Controlling meiotic double-strand break formation in mice: a web of multivalent protein-protein interactions

Abstract

A complex network of protein interactions promotes the cooperative assembly of biomolecular condensates that control the formation of meiotic DNA double-strand breaks (DSBs). In many species, programmed DSBs made by SPO11 are required for the faithful pairing and segregation of homologous chromosomes during meiosis I. DSBs must be tightly regulated to support their intended function but also to minimize the risk of genomic damage from mutation or defective DSB repair. Multiple chromosome axis-associated proteins are important for the control of meiotic DSB formation and recombination, but the underlying mechanisms of those proteins remain poorly understood. Here, we present a current view of how these proteins work together in mice, based on recent advances in the characterization of their function and structure.

Introduction

Meiosis is a specialized cell division in sexually reproducing organisms in which one round of chromosome replication is followed by two consecutive rounds of cell division (Figure 1A) [1]. Homologous chromosomes are segregated during the first division, meiosis I (MI), and sister chromatids are segregated during the second division, meiosis II (MII) [1]. The proper pairing and segregation of homologous chromosomes in meiosis I requires the generation of programmed DNA double-strand breaks (DSBs) followed by repair through homologous recombination [2, 3]. In this process, the broken chromosome uses the intact homologous chromosome as a template for repair, a mechanism that brings the homologs into close juxtaposition [1–3]. The resulting reciprocal recombination products (crossovers) contribute to temporary physical linkages between homologous chromosomes that allow alignment of chromosome pairs along the meiotic spindle (Figure 1A,B) [1, 4, 5].

Figure 1. Overview of meiosis and homologous recombination.

A. Schematic of generation of haploid gametes from a diploid cell through meiosis. B. A simplified view of key steps during homologous recombination in meiosis. DSBs are generated with SPO11 covalently bound to 5′ ends, then the ends are processed by a combination of nucleases that yield long 3′ single-stranded DNA tails and release SPO11 still bound to a short oligonucleotide. The ssDNA is used by strand exchange proteins to search for a homologous DNA to use as a template for repair. Regulated formation of branched recombination intermediates results in either a double Holliday junction, which is usually resolved to form a crossover (left branch of pathway), or completion as a noncrossover (right branch). C. Chromosome organization during meiotic prophase I is illustrated with a pair of homologous chromosomes. In leptonema, which occurs after DNA replication, each homolog comprises a pair of sister chromatids that begin to form short stretches of axis. During zygonema, the axes elongate and homologous chromosomes begin to synapse with one another as their axes become juxtaposed and are held together by the SC central region material. At pachynema, chromosomes are fully synapsed. D. An example of surface-spread chromosomes from a mouse pachytene spermatocyte, stained for axis protein SYCP3 and central region transverse filament protein SYCP1. The 19 autosomes are fully synapsed: the overlapping SYCP1 and SYCP3 signals, which are not resolved by conventional light microscopy, appear as white lines. The chromatin in the loops is not stained. The inset at right shows a magnified view of the partially synapsed X and Y chromosomes, which have homology only at the PAR. Image courtesy of Tao Li (Keeney lab, MSK).

These events occur during prophase I, which is divided into substages based on cytological descriptions of chromosome organization [1, 6, 7] (Figure 1C). In leptonema in mouse spermatocytes, chromosomes begin to assemble linear proteinaceous structures (chromosome axes) from which the chromatin of sister chromatids emanates in loops. DSBs start forming at this stage. During zygonema, the axes elongate and homologous chromosomes begin to pair and form the tripartite synaptonemal complex (SC), comprising the axis of each homolog sandwiching the SC central region spanned by transverse filament proteins. Most DSB formation is completed by this stage, although it may continue in certain genomic regions or if synapsis is delayed or defective. Pachynema is the stage when chromosomes are fully synapsed from end to end. Completion of recombination to form crossovers occurs during this stage. The SC then disassembles at diplonema as chromosomes further prepare for segregation at MI.

Although DSBs are required for faithful segregation of homologous chromosomes, they could be potentially lethal to cells because errors in DSB repair may lead to aneuploidy, mutation, apoptosis, and/or infertility [8–10]. Recent advances in understanding the regulation of meiotic DSB formation in Saccharomyces cerevisiae were comprehensively reviewed [11]. Here, we discuss the multivalent protein-protein and protein-DNA interactions governing meiotic DSB formation in mice, examining how these interaction networks drive the cooperative assembly of condensates along chromosome axes and confer emergent properties essential for spatiotemporal control of DSB formation. We will also draw on studies in other species, particularly S. cerevisiae, to provide context for the mouse system. In some cases, mechanistic details are known only in yeast. In other cases, findings first made in yeast were later confirmed in mice, although often with less molecular or mechanistic detail. Emphasis on mouse leverages recent advances and allows exploration of both the conserved core machinery and features that are absent in other taxa.

