Prm2 deficiency triggers a Reactive Oxygen Species (ROS)-mediated destruction cascade during epididymal sperm maturation in mice

Protamines are the safeguards of the paternal sperm genome. They replace most of the histones during spermiogenesis, resulting in DNA hypercondensation, thereby protecting its genome from environmental noxa. Impaired protamination has been linked to male infertility in mice and humans in many studies. Apart from impaired DNA integrity, protamine-deficient human and murine sperm show multiple secondary effects, including decreased motility and aberrant head morphology. In this study, we use a Prm2-deficient mouse model in combination with label-free quantitative proteomics to decipher the underlying molecular processes of these effects. We show that loss of the sperm’s antioxidant capacity, indicated by downregulation of key proteins like SOD1 and PRDX5, ultimately initiates an oxidative stress-mediated destruction cascade during epididymal sperm maturation. This is confirmed by an increased level of 8-OHdG in epididymal sperm, a biomarker for oxidative stress-mediated DNA damage. Prm2-deficient testicular sperm are not affected and initiate the proper development of blastocyst stage preimplantation embryos in vitro upon intracytoplasmic sperm injection (ICSI) into oocytes. Our results provide new insight into the role of Prm2 and its downstream molecular effects on sperm function and present an important contribution to the investigation of new treatment regimens for infertile men with impaired protamination. Significance statement Sexual reproduction requires the successful fertilization of female eggs by male sperm. The generation of functional sperm is a complex, multi-step differentiation process known as spermatogenesis that takes places in the male testis. One important step for physiological sperm function is the incorporation of small proteins, known as protamines into the DNA. Defects within this process are common causes of male infertility. However, the underlying molecular mechanisms still remain largely unknown, thus preventing targeted therapies. Here, we identify the molecular cascade being initiated in protamine-deficient murine sperm that ultimately impedes fertilization. Our findings have broad implications for the development of new treatment options for infertile men with faulty protamination that seek medical advice.


Introduction
Male fertility relies on the tightly regulated differentiation of spermatogonial stem cells into flagellated sperm within the seminiferous epithelium of the testis. During spermatogenesis, a series of mitotic and meiotic divisions results in the formation of haploid, round spermatids.
Extensive cytological remodeling during spermiogenesis generates the characteristic, species-specific sperm morphology. Canonical histones are replaced by testis-specific histone variants and transition proteins, which finally guide the deposition of protamines into the DNA (1). This induces a conformational change from a nucleosomal into a toroidal chromatin structure, with few remaining histone solenoid structures, thereby conferring DNA hypercondensation (2). Chromatin remodeling is pivotal for sperm function as it is supposed to protect the paternal genome from environmental noxa, to induce a transcriptionally quiescent state and to contribute to the hydrodynamic shape of the sperm head (3).
Since cell culture models for human spermatogenesis are scarce and inefficient, mouse models have contributed to a broader understanding of protamine function. Several knockout studies for both, Prm1 as well as Prm2, have shown that both isoforms are indispensable for male fertility in mice as well (24), (25), (26), (27). Similar to human sperm, impaired protamination was linked to faulty DNA compaction and increased DNA fragmentation.
Further, sperm displayed secondary defects like a decline in motility and morphological abnormalities.
In this study, we used a Prm2-deficient mouse model established by us to study the molecular origin of these defects using sperm proteomics. We show that loss of Prm2 induces a downregulation of ROS scavenger proteins like SOD1 and PRDX5. This, in turn results in oxidative stress induced DNA damage during epididymal sperm maturation. To our surprise, testicular sperm are hardly affected by oxidative damage. As a consequence, infertility of Prm2-deficient males can be overcome by ICSI of testicular sperm.

