DNA double-strand break repair pathway choice in Dictyostelium

DNA double-strand breaks (DSBs) can be repaired by homologous recombination (HR) or non-homologous end joining (NHEJ). The mechanisms that govern whether a DSB is repaired by NHEJ or HR remain unclear. Here, we characterise DSB repair in the amoeba Dictyostelium. HR is the principal pathway responsible for resistance to DSBs during vegetative cell growth, a stage of the life cycle when cells are predominantly in G2. However, we illustrate that restriction-enzyme-mediated integration of DNA into the Dictyostelium genome is possible during this stage of the life cycle and that this is mediated by an active NHEJ pathway. We illustrate that Dclre1, a protein with similarity to the vertebrate NHEJ factor Artemis, is required for NHEJ independently of DNA termini complexity. Although vegetative dclre1− cells are not radiosensitive, they exhibit delayed DSB repair, further supporting a role for NHEJ during this stage of the life cycle. By contrast, cells lacking the Ku80 component of the Ku heterodimer that binds DNA ends to facilitate NHEJ exhibit no such defect and deletion of ku80 suppresses the DSB repair defect of dclre1− cells through increasing HR efficiency. These data illustrate a functional NHEJ pathway in vegetative Dictyostelium and the importance of Ku in regulating DSB repair choice during this phase of the life cycle.


Introduction
Repair of DNA damage is crucial for maintenance of genome integrity. A particularly toxic variety of DNA lesion is the DNA double-strand break (DSB), which can be repaired by two alternative pathways: homologous recombination (HR) or nonhomologous end joining (NHEJ). HR uses sequences homologous to the damaged DNA template to facilitate repair in late S and G2 phases of the cell cycle (Branzei and Foiani, 2008). As such, it is considered an accurate mechanism for re-joining DSBs. The initiating step of HR is 5Ј to 3Ј resection of DNA termini by multiple nuclease activities, including that of the Mre11-Rad50-Nbs1 (MRN) protein complex in conjunction with CtIP, exonuclease I (Exo1) and Dna2 (Raynard et al., 2008). Subsequent recruitment of Rad51 to form a nucleoprotein filament facilitates homology searching and strand invasion. Extension of the 3Ј tail by DNA polymerases then occurs using the undamaged homologous sequence as a template and resolution of the resulting structures leads to accurate repair.
In contrast to HR, NHEJ requires no homologous DNA sequences to facilitate repair, but directly processes and re-ligates DNA termini. In vertebrates, NHEJ is the principal pathway by which DSBs are repaired in G1, although it can also function outside this phase of the cell cycle (Beucher et al., 2009;Deckbar et al., 2007;Rothkamm et al., 2003). The first step in NHEJ is recognition of DNA termini by the DNA-end-binding activity of Ku, which serves as a platform to recruit a number of different proteins to DSBs, including the DNA-dependent protein kinase (DNA-PK) catalytic subunit (DNA-PKcs). DNA-PKcs kinase activity is activated upon DNA end binding and phosphorylates a number of substrates in vivo (Jackson, 2002;Meek et al., 2004), in addition to mediating DNA end synapsis to prevent illegitimate re-joining of DNA DSBs (DeFazio et al., 2002). DNA-PK also promotes the interaction of DNA DSBs with Artemis, a protein required for repair of a subset of breaks induced by IR (Riballo et al., 2004). Artemis, in addition to a number of other factors, including variant DNA polymerases and polynucleotide kinase, is thought to process DNA DSBs to make them compatible with ligation (Lieber et al., 2008). Finally, resealing of the break is performed by the DNA ligase IV-XRCC4-XLF protein complex (Mahaney et al., 2009).
Mechanistically, NHEJ is conserved throughout eukaryotes, although the components used to achieve end joining differ between species. Until recently, DNA-PKcs was thought to be conserved only in vertebrates. However, we and others identified core components of the NHEJ pathway in the amoeba Dictyostelium, including a functional orthologue of DNA-PKcs (Block and Lees-Miller, 2005;Hsu et al., 2006;Hudson et al., 2005;Muramoto and Chubb, 2008). Other DNA repair factors absent in conventional invertebrate model organisms used to study DNA repair are also conserved in Dictyostelium, including components of the Fanconi anaemia (Zhang et al., 2009) and single-strand break repair pathways (Rajawat et al., 2007). Furthermore, Dictyostelium are highly resistant to DNA damage, including ionising radiation (Deering, 1968) and UV light (Freim and Deering, 1970). Therefore, understanding how Dictyostelium tolerates high levels of genotoxic stress will provide insights into how other organisms become resistant to DNA damage. These observations, in addition to the genetic tractability of Dictyostelium, suggest it will be an important model organism to study a variety of DNA repair pathways.
