Replication factor C1 (RFC1) is required for double-strand break repair during meiotic homologous recombination in Arabidopsis

Authors


For correspondence (e-mail jzhao@whu.edu.cn).

Summary

Replication factor C1 (RFC1), which is conserved in eukaryotes, is involved in DNA replication and checkpoint control. However, a RFC1 product participating in DNA repair at meiosis has not been reported in Arabidopsis. Here, we report functional characterization of AtRFC1 through analysis of the rfc1–2 mutant. The rfc1–2 mutant displayed normal vegetative growth but showed silique sterility because the male gametophyte was arrested at the uninucleus microspore stage and the female at the functional megaspore stage. Expression of AtRFC1 was concentrated in the reproductive organ primordia, meiocytes and developing gametes. Chromosome spreads showed that pairing and synapsis were normal, and the chromosomes were broken when desynapsis began at late prophase I, and chromosome fragments remained in the subsequent stages. For this reason, homologous chromosomes and sister chromatids segregated unequally, leading to pollen sterility. Immunolocalization revealed that the AtRFC1 protein localized to the chromosomes during zygotene and pachytene in wild-type but were absent in the spo11–1 mutant. The chromosome fragmentation of rfc1–2 was suppressed by spo11–1, indicating that AtRFC1 acted downstream of AtSPO11-1. The similar chromosome behavior of rad51 rfc1–2 and rad51 suggests that AtRFC1 may act with AtRAD51 in the same pathway. In summary, AtRFC1 is required for DNA double-strand break repair during meiotic homologous recombination of Arabidopsis.

Introduction

Meiosis is a specialized type of cell division that is conserved among most eukaryotes and is indispensable for formation of viable offspring. It consists of two successive divisions after a single round of DNA replication, giving rise to four haploid daughter cells from a single diploid parent cell. In plants, this enables the generation of microspores and megaspores, which then differentiate into pollen and the embryo sac, respectively. Unlike mitosis, meiosis I requires pairing and synapsis to ensure segregation of the homologous chromosomes, whereas meiosis II results in separation of sister chromatids (Zickler and Kleckner, 1999; Harrison et al., 2010).

In meiosis, homologous chromosome pairing and synapsis facilitate recombination, which is a prerequisite for repair of the SPO11-induced DNA double strand-breaks (DSBs) that initiate meiosis and is the source of heterosis in the next generation (Jones et al., 2003). DSBs in meiosis are repaired by homologous recombination. The SPO11 protein is a member of the type II topoisomerase family (Bergerat et al., 1997). In Arabidopsis thaliana, AtSPO11–1 and AtSPO11–2 are necessary for DSB formation (Hartung and Puchta, 2000, 2001; Stacey et al., 2006). After SPO11 removal, DSB ends are resected to generate 3′ single-strand overhangs. These become coated with DMC1 and RAD51 to enable invasion of the duplex homologous DNA sequence, which leads to D–loop formation. Subsequent DNA synthesis is primed from its 3′ end, leading to replicative extension of the D–loop in the 5′→3′ direction. This DNA synthesis creates the double Holliday junction recombination intermediate that is then resolved to form crossover and non-crossover (Shinohara et al., 1992; Dresser et al., 1997; Neale and Keeney, 2006). Previous data have confirmed that this reaction also requires the clamp-loader components proliferating cell nuclear antigen (PCNA) and replication factor C (RFC) (Wang et al., 2004; Li et al., 2009).

DNA synthesis after Rad51-mediated DNA strand invasion is a crucial step during recombination. Eukaryotic DNA replication is a highly regulated process that requires many proteins and factors to ensure this program is tightly controlled (Mendez and Stillman, 2003; Sclafani and Holzen, 2007). Moreover, in yeast, some DNA replication proteins have been found to have additional roles in checkpoint regulation and/or DNA repair. For instance, DNA polymerases δ and ε, the single-strand DNA-binding replication protein A, the clamp protein PCNA and the clamp loader RFC are also involved in DNA repair in vivo (Budd and Campbell, 1995; Soustelle et al., 2002; Corrette-Bennett et al., 2004; Stone et al., 2008). A recent study demonstrated that loading of PCNA is specifically required to initiate recombination-associated DNA synthesis in extension of the 3′ end of the invading strand in a D–loop (Li et al., 2009). RFC1 is the large subunit of RFC. All eukaryotic cells have a series of RFC or RFC-like complexes. In yeast and human cells, ELG1, a homolog of RFC1, forms a distinct RFC-like complex that is required for maintaining genome stability (Kanellis et al., 2003). In Saccharomyces cerevisiae, the Ddc1/Rad17/Mec3 complex and RAD24 are DNA damage checkpoint components that share structural and functional similarity with PCNA and RFC1, respectively. RAD24 (RAD17 in humans) acts together with the four small subunits of RFC to load the RAD17/3/1 at DNA damage sites. However, it is not known how conserved this feature is among plants. Previously, deletions of RAD17 and RAD24 in S. cerevisiae were found to delay repair of meiotic DSBs (Shinohara et al., 2003), but RFC and PCNA participation in meiosis in higher plants has not been reported.