To the molecular origin of secondary defects in Prm2-deficient sperm
Previous attempts to establish Prm2-deficient mouse lines using classical gene-targeting failed as male chimeras generated by blastocyst injection of Prm2 +/-ES-cells were sterile (24). This prohibited detailed functional studies on protamine function in mice. Recently, we reported the successful generation and establishment of Prm2-deficient mouse lines using CRISPR/Cas9-mediated gene-editing in oocytes (27). Interestingly, Prm2 +/males displayed normal fecundity with sperm being morphologically and functionally indistinguishable from wildtype sperm. This enabled the breeding of Prm2 -/males, which are infertile.
Comprehensive phenotypical characterization revealed, that the loss of Prm2 did not affect the efficiency of spermatogenesis but resulted in defective DNA integrity in more than 90% of sperm. Furthermore, 100% of sperm were immotile, displayed morphological malformations, membrane defects and a decline in viability (27) (Suppl. Fig. 1). We categorized the latter as secondary defects, as they cannot be explained by the molecular function of protamines.
Since incorporation of protamines into the sperm DNA is supposed to induce transcriptional silencing, we expected to find the underlying reason for induced secondary sperm defects rather on proteomic than on transcriptomic level. Therefore, epididymal sperm protein lysates from wildtype, Prm2 +/and Prm2 -/males were pre-fractionated by SDS gel electrophoresis and subjected to LC-MS (liquid chromatography -mass spectrometry). In total, 3318 different proteins were identified with high confidence. Following removal of non-unique peptides and single-shot proteins, 2299 proteins were considered for label-free quantification (Suppl. Fig. 2 A-B). In order to evaluate the power of the generated dataset, we compared the list of identified proteins with the compiled murine sperm proteome of four previous studies (28), (29), (30), (31) showing a coverage of 63% ( Fig. 1 A). Our analysis detected 1498 additional proteins, which had not been identified previously. Further, we also compared our dataset with the human sperm proteome, which has been studied in much more detail (32). Of note, 77% of the murine sperm proteome were overlapping with the human sperm proteome (Fig. 1 A). Thus, we concluded that the present dataset reflects a significant portion of the murine sperm proteome. Following data normalization and quality controls (see Suppl. replicates clustered strongly apart from controls, being indicative for an overall biological difference ( Fig. 1 B). This was verified by heatmap visualization of the most significantly deregulated proteins. Knockout replicates displayed a strongly altered proteomic profile and were only distantly related to control replicates as visualized by hierarchical clustering using the Euclidian distance ( Fig. 1 C). In accordance with the phenotypical characterization, a highly similar proteomic profile was observed for wildtype and Prm2 +/replicates. No proteins were found to be significantly up-or downregulated ( Fig. 2 A). However, in the knockout condition, 24 proteins were found to be significantly deregulated compared to wildtype (17 down-and 8 upregulated) (Fig. 2 A). A highly similar set of deregulated proteins was observed by pairwise comparison of knockout and heterozygous samples ( Fig. 2 A).
Many of the differentially expressed proteins, including SORD, SOD1, PRDX5 and ACR have a well-known function for male fertility and sperm motility (33) (34) (35) (36). Thus, deregulated proteins correlate with the observed defects. To investigate whether deregulations can be attributed to specific biological processes, reactome pathway analysis was performed, showing a significant enrichment of downregulated proteins in processes of energy metabolism and the detoxification of ROS (Fig. 2 B). Further, STRING analysis suggests a strong interaction among the downregulated proteins ( Fig. 2 C). Taken together, MS data imply that loss of PRM2 triggers a molecular cascade, that initiates described secondary sperm defects, which ultimately result in male infertility.

Prm2 deficiency induces a downregulation of ROS scavenger proteins
Since excessive levels of reactive oxygen species (ROS) are well-known mediators of cellular damage, including DNA damage and lipid peroxidation, the potential role of imbalanced ROS level was investigated in detail. Among the differentially expressed proteins, superoxide dismutase 1 (SOD1) and peroxiredoxin 5 (PRDX5) were identified as key players of ROS detoxification. SOD1 catalyzes the conversion of superoxide radicals into hydrogen peroxide, which is subsequently degraded into non-toxic H 2 O molecules. PRDX5 is another cytoprotective antioxidant enzyme that catalyzes the reduction of hydrogen peroxide, alkyl hydroperoxide and peroxynitrite (37). Thus, both enzymes have ROS scavenging function that might result in oxidative stress upon loss. As shown in the profile plots for SOD1 and PRDX5, the relative abundance of SOD1 and PRDX5 peptides identified by MS was strongly diminished in Prm2 -/sperm compared to controls (Fig. 3 A, Suppl. Fig. 3). To validate MS results independently, protein level of SOD1 and PRDX5 were determined by Western Blot analysis. Both, levels of SOD1 and PRDX5 were strongly decreased in Prm2 -/sperm compared to wildtype and Prm2 +/samples ( Fig. 3 B). However, in testicular lysates protein levels were not changed (Fig. 3 B). This suggests that the proposed ROS signaling cascade is first initiated during epididymal transit.

