Arabidopsis CHROMOSOME TRANSMISSION FIDELITY 7 (AtCTF7/ECO1) is required for DNA repair, mitosis and meiosis

The proper transmission of DNA in dividing cells is crucial for the survival of eukaryotic organisms. During cell division, faithful segregation of replicated chromosomes requires their tight attachment, known as sister chromatid cohesion, until anaphase. Sister chromatid cohesion is established during S-phase in a process requiring an acetyltransferase that in yeast is known as Establishment of cohesion 1 (Eco1). Inactivation of Eco1 typically disrupts chromosome segregation and homologous recombination-dependent DNA repair in dividing cells, ultimately resulting in lethality. We report here the isolation and detailed characterization of two homozygous T-DNA insertion mutants for the Arabidopsis thaliana Eco1 homolog, CHROMOSOME TRANSMISSION FIDELITY 7/ESTABLISHMENT OF COHESION 1 (CTF7/ECO1), called ctf7-1 and ctf7-2. Mutants exhibited dwarfism, poor anther development and sterility. Analysis of somatic tissues by flow cytometry, scanning electron microscopy and quantitative real-time PCR identified defects in DNA repair and cell division, including an increase in the area of leaf epidermal cells, an increase in DNA content and the upregulation of genes involved in DNA repair including BRCA1 and PARP2. No significant change was observed in the expression of genes that influence entry into the endocycle. Analysis of meiocytes identified changes in chromosome morphology and defective segregation; the abundance of chromosomal-bound cohesion subunits was also reduced. Transcript levels for several meiotic genes, including the recombinase genes DMC1 and RAD51C and the S-phase licensing factor CDC45 were elevated in mutant anthers. Taken together our results demonstrate that Arabidopsis CTF7/ECO1 plays important roles in the preservation of genome integrity and meiosis.


INTRODUCTION
Precise cell division with transmission of genetic information is a key process controlling growth and development in all eukaryotic organisms (Peters and Bhaskara, 2009). Chromosomes need to be properly replicated and condensed then attached to the spindle fibers in order to be distributed evenly among daughter cells (D ıaz-Mart ınez and Clarke, 2009). The cohesin complex is critically important for these processes. Compliance with this program ensures the timely growth and development of unicellular organisms such as yeast, and the proper formation of tissues and organs in multicellular organisms such as animals and plants (Skibbens, 2010;Wu et al., 2010).
Proteins from the STRUCTURAL MAINTENANCE OF CHROMOSOMES (SMC) family and associated non-SMC factors are essential for the regulation of higher-order chromosomal structure in eukaryotes (Schubert, 2009). The SMC complexes are mostly composed of canonical SMC proteins, which contain a globular ATPase head, and kleisin subunits that connect the two heads to form a ring that topologically embraces nascent chromatid fibers (Peters et al., 2008;Watanabe, 2012). This topological entrapment allows each chromatid to be used as a template for homology-dependent DNA repair during DNA synthesis in the Sphase (Murakami et al., 2010), and binds sister chromatids to each other for proper spindle orientation and segregation during the G 2 /M phase (Beckouet et al., 2010). Chromosome cohesion involves cohesin complexes that include SMC3, SMC1, SCC3 and one of several different kleisins (Schubert, 2009). Cohesins are also important for the repair of DNA lesions caused by exposure to radiation or chemical agents post-replication, a task performed by cohesin complexes that include SMC5, SMC6A/B and the d-kleisins NSE4A/B (Watanabe et al., 2009;Callegari et al., 2010;Kim et al., 2010a). Cohesins are also required for the exchange of nonsister chromatid segments between homologous chromosomes during meiosis (Kim et al., 2010b).
The assembly of cohesin rings around chromosomes has been extensively studied in yeast (Saccharomyces cerevisiae) and humans. In yeast the key regulator of cohesin establishment is an acetyltransferase known as Establishment of cohesion 1 (Eco1). Acetylation of key lysine residues K112, K113 and K84, K210 of SMC1 and SMC3, respectively, by Eco1 stabilizes the ring and facilitates binding to the a-kleisin, Sister chromatid cohesion 1 (Scc1), until anaphase (Beckouet et al., 2010). Then two sequential events occur, first the enzyme separase cleaves Scc1 to open the ring, followed by deacetylation of SMC1 and SMC3 by Histone lysine deacetylase 1 (Hos1) to facilitate recycling of SMC1 and SMC3 (Rivera and Losada, 2010).
In humans, point mutations in the Eco1 homolog ESCO2 lead to congenital abnormalities exemplified by Roberts syndrome (RBS). In RBS patients only 10-20% of cells show abnormal mitosis; however, all cells are hypersensitive to DNA-damaging agents and show premature centromere separation (Vega et al., 2005;van der Lelij et al., 2009;Whelan et al., 2012b). Recent studies on CTF7 in yeast and mouse Eco1 and Esco2 mutants suggest that mutations in the C-terminal acetyltransferase domain have little effect on S-phase cohesion and chromosome segregation, but increase the sensitivity to DNA-damaging agents, thereby phenocopying RBS cells (Lu et al., 2010;Whelan et al., 2012a). Mutations in the N-terminus mostly lead to defects in cohesion, and often to loss of chromosomes during mitosis (Lu et al., 2010;Whelan et al., 2012a). Moreover, in yeast it has been observed that haploid-strains defective in Eco1 are not able to sporulate, while diploid heterozygous strains are normal (Rudra and Skibbens, 2012). In mice heterozygous conditional-Esco2 mutants show no phenotype, while homozygous embryos die at the eight-cell stage (Whelan et al., 2012b). These findings have led to the suggestion that Eco1 activity is dosage-dependent (Rudra and Skibbens, 2012;Whelan et al., 2012b), a claim made earlier by Skibbens (2010) who suggested that a decrease in yeast Eco1 activity may compromise DNA repair first and chromatid pairing second.
