Nonhomologous end-joining promotes resistance to DNA damage in the absence of an ADP-ribosyltransferase that signals DNA single strand breaks

ADP-ribosylation is the addition of single or multiple ADP-ribose moieties to target proteins (D'Amours et al., 1999). ADP-ribosyltransferases (ARTs), the enzymes that catalyse this post-translational modification, are found in a variety of eukaryotes, with 22 genes containing a predicted ART catalytic domain identified in humans (Hottiger et al., 2010). ADP-ribosylation has been implicated in regulating a variety of cellular processes including DNA repair, cell growth and differentiation, transcriptional regulation and programmed cell death (Hottiger et al., 2010; Messner and Hottiger, 2011; Quénet et al., 2009).

The best defined role of ARTs is in the cellular response to DNA damage, with particular reference to DNA strand breaks. PARP1, the founding member of the ART family, is recruited to single strand breaks (SSBs), becomes activated, and modifies a number of proteins at DNA lesions, including itself (Caldecott, 2008; Krishnakumar and Kraus, 2010). Resolution of SSBs is reduced in the presence of ART inhibitors (ARTi), or following depletion of PARP1 (Ding et al., 1992; Durkacz et al., 1980; Fisher et al., 2007; Morgan and Cleaver, 1983; Schraufstatter et al., 1986). Similarly, mice disrupted in the parp1 gene are sensitive to DNA damaging agents that induce base damage or strand breakage and exhibit delayed kinetics of SSB repair (SSBR) (de Murcia et al., 1997; Le Page et al., 2003; Masutani et al., 1999; Trucco et al., 1998; Trucco et al., 1999). More recently, a second ART (PARP2) has been implicated in SSBR. Similar to parp1^–/– cells, parp2^–/– mice are sensitive to DNA damage, exhibit increased chromosome instability and a delay in repair following exposure to alkylating agents (Ménissier de Murcia et al., 2003; Schreiber et al., 2002). While the precise mechanism by which ARTs regulate SSBR remains to be defined, ADP-ribosylation of target proteins, including histones, is thought to facilitate chromatin relaxation around the break and to recruit chromatin remodelling enzymes to DNA lesions (Ahel et al., 2009; Gottschalk et al., 2009; Mehrotra et al., 2011; Timinszky et al., 2009). An additional non-mutually exclusive function of ADP-ribosylation is to enrich repair factors at DNA lesions through poly-ADP ribose (PAR) interaction domains located in these proteins (Bekker-Jensen et al., 2007; Caldecott et al., 1996; El-Khamisy et al., 2003; Iles et al., 2007; Kanno et al., 2007; Masson et al., 1998; Okano et al., 2003; Rulten et al., 2008).

In addition to regulating SSBR, ARTs are also required for resolution of DNA double strand breaks (DSBs). ARTi sensitise mammalian cells to DSBs (Audebert et al., 2004; Boulton et al., 1999) and delay repair of this variety of DNA lesion (Rulten et al., 2011). DNA DSBs can be repaired by homologous recombination (HR) or nonhomologous end-joining (NHEJ) (Branzei and Foiani, 2008). PARP1 has been implicated in HR-mediated repair of DSBs at stalled and/or damaged replication forks (Bryant et al., 2009; Sugimura et al., 2008; Yang et al., 2004), in addition to alternative NHEJ (A-NHEJ), an end-joining pathway that is activated in the absence of core NHEJ factors such as DNA-PKCs and Ku (Audebert et al., 2004; Brown et al., 2002; Robert et al., 2009; Wang et al., 2006). However, its role in classic NHEJ is less clear. While PARP1 interacts with Ku and DNA-PKcs (Ariumi et al., 1999; Galande and Kohwi-Shigematsu, 1999; Li et al., 2004; Ruscetti et al., 1998), classic NHEJ is normal in murine parp1^–/– cells (Yang et al., 2004), and depletion of PARP1 or PARP2 has little impact on DSB resolution (Rulten et al., 2011). Instead, we and others recently identified a third ART (PARP3/Adprt1a) that becomes activated in response to DNA DSBs and promotes classic NHEJ by facilitating enrichment of NHEJ factors at DNA lesions through PAR-interaction domains located in these proteins (Boehler et al., 2011; Couto et al., 2011; Loseva et al., 2010; Rulten et al., 2011).

