SEARCH

SEARCH BY CITATION

Keywords:

  • Arabidopsis;
  • control of recombination;
  • genetic crossover formation;
  • meiosis;
  • meiotic recombination

Abstract

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

Contents

 Summary523
I.Introduction524
II.The meiotic pathway: a brief overview524
III.Homologous chromosome pairing and  movement during prophase I525
IV.Meiotic DNA double-strand break formation526
V.Processing of DNA double-strand breaks528
VI.Strand exchange: the role of the RecA  homologues and their accessory proteins529
VII.Promotion of stable strand exchange532
VIII.Pathways to crossover formation532
IX.The class I pathway of meiotic recombination533
X.The class II pathway of meiotic recombination536
XI.Holliday junction (Hj) resolution537
XII.Noncrossover pathways and the crossover/ noncrossover decision537
XIII.Conclusions538
 Acknowledgements538
 References538

Summary

Meiosis is a central feature of sexual reproduction. Studies in plants have made and continue to make an important contribution to fundamental research aimed at the understanding of this complex process. Moreover, homologous recombination during meiosis provides the basis for plant breeders to create new varieties of crops. The increasing global demand for food, combined with the challenges from climate change, will require sustained efforts in crop improvement. An understanding of the factors that control meiotic recombination has the potential to make an important contribution to this challenge by providing the breeder with the means to make fuller use of the genetic variability that is available within crop species. Cytogenetic studies in plants have provided considerable insights into chromosome organization and behaviour during meiosis. More recently, studies, predominantly in Arabidopsis thaliana, are providing important insights into the genes and proteins that are required for crossover formation during plant meiosis. As a result, substantial progress in the understanding of the molecular mechanisms that underpin meiosis in plants has begun to emerge. This article summarizes current progress in the understanding of meiotic recombination and its control in Arabidopsis. We also assess the relationship between meiotic recombination in Arabidopsis and other eukaryotes, highlighting areas of close similarity and apparent differences.


I. Introduction

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

Meiosis is a key biological process that underpins sexual reproduction. During meiosis, a single round of DNA replication is followed by two rounds of nuclear division to produce four haploid gametes. Subsequent union of the male and female gametes restores the chromosome number of the zygote to the parental level, thereby ensuring the stability of the chromosome complement from generation to generation. In addition, it produces genetic variation through the formation of new combinations of alleles during meiotic recombination which occurs during prophase I.

Meiosis research dates back over a century to the early days of cell biology when developments in microscopy allowed direct observation of chromosome segregation. This led to detailed cytogenetic studies of meiosis in a diverse range of organisms, with plants playing a prominent role. Although these provided remarkable insights into meiotic chromosome organization and behaviour, advances in our understanding of the underlying molecular mechanisms that control meiotic recombination are comparatively recent. Studies in Saccharomyces cerevisiae (budding yeast), in particular, have provided much insight into our current understanding of meiosis. These have stimulated and been accompanied by research in a wide variety of other eukaryotic species, including Schizosaccharomyces pombe (fission yeast), the filamentous fungus Sordaria macrospora, invertebrates such as Drosophila melanogaster and Caenorhabditis elegans, and the mouse Mus musculus. Similarly, significant progress has been achieved in flowering plants, led by studies in Arabidopsis thaliana, accompanied by valuable insights from maize (Zea mays) and, increasingly, rice (Oryza sativa). This, combined with their amenability to cytogenetic analysis, has allowed the study of plant meiosis to continue to make an important contribution to our understanding of this fundamental biological process.

This article focuses on the machinery of meiotic recombination and its control, highlighting the contribution made by studies in Arabidopsis to the understanding of the genes and proteins required for the formation of meiotic crossovers (COs) in plants. In addition, we discuss the similarities and differences in the meiotic pathway in plants compared with other organisms. Owing to space limitations, several aspects of meiosis are not extensively discussed. Hence, for readers requiring a broader overview of the topic, we would draw their attention to the following articles and others cited within the article (Zickler & Kleckner, 1998, 1999; Hamant et al., 2006; Mezard et al., 2007; Mercier & Grelon, 2008).

II. The meiotic pathway: a brief overview

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

In flowering plants, male meiosis occurs in the anther and female meiosis in the ovary. During anther development, a group of sub-epidermal cells differentiates, forming archesporial cells, which then give rise to primary sporogenous cells. These differentiate into pollen mother cells, in which meiosis occurs, and parietal cells from which the remaining tissues of the mature anther are derived. In the female tissue, an archesporial cell is derived from a hypodermal cell at the top of the ovule primordium, forming the megaspore mother cell in which meiosis occurs. Unlike the male, where the four haploid products of meiosis all form mature pollen grains, three of the four female spores undergo programmed cell death. The remaining cell then develops into the female gametophyte (embryo sac) (Bowman, 1994; Yang et al., 1999).

During meiotic S-phase, cohesion is established between the sister chromatids by loading of the cohesin complex (reviewed in Nasmyth & Haering, 2005). At the end of G2, a proteinaceous axis is elaborated along the chromatids, such that they form a looped array that is conjoined at the loop bases (Fig. 1) (Kleckner, 2006). This denotes leptotene, the first substage of prophase I. DNA double-strand breaks (DSBs) are then formed at sites along the chromosomes. Evidence from budding yeast indicates that the actual break sites are located within chromatin loops, but become physically associated with the chromosome axis. The homologous chromosomes then begin to align along their length (Kleckner, 2006). In most species analysed thus far, this is facilitated by early stages of the recombination process, but there are exceptions, such as C. elegans (see section III for a discussion of early pairing). As recombination proceeds, the homologous chromosomes are brought into close apposition by the formation of the synaptonemal complex (SC) (Fig. 1) (Page & Hawley, 2004). This is initiated at zygotene and continues through to pachytene which is denoted by a continuous SC running the length of the fully synapsed pairs of homologous chromosomes. The SC has a tripartite structure comprising the homologous chromosome axes, referred to as lateral elements in the context of the SC, which are cross-linked in the manner of a ‘zipper’ by a transverse filament protein that polymerizes between them. Recombination is completed during pachytene. During late prophase I (diplotene/diakinesis stages), the homologous chromosomes condense and the SC breaks down. By diakinesis, the SC is fully degraded and the homologous chromosome pairs (bivalents) begin to separate, except at sites at which COs have occurred, establishing physical connections which are known cytologically as chiasmata. At this stage, sister chromatid cohesion is maintained. At the end of prophase I, the bivalents continue to condense and begin to align on the equator of the metaphase I plate. Correct alignment and subsequent accurate disjunction at anaphase I are dependent on chiasma formation and sister chromatid cohesion. At anaphase I, cohesion along the chromosome arms is lost and the homologous chromosomes separate to opposite poles; inter-sister links are preserved by maintaining centromeric cohesion. This first division is then followed by a mitotic-like second division, at which point centromeric cohesion is lost, allowing the sister chromatids to separate to form a tetrad of four haploid spores.

image

Figure 1. Chromosome organization during prophase I of meiosis. Following meiotic S-/G2-phases, the sister chromatids become conjoined along their length by a proteinaceous axis that is elaborated along the bases of the chromatin loops. This denotes the onset of leptotene. Immunolocalization on chromosome spread preparations from meiocytes using antibodies which recognize axis components, such as the cohesin AtSMC3, enable visualization of the linear axis. The homologous chromosomes then align and pair before undergoing synapsis in zygotene. During zygotene, formation of the synaptonemal complex (SC) occurs, bringing each homologous chromosome pair into close apposition. At pachytene, SC polymerization is complete along the length of the homologues (transverse and longitudinal views illustrated). SC polymerization may be monitored by immunolocalization of the transverse filament protein (AtZYP1 in Arabidopsis) which is a component of the SC central region.

Download figure to PowerPoint

In Arabidopsis, meiosis takes c. 33 h following S-phase, most of which (30 h) is taken up by prophase I (Armstrong et al., 2003). The two nuclear divisions of meiosis I and meiosis II occur within a period of just 3 h.

III. Homologous chromosome pairing and movement during prophase I

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

In most species, including Arabidopsis and other plants, completion of chromosome pairing and synapsis are dependent on recombination. However, the earliest events in the process of homologue recognition are not well understood. Nevertheless, it seems clear that the telomeres play an important role. In many species, a prominent feature of meiotic nuclei at early prophase I is the organization of the chromosomes into a ‘bouquet’ arrangement. During the bouquet stage, the telomeres cluster to a small region on the nuclear membrane near the microtubule organizing centre (MTOC) (Scherthan, 2001). Bouquet formation is also observed in many plants, but whether this involves a specific attachment region is unclear, as plants do not appear to have an MTOC (Roberts et al., 2009). It has been proposed that it is this arrangement that facilitates the earliest steps in homologous chromosome pairing (Scherthan, 2001).

Centromere pairing during early prophase I, preceding alignment of the chromosome arms, has also been reported in many species, including plants, such as onion, in which this phenomenon was first reported (Church & Moens, 1976). Importantly, however, during early prophase I, pairing occurs between nonhomologous centromeres, giving way to homologous pairing later in prophase I. Indeed, studies in wheat suggest that homologous chromosome pairing is primarily dependent on homology along the chromosome arms rather than at the centromeres (Corredor et al., 2007). Moreover, in Arabidopsis, there is no evidence of early centromere pairing. The telomeres form a nucleolus-associated cluster, rather than a classical bouquet, before undergoing homologous pairing in late G2/early leptotene (Armstrong et al., 2001). The centromeric regions do not pair until zygotene. Hence, it seems likely that early centromere pairing serves some other function rather than facilitating homologue recognition and early pairing. Stewart & Dawson (2008) have proposed that it might influence kinetochore assembly or disfavour the formation of potential deleterious centromere proximal chiasmata.

