Impeding DNA Break Repair Enables Oocyte Quality Control

SUMMARY

Oocyte quality control culls eggs with defects in meiosis. In mouse, oocyte death can be triggered by defects in chromosome synapsis and recombination, which involve repair of DNA double-strand breaks (DSBs) between homologous chromosomes. We show that RNF212, a SUMO ligase required for crossing over, also mediates oocyte quality control. Both physiological apoptosis and wholesale oocyte elimination in meiotic mutants require RNF212. RNF212 sensitizes oocytes to DSB-induced apoptosis within a narrow window as chromosomes desynapse and cells transition into quiescence. Analysis of DNA damage during this transition implies that RNF212 impedes DSB repair. Consistently, RNF212 is required for HORMAD1, a negative regulator of inter-sister recombination, to associate with desynapsing chromosomes. We infer that oocytes impede repair of residual DSBs to retain a “memory” of meiotic defects that enables quality control processes. These results define the logic of oocyte quality control and suggest RNF212 variants may influence transmission of defective genomes.

Graphical Abstract

eTOC Blurb

High quality gametes (sperm and eggs) are paramount for successful reproduction. Defective gametes cause infertility and congenital disorders. Qiao et al. describe a counterintuitive process that helps developing eggs sense whether errors have occurred by impeding the repair of lingering chromosome breaks.

INTRODUCTION

Oocyte quality and number are important determinants of reproductive success (Kerr et al., 2013). These attributes are influenced by the selective elimination of oocytes that experience problems during the early stages of meiosis in a conserved process called fetal oocyte attrition (Bolcun-Filas et al., 2014; Cloutier et al., 2015; Di Giacomo et al., 2005; Hunter, 2017; Kerr et al., 2013; Kogo et al., 2012a; Malki et al., 2014; Rinaldi et al., 2017; Wojtasz et al., 2012). In mouse, epigenetic reprogramming during meiosis derepresses LINE-1 transposons triggering perinatal loss of at least half of all fetal oocytes (Malki et al., 2014). Defects in the chromosomal events of meiotic prophase also trigger oocyte loss, typically at a later stage than LINE-1 induced attrition, as oocytes transition into quiescence before developing into primordial follicles to establish the ovarian reserve (Di Giacomo et al., 2005; Hunter, 2017). This second wave of oocyte death is mediated by interrelated pathways that signal defective interactions between pairs of homologous chromosome, i.e. defects in DSB repair and/or homolog synapsis (Bolcun-Filas et al., 2014; Cloutier et al., 2015; Di Giacomo et al., 2005; Kogo et al., 2012a; Rinaldi et al., 2017; Wojtasz et al., 2012). Canonical DNA-damage response factors signal unrepaired DSBs via the ATR and CHK2 kinases to trigger p53/p63-mediated oocyte apoptosis. Defective synapsis can also trigger oocyte death by a distinct pathway that mediates transcriptional silencing in a process termed meiotic silencing of unpaired chromatin (MSUC). Together, these early and late oocyte attrition pathways balance the quality and size of the ovarian follicle reserve to maximize reproductive success. In this study, we address why and how defective interhomolog interactions trigger oocyte apoptosis after chromosomes desynapse during the transition into quiescence. We implicate a new factor, RNF212, in the pre-follicle oocyte apoptosis pathway and show that it functions in a counterintuitive process that helps oocytes gauge whether meiotic inter-homolog interactions were defective.

RESULTS

Rnf212 Mutants Have Enlarged Oocyte Reserves

RNF212 is a RING-family E3-ligase that regulates the progression of meiotic recombination via the SUMO modification and ubiquitin-proteasome systems, and is essential for crossing over (Qiao et al., 2014; Rao et al., 2017; Reynolds et al., 2013). Rnf212 mutation in mice causes sterility in both sexes. Although early events of meiotic prophase occur efficiently in Rnf212^−/− mutants, including synapsis and DSB repair, crossing-over fails leaving homologs unconnected at metaphase I. In males, the resulting univalents are a potent trigger of apoptosis resulting in the complete absence of spermatids (Reynolds et al., 2013). Elimination of gametes was not seen in Rnf212^−/− mutant females (Figure 1, A–E). On the contrary, large numbers of oocytes were present in Rnf212^−/− mutant ovaries. This sexually dimorphic response to Rnf212^−/− mutation presumably reflects the fact that postnatal oocytes are arrested in the dictyate stage, which follows crossing over and desynapsis of homologous chromosomes, but precedes the meiosis I division. Thus, the absence of crossovers in Rnf212^−/− mutant oocytes goes undetected until they resume meiosis in the preovulatory stage of folliculogenesis, as seen for other mutants defective for late steps of crossing over (Hunter, 2017; Kan et al., 2008). Consistent with this interpretation, in vitro maturation experiments showed that oocytes from Rnf212^−/− mutants resumed meiosis efficiently, but failed to successfully execute meiosis I (Figure S1). Thus, the absence of early defects in DSB repair and homolog synapsis allows non-fertile follicle reserves to be established in Rnf212^−/− mutants.

Figure. 1. RNF212 promotes postnatal oocyte apoptosis.

(A, B) Representative wild type and (C, D) Rnf212 mutant ovary sections from 10 and 18 dpp females immunostained for p63 with hematoxylin counterstaining. (E) Oocyte counts in wild type and Rnf212−/− mutant ovaries. For immunofluorescence-stained ovary sections (1 and 4 dpp counts), MVH/SYCP3-positive cells with or without p63 staining-positive nuclei were counted. For immunohistochemistry-stained sections (10 and 18 dpp counts), p63-positive oocyte nuclei with characteristic surrounding granulosa cell layers were counted. In both cases, counts were made for every fifth section and multiplied to calculate the total number of oocytes per ovary, as described (Di Giacomo et al., 2005). For each genotype, 6-9 ovaries from 3-5 mice were analyzed. Error bars show mean ± Sd. **P ≤ 0.01, two-tailed Mann-Whitney test. (F) Representative wild-type oocyte nucleus at 2 dpp immunostained for γH2AX (green), caspase 3 (red) and chromosome axis marker SYCP2 (grey). (G) Quantification of apoptotic nuclei at successive days postpartum. For each data point, 96-109 nuclei were counted from two animals. **P < 0.01, z-test. Scale bar in F. is 10μm.

