Linking human Dead end 1 (DND1) variants to male infertility employing zebrafish embryos

Abstract

STUDY QUESTION

Is the vertebrate protein Dead end (DND1) a causative factor for human infertility and can novel in vivo assays in zebrafish help in evaluating this?

SUMMARY ANSWER

Combining patient genetic data with functional in vivo assays in zebrafish reveals a possible role for DND1 in human male fertility.

WHAT IS KNOWN ALREADY

About 7% of the male population is affected by infertility but linking specific gene variants to the disease is challenging. The function of the DND1 protein was shown to be critical for germ cell development in several model organisms but a reliable and cost-effective method for evaluating the activity of the protein in the context of human male infertility is still missing.

STUDY DESIGN, SIZE, DURATION

Exome data from 1305 men included in the Male Reproductive Genomics cohort were examined in this study. A total of 1114 of the patients showed severely impaired spermatogenesis but were otherwise healthy. Eighty-five men with intact spermatogenesis were included in the study as controls.

PARTICIPANTS/MATERIALS, SETTING, METHODS

We screened the human exome data for rare, stop-gain, frameshift, splice site, as well as missense variants in DND1. The results were validated by Sanger sequencing. Immunohistochemical techniques and, when possible, segregation analyses were performed for patients with identified DND1 variants. The amino acid exchange in the human variant was mimicked at the corresponding site of the zebrafish protein. Using different aspects of germline development in live zebrafish embryos as biological assays, we examined the activity level of these DND1 protein variants.

MAIN RESULTS AND THE ROLE OF CHANCE

In human exome sequencing data, we identified four heterozygous variants in DND1 (three missense and one frameshift variant) in five unrelated patients. The function of all of the variants was examined in the zebrafish and one of those was studied in more depth in this model. We demonstrate the use of zebrafish assays as a rapid and effective biological readout for evaluating the possible impact of multiple gene variants on male fertility. This in vivo approach allowed us to assess the direct impact of the variants on germ cell function in the context of the native germline. Focusing on the DND1 gene, we find that zebrafish germ cells, expressing orthologs of DND1 variants identified in infertile men, failed to arrive correctly at the position where the gonad develops and exhibited defects in cell fate maintenance. Importantly, our analysis facilitated the evaluation of single nucleotide variants, whose impact on protein function is difficult to predict, and allowed us to distinguish variants that do not affect the protein’s activity from those that strongly reduce it and could thus potentially be the primary cause for the pathological condition. These aberrations in germline development resemble the testicular phenotype of azoospermic patients.

LIMITATIONS, REASONS FOR CAUTION

The pipeline we present requires access to zebrafish embryos and to basic imaging equipment. The notion that the activity of the protein in the zebrafish-based assays is relevant for the human homolog is well supported by previous knowledge. Nevertheless, the human protein may differ in some respects from its homologue in zebrafish. Thus, the assay should be considered only one of the parameters used in defining DND1 variants as causative or non-causative for infertility.

WIDER IMPLICATIONS OF THE FINDINGS

Using DND1 as an example, we have shown that the approach described in this study, relying on bridging between clinical findings and fundamental cell biology, can help to establish links between novel human disease candidate genes and fertility. In particular, the power of the approach we developed is manifested by the fact that it allows the identification of DND1 variants that arose de novo. The strategy presented here can be applied to different genes in other disease contexts.

STUDY FUNDING/COMPETING INTEREST(S)

This study was funded by the German Research Foundation, Clinical Research Unit, CRU326 ‘Male Germ Cells’. There are no competing interests.

TRIAL REGISTRATION NUMBER

N/A.

Dead end, male infertility, zebrafish, azoospermia, whole-exome sequencing, teratoma, non-obstructive azoospermia

Introduction

Male infertility is a common disease, affecting about 7% of men (Krausz and Riera-Escamilla, 2018). A range of causes can lead to male infertility, with a major proportion of those traced to defects in germline development that result in lack of sperm, low sperm production, or production of defective sperm. Non-obstructive azoospermia (NOA) is the most severe form of male infertility, with no sperm in the ejaculate (Tüttelmann et al., 2018; Koc et al., 2019). At the testicular level, it manifests as a range of phenotypes within the seminiferous tubuli: the complete absence of germ cells with only the somatic Sertoli cells present (Sertoli cell only, SCO), arrest at specific maturation stages, or hypospermatogenesis, where SCO loci are adjacent to arrested tubuli, and/or full spermatogenesis. Despite their prevalence, the pathomechanistic causes of spermatogenic failure often remain unknown but are thought to be of genetic origin in the majority of severe cases (Tüttelmann et al., 2018). Recent progress in applying modern genome analysis tools has allowed the identification of genetic loci linked to the infertility phenotype, but determining the relevance of specific gene variants for the disease remains challenging (Houston et al., 2021). In some cases, patient-derived genetic data indicate specific protein-altering variants that are potentially causative for male infertility. Determining the effects such variants have on protein function can support or rule out their relevance for the disease. One possible path for analyzing genetic data obtained from infertile men is to employ animal models to investigate gene function. Such an approach allows evaluation of the activity of specific gene variants in a relevant in vivo context, where molecular targets and interaction partners are present at physiological levels. Among the available vertebrate models, the zebrafish is especially suitable for this task as the development of the germline is well characterized and is amenable to imaging and molecular manipulation.

Using such an approach, an interesting candidate gene is Dead end (DND) microRNA-mediated repression inhibitor 1 (DND1), which is primarily expressed in the testis (https://www.proteinatlas.org/ENSG00000256453-DND1/tissue). The DND1 protein was originally identified in zebrafish, where it proved to be essential for germ cell migration, viability and germ cell fate maintenance (Weidinger et al., 2003; Gross-Thebing et al., 2017). DND1 was then found to be expressed in the germline of a range of vertebrates, where it was shown to be linked to infertility and the formation of germ cell tumors (Slanchev et al., 2005; Youngren et al., 2005; Horvay et al., 2006; Suzuki et al., 2016; Niimi et al., 2019; Ruthig et al., 2019; Gross-Thebing and Raz, 2020; Imai et al., 2020; Zhang et al., 2021). Consistently, a green fluorescent protein (GFP)-tagged version of DND1 was shown to be expressed in pre-meiotic mouse germ cells (Ruthig et al., 2021). Functional analysis of mutations in the zebrafish DND1 protein highlighted key domains critical for the protein’s function, namely the conserved RNA recognition motif (RRM) and the C-terminus (Slanchev et al., 2009). Indeed, a recent study suggested that a missense variant of DND1 causes NOA, thereby leading to human male infertility (Xie et al., 2022).

