Identification of missense DMC1 variants in males with non-obstructive azoospermia

Abstract

Purpose

Human male infertility is a significant reproductive condition, with non-obstructive azoospermia (NOA) being the most severe form, resulting from impaired spermatogenesis. Many genetic variants have been identified as negatively impacting sperm development and maturation at multiple stages, leading to spermatogenic failure (SPGF). Here, we aim to study such variants, particularly those in the critical, highly conserved, meiosis-specific DMC1 (DNA meiotic recombinase 1) gene, to identify genetic candidates for male infertility and to strengthen DMC1’s existing genotype–phenotype relationships.

Methods

We used whole exome sequencing (WES) and in silico analysis to investigate select DMC1 variants in a large cohort of infertile sporadic and familial cases (n = 3150).

Results

Our familial analyses identified a homozygous DMC1 missense variant, p.Thr55Ile, in two NOA-affected male siblings. We also report additional homozygous missense variants, p.Thr164Ala and p.Tyr194Cys, and one notable, rare single heterozygous variant, p.Asp160Gly, in unrelated sporadic patients. Our 3D protein modeling indicates that each of our identified variants would significantly impact the structure and functional activity of DMC1 protein.

Conclusion

Our extensive genomic study identified three rare, recessive DMC1 variants in human NOA patients. Further, we report an alternative maturation arrest phenotype than previously observed in DMC1-related NOA. We also provide preliminary support for the possible exploration of select single heterozygous variants in the DMC1 gene, potentially expanding the male infertility field’s understanding of the disease states and inheritance patterns associated with variants in DMC1.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10815-025-03591-6.

Introduction

Infertility is a significant reproductive health issue worldwide, affecting approximately 10% of reproductive-aged males and can result from environmental exposures, lifestyle choices, chromosomal aberrations, and genetic origins [1, 2]. Azoospermia, the lack of spermatozoa in the ejaculate, is a particularly severe and clinically complex type of infertility in males and can be categorized further into obstructive azoospermia (OA) due to morphological obstruction with no impacts to spermatogenesis and non-obstructive azoospermia (NOA) due to disrupted or poor spermatogenesis without morphological obstruction [1, 2]. NOA is a heterogenous condition, occurring in approximately 0.4–2% of the general male population and approximately 10–15% among infertile males [3]. A genetic cause is presumed in a significant portion of those with severely impaired spermatogenesis, with many of these mutations occurring in meiosis-specific genes required for spermatogenic progression [1–4]. One critical feature of meiosis in spermatogenesis is meiotic recombination during meiosis prophase I to generate genetic diversity in male germ cells [5–7]. Meiotic prophase I is characterized by several processes, including DNA replication, the creation of DNA double-strand breaks (DSBs) with mismatch repair, meiotic recombination, and the separation of the synaptonemal complex (SC) [1, 6–8]. Any interference with these events may result in abnormalities of spermatogenesis and a decrease in or lack of viable sperm in semen.

Recently, biallelic and hemizygous single-nucleotide variants in several meiosis-linked genes have been identified as associated with NOA phenotypes. These genes include Meiosis 1 Associated Protein (M1AP), Testis Expressed 11 (TEX11), Meiotic Double-Stranded Break Formation Protein 1 (MEI1), Synaptonemal Complex Protein 3 (SYCP3), and Testis Expressed 15 (TEX15), among other gene candidates with varying significance of evidence of causality [2, 4, 9–13]. In this study, we applied a genomic approach and utilized in silico models to identify the likely genetic basis of NOA; our analysis identified multiple variants in the DMC1 gene within our patient cohort.

DMC1 (DNA meiotic recombinase 1) is a meiosis-specific gene which codes for an essential recombinase involved in the repair of DSBs [8, 14]. DMC1 missense and loss-of-function variants have previously been identified as likely causal for NOA in two family cases and two sporadic cases [15–17]. In functional studies of DMC1, mutant models appear to impact D-loop formation during DSB repair, preventing meiotic progression [5, 8, 18–24]. These previous mice, yeast, and other functional DMC1 models support that significant variants in DMC1 result in azoospermia and meiotic impairments [5, 8, 18–24]. Here, we report three extremely rare variants in the DMC1 gene in a large cohort of infertile males and consider potential impacts of one rare heterozygous missense variant.

