Flavonoid quercetin modulates glucose metabolism by up-regulating the methylation status of genes as Kcnj11, Gys1 and Erp29 in liver of Wistar rats

Background: Quercetin is a avonoid that exists extensively in vegetables and fruits, and has many biological activities. It is reported that quercetin participates in the regulation of glucose metabolism through various mechanisms. However, whether DNA methylation is involved in those regulatory effects remains unclear. As liver is one of the main organ involved in methyl and glucose metabolism, DNA methylation targets related to glucose metabolism were identied in liver of Wistar rats upon quercetin exposure. Methods: The rats were fed a control diet or a 0.5% quercetin-supplemented diet for 6 weeks. Arraystar Rat 4 × 180K RN4 RefSeq Promoter Arrays were used for a genome-wide survey of DNA methylation in rat livers. NimbleScan v2.5 software was used to process microarray data. DAVID software was used to perform GO and Pathway analysis of regulatory networks. Gene promoter methylation status was examined by the ChIP-quantitative PCR assay, and hepatic levels of methylated Kcnj11, Erp29, Gys1, Flot1 and GAPDH were evaluated. Gene expression was assayed by quantitative PCR, and hepatic mRNA expression levels of Kcnj11, Gys1 and Erp29 were estimated. Results: Quercetin induced specic changes in DNA methylation. A total of 1,263 differentially expressed genes were found in 22 chromosomes, particularly on chromosomes 1, 3, 5, 7, 8, and 10. According to GO functional analysis, differential genes have focused on organic substance, cellular and primary metabolic process. According to pathway analysis, the most enrichment pathways included Type 2 diabetes mellitus, insulin signaling pathway and protein processing in endoplasmic reticulum. Nineteen up-methylated genes were found among several biological pathways after quercetin treatment. Critical genes and pathways associated with glucose metabolism (Kcnj11 and Gys1) and protein processing in the endoplasmic reticulum (Erp29) were changed signicantly. Promoter methylation levels of Kcnj11, Gys1, and Erp29 were signicantly increased, and the mRNA expression of those genes signicantly decreased simultaneously upon quercetin exposure. Conclusions: Quercetin changed the promoter methylation status and expression of Kcnj11, Gys1, and Erp29, which are mainly related to glucose metabolism. The gene Kcnj11, Gys1, and Erp29 could be novel epigenetic targets of quercetin in regulating glucose metabolism.


Introduction
Epigenetic changes are heritable changes to gene expression independent of changes to the DNA sequence [1]. DNA methylation is one of important form of epigenetic modi cation, and generally occurs at the cytosine residue in cytosine-guanine dinucleotide pairs to form 5-methylcytosine [2]. Unlike genetic modi cations, epigenetic modulation is reversible and can be altered by certain factors. This characteristic of epigenetic modulation may allow regulation of physiological responses to diet and environmental stimuli [3,4]. Evidence has shown that dietary polyphenols present in fruits, vegetables and beverages are potential epigenetic regulators [5,6]. It has been reported that the compound quercetin, which belongs to the avonoid subclass of polyphenols, may affect DNA methylation in therapy for cancer and chronic in ammatory disorders [7,8].
Quercetin is a representative compound of avonoid, and is ubiquitously present in Chinese diet such as fruits and vegetables. Quercetin has been reported to exert a wide range of biological effects, including antioxidant, anticarcinogenic, anti-in ammatory, and combatting gut dysbiosis activities [9][10][11]. Besides the above properties, quercetin also participates in regulating glucose metabolism. This compound modulates hyperglycemia by improving the pancreatic enzymes activities linked with glucose metabolism in rats with diabetes [12], and alters glucose homeostasis through insulin-dependent andindependent mechanism in the brain of diabetic rats [13]. In vitro experiments demonstrated that a mixture containing quercetin had the potential to modulate cellular glucose metabolism in human HepG2 cells [14], and could stimulate insulin release in rat INS-1 beta-cells [15,16]. However, whether DNA methylation plays a role in those effects of quercetin is not clear.
The liver plays a pivotal role in controlling glucose metabolism [17]. It can store glucose in the form of glycogen with feeding, assemble glucose via the gluconeogenic pathway in response to fasting, and also participates in metabolizing and releasing glucose [18]. The liver is also an important organ in which quercetin is metabolized to its methylated compounds [19,20]. It has been demonstrated that DNA methylation patterns exhibit a tissue speci city across human and mammalian animals [21,22]. The liver has the highest DNA methylation rate in sika deer [22], while the brain contains the highest levels of DNA methylation in human [23]. Several studies have revealed that DNA methylation plays a much larger role in regulation of tissue-speci c expression for genes [24][25][26]. Otherwise, quercetin improves the ultrastructure of hepatocytes and serum markers of liver injury in diabetic rats [27], and reduces the liver damage induced by drugs [28,29], chemicals [30][31][32][33][34], and physical factors [35]. As liver is one of the main organ involved in glucose and methyl metabolism, we want to identify DNA methylation targets related to glucose metabolism in liver of rats upon quercetin exposure.
Although nutrition affected global DNA methylation status throughout lifespan in mammalians [36], limited studies have reported how dietary quercetin modulates the tissue speci c epigenetic pro le. In this study, using rat DNA methylation promoter microarrays, we performed a genome-wide survey of DNA methylation in rat livers exposed to quercetin, and evaluated epigenetic alterations in speci c gene promoters by ChIP-quantitative PCR methods. We hypothesized that quercetin would induce the methylations of genes related to glucose metabolism, which could affect gene expression in liver of Wistar rats.