Although Dictyostelium is a promising model to study DSB repair, how NHEJ and HR contribute to maintenance of genome stability and the relationship between these two pathways at different stages of the Dictyostelium life cycle remains unclear. For example, disruption of the dnapkcs and ku80 genes sensitises Dictyostelium cells to DNA DSBs and abrogates phosphorylation of the histone variant H2AX following DNA damage (Hudson et al., 2005). However, this phenotype is only evident when DNA damage is administered to germinating spores. By contrast, dnapkcs is not required for cells to phosphorylate H2AX or tolerate DSBs during vegetative cell growth, suggesting that other repair pathways predominate, at least as regards cell viability following DNA damage (Hudson et al., 2005). However, Ku and DNA-PKcs have recently been implicated in recovery from DSBs and progression of vegetative Dictyostelium through mitosis, suggesting that NHEJ does function in some respect at this stage of the life cycle (Block and Lees-Miller, 2005;Hsu et al., 2006;Hudson et al., 2005;Muramoto and Chubb, 2008). Here, we address these questions by further characterising the relative contributions of HR and NHEJ to DSB repair in vegetative Dictyostelium. We illustrate that, although HR is required for resistance of vegetative Dictyostelium cells to DSBs, NHEJ is also active during this stage of the Dictyostelium life cycle. We identify Dclre1, a protein with similarity to Artemis, as a factor required for NHEJ and show that disruption of this gene results in delayed DSB repair. By contrast, ku80cells exhibit no such defect, but instead resolve DSBs efficiently through an increased ability to perform HR. Taken together, these data demonstrate active HR and NHEJ pathways in vegetative Dictyostelium, and identify Ku as a factor that governs DSB repair pathway choice in this organism.

Homologous recombination is required to maintain cell viability following DNA DSBs in vegetative Dictyostelium
Dictyostelium ku80and dnapkcscells are not sensitive to DSBs during vegetative cell growth (Hudson et al., 2005). Vegetative Dictyostelium have no discernable G1, a short S phase and protracted G2 (Muramoto and Chubb, 2008;Weijer et al., 1984). Given that HR is active during late S and G2 phases of the cell cycle in other organisms, these data imply that HR is required for resistance to DSBs in vegetative Dictyostelium. The initiating step of HR is DNA end resection through multiple nucleases, including Mre11, Exo1 and Dna2 (Raynard et al., 2008). Therefore, to assess the contribution of HR to DSB repair in Dictyostelium, we generated a strain disrupted in the exo1 gene. Consistent with a requirement for exo1 in HR, we observe an inability of exo1cells to perform targeted HR at the cdk8 locus (Fig. 1A). Furthermore, in contrast to cells disrupted in components of the NHEJ pathway, exo1cells display increased sensitivity to DSBs administered during vegetative cell growth compared with parental Ax2 cells (Fig. 1B). These data illustrate a requirement for exo1 in HR and demonstrate that this pathway is required to maintain cell viability following DSBs administered during this stage of the Dictyostelium life cycle.

Restriction-enzyme-mediated integration of DNA into the Dictyostelium genome is dependent on NHEJ
Although ku80and dnapkcscells are not sensitive to DSBs administered during vegetative cell growth, they are sensitive to this variety of DNA damage when germinating from spores (Hudson et al., 2005). Dictyostelium spores have been reported to be in G1 phase of the cell cycle (Chen et al., 2004). It is therefore tempting to speculate that this might explain the requirement for NHEJ at this stage of the life cycle. However, accumulating evidence suggests that, similar to vegetative cell cultures, germinating Dictyostelium spores are in G2 (MacWilliams et al., 2006;Muramoto and Chubb, 2008;Weijer et al., 1984). Taken together, these data imply that, similar to humans (Beucher et al., 2009;Deckbar et al., 2007;Rothkamm et al., 2003), NHEJ is active during G2 in Dictyostelium and therefore might function during vegetative cell growth. To directly test this, we developed an assay to assess NHEJ in vegetatively growing Dictyostelium. Rejoining of linear extrachromosomal plasmids containing a selectable marker has been used to assess NHEJ efficiency in yeast and mammalian cells (see Boulton and Jackson, 1996;Fattah et al., 2010;Secretan et al., 2004;Wilson et al., 1999). However, plasmid DNA undergoes recombination events when transformed into Dictyostelium, potentially complicating this approach as regards assessing the efficiency of NHEJ (Katz and Ratner, 1988). As an alternative, therefore, we exploited the observation that cotransfection of linear plasmid DNA with the restriction enzyme used to linearise the vector results in stimulation of DNA integration into the genome by restriction-enzyme-mediated integration (REMI) (Kuspa and Loomis, 1992;Schiestl and Petes, 1991). Cotransfection of a BamHI-linearised plasmid DNA with BamHI restriction enzyme results in stimulation of DNA integration into the genome at genomic BamHI sites without loss of DNA end integrity ( Fig. 2A) (Kuspa and Loomis, 1992;Schiestl and Petes, 1991). The retention of the BamHI site upon integration, and the fact that vector DNA has limited sequence homology to the Dictyostelium genome, suggest that REMI is mediated by direct ligation of DNA termini, most probably through NHEJ. To test this hypothesis, we assessed efficiency of REMI into the genome of dnapkcsand ku80cells. Whereas transfection with BamHI alone (A)Ax2 or exo1cells were assessed for HR efficiency by measuring targeted integration of the hygromycin cassette at the cdk8 locus, as described in the Materials and Methods. Multiple transfections were performed and drugresistant clones were analysed for proper targeting by PCR. The percentage targeted HR efficiency at the cdk8 locus is calculated as the proportion of aggregate-null colonies against the total number of colonies assessed. The n number represents the total number of clones analysed. (B)Vegetative Ax2 or exo1cells were exposed to increasing concentrations of bleomycin as indicated and cell viability assessed as described previously (Hudson et al., 2005). Error bars represent the s.e.m. from three independent experiments.
