Poly(ADP-ribose) polymerases covalently modify strand break termini in DNA fragments in vitro

Abstract

Poly(ADP-ribose) polymerases (PARPs/ARTDs) use nicotinamide adenine dinucleotide (NAD⁺) to catalyse the synthesis of a long branched poly(ADP-ribose) polymer (PAR) attached to the acceptor amino acid residues of nuclear proteins. PARPs act on single- and double-stranded DNA breaks by recruiting DNA repair factors. Here, in in vitro biochemical experiments, we found that the mammalian PARP1 and PARP2 proteins can directly ADP-ribosylate the termini of DNA oligonucleotides. PARP1 preferentially catalysed covalent attachment of ADP-ribose units to the ends of recessed DNA duplexes containing 3'-cordycepin, 5'- and 3'-phosphate and also to 5'-phosphate of a single-stranded oligonucleotide. PARP2 preferentially ADP-ribosylated the nicked/gapped DNA duplexes containing 5'-phosphate at the double-stranded termini. PAR glycohydrolase (PARG) restored native DNA structure by hydrolysing PAR-DNA adducts generated by PARP1 and PARP2. Biochemical and mass spectrometry analyses of the adducts suggested that PARPs utilise DNA termini as an alternative to 2'-hydroxyl of ADP-ribose and protein acceptor residues to catalyse PAR chain initiation either via the 2',1″-O-glycosidic ribose-ribose bond or via phosphodiester bond formation between C1' of ADP-ribose and the phosphate of a terminal deoxyribonucleotide. This new type of post-replicative modification of DNA provides novel insights into the molecular mechanisms underlying biological phenomena of ADP-ribosylation mediated by PARPs.

Figure 1.

Schematic presentation of various DNA substrates used in this study.

Figure 2.

PARP-catalysed formation of high-molecular-weight (HMW) DNA products from oligonucleotide duplexes. (A) Denaturing PAGE analysis of PARP1-generated HMW products (incubation with 3′-³²P-cordycepin-labelled 40-mer nicked, gapped or recessed DNA). (B) Denaturing PAGE analysis of PARP2-generated HMW products (incubation with 5′-[³²P]labelled 40-mer nicked, gapped or recessed DNA). (C) Graphic representation of the average numerical data on the formation of the HMW products by PARPs from gapped, nicked or recessed DNA duplexes. Each bar represents PARP activity as mean ± SD from three independent experiments; (D) PARP1-catalysed PARylation in the presence of varying concentrations of NAD⁺. Fifty nM PARP1 or 10 nM PARP2 was incubated with 20 nM DNA substrate and varying concentrations of NAD⁺ (5 nM to 1 mM) for 30 min at 37°C. The reaction products were analysed as described in Materials and Methods. Arrows indicate HMW and LMW PAR–DNA products and the 21-mer free oligonucleotide.

Figure 3.

Effects of the type of DNA structures and termini on the PARP1- or PARP2-catalysed formation of PAR–DNA adducts. Fifty nM PARP1 or PARP2 was incubated with 20 nM [³²P]labelled oligonucleotide and 1 mM NAD⁺ for 30 min at 37°C. The products of the reaction were separated using denaturing PAGE and the relative amounts of the PAR–DNA products were measured. (A) Effects of duplex and single-stranded DNA structures on PARP-catalysed DNA PARylation. (B) Effects of the 3′-terminal ribonucleotide in oligodeoxynucleotides on PARP1- or PARP2-catalysed PARylation of oligonucleotides. ‘Rib’ in structures № 3, 4, 7 and 8 indicates an adenosine ribonucleotide. The data on PARP-catalysed formation of PAR-DNA products are presented as mean ± SD from three independent experiments.

Figure 4.

Analysis of the products of enzymatic digestion of the PAR–DNA adducts. (A) Graphical representation of formation and chemical structure of poly(ADP-ribose) polymers. (B) Denaturing PAGE analysis of the products of PARG- and CIP-catalysed digestion of the 5′-[³²P]labelled PAR–DNA products. (C) Denaturing PAGE analysis of the products of CIP- and SVPDE1-catalysed digestion of the PARP1-generated 3′-³²P-cordycepin-labelled PAR–DNA products. To generate the PAR–DNA products, 20 nM 3′dAM³²P-ExoA•RexT^rec was incubated with 50 nM PARP1 or PARP2 in the presence of 1 mM NAD⁺ for 30 min at 37°C. After that, the samples were heated for 20 min at 80°C and then incubated with 50 pg/μl PARG (in PARP1 buffer), 0.1 U SVPDE1 (in SVPDE1 buffer) or 10 U CIP (in CIP buffer) for 60 min or 30 min at 37°C. Arrows depict HMW PAR–DNA products and free oligonucleotides.

