Shu complex SWS1-SWSAP1 promotes early steps in mouse meiotic recombination

The DNA-damage repair pathway homologous recombination (HR) requires factors that promote the activity of strand-exchange protein RAD51 and its meiosis-specific homolog DMC1. Here we show that the Shu complex SWS1-SWSAP1, a candidate for one such HR regulator, is dispensable for mouse viability but essential for male and female fertility, promoting the assembly of RAD51 and DMC1 on early meiotic HR intermediates. Only a fraction of mutant meiocytes progress to form crossovers, which are crucial for chromosome segregation, demonstrating crossover homeostasis. Remarkably, loss of the DNA damage checkpoint kinase CHK2 rescues fertility in females without rescuing crossover numbers. Concomitant loss of the BRCA2 C terminus aggravates the meiotic defects in Swsap1 mutant spermatocytes, suggesting an overlapping role with the Shu complex during meiotic HR. These results demonstrate an essential role for SWS1-SWSAP1 in meiotic progression and emphasize the complex interplay of factors that ensure recombinase function.

In general, the work is extensive and well presented. The story will be of interest, not only because the role of these proteins is elucidated, but also because the unique phenotypes may be useful for other investigations, for example in checkpoint control (as they did here), and the role of suboptimal genomic integrity in reproduction and embryonic viability (which they developed to a lesser extent). The overall impact is viewed to be above average, as is the quality of this manuscript.  Fig. 3h,I for Mlh1 focus counts. The latter is bad, given the relatively small difference in average foci, and the fact that the mice were not fully inbred ("backcrossed for 3-6 additional generations"); there are known to be genetic differences in Mlh1 foci in different inbred strains. This shortcoming cannot be overcome by simply scoring more spreads from a single mouse.
-the basis for the smaller litter size in Chk2-/-Swsap1-/-animals should be explored. Mice with smaller oocyte pools may ovulate identical numbers (at least in early life), but may suffer from aneuploidy (associated with reduced crossovers), leading to embryonic lethality. The experiment would be relatively easy, since embryos and ovulation sites can be quantified in the same litters. This is likely, actually, and would increase the paper's impact if true. Can the authors elaborate on the effectiveness of the spindle checkpoint in general, and whether it comes into play at all in these mutants. Additionally, the Chk2-/-rescue results are only superficially interpreted. The authors state, "The lack of a complete rescue with the loss of CHK2 is likely due to the persistence of synapsis defects (open circles; Fig. 4e) and/or reduced MLH1 focus numbers ( Fig. 4e and Supplementary Fig.   5b)." The implications of this statement will not be obvious to everyone. Indeed the latter is probably incorrect, because MLH1 deficiency doesn't lead to elimination of oocytes prior to ovulation.
-The authors use MEIOB foci and H2AX foci to conclude that the Shu mutants form normal numbers of DSBs. Probably true, but a Spo11 oligo assay would be a very convincing orthogonal assay.
-Are their two mutants really the same? They are in a complex together, but they are used interchangeably here, and in certain figures there are differences between the two. What does this imply? If they are not the same, why weren't double mutant experiments performed for the Sws genes as well (combined with Chk2 and Brca2 C term).

Minor issues:
-The abstract fails to reflect the significance and context of this paper. To anyone but an afficionado, the Shu complex and BRCA2 C terminus need an explanation. I realize that there may be a word limit in the abstract, but it is important to craft it so diverse scientists can understand the implications of the work.
-Final sentence in intro, last phrase, is uninformative.
-The use of MLH1 as equivalent for crossovers is a bit misleading; yes; it's believed that MLH1 marks crossovers but to say it is a crossover isn't entirely appropriate -metaphase spreads to detect crossovers would be needed to state this definitely.
-p. 5, the statement that "Unlike later stages, early meiotic prophase cells at leptonema and zygonema are increased in Shu single-and double-mutant mice. " is misleading. I think the authors may mean that the fraction of early meiotic cells, as a proportion of all cells in mutant tubules (which lack postmeiotic stages), is higher than wild type. As stated, it sounds like they have more overall numbers of those cells. -This statement on page 6 is confusing: "Consistent with defects in DSB repair, mutant spermatocytes at early pachynema display γH2AX on autosomes, which is even more evident in early pachytene-like cells with synapsis defects (Supplementary Fig. 4d)." I'm not sure how they distinguish between H2ax staining of DSBs vs synapsi defects.
-Supplemental Figure 3E, should an analysis of chromosomal end-to-end fusions in the mutants be conducted?
-Supplemental Figure 4C: no abnormal cells are quantified in this figure as stated in the legend; only part of the symbol key is present in the figure -typo page 3, last paragraph "Suprisingly" Reviewer #2 (Remarks to the Author): Review on Abreu et al.

