The MIRA microarray does not require restriction endonuclease, antibody or bisulfite treatment of the genomic DNA and, therefore, offers many advantages over existing methods for the genome-wide screening of DNA methylation. MIRA microarrays have been used to identify candidate methylation biomarker for cancer diagnosis
[12, 25]. In this study, we applied this approach to seek candidate biomarkers for noninvasive prenatal diagnosis. Currently, noninvasive prenatal screening typically involves a combination of ultrasound tests and the measurement of non-specific maternal serum markers. These screening tests are limited to trisomies of chromosomes 21 and 18 and do not reliably diagnose or exclude these types of abnormalities. The discovery of fetal DNA in maternal plasma opened new doors for non-invasive prenatal diagnosis. However, the presence of background maternal DNA interferes with the analysis of fetal DNA which usually constitutes less than 10% of the total circulating free DNA in early pregnancy
. This has posed significant technical hurdles for the detection of fetal genetic loci that are not completely absent from the maternal genome using current PCR-based approaches
. To overcome this problem, the differential methylation between placentally and maternally derived cell-free DNA sequences has been investigated
. Many studies have shown that these epigenetic differences may serve as potential fetal molecular markers for noninvasive prenatal diagnosis
[7, 8]. However, only a limited number of genomic regions have been identified or tested so far and the majority of studies have focused only on fetal chromosomal aneuploidy detection, for example, of chromosomes 21 and 18. Here, we aimed to identify methylation biomarkers for the prenatal diagnosis of not only fetal chromosomal aneuploidies but also of monogenic diseases at a genome-wide level.
Promoter methylation plays important roles in regulating gene expression both in development and in human disease and, so far, DNA methylation studies have mainly been focused on the promoter regions of genes. Recently, methylation of the gene body (sometimes called intragenic methylation) has been reported to play a role in transcriptional regulation and efficiency
 and intragenic methylation is attracting increasing attention. Therefore, in this study, we investigated the hypermethylation of genes based on the methylation status of both the promoter and the gene body.
Large differences were observed in methylation patterns between maternal peripheral blood and placental tissue. Further analysis based on GO terms revealed that many of the differentially methylated genes were involved in regulation of transcription and multicellular organismal. This result might suggest that the differentially methylated genes contribute to the control of gene expression during embryonic development. It is well known that DNA methylation changes during embryonic development are frequent events that play major roles in regulating gene expression and other developmental processes. It is worth noting that differentially methylated genes were involved in mammary gland development (Figure
2). This finding suggested that methylation may play major roles in regulating lactation.
The bisulfite sequencing and COBRA assays both confirmed that the observed genes were hypermethylated in placental tissue and chorionic villus; however, the DNA methylation was unobvious in the amniotic fluid samples. The placental tissue samples, obtained immediately after delivery, are from a late-stage placenta and the chorionic villus tissue samples are from an early-stage placenta, the amniotic fluid is not a placental tissue. The cells in amniotic fluid are an admixture of various cells from various fetal tissues, mainly fibroblasts, epithelial cells and amniocytes that are shed from the fetus surface. Therefore, we concluded that the observed methylation differences in these samples may reflect tissue specificity rather than developmental specificity.
Previous studies have shown that the circulating fetal DNA in maternal peripheral blood are mainly from the placenta
 because the placenta is the only channel for nutrient transport between mother and fetus. The DNA in different tissues carries the same sequence information, no matter whether the sequences are methylated or not. Therefore, the detection of fetal methylated DNAs in maternal peripheral blood will yield fetal information about genetic variations that may be useful for the diagnosis of fetal genetic diseases. Furthermore, it is convenient and feasible to discriminate between fetal methylated DNA and maternal nonmethylated DNA in maternal peripheral blood based on differences in their methylation patterns. Thus, the observed methylation differences of disease-associated genes between placental tissue and maternal peripheral blood provide a foundation for developing novel methods for the detection of fetal genes in maternal peripheral blood.
We identified a large number of hypermethylated genes in fetal tissues; most of these genes have been recorded in the Online Mendelian Inheritance in Man (OMIM) database (http://www.ncbi.nlm.nih.gov/omim) where the relationship between abnormalities in these genes and diseases has been defined. For example, mutations in PITX2, a homeobox gene, are known to contribute to Axenfeld-Rieger syndrome (ARS), an autosomal-dominant developmental disorder
[31, 32]. The hypermethylated genes have great potential to be developed into molecular markers for noninvasive prenatal diagnosis of monogenic disorders. In a future study, we will use the MBD protein to enrich fetal hypermethylated DNA fragments in maternal peripheral blood and further explore the feasibility of using these hypermethylated genes as biomarkers for noninvasive prenatal diagnosis in large samples.