Materials and methods

Study cohorts

Patients were recruited from Khyber Medical University, Pakistan (n = 34 familial cases); the GEMINI Consortium (n = 1474 sporadic cases); Magee-Womens Research Institute, University of Pittsburgh, USA (n = 146 combined sporadic and familial cases); Institute of Biomedicine and Translational Medicine, University of Tartu, Estonia (n = 713 sporadic cases); Biosciences Institute, Faculty of Medical Sciences, Newcastle University, UK (n = 743 combined sporadic and familial cases); and Institute of Human Genetics, Polish Academy of Sciences, Poznan, Poland (n = 40 sporadic cases). The study was approved by the Institutional Research Ethical Board (IREB) of Khyber Medical University, Pakistan (No.KMU/IBMS/IREB/8th meeting/2024/1685) and the Institutional Review Board of the University of Pittsburgh, USA (PRO10030036). Informed consent was obtained from patients and family members. All patients were diagnosed based on American Urology Association and American Society for Reproductive Medicine guidelines [25]. Clinically, patients were evaluated for serum reproductive hormone levels, karyotyping, testicular ultrasound, and testicular biopsy. Secondary reproductive variables, such as ejaculatory problems, immunological irregularities, congenital illnesses, and environmental exposures, were ruled out. Individuals with nonidiopathic azoospermia (e.g., trauma, surgery, medication, chromosomal aneuploidies, or Y-microdeletions) or obstructive azoospermia cases were excluded from the study.

Testicular biopsy and histology

Testicular biopsies were collected from both affected siblings in the family case (INF61) at Khyber Medical University, Peshawar, Pakistan. The patients were briefed about the procedure, and written informed consent was obtained. The tissues were fixed overnight in Bouin’s solution and subsequently processed using an automated tissue processor (Leica ASP 300; Leica Biosystems, Deer Park, IL, USA). Following processing, the tissues were embedded in paraffin using an automated embedding center (Leica EG 1160, Leica Biosystems, Deer Park, IL, USA). For histological evaluation, tissue sections of 3–5 µm thickness were prepared and stained with hematoxylin and eosin (H&E) following the protocol outlined by [26]. The staining was performed using an automated H&E staining system (Leica ST5010, Leica Biosystems, Deer Park, IL, USA). The sections were microscopically examined utilizing the Nikon Eclipse 90i DS-U3 microscope (Nikon Instrument, Inc. USA).

Whole exome sequencing (WES)

DNA was extracted from peripheral blood, and WES and subsequent analysis were carried out as previously described [27]. Briefly, DNA was extracted from blood with the Gentra Puregene kit (Qiagen, USA). WES libraries were created for samples prepared in the University of Pittsburgh with the SOPHiA Whole Exome Solution (SOPHiA GENETICS, Inc., USA) and sequenced on the Illumina NovoSeq 6000 (Novogene, Sacramento, CA). Paired-end fragments were sequenced with a target read length of 150 bp and average depth coverage of ~ 110 × per target interval. Raw data quality obtained was evaluated with FastQC software (Babraham Bioinformatics, Babraham Institute, UK). FASTQ files were aligned using SOPHiA Genetics DDM platform (SOPHiA GENETICS, Inc., Boston, MA, USA) and Emedgene Illumina (Emedgene Inc., CA, USA) utilizing the DRAGEN bioinformatics pipeline version 3.6 with GRCh37/hg19 as the reference human genome (Emedgene pipeline).

