Suppression of Ku70/80 or Lig4 leads to decreased stable transformation and enhanced homologous recombination in rice

Summary Evidence for the involvement of the nonhomologous end joining (NHEJ) pathway in Agrobacterium-mediated transferred DNA (T-DNA) integration into the genome of the model plant Arabidopsis remains inconclusive. Having established a rapid and highly efficient Agrobacterium-mediated transformation system in rice (Oryza sativa) using scutellum-derived calli, we examined here the involvement of the NHEJ pathway in Agrobacterium-mediated stable transformation in rice. Rice calli from OsKu70,OsKu80 and OsLig4 knockdown (KD) plants were infected with Agrobacterium harboring a sensitive emerald luciferase (LUC) reporter construct to evaluate stable expression and a green fluorescent protein (GFP) construct to monitor transient expression of T-DNA. Transient expression was not suppressed, but stable expression was reduced significantly, in KD plants. Furthermore, KD-Ku70 and KD-Lig4 calli exhibited an increase in the frequency of homologous recombination (HR) compared with control calli. In addition, suppression of OsKu70,OsKu80 and OsLig4 induced the expression of HR-related genes on treatment with DNA-damaging agents. Our findings suggest strongly that NHEJ is involved in Agrobacterium-mediated stable transformation in rice, and that there is a competitive and complementary relationship between the NHEJ and HR pathways for DNA double-strand break repair in rice.


