Abstract
The rad-51 gene in Caenorhabditis elegans is transcribed into alternative mRNAs potentially coding three alternative protein isoforms. We have genetically modified this gene in order to investigate the potential roles of the longest isoform, namely isoform A, in genome stability. The RAD-51 isoform A appears to contribute to genome stability in late development, but is not implicated in meiosis or DNA repair in the germline. However, the RAD-51 isoform A has a pivotal role in DNA damage induced apoptosis, but not in DNA damage checkpoint activation or physiological cell death. This is a relevant new finding that improves our understanding of how DNA damage apoptosis is restricted to late pachytene stage preventing the inappropriate loss of nuclei undergoing the earlier stages of meiotic recombination, during which a large number of physiologically induced DSBs are present.
Introduction
Homologous repair is the most faithful double strand break (DSB) repair program. However, it can only occur when a homologous substrate is present: either the sister chromatid after replication or the homologous chromosome in meiosis. Single strand invasion is the essential step in initiating such a process. The key protein at this step is the RecA-like protein Rad51. This protein is conserved in all eukaryotes from fungi to humans (La Volpe and Barchi 2012).
Meiosis is the mechanism that ensures proper segregation of homologous chromosomes during eukaryotic gametogenesis in order to confer a complete haploid genome to oocytes and sperms (Kleckner 1996; Hillers et al. 2017). During meiosis the topoisomerase-like protein Spo11 physiologically induces DSBs. Homologous repair will then occurs among homologous chromosomes in order to generate cross-overs (COs) and subsequent chiasmata that are essential for generating appropriate tension between homologous chromosomes and faithful segregation at anaphase I (Kleckner 1996).
In most eukaryotes a meiosis specific RecA-like protein, Dmc1, is required in the strand invasion step during homologous recombination at meiosis (Bishop et al. 1992). Rad51 and Dmc1 cooperate in this process (Brown and Bishop 2014). Surprisingly, in some organisms, such as in Sordariomycetes, Caenorhabditis elegans, and Drosophila melanogaster, the key actor in the meiotic step of strand invasion, Dmc1, is missing and Rad51 alone seems to perform strand-invasion in meiosis.
The C. elegans germ line exhibits the complete time course of meiotic prophase I in which nuclei at the different stages of oogenesis can be easily identified based on their position and chromosome appearance. In wild-type C. elegans, RAD-51 foci, representing nascent meiotic DSBs, arise during zygotene and early pachytene stages and then decrease in number during the middle pachytene stage as meiotic DSBs repair progresses (Colaiacovo et al. 2003).
Increase and/or persistence of such foci are indicative of DNA repair defects. During early meiotic prophase there is a clear bias towards homologous recombination using the homologous chromosome as a template, thus promoting CO formation (Rosu et al. 2011). In order to perform COs, chromosomes must be aligned and stabilized by the synaptonemal complex (SC) (Colaiacovo et al. 2003). As in other organisms, there is a surveillance system in C. elegans that checks recombination intermediates and pairing formation during meiosis (Machovina et al. 2016). During meiotic prophase, a number of proteins cooperate to ensure CO formation: some are involved in promoting pairing and synapsis of homologous chromosomes (Couteau and Zetka 2005; Colaiacovo 2006; Penkner et al. 2007; Meneely et al. 2012), whereas others, known as pro-CO factors, seem to be involved in stabilizing recombination intermediates between homologous chromosomes (Zalevsky et al. 1999; Kelly et al. 2000). In absence of the SC, as in syp-2 mutants, COs do not occur. Later, residual DSBs can be repaired using the sister chromatide. At this stage DSBs repair is brc-1 dependent (Adamo et al. 2008).
By the end of pachytene, about half of the oocytes in the C. elegans germ line undergo physiological cell death (Gumienny et al. 1999). In addition, exogenously induced DNA damage or persistent DNA lesions, due to endogenous DNA repair defects, are sufficient to trigger the DNA-damage checkpoint and significantly increase germline apoptosis (Gartner et al. 2000; Schumacher et al. 2005; Adamo et al. 2008; 2010; Rutkowski et al. 2011). In addiction to the DNA damage checkpoint, a different checkpoint enhances apoptosis during pachytene in response to defects in SC assembly (Bhalla and Dernburg 2005).
