Mouse REC114 is essential for meiotic DNA double-strand break formation and forms a complex with MEI4

Mouse REC114 is essential for meiotic DNA double-strand break formation and forms a complex with IHO1. Its N-terminal region forms a Pleckstrin homology domain, while its C-terminal region is interacting with MEI4.


Introduction
The conversion from diploid to haploid cells during meiosis requires the expression of a specific and highly differentiated meiotic program in all sexually reproducing eukaryotes. Indeed, meiosis is a specialized cell cycle composed of one replication phase followed directly by two divisions. At the first meiotic division, homologous chromosomes (homologues) are separated through a process called reductional segregation. In most species, reductional segregation requires the establishment of connections between homologues. To achieve this, homologous recombination is induced during meiotic prophase to allow homologues to find each other and to be connected by reciprocal products of recombination (i.e., crossovers) (Hunter, 2015). This homologous recombination pathway is initiated by the formation of DNA double-strand breaks (DSBs) (de Massy, 2013) that are preferentially repaired using the homologous chromatid as a template. Meiotic DSB formation and repair are expected to be tightly regulated because improper DSB repair is a potential threat to genome integrity (Sasaki et al, 2010;Keeney et al, 2014). In Saccharomyces cerevisiae, several genes are essential for their formation and at least five of them are evolutionarily conserved. Spo11, Top6bl, Iho1, Mei4, and Rec114 are the mouse orthologs of these five genes (Baudat et al, 2000;Romanienko & Camerini-Otero, 2000;Kumar et al, 2010;Robert et al, 2016a;Stanzione et al, 2016;Tesse et al, 2017) and are specifically expressed in mouse meiotic cells. SPO11 is a homologue of TopoVIA, the catalytic subunit of archea TopoVI, and is covalently bound to the 59 ends of meiotic DNA breaks. This indicates that meiotic DSBs are formed by a mechanism with similarity to a type II DNA topoisomerase cleavage (Bergerat et al, 1997;Keeney et al, 1997;Neale et al, 2005). SPO11 acts with a second subunit, TOPOVIBL, homologous to archea TopoVIB (Robert et al, 2016a;Vrielynck et al, 2016). TOPOVIBL is quite divergent among eukaryotes, and in some species, such as S. cerevisiae, the homologous protein (Rec102) shares only one domain of similarity with TOPOVIBL (Robert et al, 2016a). Based on previous interaction studies between Rec102 and Rec104 (Arora et al, 2004;Jiao et al, 2003), which are both required for meiotic DSB formation, it has been proposed that the S. cerevisiae Rec102/Rec104 complex could fulfill the function of TOPOVIBL (Robert et al, 2016b).
The IHO1, MEI4, and REC114 families, which have been shown to be evolutionary conserved (Kumar et al, 2010;Tesse et al, 2017), have been studied in several organisms, including S. cerevisiae (Mer2, Mei4, and Rec114), Schizosaccharomyces pombe (Rec15, Rec24, and Rec7), Arabidopsis thaliana (PRD3, PRD2, and PHS1), Sordaria macrospora (Asy2, Mei4 ortholog not identified, and Asy3), Caenorhabditis elegans (Mer2 and Mei4 orthologs not identified, and DSB1/2), and Mus musculus. Several important properties of 1 these proteins suggest that they act as a complex. Indeed, in S. cerevisiae (Li et al, 2006;Maleki et al, 2007) and M. musculus (Stanzione et al, 2016), they were shown to co-localize as discrete foci on the axes of meiotic chromosomes which are the structures that develop at the onset of meiotic prophase and allow the anchoring of chromatin loops. Their localization is SPO11-independent, as shown for the three S. cerevisiae, for S. macrospora Asy2 (Tesse et al, 2017), for the IHO1 and MEI4 M. musculus proteins, for S. pombe Rec7 (Lorenz et al, 2006), and for C. elegans DSB1/2 Stamper et al, 2013). They appear before or at the beginning of meiotic prophase and the number of foci decreases as chromosomes synapse in S. cerevisiae (Li et al, 2006;Maleki et al, 2007) and in M. musculus (Kumar et al, 2010;Stanzione et al, 2016). In