does not result in reduced viability of dnapkcsor ku80strains (data not shown), REMI efficiency is compromised in dnapkcsand ku80cells compared with that of parental Ax2 cells (Fig. 2B). Dictyostelium Ku80 contains a C-terminal domain that is conserved only in organisms that contain DNA-PKcs and is required for recruitment of this kinase to DSBs (Falck et al., 2005). Consistent with conservation of this domain in Dictyostelium, its deletion results in reduced REMI efficiency similar to that observed in dnapkcscells (Fig. 2B). Taken together, these data illustrate that components of the NHEJ pathway are required for REMI of DNA into the Dictyostelium genome and provide a convenient assay to assess DNA end joining in this organism.

Identification of a -CASP-MBL family protein in Dictyostelium required for NHEJ
We wished to exploit our REMI assay to characterise other factors required for NHEJ. In particular, given the conservation of DNA-PKcs in Dictyostelium, we wanted to assess whether other factors thought to be conserved only in vertebrates function in the NHEJ pathway of this organism. Dictyostelium contains a -CASP (CPSF-Artemis-Smn1-Pso2) metallo--lactamase (MBL) (-CASP-MBL) family nuclease (dclre1; DDB_G0277755 at dictybase.org) that has been suggested to be an orthologue of the vertebrate NHEJ nuclease Artemis (also known as SNM1C) (Block and Lees-Miller, 2005;Hsu et al., 2006;Hudson et al., 2005;Muramoto and Chubb, 2008). Artemis possesses 5Ј to 3Ј exonuclease activity in vitro and, in association with DNA-PKcs, acquires endonuclease activity capable of cleaving 5Ј or 3Ј singlestranded DNA overhangs or DNA hairpins (Ma et al., 2002;Ma et al., 2005). Therefore, we wished to adapt our assay to assess the ability of cells to process incompatible DNA termini prior to ligation and determine whether Dclre1 is required for these events.
Consistent with a previous report (Kuspa and Loomis, 1992), we observed inefficient REMI when BamHI (5Ј overhang) is used in conjunction with plasmid DNA bearing an incompatible 3Јoverhang (PstI) at both ends ( Fig. 3A; data not shown). However, induction of plasmid integration was achieved when plasmids bearing one compatible (BamHI) and one incompatible (KpnI) 3Ј overhang were co-transfected with BamHI (Fig. 3B). REMI was still observed following removal of 5Ј-terminal phosphate groups by phosphatase treatment prior to introduction of plasmids into cells, illustrating that DNA integration of incompatible DNA ends is not achieved through concatemerisation of plasmids to produce two compatible DNA termini (Fig. 3B). Similar to compatible DNA termini, integration of incompatible DNA ends is dependent on DNA-PKcs and Ku80, illustrating a dependence on NHEJ (Fig.  3C). Therefore, Dictyostelium are capable of processing DNA ends to facilitate NHEJ-mediated repair of DSBs.
Next, we tested whether Dclre1 is required for REMI of DNA into the Dictyostelium genome. Given the substrate specificity of Artemis in vitro (Ma et al., 2002;Ma et al., 2005), we tested whether Dclre1 is required to process 3Јoverhang and hairpin DNA structures in vivo by engineering plasmids to contain either a 20-nucleotide 3Ј overhang or a hairpin structure at one end. These substrates were employed in REMI alongside plasmids containing two BamHI or BamHI-KpnI ends. All plasmids bearing incompatible DNA ends were capable of undergoing REMI in Ax2 cells, although less efficiently than plasmid DNA with compatible ends (Fig. 3D). REMI was compromised in dclre1cells, although this was independent of the type of DNA end employed in the assay (Fig. 3D). Therefore, although Dclre1 is required for NHEJ in Dictyostelium, its role is independent of DNA termini complexity.

Dclre1 is required for efficient repair of DNA DSBs during vegetative cell growth
The inability of dclre1cells to perform REMI suggests that NHEJ can function in vegetative Dictyostelium cells and contribute to repair of DSBs. However, when we assessed whether Dclre1 is required for resistance to DSBs administered during vegetative cell growth, similar to ku80and dnapkcsstrains (Hudson et al., 2005), we observed no gross sensitivity of vegetative dclre1cells to bleomycin (Fig. 4A). Although this might suggest an inactive NHEJ pathway in these cells, during the course of these experiments we noticed that dclre1cells consistently showed delayed resumption of growth following bleomycin treatment when compared with Ax2 cells. To assess this further, we examined the ability of parental Ax2 and dclre1cells to recover from transient exposure to DNA DSBs administered during vegetative cell growth. Exposure of Ax2 cells to bleomycin induces an arrest in cell growth, presumably through activation of a cell cycle checkpoint, with eventual expansion of cultures between 30 and 45 hours following removal of bleomycin (Fig. 4B). The dclre1strain exhibits a similar arrest of cell growth following administration of DSBs (Fig. 4B). However, although Ax2 and dclre1cells exhibit similar growth rates in the absence of bleomycin, dclre1cells exhibit an extended delay in recovery following a transient exposure to DSBs ( Fig. 4B; supplementary material Fig. S1). Importantly, dclre1cells do not exhibit differential sensitivity to bleomycin (Fig. 4A) compared with Ax2 controls, implying that the observed extension of bleomycin-induced growth arrest is not due to increased cell death within the dclre1culture, but might instead reflect a reduced ability to repair DNA damage.