Whole exome sequencing data analysis

WES variants were annotated using the SOPHiA Genetics DDM (SOPHiA GENETICS, Inc., Boston, MA, USA) and Emedgene Illumina (Emedgene Inc., CA, USA) platforms. All identified variants were screened using stringent criteria (Supplemental Fig. 1). Rare variants with MAF < 0.01 and of VUS or greater classification were considered [2, 28, 29]. GnomAD v4.1.0 (Broad Institute) [30] was used to determine variant allele frequency in the general population. For the clinical interpretation and classification of variants, we followed American College of Medical Genetics and Genomics (ACMG) [31] (https://www.acmg.net/) guidelines and standards, and the following bioinformatics pipelines were used: OMIM (Online Mendelian Inheritance in Man) (Johns Hopkins University) (https://www.omim.org/) [32], HGMD (Human Gene Mutation Database) (http://www.hgmd.cf.ac.uk/ac/index.php) [33], and ClinVar (NCBI) (https://www.ncbi.nlm.nih.gov/clinvar/). For the evaluation of genetic variants in murine mammals, MGI (Mouse Genome Informatics) (Jackson Laboratory, USA) (http://www.informatics.jax.org/) [34] was used. For evaluation of gene expression at the tissue level, HPA (Human Protein Atlas) (https://www.proteinatlas.org/) [35], The Genotype-Tissue Expression (GTEx Portal) (https://gtexportal.org/home/) [36], and AceView (National Center for Biotechnology Information, NCBI) (https://www.ncbi.nlm.nih.gov/) databases were used. PhyloP (Phylogenetic p-values) (Cornell University) (http://compgen.cshl.edu/phast/) [37] and Clustal Omega (EMBL-EBI) (https://www.ebi.ac.uk/Tools/msa/clustalo/) [38, 39] were used for assessing variant conservation. Combined Annotation Dependent Depletion (CADD) (University of Washington, Hudson-Alpha Institute for Biotechnology and Berlin Institute of Health) (https://cadd.gs.washington.edu/) [40], PolyPhen-2 (Harvard University) (http://genetics.bwh.harvard.edu/pph2/) [41], Sorting Tolerant From Intolerant (SIFT) (J. Craig Venter Institute) (https://sift.bii.a-star.edu.sg/) [42], MutationTaster (Charité) (https://www.mutationtaster.org/) [43], and REVEL (Rare Exome Variant Ensemble Learner) (ucsc.edu) [44] were used for prediction of consequence of amino acid changes on protein function. We utilized SpliceAI [45] and MaxEntScan (Massachusetts Institute of Technology) (http://hollywood.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq.html) [46] for evaluating the effect of variants on possible splicing alterations. WES variants were validated by Sanger sequencing utilizing BigDye® Direct Sanger Sequencing Kit (ThermoFisher Scientific). Primers were designed using Primer3Web [47–49]. The Sanger sequencing results were analyzed by SnapGene software (www.snapgene.com).

2D and 3D protein modeling

To identify and model the specific domains where the variants are located, the human DMC1 protein sequence (accession#Q14565) was retrieved from UniProt [50]. We utilized the full-length DNA-bound DMC1 post-synaptic structure determined by cryo-electron microscopy (PDBID 7c98) [51] and the co-crystal structure of an octameric ring bound to a peptide from BRCA2 (PDBID 8qqe) [52]. Three-dimensional structural models were developed using AlphaFold2 [53]. PyMOL (Schrödinger, LLC) [54] was used to generate structural model figures. Domain boundaries were individually identified by Pfam [55] and then manually curated.

Results

Homozygous missense DMC1 variant co-segregates with NOA in a family case

Two brothers (INF61-2 & INF61-3) from a Pashtun family, recruited from Khyber Pakhtunkhwa, Pakistan, presented with primary male infertility (Fig. 1). The proband (INF61-2) was 30 years old and married for 4 years, and the affected brother (INF61-3) was 28 years old and married for 3 years. Both brothers were found to have NOA on semen analysis with unremarkable semen volume, pH, and viscosity (Table 1). Physical examinations revealed no abnormalities in weight, height, or secondary sexual features for either sibling. Serum reproductive hormones were normal in the proband, while the affected brother had an elevated serum FSH level (Table 1). Testicular ultrasound of both brothers revealed unremarkable testes and epididymides with left grade I varicocele (Table 1). Testicular biopsy of the proband and the affected brother revealed spermatogenesis up to the level of spermatozoa with a Johnson’s score of 5.5/10 and 5/10, respectively (Fig. 2). Both brothers had a normal 46,XY karyotype. No Y chromosome microdeletions or vas deferens-associated mutations were identified in the affected brothers (Table 1). There was a third unaffected male sibling in this family who had a confirmed normal sperm count and morphology via semen analysis.

Fig. 1.

Pedigree of the INF61 familial case with two affected brothers diagnosed with non-obstructive azoospermia (NOA). Squares represent males and circles represent females. The cross through the father’s square indicates that the father was deceased prior to the proband’s recruitment into our study. The affected brothers are represented by filled squares. P with the arrow indicates the proband, which was the brother that was first recruited and evaluated in our study. The open squares and circles represent normal, fertile males and females of the family, respectively. Genotype c.164C > T, p.Thr55Ile shown as +/+ = wildtype; +/− = heterozygous; −/− = homozygous

Table 1.