In this study, we employed the germline of zebrafish as a sensitive in vivo model to assist in categorizing gene variants identified in infertile men as likely causative, or likely non-causative, for the disease. Focusing on the DND1 gene, we investigated the functional relevance and, thus, the pathogenicity of rare, heterozygous variants in human DND1 identified via exome sequencing of azoospermic infertile men. To this end, we examined the activity of zebrafish DND1 protein variants corresponding to variants found in human DND1 of infertile males. By employing a rapid assay we developed, it was possible to determine whether these mutations affect protein function, namely, maintaining germ cell fate and the ability of the cells to arrive at the gonad. By testing the function of the mutated proteins in the context of early development of the zebrafish germline, we could distinguish between variants that are likely to be causative for the azoospermia in infertile men and variants that do not alter DND1 protein function.

Materials and methods

Ethics

All participants gave written informed consent before inclusion in the study. Study protocols were approved by the relevant ethics committees according to the Declaration of Helsinki (Münster: Kennzeichen 2010-578-f-S, Gießen: No. 26/11). All animal procedures were performed in accordance with the regulations of the state of North Rhine-Westphalia, supervised by the veterinarian office of the city of Muenster.

Study population

The Male Reproductive Genomics (MERGE) cohort examined in this study included 1305 men, most attending the Centre of Reproductive Medicine and Andrology (CeRA), University Hospital Münster, or the Clinic for Urology, Pediatric Urology and Andrology, University Hospital Gießen. Well-known causes of male infertility had been excluded in advance, namely testicular tumors, previous radio- or chemotherapy, vasectomy, hypogonadotropic hypogonadism, as well as chromosomal aberrations such as Klinefelter syndrome and Y-chromosomal azoospermia factor (AZF) deletion. A total of 1114 of the patients showed severely impaired spermatogenesis (901 with azoospermia, 155 with cryptozoospermia (<0.1 mio. sperm/ml), 58 with severe oligozoospermia (<5 mio. sperm/ml)) but were otherwise healthy. Eighty-five men with intact spermatogenesis were included in the study as controls, the other 106 participants had unrelated infertility phenotypes. As a second control cohort, we used 6640 proven fathers who had undergone exome sequencing at the Radboudumc genome diagnostics center in Nijmegen, the Netherlands (Oud et al., 2022).

Genetic analysis

Exome sequencing was performed as described previously (Wyrwoll et al., 2021). Exome data of unsolved cases was filtered for rare (minor allele frequency (MAF) in any subpopulation (PopMax) in gnomAD database v.2.1.1 < 0.001), stop-gain, frameshift, splice site, as well as missense variants in DND1 (NM_194249.3), as described previously (Wyrwoll et al., 2022). Variants were validated by Sanger sequencing according to standard procedures. Functional domains of DND1 were obtained from SMART (http://smart.embl-heidelberg.de). Segregation analysis of the variants was performed if DNA samples from parents were available. Patients with variants in DND1 were also analyzed for relevant variants in NOA-associated genes, as previously described (Houston et al., 2022). Only a colon biopsy sample was available for genetic analysis from the father of patient M2196, which failed to yield DNA that could be amplified by PCR. We therefore adopted ancient DNA methodology, which can greatly improve sequence recovery from formalin-fixed samples (Stiller et al., 2016). We first used 10 µl DNA extract to produce a single-stranded DNA library (Gansauge et al., 2020) and then performed in-solution hybridization capture (Zavala et al., 2022) with a pair of biotinylated oligonucleotides as capture probes (Supplementary Table SI) to enrich the library for DNA fragments overlapping the location within DND1 where the patient DNA showed a single-nucleotide duplication. Paired-end sequences (2× 75 cycles) were produced on a MiSeq (Illumina Technologies, San Diego, CA, USA) and merged into full-length molecule sequences using leeHom (Renaud et al., 2014). Mapping sequences of at least 30 bp to the human reference genome (hg19/GRCh37) using burrow wheeler aligner (Li and Durbin, 2009) with ancient parameters (Meyer et al., 2012) yielded 188 unique DNA fragments with a length of least 35 bp that mapped to the position of interest, 46 of which had a mapping quality of 25 or greater. None of the 188 fragments shared the duplication observed in patient M2196 (Supplementary Fig. S1). In addition, we enriched the library using exome capture probes (Castellano et al., 2014), sequenced it on a NextSeq (Illumina Technologies, San Diego, CA, USA) and processed the sequences as described above. Maternity was confirmed with short tandem repeat analysis of 12 markers. Paternity was confirmed based on the comparison of exome sequencing data of M2196 with his father.

Testis biopsies and testicular histology

Testis biopsies from patients were performed according to Nieschlag et al. (2010) to detect azoospermia and with the aim of testicular sperm extraction (TESE). Testicular biopsies were fixed in Bouin’s solution overnight, followed by washing with 70% ethanol, and routine embedding in paraffin. Sections (5 µm) were then stained using the Periodic acid–Schiff reaction or hematoxylin staining, according to established protocols (Brinkworth et al., 1995). Slides were documented with a PreciPoint M8 Scanning Microsocope, Olympus BX61VS Virtual Slide System Axioskop (Zeiss, Oberkochen, Germany), or an Olympus BX61 microscope with Retiga 400 R camera (Olympus, Melville, NY, USA) and CellSens imaging software (Olympus, Melville, NY, USA).

Zebrafish strains and experimental replicates

Zebrafish (Danio rerio) of the AB and AB/Tüpfel long fin (TL) genetic background were used. A transgenic line (Tg(kop: EGFP-FTASE-UTR-nanos3) (er1Tg, RRID: ZDB-ALT-070406-1, Blaser et al., 2005), in which enhanced green fluorescent protein (EGFP) is expressed on the primordial germ cell (PGC) membrane was used. In a different experiment, the transgenic line Tg(-1.8unc45b: TFP) (ka14Tg, RRID: ZDB-GENO-181011-2, Etard et al., 2015) was used to express the teal fluorescent protein (TFP) protein in the developing muscle tissue. The transgenic line Tg(kop: mCherry-FTASE-UTR-nanos3) (mu6Tg, RRID: ZDB-GENO-150810-2) (Tarbashevich et al., 2015) was used for the mesoderm conversion experiment to label the PGC membrane in red. Embryos were kept in 0.3× Danieau’s solution (17.4 mM NaCl, 0.21 mM KCl, 0.12 mM MgSO4·7H₂O, 0.18 mM Ca(NO₃)₂) and maintained at 28°C. All experiments were performed with three independent replicates.