S. macrospora, where only Mer2 has been analyzed, its axis localization is also decreased at pachytene upon synapsis (Tesse et al, 2017). In C. elegans, foci of the Rec114 orthologs decrease with pachytene progression Stamper et al, 2013). Importantly, in addition to interaction with axis-associated sequences, these proteins are also detected by chromatin immuno-precipitation (ChIP) to interact with DSB sites (Sasanuma et al, 2008;Panizza et al, 2011;Miyoshi et al, 2012;Carballo et al, 2013), which makes sense given their implication in DSB formation. This interaction is weak and transient and is consistent with the tethering of chromatin loops to axes proposed to be established at or before DSB repair (Blat et al, 2002). The determinants of their localization are known to depend on several axis proteins. In S. cerevisiae, Mer2, Mei4, and Rec114 are detected particularly at domains enriched in Hop1 and Red1, two interacting meiotic-specific axis proteins (Panizza et al, 2011). The localization of Red1 depends on the meiotic-specific cohesin Rec8 through a direct interaction (Sun et al, 2015).
Molecular organization and activity of the Mer2, Mei4, and Rec114 complex has remained elusive however. Several studies have reported the interactions between these three proteins, suggesting a tripartite complex in S. pombe (Steiner et al, 2010;Miyoshi et al, 2012), S. cerevisiae (Li et al, 2006;Maleki et al, 2007), and M. musculus (Kumar et al, 2010;Stanzione et al, 2016). The current knowledge on their in vivo direct interactions is limited and based only on yeast two-hybrid assays. Mer2 plays a central role and seems to be the protein that allows the recruitment of Mei4 and Rec114 on chromosome axes. This view is based on the observation that the Mer2 orthologs Rec15 in S. pombe and IHO1 in M. musculus, interact with the axis proteins Rec10 (Lorenz et al, 2006) and HORMAD1 (Stanzione et al, 2016), respectively. In S. cerevisiae, Mer2 is necessary for Rec114 and Mei4 recruitment to the axis (Sasanuma et al, 2008) (Panizza et al, 2011). S. cerevisiae Mer2 is loaded on chromatin before prophase, during S phase, where it is phosphorylated, a step required for its interaction with Rec114 (Henderson et al, 2006;Murakami & Keeney, 2014). Thus, Mer2 coordinates DNA replication and DSB formation. Analysis of the Mer2 ortholog in S. macrospora revealed additional functions in chromosome structure (Tesse et al, 2017). Overall, it is thought that this putative complex (Mer2/Rec114/ Mei4) might directly interact with factors involved in the catalytic activity (i.e., at least Spo11/Rec102/Rec104 in S. cerevisiae) at DSB sites. Interactions between Rec114 and Rec102 and Rec104 have been detected by yeast two-hybrid assays (Arora et al, 2004;Maleki et al, 2007). Moreover, in S. pombe, an additional protein, Mde2 might bridge the Rec15/Rec7/Rec24 and Rec12/Rec6/Rec14 complexes (Miyoshi et al, 2012). However, no specific feature or domain has been identified in Mei4 or Rec114 to understand how they may regulate DSB activity. One could hypothesize that they play a direct role in activating or recruiting the Spo11/TopoVIBL complex for DSB formation. The hypothesis that these proteins might regulate DSB formation through some interactions is also consistent with the findings that Rec114 overexpression inhibits DSB formation in S. cerevisiae (Bishop et al, 1999) and that altering Rec114 phosphorylation pattern can up-or down-regulate DSB levels (Carballo et al, 2013). It is possible that Rec114 and Mei4 have distinct roles because Spo11 non-covalent interaction with DSBs is Rec114-dependent but Mei4-independent (Prieler et al, 2005), and Spo11 self-interaction depends on Rec114 but not on Mei4 (Sasanuma et al, 2007). However, in Zea mays and A. thaliana, the Rec114 homologue (Phs1) seems not to be required for DSB formation (Pawlowski et al, 2004;Ronceret et al, 2009). Here, we performed a functional and molecular analysis to determine whether mouse REC114 is required for meiotic DSB formation, and whether it interacts directly with some of its candidate partners.