Clinical parameters of the affected family members and sporadic cases

ID	Phenotype	Sperm conc. (mil/ml)	Vol (ml)	FSH (0.95–11.95 mIU/mL)	LH (1.8–8.6 mIU/mL)	Testo (1.63–34.0 nmol/L)	Testicular ultrasound	Testicular biopsy	Karyotype	AZF del
INF61-2	NOA	0	Norm	7.27	4.18	24.52	Left grade I varicocele	Spermatogenesis seen up to the level of Spermatozoa. Johnson’s score 5.5/10	46, XY	Neg
INF61-3	NOA	0	Norm	15.37↑	5.19	17.03	Left grade I varicocele	Spermatogenesis until spermatids is seen in a few tubules. Rare spermatozoa are also identified. Johnson’s score 5/10	46, XY	Neg
GEMINI-204	NOA	0	Norm	3.19	N/A	N/A	Bilateral varicocele and small testes	Maturation arrest	46, XY	Neg
GEMINI-640	NOA	0	Norm	N/A	N/A	N/A	N/A	N/A	46, XY	Neg

NOA, non-obstructive azoospermia; Vol, volume; FSH, follicle stimulating hormone; LH, luteinizing hormone; Testo, testosterone; AZF del, azoospermia factor deletion; Norm, normal; N/A, not available; Neg, negative; ­, elevated. Measuring units for FSH and LH are mIU/ml and for testosterone nmol/l

Fig. 2.

Hematoxylin and eosin staining of testicular biopsy section for affected patient in the INF61 family case. Histological examination of sections shows maturation up to the level of rare round spermatids. In the testis section, there appears to be decreased germ cell presence and apparent apoptotic germ cells gathering in the center of the tubules

To determine the likely genetic etiology of NOA, DNA samples from both affected brothers, mother, and unaffected brother were analyzed using WES. Briefly, candidate infertility variants were first determined using familial segregation data. Any genotypically feasible variants were then further interrogated based on biological feasibility (i.e., protein/RNA expression in relevant tissues, in silico predictions of amino acid change consequence and previous evidence linking them to infertility in either humans or model organisms) (Supplemental Fig. 1) [2, 28, 29]. Based on the plausible inheritance mode and biological molecular factors, we first identified homozygous missense variants in autosomal recessive (AR) candidate genes, including DMC1, ISCU, LAMA4, MUS81, PCM1, PSMD13, SOHLH1, SPATA31A1, CCDC157, BAIAP2L2, for X-linked (XL) hemizygous variants in ARMCX4 and GPRASP3, and a de novo X-linked variant in RHOXF2B in the proband, INF61-2. However, after further evaluation of these potential candidate genetic variants using familial segregation, with the infertile brother as our positive control and the fertile brother as our negative control (Supplemental Fig. 1), the homozygous variant in DMC1 was identified as the only likely candidate for NOA in the two affected brothers.

The identified rare homozygous missense variant in DMC1, c.164C > T; p.Thr55Ile, was shared between the two NOA affected brothers, not present in the unaffected brother, and was heterozygous in both the mother and sister, which indicated an AR mode of inheritance; zygosity of the variant in sampled family members confirmed via Sanger sequencing (Fig. 3). Notably, our evaluation of chromosomal regions of homozygosity found significant regions of homozygosity on Chromosome 22 (in which the DMC1 gene is located) in both affected brothers, whereas the same runs of homozygosity were not found in the unaffected brother (Supplemental Table 1).

Fig. 3.

Sanger sequencing confirmation of DMC1 variant in the INF61 family (c.164C > T; p.Thr55Ile). Both the affected brothers are homozygous for the variant (T/T) as represented by the red peak with no overlap. The unaffected fertile brother is homozygous for the wild type allele (C/C) as represented by the blue peak with no overlap, while the fertile sister is heterozygous (C/T) for the variant as represented by the overlapping red and blue peaks

The minor allele frequency (MAF) of the segregating variant, p.Thr55Ile, is 0.000001231 according to gnomAD V4.1.0 [30]. The overall allele count is 2 in the South Asian populations; one male and one female carrying the heterozygous variant; no individuals homozygous for this variant were previously reported. In silico predictors suggest consensus deleterious impacts for the identified, highly conserved variant (Table 2; Fig. 4). Histopathological evaluation of the testicular biopsy in the two affected brothers revealed spermatids in a few tubules with rare spermatozoa (Fig. 2).

Table 2.

Genetic variants in DMC1; NG_017203.2

ID	DNA coding change	Protein change (AA)	Zygosity	MAF*	In silico predictions (P, F, S, T, D, C, R)
INF61	c.164C > T	p.Thr55Ile	Hom	0.000001231	Path, 6.35, Del, Del, 23.6, 0.456
GEMINI-204	c.490A > G	p.Thr164Ala	Hom	0.000003104	Path, 6.29, Del, Del, 25.8, 0.482
GEMINI-640	c.581A > G	p.Tyr194Cys	Hom	0.000004960	Path, 6.30, Del, Del, 32, 0.676

AA, amino acid; Zygosity, inheritance pattern of the variant; Hom, homozygous; Het, heterozygous; MAF, minor allele frequency; P, polyphen; F, phyloP; S, SIFT; T, MutationTaster; C, CADD; R, REVEL; Path, pathogenic; Del, deleterious. *Minor allele frequency (MAF) values are from gnomAD v4.1.0. The overall allele count in South Asian populations is 2; one male and one female; no homozygous variants found

Fig. 4.