Cloning of DNA constructs

To obtain the DND1 protein variants for analysis in zebrafish, single-nucleotide exchanges or insertions were introduced in a Morpholino antisense oligonucleotide (MO)-resistant DND-nos 3′UTR plasmid (Slanchev et al., 2009), with site-specific mutagenesis as previously described (Edelheit et al., 2009). The choice of the amino acid to be changed relied on protein sequence alignment of the two species in the relevant region. In most cases, the relevant amino acids in the two species were identical, or belonging to the same class, except for the case of DND^R124S. The amino acid exchange in the human variant was mimicked at the corresponding site of the zebrafish protein. In all cases, the altered amino acid belongs to a different class as compared with the wild-type sequence. For the generation of the frameshift variant DND^L329PfsTer16, the nucleotide C at position 985 bp (corresponding to position 775 bp in human; primer sequence cgccgtgtctcatcctggggatccaccggtctac) was duplicated, thereby changing amino acid Leucine to Proline, and inducing a frameshift and a premature stop codon at position 1030–1032 bp. The hemagglutinin (HA) version of the DND^S212P variant was generated by performing site-specific mutagenesis on the DND-HAHA-nos 3′UTR plasmid (Slanchev et al., 2009) and the GFP-tagged version was generated by performing site-specific mutagenesis on the DND-GFP-nos 3′UTR plasmid (Weidinger et al., 2003). The mScarlet-NLS-nos 3′UTR plasmid was generated by replacing the GFP sequence of a GFP-nos 3′UTR plasmid (Köprunner et al., 2001) with that of mScarlet-NLS. To label somatic nuclei, an mcherry-nls sequence was cloned upstream of a Xenopus globin 3′UTR.

Microinjection into zebrafish embryos

Capped sense mRNAs were synthesized using the mMessage mMachine kit (Ambion) according to the protocol of the manufacturer. Embryos were injected with 1 nl of mRNA and/or MO-containing solution into the yolk at the 1-cell stage. dnd^WT-nos 3′UTR mRNA (Slanchev et al., 2009) and dnd^variant-nos 3′UTR mRNAs were injected at a concentration of 200 ng/µl. For immunostaining of the DND1 protein versions, dnd^WT-HAHA-nos 3′UTR mRNA (Slanchev et al., 2009) and dnd^S212P-HAHA-nos 3′UTR mRNA were injected at a concentration of 150 ng/µl. To quantify protein expression in PGCs relative to a nuclear marker, 150 ng/µl of either dnd^WT-gfp-nos 3′UTR (Weidinger et al., 2003) or dnd^S212P-gfp-nos 3′UTR mRNA were injected along with 70 ng of mscarlet-nls-nos 3′UTR mRNA. To label the somatic nuclei for the motility analysis, 50 ng/µl of mcherry-nls-globin 3′UTR mRNA was co-injected. For converting embryonic cells into mesodermal tissue, 100 ng/µl of cyclops-globin 3′UTR (Sampath et al., 1998) was injected along with 238 µM of sox32-targeting Morpholino (CAGGGAGCATCCGGTCGAGATACAT) (Dickmeis et al., 2001). Morpholinos targeting dnd1 (GCTGGGCATCCATGTCTCCGACCAT) (Weidinger et al., 2003) and cxcl12a (TTGAGATCCATGTTTGCAGTGTGAA) (Doitsidou et al., 2002) were injected at a concentration of 20 and 200 µM, respectively.

Whole-mount immunostaining of embryos

To visualize endogenous proteins within germ cells, 24 h post-fertilization (hpf) embryos were fixed with 4% paraformaldehyde (PFA) for 1 h at room temperature. Embryos were subsequently rinsed in 1× phosphate-buffered saline solution with 0.1% Tween (PBT) and transferred to and stored in 100% methanol. For the immunostaining, embryos were incubated in Dents solution (80% methanol, 20% dimethylsulfoxide (DMSO)) for 24 h at 4°C and were then rehydrated by 5-min washes with increasing proportions of 1× PBT in methanol (25%, 50%, 75%, and 100%). Embryos were then re-fixed in 4% PFA for 20 min at room temperature, followed by three 5-min washes in 1× PBT. Embryos were then incubated in 1× PBT + 0.3% Triton X-100 (Carl Roth) for 1 h at room temperature and subsequently incubated for 5 h in blocking solution (1× PBT + 20% goat serum + 5% DMSO) at room temperature. Next, the embryos were incubated for 24 h at 4°C in blocking solution containing antibodies directed against GFP (1:600 dilution, anti-chicken, Aves labs: SKU GFP-1020, RRID: AB_2307313) and HA tag (1:300 dilution, anti-rabbit, Sigma: catalogue no. H6908). The embryos were then incubated in blocking solution containing secondary antibodies (1:600 dilution, anti-chicken, Jackson Immuno-Research and 1:1000 dilution, anti-rabbit, Thermo Fisher Scientific: catalogue no. A-11036) for 12 h at 4°C.

Assessment of the germ cell number

The number of germ cells within the region of the developing gonad was determined in 24 hpf embryos. Germ cells were counted on one lateral side of the fish and the number was then multiplied by two, providing an estimate of the total number of cells. Embryos were fixed in 4% PFA prior to counting, to prevent an influence of the progression in development among samples.

Evaluation of variant proteins in germ cells

To quantify the amount of DND1 wild-type or variant protein, 3D stacks of PGCs co-expressing either DND^WT-GFP or DND^S212P-GFP fusion proteins and the mScarlet-NLS (red) nuclear marker were acquired, focusing on the gonad region of 24 hpf embryos. First, the cells were masked based on the mScarlet-NLS signal, which is also expressed at low amounts in the cytoplasm, thus allowing for a segmentation of the cell outline. Subsequently, the mean intensities of red nuclear signal and green DND1 protein signal were extracted in all segmented areas of the 3D stack, and an average value was calculated for both channels per embryo. To normalize this value, the segmented outlines were then shifted to background areas in the vicinity of the PGCs, the measurement repeated, and the average background intensity subtracted from the average PGC intensity. Based on these normalized values, a ratio of the DND-GFP signal and the mScarlet-NLS signal was measured.

Analysis of a muscle marker in germ cells

To compare the presence of a muscle marker protein (TFP under control of the unc45b promoter) in germ cells depleted of endogenous Dnd protein and expressing either DND^WT or DND^S212P protein variants, the ratio of the mean intensity within a PGC and the neighboring muscle-precursor cells was measured, as described previously (Gross-Thebing et al., 2017). For this purpose, the following steps were conducted using the Fiji software (National Institutes of Health, version 2.0.0-rc-43/1.51a): a region of interest (ROI) was drawn around the perimeter of the PGC and then projected onto the channel displaying the muscle marker signal. The mean intensity within the ROI was then derived. Next, the same ROI was shifted to 2–3 neighboring muscle cells, and the mean signal intensity within each of these ROIs was determined. The ratio of the mean intensity measured in the PGC and in the neighboring muscle tissue was then calculated.