Taken together these studies indicate that different ARTs respond to different varieties of DNA damage. However, given that ARTs have been implicated in regulating multiple DNA repair pathways, it is possible redundancy exists between these enzymes to allow cells to tolerate genotoxic stress. For example, reduced ADP-ribosylation is apparent in parp1^−/− cells (Shieh et al., 1998) and resolution of DNA strand breaks is only delayed in these cells (Le Page et al., 2003; Trucco et al., 1998). Further, disruption of murine PARP1 and PARP2 in combination results in embryonic lethality indicating redundancy between different ARTs during the development process and/or in maintaining genome stability (Ménissier de Murcia et al., 2003). Recently, ARTi have received much attention as therapeutic agents that specifically kill tumours defective in HR due to the inability of these cells to resolve unrepaired SSBs that are converted to DSBs during DNA replication (Bryant et al., 2005; Farmer et al., 2005; Mendes-Pereira et al., 2009; Sourisseau et al., 2010; Weston et al., 2010; Williamson et al., 2010). However, these agents inhibit multiple ARTs, including PARP1–3 (Loseva et al., 2010). Therefore, assessing how different ARTs regulate compensatory repair mechanisms following genotoxic stress will further rationalise the use of ARTi in the clinic.

Recently, we and others initiated a study of DNA repair in Dictyostelium and found it contains certain DNA repair proteins absent in other invertebrates (Block and Lees-Miller, 2005; Hsu et al., 2006; Hudson et al., 2005; Zhang et al., 2009). For example, while poly-ADP-ribosylation is absent in yeast, ART activity is evident in Dictyostelium and ADP-ribosyltransferases (Adprts) have been isolated from this organism (Kofler et al., 1993; Rickwood and Osman, 1979). Similar to vertebrates, two Dictyostelium ARTs (Adprt1b and Adprt2) promote resistance of cells to agents that induce SSBs (Couto et al., 2011; Pears et al., 2012). Further, we identified an additional ART (Adprt1a) that although dispensable for tolerance of cells to SSBs, promotes NHEJ through enhancing accumulation of Ku at DNA lesions (Couto et al., 2011; Pears et al., 2012). Taken together, these observations suggest Dictyostelium will prove an important model to assess certain DNA repair pathways, including those that employ ARTs. Here we exploit the genetic tractability of Dictyostelium to uncover novel cross-talk between ARTs following genotoxic stress. We find that disruption of the SSB-responsive ART Adprt2 results in accumulation of DNA damage and that this is signalled by the DSB-responsive ART Adprt1a. Further, we identify that Adprt1a-dependent signalling of MMS-induced DNA lesions channels repair through NHEJ and that this is required for cells to tolerate base damage or SSBs in the absence of Adprt2.

Previously, we identified that two ARTs (Adprt2 and Adprt1b) are required for cells to tolerate base damage induced by methylmethanesulfonate (MMS) (Couto et al., 2011). Therefore, we wished to investigate the impact of disrupting adprt2 on chromosome stability either in the absence or presence of this genotoxic agent. Accordingly, adprt2^– cells were left untreated or exposed to MMS and accumulation of DNA damage established by isolating chromatin fractions from cells and assessing the levels of phosphorylated H2AX (γH2AX) by western blotting. While little H2AX phosphorylation is observed in the absence of genotoxic stress in chromatin fractions prepared from parental Ax2 cells, moderate induction of γH2AX is observed following administration of MMS (Fig. 1A). In contrast to its parent strain, slightly elevated levels of γH2AX are observed in adprt2^– chromatin fractions in the absence of genotoxic agents, with robust induction of γH2AX apparent following exposure of these cells to MMS (Fig. 1A). Given that formation of DNA-damage-induced nuclear foci is a commonly used marker for post-translational modifications at or adjacent to sites of DNA damage, we additionally assessed accumulation of γH2AX nuclear foci following exposure of cells to MMS. While low levels of γH2AX-positive nuclei are apparent in Ax2 cultures in untreated cells, this is increased slightly upon exposure to MMS (Fig. 1B). In contrast, adprt2⁻ cells exhibit elevated numbers of γH2AX-positive nuclei in the absence of genotoxic stress compared to Ax2 and this is further increased following exposure to MMS. Taken together, these data illustrate that loss of Adprt2 results in accumulation of DNA damage in the absence of genotoxic stress that is further exacerbated upon exposure to agents that induce base damage and DNA SSBs.

Defective SSBR results in unrepaired SSBs that can be converted to DSBs as replication forks encounter DNA lesions during S-phase. Therefore, a possible explanation for the increased γH2AX staining in adprt2⁻ cells is that additional DNA damage is sustained during DNA replication. To test this possibility, Ax2 and adprt2⁻ cells were exposed to MMS and subjected to staining with antibodies that recognise γH2AX. In parallel, cells were left untreated and incubated with bromodeoxyuridine (BrdU) to mark cells undergoing S-phase. Cells positive for γH2AX or BrdU were scored to allow comparison of the number of cells that sustain DNA damage with those undergoing S-phase at the time of MMS administration. The numbers of BrdU- and γH2AX-positive nuclei are similar in Ax2 cultures (Fig. 2A). In contrast, while 19.8±3% of adprt2⁻ cells are undergoing DNA replication at the time of MMS treatment, 80.5±4.6% are positive for γH2AX following administration of DNA damage (Fig. 2A). The larger proportion of cells that exhibit γH2AX staining above those that are replicating at the time of MMS treatment indicates that while disruption of adprt2 results in accumulation of DNA damage, this is not restricted to S-phase, but can accumulate at other stages of the cell cycle.

