Recombination between the mouse Y chromosome short arm and an additional Y short arm-derived chromosomal segment attached distal to the X chromosome PAR

In a male mouse, meiosis markers of processed DNA double strand breaks (DSBs) such as DMC1 and RAD51 are regularly seen in the non-PAR region of the X chromosome; these disappear late in prophase prior to entry into the first meiotic metaphase. Marker evidence for DSBs occurring in the non-PAR region of the Y chromosome is limited. Nevertheless, historically it has been documented that recombination can occur within the mouse Y short arm (Yp) when an additional Yp segment is attached distal to the X and/or the Y pseudoautosomal region (PAR). A number of recombinants identified among offsprings involved unequal exchanges involving repeated DNA segments; however, equal exchanges will have frequently been missed because of the paucity of markers to differentiate between the two Yp segments. Here, we discuss this historical data and present extensive additional data obtained for two mouse models with Yp additions to the X PAR. PCR genotyping enabled identification of a wider range of potential recombinants; the proportions of Yp exchanges identified among the recombinants were 9.7 and 22.4 %. The frequency of these exchanges suggests that the Yp segment attached to the X PAR is subject to the elevated level of recombinational DSBs that characterizes the PAR.


Introduction
In a normal male mouse meiosis, the regular Spo11-dependent recombination that occurs between the X and Y chromosomes is efficiently restricted to their PAR regions (Burgoyne 1982;Kauppi et al. 2011), although X-Y synapsis may spread beyond the PAR boundary during the mid pachytene stage. The initial synapsis of the PARs protects the PAR chromatin from meiotic transcriptional silencing that is initiated at the zygotene/pachytene transition (reviewed by Burgoyne et al. 2009), but it is not widely appreciated that the subsequent spreading of synapsis may well interfere with the transcriptional silencing of X-Y chromatin proximal to the PARs. This is likely the explanation for the occasional finding of pachytene spermatocytes transcribing the X-linked gene Scml2 (Royo et al. 2010), since it is located only ∼7 Mb from the X PAR (total X length ∼177 Mb).
Although immunostaining for proteins that localize to sites of DSBs (e.g. RAD51, DMC1, RPA) have shown that the X has multiple DSBs that could enable recombination with a synapsed non-PAR Y segment lacking DSBs (Kauppi et al. 2011(Kauppi et al. , 2012Moens et al. 1997;Plug et al. 1998), translocations between the X and the Y are extremely rare. This is presumably because regions of homology are very limited This article is part of a Special Issue on BRecent advances in meiotic chromosome structure, recombination and segregation^edited by Marco Barchi, Paula Cohen and Scott Keeney Electronic supplementary material The online version of this article (doi:10.1007/s00412-015-0559-0) contains supplementary material, which is available to authorized users. and unlikely to be juxtaposed during the presynaptic homology search. This is supported by the sequence data for multicopy X and Y paralogs that indicate crossovers leading to exchange between X and Y copies are very rare despite substantial homology (Soh et al. 2014). Markers of DSBs have rarely been reported on the non-PAR Y, but it can participate in homology-driven recombination as evidenced by mouse and human Y chromosome sequence data, which implicate intrachromatid or inter-sister recombination between ampliconic repeats as responsible for the generation of rare Y deletions and rearrangements (Lange et al. 2009(Lange et al. , 2013Skaletsky et al. 2003;Soh et al. 2014). It is therefore reasonable to posit that rare DSBs in Yp could participate in rare recombination with a synaptic partner with substantial homology, provided they are juxtaposed during the presynaptic homology search.
That this may be the case is supported by the historical identification of occasional Y short arm recombinants among the progeny of males in which there was a Y short arm (Yp) derivative (Sxr a or Sxr b -see Fig. 1b, d) attached distal to the X PAR, or distal to the X and Y PARs Laval et al. 1995;McLaren et al. 1988McLaren et al. , 1992Simpson et al. 1984). While there was the expected frequent transfer of the complete Sxr segment from the Y PAR to the X PAR ( Fig. 1c) or vice versa, there were also exchanges of partial Sxr segments with Yp; the latter exchanges must involve DSBs located within Sxr or within Yp. These partial Sxr exchanges could be balanced exchanges or unbalanced exchanges; the latter are an expected consequence of the presence of substantial regions of sequence repeats on the mouse Y short arm (Soh et al. 2014).
Two of these studies are particularly relevant in the context of the new data reported here. Firstly, Simpson et al. (1984) identified among the progeny of an XSxr-YSxr male (with Sxr attached to the X and Y PARs) a mouse carrying an Sxr deletion variant that was designated Sxr b (with the original Sxr then being designated Sxr a ). Subsequent work established that the deletion resulted from a crossover located within Zfy1 of the Y-located Sxr and within Zfy2 of the X-located Sxr, resulting in a 1.38-Mb deletion removing six Yp genes and creating a Zfy2/1 fusion gene (Fig. 1d) (Decarpentrie et al. 2012;Mazeyrat et al. 1998;Simpson and Page 1991). Secondly, McLaren et al. (1992) analyzed the progeny of XSxr b-YSxr a males that enabled them to document exchanges between Sxr a and Sxr b and between Sxr b and Yp. Significantly, the Sxr a -Sxr b exchanges must have involved DSBs located within these Yp derivatives attached to the PARs. Here, we report a strikingly high frequency of Sxr b -Yp recombination in two mouse models that have Sxr b attached to the X PAR; we propose that this high frequency of DSBs is a consequence of the spreading of DSB hotspot activity from the X PAR into Sxr b .

