The Y-encoded gene zfy2 acts to remove cells with unpaired chromosomes at the first meiotic metaphase in male mice.

During male but not female mammalian meiosis, there is efficient apoptotic elimination of cells with unpaired (univalent) chromosomes at the first meiotic metaphase (MI) [1]. Apoptotic elimination of MI spermatocytes is seen in response to the univalent X chromosome of XSxr(a)O male mice [2], in which the X chromosome carries Sxr(a) [3, 4], the Y-chromosome-derived sex-reversal factor that includes the testis determinant Sry. Sxr(b) is an Sxr(a)-derived variant in which a deletion has removed six Y short-arm genes and created a Zfy2/Zfy1 fusion gene spanning the deletion breakpoint [4, 5]. XSxr(b)O males have spermatogonial arrest that can be overcome by the re-addition of Eif2s3y from the deletion as a transgene; however, XSxr(b)OEif2s3y transgenic males do not show the expected elimination of MI spermatocytes in response to the univalent [6]. Here we show that these XSxr(b)OEif2s3y males have an impaired apoptotic response with completion of the first meiotic division, but there is no second meiotic division. We then show that Zfy2 (but not the closely related Zfy1) is sufficient to reinstate the apoptotic response to the X univalent. These findings provide further insight into the basis for the much lower transmission of chromosomal errors originating at the first meiotic division in men than in women [7].

Results

Schematic diagrams of the Y gene complement of the mouse Y short arm and of Sxr and Sxr along with salient details of the mouse models analyzed in this study are provided in Figure 1. Our first objective was to fully characterize the stage of meiotic arrest in XSxrOEif2s3y transgenic males, as well as in XOSry,Eif2s3y transgenic males that have the minimum Y gene complement compatible with progression to the first meiotic metaphase. For this we have used the X-linked Eif2s3y transgene of Mazeyrat et al. [6] (this transgene is hereafter denoted by X); because it is on the X chromosome, this transgene is silenced during pachytene by ‘meiotic sex chromosome inactivation’ (MSCI [8]) thus mimicking the pachytene silencing of the endogenous Eif2s3y. In normal males the two meiotic divisions are completed during stage XII of the spermatogenic epithelium, although rare apoptotic MI spermatocytes may be retained into stage I. Histologically (Figure 2A) it appeared that in the XOSry and XO males (‘XO Eif2s3y rescue males’) a substantial number of spermatocytes in stage XII tubules are completing the first meiotic division to form interphasic secondary spermatocytes (in normal males this is a very transitory stage between the two meiotic divisions). Examination of subsequent tubule stages (e.g tubules at V-VI) revealed an accumulation of cells that had the morphology of interphasic secondary spermatocytes and the absence of cells with the nuclear morphology and size of round spermatids (insets in Figure 2A). It has previously been shown that MI and interphasic secondary spermatocytes retain the synaptonemal complex protein SYCP3 adjacent to the centromeres, SYCP3 then disappears soon after the formation of round spermatids. A combination of staining with SYCP3 and the nuclear stain DAPI thus enables interphasic secondary spermatocyte nuclei to be specifically identified [9]. SYCP3/DAPI staining was therefore carried out on spread spermatogenic cells, and interphasic secondary spermatocytes were clearly identified in the XO Eif2s3y rescue males (see Figure 2B for XOSry). However, not all the cells with the interphasic secondary spermatocyte nuclear morphology retained an SYCP3 signal; we suspect that these are secondary spermatocytes that have been retained beyond the time that SYCP3 staining would normally have disappeared. Quantification of the DAPI signal in these spread spermatogenic cells showed that all the presumptive interphasic secondary spermatocytes have the expected 2C DNA content and that no haploid spermatids (1C DNA content) are being produced (Figure 2C). These results show that in XO Eif2s3y rescue males a substantial number of spermatocytes are completing the first meiotic division and that there is then arrest at the interphase between the first and second meiotic divisions.

Figure 1

Information Pertaining to the Mouse Models

(A) The mouse Y short arm (Yp; represented to scale in the magnified view) has seven single-copy genes, two duplicated genes, and one multi-copy gene. (Note that the mouse Y chromosome has a single distal pseudoautosomal region [PAR] on the long arm.)

(B) The Y chromosome short-arm-derived Sxr sex-reversal factor, here attached distal to the PAR on the X chromosome, includes most of the Yp genes.

(C) The Sxrderived deletion variant Sxr has a >900 kb deletion removing six single-copy genes (Δ) and creating a Zfy2/1 fusion gene spanning the deletion breakpoint (†).