Another possibility is that centromeric pairing may provide a counter-balance to the rapid telomere-led chromosome movements that have been observed during mid-prophase I (Stewart & Dawson, 2008 and references therein). Telomere movements were initially observed in S. pombe and, more recently, in S. cerevisiae and C. elegans (Koszul & Kleckner, 2009 and references therein). They are mediated by components of the cytoskeleton that are directly linked via bridges that span the nuclear membrane to the chromosome ends (see next paragraph). It is proposed that they remove unwanted chromosomal connections. Rapid chromosome movements have also been observed during meiotic prophase I in maize, where they may aid homologous pairing during zygotene (Sheehan & Pawlowski, 2009).

The dependence of chromosome pairing on recombination is by no means universal. For example, pairing occurs in male Drosophila in the absence of recombination (Vazquez et al., 2002). In C. elegans homologue recognition, pairing and synapsis are independent of DSBs and subsequent recombination. Instead, homologue recognition is dependent on heterochromatic repeats that are present in the subtelomeric region at one end of each chromosome (MacQueen et al., 2005). These are referred to as homologue recognition regions or pairing centres (PCs). The proteins ZIM1-3 and HIM-8 specifically interact with one or two chromosomes via the PC to mediate homologous pairing. The PCs are tethered to the nuclear membrane, where they are linked to cytoplasmic microtubules via a nuclear membrane-spanning complex comprising the SUN/KASH proteins (reviewed in Hiraoka & Dernburg, 2009). The KASH protein ZYG-12 interacts directly with the dynein motor protein that can ‘walk’ along the microtubule. A model is proposed whereby the PCs associate in a patch on the nuclear membrane via an, as yet undetermined, mechanism (Jaspersen & Hawley, 2009; Sato et al., 2009; Baudrimont et al., 2010). This facilitates homologue recognition via the PCs and cognate PC proteins. This is counteracted by dynein which tends to pull the associations apart, such that only correctly linked homologous PCs are sufficiently robust to resist, whereas nonhomologous interactions are rapidly separated. In addition, the mechanical force generated by dynein pulling on the correctly paired PCs is thought to somehow facilitate chromosome synapsis by releasing a block imposed by SUN-1. It is worth noting that the SUN/KASH domains are widely conserved. For example, the SUN domain protein, Sad1, is known to mediate telomere clustering in fission yeast via interaction with the Rap1/Bqt1/Bqt2 complex (Chikashige et al., 2006). Arabidopsis has at least two SUN domain proteins and rice has three, although their meiotic role has yet to be established (Graumann et al., 2009). Thus, it seems likely that the telomere-led chromosome movement observed in a variety of species during meiotic prophase I may be mediated by a similar mechanism.

IV. Meiotic DNA double-strand break formation

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

Meiotic recombination mechanisms have been established primarily from studies in budding yeast. Homologous recombination leading to CO formation is initiated early in prophase I by the formation of programmed DSBs catalysed by the conserved protein, Spo11, which shares homology with the TOP6A type II topoisomerase from the archaeon Sulpholobus shibatae (Fig. 2) (Bergerat et al., 1997; Keeney et al., 1997). Following break formation, Spo11 remains covalently attached to the 5′-ends of the DNA on either side of the break site. It is removed as the DNA ends are resected to generate 3′-ended single-stranded DNA (ssDNA) required for homologous recombination. These interact with the RecA homologues Rad51 and Dmc1 to form nucleoprotein filaments. The filament on one side of the break invades the homologous duplex DNA of one of the two nonsister chromatids to form a stable single-end invasion intermediate. The displaced DNA strand forms a D-loop that extends as the invading strand polymerizes. This enables the capture of the 3′-end on the other side of the break. Subsequent ligation of the broken DNA strands results in the formation of the double Holliday junction (dHj) recombination intermediate. In budding yeast, the dHjs are resolved to form COs (Neale & Keeney, 2006). In Arabidopsis, it has yet to be established whether dHjs solely give rise to COs, or whether a proportion may be processed as noncrossover (NCO) products.

image

Figure 2. Homologous recombination during meiosis. It is believed that the meiotic recombination pathway in Arabidopsis is likely to be broadly similar to the ‘early crossover decision’ model which has been developed in budding yeast (see text for details). Several of the key recombination pathway proteins are shown at the stage at which they are active. It should be noted that these proteins work in conjunction with many others that are mentioned in the text but, for reasons of clarity, are not included in the figure. Hydrogen bonding between the single-stranded DNA during second-end capture is depicted by dots. It is considered that most noncrossovers (NCOs) arise via synthesis-dependent strand annealing. Recombination intermediates that form double Holliday junction (dHj) structures are thought to be primarily resolved as crossovers (COs). However, studies in Arabidopsis suggest that a proportion of the dHjs may undergo dissolution by *AtRMI1/*AtTop3α to form NCOs (see text for details). DSB, double-strand break.

Download figure to PowerPoint

Arabidopsis contains three Spo11 paralogues, two of which, AtSPO11-1 and AtSPO11-2, are required for meiosis (Table 1) (Hartung & Puchta, 2000, 2001; Grelon et al., 2001; Stacey et al., 2006). The catalytically active tyrosine residues of both proteins are necessary for DSB formation and probably function as a heterodimer (Hartung et al., 2007b). A time-course analysis of recombination initiation identified a significant delay between the association of AtSPO11-1 with the chromatin, early in G2, and the appearance of DSBs, marked by histone AtH2AX phosphorylation, which occurred concurrently with chromosome axis formation (Sanchez-Moran et al., 2007). These observations are consistent with the ‘tethered-loop/axis complex’ model which proposes that DSBs and axis morphogenesis are synchronized, spatially coordinated events (Blat et al., 2002; Kleckner, 2006).

Table 1.   DNA double-strand break (DSB) formation
GeneProtein activity/functionMutant phenotype
  1. Notes: (i) Despite Spo11-mediated DSB formation being highly conserved in sexually reproducing eukaryotes, the accessory proteins associated with this process appear to be diverse. In budding yeast, nine accessory proteins (Mre11, Rad50, Xrs2, Ski8, Rec102, Rec104, Rec114, Mei4, Mer2) are required for DSB formation. In Arabidopsis, AtSPO11-1/AtSPO11-2-mediated DSB formation requires AtPRD1/AtPRD2/AtPRD3. (ii) Homologues of Mre11, Rad50, Ski8 are present, but AtMRE11 and AtRAD50 are required after DSB formation (see text and Table 2) and AtSKI8 appears to have no meiotic role.

AtSPO11-1/AtSPO11-2Topoisomerase II-related proteins essential for DSB formation. It is proposed that AtSPO11-1 and AtSPO11-2 function as a heterodimerAbsence of chiasmata; univalent chromosomes observed at metaphase I
AtPRD1Required for DSB formation; biochemical function unknown, but possible functional homologue of mammalian MEI1 protein. Interaction partner with AtSPO11Absence of chiasmata; univalents at metaphase I; absence of AtDMC1 foci
AtPRD2Required for DSB formation; contains motifs present in DNA-binding proteins. Probable orthologue of MEI4 found in yeast and mouseAbsence of chiasmata; univalents at metaphase I; absence of AtDMC1 foci
AtPRD3Required for DSB formation. Homologue of rice PAIR1 geneAbsence of chiasmata; univalents at metaphase I; absence of AtDMC1 foci

In addition to Spo11, DSB formation in budding yeast requires at least nine other proteins (Mre11, Rad50, Xrs2, Ski8, Rec102, Rec104, Rec114, Mei4 and Mer2) which form several interacting subgroups (reviewed in Cole et al., 2010). The identification of Arabidopsis DSB accessory proteins has not been straightforward because of poor conservation between species. Even where sequence orthologues have been identified, they are often functionally divergent. For example, the Arabidopsis SKI8 orthologue is not required for meiosis (Jolivet et al., 2006), and AtMRE11 and AtRAD50 are not required for DSB formation.

Nevertheless, several genes required for DSB formation, AtPRD1, AtPRD2 and AtPRD3, which is related to rice OsPAIR1 (Table 1) (Nonomura et al., 2004), have been identified by fertility screens of mutant lines (De Muyt et al., 2007, 2009). These genes have a similar mutant phenotype, namely an absence of early recombination markers, such as AtDMC1 foci, no synapsis and a complete lack of chiasmata, leading to random chromosome segregation at anaphase I. Each of the three mutants can suppress the DSB repair defects of a range of meiotic mutants, such as Atrad51, indicating that they are required for DSB formation itself, rather than break repair. AtPRD2 contains several motifs found in DNA-binding proteins, suggesting that it could promote DSB formation by binding to chromosomes, whereas AtPRD1 and AtSPO11 have been found to interact, suggesting that they may be partners. This study also revealed similarity between AtPRD1 and the mammalian DSB protein, MEI1 (Libby et al., 2003), suggesting that they could be functional homologues, and, in a recent study, AtPRD2 was identified as a probable orthologue of budding yeast and mouse MEI4 (Kumar et al., 2010). The same study identified a potential Arabidopsis orthologue of REC114 as AtPHS1, already known to have a meiotic role (Ronceret et al., 2009). Cole et al. (2010) have suggested that rapid evolution of meiotic DSB accessory proteins possibly reflects co-adaptation in response to rapid changes in DNA binding specificity of the recombination hot-spot determinant, PRDM9 (Baudat et al., 2010; Myers et al., 2010; Parvanov et al., 2010), in order to counter hot-spot extinction. To date, no Arabidopsis PRDM9 orthologue has been identified.

