COM-1–deficient germ cells bear chromosomal aggregates and univalents
In order to study the meiotic functions of COM-1 we obtained two different com-1
mutant alleles previously identified by Penkner and colleagues (Figure S1
) and 
. In C. elegans
, defects in repair of meiotic DSBs can be detected relatively easily, as these often manifest as chromosomal abnormalities in diakinesis nuclei of maturing oocytes (). Wild-type diakinesis nuclei typically have six rod-shaped DAPI-stained bodies named bivalents, which represent the six pairs of homologous chromosomes, each held together by chiasmata ( and ). In the absence of meiotic DSBs (e.g. in spo-11
mutants) chiasmata are not formed, which can be detected by the presence of 12 DAPI-stained bodies, i.e.
. When meiotic DSBs are induced but not repaired, chromosomal fragmentation occurs, typically resulting in ≥12 irregularly shaped DAPI-stained bodies at diakinesis 
. Surprisingly, com-1
mutant oocytes exhibited a different chromosomal pattern: the diakinesis nuclei contained 1 to 12 DAPI-stained entities 
. We validated this finding by careful inspection of COM-1-deficient diakinesis nuclei (). These diakinesis nuclei occasionally showed chromosomal fragments, albeit only in 2% of the oocytes (). We argued that the low frequency of chromosomal fragmentation in com-1
mutants is inconsistent with a conserved role for COM-1 in SPO-11 removal, given that SPO-11-bound DSBs are refractory to repair. Based on the diakinesis studies we envisaged a different scenario in which com-1
mutants are able to repair meiotic DSBs, yet do so in an error-prone manner, ultimately resulting in chromosomal aggregates and failed chiasmata formation. Several observations supported this hypothesis: Firstly, unlike spo-11
mutant oocytes hardly ever contained exactly 12 univalents, which indicated that DSBs were induced. Secondly, all diakinesis nuclei had fewer than 12 DAPI-stained bodies and rarely contained small chromosomal fragments, arguing that most programmed DSBs are repaired. Thirdly, the diakinesis nuclei often contained more than 6 DAPI-stained bodies and frequently exhibited DAPI bodies that morphologically resembled univalents, which implied that chiasmata formation was impaired. Finally, many diakinesis nuclei had fewer than six DAPI-stained bodies, potentially reflecting chromosomal entanglements and/or fusions between non-homologous chromosomes.
Loss of cku-80 prevents chromosomal aggregation and restores chiasmata formation and embryonic survival in com-1(t1626) mutants.
Loss of cku-80 as well as γ-irradiation rescues the CO defect of com-1 mutants.
Loss of lig-4 prevents chromosomal fusion in com-1 mutants, but does not restore viability.
Loss of Ku restores chiasmata formation and viability in com-1 mutant animals
To test if the chromosomal aggregation events in com-1 mutant oocytes were due to inappropriate NHEJ activity, we crossed com-1 mutants with worms lacking the NHEJ factor CKU-80. Strikingly, cku-80 deficiency led to a >20 fold increase in viability among com-1 mutant progeny: while com-1 single mutants produced 0–2% viable embryos, com-1 cku-80 double mutants produced 30–40% viable progeny (). Moreover, nearly all hatchlings of com-1 cku-80 double mutants successfully developed into adults, while com-1 single mutant hatchlings typically died as arrested L1/L2 larvae.
To verify these observations we crossed animals carrying another allele of com-1
to worms lacking the other well-conserved Ku subunit CKU-70. The resultant com-1(t1489) cku-70
double mutants showed identical phenotypes as the aforementioned com-1(t1626) cku-80
double mutants, including elevated embryonic survival and restored larval development as compared to com-1(t1489)
single mutants (Figure S1
). We therefore conclude that com-1
deficient animals suffer from toxic Ku activity and that in the absence of Ku, COM-1 is dispensable for C. elegans
development and gametogenesis.