The histone variant H2AX is phosphorylated in response to DSB (termed -H2AX) and decay of this signal is a commonly used marker for DSB repair (e.g. Goodarzi et al., 2008;Riballo et al., 2004). H2AX phosphorylation is conserved in Dictyostelium (Hudson et al., 2005). Therefore, to test whether the delay in resumption of cell growth in dclre1cells is due to decreased efficiency of DSB repair in these cells, we compared decay of -H2AX in dclre1and Ax2 cells. H2AX phosphorylation occurs rapidly in Ax2 and dclre1cells (Fig. 4C) and, similar to other organisms, nuclear -H2AX foci become apparent throughout the Ax2 cells were transfected with vector bearing two BamHI or two PstI termini in either the absence or presence of BamHI. Following selection, blasticidin-resistant colonies were scored and REMI efficiency calculated for each strain as fold induction in the number of blasticidin-resistant colonies achieved by addition of BamHI. Data are represented as the percentage induction of a given transfection relative to that achieved with the BamHI cut vector. Error bars represent s.e.m. from three independent experiments. (B)Efficient REMI of plasmid DNA with one non-compatible and one compatible DNA end. Ax2 cells were transfected with vector bearing one BamHI and one KpnI overhang, either with or without 5Ј phosphate groups having been removed by calf intestinal phosphatase (CIP), in the absence or presence of BamHI as indicated. DNA termini introduced into cells are illustrated in the left-hand panel. Following selection, blasticidin-resistant colonies were scored. REMI efficiency was determined as in A. Data are represented as the percentage induction of a given transfection relative to that achieved with the BamHI cut vector. Error bars represent s.e.m. from three independent experiments. (C)REMI of plasmid DNA with one non-compatible DNA end is dependent on NHEJ. Ax2, dnapkcsor ku80cells were transfected with vector bearing one BamHI and one KpnI overhang, both lacking 5Ј phosphate groups, in either the absence or presence of BamHI. Following selection, blasticidin-resistant colonies were scored. REMI efficiency was determined as in A. Data is represented as the percentage induction of a given strain relative to the parental Ax2 strain. Error bars represent s.e.m. from three independent experiments. (D)Parental Ax2 or dclre1cells were transfected with plasmid DNA containing a BamHI site at one end in combination with a 4-nucleotide 3Јoverhang (KpnI), 20-nucleotide 3Ј overhang or hairpin structure at the other end, as indicated. Following selection, blasticidin-resistant colonies were scored for each transfection. REMI efficiency was calculated for each strain as fold induction in the number of blasticidin-resistant colonies achieved on addition of BamHI to transfections. Data are represented as the percentage induction of a given strain relative to the parental Ax2 strain. Error bars represent s.e.m. from three independent experiments. nucleus following exposure to DNA DSBs ( Fig. 4D; supplementary material Fig. S2). Levels of -H2AX following damage are slightly elevated in the dclre1strain (Fig. 4C), suggesting an inability of these cells to resolve DNA damage compared with Ax2 controls. In this regard, the majority of Ax2 cells are no longer -H2AX positive 56 hours following removal of bleomycin, indicating repair of DNA damage. By contrast, -H2AX foci persist in dclre1cells, indicating a reduced ability to repair DSBs (Fig. 4E). A possible explanation for these observations is that, in addition to being required for NHEJ, Dclre1 is also required for HR and that the observed defect in recovery from DSBs in dclre1cells is a consequence of loss of this pathway. However, targeted integration efficiency at the cdk8 locus is equivalent in Ax2 and dclre1cells, illustrating that dclre1cells are proficient in HR (Fig. 5B). Therefore, the observed DSB repair defect in dclre1cells reflects a reduced ability to repair DNA damage by an HR-independent mechanism. Given the decreased REMI in dclre1cells (Fig. 4), this is most likely through a compromised NHEJ pathway.

Deletion of ku80 promotes HR and suppresses the DNA repair defect of dclre1cells
Given that disruption of dclre1 results in a delay in DSB repair, we wished to establish whether loss of other NHEJ components results in a similar phenotype. In particular, given that Ku is required for recruitment of NHEJ factors to DSBs, we assessed the impact of disrupting this protein complex on DNA repair. Surprisingly, whereas dclre1cells exhibit delayed recovery from exposure to bleomycin, ku80cells recover from DSBs with kinetics similar to that of Ax2 cells (Fig. 5A). These data suggest that, in contrast to disruption of dclre1, loss of Ku has little impact on the ability of cells to repair DSBs. Consistent with this hypothesis, decay of -H2AX foci following administration of bleomycin to ku80cells is comparable to that observed in parental Ax2 controls (Fig. 5A).
