Escherichia coli YcaQ is a DNA glycosylase that unhooks DNA interstrand crosslinks

Abstract

Interstrand DNA crosslinks (ICLs) are a toxic form of DNA damage that block DNA replication and transcription by tethering the opposing strands of DNA. ICL repair requires unhooking of the tethered strands by either nuclease incision of the DNA backbone or glycosylase cleavage of the crosslinked nucleotide. In bacteria, glycosylase-mediated ICL unhooking was described in Streptomyces as a means of self-resistance to the genotoxic natural product azinomycin B. The mechanistic details and general utility of glycosylase-mediated ICL repair in other bacteria are unknown. Here, we identify the uncharacterized Escherichia coli protein YcaQ as an ICL repair glycosylase that protects cells against the toxicity of crosslinking agents. YcaQ unhooks both sides of symmetric and asymmetric ICLs in vitro, and loss or overexpression of ycaQ sensitizes E. coli to the nitrogen mustard mechlorethamine. Comparison of YcaQ and UvrA-mediated ICL resistance mechanisms establishes base excision as an alternate ICL repair pathway in bacteria.

Scheme 1.

Synthesis of NM₈.

Figure 1.

YcaQ is a Monofunctional DNA Glycosylase Specific for Cationic N-Alkylpurines. (A) Multiple sequence alignment of AlkZ/YcaQ homologs from Streptomyces sahachiroi and representative pathogens. Catalytic residues in AlkZ are labeled with asterisks. Secondary structures derived from the AlkZ crystal structure are shown above the alignment. (B) Schematic of d7mG base excision reaction. (C–F) Denaturing PAGE of 5′-FAM labeled d7mG-DNA substrate (S) and nicked AP-DNA product (P) after treatment with enzyme or buffer (mock). AP-DNA resulting from glycosylase activity was nicked by treatment with either 0.1 M NaOH (C–F) or EndoIV (C) to produce β- and β,δ-elimination products, which are quantified below each gel. (C) d7mG-DNA was incubated with (+) or without (-) YcaQ for 1 hr followed by treatment with either water, NaOH, or EndoIV. (D) Comparison of WT and mutant YcaQ activity after 1 hr. (E) Time-dependence of d7mG excision by AlkZ and YcaQ. The presence of a single product band in the AlkZ reaction is likely a result of EndoIV contamination in the work-up buffer, consistent with our previous analysis (20). (F) Structure and excision of mFaPy-dG DNA, treated with E. coli Fpg, YcaQ, or buffer for the specified time, followed by alkaline hydrolysis.

Figure 2.

AlkZ and YcaQ can unhook either side of an azinomycin B-ICL. (A) Reaction of azinomycin B with guanines on opposing strands to form an ICL. N-glycosidic bonds hydrolysed by AlkZ/YcaQ are highlighted with green and red arrows, corresponding to FAM and Cy5-labeled strands, respectively. (B) Schematic of the 26-mer oligonucleotide AZB-ICL substrate used in this study. The epoxide predominantly reacts with the FAM-labeled GCC strand and aziridine with the Cy5-labeled GGC strand. Green and red spheres represent FAM and Cy5 labels, respectively. (C) Schematic of the base excision assay used to monitor ICL unhooking activity. Glycosylase unhooking of ICL-DNA potentially forms monoadducts and AP sites on either strand. AP-sites are nicked with hydroxide to form shorter oligonucleotides through β,δ-elimination. ICL-, monoadduct (MA)-, and β,δ-elimination products can be separated by denaturing PAGE. PUA, 3′-phosphor-α,β-unsaturated aldehyde (β-elimination product), P, 3′-phosphate (β,δ-elimination product). (D-E) Denaturing PAGE of AZB-ICL, AZB-MA, and nicked AP-DNA products after treatment with heat (Δ), buffer (mock), or enzyme, followed by alkaline hydrolysis. The percent of β,δ-elimination product is quantified below the gel. Each image is an overlay of false-colored FAM (green) and Cy5 (red) fluorescence scans of the gels, in which yellow depicts coincident red and green intensity. Individual FAM and Cy5 imaged gels are shown in Supplementary Figure S2. (E) 30-min reactions between AlkZ, YcaQ and catalytic mutants for AZB-ICL-DNA. (F) Denaturing PAGE showing time-dependence of AlkZ and YcaQ unhooking of AZB-ICLs. The fraction of ICL-DNA unhooked from three independent experiments is quantified to the right (mean ± SEM). (G) Quantification of the fraction of monoadduct (MA) and β,δ-elimination (nicked product) from the mock (left), AlkZ (middle), and YcaQ (right) reactions from the gel in panel F.

