we firstly analyzed the published ChIP-seq dataset (GSE65167) (23). Bioinformatic analysis based on GO and KEGG database indicated that the Arid1a targeted genes are highly associated with lipid, fatty acid metabolic processes and PPAR signaling (Fig. 7A and B). Then we integrated the Arid1a-regulated genes, indicated by RNA-seq data, in mouse hepatocytes, with the genes containing Arid1a-binding peaks on promoter, showing that two key FAO related genes Acox1 and Cpt1a have the Arid1a-binding peaks (Fig. 7C). Notably, Cpt1a is a rate-limiting enzyme involved in mitochondrial β-oxidation, catalyzing the esterification of long-chain acyl-CoAs to L-carnitine for their transportation into the mitochondria (24). While Acox1 is the first and a rate-limiting enzyme in the peroxisomal β-oxidation which catalyzing the desaturation of very-long-chain acyl-CoAs to 2-trans-enoyl-CoAs (25). To verify the binding of Arid1a to their promoters, we performed ChIP-PCR with anti-Arid1a antibody in primary hepatocytes with or without Arid1a. Our results showed that Arid1a was present in the promoters of Acox1 and Cpt1a in hepatocytes with Arid1a, but this occupancy was decreased in Arid1a-deficient cells (Fig. 7D). We further explored whether Arid1a can epigenetically regulate fatty acid oxidation-related genes. Here, we detected the level of trimethylation of H3 lysine 4 (H3K4me3), an epigenetic marker associated with transcriptional activation, via ChIP-PCR with anti-H3K4me3 antibody, on Acox1 and Cpt1a promoters. It turned out that H3K4me3 on these regions was significantly reduced with Arid1a loss (Fig. 7E). Furthermore, Arid1a deficiency diminished the recruitment of SWI/SNF core subunit Brg1 to the promoter of Cpt1a, indicated by ChIP-PCR with anti-Brg1 antibody (Fig. 7F). Next, we conducted ATAC-seq to assess the alterations in chromatin accessibility on the promoters of Acox1, Cpt1a and Pparα. Notably, statistical analysis on the ATAC-seq data indicated that the chromatin accessibilities on promoters of Pparα, as well as Cpt1α, were significantly reduced when Arid1a was deleted in the hepatocytes (Fig. 7G). Altogether, these results indicate that Arid1a facilitates fatty acid oxidation by directly altering the epigenetic landscape of metabolism gene loci, including chromatin accessibility and local histone modification, as well as by indirectly activating downstream genes through Pparα transcription factor. Arid1a deficiency would result in transcriptional reduction of these FAO associated genes, thereby attenuating the corresponding metabolic functions that ends to hepatic steatosis, insulin resistance and NAFLD (Fig. 7H).
Discussion
As one of the most prevalent liver diseases worldwide, NAFLD is a progressive pathological condition that promotes more severe liver and metabolic dysfunction. Lipid accumulation in the liver (steatosis) is thought to be the “first hit” of NAFLD. In this study, we identify that Arid1a is a diet-sensitive subunit of the SWI/SNF chromatin remodeling complexes in the liver, which plays an important role in increasing the susceptibility to develop hepatic steatosis and insulin resistance in mice in conditions of HFD-induced NAFLD model. At the mechanistic level, hepatic Arid1a is critical for maintaining metabolic homeostasis through Pparα-mediated fatty acid oxidation in the liver. From a clinical perspective, these data indicate that targeting ARID1a might be a promising therapeutic strategy for treating NAFLD and its related metabolic disorders.
Several lines of evidences supported a pivotal role of Arid1a in metabolic regulation. First, the transcriptional profiling and pathway analysis indicated that the genes involved in lipid metabolism, especially fatty acid metabolism were among the top down-regulated by Arid1a deletion. Second, mRNA levels of the genes responsible for fatty acid oxidation were significantly reduced in Arid1a-/- livers and hepatocytes. Third, Arid1a downregulates Pparα, the major transcriptional factor for FAO (26). Consistently, Pparα in Arid1a-/- hepatocytes was sufficient to reverse the lipid accumulation and insulin resistance phenotype in vitro. Moreover, Arid1a regulates transcription activity of key metabolic genes on the local epigenetic landscape. Specifically, ARID1a deficiency leads to reduced levels of H3K4me3 and compacted chromatin structure on promoters of key metabolic genes.