Liu et al. (2010) found that the sensitivity of rfc1–1 seedlings to methylmethane sulphonate (MMS) and cisplatin increased compared with wild-type, suggesting that RFC1 is involved in DNA repair in somatic cells. However, the underlying molecular mechanism for how AtRFC1 is involved in DNA repair requires further investigation, especially in sexual cells during meiosis. Previous studies in animals and plants showed that the DSB formation and subsequent repair are required for normal meiosis to proceed. Some components of the DNA replication machinery are also required for meiosis, including DNA replication factors CDC45 and MEI1 in Arabidopsis (Grelon et al., 2003; Stevens et al., 2004). Although there is some evidence concerning the biochemical properties of AtRFC1 and some of its functions in somatic cells, the role of AtRFC1 in meiosis remains unknown.

As the rfc1–2 mutant produces fewer seeds than wild-type plants (Liu et al., 2010), we investigated the cause of sterility in the mutant plant. In the present study, we used the rfc1–2 mutant to characterize the action of AtRFC1 within sexual reproduction, especially in microspore mother cell meiosis. We found that the sterility of the rfc1–2 mutant was due to chromosome fragmentation arising from a meiosis defect. Genetic analysis revealed that AtRFC1 was required for repair of meiotic DSBs induced by AtSPO11–1, and that AtRFC1 may participate in the AtRAD51-mediated recombination intermediate repair process. Additionally, expression of AtRFC1 in meiotic cells and the protein localization in the chromosome axis are consistent with its function in meiosis.

Results

Characteristics of the rfc12 mutant

Multiple sequence alignment showed that AtRFC1, 2, 3, 4 and 5 contain eight conserved domains (I–VIII), which have orthologs and paralogs in other organisms (Figure S1). In contrast to AtRFC2/3/4/5, AtRFC1 possess additional N–terminal and C–terminal features. In the N–terminal region, a breast cancer tumor C–terminal (BRCT) motif, whose secondary structure is conserved, is located in domain I. This comprises a four-stranded parallel β–sheet flanked by three α–helices (Figure S2), which has previously been identified within numerous proteins involved in DNA repair and cell-cycle checkpoints (Yoshida and Miki, 2004).

To investigate the biological function of AtRFC1 in Arabidopsis, we characterized the AtRFC1 mutant rfc1–2 (T–DNA insertion). The EMS-mutagenized rfc1–1 mutant, which has a defect in transcriptional gene silencing, has been described previously (Liu et al., 2010). Expression analysis of AtRFC1 in rfc1–2 homozygous plants was investigated: no full-length transcript corresponding to wild-type AtRFC1 was detected in the rfc1–2 mutant; a partial transcript was observed upstream of the rfc1-2 T-DNA insertion site (Figure S3).

The plant fertility was greatly impaired in the rfc1–2 mutant (Figure S4a). Unlike wild-type, most ovules in the rfc1–2 mutant aborted at an early stage, leaving white dots that indicate seed abortion (Figure S4b,c). Statistical analyses of silique length and seed number between wild-type and rfc1–2 are shown in Figure S4(e,f). Examination of the rfc1–2 flowers revealed that they are obviously different from wild-type, as they exhibited shriveled anthers and a lack of pollen (Figure S4g,h).

To confirm that the rfc1–2 sterility phenotype was caused by the T–DNA insertion within the AtRFC1 locus, a 7.8 kb genomic fragment encompassing AtRFC1, comprising its promoter region to its 3′ UTR, was transformed into the rfc1–2 homozygote. After resistance screening, 19 transformants were obtained. Ten of these transformants were positive through PCR analysis (Figure S5), and subsequent examination revealed that fertility was restored in these lines (Figure S4d,i), demonstrating that the sterility phenotype is due to loss of the full AtRFC1 transcript.

Arrest of male and female gametophyte development in the rfc1–2 mutant

To determine the cause of silique sterility, we performed reciprocal crosses to examine genetic transmission between rfc1–2 and wild-type gametophytes. When wild-type pollen was used to fertilize rfc1–2 ovaries, 78.9 ± 5.1% of the ovaries remained unfertilized within the siliques (= 20). When wild-type stigmas were pollinated with mutant pollen, a similar result was obtained. Together, these data demonstrated that the transmission efficiency of the rfc1–2 mutation through both male and female lines was severely reduced.

Alexander's staining showed that the vast majority of pollen grains were aborted in mutant anthers compared with wild-type and the complemented mutant (Figure 1a–c). Pollen viability counts confirmed that viable pollen in rfc1–2 mutant showed a significant decrease (Figure 1d), to approximately 13.31% of wild-type (= 48, < 0.01). There were no significant differences of development in the pollen mother cell (PMC) and tetrad stages (Figure 1e,i). A phenotypic defect was first observed at the stage that the wild-type microspore undergoes polarization (Figure 1f), as a high proportion of the rfc1–2 microspores are not able to proceed to polarization (Figure 1j). The abnormal polarization phenotype was found in 2.4% (= 170) and 80.7% (= 152) of microspores from wild-type and rfc1–2 plants, respectively. The defect was maintained when pollen grains enter into the first mitotic division (Figure 1g,k). The proportions of abnormal pollen at this stage were 2.7% (= 214) and 78.6% (= 271) in wild-type and rfc1–2 plants, respectively. In the subsequent developmental stage of the rfc1–2 mutant, the arrested pollen became shrunken and was aborted (83.0%, = 223, compared to 1.5%, = 185, in wild-type, Figure 1h,l).

Figure 1.

