Meiosis is a key stage in the life cycle of all sexually reproducing eukaryotes. In plants, specialized reproductive cells differentiate from somatic tissue. These cells then undergo a single round of DNA replication followed by two rounds of chromosome division to produce haploid cells that then undergo further rounds of mitotic division to produce the pollen grain and embryo sac. A detailed cytological description of meiosis has been built up over many years, based on studies in a wide range of plants. Until recently, comparable molecular studies have proved too challenging, however, a number of groups are beginning to use Arabidopsis thaliana to overcome this problem. A range of meiotic mutants affecting key stages in meiosis have been identified using a combination of screening for plants exhibiting reduced fertility and, more recently, using a reverse genetics approach. These are now providing the means to identify and characterize the activity of key meiotic genes in flowering plants.
Received 12 June 2002; Accepted 18 September 2002
The life cycles of all sexually reproducing eukaryotes are dependent on the process of meiosis. This combines a single round of DNA replication with two rounds of chromosome division to produce haploid gametes and to allow the process of genetic recombination to occur producing novel combinations of genetic information. A co‐ordinated series of events during prophase I lead to homologous chromosome pairing, synapsis to form bivalents and recombination between non‐sister chromatids within the bivalent. Reductional division then occurs separating the homologous chromosomes during anaphase I. The second meiotic division is equational and separates the sister chromatids of each chromosome to give rise to the haploid gametes necessary for sexual reproduction.
Meiosis in flowering plants has long been a subject of research and has led to the identification of many meiotic mutants (Baker et al., 1976; Golubovskaya, 1979; Koduru and Rao, 1981; Kaul and Murthy, 1985). Until recently, it has not been possible to identify and characterize the genes affected in these mutants at a molecular level. However, this has begun to change with the identification of a number of meiotic mutants from Arabidopsis thaliana, which have provided the means to clone a range of genes. Analysis of these genes together with the development of improved cytogenetic and immunocytological techniques for Arabidopsis (Ross et al., 1996; Armstrong and Jones, 2001, 2003) are permitting more complete analyses of meiotic chromosome behaviour to be undertaken. In this article, recent progress in the isolation of meiotic mutants in Arabidopsis and the contribution this is making to the study of meiosis in plants is reviewed.
Meiosis in wild‐type Arabidopsis thaliana: a brief outline
The development of the male (anther) and female (ovule) reproductive structures from somatic tissue initiate the basic process of reproductive development in Arabidopsis thaliana.
In male development, a group of cells within the developing anther differentiate into archesporial cells. These then give rise to the primary sporogenous cells which differentiate into the pollen mother cells (PMCs) in which meiosis occurs and primary parietal cells which give rise to the tapetum, endothecium and the middle layer of the anther (Yang et al., 1999a). Each anther locule contains around 30 PMCs, which enter and proceed through meiosis with a high degree of synchrony (Armstrong et al., 2001). Meiosis in the PMC produces a tetrad of haploid microspores, which are then released from a common wall and go on to develop into the male gametophytes (pollen grains).
In the female reproductive tissues an archesporial cell at the top of the ovule primordium differentiates from a single hypodermal cell. This archesporial cell directly differentiates into the megaspore mother cell (MMC). Meiosis in the MMC results in the formation of four haploid megaspores. The three spores closest to the micropyle of the ovule undergo programmed cell death leaving the chalazal megaspore to develop into the female gametophyte (the embryo sac) (Yang et al., 1999a). It is of interest to note that in common with many other angiosperms there is a degree of asynchrony between male and female meiosis, such that by the time male meiosis is complete, female meiosis is still at prophase I (Armstrong and Jones, 2001).
The development of improved cytogenetic techniques using modified chromosome spreading and DAPI (4′,6‐diamidino‐2‐phenylindole) staining procedures have enabled the behaviour of the chromosomes during meiosis to be examined in considerable detail (Ross et al., 1996, 1997; Armstrong and Jones, 2001). Following replication and meiotic interphase the diffuse chromosomes begin to condense, establishing short stretches of chromosome axes that become discernible during the leptotene stage of prophase I. Telomere pairing is also initiated during leptotene (Armstrong et al., 2001), progressing into zygotene where the pairing of homologous chromosomes is first clearly observed. At pachytene chromosome synapsis is complete (Fig. 1a). During diplotene/diakinesis the homologous chromosome pairs condense further and desynapse leading to the visualization of discrete bivalents linked by one to three chiasmata (Ross et al., 1996), the sites of physical cross‐over that arise from the events of homologous recombination during prophase I (Fig. 1c). As meiosis proceeds from prophase I into metaphase I the highly condensed bivalents become attached to the meiotic spindle and align at the equator of the cell with their centromeres directed to opposite poles. At anaphase I homologous chromosomes migrate to the opposite poles of the cell. At telophase I (the dyad stage) the two chromosome groups partially decondense before the second meiotic division begins. The chromosomes recondense at prophase II and at metaphase II become aligned prior to the second division (Fig. 1h). At this point sister chromatid cohesion has broken down in the chromosome arms thus permitting pairs of chromatids to be observed. The chromatids are separated at anaphase II to produce four tetrad nuclei each containing a haploid chromosome content (Fig. 1j). Cytokinesis then occurs producing four haploid cells (Ross et al., 1996; Armstrong and Jones, 2001).