Nothing was known about the biological function of the Arabidopsis thaliana Eco1 homolog until recently, when Jiang et al. (2010) showed that Arabidopsis CTF7/ECO1 encodes an acetyltransferase with the ability to rescue yeast eco1 deletion mutants. Arabidopsis CTF7/ECO1 encodes a 345 amino acid protein, which contains a conserved N-terminal PIP box required to interact with the replication fork subunit PROLIFERATING CELL NUCLEAR ANTIGEN (PCNA) and a zinc finger domain, important for chromatin binding. At the C-terminus of the protein is the acetyltransferase domain, required to acetylate cohesin factors (Jiang et al., 2010;Higashi et al., 2012;Rudra and Skibbens, 2012). Heterozygous ctf7-1 mutants showed asynchronous female development, while homozygous embryos were found to arrest before or at the globular stage. Here, we report the identification and characterization of homozygous ctf7-1 and ctf7-2 T-DNA insertion mutants and show that CTF7/ECO1 is required to establish sister chromatid cohesion during male meiosis, and to allow proper cell division in vegetative tissues. We also show that CTF7/ECO1 is required for DNA repair and discuss these results in the context of a complex regulatory network.

RESULTS
Homozygous ctf7-1 and ctf7-2 plants are viable but exhibit defects in vegetative and reproductive development It was previously shown that approximately 25% of the seed in siliques of heterozygous ctf7-1 plants (ctf7-1/+) exhibit defects in zygote and embryo development including arrest by the early globular stage (Jiang et al., 2010), suggesting that inactivation of Arabidopsis CTF7 results in embryo lethality. During the analysis of segregating populations of progeny of the ctf7-1/+ (SALK_059500) and ctf7-2/+ (SAIL_1214G06) T-DNA lines (Jiang et al., 2010;Figure 1a), we identified several slow-growing dwarf plants (Figure 1b). At about the same time analysis of the subcellular localization of AtCTF7 in Arabidopsis protoplasts indicated that AtCTF7 localizes to the nucleus (Figure 1c), a result that is in agreement with the hypothesis that Arabidopsis CTF7 is an essential nuclear protein required for growth (Jiang et al., 2010). Indeed, genotyping indicated that dwarf plants were homozygous for the T-DNA insert and segregated at a very low frequency (below 4%), a rate that deviated significantly from a 1:3 Mendelian ratio (Figure 1d). The phenotypes of ctf7-1 and ctf7-2 homozygous mutants are indistinguishable, therefore it was decided to focus efforts on the characterization of ctf7-1, which had been successfully complemented using the full genomic sequence of CTF7/ECO1, plus its native promoter (Jiang et al., 2010). This line is referred to as the ctf7-1 complementation line (Com) in this paper. Quantitative real-time PCR (QPCR) showed that plants homozygous for either ctf7-1 (ctf7-1 plants) or ctf7-2 (ctf7-2 plants) contain <20% of wild-type (WT) CTF7 mRNA levels corresponding to exon 5, located downstream of the corresponding T-DNA inserts (Figure 1e). Amplification of CTF7 cDNA with primers spanning the T-DNA insert was not possible in ctf7-1 plants, suggesting that the ctf7-1 mutation gives rise to truncated versions of the transcript, a (a) (d) Less than 4% of the progeny of selfpollinated heterozygous ctf7-1 and ctf7-2 (ctf7-1/ +, ctf7-2/+) plants were homozygous (ctf7-1, ctf7-2). Segregation of progeny for both T-DNA alleles was non-Mendelian (not 1:3), and the respective P-values for the chi square test (with two degrees of freedom) were highly significant, suggesting serious developmental defects.
(e) Quantitative real-time PCR experiments with primers complementary to exons 3, 4 and 5, which flank the T-DNA inserts in ctf7-1 and ctf7-2 indicated a significant reduction in CTF7/ ECO1 expression downstream of the respective T-DNA insert, while the ctf7-1 complementation line (Com) showed values similar to wild type (WT). Results are shown as means AE SD (n = 3) from three biological samples. Asterisks represent significant differences (*P < 0.5, **P < 0.01; Student's t-test) relative to WT. (f) After anthesis, free pollen grains were easily identifiable on the surface of WT stigma, but not on ctf7-1. Distribution of the petals, sepals and anthers was also affected in ctf7-1 flowers.
(h) Compared with the WT, siliques of heterozygous ctf7-1 and ctf7-2 plants contained a higher percentage of defective/aborted seeds. Nevertheless, ctf7-1 and ctf7-2 plants only produced immature siliques without normal seeds. situation previously observed in mutations for human ESCO2 (Vega et al., 2005). Developmental defects are widespread in ctf7-1 and ctf7-2 seedlings, including defects in the distribution of leaves on the stem (e.g. phyllotaxy). In the WT all leaves are arranged in a spiral, while in ctf7-1 and ctf7-2 this arrangement shows modifications, including additional basal leaves and clusters of modified, small rosette leaves ( Figure S1a). Defects in root development were also observed, including reduction in the length of the elongation zone and root swelling ( Figure S1b).