Previous studies demonstrated that SWI/SNF family member BAF60c, recruiting other subunits, including ARID1a, interacts with USF transcription factor to lipogenic gene promoters to increase lipogenesis in response to feeding/insulin treatment (8). However, in our study, Arid1a deletion leads to steatosis by downregulating expression of these genes involved in fatty acid oxidation rather than de novo lipogenesis pathway. Especially, the expression of Cpt1a and Acox1, two key rate-limiting enzyme in fatty acid oxidation, were largely reduced as Arid1a loss. It has been reported that the disorders of hepatic fatty acid oxidation lead to massive steatosis and hypertriglyceridemia in animals. Some emerging evidences have shown that hepatic FAO is also impaired in human NAFLD (19,27,28). Moreover, deficiency of PPARα or its coactivators were also shown to affect the development of NAFLD. Compared with control mice, Pparα-deficient mice developed massive steatosis, lobular inflammation, and are more susceptible to the progression of NASH due to an impairment of mitochondrial FAO, and increment of oxidative stress and inflammation (29-31). We identified Pparα expression was downregulated in hepatocytes and mice with Arid1a deletion, whereas overexpression of Pparα restored lipid accumulation and insulin resistance caused by Arid1a deficiency. Unfortunately, although by analyzing ChIP-seq data performed by others (23), Pparα contains ARID1a-binding sites on promoter, no such direct binding of ARID1a on Pparαpromoter was identified in our experiment. In addition, we also did not detect any physical association between ARID1a and PPARα in 293T cells (data was not shown).
In summary, this study demonstrates that hepatocyte-specific inactivation of Arid1a reduced fatty acid oxidation, and aggravated diet-induced steatosis and insulin resistance in mice. Arid1a downregulates expression of the genes involved in fatty acid oxidation through Pparα and epigenetic regulation. These findings reveal a new mechanism underlying the role of Arid1a in NASH pathogenesis, and suggest a promising approach for the treatment of hepatic steatosis, fibrosis and insulin resistance by modulating Arid1a.
Materials and methods
Mice and diets
Arid1aLKO mice were generated by mating Arid1aFlox/Flox mice (kindly provided by Zhong Wang at the Cardiovascular Research Center, Massachusetts General Hospital, Harvard Medical School) with Albumin-Cre mice (from the Jackson Laboratory). Arid1aFlox/Flox animals were used as controls. Six week-old Arid1aLKO and Arid1aFlox/Flox mice were given free access to an HFD (composed of 59% fat, 15% protein, and 26% carbohydrates based on caloric content;) or CD (protein, 20%; fat, 10%; carbohydrates, 70%;) for 12 weeks. All animal experiments were approved by the Animal Care and Use Committee of Shang Hai Jiao Tong University, and all the procedures were conducted in compliance with institutional guidelines and protocols.
Blood parameters
The concentrations of the hepatic enzymes alanine amino transferase (ALT), aspartate amino transferase (AST), and alkaline phosphatase (ALP) were analyzed using a spectrophotometer (Chemix 180i, Sysmex Shanghai Ltd, Shanghai, China) according to manufacturer’s instructions. Insulin plasma levels were determined using mouse insulin ELISA kit (Millipore, St. Charles, MO, US). The levels of the serum inflammatory indicators were measured by ELISA (BD Bioscience, San Diego, CA, USA).
Lipid parameters
Commercial kits were used to measure the liver triglyceride (TG), total cholesterol (TC), and nonesterified fatty acid (NEFA) contents in the liver according to the manufacturer’s instructions (290-63701 for TG, 294-65801 for TC, 294-63601 for NEFA; Wako, Tokyo, Japan). Serum lipids were analyzed using a spectrophotometer (Chemix 180i, Sysmex Shanghai Ltd, Shanghai, China) according to manufacturer’s instructions.