These observations suggest that, in the absence of Ku, other pathways are able to restore genome integrity following DSBs. Loss of Ku in human cells promotes an alternative subpathway of NHEJ termed microhomology-mediated end joining (MMEJ), or alternative NHEJ (A-NHEJ), which involves limited resection of DNA termini to reveal short sequences of homology that are utilised for repair (Fattah et al., 2010;McVey and Lee, 2008). However, A-NHEJ will equally be able to promote integration of plasmid DNA into the genome of cells. Therefore, the inability of ku80cells to perform REMI would argue against an alternative end-joining pathway being responsible for efficient recovery from DSBs. Alternatively, loss of Ku in yeast and vertebrates renders DNA termini more susceptible to resection and promotes HR (Lee et al., 1998;Liang and Jasin, 1996). In this regard, whereas HR remains largely unaffected by deletion of dclre1, HR targeted at the cdk8 locus is elevated in ku80cells (Fig. 5B). Taken together, these data suggest that, in the absence of Ku, DSBs are repaired more efficiently by HR. One prediction of this model is that deletion of ku80 will suppress the DNA repair defect of dclre1cells. To test this, we generated a dclre1 -ku80strain and assessed its ability to repair bleomycin-induced DNA DSBs. Similar to ku80and dclre1cells, the dclre1 -ku80strain exhibits reduced REMI efficiency (supplementary material Fig. S3). Whereas dclre1cells exhibit the characteristic delay in -H2AX foci decay and cell growth following bleomycin treatment, this is not observed in the dclre1 -ku80strain. Indeed, recovery from DSBs is slightly faster in these cells when compared with either Ax2 or ku80cells ( Fig.  (A)Exponentially growing Ax2, dclre1and ku80cells were exposed to increasing concentrations of bleomycin as indicated and cell survival assessed as described in the Materials and Methods. Error bars represent the s.e.m. from three independent experiments. (B)Cells were exposed to bleomycin at a concentration of 100 mU/ml for 1 hour. Following removal of bleomycin, cell density was determined at the indicated time points. Each point represents the s.e.m. of three replicates from a single experiment. One experiment representative of more than five is shown. (C)Ax2 or dclre1cells were exposed to 100 mU/ml bleomycin. Whole-cell extracts were prepared at 0, 20, 40 and 60 minutes following administration of DNA damage and subjected to western blotting with the indicated antibodies. (D)Ax2 cells were left untreated or exposed to 100 mU/ml bleomycin. Samples were fixed as described in the Materials and Methods, and subjected to immunofluorescence using the indicated antibody or DNA visualised using DAPI. (E)Ax2 or dclre1cells were exposed to 100 mU/ml bleomycin for 1 hour. At the indicated time points following removal of bleomycin, cells were stained with antibodies raised against -H2AX and subjected to immunofluorescence. The percentage -H2AX-positive cells were scored from a population of >200 cells. One experiment is shown, representative of more than three. 5C). Therefore, deletion of Ku channels the repair of DSBs into alternative pathways other than NHEJ in vegetative Dictyostelium.

Discussion
Previously, we and others illustrated that Dictyostelium contains functional orthologues of DNA-PKcs and Ku that are required for resolution of DNA DSBs (Hudson et al., 2005;Muramoto and Chubb, 2008). Here, we further characterise their role in DSB repair by analysing the impact of disrupting these genes on REMI of DNA into the Dictyostelium genome. Although the stimulation of plasmid integration into the genome by co-transfection with restriction enzymes has been reported in a number of organisms (Kuspa and Loomis, 1992;Manivasakam et al., 2001;Schiestl and Petes, 1991), the molecular basis of this phenomenon remains unclear. REMI occurs independently of HR (Schiestl et al., 1994) and here we report that it is dependent on components of the NHEJ machinery in Dictyostelium. Furthermore, we exploit this assay to further illustrate conservation of the vertebrate NHEJ pathway in Dictyostelium by identifying a -CASP-MBL family member with similarity to Artemis that is required for NHEJ.
To our knowledge, this is the first report of a -CASP-MBL family member functioning in NHEJ outside vertebrates. Defects in human Artemis result in severe combined immunodeficiency due to a defect in processing DNA termini during V(D)J recombination (Moshous et al., 2001;Pannicke et al., 2004;Rooney et al., 2002). As Artemis is capable of processing complex DNA termini in vitro (Ma et al., 2002;Ma et al., 2005), these studies lead to speculation that this nuclease is required for processing a subset of complex DNA termini during NHEJ. However, our data illustrate that, although Dclre1 is required for NHEJ, this occurs independently of DNA end complexity. This highlights a requirement for Dclre1 beyond simple processing of DNA termini and suggests Artemis might similarly be required for aspects of NHEJ distinct from processing DNA ends. Indeed, although Artemis is required to repair a subset of potentially complex DSBs induced by IR (Darroudi et al., 2007;Riballo et al., 2004;Wang et al., 2005), more recent findings implicate this protein in facilitating NHEJ at sites close to or within heterochromatin independently of DNA termini complexity (Goodarzi et al., 2008). The assay employed here to assess NHEJ, in contrast to the usual plasmid repair assays employed in other systems, takes place in a chromatin context on one side of the break and could function in both heterochromatic and euchromatic regions of the genome. Given the genetic tractability of Dictyostelium, in addition to the similarity of Dictyostelium chromatin-modifying enzymes to those in humans (Chubb et al., 2006), it will be interesting to exploit Dictyostelium to assess the requirements for dclre1 and other factors to facilitate integration of plasmid DNA into the genome in different chromatin contexts.