Animals handling
The animal experiments were approved by the Ethical Committee of the Department of Scienti c Management of the Institute. After adaptive feeding with AIN-93M formula [37] for 5 days, 24 Wistar rats were divided randomly into control or 0.5% quercetin groups according to body weight and maintained on an AIN-93 diet (control) or 0.5% quercetin-supplemented (Sigma-Aldrich, St. Louis, MO, SA) AIN-93 diet (0.5%Q) for 6 weeks. Each group consisted of 6 female and 6 male rats, whose initial weights ranged from 180 to 200 g. Dietary intake was recorded daily and body weight weekly. The animal source, handling methods, and environmental conditions matched a previously reported protocol [20]. At the end of the experiment, all rats were fasted overnight. The animals were sacri ced by cervical dislocation. Liver tissues were sampled immediately, washed in ice-cold saline, and frozen in liquid nitrogen until use.
DNA preparation and methylated DNA immunoprecipitation-ChIP analysis Liver samples from 2 female and 2 male rats in each group were randomly selected for genome-wide methylation analysis. Genomic DNA (gDNA) extraction, puri cation, quanti cation, and quality assessments were conducted according to the procedure of Aksomics, Ltd. (Shanghai, China). Sonicated gDNA was used for immunoprecipitation with a mouse monoclonal anti-5-methylcytosine antibody

Data Normalization and Analysis
The MeDIP-chip data were analyzed by sliding-window (1500 bp) peak-nding algorithm provided by NimbleScan v2.5 (Roche-NimbleGen Inc.) from the normalized log2 ratio data. NimbleScan detects peaks by searching for at least two probes above a p-value minimum cutoff (-log10) of 2 and maximum spacing of 500 bp between nearby probes within the peaks. To compare differentially enriched regions between the 0.5%Q group and the control group, the log2 ratios were averaged and then used to calculate M0 for each probe: M0 = Average (log2 MeDIP(1%Met)/Input(1%Met) )-Average(log2 MeDIP(control)/Input(control) ). The NimbleScan sliding-window peak-nding algorithm was run on these data to nd the differential enrichment peaks (DEPs). The DEPs, identi ed by the NimbleScan algorithm, were ltered according to Microarray data processing and Gene Ontology (GO) and Pathway analysis Raw microarray data were normalized by the Bioconductor packages Ringo, limma, and MEDME. Normalized MeDIP-chip data were analyzed by NimbleScan v2.5 (NimbleGen). DAVID software was used to perform GO and Pathway analysis of regulatory networks. The GO project provides a controlled vocabulary to describe gene and gene product attributes in any organism and covers three domains: Biological Process, Cellular Component and Molecular Function. Pathway analysis is functional analysis that maps genes to KEGG pathways. The p-value (EASE-score, Fisher-Pvalue or Hypergeometric-Pvalue) denotes the signi cance of the Pathway correlated to the conditions. Lower the p-value, more signi cant is the Pathway (The recommend p-value cut-off is 0.05), using an unbiased, automated survey of published scienti c literature (Global Literature Analysis). This analysis identi es functional relations among genes, such as direct binding, up-regulation or down-regulation and also builds subnetworks of genes and cellular processes based on their interconnections.

