DNA methylation abnormalities of imprinted genes in congenital heart disease: a pilot study

Background Congenital heart disease (CHD) is resulted from the interaction of genetic aberration and environmental factors. Imprinted genes, which are regulated by epigenetic modifications, are essential for the normal embryonic development. However, the role of imprinted genes in the etiology of CHD remains unclear. Methods After the samples were treated with bisulfate salt, imprinted genes methylation were measured by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. T test and One-way ANOVA were performed to evaluate the differences among groups. Odds ratios (ORs) were performed to evaluate the incidence risk of CHD in relation to methylation levels. Results We investigated the alterations of imprinted gene germline differential methylation regions (gDMRs) methylation in patients with CHD. Eighteen imprinted genes that are known to affect early embryonic development were selected and the methylation modification genes were detected by massarray in 27 CHD children and 28 healthy children. Altered gDMR methylation level of 8 imprinted genes was found, including 2 imprinted genes with hypermethylation of GRB10 and MEST and 6 genes with hypomethylation of PEG10, NAP1L5, INPP5F, PLAGL1, NESP and MEG3. Stratified analysis showed that the methylation degree of imprinted genes was different in different types of CHD. Risk analysis showed that 6 imprinted genes, except MEST and NAP1L5, within a specific methylation level range were the risk factors for CHD Conclusion Altered methylation of imprinted genes is associated with CHD and varies in different types of CHD. Further experiments are warranted to identify the methylation characteristics of imprinted genes in different types of CHD and clarify the etiologies of imprinted genes in CHD.

genes was found, including 2 imprinted genes with hypermethylation of GRB10 and MEST and 6 genes with hypomethylation of PEG10, NAP1L5, INPP5F, PLAGL1, NESP and MEG3. Strati ed analysis showed that the methylation degree of imprinted genes was different in different types of CHD. Risk analysis showed that 6 imprinted genes, except MEST and NAP1L5, within a speci c methylation level range were the risk factors for CHD Conclusion: Altered methylation of imprinted genes is associated with CHD and varies in different types of CHD. Further experiments are warranted to identify the methylation characteristics of imprinted genes in different types of CHD and clarify the etiologies of imprinted genes in CHD.

