Research Article

A DNA topoisomerase VI–like complex initiates meiotic recombination

See allHide authors and affiliations

Science  26 Feb 2016:
Vol. 351, Issue 6276, pp. 939-943
DOI: 10.1126/science.aad5196

A partner protein for meiotic snip

Eukaryotes generate germ cells through meiotic recombination. This process initiates through breaks in genomic DNA catalyzed by the SPO11 protein. Vrielynck et al. and Robert et al. discover that SPO11, like topoisomerase VI enzymes, interacts with a partner protein (see the Perspective by Bouuaert and Keeney). This partner is required for proper meiotic recombination and is found in a wide range of eukaryotes, suggesting that it is a universal feature of the essential recombination step.

Science, this issue p. 939, 943; see also p. 916


The SPO11 protein catalyzes the formation of meiotic DNA double strand breaks (DSBs) and is homologous to the A subunit of an archaeal topoisomerase (topo VI). Topo VI are heterotetrameric enzymes comprising two A and two B subunits; however, no topo VIB involved in meiotic recombination had been identified. We characterized a structural homolog of the archaeal topo VIB subunit [meiotic topoisomerase VIB–like (MTOPVIB)], which is essential for meiotic DSB formation. It forms a complex with the two Arabidopsis thaliana SPO11 orthologs required for meiotic DSB formation (SPO11-1 and SPO11-2) and is absolutely required for the formation of the SPO11-1/SPO11-2 heterodimer. These findings suggest that the catalytic core complex responsible for meiotic DSB formation in eukaryotes adopts a topo VI–like structure.

Meiotic recombination is the key step in sexual reproduction leading to the production of haploid gametes and is therefore essential for the fertility of most eukaryotes. It is initiated by the induction of DNA double strand breaks (DSBs), which are catalyzed by the evolutionarily conserved SPO11 protein (1). SPO11 shows sequence similarities to the A subunit of the archaeal type II DNA topoisomerase, topoisomerase VI (topo VI) (2). Topo VI enzymes relax DNA supercoils by cutting both DNA strands of a DNA helix in an adenosine 5′-triphosphate (ATP)–dependent manner, forcing the passage of another DNA duplex through the generated break and eventually religating the break, regenerating an intact DNA molecule (3). Topo VI enzymes are active as heterotetramers, composed of two A and two B subunits. The two topo VIA subunits carry the catalytically active tyrosines and associate as a dimer to form the catalytic core of the topo VI enzyme (2). The topo VIB subunits contain an ATP-binding domain known as the Bergerat fold, which is characteristic of the GHKL (Gyrase, HSP90, histidine kinase, MutL) superfamily (2, 4). Topo VIB also possess a C-terminal transducer domain that transfers the conformational changes induced by ATP binding and hydrolysis from the GHKL to the catalytic A subunits (5). On the basis of the similarity between SPO11 and topo VIA, and because SPO11 can be purified covalently associated with the meiotic DSB ends (6), it was proposed that at the onset of meiotic recombination, SPO11 dimerizes to form the catalytic core responsible for DNA DSB formation. In Arabidopsis thaliana, genetic analyses revealed that two nonredundant SPO11 homologs (SPO11-1 and SPO11-2) are absolutely required for DSB formation (79), suggesting that meiotic DSBs are catalyzed by a SPO11-1/SPO11-2 heterodimer rather than a homodimer. Although most eukaryotic lineages possess at least one SPO11 protein, until now no topo VIB homolog involved in meiotic recombination had been identified. Here, we present the first characterization of a structural homolog of the archaeal topo VIB subunit required for meiotic DSB formation. We show that it mediates the formation of the SPO11-1/SPO11-2 heterodimer, changing our view of the nature of the core complex involved in meiotic DSB formation.