ChIP-quantitative PCR assay
A MeDIP assay combined with real-time quantitative PCR (qPCR) was used to evaluate the methylation status of candidate genes in the rat liver. MeDIP was performed as described above. Quantitative PCR was used to analyze the expression of puri ed DNA with an Applied Biosystems 7900 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). Expression levels of methylated Flot1, Kcnj11, Gys1, Erp29, and GAPDH were evaluated. The primers used for ChIP validation are shown in Table 1. Quantitative PCR analysis Hepatic mRNA expressions of genes were determined by quantitative PCR (qPCR) [38]. Total hepatic RNA was isolated using the TRIzol reagent. The rst cDNA strand was synthesized using a cDNA synthesis kit.
Quantitative PCR was performed using a FastStart Universal SYBR Green Master Mix kit. TRIzol reagent, cDNA synthesis kit, and FastStart Universal SYBR Green Master Mix kit were purchased from Roche, Ltd.
(Basel, Switzerland). Finally, melting curve analysis was performed by slowly cooling the PCR mixture with simultaneous measurement of the SYBR Green I signal intensity using an ABI Real-time PCR System (Applied Biosystems). The Δ threshold cycle (Ct) method was used to evaluate relative quanti cation, and GAPDH was used as a reference. The primers used for qPCR validation are shown in Table 2. The one-sided Kolmogorov-Smirnov test was applied to analyze the microarray data. Fisher's exact test was used to perform GO and Pathway analysis. ChIP-quantitative PCR and qPCR data are presented as means ± standard deviation. Statistical analysis was performed using the SPSS 10.01 software (SPSS Inc., Chicago, IL, USA). Student's t-test was used to compare differences between two groups. Differences between two groups were considered statistically signi cant at p < 0.05.

Quercetin does not change body weight or dietary intake
No signi cant difference was found in body weight or food intake between the control and quercetin groups (Fig. 1).
In the up regulation of GO biological process, differential genes have focused on cellular metabolic process and primary metabolic process (Fig. 3). In the down regulation of GO biological process, differential genes have focused on organic substance, cellular and primary metabolic process (Fig. 4). By pathway analysis of up regulation, the most enrichment pathways included Type 2 diabetes mellitus, insulin signaling pathway and protein processing in endoplasmic reticulum (Fig. 5).

Quercetin alters promoter methylation of speci c genes in biological pathways
When Peak Score was set at ≥ 3, Peak Length ≥ 500, and Peak M value higher than the median, we found that the promoters methylation status of 169 genes was changed by quercetin treatment, including 35 down-methylated genes and 134 up-methylaed genes. When those 169 genes were screened against genes in differential biological pathways, 19 up-methylated genes were identi ed after quercetin treatment ( Table 3). Five of the genes are associated with insulin and its signaling pathway, including Kcnj11 in type 2 diabetes mellitus, and Irs2, Flot1, Gys1, and Foxo1 in the insulin signaling pathway.
Erp29 is involved in protein processing in the endoplasmic reticulum. The promoter methylation levels of Kcnj11, Gys1, and Erp29 were signi cantly increased in the 0.5% quercetin group (Fig. 6), whereas the mRNA expression of those genes was notably decreased (Fig. 7). No difference was found in methylation status of the genes coding for Flot1 (Fig. 6).

