Dynamic maps of UV damage formation and repair for the human genome

Abstract

Formation and repair of UV-induced DNA damage in human cells are affected by cellular context. To study factors influencing damage formation and repair genome-wide, we developed a highly sensitive single-nucleotide resolution damage mapping method [high-sensitivity damage sequencing (HS-Damage-seq)]. Damage maps of both cyclobutane pyrimidine dimers (CPDs) and pyrimidine-pyrimidone (6-4) photoproducts [(6-4)PPs] from UV-irradiated cellular and naked DNA revealed that the effect of transcription factor binding on bulky adducts formation varies, depending on the specific transcription factor, damage type, and strand. We also generated time-resolved UV damage maps of both CPDs and (6-4)PPs by HS-Damage-seq and compared them to the complementary repair maps of the human genome obtained by excision repair sequencing to gain insight into factors that affect UV-induced DNA damage and repair and ultimately UV carcinogenesis. The combination of the two methods revealed that, whereas UV-induced damage is virtually uniform throughout the genome, repair is affected by chromatin states, transcription, and transcription factor binding, in a manner that depends on the type of DNA damage.

Keywords: DNA damage; UV; human genome; nucleotide excision repair; transcription factor.

Fig. 1.

HS–Damage-seq method. (A) Schematic of HS–Damage-seq. See Fig. S1 for details. (B and C) Gel verification of CPD (B) and (6-4)PP (C) libraries. DNA was extracted from UV-irradiated NHF1 cells at indicated time points after UV, and “−” means no UV. (D) Dipyrimidine frequencies at the predicted damage sites.

Fig. S1.

Schematic representation of HS–Damage-seq method. Genomic DNA from UV-treated cells was sonicated and subjected to end repair. Then the first adaptor was ligated to these end-repaired fragments. This step was followed by denaturation and IP with an antibody against specific DNA lesion. Next, a biotinylated primer (with a dU base) was annealed to the 3′ end of the first adaptor and extended by a DNA polymerase, which was blocked by DNA damage. The extension products were denatured and purified by streptavidin beads, then eluted by digesting the dU base. The unblocked extension products were then captured by biotinylated oligonucleotides identical to the 5′ end of the first adaptor and removed by streptavidin beads. Purified fragments were ligated to the second adaptor. Finally, the ligation products were amplified with index primers and sequenced.

Fig. 2.

DNA bulky adduct formation at TFBSs for UV-induced (6-4)PP, CPD, and cisplatin-induced d(GpG) diadduct. Cell (in vivo damaged) and naked (in vitro damaged) DNA are shown in orange and blue, respectively. The x axis is scaled to show a 50-nucleotide window centering TFBS motif. The y axis shows the damage level as reads per base per thousand mapped (onto TFBS-centered 1-kb window) reads. Each read is assigned to a single nucleotide position. Motif strand (MS) is in 5′ → 3′ direction, whereas the complementary strand (CS) is in 3′ → 5′ direction. Transcription factors presented are (A) CTCF (number of TFBSs, n = 62,002), (B) NFYB (n = 12,452), (C) POU2F2 (n = 20,390) and (D) SP1 (n = 15,877). The sequence logos were computed by using the TF-binding sites that we predict. Unbound sites for these TFs are presented in Fig. S3.

Fig. S2.

DNA damage formation at transcription factor binding sites, for UV-induced (6-4)PP, CPD, and cisplatin-induced damages. The examined transcription factors are BHLE40, EBF1, EGR1, ELF1, MAX, MEF2A, NFATC1, PAX5, RELA, RUNX3, SPI1, TBP, TCF3, YY1, and ZNF143. Cell (in vivo damaged) and naked (in vitro damaged) DNA are shown in orange and blue, respectively.

Fig. S3.

Damage formation at the unbound motif sites. Cell and naked DNA are shown in orange and blue, respectively. Motif sites are compiled from unbound (genomic regions with no called TF ChIP-seq peaks for the GM12878 cell line) for (A) CTCF, (B) NFYB, (C) POU2F2, and (D) SP1 binding site motifs. The sequence logos were computed by the ensemble of motif sites that we retrieved.

Fig. S4.

DNA damage formation at unbound transcription factor motif sites, for UV-induced (6-4)PP, CPD, and cisplatin-induced damages. Cell (in vivo damaged) and naked (in vitro damaged) DNA are shown in orange and blue, respectively.

Fig. 3.

Patterns of damage distribution as a function of time. (A) CPD and (B) (6-4)PP damage distribution within a ∼0.9-Mb region of chromosome 9 (100,137,675–101,036,695). For each panel, Top shows the Damage-seq of different time points and initial XR-seq. [1 h for CPD and 5 min for (6-4)PP] reads mapped onto the plus strand and relevant RNA-seq reads mapped on the minus strand. Bottom represents the complementary strands. The shadows highlight transcribed genes on each strand.

Fig. 4.

UV-induced damage and repair profiles at transcribed and at DNase hypersensitive regions. The y axis shows the number of reads per kilobase per million mapped reads (RPKM) and the x axis shows the relative distance to the related genomic element. (A) CPD, (B) (6-4)PP damage, (C) CPD, and (D) (6-4)PP repair distribution along with the transcription start and end sites for nonoverlapping transcripts that are longer than 15 kb (n = 5,025). Each data point is an RPKM value at a 200-bp binned window. (E) CPD, (F) (6-4)PP damage, (G) CPD, and (H) (6-4)PP repair distribution along with the DNase hypersensitivity sites (n = 162,164). The zero point on the x axis was taken as the center of the called region. Each data point is an RPKM value at a single nucleotide position. Data represented are from two merged biological replicates.

Fig. S5.

UV-induced cell and naked DNA damage profiles at transcription start and end sites.

Fig. S6.

UV- and cisplatin-induced cell and naked DNA damage profiles at DNase hypersensitivity sites.

Fig. 5.

Damage and repair as a function of chromatin states. (A) Cell/naked DNA damage ratio. The effect of genomic states on damage formation is computed by taking the ratio of DNA damage in cell to damage in naked DNA. (B) Relative ratio of total reads normalized by the total input DNA mapping onto each chromatin state. The y axis, which is grouped by chromatin states, shows the time points from 0 to 48 h for CPD and from 0 to 4 h for (6-4)PP. The x axis shows the relative damage ratio. (C) Repair normalized to corresponding damage level. Although some chromatin states have identical names, the 15 states are de novo defined by the combination of histone modifications and CTCF binding. As a result, states 4 and 6 are more enriched around TSS than states 5 and 7.