Identification of meiotic mutants in Arabidopsis
An extensive range of meiotic genes have been identified in a variety of organisms, notably budding and fission yeast (reviewed in: Roeder, 1997; Zickler and Kleckner, 1999; Dresser, 2000; Davis and Smith, 2001). Hence, one might anticipate that it should easily be possible to identify homologues of these genes in Arabidopsis. To some extent this is indeed the case, particularly for genes that encode proteins catalysing homologous recombination. For example, DMC1 (Sato et al., 1995; Klimyuk and Jones, 1997; Doutriaux et al., 1998), RAD51 (Doutriaux et al., 1998), and SPO11 (Hartung and Puchta, 2000; Grelon et al., 2001). However, it is clear that this is only possible for a subset of meiotic genes, as in many cases homologues cannot readily be identified in silico. This seems to apply particularly to genes encoding proteins involved in chromosome morphogenesis such as synapsis of homologous chromosomes. For instance, the yeast ZIP1 and rat SCP1 genes both encode proteins that comprise the transverse filaments of the synaptonemal complex. Yet, whilst they appear functionally and structurally similar they exhibit no primary sequence homology beyond that expected between two coiled‐coil proteins (Heyting, 1996). Inspection of the Arabidopsis genome for genes encoding homologues to either or both proteins reveals a range of candidates with little more than 20% sequence identity (AP Caryl, NP Jackson, FCH Franklin, unpublished data). An added complication in assigning gene function on the basis of homology arises from the finding that the Arabidopsis genome has undergone an ancient duplication (The Arabidopsis Genome Initiative, 2000). Around 60% of the Arabidopsis genome is duplicated and in many cases the genes within these duplicated regions have undergone functional divergence. Thus, even in cases where putative meiotic genes can be identified on the basis of homology there can be several potential candidates whose functions have to be established.
Both these problems may be addressed using mutation. To date, most emphasis and progress has been made as the result of screens aimed at the de novo identification of meiotic mutants and subsequent gene isolation, which will be the main focus of the rest of this article.
Meiotic mutants in Arabidopsis have been identified by screening EMS (ethyl methanesulphonate), T‐DNA and transposon mutagenized populations for plants exhibiting reduced fertility or sterility (Ross et al., 1997; Bhatt et al., 1999; Sanders et al., 1999; Siddiqi et al., 2000; Mercier et al., 2001a). Several studies based on this approach, summarized in Table 1, have resulted in the isolation of a range of meiotic mutants. In one case, as an additional screening step the inheritance of the ‘reduced‐fertility’ trait was checked over several generations. The object was to ensure that the mutant phenotype did not segregate as fertile/reduced fertility, as this would be indicative of reduced fertility caused by heterozygosity for chromosomal rearrangement rather than by meiotic gene mutation (Ross et al., 1997). Before any putative meiotic mutant may be confirmed it is essential that it is subjected to a thorough cytogenetical examination. This requirement arises from the fact that although a screen based on a reduced fertility phenotype will indeed identify meiotic mutants, a wide range of other mutants defective in various aspects of reproductive development may be expected to possess a very similar phenotype. This has proved to be the case with mutants affecting a range of processes from sporocyte initiation (Yang et al., 1999a) to anther dehiscence (Sanders et al., 1999) being identified.
Identification of meiotic genes using reverse genetics
An alternative approach to screening for meiotic mutants representing unknown genes is to generate or identify mutations in known genes which either show homology to meiotic genes from other species or have an expression pattern suggestive of a meiotic role. As discussed above, some Arabidopsis meiotic genes will have homology to known meiotic genes from other species. In many instances this will be reflected in an essentially identical biological function. However, since there may be exceptions it will be important to identify and characterize mutant alleles of these genes.
This is exemplified by the ArabidopsisASY1 gene which, despite exhibiting some homology to the yeast HOP1 gene, appears to exhibit differences when examined at a functional level (Hollingsworth et al., 1990; Smith and Roeder, 1997; Kironmai et al., 1998; Caryl et al., 2000; Armstrong et al., 2002).
Although expression of many Arabidopsis meiotic genes can be detected in somatic tissues, some such as DMC1, are up‐regulated during meiosis (Klimyuk and Jones, 1997) or possess meiosis‐specific splice variants such as SYN1/DIF1 (Bai et al., 1999). This suggests that it may be possible to identify genes with meiotic roles by expression profiling using microarray technology.
Identification of knockouts in genes identified on the basis of sequence homology or expression pattern can be done by screening the sites of T‐DNA and transposon insertions in individual lines from tagged populations. These include the Feldmann (Feldmann and Marks, 1987), INRA‐Versailles (Bechtold et al., 1993), SALK (Salk Institute Genomic Analysis Laboratory, http://signal.salk. edu) and SAIL (Syngenta Arabidopsis Insertion Library, http://www.nadii.com/pages/collaborations/garlic_files/GarlicDescription.html) T‐DNA insertion lines and the IMA (Parinov et al., 1999) and SLAT (Tissier et al., 1999) transposon insertion lines. Several approaches can be used to screen these tagged populations. PCR using gene‐ and tag‐specific primers can be performed on DNA pools made from the tagged population. Alternatively, inverse PCR (IPCR) can be used to amplify plant DNA adjacent to the tag which can then be blotted and screened by hybridization to gene‐specific probes. More recently, IPCR generated tags have been systematically sequenced to provide a sequence database that can be queried for knockouts in a gene of interest. An approach has also been developed to allow high throughput screening of chemically mutagenized DNA. In denaturing high power liquid chromatography (DHPLC) plants are treated with a chemical mutagen such as EMS. The gene of interest is then amplified by PCR, and the resulting products resolved by HPLC where point mutations can easily be identified (McCallum et al., 2000). The identification of gene knockouts in Arabidopsis is reviewed in greater detail in Bouché and Bouchez (2001).
If a particular gene knockout cannot be identified it is possible to generate one. Directed gene disruption by homologous recombination, which is widely used in yeast and mouse would, in theory, be the best approach, but, unfortunately, this is not yet a realistic option in plants (Mengiste and Paszkowski, 1999). However, several alternative methods do exist, collectively known as post‐transcriptional gene silencing; these include antisense RNA, cosuppression and RNAi. These phenomena depend on the generation or introduction of double‐stranded RNA based on the target gene sequence. This results in the degradation of the corresponding mRNA, and in some cases the methylation of the nuclear DNA with the result that gene expression is reduced or eliminated (reviewed in Bourque, 1995; Wang and Waterhouse, 2001; Hannon, 2002).
Meiotic mutants: from phenotype to gene
A variety of Arabidopsis meiotic mutants have now been isolated. In a number of cases the corresponding gene has been cloned and further characterization carried out. In the following section current progress is reviewed. In general, an attempt was made to categorize the mutants on the basis of their meiotic phenotype. In some cases, this has been made difficult through a lack of data or a complex meiotic phenotype, hence in these cases the categorization may need to be re‐evaluated when relevant new data emerges.