Glucose and insulin tolerance tests
Glucose tolerance tests (GTT) were performed in mice that were fed either CD or HFD for 12 weeks. One week later, the same mice were used for insulin tolerance tests (ITT). For GTT, mice were fasted for 16 h. After measuring the baseline blood glucose level via a tail nick using a glucometer, 1.5g/kg glucose was administered via intraperitoneal injection, and glucose levels were measured 0,15,30,45,60,90 and 120 minutes after glucose injection. For ITT, 6-h fasted mice were injected intraperitoneally with recombinant human insulin at 1U/kg and their blood glucose concentrations were determined 0,15,30,45,60,90 and 120 minutes after insulin injection.
Quantitative real-time PCR analysis
RNA extracted from liver or hepatocytes were subjected to reverse transcription and subsequent PCR using a real-time PCR system (ABI, Carlsbad, CA). PCR primer sequences are listed in Supplemental Table 1. Expression of the respective genes was normalized to β-Actin as an internal control.
Primary hepatocyte isolation, cultivation and treatment
Primary hepatocytes were isolated from 6- to 8-week-old mice following the steps reported previously (32). Hepatocytes were cultured in Dulbecco’s modified Eagle’s medium contained 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin in a 5% CO2/water-saturated incubator at 37°C. To stimulate insulin signaling, hepatocytes were stimulated with insulin 10 nM for indicated time. For lipid accumulation in vitro, hepatocytes were treated with 0.2 mM palmitate acid (PA, Sigma, St. Louis, MO, USA) or 0.08 mM oleic acid (OA, Sigma, St. Louis, MO, USA) for 24h.
Plasmid Construction and Transfection
The plasmids encoding T7-ARID1a was purchased from Addgene (Cambridge, MA, USA). The plasmid expressing flag-tagged human PPARα was constructed by cloning the mouse PPARα cDNA into the pcDNA3.1 vector. Transfection assays for plasmids were performed using Lipofectamine 2000 (Invitrogen; Carlsbad, CA, US) according to the manufacturer’s protocol.
Histological analysis
Sirius red (Sigma, St. Louis, MO, USA) staining was performed using paraffin-prepared liver sections to determine the fibrosis of the tissues. Liver sections were embedded in paraffin and stained with hematoxylin and eosin (H&E) to visualize adipocytes and inflammatory cells in the tissues. Frozen liver sections of 10mm were fixed in formalin and then rinsed with 60% isopropanol. After staining with freshly prepared Oil-red O solution (Sigma, St. Louis, MO, USA) for 10min, the sections were rinsed again with 60% isopropanol. Sections or cells were analyzed by inverted microscope (Olympus, Tokyo, Japan).
Western blotting
After extracted from liver samples or cultured cells, each protein sample (50 mg) was subjected to SDS/PAGE and transferred to NC membrane (Millipore, Bedford, MA, USA). Then the corresponding primary and secondary antibodies were incubated to visualize the protein. Antibodies used in western blot analysis included anti-ARID1a (12354, CST, Billerica, MA, USA), anti-phospho-AKTS473 (9271, CST, Billerica, MA, USA), anti-phospho-GSK3β (9315, CST, Billerica, MA, USA), anti-AKT (9272, CST, Billerica, MA, USA), anti-GSK3β (9323, CST, Billerica, MA, USA), anti-PPARα (15540-I-AP, ProteinTech, Chicago, IL, USA), anti-β-actin (2381, (Sigma, St. Louis, MO, USA)).
ChIP
Chromatin immunoprecipitation (ChIP) assay was performed according to the protocol developed by Upstate Biotechnology as described (11). Briefly, chromatin lysates were prepared from hepatocytes following crosslinking with 1% formaldehyde. The samples were precleared with Protein-G agarose beads and immunoprecipitated using antibodies against ARID1a (sc32761-x, Santa Cruz, California, USA), H3K4me3 (Abcam, 8580, Cambridge, UK),BRG1 (sc17796, Santa Cruz, Cruz, California, USA) or control IgG in the presence of BSA and salmon sperm DNA. Beads were extensively washed before reverse crosslinking. DNA was purified using a PCR Purification Kit (Tiangen,Beijing,China) and subsequently analyzed by qPCR using primers located on the Acox1 and Cpt1a promoter.