Rec114-null mutant mice are deficient in meiotic DSB formation
We analyzed mice carrying a null allele of Rec114. In the mutated allele (here named Rec114 − and registered as Rec114 tm1(KOMP)Wtsi ), exon 3 and 4 were deleted, and a lacZ-neomycin cassette with a splice acceptor site was inserted upstream of this deletion. This deletion includes the conserved motifs SSM3, 4, 5, and 6 (Kumar et al, 2010;Tesse et al, 2017) (Figs 1A and S1). We analyzed the cDNA expressed from testes of Rec114 −/− mice and showed that the mutant allele is transcribed ( Fig S2A). However, because of the presence of a splice acceptor site in the cassette (EnSA, Fig S1), in the cDNA of Rec114 −/− mice, exon 2 is fused to DNA sequences from the cassette, themselves fused to exon 5 and 6 but out of frame ( Fig S1C). This cDNA from Rec114 −/− mice thus encodes for a putative protein containing exon 1 and 2 and lacking all other exons. We conclude that this allele is likely to be a null mutant. We confirmed the absence of detectable REC114 protein in Rec114 −/− mice by Western blot analysis of total testes extracts and after REC114 immunoprecipitation (Figs 1B and S2B). Heterozygous (Rec114 +/− ) and homozygous (Rec114 −/− ) mutant mice were viable. We also generated from this allele, another mutant allele (named and registered as Rec114 del ) without the insertion cassette. We performed all subsequent analyses using mice with the Rec114 − allele unless otherwise stated, and confirmed several phenotypes in mice carrying the Rec114 del allele.
To monitor the consequences of REC114 absence on gametogenesis, we performed histological analysis of testes and ovaries. Spermatogenesis was altered in Rec114 −/− adult male mice, as indicated by the presence of major defects in testis tubule development compared with wild-type (Rec114 +/+ ) mice ( Fig 1C). Specifically, in Rec114 −/− animals the tubule diameter was smaller and tubules lacked haploid cells (spermatids and spermatozoa). In these tubules, the most advanced cells were spermatocytes, although some were also depleted of spermatocytes. Testis weight was significantly lower in Rec114 −/− than wild-type mice ( Fig S2C). In ovaries from Rec114 −/− mice, oogenesis was significantly affected, as indicated by the strongly reduced number of primordial follicles at 2 wk postpartum and their near absence at 8 wk (Fig 1D and E). In ovaries from 2 wk old Rec114 −/− mice, a significant number of secondary follicles are detected and differentiated from the first wave of follicle growth. These follicles have not been subject to elimination as observed in other DSB-deficient mice (Di Giacomo et al, 2005). Consistent with these gametogenesis defects, Rec114 −/− males and females were sterile. Indeed, mating of wild-type C57BL/ 6 animals with Rec114 −/− males and females (n = 3/sex) crossed for 4 months yielded no progeny.
phenotypes, indicating that the cassette present in the Rec114 − allele does not cause the observed meiotic defects (Fig S7).
In vivo REC114 interacts with MEI4 and these proteins display a mutually dependent localization REC114, MEI4, and IHO1 co-localize on the axis of meiotic chromosomes, and IHO1 is needed for MEI4 loading (Stanzione et al, 2016). First, we tested whether IHO1 loading required REC114 and MEI4. This was clearly not the case because IHO1 localization was similar in wild type and in Rec114 −/− and Mei4 −/− spermatocytes (Figs 3A and S8A). This observation is consistent with a role for IHO1 in REC114 and MEI4 recruitment.
