GR-mediated transcriptional regulation of m6A metabolic genes contributes to diet-induced fatty liver in hens

Background Glucocorticoid receptor (GR) mediated corticosterone-induced fatty liver syndrome (FLS) in the chicken by transactivation of Fat mass and obesity associated gene (FTO), leading to demethylation of N6-methyladenosine (m6A) and post-transcriptional activation of lipogenic genes. Nutrition is considered the main cause of FLS in the modern poultry industry. Therefore, this study was aimed to investigate whether GR and m6A modification are involved in high-energy and low protein (HELP) diet-induced FLS in laying hens, and if true, what specific m6A sites of lipogenic genes are modified and how GR mediates m6A-dependent lipogenic gene activation in HELP diet-induced FLS in the chicken. Results Laying hens fed HELP diet exhibit excess (P < 0.05) lipid accumulation and lipogenic genes activation in the liver, which is associated with significantly increased (P < 0.05) GR expression that coincided with global m6A demethylation. Concurrently, the m6A demethylase FTO is upregulated (P < 0.05), whereas the m6A reader YTHDF2 is downregulated (P < 0.05) in the liver of FLS chickens. Further analysis identifies site-specific demethylation (P < 0.05) of m6A in the mRNA of lipogenic genes, including FASN, SREBP1 and SCD. Moreover, GR binding to the promoter of FTO gene is highly enriched (P < 0.05), while GR binding to the promoter of YTHDF2 gene is diminished (P < 0.05). Conclusions These results implicate a possible role of GR-mediated transcriptional regulation of m6A metabolic genes on m6A-depenent post-transcriptional activation of lipogenic genes and shed new light in the molecular mechanism of FLS etiology in the chicken.


Background
Fatty liver syndrome (FLS) is a metabolic disease mostly observed in laying hens, which is characterized by increased lipid accumulation in the liver [1]. FLS occurs at the rate between 4% and 20% in chickens kept in intensive systems, which may cause dramatic drop in egg production and increased mortality, leading to considerable economic losses [2,3]. Several factors have been reported to contribute to the development of FLS, including genetics, environment, nutrition, toxic substances, and hormones [4,5]. Among these, nutrition is considered the main cause of FLS in the modern poultry industry. Nutritionally over-fed laying hens are at risk for developing FLS [4], and a high-energy maize diet produces a higher incidence of FLS than a low-energy barley diet [6]. Also, high-energy low-protein (HELP) diet is used to establish a model of FLS in some previous publications [7][8][9]. FLS is caused primarily by an imbalance of hepatic energy influx and efflux. Glucocorticoids (GC) play an important role in hepatic metabolic homeostasis.
Chronically elevated GC level is a common feature of fatty liver in humans and animal models [10,11]. The actions of GC are primarily mediated by glucocorticoid receptor (GR) [12]. Over-activation of GR pathway leads to transcriptional up-regulation of lipogenic genes, causing hepatic steatosis [13]. In chickens, corticosterone (CORT) is the main active form of GC [14]. We have shown previously that excessive CORT administration causes FLS in chickens, which is characterized by excessive lipid accumulation in the liver [15][16][17]. GR is upregulated in the liver of FLS chickens [15][16][17], yet it remains unknown how GR contributes to diet-induced FLS in the chicken.
Three families of protein components are involved in the dynamic m 6 A methylation. The methyltransferases complex formed by methyltransferase-like 3 (METTL3), methyltransferase-like 14 (METTL14), and Wilms' tumor 1-associating protein (WTAP) [18][19][20] are "writers" to transfer methyl group to the adenosine of the consensus m 6 A motif in the target RNA. The two demethylases, termed as "erasers", fat mass and obesityassociated protein (FTO) and α-ketoglutarate-dependent dioxygenase AlkB homolog 5 (ALKBH5) [21,22], are responsible to remove the methyl group from the m 6 A. The "readers", such as YTH-domain family 1-3 (YTHD F1-3), can recognize this methylation and regulate the RNA metabolism including stability or translation efficiency of the mRNA [23][24][25][26]. m 6 A modification plays a key role in lipid accumulation and energy metabolism [27,28]. Recently, GR is reported to mediate corticosterone-induced fatty liver in the chicken by transactivation of FTO, leading to demethylation of m 6 A and post-transcriptional activation of lipogenic genes such as sterol regulatory element-binding protein-1 (SREBP1), fatty acid synthase (FASN) and stearoyl-CoA desaturase (SCD) [17]. However, it remains unclear what specific sites of lipogenic genes are demethylated and which m 6 A binding protein is involved in mediating m 6 A-dependent post-transcriptional activation of lipogenic genes.
Therefore, this study was aimed to investigate whether GR and m 6 A modification are involved in HELP dietinduced FLS in laying hens, and if yes, what specific m 6 A sites of lipogenic genes are modified and how GR mediates m 6 A-dependent lipogenic gene activation in HELP diet-induced FLS in the chicken.