When exogenous DNA damage occurs, the DNA damage checkpoint induces cell cycle arrest at the G2 phase in the premeiotic proliferative region of the germ line (Schumacher 2001 et al.; Moser et al. 2009) to allow DNA repair to occur. The DNA damage checkpoint requires the 9-1-1 complex, a DNA damage–sensor complex that contains the conserved proteins RAD-9, RAD-1, and HUS-1 (Hofmann et al. 2002). Many proteins are involved in the activation of such sensor complex such as, for example, the ZTF-8 protein directly interacting with RAD-1. In order to fulfill this role, ZTF-8 needs to be SUMOylated (Kim and Colaiacovo 2014, 2015). Following 9-1-1 loading, activation of the p63-like protein CEP-1 is required to elicit apoptosis (Schumacher et al. 2001). Important to note that germline apoptosis only occurs in nuclei in the late pachytene stage of meiosis suggesting that there must be specific signals that make late pachytene nuclei competent for apoptosis and that prevent apoptosis at earlier meiotic stages. Restricting DNA damage-induced apoptosis to late pachytene stage could prevent the accidental loss of nuclei undergoing the first steps of meiotic recombination, during which DSBs are present (Colaiacovo et al. 2003, Adamo et al. 2012). Some genes, whose action is essential for apoptosis induction in presence of DNA damages during C. elegans gametogenesis, appear to act downstream of the proapoptotic signaling (Silva et al. 2013, Ackermann et al. 2016). Some of these genes are also involved in CO formation such as msh-4, msh-5, and zhp-3 (Silva et al. 2013, Jantsch et al. 2004). Another gene, involved in DNA damage induced apoptosis, is the ufd-2 gene coding for the E4 ubiquitin ligase UFD-2. In its absence, RAD-51 foci persist after pachytene, and DNA damage-induced apoptosis is prevented (Ackermann et al. 2016). Interestingly, nuclei undergoing apoptosis in late pachytene appear overloaded of the RAD-51 protein although DNA repair would be useless at this stage (Colaiacovo et al. 2003).
By the diakinesis stage, chromosomes are highly structured and appear as six DAPI stained bodies per nucleus. Each body corresponds to a couple of homologous chromosomes held together by chiasmata that are the result of the occurred COs. In CO defective mutants, such as syp-2, we can observe a vast majority of diakinesis nuclei containing twelve DAPI stained bodies corresponding to the 12 univalent chromosomes.
C. elegans rad-51 gene inactivation leads to a severe phenotype: unstructured chromosomes at diakinesis, reduced brood-size, high levels of germline apoptosis, and 100% embryonic lethality at the second generation (Gartner et al. 2000; Rinaldo et al. 2002; Alpi et al. 2003).
When the C. elegans rad-51 gene was first isolated (Rinaldo et al. 1998), it was shown that the C. elegans RAD-51 is positioned in a phylogenetic tree (derived by bootstrap analysis) at an intermediate branch between the eukaryotic Rad5l group and that of the eukaryotic Dmc1. The Rad5l fungal sequences surprisingly seem to score much better than the C. elegans RAD-51 sequence against the vertebrate Rad51. Although the C. elegans RAD-51 protein is more similar to members of the Rad51 family than to the Dmc1 family, some conserved Dmc1-specific amino acids are retained suggesting that the C. elegans RAD-51 may represent a hybrid performing the activities of both these strand-invading proteins. Therefore, in C. elegans a single gene locus is responsible for all the functions ascribed to the two loci in fungi as well as in vertebrates. The rad-51 genomic locus, composed of eight exons, was shown to be transcribed into two alternative messanger RNA (Rinaldo et al. 1998). More recent evidences suggest the presence of a third intermediate transcript. The second exon is only present in the longer transcript. At the time of the isolation of the gene, the authors hypothesized that the two different isoforms coded by this gene might be used at different stages and/or in different tissues (Rinaldo et al. 1998). The extra amino acids at the amino-terminal end of the long isoform might confer a different tissue-specificity or function to the coded protein. We here investigate this possibility.