Cross-species conservation assessment of DMC1 protein amino acid sequence. Seq alignment shows high conservation of the substituted amino acids in the DMC1 protein sequence. Asterisk represents perfect conservation across all species; colon represents near-perfect conservation across all listed species; blank space represents lack of conservation at position. The red boxes represent positions of the missense variants reported in this study. Created using Clustal Omega (EBI) [38, 39]

DMC1 RNA and protein expression profile

After identifying the likely causal variant in the family case, we sought to understand the DMC1 protein and assess whether we would expect to find genetic variations in this gene as likely causal for infertility in other cases beyond the INF61 family. First, we assessed DMC1 protein expression, which exhibits predominant expression in the testes, with the highest RNA single-cell specificity observed in spermatocytes [56]. We then analyzed a single-cell RNA sequencing dataset from cryopreserved normal adult human testis to examine the germ cell-specific expression profile of DMC1 (Supplemental Fig. 2). Global analyses were conducted using results from 35,941 transcripts obtained from 13,597 cells to identify human testicular cell types at different stages [56]. Sequencing analyses revealed high expression levels of DMC1 mRNA in differentiating spermatogonia, with low, scattered expression observed in other germ cell types (Supplemental Fig. 2). DMC1 protein is highly expressed in differentiating spermatogonia, with low-scattered expression in undifferentiated spermatogonia, spermatocytes, and spermatids, as supported by data from major human protein databases, including the Human Protein Atlas (HPA) with a transcript per million (TPM) value of (11.5), Genotype-Tissue Expression (GTEx) with a TPM value of (13.70), and AceView records revealing the highest expression level of DMC1 in testicular tissue.

Genomic study of sporadic cases revealed two additional homozygous DMC1 variants linked to NOA

Given our findings in the familial case and our assessment of DMC1 protein/RNA expression, we expanded our WES analysis to the sporadic and other familial cases from our Pittsburgh study cohort and replication in our collaborators’ cohorts, which revealed two additional recessive DMC1 variants (Supplemental Fig. 3). No other likely candidate variants for infertility were identified in these two cases. The GEMINI-204 case had non-obstructive azoospermia and normal serum FSH levels. The patient presented with bilateral varicocele and small testes as observed on testicular ultrasound. Testicular biopsy indicated maturation arrest. WES identified a homozygous missense variant in DMC1; c.490A > G; p.Thr164Ala. In our third case, GEMINI-640 presented with NOA. While additional clinical data was unfortunately unavailable for this patient, WES analysis led to the identification of a homozygous missense variant in DMC1; c.581A > G; p.Tyr194Cys (Table 2), as previously identified but not evaluated by Nagirnaja et al. [3].

For the sporadic cases, the MAFs for the variants p.Thr164Ala and p.Tyr194Cys are 0.000003104 and 0.000004960, respectively, according to gnomAD V4.1.0. Like the family case variant, these two variants were not observed in a homozygous state in gnomAD. Additionally, in silico prediction softwares also suggest consensus deleterious impacts from the conserved variants in the sporadic patients (Table 2; Fig. 4). Both identified variants were assessed via in silico analyses based on ACMG variant classification guidelines (Table 2).

Computational model for deleterious impact of the DMC1 variants on protein

To evaluate the impact of the identified variants on the DMC1 meiotic recombinase, we performed in silico analyses to model the expected consequences of amino acid changes from these missense variants. Variant p.Thr55Ile, which co-segregated in two NOA affected siblings in the family case, lies in the N terminal Helix-hairpin-helix (HHH) 5 domain of the DMC1 protein. HHH is a widespread motif involved in non-sequence-specific DNA binding (Fig. 6). The other variants, p.Thr164Ala & p.Tyr194Cys, identified in the two unrelated sporadic NOA cases, both lie in the DNA repair/RAD51 domain of the DMC1 protein, which is crucial for the assembly of RAD51, promoting homologous pairing and strand exchange reactions [57, 58] (Fig. 5).

Fig. 6.

