Active N6-Methyladenine Demethylation by DMAD Regulates Gene Expression by Coordinating with Polycomb Protein in Neurons

Abstract

A ten-eleven translocation (TET) ortholog exists as a DNA N⁶-methyladenine (6mA) demethylase (DMAD) in Drosophila. However, the molecular roles of 6mA and DMAD remain unexplored. Through genome-wide 6mA and transcriptome profiling in Drosophila brains and neuronal cells, we found that 6mA may epigenetically regulate a group of genes involved in neurodevelopment and neuronal functions. Mechanistically, DMAD interacts with the Trithorax-related complex protein Wds to maintain active transcription by dynamically demethylating intragenic 6mA. Accumulation of 6mA by depleting DMAD coordinates with Polycomb proteins and contributes to transcriptional repression of these genes. Our findings suggest that active 6mA demethylation by DMAD plays essential roles in fly CNS by orchestrating through added epigenetic mechanisms.

Keywords: 6-methyladenine; Polycomb; epigenetics; neuron.

Figure 1.. DMAD demethylates intragenic 6mA in Drosophila brains.

(A) qRT-PCR across fly tissues revealed high levels of DMAD expression in heads compared to other somatic tissues (n=3).(B) High resolution HPLC quantification of 5hmC and 6mA in Drosophila brains in the presence (Ctrl) and absence of DMAD (DMAD-null). 5hmC/total C or 6mA/total A are shown as percentage per million nucleotide (ppm). There was a ~3.6-fold increase in 6mA with DMAD depletion while 5hmC was undetectable (n=2).(C) Dot blots using an antibody specific for 6mA confirmed the accumulation of 6mA in DMAD-null fly brains.(D) Confocal images of adult brains stained with anti-Elav (Green), anti-DMAD (Red) and anti-6mA (Purple) in the background of elav-Gal4 alone (i & iii) or in combination with UAS-DMAD miRNA (DMAD-nKD) (ii & iv). Enlarged views of arrowed area in iii and iv are shown below. Compared to control brains (elav-Gal4 alone), DMAD knockdown significantly increased nuclear 6mA levels in elav-expressing cells.(E) Genomic annotation of gain-of-6mA regions in DMAD-null fly brains revealed their intragenic characteristics, A heatmap shows the enrichment of each genomic feature versus expected values.(F) Plot of the global transcriptome in control and DMAD-null fly brains obtained from RNA-seq (n=2). Genes bearing gain-of-6mA regions in their gene bodies were highlighted (up-regulated genes in pink and down-regulated genes in purple).(G) Gene Ontology analysis was performed on a subset of downregulated genes in Figure 1F (purple). Log2 fold change [(KD: WT) < −0.5] was applied as the threshold cut-off. Several biological processes involved in neurodevelopment and neuronal functions were enriched and are highlighted in red. Data are represented as mean ° SEM.

Figure 2.. 6mA is enriched at Polycomb-binding sites in neuronal cells.

(A) qRT-PCR validated a 70% reduction in DMAD mRNA levels after double-strand siRNA knockdown in BG3C2 cells.(B) Western blot using a DMAD-specific antibody confirmed effective DMAD knockdown in BG3C2 cells.(C) Dot blots using a 6mA-specific antibody confirmed 6mA accumulation in the absence of DMAD in BG3C2 cells.(D) GO analysis showed specific enrichment for neurodevelopment and neuronal functions from downregulated genes carrying intragenic BG3C2 gain-of-6mA regions.(E) Average fold change in 6mA mapped reads versus non-enriched input DNA was calculated for various binned ChIP-chip regions of epigenetic regulators available from the modENCODE database. Average fold change is plotted in Heatmap view. Red (fold change >1) indicates enrichment over input while blue (fold change <1) indicates depletion. 6mA was explicitly enriched at Polycomb protein binding sites. Enrichment and depletion were significant with p-value < 0.001, Welch Two sample t-tests. Data are represented as mean ° SEM.

Figure 3.. DMAD demethylates intragenic 6mA in cis and coordinates with Trithorax-related protein Wds to regulate gene expression.

(A) Average fold changes in 6mA reads between DMAD-KD and control were calculated for DMAD-occupied gene introns, exons, and untranslated regions. Enrichment or depletion of 6mA on these genomic regions were significant (p<0.001 or p<0.05, Welch Two sample t-tests) (B) Intragenic distributions of DMAD on DMAD-bound downregulated genes bearing accumulation of 6mA upon DMAD knockdown was demonstrated proportionally. DMAD showed strong intronic enrichment on these genes. (C) The genes in (B) were further subcategorized based on the DMAD intragenic association. Gain-of-6mA distributions on each subset of genes were calculated, and the percentages shown in the pie chart. (D) The percentage of genes with both DMAD and gain-of-6mA binding to the same genomic feature were calculated.(E) Venn diagram shows substantial and significant overlap between DMAD and Wds ChIP-chip data (Binomial tests, p<0.001).(F) Co-immunoprecipitation experiments indicated a physical interaction between DMAD and Wds. (G) Downregulated genes in the DMAD-KD cells showed significant overlapping with downregulated genes in the absence of Wds. Chi-square tests were performed. RNA-seq were performed in triplicates. RPKM fold changes <−0.1 were included.(H) Substantial overlap between DMAD and Wds ChIPseq peaks suggested their functional coordination in regulating gene expression.(I) Average fold change in Pc and Wds reads of DMAD-KD over control were calculated for both DMAD-binding sites and gain-of-6mA regions to explore Pc and Wds dynamics in these regions. Heatmap demonstrated a general decrease in both Pc and Wds at DMAD binding sites when DMAD is depleted. A specific and significant increase in Pc binding on gain-of-6mA regions with DMAD depletion was found. Welch Two sample t-tests, p-values were indicated. Data are represented as mean ° SEM.

Figure 4.. DMAD and 6mA coordinate with Trithorax and Polycomb.

(A) Average normalized reads (per million) dynamics in Pc, Wds, H3K27me3, and H3K4me3 at DMAD/Wds binding sites in genes downregulated in the absence of DMAD compared to the control are shown. Welch Two sample t-tests, p-values were indicated.(B) Average normalized reads (per million) dynamics in Pc, Wds, H3K27me3 and H3K4me3 at gain-of-6mA regions in intragenic regions of genes identified in Figure 4A that were bound by DMAD/Wds. Welch Two sample t-tests, p-values were indicated.(C) Loci from Figure 4A and 4B were further tested by qPCR for expression changes in the absence of DMAD, Wds, Pc, or combined depletion of DMAD and Pc, as well as DMAD and Wds. t-tests were performed. *, p<0.05; **, p<0.01; ***, p<0.001; ****, p<0.0001.(D) In vitro 6mA-Pc binding assays were performed to confirm direct correlation between 6mA and Pc.(E) Pc binding kinetics to control and 6mA-modified probes showed that Pc displayed stronger binding to 6mA-modified DNA probes, as measured by fluorescence polarization assays. (F) DMAD binds to a group of genes involved in neurodevelopment and neuronal functions. These genes are directly targeted by the Trithorax protein Wds to maintain an active transcription profile. Additionally, DMAD actively demethylates intragenic 6mA. In the absence of DMAD,Wds binding is reduced at these loci, and accumulation of intragenic 6mA recruits Polycomb proteins. Data are presented as mean ± SEM.