NCOMMS-18-04038
The paper by Abreu et al. describes the characterization of knockout (KO) mice for two genes, SWS1 and SWSAP1, which plays a critical role in homologous recombination in somatic cells, in meiotic recombination. SWS1, which forms a complex with SWSAP1, is an orthologue of Shu2/Sws1 proteins in yeasts, which plays a role in homologous recombination and DNA repair, particularly by promoting the assembly of Rad51 protein on single-stranded (ss) DNAs. SWSAP1-SWS1 belongs to a new class of a RAD51 mediator, which may be different from RAD51 paralogs; e.g. RAD51B, -C, -D, XRCC2 and -3. The authors created two independent KO mice for both of the genes by using TALEN-mediated gene editing technology. The mutant mice are not lethal which is clearly different from phenotypes of RAD51 paralogs KO mice with embryonic lethality. SWS1 and SWSAP1 KO mice are both sterile in male and female. Cytological analysis of chromosome spreads from testis and ovary shows big reduction of number of RAD51 and DMC1 foci without affecting the number of MOEB1 foci, which marks early recombination intermediates than RAD51/DMC1. Rescued of the infertility by CHK2 deletion and crossover homeostasis is also analyzed in the SWS1 and SWSAP1 KO mice. Moreover, the authors showed that Brca2 C-terminal regions encoded by exon 27 is important for RAD51/DMC1 assembly in the absence of SWS1/SWSAP1. All of the experiments in the paper were conducted well, and the data are in high quality, very much convincing and presented in a good shape. The results in the paper are of great interest to not only researchers in meiosis and 4.
Page 6, second paragraph, line 8-9: the increased number of MEIOB foci does not necessarily come from unstable RAD51/DMC1 binding, rather longer life-span of MEIOB-containing foci possibly due to impairment of downstream events. It is better to rephrase the sentence here. Figure 2b, % of stages: The total % in the mutant mice does not reach 100%, while that in wild type is ~100%. Additional information should be presented in the legend 6. Figure 2c and Supplementary Figure 3d: In mutant mice, there Y-shaped synaptic chromosomes, in which all three parts were stained with both SYCP1 and -3 (triradial synapsis?). Is this typical in the mutant mice? How the authors explain these abnormal pairing in three ways? For me, it is hard to explain this abnormal synapsis since this structure does not have any unsynaptic chromosomes stained only for SYCP3. 7. Figure 2f and 2g (and 5e and 5g): What is "abnormal cells" in graph? If this comes from Figure 2b (5b), there are no data for focus counts in abnormal class in "early zygonema" (Figure 2b shows that 1/3-1/2 should be abnormal class). Need some explanation.

8.
Supplementary Figure 3d, bottom left image: This should contain a blue color for DAPI. It would be great if the authors add the blue or enhance it more (and delete white lines in sws1 swsap1 mouse or show what the line means).

MLH1 counts: Additional mice
In general, the work is extensive and well presented. The story will be of interest, not only because the role of these proteins is elucidated, but also because the unique phenotypes may be useful for other investigations, for example in checkpoint control (as they did here), and the role of sub-optimal genomic integrity in reproduction and embryonic viability (which they developed to a lesser extent). The overall impact is viewed to be above average, as is the quality of this manuscript.
AU: We are pleased that the reviewer is positive.