DNA-bound DMC1 post-synaptic filament cryo-electron microscopy 3D model. A The 3D model of one DMC1 oligomer form bound to DNA and AMP-PNP (PDBID 7c98 [51]) and rotated 90 degrees. B A zoomed-in depiction of a section of the interface where all four amino acid positions of interest are seen. C Zoomed-in visualization of the ion-binding pocket in the ATPase domain ATP-binding site including amino acids D160 and T164. D Visualization of amino acid positions T55 and Y194, showing how they are on different sides of the interface between two DMC1 monomers and help facilitate interaction between the two different monomers

Fig. 5.

2D schematic of the 340 amino acid human DMC1 protein. Helix-hairpin-helix (HHH), shown in light blue, is a motif involved in non-sequence-specific DNA binding. The majority of HHH motifs function as DNA-binding modules. The overlapping DNA repair/RAD51 domains, shown in pink, are thought to play an important role in the assembly of Rad51 that promotes homologous pairing and strand exchange reactions. The product of RecA_ATP binding domain, shown in orange, is a multifunctional enzyme that plays a role in homologous recombination, DNA repair, and induction of the SOS response. In homologous recombination, the protein functions as a DNA-dependent ATPase, promoting synapsis, heteroduplex formation, and strand exchange between homologous DNAs. The missense variant (p.Thr55Ile) identified in the two affected brothers in the INF61 familial case is in the HHH_5 domain. The other missense variants in the sporadic cases, p.Thr164Ala and p.Tyr194Cys, are in the DNA repair/Rad51 domain. Likely pathogenic variants of DMC1 identified are listed above the protein line. Variants shown below the protein line are previously reported [3, 15–17].

To develop hypotheses for the underlying structural bases of the mutations yielding the observed phenotypes, we examined two DMC1 structures, representing different functional conformations observed for DMC1 [59] and other RAD51 homologs [60–62]. Typical to this family of enzymes, the fold of these ATP-dependent enzymes contains canonical Walker A and Walker B motifs within the AAA + ATPase domain [63] and ions bound in the magnesium-binding sites; a nucleotide analog is observed in the post-synaptic structure [51]. The monomers in the filament and octameric ring foster different oligomeric interactions with each other. All residues of interest lie on one interface in the post-synaptic filament structure; amino acid residues Thr164 and Tyr194 from one oligomer interact with amino acid residue Thr55 from the neighboring oligomer in the post-synaptic structure (Fig. 6B). In the octamer, however, residues Thr164 and Tyr194 are near an alternative interface with the neighboring DMC1 oligomer; in subunits where residue Thr55 is visible, it interacts with residues near His211 of a neighboring DMC1 oligomer (Fig. 7B).

Fig. 7.

DMC1 bound to BRCA2 peptide octameric ring co-crystal structure 3D model. A The octamer structure DMC1 forms with BRCA2 peptide, magnesium, and chlorine ions interacting with DMC1 (PDBID 8qqe [52]). B Visualization of amino acid T55 of one monomer as it is closely positioned to position H211 from another monomer. C Visualization showing the interface with amino acids D160, T164, and Y194, showing a hydrogen bond between Y194 and E258

All mutations of interest would be predicted to affect complex assembly and/or ATP binding to the recombinase. In the octameric ring (Fig. 7B), the side chain of amino acid Thr55 sits in a polar pocket supported by Arg57 and His211 of the neighboring subunit. Therefore, in this conformation, mutation p.Thr55Ile would be predicted to disrupt this interface as the hydrophobic side chain of an isoleucine would not be expected to interact with those polar residues. Of note, this interface could affect the binding path of other proteins that interact with this portion of the ring, including the BRCA2 peptide, thereby disrupting protein functional capacity. In the filament conformation of the protein, Thr55 approaches Asp232 and Tyr194 of the next monomer in the filament; mutation p.Thr55Ile would be predicted to disrupt this interface as well. Residue Tyr194 plays an important role in facilitating inter-monomer interactions in both conformations of the enzyme. In the octameric form, Tyr194 participates in a hydrogen bond with Glu258 from the neighboring DMC1 monomer [22] (Fig. 7C). In the filament form, the side chain of Tyr194 hydrogen bonds with the side chain of Lys49 of the neighboring oligomer (Fig. 6D). Therefore, mutation p.Tyr194Cys would be expected to destabilize complex formation for both conformers. Residue Thr164 coordinates the calcium ion in the ATP-binding site (Fig. 6C). Therefore, loss of the polar side chain at Thr164 would destabilize ion binding in this site, thereby negatively impacting the activity of the ATPase function of the enzyme.

