A Rad50-null mutation in mouse germ cells causes reduced DSB formation, abnormal DSB end resection and complete loss of germ cells

DNA double-strand break (DSB) represents one of the most deleterious types of DNA damage in cells, posing a significant threaten to genetic stability and promoting tumorigenesis (Scully et al., 2019; Machour and Ayoub, 2020). DSBs arise from exposure to genotoxic chemicals and ionizing radiation, but they can be integral to various biological processes, such as meiotic recombination, and can also arise from endogenous breaks during DNA replication in mitosis (Blackford and Stucki, 2020; Keeney et al., 1997; Sun et al., 1989; Restier-Verlet et al., 2023; Sakasai et al., 2023). The MRE11-RAD50-NBS1/Xrs2 (MRN/X) protein complex with MRE11 and RAD50 as the catalytic components, governs DNA damage response (DDR) signaling, and is responsible for the recognition, DNA end resection and DNA repair of DSBs (Syed and Tainer, 2018; Lavin, 2007). In both yeast and mammals, MRE11 exhibits endonucleolytic cleavage of the 5′ terminated DNA strand in the vicinity of the DSBs and exonucleolytic resection from 3′ to 5′ towards the DSB ends (Trujillo et al., 1998; Usui et al., 1998). RAD50 is a structure maintenance of chromosome (SMC)-related protein that contains one ATPase domain at its N- and C-terminal ends, respectively, Zn hook and anti-parallel coiled coils (de Jager et al., 2001). RAD50 plays a crucial role in facilitating the MRE11 nuclease activity on DSBs by ATP binding and hydrolysis (Lammens et al., 2011; Williams et al., 2011; Majka et al., 2012). NBS1, which contains a forkhead-associated (FHA) and two BRCT (BRCA1 carboxy-terminal) domains at its N-terminal, facilitates the nuclear localization of the MR complex, and potentiates the endonuclease activity of MRE11in mammals (Desai-Mehta et al., 2001; Syed and Tainer, 2018).

Spermatogenesis is a highly complex and specialized developmental process, consisting of three successive phases: mitosis, meiosis and spermiogenesis (Oatley and Brinster, 2008). At the beginning of mouse spermatogenesis, spermatogonial stem cells (SSCs) either undergo self-renewal to maintain the stem cell pool or differentiate into undifferentiated progenitor spermatogonia. These progenitor spermatogonia then progress into A1 spermatogonia in response to retinoic acid, and subsequently develop into type B spermatogonia through five successive mitotic divisions (de Rooij, 2001). The type B spermatogonia further differentiate into preleptotene spermatocytes, which undergo one round of meiotic DNA replication to enter meiotic prophase I (Zickler and Kleckner, 2015). During meiotic prophase I, chromosomes undergo homologous pairing, synapsis and recombination (Hernández-Hernández et al., 2009). After completing meiotic prophase I, spermatocytes undergo two consecutive rounds of chromosome segregation to generate haploid round spermatids (Wassmann, 2013). The resulting round spermatids undergo dramatic morphological and molecular changes, and ultimately mature into spermatozoa, a process known as spermiogenesis (Tanaka and Baba, 2005).

Programed DSBs catalyzed by the topoisomerase II-like enzymes SPO11 and TOPVIBL initiate meiotic recombination (Keeney et al., 1997; Robert et al., 2016). After DSB formation, the MRN/MRX complex, along with EXO1 and DNA2, cleave the SPO11-DNA to generate 3′ single-stranded DNA (ssDNA) ends, which are prerequisite for meiotic DSB repair (Gray and Cohen, 2016; Tisi et al., 2020; Zhao et al., 2020). In addition to its role in meiosis, the MRN/MRX complex has also been shown to function in DNA repair during DNA replication in mitotic cells (Petsalaki and Zachos, 2020; Clerici et al., 2006). In hypomorphic mutants of each member of the Saccharomyces cerevisiae MRX complex, Spo11 remains covalently bound to the DSB termini during meiotic prophase I (Mimitou and Symington, 2009; Alani et al., 1990; Nairz and Klein, 1997). In mammals, however, the situation becomes more complex. For example, mice with hypomorphic mutations in Mre11 (Mre11^ATLD1/ATLD1) and Nbs1 (Nbs1^δB/δB) exhibited mild meiotic phenotypes (Cherry et al., 2007). Similarly, despite growth defects, cancer predisposition and progressive hematopoietic stem cell failure, hypomorphic Rad50 mutant (Rad50^s/s) males and females are fertile (Bender et al., 2002). Likewise, Rad50^+/46 mice, which have reduced Zn²⁺ affinity and dimerization efficiency, exhibited hydrocephalus, liver tumorigenesis and defects in primitive hematopoietic cells and male germ cells before meiosis (Roset et al., 2014). Additionally, null mutations in any member of the MRN complex result in early embryonic lethality in mice (Luo et al., 1999; Xiao and Weaver, 1997; Zhu et al., 2001; Roset et al., 2014). Consequently, the precise role of Rad50 in mammalian meiotic recombination and mitosis has yet to be fully understood. More specifically, its in vivo function remains to be explored, although the role of the MRN on DSB metabolism has been studied extensively in vitro (Anand et al., 2016, 2019).

