MutSgamma promotes meiotic recombination and homolog pairing in mouse spermatocytes

Abstract

DNA repair by homologous recombination is required for parental chromosomes (homologs) to accurately segregate during mammalian meiosis. Meiotic recombination promotes but also relies upon pairing between homologs. This mutual dependence and the differential reliance between recombination and pairing in well-studied organisms have been difficult to deconstruct in the mammalian context. In budding yeast, MutSgamma, a heterodimer between MSH4 and MSH5 promotes crossover-specific recombination by protecting precursors, and in many organisms plays roles in pairing and synaptonemal complex formation. We use recombination and cytological assays to infer the role of MutSgamma in mouse spermatocytes. We find in 2 alleles of Msh5—a null and one bearing a mutation in its ATPase domain, that spermatocytes are severely compromised for recombination producing only a small fraction of noncrossovers. However, they are more proficient in interhomolog pairing particularly on the longer chromosomes than spermatocytes lacking meiotic recombination entirely. We propose that MutSgamma plays an earlier role in mouse than in budding yeast to stabilize D-loops upstream of all interhomolog recombination. Further, that nascent recombination interactions can promote successful interhomolog pairing despite not completing recombination.

Introduction

Errors in meiotic chromosome segregation are the leading cause of infertility, miscarriages, and developmental disorders (Hassold and Hunt 2001). During meiosis, homologous recombination between parental chromosomes (homologs) is required for their accurate segregation. In many organisms, including humans and mice, meiotic recombination is also required for homologs to find each other in the nucleus and to pair along their lengths (Baudat et al. 2000; Mahadevaiah et al. 2001; Tsubouchi and Roeder 2003; Zickler and Kleckner 2015). As such, interhomolog recombination and pairing are mutually dependent upon each other. However, which steps in recombination are required for pairing and how much interhomolog pairing is necessary for proficient recombination is not well understood, especially in mammals.

Most recombination in mammals occurs at genomic regions with high DNA double-strand break (DSB) frequency called hotspots (Cole et al. 2014; Cole and Jasin 2011; Jeffreys et al. 2004; Jeffreys and May 2004; Wu et al. 2010). The topoisomerase-like protein, SPO11, generates meiotic DSBs (Hunter 2015; Johnson et al. 2021; Keeney 2008; Keeney et al. 1997; Lam and Keeney 2014; Yamada et al. 2020), which are subsequently resected generating 3′ single stranded DNA. RAD51 and DMC1 bind these resected tracts to promote strand exchange with the intact duplex of the homolog creating a transient intermediate called the displacement loop (D-loop) (Fig. 1a) (Brown and Bishop 2014; Cloud et al. 2012; Hinch et al. 2020; Lao et al. 2013). Most (∼90%) DSBs in mammals are repaired as noncrossovers—a short stretch of DNA bearing the donor genotype—by synthesis-dependent strand annealing (SDSA). SDSA involves extending the 3′ end by DNA synthesis followed by ejection and annealing to the opposite 3′ end for filling in and ligation (Allers and Lichten 2001; McMahill et al. 2007). The second most common (∼8%) repair pathway in mammals is double-strand break repair (DSBR). DSBR involves conversion of the D-loop into the single-end invasion intermediate (SEI) that can then engage the second 3′ end to form a double Holliday junction (dHJ) (Allers and Lichten 2001; Bishop and Zickler 2004; Borner et al. 2004; Bzymek et al. 2010; Hunter and Kleckner 2001; McMahill et al. 2007; Premkumar et al. 2023; Schwacha and Kleckner 1995; Szostak et al. 1983). In meiosis, dHJs are most frequently resolved as crossovers—the reciprocal exchange of homolog arms—by the putative MutLgamma resolvase complex, which includes a heterodimer between MLH1 and MLH3 (Baker et al. 1996; Cannavo et al. 2020; Guillon et al. 2005; Kulkarni et al. 2020; Svetlanov et al. 2008). In mouse spermatocytes, we have shown that MLH3 plays an additional structural role in converting the SEI into a dHJ (Premkumar et al. 2023). In budding yeast, formation of crossovers by DSBR requires ZMM-dependent protection of SEIs from dissociation of the 3′ end (De Muyt et al. 2012; Jessop et al. 2006; Kaur et al. 2015; Marsolier-Kergoat et al. 2018; Oh et al. 2007; Oke et al. 2014; Tang et al. 2015). ZMMs are evolutionarily conserved between yeast and mammals and include MutSgamma, the MSH4/MSH5 heterodimer. Purified MutSgamma has high affinity for D-loop-like structures and HJs. Binding to HJs stimulates MutSgamma ATPase activity and its formation into a sliding clamp that could surround and protect annealed duplex structures (Snowden et al. 2004). In this manner, MutSgamma could promote the protection of recombination intermediates that are substrates for MutLgamma, consistently mutations altering MutSgamma ATP binding negatively affect the formation of MutLgamma-dependent crossovers (Milano et al. 2019; Nishant et al. 2010).

Fig. 1.

Model and cytological characterization of MutSgamma mutants. a) The prevailing model of possible roles of MutSgamma during meiotic recombination based upon work in mouse and other organisms (see Introduction). Double-strand breaks (DSBs) are resected, and the 3′ end can anneal to the intact homolog, creating a displacement loop intermediate, the D-loop. The most common pathway is synthesis-dependent strand annealing (SDSA), which exclusively results in short noncrossovers. The second most common pathway is DSB repair (DSBR), where MutSgamma protects the D-loop to form a single-end invasion (SEI) intermediate. The SEI can then capture the second 3′ end to create a double Holliday junction (dHJ) that MutSgamma also protects. Finally, the DSBR pathway exclusively generates MutLgamma-dependent crossovers. b) Representative immunofluorescent images of WT and both MutSgamma mutants in leptonema (L), zygonema (Z), pachynema (P) or pachynema-like (P-like), and diplonema (D), and metaphase I (MI) stages of meiotic prophase I. Cells were stained with SYCP1, SYCP3, and gammaH2AX. c) Quantification of B) showing the percent (±standard deviation, SD) of nuclei of the indicated genotypes at each stage of meiotic prophase. Raw numbers available in Supplementary Table 2. d) Representative immunofluorescent images of WT and both MutSgamma mutants at P or P-like stage. Cells were stained with SYCP1 and SYCP3. The scatter plot shows the length of the SYCP1 stain in microns for each nucleus (n = 15) analyzed with mean and SD. e) Representative immunofluorescent images of WT and both MutSgamma mutants at Z. Cells were stained with RPA2 and SCYP3. The scatter plot shows the number of RPA2 foci for each nucleus with mean and SD at Z and P. One hundred twenty nuclei were analyzed for each genotype. P-values for all graphs were determined using Fisher's exact test, 2-tailed, and corrected for multiple comparisons using Bonferroni (significant if <0.017). ****P < 0.0001; ***P = 0.0005; **P < 0.0030; *P < 0.017.