The morphology of ctf7-1 flowers was also abnormal, with enlarged papillae, defective anthers and a very limited amount of pollen ( Figure 1f). Crossing experiments showed that mutant stigmas allowed the germination of WT pollen, but no mature siliques were recovered after pollination, suggesting that both male and female gametophytes are defective in ctf7-1 (Figure 1g). No defects were observed in the germination of pollen from the ctf7-1 complementation line or ctf7-1/+ plants (Figure 1g), suggesting that ctf7-1 heterozygous plants are not defective in the development of mature microspores.
Anther development in Arabidopsis starts with the formation of bilateral primordia that features locules, and vascular tissue (stages 1-4). Once the primordia are established, archesporial cells within the anther give rise to the endothecium, middle layer, tapetum and pollen mother cells (stage 5). At this stage ctf7-1 anthers already appear smaller than in the WT (Figure 2a). Alterations in microsporogenesis were first observed during stages 6 and 7 when normally meiosis is completed and tetrads are formed. Irregular division of pollen mother cells (stage 6) and the formation of irregular tetrads (stage 7) were observed in ctf7-1 anthers. Anthers in ctf7-1 remained smaller than in WT as development continued, and development of the connective tissue appeared compromised as well. After the release of microspores at stage 8, the anthers of WT and ctf7-1 continued to develop in a similar fashion; the microspores became vacuolated and the tapetum degenerated (stages 9 and 10). However, at stage 11, when microspores normally enter into mitosis and the stomium and septum degrade, ctf7-1 microspores appeared shrunken and there was no noticeable degradation of the stomium and septum. By stage 12, WT anthers contained fully developed tricellular pollen and the anthers were undergoing anthesis. In contrast, most ctf7-1 pollen appeared dead and no opening of the stomium and septum was observed.
Finally, severe defects in the ultrastructure of ctf7-1 and ctf7-2 pollen were observed, including the relative absence of an electron-dense cytoplasm, vegetative nuclei and sperm cells (Figure 2b). These results demonstrate that microsporogenesis and anthesis are defective in ctf7 homozygous mutants.
ctf7-1 male meiocytes display defects in chromosome condensation, sister chromatid cohesion and the distribution of cohesin proteins Based on observations in other systems (Baudrimont et al., 2011;Rudra and Skibbens, 2012), we expected that inactivation of CTF7 should block the establishment of sister chromatid cohesion and result in meiotic defects. In order to investigate this possibility we analyzed meiotic chromosome spreads in ctf7-1 plants. Alterations were observed from the earliest stages examined, with the first noticeable difference between ctf7-1 and WT plants being the presence of fewer meiocytes overall throughout meiosis. It is not clear if this is due to the fact that the plants are smaller and less healthy, or if some ctf7-1 microsporocytes arrest and abort prior to meiosis. Some variability was also observed in the phenotypes at different stages of meiosis, with some meiocytes appearing relatively normal; however, most meiocytes shared common phenotypes, which are described below. During pre-leptotene, WT chromosomes showed faint labeling of chromosomes with chromocenters that stain deeply (Ross et al., 1997; Figure S3a for WT), while in ctf7-1 plants no chromosome axes were recognizable and the chromocenters stained very faintly ( Figure 3a). Similar to WT, chromosome condensation was observed during leptotene in ctf7-1, although at somewhat reduced levels (Figures 3b and S3b for WT). During zygotene, chromosome alignment was reduced in ctf7-1 (Figures 3c and S3c for WT) and ultimately a mixture of unpaired and unevenly paired chromosomes were observed at pachytene (Figures 3d S3d for WT). During diplotene, a decondensed mass of chromatin was typically observed in ctf7-1; no individual separated chromosomes were visible (Figures 3e and S3e for WT). In contrast to the five bivalents observed in WT at diakinesis (Mercier et al., 2005; Figure S3f for WT) a mixture of uncondensed chromatin, unpaired chromosomes and possibly some bivalents were observed in ctf7-1 (Figure 3f). Beginning at diplotene and diakinesis and continuing through meiosis II, ctf7-1 meiocytes typically appeared less condensed than their WT counterparts. A relatively small number of cells (10%) also appeared to contain extra chromosomes (Figures 3e-g and S3e-g for WT), although it is not clear whether this was caused by defects in chromosome segregation or DNA replication.
