INTRODUCTION

Meiosis is one of the crucial processes during the life of flowering plants. During meiosis, the chromosome number is reduced by half, leading to the formation of haploid gametes that eventually fuse and restore ploidy in the following generation (Cromer et al., 2012). Meiosis is different from somatic cell division (mitosis) in several ways (see Table 1); for instance, unlike mitosis, which produces two diploid cells, meiosis involves two rounds of nuclear division, meiosis I and II, following a single round of DNA replication, thus producing four haploid nuclei (Wang and Copenhaver, 2018). However, meiosis II is similar to mitosis with segregating sister chromatids (see Table 1), while meiosis I is unique and involves segregation of homologous chromosomes (Wang and Copenhaver, 2018).

To accurately perform reductional division, specific meiotic steps include homolog pairing, the formation of the synaptonemal complex (SC), a tripartite protein structure, and the maturation of recombination intermediates into crossovers (COs), that are visualized as physical attachments (chiasmata) between homologs (see Fig. 1). These events ensure the proper segregation of homologs to opposite poles at anaphase I (d'Erfurth et al., 2010). In higher eukaryotes, the core cell cycle machinery is shared between meiosis and mitosis. Most notably, cell cycle entry and progression are determined by the activity of cyclin-dependent kinases (CDKs) and associated A- and B-type cyclin subunits that establish activity and specificity (Bulankova et al., 2013). In eukaryotes, CDK activity peaks during the M-phase, when chromosomes are attached to the microtubule and align at the mid-cell plate. Subsequent activation of the Anaphase-Promoting Complex (APC/Cyclosome [APC/C]) initiates proteolytic destruction of A- and B-type cyclins and allows for chromosome segregation (Bulankova et al., 2013) (see Fig. 2). The Arabidopsis APC/C is a conserved multisubunit cullin-based RING E3 ubiquitin ligase complex (d'Erfurth et al., 2010; Choi et al., 2014) composed of at least 11 core subunits (Bulankova et al., 2013) that plays an essential role during mitosis, meiosis and postmitotic cell differentiation (Eloy et al., 2012). In eukaryotic organisms, E3 ubiquitin ligases act as mediators between E1 ubiquitin activation, E2 ubiquitin conjugation enzymes and degradation by the 26S proteasome (Choi et al., 2014). They also provide specificity for the substrate (Bulankova et al., 2013). Ubiquitination takes place when an E3 ligase enzyme binds to both the substrate and an E2 thioesterified with ubiquitin (Deshaies and Joazeiro, 2009), bringing them close enough so that the ubiquitin is transferred from the E2 to the substrate via a covalent E3-ubiquitin thioester intermediate (Deshaies and Joazeiro, 2009). Eukaryotes have two major types of E3 ligases with an HECT and a RING domain, respectively. RING domain ligases feature conserved cysteine and histidine residues that form an interleaved structure with two zinc coordination sites for protein interactions (Deshaies and Joazeiro, 2009).

Table 1.

Key differences between mitosis and meiosis.

Figure 1.

Meiotic chromosome morphology in Col-0 and the atcdc20.1 mutant as observed by DAPI staining. The five wild-type bivalents are aligned regularly at the equatorial plate during metaphase I (A), and segregate equally into four nuclei during anaphase II (B), while the five atcdc20.1 bivalents are not well aligned during metaphase I (C), and form multiple nuclei at anaphase II (D). Bar = 5 µm.

Plant A-type cyclins, especially Arabidopsis CYCA A2 (CYCA2;1: At5g25380) and CYCA2;3 proteins (At1g15570) (see Table 2 and Fig. 2), show important functional specificity during mitotic cell cycle progression and control of ploidy (Eloy et al., 2012). The meiotic plant A-type cyclin CYCA1;2/TARDY ASYNCHRONOUS MEIOSIS (TAM; At1g77390) (see Table 2 and Fig. 2) is essential for the transition between the first and second meiotic division, and defects in its function lead to exit from meiosis after prophase I (d'Erfurth et al., 2010) and the formation of diploid gametes (d'Erfurth et al., 2010).