One such mammal-specific feature is the pseudoautosomal region (PAR) on the X and Y chromosomes (Figure 1D). In contrast to autosomes, which are homologous along their entire length, sex chromosomes in males of most eutherian mammals share only a small homologous region, the PAR, which is around 700 kb long in mice [12–14]. DSB formation in the PAR is challenging. Genome wide, a DSB forms every ~10 Mb on average, whereas at least one DSB, interhomolog recombination event, and crossover must form in the <1 Mb PAR to ensure faithful segregation of sex chromosomes [15]. Failure to meet this requirement frequently leads to spermatocyte apoptosis and infertility, or to X-Y mis-segregation that produces aneuploid sperm that lead to infertility and, in humans, to Klinefelter syndrome [15, 16]. X-Y segregation succeeds in the vast majority of spermatocytes, indicating that male germ cells have evolved strategies to overcome the inherent challenges of PAR DSB formation and recombination, including the formation of DSBs at a frequency per Mb that is much higher than on autosomes [17, 18]. The multiprotein networks we will discuss here are at the heart of these adaptive strategies.

Another important difference between yeast and mice concerns the mechanism of targeting of DSB formation to specific genomic locations. In both species, DSBs form preferentially in small regions (typically under 250 bp wide) called hotspots [19]. In yeast, most hotspots correspond to gene promoters [20, 21], but in many vertebrates including most mammals, hotspots instead are defined by the PRDM9 protein [22, 23]. PRDM9 contains an array of zinc fingers that confers sequence-specific DNA binding along with a methyltransferase domain that methylates histone H3 on lysines 4 and 36 [24]. DSBs occur preferentially at PRDM9 binding sites normally, but default to gene promoters and other functional elements enriched for histone H3 lysine 4 methylation when PRDM9 is mutated [25]. Histone methylation is essential for PRDM9 activity [26] and the chromatin remodeler HELLS is an essential partner of PRDM9 for DSB targeting [27, 28], but precisely how site-specific modification of chromatin by PRDM9/HELLS targets SPO11 activity remains incompletely understood. The multiprotein networks described below are important contributors [29].

Meiotic DSB formation: Foundational principles

The understanding of meiotic DSB formation derives largely from studies in budding yeast, where the evolutionarily conserved topoisomerase relative Spo11 and nine accessory proteins collaborate to form DSBs [3, 11]. These proteins can be divided into three groups based on their physical interactions and molecular functions, namely Spo11–Ski8–Rec102–Rec104 (the Spo11 core complex that carries the catalytic activity of DNA cleavage), Rec114–Mei4–Mer2 (accessory factors for DSB formation), and the Mre11–Rad50–Xrs2 complex (MRX, which also functions in DNA end processing and regulatory responses to DSBs) (Figure 2A). All of these proteins are essential for DSB formation in yeast. The Spo11 core complex and Rec114-Mei4-Mer2 are conserved in most species where they have been tested, including fungi, plants, and mammals [3, 30]. By contrast, the only other species where MRX is thus far known to be needed for DSB formation is Caenorhabditis elegans (which requires MRE-11 and RAD-50 [31, 32]); MRX is dispensable for DSBs in most species tested (Schizosaccharomyces pombe, Tetrahymena thermophila, Coprinopsis cinerea, Drosophila melanogaster, Arabidopsis thaliana, and likely mouse) [3, 30, 33, 34].

Figure 2. Proteins important for DSB formation and the loop-axis structure of meiotic chromosomes.

A. DSB proteins in yeast and mice. In S. cerevisiae, three subgroups (Spo11–Ski8–Rec102–Rec104; Rec114–Mei4–Mer2; and Mre11–Rad50–Xrs2) are essential for DSB formation. In M. musculus, REC114, MEI4, IHO1 (Mer2 ortholog), and MEI1 are similarly required. ANKRD31 is a vertebrate-specific factor that supports normal DSB number and location but is not strictly required for DSBs to form. The MRE11–RAD50–NBS1 complex appears to be dispensable for DSB formation in mice. Corresponding yeast and mouse proteins are shown in the same color to indicate homology. Proteins shown in white are either not found in the other species (Rec104, MEI1, ANKRD31) or are present in mice but apparently not required for DSB formation (Ski8, MRX complex). B. Tethered loop-axis complex model. Left panel: A chromatin loop extending from the linear proteinaceous axis, with part of the DSB-forming machinery assembling on the axis and PRDM9 marking a hotspot on the loop. No DSB formation has occurred at this stage. Right panel: The DSB factors assembled on the axis capture the hotspot and tether it to the axis. This capture enables axis-associated DSB formation by SPO11 within the loops. The two panels illustrate the transition from an untethered to a tethered state.

Studies in mice have identified corresponding DSB factors, with noteworthy differences (Figure 2A). Mouse SPO11 interacts with TOP6BL (structurally similar to yeast Rec102) [35–39] and DSB formation further requires REC114 and MEI4 (orthologs of the S. cerevisiae proteins of the same names) plus IHO1 (Mer2 ortholog), which assemble on chromosome axes [35, 40–43]. Rec104 is thus far only found in the yeast family Saccharomycetaceae [36], while Ski8 orthologs are dispensable for DSB formation in A. thaliana [44] and have not been reported to be required for DSB formation outside of fungi [3]. In mice, MEI1 is also essential for DSB formation [45]. MEI1 homologs are found in other vertebrates and in plants (exemplified by A. thaliana PRD1), but are currently unidentified in fungi or other lineages [45, 46]. An additional relevant factor is vertebrate-specific ANKRD31, which was identified as a direct REC114 interactor that, although not essential for DSB formation, serves as a linchpin in regulating the number, timing, and locations of DSBs [47, 48].