Mouse breeding
Mice were produced on an outbred albino MF1 (NIMR stock) background. XSxr b Y males (Fig. 2a) were generated by mating XYSxr b males (Sxr b attached to the Y PAR) to females carrying the X-autosome translocation T(X;16)16H (T16H/X females). PCR genotyping for Sxr b markers was then used to identify T16H/XSxr b mice that have developed as females-these mice can be female despite the presence of Sry in Sxr b , because the presence of the T16H translocation ensures that the inactive X is always the X carrying Sxr b (Cattanach et al. 1982;McLaren and Monk 1982). The T16H/XSxrb females were then mated to XY males, and their XSxr b Y sons were identified as fertile males (male T16H carriers are sterile) positive for Sxr b (PCR Sxr b , Table 2). XSxr b Y* males (Fig. 3a) were originally generated for another study (Vernet et al. 2014). They were produced in the same way as XSxr b Y males except that the T16H/XSxr b females were mated to XY* males (Burgoyne et al. 1998;Eicher et al. 1991).
The XSxr b Y and XSxr b Y* males were mated to XX females in order to produce the progeny to be screened for recombinants. In most cases, the females mated to XSxr b Y* males were homozygous for the X-linked coat marker Patchy fur (Paf) since this provides an independent check on some of the results obtained from the PCR genotyping (see below). Importantly, it enables the positive identification of XO female progeny as hemizygous Paf females, which are known to be produced at an elevated frequency when the Y* chromosome is present (Burgoyne and Evans 2000;Eicher et al. 1991).
Screening for sex chromosomally recombinant offspring (Tables 1, 2) The initial screen was based on phenotypic markers: 1. Fur phenotype (when the mothers were Paf homozygotes) A few XSxr b Y males and nearly all XSxr b Y* males were mated to females homozygous for the X-linked Patchy fur (Paf) mutation (Lane and Davisson 1990). At 10-15 days old, three fur phenotypes are clearly distinguishable (Burgoyne and Evans 2000) that are independent of their Yp/Sxr status: wild type (wt) that identifies X Paf Y* in which the effect of the Paf mutation is masked by the presence of the wild type allele on the Y* chromosome, Bpatchy fur^(Paf/+) due to the very sparse hair in regions where the X Paf is expressed as in X Paf X and X Paf X Y ), almost nude (Paf) due to the absence of a paternally derived X as in X Paf O. 2. Gender Based on external examination at weaning. 3. Testis size Small 15-26-mg/testis (S) that is associated with the presence of two X chromosomes and/or with the absence of the spermatogonial proliferation gene Eif2s3y that is deleted in Sxr b ; Medium/Large 70-120mg/testis (M/L).
PCR analysis using DNA samples was used to confirm and extend diagnoses of sex chromosome complement based on phenotypic markers. They were also used to detect Yp/Sxr b recombinants based on sequence information for Sxr b and Yp (Decarpentrie et al. 2012;Mazeyrat et al. 1998;Soh et al. 2014); these were designed to enable the detection of recombination events in homologous regions that would be expected to promote synapsis when juxtaposed during the homology search phase (Supplemental Fig. S1). Primer sequences are in   Mahadevaiah et al. (1998) the restriction enzyme RsaI generating a 224 bp Sxr b -specific fragment. 2. PCR Yp To detect the presence of Yp or Sxr a , we used primers Smc1/Smc2 that amplify a 300-bp fragment of Yp-linked Kdm5D (aka Smcy) and a 330-bp fragment of X-linked Kdm5C (aka Smcx) that acts as an amplification control.