Newly replicated sister chromatids are associated from their formation at meiotic S phase through to their disjunction at anaphase II. Sister chromatid cohesion is maintained by a multi‐protein cohesin complex (reviewed in van Heemst and Heyting, 2000; Lee and Orr‐Weaver, 2001). At least two Arabidopsis meiotic mutants have been described that are proposed to play a role in establishing or maintaining sister chromatid cohesion.
Three mutant alleles of the SWITCH1(SWI1)/DYAD gene have been isolated. All three alleles show similar defects in female meiosis. Ten univalents are observed at metaphase I which segregate evenly into two daughter cells (a mitosis‐like division). Agashe et al. (2002) report that in the dyad mutant GUS (β‐glucuronidase) expression is driven from a DMC1 promoter during female meiosis suggesting that this division may be meiotic rather than mitotic in nature. Subsequently, these cells can undergo either a second mitosis like division, or an aberrant second division with uneven chromosome segregation giving rise to a variable number (1–5) of daughter cells (Motamayor et al., 2000; Siddiqi et al., 2000; Mercier et al., 2001a). These cells then degenerate with no embryo sac produced in the mutant.
Male meiosis is unaffected in swi1‐1 and dyad (Motamayor et al., 2000; Siddiqi et al., 2000), but severely affected in swi1‐2. No true leptotene, zygotene or pachytene stages are observed in swi1‐2, and no evidence of synapsis is seen. Ten univalents are initially observed during meiosis I followed by a premature loss of sister chromatid cohesion. This begins along the chromosome arms and later at the centromeres resulting in the production of 20 chromatids before the end of meiosis I. Chromosome condensation is also seen to be incomplete/reduced. Uneven segregation of the chromatids during the first and second meiotic divisions results in the production of 1–10 unbalanced nuclei which subsequently degenerate (Mercier et al., 2001a).
The swi1‐1 allele is not completely penetrant as 3% of the wild‐type number of embryo sacs are observed (Motamayor et al., 2000; Mercier et al., 2001a). Heterozygotes between swi1‐1 and swi1‐2 show an intermediate male meiosis phenotype (Mercier et al., 2001a). A male‐sterile, female‐fertile mutant, ms4 identified by Chaudhury et al. (1994) also results in the production of dyads, which persist until late stages of anther development then degenerate.
The SWI1 gene produces two alternatively spliced transcripts encoding proteins of 578 and 635 amino acids, respectively, differing only at the N termini. The protein shows no clear homology to any other sequence in the database (Mercier et al., 2001a). The swi1‐1 allele, which affects only female meiosis, was caused by a T‐DNA insertion just after the transcriptional start site of the long transcript. This presumably results in no protein being produced, or occasional secondary translational initiation may produce small quantities of wild‐type Swi1 protein (Mercier et al., 2001a). The dyad allele results from a frameshift mutation after amino acid 505 (and terminates after a further 9) (Agashe et al., 2002). The swi1‐2 allele, which affects both male and female meiosis, was generated by EMS and has a stop codon after amino acid 389 (of the long transcript).
The Swi1 protein has been localized to the nucleus during premeiosis and very early prophase using a green fluorescent protein (GFP) fusion construct driven by the SWI1 promoter. The GFP signal suggests that the Swi1 protein co‐localizes with DNA (Mercier et al., 2001a). It is proposed that the Swi1 protein promotes the establishment of sister chromatid cohesion, but is not responsible for its maintenance, since the protein does not persist through meiosis I. That the sister chromatids remain associated at the centromeres in the swi1‐2 mutant rather than immediately disassociating suggests that an additional early acting, transient cohesion is promoted by factors independent of Swi1 (Mercier et al., 2001a).
Two groups have described mutations in a second gene that has been associated with chromosome cohesion (Bai et al., 1999; Bhatt et al., 1999). Mutations in the SYN1/DIF1 gene result in a complex meiotic phenotype that affects both male and female fertility. A defect can initially be detected as early as leptotene when chromosome condensation appears irregular (Bai et al., 1999), however chromosomes appear to be normally synapsed during pachytene (Bhatt et al., 1999). Extensive chromosome fragmentation is clearly observed by anaphase I. Chromosome segregation fails, with some chromosomes and acentric chromosome fragments failing to attach to the meiotic spindle (Peirson et al., 1997; Bai et al., 1999). Some univalent chromosomes are also observed at post‐pachytene stages in syn1/dif1 mutants, suggesting that sister chromatid cohesion may be compromised (Bhatt et al., 1999). Further acentric fragments together with chromosome bridges are seen at anaphase II. Subsequent non‐disjunction leads to the production of up to 8 polyads containing variable amounts of DNA. A detailed examination of female meiosis has not been made in the syn1/dif1 mutants, but the ovules do not contain an obvious embryo sac and excess callose is seen deposited around the gametocyte (Bhatt et al., 1999).
The gene responsible for the syn1/dif1 phenotype has been cloned. It has homology to the RAD21 (largely mitotic)/REC8 (meiotic) cohesin family involved in sister chromatid cohesion. In yeast mitosis six genes are known to be required to establish sister chromatid cohesion, the protein products of four of these going on to make up a cohesion complex with the DNA. In Xenopus oocytes the cohesion complex contains five proteins, three of which show homology to proteins in the yeast complex (Lee and Orr‐Weaver, 2001; Dong et al., 2001). In Schizosaccharomycespombe, rec8 mutants result in impaired meiotic chromosome pairing, linear element disruption, reduced recombination and premature sister chromatid separation during meiosis (Molnar et al., 1995; Watanabe and Nurse, 1999).
Two alternately spliced SYN1/DIF1 transcripts have been identified encoding proteins of 627 amino acids (expressed at low levels in all tissues) and 617 amino acids (expressed in buds only). Multiple mutant alleles have been identified; these include a T‐DNA (Bai et al., 1999) and five transposon insertion lines (Bhatt et al., 1999). The Syn1/Dif1 protein contains two potential nuclear localization signals and several PEST sequences typically found in proteins that are subject to rapid turnover (Bai et al., 1999; Bhatt et al., 1999). Some evidence that Syn1/Dif1 is acting as a cohesin is provided by the fact that the protein can partially substitute for the Mcd1 mitotic cohesin in yeast complementation tests (Dong et al., 2001).