Meiotic recombination promotes the formation of and is, in turn, regulated by the synaptonemal complex (SC), a multi-protein higher-order structure that spans the interface between homologs in meiosis. In many organisms, ZMMs are required for the robust formation of the SC also known as synapsis (Agarwal and Roeder 2000; Dresser et al. 1994; Fung et al. 2004; Higgins et al. 2005; Lynn et al. 2007; Novak et al. 2001; Padmore et al. 1991; Page and Hawley 2004; Pyatnitskaya et al. 2019; Rockmill et al. 2003; Sym et al. 1993; Voelkel-Meiman et al. 2016; Voelkel-Meiman et al. 2013; Zickler and Kleckner 2015). Intriguingly, comparisons between MutSgamma mutants in model organisms reveal different dependencies for the proficiency of pairing, synapsis, and crossover recombination. For example, in budding yeast, MutSgamma mutants show ∼50% synapsis, have higher frequencies of SDSA, and lack MutLgamma-dependent crossovers (Borner et al. 2004; De Muyt et al. 2012; Hollingsworth et al. 1995; Hunter and Borts 1997; Jessop et al. 2006; Khazanehdari and Borts 2000; Marsolier-Kergoat et al. 2018; Nishant et al. 2010; Novak et al. 2001; Oh et al. 2007; Ross-Macdonald and Roeder 1994; Wang et al. 1999). In Sordaria macrospora lacking MSH4, synapsis and crossovers are compromised (Storlazzi et al. 2010), but homologs pair along their lengths proficiently. In plants, MutSgamma mutants also lack MutLgamma-dependent crossovers, but synapsis is delayed or not severely affected (Higgins et al. 2004; Higgins et al. 2008; Luo et al. 2013; Zhang et al. 2014). In Caenorhabditis elegans, only crossovers are severely affected when MutSgamma is disrupted (Kelly et al. 2000; Woglar and Villeneuve 2018; Zalevsky et al. 1999). MutSgamma mutants may have the strongest phenotype in mice with highly compromised synapsis, loss of both MutLgamma-dependent and -independent crossovers, apoptosis in mid-prophase I, and infertility (de Vries et al. 1999; Edelmann et al. 1999; Kneitz et al. 2000). Those bearing an amino acid modification (G596A) in the MSH5 ATPase Walker Type A motif are infertile but have a slightly milder phenotype than Msh5^−/− and Msh4^−/−spermatocytes (Milano et al. 2019) with compromised SC formation associated with few MutSgamma foci. Occasionally, a small fraction of Msh5^G596A spermatocytes can survive to metaphase I but completely lack crossovers. Based on these previous studies, it is thought that MutSgamma could play an analogous role in stabilizing recombination intermediates in mouse. However, the relationship between recombination and pairing is poorly understood, making it difficult to define the role(s) that MutSgamma plays in these interdependent processes.

We used recombination and cytological analysis to infer the role of MutSgamma in the protection of recombination intermediates in mouse and how that protection influences interhomolog pairing. We find that Msh5^−/− and Msh5^G596A spermatocytes are severely compromised for all meiotic recombination with no evidence of crossover intermediates and a low frequency of SDSA-dependent noncrossovers. However, they are more proficient in interhomolog pairing than recombination, with substantially more pairing than spermatocytes lacking SPO11. We propose that unstable interhomolog intermediates that fail to complete recombination can contribute to pairing, at least in mouse spermatocytes. Overall, our data suggest that MutSgamma protects an earlier recombination intermediate, such as the nascent D-loop upstream of both SDSA and DSBR. We propose that MutSgamma, therefore, plays an earlier role than expected during mammalian meiotic recombination.

Results and discussion

Meiotic prophase progression and the synaptonemal complex are disrupted in MutSgamma mutants

In order to determine which steps of meiotic recombination depend upon MSH5 and MutSgamma activity in the mouse, we backcrossed the Msh5^−/− and Msh5^G596A alleles into the C57BL6/J (B) background and separately into the DBA/2J (D) mouse backgrounds to generate BxD F1 hybrids for recombination analysis. In the BxD background, homozygous Msh5^−/− and Msh5^G596A spermatocytes have high expression of the DNA damage marker gammaH2AX (Rogakou et al. 1998) that is indistinguishable from WT in leptonema but maintain high levels through zygonema and into a pachynema-like stage suggesting compromised DSB repair (Fig. 1b). Similar to the previously described Msh5^−/− phenotype in the B × 129/Sv mixed background (Barchi et al. 2005) and Msh5^G596A in the B background (Milano et al. 2019), both mutant spermatocytes lack accumulation of gammaH2AX into the sex body. They also fail to accumulate gammaH2AX into a pseudo-sex body not associated with the sex chromosomes that can be seen in some recombination mutants like Spo11^−/− (Barchi et al. 2005). Unlike Msh5^G596A spermatocytes in the B background, there was no progression past the pachynema-like stage in the BxD F1 hybrid (Fig. 1c). Thus, dependence upon MutSgamma is affected by strain background, perhaps owing to the presence of polymorphisms between homologs in the F1 hybrid.

There were more spermatocytes at leptonema and zygonema and fewer in a pachynema-like stage in Msh5^−/− and Msh5^G596A compared to WT. Detection of SC formation was assessed by staining for the lateral (SYCP3) and central (SYCP1) elements. The SC is severely compromised in both alleles (Fig. 1d) and is shorter than WT (279.5 ± 29.2 µm). The total length of SYCP1 staining was longer in the Msh5^G596A (129.5 ± 63.4 µm) as compared to the Msh5^−/− (60.9 ± 66.1 µm), suggesting that spermatocytes with compromised MSH5 ATPase activity may progress further through pachynema or are more proficient at SC formation than those lacking MSH5 entirely. However, without in situ hybridization, it is difficult to tell whether any particular SC is between homologous or nonhomologous regions.

Frequency and dynamics of RPA foci are similar between wildtype and MutSgamma mutants

MutSgamma is thought to stabilize interhomolog associations during DSB repair (Snowden et al. 2004). RPA2 foci are associated with ssDNA after 3′ resection. RPA is then replaced by RAD51 and DMC1 to mediate strand exchange. During meiotic prophase, RPA2 foci are also associated with the D-loop upon annealing of the broken 3′ end (Hinch et al. 2020). Generally, RPA2 foci showed similar dynamics in both mutants compared to wildtype (Fig. 1e). We found that the overall frequency of RPA2 foci in zygonema was slightly higher in Msh5^G596A (320 ± 69) than WT or Msh5^−/− spermatocytes (289 ± 57 and 298 ± 76, respectively). Whereas in pachynema/pachynema-like, RPA2 foci were slightly higher in Msh5^−/− (94 ± 41) compared to WT or Msh5^G596A spermatocytes (77 ± 30 and 80 ± 31, respectively). Similar frequencies of RPA2 foci in zygonema may reflect similar frequencies of D-loop formation or a longer half-life of resected 3′ ends associated with RPA when MutSgamma is compromised. However, in Msh5^G596A and Msh5^−/− spermatocytes, RAD51 foci form with similar dynamics and total number as WT (Edelmann et al. 1999; Milano et al. 2019) suggesting proficient exchange of RPA for RAD51 when MutSgamma is compromised. As such, the disappearance of RPA2 foci in pachynema-like cells in Msh5^G596A and Msh5^−/− spermatocytes suggests either repair or destabilization of the D-loops, however cytological analysis fails to distinguish between these states necessitating molecular analysis of recombination outcomes.

Interhomolog recombination is severely compromised but can be detected in MutSgamma mutants

In order to compare recombination between WT and MutSgamma mutants, we isolated spermatocytes using flow cytometry by labeling testicular cells with Hoechst 33342 (Supplementary Fig. 1). Hoeschst 33342 binds DNA and emits a blue and red fluorescence. The blue fluorescence is proportional to the concentration of DNA, allowing 4C cells engaged in meiotic prophase I to be distinguished. The red fluorescence signal strengthens as chromosomes compact in later meiotic prophase (Zickler and Kleckner 2023), distinguishing between early and late 4C spermatocytes (Patel et al. 2019). Both MutSgamma mutants lack spermatids (1C), secondary spermatocytes (2C), and 4C spermatocytes with high levels of red fluorescence consistent with their inability to progress past a pachynema-like stage (de Vries et al. 1999; Edelmann et al. 1999; Milano et al. 2019) (Supplementary Fig. 1).

