|Home | About | Journals | Submit | Contact Us | Français|
Mammalian ZFY genes are located on the Y chromosome, and code putative transcription factors with 12–13 zinc fingers preceded by a large acidic (activating) domain. In mice, there are two genes, Zfy1 and Zfy2, which are expressed mainly in the testis. Their transcription increases in germ cells as they enter meiosis, both are silenced by meiotic sex chromosome inactivation (MSCI) during pachytene, and Zfy2 is strongly reactivated later in spermatids. Recently, we have shown that mouse Zfy2, but not Zfy1, is involved in triggering the apoptotic elimination of specific types of sex chromosomally aberrant spermatocytes. In humans, there is a single widely transcribed ZFY gene, and there is no evidence for a specific role in the testis. Here, we characterize ZFY transcription during spermatogenesis in mice and humans. In mice, we define a variety of Zfy transcripts, among which is a Zfy2 transcript that predominates in spermatids, and a Zfy1 transcript, lacking an exon encoding approximately half of the acidic domain, which predominates prior to MSCI. In humans, we have identified a major testis-specific ZFY transcript that encodes a protein with the same short acidic domain. This represents the first evidence that ZFY has a conserved function during human spermatogenesis. We further show that, in contrast to the full acidic domain, the short domain does not activate transcription in yeast, and we hypothesize that this explains the functional difference observed between Zfy1 and Zfy2 during mouse meiosis.
In eutherian (placental) mammals, ZFY genes appear to play an important role in male development, as they have been found on the Y chromosome in many species including humans and mice (1). ZFY genes have a widely expressed X-linked homologue, ZFX, that codes a highly similar protein; this has been shown to be a regulator of self-renewal in embryonic and haematopoeitic stem cells (2). In metatherian (marsupial) mammals, genes related to ZFX and ZFY are autosomal (3), and it is thought that ZFX and ZFY originated more than 100 million years ago, after the separation of eutherian and metatherian lineages, by translocation of an autosomal segment to the pseudo-autosomal region in a common ancestor of extant eutherians (4). ZFY and ZFX genes code putative transcription activators, characterized by a large N-terminal acidic (activating) domain (approximately 360 amino acids) and a C-terminal DNA-binding domain of 12–13 Cys2-His2 zinc fingers, separated by a short basic nuclear localization signal (5,6). The DNA targets and protein partners of ZFY and ZFX remain to be determined.
There are two ZFY genes on the mouse Y chromosome, Zfy1 and Zfy2 (94% amino acid identity), both of which are transcribed primarily in germ cells in the post-natal testis, but have also been described in some fetal tissues (7–10). Post-natal transcripts from the Zfy1 and Zfy2 genes are first readily detected in the testis around the time that the germ cells enter meiosis, between 7 and 10 d.p.p. (days postpartum) (9,11). RNA FISH (fluorescence in situ hybridization) detecting ongoing transcription has documented the transcriptional silencing of Zfy1 and Zfy2 as they enter pachytene, a manifestation of meiotic sex chromosome inactivation (MSCI), and has identified Zfy2 transcription in Y-bearing round spermatids (12–14). Reflecting this resumption of Zfy2 transcription in spermatids, Zfy2 transcript levels have been observed to increase between 22 and 24 d.p.p. as spermatids appear and develop (11). Strong Zfy2 transcription in spermatids is directed by an ‘acquired’ spermatid-specific promoter derived from an X-linked CYPT gene (11). The CYPT genes form a spermatid-specific gene family (15). The CYPT exon of Zfy2 is most likely derived from Cypt1, as it is flanked by 12.1 kb of a sequence (5′: 7.8 kb; 3′: 4.3 kb) with approximately 90% nucleotide identity to the genomic region surrounding Cypt1 (Genome assembly NCBI37—chrX: 16087265–16107151). Zfy1 does not have an upstream copy of the CYPT promoter, and the presence of Zfy1 transcripts in spermatids remains to be determined. Furthermore, the nature of the promoter used by Zfy2 for its early expression and the structure of the Zfy1 and Zfy2 transcripts produced at the different stages of spermatogenesis have not been determined.
In mice, the Zfy genes are affected by the ~1.3 Mb Sxrb deletion that occurred in the mouse Y short-arm-derived ‘sex-reversal factor’ Tp(Y)1CtSxr−a (16) via an illegitimate recombination between the Zfy1 and Zfy2 genes; this created a transcribed Zfy2/1 fusion gene composed of the promoter region, 5′ UTR (untranslated region) and first coding exon of Zfy2 fused in intron 5 to the remaining six coding exons of Zfy1 (17). This effectively puts the Zfy1 coding region under the control of the Zfy2 promoter (18). For this reason, it has been difficult to assess how ZFY function has been affected by the deletion, if at all. Recently, however, we have shown that ZFY is involved in the apoptotic elimination of spermatocytes in two types of sex chromosomally variant males. First, we have shown that in XYY males the apoptotic elimination of spermatocytes during the mid pachytene stage is due to the mis-expression of Zfy1 and Zfy2; both genes are silenced during pachytene in normal males but escape silencing when the two Y chromosomes are synapsed (13). Second, Zfy2 (but not Zfy1 or the Zfy2/1 fusion gene) has been shown to be required for the efficient apoptotic elimination of spermatocytes at the first meiotic metaphase (MI) that occurs in response to the X univalent in XSxraO males that carry the sex reversal factor Tp(Y)1CtSxr−a attached to the pseudoautosomal regions (PAR) of the X chromosome (18).