Introduction
Plant cells are continually exposed to endogenous and exogenous genotoxic stresses, such as reactive oxygen species and UV light, which lead to the accumulation of numerous types of DNA damage, including cross-linking of DNA, base oxidation or alkylation, mismatch of bases, DNA single-strand breaks and DNA double-strand breaks (DSBs). DSBs are amongst the most serious types of DNA damage in living cells and can lead to cell death if not repaired. There are at least two repair pathways for DSB repair: nonhomologous end joining (NHEJ), which involves rejoining of the broken DNA ends; and homologous recombination (HR), which is an accurate pathway that uses homologous DNA sequences from the sister chromatid as a template. NHEJ is used preferentially to deal with DSBs in higher eukaryotes, including higher plants (Mladenov & Iliakis, 2011;Symington & Gautier, 2011;Waterworth et al., 2011), whereas HR is the main DSB repair pathway in bacteria and yeast (Pâques & Haber, 1999;Aylon & Kupiec, 2004).
The NHEJ pathway in vertebrates is thought to be as follows. A Ku70/Ku80 complex binds initially to two DNA ends at the DSB site, and then recruits a DNA-dependent protein kinase (DNA-PKcs), which has not been identified in plants. DNA-PKcs phosphorylates and activates many proteins, including nuclease and itself. Ultimately, the Lig4-Xrcc4 complex rejoins the two DNA ends of the break (Mladenov & Iliakis, 2011;Symington & Gautier, 2011). Many proteins involved in NHEJ found in mammalian cells have also been identified in Arabidopsis and rice (reviewed by Singh et al., 2010;Edlinger & Schlögelhofer, 2011), including Arabidopsis Ku70 (Tamura et al., 2002), rice Ku70 (Hong et al., 2010), Arabidopsis Ku80 (Tamura et al., 2002), Arabidopsis DNA ligase 4 (Lig4) (West et al., 2000) and Arabidopsis Xrcc4 (West et al., 2000). InArabidopsis,mutantsoftheku70,ku80andlig4geneshavebeen shown to display hypersensitivity to DSB-inducing agents, including c-irradiation, methyl methanesulfonate, ionizing radiation and bleomycin (vanAttikumet al.,2003;Friesner&Britt,2003;Gallego et al., 2003;Li et al., 2005;Hong et al., 2010;Wang et al., 2010), indicatingthat Ku70/Ku80and Lig4 proteins play animportantrole in DSB repair in Arabidopsis. In addition, recent studies in mammalian cells have shown that DSBs can be rejoined in the presence of chemical inhibition or mutation of key NHEJ factors, Re-use of this article is permitted in accordance with the Terms and Conditions set out at http://wileyonlinelibrary.com/onlineopen#OnlineOpen_Terms such as DNA-PKcs, Ku70, Ku80, Lig4 and XRCC4, suggesting the existence of backup pathways for NHEJ (Mladenov & Iliakis, 2011;Symington & Gautier, 2011). Such a backup NHEJ pathway may utilize poly(ADP-ribose) polymerase-1, MRN, histone H, DNA ligaseIIIandXRCC1(Chenget al.,2011;Mladenov&Iliakis,2011;Symington & Gautier, 2011). In Arabidopsis, several studies have also provided evidence for the existence of Ku-and Lig4-independent pathways, and have identified proteins, such as AtLig1 and AtXRCC1,thatfunctioninthesepathways (Charbonnelet al.,2010(Charbonnelet al., , 2011Waterworth et al., 2009Waterworth et al., , 2011. Agrobacterium-mediated plant genetic transformation uses the transferred DNA (T-DNA) region on a binary plasmid harbored by Agrobacterium tumefaciens as a vector for genetic engineering. DNA repair pathways have been thought to be involved in the integration of T-DNA into the plant genome, and two major models for this integration have been proposed. The firstthe DSB repair modelhypothesizes that single-strand T-DNAs imported into the plant cell nucleus by the virulence protein complex are replicated to a double-stranded form and are subsequently integrated into DSBs in the host genome (Gelvin, 2010;Pitzschke & Hirt, 2010;Magori & Citovsky, 2011). By contrast, the secondthe strand-invasion modelassumes that the 3′ end of single-stranded T-DNA finds a microhomology to plant DNA and invades the target site host DNA. The VirD2-attached 5′ end of the T-DNA binds to a nick in the plant DNA and is ligated. The complementary strand of the T-DNA is synthesized, resulting in integration of a double-strand copy of the T-DNA into the plant genome (Gelvin, 2010;Pitzschke & Hirt, 2010;Magori & Citovsky, 2011).
In Arabidopsis, mutation of either the AtKu80 or AtLig4 gene caused a decrease in the frequency of stable T-DNA integration following an in planta floral dip transformation assay (Friesner & Britt, 2003). Involvement of AtKu80 in T-DNA integration was also observed in a root tumorigenesis assay (Li et al., 2005). By contrast, no decrease in T-DNA integration was observed in AtKu80 mutants using the in planta floral dip transformation assay (Gallego et al., 2003). Furthermore, the AtLig4 mutant was not impaired in T-DNA integration using either the in planta floral dip or tumorigenesis assay (van Attikum et al., 2003). The discrepancy between these previous reports might be attributable to differences in the transformation mechanisms between germline and somatic cells, and could also depend on the physiological condition of the plant material used for the experiments. Thus, evidence for the involvement of the NHEJ pathway in Agrobacterium-mediated T-DNA integration into the plant genome remains inconclusive.
ThechoiceofNHEJandHRpathwaysforDSBrepairineukaryotes depends on the cell type, cell cycle stage and complexity of the DNA end (Shrivastavet al.,2008;Heyeret al.,2010;Symington&Gautier, 2011). In addition, disruption of the NHEJ pathway has been shown to lead to an increased frequency of HR in fungi (Ninomiya et al., 2004;Villalbaet al.,2008),mammals(Lianget al.,1996Pierceet al., 2001;Allen et al., 2003) and plants (Gallego et al., 2003), suggesting that the NHEJ pathway competes with the HR pathway for DSB repair. However, the AtKu80 mutant did not show an enhanced frequency of intrachromosomal HR (Gallego et al., 2003).
We have developed a stable and efficient Agrobacteriummediated transformation system for rice . Most recently, we have further constructed a sequential monitoring system for stable transformation by visualizing cells in which T-DNA is successfully integrated into the rice genome using a nondestructive and highly sensitive visible marker in rice (H. Saika et al., unpublished). In this report, we investigated the involvement of the NHEJ pathway in an Agrobacterium-mediated stable transformation system, and the choice between NHEJ and HR for DSB repair in rice plants. The evidence presented here indicates that the NHEJ pathway participates in Agrobacterium-mediated stable transformation in rice, and that this pathway possibly competes with the HR pathway for DSB repair. Alternatively, an increase in the HR pathway might result from genome instability and the up-regulation of HR genes derived from the suppression of the NHEJ pathway in rice somatic cells.
The LU-UC recombination substrate was constructed as follows. The 120-bp artificial synthesized fragment containing a multicloning site and two I-SceI recognition sites (AscI-SacI-I-SceI-AvrII-AatII-I-SceI-Csp45I-PacI) was cloned into the AscI/PacI site of pZAmI, which is a derivative of pPZP201 (Hajdukiewicz et al., 1994), with a herbicide bispyribac sodium resistance cassette (rice acetolactate synthase (ALS) promoter 3 kb + mutant ALS gene (W548L/S627I) (Shimizu et al., 20052005) + ALS terminator 0.7 kb). The full length and 5′ region of the luciferase (LUC) cDNA were amplified by PCR using pSP-luc+NF vector (Promega) as a template and the following primer pairs: full length (LUC-F 5′-tctagaatggtcaccgacgccaaaaacat-3′ (XbaI site in italics) and LUC-R 5′-gagctcttacacggcgatctttccgcc-3′ (SacI site in italics)) and 5′ region (LUC-F and LUC-R 891 5′-gagctcggcgaagaaggagaatagggtt-3′ (SacI site in italics)). The full length and 5′ region of LUC cDNA were cloned into the XbaI/SacI site between the maize ubiquitin1 promoter + first intron and the transcription terminator of the Arabidopsis ribulose-bisphosphate carboxylase small subunit (rbcS) gene in pENTR (Life Technologies), yielding the vectors pPubi: LUC and pPubi:LU, respectively. The fragment containing the maize ubiquitin1 promoter + first intron and 5′ region of LUC cDNA in the pPubi:LUC vector was digested with AscI/SacI and cloned into pZAmI, yielding pZAmLU. A 3.2-kb fragment containing the promoter sequence of rice elongation factor 1, mutant codA gene (D314A) (Mahan et al., 2004) and transcription terminator of the rice glycerol 3-phosphate dehydrogenase gene was amplified from pE_Pef:codAm:Tg3p vector (K. Osakabe et al., unpublished) using primers 5′-cctaggaagctttataatcccg tgcg-3′ (AvrII site in italics) and 5′-gacgtcgaattccaagcaccaccgcga-3′ (AatII site in italics). The amplified fragment was integrated into the AvrII/AatII site of pZAmLU, yielding pZAmLUcodA. A fragment containing the 3′ region of LUC cDNA (1485 bp) and the transcription terminator of the rbcS gene in pPubi:LUC was digested with Csp45I (internal restriction site)/PacI and integrated into pZAmLUcodA, yielding pZAmLUcodAUC. For construction of the I-SceI expression plasmid, the yeast I-SceI gene was cloned into the SpeI/SacI site between the 2 9 CaMV35 promoter + Ω and the transcription terminator of rice heat shock protein 17.3 of pZDB (K. Osakabe et al., unpublished).