Spermatocytes at the latest stage of meiotic prophase were isolated, and high molecular weight DNA was extracted for recombination analysis (Supplementary Fig. 2). Multiple PCRs were seeded in parallel with ∼20–40 genomes per reaction (Cole et al. 2014; Cole and Jasin 2011) to assess interhomolog recombination. The amplification reaction contains 1 allele-specific primer (ASP) to the B chromosome and the other nonspecific (universal, U). The combination of primary and secondary nested allele-specific PCR non-selectively amplifies the B parental configuration along with B to D crossovers and B noncrossovers. Genotyping the amplified products allows calculation of the frequency of (1) crossovers, (2) singleton noncrossovers that involve gene conversion of only a single polymorphism, and (3) co-converted noncrossovers that involve gene conversion of multiple and often adjacent polymorphisms (Premkumar et al. 2023; Zelazowski et al. 2017).

Recombination frequencies were assessed at 2 meiotic DSB hotspots: A3 and 59.5 located on chromosome 1, the longest, and chromosome 19, the shortest autosomes, respectively (Cole et al. 2010; Kelmenson et al. 2005; Wu et al. 2010; Zelazowski et al. 2017) (Fig. 2). The majority of SPO11-induced DSBs occur within the central ∼500 bp of the hotspots as determined by SPO11 oligonucleotide sequencing (Lange et al. 2016) (Fig. 2). In BxD F1 hybrids, A3 is a symmetrical hotspot that is equally likely to receive DSBs on either parental chromosome (Cole et al. 2010), and 59.5 is an asymmetrical hotspot with over 90% of DSBs incurred on the B chromosome (Zelazowski et al. 2017). At both hotspots, crossovers, singleton noncrossovers, and co-conversions can be readily detected in WT (Fig. 2, Supplementary Fig. 3, and Supplementary Table 1).

Fig. 2.

MutSgamma mutants are compromised at meiotic recombination. a) Representative population of co-conversions and singleton noncrossovers for the A3 hotspot isolated from WT, Msh5^G596A, and Msh5^−/− 4C spermatocytes. Top, the SPO11 oligonucleotide map in reads per million (Lange et al. 2016). The square or rectangle plots the minimum, and the line plots the maximum possible gene conversion tract. The top x-axis marks the polymorphisms analyzed, and the bottom x-axis marks the relative length and position within the hotspot, starting at 0 kb. The vertical dotted line marks the center of the hotspot, and the horizontal dashed line separates the singleton noncrossover and co-conversion events. The total number of events plotted (n_plot) or the total number of observed (n_tot) is shown at the bottom right. The total frequency ± standard deviation of singleton noncrossover and co-conversion events per 10,000 molecules is shown at the top left. b) Representative population of co-conversions and singleton noncrossovers for the 59.5 hotspot as in a). P-values were determined using Chi-square with Yates correction, 2-tailed.

Both MutSgamma mutants have severely compromised interhomolog recombination (Fig. 2 and Supplementary Table 1). As expected from previous cytological analyses of Msh5^−/− and Msh5^G596A spermatocytes (de Vries et al. 1999; Edelmann et al. 1999; Mahadevaiah et al. 2001; Milano et al. 2019), no crossovers were detected in either mutant despite high frequencies observed in WT littermates (Supplementary Table 1). The lack of crossovers in the MutSgamma mutants could be due to a failure to proceed to mid-pachynema when crossovers are likely formed (Wiltshire et al. 1995). Alternatively, as has been shown in budding yeast, MutSgamma is required for protection of the SEI crossover-specific intermediate and subsequently for most dHJs (Borner et al. 2004; De Muyt et al. 2012; Jessop et al. 2006; Kaur et al. 2015; Oh et al. 2007; Tang et al. 2015). We have previously shown that signatures of dissolved SEI- and dHJ-like intermediates can be detected in mutant mouse spermatocytes compromised in MutLgamma-dependent crossing over (Premkumar et al. 2023). In these mutants, significantly longer co-converted noncrossovers averaging greater than ∼300 and ∼600 bp can be detected that are consistent with dissolved SEI- and dHJ-like intermediates, respectively. We do not observe any co-conversions in the MutSgamma mutants, suggesting that they fail to form crossover-specific intermediates.

In WT, shorter co-conversions are common, making up 40% and 14% of noncrossovers at A3 and 59.5 and averaging 93 ± 56 and 232 ± 192 bp, respectively (Fig. 2) (Cole et al. 2010; Premkumar et al. 2023; Zelazowski et al. 2017). While some may be derived from dissolved intermediates as described for MutLgamma mutants, their length and frequency (particularly at A3) suggest derivation from SDSA-dependent gene conversion that happens to incorporate more than 1 polymorphism (Cole et al. 2014). Co-conversions were not seen in the MutSgamma mutants, suggesting that co-conversions, like crossovers, may occur at later stages of meiotic prophase than are found in Msh5^G596A and Msh5^−/−. Alternatively, co-conversions may be dependent upon robust MutSgamma activity. Taken together, the absence of long and short co-conversions in the MutSgamma mutants suggests that they do not generate SEI- or dHJ-like intermediates, consistent with the role of MutSgamma in budding yeast, but also with the model that these intermediates may form later in meiotic prophase than early pachynema when spermatocytes with compromised MutSgamma apoptose.

In contrast to co-conversions, singleton noncrossovers were observed in both mutant alleles, although at much lower frequencies ranging from 1 to 10% of that found in WT. At A3, noncrossovers were only detected in Msh5^−/− spermatocytes, but the frequency was not significantly different from that of Msh5^G596A (Fig. 2a). The 59.5 hotspot has a higher frequency of recombination than A3, allowing the isolation of more recombinants and the detection of events in both MutSgamma mutants. Msh5^G596A had a 10-fold higher frequency of singletons than Msh5^−/− spermatocytes (Fig. 2b).

Combined, interhomolog recombination is severely compromised in both mutants, which completely lack crossovers and co-conversions and have at most 10% of singleton noncrossovers. This phenotype suggests an early and essential role of MutSgamma in promoting all interhomolog recombination in mouse spermatocytes, likely by protecting an intermediate upstream of both SDSA and DSBR, such as nascent interhomolog D-loops (Supplementary Fig. 4). Congruently, MutSgamma binds to D-loop substrates with high affinity, even if it prefers HJs (Snowden et al. 2004). It is plausible that in mouse spermatocytes, MutSgamma stabilizes interhomolog D-loops prior to the formation of SEIs. Alternatively, MutSgamma may be required to protect intermediates undergoing SDSA. If so, gene conversion tracts might be shortened when MutSgamma is compromised making them impossible to detect at high frequency in our assays. In addition, our assay cannot detect nonallelic homologous recombination, which could be common in MutSgamma mutants.

