Identification of ASYNAPTIC4, a Component of the Meiotic Chromosome Axis

ASYNAPTIC4 is required for normal meiotic recombination and synapsis in Arabidopsis thaliana.

Abstract

During the leptotene stage of prophase I of meiosis, chromatids become organized into a linear looped array via a protein axis that forms along the loop bases. Establishment of the axis is essential for the subsequent synapsis of the homologous chromosome pairs and the progression of recombination to form genetic crossovers. Here, we describe ASYNAPTIC4 (ASY4), a meiotic axis protein in Arabidopsis (Arabidopsis thaliana). ASY4 is a small coiled-coil protein that exhibits limited sequence similarity with the carboxyl-terminal region of the axis protein ASY3. We used enhanced yellow fluorescent protein-tagged ASY4 to show that ASY4 localizes to the chromosome axis throughout prophase I. Bimolecular fluorescence complementation revealed that ASY4 interacts with ASY1 and ASY3, and yeast two-hybrid analysis confirmed a direct interaction between ASY4 and ASY3. Mutants lacking full-length ASY4 exhibited defective axis formation and were unable to complete synapsis. Although the initiation of recombination appeared to be unaffected in the asy4 mutant, the number of crossovers was reduced significantly, and crossovers tended to group in the distal parts of the chromosomes. We conclude that ASY4 is required for normal axis and crossover formation. Furthermore, our data suggest that ASY3/ASY4 are the functional homologs of the mammalian SYCP2/SYCP3 axial components.

Meiosis is the specialized cell division that produces the haploid cells from which the gametes will be generated. In most organisms, this reduction in ploidy is achieved by first segregating the homologous chromosomes from each other (meiosis I), then by separating the sister chromatids at meiosis II. The correct meiotic course relies on a series of coordinated mechanisms that take place during meiotic prophase I. They include the organization of sister chromatids along a common proteinaceous axis (the axial element [AE]), the pairing and the synapsis of these axes, recombination, and the formation of at least one crossover (CO) per homologous pair (Zickler and Kleckner, 1999).

The AEs are assembled early during meiotic prophase I, defining the leptotene stage. Then, axes from the homologous chromosomes become connected by the polymerization of the central element of the synaptonemal complex (SC), forming the lateral elements of the SC. The polymerization of the SC is complete by pachytene, a stage at which the maturation of recombination intermediates into COs is achieved, at least in Saccharomyces cerevisiae (Zickler and Kleckner, 1999). Next, the central element of the SC is disassembled while the chromosome axis participates in the dramatic chromosome condensation that occurs during the remaining steps of meiotic prophase I (diplotene, diakinesis).

Therefore, a defining feature of meiotic chromosomes is that sister chromatids share a chromosome axis to which they are anchored, forming regular arrays of chromatin loops. Because most of the recombination proteins are axis associated, it has been proposed that meiotic chromosome axes form a scaffold on which meiotic recombination takes place (Blat et al., 2002; Panizza et al., 2011). Notwithstanding these structural roles, chromosome axes also appear highly flexible and dynamic. Their physical association with the chromosomes depends on and is responsive to underlying transcriptional activity (Sun et al., 2015). Some of their components are displaced upon synapsis and during recombination, where there is a requirement for localized axis exchange at CO sites.

Chromosome axes are composed of various protein families (Zickler and Kleckner, 1999). Cohesins (and notably the meiosis-specific Rec8 protein) as well as cohesin-associated factors such as the condensins are key components of the AEs. Cohesins form ring-shaped complexes that associate sister chromatids together after replication and that, in S. cerevisiae, anchor the other AE proteins to chromatin (Sun et al., 2015). The HORMA domain proteins (Hop1 in S. cerevisiae, HormaD1 and HormaD2 in mammals, ASY1 [ASYNAPTIC1]/PAIR2 [HOMOLOGOUS PAIRING ABERRATION IN RICE MEIOSIS2] in plants, and HIM-3 [HIGH INCIDENCE OF XO MALES-3], HTP-1, HTP-2, and HTP-3 [HIM THREE PARALOG-1, -2, and -3] in Caenorhabditis elegans) also represent major components of the meiotic chromosomal axes that, in C. elegans, constitute the linker between the cohesins and the SC central element (Pattabiraman et al., 2017). In several organisms, including Arabidopsis (Arabidopsis thaliana), their axis association is negatively regulated by synapsis (Börner et al., 2008; Wojtasz et al., 2009; Lambing et al., 2015). The last class of known AE proteins contains the S. cerevisiae Red1 (REDUCTIONAL DIVISION1), the mouse SYCP2 and SYCP3 (SCP2 and SCP3 in rat; SYNAPTONEMAL COMPLEX PROTEINS2 and -3), and the plant ASY3/PAIR3/DSY2 (DESYNAPTIC2; Wang et al., 2011; Ferdous et al., 2012; Lee et al., 2015). All these proteins are meiosis-specific components of the AE. Red1, SYCP2/SCP2, and ASY3/PAIR3 are large proteins that show limited sequence similarities, suggesting that they could be distantly related (Offenberg et al., 1998; Ferdous et al., 2012). Concerning the mammalian SYCP3/SCP3, they are small proteins that show sequence similarities with SYCP2/SCP2, with which they interact through their coiled-coil regions. They are thought to represent key structural components of the mammalian meiotic chromosome axes, since, notably, they form multistranded fibers that mimic the AEs when expressed ectopically in somatic cells (Yuan et al., 1998; Pelttari et al., 2001). In addition, structural resolution of human SYCP3 revealed that it forms elongated helical tetrameric structures that self-assemble into AE-like fibers that possess the intrinsic capacity of mediating double-stranded DNA compaction (Syrjänen et al., 2014, 2017).