In this study, we employed conditional inactivation of Rad50 in germ cells to define its in vivo role in male germ cell development. Using END-seq, we report here that null mutation of Rad50 in mouse spermatocytes causes defects in both DSB formation and DSB end resection, leading to disruption of meiotic recombination and synapsis. We further identify that inactivation of Rad50 in gonocytes results in the loss of spermatogonial stem cells (SCCs) due to defective DSB repair. These findings indicate that RAD50 participates in mammalian germ cell development.

Owing to the embryonic lethality caused by disruption of Rad50 (Luo et al., 1999), we conditionally deleted Rad50 in advanced germ cells using a Stra8-GFPCre line (Rad50^f/Δ, Stra8-GFPCre; hereafter referred to as Rad50-sKO) to study the function of Rad50 in germ cells. We found that both Rad50-sKO males and females were infertile (Fig. S1A,B). The size and relative weight of testes from 8-week-old Rad50-sKO mice were dramatically reduced (Fig. 1A,B). Hematoxylin and Eosin staining showed the absence of post-meiotic spermatids in Rad50-sKO seminiferous tubules, with numerous vacuoles or atrophic germ cells present in these tubules, which were rarely observed in control testes at the corresponding age (Fig. 1C). Of note, spermatocytes in Rad50-sKO testes underwent apoptosis at the seminiferous epithelial stage IV, which is equivalent to mid-pachytene spermatocytes (Fig. 1C) and was identified by the presence of intermediate spermatogonia (Ahmed and de Rooij, 2009), suggesting meiotic prophase I defects. Apoptotic spermatocytes were further confirmed by terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) staining (Fig. 1D,E).

Furthermore, we found a lack of oocytes in secondary follicles within the ovaries of adult Rad50-sKO mice, rendering them infertile, whereas the control mouse ovaries contained abundant oocytes and follicles (Fig. 1F,G). We then examined newborn (postnatal day 0.5; 0.5 dpp) and 2-week-old females. Our analyses revealed a substantial presence of oocytes in both age groups of Rad50-sKO females (Fig. S1C,D), suggesting that Rad50 may play a role in the sex-specific differences observed during meiosis. It will be of interest to test this in the future. Collectively, these results demonstrate that infertility in Rad50-sKO mice results from meiotic defects.