Quantification of oocyte numbers revealed that oocyte reserves were significantly larger in Rnf212^−/− mutants than in their wild-type counterparts. Ovaries dissected from 1, 4, 10 and 18 day old animals (day post-partum, dpp) were fixed, sectioned and immunostained for oocyte-specific markers (mouse vasa homologue, MVH; synaptonemal complex protein 3, SYCP3; and/or p63)(Figure 1, A–D, and Figure S2). At 1 and 4 days postpartum (dpp), oocyte numbers in Rnf212^−/− mutant ovaries were indistinguishable from their wild-type counterparts (Figure 1E). However, the ovaries of 10 and 18 dpp Rnf212^−/− mutants contained 45-57% more oocytes than wild-type ovaries.

RNF212 Promotes Physiological and Pathological Oocyte Apoptosis

One explanation for the larger oocyte pools of Rnf212^−/− mutants is that RNF212 promotes oocyte apoptosis. To address this possibility, we immunostained early postnatal oocytes from 0.5, 1.0 and 2.0 dpp animals for the DNA-damage marker, phosphorylated histone H2AFX (γH2AX), and the apoptosis execution factor, Caspase 3 (Figure 1F and 1G). A subset of oocyte nuclei was identified with pan-nuclear staining for both γH2AX and Caspase 3, which is diagnostic of apoptotic nuclei (Harada et al., 2014) (Figure 1F). In wild-type animals, the fraction of apoptotic oocytes increased ~4-fold between 0.5 and 2.0 dpp (from 6 to 21%) indicating elevated apoptosis as oocytes transitioned into dictyate (Figure 1G). This increase was not seen in the Rnf212^−/− mutant, for which levels of apoptotic nuclei remained low and relatively constant, at ~5% of total oocytes, between birth and 2 dpp. We conclude that RNF212 facilitates the apoptosis of a subset of early postnatal oocytes.

To test the idea that RNF212 promotes apoptosis of oocytes that have experienced errors in meiotic prophase I, we asked whether Rnf212 mutation could rescue the wholesale elimination of oocytes caused by Spo11 and Msh4 mutations (Figure 2). Spo11^−/− mutants fail to initiate meiotic recombination resulting in severely defective homolog synapsis and almost complete elimination of the oocyte pool (Figure 2A and 2C) (Baudat et al., 2000). Consistent with previous analysis (Di Giacomo et al., 2005), only a very small number of oocytes survived in Spo11^−/− mutants at 18 dpp, less than 7.5% of wild type (Figure 2C and 1E). However, in Spo11^−/− Rnf212^−/− double mutants, large numbers of oocytes were present, averaging 40% of the oocyte numbers seen in Rnf212^−/− single mutants (Figure 2B, 2C and 2G). More complete restoration of oocyte pools by Rnf212 mutation was seen in the Msh4^−/− mutant background. The absence of MSH4 causes severe defects in both DSB repair and homolog synapsis resulting in the complete elimination of oocytes within 4 days of birth (Kneitz et al., 2000)(Figure 2 D and 2F). Strikingly, the ovaries of Msh4^−/− Rnf212^−/− double mutants contained high numbers of oocytes, averaging 75% of those in the Rnf212^−/− single mutant (Figure 2E, 2F and 2G). Moreover, small numbers of oocytes were able to escape apoptosis even when the Rnf212 mutation was heterozygous in the Msh4^−/− background (Figure S3). Thus, the complete elimination of oocytes in Msh4^−/− mutants is sensitive to the level of RNF212.

Figure 2. RNF212 is required for wholesale oocyte apoptosis in response to defects in early meiotic prophase.

(A) Representative Spo11^−/− and (B) Spo11^−/− Rnf212^−/− 18 dpp ovary sections immunostained for p63 with hematoxylin counterstaining. (C) Oocyte counts in Spo11^−/− and Spo11^−/− Rnf212^−/− ovaries at 18 dpp. (D) Representative Msh4^−/− and (E) Rnf212^−/− Msh4^−/− 18 dpp ovary sections. The dashed line highlights the degenerate ovary in the Msh4^−/− mutant. (F) Oocyte counts at 18 dpp. (G) Oocyte counts at 18 dpp in wild type, Rnf212^−/−, Spo11^−/− Rnf212^−/−, and Msh4^−/− Rnf212^−/−. ***P ≤ 0.001, two-tailed Mann-Whitney test. Error bars show mean ± SD. Scale bars for ovary sections are 300 μm.

RNF212 Modulates DNA Damage Levels in Early Postnatal Oocytes

The data above indicate that RNF212 promotes apoptosis of oocytes that have experienced defects in meiotic prophase I. RNF212 could influence the primary signaling of defects in recombination and synapsis, for example by altering DSB repair. Alternatively, RNF212 could be a component of downstream signaling pathways that ultimately trigger apoptosis. To distinguish these possibilities, we monitored levels of DNA damage in early postnatal oocyte nuclei via γH2AX immunostaining (Figure 3A and 3B). During meiosis, γH2AX decorates the chromatin around DSBs and accumulates at regions of chromosomes that fail to undergo synapsis (Cloutier et al., 2015; Mahadevaiah et al., 2001). We reasoned that if RNF212 functions solely in downstream steps to signal damage or trigger apoptosis, presumptively defective oocytes with high levels of γH2AX should be present at highly elevated levels in the ovaries of Rnf212^−/− mutants. Oocytes from 1 dpp mice were found to be in one of three stages: late pachytene, diplotene or early dictyate (Figure 3A). In late pachytene nuclei, an average of 40 ± 15 γH2AX chromatin “flares” were detected, decreasing to 21 ± 11 in diplotene, and just 6 ± 3 as oocytes transitioned into dictyate (Figure 3A and 3B). Contrary to expectations, numbers of γH2AX flares were ~2-fold lower at all stages in the early postnatal oocytes of Rnf212^−/− mutants implying lower DSB levels. Rnf212^−/− oocytes entering dictyate contained an average of only 2.7 ± 1.4 γH2AX flares, compared to 6 ± 3 in wild-type cells; moreover, 20% of nuclei had no staining at all, compared to just 6% of wild-type oocytes (Figure 3A and 3B). Qualitatively, the γH2AX signals detected in Rnf212^−/− oocytes were generally fainter and more compact than those seen in wild type.