Mutants defective in any component of the AE exhibit substantial perturbation of the meiotic recombination process. The plant HORMA domain-containing protein ASY1 is not required for normal double-strand break (DSB) formation but for DMC1 stabilization on recombination sites (Armstrong et al., 2002; Sanchez-Moran et al., 2007). In consequence, in asy1 mutants, meiotic DSBs are repaired predominantly using a sister chromatid as template, as is the case in a dmc1 mutant, provoking a shortage in CO formation (Sanchez-Moran et al., 2007). The axial protein ASY3/PAIR3/DSY2, on the other hand, is required for normal levels of DSB formation in Arabidopsis and in maize (Zea mays; Ferdous et al., 2012; Lee et al., 2015). It is also required for normal ASY1 assembly onto the chromosome axis, and it interacts with ASY1 (Ferdous et al., 2012; Lee et al., 2015) and with ZYP1 (Lee et al., 2015).

In this article, we identify ASY4, a short coiled coil-containing protein showing similarity with the ASY3 C-terminal coiled-coil region. We show that ASY4 is an axis-associated protein that interacts with ASY1 and ASY3. We also found that ASY4 is required for normal ASY1 and ASY3 localization, for full synapsis, and for CO formation.

RESULTS

Identification of ASY4, a Meiotic Gene with Similarity to ASY3

A BLASTP search against the Arabidopsis genome using ASY3 (At2G46980) as a query identified the uncharacterized At2g33793 protein (hereafter called ASY4) as showing 29% identity and 45% similarity with 142 amino acids of the C-terminal region of ASY3 (Fig. 1). While ASY3 is a large protein (793 amino acids, 88 kD), ASY4 is only 212 amino acids long (25 kD). Its sequence does not contain any known functional domains, and most of ASY4 is predicted to form coiled coils (amino acids 71–183; Fig. 1). Homologs of ASY4 can be identified in Tracheophyta sequenced genomes (which include flowering plant genomes and Sellaginella moellendorffii). Outside Tracheophyta, an ASY4 homolog is found in Marchantia polymorpha but not in mosses. Reverse transcription PCR on Arabidopsis cDNAs isolated from different organs from wild-type plants showed that ASY4 is expressed predominantly in flower buds (Supplemental Fig. S1).

Figure 1.

Schematic representation of the ASY4 protein and the ASY4 gene. A, The ASY4 protein shows similarities with the ASY3 C-terminal region (dashed lines). Predicted coiled coils of both proteins are indicated by gray boxes. Numbers refer to amino acids. B, The ASY4 open reading frame and the positions of the T-DNA insertion in asy4-1 and asy4-2 mutants. Exons are shown as gray boxes.

To analyze ASY4 function, we characterized two independent mutant lines in At2g33793. One was available in the public databases: line SK22114 (stock CS1006148, later referred to as asy4-1). The second one (asy4-2) was isolated by PCR screening of Max Planck Institute for Plant Breeding Research Arabidopsis T-DNA insertion mutants (Ríos et al., 2002). Insertions in the asy4-1 and asy4-2 mutants are located in the fourth and fifth exons of ASY4, respectively, and are associated with deletions of 17 and 19 bp, respectively (Fig. 1; Supplemental Fig. S2). Residual transcription corresponding to the 5′ end of the gene can be detected in both mutants (Supplemental Fig. S1). They could potentially generate a C-terminally truncated protein of 92 or 106 amino acids, respectively.

Both asy4 mutants investigated in this study showed normal vegetative growth (data not shown) but fertility defects (Supplemental Fig. S3) that correlated with meiotic defects (Fig. 2). During prophase I in wild-type meiosis, the 10 Arabidopsis chromosomes condense and recombine, resulting in the formation of five bivalents, each consisting of two homologous chromosomes attached to each other by sister chromatid cohesion and chiasmata (the cytological manifestation of COs), which become visible at diakinesis. Synapsis (the close association of two chromosomes mediated by the SC) begins at zygotene and is complete by pachytene. At metaphase I, the five bivalents are easily distinguishable and aligned on the metaphase plate. During anaphase I, each chromosome separates from its homolog, leading to the formation of dyads corresponding to two pools of five chromosomes. The second meiotic division then separates the sister chromatids, generating four pools of five chromosomes, which gives rise to tetrads of four haploid daughter cells. In asy4 mutants, each of these meiotic stages can be identified, although full synapsis was not detected. Moreover, the presence of univalent chromosomes at diakinesis and unbalanced tetrads (illustrated for the asy4-1 mutant in Fig. 2) indicates a defect in CO formation.

Figure 2.

ASY4 is required for normal meiosis. A to J, 4′,6-Diamino-phenylindole (DAPI) staining of meiotic chromosomes in the wild type (A, C, E, G, and I) and the asy4-1 mutant (B, D, F, H, and J). A and B, Leptotene. C, Pachytene. D, Partial synapsis typical of the defects of synapsis observed in asy4 mutants. E and F, Diakinesis. G and H, Metaphase I. I and J, End of anaphase II. u, Univalent; *, rod bivalent. Bars = 5 μm. K, Quantification of the number of chiasma that can be identified at metaphase I (MCN) in both asy4 mutants as well as in a series of mutants and multimutants. Numbers give the average MCN per cell. The detailed data set can be found in Supplemental Figure S4.