We then tested whether MEI4 and REC114 regulated each other's localization. MEI4 forms 200-300 foci on meiotic chromosome axes at leptonema. Then, the focus number progressively decreases as cells progress into zygonema, and MEI4 becomes undetectable at pachynema. This decrease of MEI4 foci during meiotic progression is directly correlated with synapsis formation (MEI4 foci are specifically depleted from synapsed axes) and with DSB repair (MEI4 foci are excluded from DMC1 foci) (Kumar et al, 2010). At leptonema, the number of axis-associated MEI4 foci was reduced 3-to 4-fold in Rec114 −/− spermatocytes and oocytes (Figs 3B and C, S8B, and S9A), and their intensity was significantly decreased (by 1.75-fold in spermatocytes and by 1.9-fold in oocytes) compared with wild-type controls (Fig S9B). The MEI4 signal detected in Rec114 −/− gametocytes was higher than the nonspecific background signal observed in Mei4 −/− spermatocytes (Fig 3C). This suggests that REC114 contributes to, but it is not essential for MEI4 focus formation on meiotic chromosome axis.
REC114 foci co-localize with MEI4 and, like MEI4 foci, their number is highest at leptonema and then progressively decreases upon synapsis (Stanzione et al, 2016). We thus tested whether REC114 foci required MEI4 for axis localization. At leptonema, few axis-associated REC114 foci above the background signal could be detected in Mei4 −/− spermatocytes, where their number was reduced by more than 10-fold compared with wild-type cells (Fig 4A  and B). However, this low level of REC114 foci in Mei4 −/− was still significantly higher compared with the number in Rec114 −/− gametocytes (Fig 4B). REC114 foci were not reduced in Spo11 −/− mice (Fig 4B), as previously reported for MEI4 foci (Kumar et al, 2010). This indicates that these proteins are loaded on the chromosome axis independently of SPO11 activity. Overall, MEI4 and REC114 are reciprocally required for their localization.
MEI4 and REC114 co-localization, their mutual dependency for robust localization, and their interaction in yeast two-hybrid assays (Kumar et al, 2010) strongly suggested that these two proteins interact directly or indirectly in vivo. Indeed, we could detect REC114 after immunoprecipitation of MEI4 in extracts from wild-type and Spo11 −/− mice (Fig 4C). These assays were performed using protein extracts from 14 dpp mice where cellular composition in the testes are similar in wild type and mutants deficient for DSB formation. In this assay, we observed that the amount of MEI4 recovered after immunoprecipitation with the anti-MEI4 antibody is reduced in extracts from Rec114 −/− . This observation may indicate a destabilization of MEI4 in the absence of REC114. In principle, this could also be analyzed by detecting proteins from a total extract, but MEI4 is not detectable in such extracts in our conditions. As REC114 was detected in total protein extracts, we could observe that the REC114 level is not altered in the absence of MEI4 (Figs 1 and S9C). Interestingly, IHO1 also was immunoprecipitated by the anti-MEI4 antibody suggesting that MEI4 interacts with both IHO1 and REC114 (Fig 4C). These three proteins could be a part of the same complex, or form two independent complexes. Although MEI4 and IHO1 did not interact in a yeast two-hybrid assay (Stanzione et al, 2016), the detection of IHO1 after immunoprecipitation of MEI4 in Rec114 −/− extracts ( Fig 4C) suggests a direct or indirect interaction between IHO1 and MEI4 in mouse spermatocytes. We note that after immunoprecipitation of MEI4, the level of IHO1 is not reduced in Rec114 −/− compared with the wild type, whereas the level of MEI4 is. Formally, this could be explained by an excess of MEI4 relative to the IHO1 protein available for interaction with MEI4. Alternatively, changes in protein complex structure in the absence of REC114 could lead to modification of interactions and their recovery in the conditions used for immunoprecipitation.