Figure. 3. RNF212 impedes DNA-damage repair in early-postnatal oocytes.

(A) Representative 1 dpp oocyte nuclei at the indicated stages immunostained for chromosome axis marker SYCP3 (red) and γH2AX (green). Arrow indicates an individual γH2AX flare. (B) Quantification of γH2AX flares in oocyte nuclei at successive prophase substages. All counts are for 20-40 nuclei from two independent animals. (C) Representative 1 dpp oocyte nuclei at the indicated stages immunostained for chromosome axis marker SYCP3 (red) and RPA (green). Arrows highlight individual RPA foci. (D) Quantification of RPA foci in 1 dpp oocyte nuclei at successive prophase substages. All counts are for 20-35 nuclei from two independent animals. (E) Experimental regimen to monitor the acute response to irradiation-induced DNA damage. (F) Representative oocyte nuclei 2 hrs after irradiation immunostained for chromosome SYCP3 (red) and γH2AX (green). (G) Representative oocyte nuclei 24 hrs after irradiation. (H) Quantification of γH2AX immunostaining areas in oocyte nuclei at indicated times after irradiation. All counts are for 20-25 nuclei from two independent animals. LP, late pachynema; ED, early diplonema; LD, late diplonema; Di, early dictyotene. **P ≤ 0.01, ***P ≤ 0.001, two-tailed Mann-Whitney test. Error bars show mean ± SD. Scale bars, 10μm.

Given that γH2AX can mark both DSBs and non-DSB lesions during meiosis (Cloutier et al., 2015), we also immunostained oocytes from 1 dpp animals for Replication Protein A (RPA), which tightly binds regions of single-stranded DNA created during replication, repair and recombination (Ribeiro et al., 2016)(Figure 3C and 3D). RPA-staining foci were detected in similar numbers to γH2AX flares, suggesting that the vast majority of γH2AX signals in wild-type and Rnf212^−/− oocytes mark sites of DNA lesions, very likely DSBs. Moreover, like γH2AX flares, numbers of RPA foci were reduced ~2-fold in Rnf212^−/− mutant oocytes.

Together, these data imply that RNF212 modulates the generation or processing of DNA damage that leads to oocyte apoptosis. To distinguish these possibilities, the processing of exogenous DNA damage was examined by exposing 1 dpp wild-type and Rnf212^−/− animals to 1 Gy γ-irradiation, estimated to inflict ~20 DSBs per nucleus (Figure 3E)(Rinaldi et al., 2017; Suh et al., 2006). Post-irradiation γH2AX signals were then quantified in diplotene and early dictyate nuclei after 2 and 24 hours (Figure 3F–H). After 2 hrs, wild-type nuclei had developed pan-nuclear γH2AX chromatin staining signals reflecting extensive DNA damage. In stark contrast, only axis-associated γH2AX flares and puncta were observed in Rnf212^−/− oocytes, similar to those seen in unirradiated oocytes, and total γH2AX staining areas were reduced at least 2-fold relative to wild type (Figure 3A). At 24 hrs, pan-nuclear staining persisted in wild-type oocytes and γH2AX staining areas were indistinguishable from those at 2 hrs (P=0.53, unpaired t-test) indicating unrepaired damage. In Rnf212^−/− oocytes, γH2AX staining areas at 24 hrs were significantly reduced relative to 2 hrs (P=0.024, unpaired t-test) and residual γH2AX signals comprised mostly very small foci, some clustered and some dispersed, many of which were not axis associated (Figure 3G). Thus, the processing of radiation-induced DNA damage differs dramatically in wild-type and Rnf212^−/− mutant oocytes, with much more efficient and rapid repair occurring in the latter.

RNF212 Impedes DNA Damage Repair in Meiotic Mutants

The inference that RNF212 impedes DSB repair in early postnatal oocytes was further explored by quantifying DNA-damage markers in oocytes from the meiotic mutants, Spo11^−/− and Msh4^−/−, and the corresponding double mutants with Rnf212^−/− (Figure 4 and Figure S4). The defective homolog synapsis that occurs in Spo11^−/− and Msh4^−/− mutants makes it hard to distinguish oocytes in late pachytene from those in diplotene. Therefore, we could assign only two categories of oocyte nuclei in these mutants: “early” nuclei had bright, contiguous SYCP3-staining axes and were inferred to be in pachytene- and diplotene-like stages; and “late” nuclei had fainter, fragmented SYCP3 axes and were inferred to be either early dictyate or early apoptotic oocytes (Figure 4A). In early-stage Spo11^−/− oocytes from 1 dpp animals, γH2AX staining was present as numerous large flares, encompassing long portions of individual chromosome axes indicative of the extensive transcriptional silencing that occurs in response to asynapsis called meiotic silencing of unpaired chromatin (MSUC)(Baarends et al., 2005; Cloutier et al., 2015; Kouznetsova et al., 2009; Turner et al., 2005). 88% of Spo11^−/− oocytes showed this pattern, while 12% had pan-nuclear staining (Figure 4A, 4B and 4C, and Figure S4). Much more extensive γH2AX staining was seen in late-stage Spo11^−/− mutant oocytes; more than 91% showed pan-nuclear staining, while the remainder had a pattern of extended γH2AX flares. Mutation of Rnf212^−/− in the Spo11^−/− background greatly reduced the extent of γH2AX staining. In late-stage Rnf212^−/−Spo11^−/− oocytes, the fraction of nuclei with pan-nuclear staining was reduced by half (45% versus 91% in the Spo11^−/− single mutant) with a corresponding increase in nuclei with γH2AX flares. Moreover, for both early and late-stage Rnf2l2^−/− Spo11^−/− oocytes containing γH2AX flares, the total staining areas were greatly reduced, by 5-fold, relative to those of nuclei from the Spo11^−/− single mutant (Figure 4E and Figure S4). Rnf2l2 mutation also diminished γH2AX signals in Msh4^−/− mutant oocytes, but the effects were more extreme (Figure 4A, 4B and 4C, and Figure S4). 85% of early-stage and 91% of late-stage oocytes from the Msh4^−/− mutant showed pan-nuclear γH2AX staining. Oppositely, ~80% of oocytes from Rnf2l2^−/− Msh4^−/− double mutants had the γH2AX flare-staining pattern, with greatly reduced (30-fold) staining areas compared to those of Msh4^−/− single mutant nuclei. Thus, Rnf2l2 mutation diminishes the very high levels of γH2AX that occur in early postnatal oocytes of both Spo11^−/− and Msh4^−/− mutants.