During metaphase I a mass of DNA, possibly chromosomes, congregated at the equatorial plane in ctf7-1; (a) Sections of developing anthers revealed that ctf7-1 had multiple anther developmental defects, including: a reduction in anther size (all stages), unsynchronized release of free tetrads (stage 7), release of irregular microspores (stage 8), lack of pollen mitosis II (stage 10) and failure of the septum (Sn) and stomium (St) to degenerate during anthesis (stage 12). (b) Left, transmission electron microscopy showed that homozygous ctf7-1 and ctf7-2 pollen grains were smaller, lacked cytoplasm, a vegetative nucleus and sperm cells. Right, staining with DNA dye 4′,6-diamidino-2-phenylindole (bottom) confirmed that ctf7-1 and ctf7-2 pollen lack identifiable vegetative nuclei and sperm cells, unlike wild-type (WT) pollen where vegetative nuclei and sperm cells were clearly observed. T, tapetum; PMC, pollen mother cell; StR, stomium regium, E, epidermis; En, endothecium; ML, middle layer; V, vascular tissue; C, connective tissue; MC, meiotic cell; Td, tetrad; Msp, microspore; Sm, septum; PG, pollen grain; Fb, fiber bands, St, stomium. Scale bars: in (a), 25 lm for stages 5-8 and 50 lm for stages 9-12; in (b) left 50 lm and right 2 lm. To further investigate meiosis, and in particular sister chromatid cohesion and chromosome pairing, in situ hybridization was conducted with the 180 bp centromere (CEN) repeat as a probe (Armstrong et al., 2001) (Data S1). In leptotene, approximately 10 unpaired and welldispersed CEN signals were observed in WT (11 AE 2, n = 10), while in ctf7-1 irregular CEN signals were typically observed ( Figure S4a). By zygotene, the number of CEN signals was reduced to approximately 5 AE 1 (n = 10) in WT. In ctf7-1 roughly double the number of CEN foci (12 AE 1, n = 7) were observed, consistent with a defect in synapsis ( Figure S4b). Increased CEN foci (8 AE 1, n = 4), were also observed during pachytene in ctf7-1; in addition the signals were more dispersed and less well defined than in WT, often appearing not as discrete foci but rather as long extended segments. Five easily identifiable CEN foci were observed in WT cells at diakinesis; however, 20 or more CEN signals were typically found in ctf7-1 ( Figure S4d).
During late metaphase I/early anaphase I, five pairs of CEN signals (9.1 AE 1, n = 10) were observed in WT, while over 20 CEN signals appeared randomly dispersed around the nucleus in ctf7-1 ( Figure S4e). By late anaphase I masses of DNA, some without CEN signals, some with two signals, and clusters of CEN signals were observed in ctf7-1 ( Figure S4f). Later in development, ctf7-1 microspores containing varying numbers of CEN signals, some with 20 or more signals, could be observed ( Figure S4g). 'Extra' CEN signals were also observed in interphase nuclei of some anther somatic cells of ctf7-1 plants ( Figure S4h). In contrast WT microspores and interphase anther cells always contained five and 10 CEN foci, respectively.
We next investigated the loading and distribution of the SYN1 and SMC3 cohesin proteins on chromosomes of ctf7-1 meiocytes. As has been demonstrated previously, SYN1 and SMC3 display similar distribution patterns on WT meiotic chromosomes (Yang et al., 2011a,b;Figures 4 and S5 for WT). Diffuse nuclear labeling is observed at interphase. Beginning at early leptotene and extending into zygotene both proteins decorated the developing WT chromosomal axes. During late zygotene and pachytene the proteins lined the synapsed chromosomes. As meiosis progressed from diplotene to diakinesis the chromosomeassociated cohesin signals became progressively weaker and more diffuse.
Similar to our observations in the chromosome spreading and CEN fluorescence in situ hybridization experiments, approximately half of the ctf7-1 meiocytes displayed relatively normal SYN1 and SMC3 labeling patterns. However, in most cells the labeling for both proteins was very weak and irregular (Figures 4 and S5). Very little SMC3 and SYN1 signal was present in the nucleoplasm at interphase (Figures 4a and S5a for WT). Diffuse labeling of the chromatin was first observed during leptotene (Figures 4b and S5b for WT), with some labeling of threadlike structures in some cells during early zygotene (Figures 4c and S5c for WT). However, in most cases the SYN1 and SMC3 signals were diffuse and became progressively weaker as meiosis progressed into pachytene, diplotene and diakinesis stages (Figures 4d-g and S5d-g for WT).
Several key genes for DNA repair and cell cycle progression are upregulated in ctf7-1 plants In addition to its critical role in the establishment of sister chromatid cohesion during DNA replication, CTF7/ECO1 may be involved in DNA repair and cell cycle progression (Lu et al., 2010;Lyons and Morgan, 2011). We therefore investigated the effect of the ctf7-1 mutation on the expression of a number of genes required for DNA repair and cell cycle progression. A pathway analysis using AraNet (Lee et al., 2010) suggested a strong functional linkage for these genes (P = 1.054 9 10 À82 ). The relative expression levels of the selected genes were measured in triplicate through QPCR. As shown in Figure 5, large and statistically significant increases (greater than four-fold) in transcript levels were observed for several genes in ctf7-1 plants, including: ATM (a kinase), BRCA1 (a ubiquitin ligase), PARP2 (a polymerase), RAD51 (a gene involved in homology-dependent DNA repair), CYCB1;1 (a G 2 /M checkpoint gene) and TOPOII-a (a topoisomerase) (Xie and Lam, 1994;Garcia et al., 2003;Preuss and Britt, 2003;Takahashi et al., 2010;Thomson and Guerra-Rebollo, 2010). Smaller increases (approximately two-fold) were observed for SMC5, SMC6B and SRS2 (a helicase) (Ira et al., 2003;Watanabe et al., 2009), while no significant change was observed for ATR (a kinase involved in single-strand break repair) (Yoshiyama et al., 2009), the M-phase checkpoint genes MAD2 and NQK1  and the gene CDKA1, which regulates the transition from mitosis to endocycle (Dissmeyer et al., 2007). Expression of other topoisomerases was not detected in either WT or ctf7-1 samples, including TOPOI-a, and TOPOI-b (Takahashi et al., 2002).