V. Processing of DNA double-strand breaks

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

Removal of Spo11, which remains covalently attached to the 5′ DNA end on each side of a DSB, and strand resection are carried out in two distinct steps mediated by the hinge-like Mre11-Rad50-Xrs2/Nbs1 complex acting together with Com1/Sae2/CtIP (MRX and Com1/Sae2 in budding yeast, MRN and CtIP in mammals) (Table 2) (Mimitou & Symington, 2009). A single-strand nick on each side of the DSB releases Spo11 attached to a short oligonucleotide (Neale et al., 2005). The asymmetry of this endonucleolytic cleavage may be a mechanism for defining which 3′-ended ssDNA strand carries out the initial invasion step during homologous recombination. In contrast with the requirement of MRX for DSB formation in yeast, the role of MRX/N in end processing appears to be evolutionarily conserved. In Arabidopsis, both AtMRE11 and AtRAD50 have a role in somatic DNA damage repair (Gallego et al., 2001; Bundock & Hooykaas, 2002; Bleuyard et al., 2004; Puizina et al., 2004). During meiosis, Atmre11 and Atrad50 mutants exhibit a failure of chromosomes to pair and synapse and severe DNA fragmentation (Bleuyard et al., 2004; Puizina et al., 2004). These defects are consistent with a role for these proteins in early prophase I and, indeed, substantial suppression of the fragmentation phenotype in Atmre11-3 Atspo11-1-1 double mutants indicates that AtMRE11 is required for repair, rather than induction, of programmed DSBs (Puizina et al., 2004). Consistent with their function as part of a complex, AtMRE11 and AtRAD50 have been shown to interact in vitro (Daoudal-Cotterell et al., 2002).

Table 2.   Processing of DNA double-strand breaks (DSBs)
GeneProtein activity*/functionMutant phenotype
  1. *Proposed biochemical activity on the basis of studies of the corresponding proteins in other species. See text for further details.

AtMRE11 AtRAD50 AtNBS1 (MRN complex)Function as a complex acting with AtCOM1. Removal of AtSPO11 from 5′ DNA end on each side of the DSB; resection of 5′ DNA to leave 3′ single-stranded DNA tails. In budding yeast, complex comprises Mre11/Rad50/Xrs2. Xrs2/Nbs1 is responsible for localization of the MRX/N complex and signal transduction activityFailure of chromosome alignment and synapsis. Extensive chromosome fragmentation observed at metaphase I
AtCOM1Processing of DSBs. Referred to as Sae2 in budding yeast. Interaction partner with SPO11. Possible functional homologue of the mammalian MEI1 protein required for DSB formationExtensive chromosome fragmentation observed at metaphase I

Xrs2/Nbs1, responsible for the localization and signal transduction activity of MRX/N, displays the greatest sequence divergence of all the complex members. Neverthe-less, the Arabidopsis AtNBS1 protein shares several sequence motifs with orthologues in other species, including an N-terminal forkhead-associated (FHA) domain, a weak BRCA1 C-terminus (BRCT) domain, a consensus Ser-Gln ATM phosphorylation site and a C-terminal MRE11 interaction region containing a FKXFXK motif (Akutsu et al., 2007; Waterworth et al., 2007). Analysis of an Atnbs1-1 mutant revealed a role for AtNBS1 in DNA cross-link repair, but suggested that any meiotic role is non-essential, with the caveat that the Atnbs1-1 allele expresses a short 5′ transcript comprising the FHA domain and half of the BRCT domain (Waterworth et al., 2007).

Mutations in AtCOM1, the Arabidopsis homologue of COM1/SAE2/CtIP, give rise to a similar meiotic phenotype as AtMRE11 and AtRAD50 (Uanschou et al., 2007). In addition, detailed cytological analysis of Atcom1-1 mutant lines revealed an accumulation of AtSPO11-1 during prophase I, failure to form AtRAD51 foci despite the presence of unrepaired DSBs and failure of an Atdmc1 mutation to rescue the fragmentation phenotype of Atcom1-1, all of which are consistent with AtCOM1 acting downstream of AtSPO11-1 and upstream of AtDMC1 to enable regular turnover of AtSPO11-1 and processing of DSBs.

The precise mechanisms of DSB processing are not yet fully understood. In budding yeast, MRX and Com1/Sae2 are both required for Spo11 removal (Longhese et al., 2010). mre11 mutant alleles with impaired nuclease activity allow DSB formation, but not Spo11 removal (Furuse et al., 1998; Tsubouchi & Ogawa, 1998), implying that Mre11 is responsible for cleavage. However, Com1/Sae2 also possesses an ssDNA endonuclease activity (Lengsfeld et al., 2007), raising the possibility that it may cleave close to the DNA break to provide an efficient ‘clean-ended’ substrate for resection by exonucleases (Longhese et al., 2009). In budding yeast, Spo11 removal to allow the initiation of resection and ‘licensing’ of homologous recombination has recently been shown to require Cdk1/Cdc28-dependent phosphorylation of the Ser-267 residue of Sae2 (Manfrini et al., 2010). As Cdk1 activity also directly regulates DSB formation through the phosphorylation of Mer2/Rec107 (Henderson et al., 2006), this provides a mechanism for coordinating DSB resection and progression through prophase I. ClustalW2 alignment of Sae2 and AtCOM1 shows conservation of Ser-267 at position 394 in AtCOM1, although it remains to be seen whether this is a genuine phosphorylation target (K. Osman and F. C. H. Franklin unpublished; Supporting Information Fig. S1). Although an Arabidopsis CDK1/CDC28 sequence orthologue, AtCDKA;1, has been implicated in meiosis (Dissmeyer et al., 2007), there is currently no evidence linking it to meiotic DSB initiation or processing. Although the MRX complex possesses exonuclease activity, it is not thought to be a major contributor to secondary end processing (Mimitou & Symington, 2009). In this context, the budding yeast helicase Sgs1 and the nucleases, Exo1 and Dna2, have been shown to participate in the lengthening of 5′–3′ resection tracts during meiosis, possibly acting redundantly (Manfrini et al., 2010). Interestingly, two biochemical studies have defined Dna2/Sgs1/RPA as a minimal protein complex capable of end resection, with the Top3α–Rmi1 heterodimer and the MRX complex acting as important stimulatory components (Cejka et al., 2010; Niu et al., 2010).

VI. Strand exchange: the role of the RecA homologues and their accessory proteins

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

Following DSB processing to leave 3′-ended ssDNA tails on either side of the break, RecA-related recombinases are loaded to form nucleoprotein presynaptic filaments which can then invade duplex DNA to carry out homology searches and form joint molecules to initiate strand exchange. The cooperation of two recombinases is required to achieve efficient meiotic recombination: Rad51, which also functions in mitotic recombination, and Dmc1, which is only involved in meiosis and appears to have a specific role in recombination between homologues (Bishop et al., 1992; Shinohara et al., 1992).

Like vertebrates, Arabidopsis possesses a single DMC1 gene and six RAD51 paralogues: AtRAD51; AtRAD51B; AtRAD51C; AtRAD51D; AtXRCC2; AtXRCC3 (Table 3) (Klimyuk & Jones, 1997; Doutriaux et al., 1998; Osakabe et al., 2002; Bleuyard et al., 2005). Of the six, only AtRAD51, AtRAD51C and AtXRRC3 are required for meiosis (Bleuyard & White, 2004; Li et al., 2004, 2005; Bleuyard et al., 2005). Disruption of AtRAD51 is consistent with its role as a meiotic recombinase. Mutant plants exhibit normal somatic growth but are sterile as a result of an inability to repair AtSPO11-induced DSBs (Li et al., 2004). Mutants lacking AtDMC1 are unable to form COs and their chromosomes segregate randomly as univalents at anaphase I, reflecting its role in inter-homologue recombination (Couteau et al., 1999). An AtRAD51-dependent absence of chromosome fragmentation in these mutants suggests that DSBs are repaired using the sister chromatid as a template (Siaud et al., 2004). Interactions have been demonstrated between AtRAD51C–AtXRCC3 and AtRAD51–AtXRCC3 (Osakabe et al., 2002), consistent with human studies proposing that RAD51C and XRCC3 form a complex which may include RAD51 (Liu et al., 2002 and references therein). Disruption of either AtRAD51C or AtXRCC3 leads to an AtSPO11-dependent fragmentation phenotype indicative of a role in meiotic DSB repair (Bleuyard & White, 2004; Bleuyard et al., 2004, 2005; Abe et al., 2005; Li et al., 2005; Vignard et al., 2007).

Table 3.   DNA strand exchange: RecA homologues and accessory factors
GeneProtein activity*/functionMutant phenotype
  1. DSB, double-strand break.