Transcriptome analysis of germ cells derived from mesoderm-converted embryos

To convert embryonic cells into mesodermal tissue, 1-cell stage embryos were injected with cyclops-globin 3′UTR mRNA and sox32 Morpholino (Aoki et al., 2002; Krieg et al., 2008). For expression of DND1 variants, dnd1 Morpholino and dnd1 variant mRNA were co-injected. Embryos were dechorionated at 9–10 hpf and washed 3× in egg water with methylene blue. Embryos were then dissociated in a double volume of Ca⁺/Mg²⁺-free dissociation buffer (Gibco), passed through a 40 µm nylon mesh (Falcon) into a 1× PBS solution and kept on ice until the fluorescence activated cell sorting (FACS) procedure. FACS was then performed using a FACSAria cell sorter from BD Biosciences equipped with a 70-µm nozzle. PGCs were sorted, based on red membrane marker protein expression, into 100 µl of 1× PBS, supplied with 3× volume of TRIzol and shock-frozen in liquid nitrogen until RNA extraction. The number of sorted cells obtained from two repeats was 6603 for Dnd MO + dnd^WT mRNA, 5086 for Dnd MO + dnd^S212P mRNA, and 4394 for Dnd MO + control mRNA. RNA extraction and sequencing were performed by Macrogen Europe. Briefly, a NovaSeq SMARTer Ultra low input mRNA (poly A selection, NexteraXT) library was prepared, and next-generation sequencing was performed on an Illumina NovaSeq 6000 platform (NovaSeq 6000 S4-100 PE). On average per sample, 3.96-GB data output were obtained, with 90.2% categorized as >Q30. The total read number obtained on average per sample was 52 137 580. Analyzing and presenting the data provided by Macrogen Europe was performed using the R software version 4.2.1 (R Core Team, 2021). Significantly differentially expressed genes (DEGs, P-value <0.05) were identified using DESeq2 (Love et al., 2014) and plotted using the R package EnhancedVolcano (Blighe et al., 2022). A Gene Ontology over-representation analysis was conducted to determine the gene enrichment on the biological pathways using ClusterProfiler (Wu et al., 2021).

Evaluation of germ cell motility

The dynamic behavior of PGCs in embryos depleted of endogenous DND1 protein and expressing either DND^WT or DND^S212P protein versions was defined as ‘motile’ or ‘stationary’ based on the movement of the PGC relative to neighboring somatic nuclei. Subsequently, the percentage of motile cells was calculated. See next section for details of image acquisition of the time-lapse videos.

Microscopy, image, and statistical analysis

Imaging and counting of germ cells in the gonad region were conducted using a Yokogawa CSU-X1 Spinning Disk microscope equipped with a 10× and 63× water-immersion objective (Zeiss) and connected to a Hamamatsu Orca Flash 4.0 V3 camera. The VisiView software (Visitron, version 4.0.0.14) was used to control the microscopy setup. To quantify protein amounts in PGCs in the gonad region, 3D stacks of 4 µM Z-distance were acquired and deconvolved raw images were used for the analysis. Images were then subjected to a background subtraction using the Fiji software. Time-lapses were acquired on a Yokogawa CSU-X1 Spinning Disk microscope as well, using a 10× water-immersion objective (Zeiss). For the time-lapse movies, images were acquired at 2-min intervals over a period of 2 h using an exposure of 300 ms for both 488- and 561-nm lasers. Five focal planes were acquired with a Z-distance of 40 µm between them. For the analysis of cell motility and processing of images, a maximum intensity projection was created. Images were processed using the Fiji software. Reduction in signal noise in movies was achieved by applying background subtraction and a Gaussian filter. For the muscle marker expression analysis in PGCs and for monitoring protein expression and subcellular localization of the DND^WT and DND^S212P proteins, a confocal microscope was used (LSM 710, Zeiss) equipped with 488- and 561-nm lasers, and 40× and 63× water-immersion objectives (Zeiss). The microscopy setup was controlled by the ZEN software (Zeiss, version 2010B SP1, 6.0). Images for these experiments were deconvolved using the Huygens Professional version 19.04 software (Scientific Volume Imaging, The Netherlands, http://svi.nl), employing the CMLE algorithm, with a maximum of 40 iterations. Statistical analysis was performed using the Prism software (version 9.4.0, Graphpad). For each experiment, three biological replicates were conducted. Significance was determined using the Mann–Whitney U-test for non-parametric data sets. P values less than 0.05 were considered to be statistically significant.

Results

The spectrum of rare DND1 variants in infertile men

In the exome sequencing data from the MERGE study, we identified four heterozygous variants in DND1 (three missense and one frameshift variant, Fig. 1A) in five unrelated patients (Table I). No men with homozygous or two heterozygous variants were identified. Four patients were azoospermic, and one displayed severe oligozoospermia (<5 million. sperm/ml). The testicular histological phenotype in the two men who underwent a biopsy was SCO and hypospermatogenesis, respectively (Supplementary Fig. S2). All patients presented either small or borderline-sized testes, elevated FSH, and/or low testosterone levels (Table I), all of which are signs of broad testicular dysfunction.

Figure 1.

Heterozygous DND1 variants identified in infertile men. (A) Functional domains of Dead end (DND) microRNA-mediated repression inhibitor 1 (DND1: NP_919225) protein were obtained from SMART (http://smart.embl-heidelberg.de/). dsRBD: double-strand RNA-binding domain; RRM: RNA recognition motif. The pinheads show variants found in infertile patients of the Male Reproductive Genomics (MERGE) cohort. Gray pinheads indicate rare variants without functional effect, red pinheads indicate rare variants with pronounced functional effect in zebrafish experiments. (B) Hematoxylin and eosin staining of testis biopsies. Upper row: histological images of testis from a man with intact spermatogenesis as control. Lower row: images from patient M2196 with Sertoli cell-only phenotype. ES: elongating spermatid; SC: sertoli cell; SPC: spermatocyte; SPG: spermatogonium; RS: round spermatid. (C) Pedigree of patient M2196. The patient carries the heterozygous frameshift variant c.775dup in DND1, while his parents and sister are unaffected. *Sanger sequencing of the father’s DNA was not possible and therefore paired-end sequencing of formalin-fixed colon biopsy DNA was performed (Supplementary Fig. S1).