Animals and treatment
Forty-eight Hy-Line Variety Brown laying hens (260 days of age, 1.69 ± 0.09 kg in body weight) were raised in the animal house of Nanjing Agricultural university, with the room temperature at approximately 24°C, and the light regime of 16 L: 8D. Three hens were housed in each cage (60 cm × 46 cm × 44 cm) equipped with a nipple drinker. Hens were randomly divided into control (CON, twenty-four chickens in eight cages) and highenergy low protein diet (HELP, twenty-four chickens in eight cages) groups, fed control diet (2,610 kcal/kg metabolizable energy, 16.9% crude protein) and HELP diet (3,100 kcal/kg metabolizable energy, 12.1% crude protein), respectively, for 12 weeks. The ingredient and calculated composition of the diets used in the current study are presented in Table 1. Hens were subjected to feed restriction (110 g per hen per day) with free access to water throughout the experiment. After 12 weeks of dietary treatment, 1 hen from each cage was randomly selected and killed by rapid decapitation that is considered acceptable for euthanasia of birds according to American Veterinary Medical Association (AVMA) Guidelines for the Euthanasia of Animals: 2013 Edition. Liver samples were rapidly frozen in liquid nitrogen and kept at − 80°C for further analysis. Among eight hens from HELP group, six hens were diagnosed as FLS according to the hepatic triglyceride (TG) content. So, the sample size was adjusted of both CON and HELP group to six. We conducted a statistical assessment with the replicate number by using G*Power 3.1.9.2 with power (1-β) set at 0.95 and α = 0.05. According to the TG content in the liver (CON = 21.62 ± 2.85 mg/g; HELP = 42.45 ± 7.62 mg/g), we have calculated the effective size d = 3.62. A sample size of 8 participants (4 per group) was needed. Therefore, 12 participants (6 per group) provide sufficient power to study the molecular mechanism underlying the diet-induced FLS in the present study.

Histological evaluation
To visualize the hepatic fat droplets, fresh frozen liver samples were embedded in optimal cutting temperature (OCT) compound and sliced into 8 μm sections. The frozen sections were stained with oil red O (Sigma Aldrich, Saint Louis, MO, USA) for 30 min, counterstained with H&E for 30 s, then mounted in neutral resin. The slides were observed by using an optical light microscope (Olympus-BX53, Tokyo, Japan).

Determination of triglyceride content in liver
TG content in liver was measured by using TG assay kits (E1013, Applygen Technologies Inc., Beijing, China) following the manufacturer's instructions. Briefly, 50 mg of frozen liver sample was homogenized in 1 mL of isopropanol manually in a glass homogenizer with 10 passes on ice, incubated at 4°C for 10 min. The supernatants were collected and used to measure the hepatic TG contents following the instruction of the TG assay kit.