Materials and Methods
Strains
C. elegans strains (listed in Supplemental Material, Table S1) were cultured at 20°C on NGM plates with Escherichia coli OP50 as food source according to standard methods (Brenner 1974).
PCR primers
The PCR to identify rad-51(A-) mutation are performed with three primers:
Upper rad-51: 5’TATGGGACAATCTTGGGG3’
Upper rad-51(A-): 5’ACGAAGTTATGGCTGCTCG3’
Lower rad-51: 5’ACGAAGTTATGGCTGCTCG3’
The PCR to identify brc-1(tm1145) mutation are performed with two primers:
Upper brc-1: 5’TGTCGCATCGTCGGCATTAA3’
Lower brc-1: 5’AATATAGGCACCGGCGGGGA3’
Screening of laying worms
Worms were isolated and cloned during the L4 larval state on Petri plates and left at 20°C to lay eggs for 3 days. They were transfered every 12 hours to lay eggs onto fresh plates until the deposition of non-fertilized oocytes. Each plate content was monitored every for 24/72 hours to analyze the following parameters among the progeny: embryonic lethality, presence of males, aberrant phenotypes and larval arrests. The percent of embryonic lethality is calculated as the ratio of unviable eggs on laid eggs. The percentage of males, aberrant phenotypes and larval arrests is calculated as the ratio of males/aberrant/larval phenotype on the hatched eggs.
Number of scored P0 worms: wt: 20; rad-51(A-): 12; cep-1: 8.
Larval growth assay
To analyze the larval growth rate after IR exposure, nematodes were synchronized with Alkaline Hypochlorite solution 1:1 (NaOH 1.6M, NaClO 0.6%) and embryos were transferred on seeded plates. Following the hatching of eggs the L1 worms are exposed to 120 Gy γ-rays, using a Caesium-137 source and observed after 53 hrs to analyze two different parameters: larval growth rate and the frequency of aberrant phenotypes.
Number of scored worms: wt 0 Gy: 105; rad-51(A-) 0 Gy: 114; wt 120 Gy: 97; rad-51(A-) 120 Gy:65
Immunostaining of meiotic nuclei
Gonads from 20 hrs post L4 adults were dissected in M9 solution (0.3% H2PO4, 0.6% Na2HPO4, 0.5% NaCl and 1mM MgSO4). Slides were frozen in liquid nitrogen, then immersed at −20°C in methanol, methanol/acetone (1:1) and acetone respectively for 5 min, and washed three times in PBS for 5 min each time. Slides were left in 3% BSA in PBS for 30 min at 37°C in a humid chamber. Primary rabbit antibody α-RAD-51 was diluted 1:200 in Ab buffer (1% BSA, 0.1% Tween-20, 0.05% sodium azide in 1X PBS). Slides were incubated for 90 min at room temperature followed by three washes in PBS for 5min. Secondary antibody Texas red anti-rabbit was diluted 1:400 in Ab buffer. Slides were incubated for 60 min in darkroom at room temperature, followed by three washes in PBS + 0.1% Tween-20 for 5 min. Slides were mounted with Prolong Gold Antifade reagent with DAPI (Life Technologies).
To perform the immunostaining after IR exposure, young adult worms were transferred on seeded plates and exposed to 120 Gy γ-rays, using a Caesium-137 source. Following the treatment, worms were incubated for 60 min at 20°C before the gonads dissection. The quantitative analyses of RAD-51 foci were performed by dividing the germ line into 6 zones (gonadal tip, mitotic zone, transition zone, early pachytene, middle pachytene and late pachytene stage), in accordance with their cytological features.