In mouse, it is estimated that ∼5% of crossovers are formed via structure-selective endonucleases independently of MutLgamma (Guillon et al. 2005; Premkumar et al. 2023; Svetlanov et al. 2008; Zelazowski et al. 2017). Our model that MutSgamma is required to stabilize an intermediate upstream of the SDSA and DSBR bifurcation (Fig. 1a) is consistent with the absence of any crossovers in the rare Msh5^G596A spermatocytes that proceed to metaphase I (Milano et al. 2019). In mouse spermatocytes, the higher number of MutSgamma foci (∼150 in zygonema) than crossovers (∼26) is consistent with their formation at recombination intermediates primarily destined to become noncrossovers (Kneitz et al. 2000). This is in marked contrast to budding yeast, in which ZMM (including MutSgamma) protects crossover-specific intermediates (Supplementary Fig. 4) from the RecQ helicase complex Sgs1–Top3–Rmi1 (STR) (Kaur et al. 2015; Tang et al. 2015). Dissociation of nascent interhomolog joint molecules by STR is required for SDSA in budding yeast. Consequently, loss of MutSgamma leads to higher levels of SDSA-dependent noncrossovers (Cole et al. 2014; Marsolier-Kergoat et al. 2018). Perhaps in the mammalian setting, MutSgamma is required to stabilize/protect most interhomolog recombination intermediates. The dissociation of these intermediates to form regulated noncrossovers in wildtype could be accomplished through the concerted action of multiple, independent RecQ helicase and/or Zip3 ortholog complexes found in mammals (Constantinou et al. 2000; Holloway et al. 2010; Horan et al. 2024; Hunter 2008; Ito et al. 2025; Moens et al. 2000; Qiao et al. 2014; Reynolds et al. 2013 ; Sakamoto et al. 2001). Alternatively, higher levels of SDSA in budding yeast could reflect a failure of the homologs to fully engage with each other prompting higher levels of DSBs (Thacker et al. 2014). While this may be conserved in mouse, we see no evidence of higher RPA foci (Fig. 1d) or SDSA (Fig. 2), which may reflect our inability to detect very short gene conversion tracts when MutSgamma is compromised.

Does MutSgamma also protect SEI-like intermediates in mouse spermatocytes? There are almost 3-fold more crossovers than noncrossovers at 59.5 in wildtype. If we assume that nascent D-loops have a similar chance to form SEI-like intermediates when MutSgamma is compromised as in wildtype, we would expect to see as many as ∼28 co-conversions in the Msh5^G596A spermatocytes instead of none (P < 0.0001, Fisher's exact test, 2-tailed). This finding suggests that MutSgamma may stabilize later intermediates in the DSBR pathway. However, co-conversions derived from dissolved SEI-like intermediates may not be detectable in early pachynema, necessitating further experiments to definitively determine which downstream intermediates are protected by MutSgamma in mouse.

Spermatocytes with compromised MutSgamma are better at interhomolog pairing than those lacking SPO11

Noncrossovers can be detected in the MutSgamma mutants, suggesting that bona fide homolog association can occur despite severely compromised SC formation. An early role for MutSgamma in mouse spermatocytes may be the stabilization of interhomolog intermediates to promote homolog pairing. Homolog pairing or co-alignment precedes SC formation in Sordaria, with MutSgamma required for close juxtaposition between homologs prior to synapsis (Storlazzi et al. 2010; Zickler and Kleckner 2015). In WT, the SC is most often observed between homologs, but when meiosis is compromised, the SC often forms between nonhomologous regions including in MutSgamma mutants (Barchi et al. 2005; de Vries et al. 1999; Dresser et al. 1994; Edelmann et al. 1999; Higgins et al. 2004; Higgins et al. 2005; Hollingsworth et al. 1995; Jones et al. 2024; Kneitz et al. 2000; Novak et al. 2001; Ronceret et al. 2009; Storlazzi et al. 2010; Tsubouchi and Roeder 2003; Zickler and Kleckner 1999).

To determine the requirement for MutSgamma in interhomolog pairing, we used immunofluorescent detection of SYCP1 and SYCP3 in conjunction with fluorescence in situ hybridization (FISH) with whole chromosome paints to detect the longest (Chr 1 and 2) and shortest autosomes (Chr 18 and 19), and the sex chromosomes (Chr X and Y) (Fig. 3a–c). We assessed pairing proficiency at pachynema or pachynema-like stages in WT, Msh5^G596A, and Msh5^−/− along with spermatocytes lacking induction of meiotic recombination (Spo11^−/−).

Fig. 3.

Interhomolog pairing is disrupted in MutSgamma mutants but not as severely as Spo11^−/−. a) Representative FISH images of WT, Msh5^G596A, Msh5^−/−, and Spo11^−/− mutants at pachynema or pachynema-like stage. Cells were stained with SYCP3, chromosome 18, and chromosome 19. b) Representative FISH images of WT, Msh5^G596A, Msh5^−/−, and Spo11^−/− mutants at pachynema or pachynema-like stage. Cells were stained with SYCP3 , chromosome X , and chromosome Y. c) Representative FISH of Msh5^−/− mutant at pachynema-like stage showing an example of paired (arrow) and unpaired chromosomes (arrowhead). Cells were stained with SYCP3, SYCP1, chromosome 1, and chromosome 2. The top image shows SC formation via immunofluorescence with a zoomed in portion below. Below that, the SYCP1 staining is traced. The FISH staining of the same cell above is shown, with the same zoomed in portion below including the SYCP1 trace. d) Bar graph showing the proportion of nuclei with fully paired homologs at the pachynema/pachynema-like stage for each chromosome or chromosome pair (Chr X and Y) analyzed. For each genotype, 3 independent animals and 45 cells were analyzed. P-values were determined by Fisher's exact test, 2-tailed. ****P < 0.0001; *P < 0.05.

The longer autosomes were more likely to be paired when MutSgamma is compromised than in Spo11^−/− (Fig. 3d). By contrast, the shorter autosomes were similar between MutSgamma mutants and Spo11^−/− with only a slight increase in interhomolog pairing at chromosome 19 in Msh5^−/− compared with Spo11^−/−. No interhomolog pairing was observed between sex chromosomes in any of the mutants, consistent with the observation that X and Y chromosomes are late to pair (Kauppi et al. 2011) and most frequently mis-segregate when meiotic recombination is compromised (Kauppi et al. 2012). Interhomolog pairing is highly sensitive to compromised MutSgamma activity. For example, despite longer SCs and higher recombination frequency in Msh5^G596A vs Msh5^−/−, no significant difference in interhomolog pairing was detected between the mutants. Although there may be a trend of less pairing in Msh5^−/− than Msh5^G596A spermatocytes, the number of cells needed to confirm this finding is labor and cost-prohibitive with relatively limited probative value. Taken together, longer chromosomes appear most adept at interhomolog pairing in the absence of robust MutSgamma activity. Perhaps the higher total number of meiotic DSBs on the longer autosomes provides more opportunity for successful interhomolog interactions when MutSgamma is compromised (Kauppi et al. 2013). Importantly, interhomolog pairing, as assessed by whole chromosome FISH, appears more robust in the MutSgamma mutants than interhomolog recombination at specific hotspots on those chromosomes. For example, chromosome 1 was paired in 41% and 21% of Msh5^G596A vs Msh5^−/− pachynema-like cells, but we detected <0.8% and 4.5% of wildtype recombination at the A3 hotspot, respectively (Supplementary Table 1). In the absence of robust MutSgamma activity in mouse spermatocytes, it is possible that nascent D-loops can initiate pairing on a chromosome-wide basis, particularly for the longer autosomes, even if they fail to generate noncrossovers.