Figure. 4. RNF212 impedes DNA-damage repair in Spo11 and Msh4 mutants.

(A) Representative 1 dpp oocyte nuclei from the indicated strains immunostained for chromosome axis marker SYCP3 (red) and γH2AX (green). (B) Quantification of γH2AX flares and pan nuclear γH2AX staining in 1 dpp oocytes. All counts are for 90-110 nuclei from two independent animals. (C) Quantification of γH2AX immunostaining areas in 1 dpp oocytes. All counts are for 20-30 nuclei from two independent animals. (D) Representative 1 dpp oocyte nuclei from the indicated strains immunostained for chromosome axis marker SYCP3 (red) and RPA (green). Arrows highlight individual RPA foci. (E) Quantification of RPA foci in 1 dpp oocytes from the indicated strains. All counts are for 20-30 nuclei from two independent animals. χ² contingency test was used for analysis of data in (B); two-tailed Mann-Whitney test for (C) and (E). n.s. not significant (P=0.17), **P ≤ 0.01, ***P ≤ 0.001. Error bars show means ± SD. Scale bars, 10μm.

Supporting the interpretation that Rnf2l2 mutation is reducing the level of DNA damage present in early postnatal oocytes of Spo11^−/− and Msh4^−/− mutants, RPA foci showed changes similar to those seen for γH2AX (Figure 4D and 4E). In Spo11^−/− and Msh4^−/− single mutants, early-stage oocytes contained an average of 50 ± 53 and 77 ± 42 RPA foci, respectively. The presence of high levels of Spo11-independent DSBs in Spo11^−/− mutant oocytes was previously documented and shown to make a significant contribution to oocyte apoptosis in this asynaptic background via the DNA damage response (Carofiglio et al., 2013; Carofiglio et al., 2018; Rinaldi et al., 2017). Although the origin of these breaks is currently unknown, a likely source is LINE-1 activity (Malki et al., 2014). In contrast to wild-type animals, in which RPA focus numbers decreased as oocytes transitioned into dictyate (Figure 3C and 3D), foci further increased in late-stage Spo11^−/− and Msh4^−/−oocytes (Figure 4D and 4E), mirroring the transition to pan-nuclear γH2AX staining in these mutants (Figure 4A and 4B). Rnf212^−/− mutation decreased RPA focus numbers by 3 to 4 fold in both Spo11^−/− and Msh backgrounds. Thus, consistent with the idea that RNF212 impedes DSB repair in early postnatal oocytes, the presence of RNF212 greatly enhances DNA damage levels in meiotic mutants defective for homolog synapsis and DSB repair.

RNF212 is Required for HORMAD1 to Associate With Desynapsing Chromosomes as Oocytes Transition into Quiescence

The ability of Rnf212 mutation to suppress the oocyte death of recombination and synapsis mutants is shared by mutation of the Hormad1 gene that encodes a meiosis-specific HORMA (Hop1, Rev7 and Mad2)-domain protein with central roles in regulating the events of meiotic prophase I (Daniel et al., 2011; Kogo et al., 2012b; Shin et al., 2013). HORMAD1 and its orthologs have been implicated in meiotic DSB formation, homolog pairing and synapsis, checkpoint signaling, transcriptional silencing (MSUC), and biasing meiotic recombination to occur between homologs by impeding inter-sister DSB repair (Carballo et al., 2008; Rinaldi et al., 2017; Royo et al., 2013; Shin et al., 2010; Stanzione et al., 2016; Vader and Musacchio, 2014). HORMAD1 is initially associated with unsynapsed chromosome axes during early prophase I, but then becomes depleted as homologs synapse, before re-associating with desynapsing axes during diplonema (Fukuda et al.; Niu et al., 2005; Wojtasz et al., 2009)(Figure 5A).

Fig. 5. RNF212 promotes association of HORMAD1 with desynapsed homolog axes.

(A, B) Representative 1 dpp oocyte nuclei from the indicated strains immunostained for SYCP3 (red) and HORMAD1 (green). LP, late pachynema; ED, early diplonema; LD, late diplonema; Di, early dictyotene. (C) Quantification of HORMAD1 signal intensity ratios in the indicated strains. All counts are for 80-110 nuclei from two independent animals. For each nucleus, total signal for HORMAD1 was divided by total signal for SYCP3: ratio of ≤0.2, low intensity; 0.21-0.4, medium intensity; ≥0.41, high intensity (Figure S5). ***P < 0.001, G-test. Scale bars, 10μm.