The reduction in chiasma number observed in asy4 meiocytes was quantified at the transition between metaphase I and anaphase I by estimating the number of chiasma based on bivalent shape. Rod bivalents reflect the occurrence of a minimum of one chiasma on a single chromosome arm pair, whereas ring bivalents reflect the occurrence of at least one chiasma per chromosome arm. This estimation provides a minimum chiasma number (MCN; as defined by Jahns et al. [2014]), because multiple chiasmata on a single bivalent arm cannot generally be discriminated from single chiasma. In both asy4 mutants, the MCN is decreased significantly in comparison with the wild type, with the asy4-1 allele being the most affected, showing an average of 5.9 ± 1.5 MCN per cell (in the wild type, the mean number of MCN per cell is 8.9 ± 0.89; Student’s t test, P < 0.0001; Fig. 2; Supplemental Fig. S4). Therefore, all subsequent analyses were conducted with the asy4-1 mutant.

This phenotype of a decrease in chiasma formation associated with abnormal synapsis has been described previously for mutants defective in axis formation typified by the asy1 and asy3 mutants (Armstrong et al., 2002; Ferdous et al., 2012). Therefore, we analyzed the epistatic relationships between these various mutations. This revealed that, in terms of chiasma level, the asy1 mutation is epistatic to the asy3 and asy4 mutations, with asy1 asy3 and asy1 asy4 double mutant combinations showing only 2 MCN per cell (Fig. 2; Supplemental Fig. S4). When analyzing the double mutation asy3 asy4, however, we found that the average number of chiasmata per cell is intermediate between the asy3 and asy4 mutations (4.1 ± 1.3 MCN per cell) and significantly different from each single mutant (one-way ANOVA, P < 0.0005).

asy4 Mutants Are Defective in Meiotic Recombination

In order to understand the origin of the reduced chiasma formation observed in the asy4 mutants, we investigated meiotic recombination in further detail. First, we immunolocalized DMC1, a meiosis-specific recombinase that forms foci at recombination sites. In the wild type, DMC1 foci appear at late leptotene/early zygotene, reaching an average of 240 foci per nucleus (Chelysheva et al., 2007). In the asy4-1 mutant, we counted an average of 222 ± 107 (n = 15) foci per cell, suggesting that early recombination events are not affected in asy4 (Supplemental Fig. S5). We then immunolocalized the ZMM protein MSH5, a MUTATOR S homolog, that is involved in the stabilization of progenitor double-Holliday Junctions and HEI10 (ENHANCER OF CELL INVASION NO. 10), which has been shown to mark a subset of recombination intermediates that are channeled into the ZMM pathway (Snowden et al., 2004; Higgins et al., 2008; Chelysheva et al., 2012). MSH5 foci were detected in the wild type and the asy4-1 mutant at late leptotene/early zygotene (Fig. 3, A and B). No significant difference in the number of foci was observed (wild type = 110.9 ± 38.61, n = 15; asy4-1 = 121.1 ± 29.55, n = 15; Mann-Whitney U test, P = 0.3835). This implies that recombination in the asy4-1 mutant progresses beyond DMC1-catalyzed strand invasion. HEI10 is loaded early during prophase I on a large number of recombination sites, forming foci of different sizes on chromosomes. As meiosis progresses, HEI10 foci become brighter and associated with the central element of the SC (ZYP1; Fig. 3C). During pachytene, a limited number of these foci remain at sites that correspond to class I COs, where they colocalize with MLH1 until the end of prophase (data not shown). In the asy4-1 mutant, the HEI10 dynamics were similar to those in the wild type, with foci of mixed sizes colocalizing with ZYP1 while synapsis progresses (Fig. 3D). However, ZYP1 staining was very limited, never progressing to full synapsis, confirming the chromosome synapsis defects detected after DAPI staining of the chromosomes (Fig. 2). In consequence, the pachytene-like HEI10 foci observed on the partially synapsed nuclei were decreased strongly in comparison with the wild type (Fig. 3D).

Figure 3.

The asy4 mutant is defective in recombination, axis biogenesis, and synapsis. A and B, Dual ASY1 and MSH5 immunodetection. Colors are as follows: ASY1 (green), MSH5 (red), and DAPI (blue). Images are single frames from mid Z-stack. Bars = 2 μm. C and D, Dual ZYP1 and HEI10 immunodetection together with DAPI (blue) on male meiocytes at a similar stage. Bars = 2 μm. E to H, Dual ASY1 (green) and ZYP1 (red) immunodetection. Arrows indicate synapsed regions where ASY1 is depleted in the wild type (WT) but not in the asy4-1 mutant. Bars = 2 μm.

