DNA damage threatens the integrity of the genome and has potentially lethal consequences for the organism. Plant DNA is under continuous assault from endogenous and environmental factors and effective detection and repair of DNA damage are essential to ensure the stability of the genome. One of the most cytotoxic forms of DNA damage are DNA double-strand breaks (DSBs) which fragment chromosomes. Failure to repair DSBs results in loss of large amounts of genetic information which, following cell division, severely compromises daughter cells that receive fragmented chromosomes. This review will survey recent advances in our understanding of plant responses to chromosomal breaks, including the sources of DNA damage, the detection and signalling of DSBs, mechanisms of DSB repair, the role of chromatin structure in repair, DNA damage signalling and the link between plant recombination pathways and transgene integration. These mechanisms are of critical importance for maintenance of plant genome stability and integrity under stress conditions and provide potential targets for the improvement of crop plants both for stress resistance and for increased precision in the generation of genetically improved varieties.
DNA double-strand break (DSB) repair mechanisms are important for genome stability, fertility and genetic diversity and also mediate the integration of transgenes into the plant genome. Recombination pathways are therefore of interest to biotechnologists as important targets for crop improvement. Recent advances have furthered our understanding of the sources of DSBs in plants and the responses and repair mechanisms plants use to minimize the impact of DNA damage on growth and development. In addition, manipulation of the plant genome and DSB repair pathways has led to improved precision of transgene integration in Arabidopsis and crop plants.
II. Sources of DNA damage
Plant cells are subject to high levels of DNA damage resulting from dependence on sunlight for energy and the concomitant exposure to environmental stresses including UV-B, ozone, desiccation and rehydration, and air and soil pollutants including heavy metals. These agents cause a range of DNA damage products including single-strand DNA breaks (SSBs) and DSBs. DSBs also occur spontaneously, arising from defects occurring during DNA replication, including collapsed replication forks, replication past an SSB and steric stresses as DNA is unwound (Kuzminov, 2001). Although chromosomal breaks present a serious threat to the genomic integrity of plants for most of their life cycle, they also play essential roles in meiotic recombination. In the early stages of gamete formation the programmed introduction of DSBs initiates homologous chromosome pairing and recombination. This area has seen great advances over the last decade and readers are directed to recent reviews (Edlinger & Schlogelhofer, 2011; Osman et al., 2011).
III. The toxic effects of DSBs
DSBs have extreme detrimental effects on plant growth, with particularly severe effects on actively dividing cells. DNA replication or progression through mitosis in the presence of a DSB can lead to loss of chromosome fragments. Cellular production in plants largely originates from a pool of undifferentiated dividing stem cells localized to plant meristems (Sablowski, 2011). As stem cells have the capability to progress through multiple rounds of replication, inherited genetic defects result in clonal populations of mutant cells. Mutations, in the form of deletions, insertions and chromosome fusions, have the potential to inhibit plant growth as a result of inhibition of transcription and loss of cell viability. Incorrect genome repair can lead to chromosome fusions, often producing dicentric chromosomes and anaphase bridges, as identified in early cytogenetic studies. Damage occurring in vegetative tissues can produce genetic changes in plant populations; meristem cells give rise to the germ line at a relatively late stage of plant development and mutations in stem cells can be passed on to the next generation. This is illustrated by studies in which irradiation of Arabidopsis with environmental levels of UV-B resulted in increased somatic homologous recombination in vegetative tissues, leading to recombination events that were inherited by the next generation of plants (Ries et al., 2000).
Early detection, cell cycle arrest and rapid repair of DSBs are essential to mitigate against the toxic effects of DSBs, pausing cell cycle progression to allow time for repair before DNA replication or mitosis. Arrested growth brought about by the DNA damage checkpoint is clearly exemplified by the ‘gamma-plantlet’ phenotype of the uvh1 DNA repair defective mutant line which is deficient in the XPF endonuclease. After high doses of γ-rays, which cause an array of damage products including DSBs, uvh1 seeds are able to germinate but these seedlings experience a delay in the production of true leaves as a result of transient cell cycle arrest of meristematic cells (Preuss & Britt, 2003). After higher levels of DNA damage, production of true leaves is permanently lost and meristem cells cease cell proliferation, through either differentiation and loss of meristematic identity or the terminal solution of programmed cell death (Ricaud et al., 2007; Fulcher & Sablowski, 2009). Significant progress has been made over recent years in our understanding of the mechanisms and regulation of the DNA damage response in plants. These processes are initiated by detection of the DSB, followed by signalling to the mechanisms that regulate cell cycle progression, programmed cell death and DNA repair pathways.
IV. Detection of DSBs
Recognition of DNA damage is an essential step in repair. It is likely that DSBs are detected by multiple mechanisms which precede signalling, initiation of repair pathways and activation of cell cycle checkpoints. Candidate DSB detection factors in plants include the complex of KU70 and KU80 (Lieber, 2010). The KU complex has high affinity for exposed DNA ends and is a core component of the nonhomologous end-joining (NHEJ) pathway, although dispensable for DNA damage signalling (West et al., 2004). A multifunctional complex of three proteins, MRE11, RAD50 and NBS1 (the MRN complex), also is important in DSB detection as initially demonstrated in mammals (Lee & Paull, 2004). This complex has roles in DNA damage signalling and is involved in both of the major DSB repair pathways, NHEJ and homologous recombination (HR). An important function of the MRN complex is carried out by the C-terminal domain of NBS1, which recruits the protein kinase ataxia telangiectasia mutated (ATM) to the site of DNA damage, resulting in activation of its kinase activity (Falck et al., 2005). The role of the MRN complex in DNA damage detection and activation of ATM-mediated signalling is conserved in plants (Amiard et al., 2010). The mechanisms of DSB detection influence the repair pathways used, as discussed in Section VIII.5 in the context of DSB repair pathways.