The role of HORMAD1 upon its reassociation with desynapsed diplotene chromosomes is unknown, but its timing and location suggest the potential to influence the repair of residual DSBs between sister chromatids. Therefore, we asked whether Rnf212 mutation affects the reassociation of HORMAD1 with diplotene and early dictyate chromosomes (Figure 5A). Immunostaining of oocyte chromosome spreads from 1 dpp wild-type animals revealed strong association of HORMAD1 with desynapsed homolog axes in early diplotene nuclei (Figure 5A, top row). In nuclei where desynapsis was more advanced, extensive association of HORMAD1 was seen along the lengths of all desynapsed axes. This extensive HORMAD1 staining pattern persisted into early dictyate, even in nuclei where the SYCP3-staining axes had become faint and highly fragmented. In Rnf212^−/− mutant oocytes, association of HORMAD1 with diplotene and dictyate chromosomes was greatly diminished (Figure 5A, bottom row). To quantify this defect, we binned nuclei into three HORMAD1 staining-classes based on their relative signal intensities (Figure 5C and Figure S5). In wild-type nuclei from 1 dpp animals, 61, 23 and 16% of nuclei showed high, medium and low intensity HORMAD1 staining, respectively; compared to 20, 47 and 33% of Rnf212^−/− mutant nuclei (P<0.001, G-test). We also examined HORMAD1 dynamics in 1 dpp Msh4^−/− mutants, in which all oocytes are destined for apoptosis, and in Rnf212^−/− Msh4^−/− double mutants where most oocytes survive (Figure 5B and 5C). High-intensity HORMAD1 staining was observed in 85% of Msh4^−/− oocyte nuclei (both early and late stage nuclei, defined by the criteria described above). Again, chromosomal association of HORMAD1 was greatly diminished by Rnf212^−/− mutation; just 24% of nuclei from Rnf212^−/− Msh4^−/− double mutants showed high intensity HORMAD1 staining (P<0.001, G-test). Thus, RNF212 promotes or stabilizes the association of HORMAD1 along desynapsed homolog axes both in wild-type oocytes and in mutants such as Msh4^−/−that experience severe defects in synapsis and recombination.

RNF212 Enhances DSB-Induced Apoptosis as Oocytes Transition into Quiescence

The data above suggest that the transition into quiescence, from the diplotene to dictyate stages, marks a critical oocyte quality-control juncture where the success of meiotic prophase is determined by assessing levels of residual DNA damage. To test this interpretation we asked if exogenous DSBs, inflicted specifically during the early postnatal period, could trigger oocyte apoptosis and whether RNF212 enhanced this process (Figure 6). Animals were exposed to γ-irradiation at different times during the two-day window after birth when oocytes transition into dictyate. Previous studies showed that exposure of 5 dpp mice to 0.45 Gy, estimated to induce ~10 DSBs per nucleus, efficiently triggered apoptosis (Rinaldi et al., 2017; Suh et al., 2006). Therefore, we initially exposed wild-type and Rnf212^−/− littermates, at 0.5, 1 or 2 dpp, to 0.45 Gy. Oocyte numbers from irradiated mice were then quantified at 10 dpp and compared to unirradiated controls (Figure 5A-D). At 0.5 dpp, oocytes were largely resistant to radiation-induced apoptosis. Oocyte numbers in both wild-type and Rnf212^−/− animals were similar to those of unirradiated wild-type controls P> 0.58; Figure 6D), but still smaller than those of unirradiated Rnf212^−/− animals. In contrast, oocytes at 1 and 2 dpp were highly sensitive to radiation-induced apoptosis. Survival of wild-type oocytes averaged only 16% and 7% of unirradiated controls following exposure to 0.45 Gy at 1 and 2 dpp, respectively (Figure 6D). Survival of Rnf212^−/− oocytes irradiated at 1 and 2 dpp was slightly higher than wild type, averaging respectively 21% and 10% of the Rnf212^−/− mutant unirradiated controls (or 39% and 20% of wild-type unirradiated controls). Thus, at 0.5 dpp, both wild-type and Rnf212 oocytes are largely resistant to a dose of irradiation that becomes a potent trigger of apoptosis at 1 and 2 dpp, as oocytes transition into dictyate.

Figure. 6. RNF212 sensitizes early-postnatal oocytes to radiation-induced apoptosis.

(A) Experimental regimen for irradiation experiments. (B, C and E, F) 10 dpp ovary sections from Rnf212^+/+ and Rnf212^−/− females following irradiation with 0.45 or 0.35 Gy at 1 dpp. (D and G) Oocyte counts following irradiation with 0.45 or 0.35 Gy, respectively. Error bars show means ± SD. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, two-tailed Mann-Whitney test. Scale bars, 300 μm.

Suh et al. (Suh et al., 2006) demonstrated a narrow dose-response relationship between γ-irradiation and oocyte apoptosis. This fact, together with the suggestion that Rnf212^−/− mutant oocytes are slightly more resistant to 0.45 Gy at 1 and 2 dpp (Figure 6D), motivated us to examine the effects of a lower dose of irradiation (Figure 6E–G). With a dose of 0.35 Gy (equivalent to ~7 DSBs per nucleus), both wild-type and Rnf212^−/− oocytes were again largely resistant to apoptosis when exposed at 0.5 dpp, but highly sensitive when exposed at 2 dpp, similar to the responses observed with 0.45 Gy. However, a striking difference in sensitivity was observed when mice were exposed at 1 dpp. While wild-type oocytes remained highly sensitive, averaging only 13% survival (relative to unirradiated wild type), Rnf212^−/− mutant oocytes were largely resistant with 68% survival relative to unirradiated Rnf212^−/− controls (Figure 6G). We conclude that RNF212 enhances DSB-induced apoptosis during a narrow window as oocytes transition from diplotene to dictyate.

DISCUSSION

Together, our data delineate a major oocyte quality control point at the diplotene-to-dictyate transition during which DSB repair is actively impeded in order to gauge whether the events of early meiotic prophase occurred successfully (Figure 7). At this stage, an unrepaired DSB is indicative of a failed inter-homolog interaction, or possibly a lesion resulting from LINE-1 activity (Malki et al., 2014). During diplotene, homologs are no longer closely juxtaposed (except very locally at crossover sites) and inter-homolog recombination factors are no longer present such that accurate repair of residual DSBs can only be achieved via recombination with the sister chromatid. By blocking inter-sister recombination, RNF212-dependent re-loading of HORMAD1 onto desynapsed homolog axes effectively acts to preserve a “memory” of meiotic prophase errors. Reinstallation of HORMAD1 could similarly impede repair of residual LINE-1 induced damage, and also reinforce transcriptional silencing via MSUC, which requires HORMAD proteins (Royo et al., 2013; Wojtasz et al., 2012). Consequently, a robust signaling response from any residual lesions will be maintained during the diplotene-to-dictyate transition enabling a meiotic quality-control decision to be made (Figure 7). Signaling above a critical threshold efficiently triggers apoptosis thereby preventing oocytes that have experienced significant recombination and synapsis errors, or high levels of LINE-1 activity, from contributing to the ovarian reserve. Consistent with the study of Rinaldi et al. (2017), we estimate that this threshold is equivalent to ~10 DSBs per nucleus (Figure 6). Importantly, our data indicate that pre-follicle apoptosis is a significant physiological quality-control pathway that removes as much as half of the oocytes that remain after LINE-1 induced early oocyte attrition (Malki et al., 2014). Together, we can estimate that fetal and pre-follicle attrition cull ~75% of all oocytes in mouse (Malki et al., 2014)(this study), mirroring the dramatic reduction in oocyte numbers seen in human females between ~20 weeks gestation and birth (Findlay et al., 2015), presumably via equivalent quality control mechanisms.