In humans, in contrast to mice, there is a single ZFY gene on the Y chromosome; furthermore, the transcription of human ZFY is ubiquitous (1). Despite its widespread expression, a rare deletion of ZFY and SRY in a woman carrying a Y;22 balanced translocation was not associated with Turner syndrome stigmata, indicating that ZFY has no critical somatic functions (19). No mutations of ZFY have been described in men and there is, therefore, no information concerning its possible contribution to human germ cell development or male fertility.
With the aim of understanding the basis of the functional difference observed between the nearly identical Zfy1 and Zfy2 genes in relation to the apoptotic response to X chromosome univalence at MI, we undertook a detailed analysis of mouse Zfy transcription during spermatogenesis. We reveal an unexpected complexity of Zfy transcription involving multiple splice variants. This led us to reassess ZFY transcripts in the human testis where we have identified a testis-specific splice-variant, structurally homologous to the major Zfy1 variant transcript in mice.
In order to define the cell types in which the Zfy1 and Zfy2 genes are transcribed in the adult testis, we carried out RNA FISH designed to identify nascent transcripts within the nucleus as they are transcribed (20) (Fig. 1). No clear hybridization was seen in nuclei of Sertoli cells or other somatic cells. Weak hybridization signals were seen in nuclei of a few A spermatogonia (Zfy1: 1/20, and Zfy2: 2/30; data not shown), but appropriate localization to the encoding genes was not confirmed by DNA FISH; no Zfy signals were detected in B or intermediate spermatogonia. Zfy1 and Zfy2 signals then began to appear during leptotene (data not shown), and robust signals were clearly present in all mid-late zygotene nuclei. As expected, no signals were detected in pachytene nuclei where the Zfy genes are silenced by MSCI. In agreement with previous data, Zfy2 transcription is resumed in (Y-bearing) round spermatids where it localizes to the post meiotic sex chromatin; however, Zfy1 is also clearly transcribed at this stage, although it was detected in a substantially lower proportion of spermatid nuclei than Zfy2 (17.5 versus 37%, which equates to 35 versus 74% of Y-bearing spermatids). The higher proportion of spermatids with a Zfy2 signal probably reflects the higher level of Zfy2 transcription from the strong Cypt-promoter in spermatids. Our results establish that both Zfy1 and Zfy2 genes are transcribed in spermatocytes prior to the onset of MSCI, and that both genes are reactivated in spermatids.
We analysed the Zfy1 and Zfy2 transcripts produced during spermatogenesis, using qualitative and quantitative reverse transcriptase polymerase chain reaction (RT-PCR) with RNAs extracted from testes at different ages after birth: 3 and 7 d.p.p. (spermatogonia only), 10, 16 and 20 d.p.p. (spermatogonia + spermatocytes only) and 27 d.p.p. and adult (spermatogonia + spermatocytes + spermatids). Zfy exons are numbered following the published genomic structure of Zfy1 and Zfy2 (21), which are shown in Figure 2A.
We first investigated the use of the different putative promoters by amplifying from the first exons of Zfy1 (exon 1b) or Zfy2 (exon 1a or 1b) to exon 5, the first coding exon (Fig. 2B). The identification of transcripts at 7 d.p.p. provides some support for the RNA FISH identification of weak Zfy1 and Zfy2 signals in some A spermatogonia. The products identified at 7 and 16 d.p.p. show that during early meiotic stages prior to the onset of MSCI, Zfy1 and Zfy2 transcripts include exon 1b (transcripts Zfy1-b and Zfy2-b). We show that Zfy2 does not, however, use the alternative non-coding exons 3 or 4, as previously described for Zfy1 (7). There is no evidence that any one 5′ UTR structure of Zfy1 is specific to a particular stage of spermatogenesis. As shown previously (11), the Zfy2 transcripts with exon 1a (Zfy2-a), produced from the spermatid-specific Cypt-promoter, are not detected at earlier stages, appearing only in the testis containing spermatids (27 and 50 d.p.p.).
These results were confirmed and extended by quantitative PCR (qPCR) (Fig. 2C). Zfy1-b transcripts were amplified with primers in exons 8 and 10, to avoid the multiple bands generated by PCR from exon 1b for Zfy1. As expected, levels of Zfy2-a (Cypt) were strong only in the testis with spermatids (27 and 50 d.p.p.). However, Zfy1-b and Zfy2-b are seen to increase between 7 and 20 d.p.p. as spermatocytes appear and develop. Zfy1-b levels continue to increase with the arrival of spermatids (27 d.p.p.), consistent with our RNA FISH data showing that Zfy1 is reactivated in spermatids. For Zfy2-b, however, levels remain constant between 16 and 50 d.p.p., suggesting that the Zfy-promoter of Zfy2 is not as strongly reactivated in spermatids as that of Zfy1.
The two Zfy genes in the mouse are generally held to differ in their 5′ UTRs, and to be transcribed from distinct promoters, with Zfy2 assumed to be transcribed exclusively from a Cypt-promoter, even though Zfy2 retains sequences nearly identical to the 5′ UTR and putative promoter of Zfy1 (11,21,22). Here, we show, however, that Zfy2 does use its Zfy1-homologous promoter, and 5′ UTR, in particular for its earlier phase of post-natal transcription, when the Cypt-promoter is silent.
