DNA methylation profile dynamics of tissue-dependent and differentially methylated regions during mouse brain development

Distinct DNA methylation profiles in NPCs with different fates

To explore DNA methylation profiles of mouse NPCs, we compared neurospheres (NSph) derived from telencephalons at embryonic days 11.5 (E11.5NSph) and 14.5 (E14.5NSph) by T-DMR profiling with restriction tag-mediated amplification (D-REAM) with mouse promoter tiling arrays covering from 6 kb upstream to 2.5 kb downstream of 30,140 gene TSSs (Ensembl Transcript IDs)[7]. The distinctive cell fates of E11.5NSph and E14.5NSph were indicated by biased expression of marker genes for neural and oligodendrocyte progenitor in the undifferentiated NSph, and those of neuronal and glial marker genes in the differentiated ones, respectively (Figure 1A and Additional file1: Figure S1). We screened genomic regions that exhibited differential MATscores[7, 21] between NSphs, which indicate differential methylation status, and obtained a total of 1,403 NSph-T-DMRs consisting of 380 E11Hypo-T-DMRs and 1,023 E14Hypo-T-DMRs, which were hyper- and hypomethylated, respectively, in E14.5NSph compared to E11.5NSph.

The localization patterns along the genome were distinct between E11Hypo-T-DMRs and E14Hypo-T-DMRs. The former exhibited bimodal distributions within 2.5 kb from TSSs and biased to the CGI genes, which contain CGIs around TSSs[7]; the latter were located 6 kb to 2 kb upstream of TSSs without any promoter type bias (Figure 1B). These findings are noteworthy because they indicate that methylation changes occur in regions around high-CpG promoters. We analyzed E11Hypo-T-DMR 3^′ downstream from the Nrcam TSS, and E14Hypo-T-DMR 5^′ upstream from the Kat5 TSS. Combined bisulfite restriction analysis (COBRA) of these T-DMRs indicated differential DNA methylation status as indicated by D-REAM (Figure 1C and1D).

Among the genes with NSph-T-DMRs, we identified human gene orthologs involved in neurological diseases, such as spinocerebellar ataxia type 1 (ATXN1 and KAT5), schizophrenia (BLOC1S1, NTNG1, and OLIG2), autism (NRCAM), and brain malformation syndrome (LIG4 and NFIA) (Additional file2: Table S1). We performed bisulfite sequencing of 1 E11Hypo-T-DMR (Atxn1) and 3 E14Hypo-T-DMRs (Lig4, Nfia, and Olig2) that were located at various relative positions from the TSSs, including one associated with an alternative TSS and those facing their TSS in proximal regions. The results showed clear differences in the DNA methylation statuses of CpG sites around the HpyCH4IV sites between the 2 types of NSphs (Figure 1E). These data indicate that methylation changes occurred in a subpopulation of NSphs in a gene-dependent manner.

Association of NSph-T-DMRs with neural development and function

DNA methylation profile of NSph-T-DMRs in the AdBr

We compared D-REAM data between NSphs and the AdBr and used K-means clustering to classify E11hypo- and E14hypo-T-DMRs into 3 clusters. In the AdBr, most E11Hypo-T-DMRs and E14Hypo-T-DMRs exhibited hypomethylation (clusters 2 and 3) (Figure 2A). Although the degrees of differences varied among genes, COBRA of NSph-T-DMRs indicated hypomethylated status at some loci as clusters 2 and 3 (e.g., Dclk1 and B3gnt2 for E11Hypo-TDMR, and Rdh5/Bloc1s1 and Mcf2l for E14Hypo-T-DMRs), and hypermethylated status at other loci as cluster 1 (e.g., Cdkn2a and Ntng1 for E11Hypo- and E14Hypo-T-DMRs, respectively) in the AdBr (Figure 2B and Additional files3 and4: Tables S2 and S3).

Among genes with cluster-1 E14Hypo-T-DMRs, we unexpectedly found that T-DMR hypermethylation was associated with higher gene expression in the brain (described later). To address this issue, we further investigated the DNA methylation status of other HpyCH4IV sites in these genes using AdBr D-REAM data and found AdBr-specific hypomethylated T-DMRs 3^′ downstream of their TSSs in Ntng1, Aldh1a2, and Accn1 (Figure 3A). It is noteworthy that all these T-DMRs were located within few kb from CGIs.

The positional changes of hypomethylated T-DMRs in a specific genomic region are summarized in Figure 3B. Bisulfite sequencing analysis of T-DMRs in the Ntng1 gene indicated hypermethylation of E14Hypo-T-DMRs at the 5^′-upstream region and hypomethylation at the 3^′ downstream of the TSS in the AdBr with unmethylated neighboring regions in all samples (Figure 3C). Quantitative reverse-transcription polymerase chain reaction (Q-RT-PCR) data indicated negative correlation between hypomethylation at distal T-DMR (region 4) in undifferentiated NSphs, and an association of gene expression in AdBr with hypomethylation of the T-DMR 3^′ downstream of the CGI (Figure 3D). These results highlight functions associated with developmental stage-dependent multiple T-DMRs in a gene region.