Male and female gametophyte development is severely impaired in the rfc1–2 mutant. (a–c) Mature anthers from wild-type, rfc1–2 and the rfc1–2 complemented line visualized by Alexander staining. Scale bars = 100 μm. (d) Numerical analysis of viable pollen between wild-type and rfc1–2 using Alexander staining. = 48 per genotype. The asterisks indicate that the difference between wild-type and rfc1–2 is highly significant, 0.01<P<0.05. (e–l) Semi-thin sections of wild-type and rfc1–2 anthers. (e, i) Tetrad stage. (f, j) Polarized microspore stage. The arrowhead indicates the absence of polarized pollen. (g, k) Bicellular pollen stage. (h, l) Mature pollen stage. The arrow indicates aborted pollen. Scale bars = 10 μm. (m–s) Ovule development in wild-type (m, o, q) and rfc1–2 (n, p, r, s) plants by confocal laser scanning microscopy. CN, chalazal nuclei; DM, Degenerated megaspore; FM, functional megaspore; MMC, megaspore mother cell; MN, micropylar nuclei. (m, n) Megaspore mother cell (MMC) stage. (o, p) Functional megaspore (FM) stage. (q) Four-nucleus stage. (r) Developing rfc1–2 female gametophyte with undivided FM cell at later stage. (s) rfc1–2 female gametophyte lacking an FM cell. Scale bars = 10 μm.

We further examined wild-type and rfc1–2 pollen grains at various stages by staining with 4′,6–diamidino-2–phenylindole (DAPI). Consistent with the previous results, most uninucleate rfc1–2 microspores did not proceed to polarization, and the subsequent stages was clearly aberrant compared with wild-type (Figure S6a–f). Numerical analysis revealed that 90.1 ± 1.8% (= 201) and 30.5 ± 3.2% (= 209) of the microspores displayed polarization in wild-type and the rfc1–2 mutant, respectively. More strikingly, in mature rfc1–2 anthers, only 6.4 ± 3.1% (= 329) tricellular pollen was observed, with the remaining 93.6 ± 1.6% aborted. In contrast, wild-type anthers exhibited 89.5 ± 2.3% (= 250) tricellular pollen (Figure S6g).

Next we analyzed the embryo sac development of the rfc1–2 mutant. Prior to meiosis, there were no distinct differences between wild-type (Figure 1m) and the rfc1–2 mutant (Figure 1n) at the stage of the megaspore mother cell. After meiosis, there were discrepancies in the shape and size of the functional megaspore cell. In the wild-type, this is a teardrop shape (Figure 1o), which was dissimilar to that of the rfc1–2 mutant ovule (Figure 1p). When the wild-type embryo sac reached the four-nucleus stage (Figure 1q), the rfc1–2 mutant remained in the functional megaspore stage (Figure 1r) and was ultimately degraded (Figure 1s). Numerical analysis showed that only 19.8 ± 2.5% (= 10 siliques) of the mutant ovules possessed a functional embryo sac.

Taken together, these observations suggested male and female gametophyte development of rfc1–2 are both impaired, which is the reason for the mutant sterility.

Male meiosis is disrupted in the rfc1–2 mutant

As the earliest appearance of an abnormal nucleus is in the microspore, we attempted to determine whether anything unusual occurred before this stage. Thus, we examined male meiosis with DAPI staining of PMC chromosome spreads. In wild-type male meiocytes, prophase I may be divided into five stages according to distinctive chromosome behavior: leptotene, zygotene, pachytene, diplotene and diakinesis. In leptotene, chromosomes were clearly single and unpaired (Figure 2a). During zygotene, synapsis began between homologous chromosomes (Figure 2b) and was completed in pachytene, with thicker and more heavily stained chromosome spreads (Figure 2c). At the diplotene stage, the homologous chromosomes desynapsed, remaining linked at the chiasmata only, and underwent condensation to form five bivalents at diakinesis (Figure 2d,e). At metaphase I, the five bivalents were arranged in a line and then separated towards opposite poles, leading to formation of dyads containing two pools of five chromosomes (Figure 2f–i). Subsequently, they performed a mitosis-like division to give rise to a tetrad (Figure 2j).

Figure 2.

DAPI staining of meiosis stages from wild-type and rfc1–2 pollen mother cells. Various stages of meiotic chromosome spreads from wild-type (a–j) and rfc1–2 (k–y) are illustrated. (a, k) Leptotene stage. (b, l) Zygotene stage. (c, m) Pachytene stage. (d, n, o, p) Diplotene stage. (e, q, r) Diakinesis stage. (f, s) Metaphase I. (g, t) Anaphase I. (h, u, v) Telophase I. (i) Interphase II. (w) Metaphase II. (x) Anaphase II. (j, y) Telophase II. Scale bar = 10μm.