Agrobacterium-mediated stable transformation assay
The Agrobacterium-mediated stable transformation assay was performed according to the method established by H. Saika et al. (unpublished). The reporter constructs used (Fig. 2a, p35Smini: Eluc vector) contain the gfbsd2 gene (Ochiai-Fukuda et al., 2006) encoding a fusion gene of gfp and bsd (herbicide Blasticidin S resistance) under the control of the constitutive rice elongation factor 1 promoter (Pef) and emerald luciferase (Eluc) gene under the control of the CaMV35S minimal promoter (m), which can express the Eluc gene only after stable integration into an enhancerlike sequence (H. Saika et al., unpublished). Rice calli from control plants transformed with pANDA empty vector and KD-OsKu70, KD-OsKu80 and KD-OsLig4 plants were induced and grown on N6D medium containing 50 mg l À1 hygromycin for 7 d at 33°C, and were infected with Agrobacterium harboring p35Smini:Eluc vector. After 3 d of co-cultivation with Agrobacterium at 22-25°C under continuous darkness on 2N6-AS medium, calli were washed and cultured on N6D medium containing 50 mg l À1 hygromycin (Wako Pure Chemical Industries, Osaka, Japan), 10 mg l À1 blasticidin S (Wako Pure Chemical Industries) and 25 mg l À1 meropenem (Wako Pure Chemical Industries). Green fluorescent protein (GFP) fluorescence and LUC luminescence were then detected and quantified sequentially in rice callus transformed with the p35Smini:Eluc reporter construct.

Observation of GFP fluorescence
GFP fluorescence images were taken using a Molecular Imager FX (Bio-Rad, USA) with an excitation wavelength of 488 nm and an emission wavelength of 530 nm.

Observation of LUC luminescence
Rice calli were treated with 0.2 mM Beetle Luciferin potassium salt (Promega). LUC luminescence images were taken using a highresolution photon counting camera (C2400-700 VIM camera, Hamamatsu Photonics, Hamamatsu, Japan) with a 15 min exposure time, and processed using Aquacosmos (Hamamatsu Photonics).