As spermatocyte apoptosis can be triggered by defects in synapsis and DSB repair, we tracked homologous synapsis on spermatocyte nuclear surface spreading with immunostaining for SYCP3 [the axial/lateral element of synaptonemal complex (SC)] and SYCP1 (the central element of the SC) (Gray and Cohen, 2016). The development of chromosome axes and the SC can help define the substages of meiotic prophase I (Gray and Cohen, 2016). In the control spermatocytes, chromosome axes start to form at the leptotene and are fully formed by late zygotene, and SC formation on autosomes is completed by pachytene (Fig. 2A). The control mice contained all substages of spermatocytes during meiotic prophase I (Fig. 2A,B). The development of the chromosome axis in leptotene spermatocytes was comparable between the control and mutant mice (Fig. 2A). Although co-immunostaining for H1T and phosphorylated H2AX at Ser 139 (γH2AX) showed the presence of mid-pachytene spermatocytes and onwards in mutants (Fig. S2A,B), we did not observe any normal pachytene or diplotene spermatocytes in the Rad50 knockout testes (Fig. 2A). By counting the number of substages of spermatocyte substages in meiotic prophase I from the control and Rad50-sKO mice, we found a significant increase in the percentage of zygotene spermatocytes in the mutant mice compared with that in the control mice (67.29±3.47% versus 17.44±1.84%; mean±s.d.; P<0.0001; two-tailed t-test; Fig. 2B). In Rad50-sKO mice, we observed abnormal chromosome structures and aberrant synaptic features at a stage referred to as ‘zygotene-like’, which corresponds to zygotene spermatocytes (based on the presence of fully formed axes; Fig. 2A,C). In Rad50 deficient spermatocytes, the formation of the SC on autosomes was never completed; many autosomes did not even partially synapse; and SCs were formed extensively between non-homologs (Fig. 2C). The abnormal nuclei displayed one of the following: (1) short synapsed SCs, (2) chromosome ‘tangles’ consisting of synapsed non-homologs and asynaptic chromosomes, (3) a mixture of synapsed non-homologs, asynaptic homologs and synapsed homologs (Fig. 2C). To more comprehensively understand synapsis defects in mutants, we conducted spermatocyte nuclear surface spreading with immunostaining for HORMAD1 and SYCP3. In control spermatocytes, HORMAD1 localized to chromosome axes during the leptotene stage, to unsynapsed axes at the zygotene stage and to the unsynapsed sex chromosome region at the pachytene stage (Fig. 2D), which is consistent with previous report (Wojtasz et al., 2009). In contrast, HORMAD1 accumulation was dramatically more pronounced on the unsynapsed axes in Rad50 mutant spermatocytes at leptotene, zygotene-like and pachytene-like stages. Interestingly, even though synapsis occurred, there was still some accumulation of HORMAD1 on the synapsed axes in zygotene-like and pachytene-like spermatocytes (Fig. 2D). Collectively, these results indicate that Rad50 loss results in defective in homologous synapsis, thereby triggering apoptosis in spermatocytes during meiotic prophase I.

Temporal alterations in synaptic aberrations in Rad50-deficient spermatocytes suggest significant DSB repair defects. We therefore performed spermatocyte nuclear spreading with immunostaining for γH2AX, a marker of ATM/ATR-dependent DDR along with SYCP3. We found that both leptotene and zygotene spermatocytes from Rad50-sKO mice exhibited higher levels of γH2AX signals across entire chromosomes (Fig. 3A). By quantification of γH2AX signal intensity, we have further confirmed the mutant phenotype (Fig. 3B). Moreover, the γH2AX signals from Rad50-sKO pachytene-like spermatocytes remained diffused on asynaptic autosomes, resembling the γH2AX staining observed at the sex body region in control pachytene spermatocytes (Fig. 3A). Additionally, abnormal chromosome structures of zygotene-like spermatocytes from Rad50-sKO mice displayed strong γH2AX signals across entire chromosomes (Fig. 3C). These results indicated that Rad50 deficiency leads to impaired repair of DSBs during mouse meiosis.

During meiotic recombination, the MRN-dependent DNA end resection, which creates 3′ single-stranded DNA ends (3′ ssDNAs), is the initial step in DSB repair (Tisi et al., 2020). The formed 3′ ssDNA is loaded with replication protein A (RPA) complex, and then replaced by DMC1 and RAD51, facilitating homology search and strand invasion (Ribeiro et al., 2016; Hinch et al., 2020). Thus, the foci of RPA, DMC1 and RAD51 can provide an estimate of the number of DSBs. Considering that RAD50 is crucial for MRE11 nuclease activity and Rad50-sKO spermatocytes show homologous synaptic defects, we proposed that RAD50 plays a role in DSB formation and repair in mice, similar to its role in yeast. To address this, we performed immunostaining for RPA, DMC1, RAD51 and SYCP3 on nuclear spreads. In control meiocytes, the foci of RPA, DMC1 and RAD51 appeared in leptotene, reached their peaks in early/mid-zygotene and then decreased during late zygotene as DSB repair progressed (Fig. 4A-F). In Rad50 deficient leptotene spermatocytes, we found a lower number of RPA [67 (17, 95) versus 7 (3, 11); M(P25, P75); P=0.000; Mann–Whitney U-test], DMC1 [41 (20, 61) versus 7 (3, 12); M(P25, P75); P=0.000; Mann–Whitney U-test] and RAD51 [45 (6, 93) versus 3 (1, 8); M(P25, P75); P=0.000; Mann–Whitney U-test] foci compared with the controls (Fig. 4B,D,F). Furthermore, the number of RPA foci [250 (232, 273) versus 90 (60, 17); M (P25, P75); P=0.000; Mann–Whitney U-test] in Rad50 mutant early/mid-zygotene were significantly decreased compared with the controls (Fig. 4B). Similarly, the number of DMC1 foci [195 (172, 210) versus 60 (21, 116); M (P25, P75); P=0.000; Mann–Whitney U-test] and RAD51 [201 (17, 21) versus 75 (20, 9); M (P25, P75); P=0.000; Mann–Whitney U-test] were markedly lower in mutant early/mid-zygotene (Fig. 4D,F). During the late-zygotene stage, Rad50 mutant spermatocytes exhibited a significant decrease in the number of RPA [185 (154, 206) versus 65 (43, 87); M (P25, P75); P=0.000; Mann–Whitney U-test], DMC1[125 (92, 149) versus 55 (33, 76); M (P25, P75); P=0.000; Mann–Whitney U-test] and RAD51 [142 (117, 158) versus 49 (33, 71); M (P25, P75); P=0.000; Mann–Whitney U-test] foci compared with the controls (Fig. 4B,D,F). Collectively, these results imply that Rad50 deficiency could affect both DSB repair and potentially DSB formation, thereby contributing to defective synapsis. Thus, Rad50 participates in meiotic recombination in mammals, similar to its role in yeast.