No apparent alterations were observed in rfc1–2 meiocytes at the early stages of meiosis from leptotene though to pachytene (Figure 2k–m). Chromosome pairing and synapsis appeared to occur normally (Figure 2m). To confirm that synapsis was complete, immunolocalization was performed on spread preparations of PMCs. The localization of a synaptonemal complex protein (ZYP1, Higgins et al., 2005) and an axis-associated protein (ASY1, Armstrong et al., 2002) in the rfc1–2 mutant was indistinguishable from that in wild-type (Figure 3a–d). This indicated that axis formation, chromosome pairing and synapsis were unaffected in the mutant. However, at diplotene in the rfc1–2 mutant, some chromosome breaks were apparent as the homologs desynapsed (Figure 2n). As the chromosomes began to condense during late diplotene, it became clear that this process was aberrant, as numerous brightly stained spots were observed (Figure 2o–q). At diakinesis, irregular bridges and fragmentations were exhibited (Figure 2r). In addition, we did not observe distinct metaphase I bivalents in rfc1–2. Instead, the bulk of the chromatin appeared as a clump at the metaphase plate (Figure 2s). These fragmentations remained during anaphase I and telophase I, and bridges between separated chromosomes were frequently observed (Figure 2t–v). Numerical analysis revealed that a high proportion of fragmentation (76%) at anaphase I occurred in the rfc1–2 mutant (Table 1). As a result, the chromosomes segregated abnormally, leading to dyads with variable DNA contents (Figure 2w). These defects were maintained during meiosis II (Figure 2x), and typical tetrads were not found as some fragments dispersed throughout the cytoplasm (Figure 2y).

Figure 3.

Immunolocalization of ASY1, ZYP1 and γH2AX in wild-type and rfc1–2 meiocytes. (a–d) ASY1 and ZYP1 localization during pachytene in wild-type and the rfc1–2 mutant by fluorescence microscopy. (e–j) Immunolocalization of γH2AX at prophase I. Scale bars = 10 μm. (k) Relative quantification of γH2AX foci for wild-type and the rfc1–2 mutant; n ≥ 10 individuals per stage.

Table 1. Frequencies of anaphase I fragmentation in wild-type and various mutants of Arabidopsis
GenotypeTotal observed number of anaphase I cellsNumber of fragmented cellsPercentage fragmentation (%)
Wild type14500
rfc1–2 806176
spo11–1–3 6500
spo11–1–3 rfc1–2 8589
rad51 5555100
rad51 rfc1–2 7070100

The rfc1–1 allele is a point mutation, and its mutation site is also located in the C–terminus (Liu et al., 2010). Similar pollen development and meiosis defects were observed (Figure S7).

These observations showed that the DNA breaks that emerge during early prophase I cause abnormal meiotic chromosome behavior subsequently. This cytological analysis in the rfc1–2 mutant suggests that the AtRFC1 gene plays a crucial role in chromosome integrity and normal progression during meiosis.

DNA double-strand breaks are maintained in the rfc1–2 mutant

The nucleosomal histone family H2A has several variants. One of these variants, H2AX, can be phosphorylated and thus referred to as γ-H2AX (Rogakou et al., 1998). A previous study showed that phosphorylated H2AX (γH2AX) accumulation was dependent on AtSPO11–1-catalyzed DSB formation during meiosis prophase, and may be regarded as a marker to detect DSBs (Sanchez-Moran et al., 2007). As some DNA breaks occurred during late prophase I in the rfc1–2 mutant, we determined the level of γH2AX and whether the DSB repair pathway was disrupted. To do this, immunolocalization studies were performed in wild-type and the rfc1–2 mutant. γH2AX foci distribution in the wild-type was previously described by Sanchez-Moran et al. (2007). Here, our examination showed >50 foci during leptotene in the wild-type (Figure 3e), which decreased sharply in zygotene (Figure 3f) such that there were < 10 foci at pachytene (Figure 3g). In contrast, the rate of decrease in γH2AX foci in the rfc1–2 mutant slowed down, and more γH2AX foci remained at pachytene (Figure 3h–j), indicating that DSBs remained during these stages. The counting results of γH2AX foci is displayed in Boxplot (Figure 3k). In wild-type, approximately 90% of the red fluorescence in leptotene disappeared on entering into zygotene, and only 1% remained in pachytene. In contrast to wild-type, signals in the rfc1–2 meiocyte decreased slowly, and more than one-third of the red fluorescence seen in leptotene remained during pachytene (Figure 3k). These results demonstrate that DSB repair is impaired in the rfc1–2 mutant, and the aberrant fragmentations present in the rfc1–2 meiocyte may be unrepaired DSBs.

Homologous chromosomes and sister chromatids separate unequally in the rfc1–2 mutant

To investigate the meiotic defects of rfc1–2 in more detail, fluorescence in situ hybridization (FISH) was performed using a centromere-specific probe that localizes to the pericentromeric heterochromatin in Arabidopsis (Ronceret et al., 2009). In wild-type, homologous chromosomes are pulled to opposite directions by spindles in telophase I, leading to the formation of two pools of five signals (Figure 4a). During meiosis II, segregation of sister chromatids produces tetrads, each of which contain five signals (Figure 4c). In the rfc1–2 mutant, the ten signals separated unequally in some cases, e.g. divided into four and six signals, respectively (Figure 4b). After meiosis II (Figure 4d), the four microspores contained varying numbers of centromere signals due to this inequality. We determined the patterns for wild-type and the rfc1–2 mutant during telophase I and telophase II. As Figure 4(e,f) shows, almost all wild-type meiocytes displayed the five/five pattern in telophase I, whereas diverse patterns occurred in the rfc1–2 mutant. Similar results were obtained for telophase II (Figure 4g,h), but the number of meiocytes showing the normal pattern in the rfc1–2 mutant was considerably reduced between telophase I (40%) and telophase II (14.8%). This suggests that separation of not only homologous chromosomes but also sister chromatids was affected. These results indicate that DNA breaks produced in the rfc1–2 mutant cause aberrant alignment and subsequent unequal separation of homologous chromosomes and non-homologous sister chromatids, giving rise to aneuploidy, which may be the reason for the rfc1–2 pollen sterility.

