Re-visiting the Protamine-2 locus: deletion, but not haploinsufficiency, renders male mice infertile

Protamines are arginine-rich DNA-binding proteins that replace histones in elongating spermatids. This leads to hypercondensation of chromatin and ensures physiological sperm morphology, thereby protecting DNA integrity. In mice and humans, two protamines, protamine-1 (Prm1) and protamine-2 (Prm2) are expressed in a species-specific ratio. In humans, alterations of this PRM1/PRM2 ratio is associated with subfertility. By applying CRISPR/Cas9 mediated gene-editing in oocytes, we established Prm2-deficient mice. Surprisingly, heterozygous males remained fertile with sperm displaying normal head morphology and motility. In Prm2-deficient sperm, however, DNA-hypercondensation and acrosome formation was severely impaired. Further, the sperm displayed severe membrane defects resulting in immotility. Thus, lack of Prm2 leads not only to impaired histone to protamine exchange and disturbed DNA-hypercondensation, but also to severe membrane defects resulting in immotility. Interestingly, previous attempts using a regular gene-targeting approach failed to establish Prm2-deficient mice. This was due to the fact that already chimeric animals generated with Prm2+/− ES cells were sterile. However, the Prm2-deficient mouse lines established here clearly demonstrate that mice tolerate loss of one Prm2 allele. As such they present an ideal model for further studies on protamine function and chromatin organization in murine sperm.


Results
CRISPR/Cas9-mediated generation of Prm2-deficient mice. Previously, it was reported that chimeric mice generated by injecting Prm2 +/− ES cells were sterile 29 . In general, ES cell lines are of male sex and mediate sex-reversal when injected into female host blastocysts. Thus, a mouse line harboring a mutation affecting the male germline, like the Prm2-heterozygosity, cannot be established by this technology, precluding detailed research of protamine biology. Since protamines are only required for male germ cell development, establishment of Prm2-deficient mouse lines should be possible from female founder animals. Therefore, we decided to apply CRISPR/Cas9-mediated gene-editing in murine zygotes to generate Prm2-deficient mice 31 .
The murine Prm2 gene consists of two exons, encoding the 106 amino acids-long precursor protein (Fig. 1a). We derived three guide RNAs (gRNAs) targeting the first exon close to the translational start site (Fig. 1a). First, the gRNAs were validated in murine ES cells. CRISPR/Cas9 induces double-strand breaks, leading to error-prone non-homologous end-joining mediated repair 32 . This results in formation of small insertions or deletions (indels), which can be detected by the 'Surveyor nuclease' , resulting in cleavage of PCR products. The gene-edited Prm2-coding region was amplified by PCR and subjected to the Surveyor assay 33 , which confirmed the functionality of derived gRNAs, as indicated by cleavage of the PCR products (Fig. 1b, cleavage products highlighted by arrowheads).