Number of scored nuclei are indicated in the legends
The quantifications of nuclei positive for RAD-51 1hr, 16hrs and 48hrs after IR were performed by immunostaining after 20 Gy γ-rays exposure of young adult worms, using a Caesium-137. Following the treatment, worms were incubated for 1hr, 16hrs and 48hrs at 20°C before the gonads dissection.
Number of scored nuclei are indicated in Table 2.
Quantitative analysis of germline apoptosis
Adult nematodes were suspended in M9 solution and stained by incubation with 33 μM SYTO-12 (Molecular probes) for 1hr and 30 min at room temperature in the dark. The worms were then transferred to seeded plates to allow stained bacteria to be purged from the gut. After 30 minutes, the animals were mounted on 2% agarose pads in 2mM levamisole. The estimation of apoptotic levels for each genotype was calculated as the average number of apoptotic nuclei per gonadal arm.
Number of scored arms are indicated in the legends.
Analysis of DAPI-stained germ lines
Adult nematodes were suspended in M9 solution on glass slides, permeabilized and fixed with 10 μl of 100% Et-OH. Finally, slides were mounted in 10 μl of DAPI (2ng/μl) diluted in M9. Number of scored nuclei are indicated in the legends of charts.
Mitotic nuclei were scored 15 h after 120 Gy γ-rays exposure of young adult worms, using a Caesium-137 Number of observed nuclei: wt 0 Gy: 85, rad-51(A-) 0 Gy:130; wt 75 Gy: 72, rad-51(A-) 75 Gy: 106.
Image collection and processing
Collection of images was performed using a Leica DM6 fluorescence microscope, Hamamatsu camera under the control of Leica LAS AF 6000 software. Images were processed using Leica LAS AF 6000 software and Image J program. Quantitative analysis of RAD-51 foci and DAPI-stained bodies along the germline were performed on z series, Optical sections were collected at 0.18 μm and 0.50 μm increments respectively.
Statistical tools
Statistical analyses of apoptosis levels and RAD-51 foci patterns were computed through t-Student test for independent samples.
The level comparison of aberrant phenotypes and larval arrests of the different genotypes were computed through χ 2 test.
Statistical analyses of nuclear diameters at the premeiotic tip were computed throught the two-tailed Mann-Whitney U test.
Statement on data availability
Strains are available upon request.
File S1 contains the Table Figures:
Table S1 (list of all strains used in this study and description rad-51(A-) mutant CRISP/Cas9-mediated genome editing); Table S2 (Statistical analysis of RAD-51 foci in meiosis: t-student); Table S3 (Statistical comparing of apoptotic nuclei in various genetic combination: t-student)
File S2 contains the Supplemental Figures:
Figure S1 (rad-51(A-) mutant by CRISP/Cas9 mediated genome editing); Figure S2 (RAD-51 A in the larval development); Figure S3 (genetic cross); Figure S4 (Immunostaining of RAD-51 in syp-2; rad-51(A-) strains).
Results
Lack of the RAD-51 isoform A affects genome stability in late development stages, but does not alter COs formation in Meiosis I
In order to understand whether the different isoforms coded by the rad-51 gene perform different functions in C. elegans, we edited the rad-51 gene by mean of the CRISPR Cas9 technology obtaining a mutant that cannot express the long RAD-51 isoform A (see Material and Methods and figure S1), but retains the ability to code for the shorter forms. We called the mutant rad-51(A-).
We observed the presence of occasional larval arrests in the mutant population, but we did not observe an increase in the level of males or dead embryos that would be suggestive of chromosome non disjunction in meiosis (Table 1). We then synchronized the nematodes and exposed the larvae at the stage L1/L2 to 120 Gy of g-rays and screened for developmental defects after 52 hrs. We observed in the rad-51(A-) mutant a significant increase in the larval arrest, eggless, and vulvaless phenotypes compared to the wild type in the same conditions (Table 2 and Figure S2) suggesting a requirement of the RAD-51 isoform A in maintainance of genome stability during late development.