1. The roles of ATM and ataxia telangiectasia-mutated and Rad3-related (ATR)
The eukaryotic response to DNA damage is regulated by the phosphoinositide-3-kinase-related protein kinases (PIKKs) ATM and ATR (Bradbury & Jackson, 2003). ATM and ATR have conserved functions in eukaryotes as master controllers of the DNA damage response, coordinating cell cycle progression and activation of DNA repair pathways (Fig. 1). ATM is principally activated by DSBs (Lee & Paull, 2005; Culligan et al., 2006). Conversely, ATR responds more strongly to replication defects, with regions of single-stranded DNA activating the kinase activity of ATR, mediated by the ATR-interacting protein (ATRIP) in both plants and animals (Zou & Elledge, 2003; Sweeney et al., 2009). Consequently, atm mutants are highly sensitive to X-rays (DSBs), whereas atr (and atrip) mutants are sensitive to the DNA replication inhibitor hydroxyurea (Garcia et al., 2003; Culligan et al., 2004). However, there is overlap between the functions of the two kinases in plants. Either gene is sufficient for the observed down-regulation of a subset of transcripts following irradiation (Culligan et al., 2006). The partial redundancy between ATM and ATR is also apparent in the meiotic phenotypes of single and double mutants, with atr displaying full fertility, atm producing c. 10% of the seeds of wild-type plants, and the atr atm double mutant showing complete sterility (Culligan & Britt, 2008). The kinase activity of ATM and ATR is important in DNA damage signalling, directly or indirectly phosphorylating several hundred target proteins in mammals, including the histone 2A isoform H2AX, Nbs1, and the checkpoint associated protein kinases Chk1 and Chk2 (Matsuoka et al., 2007). Activation of these repair pathways leads to the rapid relocalization of DSB-repair proteins to the sites of damage, cell cycle checkpoint activation and initiation of DNA repair (Shiloh, 2006). In general, most of the signalling components downstream of the PIKKs are uncharacterized in plant species, with the notable exception of the ATM-dependent rapid phosphorylation of the histone variant H2AX, which is a highly conserved eukaryotic response to DSB induction.
2. H2AX phosphorylation
In plants, studies using mitotic dividing root cells showed that H2AX is phosphorylated on Ser139 very rapidly upon DSB induction, with foci formation detectable 2 min post-irradiation in Arabidopsis and peaking at 10 min (Friesner et al., 2005; Charbonnel et al., 2010). Phosphorylated H2AX (γH2AX) serves to signal the presence of the DSB in yeast and animals and facilitates the orderly recruitment of protein mediators of the DNA damage response to the site of repair (Lukas et al., 2004). H2AX phosphorylation can extend for up to two million bases from the DSB in mammals, amplifying the DNA damage response as part of the intracellular signalling pathway (Rogakou et al., 1998). Although H2AX phosphorylation is predominantly ATM-dependent in Arabidopsis, ATR appears to be responsible for a subset of c. 10% of foci (Friesner et al., 2005). The disappearance of most foci by 2 h suggests rapid repair of DSBs and dephosphorylation or turnover of γH2AX, although persistence of some foci after this time-point is consistent with the existence of a subset of DSB ends with slower repair kinetics. Rapid γ-ray-induced H2AX phosphorylation does not occur in rad50 or mre11 mutant lines, demonstrating that the MRN complex plays an important role in DNA damage signalling in plants leading to ATM/ATR-mediated phosphorylation of H2AX (Amiard et al., 2010). However, these mrn mutants display constitutive ATR-dependent formation of H2AX foci in the absence of genotoxin treatment, indicating that accumulated DNA damage in rad50 or mre11 mutant lines activates ATR, possibly through S-phase defects or single-stranded DNA repair intermediates (Amiard et al., 2010). In yeast an important function of H2A phosphorylation is to recruit the INO80 complex (Fig. 1) (van Attikum et al., 2004) which has chromatin-remodelling activity and displaces nucleosomes, facilitating HR (Tsukuda et al., 2005; van Attikum et al., 2007) which was initially identified in a screen for plants with decreased levels of somatic HR, with ino80 mutants displaying 15% of the wild-type levels of HR in vegetative tissues (Fritsch et al., 2004). The INO80 complex comprises 12 subunits including INO80 and the actin-related protein Arp5; arp5 mutants display hypersensitivity to DNA damage (Kandasamy et al., 2009).
3. Transcriptional changes in response to DNA damage
Transcriptional changes are integral to the DNA damage response in most organisms, although the subset of induced transcripts is poorly conserved between kingdoms (Culligan et al., 2006). In plants ATM is required for the transcriptional DNA damage response, although ATR also plays a minor role in regulating gene expression. The ATM-mediated transcriptional response is highly specific to genotoxins that cause DSBs rather than a generic response to DNA damage (Molinier et al., 2005) and is characterized by the massive up-regulation of hundreds of genes, many displaying over a hundred-fold induction after a 100-Gy γ-ray dose (Culligan et al., 2006). Many induced genes encode factors involved in DNA repair processes, in particular DNA metabolism and the cell cycle, including the HR factors RAD51 and the endonuclease COM1/CtIP, components required for DNA synthesis (a replication factor A subunit, DNA polymerases and thymidine kinase) and DNA signalling factors (RAD17, breast cancer-associated gene 1 (BRCA1) and poly ADP-ribose polymerase (PARP)), and several uncharacterized genes (Ricaud et al., 2007; Uanschou et al., 2007; Culligan & Britt, 2008). NHEJ genes display little transcriptional response to DNA damage (Culligan et al., 2006), consistent with constitutive expression of the NHEJ pathway, which plays a major role in repair of DSBs in vegetative tissues. In response to DNA damage the expression of some genes is down-regulated, including genes expressed in G2 or M phase, indicative of cell cycle arrest in response to DNA damage (see Section V.1) (Culligan et al., 2006; Ricaud et al., 2007).
4. Suppressor of gamma response 1 (SOG1)
Irradiation of uvh1 (XPF endonuclease) mutant Arabidopsis seeds with high doses of gamma radiation activates a strong G2 arrest, transiently inhibiting growth and resulting in an 8-d delay in the production of true leaves. This G2 arrest at relatively low doses of X-rays presumably results from the accumulation of replication blocking DNA damage products in these plants, and is not observed in the wild type and NHEJ mutants. The genetic basis of the checkpoint mechanisms underlying the gamma-ray-induced growth arrest was investigated using a forward genetic screen. A mutant line was identified that allowed production of true leaves following irradiation, and was termed suppressor of gamma response 1 (sog1) (Preuss & Britt, 2003). The mutation responsible for the radiation-resistant growth was mapped and the gene identified as a putative transcription factor (Yoshiyama et al., 2009). sog1 mutants resemble atr mutants: both ATR and SOG1 are required for delaying leaf production after gamma irradiation of the uvh1 mutants, while ATM is not required for growth arrest (Yoshiyama et al., 2009).
Whilst there is little evidence that ATM is required for the SOG1/ATR-mediated transient growth arrest, surprisingly, SOG1 is required for the ATM-dependent transcriptional response. This indicates key roles for SOG1 in both major DNA damage signalling pathways in plants (Fig. 1). SOG1-deficient plants retain the capability to phosphorylate H2AX, indicating that ATM is activated in sog mutants and that it is the transcriptional ATM response that is specifically abrogated (Yoshiyama et al., 2009). Therefore, SOG plays analogous roles to mammalian p53, which is important for cell cycle checkpoint activation and plays a role in the mammalian transcriptional response (Stankovic et al., 2004; Elkon et al., 2005). Additional similarities between p53 and SOG1 are the requirement of both proteins for the programmed cell death response to high levels of genotoxic stress (see Section V.2).