Genomic DNA extraction and Southern blot analysis
Genomic DNA was extracted from rice calli transformed with the LU-UC recombination substrate using Nucleon PhytoPure (GE Healthcare, Little Chalfont, Buckingham, UK) according to the manufacturer's protocol, digested with HindIII and fractionated in a 1.0% agarose gel. Southern blot analysis was performed according to a standard protocol. Specific DNA probes for the codA gene were synthesized with a PCR digoxigenin (DIG) probe synthesis kit (Roche Diagnostics) according to the manufacturer's protocol, using the primers 5′-ggcccatggtgtcgaataacgctttacaaac-3′ and 5′cccgagctctcaacgtttgtaatcgatggct-3′.

Genotoxic stress treatment
Seedlings of control, KD-Ku70, KD-Ku80 and KD-Lig4 plants were grown on Murashige and Skoog medium for 7 d at 30°C under continuous light in a growth chamber, transferred to water containing 5 lM bleomycin and incubated for 12 h in the growth chamber. The aerial parts of plants were collected and immediately frozen in liquid N 2 and stored at À80°C for quantitative RT-PCR analysis and microarray analysis.

Microarray analysis
Total RNA was isolated using an RNeasy Plant Mini Kit (Qiagen), labeled using a Quick-Amp Labeling Kit (Agilent Technologies, Santa Clara, CA, USA) and hybridized to a rice 4 9 44 K custom oligoDNA microarray (Agilent Technologies) according to the manufacturer's instructions. Hybridization microarray slides were scanned with an Agilent Microarray Scanner (Agilent Technologies). The images generated were analyzed using Feature Extraction software (Agilent Technologies), applying standard normalization procedures.

Suppression of NHEJ-related gene expression causes inhibition of Agrobacterium-mediated stable transformation in rice calli
We searched for genes with high sequence homology to AtKu80 and AtLig4 in the rice genome annotation project database (http:// rice.plantbiology.msu.edu/), and identified the rice homologs of Ku80 (LOC_Os03g63920.1) and Lig4 (LOC_Os04g51700.1). We generated rice plants transformed with RNAi constructs using the 3′ untranslated region (UTR) of OsKu70, OsKu80 or OsLig4 mRNA. Among a number of T 2 lines generated, two independent lines of each transgenic rice plant with a single T-DNA insertion and reduced expression of the targeted gene were selected and named KD-OsKu70 (-1 and -4), KD-OsKu80 (-1 and -4) and KD-OsLig4 (-8 and -14), respectively ( Fig. 1a-c). T 0 plants were selfpollinated and T 1 seeds were used for further analysis.
There were no significant differences between wild-type and KD-OsKu70, KD-OsKu80 and KD-OsLig4 plants in the frequency of callus induction, callus proliferation, callus shape, vegetative growth or fertility (data not shown).
In previous reports of Agrobacterium-mediated transformation in rice calli, two patterns of expression derived from the foreign gene on the T-DNA have been detected: transient expression of the foreign gene from the unintegrated double-stranded T-DNA, and stable expression following integration of the foreign gene into the plant genome . To distinguish between transient and stable expression more clearly and sensitively, we constructed a new T-DNA vector, p35Smini:Eluc, containing the GFP gene under the control of a constitutive promoter and the emerald LUC gene under the control of a minimal promoter (Fig. 2a;H. Saika et al.,unpublished). This system allows us to evaluate Agrobacterium-mediated transformation events by visualizing stable transgene expression in rice. The GFP signal observed in rice calli at the early transformation stage (3-4 d after Agrobacterium infection) was derived from transient expression of T-DNA that was not integrated into the rice genome. The emerald LUC luminescence seen in rice calli at later transformation stages (7-9 d after Agrobacterium infection) was derived from stable expression of T-DNA that had integrated successfully into the rice genome and was expressed because of a nearby enhancer.
Furthermore, a mutant line (3A-01546; Hong et al., 2010;Fig. 3a), containing a T-DNA insertion in the OsKu70 gene (KO-Ku70), was acquired from the Rice T-DNA Insertion Sequence Database (http://www.postech.ac.kr/life/pfg/risd/) and self-fertilized to obtain a pure homozygous line. However, progeny derived from homozygous T-DNA insertion lines for the OsKu70 gene (KO-Ku70 À/À ) were not obtained. It has been reported that plants homozygous for the OsKu70 T-DNA insertion mutation display a sterile phenotype and severely retarded vegetative organ growth (Hong et al., 2010). Thus, mature seeds of the heterozygous KO-Ku70 mutant (KO-Ku70 +/À ) were analyzed for segregation of the mutant allele and transcript levels of the OsKu70 gene (Fig. 3d). The seeds were then inoculated on N6D medium to induce calli. As 7-d-old calli of KO-Ku70 À/À showed severely reduced growth compared with calli of wild-type or knockdown (KD) plants (Fig. 3b), we could not use these calli for the Agrobacteriummediated stable transformation assay. Seven-day-old rice calli of the control and KO-Ku70 +/À plants were infected with Agrobacterium harboring p35S mini:Eluc vector, and GFP fluorescence and LUC luminescence were analyzed after 3 d of co-culture and Relative mRNA levels Relative mRNA levels Relative mRNA levels elimination of the bacteria. Consistent with results from KD-Ku70, no significant differences in the transient expression of GFP between control and KO-Ku70 +/À plants were observed (Fig. 3c). However, LUC luminescence was suppressed slightly in KO-Ku70 +/À calli when compared with control calli at 9 d after infection (Fig. 3c,e). These results suggest that the suppression of NHEJ-related gene expression causes an inhibition of Agrobacterium-mediated stable transformation, but not of import of T-DNA into the nuclei of infected cells.