Arabidopsis has two other genes with homology to the RAD21/REC8 cohesins, called SYN2 (encoding an 809 amino acid protein) and SYN3 (encoding a 692 amino acids protein) (Dong et al., 2001). As yet no mutant alleles of these genes are known. Both are most highly expressed in regions of actively dividing cells suggesting a possible role in mitosis, however, it is interesting to note that neither will complement yeast cells deficient in the Mcd1 mitotic cohesin (Dong et al., 2001).
Recombination and meiotic DNA repair
Meiotic recombination has been studied extensively, notably in yeast (reviewed in Roeder 1997; Zickler and Kleckner, 1999; Sung et al., 2000). Many of the genes encoding the proteins that catalyse this process have been cloned and characterized. In general, these genes appear to be highly conserved across a wide range of eukaryotes. For example, DMC1 homologues are known in yeast (Bishop et al., 1992), mouse (Habu et al., 1996), human (Habu et al., 1996), and Arabidopsis (Sato et al., 1995). There are now several characterized examples of Arabidopsis mutants in recombination genes.
Meiotic recombination in higher eukaryotes is initiated following double‐strand break formation (DSB) by the Spo11 protein (Keeney et al., 1997). The protein is a member of a novel family of type II‐like topoisomerases (Bergerat et al., 1997). It possesses a transesterase activity that catalyses double‐strand breaks through formation of a 5′‐phosphotyrosyl linkage (Bergerat et al., 1997; Keeney et al., 1997). In addition to initiating double‐strand breaks the protein is also required for chromosome synapsis in yeast and mouse, but not in Drosophila and Caenorhabditis elegans (Giroux et al., 1989; Dernburg et al., 1998; McKim and Hayashi‐Hagihara, 1998; Romanienko and Camerini‐Otero, 2000).
Three potential homologues of SPO11 have been identified in Arabidopsis. Evidence suggests that of one these, AtSPO11‐1, is required in meiosis (Grelon et al., 2001). The gene encodes a protein of 362 amino acids and is most highly expressed in reproductive tissues (Hartung and Puchta, 2000; Grelon et al., 2001). Two mutant alleles of AtSPO11‐1 have been identified amongst reduced fertility plants isolated from T‐DNA transformed and EMS mutagenized populations (spo11‐1‐1, spo11‐1‐2, respectively) (Grelon et al., 2001).
Examination of spo11‐1‐1 revealed that the mutant formed few bivalents during both male and female meiosis, leading to meiotic non‐disjunction and the production of polyads (as opposed to tetrads) with variable DNA content during male meiosis. During female meiosis the embryo sac typically fails to differentiate. No fully synapsed chromosomes are detected at prophase I. As a result of these defects the mutant exhibits a dramatic reduction in meiotic recombination frequency (Grelon et al., 2001).
The reduced chiasma frequency (7% of wild type) is an interesting feature of spo11‐1‐1, which, based on the site of T‐DNA insertion in the first exon, is expected to be a null mutant. The corresponding Drosophila and C. elegansspo11 mutants show no recombination (McKim et al., 1998; McKim and Hayashi‐Hagihara, 1998; Dernburg et al., 1998). This suggests that, in Arabidopsis, some DSBs are formed by an alternative route, such as one of the other SPO11 homologues, or another plant specific protein(s) or environmental DNA damage (Grelon et al., 2001).
There is no compelling evidence to support a meiotic role for the two remaining ArabidopsisSPO11 homologues, AtSPO11‐2 and AtSPO11‐3. Recent yeast two‐hybrid studies have revealed that Spo11‐2 and Spo11‐3 proteins can interact with a homologue of the archaebacterial topoisomerase B subunit (AtTOP6b) also found in Arabidopsis. This finding taken in conjunction with their expression patterns suggest they primarily function as topoisomerases (Hartung and Puchta, 2001).
The yeast homologues of the Escherichia coli RecA protein, Rad51 and Dmc1 have been shown to be essential for inter‐homologue recombination and, therefore, bivalent formation (Bishop et al., 1992; Shinohara et al., 1992). In meiosis, these proteins work in conjunction to catalyse invasion of the homologous non‐sister duplex by single‐stranded resected 3′ ends from double‐strand breaks. Whilst the role of Dmc1 is specific to meiosis, Rad51 is also active in mitotic recombinational repair (Bishop et al., 1992; Shinohara et al., 1992). Both genes appear to be highly conserved in eukaryotes, including Arabidopsis (Bishop et al., 1992; Morita et al., 1993; Shinohara et al., 1993; Sato et al., 1995; Habu et al., 1996). To date there are no reports of an ArabidopsisRAD51 mutation, possibly reflecting the mitotic requirement for the protein, however a T‐DNA insertion mutant in AtDMC1 has been identified (Couteau et al., 1999). Homozygous dmc1 plants set 1.5% viable seed compared to wild type (random segregation of five chromosome pairs is expected to result in the production of about 3% viable seed) with both male and female fertility affected. In PMCs the primary cytological defect appears to be in bivalent formation with ten univalents observed at metaphase I, although occasional bivalents are seen. Subsequent non‐disjunction during meiosis I and meiosis II results in polyads with variable DNA content. Megaspore mother cells in the mutant arrest just after meiosis and fail to go on to form the 8‐nucleus embryo sac (Couteau et al., 1999). The meiotic behaviour of dmc1 highlights a feature found in other Arabidopsis mutants such as asy1 (Ross et al., 1997), syn1/dif1 (Peirson et al., 1997; Bhatt et al., 1999) and ms5/tdm1 (Chaudhury et al., 1994; Ross et al., 1997) in that, despite gross defects, meiosis does not arrest. In contrast, in other species such as yeast and Drosophila (reviewed in Zickler and Kleckner, 1999; Roeder and Bailis, 2000) such lesions often trigger meiotic arrest or in the case of mice (and perhaps C. elegans), apoptosis (Romanienko and Camerini‐Otero, 2000; Gartner et al., 2000; Roeder and Bailis, 2000). This suggests that Arabidopsis and possibly other plants lack the meiotic checkpoints found in other organisms.