To edit the Prm2 gene in murine zygotes, we microinjected in vitro transcribed single-gRNAs (sgRNAs) and Cas9 mRNA into the cytoplasm. A combination of sgRNAs 1 and 2 or sgRNAs 1 and 3 was applied to induce deletions of approximately 100 bp. From 336 injected zygotes, 245 (72.9%) survived (Table 1); 82 blastocysts were transferred and 10 viable pups were born (Table 1). Genotyping by PCR and Surveyor assay identified 4/10 pups carrying CRISPR/Cas9-induced mutations within the Prm2 locus (Fig. 1c,d). Of those, three animals (# 2, 6, 7) contained small indels, indicating that one sgRNA had catalyzed gene-editing. One female (#10) harbored the intended deletion of approximately 100 bp (Fig. 1c). Sequence analysis of genomic DNA from #10 revealed that the CRISPR/Cas9-mediated gene-editing resulted in four different alleles. Besides the wildtype allele, three alleles with deletions of 97 bp, 101 bp, and 103 bp were obtained. This suggests that gene-editing took place at the 2-cell-stage. Alleles with deletions of 97 bp and 101 bp were separated by back-crossing with C57BL/6 wildtype mice. Haploinsufficiency of Prm2 does not affect male fertility. Two mouse lines, Prm2Δ 97 (harboring a 97 bp deletion) and Prm2Δ 101 (harboring a 101 bp deletion), were established. On both alleles, a frameshift was generated, resulting in nonsense transcripts with a premature stop codon and deletion of Prm2 (Fig. 2a). In contrast to earlier reports 29 , male mice heterozygous for either Prm2Δ 97 or Prm2Δ 101 were fertile and sired offspring. Matings of males heterozygous for either Prm2Δ 97 or Prm2Δ 101 produced an average litter size of 8.6 (Fig. 2b). There was no significant difference in the average litter size among the two mouse lines (8.8 vs. 8.4). This is comparable to published data for wildtype mice with a mean litter size of 6.2 (9.5 in the plateau period of reproduction) 34,35 and our wildtype matings with an average litter size of 7.3. In contrast, male mice homozygous for either Prm2Δ 97 or Prm2Δ 101 were not able to reproduce (Fig. 2b), whereas female mice homozygous for either Prm2Δ 97 or Prm2Δ 101 were fertile and sired offspring with an average litter size of 8.5, when mated with heterozygous males (Fig. 2b). These results clearly indicate that loss of one allele of Prm2 can be tolerated, whereas complete lack of Prm2 results in male sterility.
To validate that gene-editing generated a Prm2 loss-of-function allele, we analyzed full-length Prm2 RNA levels. Since Prm2 is expressed during spermiogenesis in step 7-15 spermatids, qRT-PCR was performed on RNA isolated from murine testis 36 . In heterozygous animals, relative Prm2 mRNA expression levels were reduced by approximately 50% compared to wildtype animals (Fig. 2c). Sperm from males homozygous for either mutation did not show any Prm2 signal, indicating that the gene-editing led to successful deletion of the Prm2 gene (Fig. 2c).
Immunohistochemical staining (IHC) using an anti-PRM2 antibody on testicular tissue sections detected the protein in spermatids at stage VIII of the seminiferous epithelial cycle of both, wildtype and Prm2 +/− mice (Fig. 2d). The staining appeared less intense in heterozygous mice, suggesting a reduction in protein level due to decreased mRNA levels. In accordance with qRT-PCR data, sections from Prm2 −/− testes were negative for PRM2 at all stages of the mouse seminiferous epithelial cycle, further confirming the deletion of the gene (Fig. 2d, right).
Western blot analysis of basic nuclear proteins isolated from epididymal sperm was performed to determine protein levels of PRM2. Males heterozygous for either Prm2Δ 97 or Prm2Δ 101 displayed an approximately 50% decrease of mature PRM2 compared to wildtype controls (Fig. 2e). In Prm2 −/− mice, mature PRM2 protein was not detectable (Fig. 2e).
Prm2-deficient sperm display severe morphological defects. Because Prm2 +/− males were fertile, we were able to breed Prm2 −/− animals and analyze the effects of loss of Prm2 in more detail. Since testis weight (Fig. 3a), epididymal sperm count (Fig. 3b), and testis morphology was not changed in Prm2 +/− and Prm2 −/− animals compared to wildtype (Fig. 3c), we concluded that spermatogenesis is not affected. To investigate why Prm2-deficient male mice are infertile, we analyzed the morphology of testicular spermatids and epididymal sperm.
Electron microscopy of Prm2-deficient testicular spermatids revealed impaired nuclear matrix morphology, exhibiting a fine-grained, sometimes coarse-grained texture (Fig. 3d). Anomalies in chromatin condensation were obvious from step 12 spermatids onward. Until sperm release, the heterogeneity in chromatin packaging, including vesiculation, became more pronounced (Fig. 3e). However, sperm head elongation and shaping were not affected in testicular spermatids (Fig. 3d,e).