Background
Congenital heart disease (CHD) is the most common type of congenital malformations. The etiology for CHD is complex, due to the interaction of genetic aberration and environmental factors, profoundly in uenced by conditions in fetal life [1]. Despite the diversity of the CHD category, there appear to be shared epigenetic factors in the etiology of CHD for a couple of reasons. First, approximately 500 genes have been related to CHDs, but only accounting for 10% of CHD cases [2]. Second, recent research has shown that gene expression can be altered by chemical modi cations to the DNA and associated proteins in the nucleus, the epigenome. These modi cations can be altered in response to environmental factors, such as ambient air pollution and maternal cigarette smoking after adjusting for socioeconomic status, etc. [1].
Among the different epigenetic mechanisms, the potential relationship between abnormal DNA methylation and CHD has been increasingly recognized. DNA methylation has been found to be highly dynamic with the feature of demethylation in cardiomyocyte-associated gene sets during cardiomyocyte development [3]. Studies have shown that there are hypermethylation of myocardial-related genes in myocardial tissues of CHD, which is closely related to gene downregulation [4,5]. In patients with CHD, decreased transcriptional activity of CITE2, ZIC3, NR2F2 and BRG1 is associated with abnormal methylation [6][7][8][9]. DNA methylation has a major in uence on the establishment of imprinting markers [10]. Several human disorders have been found to be associated with varied methylation modi cations at imprinting control regions. For example, Silver-Russell syndrome is characterized by hypomethylation and Prader-Willi Syndrome by hypermethylation of the related imprinted genes [11]. Epigenetic imprinting is particularly vulnerable in early embryonic development. Exploring the epigenetic modi cation of imprinted genes may provide etiologies of congenital disorders including CHD.
In placental mammals, the imprinted genes regulate embryonic development. At present, about 94 human imprinted genes have been identi ed [12]. It has been proposed that IGF2 and GRB10 and co-regulated genes of PEG1/MEST, GTL2/MEG3, CDKN1C, PLAGL1 and DLK1 regulates embryonic growth and development by forming a proposed "imprinted gene network" [13]. The most well-studied example of involving in cardiac development is DLK1-DIO3, encoded ncRNAs, which participate in the commitment of the mesoderm to different subsets of a speci c cardiac cell lineage through many DLK1-DIO3 ncRNAs [14]. However, there has been no study on the epigenetic modi cation of imprinted genes in CHD.
Therefore, we investigated the alterations of imprinted gene methylation in patients with CHD through methylation modi cation. We selected 18 imprinted genes that are known to play an important role in early embryonic developments [15].
Altered methylation levels of imprinted genes in CHD classi cations CHD was classi ed into four groups including atrial and ventricular septal defect (ASD and VSD), valvular malformation (aortic and mitral stenosis or regurgitation), thoracic vascular malformation (coarctation of aorta, patent ductus arteriosus) and Tetralogy of Fallot (ToF). In ASD and VSD patients, 9 genes showed methylation changes, as compared with control group, whereby 2 imprinted genes (the GRB10 and MEST) were hypermethylated ( Fig. 2A Fig. S9A). In valvular malformation group, the methylations of 5 imprinted genes were lower than that of control group, including PEG10, NAP1L5, INPP5F, NESP and MEG3 ( Fig. 3; Additional le 9, Fig. S9B). In ToF group, methylation of 5 imprinted genes changed, including GRB10 with higher methylation than that in control group (  Fig. S9C). There were differences in methylation of 6 imprinted genes between thoracic vascular malformation and control groups, including hypermethylation of the GRB10 gene and hypomethylation of PEG10, INPP5F, NESP, MEG3 and KCNQ1OT1 respectively ( Fig. 5; Additional le 9, Fig. S9D). Taken together, 3 imprinted genes (PEG10, INPP5F and MEG3) were hypomethylated in all four CHD classi cations. NESP hypomethylation occurred in all CHDs except ToF, while GRB10 hypermethylation occurred in all CHDs except valvular malformation.
Risk Analysis between germline differential methylation regions (gDMRs) methylation levels of imprinted genes and CHD In order to develop a model for assessing the risk of CHD based on the amplitude of methylation levels of gDMRs of imprinted genes, CHD samples were categorized as Q1-Q4 according to the quartiles of methylation levels in control group. As shown in Table 2, the gDMRs of 8 differentially imprinted genes were further analyzed to examine the relationship between methylation changes of these genes and risk of CHD. Compared to control group, the gDMRs methylation level of NESP gene was lower in all CHDs. Methylation modi cation of two imprinted genes were elevated at 75th percentile level in CHD, with GRB10 in 21 of 25 patients and MEST in 10 of 20 patients. There were 4 imprinted genes whose methylation level of gDMRs were lower at 25th percentile level in CHD including14 of 25 cases of PEG10, 17 of 22 cases of INPP5F, 11 of 26 cases of PLAGL1 and 16 of 20 cases of MEG3 respectively. When logistic regression analysis was used with adjustment to sex and compared with control group, the risk of CHD was 21 times higher than that of control group when GRB10 methylation was higher than 44% P<0.05 . When the methylation levels of PEG10, INPP5F, PLAGL1 and MEG3 were lower than 45.2%, 70.39%, 40.76% and 41.83%, respectively, the risk of CHD increased by 16.8 P<0.05 , 11.06 P<0.01 , 11 P<0.05 and 12 P<0.01 times. Due to the small sample size, con dence intervals were wide (Table 3).