To explore molecular mechanism underlying dysregulated meiotic recombination in Rad50-deficient spermatocytes, we used END-seq, a quantitative technique that sequences DSB at nucleotide resolution and can directly detect the terminal end of physiological resection after in vitro blunting of the 3′ overhang (Paiano et al., 2020). We performed END-seq on homogeneous zygotene populations from control and Rad50-sKO mouse synchronous spermatogenic cells, and found that most END-seq peaks overlapped with previously identified hotspots defined by released SPO11-oligos (GEO: GSE84689) (Lange et al., 2016) or DMC1-ssDNAs (GEO: GSE35498) (Brick et al., 2012) (Fig. S3A,B). We therefore identified 8168 DSB sites and 6636 DSB sites in control spermatocytes and Rad50 defective spermatocytes, respectively (Fig. 5A). Notably, 96.4% (6396 out of 6636) of DSBs identified in the mutant spermatocytes overlapped with DSBs found in the control spermatocytes (Fig. 5A,B). However, 1772 DSBs in the controls were absent in the mutants (Fig. 5A). Consistent with this, Rad50 deficient leptotene spermatocytes showed significantly decreased DMC1 and RAD51 foci (Fig. 4D,F). These results suggest a reduced DSB formation in the mutant spermatocytes. Moreover, we observed that control spermatocytes show clear read-less gaps, in resections with a mean minimum distance of 644 nt (normalized by SPO11-oligos), in agreement with previously reported results (Fig. 5C,D). In contrast, there are no distinct boundaries between the short- and long-range resections in Rad50-deficient spermatocytes, indicating that some DSBs remain unresected (Fig. 5C,E). Interestingly, we found a dramatic accumulation of the central signals, which represent SPO11-bound DSBs, in mutant spermatocytes (Fig. 5C). These results reveal increased unresected DSBs in Rad50-deficient spermatocytes. Furthermore, we found a significant decrease in long-range resection distance in Rad50-deficient spermatocytes compared with that in controls [mean maximum resection length: 1923 nt (control) versus 1279 nt (mutant)] (Fig. 5C,D). Collectively, these results reveal that RAD50 promotes both short- and long-range resection, and is likely required for the formation of meiotic DSBs in mammals.

We next asked whether RAD50 is required for mitosis of male germ cells. To test this, we generated germ cell-specific Rad50 knockout mice by Vasa-Cre line (Rad50^f/Δ, Vasa-Cre; hereafter referred to as Rad50-vKO), which initiates recombinase expression in gonocytes at embryonic day 15.5 (E15.5). We found that Rad50-vKO males were sterile but exhibited normal copulating behavior. The relative weights of testes from Rad50-vKO mice were dramatically lower than those from controls at age 3 weeks, 8 weeks and 4 months (Fig. 6A,B). Histological analyses of 4-month-old Rad50-vKO testes showed degenerated seminiferous tubules with the absence of germ cells, resulting in a Sertoli cell-only syndrome (Fig. 6C). In contrast, control seminiferous tubules contained a cohort of spermatogonia, spermatocytes, spermatids and spermatozoa (Fig. 6C). Immunofluorescence staining for MVH, a germ cell marker, confirmed the complete absence of MVH-positive germ cells in Rad50-vKO seminiferous tubules from 4-month-old mutants, corroborating the loss of germ cells in mutants (Fig. 6D).

