HIPK4 is a novel gene associated with teratozoospermia and male infertility

Abstract

STUDY QUESTION

Are pathogenic variants in homeodomain-interacting protein kinase (HIPK4) associated with sperm head abnormalities that cause male infertility?

SUMMARY ANSWER

HIPK4 is a novel candidate gene associated with sperm head defects and human male infertility.

WHAT IS KNOWN ALREADY

Numerous genes have been described in which pathogenic variants cause male infertility due to multiple morphological abnormalities of the sperm flagella (MMAF), but the genetic basis of sperm head defects is less well understood.

STUDY DESIGN, SIZE, DURATION

This study included four infertile brothers displaying varying degrees of quantitatively and/or qualitatively impaired spermatogenesis, their parents, and their fertile brother. We also queried the Male Reproductive Genomics (MERGE) cohort comprising exome/genome sequencing data of >3300 men.

PARTICIPANTS/MATERIALS, SETTING, METHODS

We performed exome sequencing in all five brothers and their parents. To characterize sperm phenotypes, we carried out standard semen analysis, immunofluorescence staining, and transmission electron microscopy (TEM). Further, we evaluated the impact of the HIPK4 variant in cell culture experiments using HEK293T cells.

MAIN RESULTS AND THE ROLE OF CHANCE

By analysing the exome data, we could not identify a common genetic cause in all four affected brothers. However, one of the affected brothers was compound heterozygous for two loss-of-function variants in DNAH17 (c.1076_1077dup p.(Lys360*) and c.7752+2T>A p.?), associated with markedly reduced sperm motility and MMAF. The variants’ pathogenicity was further validated by TEM of flagellar cross sections revealing an outer dynein arm defect and axonemal disruption. In contrast, his three infertile brothers were homozygous for the start-loss variant c.1A>G in HIPK4. This gene is expressed during spermiogenesis and is reportedly involved in sperm head shaping in mice. Heterologous expression of (partial) HIPK4 variant cDNA showed that translation was being initiated at an alternative in-frame start codon located 35 amino acids downstream, resulting in an N-terminally truncated protein p.(Met1_Glu35del). The truncated HIPK4 protein lacks parts of its kinase domain and shows reduced protein stability. Corresponding with published mouse models, all three brothers displayed 100% abnormal sperm head morphology with variable defects. Importantly, one brother affected by HIPK4 variants fathered a child after successful ICSI, demonstrating a successful treatment option for HIPK4-related teratozoospermia. No further men from the MERGE cohort were affected by biallelic HIPK4 variants. Taken together, HIPK4 is an autosomal recessive candidate gene in which pathogenic variants are associated with sperm head defects and male infertility.

LARGE-SCALE DATA

The reported variants in DNAH17 and HIPK4 have been published in ClinVar.

LIMITATIONS, REASONS FOR CAUTION

Independent replication is required to assess the phenotypic spectrum and the reproductive outcome associated with biallelic HIPK4 variants and to formally establish the gene-disease relationship for male infertility.

WIDER IMPLICATIONS OF THE FINDINGS

This study raises awareness of the significant genetic heterogeneity of male infertility. The described family highlights that distinct genetic causes may underlie a seemingly similar phenotype. Exome sequencing of families is helpful to efficiently disentangle individual causes among affected family members.

FUNDING

N.N., J.R., H.O., S.L., C.F., and F.T. were supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) within the Clinical Research Unit ‘Male Germ Cells’ (CRU326, project number 329621271). R.T.-W., N.N., J.R., H.O., and F.T. were supported by the Federal Ministry of Research, Technology and Space (BMFTR) as part of the project ReproTrack.MS (grant 01GR2303). S.A.K. was supported by the DFG Clinician Scientist programme CareerS Münster (project number 493624047). A.S.G. was supported by the Medical Faculty Münster via an Innovative Medical Research (IMF) grant (GA-122104).

DISCLOSURES

The authors declare no conflicts of interest.

WHAT DOES THIS MEAN FOR PATIENTS?

The underlying causes of abnormal semen parameters leading to infertility are poorly understood leaving many men without a causal (e.g. genetic) diagnosis. This study, however, finds two different genetic causes for sperm abnormalities that lead to infertility within the same family.