Determination of corticosterone content in plasma
CORT content in plasma was measured by using chicken CORT ELISA Kit (E-EL-0160c, Elabscience, TX, USA) following the manufacturer's instructions.
Total RNA isolation and real-time PCR Total RNA was isolated from liver sample (30 mg) using TRIzol Reagent (Invitrogen, Carlsbad, CA, USA) and reverse transcribed into cDNA by using HiScript II Q RT SuperMix for qPCR (+gDNA wiper) (R223-01, Vazyme, Nanjing, China). The coding sequences were used to design specific oligonucleotide primers (GenScript Biotech Co., Nanjing, China) for PCR ( Table 2) with AceQ qPCR SYBR Green Master Mix (Q111-02, Vazyme, Nanjing, China) on the Applied Biosystems QuantStudio 6 Flex Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). The relative mRNA abundance was calculated with the 2 −ΔΔCt method using GAPDH as an internal reference.

RNA m 6 A dot blot assays
For m 6 A dot blot, 500 ng RNA sample was denatured at 95°C for 5 min and transferred onto a Hybond-N + membrane (GE Healthcare, Piscataway, NJ, USA). After UV crosslinking, the membrane was washed with TBST buffer, blocked with 5% non-fat milk, and incubated with anti-m 6 A antibody (AB151230, Abcam, diluted 1:1,000) overnight at 4°C. Then, the membrane was incubated with secondary antibody at room temperature for 2 h. The signals were visualized by the chemiluminescence system (Bio-Rad, USA) and the dot density was analyzed with Quantity One software (Bio-Rad, USA), with staining of 0.02% methylene blue (in 0.3 mol/L sodium acetate, pH = 5.2) as loading control.

SELECT for detection of m 6 A
From the m 6 A-seq database obtained in a previous study on CORT-induced FLS chickens [17], sequences with m 6 A peaks were retrieved for lipogenic mRNAs including SREBP1, SCD and FASN, and subjected to specific m 6 A site analysis with SRAMP (http://www.cuilab.cn/ sramp). One very high/high confidence m 6 A site was selected for each gene and verified by using a single-base elongation-and ligation-based qPCR amplification method (termed as "SELECT").  Table 3.

Chromatin immunoprecipitation (ChIP) assay
ChIP was carried out as previously described [30]. Briefly, 200 mg frozen liver samples were ground in liquid nitrogen and washed with PBS containing protease inhibitor cocktail (Roche, Basel, Switzerland). After cross-linking in 1% formaldehyde, the reaction was stopped with 2.5 mol/L glycine. The pellets were lysed and chromatin was sonicated to an average length of ∼ 300 bp and the protein-DNA complex was diluted in ChIP dilution buffer, incubated with 2 μg of GR antibody (sc-1004, Santa Cruz, California, USA) overnight at 4°C. A negative control was included with normal IgG or no antibody. Protein G agarose beads (sc-2003, Santa Cruz, California, USA) were added to capture the immunoprecipitated chromatin complexes. Reverse cross-linking was performed at 65°C for 5 h to release DNA fragments from the immunoprecipitated complex and DNA was purified. The putative GREs in FTO, YTHDF1 and YTHDF2 promoters were predicted using JASPAR 2020 (http://jaspar.genereg.net) [31]. Immunoprecipitated DNA was used as a template for real-time PCR. The primers used to amplify the sequences covering these putative GREs are listed in Table 2.

Statistical analysis
Differences between two groups were analyzed by t-test using SPSS 20.0 software (SPSS Inc., Chicago, IL, USA). Data are expressed as means ± SEM. Pearson correlation analysis was performed for correlation analysis. The differences were considered statistically significant when P < 0.05.

Results
Lipogenic genes are activated in the liver of chickens fed HELP diet Chickens fed HELP diet had significantly higher hepatic lipid accumulation compared with their control counterparts, as seen in Oil Red O staining (Fig. 1A) and hepatic TG (Fig. 1B) content (P < 0.05). Meanwhile, plasma CORT concentration was significantly elevated (P < 0.05) in HELP group (Fig. 1C). Moreover, hepatic expression of lipogenesis genes, such as SREBP1, FASN and SCD were significantly up-regulated (P < 0.05) at both mRNA ( Fig. 1D) and protein (Fig. 1E) levels.