We then estimated the average number of the class I COs (which rely on the ZMM pathway and are sensitive to interference from neighboring COs) in the asy4 mutant by immunolabeling chromosomes with antibodies directed against MLH1, a marker of class I COs (Fig. 4). We found that the asy4-1 mutant shows a limited but significant decrease in MLH1 foci from 11 ± 1.5 (mean ± sd; n = 60) in the wild type to 8.6 ± 2.2 (n = 147) in the asy4-1 mutant (Student’s t test, P < 0.05). We then analyzed the distribution of these foci within bivalents. We kept in our analysis all pairs of chromosome arms where at least one MLH1 foci can be observed at diakinesis. In wild-type meiocytes, the mean number of MLH1 foci per chromosome arm is 1.4 ± 0.52 (n = 180; range, 1–3), whereas in the asy4-1 mutant, it increased significantly (P < 0.0001, Student’s t test) to a mean of 1.8 ± 0.85 (n = 134), with a much greater range of values than in the wild type (1–6 compared with 1–3 in the wild type; Fig. 4B). In order to confirm these results, we analyzed the level of recombination in four genetic intervals located on chromosome 5 using the Fluorescent Tagged Lines (FTL) tool (Berchowitz and Copenhaver, 2008). For most intervals (three out of four), recombination rates decreased significantly but moderately in the asy4 mutant, reaching, on average, 75% of the wild-type level of recombination (Table 1). This effect is comparable to the decrease in chiasma number observed in asy4 mutants (Fig. 2). However, the I5b interval, which is distally located on chromosome 5, appears differentially affected, since meiotic recombination increases slightly but significantly in asy4 mutants (from 16 to 20 cM; Table 1). In conclusion, the asy4 mutation provokes a decrease in meiotic recombination, but this effect appears to vary according to the chromosomal intervals considered.

Figure 4.

MLH1 detection and quantification. A, MLH1 was immunolocalized (green) on diakinesis chromosomes from the wild type (Wt) or the asy4-1 mutant (asy4). Chromosomes were stained by DAPI (red). Bars = 5 μm. B, Left side of the graph (black dots), number of MLH1 foci per cell at diakinesis in the wild type and the asy4-1 mutant. Right side of the graph (green dots), number of MLH1 foci per chromosome arm, measured among the pairs of chromosome arms where at least one MLH1 foci can be observed at diakinesis. Asterisks indicate significant differences between the wild type and the mutant in both cases (*, P < 0.05 and ***, P < 0.0001).

Table 1. Recombination rates and interference.

Recombination rates were measured in four chromosome 5 intervals (I5a–I5d). For each interval, the map distance in centimorgan (cM) was calculated using the Perkins genetic map equation (PMID 17247336). The distance ratio compares the recombination rates between the wild type and asy4-1. The nonparental ditype (NPD) ratio and the interference ratio (IR) give the strength of interference either within the considered interval or among two adjacent intervals (no interference if the ratios are equal to 1, absolute interference if the ratios are equal to 0). Asterisks indicate significant differences between ratios and 1 (*, significant at 5% and **, significant at 1%).

Plant	Interval	No. of Tetrads	Map Distance	Distance Ratio (asy4/Wild Type)	NPD Ratio	IR
Wild type	I5a	10,303	27	–	0.3**	0.4**
	I5b	10,303	16.1	–	0.2**
	I5c	14,590	7.7	–	0.3**	0.3**
	I5d	14,590	7.4	–	0.3**
asy4-1	I5a	7,462	15.5	0.6	0.9	1.2**
	I5b	7,462	20	1.2	0.6**
	I5c	13,753	6.8	0.9	0.4**	0.7**
	I5d	13,753	5.6	0.8	0.5*

In Arabidopsis, most COs are under the control of the ZMM pathway and exhibit interference (Mercier et al., 2015). We analyzed asy4 zip4 and asy4 msh5 double mutants and found that the level of bivalent formation was reduced dramatically by more than 95% (Fig. 2), showing that, in mutants as in the wild type, most of the COs were dependent on the ZMM pathway. We then estimated the level of interference between COs in each FTL interval by calculating the ratio between the observed number of double COs and the expected number of double COs under the hypothesis of no interference (NPD ratio as defined by Snow [1979]). We observed that, in most intervals considered, in the asy4 mutants as in the wild type, the NPD ratio is smaller than 1, revealing the presence of interference between adjacent COs.

Then, the interference between COs occurring in adjacent intervals (I5a/I5b or I5c/I5d) was estimated by calculating the IR as defined by Malkova et al. (2004). The IR compares the genetic length of one interval with and without the presence of a simultaneous event in the neighboring interval. When the occurrence of a CO in one interval reduces the probability of a CO occurring in the adjacent interval, the IR is less than 1, indicating CO interference. When COs in the two adjacent intervals are independent of each other, the IR is 1, and if the presence of one CO in an interval increases the probability of an additional CO in the adjacent interval, the IR is greater than 1, indicating negative interference. IRs revealed the presence of interference between COs in the wild type (for both pairs of intervals) and for the asy4 mutants for the I5c/I5d pair of intervals (Table 1). However, for the I5a/I5b pair of intervals, the IR in asy4 mutants is above 1, suggesting that, in that chromosomal region, adjacent COs occur more frequently than in the wild type. Taken together, these data show that the asy4 mutation perturbs meiotic recombination quantitatively (by decreasing it) and qualitatively (by altering CO location).