Chromosome dynamics in Msh5^G596A are more similar to WT than Msh5^−/− spermatocytes

We sought to determine if there were differences in chromosome dynamics in Msh5^G596A vs Msh5^−/− spermatocytes, particularly in homolog pairing between the MutSgamma mutants. As meiotic prophase progresses, chromosomes compact, and homologs pair with one another, reducing the proportion of nuclear territory homologs occupy. Comparing leptonema, zygonema, and pachynema, we measured the area of each chromosome FISH signal as a fraction of the DAPI-stained area (Fig. 4a and Supplementary Fig. 5). In WT, the percent of the nucleus comprised of chromosome paint signal reduced for all chromosomes as spermatocytes progress from leptonema to pachynema (Fig. 4b). The longer autosomes showed a discrete stepwise reduction in percent nuclear occupancy. For example, chromosome 1 incorporated 27% of the nucleus in leptonema, 18% in zygonema, and 8% in pachynema. The shorter autosomes and sex chromosomes showed reduced occupancy as prophase progressed, with the biggest reduction occurring between zygonema and pachynema. For example, chromosome 19 incorporated 21% of the nucleus in leptonema, 18% in zygonema, and 2% in pachynema. The different dynamics we observe between long and short autosomes are congruent with the finding that shorter chromosomes synapse later than longer ones in meiotic prophase (Kauppi et al. 2013; Kauppi et al. 2012; Murakami et al. 2020).

Fig. 4.

Chromosome dynamics in WT, MutSgamma mutants, and Spo11^−/−. a) Representative FISH images of WT, Msh5^G596A, Msh5^−/−, and Spo11^−/− mutants at leptonema, zygonema, and pachynema/pachynema-like stages. Cells were stained with SYCP3, chromosome 1, and chromosome 2. DAPI is shown in a separate panel. The dashed gray circle represents the DAPI area, and the dashed lines represent the Chr 1 and Chr 2 areas. b) Bar graph showing the percent of the DAPI staining with chromosome signal for each chromosome or chromosome pair (Chr X and Y) with standard deviation in WT. The stages analyzed are leptonema (L), zygonema (Z), and pachynema (P). For each genotype, 3 independent animals and 45 cells were analyzed. c) Bar graph showing the percent of the DAPI staining with chromosome signal for each chromosome or chromosome pair (Chr X and Y) with standard deviation in WT, Msh5^G596A, Msh5^−/−, and Spo11^−/− mutants at pachynema or pachynema-like stage. For each genotype, 3 independent animals and 45 cells were analyzed. d) Line graph showing the percent of DAPI staining with chromosome signal for each chromosome or chromosome pair (Chr X and Y) as indicated with standard error of the mean analyzed in WT, Msh5^G596A, Msh5^−/−, and Spo11^−/− mutants at leptonema (L), zygonema (Z), and pachynema/pachynema-like (P) stages. For each genotype, 3 independent animals and 45 cells were analyzed. P-values were determined by Kruskal–Wallis and Mann–Whitney tests if the Kruskal–Wallis test showed significance. ****P < 0.0001; ***P = 0.0004; **P < 0.0079; *P < 0.04.

Contrary to WT, spermatocytes lacking SPO11 showed no difference in the percent nuclear occupancy with meiotic prophase progression (Kruskal–Wallis test, Fig. 4c and Supplementary Fig. 6). This finding in Spo11^−/− spermatocytes suggests either they lack both chromosome compaction during meiotic prophase progression and interhomolog pairing or that interhomolog pairing provides the majority of the reduction in chromosome nuclear occupancy we measure.

Chromosome dynamics in Msh5^G596A were more similar to WT than Spo11^−/− spermatocytes. The longer autosomes showed a steady reduction in percent nuclear occupancy from leptonema to pachynema. The shorter autosomes showed a marked decrease in nuclear occupancy in Msh5^G596A spermatocytes between leptonema and zygonema, whereas, in WT, a marked reduction was seen between zygonema and pachynema. The sex chromosomes in Msh5^G596A spermatocytes were more similar to the long autosomes, showing a steady decrease as meiotic prophase progressed (30% in leptonema to 23% in zygonema to 17% in pachynema) (Fig. 4c and Supplementary Fig. 6). Overall, while the trend of reduced nuclear occupancy was similar between Msh5^G596A and WT, the total reduction in area in WT was not observed in Msh5^G596A spermatocytes, with chromosomes occupying 1.6–3.5-fold more nuclear territory than WT. This result is consistent with our previous finding (Fig. 1c) that Msh5^G596A spermatocytes do not progress past the pachynema checkpoint required to achieve maximal chromosome compaction and do not entirely pair homologs (Fig. 3d).

Despite detecting no difference in interhomolog pairing at the pachynema-like stage between Msh5^G596A and Msh5^−/−, Msh5^−/− spermatocyte dynamics were more similar to Spo11^−/− than Msh5^G596A, particularly for the shorter autosomes and sex chromosomes (Fig. 4c and Supplementary 6). The longer autosomes showed a significant reduction in percent nuclear occupancy in Msh5^−/− spermatocytes, unlike Spo11^−/− spermatocytes. For example, in Msh5^−/− spermatocytes, chromosome 2 incorporated 20% of the nucleus in leptonema, 13% in zygonema, and 8% in pachynema. In Msh5^−/−, the total area of chromosome signal in pachynema was between 1.3- and 7.5-fold larger than in WT spermatocytes. Mostly, the shorter autosomes and sex chromosomes were unchanged between stages in Msh5^−/− like Spo11^−/− and unlike Msh5^G596A spermatocytes. This finding suggests that Msh5^−/− spermatocytes struggle with chromosome compaction and interhomolog pairing regarding the shorter autosomes and sex chromosomes, similar to when all meiotic recombination is lost in Spo11^−/− spermatocytes.

Intriguingly, much of the differences between WT and the mutants could be attributed to the differences observed in leptonema, with Msh5^G596A and Msh5^−/− occupying a larger and smaller fraction of the nucleus than WT, respectively (Fig. 4c). These differences are consistent with an early role for MutSgamma in promoting interhomolog pairing and/or chromosome compaction. Comparing nuclear occupancy in the pachynema and pachynema-like stages shows that MutSgamma mutants have an intermediate phenotype between WT and Spo11^−/− spermatocytes (Fig. 4d). In general, chromosome dynamics for the long autosomes were more similar between WT and both MutSgamma mutant spermatocytes (Supplementary Fig. 6). For the shorter autosomes and sex chromosomes, Msh5^G596A were more similar to WT than Msh5^−/− spermatocytes. These findings suggests that Msh5^G596A spermatocytes may have slightly better interhomolog pairing than Msh5^−/− spermatocytes, but their phenotypes are similar.

Conclusions

Based on previous studies, MutSgamma is involved in homologous recombination, interhomolog pairing, and SC formation (Borner et al. 2004; Dash et al. 2024; De Muyt et al. 2012; Higgins et al. 2004; Higgins et al. 2008; Hollingsworth et al. 1995; Jessop et al. 2006; Khazanehdari and Borts 2000; Luo et al. 2013; Marsolier-Kergoat et al. 2018; Nandanan et al. 2021; Nishant et al. 2010; Novak et al. 2001; Oh et al. 2007; Oke et al. 2014; Ross-Macdonald and Roeder 1994; Storlazzi et al. 2010; Voelkel-Meiman et al. 2016; Zalevsky et al. 1999; Zhang et al. 2014). In mouse, the exact roles of MutSgamma in recombination and how those relate to interhomolog pairing were not understood (de Vries et al. 1999; Edelmann et al. 1999; Milano et al. 2019). Combining cytological and molecular analysis, we aimed to clarify the relationship between recombination and pairing. We found that MutSgamma is required for most homologous recombination but that interhomolog pairing was disrupted to a lesser degree, particularly for the longer autosomes. Like Sordaria, we detect pairing when MutSgamma is compromised, but dissimilarly, chromosomes fail to align fully along their lengths (Storlazzi et al. 2010). We propose that, at least in mouse spermatocytes, nascent interhomolog interactions that fail to complete recombination can contribute to interhomolog pairing. We find that recombination is more dependent upon MutSgamma in mouse spermatocytes than in any of the other highly studied model organisms, perhaps owing to the large size and complexity of the mammalian genome. In plants, crossovers and synapsis are readily detected in the absence of Msh4 or Msh5 (Higgins et al. 2004; Higgins et al. 2008; Luo et al. 2013; Zhang et al. 2014), but both are severely compromised in mouse spermatocytes (de Vries et al. 1999; Edelmann et al. 1999; Kneitz et al. 2000; Milano et al. 2019). Previous analysis of mouse mutants suggested that MutSgamma plays an early and essential role in meiotic recombination. Congruently, SDSA-dependent noncrossovers are higher in budding yeast when MutSgamma is compromised (Borner et al. 2004; Dash et al. 2024; De Muyt et al. 2012; Hollingsworth et al. 1995; Jessop et al. 2006; Khazanehdari and Borts 2000; Marsolier-Kergoat et al. 2018; Nandanan et al. 2021; Nishant et al. 2010; Novak et al. 2001; Oh et al. 2007; Oke et al. 2014; Ross-Macdonald and Roeder 1994; Voelkel-Meiman et al. 2016), but we find 10–100-fold fewer noncrossovers in mouse. Therefore, MutSgamma is required for most interhomolog recombination in mouse spermatocytes and likely stabilizes and/or protects earlier recombination intermediates that are upstream of both crossover and noncrossover pathways (Supplementary Fig. 4). Taken together, our findings further support the model that MutSgamma plays an earlier role than expected during mammalian meiotic recombination.