Major issues:
-a number data points are derived from only 1 mouse, for example Sws1-/-in Fig 3e,f; Dmc1-/-in Fig 3a,b; Sws1-/-and Swsap1-/-oocytes in Fig. 3h,I for Mlh1 focus counts. The latter is bad, given the relatively small difference in average foci, and the fact that the mice were not fully inbred ("backcrossed for 3-6 additional generations"); there are known to be genetic differences in Mlh1 foci in different inbred strains. This shortcoming cannot be overcome by simply scoring more spreads from a single mouse.
AU: We addressed this by adding data from additional animals with one exception (see below). The results are comparable to that previously reported with a smaller number of mice.

Sws1 -/spermatocytes in Fig 3e,f for MLH1 focus counts:
We added data from two Sws1 mutants (∆1A and ∆1G) for a total of 3 mutants, along with their littermate controls. The mean number of MLH1 foci with the two additional mutants is comparable to that previously reported with the single ∆1A mutant (18.0 foci), as is the control mean (23.5 foci).

Swsap1 -/spermatocytes in Fig 3e,f for MLH1 focus counts:
We added data from one additional Swsap1 mutant for a total of 3 mutants. The mean number of MLH1 foci is nearly identical adding this new mutant to that previously reported for the two mutants (18.4) Fig 3a,b, 5h and Supp. Fig. 5c: We counted MEIOB foci from three additional Dmc1 mice for a total of 4 mutants. The mean number of MEIOB foci is similar to the number that we previously reported with one mouse. Fig. 3h,I for MLH1 focus counts: We harvested embryos from 5 pregnant mice and made chromosome spreads from all 17 female embryos obtained from them while the embryos were being genotyped.

Sws1 -/oocytes in
Despite this substantial effort, unfortunately none of the female embryos was mutant, precluding us from adding MLH1 counts on Sws1 oocytes. However, we have now provided data on the Chk2 crosses with the Sws1 mutant in Fig. 4a, Supp. Fig. 6a,b, and Supp. Table 3c. Notably, Sws1 Chk2 ovaries are rescued substantially in size relative to the Sws1 mutant, consistent with a DNA damage response in the single Sws1 mutant, which leads to a severe depletion in oocytes that can be dampened by the loss of CHK2. Further, Sws1 Chk2 females can give rise to live births, implying that crossover numbers are not impacted enough to preclude live births. Moreover, MLH1 foci in Swsap1 males and females are similar (18.4 and 17.3, respectively) which is also very similar for that in Sws1 males (18.0). Thus, it seems likely that Sws1 oocytes have similar numbers of MLH1 foci. In support of this is the similarity of the Sws1 and Swsap1 mutants in many other respects in males and females. Fig. 3h,I for MLH1 focus counts: We added data from two additional Swsap1 mutants for a total of 3 mutants, along with their littermate controls. The mean number of MLH1 foci is similar adding the two additional mutants to that previously reported with the single mutant (17.3 foci), as is the number for the controls (24.4 foci).