Figure 4.

Centromere fluorescent in situ hybridization on wild-type and rfc1–2 meiocytes. (a, b) Telophase I. Ten signals are separated unequally in the rfc1–2 mutant. (c, d) Telophase II. The separation of rfc1–2 tetrad centromere signals is variable compared to wild-type. Scale bars = 20 μm. (e–h) Pie charts showing the percentages of various patterns in wild-type (e, g) and the rfc1–2 mutant (f, h) during telophase I (e, f) and telophase II (g, h). = 144, 90, 124 and 108, respectively, for (e), (f), (g) and (h).

AtRFC1 expression profiles in reproductive organs

As the observations above showed that there were defects during rfc1–2 meiosis, we investigated the detailed spatial and temporal expression patterns of AtRFC1 during sexual reproductive development. RNA in situ hybridization showed that there was a basal level of AtRFC1 expression in most organs, but high levels in reproductive organs (Figure 5a). Particularly strong signals were present in male meiocytes from the PMC stage to tetrad formation (Figure 5b,c). After meiosis, the tapetum showed increased signals but expression was reduced in pollen cells (Figure 5d). In the early gynecium, AtRFC1 expression was present throughout developing ovules, with high levels in ovule primordia, megaspore mother cells and female gametophytes (Figure 5e–g). Quantitative analysis showed that there was relatively high expression of AtRFC1 in buds of anthers at stages 1–7 compared with stages 8–10 and 11–12, indicating that AtRFC1 expression decreases gradually during the process of anther development (Figure 5i). Collectively, AtRFC1 was expressed highly during reproductive development, and this expression pattern was consistent with its function in meiosis.

Figure 5.

Expression pattern of the AtRFC1 gene in floral meristem, male and female organs by RNA in situ hybridization. (a) AtRFC1 is expressed in the inflorescence meristem (IM), developing flower primordia (FP) and stamen primordial (SP). (b) AtRFC1 is abundant in the pollen mother cells (PMCs). (c) The tetrad (Td) and tapetum show obvious AtRFC1 expression. (d) AtRFC1 expression is mainly focused in the tapetum (T) in the uniucleate pollen stage. (e) During ovule formation, AtRFC1 mRNA is present in ovule primordia (OP). (f) AtRFC1 expression is detectable at the megaspore mother cell (MMC) stage. (g) Female gametophyte (FG) development. (h) Sense probe control. (i) Relative expression of AtRFC1 in three mixed buds divided on the basis of anther development stages.

The meiotic chromosome fragmentation generated in the rfc1–2 mutant depends on AtSPO11-1 and is repaired via the AtRAD51 pathway

The formation of DSBs catalyzed by SPO11 is indispensable to subsequent recombination and synapsis (Bergerat et al., 1997). To determine whether the chromosome fragmentation observed in rfc1–2 meiocytes was due to a defect in SPO11-induced DSB repair or unrepaired breaks during replication, we generated an spo11–1–3 rfc1–2 double mutant and analyzed its meiotic chromosome characteristics. Due to a dramatic decrease in DSB-mediated meiotic recombination, ten achiasmatic univalents instead of five bivalents were observed in diakinesis and metaphase I in spo11–1–3 (Figure 6a2,a3). The spo11–1–3 rfc1–2 double mutant exhibited a similar phenotype (Figure 6b1–b6) to that of the spo11–1–3 single mutant (Figure 6a1–a6): synapsis did not occur (Figure 6b1) and the fragmentation present in rfc1–2 meiocytes was absent. Instead, ten randomly distributed univalents (Figure 6b2,b3) and polyads at telophase II (Figure 6b6) were observed in the double mutant. Counting analysis showed that the spo11–1–3 mutant lacks any fragmentation (0%), and a low level of fragmentation (9%) was observed in the spo11–1–3 rfc1–2 double mutant, far below that of rfc1–2 (76%, Table 1). These results demonstrate that unrepaired DSBs in rfc1–2 are mainly induced by AtSPO11–1, and AtSPO11–1 is epistatic to AtRFC1.

Figure 6.

Chromosome behavior at male meiosis in spo11–1, spo11–1 rfc1–2, rad51, rad51 rfc1–2, mlh3 and mlh3 rfc1–2 lines. (a1–a6) spo11–1; (ba–b6) spo11–1 rfc1–2; (c1–c6) rad51; (d1–d6) rad51 rfc1–2; (e1–e6) mlh3 and (f1–f6) mlh3 rfc1–2.

A previous study in budding yeast showed that, after SPO11 removal, the 3′ strand end is coated with RAD51, which mediates subsequent single-strand invasion and double Holliday junction formation (Shinohara et al., 1992). To further assess AtRFC1 involvement in the recombination pathway, an rad51 rfc1–2 double mutant was generated and analyzed. Chromosome behavior of the homozygous double mutant was indistinguishable from that of the rad51 single mutant, with an absence of synapsis (Figure 6c1,d1), abnormal condensation of chromosomes (Figure 6c2,c3,d2,d3), prominent fragmentation (Figure 6c4,d4) and multiple brightly stained spots (Figure 6c5,c6,d5,d6). As Table 1 shows, the fragmentation level in the rad51 rfc1-2 mutant was 100% in both cases. This implies that AtRFC1 may act with AtRAD51 in processing of recombination intermediates during meiosis.