Next, epididymal sperm were co-stained with PNA-FITC to label the acrosome, MitoTracker to label mitochondria in the midpiece of the flagellum, and DAPI to label the DNA in the sperm head (Fig. 3h). Wildtype mice and mice heterozygous for Prm2Δ 97 or Prm2Δ 101 showed the typical sickle-shaped head with the acrosome lining the sperm head, whereas Prm2-deficient sperm lacked the acrosome and the midpiece of the flagellum was wrapped around the sperm head (Fig. 3h). The majority of Prm2-deficient sperm showed this phenotype, which was confirmed by electron microscopy (Fig. 3f). On the ultrastructural level, Prm2-deficient sperm displayed detachment of the acrosome at the acrosome-nuclear interface, giving rise to an abnormally shaped head with an undulating plasma membrane (Fig. 3f). In line with observations from testicular spermatids, chromatin integrity was highly disturbed in epididymal sperm (Fig. 3f). Quantification of chromatin integrity revealed that around 80% of Prm2-deficient sperm showed intermediate to low chromatin density. No difference was observed between wildtype and Prm2 +/− animals with around 80% of sperm showing high chromatin integrity (Fig. 3g). Thus, loss of Prm2 not only affects DNA integrity in the sperm head, but also bending of the flagellum and anchorage of the acrosome. To analyze whether a loss of DNA integrity makes sperm more susceptible to DNA damage, we subjected genomic DNA from sperm to agarose gel electrophoresis. While a band indicative of high molecular weight DNA suggested little to no DNA damage in sperm of Prm2 +/+ and Prm2 +/− animals, the majority of sperm DNA from Prm2-deficient mice was strongly degraded into fragments of around 250 bp in size (Fig. 3i). Next, sperm were treated with a strong acid to induce denaturation of DNA, followed by staining with acridine orange and FACS analysis, referring to the sperm chromatin structure assay (SCSA). Intercalation of acridine orange into intact, double-stranded DNA results in emission of green fluorescence, whereas incorporation into damaged,  single-stranded DNA results in red fluorescence. In accordance with the results from the agarose gels, red fluorescence intensity was strongly increased in Prm2-deficient sperm compared to Prm2 +/+ and Prm2 +/− animals (Fig. 3j). In total, more than 80% of Prm2-deficient sperm cells showed intense red fluorescence indicating a strong DNA damage. Although there is a slight increase in sperm damage in Prm2 +/− animals compared to wildtype mice, more than 90% of Prm2 +/− sperm presented intact (Fig. 3k).  Prm2-deficient sperm are immotile. To reveal whether sperm motility in addition to sperm morphology was affected by loss of Prm2, we isolated wildtype, heterozygous, and Prm2-deficient sperm and analyzed the motility of freely swimming sperm using computer-assisted semen analysis (CASA) and the flagellar waveform of tethered sperm. While motility parameters of freely swimming sperm form wildtype and heterozygous Prm2Δ 97 or Prm2Δ 101 animals were not significantly different ( Table 2, Fig. 4a), homozygously Prm2-deficient sperm were completely immotile (Fig. 4b). Wildtype and heterozygous mice displayed a symmetrical flagellar waveform, which was absent in Prm2-deficient sperm (Fig. 4b). Sperm function, in particular sperm motility, crucially relies on Ca 2+ signaling [37][38][39] . Thus, we analyzed Ca 2+ signaling in Prm2-deficient mice using the kinetic stopped-flow technique to measure changes in the intracellular Ca 2+ concentration in sperm populations loaded with the Ca 2+ dye Cal520-AM. We applied a number of different stimuli to evoke a Ca 2+ influx in sperm. Stimulation with K8.6, a buffer that depolarizes and alkalizes the cell and thereby opens the principal Ca 2+ -channel in mammalian sperm, CatSper, evoked a Ca 2+ response in wildtype and Prm2-heterozygous, but not in deficient mice (Fig. 4c). Similarly, alkalization of sperm using NH 4 Cl and direct opening of CatSper using 8-Br-cAMP evoked a Ca 2+ response in wildtype and Prm2-heterozygous, but not in deficient sperm (Fig. 4c). As a control, we applied the Ca 2+ -ionophore ionomycin, which resulted in Ca 2+ influx in controls and Prm2-heterozygous, but not in Prm2-deficient sperm (Fig. 4c). To test whether Prm2-deficient sperm were loaded with Cal520-AM, we compared Prm2-heterozygous and deficient sperm using fluorescence microscopy (Fig. 4d). Heterozygous sperm were loaded with the dye, showing prominent fluorescence in the midpiece, whereas loading of Prm2-deficient sperm was negative (Fig. 4d). Of note, low levels of fluorescence could also indicate that the intracellular Ca 2+ concentration in Prm2-deficient sperm is extremely low. However, this can be excluded since treatment with ionomycin did not evoke a Ca 2+ response. Together with the results obtained by transmission electron microscopy (TEM), this suggests that Prm2-deficient sperm have severe plasma membrane defects, which might prevent dye loading.