Given the developmental defects observed in ctf7-1 meiocytes the transcript level of highly expressed meiotic genes (Yang et al., 2011b) was measured by QPCR as well. As shown in Figure 6, statistically significant increases in transcript levels were observed for ATM and ATR, BRCA2B, RAD51C, DMC1, SMC1 and SMC3, and CDC45, a gene that codes for an S-phase licensing factor and is required for meiosis (Stevens et al., 2004). No significant increase was observed for the mitotic checkpoint gene BUB1.3. Expression of several other genes was not detected, including  . SYN1 exhibits an altered distribution pattern in homozygous ctf7-1 male meiocytes. Merged images of 4′,6-diamidino-2-phenylindole stained chromosomes (red) and SYN1 (green) are shown. (a) Interphase. SYN1 is distributed throughout the nuclei of wild type (WT) male meiocytes; in contrast little or no SYN1 was observed within the nuclei of ctf7-1. (b) Leptotene. SYN1 decorated WT chromosome filaments as they started to condense, while in ctf7-1 only labeling is weak and diffuse. (c), (d) Early and late zygotene. SYN1 decorates WT chromosome axes as chromosomes start to synapse. (e) Pachytene. SYN1 lined the synapsed WT bivalents, but in ctf7-1 the labeling remained weak and diffuse. (f) Diplotene. SYN1 continues to label WT bivalents, while in ctf7-1 both proteins appear as scattered, punctuate foci. (g) Diakinesis. Chromosome-associated SYN1 begins to weaken in WT and is absent in ctf7-1 nuclei. Scale bars = 10 lm. The ability of ctf7-1 plants to repair DNA double breaks was tested using the comet assay (Kozak et al., 2009) (Data S2), which has been employed in Arabidopsis mutants with defects in either chromosome cohesion or DNA repair proteins . Seven-day-old WT and ctf7-1 seedlings were exposed to a bleomycin solution (50 lg ml À1 ) for 1 h and the percentage of DNA present in nuclei tails after recovery times of 0, 30 and 60 min was used to estimate the level of double-strand breaks remaining in each sample ( Figure S6). In WT most double-strand breaks were repaired after 30 min and only approximately 29% remained after 1 h. In contrast approximately 79% of all double-strand breaks remained unrepaired after 1 h in ctf7-1 plants, indicating that CTF7/ECO1 is required for DNA repair in Arabidopsis.
ctf7-1 and ctf7-2 plants are defective in mitotic cell cycle progression The extreme dwarf phenotype of ctf7-1 and ctf7-2 plants suggested that, similar to the situation in yeast (Moldovan et al., 2006), Arabidopsis CTF7 plays an important role in cell cycle progression. To determine if ctf7-1 and ctf7-2 cells show cell cycle arrest, we analyzed the morphology and density of pavement and stomata cells in the first true leaves of 7-day-old seedlings by cryo-scanning electron microscopy (cryo-SEM). In WT, pavement and stomata cells were small (Figure 7a), developed at a density of 1070 and 590 cells mm À2 (Figure 7b) and covered an approximate area of 892 lm 2 (for pavement cells) and 88.0 lm 2 (for stomata) (Figure 7c). Similar values were observed for the ctf7-1 complementation line (Com) (Figure 7a,b). In cells of ctf7-1 and ctf7-2 there was a significant reduction in the density of both pavement cells (580 cells mm À2 ), and stomata (250 cells mm À2 ) (Figure 7b), accompanied by a significant increase in the average area of pavement cells (1300 lm 2 per cell) ( Figure 7c); however, no change was observed in the area of stomatal cells under cryo-SEM.
Mitotic cell cycle progression was further examined by analyzing the DNA content of leaf cells. Intact nuclei were isolated from the first leaves of 7-day-old plants, followed by flow cytometry. In WT and complementation line samples, approximately 71-73% of all nuclei had a haploid DNA content of 2, '2C', which reflects normal entry into mitosis and cell division (Figure 7d). Approximately 16-17% of WT and complementation line nuclei showed a 4C value, which represents those cells that have completed DNA replication but have not entered the G 2 /M phase. In contrast, 55-56% of ctf7-1 and ctf7-2 nuclei showed a 4C value (Figure 7d), suggesting a defect in the ability of cells to advance into M-phase after DNA replication.
ctf7-1 and ctf7-2 plants are defective in mitotic chromosome segregation In order to obtain direct evidence of the role of Arabidopsis CTF7/ECO1 in mitotic chromosome segregation, root tips were excised from 2-week-old seedlings corresponding to WT, Com, ctf7-1 and ctf7-2. As observed in Figure 8, it was possible to observe cell division from interphase to telophase. In both WT and Com, 10 chromocenters were observed during interphase (Figure 8a,e), which corresponds to a diploid number of chromosomes, while in Complementary DNAs from 1-week-old wild-type (WT) and ctf7-1 seedlings were generated and used in quantitative real-time PCR. Transcript levels of ATM, PARP2, BRCA1, RAD51, SMC5, TOPOII-a and CYCB1;1 are elevated in ctf7-1. Data are shown as means AE SD (n = 3) from three biological samples. Asterisks represent significant differences (*P < 0.5, **P < 0.01, ***P < 0.001; Student's t-test) relative to WT.  (Figure 8l,p). In fact chromosome segregation in homozygous ctf7-1 and ctf7-2 mitotic cells is statistically different from WT (P < 0.001, Student's t-test) and shows defective chromosome segregation (Figure 8q). Taken together these results suggest that CTF7/ECO1 plays a critical role in the regulation of chromosome segregation during mitotic cell division.

