Common and Unique Genetic Interactions of the Poly(ADP-ribose) Polymerases PARP1 and PARP2 with DNA Double-Strand Break Repair Pathways

Abstract

In mammalian cells, chromatin poly(ADP-ribos)ylation (PARylation) at sites of DNA Double-Strand Breaks (DSBs) is mediated by two highly related enzymes, PARP1 and PARP2. However, enzyme-specific genetic interactions with other DSB repair factors remain largely undefined. In this context, it was previously shown that mice lacking PARP1 and H2AX, a histone variant that promotes DSB repair throughout the cell cycle, or the core nonhomologous end-joining (NHEJ) factor Ku80 are not viable, while mice lacking PARP1 and the noncore NHEJ factor DNA-PKcs are severely growth retarded and markedly lymphoma-prone. Here, we have examined the requirement for PARP2 in these backgrounds. We find that, like PARP1, PARP2 is essential for viability in mice lacking H2AX. Moreover, treatment of H2AX-deficient primary fibroblasts or B lymphocytes with PARP inhibitors leads to activation of the G2/M checkpoint and accumulation of chromatid-type breaks in a lineage- and gene-dose dependent manner. In marked contrast to PARP1, loss of PARP2 does not result in additional phenotypes in growth, development or tumorigenesis in mice lacking either Ku80 or DNA-PKcs. Altogether these findings highlight specific nonoverlapping functions of PARP1 and PARP2 at H2AX-deficient chromatin during replicative phases of the cell cycle and uncover a unique requirement for PARP1 in NHEJ-deficient cells.

Graphical abstract

1. Introduction

Poly(ADP-ribos)ylation (PARylation) is a ubiquitous posttranslational modification that regulates DNA damage signaling and repair, transcription and other cellular processes [1-3]. Among six bona fide cellular poly(ADP-ribose) polymerases (PARPs), PARP1 and PARP2 are nuclear proteins recruited to sites of DNA breaks to promote their repair [4]. While PARP2 provides the bulk of PARP activity in plants [5], PARP1 is thought to generate most PAR within mammalian cells in response to DNA damage [6-8]. PARP1 and PARP2 share a highly homologous carboxi-terminal PARP domain with bona fide polymerase activity [9] but display unique amino-terminal DNA binding domains. In this context, their substrates only partially overlap [10, 11] and the analysis of Parp2^−/− mice has revealed unique functions for PARP2 in lymphoma suppression, spermatogenesis, metabolism and other processes ([12-14]; reviewed in [15]). Furthermore, depletion of both PARP1 and PARP2 leads to early embryonic lethality in the mouse [16], indicating that PARP2-dependent PARylation is not a mere redundant activity.

In the context of single-strand breaks (SSB) repair, both PARP1 and PARP2 PARylate polβ and XRCC1 to promote ligase III-dependent repair [7, 17]. In line with these findings, Parp1^−/− or Parp2^−/− mouse embryonic fibroblasts (MEFs) show delayed kinetics of SSB repair, radiosensitivity, hypersensitivity to alkylating agents and increased frequency of Sister Chromatid Exchanges [7, 16]. These functions of PARP1 and PARP2 in SSB repair have formed the basis for the use of PARP inhibitors (PARPi) as sensitizers to radiation or chemotherapy and for the treatment of cancers with deficiencies in HR factors [18-20].

PARP1 and PARP2 are also recruited to chromatin at sites of DSBs [21], where they promote repair via PARylation of repair factors, histones and chromatin remodeling activities [10, 11, 21-24]. However, PARP1 and PARP2 are recruited to these lesions with different kinetics [25] and recognize differentially processed broken DNA ends [26], suggesting that they may have evolve specialized roles at DSBs. Specifically in the context of the DNA Damage Response (DDR), PARylation was previously shown to cooperate with γ-H2AX to promote BRCA1 recruitment to DSBs [27]. Consistent with this finding, we previously reported that mice lacking PARP1 and H2AX are not viable [28]. Although a role for PARP2 in this setting has not been reported, the observation that loss of PARP2 leads to embryonic lethality in ATM-deficient mice [29] suggests that both factors may become essential in DDR-deficient cells. In addition, PARP1 has been shown to play essential roles at DSBs arising in cells deficient in repair per se. Specifically, PARP1 disruption markedly aggravates the phenotypes of mice deficient for the NonHomologous End-Joining (NHEJ) factors Ku80 [30] or DNA-PKcs [31, 32]. Whether PARP2 may play a similar role in this setting has not been reported.