Chromosome defects in rfc1–2 are independent of AtMLH3

In recombination, another key event is crossover regulation, which is resolved from the double Holliday junction recombination intermediate. Eukaryotic homologs of MutS and MutL in A. thaliana were previously shown to be involved in crossover formation. In order to understand whether the chromosome phenotypes of rfc1–2 are related to crossover regulation, we analyzed the involvement of one MutL protein: AtMLH3. The result showed that loss of AtMLH3 in the mlh3 mutant did not affect synapsis, but some homologous chromosome pairs lacked chiasmata and were present as univalents at diakinesis and metaphase I, and chromosome mis-segregation occurred at the first meiotic division (Figure 6e1–e6). When we analyzed the mlh3 rfc1–2 double mutant, we found that it behaves more like the single rfc1–2 mutant: many fragmentations were present during the division process (Figure 6f1–f6). Thus, the results suggest that AtRFC1 is not involved in crossover regulation. Therefore, we conclude that the meiosis defect in rfc1–2 is independent of AtMLH3 function.

AtRFC1 protein localization in meiocytes of wild-type, spo11–1 and rad51

As loss of AtRFC1 caused meiotic chromosome abnormalities, we used an immunofluorescence technique to investigate the intracellular distribution of AtRFC1 on PMCs using anti-AtRFC1 antibody. A Western blot experiment was performed to test the antibody specificity (Figure S8). When the anti-AtRFC1 antibody was applied to a chromosome spread for rfc1–2 meiocytes, no signal was detected (Figure S9). In wild-type, chromosomes lack AtRFC1 foci at early prophase I (Figure 7a–d). When the meiotic chromosomes entered zygotene and pachytene, AtRFC1 signal was mainly detected along the chromosome axes (Figure 7e–l). This signal persisted through zygotene and pachytene, during which partial co-localization with the axis-associated protein ASY1 was observed.

Figure 7.

AtRFC1 protein localization to meiotic chromosomes in wild-type, spo11–1 and rad51. (a–l) Immunolocalization of AtRFC1 in wild-type meiocytes using anti-AtRFC1 antibody. (a–d) Leptotene. (e–h) Zygotene. (i–l) Pachytene. (m–t) AtRFC1 signal in spo11–1 (m–p) and rad51 (q–t) meiocytes at zygotene.

Previously, it has been shown that Ddc1 protein, a PCNA ortholog in yeast, requires formation and processing of DSBs for its localization to meiotic chromosomes (Hong and Roeder, 2002). To determine whether AtRFC1 localization had a similar dependence, AtRFC1 foci were examined in spo11–1 meiocytes. AtRFC1 foci were absent in the spo11–1 mutant compared with wild-type (Figure 7m–p). Additionally, we compared the relative expression level of AtRFC1 in wild-type and spo11–1 inflorescences, and found that AtRFC1 transcripts were reduced in the spo11–1 mutant (Figure S10). Therefore, our results suggest that DSBs are required for normal localization of AtRFC1 to meiotic chromosomes. As the above results suggest that AtRFC1 and AtRAD51 function in the same pathway, we analyzed the AtRFC1 localization in rad51 meiocytes. As Figure 7(q–t) shows, the AtRFC1 signal distribution did not change in rad51 compared with wild-type. The results suggest that loss of AtRAD51 does not affect recruitment of AtRFC1 to the chromosome.

Discussion

In the present study, we found that AtRFC1, the largest subunit of the replication factor C complex, contains a BRCT motif, and is involved in maintaining meiotic chromosome stability. The sterility of the rfc1–2 mutant was due to defects in the male and female meiocytes. In particular, meiotic chromosome behavior was abnormal, with fragmentation emerging as early as diplotene and remaining in subsequent stages. Due to the serious damage in rfc1–2 meiocytes, homologous chromosomes and sister chromatids separated unequally, ultimately leading to gametophyte sterility. In addition, the fact that the spo11–1–3 mutation suppresses the rfc1–2 phenotype indicates that AtRFC1 has a meiotic function following DSB formation. Moreover, we also found that AtRFC1 may plays roles during AtRAD51-mediated formation of recombination intermediates. The AtRFC1 localization on zygotene and pachytene was consistent with its role in homologous recombination.

The AtRFC1 C-terminus may be required for the DNA damage response

In the rfc1–2 mutant, the T–DNA insertion is located at the C–terminus of AtRFC1 and disrupts transcription of the C–terminus. Similarly, the amino acid mutation in the rfc1–1 mutant is also located in the C–terminal, very close to the rfc1–2 mutation. As both mutants displayed a defect in DNA repair during meiosis, we speculate that the C–terminal region of AtRFC1 may be responsible for DNA damage response during meiosis. Analysis of Elg1, one of the RFC1 paralogs in S. cerevisiae, revealed the possible roles of various Elg1 regions. When the C–terminal 60 amino acids were removed, spontaneous DNA damage increased, indicating an independent role of the C–terminus (Davidson and Brown, 2008). Therefore, disruption of the C–terminus of the rfc1–1 and rfc1–2 mutants may lead to an impaired DNA damage response and a reduction of transcripts upstream of the T–DNA insertion in rfc1–2. The regions required for complex formation and chromatin binding were not severely affected, thus the rfc1–1 and rfc1–2 mutants were not lethal.