Meiotic DSB formation and recombination are associated with the higher order organization of chromosomes into chromatin loops that are anchored along a linear proteinaceous axis (Figure 2B) [21, 49–55]. In yeast, the axis includes cohesin complexes (including those containing the meiosis-specific kleisin subunit Rec8), the HORMA-domain protein Hop1, and coiled-coil protein Red1 [56–58]. In mice, axes similarly contain cohesin complexes (including those with meiosis-specific kleisins REC8 and RAD21L), Red1 homologs SYCP2 and SYCP3, and Hop1 homologs HORMAD1 and HORMAD2 [59–62]. In yeast, DSBs mostly take place within chromatin loops, rather than on DNA that is defined as axis-associated based on the preferred binding of Rec8 or other axis proteins in chromatin immunoprecipitation experiments [21, 52–55]. However, cytological markers of recombination locate on the axes in most species, suggesting that recombination occurs in the context of the axes [53, 54, 63, 64]. The “tethered loop-axis complex” (TLAC) model resolves this apparent disparity in the locations of DSBs [51, 53, 54]. This model proposes that pro-DSB factors assemble on axes and then capture DNA segments that are in chromatin loops, thereby tethering potential DSB sites to the axis prior to DSB formation (Figure 2B).

Recent studies (described in detail below) have revealed that these pro-DSB factors assemble in vitro—and likely in vivo—as biomolecular nucleoprotein condensates. Such condensates are membrane-less structures that concentrate specific proteins and nucleic acids through a process sometimes but not always driven by phase separation [65–68]. These structures form when multivalent interactions reach a critical threshold concentration, leading them to spontaneously demix from the surrounding dilute phase into a dense phase [67, 68]. Many condensates exhibit characteristic properties including dynamic internal rearrangement, reversibility, and the ability to fuse [65–69]. In the context of meiosis, nucleoprotein condensate formation is proposed to be a key organizing principle for assembling the DSB machinery, although the precise biophysical nature of these condensates remains poorly understood [69–72].

The molecular mechanisms proposed to underlie chromatin loop tethering are evolutionarily conserved. In yeast in early prophase I, Rec114, Mei4, and Mer2 form foci that interact with axis components and with the histone H3 lysine 4-methyl binding protein Spp1; simultaneous binding to both the axis and to histone modifications that are enriched at the gene promoters where DSBs most often form may provide a loop tethering function [73–75]. In vitro, multivalent interactions drive the co-condensation of Rec114, Mei4 and Mer2 with DNA into nucleoprotein clusters that can recruit Spo11 complexes [69]. Mouse REC114, MEI4, and IHO1 may form analogous condensates, as inferred from their ability to form foci on chromosomes [29, 41, 43, 70], with Spp1 homolog CXXC1 proposed to mediate loop-axis connectivity by bridging between IHO1 and the histone methylation marks placed by the PRDM9 methyltransferase to define DSB hotspots [76].

Assembly via condensate formation is dramatically accentuated in the mouse PAR, which becomes highly enriched for pro-DSB proteins REC114, MEI4, MEI1, ANKRD31 and IHO1 (hereafter abbreviated as RMMAI) from leptonema to late zygonema [77]. These proteins coalesce into “blobs” that are larger and brighter than smaller foci on autosomes as visualized by immunofluorescence microscopy, and these structures are essential for ensuring DSB formation within the small PAR [77].

Mouse meiotic DSB proteins and their interactions

SPO11–TOP6BL: the catalytic core of meiotic DSB formation.

Topoisomerase VI is a type IIB topoisomerase first identified in archaea [78]. It is an A₂B₂ heterotetramer, with the A subunit performing DNA cleavage while the B subunit undergoes ATP-dependent conformational changes that govern the topoisomerase catalytic cycle [79–82]. SPO11, a eukaryotic homolog of Top6A, serves as the catalytic subunit of the meiotic DNA cleavage activity [78, 83–87]. A potential homolog of the Top6B ATPase subunit had remained unknown until a forward genetic screen in A. thaliana identified recombination-defective mutations in the mTopVIB gene, leading in turn to homology-based identification of TOP6BL protein in mice [35, 40]. The “transducer” domain of TOP6BL interacts directly with an N-terminal segment in SPO11 [35, 36, 40, 88]. Although TOP6BL retains a GHKL ATPase-homologous domain (so named because of its presence in DNA Gyrase, Hsp90, histidine kinase, and MutL [89]), key residues important for ATP binding and hydrolysis are missing and ATP does not detectably affect TOP6BL properties in vitro, so there is currently no evidence for a role of ATP [37, 87, 90–92]. In yeast, the Top6B homolog is Rec102, which (like homologs in S. pombe and D. melanogaster) lacks a GHKL domain entirely [35, 36, 40]. Perhaps ATP’s modulatory function has been replaced by interactions with other factors, such as REC114 [91].