To better define common and unique functions for PARP1 and PARP2 at sites of DSBs, we have generated mice lacking histone H2AX or the NHEJ factors Ku80 and DNA-PKcs in a PARP2-deficient background. Their analyses demonstrate essential roles for both PARP1 and PARP2 at H2AX-deficient chromatin, while PARP1 is uniquely required at NHEJ-deficient chromatin.

2. Materials and Methods

2.1. Mice

Mice deficient for H2AX [33], PARP1 [34], PARP2 [16], DNA-PKcs [35] or Ku80 [36] were previously described. For experiments with primary lymphocytes, mice were euthanized at 2-4 months of age. For experiments involving lymphomas, the health of mice cohorts was monitored clinically and mice were euthanized when moribund. All mouse experiments included similar numbers of males and females and were conducted in accordance with Institutional Animal Care and Use Committee (IACUC)-approved protocols. Mouse genotypes from tail biopsies were determined using real time PCR with specific probes designed for each gene and validated on positive and negative control tail DNA (Transnetyx, Cordova, TN).

2.2. Cells

Mouse embryonic fibroblasts (MEFs) were obtained from timed matings at E13.5 following standard procedures. B-lineage splenocytes were purified from the spleen by negative selection with CD43 beads (Miltenyi) and activated by incubation with the B cell mitogens α-CD40 antibody (1 μg/mL, BD Pharmingen) and IL-4 (20 ng/mL, R&D Systems), as described [37]. Fresh lymphoma tissue was disaggregated mechanically and cultured briefly in the presence of IL-2 (100 IU/mL) and IL-7 (5 ng/mL), as described [32].

2.3. Drug treatments

Olaparib/AZD2281 (Selleckchem, S1060) and veliparib/ABT-888 (Enzo Life Sciences, ALX-270-444-M001) were stored at −80° C in single use aliquots.

2.4. Telomere FISH

Activated B cells or primary MEFs were incubated in colcemid (KaryoMAX, Gibco), swollen in 30 mM sodium citrate, fixed in methanol/acetic acid (3/1), hybridized with a telomere PNA probe and imaged as described [37]. At least 30 metaphases per sample were scored for chromosomal aberrations.

2.5. Cell cycle analysis

Cells were fixed in cold 70% ethanol, permeabilized in Triton-X, digested with RNAse A and stained with propidium ioidide (PI), as described [38]. Data was acquired using a FACSCalibur and analyzed with FlowJo software.

2.6. Statistical analysis

To determine statistical significance, we performed Student's T test on 3-5 data points per experiment, from 3-5 independent experiments. P≥0.05, not significant (n.s.); P<0.05, *; P<0.01,**; P<0.001,***).

3. Results and Discussion

3.1. PARP2 is required for viability of H2AX-deficient mice

We previously showed that concomitant loss of PARP1 and H2AX results in embryonic lethality in the mouse [28]. To determine whether this genetic interaction is PARP1-specific, we interbred Parp2^+/−/H2afx^+/− mice and genotyped 107 pups from Parp2^+/−/H2afx^+/− intercrosses (n=18 litters; age at genotyping, 7-10 days). We failed to observe Parp2^−/−/H2afx^−/− pups, while all remaining genotypes were observed at near Mendelian ratios (Table 1). Additional breedings of Parp2^−/−/H2afx^+/− or Parp2^+/−/H2afx^−/− females to Parp2^+/−/H2afx^+/− males similarly failed to yield double mutants (not shown). In contrast, Parp2^−/−/H2afx^+/− and Parp2^+/−/H2afx^−/− mice were viable and showed no additional phenotypes on growth and development over their Parp2^−/− or H2afx^−/− control littermates (not shown). Moreover, cohorts of Parp2^−/−/H2afx^+/− (n=12 mice) and Parp2^+/−/H2afx^−/− mice (n=9 mice) followed for an average of 9 months (range, 6-13 months) appeared healthy and elective necropsies didn't reveal the appearance of tumors or other relevant macroscopic abnormalities (not shown). We conclude that both PARP1 and PARP2 are essential for embryonic development in an H2AX-deficient background, paralleling previous observations in the absence of ATM [29]. Together, these findings underscore unique functions for PARP2 at DSBs that cannot be compensated by PARP1.

Table 1.