REC114 and MEI4 form a stable complex
These in vivo assays suggested that REC114 and MEI4 directly interact. To test whether these two proteins interact and form a stable complex in vitro, we produced recombinant full-length REC114 in bacteria (Fig 5). However, we could not produce recombinant fulllength MEI4 or its N-terminal or C-terminal domains alone. Conversely, when co-expressed with REC114, the N-terminal fragment (1-127) of MEI4 was soluble and could be co-purified with REC114 on Strep-Tactin resin (Fig 5A, lane 4), providing the first biochemical evidence of a direct interaction between REC114 and MEI4. To identify the REC114 region that interacts with MEI4, we produced the N-terminal domain and a C-terminal fragment (residues 203-254) of REC114 and found that the REC114 C-terminal region (but not the N-terminal domain) was sufficient for binding to MEI4 (Figs 5A, lanes 5, 6 and S10). Finally, we could purify the MEI4 (1-127) and REC114 (203-254) complex and show that the two proteins co-eluted as a single peak from the Superdex 200 gel filtration column (Fig 5B and C).

Rec114 contains a pleckstrin homology (PH) domain
To gain insights into the structure of mouse REC114, we produced the full-length protein in bacteria. Then, using limited trypsin proteolysis, we identified a stable fragment (residues 15-159) that was suitable for structural analysis. We determined the crystal structure of this REC114 N-terminal region at a resolution of 2.5Å by SAD using the selenomethionine (SeMet)-substituted protein. The final model, refined to an R free of 30% and R-factor of 25% included residues 15-150 (Table 1).
Unexpectedly, the structure revealed that REC114 (15-150) forms a PH domain, with two perpendicular antiparallel β-sheets followed by a C-terminal helix (Fig 6). Several residues are disordered in loops between the β strands. In the SeMet protein dataset that we solved at 2.7Å, the crystallographic asymmetric unit contained two REC114 molecules, but the position of β2 that packs against β1 in one of the molecules was shifted by three residues. A Protein Data Bank search using the PDBeFold server at EBI revealed that REC114 (15-150) was highly similar to the other PH domains and that the N-terminal domain of the CARM1 arginine methyltransferase (PDB code 2OQB) was the closest homologue ( Fig S11).
Mapping the conserved residues to the protein surface revealed that both β-sheets contained exposed conserved residues that could be involved in protein interactions with REC114 partners ( Fig  S12). In the crystal, the PH domain formed extensive crystallographic contacts with a symmetry-related molecule. Indeed, this interface, judged as significant using the PDBePisa server, buried a surface of 764Å 2 and included several salt bridges and hydrogen bonds formed by the well-conserved Arg98 of β6, and Glu130 and Gln137 of α1 (Fig S13A). To test whether REC114 dimerized in solution, we analyzed the PH domain by size-exclusion chromatographymultiple angle laser light scattering (SEC-MALLS) that allows measuring the molecular weight. Although the monomer molecular mass of the fragment was 16 kD (32 kD for a dimer), the MALLS data indicated a molecular weight of 24.7 kD for the sample at the concentration of 10 mg/ml. When injected at lower concentrations, the protein eluted later and the molecular weight diminished (21 kD at 5 mg/ml) (Fig S13B). These results could be explained by a concentration-dependent dimerization of the PH domain with a fast exchange rate between monomers and dimers that co-purify together during SEC. The validation of the monomer interface and the significance of this putative dimerization will be interesting to evaluate.

Discussion
Previous studies in yeast have shown that the putative complex involving S. cerevisiae Rec114, Mei4, and Mer2 is essential for meiotic DSB formation. Their transient localization at the DSB sites (observed by ChIP) suggests that, in yeast, this complex may play a direct role in promoting DSB activity (Sasanuma et al, 2007(Sasanuma et al, , 2008Panizza et al, 2011;Carballo et al, 2013). Several studies have shown the evolutionary conservation of these three partners. In mammals, MEI4 and IHO1 (the Mei4 and Mer2 orthologs, respectively) are required for meiotic DSB formation (Kumar et al, 2010;Stanzione et al, 2016). Here, we show that REC114 function in the formation of meiotic DSBs is conserved in the mouse. Moreover, we provide the first direct evidence of the interaction between REC114 and MEI4 and identified a potential interaction domain in REC114 that includes previously identified conserved motifs.