We further questioned the efficiency of CO and chiasma formation in our mutant by cytogenetic analysis: rad-51(A-) shows six DAPI stained bodies (corresponding to the wildtype six bivalents) in all the diakinesis nuclei (Figure 1A). Furthermore, we assayed the frequency and distribution of RAD-51 foci along the adult hermaphrodite gonad by mean of immunolocalization and we observed no significant differences from the wild type strains (Figure 1B and Table S2). RAD-51 foci start to appear on chromosomes in the transition zone, increase in early pachytene and are unloaded in middle to late pachytene, as in the wild type. Therefore, CO formation does not seem to be affected by the lack of RAD-51 isoform A.
RAD-51 isoform A is not involved in inter-sister DNA repair during meiosis
In wildtype C. elegans, one and one only CO per pair of homologous chromosomes is observed. However, as in other eukaryotes, each homologous pair is subject to more than one DSB performed by SPO-11. It has been reported that meiotic DSBs are about twice the number of the final COs in C. elegans (Mets and Meyer 2009).
Some DNA repair genes such as brc-1 and fcd-2 are thought to be involved in meiosis in the intersister homologous repair of those DSBs that have not been resolved as COs (Adamo et al. 2008; 2010). We wondered whether RAD-51 isoform A might be loaded at such a step and might be involved in such a pathway. It has been demonstrated that, in absence of the synaptonemal complex, BRC-1 is necessary to repair the DSBs that could have not been resolved as COs: in fact, while in the syp-2 single mutant about 80% of diakinesis nuclei show 12 DAPI stained bodies corresponding to the 12 univalents, in the double mutant syp-2; brc-1 only a third of the nuclei have 12 DAPI stained bodies, while the large majority shows fragmented chromosomes at diakinesis (Adamo et al. 2008). We therefore assayed the number of diakinesis DAPI stained bodies in the double mutant syp-2; rad-51(A-). In this background most diakinesis nuclei show 12 DAPI stained bodies (Figure 2A) and a very small proportion (about 15%) shows chromosome fragmentation similarly to what observed in the syp-2; cep-1 germline (Figure 2B).
We can conclude that rad-51(A-) mutant is able to accomplish COs formation and inter-sister homologous repair during meiosis.
RAD-51 isoform A is necessary for DNA damage apoptosis in response to exogenously induced DSBs
In order to assay the sensitivity to exogenous damages in the rad-51(A-)mutant, we treated it with 120 Gy of γ-rays along with the wild type and the cep-1 strains. As observed for other phenotypes after γ-ray treatment, in the rad-51(A-) mutant RAD-51 foci appear at very high level from the mitotic tip of the gonad up to the end of the pachytene stage (Figure 3). However, unlike what occurs in wild type, in 51(A-). as well as in cep-1, no increase in apoptosis levels, compared to untreated nematodes, was observed (Figure 4A and Table S3). A possible explanation for this phenomenon is that the long isoform A, coded by the rad-51 gene, may be necessary to activate the DNA damage dependent apoptosis pathway. However, this is in contrast with what has been observed so far in experiments on gonads in which all the isoforms of the RAD-51 protein were absent. RNA-interference-mediated gene inactivation of the rad-51 locus, in fact, has been shown to induce, even in absence of gamma rays, a high level of DNA-damage-dependent apoptosis that can be suppressed by the cep-1 mutation (Gartner et al. 2000). Similarly, the rad-51(lg8701) mutant strain, carrying a large deletion of the locus and therefore defective for all the isoforms, also shows a high level of apoptosis (Alpi et al. 2003; Silva et al. 2013). We therefore tried to explore this contradiction. The rad-51(lg8701) mutant in homozygousis produce a 100% lethal progeny, therefore only the F1 progeny of heterozygous parents can be studied. We wondered if a relative small amount of maternal RAD-51 isoforms might be present in the adult gonad. We therefore performed a RAD-51 immunolocalization on the rad-51(lg8701) F1 progeny of heterozygous parents and detected a very low, but significant amount of RAD-51 foci in the gonad (Figure 4B). We hypothesised that the relative small amount of maternal RAD-51 isoforms B and C might be insufficient to repair the DNA damages, however the traces of RAD-51 isoform A may still allow the damage detection and the activation of the DNA damage dependent apoptosis pathway. In order to demonstrate that this small amount of maternal protein may be necessary and sufficient to activate the DNA damage dependent apoptosis, we assayed the apoptosis of the rad-5 (lg8701) homozygous progeny of rad-51 (lg8701);rad-51(A-) heterozygous hermaphrodite parents (see figure S3 for details of the genetic cross). These nematodes could have inherited the RAD-51 short isoforms B and C by maternal effect, but could not have inherited the long isoform A that has not been expressed for two generations. These nematodes, unlike what observed so far either in the rad-51RNAi or in the rad-51(lg8701) knockout strain at the F1, only show (low?)basal level of apoptosis supporting our model (Figure 4C and Table S3).