Figure 3.

YcaQ unhooks nitrogen mustard ICLs. (A) 5-atom nitrogen mustard (NM₅) ICL formed between mechlorethamine and guanines located on opposite DNA strands. (B-C) Denaturing PAGE of NM₅-ICL substrate, monoadduct, and nicked AP-DNA products after treatment with WT or mutant YcaQ (B), or with YcaQ, human AAGΔ83, E. coli AlkA, or S. pombe Mag1 (C), followed by treatment with EndoIV. The percent of unhooked ICL is quantified below the gel. Images are false colored overlays of individual FAM and Cy5 scans of the gel, which are shown in Supplementary Figure S3. (D) Denaturing PAGE showing time-dependence of AlkZ and YcaQ unhooking of NM₅-ICLs with a hydroxide workup. The plot quantifies the decrease in the fraction of NM₅-ICL (mean ± SEM, n = 3). (E) Quantification of the fraction of monoadduct (MA) and β,δ-elimination (nicked AP-DNA product) from the mock (left), AlkZ (right), and YcaQ (middle) reactions from the gel in panel D.

Figure 4.

NM-ICL unhooking by YcaQ does not depend on a kinked duplex. (A) Structure and ICL formed from 8-atom nitrogen mustard (NM₈) derivative. (B) Denaturing PAGE of the NM₈-ICL-DNA substrate and nicked abasic-DNA products after treatment with buffer (mock), AlkZ or YcaQ for the specified time followed by alkaline hydrolysis. Heat-mediated depurination (lanes 2–3) serve as a positive control for excision products. Individual FAM and Cy5 imaged gels are shown in Supplementary Figure S4. (C) Quantification of the fraction of ICL from three separate experiments (mean ± SEM).

Figure 5.

YcaQ and AlkZ create opposing AP sites. (A) Native PAGE analysis of DNA products formed from YcaQ incubation with NM₅-ICL and AlkZ incubation with AZB-ICL DNA substrates. AP sites formed from glycosylase activity were nicked by incubation with EndoIV prior to loading the DNA. Double-stranded standards for the double-nicked excision products are shown in lanes 1–2. The gel is a false-colored composite of the individual FAM- and Cy5-imaged gel, which are shown in Supplementary Figure S5A. (B) Quantification of the total fluorescent signal for the fraction of ICL, single- and double-nicked EndoIV cleavage products from the gel in panel A. (C) Denaturing PAGE of d7mG/AP- and d7mG/dU-DNA substrates (S) and nicked AP-DNA products (P) after treatment with (+) or without (–) YcaQ followed by alkaline hydrolysis. Only the FAM-d7mG strand is visualized on the gel. The percent of β,δ-elimination products is quantified below.

Figure 6.

Deletion and overexpression of YcaQ sensitizes E. coli to mechlorethamine. (A) Colony dilution assay for E. coli deletion strains exposed to increasing concentrations of mechlorethamine (HN2). Values are mean ± SEM (n = 3). Percent (%) survival is relative to untreated cells. (B) EC₅₀ values derived from data in panel A. Significance values were calculated using a one-way ANOVA (*P = 0.0140; **P = 0.0085; ***P < 0.0001; n.s, not significant). (C) Colony dilution assay for E. coli strains complimented with YcaQ or empty vector. Values are mean ± SEM (n = 3). (D) Quantification of the data shown in panel D. One-way ANOVA values: ***P = 0.0004, n.s., not significant. (E) Colony dilution assay showing the effect of mechlorethamine (HN2) on wild-type E. coli overexpressing YcaQ variants and/or EndoIV. (F) Quantification of the data shown in panel E. One-way ANOVA significance values: *P = 0.0088; **P = 0.0013; ***P < 0.0001; n.s, not significant. (G) MMS EC₅₀ values (mM) for various E. coli deletion (Δ) or over-expression (OE) strains. (H) qRT-PCR results of DNA repair genes after treatment of E. coli with 5 mM MMS or 200 μM mechlorethamine (HN2) for 2 hr. Average ± SEM for three biological replicates.