The role of plant B-type cyclins is less clear, although 11 tentative genes were believed to exist in Arabidopsis at one point (Cromer et al., 2012). Functional characterization of B-type cyclin CYCB3;1 (At1g16330) (see Table 2 and Fig. 2) indicated that this protein has a role in spindle organization and cell wall formation in male meiocytes and that it cooperates with plant-specific cyclin SOLO DANCERS (SDS: At1g14750) (see Table 2 and Fig. 2), which has a role in meiotic recombination (Azumi et al., 2002). In eukaryotes, both A- and B-type cyclins are targeted for proteolysis by the APC/C via recognition of two specific amino acid motifs: 1) destruction (D) box and 2) KEN box (Eloy et al., 2012).

Activation of the Drosophila APC/C requires the activity of Cdc20/Fizzy and Cdh1/Fizzy-related proteins, known in plants as CELL CYCLE SWITCH 52 (CCS52A1: At4g22910, CCS52A2: At4g11920, and CCS52B: At5g13840) (Heyman et al., 2011) (see Table 2). CELL DIVISION CYCLE 20 (CDC20) is activated from late G2 phase onward, but anaphase to early S phase activation depends on CADHERIN 1 PRECURSOR (CDH1) (Heyman and De Veylder, 2012). Nonetheless, how Arabidopsis CCS52A1, CCS52A2, and CCS52 operate during plant meiosis is entirely unclear (Heyman et al., 2011).

The Arabidopsis APC/C itself is negatively regulated by several proteins, including the regulator of male gametogenesis and A-type cyclin stabilizer SAMBA (At1g32310) (Eloy et al., 2012) as well as the meiotic regulator OMISSION OF SECOND DIVISION 1 (OSD1: At3g57860) (d'Erfurth et al., 2010) (see Table 2 and Fig. 2), the presumed homolog of protein Mes1 in fission yeast and Emi2/Erp1 in metazoans (Cifuentes et al., 2016). The OSD1 amino acid sequence shares with Mes1 having two putative APC/C degradation motifs, a D-box (residues 104-110) and a KEN-box (residues 80-83), which have been shown to determine Mes1 function (Iwata et al., 2011). The primary function of Emi1 is to inhibit the APC/C-CDH1/FZR complex, terminating DNA replication and enabling transition from the G2 to M phase (Iwata et al., 2011), whereas Emi2 inhibits the APC/C-CDC20 complex, enabling progression through meiosis (Iwata et al., 2011).

Figure 2.

Schematic representation of protein interactions between ubiquitination regulators during meiosis in Arabidopsis thaliana. Abbreviations and notes: spindle assembly checkpoint (SAC). SDS, CYCA and CYCB are plant cyclins. CDC20.1 and CCS52 are APC/C activators. Notes: SAC means Spindle Assembly Checkpoint. Red crosses indicate that the corresponding processes or pathways are blocked. Arrowheads indicate activation of processes. Inhibition means that targeting induces either proteasomal degradation, phosphorylation or downregulation of activity.

The APC/C is also believed to interact directly with protein THREE DIVISION MUTANT1/MALE STERILE 5 (TDM1/MS5: At4g20900) (see Table 2 and Fig. 2), a homodimer that may regulate termination of meiosis at the end of the second meiotic division and whose activity during the first division may be inhibited by CDKA;1-CYCA1.2/TAM-mediated phosphorylation at T16 (Cifuentes et al., 2016). TDM1 is believed to share limited structural similarity with the APC/C subunit CDC16/Cut9/APC6 and contains a tetratricopeptide repeat (TPR) domain known to mediate protein-protein interactions (Cifuentes et al., 2016).