Prm1 expression is maintained in Prm2-deficient mice. Next, we analyzed the effects of Prm2-deficiency on the overall histone-to-protamine exchange dynamics. Western blot analysis of basic nuclear proteins isolated from epididymal sperm revealed weak levels of histone H3 in wildtype sperm (Fig. 5a). This is in accordance with previous studies, which described that around 1% of histones remain bound to sperm chromatin after histone-to-protamine exchange 40 . Interestingly, an increased level of H3 was observed in sperm from Prm2 +/− animals, whereas no H3 was detectable in Prm2-deficient sperm (Fig. 5a). No differences in H3 level were observed in protein extracts from testicular tissue (Fig. 5a). IHC stainings gave evidence that the histone replacement is functional. While spermatogonia, round spermatids, and elongating spermatids stained positive for H3, fully elongated spermatids displayed no or only very weak staining (Fig. 5b). Immunoblotting against TNP1 revealed a complete absence in epididymal sperm (Fig. 5a). As a control, presence of TNP1 in testicular tissue was shown. In accordance with this, IHC staining revealed expression of TNP1 in elongating spermatids. In elongated spermatids, where transition proteins are assumed to be completely exchanged by protamines, staining for TNP1 was only observed in residual bodies of Prm2 +/+ and Prm2 +/− spermatids (Fig. 5b). Interestingly, sections of Prm2-deficient animals showed an additional staining within elongated spermatids (Fig. 5b), suggesting a disturbed TNP1 exchange.
The importance of the species-specific ratio of protamines for male fertility suggests that Prm1 and 2 expression levels are tightly controlled. Therefore, we next analyzed the effects of Prm2-deficiency on Prm1 gene expression. qRT-PCR analysis revealed no significant differences in testicular Prm1 mRNA levels in males heterozygous or deficient for Prm2 compared to wildtype controls (Fig. 5c). IHC staining detected PRM1 protein in elongating spermatids of wildtype, heterozygous, and homozygous Prm2Δ 97 and Prm2Δ 101 animals (Fig. 5b). In the seminiferous epithelial cycle, cells of step 10 were the first cells expressing PRM1. Signal intensity was highest in step 12-15 elongated spermatids, marking the replacement of transition proteins by protamines. Western blot analysis of basic nuclear proteins isolated from epididymal sperm further demonstrated the successful incorporation of PRM1 into sperm chromatin (Fig. 5d). Interestingly, PRM1 protein level were slightly decreased in Prm2 +/− animals, but increased in Prm2-deficient sperm compared to controls (Fig. 5d). Of note, for western blot analysis the basic proteins like the protamines were isolated from precipitated DNA. The fact that we could detect PRM1 in PRM2-deficient sperm suggests that PRM1 is incorporated into sperm chromatin independent of PRM2.