The number of γH2AX-positive Ax2 cells is similar to those undergoing DNA replication at the time of exposure to MMS (10.32±1.5 and 7.5±2% respectively; Fig. 2A). An interpretation of these data are that in contrast to adprt2⁻ cells, Ax2 cells accumulate damage during DNA replication following exposure to MMS. To assess this possibility, we induced Ax2 and adprt2⁻ cells to synchronously pass through the cell cycle following release from growth arrest induced by cold shock (Araki et al., 1994), and administered DNA damage to cell populations enriched in S-phase. Release of Ax2 and adprt2⁻ cells from cold shock induces synchronous cell division between 2 and 4 hours following temperature shift (Fig. 2B). Dictyostelium have no discernible G1, a short S-phase and protracted G2 (Muramoto and Chubb, 2008; Weijer et al., 1984). Accordingly, we predict that immediately following cell division cells will enter S-phase. Consistent with this and previous reports (Araki et al., 1994; MacWilliams et al., 2006; Strasser et al., 2012), we observe increased numbers of BrdU-stained cells immediately following cell division between 3 and 4 hours after release from cell cycle arrest (data not shown and Fig. 2C). The number of BrdU-positive cells is subsequently reduced to levels comparable with asynchronous cell cultures 6–7 hours following release from cold shock and prior to the second synchronous cell division. The percentage of adprt2⁻ cells that exhibit nuclear γH2AX staining following exposure to MMS is not significantly different in asynchronous, S-phase, or pre-mitotic cultures. The absence of a further increase in the number of nuclei that exhibit MMS-induced H2AX phosphorylation in S-phase-enriched cultures further supports the hypothesis that DNA damage is not restricted to S-phase in adprt2⁻ cells (Fig. 2D). In contrast, while asynchronous Ax2 cells exhibit relatively low levels of nuclear γH2AX staining following exposure to MMS, this is increased in S-phase-enriched cultures, with a subsequent reduction in pre-mitotic cells. In all instances, the percentage of Ax2 nuclei that exhibit γH2AX staining does not differ significantly from those undergoing S-phase (P>0.1). The similar profiles of cells that accumulate γH2AX with those undergoing DNA replication suggests that DNA damage occurs predominantly during S-phase in these cells. Taken together, we conclude that while MMS induces DNA damage during S-phase in Ax2 cells, disruption of adprt2 renders cells more susceptible to accumulating DNA damage throughout the cell cycle.

Although adprt2⁻ cells are sensitive to agents that induce SSBs, ADP-ribosylation is evident in these cells (Couto et al., 2011). These data indicate that DNA damage observed following administration of MMS to adprt2⁻ cells is signalled by ARTs other than Adprt2. In addition to Adprt2, Adprt1b and Adprt1a also function in signalling DNA damage in Dictyostelium cells (Couto et al., 2011). To test whether either of these proteins is responsible for the observed ADP-ribosylation in adprt2⁻ cells, we attempted to disrupt adprt1b or adprt1a in combination with adprt2. Despite repeated attempts, we have been unable to disrupt adprt2 and adprt1b in combination. In contrast, we were successful in generating an adprt1a⁻adprt2⁻ strain. Ax2, adprt1a⁻adprt2⁻ or the relevant single mutants were exposed to MMS and the ability of cells to ADP-ribosylate proteins assessed by subjecting cells to immunofluorescence using PAR-specific antibodies. Robust nuclear ADP-ribosylation is observed in Ax2, adprt1a⁻ and adprt2⁻ cells in response to MMS (Fig. 3A). Strikingly, however, ADP-ribosylation following exposure to MMS is abrogated in the adprt1a⁻adprt2⁻ strain. A similar abrogation of ADP-ribosylation is observed in adprt1a⁻adprt2⁻ cells exposed to H₂O₂ (data not shown). Therefore, in the absence of Adprt2, Adprt1a undertakes a critical role in signalling base damage and DNA SSBs.

Although disruption of the adprt2 gene induces sensitivity to agents that induce SSBs, our observations that Adprt1a is required for MMS-induced ADP-ribosylation in these cells raises the possibility that alternate pathways exist that allow cells to tolerate genotoxic stress in the absence of Adprt2. To test this possibility, we assessed the impact of disrupting adprt1a and adprt2 in combination on the ability of cells to tolerate exposure to increasing concentrations of MMS. As reported previously (Couto et al., 2011), adprt1a⁻ cells tolerate exposure to MMS equivalent to the parental Ax2 strain, while adprt2⁻ cells are sensitive to this agent. Strikingly, however, the adprt1a⁻adprt2⁻ strain exhibits increased sensitivity to MMS above that of the single mutants (Fig. 3B). A similar increased sensitivity of the adprt1a⁻adprt2⁻ strain was observed in response to H₂O₂ (data not shown). Taken together, these data illustrate that in the absence of Adprt2, Adpt1a is required to ADP-ribosylate substrates and promote resistance of cells to agents that induce base damage and/or DNA SSBs.