Discussion
It is well known that the epigenetic regulation of imprinted genes is involved in many embryonic developmental processes, including cardiac development [16,17].
At present, DNA methylation disorder is considered as one of the important factors in the occurrence and development of CHD [4]. However, the speci c role of the imprinted gene methylation regulation on cardiac development remains unclear. In this study, we found for the rst time the abnormal gDMR methylation changes of 8 imprinted genes (GRB10, MEST, PEG10, NAP1L5, INPP5F, PLAGL1, NESP and MEG3) in patients with CHD. The methylations of imprinted genes varied with the heterogeneity of CHDs. Six of them were associated with signi cantly higher risk of CHD except MEST and NAP1L5.
The etiology of CHD is complex, involving the interaction of environmental and genetic factors, leading to developmental phenotypes that regulate morbidity and severity. The potential relationship between alterations of genomic methylation and CHD is increasingly recognized. The dynamic changes of DNA methylation in cardiomyocyte-related gene sets during cardiac development suggest that DNA methylation modi cation is essential for the occurrence of cardiac diseases [3,4]. One study has con rmed that there are signi cant differences in methylation of genes associated with muscle contraction and cardiomyopathy in CHD [4]. Genomic imprinting is a process of epigenetic modi cation, and loss of imprinting may lead to abnormal embryonic development [13]. DMR plays an important role in the establishment of gene imprinting by targeting imprinted gene. Deletion or abnormal methylation of DMR might result in the expression disorder of the imprinted gene cluster [18]. Speci cally, we selected DMRs in the imprinting control region of imprinted genes studied, i.e., gDMRs, which are established during gametogenesis and play a key role in the imprinting of imprinted gene clusters [18,19]. The small changes in gDMRs found in our study might be relevant in the etiology of CHD, which was supported by the strati ed analysis. The relatively wide con dence interval in the risk analysis was due largely to the small sample size in this pilot study.
Some imprinted genes have key functions in uencing the proper embryonic growth and development by forming "imprinted gene network", including GRB10 and MEST as the core molecules [13]. GRB10, as an important growth-limiting factor, has a wide range of effects on embryonic development. Maternal GRB10 knockout results in embryonic overgrowth, and the change of ICR of GRB10 gene results in obvious dysplasia [20,21]. Previous studies have found an increase methylation of GRB10 in fetal samples of spontaneous abortion [22]. It has been found that the Ddc gene in mice, involved in the development of trabecular cardiomyocytes of the embryonic and neonatal heart, displays tight conserved linkage with the methylated GRB10 gene [23]. In the present study, we found that the gDMR methylation of GRB10 imprinted gene in CHD was increased except for valvar defects. It is hinted the methylation disorder of GRB10 may be involved in the development of heart through the interaction with Ddc. Hypermethylation of GRB10 might inhibit its expression and regulate the growth through insulin pathway such as IGF2, potentially leading to cardiac dysplasia. MEST is a maternal imprinted gene and widely expressed throughout the embryonic period [24]. Mest-knockout mice exhibited embryonic and placental growth retardation and postnatal growth inhibition, while loss of the imprinted gene resulted in postnatal weight gain and multiple organ hypertrophy [25,26]. King et al. have found that Mest speci cally expressed myocardial trabeculae in developing atria and ventricles of mice. Mest-knockout mice showed subtle changes of myocardial trabeculae, which displayed an increase in thickness and reduction in density of the compact myocardium, similar to that in human heart disease, which hinted MEST is closely related to heart disease [27]. The increased methylation of gDMR of MEST imprinted gene in CHD was also found in our data, especially in ASD and VSD. Potentially, MEST expression disorder via altered methylation of the gene might be involved in the occurrence of CHD by affecting the development of myocardial trabecula.
PEG10, a paternally expressed imprinted gene, is expressed in embryonic tissues and placenta, and participates in cell proliferation, differentiation and apoptosis [28,29]. The Peg10-knockout mice exhibited early embryo death and placental defects [30]. A study has found that the expression of PEG10 in the heart was very low, indicating that trace PEG10 could maintain the normal development of the heart [28]. On the contrary, we found abnormal hypomethylation of PEG10 in CHD. Overexpression of PEG10 may be involved in the invasion and metastasis of malignant tumors, such as hepatocellular carcinoma and endometrial cancer etc., through the epithelial mesenchymal transition (EMT) [31,32]. Therefore, the low methylation of PEG10 might cause abnormal elevation of expression, and induce abnormal migration of related cells through EMT pathway, potentially leading to abnormal cardiac development.
INPP5F encodes inositol 1,4,5-trisphosphate (InsP3) 5-phosphatase, which is an important functional endogenous regulator. It has been found that INPP5F regulated the size of cardiac myocytes and cardiac stress response, increased the hypertrophy and activation of the fetal gene program in the stress response in Inpp5f-knockout mice, and decreased the hypertrophic response in Inpp5f-overexpressed mice [33,34]. INPP5F can be used as a negative feedback regulator of insulin signal and downregulation of INPP5F in diabetes mellitus has cardio protective effect [35]. Consistent with previous studies, we found that INPP5F methylation was reduced in a variety of CHDs and increased CHD risk, suggesting that INPP5F might be a protective factor for cardiac development.
MEG3 is a maternal-expressed imprinted gene that is expressed in many normal tissues. It inhibits the proliferation of tumor cells in vitro. It also interacts with tumor suppressor p53 to regulate the expression of p53 target genes [36]. Consistent with this effect, MEG3 has an impact on cardiac remodeling induced by pathological hypertrophy by regulating the binding of p53 to other gene promoters [37]. Other studies have shown that MEG3 participates in the development of myocardial brosis and prevents myocardial remodeling by regulating the production of MMP-2 by CFs in vitro and in vivo [38]. In consistency to previous studies, our results also found that methylation of MEG3 was reduced and associated with the increased risk of CHD.
We also detected hypomethylation of gDMRs of other 3 imprinted genes including NAP1L5, PLAGL1 and NESP in CHD. Strati ed analysis revealed that different CHDs showed speci c methylation patterns, such as hypomethylation of the 3 genes in ASD and VSD, hypomethylation of NAP1L5 and NESP in valvular malformation, hypomethylation of NESP in thoracic vascular malformation, and no signi cant change in ToF.
As a pilot study, it is subject to a couple of limitations. First, we studied a small and heterogeneous group of patients with CHD. A large sample size is warranted to identify the methylation characteristics and risk factors of imprinted genes in relation to the types of CHD. Second, the causal relationship between methylation alterations of imprinted genes and CHD remains to be explored in further experimental studies to examine their potential impact on RNA expression and CHD development on cellular and organ levels in animal models.