In contrast to the diakinesis nuclei of com-1
single mutants, which hardly ever contain six DAPI-stained bodies, 70% of diakinesis nuclei of com-1 cku-80
double mutants had the wild-type set of six bivalents (). We obtained similar results for com-1 cku-70
double mutants (Figure S1
). The fact that Ku deficiency restored bivalent formation in com-1
mutant animals implies that both the univalents and the chromosomal aggregates in com-1
deficient oocytes were due to Ku-mediated NHEJ. These observations also demonstrate that COM-1 is not required for chiasma formation per se
. Notably, both bivalent formation and embryonic viability in com-1 cku-80
double mutants were completely spo-11
-dependent (), which indicates that chiasma formation in com-1
mutants occurs at programmed DSBs and not at spontaneous DSBs.
Based on these diakinesis studies we conclude that i) COM-1 is crucial to prevent NHEJ activity in meiotic cells; ii) Ku can act efficiently on meiotic DSBs (at least when COM-1 activity is perturbed); iii) a com-1-independent mechanism exists that is able to convert SPO-11-induced DSBs into proper chiasmata, and iv) in contrast to Sae2/Ctp1 in yeast, COM-1 is not required for SPO-11 removal in C. elegans.
Ku prevents CO formation in com-1 mutant germlines
single mutants fail to adequately form chiasmata and this defect can be restored by Ku loss (), we reasoned that Ku might obstruct CO formation. In C. elegans
, exactly one CO occurs per homolog pair and these presumptive CO sites can be visualized by specific recruitment of the fusion protein ZHP-3::GFP at late pachytene/diplotene stage 
. As shown in , wild-type animals had six ZHP-3::GFP foci in nearly all diplotene nuclei. In contrast, com-1
single mutants on average had only two ZHP-3::GFP foci per diplotene nucleus () and often exhibited persistent ZHP-3::GFP localization along the full length of the synaptonemal complex (SC) – a localization pattern characteristic of CO failure 
Importantly, loss of cku-80 alleviated the ZHP-3::GFP localization defect of com-1 mutant germlines: virtually all diplotene nuclei of com-1 cku-80 double mutants had the normal complement of six ZHP-3::GFP foci (). We conclude that COM-1 is not needed for CO formation per se, yet COM-1 is essential to prevent interference by Ku and hence is critical for CO assurance.
The CO defect of com-1 mutants is due to a scarcity of accessible DSBs
We hypothesized that Ku binds DSB ends and blocks DNA end resection and subsequent meiotic recombination. In order to test if the CO defect observed in com-1 mutants is due to an insufficient number of DSBs available for HR, we subjected these animals to ionizing radiation (IR) to introduce additional DSBs. 70 Gy of IR did not alter the number of COs in wild-type animals: six ZHP-3::GFP foci were present per diplotene nucleus, irrespective of IR treatment (). Strikingly, 70 Gy of IR substantially increased CO formation in com-1 mutant animals: while mock-treated com-1 mutants had on average only two ZHP-3::GFP foci per diplotene nucleus, irradiated com-1 mutants commonly contained six foci ().
Previous studies have shown that IR can increase CO frequencies only when meiotic DSBs are limiting, e.g.
. This effect is attributed to CO homeostasis mechanisms that ensure that meiotic cells receive at least one and only one CO per homolog pair 
. Our results imply that in the absence of com-1
CO homeostasis mechanisms are still active and encourage the formation of the obligate COs, yet the substrates to do so are limited. A recent dose-response study estimated that 10 Gy of IR resulted in ~4 DSBs per chromosome pair, which was sufficient to consistently induce six CO foci in spo-11
. We exposed com-1
mutants to 10 Gy, 50 Gy and 70 Gy of IR and found that only 70 Gy resulted in a robust induction of six ZHP-3::GFP foci ( and Figure S2
). The observation that 10 Gy of IR was not sufficient to induce six CO foci in com-1
mutants, suggests that Ku can also hijack SPO-11-independent DSBs. In support of this notion, IR resulted in increased levels of chromosomal aggregation in com-1
deficient oocytes () and 
. Given the relatively high IR dose needed to allow six CO foci to be formed in COM-1-deficient animals, we propose that IR alleviates the CO defect, not because it introduces SPO-11-independent DSBs, but because it can introduce a total number of DSBs that exceeds the capacity of available Ku, leaving a subset of DSBs unblocked and available for HR.
We conclude that both IR treatment and Ku deletion alleviated the CO deficit in com-1 mutant animals, yet only Ku deletion restored the bias towards HR-mediated DSB repair.