To investigate coding exon usage for Zfy1 and Zfy2, we performed RT-PCR, on testis RNA from normal XY males, with primers in the first and last coding exons (Fig. 3A). Unexpectedly, the major product for Zfy1 was ~500 bp smaller than predicted. The expected size was observed for Zfy2. Sequencing revealed the major pre-MSCI Zfy1 product (16 d.p.p.) to have spliced out exon 6, a 552 bp exon that codes for half the acidic domain. Sequencing the major product for Zfy2 revealed it to include the full exon 6. The lower of the two strongest minor Zfy2 products at ~900 bp lacks exon 6, whereas the upper product is a doublet that retains 111 or 123 bp of exon 6 (Fig. 3B).
We then zoomed in on this region, using Zfy1- or Zfy2-specific primers in exon 5 and 7, and confirmed that Zfy1 transcripts predominantly lack exon 6, whereas the major Zfy2 transcripts retain exon 6. This is especially evident at 16 d.p.p. (Fig. 3B). We performed qPCR analysis with primer pairs specific for the splice variants with or without exon 6 and specific for Zfy1 or Zfy2 (Fig. 3C). This revealed that, for Zfy1, the transcript with exon 6 is present at low levels until 20 d.p.p., but increases more than 4-fold between 20 and 27 d.p.p., coincident with the progression of germ cells into spermiogenesis. On the contrary, the concentration of Zfy1 transcript without exon 6 reaches its peak between 16 and 20 d.p.p. prior to the development of spermatids. For Zfy2, the transcript with exon 6 increases between 10 and 20 d.p.p., as spermatocytes accumulate, and further increases between 20 and 27 d.p.p., when the strong Cypt promoter becomes active. The Zfy2 transcript without exon 6 can be seen to increase between 20 and 27 d.p.p., but is expressed at extremely low levels at 7–20 d.p.p., when meiotic stages appear and develop in the testis.
These results show that, in spermatocytes, prior to the onset of MSCI, Zfy transcription is strongly characterized by the differential splicing of exon 6 that is predicted to generate two ZFY protein isoforms with very different acidic domains. The quantitative real-time reverse transcription PCR (RT-qPCR) analysis also shows that the quantity of Zfy1 transcripts with exon 6 increases in spermatids, suggesting that the predominance of Zfy1 transcripts without exon 6 may be specific to spermatocytes.
We also found gene- and stage-specific differences in polyadenylation (polyA) site usage by 3′ RACE (Fig. 4A). Sequencing the major 3′ RACE products showed that Zfy1 uses two polyA sites, one major (3′ UTR: 196 bp) and the other minor (3′ UTR: 124 bp). The major site is 14 bp downstream of a canonical AATAAA motif. The minor site is 15 bp after a non-canonical AATATAAA site.
Zfy2 also uses two polyA sites. However, a comparison of the 3′ UTRs from Zfy1 with those of Zfy2 reveals that the canonical AATAAA polyA signal used by Zfy1 is absent from Zfy2. Instead Zfy2 uses an upstream site corresponding to the minor Zfy1 site (3′ UTR: 124 bp) for its late transcripts (mostly Cypt transcripts) in spermatids, and a downstream site (3′ UTR: ≥284 bp) for its pre-MSCI transcripts, giving it a 3′ UTR that is longer than Zfy1 at the early stage. The major downstream site used by Zfy2 is difficult to define because the major Zfy2 3′ RACE product, at 16 d.p.p., has its 3′ end within a genomic polyadenosine tract (dA14) (Fig. 4A). The true polyA site of the Zfy2 meiotic transcript must, therefore, be situated further downstream.
Conventional PCR from exon 5 to a primer specific for each 3′ UTR in exon 11 of Zfy1 or Zfy2 amplified fragments with and without exon 6, showing that the inclusion or exclusion of exon 6 is not associated with a particular 3′ UTR for either Zfy1 or Zfy2 (Fig. 4B and C). RT-qPCR shows that the levels of Zfy1 transcripts with the shorter 3′ UTR increase between 16 and 27 d.p.p., whereas the levels of the medium 3′ UTR increase between 7 and 16 d.p.p. and then remain constant (Fig. 4B). This is also indicated by the 3′ RACE result for Zfy1 (Fig. 4A), where the short product, at 27 and 50 d.p.p., appears stronger, relative to the medium product, than it does at 7 and 16 d.p.p.. This suggests that the usage of the minor polyA site may increase in spermatids for Zfy1, as is observed for Zfy2.
Although human ZFY also uses alternative polyA sites and the non-canonical AATATAAA site (23), only a short stretch of nucleotide homology remains between the mouse and human 3′ UTR (61 bp with 79% nucleotide identity), beginning, respectively, at 23 and 24 bp after the stop codon. This homology does not include the polyA signals. There is, therefore, no strong evidence that ZFY polyA choice is achieved by a conserved mechanism.
The Sxrb deletion is known to have arisen by unequal exchange between intron 5 of Zfy1 (20 705 bp) and intron 5 of Zfy2 (11 556 bp) (17). The exact breakpoints, resulting in the generation of the Zfy2/1 fusion gene, have not been determined. Since the breakpoints map to the intron that precedes the alternatively spliced exon 6, we localized the breakpoints precisely using Zfy1- and Zfy2-specific STS markers from intron 5 (data not shown), and then amplified and sequenced the Sxrb junction fragment (Fig. 5A and B). This revealed that the Sxrb deletion occurred by an illegitimate cross-over event between Zfy1 and Zfy2 in 95 bp of identity. Based on the position of the breakpoints on the mouse Y chromosome contig NT_078925, we estimate the length of the Sxrb deletion interval to be 1 299 537 bp.