Although it is clear that Dictyostelium utilises components of the NHEJ machinery hitherto thought to be conserved only in vertebrates, to date it has not been exploited extensively to study DSB and other repair pathways. Consequently, little is know about how this organism combats DSBs at different stages of its life cycle. Previously, we demonstrated a reliance on NHEJ to combat DSBs administered as cells germinate from spores (Block and Lees-Miller, 2005;Hsu et al., 2006;Hudson et al., 2005;Muramoto and Chubb, 2008). Here, we extend these studies to assess the contribution of HR and NHEJ to DSB repair during vegetative cell growth. We find that, similar to mutation of other NHEJ components (Block and Lees-Miller, 2005;Hsu et al., 2006;Hudson et al., 2005;Muramoto and Chubb, 2008), disruption of dclre1 , ku80and dclre1cells were treated with 100mU/ml bleomycin for 1 hour. Following removal of bleomycin, cells were resuspended at a density of 1ϫ10 6 cells/ml in HL5. Cell density was determined at the indicated time points (left-hand panel). In parallel, cells were fixed, stained using antibodies raised against -H2AX and subjected to immunofluorescence. The percentage -H2AXpositive cells were scored from a population of >200 cells (right-hand panel). One experiment is shown, representative of at least three. (B)Ax2, dclre1and ku80cells were assessed for HR efficiency by measuring targeted integration of the blasticidin-resistance cassette at the cdk8 locus. Multiple transfections were performed and drugresistant clones were analysed for proper targeting by PCR. The percentage targeted HR at the cdk8 locus is calculated as the proportion of aggregate-null colonies against the total number of colonies assessed. The n number represents the total number of clones analysed. (C)AX2, dclre1 -, ku80and ku80 -dclre1cells were treated with 100mU/ml bleomycin for 1 hour. Following removal of bleomycin, cells were resuspended at a density of 1ϫ10 6 cells/ml in HL5. Cell density was determined at the indicated time points (upper panel). In parallel, cells were fixed and stained using antibodies raised against -H2AX at the time points indicated. Cells that display -H2AX staining were identified by immunofluorescence. The percentage -H2AX-positive cells were scored from a population of >200 cells. One experiment is shown, representative of at least three.
does not sensitise vegetative cells to bleomycin, indicating that the NHEJ pathway is not required to maintain cell viability following DSBs in Dictyostelium. Instead, the radiosensitivity of exo1cells illustrates that HR is required for this process. This is consistent with the cell cycle distribution of vegetative Dictyostelium, with the majority of cells being in G2 (Muramoto and Chubb, 2008;Weijer et al., 1984), a stage of the cell cycle that would support HR (Branzei and Foiani, 2008). Thus, it would appear that, similar to yeast, HR is the predominant pathway utilised to combat DSBs in vegetative Dictyostelium, at least as regards maintaining cell viability following DNA damage. These data also imply that HR is the predominant pathway by which DSBs are repaired during G2 in Dictyostelium, in marked contrast to vertebrate cells, which rely principally on NHEJ in G1 and G2 (Beucher et al., 2009;Deckbar et al., 2007;Rothkamm et al., 2003). The reliance of Dictyostelium on HR might go some way to explaining the relatively high efficiency of gene disruption by targeted HR in this organism compared with vertebrates and its high resistance to ionising radiation (Deering, 1968).
What is striking about our data, however, is that, although NHEJ is not required for resistance of vegetative Dictyostelium to DSBs, NHEJ is functioning in some respect to combat this variety of DNA damage during this phase of the life cycle. This conclusion is based on our observations that: REMI of plasmid DNA into the genome is dependent on components of the NHEJ pathway; and disruption of a factor required for NHEJ-mediated REMI (dclre1) results in a delay to the repair of DSBs administered to vegetative cells. Therefore, both HR and NHEJ are capable of repairing DSBs in vegetative Dictyostelium. Interestingly, in contrast to dclre1cells, we observe that, although Ku80 is required for efficient NHEJ in vegetative Dictyostelium, recovery from DSBs is largely unaffected in ku80cells. Ku80 has recently been implicated in recovery from DSBs and progression of vegetative Dictyostelium through mitosis (Muramoto and Chubb, 2008). Although this study seems to conflict with our observations, it should be noted that our studies assess recovery from DSBs at later time points than those analysed by Muramoto and Chubb. Studies in human cells reveal that the G2-M checkpoint is released prior to completion of DSB repair . One potential explanation for these discrepancies is that, although Ku is required for maintenance of G2-M arrest, ku80cells do eventually pass through this checkpoint with persistent DSBs that are resolved in the subsequent cell cycle. This is a particularly attractive hypothesis given that DSBs will probably be repaired by HR in the subsequent S-G2 and that ku80cells exhibit elevated levels of recombination (Fig. 5B). In both studies, however, a crucial role for Ku in facilitating DSB repair during G2 phase of the cell cycle is apparent.