The asy4 Mutation Is Associated with Axis Defects

We investigated the behavior of several components of the meiotic chromosome axis (ASY1, ASY3, REC8, and SCC3) in the asy4 mutant in comparison with the wild type (Figs. 3 and 5; Supplemental Fig. S6). ASY1, ASY3, REC8, and SCC3 are detected during meiotic prophase I and exhibit different dynamics as meiosis progresses (Armstrong et al., 2002; Cai et al., 2003; Chelysheva et al., 2005; Ferdous et al., 2012). At leptotene, immunodetection showed that all these proteins decorate meiotic chromosomes, revealing signals all along the typical thread-like chromosomal axis. As synapsis proceeds and the central element connects the AEs of the homologous chromosomes, ASY1 is depleted from the axis and, consequently, the ASY1 signal appears faint and fuzzy (Fig. 3, arrows; Supplemental Fig. S6). ASY3, REC8, and SCC3 also mark the chromosome axes, but contrary to ASY1, they are not removed during synapsis (Fig. 5; Supplemental Fig. S6). In the case of the cohesins REC8 and SCC3, no obvious modification in their pattern could be detected (Fig. 5; Supplemental Fig. S6). The two axis-associated proteins ASY1 and ASY3 are loaded onto the chromosome axis, and chromosome threads typical of leptotene stages can be seen. However, ASY1 and ASY3 signals adopt an abnormally patchy and lumpy aspect (Figs. 3 and 5) that is very unlikely to originate from the synapsis defect of asy4 mutants; instead, it suggests that ASY4 is required for normal chromosome axis structure. In addition, we observed no displacement of ASY1 from the synapsed chromosome axes (Fig. 3, zoom), revealing abnormal axis dynamics. We investigated the chromosome axis further by silver staining of chromosome spreads and wide-field microscopy observation as described by Armstrong and Jones (2001). This chromatin staining permits the detection of the meiotic chromosome axis from leptotene to the end of meiosis. In the asy4 mutant as well as in asy3 asy4 and asy1 asy3 asy4 multiple mutants, no modification of the silver-stained axis could be detected (Fig. 5), suggesting that, even if axis composition and/or dynamics is affected in asy4, at this level of resolution, the overall structure of the axis appears physically intact.

Figure 5.

Chromosome axis investigation. A to D, Dual ASY3 (green) and REC8 (orange) immunolocalization on wild-type (Wt; A and B) or asy4-1 mutant (asy4; C and D) male meiocytes. E and F, Silver staining of wild-type and triple asy1 asy3 asy4 mutant male meiocytes. Bars = 2 μm.

ASY4 Is an Axis-Associated Protein

To examine the cellular localization of ASY4, we used fluorescent protein tagging. An ASY4-eYFP construct was produced and introduced into homozygous asy4-1 plants, the most severely affected mutant background. Seed counts were performed on siliques from T2 generation plants (Supplemental Fig. S7). Fertility levels across the transformant lines were wide ranging, from those similar to the asy4-1 mutant to a line that was not significantly different from the wild type (line 165.15, subsequently referred to as asy4-1::ASY4eYFP; Supplemental Fig. S7). Analysis of DAPI-stained chromosome spreads of asy4-1::ASY4eYFP male meiocytes from T3 plants at metaphase I revealed a chiasma frequency of 7.7 ± 1.1 (n = 75). This was significantly higher than in the asy4-1 mutant (5.9 ± 1.43 [n = 64]; Mann-Whitney U test, P < 0.01). However, it was slightly lower than in the wild type (8.6 ± 0.83 [n = 28]; Mann-Whitney U test, P < 0.01; Fig. 2; Supplemental Fig. S7). In addition, occasional seed gaps in its siliques were apparent, suggesting that fertility was not restored completely (Supplemental Fig. S7).

Examination of the anthers from asy4-1::ASY4eYFP plants using epifluorescence microscopy confirmed the expression of the tagged gene within male meiocytes (Supplemental Fig. S7). Localization of ASY4eYFP was then investigated in prophase I chromosome spread preparations by direct fluorescence combined with immunostaining of the chromosome axis protein, ASY1, and the SC protein, ZYP1. This revealed that ASY4 localizes as a linear, axis-associated signal at leptotene, where it follows the localization pattern of ASY1 with alternating regions of high and low intensity (Fig. 6). However, in contrast to ASY1, which becomes depleted from the axes as zygotene progresses, it persists on synapsed regions of the chromosomes (Fig. 6). In this respect, its behavior is similar to that of ASY3, REC8, and SCC3.

Figure 6.

Localization of ASY4eYFP in prophase I chromosome spreads of asy4-1::ASY4eYFP plants. A, Wild-type (Col 0) zygotene showing the absence of eYFP fluorescence. B and C, asy4-1::ASY4eYFP leptotene (B) and asy4-1::ASY4eYFP zygotene (C). D, Details of the ASY4eYFP fluorescence present on the axis in regions of intense ASY1 staining (unsynapsed) and ZYP1 staining (synapsed). Note the reduction in intensity of ASY1 signal in synapsed regions (white arrowheads). E, ASY4eYFP fluorescence is not uniform and alternates between regions of high (yellow arrowheads) and low intensity. Colors are as follows: ZYP1 (blue) and ASY1 (red) immunostaining with ASY4-eYFP fluorescence (green). Bars = 5 µm.