Oxidative stress causes DNA damage during epididymal sperm maturation
To prove the hypothesis that excessive ROS levels are the cause of secondary sperm defects, testicular and epididymal tissue sections were stained against 8-OHdG (8-hydroxy- an enzyme-linked immunosorbent assay (ELISA) was performed. The concentration of 8-OHdG was highly significantly increased from 1.449 ng/ml in DNA from wildype sperm to 3.256 ng/ml in Prm2 -/sperm ( Fig. 4 B). Only a slight, but non-significant increase in 8-OHdG concentrations was observed between wildtype and Prm2 +/sperm. Taken together, the results clearly show that Prm2 deficiency leads to an imbalance in the ROS scavenging system, supposedly leading to elevated ROS levels triggering a reactive oxygen-mediated destruction cascade in sperm, which is exerted during epididymal sperm maturation.

Secondary sperm defects arise during epididymal maturation
To verify that other secondary sperm defects can be also assigned to the proposed ROS pathway, these defects are expected to appear and intensify during epididymal sperm maturation as well. The initial phenotypical characterization showed severe acrosomal malformations in Prm2-deficient cauda epididymal sperm (27). Here, periodic-acid-Schiff (PAS) staining, which stains glycoprotein-rich acrosomal structures pink, was performed on testicular tissue sections. Acrosomal vesicles were first detected in round spermatids ( Fig. 5 A, arrowheads). In the course of spermiogenesis, vesicles elongated and finally capped the sickle-shaped head of step 16 spermatids of all genotypes (Fig. 5 A, arrows). This indicates that acrosome biogenesis is not impaired upon loss of Prm2 and that defects are inevitably acquired during epididymal transit.
Further, we analyzed the effects of Prm2 deficiency on sperm nuclear morphology. The characteristic sickle-shaped head morphology was observed in hematoxylin stained stage VII/VIII seminiferous tubules of all genotypes (Fig. 5 B). Quantification of sperm head length and width revealed no significant differences between Prm2-deficient step 16 testicular spermatids and controls (Fig. 5 B), thus indicating that Prm2 is dispensable for proper shaping of the sperm head during spermiogenesis. However, severe changes in sperm head size and nuclear morphology were observed in DAPI stained cauda epididymal sperm (Fig. 5 C). In general, most Prm2-deficient sperm appeared smaller in size and displayed loss of the characteristic sickle-shape compared to controls (Fig. 5 C). However, morphological changes were variable, resulting in a highly heterogeneous sperm population (Fig. 5 C). To quantify changes in sperm head morphology, an automated, high-throughput analysis was performed.
In line, the median sperm perimeter decreased from 21.13 µm (95% CI 21.09 µm ± 0.08) to 14.21 µm (95% CI 14.76 µm ± 0.25) (Suppl. Fig. 4). Both, a decline of sperm length and width resulted in a decrease of sperm area and perimeter. Further, loss of the characteristic sperm hook caused an increase in circularity and ellipticity of Prm2-deficient sperm compared to controls (Suppl. Fig. 4).
Taken together, our analyses clearly demonstrate, that the functional and morphological sperm defects observed are almost exclusively acquired during epididymal sperm maturation and correlate with the initiation of the postulated ROS cascade.
Heatmap visualization of the top 50 differentially expressed (DE) genes and hierarchical clustering using the Euclidian distance clearly separated samples of Prm2 -/males from controls as also verified by PCA analysis (Fig. 6 A-B). Most DE genes were stronger expressed in Prm2 -/testes compared to controls. Mapping of upregulated genes to their chromosomal localization, showed a random distribution throughout the whole genome and no enrichment at distinct loci (Suppl. Fig. 5 A). This indicated that the protamine-mediated gene silencing during spermiogenesis might be incomplete upon loss of Prm2. In total, 81 genes were found to be upregulated and 13 genes to be downregulated in knockout testes compared to wildtype (Fig. 6 C). Among the upregulated genes, well-known regulators of protamine expression and translational activation were identified, e.g. Y-box binding protein 2 (Ybx2) and TAR binding-protein 2 (Tarbp2), probably resembling a compensation mechanism due to the lack of Prm2. However, gene ontology analysis on up-and downregulated gene sets did not identify enrichment of any biological processes, which might point towards secondary defects observed in the epididymis (Suppl. Fig. 5 B). These results support the notion, that the induced ROS cascade is not initiated until sperm release from the seminiferous epithelium of the testis.