Tissue samples, neurosphere culture, and immunocytochemistry

All experiments using mice were carried out according to the institutional guidelines for the care and use of laboratory animals (Graduate School of Agricultural and Life Sciences, the University of Tokyo). Pregnant C57BL/6 N mice were euthanized, and fetuses were recovered in ice-cold phosphate-buffered saline (PBS) containing 0.6% glucose. For neurosphere culture, dissected telencephalons were dispersed, and were suspended in progenitor cell culture medium: Dulbecco’s modified Eagle’s medium (DMEM)/F12 (1:1) containing 5.5 mM HEPES, 2 mM L-glutamine, 1/50 volume of B-27 Supplement (Invitrogen), 20 ng/ml epidermal growth factor (EGF), 20 ng/ml human basic fibroblast growth factor (bFGF) (PeproTech), and 5 μg/ml heparin. Cells were seeded into a petri dish and cultured for 6 days, replacing half of the medium with fresh medium at day 3. To induce differentiation, cells were dispersed, suspended in differentiation medium (progenitor cell culture medium without EGF, bFGF, and heparin), and seeded onto poly-L-lysine- and laminin-coated coverslips for immunocytochemistry. After 4 days of culture, cells were fixed with 4% paraformaldehyde and stained with a monoclonal antibody against βIII-tubulin (Covance) and a rabbit polyclonal antibody against glial fibrillary acidic protein (DAKO).

Genomic DNA extraction

Cells (1.5 × 10⁶) were incubated in 200 μl of lysis solution (10 mM Tris–HCl (pH 8.0), 5 mM EDTA, 200 mM NaCl, 0.2% SDS and 200 μg/ml proteinase K) at 55°C for 30 min. The samples were extracted with phenol/chloroform/isoamyl alcohol (PCI; 25:24:1), incubated with RNase for 30 min, and extracted again with PCI. Genomic DNA was precipitated with ethanol and dissolved in 20 μl of TE (pH 8.0).

D-REAM

D-REAM was performed as previously described[7]. Genomic DNA (5 μg) was digested with HpyCH4IV (New England Biolabs), extracted with PCI and chloroform, ethanol-precipitated, and dissolved in TE (pH 8.0). DNA sample (50 ng) was ligated to the R-adaptor pair using T4 DNA ligase (New England Biolabs) at 16°C overnight. After the 5^′-overhang of the adaptor was filled in with Klenow Fragment, the DNA was digested with TaqI at 65°C for at least 1 h and purified with a Microspin S-300 HR Column (GE Healthcare). The TaqI ends of the DNA were ligated to the N-adaptor pair. The resulting DNA sample was purified with the Wizard SV Gel and PCR Clean-up System (Promega) and amplified with the R18 and N18 primers and Immolase Taq DNA polymerase (Bioline) under the following conditions: denaturation at 95°C for 7 min, followed by 20 cycles of 30 sec at 95°C, 30 sec at 62°C, and 2 min at 72°C. DNA was purified with MinElute PCR Purification Kit (Qiagen), and 7.5 μg of DNA was used for microarray analysis. Microarray analysis was conducted with GeneChip System (Affymetrix), and all procedures were done according to the Affymetrix Chromatin Immunoprecipitation Assay Protocol. DNA samples were labeled with the GeneChip WT Double-Stranded DNA Terminal Labeling Kit and hybridized with GeneChip Mouse Promoter 1.0R Arrays. Arrays were stained and washed with GeneChip Fluidics Station 450 and scanned with GeneChip Scanner 3000 7 G. The instruments were operated with GeneChip Operating Software version 1.4. D-REAM data obtained in this study have been deposited in the ArrayExpress database (accession number E-MTAB-1150). The D-REAM dataset of the adult whole cerebrums (AdBr), which were obtained from 13 week-old male mice, is reported previously[7].

Data analysis

D-REAM data for two experiments were obtained for each NSph. The data were visualized using the Integrated Genome Browser (Affymetrix). Ensembl Transcript IDs (release 46) associated with T-DMR were obtained using BioMart[39] and Galaxy website[40]. Distribution analysis was conducted with the R software package. K-means clustering of MATscores was performed with the MultiExperiment Viewer (MeV in TM4 Microarray Software Suite)[41]. Gene Ontology analysis was conducted using the DAVID Bioinformatics Resources website[42].

Combined bisulfite restriction analysis (COBRA) and sequencing

PstI- or EcoRV-digested genomic DNA (3 μg) was denatured with 0.3 M NaOH. Sodium metabisulfite (pH 5.0) and hydroquinone were added to final concentrations of 2.0 M and 0.5 mM, respectively. The reaction mixtures were incubated in the dark at 55°C for 16 h. The DNA was purified with the Wizard DNA Clean-up System (Promega), treated with 0.3 M NaOH at 37°C for 15 min, and ethanol-precipitated. The DNA was dissolved in 20 μl of TE (pH 8.0). After the bisulfite reaction, the unmethylated CpGs are converted to uracil-phosphate-guanines (UpGs), whereas the methylated CpGs remain intact. One-hundredth to 1/20 amount of the DNA was used for PCR with Immolase Taq DNA polymerase. For COBRA, one-tenth of the PCR product was digested with HpyCH4IV at 37°C overnight and electrophoresed with untreated control in a 2% agarose gel. For sequencing, PCR product was purified with the Wizard SV Gel and PCR Clean-up System (Promega) and cloned into pGEM-T Easy Vector (Promega). Up to 16 clones were sequenced. Primer sets used are listed in Additional file5: Table S2.

RT-PCR

Total RNA was extracted with the TRIzol Reagent (Invitrogen), and 1 μg of total RNA was subjected to reverse transcription using the Superscript II First-strand Synthesis System (Invitrogen). One-hundredth of the cDNA was used for PCR with Immolase Taq DNA polymerase under the following conditions: denaturation at 95°C for 7 min and 23 or 35 cycles of 30 sec at 95°C, 30 sec at 62°C, and 30 sec at 72°C (Additional file5: Table S2). Quantitative RT-PCR was carried-out on Bio-Mark HD system (Fluidigm) using Universal probes (Roche Applied Science) for monitoring amplifications (detailed in Additional file6). Makers were selected according to the previous report[43].