Properties of REC114
Our study revealed that REC114 N-terminus is a PH domain that is composed of two sets of perpendicular anti-parallel β-sheets followed by an α helix. This domain is present in a large family of proteins with diverse biological functions and is mostly involved in targeting proteins to a specific site and/or in protein interactions. A subset of these proteins interacts with phosphoinositide phosphates (Lietzke et al, 2000;Lemmon, 2003). Several conserved positively charged residues in the two β sheets, β1 and β2, important for the interaction are not present in REC114. However, subsequent studies revealed interactions between the PH domain and a variety of different partners, in some cases by binding to phosphotyrosine-containing proteins or to polyproline (Scheffzek & Welti, 2012). Therefore, the REC114 PH domain could be a platform for several interactions, some of which could involve phosphorylated serine or threonine residues because it has been shown that ATR/ATM signaling through phosphorylation of downstream proteins regulate meiotic DSB activity (Joyce et al, 2011;Lange et al, 2011;Zhang et al, 2011;Carballo et al, 2013;Cooper et al, 2014).
In terms of conservation of the REC114 primary sequence, it is remarkable that most of the previously described conserved motifs (SSM1 to 6) are structural elements (β sheets and α helices) within this PH domain and are readily identified in many eukaryotes (Kumar et al, 2010;Tesse et al, 2017). At the REC114 C-terminus, the SSM7 motif overlaps with a predicted α helical structure and is less well conserved. Moreover, its presence remains to be established in several species (Tesse et al, 2017). In this study, we demonstrated that this C-terminal domain directly interacts with MEI4, suggesting that this SSM7 region is evolutionarily conserved. The N-terminal domain of MEI4 that interacts with REC114 has a predicted α helical structure and includes two conserved motifs (Kumar et al, 2010).

Interaction of REC114 with the chromosome axis
A previous study showed that IHO1 is required for MEI4 and REC114 focus formation on axis and that it directly interacts with REC114 by two-hybrid assay (Stanzione et al, 2016). As IHO1 interacts with HORMAD1, IHO1 could act as a platform to recruit REC114 and MEI4. Such a mechanism would be similar the one identified in S. cerevisiae for the recruitment of Rec114 and Mei4 by Mer2 (Henderson et al, 2006;Panizza et al, 2011). In agreement with this hypothesis, IHO1 association with the chromosome axis is not altered in the absence of MEI4 or REC114, similar to that observed in S. cerevisiae (Panizza et al, 2011). Therefore, IHO1 could recruit REC114 by direct interaction, and this should allow MEI4 recruitment. Alternatively, we suggest a mechanism where REC114/MEI4 would be recruited as a complex to the axis as we observed a mutual dependency between these two proteins for their axis localization: the formation of REC114 axis-associated foci is strongly reduced in the absence of MEI4 and reciprocally. The residual REC114 foci detected in the absence of MEI4 do not appear to be able to promote DSB formation as DSB repair foci are abolished in Mei4 −/− mice similar to Spo11 −/− mice, and thus, suggesting an active role for the REC114/MEI4 complex. MEI4 may also be able to interact (directly or indirectly) with IHO1 or with axis proteins independently from REC114 at least in an Rec114 −/− genetic background as weak MEI4 axis-associated foci were observed in Rec114 −/− spermatocytes and oocytes and because IHO1 protein was detected upon immunoprecipitation of MEI4 in Rec114 −/− spermatocyte extracts. The details of these interactions and their dynamics during early meiotic prophase remain to be analyzed. Overall, the IHO1/MEI4/REC114 complex is expected to be the main component for the control of SPO11/TOPOVIBL catalytic activity. It may be essential for turning on and off the catalytic activity. In S. cerevisiae, it has been proposed that the local control of meiotic DSB formation is constrained by the chromatin loop organization and involves Tel1 (ATM) (Garcia et al, 2015) and possibly also the Mer2/Mei4/Rec114 (IHO1/MEI4/REC114) complex. Indeed, S. cerevisiae Rec114 shows that Tel1/Mec1-dependent phosphorylation is associated with down-regulation of DSB activity (Carballo et al, 2013). The IHO1/MEI4/REC114 complex could be a limiting factor for DSB formation. In agreement, we noted that the number of cytologically detectable foci is of the same order (about 200) as the number of DSB events measured by the detection of DSB repair proteins. The shutting off of DSB formation that correlates with synapsis between homologues (Thacker et al, 2014) could be the direct consequence of the removal of the Hop1 (or HORMAD1 in mice) axis protein, resulting in the displacement of the Mer2/Mei4/ Rec114 (IHO1/MEI4/REC114 in mice) complex from the axis. Additional studies on the protein-protein interactions and posttranslational modifications will help to understand these important steps for the regulation of meiotic DSB formation.