MULTIPLE ROLES OF CDKA;1 IN ARABIDOPSIS

In yeast and mammals, CDK activity by Cdc2/Cdc28 and Cdk1 is indispensable for cell cycle progression (Nowack et al., 2012). The Arabidopsis functional homolog of Cdc2/CDC28 and Cdk1 is CDKA;1 (At3g48750) (see Table 2 and Fig. 2), and it can complement yeast cdc2 and cdc28 mutants (Nowack et al., 2012). Arabidopsis cdka;1/- and CDKA;1-overexpression mutants show defects in pollen mitosis I, defective embryogenesis, reduced root stem growth, underbranched trichomes, and cell cycle arrest at the G2 phase (Nowack et al., 2012), which suggest control of the M and S phases. CDKA;1 protein is also believed to form a complex with CYCA1.2/TAM that phosphorylates TDM1 at residue T16 and leads to exit from the first meiotic division (Cifuentes et al., 2016). TAM is also involved in exit from the second division and organization of the meiotic spindle formation and chromosome condensation and may be the actual Arabidopsis homolog of the yeast APC/C CDC16/Cut9/APC6 subunit (Cifuentes et al., 2016).

THE ROLE OF OSD1 DURING MALE MEIOSIS

Arabidopsis OMISSION OF SECOND DIVISION (OSD1, At3g57860, also known as GIGAS and UVI4-Like) is a regulator of the APC/C and presumably operates by controlling the degradation of cyclins to influence exit from mitosis or meiosis (Crismani et al., 2013; D'Erfurth et al., 2009) (see Fig. 2). OSD1 has a role in regulating endomitotic proliferation in vegetative tissues and female gametophyte development (d'Erfurth et al., 2010). However, characterization of both osd1-1 and osd1-2 progenies indicated that they are all tetraploid and triploid, indicating that OSD1 functions to control the transition from first to second meiotic division (D'Erfurth et al., 2009). Alignment of amino acid sequences indicated that OSD1 shares limited homology with the APC/C inhibitor Mes1 from Schizosaccharomyces pombe (d'Erfurth et al., 2010), including a D-box (residues 104-110, RxxLxx[LIVM]) and a GxEN/KEN-box (residues 80-83), which have been shown to determine Mes1 function. Moreover, OSD1 shares a C-terminal MR-tail with Mes1, which is a methionine and arginine sequence that in kinase Nek2a is essential for binding and inhibition activities against the APC/C; a similar RL-tail in vertebrate Emi2 is essential for inhibiting APC/C during meiosis (Cromer et al., 2012).

Table 2.

Arabidopsis thaliana genes presumed to be involved in meiotic ubiquitination of proteins.

(continued)

Moreover, results from yeast two-hybrid assay and tandem affinity purification experiments indicate that Arabidopsis OSD1 interacts directly with APC/C activators CDC20.1 (At4g33270), CDC20.5 (At5g27570), CCS52A1 (At4g22910), CCS52A2 (At4g11920) and CCS52B (At5g13840) (see Table 2 and Fig. 2), possibly through its D-box and the MR-tail (d'Erfurth et al., 2010; Iwata et al., 2011) but not with APC/C core units APC2, APC7, APC10, CDC27a, and CDC27b (also known as HOBBIT/APC3b) (Iwata et al., 2011 ; Cifuentes et al., 2016). Crossing of osd1-3, cyca1;2/tam-2 and tdm-3 indicated that the corresponding wildtype loci cooperate in the control of the meiotic cell cycle, with OSD1 and TAM controlling the first- to second-division transition and TDM controlling exit from the second meiotic division (d'Erfurth et al., 2010) (see Fig. 2). Additional layers of genetic regulation involve the following:

In rice, combining the Ososd1-1 mutant allele (Os02g37850) with mutant alleles for REC8 (required for orientation of kinetochores; Os05g50410) (see Table 2), and PAIR1 (required for formation of DNA double-strand breaks, Os03g01590) (see Table 2) led to the formation of male and female clonal diploid gametes, a finding that may contribute greatly to plant breeding (Mieulet et al., 2016).