Next, we considered the type of DNA lesions induced by MMS in adprt2⁻ cells. One possibility is that in the absence of Adprt2, Adprt1a undertakes a hitherto unrecognised role in signalling SSBs. However, given that Adprt1a is required to ADP-ribosylate proteins following DNA DSBs (Couto et al., 2011), we also considered the possibility that deregulated SSBR results in formation of DNA DSBs that are subsequently signalled by Adprt1a. To test this hypothesis we compared enrichment of HR and NHEJ factors at DNA lesions in adprt2⁻ and parental Ax2 cells following exposure to MMS. Initially, we assessed the ability of an HR factor to recognise DNA lesions by expressing FLAG-tagged Rad51 in Ax2 and adprt2⁻ cells. Compared to untreated cultures, increased levels of FLAG–Rad51 are apparent in chromatin fractions prepared from Ax2 cells treated with MMS (Fig. 4A; supplementary material Fig. S2A). In addition to inducing DNA damage during DNA replication, high doses of base damage can produce a DNA DSB when two lesions in close proximity are processed. This, taken together with our observations that low levels of H2AX phosphorylation are observed in Ax2 cells following exposure to MMS (Fig. 1A,B), suggest that enrichment of FLAG–Rad51 in chromatin prepared from these cells is a result of DNA DSBs induced by prolonged exposure of cells to this agent. In contrast to Ax2 cells, no significant enrichment of FLAG–Rad51 is observed in chromatin fractions following exposure of adprt2⁻ cells to MMS, indicating that HR is not engaged in these cells in response to this agent (Fig. 4A; supplementary material Fig. S2A). In contrast, while no gross enrichment of Ku80 is observed in chromatin fractions prepared from Ax2 cells following treatment with MMS, enrichment of Ku80 in chromatin is observed in adprt2⁻ cells (Fig. 4B; supplementary material Fig. S3A). Taken together, these data indicate that in the absence of Adprt2, a proportion of MMS-induced DNA lesions are converted to DNA DSBs that are subsequently recognised by the NHEJ machinery.

The lack of FLAG–Rad51 recruitment to MMS-induced DNA damage in adprt2⁻ cells is surprising given that this factor is recruited to DNA lesions in Ax2 cells (Fig. 4A; supplementary material Fig. S2A). One interpretation of these data are that in the absence of Adprt2, MMS-induced DNA damage is specifically recognised by components of the NHEJ machinery. However, given that ARTs have been implicated promoting HR at stalled and/or collapsed replication forks (Bryant et al., 2009; Sugimura et al., 2008; Yang et al., 2004), an alternative possibility is that this observation reflects a defect in the HR pathway in adprt2⁻ cells, as opposed to an inherent bias towards NHEJ. However, compared to untreated control cultures, FLAG–Rad51 is enriched in chromatin fractions prepared from Ax2 and adprt2⁻ cells exposed to the DSB-inducing agent phleomycin (Fig. 4C; supplementary material Fig. S2B). Further, integration of a hygromycin resistance cassette using targeted HR is apparent at the cdk8 locus in 9.4±2.9% of adprt2⁻ cells that have integrated the knockout construct into their genome. Taken together, these data indicate that adprt2⁻ cells are able to perform HR and gene conversion events at canonical DNA DSBs. Therefore, in the absence of Adprt2, MMS-induced DNA lesions are converted to DNA DSBs with increased frequency and are specifically channelled through the NHEJ pathway as opposed to HR.

Although disruption of the adprt2 gene induces sensitivity to DNA SSBs, our observations that the adprt1a⁻adprt2⁻ strain exhibits abrogated ADP-ribosylation and increased sensitivity to MMS indicates that Adprt1a regulates alternate repair pathways to confer resistance of adprt2⁻ cells to this agent. Adprt1a promotes NHEJ by facilitating enrichment of Ku at DNA DSBs (Couto et al., 2011; Pears et al., 2012). Given that Ku is enriched at MMS-induced damage in adprt2⁻ cells (Fig. 4B; supplementary material Fig. S3A), we hypothesised that NHEJ may contribute to resistance of these cells to MMS. A prediction of this model would be that Adprt1a is similarly required to enrich Ku at MMS-induced DNA damage in adprt2⁻ cells. To test this possibility, we assessed the impact of disrupting adprt1a on the enrichment of Ku at MMS-induced DNA lesions in adprt2⁻ cells. As expected, Ku80 is not enriched in chromatin fractions prepared from Ax2 and adprt1a⁻ cells following exposure to MMS, while elevated levels of Ku80 are apparent in chromatin prepared from adprt2⁻ cells following DNA damage. Interestingly, however, disruption of adprt1a in the adprt2⁻ strain results in a decrease in Ku80 enrichment in chromatin following exposure to MMS to levels comparable with those observed in Ax2 cells (Fig. 5A; supplementary material Fig. S3B). Therefore, in the absence of Adprt2, Adprt1a is required to facilitate assembly of NHEJ factors at MMS-induced DNA lesions.