Studies in mice have revealed additional complexity for mammalian SPO11. There are two major isoforms of SPO11 generated by alternative splicing, namely SPO11α (exon 2 skipped) and SPO11β (exon 2 included) [86, 93]. The sequence absent from SPO11α includes the TOP6BL-interaction surface. This interaction surface was originally inferred from crystal structures of the Top6A–Top6B interface [81] and was supported more recently by cryo-EM structures of yeast Spo11 core complexes and AlphaFold predictions of mouse complexes [36, 37, 39]. Accordingly, only SPO11β is competent to stably bind to TOP6BL [35, 91]. SPO11β is the predominant isoform expressed in early spermatocytes, and transgenic mice expressing only this isoform display nearly normal DSB levels genome wide, but have a marked defect in forming DSBs in the PAR [17, 85, 93–96]. SPO11α, in contrast, is present in low amounts in early prophase I and becomes prevalent only after early pachynema [95]. Transgenic mice expressing only SPO11α are similar to Spo11 null mutants (no detectable DSBs), but concomitant expression of both SPO11β and SPO11α rescues the PAR DSB defect of SPO11β-only mice [96]. It is currently unclear how SPO11α performs this role in the PAR.

Given the structural similarity of SPO11 to the Top6A subunit of archaeal TopoVI, which functions as a dimer, SPO11 is predicted to form a similar dimer in which a tyrosine (Y139 in mouse SPO11) of each monomer breaks one of the two DNA strands to generate a DSB [97]. This inference is supported by recent AlphaFold predictions of the mouse SPO11 dimer [37–39]. However, biochemical and structural studies by cryo-EM with purified yeast Spo11 core complexes revealed a 1:1:1:1 stoichiometry (Spo11:Ski8:Rec102:Rec104) with no reported catalytic activity in vitro [36, 69]. It has therefore been argued that Spo11 dimerization could be an important regulator of DSB formation [11, 36], consistent with earlier suggestions based on Spo11-Spo11 co-immunoprecipitation experiments in yeast [98].

The model that SPO11 catalytic activity depends on regulated dimerization has now been supported by the reconstitution of SPO11-dependent DNA cleavage in vitro with purified mouse SPO11–TOP6BL complexes, which have a 1:1 stoichiometry in solution [37–39]. Evidence that catalytic activity is strictly dependent on SPO11 dimerization includes the findings that cleavage involves formation of hybrid active sites from two SPO11 monomers (Figure 3) [37, 39], and that artificially dimerizing SPO11 promotes DNA cleavage [37]. It was proposed that the weak intrinsic dimerization exhibited by SPO11 coupled with its high-affinity DNA binding creates a “kinetic trap” that favors dispersed binding of inactive monomers to the chromatin over formation of catalytically competent dimers [37, 39]. Consequently, SPO11 is proposed to require accessory factors to promote its dimerization at the right sites and times, a dependency that intrinsically limits the risk of closely spaced DSBs known to cause deletions and other deleterious genomic rearrangements [37, 39, 99, 100]. This arrangement provides a fundamental mechanism for the control of DSB formation via multivalent interactions. Surprisingly, SPO11 alone, without TOP6BL, is capable of DSB formation in vitro [39], so the specific function(s) of TOP6BL that makes it essential for SPO11 activity in vivo remains unknown.

Figure 3. SPO11 dimers form hybrid active sites for DNA cleavage.

Wild-type SPO11 dimers generate DNA double-strand breaks through hybrid active sites assembled from residues contributed by both SPO11 monomers. Mutant A lacks the catalytic tyrosine (Y138 in mouse), whereas Mutant B lacks a residue important for binding a pair of divalent metal ions needed for strand cleavage (E224). Neither mutant cleaves DNA on its own. However, mixing the two mutant proteins allows the formation of heterodimers that retain catalytic activity through complementation in trans. These heterodimers produce single-stranded DNA nicks because they contain one fully functional active site.

REC114, MEI4, and IHO1: a cooperative scaffold for DSB machinery assembly.

In yeast, Rec114, Mei4, and Mer2 form a functional subgroup that have correlated chromatin immunoprecipitation (ChIP) patterns and that partially colocalize on chromosomes in foci in leptonema before DSB formation [53, 73, 101, 102]. Pairwise interactions among these proteins were demonstrated by yeast-two-hybrid and co-immunoprecipitation assays in both budding and fission yeasts [73, 101–104]. Biochemical studies with purified recombinant proteins further revealed that they form two separate complexes in vitro, namely, Rec114–Mei4 (a 2:1 heterotrimer) and Mer2 (a homotetramer) [69, 72, 105–107]. AlphaFold models confirmed by NMR spectroscopy data established the structure of the Rec114–Mei4 trimer interface, in which the Rec114 C termini form intertwined alpha helices that embrace an alpha helix from the Mei4 N terminus [72, 106].

Rec114, Mei4, and Mer2 undergo multivalent interactions that drive their co-condensation into assemblies containing hundreds or even thousands of proteins in the presence of plasmid DNA [69, 72, 106]. Rec114–Mei4 trimers and Mer2 homotetramers can each bind DNA and form condensates separately, but they assemble nearly stoichiometric mixed condensates if incubated with one another along with DNA [69]. These condensates may correspond to the cytologically visible foci observed in vivo [41, 73, 102, 108], because mutations that impair condensate formation in vitro also reduce chromatin associated foci in vivo and severely compromise DSB formation [69]. Surprisingly, purified recombinant protein fragments of Rec114 and Mei4 encompassing just the trimerization domain are sufficient in vitro to form nucleoprotein condensates (albeit less efficiently than the full-length proteins) and can bundle DNA helices side-by-side in single-molecule optical tweezers experiments [72]. However, the detailed internal organization of the condensates formed by Rec114, Mei4, and Mer2 proteins remains as important questions for future investigation.