The Sxrb junction fragment was confirmed using a PCR RFLP assay (Fig. 5C). This generated a specific RsaI fragment only when the Sxrb interval was present. This will be a useful screening tool for the Sxrb interval in general, and especially for the identification of XYSxrb carrier males, previously identified through extensive breeding tests.
To better define the consequences of the Sxrb deletion for Zfy expression, we used RT-PCR to analyse the early and late transcripts expressed from the Zfy2/1 fusion gene, in XESxrbO (Zfy2/1 only) and XYSxrb (Zfy1, Zfy2 and Zfy2/1) mice, in particular with reference to the presence or absence of exon 6. XYSxrb mice have full spermatogenesis, whereas in XESxrbO mice, interphasic secondary spermatocytes block prior to the second meiotic division (18,24). The block in XESxrbO mice is leaky, however, and between 30 and 60 d.p.p. abnormal sperm appear in testis tubules (24).
Using a Zfy2-specific primer for exon 5 and a Zfy1-specific primer from exon 10, we specifically amplified the Zfy2/1 fusion gene from XYSxrb and XESxrbO testes at early (10–21 d.p.p.) and late (27–73 d.p.p.) stages (Fig. 6A and B). We detected Zfy2/1 transcripts at both these stages, in Sxrb mice, but not in normal XY males. At early stages, the transcript without exon 6 predominates, showing that the fusion gene is spliced like Zfy1, in spermatocytes. The similarity of the Zfy2/1 and Zfy1 profiles at later stages suggests that this is also the case in spermatids.
Amplification of the Zfy2/1 gene from staged XESxrbO testes with primers in exon 1a or 1b and exon 5 shows the fusion gene to be transcribed from both Zfy2 promoters (Fig. 6B). Transcripts with exon 1a (Cypt) are not detected at early stages (Fig. 6B and C). This is the transcript that was originally detected in adult XYSxrb mice, using primers in exon 1a and exon 6, when the Sxrb deletion mechanism was first described (17). It is noteworthy that Cypt promoter-driven transcripts are readily detected in 30 d.p.p. testis from XESxrbO males (Fig. 6B), because spermatids are not observed by histological examination of XESxrbO testis at this stage (18,24). This is, therefore, an indication that, in XESxrbO mice, certain spermiogenic processes required for the activation of the spermatid-specific Cypt promoter continue independently of the secondary spermatocyte block.
In XYSxrb males, the presence of the Zfy2/1 fusion gene provides an opportunity to investigate how exon 6 of Zfy1 is spliced in spermatids. Using primers specific for exon 1a of Zfy2 and exon 8 of Zfy1, Zfy2/1 fusion gene transcripts were amplified with and without exon 6 (Fig. 6C), showing that the splicing out of exon 6 does occur in spermatids as well as spermatocytes.
The 3′ RACE analysis (Fig. 6D) shows that the Zfy2/1 fusion gene transcripts predominately use the canonical AATAAA polyA signal, and not the upstream AATATAAA signal used by Zfy2 in spermatids.
In conclusion, we show that, besides the loss of the Zfy2 coding region, one of the principal consequences of the Sxrb deletion for the combined expression of the Zfy genes, is a reduced level of transcripts coding the full-length Zfy acidic domain, in spermatocytes and spermatids.
There is a single widely expressed ZFY gene in human suggesting that ZFY regulation may vary greatly between humans and mice. We nevertheless investigated the splicing of human ZFY in human testis. In human ZFY, it is exon 3 that corresponds to exon 6 in mouse Zfy1. Using RT-PCR with primers in exon 2 and exon 4, we identified a transcript lacking exon 3, present in the testis but not in the other tissues tested (Fig. 7A). As in the mouse, exon 3 is the second coding exon. It is 573 bp in length and codes for around half of the putative activating domain of ZFY.
To investigate the possibility that, like its mouse counterpart, the ZFY transcript lacking the second coding exon (exon 3) is expressed in germ cells as they enter meiosis, we tested for its expression in a testicular biopsy from two azoospermic men with a histologically determined meiotic maturation arrest. One, Ste-363, with a deletion of the AZFb interval of the Y chromosome, has a pachytene block, whereas the other, Apop12, with no detectable Y chromosome deletion has an earlier block at the leptotene/zygotene transition. Expression of the ZFY transcripts with and without exon 3 was tested by RT-PCR with primers in exon 2 and exon 4. Transcripts known to be specific to spermatids (PRM1 and TNP1) were also tested to confirm the absence of post-meiotic germ cells, and the HPRT gene was tested as a positive control. RT-PCR analysis showed the presence of ZFY transcripts with and, importantly, without exon 3, in the absence of the spermatid markers PRM1 or TNP1 (Fig. 7B). These results provide evidence that, like Zfy1 in mice, ZFY in humans produces a transcript in germ cells prior to the onset of MSCI that lacks the 573 bases of exon 3 (mouse exon 6).
We characterized indirectly the ZFY transcripts present in human spermatids by the analysis of RNA extracted from purified spermatozoa. During spermiogenesis, transcription ceases when histones are removed from the chromatin as spermatids begin to elongate. A proportion of mRNAs already present in spermatids are nevertheless retained in the mature spermatozoa, representing a sampling of the spermatid transcriptome (25). We tested for ZFY transcripts by RT-PCR with primer combinations specific for transcripts with or without exon 3 (Fig. 7C), and detected both transcripts in spermatozoa RNA. However, compared with the whole testis, where the product from the transcript with exon 3 is the stronger, it is the product from the transcript without exon 3 that predominates in spermatozoa, indicating that the transcript without exon 3 is a major ZFY transcript in spermatids.