DISCUSSION
Accumulating evidence suggests that assembly of cohesion rings around nascent chromatids allows efficient DNA repair during mitosis by guaranteeing the availability of an intact template . In yeast and mammals, a direct link exists between acetylation of cohesin rings and cohesion (Peters and Bhaskara, 2009). Acetylation of cohesin rings occurs at conserved lysine residues in SMC proteins during S phase and to a lesser extent during the G 2 /M phase (Onn et al., 2009). Further, establishment of cohesion requires functional interactions with subunits of the replication fork (Sherwood et al., 2010). In yeast this process is regulated by the acetyltransferase Eco1, which targets both SMC subunits and the PCNA subunit of the replication fork (Sherwood et al., 2010).
In contrast to budding yeast, very little is known about the role of Arabidopsis CTF7/ECO1. In this study we characterized the roles of Arabidopsis CTF7/ECO1 in both mitosis and meiosis by characterizing plants homozygous for a T-DNA insertion in AtCTF7/ECO1 (ctf7-1 and ctf7-2). Given that CTF7 is a single-copy gene in Arabidopsis and complete inactivation of CTF7 is typically lethal, our ability to obtain plants homozygous for the T-DNA insertion was unexpected. The presence in ctf7-1 and ctf7-2 plants of relatively normal levels of transcript having the potential to encode the N-terminus of the protein, and reduced but detectable levels of RNA encoding the C-terminus (which comprises the acetyltransferase domain), raises the possibility that truncated forms of the protein may be produced in some cells, which allows some nuclear division. While further experiments are required to determine if partially functional forms of the protein are formed, the possibility that low levels of CTF7 activity are present in at least some cells is consistent with the fact that we observe severe defects in development and DNA repair first, and somewhat milder defects in chromosome cohesion during nuclear division. A dosage effect has been observed for cohesins in other organisms (Rudra and Skibbens, 2012), and this also appears to be the case for Arabidopsis. Plants heterozygous for ctf7-1 and ctf7-2 showed defects during female gametophyte development, which requires three rounds of mitosis ( Figure S1c), but no developmental defects were found in microspores, which require only two rounds of mitosis ( Figure S2b) (Chang et al., 2011). Also no obvious defects in vegetative growth were detectable in heterozygous plants, which is consistent with the hypothesis that a small reduction in acetyltransferase activity does not have a significant impact on either cohesion or cell division (Whelan et al., 2012a). However, ctf7-1 homozygous plants had severe defects in male gametophytic development, including defects in anther development (Figure 2), defects in SMC3 and SYN1 binding to meiotic chromosomes (Figures 4 and S5) and dramatically reduced cohesion at centromeres ( Figure S4). Alterations in cohesion distribution can affect synaptonemal complex formation and impair RAD51-mediated formation of chiasmata (Longhese et al., 2009), resulting in defective pollen Figure 8. Staining with 4′,6-diamidino-2-phenylindole reveals defective chromosome segregation in homozygous ctf7-1 and ctf7-2 mitotic root tip cells.
Interphase cells in both wild type (WT) (a) and the ctf7-1 complementation line (Com, e) show approximately 10 chromocenters corresponding to a diploid number of chromosomes. In cells of ctf7-1 (i) and ctf7-2 (m) the chromocenters often appeared decondensed and in some instances in excess of 10 chromocenters are observed. In metaphase, WT (b) and Com (f) cells typically show 10 condensed chromosomes; however, chromosomes in cells of ctf7-1 (j) and ctf7-2 (n) appear less condensed and irregularly shaped. In anaphase, WT (c) and Com (g) cells show even segregation of chromosomes, and individual chromosomes can be recognized, while chromosomes in cells of ctf7-1 (k) and ctf7-2 (o) appear intertwined and stretched; condensation is also defective. In telophase, WT (d) and Com (h) cell chromosomes segregate into two well-condensed masses of equal size, but in ctf7-1 (l) and ct7-2 (p) chromosome bridges persist and the chromosomes often recondense into a single unevenly shaped mass of DNA (l, p). Chromosome segregation in homozygous ctf7-1 and ctf7-2 mitotic cells is statistically different from WT and shows defective chromosome segregation (q). Data are shown as means AE SD (n = 100) from 40 biological samples. Asterisks represent significant differences (***P < 0.001; Student's t-test) relative to WT. Scale bars = 10 lm. formation and sterility, as we observe in ctf7-1 (Figures 2  and 1h,i).
Homozygous ctf7-1 and ctf7-2 plants are dwarf (Figure 1c), suffer from cell cycle arrest (Figure 7d) and are unable to segregate chromosomes properly during mitotic cell division (Figure 8). Moreover, leaf cells from ctf7-1 plants are unable to efficiently repair DNA double-strand breaks ( Figure S6) and contain elevated levels of transcripts of genes required for double-strand break repair ( Figure 5), such as the effector kinase gene Ataxia Telangietasia Mutated (ATM), polymerase PARP2, recombination mediator RAD51, cohesin subunit SMC5 and checkpoint regulators such as BRCA1 and CYCB1;1 that trigger arrest at the G 2 /M phase (Garcia et al., 2003;Schubert et al., 2009;Yoshiyama et al., 2009). The anthers of ctf7-1 plants contained elevated transcript levels for a number of genes, including the ATR kinase, the meiotic recombinase gene DMC1 and its associated factor RAD51C, cohesin subunits SMC1 and SMC3, BRCA2B and the Minichromosome Complex Maintenance (MCM) subunit gene CDC45, which is required for proper meiotic entry into S phase (Stevens et al., 2004). ATR is required for the loading of meiotic recombinases (Kurzbauer et al., 2012). Therefore elevated transcript levels of ATR, RAD51C and DMC1 suggest that ctf7-1 plants may experience DNA recombination stress. Interestingly, changes in the activity of CDC45, RAD51C and BRCA2 have all been linked to chromosomal fragmentation due to pre-replicative stress and the failure to perform homologous recombination (Stevens et al., 2004;Abe et al., 2005;Kurzbauer et al., 2012), a phenotype similar to what we observe in ctf7-1 meiocytes.