Given that Adprt1a is required to promote resistance of adprt2⁻ cells to MMS, our observations that Adprt1a promotes enrichment of Ku at DNA lesions suggests that this resistance is achieved by engaging the NHEJ pathway. To test this possibility, we generated a strain in which the adprt2 gene is disrupted in combination with DNA-PKcs, a component of the core NHEJ pathway conserved in Dictyostelium (Hudson et al., 2005). Importantly, dnapkcs⁻adprt2⁻ cells are not sensitive to DNA DSBs above parental Ax2 cells or single mutants (Fig. 5B). Given that disruption of HR is either lethal to cells, or induces sensitivity to DSBs in vegetative Dictyostelium (Hsu et al., 2011), these data indicate that similar to the adprt2⁻ strain, dnapkcs⁻adprt2⁻ cells are competent for HR. Strikingly, however, while disruption of dnapkcs alone does not result in sensitivity of cells to MMS, loss of this gene in combination with adprt2 further sensitises these cells to MMS (Fig. 5C). Taken together, these data indicate that in the absence of an effective SSBR pathway, DNA lesions are converted to DSBs that are subsequently detected and resolved by the NHEJ pathway.

Previously, we illustrated that Adprt1b and Adprt2 are required for tolerance of Dictyostelium cells to agents that induce base damage and DNA SSBs (Couto et al., 2011; Pears et al., 2012). Here we extend these studies and report that loss of Adprt2 results in genome instability leading to conversion of DNA SSBs to DSBs that are subsequently repaired by NHEJ. Compared to parental Ax2 cells, adprt2⁻ cells exhibit elevated levels of H2AX phosphorylation in the absence of genotoxic stress that is further increased when cells are exposed to agents that induce base damage and/or DNA SSBs. Mechanistically, this increased DNA damage may occur through several distinct but non-mutually exclusive mechanisms. For example, DNA damage may occur as a consequence of unrepaired SSBs being converted to DSBs as replication forks encounter DNA lesions during S-phase. Alternatively, the inability of cells to repair SSBs may result in formation of DSBs either as a consequence of two unrepaired DNA lesions occurring in close proximity, or through deregulated repair leading to illegitimate processing of DNA damage. Interestingly, we observe low levels of H2AX phosphorylation in parental Ax2 cells following exposure to MMS. The number of cells that exhibit this damage is equivalent to those undergoing DNA replication at the time of exposure to DNA damage. Further, we observe a proportional increase in the number of nuclei that exhibit γH2AX staining in response to MMS with those undergoing S-phase in synchronous cell cultures. This correlation is consistent with Ax2 cells accumulating additional DNA damage during DNA replication, as opposed to illegitimate processing of DNA damage during other phases of the cell cycle. In contrast, adptrt2⁻ cells exhibit a greater number of cells that exhibit H2AX phosphorylation than those undergoing DNA replication in the absence or presence of MMS (Fig. 2A–C). In addition, the number of γH2AX-positive cells does not further increase when cell populations are enriched in S-phase prior to exposure to MMS (Fig. 2C,D). This lack of an obvious correlation between the increased H2AX phosphorylation and DNA replication in adprt2⁻ cells illustrates that in contrast to Ax2 cells DNA damage is not restricted to S-phase but can occur at other stages of the cell cycle. Given that vegetative Dictyostelium cells have no discernible G1 (Muramoto and Chubb, 2008; Weijer et al., 1984), we believe this occurs in G2-phase of the cells cycle, or as cells undergo mitosis.