MMD1/DUET IS A KEY REGULATOR OF MEIOTIC PROGRESSION

Arabidopsis MALE MEIOCYTE DEATH1/DUET (MMD1/DUET: At1g66170) (see Table 2) is involved in the development and viability of meiocytes, as seen in mmd1 male meiocytes by cytoplasmic shrinkage and DNA fragmentation at diakinesis (Reddy et al., 2003; Yang et al., 2003), and its expression peaks at late diplotene during male meiosis (Andreuzza et al., 2015). Biochemical characterization of this protein indicated that MMD1/DUET is able to bind in vitro and in vivo to H3K4me2, H3K4me3 (and possibly H3K9me2 and H3S10ph) via its Plant Homeo Domain (PHD) finger (amino acids 606 to 656) and that it activates transcription of APC/C-interacting protein TDM1 (At4g20900) (see Fig. 2) (Andreuzza et al., 2015; Wang et al., 2016). Chromatin immunoprecipitation (ChIP) experiments showed that MMD1/DUET binds to the TDM1 promoter, whereas transcriptional analyses showed that the onset of TDM1 expression coincides with the timing of DUET expression, which suggests that TDM1 is indeed a direct target of MMD1 during male meiosis (Andreuzza et al., 2015). Analysis of gene expression profiles in mmd1/duet meiocytes by quantitative PCR (qPCR) indicated significantly reduced transcription of TDM1 along with other 34 meiotic genes, a mild decrease in OSD1, and no change in expression of TAM (Andreuzza et al., 2015). RNA sequencing and qPCR results revealed that up to 756 genes may show transcriptional regulation in mmd1 meiocytes, including several condensin genes (CAP-D2: At3g57060, CAP-D3: At4g15890, CAP-H: At2g32590, and CAP-H2: At3g16730) (Wang et al., 2016). Therefore, MMD1 may affect meiotic chromosome condensation by regulating the expression of condensin I and II complexes (Wang et al., 2016). Notably MMD1/DUET is a versatile protein also involved in microtubule organization, homologous recombination and chromosome condensation at prophase and metaphase I (Wang et al., 2016).

APC/C ACTIVATOR CDC20.1 HAS A ROLE IN MEIOSIS

Tandem affinity purification (TAP) results indicated that Arabidopsis APC/C contains at least 11 core subunits, all of them are required for the ubiquitin transfer reaction (Heyman and De Veylder, 2012). The processing and timely activation of the entire complex depends on docking by the CDC20 and CCS52 APC/C activator subunits as well as the APC10 co-activating subunit (see Fig. 2) (Heyman and De Veylder, 2012). Both CDC20 and CCS52 contain seven WD40 repeats that facilitate protein-protein interactions; each targets proteins that contain a D-box sequence (RxxLxxxN/Q) (Kevei et al., 2011). CDC20 is believed to be actively sequestered by spindle assembly checkpoint (SAC) proteins such as protein kinase Aurora B to allow for correction of errors in the attachment of kinetochores to the microtubules (Niu et al., 2015). Cytological observation of anthers from T-DNA insertional alleles cdc20.1-3 and cdc20.1-4 corresponding to Arabidopsis CDC20.1 (At4g33270) (see Table 2) showed that anthers contained mostly unviable microspores and meiocytes mostly developed into polyads (see Fig. 1). Moreover, mutant meiocytes showed defects in the alignment of bivalents and segregation of chromosomes at metaphase I and anaphase I and II (see Fig. 1), possibly caused by poor alignment of chromosomes, as suggested by unequal distribution of the kinetochore marker HISTONE H3-LIKE CENTROMERIC PROTEIN (HTR12, At1g01370) (Niu et al., 2015). Analysis of the Aurora marker H3S10ph at centromeres indicates that in the mutant, its distribution is diffuse during diakinesis, which suggests that CDC20.1 is required for proper localization of Aurora 1. A similar phenotype is observed in meiocytes from transgenic ProDMC1:Aurora1RNAi plants (Niu et al., 2015), further supporting the idea that CDC20.1 is required for meiotic chromosome segregation likely through influencing Aurora localization.