Mendelian ratios from intercrosses between mice heterozygous for PARP2 (Parp2) and DNA repair factors H2AX (H2afx), DNA-PKcs (Prkdc) or Ku80 (Xrcc5)

	Parp2+/−/H2afx+/− intercrosses			Parp2+/−/Prkdc+/− intercrosses			Parp2+/−/Xrcc5+/− intercrosses
Parp2 genotype	H2afx genotype	Observed (%)	Expected (%)	Prkdc genotype	Observed (%)	Expected (%)	Xrcc5 genotype	Observed (%)	Expected (%)
	H2afx+/+	4 (3.7)	6.7 (6.25)	Prkdc+/+	2 (4.7)	2.7 (6.25)	Xrcc5+/+	14 (7.1)	12.4 (6.5)
Parp2+/+	H2afx+/−	16 (15.0)	13.4 (12.5)	Prkdc+/−	3 (7.0)	5.4 (12.5)	Xrcc5+/−	28 (14.1)	24.8 (12.5)
	H2afx−/−	5 (4.7)	6.7 (6.25)	Prkdc−/−	0 (0.0)	2.7 (6.25)	Xrcc5−/−	10 (5.1)	12.4 (6.5)
	H2afx+/+	17 (15.9)	13.4 (12.5)	Prkdc+/+	5 (11.6)	5.4 (12.5)	Xrcc5+/+	28 (14.1)	24.8 (12.5)
Parp2+/−	H2afx+/−	39 (36.4)	26.8 (25)	Prkdc+/−	13 (30.2)	10.8 (25)	Xrcc5+/−	63 (31.8)	49.5 (25)
	H2afx−/−	8 (7.5)	13.4 (12.5)	Prkdc−/−	4 (9.3)	5.4 (12.5)	Xrcc5−/−	16 (8.1)	24.8 (12.5)
	H2afx+/+	6 (5.6)	6.7 (6.25)	Prkdc+/+	1 (2.3)	2.7 (6.25)	Xrcc5+/+	10 (5.1)	12.4 (6.5)
Parp2−/−	H2afx+/−	12 (11.2)	13.4 (12.5)	Prkdc+/−	11 (25.6)	5.4 (12.5)	Xrcc5+/−	25 (12.6)	24.8 (12.5)
	H2afx−/−	0 (0)	6.7 (6.25)	Prkdc−/−	4 (9.3)	2.7 (6.25)	Xrcc5−/−	4 (2.0)	12.4 (6.5)
	Total	107		Total	43		Total	198

3.2. H2AX suppresses PARPi-induced genomic instability during replication

To gain insights into the mechanisms leading to embryonic lethality in mice lacking H2AX and either PARP1 or PARP2, we quantified genomic instability in metaphase spreads of wt and H2afx^−/− activated B cells treated with PARPi olaparib [39] or veliparib [40], at concentrations known to inhibit both PARP1 and PARP2 (Table 2, Fig. 1). In HR-deficient cells, PARPi induce breaks during replication [18, 19], visualized as chromatid-type breaks in the first mitosis after treatment. We observed no statistically relevant increase in genomic instability H2afx^+/+ B cells after PARPi exposure (Table 2; Fig. 1A). In contrast, both olaparib and veliparib induced a statistically significant increase in the frequency of metaphases containing aberrations and in the number of aberrations per cell in H2afx^−/− B cells (Table 2 for quantification and p values; Fig. 1A). Importantly, most chromosomal aberrations in PARPi-treated H2afx^−/− cells were chromatid-type (60% and 75% of all breaks after olaparib or veliparib, respectively; Table 2; Fig. 1B for examples). These observations were not specific to B cells, because concanavalin A-activated H2afx^−/− T cells similarly harbored frequent chromatid-type chromosomal breaks after exposure to olaparib (Table S1).

Table 2.

Analysis of genomic instability via telomere-FISH in PARPi-treated H2afx^−/− and control primary cells