Table I

Clinical and genetic data from infertile patients from the Male Reproductive Genomics (MERGE) cohort carrying variants in DND1.

Subject ID	Age (years); country of origin	Semen analysis/histology	Testicular volume (left/right) [ml]	FSH [IU/l]	LH [IU/l]	T [nmol/l]	Variant (cDNA)	Variant (protein)	Variant (zebrafish)	Inheritance	CADD; PP2/MT/SIFT	MAF
M1284	36; Kazakhstan	Azoo/NA	9/8	22.8	7.1	13	c.364C>T	p.Pro122Ser (P122S)	R124S	Maternal	3.9; B/Tol/N	0.0002
M2030	39; Germany	SevOligo/NA	13/15	7.6	2.7	5.3				Unknown
M1656	20; Kosovo	Azoo/NA	10/10	8.1	4.2	23.2	c.629A>C	p.Gln210Pro (Q210P)	S212P	Maternal	10.4; B/Tol/N	–
M2196	42; Germany	Azoo/SCO	5/7	32.0	7.7	8.83	c.775dup	p.Leu259ProfsTer60 (L259PfsTer60)	L329PfsTer16	de novo	NA	–
M2137	37; Germany	Azoo/Hypo	13/13	17.7	4.2	17.1	c.827A>C	p.Lys276Thr (K276T)	H346T	Maternal	23.6; D/D/D	–

Azoo: azoospermia; B: benign; CADD: Combined Annotation-Dependent Depletion; D: damaging/deleterious/disease causing; DND1: Dead end; MT: MutationTaster; ND: not done; Hypo: hypospermatogenesis; MAF: minor allele frequency from gnomAD v2.1.1 (Popmax); N: neutral; NA: not available; PP2: PolyPhen-2; SCO: Sertoli cell-only phenotype; SevOligo: severe oligozoospermia; SIFT: sorting intolerant from tolerant; T: testosterone; Tol: tolerated.

Reference values: testicular volume >12 ml each, FSH: 1–7 IU/l, LH: 2–10 IU/l, T < 12 nmol/l; bold font marks values outside the normal range.

Among the four identified variants, one single-nucleotide duplication (c.775dup) results in a frameshift with a premature stop codon (p.Leu259ProfsTer60). The carrier M2196 was azoospermic and presented predominant SCO (9% tubular shadows, 90% SCO, 1% spermatogonia/primary spermatocytes; Fig. 1B) and reduced testicular volumes (5 and 7 ml). FSH level was substantially increased (32 IU/l), LH level was in the normal range, and testosterone was reduced (8.8 nmol/l; Table I). Segregation analysis revealed the variant as de novo (Fig. 1C); maternity and paternity have both been confirmed. Except for p. Pro122Ser, which is extremely rare in the gnomAD database (MAF = 0.0002), all other identified variants are absent.

Zebrafish germ cells expressing orthologues of human DND1 variants fail to reach the gonad

To determine whether the DND1 variants identified in infertile men are impaired to a level that lies within the range relevant for supporting germ cell development, we evaluated their function in the context of zebrafish germline development and behavior. Here, we made use of the zebrafish version of the protein to obtain high sensitivity and introduced amino acid changes, which we considered to be analogous to the alterations found in the human variants (Fig. 2A). Specifically, we compared the function of the wild-type zebrafish DND1 protein with that of a protein altered at the position homologous to that of the human variant. The identification of the amino acid in the zebrafish to be altered was based on previous studies that have shown that the DND1 protein is conserved among vertebrates (Weidinger et al., 2003). In this way, we were able to map variants identified in the human protein sequence to specific amino acids in the zebrafish orthologue. The four variants we studied are located within phylogenetically conserved domains of the protein, which include the two RRM motifs and the double-stranded RNA-binding domain at the C-terminus (Fig. 2A and Supplementary Fig. S3). We reasoned that sequence alterations within these conserved domains could affect protein function in different organisms. Following this logic, we identified the amino acid changes caused by the respective mutations in each of the human variants and introduced the same changes into the zebrafish DND1 protein (Fig. 2A). Importantly, while the amino acids at the relevant positions can differ between human and zebrafish even within the conserved domains, they predominantly belong to the same class and therefore share structural and chemical similarity. For example, the glutamine (Q) at position 210 in the human protein corresponds to a serine (S) at position 212 in zebrafish (Supplementary Fig. S3), both classified as neutral amino acids. In the identified human variant Q210P, glutamine is changed to the structurally different proline (P). The same change was also introduced in the zebrafish protein, with an expected similar outcome concerning protein structure.

Figure 2.

Zebrafish germ cells expressing orthologues of patient-mirrored DND1 protein variants are less effective in reaching the gonad region in live embryos. (A) Schematic diagram of the Dead end protein (DND). An alignment between the human (upper row) and zebrafish (lower row) proteins is presented for the regions in which amino acid changes were identified in azoospermic men. The amino acids exchanged in the patients are labeled (magenta, arrow pointing up), with the corresponding exchanges introduced into the zebrafish protein indicated in the lower row (magenta, arrow pointing down). The magenta arrowheads indicate the start of a frameshift caused by a single base pair duplication. (B) Overview of the experimental setup employed for evaluating the activity of DND1 protein versions. One-cell stage embryos were injected with a morpholino antisense oligonucleotide (MO) that inhibits the translation of the endogenous dnd mRNA, resulting in lack of germ cells at the gonad in 24 h post-fertilization (hpf)-old embryos (condition a, negative control). MO-resistant mRNAs encoding the wild-type (wt) DND1 protein (condition b), or the human-mirroring variant protein (condition c) were co-injected with the MO to evaluate the efficiency of the DND1 protein variants in supporting the arrival of germ cells to the gonad. (C and D) The number of primordial germ cells (PGCs) located in the region of the gonad in 24 hpf embryos depleted of endogenous DND1 and injected with mRNA encoding for different variants of the protein. The activity of four variant proteins is compared with that of the wild-type protein (WT). The graph in (C) summarizes the activity of each of the variant proteins in supporting arrival of PGCs at their target, relative to that observed upon injection of mRNA encoding for the wild-type form (set to 100%). The graphs in (D) present the individual data points for dnd^S212P and dnd^L329PfsTer16, which exhibited significantly reduced protein activity. A Mann-Whitney U test was performed to evaluate the statistical significance. **P = 0.002, ****P < 0.0001. Representative images of embryos expressing dnd^WT or dnd^S212P, along with a green PGC marker are presented at the bottom panels. The red boxes mark a magnified region shown in the lower panels. The blue boxes mark the developing gonad region and the arrowheads point at ectopic PGCs. n (embryos) = 61 (dnd^WT), 46 (dnd^R124S), 75 (dnd^S212P), 44 (dnd^H346T), 75 (dnd^L329PfsTer16), and 46 (ctrl.). N = 3.