Blastocysts stage embryos can be derived from ICSI of Prm2-deficient testicular sperm
Since Prm2-deficient testicular spermatids neither displayed oxidative DNA damage nor severe morphological abnormalities, we tested their fertilization potential in vitro. Testicular sperm were extracted from the seminiferous tubules and injected into unfertilized wildtype oocytes using ICSI. Of note, 10% of oocytes injected with Prm2-deficient sperm developed into blastocyst stage embryos within four days after injection (Fig. 7 A-B). Genotyping revealed all embryos to be heterozygous, indicating that Prm2-deficient sperm had fertilized the Prm2 +/+ oocyte. (Fig. 7 C). Control injections with wildtype testicular spermatids displayed similar efficiencies for the induction of blastocyst stage embryos (Fig. 7 B). Thus, Prm2deficient testicular sperm appear not only morphologically intact but also functionally capable to overcome infertility of Prm2 -/males using ART.

Discussion
In the present study we utilized sperm proteomics to investigate the molecular origin of secondary sperm defects arising in consequence of abnormal sperm protamination. We demonstrate that loss of the sperm`s antioxidant capacity induces a ROS destruction cascade during epididymal sperm maturation, ultimately causing oxidative DNA damage.
The connection of impaired sperm protamination and male subfertility is well-known for more than three decades. However, the molecular processes triggered by this impairment, finally leading to infertility, are mostly unknown, although being essential for the development of  (42). Further, Prm2-deficient sperm display a strongly reduced head size, which has also been reported for sperm from Prm2 chimeric mice (43). This is unexpected in light of the DNA condensing function of protamines. However, it is in accordance with a Gpx5-deficient mouse model for oxidative stress, showing a decreased sperm volume and surface area as well (44). Increased ROS levels result either from an increased production or a diminished antioxidant capacity. In Prm2-deficient sperm, significantly lower levels of the ROS scavenger enzymes SOD1 and PRDX5 are observed.
SOD1 catalyzes the detoxification of superoxide radicals into hydrogen peroxide and oxygen.
Studies in knock-out mice revealed that Sod1-deficient male mice remain fertile, however, sperm show impaired motility (45), (36). In consequence, in vivo such sperm cannot compete with wildtype sperm and display reduced fertilization ability in IVF experiments (46), (36). A correlation between SOD1 activity and sperm motility as well as DNA fragmentation is also reported for human sperm (47), (48). Alvarez et al. discuss superoxide dismutase (SOD) to be a key player for preventing human sperm from lipid peroxidation (42). PRDX5 belongs to the class of peroxiredoxins, which act downstream of SOD and catalyze the detoxification of hydrogen peroxide as well as hydroperoxides and peroxynitrites (49). Six peroxiredoxin isoforms have been identified, with an isoform-specific cellular localization. In humans, PRDX5 is enriched in the postacrosomal region of the sperm head and the mitochondrial sheath (50). Inhibition of peroxiredoxin activity in murine sperm caused increased ROS levels, thereby underlining its role as antioxidant (35). In mice, impaired peroxiredoxin activity results in DNA fragmentation and adversely affects sperm motility, viability, fertilization capacity and early embryonic development in mice (35). In accordance, decreased peroxiredoxin levels are associated with male infertility and oxidative stress-mediated sperm damage in humans (51).
Taken together, the identified molecular changes in the proteome of Prm2-deficient sperm nicely delineate the underlying impact of ROS on sperm morphology and function -but how is Prm2 deficiency sensed and the ROS cascade induced? Interestingly, PRM2 and SOD1 share certain structural similarities. Both proteins harbor a zinc binding-site and are characterized by the formation of intramolecular disulfide bridges. Zinc incorporation into PRM2 is proposed to stabilize the chromatin structure of testicular sperm. However, during epididymal sperm maturation and enhanced disulfide bonding of protamines, zinc is released, as shown in stallion (52). For SOD1, binding of zinc is essential for enzyme stability and function (53), (54). It is tempting to speculate that zinc ions released from PRM2 might confer stability of SOD1 in the epididymis, thereby ensuring its proper antioxidant function.  (43). This confirms that the decline in chromatin integrity during epididymal sperm maturation directly impacts on early embryonic development. Similarly, ICSI with testicular sperm from Tnp1 -/-Tnp2 +/males yielded higher implantation rates and more offspring than obtained by ICSI of Tnp1 -/-Tnp2 +/cauda epididymal sperm (58). In men, impaired sperm protamination is associated with recurrent miscarriages (59) and decreased fertilization rates in ART programs (21), (22), (23). Thus, in light of the molecular mechanism provided in this study, ICSI with testicular sperm presents as only treatment option for subfertile men with impaired protamination and severe DNA fragmentation. In fact, a recent study proves, that testicular sperm from patients with impaired protamination are superior to epididymal sperm in terms of fertilization rate and pregnancy outcome following ICSI (60).  (27).