Genotype	Treatment	# mice	# metaphases	# metaphases with aberrations (%)	# aberrations per metaphase (# chromatid breaks per metaphase)	p valuesa
Activated B cells
H2afx+/+	vehicle	5	150	11 (7.3%)	0.09 (0.01)
	olaparib (1 μM)	5	150	28 (18.7%)	0.31 (0.1)	p=0.18/0.17/0.20
	veliparib (2.5 μM)	3	90	13 (14.4%)	0.3 (0.1)	p=0.21/0.36/0.26
H2afx−/−	vehicle	5	150	55 (36.7%)	0.76 (0.4)
	olaparib (1 μM)	5	158	115 (74.1%)	2.4 (1.4)	p=0.007/0.03/0.02
	veliparib (2.5 μM)	3	90	62 (68.9%)	1.7 (1.3)	p=0.004/0.01/0.008
Primary MEFs
H2afx+/+	vehicle	3	90	8 (8.9%)	0.1 (0.0)
	olaparib (0.5 μM)	3	90	42 (46.7%)	0.8 (0.4)	p=0.01/0.12/0.05
	olaparib (1 μM)	3	90	55 (61.1%)	1.7 (1.3)	p=0.01/0.01/0.05
H2afx+/−	vehicle	4	120	30 (25.0%)	0.4 (0.2)
	olaparib (0.5 μM)	4	120	65 (54.2%)	1.4 (0.9)	p=0.01/0.004/0.02
	olaparib (1 μM)	4	120	75 (62.5%)	1.9 (1.6)	p=0.002/0.02/0.02
H2afx−/−	vehicle	3	67	42 (62.7%)	1.3 (0.4)
	olaparib (0.5 μM)	3	90	84 (93.3%)	4.0 (2.9)	p=0.001/0.009/0.009
	olaparib (1 μM)	3	90	85 (94.5%)	5.6 (4.0)	p=0.003/0.001/0.001

^athe three indicated p values correspond to: % metaphases with aberrations / #chromosome breaks per metaphase / #chromatid breaks per metaphase, respectively.

Figure 1.

H2AX suppresses replication-dependent chromosomal breaks induced by PARP inhibitors in a lineage and gene-dose dependent manner. (A-B) H2afx^−/− and control H2afx^+/+ B cells were purified from the spleen, activated in vitro with cytokines and treated with olaparib (1 μM) or veliparib (3 μM) for 24 hours prior to metaphase analyses by telomere FISH. The number of chromosomal breaks per metaphase is shown and subdivided as “chromosome-type” or “chromatid-type”. Representative examples of complex “chromatid-type” rearrangements observed in H2AX-deficient cells after olaparib treatment are shown in B. (C-D) Primary P₀-P₁ MEFs of the indicated genotypes were treated with either 1 μM olaparib for 24 hours (day 1) or 48 hours (day 2) and harvested for cell cycle analysis after staining with propidium iodide (PI). The percentage of cells with 2N (G1), 2N>4N (S), 4N (G2) and >4N DNA content is shown in C. Each bar represents the average and standard deviation of three MEF cultures derived from 3 independent embryos. Representative examples for each genotype after either 1 day or 2 days of treatment are shown in D. (E-F) Primary P₀-P₁ MEFs of the indicated genotypes were treated with either 0.5 or 1 μM olaparib for 24 hours and metaphases were analyzed by telomere FISH. Bars represent the average and standard deviation of three independent cultures per genotype. Representative examples of chromosomal aberrations observed in wt and H2afx^−/− MEFs exposed to olaparib for 24 hours are shown in F. Yellow arrows point to chromatid-type breaks.

To extend these analyses to a nonlymphoid lineage, we also examined the effect of PARPi on cell cycle distribution and genomic integrity in primary H2afx^−/− and control wt MEFs (Table 2, Figure 1C-F). Because H2afx^−/− MEFs senesce prematurely [41], these studies were done at early passage (P1), when the cell cycle distribution was similar across genotypes (Fig. 1C; “day 0”). Exposure to 1 μM olaparib led to a marked increase in the frequency of cells with 4N DNA content across all genotypes (Fig. 1C-D, “day 1”). However, while cell cycle distribution of wt MEFs had returned to baseline at day 2, persistent accumulation of cells with 4N DNA content was evident in H2afx^−/− cultures (Fig. 1C-D; “day 2”). Because these experiments were done in primary cells with intact checkpoints, these observations strongly suggest persistent DNA damage in the absence of H2AX. In support of this notion, telomere FISH of metaphase spreads in the same cells revealed a marked, dose-dependent increase in the frequency of olaparib-induced chromosomal aberrations in H2afx^−/− MEFs relative to wt controls (Table 2, Fig. 1E for quantification; Fig. 1F for examples). Similar to our observations in H2afx^−/− B cells, olaparib-induced chromosomal aberrations were mostly chromatid-type (Table 2, Fig. 1E-F).