The AtDMC1 gene which has been cloned encodes a 344 amino acid protein with 51.8% identity, 71.1% similarity to the yeast Dmc1 protein (Sato et al., 1995; Klimyuk and Jones, 1997; Doutriaux et al., 1998). AtDMC1 expression can be detected in non‐meiotic tissues by RT‐PCR (Klimyuk and Jones, 1997) and Northern blots (Doutriaux et al., 1998). However, RNA in situ hybridization and promoter GUS fusion experiments suggest high level expression is confined to meiotic tissues (Klimyuk and Jones, 1997).
The Rad50 protein is essential for DSB repair in yeast and is, therefore, required both for repair of DNA damage and for meiosis (Alani et al., 1989) and is also involved in telomere maintenance (Bertuch and Lundblad, 1998). A mutant in the ArabidopsisRAD50 gene has been identified. Homozygous plants are sterile, but no detailed analysis of meiosis in the mutant has been presented (Gallego et al., 2001). It is not clear whether the phenotype affects both male and female meiosis, although this would be expected to be the case. Suspension cell cultures derived from the mutant plants show hypersensitivity to the DNA damaging agent MMS (methyl methanesulphonate) suggesting the RAD50 gene also has a role in DNA repair in Arabidopsis (Gallego et al., 2001). Cultured mutant cells undergo reduction in telomere length, although telomerase activity levels are the same as observed in wild‐type cultured cells (Gallego and White, 2001). Mutant plants display somatic hyper recombination (8–10‐fold higher than that observed in heterozygotes) reminiscent of that seen in yeast rad50 mutants (Gottlieb et al., 1989; Gherbi et al., 2001). Mutant plants grown on agar also appear stunted compared to wild type although this phenotype is not apparent in soil‐grown plants. Nevertheless, it suggests that RAD50 also has a role in mitotic growth (Gallego et al., 2001).
The RAD50 gene encodes a 1316 amino acid protein with 29.8% identity to the human and 27.3% to the yeast Rad50 proteins, respectively, and is expressed in all tissues tested (by Northern blotting), and at a higher levels in tissues containing rapidly dividing cells. The T‐DNA insertion in the rad50 mutant line is predicted to cause premature truncation of the protein resulting in the loss of the last 266 amino acids and so may not represent a null mutant (Gallego et al., 2001).
Other Arabidopsis mutants have been described that are potentially mutated in genes affecting meiotic recombination. Masson and Paszkowski, (1997) reported recombination rates in three X‐ray‐sensitive mutants isolated from an EMS mutagenized population, which were also hypersensitive to the DNA damaging agents MMS and mitomycin C. The xrs11 mutant showed normal levels of extrachromosomal recombination, but was defective in X‐ray‐mediated stimulation of homologous recombination in such a manner as to suggest a signalling failure may be involved. Meiotic recombination in this mutant was not tested (Masson and Paszkowski, 1997). The xrs9 mutant showed reduced fertility (50% of wild type), together with reduced extrachromosomal and meiotic recombination. In this it resembles yeast mutants of the RAD52 epistasis group (Masson and Paszkowski, 1997). The xrs4 mutant showed reduced extrachromosomal recombination, but was found to exhibit meiotic hyper‐recombination. This mutant is, however, only 10% as fertile as the wild type (Masson and Paszkowski, 1997).
The mei1 (also known as mcd1) mutant was isolated as a tagged mutant from a T‐DNA transformed population. This mutant displays a DNA fragmentation phenotype when examined at a cytological level, indicating that it is unable to repair double‐strand breaks. The mutant exhibits a male‐sterile (He et al., 1996) or male‐reduced‐fertility phenotype (Ross et al., 1997) (presumably dependent on the plant growth conditions). Premeiosis and early prophase I in the mutant are apparently normal, however, by early diplotene extensive chromosome fragmentation can be observed. By metaphase I (Fig. 1f) the degree of fragmentation can be seen to vary from cell to cell, with some univalent chromosomes also present. At telophase I some of the acentric chromosome fragments are excluded from the daughter nuclei which can be clearly seen at the dyad stage (Fig. 1g). Further fragmentation may also occur during the second meiotic division. The products of male meiosis are up to eight microspores with variable DNA content, some chromatin fragments remain in the cytoplasm and never reach the nuclei (He et al., 1996; Ross et al., 1997).
It is unclear at this stage which gene has been affected in the mei1 mutant. He and Mascarenhas (1998) report that the MEI1 gene is composed of three exons encoding an 89 amino acids protein with weak homology to human acrosin‐trypsin inhibitor (20% identity, 42% similarity). RT‐PCR and RACE‐PCR experiments support the existence of this gene transcript (He and Mascarenhas, 1998). However, a subsequent study suggests that the gene encodes a larger predicted protein (of between 783 and 842 amino acids) (Yang and McCormick, 2002). This predicted protein contains three BRCT domains, which are found in proteins involved in double‐strand break repair in yeast and mouse meiosis (Herrmann et al., 1998; García‐Díaz et al., 2000). However, the issue is complicated by the fact that the two proteins are predicted to be encoded on complementary DNA strands at the MEI1 locus.
In yeast and other organisms several genes with homology to the E. coli mismatch repair genes have been demonstrated to be essential for the repair of double‐strand breaks induced during meiotic recombination (Roeder, 1997). Mouse knockouts of mismatch repair genes also show a range of meiotic phenotypes (Baker et al., 1995, 1996; Edelmann et al., 1996). The Mlh1 protein, a homologue of the E. coliMutL gene, has been shown to promote crossing over during meiosis in yeast (Hunter and Borts, 1997). Arabidopsis possesses an MLH1 homologue encoding a 737 amino acid protein which shows 56% similarity to the human protein (Jean et al., 1999). A T‐DNA insertion in the promoter of MLH1 has been identified. However, transcription levels in this mutant are not affected, as the T‐DNA sequences appear to supply an alternative promoter and the line has no mutant phenotype (Jean et al., 1999). Arabidopsis has two other MutL homologues, PMS1 and MLHX (Jean et al., 1999) as well as homologues of the E. coliMutS gene termed MSH2, MSH3, MSH6‐1, and MSH6‐2 (Adé et al., 1999). No mutants in these gene have been reported, however work on the Msh2 protein suggests that it is a functional MutS homologue (Adé et al., 2001).