Considering the similarity between the ASY3 and ASY4 protein sequences, the axial association of these two proteins (Ferdous et al., 2012; this study), and the perturbed ASY1 and ASY3 signals observed in the asy4 mutant, we investigated whether these proteins interact physically. An interaction between ASY1 and ASY3 has already been demonstrated for Brassica oleracea and Arabidopsis proteins either in planta by coimmunoprecipitation of ASY3 from anthers by antibodies directed against ASY1 or in yeast two-hybrid experiments using the Arabidopsis proteins (Ferdous et al., 2012). Here, we used bimolecular fluorescence complementation (BiFC) assays in leaf epidermal cells of Nicotiana benthamiana plants (Hu et al., 2002). Fusion proteins with complementary YFP truncations (YFP^N + YFP^C) were coinfiltrated in N. benthamiana leaves expressing a cyan fluorescent protein (CFP) nuclear marker. As shown in Figure 7 and Supplemental Figure S8, this assay revealed interactions among the three ASY proteins and also self-interaction of these three proteins. The YFP signal recovered in these experiments using ASY3 or ASY4 fusion proteins revealed nonuniform nucleus-targeted signals, suggesting that these proteins when overexpressed in plant cells form nuclear aggregates. Yeast two-hybrid experiments confirmed ASY3-ASY4 interactions as well as ASY3-ASY3 and ASY4-ASY4 self-interactions (Supplemental Fig. S9).

Figure 7.

Split-YFP assays in N. benthamiana epidermal cells. N. benthamiana epidermal cells were coinfiltrated with Agrobacterium tumefaciens cultures expressing two complementary YFP fusions (N- or C-terminal truncations, YFPN or YFPC). Nuclei are identified by a constitutively expressed fluorescent nuclear protein (H2B-CFP, here shown in red). The interaction between the two tested proteins revealed a YFP signal (green). For each interaction tested, a negative control corresponding to the coinfiltration of one of the fusion proteins of interest with the complementary YFP moiety fused with an unrelated protein (Antirrhinum majus MADS box transcription factor DEFICIENS [DEF] or GLOBOSA [GLO]). The complete set of split-YFP data can be found in Supplemental Figure S8. Bars = 25 µm.

DISCUSSION

Conserved Composition of the Meiotic Axial Element in Spite of Poor Primary Protein Sequence Conservation

We identified ASY4, which shows sequence similarity with the ASY3 C-terminal region and is closely related with two of the known plant axial components, ASY1/PAIR2 and ASY3/PAIR3/DSY2. The three proteins interact together, and ASY4 is required for normal loading and/or stabilization of ASY1 and ASY3 onto chromosomes. We also found that an ASY4-eYFP fusion protein is axis associated, leading us to conclude that ASY4 is a component of the meiotic chromosome axis.

The link between ASY3 and ASY4 can be viewed as a parallel with those existing between the mammalian SYCP2/SCP2 and SYCP3/SCP3: ASY3 and SYCP2/SCP2 are large proteins that show limited sequence similarities with the small coiled-coil proteins ASY4 and SYCP3/SCP3, respectively (as an example, SCP3 shows 19% amino acid identity and 47% amino acid similarity with the last 163 amino acids of SCP2); ASY3 and ASY4 interact together (this study) as well as the mammalian SYCP3 and SYCP2 (Yang et al., 2006); all these proteins are axis-associated proteins (Offenberg et al., 1998; Schalk et al., 1998; Yang et al., 2006; Ferdous et al., 2012; this study). In addition, limited sequence similarities can be detected between ASY3/SYCP2 and the S. cerevisiae Red1 axial component (Offenberg et al., 1998; Ferdous et al., 2012). The close interconnection between these proteins and the HORMA domain-containing protein ASY1 in plants (Wang et al., 2011; Ferdous et al., 2012; Lee et al., 2015; this study) and HormaD1 and HormaD2 in mammals (Wojtasz et al., 2009) suggest that, altogether, they form a protein complex crucial for the biogenesis of the meiotic chromosome axis scaffold. Taken together, these data suggest that ASY3/ASY4 are the functional homologs of the mammalian SYCP2/SYCP3. It is interesting that these proteins of the AE as well as those that form the CE of the SC are very poorly conserved at the sequence level but that all show the same structure and assembly characteristics (Fraune et al., 2016). This limited sequence conservation among SC proteins from different species probably is due to rapid sequence divergence, as has been observed for plant and mammalian SC proteins (Ferdous et al., 2012; Fraune et al., 2016).

ASY4 Is Required for Normal Meiotic Recombination

According to chiasma and MLH1 foci counting and to genetic measurement of recombination using FTL lines, CO formation is reduced by a factor of 1.5 in asy4 mutants. This occurred with a clear decrease in HEI10 and MLH1 foci at late prophase I and diakinesis, showing that ASY4 is required for normal recombination. It should be noted that the CO decrease observed in asy4 is lower than the one associated with disruption of either of the two ASY4 partners, ASY1 and ASY3. In terms of chiasma level, the asy1 mutation is the most affected and is epistatic to asy3 and asy4. This suggests that, among the three axis components ASY1, ASY3, and ASY4, the HORMA domain-containing protein ASY1 is a key player, while ASY3 and ASY4 could be seen as accessory proteins. Nevertheless, we cannot exclude the possibility that the partially penetrant phenotype of asy4 is due to leaky mutations, since we could detect the transcription of the 5′ end of the gene in both mutants.