To examine whether the mutations that mirror those found in patients affect the function of the protein, we employed the following assay. We injected zebrafish embryos with MO (Nasevicius and Ekker, 2000) that block the translation of the endogenous dnd mRNA, which is known to result in a lack of PGCs within the developing gonad (Weidinger et al., 2003). To reverse this phenotype, we simultaneously co-injected dnd mRNA that was mutated within the MO-binding site, rendering it resistant to the MO (Fig. 2B). Accordingly, we injected MO-resistant mRNAs encoding for the wild-type and specific DND1 protein variants along with the MO targeting the endogenous dnd mRNA. We then assessed the number of PGCs located in the gonad region at the end of the first day of development (Fig. 2C and D). Using this approach, we found that two of four DND1 protein variants were significantly less effective in supporting the arrival of PGCs at the developing gonad region (Fig. 2C). Specifically, only 24% of germ cells expressing the zebrafish protein variant DND^L329PfsTer16 (corresponding to the DND^L259PfsTer60 variant human protein) reached at the gonad, consistent with the testicular phenotype of SCO of patient M2196 and the observed hormonal disturbances (Fig. 2D and Table I). Together with the fact that this gene variant arose de novo, these findings underline the importance of de novo mutations in human male infertility (Oud et al., 2022). Only 62% of PGCs expressing the zebrafish protein variant DND^S212P (corresponding to the DND^Q210P variant human protein) were present at the target, in line with borderline clinical features of NOA in patient M1656 displaying azoospermia (Fig. 2D). Importantly, PGCs expressing these DND1 variants were frequently found in ectopic locations within the embryo, a phenotype observed in DND1-depleted embryos (Gross-Thebing et al., 2017; Fig. 2D, bottom panels). Both DND1 variants are located within conserved domains of the protein. Specifically, the DND^S212P variant carries an amino acid substitution within the second RRM domain, and the DND^L329PfsTer16 variant carries a one-base duplication causing a frameshift at the beginning of the double-strand RNA-binding domain (dsRBD; Fig. 2A and Supplementary Fig. S3). The significant effect of both variants is comparable with earlier experimental findings in zebrafish and rats showing that point mutations and truncations within those domains cause a strong reduction or even a complete loss of protein activity (Slanchev et al., 2009; Liu and Collodi, 2010; Northrup et al., 2012). The findings we obtained from the zebrafish assay thus point at two specific gene variants as the likely primary cause for infertility in the respective patients, with the activity of the other two variants unaffected.

The DND^S212P variant protein is expressed at a reduced level in the germline

Of the two dnd variants that had a significant effect on the presence of germ cells in the gonad, we focused on the DND^S212P variant. In contrast with the DND^L329PfsTer16 variant, in which a frameshift deletes the whole dsRBD domain, the DND^S212P variant carries a single-nucleotide exchange that leads to the conversion of a single amino acid. Unlike for a deletion of a whole protein domain, as in the DND^L329PfsTer16 variant, the effect of the alteration of one amino acid on protein activity is more difficult to predict. While initially being classified as a variant of uncertain significance (VUS), the clear reduction of DND^S212P-expressing germ cells reaching the zebrafish gonad prompted us to evaluate the levels and subcellular localization of this protein variant (Fig. 3A–C). Interestingly, when evaluating the expression of HA-tagged DND1 proteins in otherwise untreated wild-type embryos, we found that the subcellular localization of the variant protein was not affected (Fig. 3A). Both the wild-type and variant protein specifically localized to germ granules, and to a lesser extent to the nucleus of PGCs, consistent with the previously known distribution pattern of DND1 protein (Weidinger et al., 2003; Slanchev et al., 2009). However, we noted a striking difference in the level of the protein, which was strongly reduced in the variant compared to the wild type. Specifically, the level of GFP-tagged DND1 proteins normalized to the signal of nuclear mScarlet revealed that the amount of the DND^S212P-GFP protein was lowered by 52% (GFP/mScarlet ratio of DND^WT = 0.054, DND^S212P = 0.026) as compared with the wild-type DND-GFP protein (Fig. 3B and C). These findings are consistent with the idea that the variant linked to male infertility reduces the stability of the protein.

Figure 3.

The DND^S212P variant protein is less efficiently expressed in zebrafish germ cells. (A) The expression and subcellular localization of the wild-type Dead end protein (DND) and the S212P variant within primordial germ cells (PGCs) of wild-type zebrafish embryos. HA-tagged DND1 proteins were visualized using anti-HA tag antibodies. Arrows indicate protein signal in germ granules. (B) Comparison of DND^WT-GFP and DND^S212P-GFP expression in relation to a germline-specific nucleus marker (mScarlet-NLS, directed by a nos 3′UTR element) in PGCs located in the gonad region. The acquired image stacks are shown as maximum intensity projections. (C) Ratio of the mean fluorescence intensity of DND-GFP or DND^S212P-GFP and mScarlet-NLS in PGCs. The measurement was conducted on segmented PGCs across a 3D image stack of the gonad region. The statistical significance was determined using the Mann–Whitney U test. ****P < 0.0001. n (dnd^WT) = 34 embryos and n (dnd^S212P) = 36 embryos. N = 3. GFP: green fluorescent protein; HA: hemagglutinin; NLS: nuclear localization sequence.

Taken together, the zebrafish assay allowed us to identify two DND1 variants with a likely effect on human fertility and define two others (which did not impair germ cell arrival at the gonad) as variants that are unlikely to be responsible for infertility in the corresponding men.