To evaluate the extent of germ cell loss in the different age of mutants, we examined spermatogenesis in 2-, 3- and 8-week-old testes. Two-week-old Rad50-vKO testes exhibit mildly differences in the abundance of spermatogenic cells compared with the controls. However, histological analyses of Rad50-vKO testes showed the absence of spermatocytes and spermatids at 3 and 8 weeks of age, contrasting with the expected progression observed in control testes (Fig. 6E). Immunostaining for STRA8 and γH2AX was performed to assess spermatogonial differentiation in Rad50-vKO mice. We found that the number of A1 spermatogonia, which are likely STRA8⁺γH2AX⁻ cells (Chen et al., 2018; Zhou et al., 2008; Hamer et al., 2003; Kirsanov et al., 2023; Anderson et al., 2008), in 2- and 3-week-old mutants was similar to that in the controls (Fig. S4A,B). However, the mutant testes displayed no STRA8-positive cells at 8 weeks, whereas control testes exhibited numerous seminiferous tubules with STRA8-positive cells (Fig. S4A,B). These results indicate that Rad50 deficient spermatogonia can initially differentiate into A1 spermatogonia, but an age-dependent loss of A1 spermatogonia occurs. This progressive depletion of A1 spermatogonia in Rad50-vKO mice may be attributed to the loss of undifferentiated spermatogonia. To investigate this hypothesis, we examined the expression of LIN28, an undifferentiated spermatogonial marker, and SOX9, a Sertoli cell marker, in both control and Rad50-vKO testes. We observed a significant and progressive decrease in the ratio of LIN28-positive cells to SOX9-positive cells in Rad50-vKO testes, starting at 2 weeks of age, whereas no such difference was observed at 1 week of age (Fig. 6G). The phenotype became more severe by 3 weeks of age (Fig. 6F,G). Collectively, these findings demonstrate that a Rad50 deficiency leads to a progressive failure of the undifferentiated spermatogonia growth, resulting in complete loss of male germ cells and impaired spermatogenesis.

Previous studies have indicated that DSBs generated from mitotic S phase of the cell cycle was preferentially repaired by homologous recombination (HR), suggesting that the recombinational DNA repair function of the MR might be required for DNA replication (Swift and Azizkhan-Clifford, 2022; Karanam et al., 2012). To investigate the mechanism responsible for the complete loss of Rad50-deficient spermatogonia, we checked the presence of γH2AX, a widely recognized indicator of DNA damage. Immunofluorescence staining for γH2AX and PLZF, a spermatogonia marker, was performed on testicular sections from both 3-week-old control and Rad50-vKO mice. Notably, we found the ratio of γH2AX positive undifferentiated spermatogonia (PLZF-positive cells) to PLZF-positive cells per tubule was significantly higher in Rad50-vKO mice compared with the controls and γH2AX signals in the undifferentiated spermatogonia of Rad50-vKO mice was stronger (Fig. 6H-J). This suggests a failure in DNA repair within the mutant spermatogonia. To further investigate whether DNA repair failure in spermatogonia leads to cell cycle arrest during DNA replication, we administered a brief 5-ethynyl-2′-deoxyuridine (EdU) pulse to both 3-week-old control and Rad50-vKO mice. Remarkably, we found that the ratio of both PLZF- and EdU-positive cells to PLZF-positive cells per tubule was significantly higher in Rad50-vKO mice compared with the controls (Fig. 6K,L). Moreover, we observed an increased number of undifferentiated spermatogonia in the S phase showing γH2AX signals (identified as EdU⁺PLZF⁺γH2AX⁺ triple-positive cells) in Rad50 mutant mice compared with the controls (Fig. S4C,D). Strikingly, the TUNEL assay conducted on testicular sections from 3-week-old control and Rad50-vKO demonstrated that PLZF-positive spermatogonia underwent apoptosis in Rad50-deficient mice (Fig. 6M,N). Taken together, these results indicate that Rad50-deficient undifferentiated spermatogonia perish due to unresolved DNA damage arising during DNA replication, ultimately resulting in the complete loss of spermatogonia.

Herein, we employed END-seq, a technique for the direct detection of meiotic DSBs and their resection (Paiano et al., 2020), and conditionally deleted the Rad50 allele in germ cells to investigate the precise in vivo role of Rad50 in spermatogenesis. In sharp contrast to partial disruption of RAD50 function, which leads to defects in mammalian male germ cells before meiosis (Roset et al., 2014), we demonstrate that a complete null mutation in RAD50 results in meiotic defects caused by defective DSB formation and impaired DSB end resection. Furthermore, we show that Rad50 is involved in repairing spontaneous DNA damage, and thus that progressive loss of undifferentiated spermatogonia in Rad50 mutants results from failure of DNA repair during DNA replication.