5. The 9-1-1 damage response
The up-regulation of the AtRAD17 transcript in irradiated seedlings suggested roles for this protein in the plant DNA damage response. AtRAD17 is the homologue of Schizosaccharomyces pombe Rad17, which is involved in DNA damage signalling. Schizosaccharomyces pombe Rad17 functions analogously to the PCNA clamp loader replication factor C (RFC), loading the proliferating cell nuclear antigen (PCNA)-like protein complex of Rad9–Hus1–Rad1 (the 9-1-1 complex). The similarities between the Rad17/9-1-1 and RPC/PCNA are great; structurally the complexes are much alike and many of the protein interactions are interchangeable. The 9-1-1 complex is loaded onto damaged DNA and helps integrate phosphoinositide 3-kinase (PI3-kinase) signalling and downstream cell cycle checkpoint targets. In Arabidopsis, both AtRAD17 and the plant homologue of RAD9 are involved in recovery from DNA damage, with mutants displaying hypersensitivity to bleomycin and mitomycin C and decreased rates of DSB repair (Heitzeberg et al., 2004).
V. Growth responses to genotoxic stress
1. Cell cycle regulation in the response of plant cells to DNA damage
Control of cell cycle progression is intimately linked to the DNA damage response through the activation and release of checkpoints, regulating both entry to and progression through DNA replication, and entry to mitosis (Fig. 2). DNA damage-induced cell cycle arrest is important for survival in yeast and animals. Checkpoint mutants progress through the cell cycle in the presence of DNA damage, which results in growth suppression and hypersensitivity to genotoxins. This is typified by the poor growth of atr mutants in the presence of hydroxyurea, a chemical that causes DNA replication defects by inhibiting the enzyme ribonucleotide reductase, depleting the dNTP pool (Culligan et al., 2004).
The core cell cycle machinery is relatively well conserved across eukaryotes and is controlled by complexes of cyclin and cyclin-dependent kinases (CDKs) (Dewitte & Murray, 2003; Inze & De Veylder, 2006). Whilst yeast has a single CDK activity, plants possess multiple forms of this enzyme (Francis, 2007) including CDKA;1, which acts at the G1/S transition and in G2. CDKA;1 contains conserved motifs important for CDK function in other organisms, including the regulatory phosphorylation sites Thr-14 and Tyr-15 and further levels of control include regulation by the coordinated synthesis and degradation of cyclin proteins (Inze & De Veylder, 2006). Phosphorylation of CDK plays important roles in cell cycle control and is controlled by the activities of protein kinases and phosphatases.
In mammals, the phosphorylation state of CDK is controlled by the protein phosphatase Cdc25, which is a substrate of the Chk1 and Chk2 kinases under the control of the ATM- and ATR-mediated DNA damage signalling pathway (Falck et al., 2001; Sorensen et al., 2003; Jazayeri et al., 2006). Although a plant CDC25 homologue has been identified, analysis of Arabidopsis plants deficient in or overexpressing CDC25 demonstrated little or no alteration in the sensitivity to DNA replication stress, indicating that CDC25 does not play a crucial role in the control of cell cycle progression in plants (Dissmeyer et al., 2009; Spadafora et al., 2011). In plants, phosphorylation of CDKA;1 is controlled by the protein kinase WEE1 (Fig. 2), a conserved component of cell cycle control that inhibits CDKA;1 kinase activity and cell cycle progression (De Schutter et al., 2007). Plants lacking WEE1 fail to undergo cell cycle arrest when exposed to the DNA replication blocking agent hydroxyurea, leading to hypersensitivity to the genotoxin. By contrast, overexpression of WEE1 results in phenotypes consistent with permanent activation of cell cycle checkpoints, including cell cycle arrest, differentiation of stem cells and resulting shrinkage of the meristem (De Schutter et al., 2007; Ricaud et al., 2007). A reduction in cell proliferation is also observed in plants in which Thr-14 and Tyr-15 of CDKA;1 are substituted with Asp and Glu, respectively, to mimic phosphorylation and constitutive checkpoint activation (Dissmeyer et al., 2009). These results all point to a key role for WEE1-mediated CDKA;1 phosphorylation in the control of the cell cycle in response to replication stress (Dissmeyer et al., 2009). Surprisingly, WEE1 is not required for the response to DNA damage or for normal cell cycle progression (Cools et al., 2011), and the nature of DNA damage checkpoints in plants remains to be fully elucidated.
In response to DNA damage, the G2-associated cyclins (CYC) and cyclin dependent kinases (CDK) CYCB1;2, CDKB1;2 and CDKB2;1 are down-regulated within 8 h of gamma irradiation, and reduced expression continues for 24 h subsequent to DNA damage (Culligan et al., 2006). In marked contrast, the G2/M phase-specific CYCLINB1;1 displays a rapid ATM-dependent transcriptional induction within 1 h of DNA damage, peaking at c. 8 h and finally reducing by 24 h (Culligan et al., 2006; Ricaud et al., 2007). CYCLINB1;1 protein levels were also found to increase and remain high up to 4 d post irradiation. Unlike the ATM-dependent CYCLINB1;1 transcript induction, appearance of the protein was largely dependent on ATR, with a small contribution from ATM (Culligan et al., 2006). These results, together with flow cytometry data, were consistent with ATR-dependent cell cycle arrest at both G1 and G2 stages in irradiated plants (Hefner et al., 2006; Yoshiyama et al., 2009) whilst microarray analysis pointed to arrest in S-phase in response to DNA damage (Takahash et al., 2010).
2. Programmed cell death
Programmed cell death (PCD) is an in-built mechanism for cellular destruction found in multicellular organisms and, in response to DNA damage, functions to minimize genetic defects that would otherwise lead to the production of clonal populations of mutant cells. Failures in this system can result in uncontrolled cellular proliferation and cancerous cellular growth in mammals. Growth of undifferentiated dividing cells (callus) is not generally observed in plants following genotoxin treatment. However, disruption of meristems and abnormal growth can occur, in particular after irradiation of DNA repair-deficient mutants (Ricaud et al., 2007; Abe et al., 2009; Charbonnel et al., 2011). Genome maintenance in plant vegetative meristems is of particular importance as these tissues eventually give rise to the germline. Mutations arising in apical meristem cells can be passed on to the next generation, although DNA damage-induced growth arrest to allow time for DNA repair has to be balanced against the requirement for continued growth in a range of environmental conditions in the competition for sunlight.
