In order to examine the role of the 9-1-1 complex during mammalian meiosis, we analyzed mice with germ cell-specific Hus1
deletion and evaluated the localization of the RAD9 and RAD1 subunits to normal and aberrant meiotic chromosomes. Most notably, we report that the RAD9 and RAD1 subunits of the 9-1-1 complex exhibit overlapping yet distinct localization patterns along mammalian meiotic chromosomes and are differentially affected by loss of the HUS1 subunit. RAD1 forms more foci along chromosome cores than RAD9 (this study) or RAD51 
, and we show that it preferentially localizes to the axial elements of asynapsed chromosomes or autosomes included within the sex body domain in a manner that is independent of Hus1
. Previous electron microscopy analysis indicated that while RAD1 localizes to meiotic chromosomes cores, it does not colocalize with DMC1 
, which is consistent with our finding of RAD1 at asynaptic sites. We show here that only a subset of RAD1 foci colocalize with RAD9, raising the possibility that RAD1 is present both at DSBs, as part of the canonical 9-1-1 complex, and at other chromosomal sites.
We propose a model in which meiotic genome maintenance involves several distinct checkpoint clamp complexes, some of which function in DSB repair, and others of which carry out separate functions, possibly including ATR activation through TOPBP1 in response to asynapsis. Of particular note are the previous findings that the mouse and human genomes contain paralogs of RAD9 and HUS1, termed RAD9B and HUS1B, respectively, that are highly expressed in the testis 
. Given that RAD1 localized to some chromosomal sites that lacked RAD9 and did so following genetic ablation of Hus1
, it is tempting to speculate that RAD9B and HUS1B associate with RAD1 to form an alternative 9-1-1-like heterotrimer, although it remains possible that RAD1 localizes to chromosomes on its own or in conjunction with other cofactors. In this regard, it is worth noting that a substantial portion of RAD1 protein in cultured human cells exists in monomeric form, independent of the heterotrimeric 9-1-1 complex 
. Overall, our results are consistent with formation of the canonical 9-1-1 complex composed of RAD9, RAD1, and HUS1 at meiotic DSBs, as well as with RAD1 assembling on chromosomes independently of HUS1 and RAD9 in patterns overlapping with TOPBP1 and ATR.
Despite a well-established role for 9-1-1 in ATR activation in somatic cells, we observed unperturbed localization of the 9-1-1-interacting, ATR-activating TOPBP1 protein and similarly no change in RNA Polymerase II exclusion from the XY body domain in Hus1
CKO spermatocytes. These findings indicate that ATR-dependent responses to unsynapsed chromatin, including meiotic silencing via MSUC and MSCI, do not require HUS1, which is in distinct contrast to the canonical model of mammalian ATR activation in which 9-1-1-mediated TOPBP1 recruitment to damaged DNA allows TOPBP1 to contact and activate ATR. Since RAD1 localization to asynapsed meiotic cores was also unperturbed in the absence of Hus1
, it is possible that a RAD1-containing complex functions along with TOPBP1 to activate ATR at such sites, in a manner that may be at least in part distinct from established mechanisms for 9-1-1 loading and function at DNA damage sites. The meiosis-specific HORMAD1 and HORMAD2 proteins are required for ATR recruitment to sites of unsynapsed chromatin 
, and together with BRCA1, which is also required for ATR recruitment 
, these proteins could provide an alternative mechanism for ATR activation that is independent of 9-1-1 and typical substrates for 9-1-1 loading. We favor the idea that alternative complexes involving RAD1 and possibly RAD9B and HUS1B might engage in a 9-1-1-like interaction with TOPBP1. It remains to be determined whether the recruitment of such a complex to asynaptic sites is facilitated by particular DNA structures or binding partners that would not attract conventional 9-1-1 complexes.
deletion in germ cells resulted in numerous chromosomal defects, including asynapsis, unrepaired meiotic DSBs, X-autosome end-to-end fusions, autosomes incorporated into the sex body domain (which presumably are inappropriately transcriptionally silenced), and SC ruptures. We propose that some of these defects (asynapsed chromosomes, X-autosome fusions, and autosomes incorporated into the sex body domain) lead to loss of cells during the pachytene stage in Hus1
CKOs, consistent with an absence of cells with such abnormalities in diplotene (), whereas other defects (unrepaired DSBs, extended sex body domains, and SC ruptures; ) either may not be actively monitored or may normally require HUS1 for checkpoint-dependent clearance and therefore persist beyond pachytene into diplotene in Stra8-Cre Hus1
CKOs. The latter set of phenotypes was less prominent in Spo11-Cre Hus1
CKO mice, with fewer defective cells persisting into diplotene, suggesting that HUS1-dependent checkpoint monitoring may be an earlier function that is less affected by mid-prophase I loss of Hus1
. Notably, SC ruptures were still prominent in diplotene cells of both Stra8-Cre
and Spo11-Cre Hus1
CKOs, affecting one-third of the diplotene cell population. Thus, HUS1 is required for some aspect of chromosome integrity that affects the SC, the failure of which may contribute to the high level of metaphase germ cell loss. This SC rupture phenotype (; ) is particularly striking and may be a key driver of the fertility defect observed following Hus1
loss, as no significant change was observed in the number of MLH1 foci in Hus1
mutant pachytene spermatocytes or in the number of bivalents seen at diakinesis (Figure S6
). The molecular cause of the SC rupture defect remains unresolved, although it could be related to incomplete repair of meiotic DSBs.