Discussion
Our genome-wide analysis of the DNA methylation landscape in rats revealed that some genes were changed signi cantly by quercetin. According to GO functional analysis, differential genes have focused on organic substance, cellular and primary metabolic process. According to pathway analysis, the most enrichment pathways of Type 2 diabetes mellitus, insulin signaling pathway and protein processing in endoplasmic reticulum were changed by quercetin. Differentially methylated CpG sites were upmethylated in response to quercetin. Those sites were mainly located in genes involved in insulin metabolism, including Kcnj11, Irs2, Gys1, Flot1, and Foxo1. These results suggest that quercetin may regulate peptide hormone metabolism, especially insulin metabolism, by epigenetic mechanisms.
In the present study, we found for the rst time that Kcnj11, Gys1, and Erp29 were signi cantly upmethylated, while their mRNA expression levels were down-regulated in rat livers. Kcnj11, the potassium inwardly-rectifying channel, subfamily J, member 11 gene, encodes Kir6.2, a subunit of ATP-sensitive potassium (K ATP ) channels in mammalian cells. In pancreatic beta cells, K ATP channels play a pivotal role in glucose-stimulated insulin secretion [39][40][41]. It has been reported that Kcnj11 gene methylation might be associated with type 2 diabetes mellitus (T2DM). In patients with T2DM, one or more locus of the Kcnj11 gene was signi cantly up-methylated, and this alteration may mimic genetic defects [42,43]. However, experiments in vitro got different results. Insulin secretion was continuously increased in Kir6.2 −/− beta cells under unstimulated conditions [44]. Increased insulin secretion was also observed in cultured islets isolated from Kir6.2 −/− mice [45]. In vivo experiments con rmed that the hyperinsulinemic phenotype was based on incomplete loss of K ATP in beta cells [46]. For example, methylation at a speci c locus of the Kcnj11 gene reduced expression of the Kir6.2 subunit. In our experiment, we found that hepatic Kcnj11 was up-methylated and its mRNA expression down-regulated after quercetin treatment, indicating that Kcnj11 expression was incompletely lost, which might account for the hyperinsulinemic phenotype. Actually, quercetin stimulated insulin release in rat INS-1 beta-cells [15,16], and transient K ATP channel inhibition induced by DNA methylation may be one of the mechanisms.
The Erp29 gene encodes endoplasmic reticulum protein 29 (ERp29). ERp29 is a molecular chaperone that plays a pivotal role in protein secretion, folding, and tra cking. ERp29 is highly expressed in secretory tissues, including in the pancreas and liver [47]. High ERp29 protein expression was found in islets of transgenic MKR mice, and was associated with T2DM development [48]. Our results show that quercetin epigenetically down-regulates Erp29 expression in the rat liver. Therefore, we inferred that the Erp29 gene may be an important target of quercetin in preventing T2DM development. The Gys1 gene encodes glycogen synthase, which catalyzes the key step of glycogen synthesis in liver and muscle cells [49,50]. Under fasting conditions, lower glycemia and higher insulinemia were found in Gys1-knockout mice.
However, under normal conditions, the mice showed no abnormalities in glucose tolerance, insulin secretion, or basal glycemia [51]. In our research, we found that the Gys1 gene was epigenetically downregulated in healthy rats upon quercetin treatment. Considering the results from Gys1-knockout mice, we speculate that the effects of quercetin on the Gys1 gene may not cause adverse effects to healthy rats.

Conclusions
In the present study, we showed that quercetin is involved in the regulation of glucose metabolism by regulating the methylation level of Kcnj11, Gys1, and Erp29 genes, and also modulating related glucose metabolism pathways. Quercetin may be helpful to maintain the balance of blood glucose, and consumption of functional food and nutraceutical rich in quercetin could be a cheap and affordable method for the improvement of hyperglycemia.
We reported for the rst time that the effects of quercetin on methylation status of genes participating in the glucose metabolism in Wistar rats. To our knowledge, intervention studies of dietary factors on DNA methylation patterns in humans are very limited besides the study on folic acid [52]. Our results will provide a reference for future studies that examining the interplay of epigenetics and environmental factors in humans.