ETS transcription factors induce a unique UV damage signature that drives recurrent mutagenesis in melanoma

Abstract

Recurrent mutations are frequently associated with transcription factor (TF) binding sites (TFBS) in melanoma, but the mechanism driving mutagenesis at TFBS is unclear. Here, we use a method called CPD-seq to map the distribution of UV-induced cyclobutane pyrimidine dimers (CPDs) across the human genome at single nucleotide resolution. Our results indicate that CPD lesions are elevated at active TFBS, an effect that is primarily due to E26 transformation-specific (ETS) TFs. We show that ETS TFs induce a unique signature of CPD hotspots that are highly correlated with recurrent mutations in melanomas, despite high repair activity at these sites. ETS1 protein renders its DNA binding targets extremely susceptible to UV damage in vitro, due to binding-induced perturbations in the DNA structure that favor CPD formation. These findings define a mechanism responsible for recurrent mutations in melanoma and reveal that DNA binding by ETS TFs is inherently mutagenic in UV-exposed cells.

Fig. 1

Genome-wide map of CPD lesions reveals that CPDs are elevated at active TFBS. a Schematic diagram of the CPD-seq method for mapping CPD lesions at single nucleotide resolution. ‘T = C′ indicates a CPD lesion at TC dipyrimidine. Oligonucleotide adapters are indicated in green and purple; ‘NNNNNN′ indicates a random DNA hexamer. A 3′ hydroxyl is indicated with OH, while ‘dd’ indicates a dideoxy 3′ end. The CPD lesion is cleaved with T4 endonuclease V and apurinic/apyrimidinic endonuclease (APE1) to generate a free 3′ hydroxyl immediately upstream of the CPD lesion, which is ligated to an adapter and sequenced. b Mutation density surrounding active promoter-proximal TFBS from 184 sequenced melanoma tumors. Observed mutation density (i.e., in melanoma tumors) was analyzed adjacent to known TFBS located in promoter-proximal regions (up to 2500 bp upstream of transcription start site) that were considered active (i.e., overlapping with melanocyte DNase I-hypersensitivity (DHS) regions) for 82 distinct TFs. Expected mutation density was determined from the corresponding DNA sequences surrounding each active promoter-proximal TFBS, based on the trinucleotide mutation signature frequencies for all promoter-proximal regions. c Same as part (b), except mutations were analyzed adjacent to promoter-proximal TFBS that were considered inactive (i.e., not overlapping with melanocyte DHS regions). d Average number of CPD lesions (per TFBS) adjacent to active promoter-proximal TFBS. CPD lesions were mapped using CPD-seq from UV-irradiated NHF1 cells (100 J m⁻²) or isolated NHF1 DNA that was UV-irradiated (80 J m⁻²) in vitro (naked DNA). Cellular DNA was harvested immediately after UV irradiation, so essentially no repair was allowed to occur. e Same as in part (d), except CPD lesions were analyzed adjacent to inactive promoter-proximal TFBS

Fig. 2

Elevated CPD lesions and mutation density are associated with ETS TFBS. a Enrichment of CPD levels following UV irradiation in cells relative to naked DNA at active promoter-proximal TFBS for each individual TF. The scaled CPD enrichment was plotted against the total number of CPD lesions present in the set of TFBS for each TF. Only CPD lesions located in the core of the TFBS (−4 to +4 bp relative to the TFBS midpoint) were analyzed. ETS family TFs are in red, NFYA and NFYB are in blue, Fos and Jun TF family members are in green. b Same as in part (a), except only CPD lesions associated with flanking DNA (−100 to −5 bp and +5 to +100 bp relative to the TFBS midpoint) were analyzed. c Same as in part (a), except only mutagenic CPDs (mCPDs), which are CPD lesions at TC, CT, and CC dinucleotides, were analyzed. d Same as part (c), except only mCPD lesions associated with flanking DNA (−100 to −5 bp and +5 to + 100 bp relative to the TFBS midpoint) were analyzed. e Enrichment of melanoma mutations at the core of active promoter-proximal TFBS (−4 to +4 bp relative to the TFBS midpoint) for each TF. Mutation enrichment was calculated from the ratio of observed mutations in 184 melanoma tumors relative to the expected number of mutations based on DNA sequence (see Fig. 1b legend). ETS family TFs are in red, NFYA and NFYB are in blue, and Fos and Jun TF family members are in green. f Mutation enrichment strongly correlates with mutagenic CPD (mCPD) enrichment at active TFBS (two-sided Spearman correlation coefficient = 0.47, P = 0.0002). Log–Log plot of mutation enrichment relative to mCPD enrichment for the core of active promoter-proximal TFBS for each TF. Only TFs that passed the inclusion criteria for parts (c) and e were plotted

