High resolution mapping of modified DNA nucleobases using excision repair enzymes

Abstract

The incorporation and creation of modified nucleobases in DNA have profound effects on genome function. We describe methods for mapping positions and local content of modified DNA nucleobases in genomic DNA. We combined in vitro nucleobase excision with massively parallel DNA sequencing (Excision-seq) to determine the locations of modified nucleobases in genomic DNA. We applied the Excision-seq method to map uracil in E. coli and budding yeast and discovered significant variation in uracil content, wherein uracil is excluded from the earliest and latest replicating regions of the genome, possibly driven by changes in nucleotide pool composition. We also used Excision-seq to identify sites of pyrimidine dimer formation induced by UV light exposure, where the method could distinguish between sites of cyclobutane and 6-4 photoproduct formation. These UV mapping data enabled analysis of local sequence bias around pyrimidine dimers and suggested a preference for an adenosine downstream from 6-4 photoproducts. The Excision-seq method is broadly applicable for high precision, genome-wide mapping of modified nucleobases with cognate repair enzymes.

Figure 1.

Excision-seq methods for mapping modified nucleobases in genomic DNA. (A) In “predigestion” Excision-seq for uracil, uracil-containing DNA is cut with a base excision repair enzyme (e.g., E. coli UDG, red). Released fragments are end-repaired, A-tailed, ligated to adaptors, and PCR amplified. Sequences derived from this library identify the positions of modified bases (e.g., one base upstream of the 5′-most position of the read). (B) In “post-digestion” Excision-seq for uracil, DNA is sheared mechanically, then treated by standard polishing and ligation. A base excision enzyme cleaves one or both strands containing modified bases. Intact strands remaining after digestion are PCR amplified and sequenced. (C) Genomic DNA isolated from dut ung E. coli is digested by UDG and T4 endonuclease IV (cf. lanes 3 and 4), whereas genomic DNA isolated from a wild-type strain is not digested (lanes 1 and 2). (D) Normalized coverage from shotgun sequencing of mechanically sheared genomic DNA (gray, reads per million [RPM]) and predigestion Excision-seq for uracil (blue, RPM) for a 2.8-Mb region of the E. coli chromosome. GC-content and the positions of protein-coding genes are plotted below. Uracil content is lowest in a region centered on the origin of replication, encompassing ∼200 kb of DNA.

Figure 2.

Excision-seq mapping of uracil content in the budding yeast genome. (A) Normalized frequency of nucleotides relative to mapped positions of sequences from predigestion Excision-seq libraries for mapping S. cerevisiae uracil content. Position 0 corresponds to the mapped position of the 5′ end of 11,326,044 sequencing reads; negative numbers are upstream. Frequencies were normalized to genomic mononucleotide frequencies. (B) Normalized frequency of nucleotides relative to mapped positions of sequences from post-digestion Excision-seq libraries for mapping S. cerevisiae uracil content. Position 0 corresponds to the mapped position of the 5′ end of 2,939,357 sequencing reads (frequency normalization and colors identical to A). (C) Boxplot comparing post-digestion Excision-seq mapping of uracil in S. cerevisiae to replication timing (T_rep) (Raghuraman et al. 2001). Mean signals were calculated in 500-bp windows across the genome. (D) Boxplot comparing mean genomic AT content in S. cerevisiae to replication timing; 500-bp regions are identical to C.

Figure 3.

Excision-seq maps of uracil content in the budding yeast genome. (A) Data showing the entire yeast chromosome 4 for dut1-1 ung1∆ yeast using post-digestion Excision-seq (red, reads per million [RPM]), predigestion Excision-seq (blue, RPM), post-digestion Excision-seq data for ung1∆ yeast treated with 5-fluorouracil (5-FU) (green, RPM), single-stranded DNA accumulation caused by hydroxyurea treatment of a rad53 yeast strain (purple, arbitrary units) (Feng et al. 2006), replication timing data (T_rep, minutes replicated after G₁ release) (Raghuraman et al. 2001) (gray), annotated origins of replication (Nieduszynski et al. 2007), ORC chromatin immunoprecipitation signals (Eaton et al. 2010) (brown, coverage), and labeled segments from an eight-state DBN segmentation (Hoffman et al. 2012) incorporating replication timing (Yabuki et al. 2002) and post-digestion Excision-seq mapping of uracil. (B) A 450-kb region of chromosome 4 highlights patterns of uracil incorporation in early-replicating origins (ARS418 and ARS428), as well as uracil depletion in late-replicating regions. (C) Correspondence of peak widths between post-digestion Excision-seq (red) and ssDNA accumulation (Feng et al. 2006) (purple) at three early-replicating origins in a 100-kb region of chromosome 3. (D) Post-digestion Excision-seq measurement of uracil content for 50 early-replicating origins. Lagging strands have ∼1.3-fold higher relative coverage than leading strands in post-digestion Excision-seq data, reflecting increased uracil content in leading strands. (E) A 15-kb region of chromosome 4 highlights patterns of uracil incorporation at the early-replicating origin ARS428.

Figure 4.

Excision-seq mapping of dipyrimidines in the budding yeast genome. (A) In predigestion Excision-seq for pyrimidine dimers, UV-damaged DNA is cleaved with UVDE, releasing double-stranded fragments with five dipyrimidines (red). Fragments are treated with CPD or 6-4pp photolyase enzymes, repairing dipyrimidines to “mono” pyrimidines, and yielding ends compatible with polishing, ligation, and PCR. (B) Analysis of sequencing libraries treated with CPD or 6-4pp photolyase prepared from UV-irradiated DNA showed an enrichment of sequence reads with dipyrimidine ends (red text) relative to genomic dinucleotide content and recapitulated the known specificity of the photolyase enzymes (Chowdhury and Guengerich 2008). (C) Frequency of nucleotides relative to mapped positions of sequences from predigestion Excision-seq libraries for mapping cyclobutane dimers in S. cerevisiae. Position 0 corresponds to the mapped position of the 5′ end of 5,063,196 sequencing reads. (D) Frequency of nucleotides relative to mapped positions of sequences from predigestion Excision-seq libraries for mapping 6-4 photoproducts in S. cerevisiae. Position 0 corresponds to the mapped position of the 5′ end of 3,655,251 sequencing reads.