Mouse strains
The non-conditional Rec114 mutant allele is referenced as 2410076I21Rik tm1(KOMP)Wtsi and named Rec114 −/− in this study. The Rec114 del allele, in which the inserted lacZ-Neo cassette was deleted, was obtained by expression of Flp in mice carrying Rec114 −/− . These mice are in the C57BL/6 background. The Mei4 −/− , Spo11 −/− , and Spo11 YF/YF strains were previously described (Baudat et al, 2000;Kumar et al, 2010;Carofiglio et al, 2013). All animal Residues that are 100% conserved are in solid green boxes. The secondary structures of REC114 are shown above the sequences. The previously identified conserved motives, SSM1 to SSM6 (Kumar et al, 2010;Tesse et al, 2017) experiments were carried out according to the CNRS guidelines and approved by the ethics committee on live animals (project CE-LR-0812 and 1295).

cDNA analysis
Total RNA was extracted from two testes of young (14 dpp) wild type or Rec114 −/− mice using the GeneJet RNA Purification Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. One μg of total RNA was treated with RQ1 RNAse-free DNase (Promega) 30 min at 37°C to degrade genomic DNA. 1 μg of total RNA was used for cDNA synthesis, using random primers and the Transcriptor First Strand cDNA Synthesis Kit (Roche) according to the manufacturer's instructions. Each PCR was performed using 0.05 μl of cDNA and KAPA Taq PCR kit (Sigma-Aldrich). Oligonucleotides sequences are provided in the Table S1. PCR cycling conditions were 3 min at 95°C, 35 cycles for 16U22/232L22 and 695U22/806L25 or 40 cycles for 16U21/644L22 and 16U21/688L24 with 15 s at 95°C, annealing for 15 s at 53°C, 51°C, 64°C, and 59°C for 16U22/232L22, 695U22/806L25, 16U21/644L22, and 16U21/688L24, respectively, and extension for 30 s at 72°C, followed by 5 min at 72°C at the end of cycles. Amplified products were separated on 2% agarose gels. PCR products obtained from the oligonucleotide combinations 16U21/644L22 or 16U21/688L24 were run in a gel and the most abundant product purified and sequenced.

Histology and cytology
Testes and ovaries were fixed in Bouin's solution (Sigma-Aldrich) at room temperature overnight and for 5 h, respectively. After dehydration and embedding in paraffin, 3-μm sections were prepared and stained with periodic acid-Schiff for testes and with hematoxylin and eosin for ovaries. Image processing and analysis were carried out with the NDP.view2 software (Hamamatsu).
Spermatocyte and oocyte chromosome spreads were prepared by the dry-down method (Peters et al, 1997).

Image analysis
γH2AX was quantified using Cell profiler 2.2.0. The total pixel intensity per nucleus was quantified. The intensity of MEI4 foci was the mean pixel value within a focus. Axis-associated MEI4 foci were determined by co-labelling with SYCP3.