The corresponding mouse proteins display both conservation and vertebrate-specific elaboration. Mouse REC114, MEI4, and IHO1 are all essential for DSB formation, interact with one another, at least partially colocalize on the axes of meiotic chromosomes before DSB formation, and are interdependent for chromosome association [29, 41, 43, 108, 109]. MEI1 also appears to be part of this subgroup on the basis of its localization and physical and functional interactions, but it is currently less well characterized [42, 45, 77].

The mouse REC114–MEI4 interface is predicted by AlphaFold to be similar to the interface between their yeast orthologs [72, 106, 110]. Mouse IHO1, like yeast Mer2, forms homotetramers through a central helical region that assembles a four-stranded parallel coiled-coil, as deduced from AlphaFold models supported by hydrodynamic analysis and small-angle x-ray scattering [69, 105, 106, 110]. IHO1 also interacts with REC114 (as shown by yeast two-hybrid analysis, co-purification, isothermal titration calorimetry, and AlphaFold modeling), and the N terminus of IHO1 is sufficient for the interaction with the REC114 N-terminal pleckstrin homology (PH) domain (Figure 4) [43, 110].

Figure 4. Protein-protein interaction map.

Protein domains and interacting segments (defined in most cases by direct structure determination or yeast two-hybrid assays) are shown in color. Proteins are drawn approximately to scale. The molecular details of some interactions remain to be fully determined.

Consistent with the condensation behavior observed in yeast, REC114 and MEI4 form discrete foci on mouse chromosome axes, and IHO1 also forms foci but, unlike REC114 and MEI4, localizes more extensively along the axes in a more continuous, thread-like distribution [29, 41, 43, 70, 108]. It remains to be determined whether the mouse proteins do indeed form nucleoprotein condensates in vitro, but it appears likely that REC114–MEI4 trimers and IHO1 tetramers work together to form a cooperative scaffold that drives assembly of higher order axis-associated complexes [11, 37, 39, 69, 70, 99]. This scaffold may serve to enhance interaction specificity amongst its components while concentrating client proteins such as TOP6BL and SPO11, thereby promoting efficient and accurate DSB formation at specific genomic regions [11, 29, 37, 39, 69, 70, 99].

TOP6BL–REC114: coupling the catalytic core to condensates.

Since REC114, MEI4, and IHO1 are essential for SPO11 catalytic activity in vivo, they must be functionally coupled with the catalytic SPO11–TOP6BL complex. Such coupling is mediated by direct physical interactions. For example, the yeast Spo11 core complex can be recruited to Rec114–Mei4–Mer2 nucleoprotein condensates in vitro through two types of interaction: one depends on Mer2 and the other involves interactions between the PH domain of Rec114 and both Rec102 and Rec104 [69].

In mice, a critical link connecting SPO11–TOP6BL complexes with REC114, MEI4, and IHO1 is established by direct interaction between the PH domain of REC114 and the C terminus of TOP6BL (Figure 4), as demonstrated by yeast-two-hybrid assays and in vitro binding experiments with recombinant proteins [92]. An x-ray crystal structure revealed a conserved interaction interface centered on TOP6BL W562 and REC114 L104, and point mutations at these residues abolished the interaction in vitro [92]. The functional significance of this interaction was demonstrated by mouse models with Top6bl^W562A/W562A point mutation and Top6bl^{Δ17Ct/Δ17Ct} truncation, in which the interaction is disrupted without significantly altering the structure of TOP6BL [92]. These mutants show reduced DSB activity in both oocytes and spermatocytes, with a particularly marked decrease in sub-telomeric regions in spermatocytes. Therefore, the TOP6BL–REC114 interaction is an important node that connects two meiotic DSB complexes and ensures the correct spatiotemporal efficiency of DSB formation.

ANKRD31–REC114: a tunable regulator of DSB number, timing and location.

The evolutionary complexity of meiotic DSB control is exemplified by ANKRD31, a vertebrate specific protein that controls DSB timing, number, and location [47, 48]. ANKRD31 is essential for targeting DSBs to the PARs of sex chromosomes, and is partially required for PRDM9-dependent targeting of DSBs across the rest of the genome. These two functions are inferred from the fact that Ankrd31 null mutants have greatly reduced DSBs on the PAR, and they form DSBs in the rest of the genome at both PRDM9-directed hotspots and so-called “default” hotspots, which are transcription promoters and other functional elements that are targeted by SPO11 when PRDM9 is not active [25, 47, 48]. Roles in the PAR are discussed further later; why ANKRD31 is needed for normal PRDM9 hotspot activity is currently unknown.

