PARP3 is a sensor of nicked nucleosomes and monoribosylates histone H2B(Glu2)

Abstract

PARP3 is a member of the ADP-ribosyl transferase superfamily that we show accelerates the repair of chromosomal DNA single-strand breaks in avian DT40 cells. Two-dimensional nuclear magnetic resonance experiments reveal that PARP3 employs a conserved DNA-binding interface to detect and stably bind DNA breaks and to accumulate at sites of chromosome damage. PARP3 preferentially binds to and is activated by mononucleosomes containing nicked DNA and which target PARP3 trans-ribosylation activity to a single-histone substrate. Although nicks in naked DNA stimulate PARP3 autoribosylation, nicks in mononucleosomes promote the trans-ribosylation of histone H2B specifically at Glu2. These data identify PARP3 as a molecular sensor of nicked nucleosomes and demonstrate, for the first time, the ribosylation of chromatin at a site-specific DNA single-strand break.

Figure 1. PARP3 promotes chromosomal SSBR and is stimulated by SSBs with canonical termini.

(a) Wild-type (WT) DT40 cells, PARP3^−/− (PARP3 KO) DT40 cells, and PARP3^−/− DT40 cells stably transfected with either empty vector (vector) or vector encoding human recombinant PARP3 (hPARP3) were treated with the indicated doses of γ-rays and survival calculated in clonogenic assays. Data are the mean (±s.e.m.) of three independent experiments. Where not visible, error bars are smaller than the symbols. (b) WT, PARP1^−/−, or PARP3^−/− DT40 cells were treated on ice with γ-rays (20 Gy) and incubated for the indicated times to allow repair. DNA strand breaks were quantified (tail moment) by alkaline comet assays. Data are the average tail moment of >50 cells per sample and are the mean of three independent experiments (±s.e.m.). (c) WT, PARP3^−/−, or derivatives of PARP3^−/− DT40 cells complemented with empty vector or hPARP3 were treated on ice with γ-rays (20 Gy) and incubated for the indicated times to allow repair. DNA strand breaks were quantified as above. (d) WT, KU70^−/−, or XRCC3^−/− DT40 cells were treated on ice with γ-rays (20 Gy) and incubated for the indicated times to allow repair. DNA strand breaks were quantified as above. ANOVA was employed to compare mutant DT40 for significant differences with WT (**P<0.01. ‘ns'; not significant). Data are the mean (±s.e.m.) of three independent experiments. (e) hPARP1 and/or hPARP3 (50 nM) was incubated with 12.5 μM biotin-NAD⁺ and 200 ng uncut or nicked plasmid (nicked with Nt.BsmA1; nick concentration of 32 nM) that was pretreated or not as indicated with CIP to dephosphorylate 5′-termini. Reaction products were separated by SDS–PAGE and blotted with streptavidin-HRP. (right) Aliquots of uncut, nicked, and linear plasmid were analysed by agarose gel electrophoresis and staining with ethidium bromide. ANOVA, analysis of variance; CIP, calf intestinal phosphatise; HRP, horseradish peroxidase

Figure 2. Mechanism of DNA break binding by the cPARP3 WGR domain.

(a) Schematic representation of PARP3 domains with the human (top) and chicken (bottom) amino acid positions indicated. The WGR domain is shown in expanded format below showing the HSQC perturbed residues of chicken PARP3 (bottom) and the equivalent residues in human PARP3 (top). (b) Overlay of ¹H–¹⁵N HSQC NMR full spectra for cPARP3^1–169 in the absence (blue) or presence (red) of oligodeoxyribonucleotide duplex harbouring a 5′-phosphorylated nick (protein:DNA ratios of 1:0 and 1:2, respectively). (c) Expanded view of the small boxed region shown in b demonstrating the chemical shifts induced in cPARP3^1–169 by different concentrations of nicked DNA. Protein:DNA ratios were 1:0 (that is, no DNA; blue), 5:1 (magenta), 1:1 (green) and 1:2 (red). (d) Map of significant chemical shifts induced in cPARP3^1–169 by DNA duplex harbouring a 5′-phosphorylated nick (>0.1  p.p.m) or 10-bp 3′-overhang with a recessed 5′-phosphorylated terminus (>0.04), surface modelled using CS-Rosetta. Residues with a significant chemical shift in the presence of either a nick (blue) or 3′-overhang (green) or both (red) are indicated. (e) Electrostatic surface of modelled cPARP31–169 with nicked DNA. (f) Model of cPARP3^1–169 with nicked DNA, depicting residues with significant chemical shifts as above. (g) Model of cPARP3^1–169 with nicked DNA lacking the strand located upstream (5′) of the nick (that is, harbouring a DSB with 10-bp 3′-overhang). Residues exhibiting a significant chemical shift are indicated as above.

Figure 3. The PARP3 DNA-binding interface is required for PARP3 stimulation and accumulation at chromosome DNA damage.