26S PROTEASOME AND E3 UBIQUITIN LIGASES AFFECT MEIOTIC RECOMBINATION

The 26S proteasome plays an essential role in meiosis. Both pairing and chromosome segregation require proper proteasome function during budding yeast meiosis and mouse spermatogenesis (Ahuja et al., 2017; Rao et al., 2017). In Saccharomyces cerevisiae, the proteasome regulates axis morphogenesis and synapsis, disassembling non-homologous SCs from clustered centromeres at early prophase I (Ahuja et al., 2017). Additionally, the proteasome is needed for removing SCs at the end of prophase I. In mice, the ubiquitin-proteasome system controls meiotic recombination indirectly by influencing the turnover of several recombination factors (Rao et al., 2017).

Humans have two important E3 ligases that control meiotic recombination: RING FINGER PROTEIN 212 (RNF212) and HUMAN ENHANCER OF INVASION CLONE 10 (HEI10) (Kong et al., 2008). RNF212 is the homologue of yeast Zip3 and worm ZHP-3, a SMALL UBIQUITIN-LIKE MODIFIER (SUMO) RING-finger ligase (Kong et al., 2008; Zhang et al., 2017). HEI10 is a ubiquitin E3 ligase important for CO designation and cell cycle (Toby et al., 2003). RNF212 and HEI10 show haploinsufficiency and play antagonistic roles in CO formation. RNF212 stabilizes recombination factors in the meiotic axes and HEI10 limits the colocalization of RNF212 and pro-CO factors, thereby allowing for resolution of recombination intermediates (Reynolds et al., 2013).

What happens in Arabidopsis? Most proteins involved in the ubiquitination pathway correspond to RING E3 ligases that specifically recognize target proteins (Mazzucotelli et al., 2006). Although no homolog of RNF212 has been found, Arabidopsis HEI10 (At1g53490) (see Table 2) has an essential role during meiosis, because it is required for most COs between homologs (∼85%) (Chelysheva et al., 2012). The HEI10 E3 ligase is a ZMM protein (from the proteins Zip1,2,3,4; Msh4,5; Mer3, implicated in interference-sensitive COs) that appears as numerous foci on the chromosome axes during early prophase I and then is maintained at only ∼10 sites, co-localizing with the CO marker MUTL HOMOLOG 1, COLON CANCER, NONPOLYPOSIS TYPE 2 (MLH1) at pachytene (Chelysheva et al., 2012). The corresponding mutant displayed fertility defects that are due to the presence of univalents at metaphase I (Chelysheva et al., 2012). HEI10 is believed to interact directly with MHS5, to stabilize MSH4 and MSH5, and to regulate the recruitment of MLH1 and MLH3 (see Fig. 2) (Ziolkowski et al., 2017). The HEI10 homolog (Os02g13810) was also identified in rice, a monocot (Wang et al., 2012), the oshei10 mutant shows a similar meiotic defect to the Arabidopsis hei10 mutant, indicating that the HEI10 function in meiotic crossover formation is highly conserved (Wang et al., 2012). Recently, natural polymorphisms in the coding sequence of HEI10 were found associated with variations in chiasma frequencies between different Arabidopsis accessions (Ziolkowski et al., 2017). Furthermore, HEI10 was characterized as haploinsufficient in heterozygotes, and having extra HEI10 copies led to double the number of MLH1 recombination foci, more compact bivalents, and increased subtelomeric recombination (Ziolkowski et al., 2017). This suggests its use as a tool for increasing recombination in key crop genomes, especially by combining it with mutation of the anticrossover helicase recq4a recq4b (RECQ4A: At1g10930, RECQ4B: At1g60930) (see Table 2) (Serra et al., 2018b). In the latter mutant background, unrepaired joint DNA molecules (D loops and double Holiday Junctions) are likely to persist and be repaired as non-interfering class II crossovers (Serra et al., 2018b). Moreover, Arabidopsis anticrossover pathways are not completely redundant (Serra et al., 2018b; Ziolkowski et al., 2017); hence, additional mutations in FANCM (At1g35530), FIGL1 (At3g27120), FANCD2 (At1g48360) and MSH2 (At3g18524) (see Table 2), may further enhance crossover formation (Serra et al., 2018b; Kurzbauer et al., 2018; Serra et al., 2018a; Girard et al., 2015; Crismani et al., 2012; Fernandes et al., 2018). The efficiency of such mutations in causing an increase in recombination appears to be universal, as shown by the characterization of FANCM, RECQ4 and FIGL1 mutants in several crop species such as rice (Oryza sativa), pea (Pisum sativum) and tomato (Solanum lycopersicum) (Mieulet et al., 2018).