Germ cells expressing the DND^S212P variant exhibit defects in fate maintenance

The strong phenotype observed in PGCs expressing the DND^S212P protein variant (hereafter dnd^S212P PGCs) prompted us to assess another feature of DND1-depleted germ cells, namely the defects in germ cell fate maintenance and trans-differentiation. To this end, we investigated the expression of soma-specific mRNAs in wildtype and dnd^S212P PGCs. Since zebrafish PGCs are specified at different positions within the embryo and since DND1 is required for their motility, PGCs depleted of DND1 do not reach the gonad but are located throughout the embryo. Therefore, to compare dnd^S212P PGCs with wild-type PGCs, we inhibited translation of the mRNA encoding the guidance cue CXCL12A, which resulted in the equal dispersal of both control and dnd^S212P PGCs across the embryo. We have previously shown that ectopic PGCs maintain their germ cell fate, as judged by cellular behavior and gene expression (Doitsidou et al., 2002; Gross-Thebing et al., 2017). Accordingly, we injected embryos with Morpholinos inhibiting the translation of mRNAs encoding CXCL12A and DND1 and co-injected MO-resistant dnd1 mRNA encoding either wild-type or variant protein. We then focused on the developing muscle, as it represents the largest tissue during early development, hence containing a large fraction of the ectopic PGCs (Fig. 4A). Using a transgenic line expressing the TFP under the muscle-specific promoter unc45b, we compared TFP expression in wild-type PGCs with that in dnd^S212P PGCs, both depleted of endogenous Dnd1 protein (Fig. 4B). The comparison was performed by determining the ratio of the muscle marker signal intensity within the PGC and within the surrounding muscle-precursor cells. Strikingly, we found that dnd^S212P PGCs showed significantly elevated levels of the muscle marker compared to control PGCs (Fig. 4B, right graph), indicating that they fail to retain their germ cell fate, a phenotype originally observed for a complete knockdown of DND1 (Gross-Thebing et al., 2017).

Figure 4.

Zebrafish germ cells expressing the DND^S212P variant express abnormal high levels of somatic genes. (A) Overview of the experimental setup employed for evaluating the activity of the Dead end DND^S212P protein in maintaining primordial germ cell (PGC) fate. One-cell stage embryos were co-injected with morpholino antisense oligonucleotide (MO) inhibiting Cxcl12a expression, a dnd MO targeting the endogenous dnd mRNA, and a MO-resistant mRNA encoding the wild-type DND1 protein (WT), or for the DND^S212P protein. At 24 h post-fertilization (hpf), PGCs (red) located within the TFP-expressing muscle tissue (cyan) were analyzed. (B) The expression of a muscle-specific marker (TFP under the control of the unc promoter, cyan) was followed in PGCs (red membrane). Unlike PGCs expressing the wild-type protein (WT, a–c), PGCs expressing the DND^S212P version (d–i) often expressed the somatic marker, indicating defects in cell fate maintenance. Dashed lines in the middle panels mark the outlines of the PGCs. Each outlined PGC is numbered, allowing the intensity value data points to be identified in the graph on the right. The graph shows the ratio of the mean fluorescence intensity of muscle-specific TFP protein within PGCs and neighboring muscle tissue. Black numbered boxes indicate values for PGCs shown in the left image panels. A Mann–Whitney U test was performed to evaluate the statistical significance. **P = 0.002. n (dnd^WT) = 27 cells from 16 embryos and n (dnd^S212P) = 56 cells from 22 embryos. N = 3. (C) A scheme showing the experimental setup for the transcriptome analysis of dnd^WT and dnd^S212P PGCs. One-cell stage embryos were injected with mesoderm-converting factors, dnd MO and MO-resistant mRNA encoding the wild-type or variant DND1 protein. At 10 hpf, PGCs were sorted by FACS and their transcriptome analyzed. (D) Volcano plot showing up- and downregulated genes in dnd^S212P PGCs compared to dnd^WT PGCs. (E) Gene Ontology over-representation analysis of biological pathways related to upregulated genes in dnd^S212P PGCs (left plot) or Dnd KD PGCs (right plot) as compared to dnd^WT PGCs. KD: knockdown; TFP: teal fluorescent protein.

To further substantiate these findings, we conducted a transcriptome analysis of dnd^WT and dnd^S212P PGCs using the following setup: we converted the embryonic cells into cells of a single-germ layer by injecting mesoderm-conversion factors into the fertilized egg (Aoki et al., 2002; Krieg et al., 2008; Fig. 4C). This uniform environment simplifies comparison of the dnd^WT and dnd^S212P PGC transcriptome, as the trans-fating primarily involves mesoderm-expressed genes. Using this approach, we identified DEGs in dnd^S212P PGCs compared to wild type (Fig. 4D). Most of the upregulated genes function in pathways regulating somatic development (Fig. 4E, left). Importantly, we detected a similar pattern of upregulation in PGCs lacking DND1 expression, which is in line with previous findings (Gross-Thebing et al., 2017; Fig. 4E, right). The stronger changes in the gene expression pattern under the full knockdown conditions is consistent with the partial loss-of-function of the variant protein, a further demonstration of the sensitivity of the zebrafish-based approach.

Our findings concerning the maintenance of PGC fate show that the zebrafish homologue of the Q210P variant identified in human patient M1656 exhibits reduced activity, suggesting a causative link to male infertility.

Germ cells expressing the DND^S212P variant exhibit defects in cell migration

A common feature of PGCs in different organisms is their motility (Grimaldi and Raz, 2020). This behavior is not observed in DND1-deficient PGCs, which do not perform active migration and remain stationary (Weidinger et al., 2003; Goudarzi et al., 2012). To investigate whether the DND^S212P protein can support the cells’ motility, we monitored PGC migration between 8 and 10 hpf (Fig. 5A). To this end, we followed the motility of PGCs depleted of the endogenous DND1 protein and expressing either wild-type DND or the DND^S212P protein (Fig. 5B and C and Video 1). Using this approach, we found that the motility of PGCs expressing the DND^S212P variant was strongly reduced (Fig. 5C). While most of the dnd^WT PGCs exhibited active migration within a 2-h time interval, most of the dnd^S212P PGCs remained stationary (Fig. 5B and C and Video 1). These migration defects are consistent with the finding that a significant number of the dnd^S212P PGCs failed to reach the gonad (Fig. 2C and D). Together with the aberrant expression of somatic genes in dnd^S212P PGCs (as extrapolated from the evaluation of muscle gene expression, as well as the transcriptome analysis in Fig. 4), our findings show that zebrafish dnd^S212P PGCs fail to maintain germ cell fate, undergo trans-differentiation into somatic cells, and exhibit abnormal migratory behavior. In conclusion, DND^S212P protein-expressing zebrafish PGCs exhibit defects that are characteristic of DND1 deficiency, suggesting that the azoospermia observed in patient M1656 results from the reduced activity of the corresponding human DND1^Q210P protein variant.

Video 1.

Representative examples of motile and stationary zebrafish germ cells. The movie on the left shows a germ cell (red) that migrates relative to neighboring somatic nuclei (cyan) and that on the right shows a cell that does not move with respect to adjacent cells. Reference somatic nuclei are labeled with asterisks. Images were acquired at time intervals of 2 min. The zebrafish germ cells are defined and quantified in Fig. 3.

Figure 5.