Ionizing radiation triggers the DNA damage checkpoint in the mitotic proliferative region of the rad-51(A-)gonad
The nuclei in the distal tip of the C. elegans gonad are engaged in mitotic proliferation. They respond to exogenous DNA damage triggering a temporary block in cell cycle progression to allow DNA repair. As a consequence of the cell cycle arrest, nuclei in the mitotic region of the germ line of irradiated worms appear enlarged and their overall number is reduced. In order to understand if the RAD-51 isoform A is involved in checkpoint activation or in apoptosis induction, we measured the nuclei diameters in the mitotic region before and after irradiation at 75 Gy. As shown in figure 5A, the rad-51(A-) mutant is proficient in activating the DNA damage response following IR. In fact, the average diameter of mitotic nuclei increases to twice their original size as is also observed in the wild-type control (Figure 5B).
Since the DNA damage checkpoint seems to be activated in irradiated rad-51(A-) gonads, we wondered whether failure in apoptosis execution might be associated with a delay in RAD-51 foci un(?)loading in late pachytene as observed in the ufd-2 mutant by Ackermann et al. 2016. We therefore performed RAD-51 immunolocalization experiments on nematodes that have been irradiated at 20 Gy, and we analysed the kinetics of RAD-51 disassemby on pachytene chromosomes one hour after IR. Both the wild type and the rad-51 (A-) mutant gonads show about 100% of the nuclei in late pachytene to be positive for RAD-51 foci. However, more than 90% of the late pachytene rad-51 (A-) mutant nuclei still shows RAD-51 foci 16 hrs after IR unlike what is observed in wild type, where about a third of the pachytene nuclei are devoid of RAD-51 foci. Eventually, 48 hrs after IR, in both strains less than 50% of the pachytene nuclei show detectable foci (Figure 5 C and D).
RAD-51 isoform A is necessary to induce apoptosis in response to endogenous DNA damages
The brc-1 mutant shows a permanence of RAD-51 foci in middle to late pachytene thought to represent stalled intersister homologous repair intermediates (Adamo et al. 2008 and Figure 6A). We analysed the RAD-51 foci appearance/unloading trend in the double mutant brc-1;rad-51(A-) and found that the level of RAD-51 foci in the double mutant is higher in the transition zone, but lower in pachytene compared to the brc-1 mutant control (Figure 6A and Table S2). Furthermore, the brc-1;rad-51(A-) double mutant shows six DAPI stained bodies at diakinesis similarly to the wildtype, the rad-51(A-) and the brc-1 single mutants (Figure 6B).
In absence of the synaptonemal complex, RAD-51 foci are present at high level on chromosomes during middle to late pachytene (Colaiacovo et al. 2003). The double mutant syp-2;rad-51(A-) shows a similar trend to the single mutant syp-2 (Figure S4).