RNA-Seq
Total RNA-seq was performed on three independent biological replicates per condition. Samples from different conditions were processed together to prevent batch effects. For each sample, 10 million cells were used to extract total RNA and produce RNA-seq libraries. Raw sequencing reads were trimmed by Trimmomatic (v0.36) to remove adapters and low quality sequences. The filtered reads were aligned to the GRCm38 UCSC annotated transcripts via Tophat (v2.1.0)56. Transcripts were then assembled, counted and normalized with the Cufflinks suit (v2.2.1)57. Differentiated expressed genes were analyzed by Cuffdiff in the Cufflinks suit, using p value <0.05 and fold change >1.5 as cutoff. Heatmaps were generated by using R package pheatmap (1.0.10).
Gene ontology and KEGG pathway enrichment analyses were performed using DAVID (v6.8) with the differentially expressed gene lists. Gene set enrichment analysis (GSEA v3.0) was performed using the normalized expression values. The raw sequencing data of RNA-seq has been deposited to NCBI GEO database with the accession number (GSE….)
ATAC-seq
Cells were harvested and subjected for ATAC-seq analysis as previously described (33). Briefly, 50, 000 cells were washed with cold PBS and cell membrane was lysed in lysis buffer (10 mM Tris-HCl, pH 7.4, 10 mM NaCl, 3 mM MgCl2, 0.1% Nonidet P-40). After centrifugation, cell pellets were resuspended and incubated in transposition mix containing Tn5 transposase (Illumina, California, USA) at 37 °C for 30 minutes. Purified DNA was then ligated with adapters and amplified 11 total cycles. Libraries were purified with AMPure beads (Beckman Coulter Inc, Brae, CA, USA) to remove contaminating primer dimers. Libraries were then sequenced on Illumina X-Ten system with 150 bp paired-end sequencing strategy.
For ATAC-seq data analysis, Illumina adapters and low-quality reads were trimmed by Trimmomatic (v0.36) (Trimmomatic: a flexible trimmer for illumina sequence data). Afterwards, all reads for each sample were combined and aligned to mouse reference build GRCm38 using bwa (v0.7.11) with default settings. More than 80 million pair-ended sequencing reads were obtained from each library of the MEF cells and 40 to 65 million reads were obtained from the libraries of the primary mouse liver cells. Low quality reads (mapping quality < 20) and reads mapping to mitochondrial DNA were removed using samtools (v1.6). Duplicates were excluded using Picard Tools (v1.4.5). Between 49 and 61 million high-quality reads per sample that mapped to genomic DNA were obtained for MEF cells. Between 11 and 25 million high-quality reads per sample that mapped to genomic DNA were obtained for mouse primary liver cells. All mapped reads were offset by +4bp for the +strand and -5bp for the –strand (Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position) using bedtools (v2.25.0). Libraries of the same condition were merged for subsequent analyses. Peaks were called for each sample using MACS2 (Model-based analysis of ChIP-Seq (MACS)) (v2.1.1.20160309) using parameter “-–nomodel –shift 100 –extsize 200”. The peaks with p<0.01 for each group in the same cell line were merged for searching differential peaks. Differential peak calling, as well as peak annotation and motif analysis were performed using HOMER (v4.9.1) with default settings. Fold change larger than 2 was set as the cutoff for differential peak identification. Sequencing signals were generated by transforming the mapping files into bigwig tracks, which was visualized in the Integrated Genomic Viewer (IGV). Signal heat maps centered around peaks were generated using deep Tools (v 2.5.3). All genomic datasets were deposited in GEO datasets with the accession number (GSE……).
Statistics
All statistical analyses were performed using Graphpad prism software (version 19.0), and all data are expressed as the mean ± standard deviation. For data with a normal distribution and homogeneity of variance, two-tailed Student t tests were used to evaluate significant differences between two groups, P values less than 0.05 were considered significant. * P < 0.05; ** P < 0.01; *** P < 0.001.
Acknowledgments
We sincerely thank Zhong Wang professor at the Cardiovascular Research Center of Harvard Medical School, for providing Arid1aF/F mice in this study. This work was supported by National Natural Science Foundation of China (81672772 and 81472621), China National Science and Technology Major Project for Prevention and Treatment of Infectious Diseases (grant no. 2017ZX10203207) and National Program on Key Research Project of China (grant no. 2016YFC0902701)