Gene and transposable element methylation in great tit (Parus major) brain and blood

Background: Studies on vertebrate DNA methylomes have revealed a regulatory role of tissue specific DNAmethylation in relation to gene expression. However, it is not well known how tissue-specific methylation variesbetween different functional and structural components of genes and genomes. Using whole-genome bisulfitesequencing data we here describe both CpG and non-CpG methylation profiles of whole blood and brain tissue inrelation to gene features, CpG-islands (CGIs), transposable elements (TE), and their functional roles in an ecologicalmodel species, the great tit (Parus major).Results: We show that hypomethylation at the transcription start site (TSS) is enriched in genes with functionalclasses that relate directly to processes specific to each tissue type. We find that 6877 (~21 %) of the CGIs aredifferentially methylated between blood and brain, of which 1186 and 2055 are annotated to promoter andintragenic regions, respectively. We observe that CGI methylation in promoter regions is more conserved betweentissues compared to CGI methylation in intra and inter-genic regions. Differentially methylated CGIs in promoterand intragenic regions are overrepresented in genomic loci linked to development, suggesting a distinct role forCGI methylation in regulating expression during development. Additionally, we find significant non-CpGmethylation in brain but not in blood with a strong preference for methylation at CpA dinucleotide sites. Finally,CpG hypermethylation of TEs is significantly stronger in brain compared to blood, but does not correlate with TEactivity. Surprisingly, TEs showed significant hypomethylation in non-CpG contexts which was negatively correlatedwith TE expression.Conclusion: The discovery that TSS methylation levels are directly linked to functional classes related to each tissueprovides new insights in the regulatory role of DNA-methylation patterns. The dominant sequence motifs for brainnon-CpG methylation, similar to those found in mammals, suggests that a conserved non-CpG regulatorymechanism was already present in the amniote ancestor. The negative correlation between brain non-CpGmethylation and TE activity (not found for CpG methylation) suggests that non-CpG is the dominant regulatoryform of methylation in TE silencing.