Figure. 7. Logic of pre-follicle oocyte quality control.

Timeline of chromosomal and DNA events during oocyte meiotic prophase. Magenta lines indicate homolog axes associated with HORMAD1. Red “explosions” represent unrepaired DSBs. Transparent boxes indicate repair pathways that are unavailable following desynapsis in diplotene. Inter-sister repair is actively impeded by RNF212-dependent reassociation of HORMAD1 with desynapsed homolog axes. Consequently, residual DSBs persist and robustly signal via the DNA-damage response pathway. The integrated level of damage signaling from all residual DSBs dictates whether an individual oocyte undergoes apoptosis or survives to form a primordial follicle and become established in the ovarian reserve.

Analysis of recombination intermediates in budding yeast has provided direct evidence that Hop1^HORMAD negatively regulates inter-sister recombination as part of a phospho-kinase signal-transduction pathway that responds to DSBs (Humphryes and Hochwagen, 2014; Lao and Hunter, 2010). Following DSB-dependent phosphorylation by PI3K-like sensor kinases, Mec1^ATR and Tel1^ATM, Hop1^HORMAD acts as a mediator for activation of the effector kinase, Mek1, which directly targets components of the recombination and cell-cycle machinery (Callender et al., 2016; Carballo et al., 2008; Prugar et al., 2017). Two lines of evidence in mouse indicate that mammalian HORMADs also negatively regulate inter-sister recombination; first, repair of irradiation-induced DSBs is accelerated when HORMADs are absent; and second, DSB levels are diminished in asynaptic Spo11^−/− mutant meiocytes in which inter-sister recombination is likely the only repair option (Carofiglio et al., 2018; Rinaldi et al., 2017; Shin et al., 2013).

HORMAD1 initially loads onto leptotene chromosomes independently of RNF212 (data not shown). After homolog pairing has been achieved, we propose that nascent interhomolog interactions are “locked in” by synaptonemal complexes and associated meiosis-specific recombination factors (ZMMs)(Hunter, 2015; Lynn et al., 2007). Consequently, the role of HORMAD proteins in preventing intersister recombination is no longer necessary. Indeed, dissociation of HORMADs as synapsis ensues may enable inter-sister repair of DSBs that either failed to engage a homolog template, or that engaged in interhomolog DNA strand exchange to facilitate pairing, but did not complete repair. Estimates of unrepaired DSB numbers in the Trip13 mutant, which fails to remove HORMADs from homolog axes upon synapsis, implies that up to a quarter of DSBs may utilize inter-sister recombination for their repair (Rinaldi et al., 2017). Our observation that late pachytene cells are relatively insensitive to DSB-induced apoptosis (Figure 6)(Hanoux et al., 2007) could be explained, at least in part, by a larger capacity to repair DSBs via inter-sister recombination following the dissociation of HORMAD proteins. As such, many more DSBs would be required to reach the signaling threshold required to induce apoptosis at this stage. In addition, the signaling threshold required to induce apoptosis during pachytene may be increased in order to prevent aberrant culling of oocytes that are still completing recombination (Hanoux et al., 2007; Kim and Suh, 2014). The window in which RNF212 enhances DSB-induced apoptosis also likely reflects the timing of expression of the transactivating p63 isoform, TAp63, and components of the apoptotic machinery. As oocytes progress into dictyate, TAp63, becomes fully expressed and activatable by phosphorylation rendering primordial follicles exquisitely sensitive to DNA damage (Kim and Suh, 2014; Livera et al., 2008; Suh et al., 2006). Similarly, expression of the initiator caspase, caspase-2, correlates with the onset of radiosensitivity in early postpartum oocytes (Hanoux et al., 2007).

RNF212 is known for its essential role in meiotic crossing over (Qiao et al., 2014; Rao et al., 2017; Reynolds et al., 2013). At the cytological level, RNF212 promotes chromosome-associated SUMO conjugation during early meiotic prophase and renders the progression of recombination beyond nascent strand-exchange intermediates dependent on the action of the ubiquitin-proteasome system, mediated specifically by a second RING ligase, HEI10. RNF212 initially localizes as numerous puncta along synapsed chromosomes during zygotene and early pachytene, before accumulating specifically at future crossover sites (Reynolds et al., 2013). However, RNF212 is not detected along desynapsing chromosomes during diplotene, i.e. during the time it ostensibly mediates the reassociation of HORMAD1. This observation suggests that RNF212 may function in the nucleoplasm to license HORMAD1 to reassociate with homolog axes upon desynapsis. Alternatively, such a licensing step could occur much earlier, in early-mid pachytene when RNF212 localizes along synaptonemal complexes, but HORMAD1 loading would be contingent upon subsequent desynapsis. Possibly, RNF212 prevents HORMAD1 from being degraded and/or antagonizes the action of TRIP13, an AAA+ ATPase required to dissociate HORMADs from synapsed homolog axes (Li and Schimenti, 2007; Roig et al., 2010; Wojtasz et al., 2009; Ye et al., 2017). Given that low levels of chromosome-associated HORAMD1 are still detected in Rnf212 mutant oocytes during diplotene and dictyate (Figure 5), RNF212 may primarily affect the ability of HORMAD1 to oligomerize, a fundamental attribute of HORMAD proteins that is central to their function (Kim et al., 2014; Rosenberg and Corbett, 2015; Ye et al., 2017).