Discussion
In the present study, we used CRISPR/Cas9 gene-editing to generate Prm2-deficient mice to investigate the role of PRM2 in controlling chromatin hypercondensation and fertility. We demonstrate that Prm2-heterozygous  mice remain fertile with sperm being morphologically and functionally indistinguishable from wildtype sperm. However, lack of Prm2 causes infertility with severe defects in sperm head morphology and sperm motility. Although the role of protamines for sperm function has been described many times for different species 41 , our observation that loss of one Prm2 allele does not affect male fertility in mice is contradictory to previous studies. Cho et al. described that already chimeras produced by injecting Prm2 +/− ES cells were sterile 29 . We hypothesize that these fundamental differences between our study and the observations made by Cho et al. might be due to the different experimental approaches used to generate Prm2-knockout alleles. Cho et al. applied a classical gene-targeting approach in ES cells to disrupt the Prm2 coding region by insertion of a Pgk-neo cassette 29 . This cassette might have affected the expression in the tightly regulated 11.7 kb spanning gene cluster encoding for Prm1-Prm2-Tnp2 12 . Insertion of the Pgk-neo cassette into the Prm2 coding region might have resulted in silencing of the neighboring Prm1 gene. In fact, such a trans-silencing has been described for the knockout of the herculin (Myf6) gene. Here, the Pgk-neo cassette introduced by homologous recombination to disrupt the Myf6 locus silenced expression of the nearby myogenic factor 5 (Myf5) gene 42,43 . We propose that the trans-silencing of the Prm1 allele may be caused by the Pgk-neo cassette inserted into the Prm2 allele. This assumption is supported by the fact that testes of chimeric mice generated by injecting Prm2 +/− ES cells displayed a strong reduction in PRM1 protein level 29 . In our model, qRT-PCR clearly demonstrates that expression of Prm1 is maintained in Prm2 +/− males, while only a mild decrease in protein level is observed.
Further, the hypothesis that physiological PRM2 protein levels are required for proper incorporation of PRM1 into sperm chromatin has to be re-visited 29 , since sperm of Prm2 +/− and Prm2 −/− males displays certain degrees of DNA-hypercondensation indicative for DNA-protamine interaction. Here, we utilized the CRISPR/Cas9 system to generate Prm2-deficient mice. This state-of-the-art technique allows to efficiently edit genes without introducing additional DNA elements like loxP sites and/or antibiotic resistance cassettes 31 . Off-target activity is a highly discussed issue in the field of CRISPR/Cas9-mediated genome editing. Nonetheless, different approaches analyzing potential off-target effects in genetically altered mice and ES cells, which closely resemble the situation in zygotes, showed either no or only a very low incidence for CRISPR/ Cas9-mediated off-target activity 44,45 , when compared with analyses carried out (mainly) in cancer cell lines, which are hallmarked by genomic instability [46][47][48] . To reduce the risk for off-target events, we selected only gRNAs with very low predicted off-target activity. Furthermore, if possible we used 18-mer guide sequences instead of 20-mer as this was shown to further decrease the frequency of Cas9-mediated off-target events 49 . In addition, the Prm2Δ 97 and Prm2Δ 101 alleles were backcrossed with wildtype mice. Thus, potential off-target effects, which might have affected loci on other chromosomes, are lost during further breeding.
We established two mouse lines, Prm2Δ 97 and Prm2Δ 101 , with N-terminal deletions of 97 and 101 bp, respectively. Both mutations did not interfere with or deleted regulatory regions and left the overall structure of the gene cluster intact. Male mice heterozygous for the respective mutation were fertile and sired offspring with an average litter size of 8.6, which is within the range of the reproductive performance of wildtype mice with a mean litter size of 6.2 and a maximum mean litter size of 9.5 in the plateau period of reproduction 34,35 . This clearly indicates that deletion of one Prm2 allele does not affect male reproductive performance in mice. The typical apical hook-like structure of sperm heads and intact sperm motility further suggests that sperm physiology and morphology are not affected.
This raises the question whether disturbed chromatin integrity or the loss of sperm motility and detachment of the acrosome is the main reason for infertility of knockout mice. The latter is supported by Takeda et al., who were able to generate offspring from Prm1 +/− males by in vitro fertilization of zona-free oocytes, although chromatin integrity of those sperm was severely affected 50 . Nontheless, strong sperm DNA degradation observed in Prm2-deficient animals most likely prohibits successful reproduction.