Swsap1 -/oocytes in
-the basis for the smaller litter size in Chk2-/-Swsap1-/-animals should be explored. Mice with smaller oocyte pools may ovulate identical numbers (at least in early life), but may suffer from aneuploidy (associated with reduced crossovers), leading to embryonic lethality. The experiment would be relatively easy, since embryos and ovulation sites can be quantified in the same litters. This is likely, actually, and would increase the paper's impact if true. Can the authors elaborate on the effectiveness of the spindle checkpoint in general, and whether it comes into play at all in these mutants.
AU: To address the basis for the smaller litter size in the Swsap1/Chk2 double mutants, we sacrificed pregnant females at 12.5 dpc (or in a few cases adult virgin mice) and counted corpora lutea in the ovary as a measure of ovulated oocytes and also the total implantation sites in the uterine horns, including normal embryos and resorbed embryos. Both the number of ovulated oocytes and the number of implantation sites are marginally reduced (~20%) in the Swsap1/Chk2 double mutants, although neither reaches statistical significance. The number of embryos is reduced (38%) and the number of resorbed embryos is increased about 3-fold, both of which reach statistical significance. These results suggest that the embryo lethality is increased, possibly due to reduced crossovers, but that reduced numbers of ovulated oocytes may also contribute. These data are presented in Supp. Fig. 6d.
Additionally, the Chk2-/-rescue results are only superficially interpreted. The authors state, "The lack of a complete rescue with the loss of CHK2 is likely due to the persistence of synapsis defects (open circles; Fig. 4e) and/or reduced MLH1 focus numbers ( Fig. 4e and Supplementary Fig. 5b)." The implications of this statement will not be obvious to everyone. Indeed the latter is probably incorrect, because MLH1 deficiency doesn't lead to elimination of oocytes prior to ovulation. AU: We thank the reviewer for this last point and have removed the quoted statement. Furthermore, we have added sentences to further the interpretation: "Thus, CHK2 is critical for oocyte elimination in the Shu mutants. The rescue by CHK2 ablation is likely better in the Shu mutants than that observed in Dmc1 mice because DSB repair is more proficient, consistent with previous evidence that DSB load is a determinant of oocyte elimination (Rinaldi et al 2017)." -The authors use MEIOB foci and H2AX foci to conclude that the Shu mutants form normal numbers of DSBs. Probably true, but a Spo11 oligo assay would be a very convincing orthogonal assay.
AU: While the SPO11-oligo assay would further support the conclusions that DSB levels are unaffected, we feel that the gH2AX data is sufficient to make this point.
-Are their two mutants really the same? They are in a complex together, but they are used interchangeably here, and in certain figures there are differences between the two. What does this imply? If they are not the same, why weren't double mutant experiments performed for the Sws genes as well (combined with Chk2 and Brca2 C term).
AU: Every indication is that the Sws1 and Swsap1 (and double) mutants have similar or identical phenotypes: from gross phenotypes like viability, infertility in both sexes, similar reduction in testis and ovary weights, and arrest stages, to more subtle phenotypes like RAD51, DMC1, MEIOB, MSH4, MLH1 focus reductions. We focused on SWSAP1 initially because it is considered to be a RAD51 paralog and, moreover, double mutant analysis is costly both in effort and funds, especially when one of the mutants is infertile. We certainly appreciate the point that double mutant analysis could uncover unexpected phenotypes that distinguish the two mutants, and thus attempted to generate Sws1/Chk2 and Sws1/Brca2 C term double mutants.

Sws1/Chk2:
We were successful in generating Sws1/Chk2 double mutant mice and now provide a detailed analysis of them, essentially recapitulating the data we obtained with Swsap1/Chk2 double mutants. Thus, we observed that there was a substantial rescue of ovary weight in the Sws1/Chk2 double mutants (Fig. 4a, Supp. Fig. 6a) accompanied by the presence of follicles at different stages of development (Fig. 4d, Supp. Fig. 6b), as well as a rescue of fertility (Supp . Table 3c). In males, testis weights were not rescued (Fig. 4b, Supp. Fig. 6e), but post-meiotic cells were observed in some tubules (elongating spermatids, Supp. Fig. 6f). Thus, Sws1/Chk2 analysis recapitulates the Swsap1/Chk2 partial rescue, demonstrating that CHK2 plays a critical role in oocyte death and a minimal, but detectable, role in halting spermatocyte progression.

Sws1/Brca2 C term:
While we were successful with Sws1/Chk2, and obtained similar results as with Swsap1/Chk2, we have been unsuccessful as yet in obtaining Sws1/Brca2 C term double mutants. Given that the additional analysis we have performed in response to the reviewer's comments demonstrates similar phenotypes for both the Sws1 and Swsap1 mutants, we expect that the Sws1/Brca2 C term double mutants will behave similarly to the Swsap1/Brca2 C term double presented in Fig. 5. Still, there is the possibility that it will behave differently, but even if that would turn out to be the case, we believe it would not substantially change any of the central conclusions of the paper. We have modified the text to be careful to note that the results apply specifically to the Swsap1/Brca2 C term double mutant, and we have also softened any more general conclusions by referring to an "intact" Shu complex or by using the words "suggest" or "suggesting".