In a screen for A. thaliana mutants with reduced fertility, we identified two allelic lines, mtopVIB-1 and mtopVIB-2, that show a quasi-sterile phenotype (4% of the wild-type level of seeds) (fig. S1, A and B) correlated with male and female gamete developmental defects (fig. S1, C to E). Staining mtopVIB male meiocyte chromosome spreads with 4′,6-diamidino-2-phenylindole (DAPI) showed a considerable reduction in bivalent formation at metaphase I (Fig. 1, A and B, and figs. S2 and S3); in wild type, five bivalents were observed in 100% of cases (n = 94 cells), whereas only univalents were observed in 86% of mtopVIB-1 and 100% of mtopVIB-2 meiocytes (n = 141 and n = 52 cells, respectively) (fig. S3). In addition, synapsis of homologous chromosomes was completely abolished in mtopVIB (n = 47 cells) (Fig. 1, C and D), and no signal above background could be detected for the meiotic recombinase DMC1 (n = 29 cells) (Fig. 1, E and F). Meiotic recombination rates also fell considerably in mtopVIB, to only a few percent of the wild-type level, which is an effect comparable with that measured for the spo11-1 mutation (table S1). These meiotic defects observed in mtopVIB mutants could be indicative of either defects in meiotic DSB formation (as in spo11 mutants) or in DSB repair, using the homologous chromosome as the repair template (as seen in the dmc1 mutants) (10). To distinguish between these two hypotheses, we introduced the mtopVIB mutation into two DSB-repair–deficient mutants, rad51 and mre11 (11, 12). We observed that the mtopVIB mutation suppressed all the fragmentation defects observed in these backgrounds (Fig. 1 and fig. S2). Taken together, these data show that mtopVIB mutants display the same meiotic defects as both spo11-1 (7) and spo11-2 mutants (8, 9) and establish an essential role for MTOPVIB in meiotic DSB formation.

Fig. 1 mtopVIB mutants are DSB-deficient.

(A and B) DAPI staining of meiotic chromosomes during the metaphase I/anaphase I transition shows that in (A) wild type (Wt), the 10 A. thaliana chromosomes associate into five bivalents that align at the metaphase plate. (B) In mtopVIB (mtopVIB-2), 10 randomly arranged univalents are observed. Scale bar, 10 μm. (C and D) Co-immunolocalization of the axis protein ASY1 (green) and the synaptonemal complex transverse filament protein ZYP1 (red). (C) In wild-type pachytene cells, ZYP1 connects the axial element of the homologous chromosomes along their whole length, as revealed by the perfect colocalization of ZYP1 with the axis protein ASY1. (D) In mtopVIB, no ZYP1 staining is detected, showing a complete absence of synapsis (mtopVIB-2). Scale bar, 2 μm. (E and F) Co-immunolocalization of the meiotic-specific cohesin REC8 (red) together with the meiotic specific recombinase DMC1 (green). (E) In wild type (Wt), ~250 recombination sites are detected by using antibodies against DMC1, whereas (F) none are visible in mtopVIB (mtopVIB-2). Scale bar, 5 μm. (G and H) DAPI staining of meiotic chromosomes during the metaphase I/anaphase I transition in (G) mre11-4 (mre11) and (H) mtopVIB-1mre11-4 (mtopVIBmre11). (G) The mre11 mutant, which is defective in meiotic DSB repair, shows severe chromosome fragmentation at the first meiotic division, but (H) this phenotype is completely abolished in the mtopVIB mutant background. Scale bar, 10 μm.

Immunolocalized MTOPVIB forms numerous foci in meiotic cells from leptotene (172 ± 9 foci, mean ± SEM, n = 30 cells) to pachytene, with a slight but significant increase (219 ± 11, n = 38 cells, P = 0.002 unpaired t test) (Fig. 2, A to F and J). The MTOPVIB signal was not detected in G2/early leptotene cells or in somatic nuclei (Fig. 2, A, E, and F) and only partially colocalized with the axis signal. We also examined MTOPVIB foci in a series of spo11 mutants characterized by a near complete absence of recombination (fig. S3) (8, 9). Foci were not detected above the background level in spo11-1 mutants but were still present in the three different spo11-2 alleles investigated (Fig. 2, H to J); because spo11-2 mutants are DSB-defective, this suggests that MTOPVIB foci do not mark active DSBs, at least in these backgrounds. Whether this is the case in wild type remains to be seen. Further, if SPO11-1—but not SPO11-2—is essential for MTOPVIB loading onto chromosomes, then this is the first evidence of a functional difference between SPO11-1 and SPO11-2 during meiotic recombination initiation.

