Mammalian N1-adenosine PARylation is a reversible DNA modification

Abstract

Poly-ADP-ribosylation (PARylation) is regarded as a protein-specific modification. However, some PARPs were recently shown to modify DNA termini in vitro. Here, we use ultrasensitive mass spectrometry (LC-MS/MS), anti-PAR antibodies, and anti-PAR reagents to show that mammalian DNA is physiologically PARylated and to different levels in primary tissues. Inhibition of PAR glycohydrolase (PARG) increases DNA PARylation, supporting that the modification is reversible. DNA PARylation requires PARP1 and in vitro PARP1 PARylates single-stranded DNA, while PARG reverts the modification. DNA PARylation occurs at the N1-position of adenosine residues to form N1-Poly(ADP-ribosyl)-deoxyadenosine. Through partial hydrolysis of mammalian gDNA we identify PAR-DNA via the diagnostic deamination product N1-ribosyl-deoxyinosine to occur in vivo. The discovery of N1-adenosine PARylation as a DNA modification establishes the conceptual and methodological framework to elucidate its biological relevance and extends the role of PARP enzymes.

Fig. 1. Mouse and human genomic DNA is PARylated.

Dot blot analysis for PARylation of mESC, HEK293T and plasmid DNA (a), mESC DNA treated with PARG (b), DNA purified from mESCs treated with the PARG inhibitor PDD00017273 (c), DNA purified from mESCs treated with the PARP inhibitor olaparib (d), genomic mESC DNA after IP with IgG or anti-dsDNA antibody (e). gDNA was serially diluted (2×) in a and c, d; MB, methylene blue. f Southwestern blot analysis for PARylation of mESC DNA treated with EcoRI, MseI, Proteinase K, and PARG as indicated. Length of marker DNA is shown on the left. Southwestern is representative of three independent experiments with similar outcomes. g LC-MS/MS quantification of ribosyl-adenosine (R-Ado) on HEK293T DNA treated as indicated. Samples were repurified after enzyme treatments by a second column-based DNA purification to remove any PAR and DNA monomers. Dashed line, limit of detection (LOD). Data are presented as mean values ± s.d. of three biological replicates; n.d., not detected. h–j LC-MS/MS quantification of R-Ado (h) of mESC DNA isolated by two different DNA preparation kits (Qiagen Blood & Cell culture DNA kit or Macherey-Nagel NucleoBond HMW DNA kit; data are presented as mean values ± s.d. of three biological replicates; n.s., not significant by two-sided, unpaired t test for unequal variances), i of DNA from the indicated adult female mouse tissues (data are presented as mean values ± s.d. of three individual mice), j of DNA from mESCs treated with increasing amounts of the PARG inhibitor PDD00017273. R-Ado levels from mock-treatment is arbitrarily set to 1 (data are presented as mean values ± s.d. of three biological replicates; indicated p values and not significant (n.s.) as by two-sided Dunnett’s test). Source data are provided as a Source Data file.

Fig. 2. DNA PARylation requires PARP1 but is not triggered by strand breaks in vivo.

LC-MS/MS quantification of ribosyladenosine (R-Ado) in 293T (a) and mESC (b) DNA after siRNA depletion of selected PARPs. Data are presented as mean values ± s.d. of three biological replicates; indicated p values as by two-sided Dunnett’s test. c γH2AX immunofluorescence of DIvA cells in which AsiSI nuclease expression was induced by 4-OHT treatment. DNA is stained with DAPI. Immunofluorescence is representative of three independent experiments with similar outcomes. d Dot blot analysis for PARylation of DNA from AsiSI nuclease expressing DIvA cells treated as in c with serially diluted (2×) DNA. e Top: γH2AX ChIP-qPCR analysis of selected genomic loci (shown at the bottom) of DIvA cells previously described to be associated with (+) or without (−) γH2AX upon 4-OHT treatment to induce AsiSI nuclease. Bottom: PAR-DIP-qPCR analysis of the same loci as used for top panel. Target regions are named by chromosomal location of the AsiSI recognition site in the human hg38 reference assembly. Data are presented as mean values ± s.d. of three biological replicates; indicated p values and not significant (n.s.) over mock as by two-sided, unpaired t test for unequal variances. Source data are provided as a Source Data file.