Meiosis is a specialized cell division to ultimately produce haploid gametes from diploid germ cells in all sexually reproducing organisms (Handel and Schimenti, 2010). Meiotic recombination, the most prominent event in meiosis, creates genetic diversity and ensures accurate homologous chromosome (homolog) segregation at the first meiotic division (Baudat et al., 2013). This unique process involves two temporally coupled events: the formation of DSBs and their subsequent processing (Keeney et al., 1997; Gray and Cohen, 2016; Tisi et al., 2020). DSB formation is triggered by Spo11, an enzyme that is conserved across species (Keeney et al., 1997). Genetic studies in yeast have unveiled that Spo11 activity relies on the involvement of at least nine additional factors, including the MRX complex (Roeder, 1997; de Massy, 2013; Alani et al., 1990; Johzuka and Ogawa, 1995). For example, rad50 regulates DSB formation in S. cerevisiae (Alani et al., 1990). In our own study, we have demonstrated that the loss of Rad50 in mouse spermatocytes results in impaired formation of DSBs characterized by a considerable reduction in the number of DSBs (Fig. 5), although the phenotype is not as severe as that seen in yeast null rad50 mutants (Hayashi et al., 2007). Thus, the role of RAD50 in DSB formation exhibits similarities between mammals and yeast. Spo11-induced DSBs are then repaired by HR, which involves nucleolytic degradation of the 5′-terminated strands (end resection) to generate 3′-ssDNA tails (Sun et al., 1991, 1989). This crucial step begins with a short-range resection catalyzed by the MRN/X complex, followed by a long-range resection facilitated by EXO1/DNA2 (Gray and Cohen, 2016; Mimitou and Symington, 2009; Tisi et al., 2020). It has been previously reported that in non-null rad50s mutants, meiosis-specific DSBs accumulate and the ends of the broken DNA molecules are not processed during yeast meiosis (Cao et al., 1990; Sun et al., 1991). Consistent with this, we observed a marked increase in unresected DSBs in Rad50-deficient mouse spermatocytes (Fig. 5). Specifically, we showed that Rad50 deficient spermatocytes displayed a substantial decrease in minimum and maximum resection endpoints assessed by END-seq (Fig. 5), indicating that RAD50 facilitates MRE11 nuclease activity and EXO1/DNA2 resection activity during mammalian meiosis. In agreement with the impairments in DSB formation and processing, we observed various synaptic failures, including asynapsis or incomplete synapsis between homologs, as well as synapsis between non-homologs, in null Rad50 mutant mice. The prominent reason for synaptic failures may be attributed to defective DSB repair, as evidenced by the higher levels of γH2AX signals observed in Rad50 mutant spermatocytes. In addition, Rad50 mutant mice exhibited a 21.7% reduction (1772 out of 8168) in DSB levels compared with controls, a reduction similar to that observed in Spo11^+/− spermatocytes (Cole et al., 2012). This reduced level of DSBs alone is not sufficient to cause synaptic failures in Spo11^+/− mice (Kauppi et al., 2013), suggesting a potential equivalence to Rad50 mutant mice.

Considering the contrasting phenotypic difference in Rad50s during mouse and yeast meiosis, it appears that the domains of Rad50 involved in the MR complex are not conserved, whereas the overall Rad50 protein is conserved in evolution. Despite significant progress in recent years regarding the understanding of the biochemical and structural properties of the MR complex, the interplay between RAD50 and MRE11 remains poorly understood. The phenotypes of mouse Rad50-depleted germ cells were shown here to be more severe than those of hypomorphic Rad50 mutations reported previously (Bender et al., 2002). Additionally, mutations targeting the invariant cysteines of the Rad50 Zn hook domain phenocopied the null Rad50 mutation in somatic cells, whereas heterozygous hook domain mutations exhibited normal meiosis (Roset et al., 2014). These observations support the notion that RAD50 is integral to the functions of the MR complex. However, the specific domains of RAD50 involved and how they collaborate with MRE11 to assemble an active DNA-processing complex in male germ cells remain unclear. To gain a comprehensive understanding of this process, it would be intriguing to further investigate germ cell-specific deletion of each domain in RAD50. In addition, although female mutants were infertile, there was no evident loss of oocytes in 2-week-old RAD50 mutant females. A previous study demonstrated that extensive defects in DSB repair in Dmc1^−/− mutants resulted in nearly complete oocyte loss by 18 dpp (Di Giacomo et al., 2005). It is conceivable that unresected DSBs in mutant oocytes could lead to a delayed DDR stemming from unprocessed DSBs. Exploring this possibility will be of interest in future investigations.