To investigate the functional differences between the two isoforms of the ZFY genes in mice and humans, we exploited the fact that the acidic domains of Zfy2 and Zfx have been shown to activate transcription in the yeast Saccharomyces cerevisiae when fused to the DNA-binding domain of the Gal4 transcriptional activator (5). We, therefore, generated six constructs expressing the long or the short acidic domains of mouse ZFY1, mouse ZFY2 and human ZFY fused to the Gal4 DNA-binding domain in the vector pGBKT7 (Clontech). We also generated a construct with the yeast Gal4 acidic domain fused to the Gal4 DNA-binding domain as a positive control, and the Gal4 DNA-binding domain alone (empty vector) as a negative control. These constructs were transformed into the Y187 strain of yeast containing a LacZ (β-galactosidase) gene under the control of the Gal4-dependent GAL1 promoter, and we estimated transcriptional activation by measuring β-galactosidase activity.
Our results with two independent transformants for each construct show a clear difference between the long and the short acidic domains (Fig. 8): whereas the long acidic domain encoded by ZFY, Zfy2 and Zfy1 activated β-galactosidase production like the acidic domain of Gal4, none of the three short acidic domains activated at all. We were able to confirm the expression of all the ZFY-Gal4 fusion proteins by western blot analysis, although ZFY2-long was extremely faint (Supplementary Material, Fig. S1) and the Gal4-AD fusion protein was not detected at all. We suspect that this may be the result of selection against high expressing clones, since it was the ZFY2-long and Gal4-AD fusion constructs that induced the highest β-galactosidase activity, and these yeast transformants alone gave very small colonies and grew very slowly in liquid culture.
ZFY was isolated from the human Y chromosome, over 20 years ago, and was the first gene to be found conserved on the mouse and human Y chromosome (1,8). Most of the early characterization of the two mouse Zfy genes focused on their embryonic expression in the developing gonad (7,9,26), with the aim of testing the hypothesis that ZFY was the primary testis determinant (1). With the discovery that testis determination was performed by SRY, in mice and humans (27,28), interest in ZFY waned. As a consequence, the post-natal transcription of Zfy1 and Zfy2 has not been fully explored.
Our study builds on this previous work to reveal unsuspected intricacies in Zfy transcription during mouse spermatogenesis in XY males and in males carrying the Sxrb deletion. Our study is the first detailed analysis of Zfy transcripts during the early phase of its expression prior to the onset of MSCI (pre-MSCI).
The principal finding of our study is that mouse Zfy1 and human ZFY produce a previously unknown major splice variant in the testis that results in the exclusion of the large second coding exon (exon 6 in Zfy1) that represents a quarter of the full-length ZFY protein and a half of the acidic domain. This splice variant has not been found among ZFX transcripts, including ESTs, in mice or humans (6,29). Only one ZFX transcript without the second coding exon has been identified, but it is from a human lymphoblastoid cell line, and its splicing is different from the short ZFY transcripts since the first coding exon is also spliced out (6). For the mouse, our own RT-PCR analyses (data not shown) with primers specific for the first and last coding exons of Zfx confirmed this, showing that the transcript encoding the full ZFX acidic domain predominates in the testis. A secondary Zfx transcript was detected that lacks the 150 nucleotide third coding exon. This latter transcript certainly gave rise to the Zfa retrogene which lacks the same exon. The production of the short acidic domain through the splicing out of the large second coding exon would, therefore, appear to be an adaptation that is essentially specific to ZFY transcripts in the testis.
Based on their structure, and the fact that the acidic domains of mouse ZFY2 and ZFX have been shown to activate transcription in yeast (5), ZFY proteins are predicted to bind to specific sites in the genome via their zinc finger domain and subsequently recruit regulatory or transcriptional machinery through interactions with their negatively charged acidic domain. We have now shown that, as for ZFY2, the long acidic domain of mouse ZFY1 and human ZFY can activate transcription at Gal4-responsive promoters in yeast. Importantly, however, we show that the short acidic domain from ZFY1, ZFY2 or human ZFY does not activate transcription in yeast.
In yeast, Gal4 is known to activate transcription through the recruitment of a large pre-initiation complex including histone acetyltransferases (HAT) and the TATA-binding protein to the promoter region of target genes, with recruitment dependent on a direct interaction between the Gal4 acidic domain and the Tra1 protein (30). Tra1 is a member of the ATM superfamily, and has been conserved during the evolution of eukaryotes, and its mammalian homologue is TRRAP (transformation/transcription domain-associated protein). TRRAP is common to many HAT complexes and is required for their recruitment to the chromatin during transcription, replication and DNA repair (31). Thus, our data support a basic model in which ZFY-long, but not ZFY-short, recruits TRRAP-containing complexes to specific sites on meiotic chromosomes. ZFY-short still maintains a large acidic domain of 183 amino acids with a net charge of −32 that may recruit distinct complexes to the chromatin during spermatogenesis.