Taken together, these results suggest that CTF7/ECO1 is extremely important for mitotic cell cycle progression, meiosis, mitosis and DNA repair in Arabidopsis. The dramatic developmental defects observed in ctf7-1 plants are not observed in humans containing ESCO mutations, which induce defects in cohesion and DNA damage repair but not in chromosome segregation (van der Lelij et al., 2009). A role for ESCO2 in human meiosis has remained mostly hypothetical (Hogarth et al., 2011). Likewise, in Drosophila melanogaster, Eco1 homologs are required for checkpoint activation and chromatid cohesion (Williams et al., 2003), but a role in meiosis has not been demonstrated (Pimenta-Marques et al., 2008).
Interestingly, somatic cells of ctf7-1 plants contained transcript levels for topoisomerase II-a, which is the only eukaryotic enzyme known to mediate topological responses to a loss of cohesion by controlling the amount of catenation in replicated DNA (Ryu et al., 2010). Upregulation of this gene raises the possibility that loss of CTF7-dependent cohesion may induce an increase in the activity of topoisomerase II-a as a compensatory process that restores a certain degree of cell division and allows Arabidopsis ctf7-1 mutants to remain viable, as has been reported in human HeLa cells defective in cohesion (D ıaz-Mart ınez et al., 2006). In fact, a positive reinforcing relationship between DNA catenation and cohesin-mediated cohesion has been proposed in yeast (Farcas et al., 2011).
Finally, some of the observed developmental defects may be the result of impaired acetylation of proteins other than SMC1 and SMC3. The Arabidopsis genome harbors several SMC-like genes that have been show to affect organ development, gene expression and chromatin compaction (Schatlowski et al., 2010;Moissiard et al., 2012) and whose relationship with CTF7/ECO1 has not yet been determined. For instance the condensin subunit SMC2A/CAP-E1 has been shown to play an important role in meiosis (Siddiqui et al., 2003), while condensin subunits HEB1 and -2 directly influence root and root hair development under boron stress (Sakamoto et al., 2011).
In conclusion we have shown that Arabidopsis CTF7/ ECO1 plays critical roles in meiosis, mitosis and DNA repair and is essential for microsporogenesis and anther development. While further work is required to dissect the regulatory mechanism of Arabidopsis CTF7/ECO1 the availability of plants homozygous for CTF7 mutations provides a valuable system for studying this important enzyme.

Plant material and growth
The WT (ecotype Columbia) and T-DNA insertional lines, SALK_059500 (ctf7-1), and SAIL_1214G06 (ctf7-2) were obtained from the Arabidopsis Biological Resource Center (Columbus, OH, USA; http://abrc.osu.edu). Seeds were surface sterilized in 30% sodium hypochlorite and germinated on half-strength Murashige and Skoog medium without sucrose, followed by stratification at 4°C for 96 h in the dark. Seedlings were grown at 21°C under a 16-h photoperiod and 60% relative humidity for approximately 5 days after the emergence of the radicle. Seedlings were then genotyped following emergence of either cotyledons or first true leaves and used for the experiments. The remaining seeds were germinated on soil to further characterize the phenotype of the mutants.

Molecular analysis of CTF7/ECO1
The Arabidopsis At4g31400 locus was first selected for study during a screen for potential gametophytic mutants. The T-DNA lines SALK_059500 (ctf7-1) and SAIL_1214G06 (ctf7-2) were selected in this study, since previous work showed that both independent lines display a similar phenotype and line ctf7-1 could be successfully complemented by transformation with the Gateway binary vector pFGC5941 carrying the full genomic sequence of At4g31400 (Jiang et al., 2010). Genomic DNA was extracted from segregant ctf7-1 and ctf7-2 T 3 seedlings. Plants were genotyped with specific primer pairs for their corresponding T-DNA inserts and WT locus. At least five independent complementation lines were analyzed using specific primers for the ctf7-1 insertion, for the pFGC5941 vector and for exons 3, 4 and 5 of CTF7/ECO1. Resistance to Basta, and plant fertility were also analyzed. The complete list of primers used can be found in Table S1. Morphological characterization of ctf7-1 and ctf7-2 Images of seed set were recorded after dissection of at least 30 siliques from 7-week-old plants under a Carl Zeiss stereo Lumar V12 fluorescence stereomicroscope (Carl Zeiss, http://microscopy. zeiss.com/microscopy) connected to a Carl Zeiss AxioCam MRc5 CCD unit. Whole anther morphology was analyzed by staining with Alexander staining for 24 h at 50°C (Alexander, 1969) or by sectioning followed by staining with toluidine blue. Pollen viability was analyzed by releasing mature pollen grains into fluorescein diacetate (FDA) (Sigma-Aldrich, http://www.sigmaaldrich.com) solution for 15 min in the dark followed by observation under an Olympus BX51 epifluorescence microscope coupled to an Olympus DP70 CCD unit (Olympus, http://www.olympus-global.com/en/ corc/company/lifescience). Analysis of in vivo pollen tube growth was performed as described by Szumlanski and Nielsen (2009) and emasculated pistils from the WT and homozygous ctf7-1 (ctf7-1) were cross-pollinated and collected after 24 h. Tissue was obtained from 5-week-old plants, with three biological replicates and at least 100 pollen grains per replicate were used to estimate pollen viability.