PCR Yq
To detect the presence of the long arm of the Y (Yq), we used primers PC11fp2F and R that amplify a 237-bp fragment of Ssty2 (present in multiple copies) and primers Sstxfp10 and Sstxrp13 that amplify a 350-bp fragment of X-linked Sstx (present in multiple copies- Soh et al. 2014) that acts as an amplification control.

PCR X vs Yp
To double-check some genotyping, we performed a qualitative ratio measurement of the X and Yp / Sxr a fragments. Primers o4110 and o4111 (see Table 2) amplify a 161-bp fragment of X-linked Ddx3x and a 128bp fragment of Yp-linked Ddx3y. The difference in band intensity gives information about the X vs Yp/Sxr a dose. 5. PCR new Sxr b To confirm the products of crossover 3 (XΔSxr b and Y ΔΔSxra ) that creates a new Zfy2/1fusion gene (Supplemental Fig. S1), we have designed several PCR/RFLP tests allowing amplification of both Zfy1 and Zfy2 but, thanks to specific SNPs, digestion of only one (primers o001/o002, o003/o004, o005/o006, and o021/o022). 6. Y* X qPCR (Vernet et al. 2014). To detect the presence of Y* X , we utilized qPCRs for X-linked Prdx4 (absent in Y* X ), Amelx (present in Y* X ) and Myog (on chromosome 1) for normalization.

Immunostaining to identify X-Y bivalent configurations
In XY males, the frequency of X and Y univalence at pachytene is <5 % but it is usually in excess of 20 % when Sxr is present. In an extensive study involving males with Sxr attached to the X and/or Y PAR, the X and Y were separated in 26.5-86.9 % of pachytene spermatocytes as compared to 3.5 % in XY controls (Tease and Cattanach 1989). However, in the absence of synapsis, no crossover can form, and the resulting univalence at the first meiotic metaphase (MI) leads to a ∼97 % efficient elimination of the spermatocytes by the MI spindle checkpoint (Burgoyne et al. 1992(Burgoyne et al. , 2009Sutcliffe and Burgoyne 1989;Vernet et al. 2011Vernet et al. , 2014. Thus, these spermatocytes with X and Y univalence make a negligible contribution to the offspring. Our interest was therefore restricted to pachytene spermatocytes with X-Y bivalents.

Sxr b
XSxr b Y*  11.1% To investigate the X-Y bivalent configurations, we used immunostaining of surface spread spermatocytes as previously described (Turner et al. 2004) using the following antibodies: Guinea pig polyclonal anti-SCP3 (dilution 1:400, gift from James M Turner), mouse monoclonal anti-γH2AX (dilution 1:100, Millipore cat. No. 05-636), Human anti-CREST (dilution 1:50, gift from William Earnshaw). Images were captured on an Olympus IX70 inverted microscope. Each fluorochrome image was captured separately as a 12-bit source image by using a computer-assisted (Deltavision) liquid-cooled CCD (Photometrics CH350L; Sensor: Kodak KAF1400, 1317×1035 pixels). A single multiband dichroic mirror was used to eliminate shifts between different filters. Captured images were processed with Fiji and Adobe Photoshop CS5.1.