Fig. 2 MTOPVIB forms foci on meiotic chromosomes from leptotene to pachytene.

(A to J) Co-immunolocalization of the REC8 cohesin (red signal) and MTOPVIB (green signal) in [(A) to (F)] wild-type Col-0, (G) mtopVIB-2, (H) spo11-1-3, and (I) spo11-2-3 mutants. (E) DAPI staining is shown in white. The MTOPVIB signal was not detected in (A) G2/early leptotene cells but forms numerous foci associated with the REC8 signal at (B) leptotene, (C) zygotene, and (D) pachytene. DAPI staining in (E) reveals five nuclei, four of which are not labeled, with either REC8 or MTOPVIB (F) showing that MTOPVIB is specifically expressed in meiocytes. No MTOPVIB signal was seen in (G) meiocytes from the mtopVIB mutant, or in (H) the spo11-1 mutant, but (I) foci were visible in spo11-2. (J) The number of MTOPVIB foci detected in wild type (L/Z, leptotene or early zygotene; Z/P, late zygotene or pachytene), in mtopVIB (mtopVIB-2), and in spo11 mutants (spo11-1-3, spo11-2-2, spo11-2-3, and spo11-2-4). P values from an ordinary one-way analysis of variance are indicated. ns, not statistically different. ***P < 0.0001. Scale bar, 5 μm.

The MTOPVIB gene (At1g60460) (gene and mutant molecular characterization are provided in fig. S4) encodes a 493-amino-acid protein with clear orthologs among the flowering plants (Magnoliophyta) (Fig. 3A and fig. S5A). The MTOPVIB protein family is characterized by four highly conserved motifs (b1 to b4) (Fig. 3A and fig. S5A). The MTOPVIB protein sequence has no obvious similarity to known functional domains. However, HHpred searches using this protein family as a query revealed clear structural homology with the two archaeal topo VIB proteins with known crystal structures [Sulfolobus shibatae, PDB code 2ZBK, probability = 98.5%, expect value (E) = 4.2 × 10–6 and Methanosarcina mazei, PDB code 2Q2E, probability = 95.0%, E = 0.0094]. On the basis of sequence and structure comparisons, MTOPVIB proteins comprise four domains (Fig. 3B and fig. S5B). Overall, the structure of the first domain, the GHKL/Bergerat fold, is highly conserved (Fig. 3C). Among the motifs that constitute the GHKL signatures (Fig. 3C and fig. S5B) (2), two correspond to MTOPVIB motifs b1 and b2 and include the conserved B2 motif DxGxG, which is important for ATP-binding (13). However, the glycine-rich Bergerat fold motif B3 localized in the ATP-lid is poorly conserved, and only the first glycine of this motif is present (Fig. 3C and figs. S5B and S6C), suggesting that the ATP hydrolysis mechanism could be different in MTOPVIB. The second MTOPVIB domain is a small domain (SmD), the borders of which are more difficult to define, and structural similarity with the archaea topo VIB H2TH domain is low (Fig. 3 and figs. S5B and S6A). The third domain is the transducer domain. This long α-helix structure is shorter than in the topo VIB subunits but still very well conserved (Fig. 3C and fig. S6, A and D). This domain contains the strictly conserved W (dWxxY motif) and another conserved region (b4) (Fig. 3), which are both present in the topo VIB family (fig. S5B). The most conserved region (b4) in the MTOPVIB transducer domain corresponds to the region of interaction between the A and B subunits of the topo VI heterotetramer (Fig. 3 and fig. S5B) (14). Last, the MTOPVIB C-terminal extension is different from those found in some topo VIB proteins (10).

Fig. 3 Sequence and model structure of A. thaliana MTOPVIB.