Loss of LIG-4 does not restore viability of com-1 mutants
mutant animals Ku causes two problems: defective CO formation and chromosomal aggregation. We next set out to determine how Ku exerts these toxic effects. In classical NHEJ, Ku blocks DNA end resection, stabilizes the break ends and recruits the downstream factor LIG-4, which subsequently seals the DSB 
. To assess if the Ku complex could be toxic independent of LIG-4-mediated ligation, we made com-1 lig-4
double mutants and compared those to com-1 cku-80
and com-1 cku-70
double mutants. Interestingly, unlike cku-70
, the introduction of a lig-4
null allele did not rescue progeny survival of com-1
mutants (). Since either lig-4
loss prevents NHEJ, blocking NHEJ per se
is not sufficient to restore viability in com-1
mutants. We therefore infer that Ku has toxic activities that are independent of NHEJ-mediated fusion.
Consistent with that notion, diakinesis nuclei of com-1 lig-4 double mutants often showed more than six DAPI-stained bodies, indicating that CO formation remained perturbed (). While lig-4 deletion did not restore the CO defect, it did prevent chromosomal aggregation: in contrast to com-1 single mutants, the diakinesis nuclei of com-1 cku-80 and com-1 lig-4 double mutants rarely had fewer than six DAPI-stained bodies ( and ). These observations indicate that chromosomal aggregation in com-1 mutants mainly depends on classical NHEJ.
Notably, diakinesis nuclei of com-1 lig-4 double mutants frequently contained small DAPI-stained fragments, which are indicative of persistent DSBs (). We next established that these chromosomal fragments were the consequence of defective repair of programmed SPO-11-induced DSBs (and not of spontaneous DSBs): com-1 lig-4 spo-11 triple mutant animals exhibited 12 intact univalents at diakinesis and no fragmentation (). Together, these results strongly suggest that in COM-1-deficient animals, Ku promotes LIG-4-mediated fusions and that in the absence of LIG-4 the Ku-bound DSBs remain unrepaired. We therefore propose that COM-1 needs to prevent Ku activity not only because Ku promotes classical NHEJ at meiotic DSBs, but mainly because Ku forestalls meiotic recombination directly.
Ku acts at early/mid pachytene stage and blocks the formation of RAD-51 foci
We next determined how and when Ku prevents meiotic recombination. Based on their homologous counterparts, we expect CKU-70/CKU-80 to block DNA end resection. This scenario is consistent with the reported defect in RAD-51 recruitment in COM-1-deficient germlines 
. Meiotic recombination is initiated in the transition zone where RAD-51-coated recombination intermediates become visible as distinct foci 
. In wild-type worms, the number of RAD-51 foci peaks at early/mid pachytene stage (, zone 4+5) and as repair progresses, these RAD-51 foci disappear by late pachytene stage (, zone 6+7) 
. In com-1
single mutants, however, we could not detect the typical rise of RAD-51 foci in early/mid pachytene nuclei, suggestive of a defect early in meiotic recombination (, zone 4+5). Strikingly, this defect was relieved by cku-80
loss: com-1 cku-80
double mutants did show the strong increase in RAD-51 foci at early/mid pachytene stage (, zone 4+5). These results demonstrate that, in the absence of COM-1, CKU-80 prevents efficient formation of RAD-51-coated HR intermediates, likely by inhibiting DNA end resection. Moreover, they reveal that CKU-80 can already act at early pachytene stage, which paradoxically is the stage where programmed DSBs need to be channeled into HR.
Loss of cku-80 restores RAD-51 recruitment to meiotic DSBs in com-1 mutant germlines.
While com-1 cku-80 double mutant germlines were proficient in RAD-51 loading, we noted a mild delay in RAD-51 focus formation compared to cku-80 single mutant controls (, zone 4+6). This delay suggests that COM-1 may also be required for efficient DNA end resection and thus the timely formation of interhomolog COs.
COM-1 and EXO-1 act redundantly to promote meiotic recombination
To find the factors responsible for COM-1-independent meiotic recombination, we searched for genes known to have overlapping functions with COM-1 or its homologs. In yeast, the sensitivity of Sae2-deficient mitotic cells to DSB-inducing agents can be rescued by overexpressing the 5′-3′ exonuclease Exo1 
. Furthermore, Exo1 transcription is highly induced during yeast meiosis and Exo1 promotes CO formation 
, making Exo1 a suitable candidate for enabling com-1
-independent CO formation.