Some protection of cell lineages may be provided by the organization of the meristem. Root and shoot meristems consist of a pool of slowly dividing stem cells, the quiescent centre (QC) and the central zone (CZ), respectively, surrounded by more rapidly dividing initials that give rise to organs (Vernoux et al., 2000). The slower divisions of the QC and CZ may provide more time to repair the genome before cell division or DNA replication and thus ensure that these cells survive acute genotoxic stress and maintain a pool of cells for the generation of more rapidly dividing meristem cells (Fulcher & Sablowski, 2009). Genome integrity may be preserved in shoot apical meristem (SAM) cells by the elevated expression of genes involved in DNA metabolism, replication and repair (Yadav et al., 2009), also reflecting an increased requirement for repair during DNA replication. Vegetative meristem cells also display a lower threshold for the induction of cell death in response to genotoxins compared with cells of differentiated tissues; this may represent additional quality control mechanisms to preserve the quality of the stem cell population (Curtis & Hays, 2007; Ricaud et al., 2007; Fulcher & Sablowski, 2009; Furukawa et al., 2010). In particular, the rapidly dividing root meristem initials and their early descendents display a higher incidence of programmed cell death in response to genotoxic stress than surrounding vegetative cells (Ricaud et al., 2007; Fulcher & Sablowski, 2009). Cell death is a programmed event, rather than a consequence of DSB cytotoxicity, as the death of root initials and the shoot apical meristem was dependent on DNA damage signalling, largely mediated by ATM and SOG1 (Fulcher & Sablowski, 2009; Furukawa et al., 2010). PCD required de novo protein synthesis, and, whilst the mechanism remains to be elucidated, it is apparent that it is distinct from mammalian apoptosis and involves autolysis (Fulcher & Sablowski, 2009).
VI. Chromatin structure and DSB repair
Chromatin structure is important for genome stability and is established through nucleosome assembly during the S phase and repair synthesis and is further modified by epigenetic maintenance. The deposition of H3 and H4 on to newly synthesized DNA during the first stage of de novo chromatin assembly requires the evolutionarily conserved CHROMATIN ASSEMBLY FACTOR 1 (CAF-1), a heterotrimeric complex comprising FASCIATA 1 (FAS1, FAS2) and MSI1 subunits (Kirik et al., 2006). Depletion of CAF-1 causes defects in meristem organization resulting in characteristic growth fasciation, also observed in mre11, brca2 mutants and in wild-type plants after high doses of irradiation (Bundock & Hooykaas, 2002; Abe et al., 2009). Chromatin assembly defects result in hypersensitivity to genotoxins, increased basal levels of DSBs and constitutive activation of the DNA damage response including H2AX phosphorylation and RAD51 induction (Endo et al., 2006; Ramirez-Parra & Gutierrez, 2007). This results in a large increase in spontaneous intrachromosomal recombination with 40-fold increases in HR rates observed in fas1 and fas2 mutants (Takeda et al., 2004; Endo et al., 2006). Defects in chromatin assembly also lead to loss of transcriptional gene silencing (TGS) as a result of destabilization of heterochromatin, observed in mutants in the CAF-1 complex and the putative chromatin assembly factor BRU1 (Takeda et al., 2004; Ramirez-Parra & Gutierrez, 2007). DNA replication mutants, including plants deficient in REPLICATION PROTEIN A (RPA1), DNA POLYMERASE ε and TOPOISOMERASE VI, share many features of chromatin assembly mutants, including constitutive activation of the DNA damage response and, in some mutants, loss of TGS (Elmayan et al., 2005; Breuer et al., 2007; Yin et al., 2009).
1. Remodelling activities
Chromatin-remodelling activities play key roles in eukaryotic DNA repair and replication and require the activity of some of the 41 members of the SWItch/Sucrose NonFermentable (SWI2/SNF2) superfamily present in Arabidopsis (Knizewski et al., 2008; Clapier & Cairns, 2009). The Arabidopsis SWI2/SNF2 protein RAD54 is a chromatin-remodelling factor with important roles in HR. AtRAD54 can complement the genotoxin hypersensitivity of yeast rad54 mutants and also confers enhanced resistance to γ-irradiation when overexpressed in plants (Klutstein et al., 2008). Yeast RAD5 is a SWI2/SNF2 protein with roles in post-replication repair. Arabidopsis has two RAD5 homologues, RAD5A and B; rad5a mutants are hypersensitive to methyl methanesulfonate (MMS) and mitomycin C and are deficient in the induction of HR upon genotoxin treatment observed in wild-type lines (Chen et al., 2008). The function of RAD5 is to facilitate recombination, especially at blocked replication forks. After collapse of a replication fork, a double-stranded DNA end is formed and HR plays important roles in mediating repair of these substrates to re-establish the replication fork (Mannuss et al., 2010). At least three different pathways mediate these repair processes and involve RAD5A, the nuclease MUS81 and the helicase REQ4A (Mannuss et al., 2010). The close link between chromatin structure and repair was illustrated by a screen of 14 Arabidopsis SWI2/SNF2 RNAi lines that identified 11 remodelling genes required for resistance to genotoxins (Shaked et al., 2006). Plants deficient in representatives from most SWI2/SNF2 subfamilies exhibited hypersensitivity to genotoxins, including PICKLE/GYMNOS, PIE/SCRAP and DEFICIENT IN DNA METHYLATION (DDM1) (Shaked et al., 2006).
2. Structural maintenance of chromosomes proteins and DSB repair
Structural maintenance of chromosomes (SMC) proteins are large ATPases that play critical roles in chromatid association, chromosome condensation and DNA repair, and members include the MRN complex component RAD50. SMC proteins assemble in different heterodimeric combinations which form part of larger multi-subunit protein complexes (Jessberger, 2002; Lehmann, 2005). The cohesin complex includes SMC1 and 3 and maintains sister chromatid cohesion (Jessberger, 2002). Deficiency in a subunit of the Arabidopsis cohesin complex, RAD21.1, results in hypersensitivity of the mutant plants to genotoxins and a decrease in the rate of DSB repair (da Costa-Nunes et al., 2006; Kozak et al., 2009). Cohesin establishment requires additional factors, two of which, E2F target gene 1 (ETG) and CTF18, have been identified in Arabidopsis (Takahashi et al., 2010). Mutants in these genes result in a partial loss in chromatid cohesion, with a more severe effect in the double mutant line. etg mutants display a constitutive DNA damage response, with up-regulation of DNA damage-inducible transcripts, and both this response and the accumulation of DNA damage were enhanced in the etg ctf18 double mutants. These results point to a requirement for sister chromatid cohesion in maintaining genome stability in plants (Takahashi et al., 2010). Sister chromatids display increased alignment following induction of DNA damage, suggesting roles for cohesion in DNA repair pathways (Watanabe et al., 2009). This DNA damage-induced cohesion is stimulated by the SMC5-6 complex, which has also been shown to promote recombination between sister chromatids in yeast cells (De Piccoli et al., 2006; Watanabe et al., 2009). Arabidopsis mutants in the SMC6 homologue MIM (hypersensitive to MMS, Irradiation and MMC) display greatly reduced rates of DSB repair, reduced intrachromosomal homologous recombination and increased sensitivity to a broad range of genotoxins (Mengiste et al., 1999; Kozak et al., 2009).