Suppression of NHEJ-related gene expression leads to enhanced frequency of HR
To investigate whether suppression of the NHEJ pathway causes an increase in HR frequency, we used an HR assay system that permits recombination events to be visualized as luminescence from a reconstituted recombination substrate locus. This recombination substrate consists of two partially duplicated fragments of the LUC gene (LU-UC) interrupted by a cytosine deaminase (codA) expression cassette and two recognition sites for meganuclease, I-SceI (Fig. 4a). These fragments of the LUC gene share a duplicated 720-bp region in tandem arrangement. We generated rice plants transformed with recombination substrate (LU-UC), and evaluated the number of integration loci of the transgene. Southern blot analysis with a CodA probe showed that almost all transformants contained a single copy or two copies of the T-DNA (data not shown). Transgenic lines with a single copy (LU-UC lines 8) and two copies (LU-UC lines 9) of the LU-UC recombination substrate were selected (Fig. 4b), and were transformed with Agrobacterium to introduce pANDA empty vector (as a control) or the KD constructs of OsKu70 or OsLig4 (Fig. 4c).

New Phytologist
Mature seeds of T 1 control, KD-Ku70 and KD-Lig4 transgenic (T 2 LU-UC transgenic) plants were grown and selected on N6D medium containing hygromycin and bispyribac without 5-fluorocytosine, which is converted to toxic 5-fluorouracil by codA. The negative selection by codA results in the preferential growth of calli in which the codA expression cassette has been removed from the recombination substrate locus. After 4 wk of selection, LUC luminescence derived from reconstituted recombination substrate was analyzed on transgenic calli. KD-Ku70 and KD-Lig4 transgenic calli containing two copies of the recombination substrate showed a two-to three-fold increase in the frequency of HR compared with control calli (Fig. 4d,e). However, LUC luminescence was not detected on transgenic calli containing a single copy of LU-UC (data not shown).
Next, we investigated the effect of suppression of NHEJ on the frequency of DSB-inducible HR using transient I-SceI expression. To introduce DSBs at the two I-SceI sites flanking the codA expression cassette of LU-UC recombination substrates by inducing transient I-SceI expression, 4-wk-old control, KD-OsKu70 and KD-OsLig4 transgenic calli containing LU-UC recombination substrates were infected with Agrobacterium harboring the I-SceI expression vector (Supporting Information Fig. S1a). LUC luminescence derived from DSB-inducible reconstituted recombination substrate was analyzed 5 d after infection, at which point T-DNA is largely unintegrated into the rice genome . We found no difference in LUC luminescence between control, KD-OsKu70 and KD-OsLig4 calli (Fig. S1b).