Protamines are expressed in haploid spermatids, hence in Prm2 +/− mice, only half of the spermatids express Prm2. However, IHC staining revealed PRM2 protein in all spermatids of seminiferous tubules. While transcript or protein sharing among syncytial spermatids has been discussed in general, the IHC stainings shown here clearly demonstrate that such sharing exists for Prm2 29,51-53 .
Of note, knockout male mice heterozygous for the transition proteins TNP1 and TNP2 display preserved fertility as well, while complete deletion of both alleles resulted in subfertility 54,55 .
Prm2-heterozygous mice display reduced levels of PRM2 protein, compared to controls, and show maintained expression of PRM1. We demonstrate that these mice sire normal litter sizes indicative of full fertility. Western blot showed that in Prm2-heterozygous mice deletion of one allele of Prm2 not only led to a reduction of PRM2 protein but also caused a moderate decrease of PRM1 protein level. This is suggestive for a PRM1/2 coregulation in order to maintain the mouse specific PRM1/2 ratio. Recent analyses revealed that PRM1/PRM2 ratio range from 65% (M. castaneus and M. domesticus) to 99% (M. macedonicus and M. spicilegus) in different mouse subspecies 56 . This indicates that mice might be able to tolerate changes in the PRM1/2 ratio to some extent. This would be in contrast to the situation in humans, where changes of the PRM1/2 ratio are associated with reduced fertility. Further, the elevated levels of histones detected in such animals that remain bound to sperm chromatin might compensate for the lower total amount of protamines.
Prm2-deficient sperm displayed upregulation of PRM1. We hypothesize that the higher PRM1 levels are an attempt to compensate for the loss of PRM2. A similar compensatory effect has been described for Tnp2-deficient mice, which showed enhanced Tnp1-expression 55 . Sperm from Prm2 −/− animals lost the characteristic hook-like structure: sperm exhibited round heads and were immotile. Interestingly, a recent study suggests that the cleaved N-terminal part of PRM2 rather than the chromatin-bound mature PRM2 is involved in controlling sperm head morphology 57 . However, the underlying molecular mechanism is not known. In TEM analysis, we were able to show a detachment of the acrosomal cap and a less-condensed chromatin. Similar observations were made for protamine-deficient human sperm 58 . Whereas the reason for the defect in chromatin packaging is obvious, the molecular mechanisms underlying the ultrastructural change in acrosomal cap formation and attachment to the nucleus need to be addressed in further studies.
But why are sperm from Prm2-deficient mice immotile? Once histones are replaced by protamines, transcription is silenced 6 . Thus, transcriptional regulation of other genes by Prm2 seems unlikely. Further, in Prm2-deficient males, development of testicular spermatids is not affected and sperm counts are comparable. Profound morphological aberrations in Prm2-deficient mice are first apparent in the epididymis with detachment of acrosome and bending of the flagellum towards the head. In parallel, treatment with Ca 2+ ionophores indicates that epididymis derived Prm2-deficient sperm presents with defects in membrane function. It remains to be determined whether this is a defect in the self-assembly of sperm or a built-in quality control mechanism triggering a self-destructing program. Of note, impaired sperm motility is a common phenotype of Prm1, Prm2, Tnp1, and Tnp2 knockout mice 29,50,54,55,59 .
With Prm2Δ 97 and Prm2Δ 101 mouse lines, we have established, for the first time, a robust and reliable model for functional studies on protamine-induced chromatin hypercondensation during spermiogenesis. These mice allow for a detailed investigation of basic regulatory mechanisms of haploid gene expression in sperm. Shedding light on these molecular mechanisms is an absolute requirement for a better understanding of aberrant protamine expression in subfertile men.

Material and Methods
Ethics statement. All animal experiments were conducted according to German law of animal protection and in agreement with the approval of the local institutional animal care committees (Landesamt für Natur, Umwelt und Verbraucherschutz, North Rhine-Westphalia, approval ID: AZ84-02.04.2013.A429).