Interestingly, we observed that the decrease in recombination observed in asy4 mutants is differentially distributed within the genome, since we found that one interval out of four tested (I5b) revealed an increase in CO level (from 16 to 20 cM). This could be related to the distal location of this interval on chromosome 5 and to the observation that the vast majority of chiasma are terminally located in asy3 and asy1 mutants (Ross et al., 1997; Ferdous et al., 2012). Two other findings of our study confirm that CO location is modified in asy4. First, despite the average decrease in MLH1 foci in asy4 mutants, we detected an increased number of MLH1 foci per chromosome arm in comparison with the wild type, with up to six foci in the same arm in the asy4 mutant, while we never observed more than three per chromosome arm in the wild type. Second, we found an IR greater than 1 for one pair of intervals tested by FTL (I5a/I5b). This latter result involves the I5b terminally located interval on chromosome 5, suggesting that the two phenomena may be connected and that, in the asy4 mutants, COs are not only decreased but also tend to group in the distal parts of the chromosomes. In this regard, it is interesting that we reported recently that, in Arabidopsis as in most species, synapsis is initiated preferentially from the distal parts of the chromosomes (Hurel et al., 2018). If this also is the case in the asy4 mutants, the limited number of ZYP1-labeled central elements on which recombination events appear to be restricted (according to HEI10 labeling; Fig. 3) are expected to be predominantly distally located. This could explain why we observed a bias in the location of the COs in the asy4 mutants. Further studies will be required to confirm these observations on a genome-wide scale and to understand the mechanisms involved.

According to our study, the decrease in CO formation measured in asy4 is not correlated with a decrease in the overall number of early initiation events, since the number of DMC1 and MSH5 foci was unchanged in the asy4-1 mutant in comparison with the wild type. It is interesting that the role in recombination of the three ASY proteins can be differentiated: ASY1, like ASY4, is not required for normal DSB formation but, contrary to ASY4, is mandatory for the formation of stable DMC1 nucleofilaments (Sanchez-Moran et al., 2007), while ASY3 is required at the step of DSB formation (Ferdous et al., 2012). Chromosome fragmentation was not detected in asy4, showing that the DMC1-labeled recombination events are eventually repaired, using either the sister chromatid or the homologous chromosome as a template. Since the number of MSH5 foci at early/mid prophase I appeared normal in the asy4-1 mutant, it would seem likely that recombination proceeds beyond the initial strand invasion stage. This would imply that CO designation, which occurs in early prophase I (Lambing et al., 2017), is normal in the mutant but that a proportion of the designated intermediates fail to mature into COs, consistent with the observed reduction in MLH1 and HEI10 foci. The defect in SC polymerization observed in asy4 may result in CO designated recombination intermediates that lie within regions of the homologs that remain asynaptic failing to form COs. Establishing the exact relationship between the loss of ASY4 and the defect in SC formation will be the target of future investigation.

MATERIALS AND METHODS

Plant Materials and Growth Conditions

The asy4-1 mutant (SK22114, CS1006148) was available in public databases and was provided by the Nottingham Arabidopsis Stock Centre (http://arabidopsis.info/; Scholl et al., 2000). The asy4-2 mutant (line 65433) was identified through a PCR-based screen of the Koncz collection (Ríos et al., 2002). Other mutant alleles used in this study are asy1 (SALK_046272, N546272), asy3 (SALK_143676, N643676), dmc1 (SAIL_170_F08, N871769), mer3 (SALK_091560, N591560), mlh1 (SK25975, N1008089), msh5 (SALK_026553, N526553), rad51 (GABI_134A01), and zip4 (SALK_068052, N568052). Genotyping conditions and primer sequences are given in Supplemental Tables S1 and S2).

Arabidopsis (Arabidopsis thaliana) and Nicotiana benthamiana plants were grown in the greenhouse (photoperiod of 16-h day/8-h night, temperature of 20°C day and night, and humidity of 70%; photoperiod of 13-h day/11-h night and temperature of 25°C day and 17°C night, respectively).

Clone Construction

ASY4 cDNA was amplified from flower bud cDNA (Columbia-0) after two rounds of nested PCR (PCRI, AtASY4RTF and AtASY4RTR; PCRII, AtASY4attB1 and AtASY4attB2; Supplemental Table S1) and cloned into pDONR207 (Invitrogen) following the manufacturer’s instructions. The generated entry vector was sequenced and used to transfer ASY4 cDNA into the yeast two-hybrid expression vectors pDEST-GADT7 and pDEST-GBKT7 (Rossignol et al., 2007). To generate the C-terminal split-YFP clones (Azimzadeh et al., 2008), a version of the cDNA without a stop codon was amplified beforehand using primers AtASY4attB1 and AtASY4-attB2wostop (Supplemental Table S1). Similar approaches were undertaken for ASY1 and ASY3 cDNAs except using primers AtASY1-attB1, AtASY3-attB1, AtASY3-attB2, AtASY3-attB2wostop, and AtASY1-attB2 (Supplemental Table S1).

Yeast Two-Hybrid Assays

Yeast two-hybrid assays were carried out using the GAL4-based system (Clontech). SV40 Antigen T and p53 protein were used as positive controls. Yeast plasmids were introduced into AH109 or Y187 strains by lithium acetate transformation following the protocol in the MATCHMAKER GAL4 Two Hybrid System 3 manual (Clontech). After mating in appropriate pairwise combinations, the resulting diploid cells were selected on synthetic drop-out (SD) medium lacking a combination of amino acids, driven by the auxotrophy genes carried by the cloning vectors. Protein interactions were assayed by growing diploid cells on SD-LWH and SD-LWHA.