Effect of suppression of NHEJ-related gene expression on induction of DSB-inducible genes
The enhancement of HR frequency observed in KD-OsKu70 and KD-OsLig4 calli could be explained by competition for DNA repair between HR and NHEJ at the sites of DSBs. However, it could also be explained by the induction of HR-related gene expression. We analyzed the effect of suppression of NHEJ-related gene expression on the transcript levels of OsRad51A2, which plays a crucial role in HR repair of DSBs (Rajanikant et al., 2008), and OsBRCA1, which colocalizes with Rad51 at DSB sites and activates HR via Rad51 New Phytologist (Lafarge & Montané, 2003). Seven-day-old control, KD-OsKu70, KD-OsKu80 and KD-OsLig4 rice plants were either treated with 5 lM bleomycin, which generates DSBs and DNA single-strand breaks, or left untreated for 12 h. No significant differences were observed in the levels of OsRad51A2 and OsBRCA1 transcripts between controls and transgenic plants without bleomycin treatment (Fig. 5a,b), whereas, on treatment with 5 lM bleomycin for 6 and 12 h, transcripts of both genes increased markedly in transgenic relative to control plants (Fig. 5a,b). Suppression of OsKu70 expression showed the highest transcript levels of HR-related genes at 12 h after treatment with bleomycin. We further analyzed the transcriptional responses to bleomycin treatment of OsKu70/80 and OsLig4 suppression plants using an Agilent rice 44K microarray. The expression levels of HR-related genes are shown in Table 1. Transcript levels of not only OsBRCA1 and OsRad51A2, but also OsDMC1A and OsDMC1B, the meiotic Rad51 paralog (Metkar et al., 2004), were induced in KD-OsKu70 and KD-OsLig4 rice plants relative to control plants under treatment with bleomycin (Table 1). In addition, under DNA-damaging conditions, the expression of OsRad54, one of the key proteins necessary for HR and DNA repair (Osakabe et al., 2006), and OsTOP3 (putative topoisomerase 3) (Singh et al., 2010) was increased in KD-OsKu70 plants. Putative OsBARD1a BRCA1-associated RING domain protein (Singh et al., 2010) was up-regulated in KD-OsLig4 plants, but not in control plants, whereas no differences in the expression of HR-related genes other than OsBRCA1 were noted between KD-OsKu80 and control plants.

Discussion
Here, we have demonstrated that the suppression of NHEJ-related gene expression causes decreased Agrobacterium-mediated stable transformation of the rice genome, especially in KD-OsLig4 (Figs 2,  3). Lig4 proteins interact with XRCC4 and catalyze the ligation step of DSBs in the NHEJ pathway (Mladenov & Iliakis, 2011;Symington & Gautier, 2011). The most generally accepted mechanisms of T-DNA integration into the plant genome are the strand-invasion model and the DSB repair model depending on the NHEJ pathway. The results presented here suggest that NHEJ is the major pathway of T-DNA integration into the rice genome, at least when rice scutellum-derived calli are used for transformation, and that the OsLig4 protein may play an important role in this process.
The decreased stable transformation observed in KD-OsKu70 and KD-OsKu80 plants and OsKu70 +/À plants is in accordance with a previous report (Li et al., 2005), in which a decreased frequency of T-DNA integration was observed in atku80 knockout mutants of Arabidopsis by root tumorigenesis assay. In addition, mRNA levels of OsKu70 and OsRad51A2, which are well-known genes associated with DNA damage, are higher in calli than in leaves, roots and anthers . Yadav et al. (2009) have revealed an enrichment of DNA repair gene expression in Arabidopsis shoot apical meristem stem cells. Furthermore, consistent with previous observations (Hong et al., 2010), we found that knockout of the OsKu70 gene resulted in a growth defect of calli (Fig. 3b), but had no effect on the growth of intact plants. Summarizing these results, DSBs arise constantly in rice callus as a result of active cell division. These DSBs are repaired mainly by the NHEJ pathway. DSBs are toxic for rice callus if not repaired correctly by the NHEJ pathway. T-DNA has been shown in several cases to be inserted into DSB sites (Salomon & Puchta, 1998). Thus, T-DNA is most likely to be integrated into the rice genome via NHEJ using endogenously induced DSBs.
There was no correlation between the transcript levels of Ku80 and LUC luminescence derived from T-DNA stable transformation in KD-OsKu80 rice calli (Figs 1b, 2d,e). It has been shown that the Ku70/80 protein functions as a heterodimer to repair DSBs, with the two constituents stabilizing each other (Smider et al., 1994). Accordingly, these results may be attributed to a decrease in protein levels of both OsKu80 and OsKu70 caused by the down-regulation of the expression of OsKu80 mRNA in KD-Ku80 rice calli.
We also detected decreased Agrobacterium-mediated stable transformation in KD-OsLig4 rice callus. However, it has also been reported that the absence of AtLig4 does not affect the frequency of T-DNA integration in an in vitro root tumorigenesis assay (van Attikum et al., 2003). Although we cannot rule out the possibility that AtLig4 and OsLig4 act differently in each species in the process of Agrobacterium-mediated stable transformation, differences in the assay system itself could lead to different results. In the root tumorigenesis assay, the efficiency of T-DNA integration was evaluated by tumor formation in roots that were grown via cell division and cell expansion for 4-5 wk after Agrobacterium infection; many environmental factors are involved in this process. Our Agrobacterium-mediated stable transformation assay system allows transient and stable expression to be visualized using a sensitive reporter gene, and thus stable transformation events can be detected rapidly (within 7-10 d after Agrobacterium infection).
It has been shown recently in Arabidopsis that the epidermal cells of the root tip are enlarged and show increased nuclear DNA content in response to DSBs derived from treatment with the genotoxic agent zeocin, indicating that DSBs induce the onset of the endocycle (Adachi et al., 2011). In other words, DNA-damaged cells in Arabidopsis could select the repair of DNA damage, the induction of cell death or, alternatively, the induction of the endocycle to avoid cell death. We speculated that a fraction of cells with integrated T-DNA could enter the endocycle as a result of DSB signals, as DSBs have been reported to be major integration sites of T-DNA (Salomon & Puchta, 1998). In contrast with Arabidopsis, the endocycle has never been reported in rice except in the endosperm. We have recently confirmed that this is true, even after genotoxic stress treatment inducing DSBs (Endo et al., 2012). According to this concept, the frequency of T-DNA integration could be underestimated in Arabidopsis, as AtKu70/80 or AtLig4 mutants accumulate DSBs as a result of insufficient NHEJ repair.
Decreased T-DNA integration has also been reported in atku80 and atlig4 mutants when analyzed using an in planta floral dip transformation assay (Friesner & Britt, 2003), whereas no differences were detected between atku80 mutants and wild-type plants (Gallego et al., 2003). Furthermore, Ziemienowicz et al. (2008)