While adprt2⁻ cells are sensitive to agents that induce base damage and SSBs, ADP-ribosylation remains intact in these cells (Couto et al., 2011). These data illustrate that in the absence of Adprt2, other ARTs signal DNA damage induced by MMS and H₂O₂. This is reminiscent of the situation in vertebrates. For example, depletion of PARP1 similarly results in increased levels of H2AX phosphorylation following exposure to H₂O₂ (Woodhouse et al., 2008) and PARP2 has been implicated in catalysing the reduced levels of ADP-ribosylation following strand breakage in parp1^−/− cells (Amé et al., 1999). Unfortunately, the inability to disrupt murine parp1 and parp2 in combination prohibits testing this model in defined genetic backgrounds (Ménissier de Murcia et al., 2003). However, depletion of PARP2 alone or in combination with PARP1 does not dramatically impact on the ability of human cells to repair DNA strand breaks (Fisher et al., 2007), suggesting ARTs other than PARP2 are responsible for signalling DNA damage in the absence of PARP1. Unfortunately, we have similarly been unable to disrupt the two Dictyostelium SSB responsive ARTs (Adprt1b and Adprt2) in combination. Therefore, we have been unable to test whether these ARTs act together to signal DNA SSBs. Recently, however, we identified an additional ART required for resolution of DNA DSBs (Adprt1a), raising the possibility of further cross-talk between ARTs in the DDR (Couto et al., 2011). In this regard, we find that disrupting adprt1a and adprt2 in combination abrogates ADP-ribosylation in response to MMS and H₂O₂ and further sensitises cells to these agents. These data illustrate that in the absence of Adprt2, Adprt1a undertakes a hitherto unrecognised role of signalling DNA damage to maintain cell viability following base damage and/or SSBs. It should be noted that similar to Dictyostelium Adprt1a, PARP3 was recently identified as an additional ART that ADP-ribosylate proteins at DNA lesions to facilitate NHEJ (Rulten et al., 2011). It is interesting to speculate, therefore, that similar cross-talk between PARP1 and PARP3 is possible when signalling DNA damage and that this ART is responsible for the residual ADP-ribosylation observed in parp1^−/− cells. In this regard, PARP1 and PARP3 act synergistically in response to IR in mouse and human cells (Boehler et al., 2011). In light of our data it will be of interest to more formally test the relationship between PARP1 and PARP3 in signalling DNA damage to facilitate repair following DNA SSBs and other varieties of DNA damage.

Several possibilities could explain how Adprt1a-mediated signalling of DNA damage promotes resistance of adprt2⁻ cells to agents that induce base damage and DNA SSBs. For example, Adprt1a may act analogous to Adprt2 and undertake a hitherto unrecognised role in signalling DNA damage to promote canonical SSBR pathways. However, given that H2AX phosphorylation is a commonly used marker for DSB induction, we reasoned that the robust induction of H2AX phosphorylation in adprt2⁻ cells may indicate formation of DNA DSBs in these cells. Our observations that Adprt1a is required to signal DNA DSBs (Couto et al., 2011), in addition to our findings that Adprt1a signals MMS-induced DNA damage in the absence of Adprt2, prompted us to assess whether SSBs are converted to DSBs that are subsequently recognised by components of the DSB repair machinery. We find that while adprt2⁻ cells exhibit slightly elevated levels of Rad51 in chromatin in the absence of genotoxic stress, this is not further increased when robust induction of H2AX phosphorylation is observed in these cells following exposure to MMS. Instead, we observe enrichment of Ku at DNA lesions. Taken together these data indicate that deregulated signalling of SSBs in the absence of Adprt2 results in DSB formation that is subsequently recognised by components of the NHEJ pathway.

It is important to note that vertebrate ARTs have been implicated in promoting HR, particularly at DSBs generated at stalled replication forks (Bryant et al., 2009; Sugimura et al., 2008; Yang et al., 2004). Given that Adprt2 is acting in a manner analogous to PARP1, at least as regards signalling base damage and DNA SSBs, an important consideration is that the absence of FLAG–Rad51 recruitment to MMS-induced DNA lesions in adprt2⁻ cells reflects an inherent defect in HR, as opposed to a preferential engagement of the NHEJ machinery. However, we are able to disrupt genes by targeted HR in the adprt2⁻ strain, and Rad51 is recruited to canonical DNA DSBs induced in these cells using phleomycin. Further, while disruption of HR genes in Dictyostelium results in either cell death (Hudson et al., 2005; Zhang et al., 2009) or sensitivity to DNA DSBs (Hsu et al., 2011), no such phenotype is observed in adprt2⁻ cells (Couto et al., 2011). Therefore, we believe adprt2⁻ cells are competent for HR and favour the possibility that the increased formation of DSBs observed in these cells in response to MMS are specifically channelled through the NHEJ pathway. In this regard, it is interesting to note that our work in Dictyostelium (Hsu et al., 2011) and that of others in vertebrates (Shibata et al., 2011) suggests that NHEJ is attempted first at DSBs and if ineffective HR is engaged. An interesting possibility, therefore, is that DSBs generated as a consequence of ineffective SSBR in adprt2⁻ cells are first detected and repaired efficiently by the NHEJ machinery, rendering the HR pathway redundant or of limited importance in this context.

Taken together, our data point to alternative ARTs signalling DNA damage in the absence of canonical pathways used to repair a specific variety of DNA lesion. Specifically, we propose that in the absence of Adprt2, MMS-induced DNA lesions are converted to DSBs that are subsequently channelled through the NHEJ pathway. Adprt1a is required to enrich and/or retain Ku at DNA lesions to facilitate NHEJ (Couto et al., 2011). We similarly find that in the absence of Adprt2, Adprt1a is required for optimal enrichment of Ku at MMS-induced DNA lesions. These data, taken together with our observations that Adprt1a is required for adprt2⁻ cells to tolerate SSBs, suggest that in the absence of effective SSBR, DNA lesions are converted into DNA DSBs that are subsequently repaired by NHEJ to promote cell survival. In support of this model, we find that disruption of a canonical NHEJ factor (DNA-PKcs) in the adprt2⁻ strain further sensitises cells to MMS. Intriguingly, the dnapkcs⁻adprt2⁻ strain is more sensitive to MMS than the adprt1a⁻adprt2⁻ strain (Fig. 5C). In this regard, it is noteworthy that although the adprt1a⁻ strain is defective in NHEJ, it does not exhibit complete abrogation of Ku enrichment at DSBs (Couto et al., 2011). Given the critical role of DNA-PKcs in NHEJ, a possible explanation for the increased sensitivity of the adprt2⁻dnapkcs⁻ strain to MMS is that it has a more severe NHEJ defect than the adprt1a⁻adprt2⁻ strain.