  2. Notes: (i) Six RAD51 paralogues (AtRAD51, AtRAD51B, AtRAD51C, AtRAD51D, AtXRCC2 and AtXRCC3) are present in Arabidopsis, of which AtRAD51, AtRAD51D, AtXRCC3, together with AtDMC1, have a meiotic role. (ii) It is still unclear how Dmc1 and Rad51 work in conjunction to promote first strand exchange. It has been suggested that they may bind to the DNA tails on either side of the break site and this may establish the asymmetry of the strand invasion process. (iii) Several important accessory proteins that mediate strand exchange in budding yeast have not yet been identified/characterized in plants. These include: Rad54, Rdh54/Tid1, the Rad52 epistasis group (Rad52, Rad55 and Rad57) and Hed1 (see text for details).

  3. *Proposed biochemical activity on the basis of studies on the corresponding proteins in other species. See text for further details.

AtRAD51/AtDMC1Homologues of the bacterial RecA protein catalyse strand exchange. AtRAD51 is present in mitotic and meiotic cells, whereas AtDMC1 is meiosis specific The recombinases load on to the 3′-ssDNA tails to form nucleoprotein presynaptic filaments which invade homologous duplex DNA to form stable joint moleculesAtrad51: Chromosome fragmentation at metaphase I. No chromosome synapsis Atdmc1: Absence of chiasmata; univalent chromosomes observed at metaphase I, but DSBs are repaired. No chromosome synapsis
AtRAD51CRequired for DSB repairChromosome fragmentation at metaphase I. No synapsis
AtXRCC3Required for DSB repairChromosome fragmentation at metaphase I. No synapsis
AtRPA1aRegulation of presynaptic filament assembly? Second-end capture? Yeast and mammals possess a single RPA large subunit gene, whereas five are found in Arabidopsis. AtRPA1a is present throughout prophase I of meiosis, but its loss does not appear to affect early recombination events, suggesting some redundancy with other AtRPA1 family members. Evidence suggests that AtRPA1a is essential at a later stage in recombination, most probably second-end captureReduced chiasma formation
AtBRCA2Recruitment of AtRAD51/AtDMC1 to nascent presynaptic filament. Role may be analogous to Rad52 in budding yeastUnivalents at metaphase I, some chromosome fragmentation and mis-segregation
AtMND1/AtHOP2Stabilization of presynaptic filament and promotion of duplex capture. Act as a complex, interaction demonstrated by yeast two-hybrid analysisFailure of chromosome pairing and synapsis. Defect in DSB repair

The activity of the strand exchange proteins is influenced by a number of accessory proteins. Replication protein A (RPA) is a heterotrimeric ssDNA-binding protein with multiple roles in DNA metabolism (reviewed in Wold, 1997), including meiotic recombination (Soustelle et al., 2002). Biochemical studies in yeast have demonstrated that the assembly of Rad51 filaments on ssDNA is modulated by RPA which can have opposing effects depending on the context (reviewed in San Filippo et al., 2008). It stimulates the assembly of the presynaptic filament by removing secondary structures within the DNA and sequestering the ssDNA formed during the strand exchange reaction (Sugiyama et al., 1997; Eggler et al., 2002; Van Komen et al., 2002). Conversely, prior saturation of the ssDNA with RPA, as is likely to be the case in vivo, strongly suppresses the ssDNA-dependent ATPase and recombinase activities of Rad51 and Dmc1 (Sung, 1997a,b; New et al., 1998; Shinohara & Ogawa, 1998; Yang et al., 2005; Haruta et al., 2006). The high binding affinity of RPA for ssDNA effectively excludes loading of the recombinases and must be overcome with the help of recombination mediators (see the following page) in order to assemble a functional presynaptic filament. RPA has also been found to have a later role in recombination, participating in Rad52-mediated DNA annealing during second-end capture (Wang & Haber, 2004; Sugiyama et al., 2006).

In budding yeast and mammals, the largest (c. 70-kDa) subunit of RPA is encoded by a single-copy, essential gene, RFA1 and RPA1, respectively. By contrast, plants contain multiple copies of this gene. Arabidopsis has five paralogues of RPA1 and two paralogues of the gene encoding the c. 32-kDa subunit, RPA2 (Shultz et al., 2007), whereas rice has been reported to have three copies of RPA1 (Ishibashi et al., 2006). A proteomics study implicated one of the Arabidopsis RPA1 paralogues, AtRPA1a, in meiosis, a role which was subsequently confirmed (Sanchez-Moran et al., 2005; Osman et al., 2009). Mutant lines exhibited a phenotype which suggested a role for AtRPA1a in second-end capture, but it was found to be dispensable for meiotic DSB repair, implying that it is not essential for RAD51 nucleoprotein filament assembly. Immunolocalization of AtRPA1a suggests that it is associated with meiotic chromosomes from leptotene until early pachytene, raising the possibility that normally, in wild-type cells, it may participate in both ‘early’ and ‘late’ stages of recombination but, in its absence, the early role may be carried out by one or more of the remaining AtRPA1 paralogues acting redundantly. In this context, it is interesting to note that T-DNA insertion mutants of each of the five AtRPA1 paralogues have been examined and none were associated with the severe reduction in fertility expected from a failure to carry out efficient RAD51-mediated meiotic DSB repair (Atrpa1a c. 30% seed set, other 4 mutant lines >90% seed set relative to wild-type) or, indeed, with any noticeable vegetative growth defect (K. Osman and F. C. H. Franklin, unpublished). This is consistent with two or more of the paralogues possessing a high degree of functional redundancy with respect to AtRAD51 filament assembly and, indeed, in essential somatic functions such as DNA replication. Taken together, these results imply that AtRPA1 protein choice in meiotic second-end capture demands a higher level of specificity than in presynaptic filament formation. This is reminiscent of observations in budding yeast, which indicated that the ssDNA-binding protein of Escherichia coli could largely substitute for RPA in presynaptic complex formation and strand exchange, but not in second-end capture (Sugiyama et al., 1997, 2006).

The efficiency of Rad51/Dmc1-mediated strand exchange is dependent on a number of accessory proteins. These include Rad54, Rdh54/Tid1 and the Rad52 epistasis group (Rad52, Rad55 and Rad57). Rad54 and Rdh54/Tid1 are members of the Swi2/Snf2 superfamily, possessing dsDNA-dependent ATPase, DNA translocase, DNA supercoiling and chromatin remodelling activities. Studies in budding yeast indicate that both proteins have multiple roles during homologous recombination and can enhance the recombinase activity of Rad51 (Symington, 2002). Rad54 has a more prominent role in promoting intrachromosomal and sister chromatid-based reactions in mitotic cells, whereas meiotic inter-homologue recombination events are more dependent on Rdh54/Tid1 (Klein, 1997; Shinohara et al., 1997; Arbel et al., 1999). rdh54 mutants exhibit severely reduced meiotic viability and rad54 rdh54 double mutants fail to form viable meiotic products or repair meiotic DSBs (Shinohara et al., 1997, 2003). Evidence suggests that Rdh54/Tid1 functions in the Dmc1-dependent pathway (Dresser et al., 1997; Shinohara et al., 2003). Recently, Hed1, a yeast meiosis-specific protein which suppresses inter-sister repair, has been shown to specifically interfere with the interaction of Rad54 with Rad51, whilst having little effect on the Rdh54–Rad51 interaction (Busygina et al., 2008).

The Arabidopsis SWI2/SNF2 family contains at least 40 members, including a RAD54-like gene (Shaked et al., 2006). Studies suggest that AtRAD54 is a genuine orthologue of the budding yeast RAD54 gene (Osakabe et al., 2006; Klutstein et al., 2008). Although the highest levels of expression of AtRAD54 are found in meiotic buds, plants lacking an intact gene were both viable and fully fertile. However, they exhibited increased sensitivity to DNA damage and reduced efficiency of somatic homologous recombination. Thus, it appears that, although AtRAD54 participates in DNA repair in somatic cells, there is currently no evidence of a meiotic role. No Arabidopsis functional equivalent of Rdh54/Tid1 has been identified to date. BlastP searches (http://blast.ncbi.nlm.nih.gov/Blast.cgi) of the Arabidopsis genome with Rad54 or Rdh54/Tid1 identified the same gene, AtRAD54, as the top hit (E = 1e−128 and E = 5e−104, respectively) (K. Osman and F. C. H. Franklin, unpublished). Apart from this gene, many other Arabidopsis SWI2/SNF2 family members share considerable sequence similarity with Rdh54/Tid1, but so far relatively few genes in the group have been functionally characterized.

The S. cerevisiae RAD55 and RAD57 genes are regarded as paralogues of RAD51 (reviewed in San Filippo et al., 2008). The proteins they encode form a stable heterodimeric complex which has ssDNA-binding activity and can interact with Rad51; although the complex itself has no recombinase activity, it appears to act as a recombination mediator, overcoming the inhibitory effect of RPA on Rad51-mediated DNA pairing and strand exchange (Sung, 1997a,b). BlastP searches of the Arabidopsis genome reveal that Rad55 shares most sequence homology with AtRAD51B (E = 6e−05) and Rad57 with AtRAD51C (E = 3e−16) (K. Osman and F. C. H. Franklin, unpublished).