The conditional mutant alleles for Rad50 were generated by the CRISPR/Cas9 technology. To generate a Rad50-floxed mouse line, in which exon 4 and 5 of the Rad50 allele is flanked by loxp sites. The resulting male founder was bred with wild-type C57BL/6J (B6) females to generate heterozygous Rad50^flox/+ mice. To obtained germ cell-specific Rad50 conditional knockout mice, the Rad50^flox/flox mice [C57BL/6J (B6) background] were first mated to Stra8-GFPCre mice (Lin et al., 2017) or Vasa-Cre mice to generate Rad50^flox/+Stra8-GFPCre or Rad50^flox/+Vasa-Cre mice, and then bred with Rad50^flox/flox mice for excising the loxP-flanked exon 4 and 5 to generate Rad50^flox/ΔStra8-GFPCre (hereafter referred to as Rad50-sKO) or Rad50^flox/ΔVasa-Cre (hereafter referred to as Rad50-vKO) mice (Fig. S5). The PCR primers for genotyping are in Table S1. All animal experiments used in this study were conducted in accordance with the guidelines in the Animal Care and Use Committee at Shanghai Institute of Biochemistry and Cell Biology, Center for Excellence in Molecular Cell Science, Chinese Academy of Science.

Spermatocyte nuclear surface spreads were performed as previously described with minor modifications (Peters et al., 1997). Briefly, testes were collected and the tunica albuginea removed in pre-cooled PBS. The testes were scattered into seminiferous tubules mechanically and incubated in hypotonic buffer [30 mM Tris (pH 8.2), 50 mM sucrose, 5 mM ethylenediaminetetraacetic acid, 17 mM trisodium citrate dehydrate, 0.5 mM dithiothreitol and 0.5 mM phenylmethylsulfony fluoride] at room temperature for 15 min. The hypotonic seminiferous tubules were then removed into 100 mM sucrose solution (pH 8.2) and made into a cell suspension with tweezers mechanically. The cell suspension (10 μl) was added to presoaked slides with 40 μl of 1% paraformaldehyde [PFA; 0.1% Triton X-100 (pH 9.2)] and incubated for at least 3 h at room temperature in wet chambers with interspace. For spermatocyte nuclear surface spread immunostaining, the slides were washed in PBS-T (PBS with 0.1% Triton X-100) for 10 min, and blocked in blocking buffer [1% bovine serum albumin (BSA) and 4% donkey serum in PBS-T] for 1 h at room temperature, and then incubated with primary antibodies overnight at 4°C. Primary antibodies used for spermatocyte nuclear surface spread immunostaining as follows: rabbit anti-SYCP3 with mouse anti-γH2AX, mouse anti-SYCP3 with rabbit anti-SYCP1, rabbit anti-DMC1 with mouse anti-SYCP3, rabbit anti-RAD51 with mouse anti-SYCP3, and rabbit anti-RPA with mouse anti-SYCP3. Subsequently, the slides were washed in PBS-T three times and incubated with corresponding secondary antibodies then mounted in Prolong Gold Antifade medium with DAPI (Molecular Probes) and imaged with fluorescence microscope (Olympus). Detailed primary and secondary antibody information is provided in Table S2.

For histological analysis, testes were fixed in Bouin's buffer overnight at room temperature, embedded in paraffin wax and sectioned at 5 μm. Sections were deparaffinized, rehydrated and stained with Hematoxylin and Eosin (H&E). For immunofluorescence analysis, testes were fixed in 4% PFA at 4°C overnight. After embedding, sectioning, removing paraffin wax and rehydration, sections were boiled in 10 mM sodium citrate buffer (pH 6.0) for 15 min and allowed to cool to room temperature. Subsequently, the sections were blocked with blocking buffer (4% donkey serum and 1% BSA in PBS-T) for 1 h at room temperature, and incubated with the primary antibodies overnight at 4°C. Primary antibodies used for immunofluorescence as follows: goat anti-LIN28 with rabbit anti-SOX9, rabbit anti-MVH, and rabbit anti-PLZF with mouse anti-γH2AX. Detailed antibody information is provided in Table S2. After washing in PBS-T three times, the slides were incubated with Alexa Fluor 488-conjugated or Alexa Fluor 594-conjugated donkey secondary antibodies for 1 h at room temperature. Sections were washed in PBS-T and mounted in Prolong Gold Antifade medium with DAPI. For TUNEL staining, PFA-fixed sections were deparaffinized and rehydrated, and apoptosis was detected using a TUNEL BrightGreen Apoptosis Detection kit (Vazyme) according to the manufacturer's instructions.