Because H2AX haploinsufficiency compromises DSB repair in some settings [33, 42], we also analyzed PARPi-induced genomic stability in H2afx^+/− B cells and MEFs. We observed a small but statistically significant increase in the frequency of chromosomal aberrations in H2afx^+/− MEFs, albeit only at the lower doses of olaparib (0.5 μM). In contrast, olaparib did not induce genomic instability in H2afx^+/− B cells at the doses used here (data not shown). Altogether, these data suggest that decreased H2AX nucleosomal density may be sufficient to modestly sensitize some lineages to PARP inhibition.

Overall, these data is in line with previous findings in chicken DT40 cells [20] and mouse ES cells [43] that deficiency for H2AX sensitizes cells to PARP inhibitors and demonstrates for the first time that hypersensitivity relates to increased genomic instability. Moreover, cell cycle and cytogenetic analyses demonstrate that all or most PARPi-dependent chromosomal instability in H2AX-deficient cells occurs in the context of replication, leading to frequent chromatid-type aberrations. These findings suggest that PARP1 and PARP2-dependent PARylation and γ-H2AX provide compensatory mechanisms specifically in S and/or G2/M phases of the cell cycle. These findings were unanticipated, because H2AX-deficient cells show primarily a defect in NHEJ [37]. Similarly, treatment of ATM-deficient cells with PARP inhibitors leads to replication-dependent DSBs [44]. Altogether, these findings provide a rationale for the development of PARP1- or PARP2-specific inhibitors for the treatment of cancers lacking ATM [45-50] and/or haploinsufficient for H2AX [51].

3.3. Unlike PARP1, PARP2 does not compromise growth and development of DNA-PKcs-deficient mice

Loss of PARP1 results in embryonic lethality or severe growth defects in NHEJ-deficient backgrounds [30, 31, 38], but a role for PARP2 in this context has not been investigated. To address this issue, we first generated mice doubly deficient for PARP2 and DNA-PKcs, a noncore NHEJ factor. In marked contrast to the severe growth retardation observed in Parp1^−/−/Prkdc^−/− mice [31, 32], Parp2^−/−/Prkdc^−/− mice were born at mendelian ratios and showed no additional phenotypes in growth and development over single mutants (Table 1, Fig. 2A).

Figure 2.

Phenotypic analysis of mice deficient for DNA-PKcs and PARP2. (A) Weight of mice deficient for DNA-PKcs and/or PARP2. Bars represent the average and standard deviation of 3-5 8-12 week-old females. Similar observations were made for males (not shown). (B) Kaplan Meier analyses of cohorts of Parp2^−/−/Prkdc^+/− (n=17), Parp2^+/−/Prkdc^−/− (n=16) and Parp2^−/−/Prkdc^−/− (n=15) mice followed for one year.

PARP2 was previously shown to promote survival of thymocytes by suppressing apoptosis in response to programmed DSBs [52]. Moreover, loss of PARP1 markedly increases the penetrance and decreases the latency of thymic lymphomas in DNA-PKcs-deficient mice [31, 32], but a similar role for PARP2 has not been examined. To address this question, we followed cohorts of Parp2^−/−/Prkdc^−/− and control Parp2^+/−/Prkdc^−/− and Parp2^−/−/Prkdc^+/− littermates until moribund. In contrast to Parp1^−/−/Prkdc^−/− mice, we found that the survival of Parp2^+/−/Prkdc^−/− and Parp2^−/−/Prkdc^+/− mice was comparable (Fig. 2B). Specifically, 14/15 Parp2^−/−/Prkdc^−/− mice and 15/16 Parp2^+/−/Prkdc^−/− mice died within the first year of life. The average age at death was 26.9 weeks (range, 13-40 weeks) and 24.5 weeks (range, 13-42 weeks) for Parp2^−/−/Prkdc^−/− and Parp2^+/−/Prkdc^−/− mice, respectively. Necropsies of moribund animals revealed thymic lymphoma as cause of death in approximately 7/14 and 7/15 Parp2^−/−/Prkdc^−/− and Parp2^+/−/Prkdc^−/− mice, respectively. In this subset of lymphoma-bearing mice, the average age at death was 29.2 weeks (range, 13-40 weeks; n=7) and 23.1 weeks (range, 14-27 weeks; n=7) for Parp2^−/−/Prkdc^−/− and Parp2^+/−/Prkdc^−/− mice, respectively. Consistent with previous observations in Parp2^−/− mice [16], 17/17 Parp2^−/−/Prkdc^+/− littermate mice appeared healthy at the end of the one-year observation period (Fig. 2B) and elective necropsies failed to reveal tumors or other macroscopic alterations (not shown). We conclude that, unlike PARP1, PARP2 does not mediate tumor suppression in thymocytes deficient for DNA-PKcs.