Although the tumour suppressor protein BRCA2 is absent from yeast, there is clear evidence from mammalian studies that it has recombination mediator activity, recruiting RAD51 to the nascent presynaptic filament in a role analogous to that carried out by Rad52 in budding yeast (Thorslund et al., 2007; Jensen et al., 2010). In common with Drosophila and C. elegans, Arabidopsis has no obvious RAD52 orthologue. It does, however, possess two BRCA2-like paralogues which encode nearly identical proteins with four BRC repeat motifs, and RNAi analysis has confirmed a role for AtBRCA2 in homologous recombination-dependent repair of meiotic DSBs (Siaud et al., 2004). Analyses have demonstrated physical interactions of AtBRCA2 (either paralogue) with AtRAD51 and AtDMC1, suggesting that it may function in their recruitment (Dray et al., 2006).

AtBRCA2 also interacts with two Arabidopsis paralogues of the mammalian BRCA2 regulator, DSS1, in a paralogue-specific manner. Furthermore, ternary interactions have been demonstrated; AtBRCA2 can simultaneously bind AtRAD51 and AtDSS1(I) or AtDMC1 and AtDSS(I) (Dray et al., 2006), consistent with a role for AtDSS1 in enabling AtBRCA2 to localize the recombinases to DNA break sites, as appears to be the case in mammalian cells (Gudmundsdottir et al., 2004). In addition to its BRC motifs, human BRCA2 can also bind RAD51 at an unrelated site in its C-terminus (Mizuta et al., 1997; Sharan et al., 1997). The C-terminal region also contains a site (Ser-3291) which is phosphorylated by cyclin-dependent kinases in a cell cycle-dependent fashion, rendering it unable to bind to RAD51 without significantly affecting its association with DMC1 (Esashi et al., 2005; Thorslund et al., 2007). Although the C-terminus of AtBRCA2 does not appear to interact with AtRAD51 in vitro, it does contain potential CDK target motifs and cyclin-binding motifs and so may be subject to an analogous form of regulation (Dray et al., 2006), which would have obvious implications for meiotic recombination by providing a molecular ‘switch’ to inter-homologue repair.

VII. Promotion of stable strand exchange

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

The conserved MND1/HOP2 complex is essential for meiotic recombination, functioning with DMC1 and RAD51 to achieve the timely formation of DNA intermediates (reviewed in San Filippo et al., 2008). The complex appears to act in a bipartite fashion: in the stabilization of the presynaptic filament and in the promotion of duplex capture to enhance synaptic complex formation (Chi et al., 2007; Pezza et al., 2007). In S. cerevisiae, but not in mammals, the expression of MND1 and HOP2 is meiosis specific, and the complex appears to be active only in a Dmc1-dependent pathway, suggesting that the three proteins work together as a functional unit, whereas in vitro evidence from mammals indicates that the complex may also function with RAD51 (San Filippo et al., 2008). AtMND1 and AtHOP2 share a similar mutant meiotic phenotype: an inability of homologous chromosomes to pair and synapse, possible nonhomologous interactions and a failure of meiotic DSB repair (Schommer et al., 2003; Domenichini et al., 2006; Kerzendorfer et al., 2006; Panoli et al., 2006). Physical interaction of AtMND1 and AtHOP2 has been confirmed (Kerzendorfer et al., 2006). AtMND1 starts to localize to chromosomes in early leptotene in an AtHOP2-dependent manner. This is independent of recombination initiation and the establishment of sister chromatid cohesion. Moreover, there is no evidence of preferential loading at DSB sites (Vignard et al., 2007). The functional inter-relationships between the AtMND1/AtHOP2 complex and the Rec-A-like proteins, AtDMC1, AtRAD51 and AtXRRC3, have been examined (Vignard et al., 2007). Both AtMND1 and AtHOP2 physically interact with AtRAD51 and AtDMC1, synapsis is similarly defective in Atmnd1, Atrad51, Atdmc1 and Atxrcc3, and AtMND1, AtRAD51 and AtXRCC3 act during the same step of meiosis. Furthermore, in an Atdmc1 background, AtRAD51 and AtXRCC3 cooperate in sister chromatid-mediated DSB repair in a nonredundant manner. In the wild-type, AtDMC1 loading requires AtRAD51 and AtXRCC3. In an Atmnd1 mutant, AtDMC1 foci accumulate in an AtRAD51-dependent, but AtXRCC3-independent manner, thus revealing a functional divergence between AtRAD51 and AtXRCC3. Based on these results, Vignard et al. (2007) have proposed a model whereby AtRAD51 plays a crucial role in the assembly of AtDMC1-nucleoprotein filaments, with AtMND1/AtHOP2 acting subsequently to promote homology searches. Depletion of AtMND1 would result in a reduced turnover rate and corresponding accumulation of AtDMC1 foci. AtXRCC3 may have a stabilizing role, such that its absence would lead to rapid disassembly of the filaments and reduced numbers of AtDMC1 foci. That this model is inconsistent with several studies of RAD51 and XRCC3 recruitment in mammals underlines the need for further analysis in this area.

VIII. Pathways to crossover formation

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

Following strand exchange, a proportion of the recombination intermediates (see sections IX and X) progress to form COs. To ensure accurate chromosome disjunction, most species require at least one, ‘obligate’ CO per chromosome pair, often referred to as CO assurance (Jones, 1984; Jones & Franklin, 2006; Shinohara et al., 2008). In general, COs are nonrandomly distributed and multiple COs on the same chromosome pair are spaced apart. This phenomenon is referred to as CO interference, where one CO reduces the probability that a second will occur in an adjacent region (Jones & Franklin, 2006). Interference was first observed in Drosophila during the early 20th century and remains one of the most intriguing and controversial aspects of meiosis. A variety of models have been proposed to explain how interference is imposed. These include the ‘counting model’, in which COs are separated by a fixed number of NCOs (Stahl et al., 2004). Although this model and a slightly modified version account well for CO data from a variety of species, including Arabidopsis, experimental data from budding yeast have been proven to be inconsistent. A prediction of the counting model is that a reduction in the number of recombination initiation events (DSBs) should result in a coordinate reduction of COs and NCOs. However, Martini et al. (2006) observed a tendency to maintain COs at the expense of NCOs. This phenomenon, referred to as CO homeostasis, raises doubts about the counting model, at least in its current form. The mechanical stress model posits that the pattern of distribution of COs along chromosomes is dependent on the chromosomes undergoing programmed rounds of expansion and contraction. This is proposed to generate mechanical stress along the chromosome arms that is relieved by the formation of COs (Kleckner et al., 2004). Other models have also been proposed but are not discussed here because of space limitations. An excellent comprehensive discussion of interference can be found in an article by Berchowitz & Copenhaver (2010).

CO control in Arabidopsis is tightly regulated, so that each of the five chromosome pairs always receives an obligate CO and most multiple COs exhibit interference. The number of chiasmata per nucleus in wild-type Arabidopsis falls in the range 8–12 (Higgins et al., 2004). If the numbers of COs were randomly distributed among cells, there would be a range from three to 17 per cell which, as there are five chromosome pairs, would not ensure the obligate CO. A number of meiotic genes encode proteins that are essential to maintain the obligate CO and interference-sensitive COs. These belong to the class I recombination pathway and account for most meiotic COs. However, genetic studies and experimental evidence indicate that a second pathway of meiotic recombination may also exist. These class II COs do not exhibit interference or maintain the obligate CO (Copenhaver et al., 2002; Higgins et al., 2004).

IX. The class I pathway of meiotic recombination

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

A group of recombination pathway proteins, collectively referred to as ZMM (Zip1, Zip2, Zip3, Zip4, Msh4, Msh5 and Mer3), is essential for the formation of class I interference-sensitive COs and SC formation in budding yeast (Borner et al., 2004). Putative ZMM homologues have been identified in Arabidopsis (Table 4).

Table 4.   Proteins required for crossover (CO) formation
GeneProtein activity*/functionMutant phenotype
  1. dHj, double Holliday junction; PTD, PARTING DANCERS; SC, synaptonemal complex; SUMO, small ubiquitin-like modifier.

  2. *Proposed biochemical activity on the basis of studies on the corresponding proteins in other species. See text for further details.