We have recently established that Zfy1 and Zfy2 are functionally distinct in spermatocytes, since we have found that Zfy2, but not Zfy1, is required for the efficient apoptotic elimination of MI spermatocytes with a univalent X chromosome (18). In XESxrbO mice, which express Zfy exclusively from the Zfy2/1 fusion gene, there is reduced MI apoptosis in response to the unpaired XESxrb chromosome. Since the Zfy2/1 fusion gene has the promoter, 5′ UTR and first coding exon from Zfy2, the functional difference between Zfy1 and Zfy2 at MI cannot be related to these parts of the genes. The functional differences must, therefore, be related either to the coding region (46 amino acid coding differences) or to the 3′ UTR. Here, we have shown that, in spermatocytes, the Zfy2/1 fusion gene, the only Zfy gene in XESxrbO mice, mainly produces transcripts lacking exon 6, like Zfy1. Furthermore, we show that exon 6 encodes a segment of the acidic domain that is required for at least a part of the transactivating activity of ZFY1 and ZFY2. Thus, we establish that the absence of exon 6 (183 amino acids) could underlie the functional difference between Zfy1 and Zfy2. We, therefore, hypothesize that the potentiation of MI apoptosis in response to a univalent X chromosome requires a ZFY protein with an acidic domain capable of recruiting TRRAP-containing complexes to the chromatin.
Several male mice that have an unpaired X chromosome, and different combinations of Zfy1 and Zfy2, have been created and studied. Comparisons of XSxraO (Zfy1 and Zfy2) and XEZ2O + Sry-tg (Zfy2 only), or XEZ1/UO + Sry-tg (Zfy1 only) and XEO + Sry-tg (no Zfy gene), have not, however, revealed any effect linked to the Zfy1 gene, during the first meiotic prophase (18). This raises the question of the role during meiosis of Zfy1 specifically and of ZFY proteins with the short acidic domain in general. In this regard, it is interesting to note that, of the 46 amino acid differences between Zfy1 and Zfy2, 34 are within the 363 amino acid acidic domain, and only 4 map to the 372 amino acid zinc finger domain (excluding the Mus musculus-specific six amino acid deletion of the third zinc finger in Zfy2). This indicates that the evolution of both the Zfy1 and Zfy2 zinc finger domains has been constrained, implying that ZFY1 and ZFY2 may have the same DNA targets, and that Zfy1 does have a selectable function. Our finding of a ZFY transcript coding a short acidic domain in humans is a good indication that the short isoform of ZFY and consequently mouse Zfy1 do fulfil a conserved role during spermatogenesis.
There is good evidence that the ZFY1 protein is translated in early spermatocytes, as persistent transcription of Zfy1 in spermatocytes after MSCI, from an autosomally located transgene, causes a mid-pachytene block at stage IV (13). ZFY1 could have a subtle impact on spermatocyte development, improving the efficiency or the fidelity of the spermatogenic process. Like ZFY2, ZFY1 could be involved in the regulation of the apoptotic response at MI, perhaps acting to reduce apoptosis of normal XY spermatocytes by antagonizing ZFY2 binding to its target genes. An anti-apoptotic effect for ZFY1 might be evident as increased MI apoptosis in the absence of unsynapsed chromosomes, in male mice lacking Zfy1, but expressing Zfy2.
The fact that the continued expression of either Zfy gene in spermatocytes, during MSCI, results in a mid-pachytene block (13) indicates that Zfy1 and Zfy2 could have the same mode of action at this stage, despite their differing capacity to promote apoptosis later at MI (18). This could be a consequence of the highly similar zinc finger domains of ZFY1 and ZFY2 proteins binding to the same promoter regions and reducing access to other transcription factors. However, an alternative explanation is suggested by the fact that a proportion of Zfy1 transcripts do code for the full-length acidic domain. Our RT-qPCR data for Zfy1 and Zfy2 (Fig. 3C) indicate that, during meiosis, transcripts with exon 6 are approximately 2-fold lower from Zfy1 than from Zfy2. However, the Zfy1-transgenic animals, in which the stage-IV arrest was seen, carried 14 copies of the Zfy1 transgene (13), and thus may have produced a level of the full-length protein equivalent to, or greater than, that produced by Zfy2, and it was this that triggered the arrest. This explanation is consistent with a model where it is only the ZFY protein with the full-length acidic domain that promotes apoptosis in spermatocytes. The importance of the full acidic domain to the known functions of Zfy genes could be tested with transgenes engineered to express Zfy1 with exon 6, and Zfy2 without exon 6.
Our data raise the possibility that, as regards Zfy function during meiosis in the mouse, the important distinction may not be between Zfy1 and Zfy2, but between full or partial acidic domains. Seen in this way, humans and mice may be understood to be using different means to achieve a similar expression of these two ZFY isoforms. Human ZFY expression more closely resembles that of Zfy1 than that of Zfy2, in that ZFY and Zfy1 are capable of producing transcripts with and without the second coding exon in germ cells, whereas Zfy2 almost exclusively produces full-length transcripts. This might be an indication that Zfy1 assures the core eutherian ZFY functions, whereas Zfy2 has specialized in the elimination of spermatocytes in which the X and the Y have failed to pair through their PAR. Studies in the mouse have recently shown that the reduced size of the PAR has been the selective force for several adaptations, probably established before eutherian radiation, to ensure double-strand break formation and crossing-over in the PAR (32). The mouse PAR is 700 kb (33), which is nearly four times shorter than the human PAR (2.7Mb) (34). The Zfy2 gene may, therefore, be an adaptation selected for by the shortening of the PAR in the rodent lineage.
Our finding that human ZFY produces a transcript without the equivalent of exon 6 that predominates in the testis and is present in the testis that lack post-meiotic stages establishes ZFY as a candidate gene for human male factor infertility. Interestingly, like Zfy in the mouse, Usp9y is a testis-specific gene whose transcription increases as germ cells begin to enter meiosis in the testis (35). Furthermore, Zfy and Usp9y are the only testis-specific mouse Y chromosome genes whose human orthologues are widely expressed (36). Isolated deletions of USP9Y (five cases) have only been found in the male partners of couples consulting at infertility clinics, suggesting that they do reduce male fertility (37–40). In three cases, however, the USP9Y deletion has been transmitted naturally, and the phenotype in most cases is mild; moderate oligozoopermia in two cases (37), and normospermia in one case (38), indicating that USP9Y increases the efficiency of spermatogenesis without being essential for male fertility. ZFY mutations could, therefore, have similar slight effects on male fertility.