Protein analysis
Whole testis protein extracts were prepared as described in Stanzione et al (2016).
For MEI4 immunoprecipitation, 3 μg of guinea pig anti-MEI4 antibody (Stanzione et al, 2016) was cross-linked to 1.5 mg of Dynabeads Protein A (Invitrogen) with disuccinimidyl suberate using the Crosslink Magnetic IP/Co-IP Kit (Pierce; Thermo Fisher Scientific). 3.6 mg of testis protein extract (from 14 dpp mice) was incubated with the cross-linked antibody at 4°C overnight. Beads were washed five times with washing buffer (20 mM Tris-HCl, pH 7.5, 0.05% NP-40, 0.1% Tween-20, 10% glycerol, 150 mM NaCl). Immunoprecipitated material was eluted by incubating the beads with the elution buffer (pH 2) for 5 min, and neutralized with the neutralization buffer (pH 8.5) (both buffers provided with the kit). The eluates were incubated with Laemmli loading buffer (1× final) at RT for 10 min, and divided in three aliquots, adding 10 mM DTT to one of them (for REC114 detection), followed by incubation at 95°C for 5 min. SPO11 immunoprecipitations on protein extracts from adult mice were performed as described (Pan & Keeney, 2009). Rabbit polyclonal anti-SPO11 was used for detection (Carofiglio et al, 2013).

Protein expression, purification, and crystallization
Mouse REC114 (15-159) fused to His-tag was expressed in E. coli BL21-Gold (DE3) (Agilent) from the pProEXHTb expression vector (Invitrogen). The protein was first purified by affinity chromatography using Ni 2+ resin. After His-tag cleavage with the TEV protease, the protein was further purified through a second Ni 2+ column and by size-exclusion chromatography. Pure protein was concentrated to 10 mg⋅ml −1 in buffer (20 mM Tris, pH 7.0, 200 mM NaCl, and 5 mM mercaptoethanol). The best-diffracting crystals grew within 1 wk at 20°C in a solution containing 0.25 M ammonium sulfate, 0.1 M MES, pH 6.5, and 28% PEG 5000 MME. For data collection at 100 K, crystals were snap-frozen in liquid nitrogen with a solution containing mother liquor and 25% (vol/vol) glycerol. SeMet-substituted REC114 was produced in E. coli BL21-Gold (DE3) and a defined medium containing 50 mg⋅l −1 of SeMet. SeMet REC114 was purified and crystallized as for the native protein.

Data collection and structure determination
Crystals of REC114 (15-159) belong to the space group P6 1 22 with the unit cell dimensions a, b = 107.5Å and c = 82.8Å. The asymmetric unit contains one molecule and has a solvent content of 71%. A complete native dataset was collected to a resolution of 2.5Å by the autonomous ESRF beamline MASSIF-1 (Bowler et al, 2015). The SeMet REC114 crystallized in the same conditions in the space group P4 2 2 1 2 and contained two molecules per asymmetric unit. A complete SeMet dataset was collected to a resolution of 2.7Å at the peak wavelength of the Se K-edge on the ID23-1 beamline at the ESRF. Data were processed using XDS (Kabsch, 2010). The structure was solved using SeMet SAD data. Selenium sites were identified, refined, and used for phasing in AUTOSHARP (Bricogne et al, 2003). The model was partially built with BUCCANEER (Cowtan, 2006), completed manually in COOT (Emsley et al, 2010) and refined with REFMAC5 (Murshudov et al, 1997). The model was used for molecular replacement to determine the structure using the native dataset and PHASER (McCoy et al, 2007). The native structure was finalized in COOT and refined with REFMAC5 to a final R-factor of 25% and R free of 30% (Table 1) with all residues in the allowed regions (96% in favored regions) of the Ramachandran plot, as analyzed using MOLPROBITY (Chen et al, 2010).

Strep-tag pull-down assays
MEI4 (1-127) was cloned in the pProEXHTb expression vector to produce a His-tag fusion protein. REC114 and its deletion mutants were cloned in the pRSFDuet-1 vector as Strep-tag fusion proteins. REC114 variants alone or co-expressed with MEI4 were purified using a Strep-Tactin XT resin (IBA). The resin was extensively washed with a buffer containing 20 mM Tris, pH 7.0, 200 mM NaCl, and 5 mM mercaptoethanol, and bound proteins were eluted with the same buffer containing 50 mM biotin and analyzed by 15% SDS-PAGE. The minimal REC114-MEI4 complex was then purified using the Strep-Tactin XT resin. The His-tag of MEI4 was removed with the TEV protease and a passage through a Ni 2+ column. The complex was then purified by size-exclusion chromatography.