Fig. 3

ETS family TFBS are the primary contributors to elevated CPD levels. a, b Formation of CPD (a) and mCPD (b) lesions is significantly stimulated at active, promoter-proximal ETS family TFBS (i.e., ELF1, ELK4, ETS1, and GABPA) in UV-irradiated NHF1 cells, but not when isolated NHF1 DNA was UV-irradiated in vitro (naked DNA). c Mutation density is significantly elevated at active promoter-proximal ETS family TFBS (i.e., ELF1, ELK4, ETS1, and GABPA) in melanoma tumors, correlating with the higher initial mCPD lesion density at these sites. d, e Formation of CPD (d) and mCPD (e) lesions is not significantly stimulated at active promoter-proximal TFBS after excluding ELF1, ELK4, ETS1, and GABPA binding sites. f Mutation density is only slightly elevated surrounding non-ETS family TFBS in aggregate (see inset with expanded scale). Only active promoter-proximal TFBS were analyzed. g CPD repair activity is elevated at ETS family TFBS following UV irradiation of human cells. Average CPD repair activity at 1 h repair in UV-irradiated NHF1 cells at ETS family TFBS (i.e., ELF1, ELK4, ETS1, and GABPA). CPD repair activity was calculated using the average number of XR-seq reads at locations surrounding active, promoter-proximal ETS binding sites. XR-seq reads were localized to the putative dipyrimidine lesion associated with each sequencing read. h Same as in g, except repair activity for active ETS family TFBS located outside promoter regions was analyzed. i Same as (g), except CPD repair activity was analyzed at non-ETS family TFBS

Fig. 4

CPDs and mutations are elevated at specific locations in ETS binding sites. a UV-induced CPD formation and mutation density is enriched at 1279 active, promoter-proximal TFBS for the ETS TFs ELK4, ETS1, and GABPA. TFBS were aligned based on DNA strand and location of ETS consensus sequence. The upper panel depicts the information content of the DNA sequences for the aligned TFBS, matching the known ETS consensus motif. Sequence logo was generated using the weblogo tool. The lower panel plots the mutation density for 184 melanoma tumors relative to average CPD levels following UV-irradiation of NHF1 cells and isolated DNA in vitro (naked DNA). CPD values are plotted at half integer locations, which reflect the average number of CPD lesions forming between the two adjacent nucleotides. CPD lesion density for naked DNA was scaled so that the total CPD levels in promoter-proximal regions in cells and naked DNA were equivalent. b Analysis of CPD lesion formation and mutation density at a subset of ETS TFBS (156 sites) that have a pyrimidine nucleotide at position −4 relative to the ETS motif midpoint (indicated with *), and thus can form CPD lesions between position −4/−3. Upper and lower panels are the same as in part (a), plotted for this TFBS subset

Fig. 5

Binding of ETS1 protein promotes UV damage formation in vitro. a DNA sequences of RPL13A and SDHD promoter fragments, corresponding to chromosome coordinates chr19:49990710-49990681 and chr11:111957515-111957553, respectively. Putative ETS motifs are shown in gray background and highlighted in bold. Recurrent mutated sites in melanomas are underlined. b, c Gel shift assays showing binding of purified ETS1 protein to radiolabeled RPL13A (b) and SDHD (c) promoter fragments, respectively. d A representative sequencing gel (15%) showing CPD formation in naked RPL13A (sample 1, with UV irradiation) and ETS1-bound RPL13A DNA (samples 2–5, UV irradiation). The binding products shown in part (b) were irradiated with 1KJ m⁻² of UV-C light and CPD lesions were converted to single strand breaks by T4 endonuclease V digestion. The resulting DNA breaks were separated on a 15% denaturing sequencing gel to analyze damage abundance at different locations. A negative control (naked DNA without UV irradiation) was also digested with T4 endoV to show the background level of DNA cleavage in the absence of UV-induced DNA lesions. The first lane on the left shows a 10-nt DNA ladder. e Same as in part (d), except the SDHD promoter fragment was analyzed on a 12% gel. Asterisk indicates gel running artifact caused by bromophenol blue in the gel loading buffer. Both RPL13A and SDHD CPD formation experiments were conducted at least 3 times independently with consistent results

Fig. 6

ETS binding induces a DNA conformation predisposed to form CPD lesions. a Schematic showing the key structural parameters that affect the rate of CPD formation, namely the distance (d) between the midpoints of adjacent C5–C6 bonds, and the torsion angle (η) between the adjacent C5–C6 bonds. b, c Plots of distance (d) and torsion angle (η) between adjacent C5–C6 bonds for structures of ETS bound DNA. The arrow indicates the location of the CPD hotspot at position −1/0, corresponding to position −0.5 in the plot. The mean ± SEM is plotted for data derived from thirteen crystal structures. d Structure of GABPA bound to DNA (PDB ID: 1AWC). The CPD hotspot at positions −1/0 relative to the motif midpoint is highlighted in green. Image was visualized using pymol. e Molecular dynamics simulation of GABPA (i.e., GABPα) bound to a 14 nt duplex DNA sequence containing ETS motif-2 in the RPL13A promoter. The distance (d) and torsion angle (η) between C5–C6 bonds of adjacent pyrimidines was plotted for the final 100 ns (10,000 frames). The distribution for GABPA-bound DNA is depicted in red, while simulated unbound DNA is shown in gray. Each increment in the ordinate represents 500 counts. Green lines indicate canonical values of B-form DNA

Fig. 7

ETS TFs induce a unique UV damage signature that drives mutagenesis in melanoma. The extremely mutable subset of ETS TFBS contains a pyrimidine (C or T) at position −4 relative to the ETS motif midpoint, while most ETS TFBS contain a purine (A or G) at this position. Blue bars indicate relative levels of CPD lesions following UV irradiation, while red bars indicate the frequency of mutations in melanoma tumors. ‘N’ indicates any nucleotide, while stacked nucleotides indicate degeneracy at that position in the ETS binding site motif. ETS binding site mutations may alter transcription of target genes such as SDHD. Structure of GABPA bound to DNA (PDB ID: 1AWC) was created using pymol