To begin to evaluate the extent to which mutations in ZFY are a cause of human male fertility, we have screened 520 infertile men for the deletion of the ZFY gene, using PCR tests for the two ends and the middle of the gene, but have failed to find any deletions (M.J.M. and E.S., unpublished data). The incidence of large deletions affecting ZFY is, therefore, low (< 0.2%) in our infertile population. Nevertheless, it is now established that, in the mouse, ZFY2 contributes to the stringency of the pairing checkpoint at MI (18), and ZFY might play a similar role during human spermatogenesis. If this is the case, it will be important to identify those rare men with a mutation in ZFY, as this could have consequences for the stable transmission of the genome to the next generation.
Staged testes (day of birth = 0 d.p.p.) were from: (i) XY C57BL/6 males (Charles River, France), (ii) XYSxrb males that carry the variant sex reversal factor Tp(Y)1CtSxr–b attached distally to the Y PAR, together with their XY littermates (Random-bred MF1 stock background—NIMR colony) and (iii) XESxrbO males that carry an Eif2s3y transgene inserted on the X chromosome and Sxrb attached distally to the X PAR (also on an MF1 background). The production of the XESxrbO males has been described elsewhere (18).
Testicular samples were obtained from two azoospermic patients who underwent a testicular biopsy in order to recover testicular spermatozoa within the context of an intracytoplasmic sperm injection (ICSI) attempt. Each patient had a normal karyotype. Patient Ste-363 from St Etienne has a P5/proximal-P1 deletion of the AZFb interval of the Y chromosome and has been described previously (41). Patient APOP12, from Marseille, has no detectable Y chromosome deletion. He gave his informed consent for the storage of the immature testicular cells in our ‘Germetheque’ biobank, and their use in research. A meiotic study performed on cells from Apop12 revealed an early meiotic block, with 97% of spermatocytes at the leptotene stage (C.M.-G., unpublished data). In both cases, the material was the remains of a testicular biopsy after it had been teased apart to recover spermatozoa for ICSI.
Human ejaculates were collected by masturbation from normal men, consulting with their partner for couple infertility, through the Sperm Diagnostic Unit, Biology of Reproduction Laboratory, La Conception Hospital, Marseilles, France. Men participating in this study consulted for a sperm analysis as part of the exploration of the infertility of their couple, and gave informed consent that the remaining part of the semen sample could be stored and used in research instead of being discarded. Samples were included in the ‘Germetheque’ biobank. Sperm analysis was performed according to the WHO criteria and morphology was assessed according to the French DAVID classification (Table 1). After sperm analysis, the remaining sperm was centrifuged at 3000 r.p.m. for 15min. The pellet was frozen and stored in liquid nitrogen until use.
For all mouse samples, polyA+ RNA was isolated using the μMACs mRNA Isolation Kit (Miltenyi Biotec) according to the manufacturer's protocol. For human testis biopsies, total RNA was extracted with TriPure (Roche), as previously described (41). Total human RNAs from normal human testes and other tissues were purchased from Clontech. To extract RNA from spermatozoa, pelleted samples were thawed and washed twice in 2 ml of PBS. They were then resuspended in round cell lysis buffer (0.1% SDS, 0.5% Triton X-100 in RNase-free H2O) for 10 min on ice and then washed twice in PBS. This treatment preferentially lyses somatic cells and immature germ cells while leaving spermatozoa intact (42). RNAs were extracted from 15 × 106 spermatozoa with 1 ml of TriPure (Roche) following the manufacturer's protocol. RNA was precipitated twice with isopropanol in the presence of 20 μg of glycogen (Roche). The RNA was subsequently treated with DNase I and RNase-free (Roche) and then either purified by passage through a Chroma Spin-200 + DEPC-H2O column (Clontech) (samples 08153 and 09203) or by isopropanol precipitation (sample 09175).
For the mouse, in general, 500 ng of mRNA were converted into cDNA with expand RT (Roche) and random nonamers in a 20 μl reaction volume. Less mRNA was used for the XESxrbO cDNAs, because RNA was extracted from testis fragments and yields were <500 ng, but in all these cases, the corresponding XY control cDNA was generated from an equivalent amount of RNA. Mouse cDNAs were diluted 10-fold and 1 μl, or a volume corresponding to 2.5 ng of input mRNA, used in a final PCR volume of 12 μl. PCR was performed using standard protocols.
For 3′ RACE, cDNAs were synthesized using an oligo-dT-adaptor primer (GACTCGAGTCGACATCGA-dT17). RACE PCR was performed as nested PCR with a single reverse adaptor primer (o2761) and two forward primers specific for Zfy1 (o2828 then o2851) or Zfy2 (o2827 then o2852). The first PCR was for 30 cycles and then 1 μl was used for a second PCR of 30 cycles.
For humans, except spermatozoa, 1 μg of total RNA was converted to cDNA in a 20 μl reaction, diluted 10-fold and 1 μl used for PCR in a 12 μl reaction. For spermatozoal RNA, all recovered total RNA was converted to cDNA, diluted 2-fold and 1 μl used in a 12 μl PCR reaction. RT reactions were random-primed.