Research
New Phytologist function as a ligase for T-DNA in vitro, suggesting that T-DNA integration may be catalyzed by Lig1 via the strand-invasion model in the case of in planta floral dip transformation. The presence of a micro-homology sequence in the T-DNA integration site identified via in planta floral dip transformation (Brunaud et al., 2002) also supports this idea. Micro-homology-dependent integration of T-DNA into the rice genome was also reported when scutellumderived calli were used for transformation (Sha et al., 2004). To better understand the mechanism of T-DNA integration in plants, the effect of the suppression of Arabidopsis or rice Lig1 on the efficiency of T-DNA integration via in planta floral dip or callus transformation should be investigated.
We observed a sterile phenotype and delayed callus proliferation in OsKu70 mutants. By contrast, Arabidopsis ku70, ku80 and lig4 mutant plants are viable and exhibit a fertile phenotype (Bundock et al., 2002;van Attikum et al., 2003;Friesner & Britt, 2003). In addition, no significant difference in callus formation and callus regeneration from roots was observed between Arabidopsis ku80 mutant and wild-type plants (Li et al., 2005). These differences suggest that the defect in the NHEJ pathway that depends on Ku70, Ku80 and Lig4 is more severely affected in rice than in Arabidopsis. Furthermore, it has been reported that cells with DNA damage enter into the endocycle to prevent their proliferation and cell death in Arabidopsis (Adachi et al., 2011). Consequently, mutations of