AtRFC1 is required in meiosis

Unlike animals, plants do not display arrest in prophase I or apoptosis in response to unrepaired DSBs. In plants, various mutants such as rad51 and dmc1, which have defects during meiosis, do finally complete meiosis, in contrast to their yeast and mouse counterparts (Li et al., 2004; Shinohara et al., 1992; Dresser et al., 1997). In this study, despite the presence of chromosome breakages and fragmentations, rfc1–2 meiocytes still proceeded to cytokinesis, a characteristic that is common to other meiotic mutants. Nevertheless, most meiotic mutants do not undergo subsequent mitosis, and arrest at the uninuclear pollen or functional megaspore stage. DAPI staining indicated that the remaining nuclei started to degenerate from the centrum in the following stage. This observation indicates that nucleus integrity or chromosome stability is a significant factor for subsequent mitosis initiation. In summary, we infer that the sterility in the rfc1–2 mutant is the result of aberrant chromosome content: aneuploidy or chromosome fragments.

It is generally assumed that RFC is responsible for DNA replication and act as a damage checkpoint during S phase. However, chromosome behavior appeared normal until diplotene. The pachytene chromosome behavior and localizations of ZYP1 and ASY1 were the same as for wild-type. Taken together, pairing and synapsis proceeded normally in the rfc1–2 mutant. Thus, AtRFC1 does not participate in pairing and synapsis during meiosis. However, some breakages did occur in diplotene, in contrast to some other reported meiotic mutants. Previously, an atrad51 mutant was shown to be defective in homologous pairing and synapsis (Li et al., 2004). As similar phenotypes were observed in rad51 rfc1–2 and rad51, AtRFC1 may contribute to the homologous recombination event. The immunolocalization studies showed that AtRFC1 localizes on chromosomes at zygotene and pachytene, consistent with its role in homologous recombination. In addition, studies in yeast revealed that strand invasion drives the process of synapsis (Mahadevaiah et al., 2001). Thus our finding that AtRFC1 functions with or after AtRAD51 may explain the normal synapsis. The phenotype of typical rfc1–2 pachytene is similar to that of the atm mutant of Arabidopsis (Garcia et al., 2003). As ataxia telangiectasia mutated (ATM) in humans phosphorylates a series of DSB-related proteins including Rad24 (Kastan and Lim, 2000), AtRFC1 may act as the phosphorylation target of ATM, i.e. they probably function in the same process. Therefore, the relationship between AtRFC1 and ATM requires further analysis.

AtRFC1 repairs AtSPO11–1-induced DSB via AtRAD51-mediated homologous recombination

A previous study showed that Rad24, an RFC1 homolog, was required for repair of DSBs during meiosis in S. cerevisiae (Shinohara et al., 2003). The delay in the meiotic cell cycle in the rad24 mutant depended on SPO11. We wished to determine whether the situation for RFC1 in Arabidopsis is similar. Results for the double mutant spo11–1 rfc1–2 showed that AtRFC1 acts downstream of AtSPO11, consistent with evidence concerning Rad24. By introducing the spo11–1 mutation into the rfc1–2 mutant, breakage or fragmentation was largely absent, in contrast to the single rfc1–2 mutant. As γH2AX foci were maintained at a relatively high level in rfc1–2 meiocytes, AtRFC1 was further shown to be involved in the process of repairing DSB produced by AtSPO11–1. This repair process in which AtRFC1 is involved requires its correct localization on the chromosome, because AtRFC1 foci were absent in the spo11–1 mutant. Thus, DSBs produced by AtSPO11–1 may be able to recruit AtRFC1 onto the chromosome to perform its function.

How does AtRFC1 participate in DSB repair after DSB formation? First, according to our results for rad51 rfc1–2, AtRFC1 may act after or during single-strand invasion. We infer that the subsequent repair event, including DNA synthesis, may require the RFC loading function. After DSB formation, single-strand invasion is a crucial step for intermediate formation, requiring AtRAD51, AtRAD51C and AtXRCC3 (Bleuyard et al., 2005). Another crucial event in homologous recombination is the crossover/non-crossover decision and formation, which involves MSH4/5, MLH3, etc. (Higgins et al., 2004, 2008; Jackson et al., 2006). Based on the chromosome phenotype in rfc1–2, we suggest that AtRFC1 acts after single-strand invasion, and before completion of crossover/non-crossover formation during meiosis recombination. In fact, Li et al., (2009) proposed a model whereby loading of PCNA by RFC at the D–loop recruits polymerase δ to the 3′ end of the invading strand for processive DNA synthesis. Therefore, it looks like AtRFC1 may be responsible for DNA synthesis following single strand invasion during meiosis. Second, AtRFC1 may recognize DSBs after single-strand invasion. In yeast, DSBs may be recognized by Rad24, which loads the PCNA-like complex to sites of DSB repair (Hong and Roeder, 2002). In fact, the RFC complex may recognize the transition between ssDNA and dsDNA (Lowndes and Murguia, 2000). Therefore, the re-organization involved in single-strand invasion may be sufficient to recruit other repair proteins to perform recombination, e.g. PCNA. This interaction between the clamp and clamp loader may activate additional recruitment of recombination repair proteins.

In conclusion, based on our developmental and genetic studies, Arabidopsis female and male developmental defects were the cause of sterility. Further analysis showed that AtRFC1 acted downstream of AtSPO11–1 and with AtRAD51 during meiotic homologous recombination. The genetic and functional framework developed for AtRFC1 in this study will enable determination of other components of homologous recombination-mediated DSB repair pathways that operate during meiosis in Arabidopsis.