Lack of the CEP-1 protein suppresses the apoptotic response observed in some meiotic mutants, such as brc-1 or syp-2 (Bhalla and Dernburg 2005; Adamo et al. 2008; Silva et al. 2013). Therefore, we investigated whether DNA damage induced apoptosis in meiotic mutants also requires rad-51(A-). We observe that brc-1;rad-51(A-) double mutant displays apoptotic levels similar to the wild-type controls and to the brc-1;cep-1 double mutant (see Figure 7A and Table S3). Furthermore, the level of apoptosis detected in the syp-2;rad-51(A-) double mutant is significant lower than that observed in syp-2 single mutant (Figure 7B and Table S3) and similar to the syp-2;cep-1 (and to that previously observed in syp-2;spo-11 and syp-2 msh-4 double mutants Silva et al. 2013), in which just the synapsis checkpoint is activated.
Therefore we can conclude that the RAD-51 isoform A is necessary for DNA damage dependent apoptosis in response to endogenous DNA damages.
Discussion
The recA-like protein RAD-51 is known to catalyze the strand transfer between homologous chromosomes or sister chromatids during meiosis and DNA repair. In most eukaryotes another recA like protein is also involved in this mechanism during meiosis. Such a protein is however missing in C. elegans. The C. elegans rad-51 gene is transcribed into three mRNAs potentially coding for alternative protein isoforms. In the course of our experiments we demonstrated that the RAD-51 longest isoform performs a different function from the shorter forms. In absence of the RAD-51 long isoform A, in fact, we observe correct chromosome segregation as no difference is observed in male frequency or embryonic lethality between rad-51(A-) mutant and the wild type (Figure 1 and Table 1). The only abnormal phenotype observed is larval arrest. Defects in germline proliferation and vulva development are significantly enhanced after gamma rays treatment. This observation suggests an involvement of the RAD-51 long isoform A in genome stability during larval development.
Petermann et al. 2010 have suggested that in mammalian cells restart of stalled replication forks and HR repair of collapsed replication forks require two distinct RAD51-mediated pathways. They demonstrated that stalled replication forks are efficiently restarted in a RAD51-dependent process that does not trigger homologous recombination (HR) suggesting that RAD51-mediated strand invasion supports fork restart while RAD51-dependent HR is triggered only in case of collapsed replication forks, without apparent restart. It is possible that RAD-51 isoform A may be involved during late development either in DNA repair or in restarting blocked replication forks.
The frequency and distribution of RAD-51 foci on chromosomes along the Meiosis I prophase in rad-51(A-) are indistinguishable from wild type. This seems to suggest that the isoform A is not preferentially loaded at the sites of SPO-11 induced DSBs (Figure 1B and Table S2).
In meiosis, the repair of DSBs by HR leads to the formation of COs and non crossovers(NCOs); during C. elegans oogenesis, only half of DSBs become COs: each chromosome pair has one (obligatory) CO which is required to lock the homologues together during the first meiotic division resulting in proper chromosome segregation. All the others DSBs, once the COs are produced, will be repaired using the sister chromatid as template. We have investigated on the possible involvement of the RAD-51 long isoform A during SPO-11 induced DSBs repair either in a genetic background in which the synaptonemal complex is missing (and therefore the sister chromatid is the only available template for HR) or in a genetic background where homologous DNA repair on the sister chromatid is defective. In either double mutant syp-2;rad-51(A-) and brc-1;rad-51(A-) the average number of diakinesis DAPI stained bodies are indistinguishable from those observed in the syp-2 and brc-1 single mutants. These observations furtherl support our hypothesis that the RAD-51 long isoform A is not involved in DNA repair during meiosis I.