VII. Genome stability and environmental stress
Exciting recent advances have illustrated that genetic and epigenetic changes in plant genomes have important roles in adaptation to biotic and abiotic stresses (reviewed in Boyko & Kovalchuk, 2011). Somatic homologous recombination frequencies (HRFs) are up-regulated by environmental changes, including exposure to a wide range of stresses (Fig. 3) such as salt, heavy metals, UV irradiation, drought, cold, flood, pathogens and elicitors, and HRF also varies with day length, developmental stage and tissue type (Kovalchuk et al., 2003; Boyko et al., 2005, 2006b,c; Boyko et al., 2010a). A common factor associated with many of these stresses is the generation of reactive oxygen species, although transcriptional re-programming and epigenetic changes associated with environmental stimuli may also induce changes to chromatin accessibility with knock-on effects on background recombination frequencies. Increases in HR frequencies may be a programmed response to accelerate evolutionary adaption and generate new resistance traits, enabling greater genomic plasticity in plants in response to adverse environmental conditions (Molinier et al., 2006; Boyko & Kovalchuk, 2011). For example, studies following the Chernobyl nuclear power plant accident revealed evidence of heritable adaptation of plants to increased levels of genomic stress; the progeny of plants grown in highly contaminated areas displayed increased resistance to radiomimetic agents that induce DSBs (Kovalchuk et al., 2004). Enhanced DNA DSB repair capacity in response to γ-irradiation along with increased stress tolerance was also observed in Betula verrucosa (birch) pollen collected from contaminated sites in the Chernobyl region (Boubriak et al., 2008).
Elevated levels of somatic HR frequencies in response to a variety of stresses are transmitted to the next generation, although this effect was not identified in all studies (Pecinka et al., 2009). UV-C-treated plants displayed elevated recombination that could still be detected four generations after treatment (Molinier et al., 2006). The increases in HRF were accompanied by inheritable global changes in epigenetic status passed through the female gametes (Boyko & Kovalchuk, 2010). These epigenetic changes associated with transgenerational inheritance of stress tolerance required Dicer enzymes, responsible for RNA silencing and guiding methylation patterns (Boyko & Kovalchuk, 2010; Boyko et al., 2010a). This may represent an adaptation mechanisms for plants, priming the progeny to cope with environmental stresses (Boyko & Kovalchuk, 2011). Interestingly, the pathogen-induced increase in recombination frequencies is transferred systemically through the plant (Kovalchuk et al., 2003).
The induction of intrachromosomal HR by biotic stress may also be directly linked to plant pathogen defence, as the HR protein RAD51D is required for the expression of pathogenesis-related (PR) genes and systemic acquired resistance (SAR). In the absence of RAD51D, plants are hypersensitive to DNA-damaging agents and more susceptible to disease as a result of defective PR gene induction (Durrant et al., 2007). Interestingly, the HR complex of BRCA2–RAD51 is also required for the induction of PR genes and binds to PR gene promoters when plants are exposed to the SAR elicitor salicylic acid (Wang et al., 2010). This may represent a newly identified, independent function of the HR machinery in plants, or alternatively the presence of HR proteins at the PR gene promoter may also function to safeguard the genome as chromatin accessibility (and the potential for DNA damage) increases as these genes become induced by plant pathogens.
1. Desiccation, quiescence and DNA repair in seeds
Plants display a greater resistance to DSB-inducing genotoxins than humans, with maize (Zea mays) surviving X-ray doses two to three times the 10-Gray dose lethal to humans, even though they have comparable genome sizes (Killion & Constantin, 1972). Why have plants evolved to survive levels of X-rays that would never be encountered in the environment? Radiation resistance has been observed in organisms that can withstand desiccation, including the bacterium Deinococcus radiodurans and the invertebrate bdelloid rotifers. Deinococcus radiodurans can survive dehydration to < 5% water content for up to 6 wk. After 8 d DSBs begin to accumulate and within a month the genome is severely fragmented, whereas other lethal stresses including UV, heat and starvation do not cause an equivalent DSB accumulation (Mattimore & Battista, 1996). After γ-ray doses of 7000 Gy the D. radiodurans genome is reduced to 20–30-kb fragments but is rebuilt in 1.5 h and 80–90% of the bacteria survive (Zahradka et al., 2006). Similarly, bdelloid rotifers live in freshwater pools and can tolerate desiccation as their habitat dries up; these invertebrates can withstand γ-ray doses in excess of 1000 Gy, far higher radioresistance than animals with similar sized genomes, supporting the link between desiccation tolerance and high capacity for DNA repair (Gladyshev & Meselson, 2008).
The plant’s lifecycle can involve prolonged periods of quiescence during which the plant embryo is maintained within the seed and in some cases remains viable for hundreds of years (Sallon et al., 2008). Seeds that can survive drying to low moisture content are termed ‘orthodox’ seeds (Bewley & Black, 1994). Metabolism virtually ceases in the quiescent embryo and the low activity of cellular maintenance pathways, including DNA repair, results in a decline in seed viability during storage. In addition, dehydration, dry storage and subsequent water imbibition upon germination of the seed are associated with high levels of oxidative stress and significant damage to the integrity of the genome of the embryo of the seed, especially if seeds are stored under unfavourable conditions such as high temperatures and moisture contents (Villiers, 1974; Dandoy et al., 1987; Rajjou & Debeaujon, 2008; Kranner et al., 2010). There is a strong correlation between the loss of viability of seeds in storage, chromosomal breakages and the incidence of aberrant chromosomes (Abdalla & Roberts, 1969; Roberts, 1972; Cheah & Osborne, 1978; Osborne et al., 1984). However, even high-quality (unaged) seeds display a background level of DSBs and ‘there is no threshold level of viability before chromosomal aberrations occurs’ (Dourado & Roberts, 1984). Consistent with this observation, unaged Arabidopsis seeds display transcriptional up-regulation of DSB-responsive genes in the earliest stages of germination, coincident with DNA repair synthesis (Waterworth et al., 2010). Repair of DSBs in germination is mediated in part by NHEJ and irradiated ku70 and lig4 mutant seeds germinate more slowly than wild type controls (Friesner & Britt, 2003). In addition, a DNA ligase unique to plants, termed LIG6, is important for germination and viability, especially in aged seeds. Accelerated ageing procedures reduced both the germination rate and the viability of lig4 lig6 double mutant lines, which are more severely affected by accelerated ageing protocols than the single mutant lines. This suggests that these two ligases function in distinct repair pathways, as discussed in Section VIII.2, possibly classical NHEJ (LIG4) and an alternative NHEJ pathway (LIG6), to remove DNA damage in seeds early in imbibition (Waterworth et al., 2010).