Immunoblotting
Protein lysates were separated on 12% SDS polyacrylamide gels and transferred to PVDF membranes using the Trans-Blot Turbo System (BioRad, Feldkirchen, Germany). Equal protein loading and successful blotting was verified by Coomassie Brilliant Blue staining.
Membranes were blocked in 5% non-fat dry milk powder in phosphate buffered saline with 1% Tween20 (PBS-T) for 1.5 h at room temperature. Primary and secondary antibodies were diluted in blocking solution as specified in Table 1 and incubated at 4°C overnight and 1 h at room temperature, respectively. Following incubation with Pierce Super Signal West Pico chemiluminescent substrate (Perbio, Bonn, Germany), chemiluminescent signals were detected with ChemiDocMP Imaging system (BioRad).

Mass Spectrometry
Following separation of protein extracts by SDS-PAGE, protein gels were shortly washed   (68), (69).  Epididymal sperm were analyzed using the ImageJ plugin "Nuclear morphology analysis v1.14.1" according to the developers instructions (73). Briefly, isolated sperm were pelleted by centrifugation (500g, 5 min, RT) and fixed by dropwise addition of methanol:acetic acid  (76). Pseudogenes were removed from the count matrices based on "biotype" annotation information extracted from Biomart (R-package biomaRt, (77)). Low counts were removed by the independent filtering process, implemented in DESeq2 (78). The adjusted p-value (Benjamini-Hochberg method) cutoff for DE was set at < 0.05, log 2 fold change of expression (LFC) cutoff was set at > 0.5.

Immunohistochemistry
We performed GO term and pathway overrepresentation analyses on relevant lists of genes from DE and co-expression analyses using the PANTHER gene list analysis tool with Fisher's exact test and FDR correction (79). We tested for overrepresentation based on the GO annotation database (Biological Processes) (released 2019-02-02, (80)) and the Reactome pathway database (version 58 (81)).

Intracytoplasmatic sperm injection (ICSI)
B6D2F1 females were superovulated by intraperitoneal injection of 5 i.U. Pregnant Mare Serum (PMS) and human Chorionic Gonadotropin (hCG). Oocytes were isolated from the oviducts 15 h after the last hormone injection and freed from cumulus cells by treatment with hyaluronidase. Testicular sperm were isolated from 8-13 week-old males as described (82).
Briefly, after removal of the tunica albuginea, testes were placed into 1% (m/v) PVP solution (P5288, Sigma) and cut into minute pieces. One part of the testicular suspension was throughout mixed with two parts 12% PVP solution and incubated at 16°C until injection.
ICSI was performed at 17°C using an inverted microscope (Leica) equipped with micromanipulators (Narishige, Japan) and a piezo element (Eppendorf, Hamburg, Germany). Here, testicular sperm were injected head to tail without prior removal of the sperm flagellum.

Acknowledgment
This study was supported by a grant from the German Research foundation (DFG) to HS (SCH 503/15-2) and KS (STE 892/14-2). We kindly thank the University of Bonn Core facilities for Next Generation Sequencing (NGS), Mass Spectrometry and Bioinformatics Data Analysis (CUBA) for their support. Further, we thank Gaby Beine, Angela Egert, Barbara Fröhlich, Andrea Jäger, Anna Pehlke and Susanne Steiner for excellent technical assistance.

Additional information
Competing financial interests: the authors declare no competing financial interests. LA performed bioinformatics analyses of RNAseq data. SS and HS were major contributors in writing the manuscript. All authors read and approved the final manuscript.