Figure 5.

Separation and characterisation of PAR–DNA mono-adducts by denaturing PAGE and TLC. (A) The long-run denaturing PAGE analysis of the products of SVPDE1-catalysed digestion of the 3′-³²P-cordycepin- and 5′-[³²P]labelled PAR–DNA adducts, free DNA and the PAR polymer. The samples were desalted before loading on the gel. (B) As in panel A but short-run denaturing PAGE. The samples were not desalted before loading on the gel. (C) TLC separation of the products of SVPDE1-catalysed digestion of the 3′-³²P-cordycepin- and 5′-[³²P]labelled PAR–DNA adducts, free DNA and the PAR polymer. The samples were not desalted before loading on the plate. (D) Graphical representation of chemical structures of the phosphoribosyl adenosine monophosphate adducts. The pRib-AMP adduct is generated by digestion of the [³²P]labelled PAR by SVPDE1, whereas the 2′-pRib-3′dAMP and 5′P-pRib-dG adducts are generated by digestion of the 3′-³²P-cordycepin-labelled and 5′-[³²P]labelled PAR–DNA by SVPDE1, respectively. Arrow ‘X’ indicates the putative 2′-pRib-3′dAMP adduct, ‘Y’ points to traces of a putative 3′-terminal dAMP-3′dAMP dinucleotide, ‘Z’ indicates the putative 5′P-pRib-dG adduct containing a pRib moiety covalently attached to 5′P of dGMP, ‘AM³²P’ means adenosine 5′-[³²P]monophosphate, ‘pRib-AM³²P’ stands for pRib-AMP with adenosine 5′-[³²P]monophosphate, ‘³²P-NAD⁺’ means [adenylate-³²P]NAD⁺, ‘3′dAM³²P’ is cordycepin 5′-[³²P]monophosphate, ‘dGM³²P’ denotes 2′-deoxyguanosine 5′-[³²P]monophosphate, and ³²P means free phosphate.

Figure 6.

MALDI-TOF MS analysis of the mono-ADP-ribosylated oligonucleotide product resulting from incubation of the nicked 5′-phosphorylated oligonucleotide duplex with PARP2 and PARG. The 5′-phosphorylated ExoA•RexT^nick duplex was PARylated by means of PARP2 and then treated with PARG. Bands corresponding to the mono-ADP-ribosylated 5′-phosphorylated ExoA product were purified by denaturing PAGE and mixed with cold 5′-phosphorylated ExoA•RexT^rec duplex for further analysis. (A) A MALDI-TOF spectrum of the control mock-treated phosphorylated ExoA•RexT^rec duplex; (B) MALDI-TOF spectrum of the unpurified PARP2 reaction products with addition of the purified 21-mer ExoA-mono-ADP-ribose fragment. For details see Materials and Methods.

Figure 7.

Analysis of the products of NUDT16-catalysed hydrolysis of PAR-DNA adducts generated by PARP1 and PARP2. The 10 nM 3′dAM³²P-labelled and 3′-[³²P]labelled ExoA•RexT^rec duplexes were incubated with 100 nM PARP1 and 1 mM NAD⁺, and the 40 nM 5′-[³²P]labelled ExoA•RexT^nick duplex was incubated with 100 nM PARP2 and 1 mM NAD⁺ at 37°C for 30 min. After incubation with PARPs, the samples were heated for 20 min at 80°C and the resulting [³²P]labelled HMW products were further incubated with 50 pg/μl PARG, 1 U SVPDE1, 10 U CIP or 2–20 μM NUDIX16. (A) Denaturing PAGE analysis of the products of NUDT16 and CIP catalysed hydrolysis of the PARP1-generated 3′-[³²P]labelled PAR-DNA adducts. To generate a DNA duplex containing a 3′-terminal ³²P residue, the 3′dAM³²P-labelled ExoA•RexT^rec duplex was treated with Tdp1. Lanes 1–8, 3′dAM³²P-labelled ExoA•RexT^rec duplex; lanes 9–17, 3′-[³²P]labelled ExoA•RexT^rec duplex. (B) Denaturing PAGE analysis of the products of PARG-, SVPDE1-, CIP- and NUDT16-catalysed hydrolysis of the PARP2-generated 5′-[³²P]labelled PAR-DNA adducts (lanes 1–9). Arrows indicate phosphoribosylated (Rib-p), ribosylated (Rib) and native [³²P]labelled 21-mer and 22-mer oligonucleotides, ‘*p’ stands for a labelled phosphate residue. For more details, see Materials and Methods.