Scientific RepoRts | 6:36764 | DOI: 10.1038/srep36764 Culture and transfection of mES cells. E14Tg2a mES cells (kind gift of Christof Niehrs, IMB Mainz, Germany) were maintained on gelatinized cell culture dishes with standard ES cell medium at 37 °C and 7.5% CO 2 . For transfection, 3 × 10 5 cells per well were seeded onto a gelatinized 12-well-plate in media without antibiotics. After 3 h cells were transfected with a 3:1 ratio of pX330: Lipofectamine2000, according to the manufacturers' protocol (Thermo Fisher Scientific, Waltham, USA). 8 h later media was changed to ES media to remove DNA-Lipofectamine complexes.
Surveyor assay. Two days after transfection, genomic DNA was extracted from cells using the phenol-chloroformisoamyl-alcohol (PCI) precipitation method as described previously 63 . Gene edited loci were amplified by PCR (see Supplementary Table 1 for primer sequences). Products were denaturated by heating to 95 °C for 10 min, re-annealed by stepwise cooling to room temperature (RT), and analyzed using the 'SURVEYOR Mutation Detection Kit' (Transgenomic, Manchester, GB).
In vitro transcribed single-guide RNAs (sgRNA) (50 ng/μ l each) and Cas9 mRNA (100 ng/μ l) (Sigma-Aldrich) were injected into the cytoplasm of zygotes, as described previously 31 . Surviving zygotes were cultured in KSOM medium for 3 days. Developing blastocysts were transferred into the uteri of pseudo-pregnant foster mice.
Quantitative reverse transcription-polymerase chain reaction (qRT-PCR). RNA was isolated from testis tissue with TRIzol (Life Technologies, Carlsbad, USA). RNA concentrations and purity ratios were determined by NanoDrop measurement (Peqlab, Erlangen, Germany). qRT-PCR was performed as described previously 64   Isolation of epididymal sperm. Sperm were isolated by multiple incisions of the cauda followed by a swim-out in modified TYH medium 65 or M2 medium. Sperm count was determined using a Neubauer hemocytometer. For all experiments, sexually mature males with an age of 2-5 months were used. Animals were derived from intercrosses of the F1 generation.
Isolation of sperm nuclear proteins. Nine million sperm were washed with phosphate-buffered saline (PBS, pH 7.4) containing protease inhibitors (Complete mini, Roche, Basel). Extraction of basic nuclear proteins was performed according to Deyebra and Oliva 66 . Immunoblotting. Protein extracts were subjected to 15% polyacrylamide acid-urea gel electrophoresis 67,68 and blotted onto a polyvinylidene fluoride (PVDF) membrane using the Trans-Blot Turbo system (BioRad, Munich, Germany). Equal protein loading was validated by staining with Coomassie Brillant Blue G-250. Membranes were blocked in 3% nonfat dry milk powder in Tris-buffered saline with Tween20 (TBST) for 2 h at RT and probed with primary (overnight at 4 °C) and secondary (1 h at RT) antibodies in blocking solution. Chemiluminescent signals were detected using ChemiDoc MP Imaging System (BioRad) after incubation of the membrane with PierceSuper Signal West Pico chemiluminescent substrate (Perbio, Bonn, Germany) or Westar Supernova substrate (7Bioscience, Hardtheim, Germany).