Experimental procedures

Plant materials and growth conditions

Arabidopsis mutants harboring a T–DNA insertion were obtained from the Arabidopsis Biological Resource Center: rfc1–2 (SALK_140231), mlh3 (SALK_015849), spo11–1–3 (SALK_146172) and rad51 (SALI_873_C08). Double mutants of spo11–1–3 rfc1–2 and rad51 rfc1–2 were generated by crossing the above mutants. Primers are listed in Table S1. All plants were grown in a growth chamber at 22 ± 2°C with a 16 h light/8 h dark photoperiod.

Phenotypic analysis and microscopy

Pollen viability was examined using Alexander staining (Alexander, 1969). Meiotic chromosome spreads were performed as described by Ross et al. (1996) and observed under a fluorescence microscope.

The confocal observation of ovules was performed as described by Christensen et al. (1997) with slight modifications. Inflorescences were fixed in 4% paraformaldehyde overnight at 4°C. The samples were then dehydrated through a conventional ethanol series at 20 min per step. The tissues were cleared in a solution of chloral hydrate/double-distilled water/glycerin (8:2:1 w/v/v), and the pistils or siliques were dissected using fine needles, mounted with the cleared solution, and covered with a cover glass. The tissues were observed using an Olympus (http://www.olympus-global.com) Fluoview 1000 laser scanning microscope with a 488 nm argon laser and a long-pass 530 filter.

For microscopic observation, anthers from wild-type and rfc1–2 plants were fixed in 2.5% glutaraldehyde and 2% paraformaldehyde in phosphate buffer, pH 7, for 4 h at 4°C. Then an ethanol dehydration series was performed, followed by post-fixation in xylene and embedding of samples in Spurr resin (SPI-ChemTM Low Viscosity "Spurr" Kits, http://www.2spi.com/). Semi-thin sections (approximately 1 μm) were cut using an ultramicrotome (RMC-MTX).

Complementation of the rfc1–2 mutant using a full-length AtRFC1 genomic fragment

The AtRFC1 full-length genomic fragment ranging from the promoter to the 3′ UTR was amplified from wild-type Arabidopsis using primers each plus SalI and KpnI, and then cloned into pCambia1300 vector (Cambia, http://www.cambia.org/). The floral-dip method (Clough and Bent, 1998) was used to transform rfc1–2 mutants. After screening on a hygromycin plate, positive transformants were used for subsequent analysis.

In situ hybridization

RNA in situ hybridization was performed as described previously (Brewer et al., 2006). Probes were prepared by PCR methods. An antisense probe was generated by PCR amplification following in vitro transcription. Procedures were performed according to Brewer et al. (2006).

FISH was performed as described by Fransz et al. (1998) with some modifications. After the formaldehyde fixation step, the slides were treated with 70% deionized formamide in 2× SSC for 5 min at 75°C, and dehydrated through a graded ethanol series. Approximately 40 μl of probe containing 50% deionized formamide, 2× SSC and 10% dextran sulfate was applied to the slides, which were hybridized for 18 h at 37°C in a humid box. Then slides were counterstained using DAPI. The oligonucleotide 5′-GGTTGCGGTTTAAGTTCTTATACTCAATCATACACATGAC-3′ labeled with FITC at the 5′ end was used to detect the centromere (Ronceret et al., 2009).

Fluorescence immunolocalization

The AtRFC1 C–terminal amino acid sequence, DSLRDEDGEPLADNE, was synthesized as a peptide. AtRFC1 polyclonal antibody was produced from rabbit using the synthesized peptide as the immunogen (GenScript, http://www.genscript.com/). Pre-immune and immune serum were obtained from rabbits to perform Western blotting. Immunolocalization was performed as previously described by Armstrong et al. (2002) using the following antibodies: anti-ASY1 (rabbit, 1/500 dilution), anti-ZYP1 (rabbit, 1/500 dilution), anti-AtRFC1 (rabbit, 1/100 dilution) and anti-γH2AX (phosphorylated on Ser139, rabbit, 1/200 dilution; Upstate Biotechnology, http://www.millipore.com/). For simultaneous staining of ASY1 and AtRFC1 using primary antibodies from the same species, the ASY1 antibody was covered with FITC-labeled Fab fragments before application of the AtRFC1 antibody. Quantification of the fluorescence foci was performed using ImageJ software (http://rsbweb.nih.gov/ij/).

Statistical analysis

Most statistical analyses were performed in Excel (Microsoft, http://www.microsoft.com/). Boxplots were generated in R (http://www.r&#x2013;project.org/) using the boxplot()function.

Accession numbers

The accession numbers for the proteins used in this study are At5g22010 (AtRFC1), At3g13170 (AtSPO11-1), at4g35520 (AtMLH3) and At5g20850 (AtRAD51).

Acknowledgements

We are very grateful to Chris Franklin (University of Birmingham, UK) for kindly providing the antibodies against ASY1 and ZYP1, helpful advice and manuscript revision. We thank Zhizhong Gong (China Agricultural University, China) for sharing rfc1–1 seed. This research was supported by the Major State Basic Research Program of China (2012CB944801 and 2013CB126903), the Key Grant Project of the Chinese Ministry of Education (311026) and the National Natural Science Foundation of China (30970277).

Ancillary