Mutants affected in chromosome synapsis or bivalent stability
A central feature of prophase I of meiosis is the juxtaposition and pairing of homologous chromosomes leading to their intimate synapsis at pachytene (Zickler and Kleckner, 1999). In the majority of eukaryotes synapsis is mediated by an evolutionarily conserved tripartite proteinaceous structure, the synaptonemal complex (von Wettstein et al., 1984; Zickler and Kleckner, 1999). Although, the exact function of the synaptonemal complex remains to be established it plays a key role in meiosis.
Several Arabidopsis mutants that show defects in chromosome synapsis have been described. By using detailed cytological examination these can be subdivided into two categories; asynaptic, where chromosome synapsis is never achieved and desynaptic, where chromosome synapsis occurs but is lost prematurely.
Two synaptic mutants were found as part of a screen for meiotic mutants amongst the Feldmann collection of T‐DNA transformed Arabidopsis lines (Feldmann and Marks, 1987; Ross et al., 1997). Asynaptic1 (asy1) is an asynaptic mutant that leads to a male and female reduced‐fertility phenotype. Cytological examination of spread DAPI stained pollen mother cells (PMCs) and embryo sac mother cells (EMCs) reveal that asy1 homozygotes exhibit a failure of homologous chromosome synapsis during prophase I (Ross et al., 1997; Armstrong and Jones, 2001). The mutant proceeds normally through premeiotic interphase with unpaired centromeres distributed throughout the cell and unpaired telomeres closely associated with the nucleolus. Telomere pairing is initiated in both asy1 and wild type during leptotene (Armstrong et al., 2001). Meiosis in the asy1 mutant then diverges from wild type as there is no compelling evidence of synapsis, hence normal pachytene cells are not found (Fig. 1b). At late diplotene most chromosomes are seen to be univalent (Fig. 1e) with a residual low bivalent frequency of 1.27–1.57 per cell (Sanchez‐Moran et al., 2001; Ross et al., 1997) in PMCs. The residual chiasmata are predominantly sub‐telomeric (Sanchez‐Moran et al., 2001), rather than exhibiting the normal wild‐type distribution. The lack of synapsis in asy1 results in univalent formation and meiotic non‐disjunction during meiosis I (a typical example of the uneven chromosome distribution seen after meiosis I is shown in Fig. 1i) and meiosis II giving rise to polyads containing uneven numbers of chromosomes.
The ASY1 gene has been cloned and encodes a 596 amino acid protein with significant homology to the yeast Hop1 protein over the first 250 amino acids. This region comprises a HORMA domain found in a number of chromatin interacting proteins (Aravind and Koonin, 1998). The remainder of the protein is unique with no homology to anything else in the database (Caryl et al., 2000). The ASY1 gene was found to be expressed in all tissues tested by RT‐PCR. Yeast hop1 mutants exhibit reduced levels of meiotic recombination and extremely low levels of spore viability (Hollingsworth and Byers, 1989). Hop1 protein accumulates at multiple discrete foci during prophase I of meiosis as a component of the axial elements (Hollingsworth et al., 1990; Smith and Roeder, 1997), but is lost by the time full synapsis has been achieved (Smith and Roeder, 1997) at pachytene. The Hop1 protein has DNA binding activity, and contains a non‐classical zinc finger (not seen in Asy1) around amino acid 371 of the protein (Hollingsworth et al., 1990). However, DNA binding activity, while enhanced in the presence of zinc, is still observed even in the presence of EDTA, suggesting the HORMA domain may also be involved in DNA binding (Kironmai et al., 1998).
Immunolocalization studies using an antibody raised against recombinant Asy1 protein have revealed the spatial and temporal distribution of the protein. Asy1 is first observed as punctate chromatin associated foci during meiotic interphase in PMCs. This signal becomes more continuous during leptotene and zygotene, and is seen to be associated with axial elements, but not the associated chromatin loops. The protein begins to disappear from the chromosomes as they desynapse. By late diplotene some remaining protein may be observed as aggregates no longer associated with chromatin (Armstrong et al., 2002). A more detailed examination of Asy1 localization using immunogold labelling has revealed that the protein is not continuous along the axial/lateral elements, but instead is found at discrete points along the chromatin associated with these structures (Armstrong et al., 2002). Taken together the asy1 mutant phenotype and protein localization suggests a potential role for Asy1 in homologous chromosome juxtaposition during early prophase I with a failure to complete homologous chromosome pairing leading to a failure in SC assembly. Alternatively, if transient or unstable pairing does occur in asy1 the role of Asy1 may be to recruit the bases of chromatin loops to the developing axial/lateral elements (Armstrong et al., 2002).
Desynaptic1 (dsy1) is a desynaptic mutant that leads to a reduction in both male and female fertility. The dsy1 mutant proceeds normally through the early stages of meiosis achieving apparently normal synapsis at pachytene (Fig. 1a), however, by early diplotene homologous chromosomes are seen to be loosely associated rather than associated into bivalents by chiasmata (Fig. 1d), thus suggesting a defect in chiasma formation or retention. Occasional bivalents are, however, observed (frequency 0.75 per cell). A mixture of reductional and equational division at anaphase I lead to the production of a variable number of cells containing variable amounts of DNA. The mutation affects both male and female fertility. However, the line can be maintained as homozygotes, although a high level of aneuploidy is seen in the progeny (SJ Armstrong, personal communication). The gene involved has not yet been identified, but the mutation maps to chromosome 3 between GL1 and HY2 (Ross et al., 1997).