It is still unclear why Hus1
CKO cells have a small number of persistent DSBs, and why many of these are associated with death at metaphase I. Perhaps the 4–8 remaining unrepaired breaks are not sufficient to trigger the pachytene checkpoint (or cannot, due to a requirement for HUS1), but lead to a defect that is recognized at metaphase by the spindle checkpoint. It is possible that the persistent breaks are not simple D-loop intermediates containing RAD51 and γH2AX, but are complex multi-chromatid intermediates, such as those seen in Blm/Sgs1
helicase mutants 
. The symmetrical RAD51 foci seen on either side of the chromosome cores in 9% of diplotene nuclei from Hus1
CKOs (see , asterisks) would support this, pointing to RAD51/DMC1 presence on more than one homolog during homolog separation. Clearly, the extent of the DNA repair defect in Hus1
CKO mice is less severe than that observed in other meiotic mutants. For instance, Trip13
mutant spermatocytes have been reported to exhibit between 99 and 138 RAD51 foci during pachytene compared to 11–18 foci in controls 
, whereas we observed an average of 25 RAD51 foci in early to mid-pachytene Hus1
CKOs (compared to 22 foci in controls) and 9 RAD51 foci in late pachytene (compared to 2 foci/nucleus in controls). These results are consistent with the idea that HUS1 has a relatively late and restricted role in DSB repair, perhaps related to the completion of a subset of late recombination events.
The persistent γH2AX and RAD51 foci on Hus1-
deficient meiotic chromosomes as well as the colocalization of RAD9 with RAD51 on normal chromosomes suggest that the 9-1-1 complex is critical for the efficient repair of a subset of meiotic DSBs. Several lines of evidence in somatic cells also indicate that 9-1-1 functions directly in homologous recombination (HR) repair of DSBs. In human cells, 9-1-1 is reported to physically interact with RAD51, and Rad9
knockdown results in reduced HR 
. In a separate system using human cells with conditional Rad9
repression, RAD9 enhances survival and DNA repair in response to ionizing radiation 
, and similarly, mouse Rad9−/−
ES cells are sensitive to γ-irradiation 
. Reducing Hus1
expression in mouse cells via siRNA also decreases the efficiency of HR repair 
. In contrast to the situation in yeast where 9-1-1 is proposed to be important for the loading or assembly of RAD51 complexes onto meiotic chromosomes 
, we propose that mammalian 9-1-1 may be important for later steps of meiotic recombination, since we did not observe delayed loading of RAD51, and RAD51 foci persisted in the absence of Hus1
. Among the known 9-1-1 binding partners are DNA ligase I 
and DNA Polymerase β 
, the latter of which is known to play critical roles during mammalian meiosis 
, raising the possibility that 9-1-1 may recruit Pol β and DNA ligase to recombination intermediates to complete repair. Alternatively, interactions between the 9-1-1 complex and translesion synthesis polymerases 
, which have been implicated in HR in somatic cells 
, could promote the extension of the 3′ ends subsequent to RAD51-mediated strand exchange. The requirement for 9-1-1 could be accentuated due to unique features of genome maintenance in meiotic cells. For instance, non-homologous end joining, a major mechanism for DSB repair in somatic cells, is suppressed during meiosis 
. In addition, the unique structure of SPO11-induced meiotic DSBs may create a greater demand for 9-1-1 complex-mediated repair functions than a typical mitotic DSB. The repair of most meiotic DSBs occurred normally in the absence of HUS1, suggesting that the 9-1-1 complex may be necessary to deal with only a subset of breaks, perhaps ones that prove difficult to repair because of the sequence context, chromatin structure, or other factors.
The production of haploid gametes during meiosis clearly raises challenges for genome maintenance, many of which are distinct from those in somatic cells. Based on the findings reported here, we propose that mammalian 9-1-1 components have acquired specialized roles during meiosis, with the canonical RAD9-RAD1-HUS1 complex functioning at DSBs and an alternative RAD1-containing complex functioning at sites of asynapsis. The mouse models described here represent powerful systems to elucidate how the mammalian 9-1-1 complex promotes meiotic chromosome integrity, in some cases distinct from the well-established roles of 9-1-1 in TOPBP1- and ATR-dependent checkpoint signaling, and highlight the intriguing possibility of alternative checkpoint clamps functioning in various capacities in the mammalian germline.