New Phytologist
Arabidopsis ku70, ku80 and lig4 may have little direct effect on viability and productivity. Of all rice tissues, HR events occur with highest frequency in callus tissue under normal conditions . Consistent with this observation, we detected LUC luminescence derived from reconstituted recombination substrate in both control and transgenic calli; HR activity in KD-Ku70 and KD-Lig4 transgenic calli was two-to three-fold higher than in control calli (Fig. 4d,e). Thus, the frequency of HR seemed to be enhanced by suppression of the NHEJ pathway in rice plants, whereas there was no difference in the frequency of DSB-inducible HR between control and transgenic calli (Fig. S1b). In this study, to eliminate the effect of the efficiency of T-DNA integration on I-SceI expression levels, we analyzed LUC luminescence by monitoring DSB-inducible HR at 5 d after infection with Agrobacterium harboring an I-SceI expression vector. It was assumed that LUC luminescence derived from DSB-inducible HR might not be detected correctly because I-SceI expression levels derived from free T-DNA may be very low. Alternatively, it is possible that no differences were observed in the frequency of DSB-inducible HR between control and transgenic calli because DSBs induced by I-SceI expression at the two I-SceI sites flanking the codA expression cassette of LU-UC recombination substrates were repaired via the single-strand annealing (SSA) pathway. It has been reported that DSBs flanking directly repeated sequences are repaired mainly by the SSA pathway using homologous sequences (Siebert & Puchta, 2002). In future, evaluation of the frequency of DSB-inducible HR should be performed with KD-Ku70/80 or KD-Lig4 harboring a recombination substrate and a stably integrated I-SceI expression cassette driven by a chemically inducible system. To assess HR frequency in KD-Ku70/80 or KD-Lig4 in detail, the establishment of an HR assay system in rice is also required to detect the reconstruction of the recombination substrate only via HR, for example, intrachromosomal HR reporter with indirect repeats (Gherbi et al., 2001) or interchromosomal HR reporter (Molinier et al., 2004).
A large number of genes, including HR-related genes, are upregulated in response to ionizing radiation (Nagata et al., 2005;Culligan et al., 2006;Ricaud et al., 2007). In addition, previous studies of Arabidopsis ku80 mutants have shown that several genes, including DSB repair genes, display transcriptional induction (West et al., 2004). Consistent with these findings, we demonstrate here that the transcript levels of OsRad51A2 and OsBRCA1 are higher in KD-Ku70, KD-Ku80 and KD-Lig4 plants than in control plants under conditions of DNA damage (Fig. 5).
Considering the transcriptional induction of HR-related genes in KD-Ku70, KD-Ku80 and KD-Lig4, suppression of the NHEJ pathway causes the accumulation of DNA damage under stressful conditions, such as active cell division and treatment with DNAdamaging agents, and the HR pathway is then activated and DNA damage repaired as a result of the induction of HR-related gene expression. Recently, it has been shown that Ku proteins interact with a number of proteins and function not only in the NHEJ pathway for DSB repair, but also in transcriptional regulation and signal transduction in mammals (Bertinato et al., 2003;Brenkman et al., 2010;Zhang et al., 2011;Fell & Schild-Poulter, 2012). Wang et al. (2010) have identified Arabidopsis ovate family protein 1 (AtOFP1) as a novel AtKu70-interacting protein using yeast two-hybrid screening and pull-down assay. In this report, suppression of OsKu70 expression has a major effect on HRrelated genes, such as OsRad51A2 and BRCA1 (Fig. 5, Table 1), therefore suggesting that OsKu70 may also be involved in the signaling pathway of the DSB response.
In conclusion, our results clearly show the involvement of the NHEJ pathway in the Agrobacterium-mediated stable transformation in rice and the presence of the competitive and complementary relationship between the NHEJ and HR pathways for DSB repair in rice. Although it takes a relatively long time to set up an assay system in rice relative to Arabidopsis because of its longer life-cycle, a well-established evaluation system in rice could be applied to other important crops in Gramineae.
In contrast with random T-DNA integration, sequence-specific integration of T-DNA into the endogenous homologous sequence gene targeting (GT)occurs through the HR pathway (Smithies et al., 1985;Paszkowski et al., 1988;Müller, 1999;Iida & Terada, 2004). As a result, it has been reported that GT events of endogenous genes by HR occur on the order of 0.01-0.1% compared with random integration in higher plants (Paszkowski et al., 1988;Offringa et al., 1990;Puchta et al., 1996). Our data

Research
New Phytologist indicate that suppression of the NHEJ pathway is expected to cause an increase in the occurrence of sequence-specific integration by HR and inhibition of random integration by NHEJ, resulting in a synergistic effect that will improve GT experiments. In fact, it has been reported that deletion of the Ku or Lig4 gene can increase the frequency of GT in several organisms, such as fungi (de Boer et al., 2010;Ushimaru et al., 2010;Nakazawa et al., 2011), bacteria (Zhang et al., 2012) , birds (Iiizumi et al., 2008), mammals (Bertolini et al., 2009;Fattah et al., 2008;Iiizumi et al., 2008) and plants (Tanaka et al., 2010;Jia et al., 2012). We are progressing towards the establishment of a high-efficiency GT procedure using KD rice plants targeting NHEJ-related genes. The log 10 ratios of HR-related genes in control, KD-Ku70, KD-Ku80 and KD-Lig4 plants with (12 h) or without (0 h) 5 lM bleomycin treatment compared with those in control plants without bleomycin treatment are presented. Bold characters represent genes whose expression is significantly induced (log 10 ratio > 0.3 and P < 0.01).

Supporting Information
Additional supporting information may be found in the online version of this article. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.