A clear Exo1 homolog is present in C. elegans
, F45G2.3, which we named exo-1
. We used a deletion mutant of exo-1
, which is predicted to express a severely truncated protein lacking the conserved nuclease domain (), to show that EXO-1 has a conserved role in HR-mediated DSB repair in germ cells. Firstly, exo-1
mutant germlines were hypersensitive to IR, in a manner epistatic with the well-studied HR factor brc-1
() and secondly, exo-1
mutants were hypersensitive to transposon-induced DSBs, i.e. exo-1
deficiency significantly reduced embryonic survival in animals that have elevated levels of transposition in the germline (Figure S3
). Despite the need for exo-1
in repair of ectopic DSBs, unchallenged exo-1
single mutants did not display major meiotic defects (), which suggests that EXO-1 does not act on SPO-11-induced DSBs or it operates in a redundant fashion.
EXO-1 promotes DSB repair in germ cells.
To assess if EXO-1 is responsible for COM-1-independent meiotic recombination, we created com-1 cku-80 exo-1 triple mutants and analyzed CO formation and progeny survival. In contrast to com-1 cku-80 double mutants, which have robust CO formation (), com-1 cku-80 exo-1 triple mutants fail to adequately produce COs, as illustrated by the scarcity of ZHP-3::GFP foci at diplotene () and the lack of chiasmata at diakinesis (). Consequently, com-1 cku-80 exo-1 animals typically produce aneuploid gametes and hardly any viable progeny ().
EXO-1 is required for meiotic recombination in absence of COM-1.
We next investigated how EXO-1 promotes CO formation in com-1
deficient germlines. Recently, yeast Exo1 has been shown to promote CO formation via two distinct activities: i) by performing DNA end resection and ii) by resolving CO intermediates named double Holliday Junctions (dHJs) 
. These two Exo1 activities affect HR at different steps: DNA end resection promotes the formation of RAD-51 intermediates, whereas dHJ resolution supports the clearance of RAD-51 intermediates. We found that early/mid pachytene nuclei of com-1 cku-80 exo-1
triples contained hardly any foci (), which contrasts the many RAD-51 foci observed in com-1 cku-80
double mutants (). This implies that EXO-1 promotes com-1
-independent CO formation mainly via its role in DNA end resection.
From these results it can be deduced that i) EXO-1 can act on meiotic DSBs and ii) EXO-1 and COM-1 act in parallel pathways to promote RAD-51 recruitment at early/mid pachytene stage and individually can assure timely CO formation. Furthermore, both COM-1 and EXO-1 are not essential for SPO-11 removal because we did not observe substantial chromosome fragmentation in the diakinesis nuclei of com-1 cku-80 exo-1 triple mutants. Instead, we detected six to twelve regularly shaped DAPI-stained bodies (), which suggests some degree of DSB repair.
Homolog-independent HR does not depend on COM-1 and EXO-1
germ cells switch between different DSB repair modes as they progress through meiosis 
. In the early stages of meiotic prophase, the majority of meiotic DSBs are repaired using the homologous chromosome as a template 
. At late pachytene stage this dominance is thought to be relieved, allowing homolog-independent mechanisms to repair the meiotic DSBs 
. One example that supports this notion is that mutant animals defective in interhomolog HR (e.g. syp-2
mutants) show persistent meiotic DSBs that are eventually repaired late in meiotic prophase in a rad-51
-dependent manner 
. Subsequent studies suggest that these remaining DSBs are repaired efficiently via intersister HR, ultimately giving rise to intact chromosomes at diakinesis 
To investigate the contribution of COM-1 and EXO-1 to homolog-independent HR, we quantified RAD-51 focus formation throughout the germline. com-1 cku-80 double mutants had many RAD-51 foci at early/mid pachytene stage (, zone 4+5), but very few RAD-51 foci at late pachytene stage (, zone 7), indicating that the majority of RAD-51 intermediates were resolved by that point. Conversely, com-1 cku-80 exo-1 triple mutant germlines had very few RAD-51 foci at early/mid pachytene stage (, zone 4+5), but showed many RAD-51 foci at late pachytene stage (, zone 7). This abundance of RAD-51-coated recombination intermediates at late pachytene implies that COM-1 and EXO-1 are dispensable for DNA end resection at these later stages, which suggests further redundancy and/or temporal regulation of DNA end resection during meiotic prophase. Moreover, these findings imply that intersister HR may not be affected by com-1 and exo-1 loss.