Discussion

Meiosis emerged with the rise of eukaryotes for the purpose of recombining chromosomes to improve genetic diversity and gamete fitness during reproduction. While many different genes evolved to facilitate this process, DMC1 is one of the most critical and well-conserved genes across plants, fungi, and animals, and acts as the meiotic counterpart to the well-known DNA damage response gene, RAD51 [6, 8, 18, 57, 64]. While both homologous genes broadly work to repair DSBs in dividing somatic and sex cells, in meiosis specifically, RAD51 accumulates at sites of DSBs to recruit DMC1 for d-loop formation necessary for DSB repair and the progression of meiosis [6, 57, 64]. In previous studies, missense variants in RAD51 have been indicated as likely pathogenic for various DNA repair-defective disease phenotypes [65]. Given the critical and similar role both genes serve, we postulate that we have similarly observed likely pathogenic, recessive missense variants in DMC1, as opposed to more deleterious loss-of-function (LoF) variants, as candidate variants for infertility phenotypes in our patient cohort.

While we did not perform functional studies of our identified DMC1 variants, we used information from previously generated functional models and in silico databases to support the hypothesized pathogenicity of our missense variants. In a previous study of human DMC1, researchers knocked in multiple missense variants, including p.Glu258Ala, which destabilized the DMC1 octamer and impacted protein function [22]. As our DMC1 protein modeling indicates, one of our residues of interest, Tyr194, directly interacts with Glu258, so we hypothesize that our mutation p.Tyr194Cys would likely result in a similar impairment to octamerization and protein function. Further, in a previously generated DMC1 yeast model, the hypomorphic mutation, dmc1-T159A, which was analogous to the observed human variant in our study, p.Thr164Ala, resulted in oligomerization failure and negative impacts to yeast meiotic progression [21]. Given this demonstrated defect in yeast, we hypothesize that our orthologous variant in GEMINI-204 would produce similar results in humans, triggering meiotic arrest. Finally, the pathogenicity of missense variants in DMC1 is supported by gnomAD v4.1.0, which revealed extremely low frequencies of any identified variants within the general population, suggesting that both LoF and missense variants in DMC1 are naturally selected against. In summation, we believe that our findings greatly strengthen DMC1’s classification as a clinically significant male infertility gene and provide evidence that significant missense variants may be sufficient for causing AR pathogenic NOA phenotypes.

More novelly, with consideration of the high degree of conservation across phyla of DMC1 and its known key interactors, we speculate that it may be possible that due to currently limited understanding of disease states associated with DMC1 in humans, single heterozygous variants may be more damaging than previously considered. In Saccharomyces cerevisiae models, missense variants in dmc1 have been observed with both recessive and dominant patterns of pathogenicity, depending on the conformational, structural, or functional changes introduced by modeled variants in the DMC1 protein [23]. In one study of the recessive model variant, dmc1-K69E, one mutated allele resulted in inhibition of oligomeric formation involving monomers within the protein; however, remaining WT alleles were sufficient to maintain functionality of the protein in meiosis [23]. Conversely, a dominant modeled variant, DMC1-G126D, in the putative ATP-binding site, did not affect oligomerization, but was predicted to negatively impact the activity of DMC1 in meiosis [23]. As noted in Dresser et al., the dominant DMC1 variant occurred in a conserved motif region shared with RAD51, in which a homologous dominant mutation in RAD51 at the same position, RAD51-G190D, also resulted in null protein activity for RAD51 [5, 20, 23]. Similarly, in another yeast DMC1 knock-in missense model, the dmc1-E157K allele (analogous to human Glu162 position) displayed a dominant negative effect and eliminated spore viability in a heterozygous diploid yeast model [24]. Our proposed variant, p.Asp160Gly, is in close proximity to the orthologous human position in the dominant negative-effect yeast model. Given the close proximity and our variant’s strongly predicted splice gain effect (Supplemental Fig. 4), we hypothesize that this mutation may be sufficient to cause a significant impact to this critical region, demonstrating a similar dominant negative effect as observed in the yeast model.

