Selective chemical tracking of Dnmt1 catalytic activity in live cells

Abstract

Enzymatic methylation of cytosine to 5-methylcytosine in DNA is a fundamental epigenetic mechanism involved in mammalian development and disease. DNA methylation is brought about by collective action of three AdoMet-dependent DNA methyltransferases, whose catalytic interactions and temporal interplay are poorly understood. We used structure-guided engineering of the Dnmt1 methyltransferase to enable catalytic transfer of azide tags onto DNA from a synthetic cofactor analog, Ado-6-azide, in vitro. We then CRISPR-edited the Dnmt1 locus in mouse embryonic stem cells to install the engineered codon, which, following pulse internalization of the Ado-6-azide cofactor by electroporation, permitted selective azide tagging of Dnmt1-specific genomic targets in cellulo. The deposited covalent tags were exploited as "click" handles for reading adjoining sequences and precise genomic mapping of the methylation sites. The proposed approach, Dnmt-TOP-seq, enables high-resolution temporal tracking of the Dnmt1 catalysis in mammalian cells, paving the way to selective studies of other methylation pathways in eukaryotic systems.

Keywords: 5-methylcytosine; DNA methyltransferase; cofactor selectivity; electroporation of AdoMet analogs; embryonic stem cells; enzyme engineering; epigenetic regulation; in cellulo labeling.

Graphical abstract

Figure 1

Structure-guided engineering of the Dnmt1 methyltransferase for catalytic transfer of extended functionalized groups (A) Enzymatic modification of hemimethylated CG/m⁵CG sites by the mouse Dnmt1 methyltransferase. Biological methylation using the AdoMet cofactor yields m⁵C, whereas bioorthogonal transfer of an extended side chain with a functional azide group from a synthetic cofactor analog, Ado-6-azide, yields 5-(6-azidohex-2-ynyl)cytosine (N₃-m⁵C) (shown in cyan and magenta, respectively). (B) Structure of the catalytic/cofactor-binding pocket of WT (PDB: 6w8w) and engineered Dnmt1-R1576/N1580 variants. The methyl-accepting C5 atom of the target 5-fluorocytosine residue (shown in cyan stick presentation) and the methyl-releasing sulfur atom of the bound AdoHcy (green sticks) are connected with a dashed magenta line. (C) Sequence conservation of the R1576 and N1580 residues in motif X across animal Dnmt1-like proteins and the GCGC-specific bacterial M.HhaI methyltransferase. Identical amino acids are shown in gray boxes. (D) Enhanced azidoalkylation of DNA by the engineered variants of Dnmt1 methyltransferase. HPLC-MS/MS analysis of nucleosides obtained after hydrolysis of Dnmt1-modified poly(dI-dC)·poly(dI-dC) DNA. Reactions were performed using 6 μM DNA substrate, 100 μM AdoMet or Ado-6-azide, and 400 nM Dnmt1 variant for 40 min at 37 °C. (E) Methylation activity of Dnmt1 variants toward hemimethylated and nonmethylated 25-mer DNA duplexes. Triplicate reactions containing 6 μM DNA substrate, 5.3 μM [methyl-³H]-AdoMet, and 20 nM WT or 100 nM engineered variant as shown were incubated for 40 min at 37 °C. Error bars denote ±SD.

Figure 2

Enhanced catalytic activity of the engineered Dnmt1 variants with AdoMet analogs containing sulfonium-bound propargylic side chains (A) DNA modification reactions were carried out for 1 h at 37°C using 50 nM hemimethylated 25-mer duplex CG-HM in which the unmethylated strand was 5′-³²P-labeled, 100 nM Dnmt1 and 100 μM AdoMet, or its synthetic analog (partial structures shown in top panel). Resulting modified DNA was reannealed with a 125-fold molar excess of an unmodified complementary strand, digested with the methylation-sensitive HhaI endonuclease, fractionated by denaturing PAGE and 5′-³²P-labeled strands, and visualized by autoradiography (see also Figure S4). (B) The effect of Ado-6-azide concentration on activity of Dnmt1 variants under single-turnover conditions. Reactions were carried out using 50 nM DNA, 100 nM Dnmt1, and Ado-6-azide as shown.