Mice were injected intraperitoneally with EdU (Invitrogen, 50 μg/g body weight) in PBS as previously described (Lin et al., 2017). The labeled mice were then sacrificed 2 h later and the testes were removed for further fixation and sectioning. EdU incorporation of labeled testes sections was detected using a Click-It EdU Alexa Fluor 594 Imaging Kit (Invitrogen) according to the manufacturer's protocol, then the sections were stained using rabbit anti-PLZF for further immunofluorescence analysis.

Spermatogenic cells were synchronized as previously described (Chen et al., 2018). In brief, 2-dpp Rad50-control and Rad50-sKO mice were pipette fed with WIN 18,446 (Sigma, suspended in 1% gum tragacanth, 100 μg/g body weight) for 7 consecutive days. The day after these consecutive WIN 18,446 treatments, the mice received an intraperitoneal injection of retinoic acid (RA) (Sigma; 30 μg/g body weight, suspended in DMSO). Rad50-control and Rad50-sKO mice were sacrificed at day 8 of RA treatment for enrichment of zygotene spermatocytes. Spermatogenic cell isolation was performed as previously described (Chen et al., 2018). Briefly, testes from the mice were collected in pre-cooled PBS and the tunica albuginea was removed. The testes were then incubated in collagenase type I (120 U/ml in PBS) at 34°C with gentle agitation for 3 min. The seminiferous tubules were further digested with 5 ml 0.25% trypsin and 0.1 ml DNase I (5 mg/ml in PBS) by pipetting gently several times at 34°C for 5 min, followed by termination of digestion using 0.5 ml fetal bovine serum (FBS). The dissociated cell suspension was filtered according to size using 70 μm strainer, and centrifuged at 500 g for 5 min at 4°C. After discarding the supernatant from the pellet, the cells were resuspended in PBS for a further END-seq experiment. The efficiency of synchronized spermatogenic cells was further examined by nuclear spreading, using immunofluorescence staining, and by H&E staining of testicular sections, as described above.

END-seq was performed as previously described, with modifications (Canela et al., 2019, 2016; Paiano et al., 2020). In brief, single cell suspensions of synchronized (treatment of RA for 8 days) testicular cells were washed twice in ice-cold PBS and resuspended in ice-cold cell suspension buffer. The similar number of zygotene spermatocytes from both the control and mutant mice were used for the following experiments. After being equilibrated at room temperature for 5 min, the resuspended zygotene spermatocyte-enrichment testicular cells were embedded in 0.75% agarose (final concentration) and solidified at 4°C for 15 min. Subsequently, the embedded cells were lysed and digested with proteinase K (50°C for 1 h then 37°C for 7 h). After washing with TE buffer (Tris-EDTA, pH 8.0), the plugs were then digested with RNase at 37°C for 1 h followed treatment of Exonuclease VII (NEB) at 37°C for 1 h and Exonuclease T (NEB) at 24°C for 45 min to blunt the DNA ends. After blunting, A-tails were added to the free 3′-OH of blunted ends, then ligated with END-seq hairpin adaptor 1. Subsequently, agarose plugs were melted and dialyzed, and the released DNA was sonicated to a shear DNA fragments at a length of about 170 bp. Shear DNA fragments were collected and purified, then the DNA pellet was dissolved in 70 μl TE buffer. Biotinylated adaptor 1-DNA fragment was further isolated using MyOne Streptavidin Beads. The second end was repaired and A-tails added, then ligated with END-seq hairpin adaptor 2. Both sides of hairpin were digested by using USER (NEB) and the resulting DNA fragments amplified by PCR. The libraries were then sequenced using Illumina HiSeq 2000.

All experiments in this study were performed at least three times independently. A paired two-tailed Student's test or a Mann–Whitney U-test was used to evaluate statistical significance. Data are presented as mean±s.d. *P<0.05, **P<0.01 or ***P<0.001 were considered statistically significant.

We thank the Bioinformatics Core Facility of SIBCB for END-seq data analysis.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202312.reviewer-comments.pdf