Fig 1.
Tethered oligonucleotide priming and analysis of 5hmC.
(A) Strategies of 5hmC derivatization. (B) Schematic of a conventional primer extension reaction by a DNA polymerase derived from Pfu-DNA cocrystal structure (PDB: 5OMF) (left) and a putative mechanism of tethered ODN–primed polymerase reaction (right). The chemical tether (shown as green/red line) connecting T2 in the bound DNA duplex and the 5hmC residue in gDNA strand facilitates the capture of the 5hmC in a stacked position required for the priming reaction. Beige areas show regions of extensive contacts between the Pfu polymerase and the bound DNA. (C) Left, chemical structure of the tethering linker 1 (top) derived by MM2 conformational energy minimization (S1 Fig) is shown in the context of the template DNA strand 5′..T4G5T6.. of the KOD polymerase–DNA complex (PDB: 4AIL) resembling the positions of the hmC, T1, and T2 nucleotides (bottom) in Fig 1B. Right, a large cleft of Pfu polymerase (space-fill orange) seen from the major groove side of the bound DNA duplex (blue and green strands). (D) Workflow of the hmTOP-seq procedure. (Step 1) Fragmented gDNA is tagged with an azide group through BGT-glucosylation. (Step 2) Azide-modified DNA is ligated to partially complementary adaptors. (Step 3) The ODN containing a biotin group is tethered to azide groups using click chemistry. (Step 4) The biotin-labeled fragments are captured on streptavidin beads. (Step 5) TO-primed strand extension. (Step 6) PCR amplification with Ad-A2 and Ad-TO primers containing NGS platform-specific 5′-end adaptor sequences. Unidirectional sequencing from the A adaptor sequence included in the 5′ part of Ad-TO-barcode amplification primer. 5hmC, 5-hydroxymethylcytosine; A1 and A2, strands of a partially complementary adaptor; Ad, extended sections of platform-specific adapters; BGT, β-glucosyltransferase; DBCO, dibenzocyclooctyne; eM.Sssl, an engineered version of the CG-specific DNA cytosine-5 methyltransferase M.SssI; gDNA, genomic DNA; hmC, 5-hydroxymethylcytosine; hmCG, hydroxymethylated CG site; hmTOP-seq, 5hmC-specific tethered oligonucleotide–primed sequencing; KOD, the KOD DNA polymerase from Thermococcus kodakaraensis; mCG, methylated CG site; NGS, next generation sequencing; NHS, N-hydroxysuccinimide; ODN, oligodeoxyribonucleotide; PDB, Protein Data Bank; Pfu, the Pfu DNA polymerase from Pyrococcus furiosus; TO, tethered oligodeoxyribonucleotide.
Fig 2.
Analysis of hmTOP-seq libraries in model lambda genome and mESCs.
(A) Correlation between the coverage of 2.5% and 40% pre-hydroxymethylated GCGC sites in the technical replicates of hmTOP-seq libraries of lambda bacteriophage genome. (B) Dependence of hmTOP-seq coverage on the level of hydroxymethylation of GCGC sites in bacteriophage lambda DNA. Quadratic regression was used to fit the plotted data (Y = 99.5 + 15.7X − 0.240X2). (C) Distance distribution of read start positions from a nearest CG site in the hmTOP-seq library of mESC DNA (500-ng input). (D) Correlation of 5hmCG coverage and h-density signals in replicates of hmTOP-seq libraries prepared with varying amounts of mESC DNA. h-density was computed by normalizing coverage values by the unweighted CG density as described in [25]. (E) Comparison of hmTOP-seq coverage and 5hmC percentages estimated by bisulfite-based TAB-seq. Each dot represents the average hmTOP-seq value for specific TAB-seq percentage group (97% of all CG that overlap between hmTOP-seq and TAB-seq are used for analysis). (F) Correlation between hmTOP-seq coverage at 5hmCHs in technical replicates of 500-ng input mESC DNA libraries (OR = 347, p < 2.2 × 10−16; Fisher’s exact test). (G) Correlation of 5hmC signal between mESCs hmTOP-seq and nano-hmC-Seal data (average peak region size 615 bp). Within each nano-hmC-Seal peak region, total amount of signal from both methods was square-root transformed and correlated per each autosome. The data underlying this figure are included in S1 Data. 5hmC, 5-hydroxymethylcytosine; 5hmCG, hydroxymethylated CG site; 5hmCH, hydroxymethylated CH site, where H = A, C, or T; hmTOP-seq, 5hmC-specific tethered oligonucleotide–primed sequencing; mESC, mouse embryonic stem cell; OR, odds ratio; TAB-seq, Tet-assisted bisulfite sequencing.
Fig 3.
Genomic distribution of 5hmCGs in mESCs.
(A) OR from Fisher’s exact test for enrichment of the high- (top 20%), medium- (middle 20%), and low-coverage (bottom 20%) 5hmCGs across various genomic features. Poised enhancers (“enh.”): regions with H3K4me1 mark only; active enhancers: regions with H3K4me1 and H3K27ac histone marks; active promoters: 2-kb regions upstream of the gene start that overlap H3K9ac histone mark; nonactive promoters: 2-kb upstream regions depleted in H3K9ac. All analyses are shown for a 500-ng input hmTOP-seq library. All shown enrichment values have p ≤ 1.1 × 10−6. The data underlying this figure are included in S1 Data. (B) hmTOP-seq coverage profile normalized to CG density over different gene-associated regions: upstream (2 kb), 5′ UTR, exons, introns, 3′ UTRs, and downstream (2 kb). Distribution of (C) 5hmCGs and (D) uCGs across the sense and the antisense strands of genes grouped according to their expression level. Numbers of genes in each group and p-values for the modification difference between the strands are shown above each graph. All analyses are shown for a 500-ng input hmTOP-seq library. 5hmCG, hydroxymethylated CG site; CGI, CG island; H3K4me1, histone H3 lysine 4 monomethylation; H3K9ac, histone H3 lysine 9 acetylation; H3K27ac, histone H3 lysine 27 acetylation; H3K36me3, histone H3 lysine 36 trimethylation; hmTOP-seq, 5hmC-specific tethered oligonucleotide–primed sequencing; LTR, long terminal repeat; mESC, mouse embryonic stem cell; OR, odds ratio; SINE, short interspersed nuclear element; TSS, transcription start site; TTS, transcription termination site; uCG, unmethylated CG; UTR, untranslated region.
Fig 4.
CG modification profiles at exon-intron cross-boundaries.
Distribution of 5hmCGs (A) (500-ng input DNA hmTOP-seq libraries) or uCGs (B) at both sides of the exon-intron boundary is presented for the sense and the antisense strands. The x-axis shows the distance (nt) of CGs from the boundary. p-Values indicate a difference in coverage between exonic and intronic side of the boundary for the first 25 nt. At the exon-intron boundary (left part), general 5hmCG modification levels are higher at the exonic side as compared with the intronic side for both strands. At the intron-exon boundary (right part), the antisense strand shows the same trend, whereas the sense strand shows higher general 5hmCG levels on the intronic side. The data underlying this figure are included in S1 Data. 5hmCG, hydroxymethylated CG site; nt, nucleotide; uCG, unmodified CG site.