Further along in the evolutionary phyla, in mice, a dominant, missense Dmc1 variant was created in a chemically mutagenized embryonic stem cell study [5, 20]. Most notably, this missense allele (Mei11) in mice resulted in dominant fertility defects, with male sterility and females exhibiting reduced reproductive lifespans [5]. Furthermore, Bannister et al. recreated the missense allele (HuDMC1Mei11) in human DMC1 protein, and it was determined that the mutant had a threefold decrease in single-strand DNA-binding activity [5]. Cumulatively, assessment of previously generated functional yeast and mouse models supports that any dominant effects observed in DMC1 mutant models are not caused by haploinsufficiency, but instead by dominant negative missense variants resulting in gain of function defects [5, 23, 24]. Finally, if we again look to the structurally and functionally similar DSB repair gene RAD51 as a model, identified human missense variants have been functionally demonstrated to produce dominant negative effects resulting in significant disease phenotypes [65]. We postulate that it may be possible that similar dominant effects may be seen in select, significant DMC1 variants in humans, as they function in the same complex in meiosis for DSB repair and are evolutionarily both homologs of bacterial recA [7, 18, 23, 57].

While not functionally tested, we identified a single heterozygous missense variant with a putative splice gain effect, c.479A > G;p.Asp160Gly, which could be considered a candidate variant for the less severe cryptozoospermia phenotype seen in the respective patient (Supplemental Table 2). As hypothesized above, we believe that this single heterozygous variant may be sufficient to impair spermatogenesis in our affected male patient, though less severely given the observed cryptozoospermia phenotype in this patient. This notion is further supported by protein modeling, which predicts that the p.Asp160Gly variant is likely to impact the ATP-binding domain of the DMC1 protein (Fig. 6C). Additionally, the coding mutation, c.479A > G, is predicted to have a splicing impact, possibly creating an alternative acceptor splice site which could cause a more significant impact to this critical region (Supplemental Fig. 4).

Considering all developments in the male infertility field, one begins to explore the potential for more dominant causes of NOA [1]. One convincing option may be to begin functionally evaluating identified single heterozygous variants in DMC1 and other meiosis-related genes [66]. Further, as seen in recent studies from Lillepea et al. [1], it may be possible that single heterozygous variants in conjunction with variants in other candidate male infertility genes may result in a disease phenotype, via multigenic compounding deficits on structural and mechanistic functions, which should be considered in future analyses and explored in functional models.

To our knowledge, this study represents the largest human-focused study of DMC1 variants in human male infertility. We demonstrate strong support for DMC1 as a recessive monogenic cause of male infertility in humans, and we postulate that there may be a foundation for further exploring convincing single heterozygous variants for autosomal dominance and/or multigenic effects. Additionally, we expand the current known gene-phenotype relationship for DMC1. Previously reported cases exclusively identified the complete absence of spermatozoa and post-meiotic cells in testicular biopsy [15–17]. Conversely, in the INF61 family case, few spermatozoa cells were identified, suggesting a possible variable phenotype with DMC1-related infertility, which may indicate that certain DMC1 variant positive patients may be candidates for assisted reproductive technologies.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

NU, CP, and RHG were responsible for writing the manuscript and researching the DMC1 variants. AJB was responsible for protein modeling. CP, RHG, and MM were responsible for handling the Pittsburgh blood samples. MJK, RN, SF, and NU were responsible for recruiting the patients from Pakistan. ARE, HCM, and DFC were responsible for performing the GEMINI database search of the gene variants and providing clinical patient data. AM, MO, and MK were responsible for the Polish cohort variant search and patient clinical information. KL, AD, AV, RI, and ML were responsible for the Estonian cohort. MJX and JAV were responsible for the Newcastle cohort. CAC was responsible for histological preparation and evaluation. KEO and ANY were responsible for the Pittsburgh cohort. All authors participated in editing/review of manuscript.

Funding

This study was supported by the Eunice Kennedy Shriver NICHD Grant HD080755 (ANY), the Magee-Womens Research Institute University of Pittsburgh Start Up Fund (ANY), NIH P50 Specialized Center Grant HD096723 (ANY, DC, and KO), National Science Centre in Poland, grants no.: 2017/26/D/NZ5/00789 (AM) and 2015/17/B/NZ2/01157 (MK), GM127569 (ANY), the National Health and Medical Research Council Project grant APP1120356 (DC) and Khyber medical university faculty research grant Ref no. KMU/ORIC/FARE/IPMS/028. Computational analysis was supported in part by the University of Pittsburgh Center for Research Computing through the resources provided.

Data Availability

The variants presented in this study are deposited into the ClinVar repository, accession number SUB15022908.

Declarations

Ethics Statement

The studies involving human participants were reviewed and approved by the Institutional Research Ethical Board (IREB) of Khyber Medical University (No.KMU/IBMS/IREB/8th meeting/2024/1685), the Institutional Review Board of the University of Pittsburgh (PRO10030036), and the Local Bioethics Committee of Poznan University of Medical Sciences (1,003/18). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.