When several siblings have the same medical condition, it is typically expected that they share one genetic cause.

In this study, genetic variants in one known infertility gene (DNAH17) explained infertility in only one brother. In his three other brothers, variants in the gene HIPK4 were likely responsible. The impact of the HIPK4 variant was confirmed in cell culture experiments. Morphological assessment of sperm from each of the brothers revealed differences in line with the distinct causes.

Overall, the study identifies HIPK4 as a new gene in which variants are associated with abnormal sperm morphology. The study also highlights an important message: that even within the same family, the seemingly same condition can be caused by different genes. This has implications for genetic testing and diagnosis of male infertility.

Introduction

Men in infertile couples often exhibit sperm head abnormalities and other morphological defects (teratozoospermia) (Tüttelmann et al., 2018). Yet, more than 70% of men do not receive a causal genetic diagnosis (Oud et al., 2025). While the number of genes associated with sperm flagellar defects has steadily grown during the past years, the genetic basis of sperm head abnormalities is less well understood (Beurois et al., 2020; Arora et al., 2024; Cavarocchi et al., 2025).

Abnormal semen parameters can be explained by disturbances in spermatogenesis, the complex process of male gametogenesis. Specifically, the processes of spermiogenesis, which occur after meiotic differentiation, are of special interest for examining the genetic causes of morphological sperm defects, since it is during these processes that haploid round spermatids drastically transform into elongated spermatids and finally into sperm consisting of a head, midpiece, and tail.

Sperm defects that involve certain phenotypes, such as absent, short, bent, or coiled flagella and/or irregular calibre flagella, are referred to as multiple morphological abnormalities of the sperm flagella (MMAF). Notably, MMAF are typically also associated with severely reduced sperm motility (Touré et al., 2021). Considering MMAF, more than 20 genes have reached a sufficient level of evidence to be associated with this phenotype, having at least a moderate gene-disease relationship (GDR), and are ready for clinical diagnostic analyses (Stallmeyer et al., 2025).

MMAF can be caused by variants in genes encoding axonemal proteins such as subunits of the flagellar outer or inner dynein arms (ODAs/IDAs). One example is the ODA subunit dynein axonemal heavy chain 17 (DNAH17) (Milisav and Affara, 1998). Biallelic pathogenic variants in DNAH17 cause impaired sperm motility (asthenozoospermia) and MMAF due to lack of ODAs and/or disorganization of the axonemal ultrastructure of the sperm flagellum (Whitfield et al., 2019; Song et al., 2023).

Yet, as opposed to flagellar defects, sperm head defects are so far associated with only a handful of genes that are ready to be included in diagnostic gene panels (Stallmeyer et al., 2025). These few genes are associated with specific, monomorphic sperm head phenotypes, such as macrozoospermia (AURKC), globozoospermia (DPY19L2), or acephalic sperm (PMFBP1, SUN5, TSGA10) (Stallmeyer et al., 2025).

Spermiogenesis, including sperm head shaping, requires fine-tuned regulation. Because spermatids are transcriptionally inactive, such regulation is often governed by post-translational regulation such as phosphorylation (Baker, 2016; Li et al., 2019). Here, protein kinases come into play: they use ATP to phosphorylate their targets at specific residues and can, thereby, activate or inactivate protein functions (Röhm et al., 2021). One member, the serine/threonine kinase called homeodomain-interacting protein kinase (HIPK4; Arai et al., 2007; He et al., 2010) is reportedly required for sperm head shaping via cytoskeletal remodelling and male fertility in mice (Crapster et al., 2020; Liu et al., 2022). In humans, previously published HIPK4 variants have been identified in men with azoospermia, either in a heterozygous state (Liu et al., 2022) or in a homozygous state, but have been classified as a variant of uncertain significance (Alhathal et al., 2020). Accordingly, the relevance of HIPK4 in human male infertility remains uncertain.

In this study, we present detailed genetic and phenotypic data of four brothers affected by infertility. Based on the morphological characterization of sperm morphology and in vitro experiments, we disentangled two distinct genetic causes, variants in DNAH17 and HIPK4, and, thereby, now describe a homozygous truncating variant in HIPK4 that explains sperm head defects.