(A) Multiple sequence alignment of the plant MTOPVIB protein family showing the four conserved amino acid regions in MTOPVIB (b1 to b4). Two are in the GHKL domain, one is in the small domain, and one is in the transducer domain. Identical amino acids are in red. The 19 sequences used for the alignment are listed in the legend for fig. S5A. (B) Domain organization of the A. thaliana MTOPVIB and the S. shibatae topoisomerase VIB subunit (S. shibatae TOPVIB). Homologous domains are shown in the same color: yellow, GHKL domain; orange, Helix-2Turn-Helix (H2TH) domain and small domain (SmD); brown, transducer domain; green, C-terminal (Cter) domain; The amino acid numbers indicate the domain borders. (C) Proposed model structure of A. thaliana MTOPVIB and crystal structure of S. shibatae TOPVIB. The four conserved amino acid regions of MTOPVIB are indicated by b1 to b4 and are in red. On the S. shibatae TOPVIB crystal structure, the three GHKL conserved motifs [B1 to B3, as defined in (2)] are indicated in purple. The A-B interaction surface of the transducer is indicated in dark blue. Domains are color-coded as in (B). For each structure, two different zooms of the GHKL domains (dashed boxes) are given. For MTOPVIB and S. shibatae TOPVIB GHKL zooms 1, the aspartate and the two glycines of the DxGxG motif (b2 for MTOPVIB and B2 for S. shibatae TOPVIB) are represented in small CPK. Zooms 2 correspond to the ATP-lids, in which the Cα of the aspartate and glycines of the b2 or B2 motif (DxGxG), the single glycine in MTOPVIB, and the glycines of the B3 GxxGxG motif of S. shibatae TOPVIB are represented by small spheres.

The similarity of SPO11s to the archaeal topo VIA and of MTOPVIB to the topo VIB, together with their requirement in meiotic DSB formation, prompted us to investigate whether these proteins could associate into a topo VI–like complex. The combination of yeast two-hybrid and bimolecular fluorescence complementation [BiFC, or Split–yellow fluorescent protein (YFP)] (15) assays revealed that MTOPVIB interacts directly with the winged-helix domain (WHD)–containing N-terminal regions of both SPO11-1 and SPO11-2 (Figs. 4B and 5A and fig. S7). The last 149 amino acids of MTOPVIB (MTOPVIB345-493) (Fig. 4A) are sufficient to establish the interaction with the N-terminal regions of SPO11-1 and SPO11-2 (Fig. 4B and fig. S7), which is in agreement with the known topo VI structure (figs. S5B and S6D) (14, 16). The topo VI structure also predicts that the two A subunits should self-associate, but such a direct interaction between SPO11 proteins was not found in either yeast two-hybrid or BiFC assays (Figs. 4C and 5B and fig. S7C) (8). However, coexpression of MTOPVIB in both yeast triple-hybrid and BiFC assays showed that it mediates SPO11 heterodimerization (Figs. 4C and 5B and fig. S7C). MTOPVIB could only mediate the interaction between SPO11-1 and SPO11-2 but did not facilitate the formation of either SPO11-1 or SPO11-2 homodimers (Fig. 4C and fig. S7, C and D). Thus, we propose that the catalytic complex required for meiotic DSB formation adopts a structure comparable with that of the archaeal topo VI complex, involving two MTOPVIB subunits and the two SPO11 subunits. Searches for MTOPVIB orthologs outside flowering plants identified the presence of functionally and structurally conserved homologs in distant phyla (17), suggesting that the requirement for a topo VI–like complex for meiotic DSB formation is a common feature among eukaryotes.

Fig. 4 MTOPVIB mediates the formation of a SPO11-1/SPO11-2 heterodimer in yeast cells.