(i) Class I crossover pathway
 AtMER3/RCKDNA helicase, thought to stabilize and stimulate DNA heteroduplex extensionStudies in different mutants affecting class I COs suggest that this pathway accounts for c. 85% of the total COs
 AtMSH4/AtMSH5Homologues of the bacterial MutS mismatch repair protein. Biochemical studies on hMSH4/MSH5 indicate that they form a complex that binds and stabilizes progenitor HjsReduced chiasma frequency, residual chiasmata are randomly distributed. Delay in prophase I progression
 AtZIP3Yeast Zip3 possesses SUMO-E3 ligase activity. May regulate SC polymerization to ensure it is contingent on homologue pairingReduced chiasma frequency
 AtZIP4In yeast, Zip4 functions in conjunction with Zip2 to promote polymerization of the SC transverse filament protein Zip1Reduced chiasma frequency, residual chiasmata are randomly distributed
 AtPTDUnknown, but some evidence to suggest that PTD is required to ensure that class I CO intermediates are resolved as COs (see text). Similarity to ERCC1 which forms a complex with XPF endonuclease that, in vitro, can cleave DNA structures resembling recombination intermediatesReduced chiasma frequency, residual chiasmata are randomly distributed
 AtSHOC1Putative XPF endonuclease. Interaction with PTD?Reduced chiasma frequency, residual chiasmata are randomly distributed
 AtZYP1a/AtZYP1bEncodes transverse filament protein of the synaptonemal complex. Required for controlled formation of crossoversAsynaptic. Slight reduction in COs. Nonhomologous COs occur. Duplicated genes, loss of either results in a fertility reduction of c. 30%. Orthologue in rice, ZEP1, also affects COs but, in this case, there appears to be an increase in CO frequency
 AtMLH3Thought to ensure that dHjs are resolved as COs. MutL homologue, based on evidence in other species AtMLH3 probably forms a heterodimer with AtMLH1Reduced CO formation
(ii) Class II crossover pathway
 AtMUS81Implicated in the formation of c. 30% of class II noninterfering COs. In vitro studies indicate that it can cleave nicked and intact HjsMinor reduction in CO formation, perhaps 5% of total

The MutS homologues, MSH4 and MSH5, play a key role in the promotion of CO formation in eukaryotes (Ross-Macdonald & Roeder, 1994; Zalevsky et al., 1999). In E. coli, the MutS proteins form a dimer that binds to mismatched nucleotides as part of the MutHLS system for mismatch repair (MMR) of DNA damage. However, Msh4/5 do not appear to function in MMR in yeast and lack the MutS I domain which is required to recognize DNA base mismatches (Ross-Macdonald & Roeder, 1994; Lamers et al., 2000). The MutS I domain is at the core of the dimer, and its absence is predicted to leave a hole large enough to encompass two DNA duplexes side by side (Obmolova et al., 2000). Further insight into the mechanism of the MSH4/5 dimer has come from in vitro analyses of hMSH4/5 (Snowden et al., 2004). In these assays, hMSH4/5 specifically bound to the cores of Holliday junctions (Hjs), and also to pro-Hj structures such as D-loops. The results are consistent with the hypothesis that hMSH4/5 initially stabilize a pro-Hj (single-end invasion) by a sliding clamp mechanism that embraces duplex DNA, enabling conversion into a dHj and resolution into either a CO or NCO (Snowden et al., 2004). In support of this, a yeast mutant lacking the Msh5 protein was found to be defective in the formation of single-end invasion intermediates, dHjs and COs (Borner et al., 2004).

In Arabidopsis, AtMSH4 and AtMSH5 were identified through their similarity to budding yeast and human sequences (Higgins et al., 2004; Higgins et al., 2008a). AtMSH4/5-specific antibodies co-localized on meiotic chromosome axes from leptotene to pachytene in a DSB-dependent manner. In leptotene, numerous AtMSH5 foci (c. 135) localized to the chromosomes. These decreased throughout zygotene until pachytene, where c. eight foci could be detected (Higgins et al., 2008b). These observations are in agreement with the hypothesis that a portion of recombination intermediates are directed for CO formation, whereas the others are required for pairing, chromosome alignment and synapsis initiation, and these are repaired as NCOs. In the absence of AtMSH4, AtMSH5 failed to localize to the chromosome axes and, although synapsis was completed, it was delayed. Most striking was a reduction from 9.86 chiasmata per cell in the wild-type to 1.25 (Atmsh4) and 1.15 (Atmsh5), indicating an 85% reduction in COs. This is reminiscent of yeast zmm mutants grown at 33°C, which exhibit a reduction in CO formation to c. 15% of the wild-type level (Borner et al., 2004). The ZMM-dependent COs are also subject to interference. The residual COs in Atmsh4 were not subject to interference and numbers were insufficient to ensure the obligate chiasmata. A recent study has found that normal localization of AtMSH4 is dependent on the retinoblastoma protein, RBR (Chen et al., 2011). Localization of AtDMC1 and AtRAD51 in an Arabidopsis rbr-2 mutant appears to be normal; hence, this may suggest that stable strand exchange is compromised in the absence of AtRBR. However, this has yet to be established.

In budding yeast, Mer3 is a ZMM protein epistatic to Zip1/2/3/4 and Msh4/5 (Borner et al., 2004). Mer3 is a DExH-box-type DNA helicase that unwinds duplex DNA in the 3′ to 5′ direction (Nakagawa et al., 2001). In vitro, Mer3 stimulates DNA heteroduplex extension in the direction relative to the incoming (or displaced) ssDNA (Mazina et al., 2004). Meiotic recombination-related DNA synthesis is reduced significantly in a mer3 strain compared with the wild-type (Terasawa et al., 2007). These data support the idea that the Mer3 helicase stabilizes Dmc1/Rad51-mediated nascent DNA heteroduplex molecules, thus allowing extension of the heteroduplex DNA.

In Arabidopsis, the Mer3 homologue, AtMER3 (also known as ROCK-N-ROLLERS [RCK]), is 51% similar to the budding yeast protein, and the loss of this gene led to a severe reduction in fertility, consistent with a meiotic role (Chen et al., 2005; Mercier et al., 2005). Similar to Atmsh4/Atmsh5, SC formation appeared to be normal in Atmer3/rck by light and electron microscopy (Chen et al., 2005; Mercier et al., 2005). Analysis of metaphase I chromosome spreads revealed a reduction in chiasmata from 9.2 to 2.25 (Mercier et al., 2005) and 9.83 to 3.10 (Chen et al., 2005) per cell. The residual chiasmata, although greater in number than in the Atmsh4/Atmsh5 mutants, were also shown to be interference insensitive (Chen et al., 2005; Mercier et al., 2005).

In the absence of Zip3, a small ubiquitin-like modifier (SUMO) ligase protein in budding yeast, SC assembles on chromosomes without meiotic recombination initiation and homologue pairing (MacQueen & Roeder, 2009). These data suggest that Zip3 regulates Zip1 mechanistically to ensure that SC assembly is contingent on earlier chromosomal events, such as homologue pairing. The Zip3 protein has been found to have SUMO E3 ligase activity that might regulate sumoylation of SC proteins, chromosomal axis components or discourage Zip1 stability and/or Zip1’s capacity to polymerize on meiotic chromatin (Agarwal & Roeder, 2000). Preliminary analysis of a putative AtZIP3 homologue (AT1G53490) identified by A. Barakate (University of Dundee, UK) suggests that AtZIP3 is epistatic to AtMSH4/5 in class I CO formation, but dispensable for SC formation (J. D. Higgins and F. C. H. Franklin, unpublished).

In budding yeast, Zip4 (also known as Spo22) functions with Zip2 to promote polymerization of Zip1 along chromosomes, and is epistatic to all ZMM proteins in CO formation (Tsubouchi et al., 2006). A bioinformatic analysis implicated Zip2/Zip3/Zip4 in ubiquitin labelling and protein–protein interactions, although no clear role for this protein has been determined (Perry et al., 2005). The Arabidopsis ZIP4 homologue, AtZIP4, was identified in a T-DNA insertional screen for meiotic mutants (Chelysheva et al., 2007). Analysis of the mutant revealed that AtZIP4 belongs to the same epistasis group as AtMSH4. Further-more, genetic analyses on two adjacent intervals of chromosome I established that the remaining COs in Atzip4 did not show interference. However, unlike budding yeast, polymerization of the SC central element protein AtZYP1 was not affected in an Atzip4 background, but initiation may have proceeded from fewer sites compared with the wild-type.

PARTING DANCERS (PTD) and Shortage in Chiasmata (SHOC1) are novel Arabidopsis genes epistatic to the Arabidopsis ZMMs (Wijeratne et al., 2006; Macaisne et al., 2008). AtPTD was identified as a gene with elevated expression in meiocytes (Wijeratne et al., 2006), whereas AtSHOC1 was found in a screen for plants with T-DNA insertions that showed reduced fertility and meiotic defects (Macaisne et al., 2008). The mean chiasma frequency was reduced from 9.7 in the wild-type to 2.5 in Atptd-2 and from 9.2 in the wild-type to 1.27 in Atshoc1-1 (Wijeratne et al., 2006; Macaisne et al., 2008). These studies concluded that AtPTD and AtSHOC1 were specifically required for class I CO formation in Arabidopsis. The AtSHOC1 protein showed weak similarity to a putative, previously unrecognized, XPF domain in the Zip2 protein, suggesting that AtSHOC1 and Zip2 could play similar roles in class I CO formation (Macaisne et al., 2008). Because single-end invasion is affected in zip2, AtSHOC1 may be involved in its promotion/stabilization (Macaisne et al., 2008). AtPTD encodes a protein with sequence similarity to ERCC1 proteins (Wijeratne et al., 2006). ERCC1 forms a complex with XPF that performs a single-strand 5′ incision in DNA nucleotide excision repair and participates in the cleavage of in vitro structures that resemble recombination intermediates (Sancar et al., 2004). AtPTD may act like ERCC1 as a noncatalytic subunit in a complex with another endonuclease to perform a meiotic function (Wijeratne et al., 2006). It is tempting to speculate that AtSHOC1 might fulfil the XPF catalytic function in a complex with AtPTD. This may be similar to the role of Mei-9 and ERCC1 in Drosophila meiotic CO formation (Sekelsky et al., 1995; Radford et al., 2005). Interestingly, the numbers of late recombination nodules shown by electron microscopy were not statistically significantly different in Atptd (c. 8) compared with the wild-type (c. 9) (Wijeratne et al., 2006). Late nodules mark the sites of future COs, but, in Atptd, it appears that the class I CO sites are not being resolved as COs. It is argued that AtPTD could be involved in dHj resolution and that, in the absence of AtPTD, recombination intermediates destined to form COs are repaired as NCOs (Wijeratne et al., 2006).