Yp:Sxr b recombination crossover sizes estimation and location of potential DSB hotspots
Size estimates of Yp, the Sxr a region, and the sub regions flanking the Sxr b deletion breakpoints are based on the most recent published sequence of the mouse Y chromosome (Soh et al. 2014-File 3 in Data S1) and the sequence of the Sxr b deletion breakpoint (Decarpentrie et al. 2012). They are minimum size estimates as the Yp sequence is not yet complete. Three large blocks of N's within the Yp sequence, totalling 450 kb, were excluded.
The Sxr a breakpoint has not been sequenced but is known to fall within the distal end of the 37-kb Rbmy-repeat tandem array. Based on the recent estimate of 30 Rbmy copies in Yp (Soh et al. 2014) and our previous estimate that 1 in 7 Rbmyrepeats remain in Sxr a , we have assumed that the Sxr a breakpoint lies between the fourth and fifth most distal copies of the Rbmy-repeat array.
The Yp estimate was derived from the sequence telomeric to the centromeric heterochromatin, the latter defined as bases 57238-147035 of Genbank AC175459. The Soh et al. sequence includes 10 of the estimated 30 Rbmy repeats and we therefore added 740 kb to the Soh et al. sequence length (20 times a repeat unit length of 37 kb).
The Soh et al. sequence does not include the Prssly or Teyorf1 genes, which we have mapped to the Sxr b region (unpublished data). These two genes are present in an isolated BAC sequence of 185 kb (NW_001034423). We have assumed that Prssly and Teyorf1 are located distal to Zfy1 since there is no gap in the Soh Yes Yes S small, M medium, L large, pos positive, neg negative  (Brick et al. 2012; Supplementary Data File 1: List of DSB hotspots and H3K4me3 marks). To determine which of these are located within Sxr b , the sequence coordinates were used to identify the associated DNA sequences in the mouse genome using NCBI37/mm9 (as used by Brick et al. 2012), where necessary checking these against the sequence data we have deduced for Sxr b based on the data from Soh et al. 2014.

Results and discussion
The recombination data we present here were obtained from PCR genotyping of offspring (see Materials and Methods) that derive from two crosses. Firstly, there are data for 160 offspring obtained from XSxr b Y×XX matings. Secondly, there are data for 217 offspring from XSxr a Y*×XX matings.

Recombination in XSxr b Y
As expected, PAR-PAR recombination produces X and YSxr b recombinants (52.5 % of offspring). Yp-Sxr b recombination produced the recombinants XSxr a and Y ΔSxrb (5.6 %) that are predicted for crossover 1 (Fig. 2a, b, c). Recombinants were not seen for the potential crossover 2 nor were there recombinants deriving from the potential alternative synaptic alignment illustrated in Fig. 3c. Overall, there was a deficiency of non-recombinants (n = 67, 41.9 %) as compared to recombinants (n=93, 58.1 %), and this was just significantly different from the expected 50 % of each [Binomial test (http://www.measuringu.com/onep.php); two sided p value= 0.0478]. We have no explanation for this deficiency, and it was not observed with the larger sample of data for the XSxr b Y* model (see below).