In contrast to our observations, disruption of the NHEJ pathway can supress the sensitivity of Fanconi anaemia (FA) and HR-defective cells to genotoxic agents (Adamo et al., 2010; Bouwman et al., 2010; Bunting et al., 2010; Pace et al., 2010). These data indicate that under certain circumstances NHEJ is toxic to cells. It is interesting to note, however, that the instances when NHEJ is toxic are usually associated with deregulated repair of DNA DSBs during S-phase. Our data indicate that disruption of Adprt2 leads to increased formation of DSBs throughout the cell cycle. It is interesting to speculate, therefore, that while engaging the NHEJ pathway during DNA replication is undesirable, it can be potentially beneficial during other phases of the cell cycle such as G2 and compensate to some degree with the loss of an effective SSBR pathway.

In summary we show that the sensitivity of adprt2⁻ cells to base damage and DNA SSBs is reflected in increased levels of DNA damage either in the absence or presence of genotoxic stress. DNA damage is not restricted to S-phase, but accumulates throughout the cell cycle and manifests itself as an increased frequency of DNA DSBs, the signalling of which becomes reliant on Adprt1a. Our findings additionally illustrate that while adprt2⁻ cells are competent for HR, DSBs generated in these cells following exposure to MMS are not repaired by this pathway. Instead, they are channelled into the NHEJ pathway through Adprt1a-dependent accumulation of Ku at DNA lesions. This is reflected in an increased reliance on NHEJ to tolerate base damage in the absence of effective signalling of DNA damage by Adprt2. Taken together, these data illustrate cross-talk between different ARTs to signal DNA damage and promote alternate DNA repair pathways to maintain cell viability following genotoxic stress.

Dictyostelium strains were grown using standard procedures, either axenically or on SM agar in association with Klebsiella aerogenes. The dnapkcs⁻, adprt1a⁻, adprt2⁻ and adprt1a⁻adprt2⁻ strains have been previously described (Couto et al., 2011; Hudson et al., 2005). To generate the dnapkcs⁻adprt2⁻ strain, the adprt2 disruption construct (Couto et al., 2011) was transfected into the dnapkcs⁻ strain in which the BSR cassette had been removed using cre-recombinase (Hudson et al., 2005). Following selection, blasticidin-resistant clones were isolated using standard procedures, and gene disruption verified by PCR and Southern blot (supplementary material Fig. S1). All key observations were repeated in more than one independent clone.

The cell synchronisation protocol was adapted from previously described methods (Araki et al., 1994; Strasser et al., 2012). Briefly, exponentially growing cells were seeded at between 0.75 and 1×10⁶ cells/ml and incubated in shaking suspension (220 rpm) at 9.5°C for 16 hours to induce cell cycle arrest in G2. Subsequently, cells were released from this block by raising the temperature of cultures to 22°C within 30 seconds and incubating cells in shaking suspension at 22°C. Synchronicity of cell cycle progression following temperature shift to 22°C was assessed by two criteria. The cell density of synchronised cells, alongside exponentially growing cells diluted to 0.75×10⁶–1×10⁶ cells/ml, was measured every hour using a haemocytometer to assess when cells divided. S-phase in synchronised cultures was predicted to occur following synchronous cell doubling ∼3–4 hours post-temperature shift. The timing of S-phase in synchronised cultures was also established by scoring cells for incorporation of BrdU into DNA in each experiment (see below).

Asynchronous or synchronised cell cultures were allowed to settle on coverslips for 10 minutes before removing the media and replacing it with HL5 supplemented with 100 µM BrdU either in the presence or absence of MMS. After 30 minutes coverslips were incubated for 5 minutes in ice-cold nuclear extraction buffer (10 mM Pipes, pH 6.8, 300 mM sucrose, 3 mM MgCl₂, 20 mM NaCl and 0.5% Triton X-100) and washed in TBS before being fixed with 70% ethanol for a further 5 minutes. Coverslips were finally rinsed in 100% methanol and washed twice with TBS. Coverslips were then incubated for 1 hour at room temperature with 2N HCl to denature DNA before being washed a further three times in TBS. Following blocking in 10% donkey serum diluted in TBS (D-TBS) for 1 hour at room temperature, coverslips were incubated overnight at 4°C with the BrdU antibody diluted in D-TBS. The following day, coverslips were washed three times in TBS and incubated with a Cy3 conjugated secondary antibody (Jackson Laboratories) diluted in D-TBS. Following three washes in TBS, coverslips were mounted onto glass slides with VECTASHIELD mounting medium containing DAPI (Vector Laboratories) and visualized as described in protocols for immunofluorescence.