Levels of m 6 A on specific sites of lipogenic mRNAs are decreased in the liver of HELP diet-fed chickens
Specific m 6 A sites on 3'UTR of SREBP1, SCD and FASN mRNAs were selected for site-specific m 6 A quantification by SELECT method. Cycle threshold numbers were significantly increased (P < 0.05) at potential m 6 A site (X site, Fig. 4A, C and E), but not at the negative control site (N site, Fig. 3B, D and F), indicating decreased m 6 A modification on specific site of lipogenic mRNA 3'UTRs in the liver of HELP diet-fed chickens. ChIP-PCR analysis revealed changes of GR binding on the promoter of FTO and YTHDF2 genes. Fragments containing putative GREs on the promoter of FTO (Fig. 5A), YTHDF2 (Fig. 5B) and YTHDF1 (Fig. 5C) were amplified after chromatin immunoprecipitation with GR antibody. GR binding to the fragment of FTO gene promoter (Fig. 5A) was significantly increased (P < 0.05), while that to the fragment 1 of YTHDF2 gene promoter ( Fig. 5B) was significantly decreased (P < 0.05), in the liver of HELP diet-fed chickens. In contrast, GR binding to the fragment 2 of YTHDF2 or YTHDF1 gene promoter was not affected.

Discussion
Accumulating evidences indicate that FTO-dependent RNA demethylation and nonalcoholic fatty liver disease are closely intertwined [32,33]. FTO-dependent demethylation of m 6 A leads to an increase in lipogenic expression in hepatocytes through m 6 A modification [34,35]. Previously, we used chronic administration of corticosterone (CORT) to establish an in vivo FLS model in the juvenile chickens, and to induce excessive lipid accumulation in primary chicken hepatocytes in vitro with combined treatment of oleic acid and dexamethasone (OA/DEX) [17]. In both in vivo and in vitro models, we found that GR-mediated transactivation of FTO and m 6 A demethylation contribute to lipogenic gene activation. Interestingly, HELP diet-induced FLS is also associated with global m 6 A demethylation and the activation of lipogenic genes in the liver of laying hens. It may not be totally unexpected, because all these fatty liver models, no matter how they are induced, share the same hormonal and biochemical status of elevated CORT and lipid concentration in the blood. The major mechanism by which m 6 A exerts its effects is recruiting m 6 A-binding proteins [36]. m 6 A can be recognized by proteins that contain a YTH (YT521B homology) domain [37] or alternatively by eukaryotic initiation factor 3 (eIF3) [38,39]. The functions of m 6 Abinding proteins are context-dependent, which means that different m 6 A-binding proteins bind m 6 A on different regions of mRNA to exert different functions in gene regulation [40]. Among three m 6 A-binding proteins determined in the present study, YTHDF2 was downregulated at both mRNA and protein levels, indicating YTHDF2-mediated gene regulation. YTHDF2 binds transcripts carrying m 6 A in 3'UTR to induce mRNA degradation partially through recruiting the CCR4-NOT deadenylase complex [41]. YTHDF2 was reported to selectively recognize m 6 A sites in FASN mRNA, leading to increased FASN mRNA decay and decreased FASN protein content in HepG2 cell lines [42]. Previously, we conducted a m 6 A-seq analysis to elaborate the epitranscriptomic modification of m 6 A in the liver of CORT-induced FLS chickens [17]. From this published database, we selected some CORT-responsive m 6 A peaks in 3'UTR of the lipogenic transcripts, and identified specific HELP-responsive m 6 A sites on 3'UTR of lipogenic mRNAs, including SREBP1, SCD and FASN, with SELECT analysis. It is possible that the down-regulation of YTHDF2 impairs m 6 Adependent lipogenic mRNA degradation, which leads to augmented lipogenesis and excessive lipid accumulation in the liver of FLS hens. Fig. 4 3'UTR of lipogenic mRNAs is m 6 A hypermethylated in the liver of HELP diet-fed chickens. A and B detection of m 6 A modification in SREBP1 3'UTR using SELECT, n = 6. X site was predicted by SRAMP and N site (non-modification