In the natural environment, most seeds will remain desiccated for only a brief period before becoming rehydrated in the soil where they can exist in a dormant state. DNA repair processes operate in imbibed dormant seeds (Villiers, 1974); for example, wild oats (Avena fatua) undertake efficient repair of damaged DNA over several imbibition and re-drying cycles (Boubriak et al., 1997), which is key to helping ensure their survival over long periods in the soil seed bank. Similarly, controlled imbibition or ‘osmopriming’ of seeds facilitates biochemical processes including repair of damaged DNA (Ashraf & Bray, 1993) and confers enhanced germination properties, adding commercial value to primed seed stocks. An osmopriming-like process has evolved to enable survival of the desert-dwelling plant species Artemesia sphaerocaphala, which uses a water-absorbing polysaccharide layer secreted onto the seed surface to trap night-time dew to permit sufficient rehydration of cells to allow night-time repair of damaged DNA (Huang et al., 2008). Seeds subjected to γ-irradiation were able to repair damaged DNA overnight as a result of the presence of the hydrated mucilage seed covering before re-drying in the sun next morning (Huang et al., 2008).
VIII. Mechanisms of DSB repair
DSB repair mechanisms fall into two classes: homology-dependent, termed ‘homologous recombination’ (HR), and homology-independent or illegitimate recombination (also termed ‘non-homologous end joining’ (NHEJ)). In HR a highly similar sequence in the genome is used as a template for repair, whereas NHEJ rejoins DNA ends with little sequence dependence beyond a few bases of microhomology that may serve to stabilize a joining reaction. Both pathways have the potential to be mutagenic. Illegitimate recombination is usually associated with short deletions or insertions but can result in more serious errors including chromosome fusions. Homologous recombination (HR) usually results in restoration of the original sequence but also has the potential to copy allelic genes, leading to loss of heterozygosity, or can even lead to crossing over between nonhomologous chromosomes which can result in severe genomic instability and cell death.
NHEJ is a highly conserved pathway with important roles in eukaryotes and prokaryotes (Fig. 4) (Lieber, 2010). Mammalian NHEJ mutants are often embryo-lethal whereas Arabidopsis NHEJ knockout plants are largely indistinguishable from wild-type plants under ideal growth conditions. However, more detailed analysis reveals that NHEJ mutants do display higher levels of background stress, with the permanent activation of the DNA damage response and increased spontaneous cell death in the absence of genotoxins (West et al., 2004; Fulcher & Sablowski, 2009). The products of NHEJ have been studied in plants using extrachromosomal rejoining assays, repair of endonuclease-induced DSBs, and analysis of chromosome fusions in telomerase mutants (Gorbunova & Levy, 1997; Salomon & Puchta, 1998; Heacock et al., 2004). These studies indicate the frequent use of microhomology at the break site, deletions and appearance insertions of ‘filler’ DNA indicative of DNA synthesis-dependent repair process whereby a single-stranded DNA end primes DNA synthesis using an ectopic site of the genome as a template. Eukaryotic NHEJ pathways may involve a wide range of factors, and early studies in mammals and yeast identified conserved core components including the KU70–KU80 complex and the DNA LIGASE 4–XRCC4 complex (Taccioli et al., 1994; Critchlow et al., 1997). The KU70–KU80 complex possesses a high affinity for exposed DNA ends and in yeast functions to stabilize the broken DNA ends and protect them from degradation via exonuclease activity (Boulton & Jackson, 1996). Arabidopsis ku mutants display greater dependence on the use of microhomologies between the DNA ends and a lower frequency of insertions (Heacock et al., 2004), although loss of KU did not result in increased length of deletions (Heacock et al., 2007). Damage-induced DSBs differ from those arising from endonuclease activities and damaged DNA ends will often not have the 5′ phosphate and 3′ hydroxl group required for ligation. A range of DNA end processing factors modify the DNA to produce DNA ends suitable for rejoining by DNA ligases. These factors are poorly characterized in plants; in yeast the MRE11–RAD50–XRS2 (MRX) complex is known to have important roles in the repair of nonligatable ends (Zhang & Paull, 2005). The plant MRE11–RAD50–NBS1 (MRN) complex has roles in DSB detection and signalling (Amiard et al., 2010), and mutants deficient in MRN display hypersensitivity to DNA-damaging agents (Gallego et al., 2001; Bundock & Hooykaas, 2002; Waterworth et al., 2007). MRE11 is required for the microhomology-dependent repair observed in ku70 mutants, as demonstrated in telomere fusion assays (Heacock et al., 2004).
The combination of processing activities produces DNA ends that are ligated by the LIG4–XRCC4 complex, which has DNA-binding activity and ATP-dependent ligase activity (West et al., 2000). In the absence of both LIG4 and KU70, DSB repair is less dependent on microhomologies and displays greater deletion when compared with ku70 mutants, consistent with slower repair in the absence of LIG4 leading to greater exonuclease digestion (Heacock et al., 2007). The understanding of NHEJ processes is far more advanced in mammals, where the KU complex recruits nuclease activities, polymerases, and end-processing accessory factors (Lieber, 2010). Many of these factors have no characterized homologues in plants, including Artemis and the DNA-dependent protein kinase catalytic subunit. However, a rice (Oryza sativa) homologue of DNA polymerase λ has been reported (Uchiyama et al., 2004); in mammals this polymerase is recruited to the KU complex via its BRCT domain (Ma et al., 2004) although involvement of plant DNA polymerase λ in DSB repair remains to be determined.