Materials and methods

Mice

The Institutional Animal Care and Use Committee approves our mouse studies (ACUF—00001132-RN04). Experimental animals, C57BL/6J × DBA/2J (BxD) F1 hybrid male mice, were generated by crossing inbred mouse strains C57BL/6J (B) and DBA/2J (D). Single mutants (Msh5^−/−, Msh5^G596A) were generated from independently maintained heterozygous mice in separate B and D strains that were genotyped to verify homozygosity for the A3 and 59.5 hotspots in both B and D strains. Genotyping information can be found in Supplementary Table 3.

Spermatocyte spreads and immunofluorescence

This protocol was performed as described in Premkumar et al. (2023) and Zelazowski et al. (2017). One part of or whole testis was decapsulated and transferred to 2 mL of testis isolation media (TIM) supplemented with 4 mg of collagenase [TIM: 104 mM NaCl, 45 mM KCl, 1.2 mM MgSO₄, 0.6 mM KH₂PO₄, 0.1% (w/v) glucose, 6 mM Na lactate, 1 mM Na pyruvate, pH 7.3, filter sterilized, and stored at room temperature]. The mixture was shaken for 55 min at 32°C at 500 rpm and inverted roughly every 15 min. TIM was added to a final volume of 15 mL, seminiferous tubules allowed to settle to the bottom, and the supernatant was removed using a transfer pipette. Seminiferous tubules were washed twice with 15 mL of TIM. Afterwards, the tubules were resuspended in 2 mL of TIM with 1.4 mg of trypsin and 20 μL of DNAse I (0.6 Kunits/μL). The solution was shaken for 15 min at 32°C at 500 rpm and occasionally inverted, after which 500 μL of trypsin inhibitor (10 mg/0.5 mL) was added along with 50 μL of DNAse I. Cells were separated by repeatedly pipetting with a transfer pipette for 2 min. The solution was filtered through a 70 μm filter top, and TIM was added to a final volume of 15 mL. This was spun for 5 min at 1,000 rpm in a benchtop centrifuge, and the supernatant was removed. Afterward, 15 μL of DNAse I was added to the pellet, which was resuspended by flicking the tube, and 15 mL of TIM was added. This was repeated once, after which the pellet was resuspended by flicking the tube after adding 15 μL of DNAse I. Depending on the pellet size (generally related to the genotype of the mouse), 4–12 mL of 1× PBS was added, and 1 mL was transferred to 1.5 mL tubes, which were spun for 5 min at 1,500 rpm in a microcentrifuge. The supernatant was removed, and depending on the pellet size, 80–120 μL of 0.1 M sucrose (warmed to 37°C) was added to resuspend the pellet, and the solution is incubated for 5 min at room temperature. Meanwhile, slides were prepped in a humidifying chamber with 65 μL of 1% paraformaldehyde (pH 9.2) with 0.1% TritonX-100 spread across the surface. After 5 min, 20 μL of sucrose suspension was dropped onto each slide, which were incubated in a closed chamber at room temperature for 2.5 h, then ajar for 30 min, then open for 30 min, after which the slides were rinsed once with water and twice with water containing a 1:250 Photo-Flo 200 solution. The slides were then air-dried and used immediately or stored at −80°C.

For immunofluorescent antibody staining, slides (freshly made or thawed) were incubated with 100 μL of each antibody dilution buffer (ADB) solution (ADB: 1× PBS, 3% IgG-free BSA, 0.05% TritonX-100, 10% goat serum) covered by parafilm within the humidifying chamber. Slides were blocked for 30 min at 37°C with ADB. Block was removed and replaced with primary antibody diluted in ADB overnight at room temperature. Slides were washed in Coplin jars while being agitated on a platform shaker for 5 min with 1× PBS containing 0.4% Photo-Flo 200, then for another 5 min with 1× PBS containing 0.4% Photo-Flo 200 and 0.01% TritonX-100. Slides were re-blocked with ADB for 10 min at 37°C. They were then incubated with secondary antibody diluted in ADB for 1 h at 37°C. Slides were then washed in the dark as before followed by a final wash for 1 min with water containing 0.4% Photo-Flo 200. They were air-dried in the dark and mounted with coverslips using Prolong Gold antifade with DAPI, then left flat and in the dark until ready for viewing.

FISH

After staining with immunofluorescence antibodies and imaging, the slides were baked for 1 h at 65°C. Slides were then incubated in prewarmed (37°C) 2× SSC (20×: 3 M NaCl and 0.3 M citric acid trisodium salt dihydrate) with 0.5% TritonX-100 in a glass Coplin jar at 37°C for 30 min after which they were washed with 2× SSC for 2 min at room temperature. After washing, the slides were treated with the Metasystems protease buffer for 5 min and washed again in 2× SSC for 5 min at room temperature. Slides were dehydrated in 70%, 80%, and 95% ethanol for 2 min each at room temperature and then air-dried. As the slides were treated, the Metasystems probes were prewarmed at 37°C for 5 min, aliquoted, denatured at 72°C for 10 min, and placed in a 37°C water bath to pre-anneal for 30 min to 2 h. Ten microliters of the probe mixture was applied to the slides, which were covered with a coverslip and sealed with nail polish. The slides were denatured at 75°C for 2 min and then incubated overnight in a humidifying chamber at 37°C. The next day, the slides were soaked in 50 mL of 2× SSC with 0.5% TritonX-100 to remove the coverslips. Slides were washed in prewarmed 0.4× SSC with a pH of 7–7.5 at 72°C for 2 min with no agitation. Once drained, the slides were washed in 2× SSC with 0.05% Tween 20 at room temperature for 30 s then rinsed in water, 10 μL of DAPI with antifade was added, and the slides were cover slipped and imaged after incubating for at least 10 min. The Metasystems stain and probes used were DAPI (D-0902-500-DA), chromosome 1 (D-1401-050-FI), chromosome 2 (D-1402-050-OR), chromosome 18 (D-1418-050-FI), chromosome 19 (D-1419-050-OR), chromosome X (D-1420-050-FI), and chromosome Y (D-1421-050-OR).