(D) H & E-stained stage XII-I testis tubule section from 30-day-old XSxrO and XO mice. In XSxrO there are many arrested, darkly stained (dying; stars), MI spermatocytes. In XO most of the MI spermatocytes have already divided to form interphasic secondary spermatocytes (arrow), and those that remain are lightly stained (healthy; arrowheads).

(E) Tables describing the transgenes and the mouse models analyzed in this study.

The scale bar for magnified views in (A), (B), and (C) represents 150 Kb; it represents 40 μm in (D).

See also Figure S1.

Figure 2

The Secondary-Spermatocyte Stage Is Reached in XO Eif2s3y Rescue Mice

(A) H & E-stained stage XII and V–VI testis tubule sections from 30-day-old XY, XOSry, and XO mice. In XY males the two meiotic divisions occur within stage XII. Metaphase plates (arrowheads) and a pair of interphasic secondary spermatocytes (SS) are indicated. At stage V–VI, small, round haploid spermatids (St) are seen adjacent to the tubule lumen. In XOSry and XO males metaphase plates (arrowheads) are again present at stage XII; there are also some darkly stained cells or groups of cells that are indicative of cell death. Haploid spermatids are not found at stage V–VI; instead, the cells adjacent to the lumen are larger and have the morphology of interphasic secondary spermatocytes (SS). An example of spermatids or secondary spermatocytes is represented at a higher magnification in the left bottom part of each panel. The scale bar represents 20 μm.

(B) Spread spermatogenic cells from 28- to 30-day-old XYΔSry and XOSry males stained with DAPI (blue) and an antibody against SYCP3 (green). The pachytene cells serve as a positive control for SYCP3 staining. There is no SYCP3 staining in spermatogonia. Interphasic secondary spermatocytes with bar-shaped SYCP3 staining remaining at the majority of chromocenters are found in both genotypes. An SYCP3-negative round spermatid nucleus is shown for the control male (these were not found in XOSry males). Cells with the DAPI nuclear morphology of interphasic secondary spermatocytes but with very weak (SS) or absent (SS∗) SYCP3 staining (lower right panel) are an additional abundant cell type in XOSry males.

(C) Nuclear DNA content quantitated by integrated intensity measurement of DAPI fluorescence on spread spermatogenic cells from 28- to 30-day-old XOSry, XO, and XYΔSry males. The values are adjusted to give an arbitrary value of 4 for pachytene nuclei. In XYΔSry males the spermatogonia (Sg), pachytene spermatocytes (P), secondary spermatocytes (SS), and spermatids (St) have the expected DNA content of 2C, 4C, 2C, and 1C, respectively. For XOSry and XO, “SS” includes cells with strong and weak SCP3 staining, and “SS∗” includes those with no SYCP3 staining; both have a 2C DNA content, and no cells with a 1C DNA content were identified. Two mice were analyzed per genotype, and the number of cells quantified for each group is indicated.

†XYΔSry is a fertile control male genotype that is generated by the cross that produces XOSry males.

In XSxrO mice the spermatocytes are eliminated by apoptosis at MI [3, 10, 11]. Our confirmation that in XOSry and XO males substantial numbers of spermatocytes complete the first meiotic division implies that apoptosis is delayed or markedly reduced in these XO Eif2s3y rescue genotypes. To verify this we assessed MI apoptosis by TUNEL assay on XOSry, XO and XSxrO testis sections, carried out at 30 dpp in order to avoid the secondary germ cell loss that is known to occur as the XO Eif2s3y rescue males age [6]. This revealed that apoptotic elimination at MI is considerably reduced in the XOSry and XO males, there being a significant decrease in the percentage of tubules with MI spermatocyte apoptosis together with >3-fold decrease in the number of apoptotic MI spermatocytes in tubules with MI spermatocytes (Figures 3A and 3B). Nevertheless, some MI apoptosis still occurs in XOSry and XO (Figure 3B). Although the apoptotic response to X chromosome univalence is not totally abolished, there must be a Y gene mapping to the Sxr deletion (Δ, inset in Figure 1C) that potentiates this response.

Figure 3

Markedly Reduced MI Apoptosis in XO Eif2s3y Rescue Mice

(A) TUNEL-positive (green) first-meiotic metaphases (MIs) and healthy phosphor histone H3 (pH3, red) MIs were identified by their position away from the cell basal layers. DAPI (blue) was used as a nuclear stain. The upper panels show the paucity of apoptotic cells in 30 dpp XOSry compared to XSxrO mice. Below, at higher magnification, the boxed areas from the upper panels show the marked reduction of apoptotic MIs in stage XII tubules from XOSry mice. Healthy meiotic divisions are present in both genotypes. The scale bar represents 325 μm for the upper panels and 40 μm for the lower panels.