Another desynaptic mutant dsy10 (Feldmann line 7219) is both male and female sterile. The line also exhibits asynchronous meiotic development both within and between anthers (Peirson et al., 1996). Meiosis in the mutant proceeds normally, with full synapsis achieved at pachytene. However, at diplotene homologous chromosomes appear to separate precociously, giving rise to 10 univalents by diakinesis (Cai and Makaroff, 2001). Occasional bivalents can be observed, usually joined at the centromeres. Chromosome condensation in the mutant is also less extensive than in wild type. Subsequent defects in this mutant vary between cells; some chromosomes fail to attach to the metaphase I spindle, whilst others attach and segregation occurs. Chromosome bridges are often associated with these segregating univalents. Some cells arrest after meiosis I while others enter meiosis II (Cai and Makaroff, 2001). At the tetrad stage a variety of cell types can be observed, tetrads containing a variable number of uninuclear and polynuclear microspores are observed, along with cells which arrested after the first meiotic division. In 10% of cells precocious loss of sister chromatid cohesion was observed giving rise to more than 10 univalents at diakinesis, while other chromatids remained attached only at the centromeres. Although the primary defect in this mutant appears to be desynapsis, the variable nature of the later stages may suggest that some aspect of cell cycle control has been affected (Cai and Makaroff, 2001)
Chromosome segregation mutants
In another group of meiotic mutants, the primary defect appears to be in the function or regulation of the meiotic spindle leading to defects in chromosome segregation.
A mutant with a Ds transposon insertion in the KATA/ATK1 gene exhibits reduced male fertility (Chen et al., 2002). Meiosis in PMCs appears normal up to metaphase I where chromosomes fail to align correctly at the equator of the cell. At anaphase I individual chromosome pairs may be precocious or laggard with respect to segregation, with some failing to separate at all, although some balanced dyads are generated. Organelle distribution is also more diffuse in the mutant. Chromatid segregation is also affected during the second meiotic division, giving rise to polyads with variable DNA content (Chen et al., 2002).
The meiotic spindle in the mutant appears abnormal with less tubulin staining than wild type. Also the spindle is diffuse and unfocused, especially at the poles and appears less organized during meiosis I. During meiosis II, minispindles form between the groups of chromosomes produced in metaphase I (Chen et al., 2002).
The KATA/ATK1 gene has been cloned and appears to be preferentially expressed in dividing tissues (Mitsui et al., 1993; Chen et al., 2002). The gene encodes a protein of 793 amino acids that has homology to kinesin. Kinesin is a microtubule motor protein and is one of a group of proteins that have been implicated in a complex range of processes including spindle formation, position and chromosome movement (reviewed in Sharp et al., 2000; Goldstein, 2001). KatA/Atk1 protein was found to localize to the mid‐zone of the mitotic spindle and phragmoplast (Liu et al., 1996). Recently, KATA was renamed ATK1 to avoid confusion with the KAT1 potassium channel gene (Chen et al., 2002; Schachtman et al., 1994) (although, it should be noted that this nomenclature has already been used for a homeodomain protein by Dockx et al., 1995). A second closely related homologue of KATA/ATK1 is present in the Arabidopsis genome. This gene, ATK5, is 91% similar to KATA/ATK1 at the protein level and may be the reason why no mitotic or female meiotic phenotype is seen in the katA/atk1 mutant (Chen et al., 2002). Two other kinesin‐like genes found in Arabidopsis, KATB and KATC (Mitsui et al., 1994) also show homology to KATA/ATK1 (57% and 56% amino acid sequence identity, respectively) (Chen et al., 2002).
Another mutant affecting chromosome segregation is ask1. The ask1 (SKIP1‐LIKE1) mutant isolated from a Ds transposon insertion line is male sterile, but female fertile (Yang et al., 1999b). Meiosis in pollen mother cells appears to proceed normally with chromosome pairing occurring, however, there is a failure of chromosome segregation at anaphase I. Some bivalents fail to separate entirely, with the chromosomes becoming stretched instead. Meiosis II is also affected with some sister chromatids failing to separate normally. This leads to the production of polyads of variable size and chromosome content (Yang et al., 1999b). Some ask1 mutant cells have spindles that are longer than wild type, and it is these cells which exhibit the highest degree of chromosome movement during meiosis I (Yang and Ma, 2001). The ASK1 gene also appears to have a non‐meiotic role as homozygous plants display reduced rosette size and internode length due to a reduction in cell numbers, together with alterations in floral organ development (Zhao et al., 1999).
The ASK1 gene encodes a protein of 160 amino acids (Porat et al., 1998) and was found to be a homologue of the yeast SKP1 gene which is involved in regulating the mitotic cell cycle by targeting proteins for degradation via the ubiquitin‐mediated proteolysis pathway (Yang et al., 1999b). The ask1 mutant contains a Ds transposon insertion towards the middle of the gene and is thought to be a null mutant (Yang et al., 1999b). The ASK1 gene is expressed in all tissues tested by Northern blotting, but maximum expression occurs in meristematic tissues (Porat et al., 1998). The ASK1 gene is part of a large gene family of at least nine members in Arabidopsis (Yang et al., 1999b).
Cell cycle regulation mutants
The phenotypes of another group of meiotic mutants suggest they may affect cell cycle regulation in some way.
In wild‐type Arabidopsis PMC meiosis is synchronized within the anther locules. The temperature‐sensitive tam1 (tardy asynchronous meiosis) mutant (Magnard et al., 2001) results in PMC meiosis becoming asynchronous within the anther and was isolated from an EMS mutagenized Arabidopsis population. Furthermore, dyad meiotic products or pollen (containing two gametophytes within one exine wall) are typically produced, apparently as a result of a slower meiotic division set against wall formation at the normal rate (i.e. the wall forms before metaphase II has a chance to occur) (Magnard et al., 2001). In addition, a range (2–7) of other microspore numbers are also observed, some of which contain multiple nuclei. In the quartet1‐tam1 double mutant (where the products of meiosis remain physically linked so they can be directly followed) meiosis II was found to proceed, but at a slower pace. The delay appears to be at the G2/M transition in both meiosis I and meiosis II. Defects may also be observed in chromosome condensation and segregation. The TAM1 gene maps to chromosome 1 between the SSLP marker nF22K20 and CAPS marker agp64 (Magnard et al., 2001).
Two other mutants exhibiting asynchronous meiotic development have been identified by Peirson et al. (1996). One, dsy10 (Cai and Makaroff, 2001), has been described earlier in this review. In the other line, 7593, a putative T‐DNA tagged mutant, some microspores arrest prior to meiosis, while others give rise to abnormal tetrads that follow several different paths prior to collapse. Abnormal callose deposition is observed in mutant microspores. The 7593 mutation only affects male fertility (Peirson et al., 1996).