All primers used are listed in Supplementary Material, Table S1.
Individual PCR products were cut from agarose gels and purified using NucleoSpin Extract II (Macherey-Nagel) according to the manufacturer's protocol, and sequenced using BigDye Terminator v1.1 Cycle Sequencing Kit (Applied Biosystems).
Sequences were then analysed with the Sequencher 4.2 software (Genecodes). Comparisons were made with sequences deposited in the NCBI database: mouse Zfy1 (NM009570), Zfy2 from its Cypt (AK133222) or its Zfy promoter (AK030048) and human ZFY (NM003411).
Quantitative RT-PCR was performed with the LightCycler 480 SYBR Green I Master Kit (Roche) on the LightCycler 480 (Roche). Reactions were performed in 10 μl. Amplification conditions were 1 cycle at 95°C for 5 min, followed by 43 cycles of 95°C for 10 s, 58 or 60°C for 15 s, 72°C for 15 s. Controls with H2O were negative for all tests. All reactions were carried out in triplicate per assay and Hmbs was included on every plate as a loading control. The melting curve profiles were analysed for every sample to verify the absence of non-specific products or primer dimers.
For each sample, the mean Ct for Hmbs was subtracted from the mean Ct for Zfy (ΔCt). The value for the Zfy transcript level was calculated from this ΔCt by attributing the arbitrary value of 1000 to the Hmbs transcript level, using the formula EΔCt × 1000, where E is the PCR efficiency. E was calculated for each primer pair, using a series of dilutions of template cDNA. The slope (a) of the standard curve, Ct = a(log [conc])+b, was obtained by plotting log10 cDNA concentration, [conc], against Ct, and E was calculated as 10(−1/a). For each plot, R2 was >0.99. If a reaction has 100% efficiency, E = 2. Values for E ranged from 1.88 to 2 (Supplementary Material, Table S1).
We chose the Hmbs gene as control housekeeping gene because it showed low variability in Ct between all testis stages when equivalent quantities of cDNA were used as template.
RNA FISH was performed as previously described (12,20). Wild-type testis material was from adult MF1 male mice. To provide control material to monitor probe specificity, we used 30 d.p.p. testes from XEZ1/UOSry or XEZ2OSry transgenic males; these males carry BAC transgenes for Eif2s3y (E) together with a BAC transgene for Zfy1+ Ube1y (Z1/U) or Zfy2 (Z2), on their single X chromosome, with testis development driven by an autosomally located Sry transgene (13,18). The Zfy2-specific probe used was the BAC CITB-288D7 (Research Genetics). The Zfy1-specific probe used was a modified version of BAC RP24-498K8 (CHORI) from which we had removed the entire Ube1y gene by recombineering.
ZFY-Gal4 fusion-protein constructs were made by inserting PCR-amplified cDNA segments encoding the different acidic domains into the NcoI and SalI restriction sites of pGBKT7 (Clontech), downstream of the GAL4 DNA-binding domain and the c-myc epitope tag of the vector. Long and short acidic domains were amplified together for each ZFY gene from testis cDNA, and each fragment was purified following size fractionation on an agarose gel. The GAL4 acidic domain (GAL4-AD) was integrated into pGBKT7 as a positive control and was amplified from the vector pGADT7 (Clontech). Primers used, with respectively an NcoI or SalI adaptor, were ZFY: o4108/o4109, Zfy1: o4100/o4101, Zfy2: o4102/o4103 and GAL4-AD: o4106/o4107 (Supplementary Material, Table S1). Recombinants were verified by sequencing the entire PCR product and vector flanks. Two recombinants were selected for each construct and transformed into cells from the S. cerevisiae strain Y187, in which the β-galactosidase gene is under the control of the Gal4-responsive GAL1 promoter. Single-transformed colonies were picked from SD/-trp agar plates and grown in liquid culture to an OD600 of 0.6–1.2 in SD/-trp liquid minimal medium. The β-galactosidase assay was performed on ~1 OD600 unit of the culture, using the permeabilized cell assay (43). Another 2.5 OD600 units from the same culture were used to prepare protein for standard western blot analysis using an anti-human c-myc tag antibody (Biolegend) (Supplementary Material, Fig. S1).
We are very grateful to the men who consented to their biopsy material or sperm sample being used for our research. We thank the Reproductive Biology Service at the Hôpital de La Conception for their invaluable help in collecting the human material and, in particular, Professor Jean-Marie Grillo (Service Head), Professor Marie Guichaoua and the technical staff, Cathy Metton and Marie-Josée Fays-Bernardin. Thanks also to Frédérique Lembo and Jean-Paul Borg of the Marseille Cancerology Research Centre (CRCM) for helping us set up the transactivation assay in yeast. P.S.B. is indebted to MRC NIMR Procedural Service Section for the production of GA lines and to Biological Services for animal husbandry and technical support. Thanks also to Helen Skaletsky and Jennifer Alfoldi for kindly sending us the RP24-498K8 BAC clone.
Conflict of Interest statement. None declared.
This work was supported by core funding from the Institut national de la santé et de la recherché médicale (Inserm) to M.J.M.; an Inserm-Région Provences-Alpes-Côtes d'Azur doctoral fellowship to F.D.; a French Ministry of Research and Technology doctoral fellowship to E.S.; the Medical Research Council UK (U117532009 to P.S.B., and MRC CDF to N.V.) and an EMBO fellowship to N.V. Funding to pay the Open Access publication charges for this article was provided by the UK Medical Research Council.