Materials and methods

Subjects

Because of infertility, four brothers (M865, M1344, M1670, and M1611) from a Caucasian family presented at the Centre of Reproductive Medicine and Andrology (CeRA), University Hospital Münster between 2013 and 2018. According to their self-declaration, their parents were not related. Andrological examination included a detailed anamnesis, physical examination, testicular ultrasound, hormonal analysis (FSH, LH, testosterone), and semen analysis. Routine genetic diagnostics ruled out chromosomal aberrations in all of the described men and AZF deletions in those men with sperm counts <5 × 10⁶. Subsequently, the fertile brother (M1688) was recruited for clinical and genetic diagnostics and DNA samples were obtained from the parents for segregation analysis.

Further, the Male Reproductive Genomics (MERGE) cohort was queried for additional men affected by biallelic HIPK4 variants. MERGE currently includes exome/genome sequencing data of n = 3338 men in infertile couples, including 96 with normozoospermia, while most have varying degrees of quantitatively and/or qualitatively impaired spermatogenesis (n = 3243). Most subjects have crypto- or a-zoospermia (n = 577 and 2038, respectively) and the remainder of 627 have oligo-, astheno- and/or terato-zoospermia. All individuals provided written informed consent, and the study was approved by the Ethics Committee of the Medical Faculty Münster and the Ärztekammer Westfalen-Lippe (2010-578-f-S) in accordance with the Helsinki Declaration of 1975.

Semen analysis

Repeated routine semen analyses were performed at the CeRA according to the WHO guidelines valid at the respective time (WHO, 2010, 2021). Importantly, strict criteria were always used when assessing sperm morphology after modified Papanicolaou staining. For visualization of sperm morphology, pictures were taken with a PreciPoint O8 scanning microscope system (Garching, Germany) with a 100× objective and oil.

Exome/genome sequencing

Genomic DNA extracted from peripheral blood leukocytes was used for exome and genome sequencing as described previously (Rotte et al., 2025). For exome sequencing, sample/library preparation and enrichment was done according to the protocol of Twist Bioscience’s Human Core Exome 1.3 plus RefSeq spike-ins or Human Core Exome 2.0 plus Comprehensive kit (Twist Bioscience, South San Francisco, CA, USA) or Agilent’s SureSelect human all exon kits V4, V5, and V6 (Agilent, Santa Clara, CA, USA). Genome samples were prepared using Illumina’s DNA PCR-Free library kit (Illumina, San Diego, CA, USA). For multiplexed sequencing, the libraries were index tagged using appropriate pairs of index primers. Quantity and quality of the libraries were determined with the Thermo Fisher Qubit, the Agilent TapeStation 2200 or 4200, and the Tecan (Männedorf, Switzerland) Infinite plate reader, respectively. Sequencing was performed on Illumina’s NextSeq 500, 550, 2000 or NovaSeq 6000 or X Plus systems using the corresponding reagent kits. Calling of single-nucleotide variants, small indels, and copy number variants (CNVs) was conducted using Broad’s GATK v3.8 or Illumina’s Dragen Bio-IT Platform v3.10, v4.3.6, or 4.3.13 using the reference genome version GRCh37.p13. The identified variants were then annotated using the Ensembl Variant Effect Predictor.

Variant filtering

Variants were filtered for rare (minor allele frequency [MAF] < 0.01, gnomAD database v 2.1.1 (Karczewski et al., 2020), and < 0.01 in our in-house database), coding variants. Only high-impact variants, i.e. frameshift, premature stop codon, loss of start codon, splice acceptor/donor variants, and missense variants with a CADD score (Kircher et al., 2014) ≥ 20, were further assessed. After analysing diagnostic genes associated with qualitative spermatogenic failure (astheno-, terato-zoospermia, and other or mixed phenotypes) (Stallmeyer et al., 2025) for causal sequence variants including CNVs, we performed an exome-wide analysis and focused on the homozygous, (compound) heterozygous, and hemizygous sequence variants shared by the affected brothers. We only considered high-impact variants (see above) in genes expressed in the testis and excluded genes with a broad expression pattern via visual inspection of available RNA expression data (GTEx Consortium, 2020), last accessed in October 2025. Shared heterozygous and hemizygous variants were further inspected if their MAF was < 0.001 and if they were absent in the father and the fertile brother (M1688). Likewise, biallelic variants identified in the fertile brother were excluded. The prioritized variants in HIPK4 and DNAH17 were validated by Sanger sequencing (for primer sequences, see Supplementary Table S1). Subsequently, the MERGE cohort was screened for potentially biallelic variants in HIPK4 using the same filter criteria as described above. Regions of homozygosity were visualized using the AltAF plotter (Radtke et al., 2024).