2. Novel pathways of NHEJ in plants
Pathways of DSB repair need to be able to repair a wide variety of damage products. Part of this flexibility is achieved by the recruitment of various repair factors as required, meaning that the NHEJ pathway can have a range of outcomes depending on the repair components involved. This is also illustrated by the activities of alternative NHEJ pathways for DSB repair that operate in the absence of one or more of the conserved NHEJ components, and there is much debate in the literature as to what extent these represent distinct pathways or a range of factors that can contribute through recruitment to a central mechanism (Lieber, 2010; Mladenov & Iliakis, 2011). In mammals, backup NHEJ pathways (B-NHEJ) independent of KU have been described, including microhomology-mediated end joining (MMEJ) and alternative NHEJ (alt-NHEJ or A-NHEJ). In some cases these display greater dependence on microhomologies, and identified B-NHEJ factors include the MRN complex, PARP and the XRCC1–DNA ligase 3 complex (Mladenov & Iliakis, 2011) Evidence for KU- and LIG4-independent DSB repair is also well established in plants (reviewed in Bray & West, 2005). Recently, a number of studies have identified components of these alterative end-joining pathways in plant cells and the corresponding mutant lines display slowed rates of DSB repair. Analysis of repair kinetics enables distinct phases of DSB repair corresponding to different repair pathways to be distinguished in plants. Repair kinetics have been analysed using two independent methods to quantify DNA damage: single cell electrophoresis (the comet assay) employs electrophoretic analysis of individual nuclei and image analysis to quantify levels of fragmented DNA, while immunocytochemistry detects the formation and disappearance of DNA damage foci using antisera to phosphorylated histone γH2AX. Currently technical limitations in plants mean that γH2AX detection can only be performed in M-phase cells, so caution is required in direct comparisons between repair kinetics measured by γH2AX foci and data obtained using the comet assay, which averages results from a mixed population of cells derived from whole seedlings. DSB repair in Arabidopsis displays an initial rapid phase of repair followed by a slower phase(s) of repair with half the breaks repaired within c. 6 min, as determined by the comet assay (Kozak et al., 2009), and c. 40 min, as assayed by analysis of H2AX foci (Charbonnel et al., 2010). Detailed analysis of repair kinetics in mutant lines has implicated a number of proteins in alternative DSB repair pathways in higher plants, including those also associated with plant SSB repair pathways (DNA LIGASE 1 (LIG1) and XRCC1) (Waterworth et al., 2009; Charbonnel et al., 2010). These studies also identified roles for the structural maintenance of chromosomes-like proteins MIM and RAD21.1, which may function to stabilize the DSB and are required for the initial rapid phase of repair (Kozak et al., 2009). It has also been shown that KU80 and XRCC1 act redundantly in the initial stages of repair (Charbonnel et al., 2010). Further analysis of the backup pathways in Arabidopsis investigated repair activities in plants mutated in the known DSB repair factors: KU80, XRCC2 (homologous recombination), XRCC1 (DSB repair using single-strand break repair proteins) and XPF (single-strand break repair and single-strand annealing of double-strand breaks, including MMEJ), both singly and in combination. The conclusions from this study were that there is a hierarchy of DSB repair pathways, with KU-dependent end joining being the major pathway in plants, but that in the quadruple ku80 xrcc1 xrcc2 xfp mutants an uncharacterized DSB repair pathway was active even in the absence of all four repair factors (Charbonnel et al., 2011). These studies clearly reflect the very robust mechanisms for DSB repair present in plant cells.
3. Homologous recombination
Homologous recombination involves base-pairing between DNA molecules originating from two different DNA duplexes. HR is initiated by resection of double-stranded DNA by nuclease activity to produce long 3′ single-stranded DNA ends, which are bound by HR proteins (Fig. 5). The single-stranded nucleoprotein filament is then involved in the homology search and invasion of duplex DNA (Baumann et al., 1996). The invading strand may not be identical to the homologous region, resulting in the formation of heteroduplex DNA, and mispaired bases are repaired by the mismatch repair machinery. Heteroduplex formation is controlled depending on the degree of similarity between the mismatched sequences (Opperman et al., 2004; Emmanuel et al., 2006). On finding a region of homology, recombination can proceed by DNA synthesis, copying sequence information from the invaded duplex and continued strand invasion and branch migration. The subsequent steps differ between pathways, but in the case of the synthesis-dependent strand-annealing (SDSA) pathway, the invading strand dissociates from the invaded duplex and recombines with the original chromatid (Ferguson & Holloman, 1996). An alternative outcome is that DNA synthesis continues all the way to the end of the chromosome, a process termed ‘break-induced replication’ (BIR) that differs mechanistically from SDSA (Llorente et al., 2008). The frequency of BIR in plants is unknown, but may be less than that found in yeast (Watanabe et al., 2009).
During strand invasion and subsequent DNA synthesis, one strand of the invaded duplex is displaced, forming a D-loop. When the displaced strand base-pairs with a homologous region on the other side of the original break it is termed ‘second-end capture’ and is described by the double-strand break repair model of HR. In this model, two Holliday junctions are formed, linking the chromatids together (Szostak et al., 1983). Holliday junctions can be resolved in two ways depending on which DNA strand is cut and this determines whether crossovers occur between chromosomes (Mazon et al., 2010). When a break occurs between repeated sequences, single-strand annealing (SSA) can occur, involving base-pairing of resected ends and resulting in the deletion of the region intervening the tandem repeats (Prado & Aguilera, 1995). This pathway is highly active in plants and accounts for around a third of repair events between repeats (Siebert & Puchta, 2002).
A key feature of HR pathways is the homology search which, in vitro, requires a single protein: RAD51. Arabidopsis mutants in RAD51 are sterile, indicating the essential role of RAD51 in homologous recombination (Li et al., 2004). These plants are phenotypically normal and do not display pronounced hypersensitivity to DSB-inducing treatments, including X-rays, suggestive of a minor role of HR compared with NHEJ in DSB repair. It remains possible that repair activities of RAD51 are required in specific developmental stages, and evidence from maize indicates roles during seed germination, with RAD51-deficient seeds failing to produce normal seedlings after irradiation (Li et al., 2007). The moss Physcomitrella patens, in which HR is the major pathway for repair, has two copies of RAD51, and mutation of both copies results in severe growth defects and up-regulation of NHEJ (Markmann-Mulisch et al., 2007; Schaefer et al., 2010). The various processes of strand invasion and subsequent processing of recombination intermediates require additional protein factors including the RAD51-like proteins RAD51B, RAD51C, RAD51D, XRCC2 and XRCC3, which form distinct protein complexes that help in the steps of strand invasion and later steps in recombination (reviewed in Bleuyard et al., 2005; Osman et al., 2011). Chromatin-remodelling enzymes assist in homology searching, and DNA helicases (Blanck et al., 2009) and resolvases determine the final outcome of recombination events. Over the past few years there has been significant progress in our understanding of helicase functions in plant recombination and this work has been recently reviewed (Hartung & Puchta, 2006; Knoll & Puchta, 2011).
4. Choice of repair pathway
Much of our understanding of DSB repair mechanisms in higher plants has been gained from studies of DNA break induction with the endonuclease I-SceI and can also be inferred from analysis of transgene repair and integration products and genome repair subsequent to transposon excision. Collectively, these studies clearly demonstrate that DSB repair is mediated largely by NHEJ-like mechanisms. Homologous recombination with a nonallelic (ectopic) site in the genome or with a homologous chromosome only accounts for c. 0.01% of repair events (Puchta, 1999), whilst intrachromosomal recombination is much more frequent. In the case where a break is flanked by a repeated sequence the majority of repair still occurs by NHEJ, although in this case a third of repair events occur by SSA (Siebert & Puchta, 2002).