Recombination in XSxr b Y*
The Y* chromosome has a complex PAR involving duplication and deletion of PAR DNA sequences and an X-derived centromere distal to this complex PAR (Fig. 3a) (Burgoyne and Evans 2000;Burgoyne et al. 1998;Eicher et al. 1991;Rodriguez and Burgoyne 2001). This leads to two orientations for PAR-PAR synapsis and to the production of four distinct recombinant chromosomes following PAR recombination (Fig. 3b). The opportunities for Yp-Sxr b synapsis (Fig. 3c) are the same as those in XSxr b Y. Four types of recombinant arising from crossovers 1 (20 recombinants) and 3 (4 recombinants) were identified (Fig. 3d). The predominance of recombinants involving crossover 1 is to be expected given that it covers 780 kb, which is more than 25-fold longer than for crossovers 2 and 3 (see Table 3). Crossover 4 does cover 580 kb, but we have no markers to detect recombination in this region. The frequencies of the various recombinants identified among the 217 offspring of XSxr b Y* males are illustrated in Fig. 3e.
How do the recombination data for XSxr b Y* males compare with those of XSxr b Y males? As expected, the XSxr b Y* PAR-PAR recombinants differ due to the presence of the Y* chromosome. There is no deficiency in the number of nonrecombinants (n=110, 50.7 %) relative to recombinants (n= 107, 49.3 %), but there is an increase in frequency of Yp-Sxr b relative to PAR-PAR recombinants in XSxr b Y* (11.1 %, n= 24 vs 83) as compared to XSxr b Y (5.6 %, n=9 vs 84) (P=0.02 Fisher's exact test, two-tailed). These recombinants include four with exchanges that are generated by crossover 3. The rare recombinants due to crossover 3 were not identified in XSxr b Y, but this may just reflect the smaller sample size (9 vs 24). Is Yp-Sxr b recombination potentiated by the attachment of Sxr b to the X PAR?
The mouse PAR is estimated to be only ∼700 kb (Perry et al. 2001). Nevertheless, one or two SPO11-mediated DSBs are regularly generated in the X and Y PARs (albeit later than those on autosomes); these DSBs reliably drive the synapsis and formation of a crossover necessary to ensure X and Y segregation at MI (Kauppi et al. 2011(Kauppi et al. , 2012. Yp in normal males is largely protected from recombination because SPO11-mediated DSBs are not preferentially targeted to this region and because it lacks homology with the DSB rich X and Y PARs. In males with Sxr b attached to the X PAR, there are two reasons why the Yp-Sxr b recombination is likely to be preferentially initiated by DSBs in Sxr b . Firstly, we estimate that Sxr b is 860 kb, whereas Yp we estimate to be 4 Mb; this is due to the 1.38-Mb Sxr b interstitial deletion (Δ Sxrb in Fig. 1d) and the reduced number of copies of Rbmy ). Thus, 3.14 Mb of Yp has no homology to Sxr b so that DSBs located in this 3.14 Mb will not be able to find a homologous partner during the presynaptic homology search. On the other hand, with a size of 860 kb, Sxr b DNA will only very rarely be cut by Spo11 unless this region has one or more DSB hotspots. This led us to hypothesize that Sxr b may have DSB hotspots because it is attached to the DSB-rich X PAR. Support for this hypothesis is provided by DSB mapping data (Brick et al. 2012). It has been established that the majority of DNA hotspots co-localize with the histone H3 methyl transferase PRDM9, and these authors found that the X PAR together with ∼900 kb of upstream X-specific DNA is unique in having PRDM9-independent DSB hotspots (hotspots that remain in Prdm9 knockout mice). The frequency of these hotspots declines with increasing distance from the X PAR boundary. We therefore suggest that when Sxr b is attached to the distal end of the X PAR, it is similarly bestowed with DSB hotspots that decrease in frequency with increasing distance from the PAR. If this is true, the highest density of DSBs should be in the region covered by crossover 1, reducing to a minimum in the region covered by crossover 4. However, if during the homology search, single-stranded DSB tails originating in the region covered by crossovers 3-4 do successfully invade the homologous sequences in Yp, then the resulting synaptic alignment may inhibit exchanges in the region covered by crossovers 1 and 2. We have not identified any double recombinants with PAR-PAR and Sxr b -Yp exchanges; this is presumably a reflection of the low likelihood that both regions of homology are juxtaposed during the presynaptic homology search.
Also of potential relevance to the location of DSBs within Sxr b are the 16 potential DSB hotspots identified in Yp (Brick et al. 2012; Supplementary Data File 1: List of DSB hotspots and H3K4me3 marks). However, only one (chrY:2158709-2159150) is retained in Sxr b . This is located within each of the few Rbmy genes that remain close to the Sxr b breakpoint that abuts the distal end of the PAR.