High-resolution immunofluorescence microscopy analysis

Semen was diluted to 1 million/ml in phosphate-buffered saline (PBS) and spread onto glass slides for immunofluorescence (IF) staining as described previously in Aprea et al. (2021). In cases of severe oligozoospermia, where dilution to 1 million/ml was not possible, samples were applied undiluted onto glass slides. After storage at −80°C, cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100 in PBS and blocked 2–4 h in 5% bovine serum albumin in 0.1% Triton X-100 in PBS. To stain the flagellum, sperm were incubated overnight (4 °C) with a monoclonal mouse antibody against acetylated alpha-tubulin followed by incubation with the secondary antibody for 1 h at room temperature in the dark (for antibody information, see Supplementary Table S2). Nuclei were stained with Hoechst 33342 (1:1000, Sigma, St. Louis, MO, USA). Slides were imaged with a Laser Scanning Microscope (LSM 880, Carl Zeiss Microscopy GmbH, Jena, Germany) and processed, and images were exported using the ZEISS ZEN Imaging Software 2012 (Carl Zeiss Microscopy GmbH). Figure panels were created with OMERO (Allan et al., 2012).

Transmission electron microscopy

Transmission electron microscopy (TEM) of sperm was performed as previously described (Aprea et al., 2023). Briefly, sperm were fixed in 2.5% glutaraldehyde overnight at 4 °C. After pelleting and washing with tap water, samples were incubated at room temperature for 1.5–2 h in 1% osmium tetroxide. Following dehydration in an ethanol series, samples were first transferred to 1,2-epoxypropan and then incubated in a 1,2-epoxypropan-epon mixture (1:2) at 4 °C overnight. Finally, samples were embedded in epon and dried at 55 °C. Ultrathin sections (80 nm) of samples were placed on support grids and contrasted with 8% uranyl acetate. The samples were analyzed with a transmission electron microscope Philips CM10 (Philips, Amsterdam, Netherlands) and TEM images were acquired with a Quemesa camera and the iTEM SIS image acquisition software (both from Olympus Soft Imaging Solutions (now Evident, Hamburg, Germany)).

Gene and protein expression and characterization

HIPK4 expression in human testicular tissues was evaluated in previously published bulk RNA sequencing data (Siebert-Kuss et al., 2022; European Genome-Phenome Archive: EGAD00001008652). Expression levels of HIPK4 at single-cell level were evaluated in human testis with normal spermatogenesis also using published datasets (Di Persio et al., 2021; NIH Gene Expression Omnibus (GEO): GSE153947). Expression levels were displayed as bubble plots using the DotPlot function and as feature plots using the function FeaturePlot in the Seurat R package version 5.3.0 (Hao et al., 2024).

The genomic sequence of HIPK4 (ENST00000291823.3) was accessed via Ensembl (version 115) (Dyer et al., 2025). Amino acid (aa) alignments for different species and functional annotations (domains, features) of human HIPK4 or DNAH17 were retrieved from UniProt (UniProt Consortium, 2023; see Supplementary Table S3 for the accession numbers). The common ATP-binding motif of kinase domains was added to the alignment (Röhm et al., 2021). The HIPK4 AlphaFold model (Jumper et al., 2021) was used for 3D visualization of relevant residues of HIPK4 using ChimeraX (version 1.9) (Goddard et al., 2018).