Whole-cell extracts were prepared as described previously (Couto et al., 2011). Antibodies were obtained from the following sources: Ku80 (Hudson et al., 2005), γH2AX (Abcam), H3 (Abcam), α-Actin (Santa Cruz Biotechnology, Inc.), FLAG (Sigma-Aldrich), Myc (Cell Signalling Technology) and BrdU (Roche).

Immunofluorescence was carried out as described previously (Couto et al., 2011). Briefly, cells were allowed to adhere to coverslips for 1 hour before incubation in media containing MMS at the doses and times stated in the figure legend. Coverslips were incubated for 5 minutes in ice-cold nuclear extraction buffer [10 mM PIPES (pH 6.8), 300 mM sucrose, 3 mM MgCl₂, 20 mM NaCl and 0.5% Triton X-100] and washed in TBS before being fixed with 70% ethanol for a further 5 minutes. Coverslips were finally rinsed in 100% methanol and washed twice with TBS.

For detecting PAR and γH2AX nuclear staining, coverslips were blocked for 1 hour at room temperature in 10% swine serum diluted in TBS (S-TBS) before incubation in the relevant antibody diluted in S-TBS for 1 hour at room temperature. Coverslips were washed three times in TBS, incubated at room temperature for 1 hour with TRITC-conjugated secondary antibody (Dako) diluted in S-TBS, and washed three times more with TBS. Coverslips were mounted onto glass slides with VECTASHIELD mounting media containing DAPI (Vector Laboratories) and visualized with a microscope (1X71; Olympus) at a magnification of 100× with immersion oil (Lenzol). All microscopy was performed at room temperature. All images were acquired on a C10600-10B-H camera (Hamamatsu Photonics) using HCImage Acquisition (Hamamatsu Photonics) image software and processed in Adobe Photoshop.

The entire coding sequence of the rad51-2 gene (DDB_G0273611) was amplified from cDNA using a primer that incorporates a KpnI site followed by a FLAG tag onto the initiating methionine codon of the gene (5′-ATGGATGGTACCGATTATAAAGATGATGATGATAAAATGGCATCAAGACAAAGACAAG-3′) in conjunction with a primer that incorporates a BamHI site directly adjacent to the stop codon (5′-GGATCCTTATTGTTCTTTATAATCGGAGATACC-3′).

The PCR product was purified and cloned into an intermediate vector (pJET; Thermo Scientific) by blunt-end ligation. The FLAG–Rad51 coding sequence was subsequently excised from pJET using KpnI and BamHI and cloned into the KpnI and BamHI sites present in the multiple cloning site of pDXA-3C.

Subcellular fraction was performed as described previously (Couto et al., 2011). Briefly, exponentially growing cells were seeded at 5×10⁶ cells/ml and mock-treated or incubated with MMS or phleomycin at the concentrations and times indicated. Following treatment, cells were washed in ice-cold KK2 buffer and resuspended in nuclear lysis buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 µM microcystin, 1 mM NaF, 2 mM sodium orthovanadate and protease inhibitor cocktail) containing 0.1% Triton X-100 to a final density of 3×10⁷ cells/ml. Cells were incubated for 15 minutes at 4°C and then centrifuged at 14,000 g for 3 minutes, giving rise to supernatant and a pellet. The pellet was resuspended in nuclear lysis buffer + 0.1% Triton X-100, and incubated for an additional 15 minutes at 4°C before centrifugation at 14,000 g for 3 minutes. The resulting pellet was resuspended in nuclear lysis buffer supplemented with 200 µg/ml RNase A (Sigma-Aldrich) and incubated with rotation at room temperature for 30 minutes. Following this step, samples were centrifuged as before and the pellet resuspended in SDS loading buffer containing 100 mM DTT. Fractions were analysed by western blotting using the antibodies indicated in the figure.

Sensitivity assays were performed as described previously (Couto et al., 2011). In brief, exponentially growing Dictyostelium were seeded at a density of 1×10⁶ cells/ml and incubated with the indicated concentrations of phleomycin (Sigma-Aldrich) or MMS (Sigma-Aldrich) at 100 rpm at 22°C. After 1 hour, cells were diluted to 1×10⁴ cells/ml in KK2 and 250 cells plated in duplicate onto two 140-mm Petri dishes containing a lawn of K. aerogenes in association with SM agar. Petri dishes were incubated in the dark at 22°C, the number of colonies visible 3, 4 and 5 days after plating was scored to measure survival.

We are grateful to the curators of dictyBase (http://dictybase.org) for assistance.