Minor issues:
5 -The abstract fails to reflect the significance and context of this paper. To anyone but an afficionado, the Shu complex and BRCA2 C terminus need an explanation. I realize that there may be a word limit in the abstract, but it is important to craft it so diverse scientists can understand the implications of the work. AU: As suggested, we have modified the abstract to attempt to make it more accessible to diverse scientists.
-Final sentence in intro, last phrase, is uninformative. AU: As suggested, we have modified this sentence.
-The use of MLH1 as equivalent for crossovers is a bit misleading; yes; it's believed that MLH1 marks crossovers but to say it is a crossover isn't entirely appropriatemetaphase spreads to detect crossovers would be needed to state this definitely. AU: As suggested, we have specified that MLH1 foci mark most, as opposed to all, crossovers.
p. 5, the statement that "Unlike later stages, early meiotic prophase cells at leptonema and zygonema are increased in Shu single-and double-mutant mice. " is misleading. I think the authors may mean that the fraction of early meiotic cells, as a proportion of all cells in mutant tubules (which lack postmeiotic stages), is higher than wild type. As stated, it sounds like they have more overall numbers of those cells. AU: As suggested, we have clarified this point. -This statement on page 6 is confusing: "Consistent with defects in DSB repair, mutant spermatocytes at early pachynema display γH2AX on autosomes, which is even more evident in early pachytene-like cells with synapsis defects (Supplementary Fig. 4d)." I'm not sure how they distinguish between H2ax staining of DSBs vs synapsi defects. AU: The supplementary figure (now Supp. Fig. 5d) shows that early pachytene cells with full synapsis show γH2AX on autosomes. We also observe residual γH2AX on the subset of synapsed chromosomes in the pachytene-like cells, although the γH2AX on unsynapsed regions is more prominent (as expected from MSUC). As the reviewer notes, for unsynapsed regions we cannot distinguish whether γH2AX is due to DSB repair or synaptic defects or both, but the signal on synapsed segments is strong evidence of a DSB repair defect. We modified the text to clarify.
-Supplemental Figure 3E, should an analysis of chromosomal end-to-end fusions in the mutants be conducted? AU: As suggested, we have now included this analysis in Supp. Fig. 3f and referred to this analysis in the last sentence in the text on page 5 and added to the figure legend.
-Supplemental Figure 4C: no abnormal cells are quantified in this figure as stated in the legend; only part of the symbol key is present in the figure AU: This was corrected. We thank the reviewer for finding this oversight.
The paper by Abreu et al. describes the characterization of knockout (KO) mice for two genes, SWS1 and SWSAP1, which plays a critical role in homologous recombination in somatic cells, in meiotic recombination. SWS1, which forms a complex with SWSAP1, is an orthologue of Shu2/Sws1 proteins in yeasts, which plays a role in homologous recombination and DNA repair, particularly by promoting the assembly of Rad51 protein on single-stranded (ss) DNAs. SWSAP1-SWS1 belongs to a new class of a RAD51 mediator, which may be different from RAD51 paralogs; e.g. RAD51B, -C, -D, XRCC2 and -3. The authors created two independent KO mice for both of the genes by using TALEN-mediated gene editing technology. The mutant mice are not lethal which is clearly different from phenotypes of RAD51 paralogs KO mice with embryonic lethality. SWS1 and SWSAP1 KO mice are both sterile in male and female. Cytological analysis of chromosome spreads from testis and ovary shows big reduction of number of RAD51 and DMC1 foci without affecting the number of MOEB1 foci, which marks early recombination intermediates than RAD51/DMC1. Rescued of the infertility by CHK2 deletion and crossover homeostasis is also analyzed in the SWS1 and SWSAP1 KO mice. Moreover, the authors showed that Brca2 C-terminal regions encoded by exon 27 is important for RAD51/DMC1 assembly in the absence of SWS1/SWSAP1. All of the experiments in the paper were conducted well, and the data are in high quality, very much convincing and presented in a good shape. The results in the paper are of great interest to not only researchers in meiosis and recombination fields but also to general readers.