(a) Wild-type or the indicated mutant full-length cPARP3 (300 nM) was incubated for 20 min at room temp with biotin-NAD⁺ (12.5 μM) and 200 nM of oligonucleotide duplex harbouring either a 5′-phosphorylated nick or 5′-phosphorylated DSB with 3′-overhang. Reaction products were separated by SDS–PAGE, blotted, and detected with streptavidin-HRP. Autoribosylated cPARP3 was quantified and plotted relative to that generated in reactions containing nicked duplex and wild type cPARP3. Data are the mean (±s.e.m.) from three independent experiments. (b) Time-course of wild-type or mutant cPARP3 incubated from 0 to 30 min in the same conditions as above. (c) Recombinant wild-type or mutant cPARP3 (0–0.8 μM) was incubated with a 3′-fluorescein isothiocyanate (FITC)-labeled oligonucleotide duplex harbouring a 5′-phosphorylated nick (100 nM), and protein-DNA complexes detected by EMSA. (d) Recruitment of wild-type and mutant human PARP3-GFP to sites of UVA-laser DNA damage in human U2-OS cells. (left) Representative images of WT and mutant PARP3-GFP before treatment (Unt) and 1 min after laser damage. (top right) Quantification of GFP accumulation at sites of laser damage (% increase over initial level). Data are the mean (±s.e.m.) of 25 or more cells per sample. The hPARP3 WGR mutations were Y83A, W101L and R103N and H384A/E514A in the catalytic domain (denoted ‘CM'). (e, top) PARP3^−/− DT40 cells stably transfected with either empty vector (vector) or vector encoding wild-type hPARP3 (WT) or the mutant derivatives Y83A, W101L and R103N were treated on ice with γ-rays (20 Gy) and incubated for the indicated times to allow repair. DNA strand breaks were quantified (tail moment) by alkaline comet assays. The inset is a western blot showing the expression level of wild type and mutant hPARP3 in PARP3^−/− DT40 cells. (bottom) The above DT40 cell lines were treated with the indicated doses of γ-rays and survival quantified in clonogenic assays. Data are the mean (±s.e.m.) of three independent experiments. Where not visible, error bars are smaller than the symbols. HRP, horseradish peroxidase.

Figure 4. PARP3 monoribosylates H2B in damaged chromatin.

(a, left) 10μg of the chicken chromatin employed in these experiments was fractionated by SDS–PAGE and stained with Coomassie blue. (right) One microgram of soluble MNase-treated chicken chromatin or 50-mer oligonucleotide duplex (200 nM) harbouring a nick with 3′-P/5′-OH termini was mock-treated (0) or treated with 1, 0.5 or 0.25 U T4 PNK to restore 3′-OH/5′-P termini. These DNA substrates were then incubated with 100 nM hPARP3 and 12.5 μM biotin-NAD⁺ for 30 min and biotinylated products separated by 15% SDS–PAGE and detected with streptavidin-HRP. (b) 1 μg chicken chromatin or the indicated recombinant histone was incubated with 100 nM hPARP3 in the presence of 300 nM ³²P-NAD⁺ or 12.5 μM biotin-NAD and oligonucleotide harbouring either a DSB (middle) or SSB (right) and the reaction products fractionated by 15% SDS–PAGE and detected by autoradiography or streptavidin-HRP. (left) An aliquot of the chicken chromatin and recombinant histones was fractionated by SDS–PAGE and stained with Coomassie blue. (c, left) Aliquots of recombinant histone standards were fractionated separately or together as an octamer on triton-acid urea gels and analysed by staining with Coomassie blue. (right) The products of the PARP3 ribosylation reactions conducted in b were fractionated on triton-acid urea gels and analysed by autoradiography. HRP, horseradish peroxidase.

Figure 5. PARP3 binds nicked nucleosomes and ribosylates H2B^E2.

(a) ADP-ribosylation of E2 in recombinant H2B by hPARP3. MS/MS fragmentation profile of doubly charged PE(+15)PAKSAPAPK (theoretical: 554.3115 m/z; observed: 554.3113 m/z; 0.4 p.p.m. mass error), indicating the position of the NH (+15.0109 Da) moiety on glutamate resulting from hydroxylamine derivatization of ADP-ribose. Preferred fragmentation N-terminal to prolines is observed, as expected. (b) Mutation of E2 reduces ribosylation of H2B by hPARP3. The products of ribosylation reactions containing 200 nM hPARP3, 12.5 μM biotin-NAD⁺, 100 nM nicked oligonucleotide duplex (50 bp), and 5 μM of wild-type or mutant recombinant H2B were separated on 15% SDS–PAGE gels and detected by autoradiaography. (c) PARP3 binds to nicked mononuclesomes. (left) Reconstituted mononucleosomes were assembled on intact or nicked DNA (Widom positioning sequence 601.2) and nucleosome quality assessed by native gel electrophoresis. (right) representative images of negative stained intact (top) or nicked (bottom) nucleosomes incubated with His-tagged cPARP3, and with the His-tagged protein detected by nanogold Ni-NTA. Scale bars, 50nm. (d) PARP3 ribosylates H2B^E2 in reconstituted nicked mononucleosomes. Hundred nanomolar of intact or nicked 601.2 DNA, present either as naked duplex or within reconstituted nucleosomes containing wild-type or mutant H2B, was incubated with 100 nM hPARP1 (lanes 1–6) or hPARP3 (lanes 7–12) and either 12.5 μM biotin-NAD⁺ (hPARP3) or 1.5 μM biotin-NAD⁺ (hPARP1, to encourage shorter chain modifications). MS/MS, tandem mass spectrometry.