(A) Schematic representation of the full-length (FL) and truncated proteins used in this study. MTOPVIB and TOP6B GHKL and transducer domains are indicated. SPO11’s N-terminal WHD and its C-terminal Toprim domains are shown. (B) Summary of the yeast two-hybrid assay results. Detailed results are given in fig. S7A. “++” indicates growth on SD-LWHA medium for at least one of the combinations tested. “+” indicates growth on SD-LWH medium for at least one of the combinations tested. “–” indicates that none of the combinations tested confer auxotrophy. Gray boxes indicate the interaction was not tested. (C) Summary of the yeast three-hybrid assay results. Detailed results are given in fig. S7C. Interactions among FL SPO11 proteins were tested by using the GAL4-based system (supplementary materials, materials and methods) in the presence or absence of one of the A. thaliana topo VIB subunits (light gray boxes, MTOPVIB; dark gray boxes, somatic TOP6B; white boxes, no B subunit). “–” indicates no growth on SD-LWHA, and “+” indicates growth on SD-LWHA.

Fig. 5 MTOPVIB mediates the formation of a SPO11-1/SPO11-2 heterodimer in plant cells.

(A and B) Single-section confocal images of N. benthamiana epidermal cells constitutively expressing a fluorescent nuclear protein [H2B–cyan fluorescent protein (CFP), blue channel], co-infiltrated with Agrobacterium cultures expressing two complementary YFP [N-terminal (YFPN) or C-terminal (YFPC)] fusions. A YFP fluorescence signal (right panel in each pair, green) reveals an interaction between the two tested fusion proteins. The full list of combinations tested is in fig. S7D. All images correspond to merged signals of chloroplast autofluorescence (red) with either the CFP signal (blue) or the YFP signal (green). Scale bar, 50 μm. (A) MTOPVIB interacts with SPO11-1 (left) and SPO11-2 (right) in N. benthamiana epidermal cells. (B) SPO11-1 and SPO11-2 do not interact in N. benthamiana epidermal cells except if MTOPVIB is coexpressed with the split-YFP SPO11 constructs.

All eukaryotic genomes encode at least one meiotic topo VIA–like protein—SPO11—but plant genomes are also one of the rare eukaryotic genomes to encode a topo VIB subunit homolog, TOP6B/TOPVIB (3, 18). A. thaliana top6B mutants have strong pleiotropic phenotypes that exclusively affect somatic development but not meiosis (1921). These phenotypes are indistinguishable from spo11-3 mutants, which are disrupted in the third SPO11 homolog found in the Arabidopsis genome (1921). It is therefore likely that SPO11-3 and TOP6B are two subunits of an A. thaliana somatic topo VI complex. In further yeast two-hybrid assays, we found that MTOPVIB interacts with the somatic topo VIA subunit SPO11-3, but that the somatic B subunit TOP6B cannot interact with either meiotic SPO11-1 or SPO11-2 (Fig. 4B) or mediate the formation of the SPO11-1/SPO11-2 meiotic catalytic heterodimer (Fig. 4C). This points toward an important role for the interacting surfaces of the A subunits in conferring specificity to interactions between the two A. thaliana topo VI complexes. In addition, we found that in contrast to the meiotic A. thaliana SPO11 homologs, SPO11-3 self-associates in yeast two-hybrid assays without the need for a topo VIB–like subunit (Fig. 4C), as is the case for the archaeal topo VI complexes, in which the A subunits self-associate (22).

Future studies will further characterize the biochemical activity of the meiotic DSB catalytic complex and examine how it compares in detail with classical topo VI. However, we have clearly shown that the meiotic topo VIB subunit has acquired a crucial role in mediating the assembly of the catalytic portion of the meiotic complex. Compared with the original archaeal topo VI, this represents an additional regulatory capacity, which may be crucial for avoiding erratic DSB formation in gamete mother cells.

Supplementary Materials

Materials and Methods

Figs. S1 to S7

Tables S1 to S5

References (2348)

References and Notes

  1. Acknowledgments: We are grateful to D. Gadelle, P. Forterre, R. Mercier, and R. Kumar for helpful discussions and constructive reading of the manuscript. We thank O. Grandjean and L. Gissot from the Imaging and Cytology platform of IJPB for providing help and technical support for the BiFC experiments. We also thank F. Hartung for providing the TOP6B cDNA and L. Gebbie (LKG Scientific Editing & Translation) for correcting the manuscript. IJPB benefits from the support of the Labex Saclay Plant Sciences–SPS (ANR-10-LABX-0040-SPS).
View Abstract

Navigate This Article