Validation of a DNA methylation microarray for 285,000 CpG sites in the mouse genome

Abstract

Mouse has been extensively used as a model organism in many studies to characterize biological pathways and drug effects and to mimic human diseases. Similar DNA sequences between both species facilitate these types of experiments. However, much less is known about the mouse epigenome, particularly for DNA methylation. Progress in delivering mouse DNA methylomes has been slow due to the currently available time-consuming and expensive methodologies. Following the great acceptance of the human DNA methylation microarrays, we have herein validated a newly developed DNA methylation microarray (Infinium Mouse Methylation BeadChip) that interrogates 280,754 unique CpG sites within the mouse genome. The CpGs included in the platform cover CpG Islands, shores, shelves and open sea sequences, and loci surrounding transcription start sites and gene bodies. From a functional standpoint, mouse ENCODE representative DNase hypersensitivity sites (rDHSs) and candidate cis-Regulatory Elements (cCREs) are also included. Herein, we show that the profiled mouse DNA methylation microarray provides reliable values among technical replicates; matched results from fresh frozen versus formalin-fixed samples; detects hemimethylated X-chromosome and imprinted CpG sites; and is able to determine CpG methylation changes in mouse cell lines treated with a DNA demethylating agent or upon genetic disruption of a DNA methyltransferase. Most important, using unsupervised hierarchical clustering and t-SNE approaches, the platform is able to classify all types of normal mouse tissues and organs. These data underscore the great features of the assessed microarray to obtain comprehensive DNA methylation profiles of the mouse genome.

Keywords: CpG sites; DNA methylation; Mouse; epigenetics; microarray; validation.

Figure 1.

Description and technical and biological validation of the 285,000 CpG sites mouse DNA methylation microarray. (a) Genomic and functional context of the 280,754 CpG sites contained in the Infinium Mouse Methylation BeadChip microarray: Chromosome location; Infinium design chemistry (Infinium I or II) of the probes; CpG content and neighbourhood context classified in CpG Island, shore, shelf, and other (open sea); functional genomic distribution of the CpG sites classified in gene body, TSS200, TSS1500, and intergenic; distribution among ENCODE candidate cis-Regulatory Elements (cCREs) with promoter-like signature (PLS), with enhancer-like signature (ELS), with distal enhancer location (dELS), with high H3K4me3 and low H3K27ac signal (DNase-H3K4me3), and with CTCF-only elements; and association with an ENCODE representative DNase hypersensitive site (rDHS). (b) Correlation plot of the CpG methylation values to show assay reproducibility of the measurements when using technical replicates on the mouse cell lines C2C12 and HAFTL. (c) Spearman’s correlation plot of the CpG methylation values obtained from two spleen fresh frozen (FF) samples when compared with their consecutive sections that were preserved as formalin-fixed paraffin-embedded (FFPE). (d) Correlation plot of the CpG methylation values to show DNA hypomethylation events in the mouse cell lines C2C12, HAFTL, and P19 upon the use of the demethylating agent 5-aza-2’-deoxycytidine (dAZA). (e) Correlation plot of the CpG methylation values to show DNA hypomethylation events in the mouse embryonic stem cells upon genetic knock-out of the maintenance DNA methyltransferase Dnmt1.

Figure 2.

DNA methylation atlas for mouse normal tissues. (a) Unsupervised hierarchical clustering and heatmap for 56 normal primary samples from 11 distinct source types. Tissue type and development layers are shown in the distinct colours as described in the figure legends. Methylation values are displayed from 0 (green) to 1 (red). (b) DNA methylation variances between mouse normal tissues and organs are displayed as t-distributed stochastic neighbour embedding (t-SNE) of Beta values. (c) Density plot of methylation Beta values showing their distribution from 56 normal tissue samples for all 226,000 CpG dinucleotides that remain after removal of erratic probe signals, XY chromosomes probes, and genotyping probes.

Figure 3.

Mouse DNA methylation mapping according to biological sex, X-chromosome, and imprinted CpG sites. (a) Unsupervised hierarchical clustering and heatmap for 56 normal primary samples from 11 distinct source types according to CpG sites located in the X-chromosome. Methylation values are displayed from 0 (green) to 1 (red). (b) Biological sex type is shown in distinct colours as described in the figure legends. DNA methylation variances between female and male mouse samples are displayed as t-distributed stochastic neighbour embedding (t-SNE) of Beta values. (c) Density plot of methylation Beta values showing their distribution from 56 normal tissue samples for the CpG dinucleotides located at the X-chromosome. (d) Density plot of methylation Beta values showing their distribution from 56 normal tissue samples for the CpG dinucleotides located at imprinted genes.

Figure 4.

Mouse strains according to the genotyping probes of the DNA methylation microarray. (a) Unsupervised hierarchical clustering and heatmap for 56 normal primary samples from four different mouse strains according to the SNP sites included in the microarray. Mouse strain is shown in different colours as described in the figure legends. (b) SNP genotyping among the mouse strains is displayed as t-distributed stochastic neighbour embedding (t-SNE) of Beta values.