Figure 3

Cofactor selectivity of the wild-type and engineered Dnmt1 variants (A) Comparative analysis of DNA modification products by the Dnmt1 variants in the presence of AdoMet, Ado-6-azide, and their equimolar mixture under steady-state conditions. DNA modification reactions containing 6 μM poly(dI-dC)·poly(dI-dC) DNA substrate, 400 nM Dnmt1 variants, and 100 μM cofactor were incubated for 40 min at 37°C; modified DNA was hydrolyzed to nucleosides and analyzed using HPLC-MS/MS. (B) Comparative analysis of DNA modification products by the Dnmt1 variants at varied ratios of the AdoMet and Ado-6-azide cofactors under single-turnover conditions. DNA modification reactions containing 50 nM hemimethylated oligonucleotide, 100 nM Dnmt1, and 100 μM cofactor or their mixture as indicated were incubated for 60 min at 37°C. Reactions were analyzed as described in Figure 2.

Figure 4

Installation of bioorthogonal functional groups in DNA by endogenous Dnmt1-N1580A in mESC lysates and live cells (A) Strategy for selective Dnmt1-directed bioorthogonal pulse-tagging of endogenous DNA targets in a living cell. m, methylated cytosine. (B) Endogenous Dnmt1-N1580A alkyltransferase activity in cell lysate. DNA alkylation reactions were performed for 1 h at 37°C in 20 μL of corresponding mESC lysates supplemented with 100 μM Ado-6-azide and 500 ng of in vivo-hemimethylated pΔL2-14 plasmid DNA (Gerasimaitė et al., 2009), and DNA hydrolysates were analyzed for N₃-m⁵C using HPLC-MS/MS. (C) Experimental procedure for selective Dnmt1-directed pulse-tagging of endogenous DNA targets in live mESCs. (D) HPLC-MS/MS analysis of genomic cytosine modification in Dnmt1^WT and Dnmt1^N1580A mESCs after electroporation with 1 mM of Ado-6-azide. (E) Effects of cofactor concentration and postelectroporation incubation (chase) time on Dnmt1-directed intragenomic incorporation of N₃-m⁵C in mESCs. (F) Viability of WT and engineered mESCs after electroporation in the presence of 1 mM of Ado-6-azide. Cell survival was determined using an MTT assay 24 h after electroporation. Mean ± SD of at least three independent replicates. ^∗p < 0.05; n.s., not significant. See also Figure S6.

Figure 5

Analysis of genomic Dnmt1 modification sites in mESCs using Dnmt-TOP-seq (A) Distance distribution of read start positions to a nearest CpG site in the Dnmt-TOP-seq libraries prepared from WT and Dnmt1^N1580A mESCs 3 h after electroporation with 1 mM Ado-6-azide. (B) Dnmt-TOP-seq CpG modification profiles along generalized genomic elements for various gene types. Modified CpG sites were computed in the upstream (4 kb from TSS), gene body (from TSS to TTS normalized by gene length), and downstream (4 kb from TTS) regions. Processed pseudogenes are reverse-transcribed copies of mRNAs that lack introns, whereas unprocessed pseudogenes are produced by gene duplication and may contain introns. (C) Enrichment analysis of Dnmt-TOP-seq genomic elements and regulatory features. Odds ratio denote the enrichment (>1) or the depletion (<1) of particular genomic element terms in the genome-wide DNA modification profile of Dnmt1^N1580 mESC. (D) Enriched GO terms of genes containing methylated CGI promoters in Dnmt1^N1580A cells. CGIs bearing at least one modified CpG in all biological replicates were designated for the analysis. q value denotes false discovery rate. See also Figures S7 and S8.

Figure 6

Contribution of Dnmt1 to methylation of genomic CpG sites in mESCs (A and B) Comparison of Dnmt-TOP-seq CpG modification profiles (top panel in yellow) with Dnmt3a1 and Dnmt3b ChIP-seq normalized profiles (data obtained from Weinberg et al., 2019) in and around CpG islands located in promoters (2 kb upstream of protein-coding genes), intragenic and intergenic regions (A), or LINE and LTR elements (B). Profiles representing 20% slices of the most modified (top), moderately modified (mid), and least modified (bottom) regions were derived from experimental Dnmt-TOP-seq data. Bottom panel: average ChIP-seq read profiles for Dnmt3a (upper) and Dnmt3b (lower) at genomic regions selected above. (C) Validation of CGI modification profiles in H1fnt and Sfi1 genes. Columns denote read coverage or methylation levels of a particular CpG determined by Dnmt-TOP-seq (upper panel) or bisulfite sequencing (lower panel), respectively. Gaussian kernel-smoothed profiles are shown as dashed lines. Error bars denote ±SD.