5. Molecular control
In vivo, homologous recombination activity has to be carefully regulated as higher eukaryotic genomes contain highly repetitive DNA; for example in humans, repeats of the Alu element accounts for c. 10% of the genome (Smit, 1996). The activity of the major recombinase RAD51 is regulated in yeast by RAD52, whilst breast cancer-associated gene 2 (BRCA2) fulfils this function in mammals. BRCA2 binds and sequesters RAD51, regulating RAD51 single-stranded DNA binding and release and RAD51 duplex invasion (Esashi et al., 2005). The plant BRCA2 homologue also has RAD51-binding activity and is important in DNA repair (Dray et al., 2006; Abe et al., 2009), whilst no clear homologue of RAD52 has been reported in plants.
HR activity is tightly regulated depending on the cell cycle stage in yeast and mammals, with lowest activity in G1, before replication of a sister chromatid. An important aspect of this regulation is the control over whether DSB ends are resected to produce 3′ single-stranded DNA tails (Longhese et al., 2010). A strong candidate for a key regulatory factor regulation is CtIP (also known as Com1 or Sae2 in yeast), a protein that interacts with the MRN complex and initiates HR (Sartori et al., 2007). Arabidopsis com1 mutants display the sterility typical of HR-deficient plants (Uanschou et al., 2007). Yeast Sae2 activity is controlled by Cdk phosphorylation (Huertas et al., 2008); similarly, mammalian CtIP activity is controlled by phosphorylation on consensus Cdk sites (Yun & Hiom, 2009), raising the possibility that these conserved mechanisms of HR regulation also operate to control COM1 activity in plants.
6. Transgene insertion and gene targeting
In the process of Agrobacterium-mediated plant transformation, the integration step of the T-DNA requires the activities of host factors, although the mechanisms of integration remain poorly defined (Gelvin, 2010). However, there are links between DSB repair and T-DNA integration as T-DNA integrates with high frequency at breaks in the genome (Salomon & Puchta, 1998). Topologically, T-DNA integration represents a DSB repair event, although the mechanism of integration could also occur via single-strand intermediates, as detailed in several possible models (Johzuka-Hisatomi et al., 2008). Evidence for the involvement of the plant recombination pathways is provided by analysis of ku80 mutants which display significantly reduced integration frequencies in both floral dip and root transformation experiments (Friesner & Britt, 2003; Gallego et al., 2003; Li et al., 2005). T-DNA integration was found to be dependent on proteins that stabilize the transgene before integration, which could partially account for the role of KU80 (Mysore et al., 2000). Significantly, KU80 mutation in plants reduces, rather than abolishes, T-DNA integration, in contrast to yeast where NHEJ proteins are essential for T-DNA integration (van Attikum et al., 2001). In plants, LIG4 is less important for Agrobacterium-mediated transformation, with no significant reduction observed in floral dip transformation rates, although transformation of lig4 callus suggested a decrease in the random integration of double-stranded DNA and enrichment of targeting events (van Attikum et al., 2003; Tanaka et al., 2010).
Transgene integration and DSB repair are closely linked processes and numerous studies conducted 10–20 yr ago provided much insight into the mechanisms of illegitimate recombination in plants through analysis of transgene integration sites (Somers & Makarevitch, 2004). However, the predominant activity of the illegitimate recombination pathways poses a major barrier to targeted transgene integration. The most successful approach to alter plant recombination pathways was provided by overexpression of yeast RAD54 in Arabidopsis, which increased targeting c. 27-fold, possibly reflecting increased chromatin accessibility facilitating the homology search (Shaked et al., 2005). Recent work in plants and animals has shown great advances in increasing gene targeting frequencies through the targeted induction of a DSB at the pre-selected locus using high-specificity zinc finger endonucleases (Bibikova et al., 2003; Porteus & Baltimore, 2003; Shukla et al., 2009; Townsend et al., 2009). This is achieved using engineered enzymes consisting of the endonuclease domain from the restriction enzyme FokI linked to an array of zinc finger motifs (small protein domains found in many transcription factors) that recognize specific base triplet sequences. The assembly of a series of different zinc finger motifs is used to design heterodimeric endonucleases with a recognition sequence of 18–24 bases, specific to the gene to be targeted. Zinc finger technology has proved highly successful, and mirroring the early experiments with I-SceI, achieving high frequencies of targeted transgene insertion. The flexibility of this approach to target any gene of choice is illustrated by two landmark publications showing targeted transgene insertion at the inositol-1,3,4,5,6-pentakisphosphate 2-kinase encoding gene IPK1 and the acetolactate synthase genes SuRA and SuRB in maize and tobacco (Nicotiana tabacum), respectively (Shukla et al., 2009; Townsend et al., 2009).
An alternative approach has also proved highly successful in achieving high-frequency gene targeting in rice. This employs selection against random integration events by flanking regions of homology with the powerful cytotoxic diphtheria toxin gene (DT-A) (Terada et al., 2002). NHEJ-mediated plant transgene insertion results in incorporation and expression of the diphtheria gene and cell death whereas HR-mediated insertion removes the flanking DT-A region. This provides strong selection against random integration and results in reproducible targeting frequencies of c. 1–2% (Johzuka-Hisatomi et al., 2008). These advances in gene targeting have enabled a molecular analysis of targeting events, showing that strand invasion results in heteroduplex formation and mismatch correction. Several models could account for the observed recombination products, including strand invasion, branch migration and synthesis-dependent strand annealing (Iida & Terada, 2004; Johzuka-Hisatomi et al., 2008).
The last few years have seen great progression in our understanding of plant DNA repair and recombination mechanisms and the main pathways of repair are becoming well characterized. However, it is clear that plants possess alternative mechanisms of end joining and the first details of these mechanisms are being uncovered, revealing both similarities with and differences from animal and yeast DNA repair. Despite our increase in knowledge, the mechanisms of transgene integration are still largely unknown. In particular, the insertion mechanism(s) of T-DNA, which is an essential step in the transformation of plants using Agrobacterium, remains obscure, despite the importance of this process to biotechnology.
The mechanisms by which plant cells respond to genotoxic stress, including the key initial steps of DSB detection, signalling and activation of cell cycle checkpoints, are beginning to be understood. Future challenges will be to unravel how these decisions are made and elucidation of the mechanisms underlying these responses, which appear to differ greatly from mammalian equivalents. Over the coming decade significant advances in this area will inform us about how plants cope with stresses, provide possible ways to improve growth under stress conditions, and deepen our understanding of the long-term effects associated with adverse environmental conditions on plant genomes.
We thank the UK Biotechnology and Biological Sciences Research Council (grant numbers BB/H012346 and BB/G001723) and the Leverhulme Trust (F/10 105/A) for supporting the authors. We apologise to all those whose work was not cited, as a consequence largely of the broad range of the review which meant much important work was omitted or only briefly covered.