site) was negative control. C and D detection of m 6 A modification in SCD 3'UTR using SELECT, n = 6. X site was predicted by SRAMP and N site (non-modification site) was negative control. E and F detection of m 6 A modification in FASN 3'UTR using SELECT, n = 6. X site was predicted by SRAMP and N site (non-modification site) was negative control. CON = Control; HELP = High-energy low-protein. Values are means ± SEM, *P < 0.05 The molecular mechanism by which HELP-diet induces hepatic up-regulation of YTHDF2 in the chicken is unknown. YTHDF2 is negatively regulated by miR-145/miR-495/miR-493-3p at post-transcriptional level in hepatocellular carcinoma cells and prostate cancer cells [43][44][45]. Moreover, YTHDF2 can be SUMOylated in vivo and in vitro at the site of K571, which significantly increases its binding affinity with m 6 A-modified mRNAs [46]. HIF-2α is reported to transrepress YTHD F2 in hepatocellular carcinoma cells [47]. Based on the observation that YTHDF2 is down-regulated at both mRNA and protein level, we come up with a hypothesis that GR may directly transactivate YTHDF2. Silicon analysis using JASPAR online database identifies 2 putative GR binding sites for YTHDF2. These binding sites were then functionally validated using ChIP-PCR. Indeed, GR binding to YTHDF2 gene promoter is decreased in the liver of hens fed HELP diet. It remains a mystery how GR binding to the promoter of FTO and YTHDF2 genes is distinctively regulated, leading to FTO up-regulation and YTHDF2 down-regulation in HELP diet-induced FLS. Some unidentified co-factors must come into play to coordinate the down-stream effectors of GR action.
In this study, GR was upregulated at both mRNA and protein levels in the liver of hens fed HELP diet. The altered GR binding to the promoter of FTO and YTHDF2 genes indicate HELP diet-induced modulation in GR activation. GR can be activated via both ligand-dependent [48] and ligand-independent manners [49]. Previously, we reported CORT-dependent GR activation in the liver of CORT-treated chickens [16,50]. In this study, chickens are not treated with CORT and the plasma CORT level was not determined. Therefore, we cannot draw a conclusion whether the altered GR expression and binding is dependent on CORT. Nevertheless, GR can be activated by cellular stressors through p38 MAPKmediated phosphorylation of Ser134, which is a hormone-independent phosphorylation site on the human GR [51]. GR can also be activated by various stimuli in the absence of glucocorticoid ligands, such as elevated temperature, excessive inflammation, and cancer [52][53][54]. Therefore, both ligand-dependent and ligand-independent pathways are possible in HELP dietinduced alteration in hepatic GR activation.

Conclusions
The present results have shown that GR-mediated transcriptional regulation of FTO and YTHDF2 contributes to lipogenic gene activation by site-specific demethylation in HELP diet-induced chicken FLS. These findings add YTHDF2-mediated m 6 A modification as a new component of GR signaling in the regulation of fat metabolism in the liver and shed new light on developing effective therapeutic strategies in the prevention and treatment of HELP diet-induced chicken FLS.  A schematic representation of GR transcription factor binding promoter of FTO (up) and ChIP-PCR assay was used to measure the binding of GR on FTO promoter in the liver of HELP diet-fed chickens (down, n = 3). B schematic representation of GR transcription factor binding promoter of YTHDF2 (up) and ChIP-PCR assay was used to measure the binding of GR on YTHDF2 promoter in the liver (down, n = 3). C schematic representation of GR transcription factor binding promoter of YTHDF1 (up) and ChIP-PCR assay was used to measure the binding of GR on YTHDF1 promoter in the liver (down, n = 3). CON = Control; HELP = High-energy low-protein. Values are means ± SEM, *P < 0.05 TG: Triglyceride; WTAP: Wilms' tumor 1-associating protein; YTHDF1-3: YTHdomain family 1-3