Conclusion
Methylation of gDMRs of 8 imprinted genes was altered in white blood cells in patients with CHD. Further experiments are warranted to identify the methylation characteristics of imprinted genes in different types of CHD and clarify the pathogenesis mechanism of imprinted genes in CHD.

Sample Collection
The study was approved by the Ethics Committee of the Capital Institute of Pediatrics. Written informed consent was obtained from parents of the children. These samples were obtained from children with CHD who visited the Capital Institute of Pediatrics during the biennium 2014-2015, and were diagnosed as CHD by echocardiography and other related tests, ranging in age from 1 to 120 months. According to the International Classi cation Standard of Diseases, the diagnosis of existing cases was strictly classi ed, and a complete database of CHD research is established. Moreover, the children with CHD were excluded from other congenital and acquired abnormal diseases, such as congenital mental retardation, genetic metabolic diseases, etc. As control group, 28 controls of blood samples were selected from the physical examination children. They were same age and race and had no developmental deformities or other diseases. Peripheral blood samples were collected from 27 children with CHD and 28 healthy children as controls in Capital Institute of Pediatrics (Table 3). In details, 2 ml venous blood was collected into the evacuated tubes without anticoagulant (Becton Dickinson). Blood samples were immediately centrifuged at 2500 rpm for 10 min. The separated plasma was aliquot without reducing agent and stored at -20℃.

DNA extraction
Genomic DNA was extracted from peripheral blood with the Blood and Tissue DNA Kit (65904, QIAGEN, Dusseldorf, Germany) according to manufacturer's instructions. The concentration and purity of the DNA were determined by nanodrop and agarose gel electrophoresis. DNA with an OD260/280 absorbance ratio of 1.8 was used for subsequent work.

Bisul te treatment
A total of 500ng genomic DNA from each sample underwent bisul te treatment using the EZ DNA methylation kit (D5001, Zymo Research, Irvine, CA, USA) according to the manufacturer's instruction. The quality of the bisul te conversion was controlled by using PCR products that had no methyl group. Sequencing results con rmed that 96.6% of cytosine residues were converted in previous research [39].

PCR ampli cation
Eighteen imprinted genes were selected according to WAMZDEX, Gene imprint, Catalogue of Imprinted Genes and other imprinted gene analysis resources (Additional le 10, Table S1). gDMRs sequences of the selected imprinted genes were retrieved to con rm using the UCSC database (Additional le 11, Table  S2). The primers were designed for the quantitative analysis of the methylation level with Methprimer (http://epidesigner.com/). An additional T7 promoter tag was added to each reverse primer for in vivo transcription, and a 10-mer tag to the forward primer to balance the melting temperature.

Sap reaction
The sap reaction was performed to eliminate residual dNTP in PCR products to facilitate mass spectrometry detection. The speci c steps were to add 2ul sap mixture into 5ul PCR product, and to conduct vortex mixing before centrifugation. The reaction conditions were 37℃ for 20 min and 85℃ for 5 min.

Methylation analyses
As the previously published method [39], the Sequenom MassARRAY platform (CapitalBio, Beijing, China) was used to perform the quantitative methylation analysis of the imprinted genes. This system uses matrix-assisted laser desorption/ionization time-of-ight (MALDI-TOF) mass spectrometry in combination with RNA base-speci c cleavage (MassCLEAVE). A detectable pattern was then analyzed for its methylation status. The spectral methylation ratios were generated using Epityper software version 1.0 (Sequenom, San Diego, CA, USA)

Statistical analyses
Data are presented as mean and standard deviation. The methylation levels in the gDMRs of imprint genes were compared between CHD and control groups by independent samples t-test. One-way ANOVA was performed to evaluate the differences among different CHD subtypes and control groups. Odds ratios (ORs) was performed to evaluate the incidence risk of CHD in relation to methylation levels. P < 0.05 was considered signi cant. GraphPad Prism 7 software (GraphPad Software, San Diego, California, USA) was used to visually display the results of analysis. Data were stored in the EPI 3.  Others included pulmonary vascular obstruction (n=2) and congenital diaphragmatic hernia (n=1); valvular malformation included aortic (n=4) and mitral (n=3) valve stenosis or regurgitation. e, 75th percentile. Cut-off value was defined as 25th and 75th percentiles of the control group methylation level. Adjusted for sex by logistic regression. f indicates that the maximum value of the imprinting gene cases was less than 75% of control group and could not be calculated, so there were divided into two groups. e, 75th percentile. Cut-off value was defined as 25th and 75th percentiles of the control group methylation level. Adjusted for sex by logistic regression. f indicates that the maximum value of the imprinting gene cases was less than 75% of control group and could not be calculated, so there were divided into two groups.