Like yeast Hop1, HORMAD1 is important for DSB formation. Thus Hormad1 mutant phenotypes reflect both reduced DSBs and derepression of inter-sister recombination (Daniel et al., 2011; Shin et al., 2010). Rnf212 mutation does not impact the early roles of HORMAD1 in DSB formation and inter-homolog bias (Reynolds et al., 2013), but reveals a previously undefined late role as it reassociates with diplotene chromosome axes. This unanticipated role of RNF212 implicates SUMO modification in oocyte quality control and suggests that common variants of mammalian Rnf212 genes (Johnston et al., 2016; Kadri et al., 2016; Kong et al., 2008) could have compound effects on crossing over, and the size and quality of oocyte reserves.

STAR METHODS

METHODS CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact Neil Hunter, nhunter@ucdavis.edu

EXPERIMENTAL MODEL AND SUBJECT DETAILS

The oocytes analyzed in this study come from female fetuses at 18 days postcoitum (dpc) and from pups at 0.5, 1, 2, 4, 10, and 18 days postpartum (dpp). All mice were congenic with the C57BL/6J background. The Rnf212^−/−, Msh4⁻⁻, and Spo11^−/− lines were previously described (Baudat et al., 2000; Kneitz et al., 2000; Reynolds et al., 2013). Mice were maintained and used for experimentation according to the guidelines of the Institutional Animal Care and Use Committees of the University of California, Davis. Mice were housed under 12 hour dark, 12 hour light cycles and fed ad libitum with regular chow.

QUANTIFICATION AND STATISTICAL ANALYSIS

Most data are presented as means ± SEM or SD and represent at least three independent experiments. Statistical tests, P and n values are described in the figures and figure legends.

METHODS DETAILS

Ovary Processing and Staining

Ovaries were dissected from freshly euthanized animals at the indicated ages and fixed in 10% formalin overnight. After washing in 70% ethanol, ovaries were embedded in paraffin and sectioned onto glass slides (5 μm sections for ovaries from 1 and 4 dpp animals; 8 μm sections for mice older than 4 dpp). Sections were then deparaffinized, rehydrated and incubated with antigen retrieval buffer (10 mM Sodium Citrate, 0.05% Tween-20) for 50 minutes at 100°C. Slides were blocked in serum free blocking solution (Dako X0909) for one hour followed by Antibody Dilution Buffer (10% normal goat serum, 3% BSA, 0.05% Triton X-100, 0.05% sodium azide) for 15 minutes at room temperature. Incubation with primary antibodies was performed overnight at room temperature. Ovaries from 1 and 4 dpp mice were stained for immunofluorescence using mouse anti-p63 (Santa Cruz Biotechnology, sc-25268; 1:500 dilution) antibodies, rabbit anti-MVH (Abcam, ab13840; 1:500) and/or rabbit anti-SYCP3 (Santa Cruz Biotechnology, sc-33195; 1:300 dilution) antibodies. Slides were subsequently incubated with one of two pairs of goat secondary antibodies for 1 h at 37°C: (1) anti-rabbit 488 (Molecular Probes, A11070; 1:1,000 dilution) and anti-mouse 594 (Molecular Probes, A11020; 1:1,000 dilution); or (2) anti-rabbit 568 (Molecular Probes, A11036; 1:1,000 dilution) and anti-mouse 488 (A11029 Molecular Probes, 1:1000). Ovaries from mice older than 4 dpp were stained for immunohistochemistry (IHC) with anti-p63 antibody (Santa Cruz Biotechnology, sc-25268; 1:500 dilution) overnight at room temperature. Biotin-conjugated anti-mouse secondary antibody (ThermoFisher Scientific, A16076; 1:1000 dilution) was used at a 1:1000 dilution for 1 h at 37°C. IHC signals were developed with the ABC-DAB Peroxidase system (Vector Labs), counterstained with hematoxylin for 10 seconds, and mounted with Permount solution.

Irradiation

At the indicated ages, mice received a single dose of ionizing radiation (0.35 or 0.45 Gy) on a rotating turntable in a ¹³⁷Cs irradiator (dose rate 9.8 rad min⁻¹). Irradiated mice and non-irradiated controls were euthanized at 10 dpp and their ovaries harvested and processed for oocyte quantification as described above.

Oocyte culture and metaphase I chromosome spreads

Germinal-vesicle stage oocytes were collected from mature mice (2-4 months old) without prior hormonal stimulation. Only oocytes with integral cumulus cell layers were used. 20-30 oocytes were collected per mouse. Surrounding cumulus cells were mechanically removed for observation of Germinal Vesicle Breakdown (GVBD) and Polar Body Extrusion (PBE). Oocytes were manipulated in M2 medium containing 2.5 μM milrinone (Sigma-Aldrich) under mineral oil at 37°C and maturation was stimulated by milrinone washout (Yun et al., 2014). GVBD was scored every 20 minutes by manual inspection over the first 2 hours and then scored again after 3 hours. Only oocytes that underwent GVBD within 3 hours were used to quantify PBE after 15 hours of incubation. All incubation times were relative to milrinone washout. To count chiasmata, oocytes were cultured in M2 media for 7 h, and then prepared for metaphase-I chromosome spreads (Chambon et al., 2013). Acid Tyrode’s solution (Sigma-Aldrich) was applied to remove zona pellucida, and zona-free oocytes were spread in 1% paraformaldehyde plus 0.15% Triton-X-100. DNA was stained with DAPI and centromeres were immunostained with human anti-centromere antibodies (ACA; HCT-0100 ImmunoVision, 1:1000 dilution) overnight at room temperature, followed by anti-human 555 (Molecular Probes, 1: 1000 dilution) secondary antibody for 1 h at 37°C.