Fig. 3. PARP1 PARylates ssDNA at a nucleobase.

a Scheme of substrates #1–#4 used for in vitro PARylation assays. The 30 nt black strand is designated as standard oligo in the main text. ³²P, 5’-phosphate with ³²P-radiolabel. b–d, Autoradiography of denaturing PAGE of reaction products from in vitro PARylation assay with substrates shown in a in presence or absence of human PARP1, NAD⁺ and the PARP1 inhibitor olaparib. In d, substrate #3 contained either a 2′-dC or a 2′,3′-ddC terminal nucleotide and was treated with or without calf intestine phosphatase (CIP) post reaction. The autoradiographs are representative of three independent experiments with similar outcomes. e–g LC-MS/MS chromatograms of reaction products from in vitro PARylation assays in presence of native or heat-denatured PARP1: e with or without substrate #3, NAD⁺ and olaparib as indicated. Left and right panels show mass transitions corresponding to mass shifts expected for loss of a deoxyribose + ribose (m/z 385 → 137) and loss of a single deoxyribose (m/z 385 → 269) from the parental ‘nucleoside 385’. -*, heat-denatured PARP1. f with substrates #1–#4 as shown in a. Mass transition is shown as expected for loss of a deoxyribose + ribose (m/z 385 → 137) from the parental molecule ‘nucleoside 385’. g with standard substrate #3, and three additional ssDNA oligonucleotides (RexT, 40 nt, 83 nt). The signal of the lower two chromatograms is magnified on the right. h Scheme for two alternative outcomes of de-PARylation of PARylated DNA with PARG. (i) PARG hydrolyzes the ADP ribose polymer but not the terminal ADP ribose unit at the DNA base acceptor side. LC-MS/MS analysis of the reaction products would yield decreased R-Ado and unchanged ‘nucleoside 385’ levels. (ii) In addition to polymer degradation, PARG cleaves the linkage of the ADP ribose and the DNA base acceptor side resulting in both decreased R-Ado and ‘nucleoside 385’ levels. i LC-MS/MS quantification of R-Ado and ‘nucleoside 385’ on an 83mer ssDNA oligo after in vitro PARylation by PARP1 followed by mock or PARG treatment of the reaction products as indicated. Data are presented as mean values ± s.d. of three independent experiments; indicated p values as by two-sided, unpaired t test for unequal variances. j PARP1 forms ‘nucleoside 385’ with single stranded genomic DNA. LC-MS/MS chromatograms for ‘nucleoside 385’ of reaction products from in vitro PARylation. Genomic DNA from 293T cells pretreated or not with Nuclease S1 was incubated with PARP1 and NAD⁺. Source data are provided as a Source Data file.

Fig. 4. DNA PARylation occurs at the N1-position of adenosine residues in vitro and in vivo.

a LC-MS/MS chromatograms of reaction products from in vitro PARylation assays in presence of native or heat-denatured PARP1 with four 83mer ssDNA substrates each bearing one type of heavy isotope-labeled nucleoside as indicated. Top, scan for signals with m/z transitions expected for a mass shift of ‘nucleoside 385’ from each label. Bottom, detection of signals that correspond to non-labeled ‘nucleoside 385’. Note, all four substrates contain a mixture of the respective labeled and unlabeled nucleoside. b LC-MS/MS chromatograms of reaction products from in vitro PARylation assays as in a but with a ¹⁵N₅-dA-labeled 83mer ssDNA substrate. Reaction products were scanned for signals with m/z transitions that correspond to mass shifts of +5 to +1 of ‘nucleoside 385’ (from top to bottom). c Spontaneous deamination of N1-ribosyl dA (r-dA, attached ribose moiety in red) leads to N1-ribosyl-dI (r-dI, ribose moiety and O⁶ in red). d LC-MS/MS chromatograms of reaction products from in vitro PARylation assays as in a but with substrate #3 in which reaction products were either processed at 95 °C or 20 °C before mass spec analysis, and screened for signals with m/z transitions expected for N1-ribosyl-dA (r-dA, m/z 384 → 136, top) or N1-ribosyl-dI (r-dI, m/z 385 → 137, bottom). e LC-MS/MS chromatograms of N1-ribosyl-dI, (arrows, m/z 385 → 137) enriched from 5 mg pig liver and 2.5 mg mouse kidney DNA and compared to the reaction product of an in vitro PARylation assay with PARP1 on substrate #3.