ANKRD31 serves as a central hub that interacts with REC114, IHO1, and MEI1, as shown by yeast-two-hybrid assays, GST-pull down assays, and immunoprecipitation plus mass spectrometry (Figure 4) [47, 48, 70, 77]. The interaction with REC114 is currently the best understood. The C-terminal ~50 amino acids of ANKRD31 are necessary and sufficient for interaction with the PH domain of REC114, forming a stable heterodimer whose structure was determined by x-ray crystallography (Figure 4) [47, 48]. An ANKRD31 C-terminal truncation that completely eliminates the REC114 interaction mimics an Ankrd31 null, with delayed DSB formation and recombination, defects in DSB repair, and altered DSB location including frequent use of default hotspots and failure to target DSBs to the PARs [47, 48, 111]. However, a substantial but not complete disruption of the interaction (via a substitution of alanine for ANKRD31 glutamine 1831) presents a more subtle phenotype with delayed DSB formation globally, but little if any other defect in recombination or DSB locations [111]. Additionally, this missense Ankrd31 allele shows a dosage effect, wherein combining it with the null or C-terminal truncation allele results in intermediate phenotypes [111]. These findings demonstrate that the ANKRD31–REC114 interaction is not a simple on/off switch but a tunable complex, ensuring the correct number, timing and location of DSB formation and recombination.

Given its lack of predicted enzymatic domains and the multiple protein-protein interactions that ANKRD31 engages in, it has been hypothesized to serve as a scaffold to help bring its binding partners together [47, 48, 111]. It remains unclear whether it has other functions beyond this scaffolding role. Additional interacting proteins (e.g., ZMYM3 and PTIP) have been detected by immunoprecipitation plus mass spectrometry from testis extracts [77]. Further work is needed to establish the functions of these proteins and to determine whether the full catalog of ANKRD31-binding partners is known.

HORMAD1–IHO1: seeding the DSB machinery on chromosome axes.

As introduced above, spatial control of DSB formation along chromosomes involves interactions between the axis and the DSB machinery. HORMAD1, a key axis component, directly interacts with IHO1 (Figure 4) and stabilizes MEI4 focus formation on unsynapsed axes [41–43, 70]. The cytologically detected chromosomal loading of IHO1 is strongly dependent on HORMAD1, but independent of MEI4 and REC114 [41, 43]. These dependencies suggest that HORMAD1 recruits IHO1 to unsynapsed axes (“seeding”), which then drives the assembly of MEI4 and REC114 and promotes efficient DSB formation [41, 43]. Interestingly, however, IHO1 binding to DSB hotspots as detected by ChIP does not require HORMAD1 [29].

The mechanistic basis of this critical seeding function has been further elucidated by demonstration that the last 7 amino acids of IHO1 are required but not sufficient for the interaction with the HORMA domain of HORMAD1, as shown by yeast two-hybrid assays of wild-type proteins validated by point mutations that disrupt the interaction [70]. Disruption of the HORMAD1–IHO1 interaction by removing these terminal residues in vivo results in loss of HORMAD1–IHO1 coimmunoprecipitation from testis extracts and inefficient seeding of the DSB machinery on chromosome axes, leading in turn to reduced DSB formation and impaired homologous chromosome pairing [70]. The IHO1 C terminus is also phosphorylated in vivo, likely by DBF4-dependent CDC7 kinase (DDK), and this phosphorylation promotes interaction with HORMAD1 [70].

Furthermore, the axis-based recruitment of the DSB machinery is integrated with another regulatory layer to achieve robust DSB formation control. Genetic and cytological evidence shows that ANKRD31 and the HORMAD1–IHO1 interaction synergistically promote the assembly of pro-DSB factors, including ANKRD31, REC114, and MEI4 on axes [70]. The biochemical and dynamic details of the pertinent interactions remain to be determined. Nonetheless, a refined model emerges in which the HORMAD1–IHO1 interaction specifically enables the axial seeding, while ANKRD31 supports both the seeding and the subsequent growth of higher-order pro-DSB factor clustering, likely through interactions with REC114 and MEI1 [70]. This cooperative mechanism might ensure that assembly of the DSB machinery is not only efficiently initiated, but also appropriately amplified within the correct chromosome context, highlighting the multi-layered control of DSB formation.

TOP6BL, IHO1 and ANKRD31 bind to REC114 mutually exclusively.

Another intriguing feature of the assembly of the DSB machinery is its potential regulation through competitive protein interactions, which may facilitate the dynamic assembly and/or functional switching at recombination sites. TOP6BL, IHO1, and ANKRD31 all bind to the same PH domain of REC114, with biophysical measurements indicating that these interactions occur with micromolar affinities. Structural models, including AlphaFold predictions and crystallographic structures, show that all three partners engage overlapping surfaces on the REC114 PH domain (Figure 4) [47, 48, 92].

This competition is supported by multiple lines of biochemical evidence [47, 92, 110]: 1) REC114 L104D mutation reduces the binding to TOP6BL, IHO1, and ANKRD31; 2) the ANKRD31 W1842A and the corresponding TOP6BL W562A mutations each disrupt the interaction with REC114; 3) no binding between TOP6BL and REC114 is observed when REC114 is pre-saturated with ANKRD31 in vitro; and 4) gel filtration elution profiles confirm that the binding of IHO1 to REC114 is mutually exclusive to that of ANKRD31 or TOP6BL.