The experiments described here do not directly assess whether NHEJ and HR are attempted at the same DSB in G2, or whether they function independently of each other at breaks with different end complexities or in different chromatin contexts. However, it should be noted that, although dclre1cells show delayed recovery from DSBs, they do eventually repair damage and resume cell growth, an observation that is reflected in the lack of sensitivity of these cells to DSBs. A possible interpretation of these data is that NHEJ is attempted at a DSB first and, if unsuccessful, another repair pathway such as HR subsequently repairs DNA damage. For example, the delay in recovery from DSBs in dclre1cells might reflect recognition of DNA ends by Ku and an unsuccessful attempt to perform NHEJ. In this regard, Ku rapidly binds DSBs before components of the HR machinery in vertebrate cells (Kim et al., 2005) and has been proposed to protect DNA termini from resection and inhibit HR in a number of different organisms (Barlow et al., 2008;Clerici et al., 2008;Lee et al., 1998;Liang and Jasin, 1996;Pierce et al., 2001;Wu et al., 2008;Zhang et al., 2007). A prediction of this model would be that a failure to initiate NHEJ would suppress defects in 'downstream' components of this pathway. Consistent with this, we find that deletion of Ku promotes HR and rescues the delay in DNA repair seen in dclre1cells. Taken together, these data illustrate that both NHEJ and HR are active in vegetative Dictyostelium and highlight a role for Ku in determining DSB repair pathway choice in this organism.

Cell culture and strain generation
All strains were grown in association with Klebsiella aerogenes on SM agar or axenically using standard procedures.
To generate the dclre1cdisruption strain, fragments either upstream (nucleotides -865 to -155) or downstream (nucleotides +3078 to +3852) of the dclre1gene (relative to A of the initiating ATG codon as +1 and T of the TGA stop codon as +3019) were amplified from Ax2 genomic DNA by PCR. These fragments were cloned either side of a blasticidin-resistance cassette contained in the pLPBLP plasmid (Faix et al., 2004;Sutoh, 1993) using KpnI-HindIII for the upstream and PstI-BamHI for the downstream fragment. The resulting disruption construct was designed to delete the entire dclre1gene. Transfection was carried out and cells subjected to selection with 10 g/ml blasticidin the following day. Disrupted strains were isolated using standard procedures and confirmed by PCR.
To generate the exo1disruption strain, fragments upstream (nucleotides -921 to -48) or downstream (nucleotides +1990 to +2474) of the exo1gene (relative to A of the initiating ATG codon as +1) were amplified from Ax2 genomic DNA by PCR. These fragments were cloned either side of a blasticidin-resistance cassette contained within pLPBLP using KpnI-HindIII for the upstream or EcoRV-PstI for the downstream fragment. Transfection, selection and screening of strains disrupted in exo1 were performed as described above.
To remove the blasticidin-resistance cassette from ku80 -, dnapkcsand dclre1cells, the relevant strains were transformed with plasmid pDEX-NLS-cre to express Cre recombinase. Blasticidin-sensitive clones were identified as described by Faix et al. (Faix et al., 2004) and verified for removal of the cassette by PCR. Generation of ku80 -dclre1cells was achieved as above using the dclre1 disruption vector in ku80cells in which the BSR cassette had been removed.
Generation of expression constructs containing full-length Ku80 or Ku80 lacking the C-terminal DNA-PKcs interaction domain was achieved by amplifying the entire ku80 gene including the stop codon (nucleotides +1 to 2511 relative to the A of the initiating ATG codon) or the ku80 gene lacking the C-terminal 14 codons from Ax2 genomic DNA. A KpnI restriction site and Myc tag were incorporated into the primer complementary to the N-terminus of the open reading frame, and an XbaI site was incorporated into the primer complementary to the C-terminal primer. The Myc-Ku80 sequences were subsequently placed under control of a constitutive promoter by cloning into the pDXA vector through the KpnI and XbaI sites. Vectors were transfected into the relevant strain along with the helper plasmid pREP and transformants selected in G418.

Restriction-enzyme-mediated integration of plasmid DNA into the Dictyostelium genome
To generate vectors bearing restriction enzyme sites at their ends, a vector containing a blasticidin cassette (pLPBLP) (Faix et al., 2004) was cut with combinations of the appropriate restriction enzymes and treated with calf intestinal alkaline phosphatase where stated using standard procedures. DNA was purified by phenol extraction and ethanol precipitation prior to being used in transfections.