Zebrafish germ cells expressing the DND^S212P variant exhibit defects in cell migration. (A) Overview of the experimental setup employed for monitoring the motile behavior of DND^S212P-expressing primordial germ cells (PGCs). One-cell stage embryos were co-injected with a dnd morpholino antisense oligonucleotide (MO) targeting the endogenous dnd mRNA and MO-resistant mRNAs encoding the DND^WT or the DND^S212P protein, along with an mRNA encoding a fluorescent nuclear protein used to visualize somatic nuclei. The motile behavior was then assessed for a 2-hour time frame in 8 h post-fertilization (hpf) embryos. (B) The dynamic behavior of PGCs (red membrane) expressing either the DND^WT or DND^S212P protein forms was evaluated by distinguishing between motile cells (left panels) and stationary cells (right panels) that do not move relative to the neighboring somatic nuclei (cyan). Reference somatic nuclei are labeled by asterisks. (C) A graph presenting the fraction of motile PGCs expressing either the DND^WT or the DND^S212P protein. The cells were followed for 2 h and the statistical significance was determined using the Mann–Whitney U test. ****P < 0.0001. n (dnd^WT) = 339 cells from 28 embryos and n (dnd^S212P) = 460 cells from 34 embryos. N = 3. DND: Dead end protein

Discussion

In this study, we present a strategy for combining clinical and genetic patient data with rapid in vivo assays in zebrafish for investigating novel human disease candidate genes and for evaluating their relevance for fertility. Specifically, using this strategy, we were able to provide data supporting the definition or exclusion of variants in the human DND1 gene derived from patients as a likely cause for infertility. Given the broad applicability of our system to other genes and disease contexts, this work provides a valuable bridging between clinical findings and experimentation in basic cell biology. This general approach will facilitate the identification of proteins that can serve as diagnostic markers and targets for therapy in the context of fertility, as well as in the context of other diseases.

Identifying the genetic causes for male infertility caused by azoospermia in men has a clinical utility, as it may allow us to predict the success rate of TESE before surgery. The result of this study, employing the zebrafish model, points at DND1 as a novel candidate gene associated with quantitatively impaired spermatogenesis and male infertility. The model we established allows the definition of other DND1 variants as hypomorphs based on a less pronounced effect on germ cell development in zebrafish embryos, which may lead to a less severe testicular phenotype in men, such as hypospermatogenesis. The mode of inheritance for variants in DND1 is, as yet, not entirely clear. The observed/expected (o/e) ratio for LoF variants is 0.21 (gnomAD v2.1.1, https://gnomad.broadinstitute.org/) meaning that less LoF variants are present than expected in the population. But the upper bound fraction (LOEUF) for DND1 is 0.55 and, therefore, beyond the recommended cut off of <0.35. Thus, these variants underlie a certain selection pressure and this may point to intolerance owing to haploinsufficiency. However, an assumed autosomal-dominant inheritance pattern for human DND1 is in contrast to that described for knockout mice and rats. Further, a recent case report of one family proposes a homozygous missense variant as causal for human male infertility (Xie et al., 2022). That missense variant, in a homozygous state, significantly decreases protein–protein interactions (DND1—NANOS2), but it was not fully abolished. Likewise, the functional impact on germ cell migration observed in this study also does not lead to a complete loss of DND1 function. Therefore, a similar biological effect of homozygous missense and heterozygous LoF variants on human male fertility is conceivable. Despite the unclear inheritance, both studies provide evidence that a concerted DND1 function is indispensable for male fertility, also in humans. Future studies will provide the required insight into the mode of inheritance and the effect of impaired DND1 function, with implications for a clinical setting.

While it is easier to link infertility to specific protein-encoding genetic loci when an open reading frame is completely deleted (e.g., as in the case of dazl (Reijo et al., 1995)), the task is more challenging for single-nucleotide variants. Our approach not only facilitates the identification of such gene variants as likely causes of infertility but also highlights variants that are less likely to cause infertility when they show no effect on protein function in the animal assays. Our strategy provides a rapid, sensitive and cost-effective readout, as the mRNA encoding the mutated protein can be directly injected and the effect on germ cells immediately assessed.

To summarize, we have demonstrated the power of this novel zebrafish-based approach for assessing human variants in the DND1 protein. This method can also be used, either identically or with modifications, to examine other germ cell-specific genes. For example, protein variants that exhibit germ cell phenotypes during vertebrate embryogenesis (e.g., doublesex and mab-3-related transcription factor 1 (DMRT1), nanos C2HC-type zinc finger 3 (NANOS3), Tudor domain containing 7 (TDRD7), heat shock protein 90 (HSP90), and tripartite motif-containing protein 71 (TRIM71); Köprunner et al., 2001; Strasser et al., 2008; Tewes et al., 2014; Pfeiffer et al., 2018; Torres-Fernández et al., 2021) could also be examined in patient-derived sequence data and evaluated as described here. While the above-mentioned examples include genes whose functions have been previously shown in animal models as being relevant for infertility, our framework is also applicable for candidate genes identified in infertile men that have not yet been studied in animal models in this context. Lastly, our experimental scheme of combining patient and animal model-derived data is not limited to fertility studies. Rather, this rapid procedure could be applicable in other areas of medicine that do not involve germline tissues. As such, this strategy offers sensitive, cost- and time-effective options for evaluating the effect of genetic variants on protein function in disease contexts.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding authors.

Acknowledgments

We thank Celeste Brennecka for editing the manuscript and Ursula Jordan, Esther Messerschmidt, Ines Sandbote, Christina Burhöi, Luisa Meier, Sarah Nagel, and Julia Richter for technical assistance.

Authors’ roles

E.R. supervised the zebrafish part of the work. K.J.W. performed the zebrafish experiments presented in Figs 2 and 3, S.R. performed the experiments presented in Fig. 4A and B, and M.R.-F. the experiments presented in Fig. 5 and Video 1. K.J.W., K.T., and T.L. performed the experiment presented in Fig. 4C–E. C.F., J.E., and M.J.W. performed the genetic analyses and evaluation of identified variants. B.S. and M.M. performed and analyzed the NGS experiments of tumor DNA. M.S.O. analyzed the exome data from the proven father cohort. D.F. performed histological staining. A.P. and S.K. recruited and took care of the patients. C.F. and F.T. supervised the human genetics part of the study. M.S. performed FACS-sorting of zebrafish PGCs. All authors read, edited, and approved the final article.

Funding

The study was funded by the DFG Clinical Research Unit 326, Male Germ Cells (to E.R., F.T., and C.F.), IZKF Münster, and RA863/15-1 (E.R.).

Conflict of interest

The authors declare that they have no conflict of interest.

Author notes

Frank Tüttelmann and Erez Raz contributed equally to this work as senior authors.