A key unresolved question is how these mutually exclusive interactions are orchestrated in vivo to regulate DSB formation. One model proposes that ANKRD31 might act early and interact with REC114 transiently, and then it might detach from REC114, making it available for interacting with TOP6BL for DSB formation [110]. However, this model does not specifically take into account the observation that pro-DSB factors form large condensates, and that ANKRD31 promotes both the seeding and growth of these clusters [70, 112]. These observations can be unified under an alternative model in which the presence of multiple REC114 molecules within a condensate allows concurrent recruitment of TOP6BL, IHO1 and ANKRD31 even though individual REC114 molecules can bind only one partner at a time. This model also provides a scenario in which these components could be synergistically stabilized through mutual interactions.

This paradigm would enable dynamic assembly of the DSB machinery. Through this mechanism, the DSB machinery could maintain flexibility in partner selection during initial assembly while ensuring the formation of stable and functional complexes for controlling DSB formation.

Unique interactions between meiotic DSB proteins on the pseudoautosomal region

Accompanying its higher-than-average DSB frequency, the mouse PAR undergoes multiple specialized behaviors. PAR axes become unusually elongated, the axes of sister chromatids split apart to form a bubble from roughly the PAR boundary to near the telomere (the split is visible in super-resolution microscopy, but in conventional microscopy appears as axial thickening), and PAR chromatin forms loops that are much smaller than typical for autosomes or the non-PAR portions of the X and Y, as assessed by combined immunofluorescence and fluorescence in situ hybridization [17, 77]. The PAR is also highly enriched for pro-DSB proteins REC114, MEI4, MEI1, ANKRD31 and IHO1 (RMMAI) from leptonema to late zygonema, forming cytologically prominent “blobs” that are larger and brighter than the more numerous foci scattered across autosomes [77] (Figure 5). Presumably, the small foci are locations where the DSB machinery is poised to make breaks at hundreds of genomic locations, while the blobs are specialized structures that ensure efficient DSB formation on the PAR [29, 77].

Figure 5. RMMAI proteins form small autosomal foci and prominent mo2-associated blobs during early meiotic prophase.

A. Representative immunofluorescence images of mouse prophase spermatocyte chromosome spreads stained for SYCP3 (magenta) and either ANKRD31 or REC114 (green). Numerous small RMMAI foci are distributed along autosomal chromosome axes, whereas intensely stained foci (“blobs”) are observed at the PAR and other mo-2 regions. The inset shows a magnified view highlighting the size difference between the large mo-2-associated blob (arrowhead) and a smaller autosomal focus (arrow). Although only ANKRD31 and REC114 are shown here, previous studies demonstrated that REC114, MEI4, MEI1, ANKRD31 and IHO1 (RMMAI) extensively colocalize, particularly within the mo-2-associated blob [47, 48, 77, 111]. The micrographs are adapted from the authors’ prior work [111] under a Creative Commons Attribution License 4.0 (CC BY). The REC114-stained spread is presented in mirror image from the original for simplicity in panel composition. B. Schematic representation of RMMAI protein organization along chromosome axes. On autosomes, RMMAI proteins assemble into numerous small foci distributed along the chromosome axis. In contrast, the PAR contains a single large RMMAI-enriched blob. The composition and internal architecture of both autosomal foci and the PAR blob remain incompletely understood. Therefore, the enlarged views are intended only to illustrate relative enrichment and not the actual molecular arrangement of RMMAI proteins.

Several lines of evidence suggest that mo-2 minisatellite DNA sequences may be a cis-acting factor for RMMAI hyperaccumulation in the PAR [77]. These 31-bp tandem repeats compose about 22% (about 160 kb) of the entire PAR, and comparison of mouse strains with different mo-2 copy numbers showed that the number of mo-2 repeats correlates with the efficiency of ANKRD31 recruitment [77, 113–115]. Autosomally-located mo-2 minisatellite arrays also assemble RMMAI blobs, suggesting that presence of these repeats may be sufficient [77].

ANKRD31, MEI4 and REC114 are all essential trans-acting factors for mo-2-associated blob formation, axis elongation, and sister chromatid axis splitting [47, 48, 77]. Ankrd31 deficiency leads to a loss of RMMAI blobs and causes failure of X-Y pairing in > 90% of cells, whereas autosomes still pair most of the time [47, 48]. In contrast, although the absence of Hormad1 leads to decreased DSBs globally, it does not affect the recruitment of RMMAI, the remodeling of PAR axes, or the high frequency PAR DSB formation [77, 116].

Multivalent protein-protein interactions are likely to be key. For example, the ANKRD31–REC114 interaction is specifically required for mo-2-associated blob formation and X-Y pairing [111]. Both ANKRD31 and REC114 blobs are reduced in number and intensity in all the ANKRD31–REC114 interaction-deficient mutants tested, with the severity of the defect—and severity of X-Y pairing failure—correlating with the strength of interaction defect [111]. Therefore, ANKRD31 and REC114 are interdependent for the full formation of mo-2-associated blobs, and the interactions of ANKRD31 with its other partners cannot fully compensate for the loss or reduction of ANKRD31–REC114 interaction [47, 48, 77, 111]. These findings illustrate how selective interactions confer specificity to pro-DSB machinery assembly in particular genomic contexts.