(B) Quantitation of MI apoptosis was carried out on testis sections from 30-day-old XSxrO, XOSry, XO, and XYΔSry mice. The latter three genotypes have significantly lower levels of apoptosis than XSxrO. ∗p ≤ 0.05, ∗∗p ≤ 0.01, and ∗∗∗p ≤ 0.001.

Next we used transgene rescue with the XOSry model, to establish which gene(s) from Δ potentiates the apoptotic elimination at MI of cells with a univalent X chromosome. For this we initially used the Ube1y1 and Ddx3y BAC transgenic lines used by Mazeyrat et al. [6] and newly generated BAC transgenic lines for Kdm5d, Uty and Usp9y [12]. These represent all the genes (in addition to Eif2s3y) that lie completely within the deletion. In a recent study the expression of all these transgenes was validated by RT-PCR and RNA FISH, the histology of XY testes with each transgene was shown to be grossly normal, and the males were fertile [12]. We produced 29-30 day-old XOSry mice with one or more of the transgenes for comparison with the XOSry, XO and XSxrO ‘controls’. For assessing the effect of the transgene additions on the apoptotic response at MI we used the TUNEL assay together with immunostaining for phospho histone H3 (pH3) in order to provide a measure of the ratio of apoptotic to healthy MI spermatocytes, as this provides a sensitive statistic for detecting changes in apoptotic rate (for details see methods). These ratios are >3-fold reduced in XOSry (1.1 ± 0.2) and XO (1.0 ± 0.3) as compared to XSxrO (3.4 ± 0.4). Classical histology showed that for all the transgene additions secondary spermatocytes are still abundant (data not shown); the combined TUNEL/pH3 assay comparisons showed there is no significant increase in the ratio of apoptotic to healthy MI spermatocytes (left panel Figure 4A). Thus the transgenic addition of Usp9y, Ube1y1, Ddx3y, Kdm5d or Uty to XOSry mice does not reinstate the apoptotic response.

Figure 4

Zfy2 Is the Y Gene that Maps to Δ and Reinstates the Apoptotic Elimination at MI

(A) The ratio of apoptotic (TUNEL-positive) to healthy (pH3-positive) MI spermatocytes is plotted for each mouse, and the average ratio for each genotype is shown as a bar. Left panel: ratios for XSxrO, XOSry, and XO “controls” and for the XOSry mice with Δ genes added as transgenes. Usp9y, Ube1y1, Ddx3y, Kdmd25, and Uty did not reinstate the apoptotic elimination at MI. The right panel presents the results for the Zfy transgene addition: only the Zfy2 transgene reinstates the apoptotic response. Importantly, a nonfunctional Zfy2 transgene integrated at the same locus (X) is unable to reinstate the MI apoptosis.

†The XSxrO males also carry the nonfunctional Zfy2 transgene as a control for possible effects of the disruption of the Hprt locus.

††The apoptotic/healthy ratio for XOSry is significantly greater (p < 0.00001) than that for XOSry and not significantly different from that for XSxrO.

(B) H & E-stained stage XII and stage IV testis tubule sections from 30-day-old XOSry and X,Z2OSry mice. Early and late stage XII testes were differentiated by the presence of diplotene (D) spermatocytes in the former and the lack of diplotene spermatocytes with the presence of secondary spermatocytes (SS) in the latter. In XOSry testes at stage XII there is a marked increase in MI spermatocytes that are heavily stained with eosin (starred), indicative of apoptosis; at stage IV there is a dramatic reduction in the number of secondary spermatocytes in comparison to XOSry testes. This reduction is due to the increased apoptotic elimination at MI in XOSry testes. The scale bar represents 40 μm.

(C) Comparison by qRT-PCR of Zfy transcript levels in testes of different genotypes and between Zfy1 and Zfy2; the primers used were common to both Zfy transcripts (a primer list is available in Table S1). Testes were collected when mice were 17.5 days old, when no overt differences in cell population were observed between genotypes; the germ-cell-specific transcript Dazl was used for normalization. XY, XSxrO, XOSry, and XOSry provide estimates of the levels deriving from Zfy1+Zfy2 on the Y chromosome, Zfy1+Zfy2 in Sxr, Zfy1 in the X transgenic line, and Zfy2 level in X transgenic line. Relative transcript levels are expressed as the n-fold change ± SEM. The comparison of the two transgenic lines shows that Zfy1 levels were significantly higher (∗∗∗p = 0.0001) than Zfy2 levels.