RAD-51 long isoform A appears, instead, to have a fundamental role in inducing apoptosis following DNA damage. Following gamma rays treatment in rad-51(A-) mutant, as well as in wild type, the DNA damage checkpoint is activated (Figure 5A), and a large number of RAD-51 foci are loaded on chromosomes along the entire gonad (figure 3). Nevertheless, the level of germline apoptosis remains at basal levels as in rad-51(A-) or wildtype untreated hermaphrodites. These data apparently contradict what has been observed so far in the rad-51(lg8701) deletion mutant, where none of the RAD-51 isoforms are expressed (Alpi et al. 2003, Silva et al. 2013) and the level of apoptosis is at very high levels. However the rad-51(lg8701) deletion mutant produce 100% dead embryos and therefore only the first homozygous generation can be observed. We observed, by immunolocalization, that a small but significant amount of RAD-51 foci, likely of maternal origin, persists in the germline of the rad-51(lg8701) homozygous progeny of the rad-51(lg8701) IV/nT1 [let-?(m435)] (IV;V) balanced heterozygous hermaphrodites. Although the small amount of RAD-51 short isoforms is not sufficient to repair the SPO-11 induced DSBs, it is likely that the amount of RAD-51 long isoform A of maternal origin is sufficient to induce germline apoptosis. In support of such hypothesis we observed that the level of germline apoptosis drops to wild type levels in the rad-51(lg8701) homozygous progeny of a rad-51(lg8701); rad-51(A-) heterozygous hermaphrodites (figure 4C and S3) where the RAD-51 long isoform A has been absent for two generations.
In rad-51(A-) as in ufd-2 mutants RAD-51 foci have a delay in the kinetics of disassembly after gamma ray treatment. It has been hypothesized that the RAD-51 disassembly itself may trigger the induction of apoptosis (Ackerman et al. 2016). However, we cannot exclude that the delay in RAD-51 disassembly may be a consequence and not the cause of apoptosis induction. In fact, all the apoptotic corpses, not only those due to DNA damage, but even those due to physiological apoptosis, always appear overloaded with RAD-51 foci. Furthermore, in many meiotic mutant RAD-51 foci accumulate in late pachytene. They persist, for example, both in msh-4/5 mutants, where DNA damage dependent apoptosis is blocked, and in brc-1 and syp-2 mutants where high level of damage induced apoptosis is present.
The RAD-51 long isoform A is, in fact, also involved in inducing apoptosis in response to endogenously induced DNA damage such as in DNA repair defective mutants. We previously demonstrated that in the brc-1 mutant DNA repair on the sister chromatid is defective in meiosis, that a high level of HR intermediates accumulate in late pachytene, and that DNA damage apoptosis is induced (Adamo et al. 2008). In the brc-1;rad-51(A-) double mutant the level of apoptotic corpses drops to the physiological wild type level and RAD-51 foci (representing HR intermediates) do not accumulate in late pachytene chromosomes as they do in the brc-1 single mutant. Therefore, in absence of rad-51(A-), DNA damage dependent apoptosis is not induced in a brc-1 background.
In the syp-2 mutant, the synaptonemal complex is not present and COs are not formed. Although SPO-11 induced DSBs will be eventually repaired using the sister chromatid as template, RAD-51 foci accumulate on late pachytene chromosomes (Colaiacovo et al. 2003) and both the pairing checkpoint and the DNA repair checkpoint are activated with a significant increase in germline apoptosis (Bhalla and Dernburg 2005). Also in this genetic background, cep-1 dependent apoptosis disappears in absence of the RAD-51 long isoform A and only the physiological and the pairing checkpoint dependent corpses persist.
We conclude that the RAD-51 long isoform A has a key role in the induction of apoptosis downstream of the activation of the DNA damage checkpoint. Although DNA damage and repair defects can be present along the whole gonad, apoptotis is allowed to start only just before diplotene stage. Therefore, a complex pattern of interactions seems to be active. Further studies will be required in C. elegans to understand the cross-talking among the genes involved in the meiotic apoptotic induction including, beside rad-51, also msh-4/5, zhp-3 (Silva et al. 2013), and ufd-2 (Ackermann et al. 2016) and to identify the specific physical interactions and steps of the apoptotic response at which each gene is required.
Acknowledgements
We thank Adriana La Volpe, John Pulitzer for useful discussions and critical reading of the paper. Some nematode strains used in this study were provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources (NCRR). We thank Monica Colaiacovo for providing a strain. This study was started as part of the Doctoral thesis of Marcello Germoglio in Biomolecular Science, at the Second University of Naples, who was the recipient of a predoctoral fellowship from Telethon Grant GGP11076.