Many mechanistic questions remain. First, it is unknown how the PAR is recognized. Is it the mo-2 repeats themselves, or as-yet undetermined mo-2-linked sequences instead or in addition? What protein(s) provides the specific recognition? Is the mo-2 DNA sequence recognized directly, or is specificity based on an unusual DNA structure or chromatin state? PAR enrichment of heterochromatin-related histone marks and proteins may point to involvement of chromatin [77], but functional tests of this have not been reported. Second, the biophysical basis of the unusual PAR ultrastructure is unknown. Are the cytological blobs just larger versions of the foci that support DSB formation across the genome, or are they some sort of “hyper condensate” in which exceptionally high concentrations of condensate-forming activities, and/or incorporation of PAR-specific factors, give distinct emergent properties? How does the accumulation of RMMAI proteins result in PAR chromatin condensation, formation of short chromatin loops arrayed on long axes, and sister axis splitting? And how do these changes ensure DSB formation? Third, it is not known whether the behaviors of the mouse PAR are evolutionarily conserved. Eutherian PARs evolve rapidly, and they are subject to both decay and replenishment by translocation of autosomal segments, fostered at least in part by the PAR’s high recombination activity [117]. It will be interesting to determine whether RMMAI protein enrichment and higher order chromosome structural dynamics are similar across species with very different PAR DNA sequences, or if different molecular strategies have evolved in different lineages to solve the X-Y pairing and recombination challenges.

Conclusions

Recent studies have unveiled perspectives into how multivalent protein-protein and protein-DNA interactions orchestrate the assembly of DSB-promoting condensates that have emergent properties essential for robust and precise regulation of meiotic DSB formation. Furthermore, the adaptability of this interaction network enables context-specific regulation, as exemplified by the distinct hyperaccumulation of RMMAI in the PAR. The biophysical properties of these condensates remain largely unexplored, including their internal organization and hierarchy of protein recruitment. How the mutually exclusive interactions among REC114, TOP6BL, IHO1 and ANKRD31 are spatially coordinated within the condensates to achieve dynamic regulation is also unresolved. It is further unclear whether the weak dimerization of SPO11 is an evolutionarily tunable property that may differ across species, and how this property integrates with the clustering functions of the RMM scaffold to control DSB frequency and spacing.

Furthermore, control of mammalian DSB formation via posttranslational protein modifications is certain to be a critical area of research that is largely untapped at present. The yeast DSB-promoting proteins are subject to various posttranslational modifications, including essential activating phosphorylation of Mer2 by the cell cycle-regulating kinases CDK and DDK [101, 118, 119]; apparently inhibitory phosphorylation of Rec114 by Mec1 and/or Tel1 (DSB-responsive kinases related to mammalian ATR and ATM, respectively) [120]; phosphorylation of Rec104 of unknown function and by an unknown kinase(s) [121]; and essential activating sumoylation of Mer2 [122]. The molecular impacts of these yeast modifications on the DSB formation mechanism are poorly understood. In mouse, no defined DSB-modulating posttranslational modifications are yet known aside from the DDK-dependent phosphorylation of the IHO1 C terminus mentioned above [70]. However, ATM regulates SPO11 activity through a negative feedback mechanism in mice, similar to budding and fission yeasts and Drosophila [123–126], and inhibiting DDK appears to downregulate DSB formation independent of the IHO1 C terminus in mouse spermatocytes [70], so there is strong supporting evidence that phosphorylation (and perhaps other modifications) is a key regulator of DSBs in mammals.

Addressing these questions will require integrated approaches combining structural biology, biophysics, in vitro reconstitution with purified recombinant proteins, and in vivo functional assays. The cooperative assembly framework provides a foundation for future experiments aimed at a deeper understanding of the mechanism and regulation of DSB formation.

Abbreviations and protein names

ANKRD31

Ankyrin repeat domain-containing protein 31

ChIP

Chromatin immunoprecipitation

CXXC1

CXXC zinc finger protein 1

DSB

Double-strand break

GHKL

ATPase domain found in DNA gyrase, Hsp90, histidine kinase, and MutL

HELLS

Helicase, lymphoid specific

HORMA domain

Protein domain found in Hop1, Rev7, and Mad2

HORMAD1

HORMA domain-containing protein 1

IHO1

Interactor of HORMAD1 (Mer2 ortholog)

MEI1

Meiosis-defective 1

Mei4/MEI4

Meiosis-specific 4

Mer2

Meiotic recombination 2

MI

Meiosis I

MII

Meiosis II

Mre11/MRE11

Meiotic recombination 11

MRX

Mre11-Rad50-Xrs2 complex

PAR

Pseudoautosomal region

PH

Pleckstrin homology

PRDM9

PR domain containing 9

Rad50

Radiation sensitive 50

Rec102

Meiotic recombination 102

Rec104

Meiotic recombination 104

Rec114/REC114

Meiotic recombination 114

RMMAI

REC114, MEI4, MEI1, ANKRD31, and IHO1

Ski8

Superkiller 8

Spo11/SPO11

Sporulation 11

TLAC

Tethered loop-axis complex

Top6A

Topoisomerase VI subunit A

Top6B

Topoisomerase VI subunit B

TOP6BL

Topoisomerase VI subunit B-like

TopoVI

Topoisomerase VI

Xrs2

X-ray sensitive 2

Data Availability Statement

Data sharing is not applicable to this article as no new data were generated or analyzed for this study.