To generate vectors with a 20-nucleotide 3Јoverhang or hairpin ends, adaptor oligonucleotides (see below) were ligated to one end of pLPBLP. The pLPBLP vector was cut with SalI and the 5Ј phosphate groups removed using CIP to prevent re-circularisation of the vector. Linearised pLPBLP was ligated to the appropriate adaptor oligonucleotide bearing a compatible XhoI end. The products were then digested with BamHI to provide a compatible terminus at the opposite end of the vector. The final products were resolved by agarose gel electrophoresis and the fragments containing the ligated oligonucleotides identified by a shift in their mobility upon electrophoresis. The relevant fragments were isolated from gels and recovered using a QIAGEN gel extraction kit.
Equal molar ratios of the oligonucleotides were mixed and heated at 95°C for 20 minutes and then cooled to room temperature to allow annealing.

Sensitivity of strains to DNA damage
Exponentially growing Dictyostelium cells were harvested and resuspended at a density of 5ϫ10 6 cells/ml in 1 part HL5 to 5 parts KK2. Bleomycin sulfate (Sigma) was added to cells at the indicated concentrations in shaking suspension (160 rpm at 22°C) for one hour before dilution in KK2 to a density of 1ϫ10 4 cells/ml. Replicates of 200 cells were plated on 15 cm SM agar plates in association with K. aerogenes. Survival was judged by plaque formation 3 and 4 days after plating.

Immunofluorescence
Exponentially growing Dictyostelium were harvested and resuspended at a density of 5ϫ10 6 cells/ml in 1 part HL5 to 5 parts KK2. Bleomycin sulfate (Sigma) was added to a final concentration of 100 mU/ml in shaking suspension (100 rpm at 22°C) for one hour. Cells were washed three times with HL5 before resuspension at a density of 1ϫ10 6 cells/ml in HL5 with shaking (100 rpm at 22°C). 1ϫ10 6 cells were removed at the time points indicated and left to adhere to polylysine-coated cover slips for 5-10 minutes prior to incubation in pre-extraction buffer (10 mM PIPES pH 6.8, 300 mM sucrose, 3 mM MgCl 2 , 20 mM NaCl, 0.5% Triton X-100) for 5 minutes. Cells were washed in Tris-buffered saline (TBS) and fixed for 5 minutes in 70% ethanol. After removal of ethanol, cover slips were washed once in ice-cold methanol, twice in TBS and blocked with 10% swine serum in TBS for 1 hour. Samples were subsequently incubated for 1 hour with anti--H2AX (Abcam) diluted in blocking solution and washed three times with TBS before incubation for 1 hour with anti-rabbit TRITC (DAKO) diluted in blocking solution. Cover slips were washed three times with TBS and mounted using Vectorshield containing DAPI (Vector Labs). Immunofluorescence was detected using an Axioscope 2 fluorescent microscope equipped with Axiovision imaging software (Zeiss).

Cell proliferation following treatment with bleomycin
Exponentially growing Dictyostelium cells were harvested and resuspended at a density of 5ϫ10 6 cells/ml in 1 part HL5 to 5 parts KK2. Bleomycin sulfate (Sigma) was added to a final concentration of 100 mU/ml in shaking suspension (100 rpm at 22°C) for one hour. Cells were washed three times with HL5 and resuspended at a density of 1ϫ10 6 cells/ml in HL5. Cell density was determined microscopically at the indicated time points. At least two independent clones of a given strain were analysed for a phenotype.

Targeted HR efficiency at the cdk8 locus
The cdk8 knockout plasmid (Lin et al., 2004) was digested with KpnI and NotI to liberate a fragment of DNA containing regions homologous to the D. discoideum cdk8 gene flanking either the BsR or hygromycin-resistance cassettes, and purified using standard procedures.
Cells in the exponential phase of growth were transfected with 7 g DNA using standard procedures, serial diluted in HL5 following a short recovery period and plated out in 96-well plates. After 24 hours, 10 g/ml blasticidin was added and plates incubated in the dark at 22°C for 14 days. Following selection, clonal suspensions of blasticidin-resistant transformants were spotted onto SM-agar containing a lawn of K. aerogenes. After 5-6 days, plaques were large enough for phenotypic analysis. Targeted integration at the cdk8 locus is indicated by an aggregation-defective phenotype. Targeted disruption of the cdk8 locus only in aggregation-deficient colonies was confirmed by PCR analysis of genomic DNA isolated from randomly selected colonies.
To perform the HR assay in the exo1strain, the cdk8 knockout vector was modified to confer resistance to hygromycin. This was achieved by cutting the cdk8 knockout vector with BamHI to remove the BsR cassette and replacing it with the BamHI-excised hygromycin-resistance cassette from the pHygTm(plus)/pG7 vector (a kind gift from Jeffrey Williams, University of Dundee, Scotland). Cells were transfected and subjected to selection with 40 g/ml hygromycin the next day. Phenotypic analysis of hygromycin-resistant colonies was performed after 14 days as described above.
We are grateful to the curators of dictyBase (dictybase.org) for assistance. This work was supported by Cancer Research UK (www.cancerresearch.org.uk; grant C1521/A8182), the BBSRC (www.bbsrc.ac.uk; grant BBS/B/08051) and the Oxford University Clarendon Scholarship Fund.