See also Figure S2.

In XO males Zfy1 and Zfy2 have been replaced by a transcribed Zfy2/1 fusion gene spanning the Δ breakpoint [5]. This fusion gene has all the promoter elements of Zfy2 but has the terminal exons 6-11 of Zfy1 that encompass all but the first 20 amino acids of the open reading frame. It is therefore expected to express, with the pattern of expression of Zfy2, a protein identical to ZFY1 aside from the 16th amino acid where a leucine is replaced by a phenylalanine (Figure S1). Prior to MI Zfy1 and Zfy2 have similar expression patterns [13, 14] so it is possible that the impaired MI apoptosis in the Eif2s3y rescue males is a consequence of a specific loss of Zfy2 function; we therefore decided to introduce a Zfy2 transgene into the XOSry mice.

Pronuclear injection of a Zfy2 BAC leading to autosomal integration of the transgene proved to be incompatible with male fertility due to total apoptotic elimination of mid pachytene spermatocytes. This pachytene apoptosis is a consequence of the mis-expression of Zfy2 during pachytene; Zfy2 is silenced by MSCI when on the Y-chromosome [12]. To overcome this problem, we used cassette mediated exchange (CME) into the Hprt locus [15] to generate X-linked Zfy2 transgenic lines in which the transgene would be subject to MSCI [12]. We generated two such lines and the carrier males were fertile; Hprt transcript analysis confirmed the expected CME disruption of the Hprt locus. However, only one of the lines expresses Zfy2 (Figure S2), while in the second line the BAC proved to have a deletion removing the region encoding the Zfy2 open reading frame (data not shown). We denote these functional and nonfunctional X-linked transgenes as X and X respectively; the nonfunctional Zfy2 transgene provides a control for the loss of Hprt function associated with the CME insertion into the Hprt locus.

The histology of XOSry mice with the expressing Zfy2 transgene (XOSry) showed a major reduction in the number of secondary spermatocytes, in conjunction with a marked increase in apoptotic MI spermatocytes (Figure 4B). We assessed apoptotic/healthy MI ratios using the TUNEL/pH3 assay and the ratio for XOSry mice (3.7 ± 0.3) was now equivalent to that obtained for XSxrO mice (3.4 ± 0.4), while the X transgene had no effect (Figure 4A). Thus Zfy2 reinstates the efficient apoptotic response to a univalent X chromosome. Subsequently a transgenic line with a single copy of a Zfy1/Ube1y1 BAC integrated on the X (X) by chance, was obtained following pronuclear injection [12]. We established by RT-PCR that this transgene expresses Zfy1 (Figure S2), and by qRT-PCR using Zfy2/1 common primers on testes from 17.5 day-old XOSry and XOSry males that Zfy1 is present at a higher level than Zfy2 (Figure 4C). Despite this higher expression, Zfy1 does not reinstate the apoptotic response (Figure 4A). Thus there is a specific need for Zfy2 for an efficient apoptotic response.

Discussion

The efficient apoptotic elimination of MI spermatocytes with a univalent sex chromosome that fails to achieve bipolar attachment to the spindle (as in XSxrO males [2, 10, 11]) is widely assumed to be triggered by a spindle assembly checkpoint (SAC) monitoring the bipolar attachment of bivalents to the spindle [1]. This apoptotic response is male-specific as evidenced by the completion of both meiotic divisions in XO female mice [16]. Here we show that in mice the Y-encoded gene Zfy2 (encoding a putative transcription factor) is required for this efficient apoptotic elimination at MI. It is interesting that Zfy2 (together with Zfy1) was recently also shown to be responsible for the stage IV pachytene apoptosis associated with a failure to silence the Y chromosome during pachytene [12].