3.4. PARP2 is dispensable for the fusion of DNA-PKcs-deficient telomeres

DNA-PKcs functions to suppress telomere fusions in mouse and human cells [53, 54] and, consistent with these previous observations, control Parp2^+/−/Prkdc^−/− thymic lymphomas accumulate telomere fusions (Fig. S1A for quantification; Fig. S1B for representative metaphases). Because PARP2 binds the shelterin component TRF2 [55] and suppresses chromosomal translocations of programmed DSBs in B cells undergoing Class Switch Recombination (CSR) [56], we investigated whether PARP2 functions to suppress the fusion of DNA-PKcs-deficient telomeres. To this end, we quantified telomere fusions in five Parp2^−/−/Prkdc^−/− thymic lymphomas (Figs. S1A-B). Three out of five double mutant tumors contained telomere fusions in all or most cells, while the remaining two tumors contained fusions in a smaller fraction of cells. Moreover, the frequency and patterns of telomere fusions in Parp2^−/−/Prkdc^−/− thymic lymphomas were similar to those observed in Parp2^−/−/Prkdc^−/− lymphomas.

We previously showed that most Parp1^−/−/Prkdc^−/− lymphomas harbor oncogenic mutations in p53, leading to aneuploidy and amplification of telomere fusions [32]. In contrast, cytogenetic analysis of Parp2^−/−/Prkdc^−/− lymphomas revealed that all tumors were euploid or near-euploid, similar to control Parp2^-+-/Prkdc^−/− tumors (Fig. S2A for quantification; Fig. S2B for examples of representative metaphases). These data strongly suggest that PARP2 does not play a significant role in the suppression of oncogenic p53 mutations in DNA-PKcs-deficient thymocytes.

3.5. Mice lacking PARP2 and Ku80 are viable and show no significant additional phenotypes in growth and development

Because DNA-PKcs is required for a subset of NHEJ reactions, we next extended our analysis to a core NHEJ factor, Ku80 (gene symbol, Xrcc5) [57]. To this end, we bred Parp2^+/−/Xrcc5^+/− males and females to generate double mutants (expected ratio, 1/16). Genotyping of n=198 pups at day 7 of life revealed similar numbers of Ku80-deficient mice regardless of Parp2 copy number (Table 1). Like control Parp2^+/+/Xrcc5^−/− mice [36], Parp2^+/−/Xrcc5^−/− and Parp2^−/−/Xrcc5^−/− littermate mice were markedly growth retarded and about half of the mice died prior to weaning. However, weaned double mutant mice appeared robust and a few mice survived when followed for up to six months (not shown). Together with our analysis of Parp2^−/−/Prkdc^−/− mice above and previous observations of early embryonic lethality in Parp1^−/−/Xrcc5^−/− mice [30], these findings indicate that PARP1 plays unique, nonoverlapping functions in the signaling and/or repair of DSBs that arise in the absence of NHEJ. In this regard, PARP1 has been implicated in end-joining via an alternative, ligase III-dependent pathway (A-NHEJ) [58-60] and cooperates with DNA-PKcs in the restart of a subset of stalled replication forks [61]. In the future, it will be important to determine PARP1 unique substrates and functions in these contexts.

4. Conclusions

PARP2 is closely related to PARP1, yet their unique amino-terminal domains confer them with unique substrates and functions. To define these in the context of DNA double-strand break (DSB) repair, we have generated here three novel mouse models of PARP2 deficiency, Parp2^−/−/H2afx^−/−, Parp2^−/−/Prkdc^−/− and Parp2^−/−/Xrcc5^−/− mice. The analyses of these compound models have uncovered a novel essential function for PARP2 in concert with H2AX during replicative phases of the cell cycle. In contrast, we find that the genetic interaction between PARP1 and NHEJ, a DSB repair pathway that operates primarily in pre-replicative phases of the cell cycle, is not shared with PARP2. Overall, these findings have broad implications for understanding how PARylation cooperates with other cellular pathways to maintain genomic integrity, thereby suppressing organismal phenotypes of aging and cancer. In the future, the analyses of novel genetic models for conditional inactivation of PARP1 and/or PARP2 in somatic cells deficient for H2AX or NHEJ factors will further define the mechanisms underlying their common and unique functions.