Another class of cell cycle mutant is represented by ms5 (also known as tdm1 and pollenless3). This was first identified as an EMS mutant in which microsporocytes initially appeared normal, but soon degenerated (Chaudhury et al., 1994). A more detailed examination of the cytology revealed that the pollen mother cells appear to undergo two rounds of normal meiotic division. However, at the end of meiosis II the cell attempts a third division without any further DNA replication. The chromosomes recondense and appear to become attached to a spindle (determined on the basis of organelle exclusion). The chromosomes then become extended and stretched (Fig. 1k). Eventually interphase nuclei reform containing random groups of chromatids (Fig. 1l), giving rise to approximately eight microspores with variable DNA content (Ross et al., 1997). Both the EMS ms5‐1 and T‐DNA ms5‐2/tdm1 alleles are sterile as homozygotes, but the latter is semi‐dominant showing reduced fertility as a heterozygote under certain environmental conditions (Ross et al., 1997; Glover et al., 1998; Luong, 1999). A third allele pollenless3‐2 is a deletion isolated from a T‐DNA transformed population (Sanders et al., 1999).
The MS5 gene encodes a 434 amino acids protein that has no clear homology with other known proteins. However, stretches of the protein have limited similarity to the synaptonemal complex protein SCP1 from rat, and the regulatory subunit of a cyclin‐dependent kinase from Xenopus (Glover et al., 1998). The ms5‐1 allele is reduced to 129 amino acids (similar in size to pollenless 3‐2) and the ms5‐2 allele is reduced to 322 amino acids (Glover et al., 1998; Sanders et al., 1999). MS5 expression is detected both in meiotic and non meiotic tissues by RT‐PCR, but RNA in situ hybridization detected MS5 transcripts only in meiotically dividing cells and indicated that this RNA is lost prior to tetrad formation (Sanders et al., 1999). Two other genes with homology to MS5 (POLLENLESS3‐LIKE1 and POLLENLESS3‐LIKE2) are also present in Arabidopsis (Sanders et al., 1999).
To date, relatively few Arabidopsis meiotic mutants have been both described and the corresponding genes cloned, compared to organisms such as Saccharomyces cerevisiae. The genes that have been identified are, however, involved in a diverse range of meiotic processes (chromosome cohesion, recombination, synapsis, chromosome segregation, and cell cycle regulation). The studies described in this article therefore highlight the value of Arabidopsis as a model system
The identification of meiotic mutants based on a reverse genetics approach to identify T‐DNA and transposon tagged derivatives of known meiotic genes from other species provides an excellent route to move the study of plant meiosis to a more detailed level. Further application of this approach will undoubtedly enable the identification of other meiotic genes in the future, especially as additional tagged Arabidopsis populations have recently become available to the research community. Other reverse genetics techniques such as antisense RNA and RNAi may also be useful in investigating the function of meiotic gene homologues from other species when tagged mutants are not available. However, this is not to say that screening for reduced fertility mutants has become redundant as some meiotic genes may not exhibit sufficient homology at the level of primary amino acid sequence to be readily identified by homology searches in silico, or may be unique to meiosis in plants. Moreover, both the meiotic mutants and corresponding genes can be used in conjunction with approaches such as transcriptomics, proteomics and interaction cloning to provide yet another powerful route towards the isolation of additional plant meiotic genes.
Using these cloned genes it is possible to investigate the molecular mechanisms that underpin plant meiosis. This information may be combined with the detailed cytogenetic data that has been obtained over many decades and recent additional information such as a timeline for Arabidopsis meiosis established using BrdU (bromodeoxyuridine) labelling of meiotic S‐phase cells (SJ Armstrong, personal communication). An example of how this approach may be used to dissect the interrelationship between different aspects of meiosis may be seen in Armstrong et al. (2001) where telomere pairing in the asynaptic mutant asy1 was examined. This revealed that while early telomere pairing occurs normally in the asy1 mutant this does not, however, progress to complete pairing and synapsis. This clearly implies that early pairing events are mechanistically distinguishable from pairing progression and subsequent synapsis.
The application of immunocytology has been used in a variety of species to investigate the distribution of meiotic proteins. As mentioned earlier this is beginning to be applied to meiotic proteins from Arabidopsis. This is an important tool as it may be used to investigate the distribution of meiotic proteins on individual chromosomes. Potential co‐localization of meiotic proteins may be used to infer protein–protein interactions that can then be investigated in more detail. Protein localization can also be helpful in developing hypotheses about gene function. GFP fusion proteins such as SWI1‐GFP can also be useful in determining protein localization. They also have the potential to allow real time visualization of meiosis in plants.
This work was supported by the Biotechnology and Biological Research Council (BBSRC).
|Number of reduced fertility mutants identified||Number of potential meiotic mutants||Mutagen used in the screen||References|
|6||2||T‐DNA||Dawson et al., 1993|
|40||na||EMS||Chaudhury et al., 1994|
|20||na||T‐DNA||Glover et al., 1996|
|140||16a||T‐DNA||Peirson et al., 1996|
|8||6||T‐DNA||Ross et al., 1997|
|44||26||T‐DNA||Sanders et al., 1999|
|855||359||EMS||Sanders et al., 1999|
|46||na||EMS||Siddiqi et al., 2000|
|17||7b||EMS||Mercier et al., 2001a|
|791||32||T‐DNA||Mercier et al., 2001b|
|Number of reduced fertility mutants identified||Number of potential meiotic mutants||Mutagen used in the screen||References|
|6||2||T‐DNA||Dawson et al., 1993|
|40||na||EMS||Chaudhury et al., 1994|
|20||na||T‐DNA||Glover et al., 1996|
|140||16a||T‐DNA||Peirson et al., 1996|
|8||6||T‐DNA||Ross et al., 1997|
|44||26||T‐DNA||Sanders et al., 1999|
|855||359||EMS||Sanders et al., 1999|
|46||na||EMS||Siddiqi et al., 2000|
|17||7b||EMS||Mercier et al., 2001a|
|791||32||T‐DNA||Mercier et al., 2001b|
a From 50 lines which were studied in detail.
b Number of complementation groups identified in the screen.