Presynaptic telomere congregation and the incidence of Yp-Sxr b recombination In the 1970s, Solari carried out painstaking electron microscopic reconstructions of the X-Y bivalent at pachytene that served to identify the juxtaposition of what we now know to be the X and Y PAR telomeric ends at the nuclear membrane attachment site and the presence of a synaptonemal complex between the PAR axes; the non PAR ends of the X-Y bivalent although attached to the nuclear membrane were not closely associated (Solari 1970(Solari , 1974. The association of the PAR ends is now widely accepted to be initiated prior to synapsis by the congregation of the telomeric ends of all chromosomes during zygotene (bouquet formation), together with dynamic chromosome movements during the bouquet phase that can bring homologous segments in sufficiently close proximity to enable homologous DNA strand invasion and synapsis (Scherthan 2001;Shibuya et al. 2014).
How might the telomeric congregation and homology recognition phases be affected in XSxr b Y and XSxr b Y* males? In XSxr b Y the XSxr b , telomere and adjacent Sxr b DNA sequences match those of Yp rather than those of Y PAR. This will provide the opportunity for homology recognition and synapsis of Sxr b with Yp-examples of such bivalents are seen in pachytene spermatocyte spreads (Fig. 4a). Nevertheless, the majority of synaptic associations involve the X PAR/Sxr b and Y PAR ends (Fig. 4b) and the recombination data document frequent PAR-PAR recombination (Fig. 2c). This is unsurprising since the Sxr b segment is short (860 kb) so that the X and Y PARs are still close to the membrane attachment sites where the dynamic chromosome movements during the bouquet phase could also lead to PAR-PAR homology recognition.
In XSxr b Y* the PAR/Sxr b and Yp, telomeric ends are the same as those of XSxr b Y thus enabling homology recognition and crossing over. However, there is one feature of Y* that might increase the frequency of Yp-Sxr b exchanges relative to PAR-PAR exchanges. The PAR end of Y* terminates in an X-derived centromeric region that includes some copies of a multi copy X sequence DXHXF34 (Laval et al. 1997;Rodriguez and Burgoyne 2001) that derives from a multi copy X gene that we have termed Sstx (Touré et al. 2004); recent sequence information from Soh et al. (2014 -see footnote to their Figure 5) has indicated that the majority of copies of Sstx are located adjacent to the X centromere. This is expected to promote homology recognition and synapsis of the PAR/centromeric end of Y* with the centromeric end of the X chromosome-examples of this can be seen in pachytene spermatocyte spreads (Fig. 4c). This could lead to crossing over adjacent to the centromeres within the Sstx repeat but we have not attempted to detect this. We identified three other pachytene configurations that we interpret as resulting from Yp-Sxr b synapsis (Fig. 4d), Yp-Sxr b synapsis and X centromere-Y* PAR/centromere synapsis (Fig. 4e) and PAR-PAR synapsis (Fig. 4f). We therefore wondered whether Y* promotion of alternative synaptic configurations to PAR-PAR synapsis might be responsible for the significant increase in the frequency of Yp-Sxr b r e c o m b i n a n t s r e l a t iv e t o PA R -PAR recombinants. In order to test this, we classified the synaptic configurations of 99 pachytene spermatocyte spreads from each genotype, which yielded the following results: XSxr b Y-78 PAR-PAR, 21Yp-Sxr b ; XSxr b Y*-77 PAR-PAR, 18 Yp-Sxr b (the remaining 4 had the X centromere associated with X-derived centromere of Y* as in Fig. 4c). From this, it is clear that the presence of Y* is not promoting Yp-Sxr b synapsis at the expense of PAR-PAR synapsis. A plausible alternative explanation is that synapsis of the complex Y* PAR with the X PAR is less likely to result in a crossover because in either the staggered or parasynapsed orientations there is a reduced PAR segment available for homologous synapsis. In those that fail to form a crossover, the X and Y will separate at the end of prophase, be subject to SAC-triggered apoptotic elimination at MI and will thus not contribute to the progeny (Burgoyne et al. 2009;Vernet et al. 2011Vernet et al. , 2014. Each panel has a diagram of the predicted crossover event, a low magnification view showing the X-Y bivalent in a γH2AX-stained (red) sex chromatin cloud together with near-by autosomes, a higher power view of the X-Y bivalent with the γH2AX staining removed, and a drawing of the deduced X-Y bivalent configuration highlighting the centromere positions. Note that the red CREST staining of active centromeres often appears as yellow where it overlaps the green chromosome axis. (a) XSxr b Y with Yp-Sxr b synapsis. (b) XSxr b Y with PAR-PAR synapsis. (c) XSxr b Y* with centromerecentromere association-this is likely to be due to synapsis driven by homology of Sstx sequences adjacent to the X and Y*(X-derived) centromeres. (d) XSxr b Y* with Yp-Sxr b synapsis. (e) XSxr b-Y* with centromere-centromere and Yp-Sxr b synapsis. (f) XSxr b-Y* with PAR-PAR synapsisparasynapsis and staggered synapsis cannot be differentiated at this resolution. The staggered configuration is indicated in the black square