BiFC

Protein interactions were tested in planta using BiFC assays (Hu et al., 2002) in leaf epidermal cells of N. benthamiana plants expressing a nuclear CFP fused to histone 2B (Martin et al., 2009). For each protein, four expression vectors were produced, generating inactive N and C termini of the YFP (YFP^N and YFP^C) fused with the target sequence in N or C termini. Combinations bringing together the two YFP complementary regions (YFP^N + YFP^C) were coinfiltrated in N. benthamiana leaves as described (Azimzadeh et al., 2008; Vrielynck et al., 2016).

Bioinformatics

PSI BLAST on the nonredundant protein sequences database using ASY3 as a query picked up At2g33793 at the first round of iteration with its C-terminal region (amino acids 636–777), where coiled coils lie (amino acids 625–785, according to Ferdous et al. [2012]). BLASTP and TBLASTN on plant sequenced genomes present in the phytozome 12 database (Blosum45) were conducted to identify homologs.

Recombination Measurement

We used the FTLs described by Berchowitz and Copenhaver (2008) to estimate recombination rates in four different genomic intervals (I5a, I5b, I5c, and I5d). We generated plants that were homozygous for the quartet mutation, heterozygous for pairs of linked fluorescent markers RY/++ (I5a and i5d) or YC/++ (I5b and I5c; R = red, Y = yellow, and C = cyan) and either wild type or homozygous for the asy4-1 mutation. Tetrad analyses were carried out as described by Berchowitz and Copenhaver (2008) on tetrads where each fluorescent marker segregated correctly.

Fluorescent Protein Tagging

The ASY4 genomic locus, comprising 1,835 bp upstream of the start codon to 502 bp downstream of the stop codon and including all introns and untranslated regions, was amplified with the primers At2g33793-P9 and At2g33793-P10 (Supplemental Table S1). The eYFP sequence was inserted in frame at amino acid position 202, downstream of the predicted coiled-coil region and close to the C terminus. The construct was inserted into the p35-Nos-BM cloning vector using SfiI sites incorporated into the primers. The resulting expression cassette was subcloned via SfiI into the pLH9000 binary vector and used for Agrobacterium tumefaciens-mediated transformation of plants using the floral dip method. Transformants were selected on kanamycin (50 µg mL⁻¹) Murashige and Skoog medium (Murashige and Skoog, 1962).

Cytological Procedures

Meiotic chromosome spreads were DAPI stained as described previously (Ross et al., 1996) or silver nitrate stained as described by Armstrong et al. [2001]). Immunostaining of male meiotic spreads was carried out as described (Armstrong and Osman, 2013; Chelysheva et al., 2013). Antibodies used for immunolocalization were anti-ASY1 (rat, 1:1,000 dilution; Armstrong et al., 2002), anti-AtZYP1 (rabbit, N-terminal antibody amino acid residues 1–415, 1:500 dilution; Higgins et al., 2005), anti-ASY3 (rabbit, 1:250 dilution; Ferdous et al., 2012), anti-REC8 (rat, 1:250 dilution; Cromer et al., 2013), anti-DMC1 (rat, 1:20 dilution; Vignard et al., 2007), anti-MSH5 (rabbit, 1:200 dilution; Higgins et al., 2008), anti-MLH1 (rabbit, 1:200 dilution; Chelysheva et al., 2013), and anti-HEI10 (rabbit, 1:250 dilution; Chelysheva et al., 2012).

Image Analysis

asy4-1::ASY4eYFP zygotene male meiocyte nucleus images were captured with a Nikon 90i, 100× objective as a Z-stack. The green channel (eYFP) was processed as an average intensity projection using Fiji, due to more rapid bleaching of eYFP relative to the red (TX red-ASY1) and blue (Alexa350-ZYP1) channels, which were processed as maximum intensity projections. Columbia-0 was imaged using the same exposure times and processed in the same way. MSH5 foci were scored using Z-stack images and Mexican Hat deconvolution as described (Ferdous et al., 2012).

Accession Numbers

Sequence data from this article can be found in the GenBank/EMBL libraries under accession numbers At2G46980 (ASY3) and At2g33793 (ASY4).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. ASY4 expression.

• Supplemental Figure S2. ASY4 coding sequence and mutations.

• Supplemental Figure S3. A decrease in fertility is associated with asy4 mutations.

• Supplemental Figure S4. Average MCN measured in metaphase I cells.

• Supplemental Figure S5. DMC1 foci formation.

• Supplemental Figure S6. ASY1, REC8, and SCC3 immunolocalization.

• Supplemental Figure S7. asy4-1 complementation with an ASY4::eYFP fusion.

• Supplemental Figure S8. BiFC results.

• Supplemental Figure S9. Yeast two-hybrid assay results.

• Supplemental Table S1. Primer sequences.

• Supplemental Table S2. Mutant lines and genotyping conditions.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• DMC1 Gramene: AT3G22880

• DMC1 Araport: AT3G22880

• MLH1 Gramene: AT4G09140

• MLH1 Araport: AT4G09140

• SCC3 Gramene: AT2G47980

• SCC3 Araport: AT2G47980

• SCP2 Gramene: AT5G42890

• SCP2 Araport: AT5G42890

• asy1 Gramene: AT1G67370

• asy1 Araport: AT1G67370

• msh5 Gramene: AT3G20475

• msh5 Araport: AT3G20475

• zip4 Gramene: AT1G10970

• zip4 Araport: AT1G10970