Measuring recombination through the use of Col/Ler chromosome substitution lines (CSLs) expressing different colors of fluorescent proteins in pollen (Berchowitz and Copenhaver, 2008; Yelina et al., 2013), led to the determination that HEI10 is part of an Arabidopsis quantitative trait loci (QTL) named rQTL1 that positively affects the formation of MLH1 recombination foci along the meiotic synaptonemal axis (Ziolkowski et al., 2017). Proteins from the HEI10 family possess N-terminal RING domains, central coiled-coil domains and C-terminal regions of unknown function, but are suspected to play a role in substrate/target recognition (Ziolkowski et al., 2017). An R264G polymorphism in the C-terminus of HEI10 is believed to promote recombination, perhaps by promoting protein function or expression timing (Ziolkowski et al., 2017). Crossover modifier loci such as HEI10 may affect genetic adaptation to diverse environments and conditions (Ziolkowski et al., 2017).

Polymorphisms have a suppressive effect in recombination in fancm, figl, and recq4a recq4b backgrounds and even in plants transformed with additional copies of HEI10. In such plants, recombination occurs preferably within gene transcribed regions and is reduced in highly polymorphic intergenic regions (Serra et al., 2018a). Likely mechanisms are: 1) MSH2-mediated mismatch recognition, which may lead to dissolution of strand-invasion events at the megabase level (see Table 2) (Serra et al., 2018b; Emmanuel et al., 2006), and 2) the presence of the chromatin marker H3K4me3, which in budding yeast is known to interact with meiotic Mer2 (Serra et al., 2018a).

INTERACTION BETWEEN RING LIGASES AND AXR1

AUXIN RESISTANT1 (AXR1: At1g05180) (see Table 2) is a RELATED TO UBIQUITIN1 (RUB1: At1g31340)-activating enzyme. Rubylation is a modification of Cullin-RING E3 ligases that may lead to protein activation (Jahns et al., 2014). The axr1 mutant displays defects in auxin responses due to impaired degradation of an AUX/IAA repressor (Del Pozo et al., 1998). Notably during meiosis, the absence of AXR1 produces defects in SC formation and reduced bivalent formation. This reduction may be due to a variation in chiasma localization without affecting class I CO frequency (Jahns et al., 2014). AXR1 may operate as an E1 enzyme that modulates the activity of Cullin RING Ligase 4 (CRL4: At1g60986), a protein associated with DNA repair in plants and humans (Jahns et al., 2014).

CONCLUSIONS

Meiosis involves numerous and fine-tuned chromosomal processes that must be regulated in a coordinated fashion. Ubiquitinated targets may include proteins involved in the maintenance of chromosome structure, meiotic recombination, axis assembly and SC formation. Future characterization of key proteins such as HEI10 should provide a better picture of the interplay between ubiquitination, ubiquitin modifiers, meiotic recombination, adaptation and evolution in higher plants. These studies may lead to the development of methods to increase homologous recombination, or to produce gametes with specific genetic make-up (i.e. diploid and clonal) that might accelerate breeding in important crops such as rice. However, this approach remains mostly unexplored in genomes that are much larger than in Arabidopsis (Lambing et al., 2017).