Immunohistochemistry. Bouin's fixed testicular tissue was processed in paraffin wax. Immunohistochemistry (IHC) against PRM1 was performed as described previously 64 using the Autostainer 480 S (Medac, Hamburg, Germany). For IHC of PRM2, sections were treated for 10 min at RT with decondensing-mix containing 25 mM dithiothreitol, 0.2% Triton X-100 and 200 IU heparin/ml in PBS to enhance antigen accessibility for the primary antibody 69 . Subsequently, sections were washed in 0.02 M PBS (pH 7.4), boiled for 20 min in sodium citrate buffer, treated for 20 min with 3% H 2 O 2 in methanol and blocked for 20 min with 5% bovine serum albumin (BSA) in PBS. Incubation with the primary antibody was performed overnight at 4 °C, with biotinylated goat-anti-mouse secondary antibody (DAKO) and the Vectastain ABC Kit (Vector Laboratories, Burlingame, CA, USA) for 1 h at RT each. Immunoreaction was visualized using AEC (DAKO). Finally, sections were counterstained with hematoxylin and covered with glycerin gelatin. Electron microscopy. Testis  2% glutaraldehyde just before usage) via the heart were immersion-fixed in Yellow-Fix for 24 h at 4 °C. Smaller samples rinsed in PBS buffer (0.1 M) were then immersion-fixed in 1% osmium tetroxide at 4 °C for 2 h and rinsed in buffer again. Subsequently, the tissue was dehydrated and embedded in Epon. Ultrathin-sections were picked up on grids, stained with lead-citrate (0.2%) and examined with a Zeiss EM 109 (Zeiss, Oberkochen, Germany).

Assessment of sperm DNA damage.
For isolation of sperm genomic DNA, ten million sperm were resuspended in 500 μ l lysis buffer (10 mM Tris HCl, 25 mM EDTA, 1% SDS, 75 mM NaCl) and supplemented with 21 μ l 1 M DTT, 2,5 μ l Triton x-100 and 40 μ l Proteinase K (10 mg/ml). The suspension was incubated at 55 °C for 2 h following PCI precipitation of DNA. SCSA-like analysis was performed referring to Evenson et al. 70 .
Measurement of changes in sperm intracellular Ca 2+ . Sperm were isolated by incision of the cauda followed by a swim-out in modified TYH medium 65 . Sperm were loaded with the fluorescent Ca 2+ indicator Cal-520, AM (AAT Bioquest, Sunnyvale, USA) (5 μ M) in the presence of Pluronic F-127 (0.05% vol/vol) at 37 °C for 45 min. After incubation, excess dye was removed by three centrifugation steps (700 × g, 7 min, RT). The pellet was resuspended in TYH and equilibrated for 5 min at 37 °C. Pictures of loaded sperm were taken on a confocal microscope (FV1000; Olympus). Changes in [Ca 2+ ] i were recorded in a rapid-mixing device in the stopped-flow mode (SFM400; Bio-Logic, Claix, France) as previously described 65 . Data acquisition was performed with a data acquisition pad (PCI-6221; National Instruments, Austin, USA) and Bio-Kine software v. 4.49 (Bio-Logic). Ca 2+ signals are depicted as the percent change in fluorescence (Δ F) with respect to the mean of the first three data points recorded immediately after mixing (F 0 ), that is, when a stable fluorescence signal was observed. The control (buffer) Δ F/F 0 signal was subtracted from compound-induced signals.

Motility analyses.
Sperm motility was studied in shallow observation chambers (depth 150 μ m) using an inverted microscope (IX71; Olympus) equipped as described previously 65 .
To analyze the flagellar beat, cells were tethered to the glass surface by adjusting the BSA concentration in the buffer from 3 to 0.3 mg/ml. Cells that had their head attached to the glass surface and had a free beating flagellum were selected for imaging. Images were collected at 200 frames per second using a CMOS camera (Dimax; PCO, Kelheim, Germany). Quantification of the flagellar beat was performed using custom-made programs written in Java, which can be made available upon request. The program identified a point on the flagellum 25 μ m apart from the center of the mouse head within every time frame. The flagellar beat parameters were determined within a time window of 1 s after each frame. The first point in this time window was chosen as reference point and the distance between the successive points found on the flagellum and the reference point was monitored. This distance varied in a sinusoid-like manner in time and the beat frequency was determined as the maxima in the Fourier spectrum.
For analyzing the swimming parameters of sperm cells, the BSA concentration in the buffer was 3 mg/ml, resulting in most of the cells swim freely at the glass surface. Images were collected at 50 frames per second (Dimax). Quantification of the swimming parameters was performed the Casa program 71 adapted to MATLAB (MathWorks. Natick, USA).