The remaining ZMM protein Zip1 encodes the transverse filament of the SC (Fig. 1) (Sym et al., 1993). It comprises an extended coiled-coil region flanked by globular domains at the N- and C-termini. Unlike some of the other ZMM proteins, the protein is only poorly conserved in different species at the level of the primary amino acid sequence. At first sight, this was rather surprising as ultrastructural studies suggest a high degree of similarity in SC structure in different species. Subsequently, following isolation of Zip1 orthologues from a range of species, it is now apparent that, despite their primary sequence divergence, the proteins have conserved physicochemical properties and domain structures. Arabidopsis contains two Zip1 orthologues, AtZYP1a and AtZYP1b, that are located within 2 kb of each other on chromosome 1 (Higgins et al., 2005). Analysis of T-DNA insertion lines suggests that they are functionally redundant, although a knockout of either gene results in a reduction in fertility of c. 30% compared with the wild-type. Immunolocalization studies indicate that AtZYP1 localizes as small foci to numerous sites on chromatin during leptotene before SC polymerization. At zygotene, as synapsis initiates, the AtZYP1 signal begins to linearize from c. 30 sites, extending along and between the axes of the homologues, bringing them into close apposition. At pachytene, the protein forms a continuous linear signal along the length of the synapsed homologues. Although Zip1 was initially investigated for its role in SC formation, it is now clear that, through its function as a ZMM protein, it plays an important role in CO formation in budding yeast (Storlazzi et al., 1996; Borner et al., 2004). The finding that AtZYP1 foci appeared at the same time as AtMSH4/AtMSH5 in leptotene, well before SC formation, suggested that it may also have a role in recombination (Higgins et al., 2005). This appears to be the case, but, unlike budding yeast, where the loss of Zip1 results in a substantial reduction in CO formation, Arabidopsis lines lacking AtZYP1 show a milder effect on chiasma frequency. However, extensive aberrant recombination occurs, resulting in the formation of multivalents, homologous and nonhomologous bivalents, and univalents at metaphase I. This indicates that the loss of the protein has resulted in a loss of CO control. Interestingly, analysis of rice plants lacking the Zip1 orthologue ZEP1 revealed a marked increase in chiasma frequency, again suggesting that the protein plays a role in the controlled formation of COs (Wang et al., 2010). Thus, despite these different effects, it seems likely that, in addition to their contribution to the structural organization of the SC, these plant Zip1 orthologues also perform an important function in the control of meiotic recombination.

The prokaryote MMR MutHLS system is essential for the preservation of genome integrity by contributing to post-replicative DNA repair and impeding recombination between divergent sequences (Dion et al., 2007). The MutS dimer recognizes and binds to DNA base mismatches and then recruits the MutL dimer. In this conformation, the active MutH protein nicks the newly synthesized strand, facilitating its removal and insertion of the correct base by de novo DNA synthesis. The eukaryotic MutL homologues (MLH) (sometimes named PMS for post-meiotic segregation) have evolved to play crucial roles in both DNA MMR and meiotic recombination. The corresponding proteins (MLH2/3, PMS1/2) form heterodimers that are always anchored by an MLH1 monomer (Li & Modrich, 1995; Raschle et al., 1999; Lipkin et al., 2000; Kadyrov et al., 2007). The human MutLα (MLH1–PMS2 heterodimer) harbours a latent endonuclease that is dependent on the integrity of a PMS2 metal-binding DQHA(X)2E(X)4E motif (Kadyrov et al., 2006, 2007). Amino acid substitutions within the DQHA(X)2E(X)4E motif of yeast Pms1 inactivated the endonuclease activity and conferred high mutation rates (Kadyrov et al., 2007). In yeast, mutations in the Mlh3 putative endonuclease domain motif conferred mlh3-null-like defects with respect to meiotic spore viability and crossing over (Nishant et al., 2008). The Mlh3 endonuclease activity could be important for dHj resolution to promote CO products during meiotic recombination (Nishant et al., 2008).

In Arabidopsis, both AtMLH1 and AtMLH3 have a role in meiosis. AtMLH1 is also required for recombination in somatic cells (Dion et al., 2007). The role of AtPMS1 appears to be restricted to somatic cells, where it is involved in DNA MMR and the restriction of recombination between homeologous sequences (Jean et al., 1999; Alou et al., 2004; Jackson et al., 2006; Dion et al., 2007; Li et al., 2009). AtMLH3 was identified by sequence homology compared with yeast and mouse (Jackson et al., 2006). In wild-type pachytene nuclei, AtMLH1 and AtMLH3 formed c. nine foci that co-localized. Immunocytological studies indicated that chromosome pairing and synapsis proceeded normally in an Atmlh3 mutant. Localization of AtMSH4 also occurred, suggesting that dHjs are formed, but, in the absence of AtMLH3, AtMLH1 fails to localize normally. Loss of AtMLH3 resulted in a c. 60% reduction in chiasmata, suggesting that the outcome of dHj resolution was biased in favour of a non-CO outcome by a ratio of c. 2 : 1. The data are compatible with a model in which the MutL complex imposes a dHj conformation that ensures CO formation (Franklin et al., 2006). A bromodeoxyuridine (BrdU) pulse labelling experiment revealed that, as a result of the Atmlh3 mutation, prophase I was extended by 25 h compared with the wild-type (Jackson et al., 2006). These data suggest that a checkpoint surveillance system is monitoring DSB repair in Arabidopsis meiosis.

AtMLH3 also contains the metal-binding motif (DQHAADERIRLE) at 944–955 residues, but it is absent in the noncatalytic AtMLH1 protein. It is tempting to speculate that this motif, required for endonuclease activity and wild-type levels of COs in yeast, is essential to impose the asymmetrical nicking of a dHj to ensure resolution into a CO.

An analysis of the Atrpa1a mutant (see section VI) revealed that AtRPA1a plays an essential role in the formation of class I COs (Osman et al., 2009). The mean chiasma frequency in Atrpa1a was found to be 3.98 per cell, compared with 9.86 in the wild-type. In a double Atmsh4/Atrpa1a mutant, the chiasma frequency was reduced to 1.08, indistinguishable from that of the Atmsh4 single mutant, suggesting that two to three residual chiasmata in the Atrpa1a mutant were from the class I pathway. An Atrpa1a/Atmlh3 double mutant gave a mean chiasma frequency of 1.76, suggesting that c. two residual chiasmata in the Atrpa1 single mutant were dependent on AtMLH3. These data suggest that AtRPA1a is required after AtMSH4 and before AtMLH3 in the class I pathway. In vitro biochemical studies of the RPA protein from the budding yeast mutant rfa1-t11 have revealed that the protein is defective in promoting Rad52-mediated strand annealing and second-end capture following D-loop formation (Sugiyama et al., 2006). It is therefore conceivable that strand annealing and second-end capture are substantially affected by the loss of AtRPA1a during meiosis in Arabidopsis.

X. The class II pathway of meiotic recombination

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

In budding yeast, c. 15% of COs form via a second recombination pathway (de los Santos et al., 2003; Borner et al., 2004). These class II noninterfering COs are dependent, at least in part, on the Mus81–Mms4 (Eme1 in fission yeast) heterodimer (de los Santos et al., 2003). In S. pombe (fission yeast), Mus81–Eme1 is required for most, if not all, meiotic COs, which, interestingly, do not exhibit CO interference (Smith et al., 2003).

The existence of two pathways of meiotic recombination in Arabidopsis was initially inferred from genetic data (Copenhaver et al., 2002). This was supported by the analysis of an Atmsh4 mutant (see section IX), which revealed that c. 15% of the chiasmata/COs were not AtMSH4 dependent (Higgins et al., 2004). Their numerical distribution fitted a Poisson distribution, implying that they were not subject to CO interference. They ranged in number from zero to seven, with a mean of 1.56 and the major class being one chiasma per cell, and thus were not sufficient to ensure the obligate CO. These data were supported by the analysis of the other Arabidopsis ZMM proteins.

In Arabidopsis, AtMUS81 was originally investigated for its role in somatic DNA repair (Hartung et al., 2006). A biochemical analysis of the AtMUS81–AtEME1 heterodimer, which has two EME1 homologues (EME1a&b), showed that they cleaved intact and nicked Hjs (Geuting et al., 2009). The meiotic role of AtMUS81 was studied using genetic and cytological methods (Table 4) (Berchowitz et al., 2007; Higgins et al., 2008a). An AtMUS81 antibody revealed that protein loading was dependent on DSB formation, concomitant with AtRAD51, before numbers of foci decreased as prophase I progressed (Higgins et al., 2008b). In contrast with AtMLH1/3, that mark class I recombination sites, AtMUS81 seems to associate with every recombination site, yet may only be functionally required at very few of these. Conceivably, AtMUS81 may not be unique in this respect and it could apply to other proteins, thereby ensuring that recombination intermediates can be processed in an efficient and timely manner. There were no apparent defects in reproductive development in Atmus81, but an Atmsh4/Atmus81 double knockout revealed a statistically significant reduction in COs from 1.25 in an Atmsh4 single mutant to 0.8 chiasmata per cell. A moderate decrease in COs was found in genetic intervals on chromosomes 1 and 3 in an Atmus81 background (Berchowitz et al., 2007). Therefore, AtMUS81 is not solely responsible for the generation of COs in the class II pathway, suggesting that other proteins are required.