Oocyte nuclei spreads and immunofluorescence cytology

Ovaries were dissected from freshly euthanized animals and processed for surface spreading of oocyte chromosomes as described (Sun and Cohen, 2013). Immunofluorescence staining was performed using the following primary antibodies with incubation overnight at room temperature: mouse anti-SYCP3 (sc-74568 Santa Cruz, 1:200 dilution), rabbit anti-SYCP3 (sc-33195 Santa Cruz, 1:300), guinea pig anti-SYCP2 (a gift from Christer Höög, 1:300) (Kouznetsova et al., 2005), rabbit monoclonal anti-RPA32 (ab76420 Abcam, 1:100), rabbit anti-REC8 (a gift from Scott Keeney, 1:300), mouse monoclonal anti-γH2AX (05-636 Millipore, 1:500), rabbit polyclonal anti-γH2AX (A300-081A Bethyl laboratories, 1:300), goat polyclonal anti-Caspase3 (L-18 Sc-1225, 1:100), rabbit anti-HORMAD1 (a gift from Attila Toth, 1:250)(Wojtasz et al., 2009). Slides were subsequently incubated with the following goat secondary antibodies for 1 h at 37 °C: anti-rabbit 488 (A11070 Molecular Probes, 1:1000 dilution), anti-rabbit 568 (A11036 Molecular Probes, 1:2000), anti-mouse 555 (A21425 Molecular Probes, 1:1000), anti-mouse 594 (A11020 Molecular Probes, 1:1000), anti-mouse 488 (A11029 Molecular Probes, 1: 1000), or anti-guinea pig fluorescein isothiocyanate (106-096-006 FITC, Jackson Labs, 1:200). Coverslips were mounted with ProLong Gold antifade reagent (Molecular Probes).

Quantification and Statistical Analysis

Immunofluorescence-stained ovary sections and chromosome spreads were imaged using a Zeiss AxioPlan II microscope with a 20× Plan Apochromat 0.45 NA, or 63× Plan Apochromat 1.4 NA objective and EXFO X-Cite metal halide light source. Images were captured with a Hamamatsu ORCA-ER CCD camera and processed using Volocity software (Perkin Elmer). Immunohistochemistry-stained ovary sections were imaged using a ScanScope digital scanner (Aperio) and processed using ImageScope software. All comparisons were made between datasets obtained from animals that were either littermates, or matched by age. Two observers performed all quantitative analyses; the second observer was blind to which group/genotype was being analyzed. For immunofluorescence-stained ovary sections, MVH/SYCP3-positive oocytes with or without p63 stained nuclei were counted. For immunohistochemistry-stained ovary sections, p63-positive oocyte nuclei with characteristic surrounding granulosa cell layers were counted. In both cases, counts were made for every fifth section, as described (Di Giacomo et al., 2005). Counts were multiplied by five to calculate the total number of oocytes per ovary. Control counts for every ovary section confirmed the accuracy of this approach. For each genotype or condition, 3-8 ovaries were analyzed. For quantifications of spread oocyte nuclei, images of 90-160 nuclei from at least two animals were analyzed. SYCP3-staining nuclei were staged using standard criteria. In late pachytene, homologs are still fully synapsed and SYCP3 is present as 20 contiguous lines; during diplotene homologs desynapse and SYCP3 lines split; and as cells enter dictyate SYCP3 lines begin to dissolve and fragment, and are often accompanied by bright aggregates of SYCP3. To quantify high-density γH2AX and HORMAD1 signals, intensity ratios were calculated: total signal for γH2AX or HORMAD1 was divided by the total signal for SYCP3.

A. Data and Software Availability

B. Additional Resources

N/A

KEY RESOURCE TABLE

Antibodies
mouse anti-p63	Santa Cruz Biotechnology	sc-25268
rabbit anti-MVH	Abcam	ab13840
rabbit anti-SYCP3	Santa Cruz Biotechnology	sc-33195
anti-rabbit 488	Molecular Probes	A11070
anti-mouse 594	Molecular Probes	A11020
anti-rabbit 568	Molecular Probes	A11036
anti-mouse 488	Molecular Probes	A11029
Biotin-conjugated anti-mouse secondary antibody	ThermoFisher Scientific	A16076
human anti-centromere antibodies	ImmunoVision	ACA; HCT-0100
anti-human 555 secondary antibody	Molecular Probes	A-21433
mouse anti-SYCP3	Santa Cruz Biotechnology	sc-74568
guinea pig anti-SYCP2	a gift from Christer Höög	N/A
rabbit monoclonal anti-RPA32	Abcam	ab76420
rabbit anti-REC8	a gift from Scott Keeney	N/A
mouse monoclonal anti-γH2AX	Millipore	05-636
rabbit polyclonal anti-γH2AX	Bethyl laboratories	A300-081A
goat polyclonal anti-Caspase3 (L-18)	Santa Cruz Biotechnology	Sc-1225
rabbit anti-HORMAD1	a gift from Attila Toth	N/A
anti-mouse 555	Molecular Probes	A21425
anti-guinea pig fluorescein isothiocyanate	Jackson Labs	106-096-006 FITC
Chemicals, Peptides, and Recombinant Proteins
Protein Block, Serum Free, Blocking, Unconjugated, Liquid form, Immunohistochemistry, 110 mL	Agilent Dako	X090930-2
ABC-DAB Peroxidase system	Vector Laboratories	Cat# PK-6100
Tyrode’s Solution, Acidic	Sigma-Aldrich	Cat# T1788
milrnone	Sigma-Aldrich	Cat# 78415-72-2
3% BSA	Thermo Fisher Scientific	Cat# AM2616
ProLong Gold antifade reagent (Molecular Probes)	Thermo Fisher Scientific	Cat# P36930
M2 medium	Sigma-Aldrich	Cat# M7167

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Software and Algorithms
Volocity 6.3	Perkin Elmer	N/A
Aperio ImageScope software	Leica	N/A

Highlights.

RNF212, a SUMO ligase required for crossovers, also mediates oocyte quality control

RNF212 promotes apoptosis of oocytes that experienced defects in meiotic prophase I

RNF212 works with HORMAD1 to impede repair of residual DSBs as chromosomes desynapse

Impeded DSBs signal that meiosis was defective and trigger oocyte apoptosis