Analysis of subcellular localization of CTF7/ECO1
Protoplasts prepared from the leaves of 4-week old Arabidopsis plants were co-transformed with 35S:CTF7/ECO1:GFP and the nuclear marker construct 35S:ERF4:mRFP. Transformed protoplasts were observed by two-photon laser confocal microscopy and analyzed with the ZEISS LSM IMAGE BROWSER version 3.5.

Analysis of ultrastructure
The ultrastructure of mature microspores was analyzed by transmission electron microscopy on a Philips CM 100 unit (Philips Research, http://www.research.philips.com/). Samples were prepared by fixation on 2.5% gluteraldehyde and 4% paraformaldehyde, in 0.1 M sodium phosphate buffer, pH 7.0 at room temperature (25°C) for 4 h. After three 20-min buffer rinses, samples were post-fixed on osmium oxide 1%. Samples were dehydrated in an acetone series, embedded in Spurr's resin, and sectioned on a Leica Ultracut E microtome (Leica Microsystems, http://www.leica-microsystems.com/). Ultra-thin sections (70-90 nm) were stained with 6% uranyl acetate and 0.4% lead citrate. Sections were observed at 80 kV. Preparation of anther sections followed the same procedure; however, semi-thin samples (1 lm) were stained instead with 10% toluidine blue in 1% Na-Borex for 1-2 min, and observed directly under a light microscope. Developmental stages were assigned according to Sanders et al. (1999). For the analysis of leaf cell structure, the first true leaves of 1week-old seedlings were excised, frozen in liquid nitrogen and then transferred to a sample preparation chamber set at À160°C. After 5 min the temperature was raised to À85°C and the sample was sublimed for 15 min. After coating with platinum at À130°C the sample was transferred to a PP2000T Cryo-SEM System chamber (Quorum Technologies, http://www.quorumtechnologies.com/ ) and observed at À160°C on a FEI Quanta 200 scanning electron microscope (FEI, http://www.fei.com/) set at 20 kV. Results were obtained from three different biological samples containing at least three technical repeats.

Chromosome spreads
In order to analyze male meiosis, anthers were isolated and digested as described (Yang et al., 2011a,b). Following digestion, cells were transferred onto poly-L-lysine slides (Sigma-Aldrich), and covered with a cover slip. The slides were frozen on dry ice and the cover slips quickly removed. The dried slides were stained with 1.5 mg ml À1 DAPI (Vector Laboratories, http://www.vectorlabs.com/). In order to analyze mitosis, root tips from seedlings grown on agar were excised and placed within 200 ll PCR tubes. Fixation and digestion was performed within the tubes, and digestion was extended to 2 h.

Flow cytometry
Cells from fresh leaves were isolated and stained with the CyStain PI Absolute P Kit (Partec, http://www.partec.com/). Finally the nuclear suspension was run through a MoFlo XDP Laser Cell Sorter (Beckman Coulter, https://www.beckmancoulter.com/) and results were analyzed with SUMMIT V5.0 software (Beckman Coulter) from at least three different biological samples.

Quantitative real-time PCR
Total RNA was extracted with the RNAeasy Plant Minikit (Qiagen, http://www.qiagen.com/) from WT and ctf7-1 seedlings (1 week old), and from meiotic anthers collected from mature WT and ctf7-1 plants. Total mRNA from meiotic anthers was further amplified with the MessageAmp II aRNA Amplification Kit (Ambion, http://zt. invitrogen.com/). First-stranded cDNA was prepared from total RNA with the Moloney murine leukemia virus reverse transcriptase system (Promega, http://www.promega.com/) according to the manufacturer's instructions. For quantitative PCR, a Power SYBR Green I Master Mix (Applied Biosystems, http://www. appliedbiosystems.com/) was used with 150-200 nM primers, 20 ng ll À1 cDNA and 50 ll of reverse transcriptase reaction product. Reactions were run and analyzed on the AB 7500 Real Time PCR System (Applied Biosystems). Melting curve analyses and negative controls were used to exclude primer-dimer artifacts and low specificity in the amplification. Quantitative reactions were done in triplicate and averaged. Primers specific for the 3′ end of transcripts were either designed on PRIMER EXPRESS version 3.0 (Applied Biosystems) or adapted from relevant references (Preuss and Britt, 2003;Czechowkski et al., 2005;Takahashi et al., 2010). The complete list of primers used can be found in Table S1.

SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article. Figure S1. Phenotypes associated with the ctf7-1 and ctf7-2 homozygous mutants. Figure S2. Homozygous ctf7-1 and ctf7-2 pollen is not viable. Figure S3. Chromosome pairing and segregation patterns in wildtype meiocytes. Figure S4. Chromosomes in homozygous ctf7-1 male meiocytes exhibit defects in cohesion and chromosome fragmentation. Figure S5. Cohesin subunit SMC3 exhibits an altered distribution pattern in homozygous ctf7-1 male meiocytes. Figure S6. Homozygous ctf7-1 seedlings exhibit reduced DNA repair efficiency. Table S1. Primers used in this paper.
Data S1. Fluorescence in situ hybridization in meiotic chromosomes. Data S2. DNA comet assay.