Studies in budding yeast indicate that mutation of MUS81 in combination with a mutation in the RecQ family helicase SGS1 is synthetically lethal (Fabre et al., 2002; Bastin-Shanower et al., 2003). Using conditional mutants, in which expression of the genes was inactivated during meiosis, it was shown that their loss resulted in the accumulation of aberrant joint molecule recombination intermediates (Jessop et al., 2006; Oh et al., 2008). It is proposed that these arise from secondary strand invasion events. In wild-type cells, their formation is antagonized by Sgs1 and, in cases in which this fails, Mus81–Mms4 acts as a back-up to resolve these structures (Oh et al., 2008). Aberrant joint molecules that may be formed from invasion of the second end of a DNA DSB are therefore resolved as COs/NCOs in the wild-type by Mus81 or Sgs1 (Oh et al., 2008). These data suggest that the class II CO pathway is necessary to repair aberrant joint molecule structures generated by the class I pathway. Although it is likely that a similar mechanism may operate in Arabidopsis, there is currently no direct evidence. Moreover, despite the fact that the Arabidopsis RecQ helicase, AtRECQ4A, possesses many of the features of an Sgs1 homologue (Hartung et al., 2007a), it does not appear to be required for meiotic recombination (Higgins et al., 2011). However, this may reflect functional redundancy of the Arabidopsis RecQ protein family (Hartung & Puchta, 2006).

XI. Holliday junction (Hj) resolution

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

Despite the fact that the role of the E. coli Hj resolvase, RuvC, is long established (West, 1997), the identification of the corresponding eukaryotic activity has proved far from straightforward. Several proteins have been shown to possess Hj resolvase activity in vitro, but establishing their contribution in vivo has proved challenging. Mus81 (see section X), which is required for CO formation in fission yeast, is related to the XPF subunit of the ERCC1–XPF family of heterodimeric endonucleases that resolve Hjs (Boddy et al., 2001). The fission yeast mus81 mutant phenotype was rescued by a bacterial Hj resolvase RusA. However, Mus81 does not appear to be required for the formation of most COs in budding yeast and Arabidopsis. It seems that Hjs are a poor substrate for the Mus81–Mms4 complex found in budding yeast (Osman et al., 2003; Hollingsworth & Brill, 2004). It has long been known that mammalian nuclear extracts contain a Hj resolvase activity, ResA; however, it is only recently that technical advances in protein isolation have enabled the identification of the active component. Analysis of nuclear extracts from HeLa cells using mass spectrometry and tandem affinity purification from budding yeast extracts has resulted in the identification of GEN1 and Yen1, respectively (Ip et al., 2008). These orthologues promote Hj resolution in a manner analogous to the bacterial RuvC protein. Yen1 was found to act redundantly with Mus81 in budding yeast (Tay & Wu, 2010), and the expression of GEN1 in fission yeast mus81 mutant strains resulted in Hj resolution and CO formation during meiosis (Lorenz et al., 2010).

Hj resolution in plants has yet to be elucidated. Arabidopsis has two GEN1 homologues that are highly expressed in bud tissue. However, any contribution to Hj resolution during meiosis has yet to be established, particularly as a double knockout line remains fertile, suggesting possible redundancy with other proteins (J. D. Higgins and F. C. H. Franklin, unpublished). Although XRCC3 and RAD51C, both of which have been identified in Arabidopsis (see section VI) (Bleuyard & White, 2004; Bleuyard et al., 2005), were found to co-purify with ResA from mammalian extracts, subsequent analysis of GEN1 indicated that they are not essential for Hj resolution (Ip et al., 2008).

It has been speculated that the XPF–ERCC1-related AtSHOC1 and AtPTD proteins (see section IX) combine to form a complex with a role in the resolution of dHjs to form class I COs (Macaisne et al., 2008). However, this has not been confirmed.

XII. Noncrossover pathways and the crossover/noncrossover decision

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

Evidence in budding yeast suggests that strand invasion is a key control step in the repair of DSBs as CO products or NCOs (Borner et al., 2004). In Arabidopsis, meiotic recombination is initiated by c. 150 DSBs, c. 10 of which are repaired as COs and c. 140 as NCOs (Sanchez-Moran et al., 2007). To ensure a CO on each chromosome pair, DSB sites must be nonrandomly selected to mature into COs, and the remaining sites enter the NCO pathway. Studies in budding yeast suggest an early time point for the CO/NCO decision, at or just before the formation of the stable single-end invasion intermediate. This is based on the analysis of the zmm mutants, which were found to form DSBs and NCOs at normal levels at 33°C, but were defective in the formation of single-end invasion intermediates, dHjs and COs (Borner et al., 2004). It is proposed that most, if not all, NCOs arise via synthesis-dependent strand annealing (SDSA), which occurs as a bifurcation in the recombination pathway, and that the intermediates that progress to form dHjs are resolved as COs (Allers & Lichten, 2001). This has led to the development of the early decision model of meiotic recombination (Fig. 2).

As many aspects of meiotic recombination appear to be similar in plants and budding yeast, it is tempting to speculate that the CO/NCO decision also occurs at an early stage in plants. Nevertheless, direct evidence is lacking; moreover, cytological analysis of meiocytes from Arabidopsis Atrmi1/blap75 and topoisomerase3α (Attop3α) mutants indicates that these proteins may promote the dissolution of dHjs as NCOs via a hemicatenane intermediate (Fig. 2) (Chelysheva et al., 2008; Hartung et al., 2008). This suggests that, in Arabidopsis, a proportion of the dHjs may be removed via this route. One possibly relevant factor is that, in yeast, a high proportion (50%) of DSBs progress to form COs. It is thought that the ZMM-containing CO pathway complexes are the sites at which SC polymerization initiates (Fung et al., 2004). In most other species, including Arabidopsis, the proportion of COs relative to DSBs is much lower (5–10%). Yet, immunostaining in Arabidopsis meiocytes suggests that the number of AtMSH4 foci associated with stretches of SC at zygotene is c. three-fold higher than the number of COs (c. 30 vs 10), presumably to ensure timely chromosome synapsis. Assuming that the AtMSH4 foci are indicative of joint molecule recombination intermediates, all or some of which form dHjs, it is conceivable that AtRMI1/BLAP75/TOP3α is a crucial route for the removal of those that do not form COs. However, further studies are required to confirm whether this is the case, or whether the complex is simply a back-up to remove dHjs that fail to be resolved by the normal route.

XIII. Conclusions

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information

The past decade has seen a substantial increase in our understanding of meiotic recombination in Arabidopsis and, increasingly, in plants such as maize and rice. These studies have highlighted that, as in mammals, plant meiosis has much in common with budding yeast, which remains the best studied system. Nevertheless, it is clear that there are differences in both the proteins involved and the aspects of control, which require further analysis and highlight the importance of studying meiosis in a wide range of systems. Although Arabidopsis will remain a key experimental system for the resolution of many fundamental questions, it seems likely that there will also be an increasing trend towards practical application by translating the knowledge from this model to crop species. For example, in cereals and forage grasses, the distribution of COs is heavily skewed to the ends of the chromosomes, leaving large regions recombinationally silent. Although these regions are gene rich, a significant proportion of genes reside elsewhere (Mayer et al., 2009). Hence, increasing the recombination frequency or changing the CO distribution should enable genetic variation to be more readily exploitable and reduce the time required to introgress new characters.

Despite much progress, it is clear from this review that, in plants, as in other systems, many important aspects of meiosis remain poorly understood. Although, in many organisms, recombination is crucial for chromosome pairing, the early stages of homologue recognition are still obscure. The basis of CO assurance and interference remains one of the key challenges in the study of meiosis. One substantive area of progress in yeast studies is the establishment of the timing of the CO/NCO decision, which seems to occur at an early stage in the recombination pathway, at around the time of strand exchange. Nevertheless, how this is executed remains unknown. Moreover, although it appears that dHjs are predominantly resolved to form COs in budding yeast, this may not be the case in all species, including plants. One area of significant progress in yeast and mammals has been the high-resolution mapping of recombination hot-spots and the emerging knowledge about the role played by histone marks in defining them (e.g. Kauppi et al., 2004; Arnheim et al., 2007; Mancera et al., 2008; Baudat et al., 2010). Although plant studies are not as advanced, recent progress in Arabidopsis by Mezard and colleagues promises to rectify this in the near future (Drouaud et al., 2006, 2007; Mezard, 2006).

References

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. The meiotic pathway: a brief overview
  5. III. Homologous chromosome pairing and movement during prophase I
  6. IV. Meiotic DNA double-strand break formation
  7. V. Processing of DNA double-strand breaks
  8. VI. Strand exchange: the role of the RecA homologues and their accessory proteins
  9. VII. Promotion of stable strand exchange
  10. VIII. Pathways to crossover formation
  11. IX. The class I pathway of meiotic recombination
  12. X. The class II pathway of meiotic recombination
  13. XI. Holliday junction (Hj) resolution
  14. XII. Noncrossover pathways and the crossover/noncrossover decision
  15. XIII. Conclusions
  16. Acknowledgements
  17. References
  18. Supporting Information