Isolation of 4C spermatocytes by flow cytometry

Spermatocytes were isolated as described in Cole et al. (2014), Premkumar et al. (2023), and Zelazowski et al. (2017). One testis or ∼1.5 testes were decapsulated and transferred to 15 mL of Gey's Balanced Salt Solution (GBSS) containing 0.75 mg/mL of collagenase and shaken for 15 min at 33°C at 500 rpm, occasionally inverted about 10 times (roughly every 5 min). The supernatant was removed using a transfer pipette, being very careful not to remove any seminiferous tubules, after which the tubules were washed with 10 mL of GBSS, again removing the supernatant. For mutant samples, 15 mL of GBSS containing 0.75 mg/mL of trypsin and 25 μL of DNAse I (0.6 Kunits/μL) was added. For WT samples, an additional 25 μL of DNAse I was added. The solution was shaken for 15 min at 33°C at 500 rpm and occasionally inverted as before, after which 0.75 mL of Newborn Calf Serum (NCS) was added. The cells were then separated by repeated pipetting for 3 min using a transfer pipette before being filtered using a 70 μm filter top. The solution was spun for 3 min at ∼1,800 rpm in benchtop centrifuge, and the supernatant was removed. Next, 25 μL of DNAse I was added, and the pellet was resuspended by flicking the tube. The cells were then washed with 10 mL of GBSS containing 2% NCS, followed by an additional 10 μL of DNAse I. The sample was spun for 3 min at ∼1,800 rpm, the supernatant was removed, and the pellet was resuspended by flicking the tube. For WT samples, 9 mL of GBSS containing 2% NCS and 24 μL of DNAse I were added along with 18 μL of Hoechst 33342 (2.5 mg/mL in DMSO, stored at 4°C). For mutant samples, 3 mL of GBSS containing 2% NCS and 8 μL of DNAse I were added along with 6 μL of Hoechst 33342. The sample was shaken for 45 min at 33°C at 500 rpm, after which propidium iodide (1 mg/mL, stored at 4°C) was added (1.8 μL for WT or 0.6 μL for mutant). The sample was filtered again using a 70 μm filter top and transferred to collection tubes. Cells were sorted using either a BD Aria or BD Fusion flow cytometer with a UV laser (350 nm argon laser). Using blue and red fluorescence from Hoechst 33342, cells were isolated into different meiotic populations as described in the paper. After sorting, a very small portion of cells was counted with a hemocytometer to gauge the number of cells collected. To assess purity, approximately 20,000 cells were immediately washed with 1× PBS and spun for 5 min at 1,500 rpm in a microcentrifuge. The pellet was resuspended in 80 μL of sucrose warmed to 37°C and left at room temperature for 5 min. Afterward, 20 μL of the sucrose suspension was dropped onto slides already prepped with 65 μL of 1% PFA with 0.1% TritonX-100 (spread across the slides). These slides were placed in a humidifying chamber, which was left closed at room temperature for 2.5 h, then ajar for 30 min, then open for 30 min, after which the slides were rinsed once with water and twice with water containing a 1:250 Photo-Flo 200 solution. The slides were then air-dried and stained with SYCP3, SYCP1, and gammaH2AX as described in immunofluorescence staining (quantified in Supplementary Table 2). Meanwhile, aliquots of sorted cells were pelleted in a 1.5 mL tube at 750 g for 5 min, the supernatant removed, and pellets were snap-frozen in an ethanol/dry ice slurry before being transferred to −80°C for later use.

Genomic DNA isolation

The genomic DNA was isolated using the QIAamp DNA Micro kit from Qiagen. The protocol followed was the “Isolation of Genomic DNA from Urine,” starting with resuspending the pellet in 300 μL of ATL buffer (preheated at 70°C to dissolve any precipitates) and adding 20 μL of proteinase K. This solution was mixed by vortexing for 10 s. The mixture was then incubated at 56°C for 1 h while shaking at 900 rpm. After incubation, the tube was briefly centrifuged to remove any liquid inside the lid, and 300 μL of AL buffer was added along with 50 μL of room temperature 100% ethanol. This was again vortexed for 10 s and briefly centrifuged. The supernatant was then transferred to a column in a 2 mL collection tube and centrifuged at 8,000 rpm for 1 min in a microcentrifuge. The column was placed in a new collection tube, and 500 μL of AW1 buffer was added. This was centrifuged again for 1 min at 8,000 rpm, and the column was placed in a new collection tube. This step was repeated with AW2 buffer, and after each centrifugation, the flow-through was discarded. The sample was centrifuged for 3 min at full speed (14,000 rpm) to dry, and the column was transferred to a clean 1.5 mL tube. Fifty microliters of 5 mM Tris (pH 7.4) was then added, and the sample was incubated at room temperature for 5 min and then centrifuged for 1 min at full speed. The sample was ready for the recombination assay after incubation overnight at 4°C.

Determining amplification efficiency of DNA

Amplification efficiency on isolated DNA from sorted samples was performed as described in Cole and Jasin (2011), Premkumar et al. (2023), and Zelazowski et al. (2017). To determine an estimate of the DNA concentration and quality, a nanodrop was used along with running a dilution series on a 1% agarose gel against a 12 ng/μL DNA control. The number of amplifiable DNA molecules/pg was determined using 24–48 PCRs seeded with 2 amplifiable molecules per reaction. Each PCR consisted of 2 nested reactions. This was based on each sample and each hotspot, and the conditions of the PCRs were the same as those used for the recombination assay. Both rounds of PCR contained 1 ASP and 1 universal primer (Supplementary Table 4), and the amplified PCR product was run on a 1% agarose gel. To calculate the amplification adjustment factor, the number of negative wells was counted and input into the equation (−ln(total number of negative wells/total number of wells))/2. For reproducibility, the amplification adjustment factor should be between 0.2 and 0.8. If the amplification adjustment factor was over or under this range, the concentration of the DNA is recalculated and the amplification efficiency assay repeated.

Recombination assay

The recombination assay was done as described in Cole and Jasin (2011), Cole et al. (2010), Premkumar et al. (2023), and Zelazowski et al. (2017). The amount of DNA added was calculated based on the amplification efficiency assay. The input DNA was either ∼20 or 40 genomes per well, so the recombinant signal was detectable over the nonrecombinant background. The primer annealing temperature was specifically determined for each lot of 2× Q5 master mix and each batch of 11.1× buffer made. For each primary PCR reaction (total volume of 8 μL), DNA was added to 1× buffer containing 2× Q5 master mix from New England Bio Labs and 0.2 μM of each primer (1 ASP and 1 universal). These primer pairs were used to non-selectively amplify crossovers, noncrossovers, and nonrecombinant DNA. Thirty-five microliters of dilution buffer (10 mM Tris–HCl pH 7.5 and 5 μg/mL sonicated salmon sperm DNA) was added to the primary PCR plate. A total of 1.6 μL of this diluted product was then added to the secondary PCR plate. For the secondary plate, the 1× buffer consisted of 11.1× buffer (10×: 450 mM Tris–HCl pH 8.8, 110 mM (NH₄)₂SO₄, 45 mM MgCl₂, 67 mM beta-mercaptoethanol, 44 μM EDTA, 10 mM each: dATP, dTTP, dGTP, and dCTP, and 1.13 mg/mL non-acetylated BSA), 12.5 mM Tris-base, 0.2 μM of each primer, 0.25 U of Taq, and 0.05 U of Pfu polymerase to a final volume of 30 μL including the DNA input from the primary plate. The PCR program for the primary PCR at A3 was denaturation (98°C, 1 min for the first denaturation and 20 s for subsequent steps), annealing (30 s at optimized temperature), and extension (72°C, 1 min per kb). The secondary PCR for A3 was the same, except the denaturation temperature was 96°C and the extension temperature was 68°C. The PCR program for the primary PCR at 59.5 was denaturation (96°C, 1 min for the first denaturation and 20 s for subsequent steps), annealing (30 s at optimized temperature), and extension (68°C, 1 min per kb). The secondary PCR for 59.5 was the same, except the extension temperature was 68°C. For A3, the primary PCR had 27 amplification cycles, while for 59.5, the primary PCR had 26. For both hotspots, the secondary PCR had 36 cycles of amplification. The entire secondary PCR product was then dot blotted to a positively charged nylon membrane in the 96-well format and genotyped by Southern blotting using allele-specific oligos (ASO) probes.