In the mouse the region of homology that mediates crossing over between the X and Y chromosomes (this region is called the pseudoautosomal region) only encompasses ∼700 kb of DNA [17], and a recent study has shown that efficient formation of a crossover in this region (on which the XY bivalent at MI depends) requires a late acting Spo11 isoform that is dispensable for autosomal bivalent formation [18]. Therefore, might the Zfy2-dependent MI apoptotic response also be an adaptation allowing mice to cope with the risk of sex chromosome univalence? We have recently analyzed MI apoptosis in “Down syndrome mice” that carry a derivative of human chromosome 21 [19] and have found a >4-fold increase in apoptotic MI/healthy MI ratios compared to control ratios (XY = 0.30 ± 0.01; XY + h21 = 1.27 ± 0.1; NV and PSB, unpublished). Thus it is clear that the apoptotic response to univalence is not restricted to the sex chromosomes. It is well established that human trisomies (such as trisomy 21, Down syndrome) are predominantly due to maternal MI errors and that for most autosomal trisomies the incidence of these MI errors increases markedly with maternal age [7]. There is also accumulating evidence that the age-related increase is at least in part due to the progressive loss of the chiasmate links on which the bivalent association depends [20-22]; this can ultimately lead to separation into two univalents. However, it is also clear that pairs of homologous chromosomes that fail to establish a chiasmate link (non-exchange chromosomes), and thus will present as univalents at MI, are also associated with a preponderance of maternal MI errors but do not have the marked age dependence [23-25]. There is now substantial evidence for MI SAC activity in mammalian oocytes [26], so whatever the origin of the univalents it is expected that the majority will elicit an MI SAC response because they fail to achieve bipolar attachment to the spindle.

This still begs the question as to why the MI SAC response does not permanently arrest these cells at MI and thus prevent the transmission of aneuploidy. Significantly, in budding yeast engineered to have a pair of non-exchange chromosomes, there is only an approximately 1 hr SAC-mediated delay, after which the meiotic cycle resumes with the transmission of aneuploidy [27]. This parallels the lack of a meiotic block and the transmission of X monosomy by XO female mice, although no delay was detected with this single univalent [16]. During the first meiotic division in female mice depleted for the anaphase I promoting protein Cdc20, there is a high incidence of chromosome segregation errors involving one to a few bivalents, indicative of a failure to achieve bipolar attachment at MI; in this case there is an approximately 1–2 hr SAC-mediated delay, after which the meiotic divisions are completed and aneuploidy is transmitted [28]. Thus, although the MI SAC serves to allow time for all bivalents in chromosomally normal cells to achieve bipolar attachment to the spindle before proceeding to anaphase (avoiding the aneuploidy that results from premature anaphase entry), it does not permanently stall the meiotic division when faced with 1 or 2 obligate univalents or with a small number of bivalents that do not achieve bipolar attachment. Hence, the widely held view that the MI SAC should eliminate cells with 1 or a few univalents (which fail to achieve bipolar attachment) is unfounded. This view probably derives from observations of anaphase arrest after experimental interventions that result in extensive univalence, as well as from the belief that in males one or two univalents do cause a SAC-mediated arrest at MI. However, as we show here, in the absence of the apoptotic elimination, the first meiotic division is completed.

Why is it that males but not females have developed this efficient apoptotic response to MI univalence? From an evolutionary point of view it could be argued that there is not a strong selection for a mechanism to prevent monosomy or trisomy because the vast majority of monosomic and a substantial proportion of trisomic conceptions are lost very early in pregnancy [29], probably with little if any impact on subsequent fertility. In spermatogenesis cells pass through meiosis as synchronously developing cohorts of cells conjoined by intercellular bridges [30], so there may be a need to eliminate cells that begin to lag behind in their development. In this context it is significant that in mice lacking Tex14, which is essential for the formation of intercellular bridges between germ cells, females are fertile but males are sterile. The male sterility is due to apoptotic elimination of spermatocytes at the mid pachytene stage (spermatogonial apoptosis is at normal levels), and there is asynchrony in germ cell development within a tubule prior to this apoptosis [31, 32]. Mid-pachytene apoptosis is seen in the context of a number of meiotic mutants, many but not all of which also exhibit MSCI failure [33]. It has been suggested that this is a stage at which Sertoli cells “check” the spermatogenic epithelium for spermatocytes lagging behind the normal developmental schedule and trigger the apoptotic removal of the laggards [1, 34]. We suspect the MI apoptosis in response to the obligate univalent is triggered by a SAC-mediated delay.

In the mouse two Y-encoded copies of Zfy encode putative transcription factors expressed only in the testis. In man there is a single copy (ZFY) that is widely expressed, but there is a testis-specific splice variant (F. Decarpentrie and M.J.M., unpublished data). In mice, and presumably in humans, these Y-located genes are shut down in pachytene spermatocytes, and in mice we have found that transcription does not resume until after the first meiotic division. A challenge for the future will be to determine how Zfy2 (but not Zfy1) is mechanistically linked to the apoptotic response and to determine whether there is a similar link with human ZFY.