EXO-1 and COM-1 are needed for efficient interhomolog HR, but dispensable for intersister HR.
To test if intersister HR is responsible for the residual repair activity in the triple mutant, we depleted the cohesin factor REC-8, which is proposed to promote both interhomolog as well as intersister HR 
. REC-8 depletion caused extensive chromosomal fragmentation in com-1 cku-80 exo-1
triple mutants (), implying that REC-8-dependent HR is active in the absence of COM-1 and EXO-1. REC-8 depletion, however, has documented pleiotropic effects, including altered SPO-11 activity, which may affect the levels of chromosome fragmentation 
. We therefore substantiated these findings by deleting Structural Maintenance of Chromosomes 5 (smc-5)
in com-1 cku-80 exo-1
has recently been shown to be specifically required for homolog-independent (presumably intersister) HR during C. elegans
. Analogous to REC-8 depletion, deletion of smc-5
in com-1 cku-80 exo-1
triple mutants resulted in high levels of chromosome fragmentation at diakinesis (). Similar results were obtained when deleting the SMC-5 complex partner SMC-6 (Figure S3
). Together these observations strongly suggest that, while COM-1 and EXO-1 redundantly promote RAD-51 recruitment and subsequent CO formation at early/mid pachytene stage, at late pachytene stage both proteins are dispensable for RAD-51-mediated intersister HR.
Ku deficiency does not fully restore genome stability in com-1 mutants
Despite the observation that HR is active and COs are formed in germlines lacking both com-1
, progeny survival of com-1 cku-80
double mutants was not restored to wild-type levels. In fact, ~70% of com-1 cku-80
double mutant progeny died during embryonic development (). Moreover, the mutant animals that survived frequently displayed developmental abnormalities, including altered body morphology and faulty vulval development (Figure S4
). These phenotypes suggest that Ku-deficient com-1
mutants still suffered from genomic instability. In support of this notion, com-1 cku-80
and com-1 cku-70
double mutants exhibited high levels of X-chromosome non-disjunction, as revealed by a 50-fold increase in XO males among the surviving progeny (Figure S4
). Careful analysis of com-1 cku-80
deficient germlines revealed that the fidelity of meiotic DSB repair is incomplete: diakinesis nuclei of com-1 cku-80
double mutants occasionally showed chromosomal abnormalities, including unstable bivalent attachments and chromosomal aggregates ( and Figure S5
). We detected similar chromosomal aberrations in com-1 lig-4
double mutants (Figure S5
), supporting the notion that an alternative mutagenic repair pathway exists that can provoke chromosomal aggregates in germ cells devoid of classical NHEJ 
. We propose that Ku-deficient com-1
mutants still suffer from (NHEJ-independent) error-prone repair events, which cause substantial chromosomal instability and embryonic lethality.
We next addressed whether these aberrant repair events in com-1 cku-80
double mutants induced germ cell apoptosis. Interestingly, despite the high degree of chromosomal instability, the level of apoptosis was not observed to be increased in com-1
single mutant germlines 
. Although we cannot formally exclude that COM-1 by itself is required for the signaling of apoptosis, our cytological data argue that Ku blocks end resection in these animals and thus precludes the formation of ssDNA - a major trigger for the DNA damage checkpoint 
. To test this hypothesis further, we counted apoptotic cells, marked by transgenic CED-1::GFP, in com-1 cku-80
deficient germlines. We observed a mild but statistically significant increase as compared to com-1
single mutants (Figure S4
). This result may reflect inefficient repair of a fraction of DSBs in com-1 cku-80
double mutants, as was also suggested by the subtle delay in RAD-51 focus resolution during meiotic prophase (, zone 6). These phenotypes are however very mildly different from wild-type behavior 
. We thus conclude that the vast majority of meiotic DSBs are repaired effectively in com-1 cku-80
mutant germ cells, without activating the DNA damage checkpoint. The fidelity of repair, however, is clearly affected by com-1