Genotyping with ASO probes

Genotyping amplicons from the recombination assay was done as described in Cole and Jasin (2011), Cole et al. (2010), Premkumar et al. (2023), and Zelazowski et al. (2017). Initially, each ASO (both genotypes, Supplementary Table 5) was diluted to 1:100 in water in a 1.5 mL tube while in a screw cap tube, the kinase reaction solution [1× kinase buffer (70 mM Tris–HCl, pH 7.5, 10 mM MgCl₂, 5 mM spermidine trichloride, 2 mM dithiothreitol), 0.35 μL T4 polynucleotide kinase (10 U/μL), 0.2 μL of gamma-32P-ATP (10 mCi/mL), and 1 μL of ASO (8 μg/mL)] was mixed to a final volume of 10 μL per 2 blots. The ASO radiolabeled in this case is from the D genotype and this was incubated at 37°C for 1 h. The reaction was stopped by adding 20 μL of kinase stop solution (25 mM EDTA, 0.1% SDS, 10 μM ATP) and adding 20 μL of unlabeled ASO from the B genotype. The dot blot membranes were rinsed in 2× SSC and pre-hybridized for 15 min at 56°C in a rotisserie hybridization oven with 3 mL of tetramethylammonium chloride (TMAC) hybridization buffer (buffer: 3 M TMAC, 0.6% SDS, 10 mM NaPO₄ pH 6.8, 1 mM EDTA, 4 μg/mL yeast RNA, and 5× Denhardt's solution [50×: 1% (w/v) Ficoll 400, 1% (w/v) polyvinyl pyrrolidone, 1% (w/v) BSA (Fraction V) and filter sterilized)]. The buffer was then poured off and replaced with fresh TMAC hybridization buffer (2.5 mL per blot) along with 2.1 μL of sonicated salmon sperm (10 mg/mL) and rotated in the oven for 10 min at 53°C. The radiolabeled ASO probe solution is then added directly to the hybridization bottle and rotated in the oven at 53°C for 2–3 h. Afterwards, the blots were washed 3 times over 20 min with 2.5 mL per blot of TMAC wash buffer (3 M TMAC, 0.6% SDS, 10 mM NaPO₄ pH 6.8, 1 mM EDTA) at 56°C while rotating in the oven. A final wash of 4 mL per blot of TMAC wash buffer was done at 56°C for 15 min while rotating. The blots were then rinsed twice with 2× SSC. The blots were exposed to a phosphor imager screen overnight after blotting off excess liquid and wrapping the blots in plastic. Blots were stripped of the radiolabeled ASO probe by washing with boiled 0.1% SDS until the signal is gone after scanning with a Geiger counter. All volumes of liquids used increase based on the number of blots used in each hybridization bottle with a maximum recommended as 6.

Cytological analysis

For immunofluorescence experiments, all animals were BxD, and mutants and controls within an experimental set were analyzed by 1 individual to reduce inter-experimentalist variation. Only on-axis (SYCP3-associated) foci were counted.

For staging spermatocytes, cells were stained with SYCP3, SYCP1, and gammaH2AX. Leptonema cells had SYCP3, copious amounts of gammaH2AX, and little to no SYCP1. Zygonema cells had stringy SYCP3, some SYCP1, and gammaH2AX. Pachynema cells had complete SYCP3, complete SCYP1, and gammaH2AX only on the sex body (WT). Pachynema-like cells had condensed SYCP3 staining, the presence of SYCP1 (more than zygonema), and a decrease in gammaH2AX staining. diplonema cells had complete SYCP3, SYCP1 beginning to disappear from the ends of chromosomes, and gammaH2AX only on the sex body.

For RPA2 foci, SYCP3 staining was used to stage cells, and cells in mid late zygonema (∼50% of axes containing continuous SYCP3 staining) and pachynema or pachynema-like were analyzed.

For whole chromosome paints, SYCP3 and SYCP1 were used to stage cells into leptonema, zygonema, and pachynema/pachynema-like. Immunofluorescent images were taken first (see above), and the placement of cells was marked using the z-stack function in Zeiss. After re-staining with the whole chromosome paints, the same cells were imaged and overlayed on the first set using Adobe Photoshop.

For the pairing analysis, cells with fully synapsed homologs (full SYCP1 and chromosome paint) were classified as paired. Any cells that were partially synapsed and/or the chromosome paints were not aligned or together were classified as not paired. Only pachynema/pachynema-like cells were analyzed.

For the chromosome condensation and dynamics analysis, all stages collected were analyzed (leptonema, zygonema, and pachynema/pachynema-like). First, the DAPI layer was measured (area of the DAPI stain) and then each chromosome paint signal was measured separately (area of each chromosome paint) using ImageJ and the separate channels function. The percentage of the area of each chromosome paint over the DAPI stain was then calculated.

Experimental design, quantification, and statistical analysis

WT and mutants were analyzed together simultaneously to reduce inter-experimental variation for the cytological analyses and the recombination assay. All data were shown as means with standard deviation or standard error of the mean, and 95% confidence intervals as shown in figures and table. The P-value was shown when the difference was significant and was adjusted for multiple comparisons as appropriate. The tests performed were included in text and figure legends along with the number of animals analyzed as capital “N” and the number of quantified data points as small “n.” At least 3 animals were analyzed per genotype.

Caveats to this study

Recombination analysis is limited to 2 loci, the A3 and 59.5 hotspots, which share a high frequency of meiotic DSBs and a high density of polymorphisms, allowing the detection of meiotic recombinants. It should be noted that behavior at these hotspots may not reflect genome-wide recombination patterns. Where possible, we compared locus-specific analysis to global chromosome behavior using cytology, but some features, such as NCOs and nascent D-loops, are indistinguishable cytologically. Our model that MutSgamma stabilizes D-loops is inferred based upon the end products of recombination in mutants and confirmation will depend upon analysis in wildtype upon development of technologies to detect recombination precursors. Importantly, our assay cannot detect nonallelic homologous recombination which could comprise a significant fraction of events in the mutants described here.

Data availability

The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables.

Supplemental material available at GENETICS online.

Funding

This work was supported by National Institutes of Health grants R01HD098129, DP2HD087943, and R01GM148028; Cancer Prevention and Research Institute of Texas grant R1213, and the Andrew Sabin Family Foundation Fellowship (FC). We acknowledge a summer undergraduate National Institutes of Health fellowship R25CA181004 (JL) and H.E.B Fellowship (MF), and Cancer Prevention and Research Institute of Texas training grant RP170067 (RK). We acknowledge the National Institutes of Health CA16672 for support of the Research Animal Support Facility Smithville and the Cancer Prevention and Research Institute of Texas RP170628 for support of the Flow Cytometry and Cellular Imaging Core.

Author contributions

MF, LP, RK, PC, AP, and JL conducted experiments; MF and FC performed statistical analyses; and MF, LP, RK, and FC designed the experiments and analyzed the data. MF and FC wrote the paper.