Cloning and mutagenesis

HIPK4 exon 1 was amplified from genomic DNA derived from one patient homozygous for HIPK4 c.1A>G and control DNA using PrimeSTAR Max polymerase (Takara Bio, Shiga, Japan), cloned into the pcDNA3.1(+) vector (Thermo Fisher, Waltham, MA, USA) and modified by adding a C-terminal HA-tag and a stop codon. Human adult testis RNA (BioCat, Heidelberg, Germany) was used to clone the complete HIPK4 cDNA. Reverse transcription was performed with the ProtoScript® II First Strand cDNA Synthesis Kit (New England Biolabs, Ipswich, MA, USA). The QuikChange XL Site-Directed Mutagenesis Kit (Agilent) was used to introduce the variant c.1A>G into the latter pcDNA3.1(+) construct. All primer sequences can be found in Supplementary Table S1. The correct insertion into the vector was confirmed by sequencing.

Transfection of HEK293T cells

Human Embryonic Kidney cells (HEK293T, Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures, Germany) were transfected with 2 µg of the respective construct using the K2 Transfection System (Biontex, München/Laim, Germany) followed by medium exchange (Dulbecco’s Modified Eagle Medium containing 10% foetal bovine serum, Thermo Fisher) after 6 h. At 24 h (short construct) after transfection, cells were detached with ice-cold Dulbecco’s PBS (Thermo Fisher), centrifugated (5 min, 4 °C, 300g), and resuspended with lysis buffer (100 mM NaCl, 20 mM imidazole, 1% Triton X-100, 2 mM CaCl₂) containing a protease inhibitor cocktail (cOmplete™, Merck, Darmstadt, Germany). After incubation on ice for 15 min and centrifugation (15 min, 4 °C, 15 000g), the supernatant was stored at −20°C or directly used for western blot.

Western blot

Cell lysates were diluted with Laemmli buffer (Bio-Rad, Hercules, CA, USA) with dithiothreitol (DTT, Merck), denatured at 95 °C for 10 min and separated on a 12% acrylamide gel (short construct) or a 4–15% gradient gel (Mini-PROTEAN TGX Stain-free gels, Bio-Rad) (full protein). After blotting onto a PVDF membrane (Trans-Blot Turbo Transfer Pack, Bio-Rad) and blocking with 5% milk powder (AppliChem, Darmstadt, Germany) in TBST, the membrane was incubated with the primary antibody (Supplementary Table S2) at 4 °C overnight. After incubation with HRP-conjugated secondary antibodies (Supplementary Table S2), chemiluminescence was detected with Peroxidase/Luminol system (Clarity enhanced chemiluminescence substrate, Bio-Rad) using a ChemiDoc MP Imaging System (Bio-Rad).

Cycloheximide chase assay

HEK293T cells seeded in 6-well plates were transfected with 2 µg of the respective vector containing full-length HIPK4 cDNA as described above. Forty-eight hours after transfection one well per condition was treated with 2 ml medium containing 100 µg/ml cycloheximide (Carl Roth, Karlsruhe, Germany) and further incubated for 24 h. Lysates of untreated (0 h) and treated cells (24 h) were used for western blots with loading of equal volumes (10 µl). The membranes were cut horizontally to separate anti-HA and anti-GAPDH staining (Supplementary Table S2). Images were captured as described above and intensity of the bands was analyzed using Fiji (ImageJ 1.54f) software (Schindelin et al., 2012). Protein levels were measured as mean grey values within a constantly sized rectangular selection in western blots. The experiment was carried out four times independently (biological repeats) with averaging of technical repeats prior to statistical analysis.

Statistical analyses

GraphPad Prism (version 10.2.3, GraphPad Software, Boston, MA, USA) was used for visualization and statistical analysis of the cycloheximide assay. Data are presented as HIPK4 relative to GAPDH levels (HIPK4-HA/GAPDH). Normal distribution was confirmed via visual inspection of the QQ-plot. The proportion of residual protein after 24 h of incubation with cycloheximide [(HIPK4-HA/GAPDH)_24h/(HIPK4-HA/GAPDH)_0h] was compared between the wild type (WT) and mutant sample using a paired t-test with a significance level of P < 0.05 (one-tailed).

Plots showing expression of HIPK4 in bulk RNA sequencing and single-cell RNA sequencing data were generated with R (version 4.4.3) in the RStudio environment (version 2024.9.1.394, R-Tools Technology Inc., Richmond Hill, ON, Canada).

Variant classification and gene-disease relationship assessment

Variants were classified according to the guidelines of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG-AMP, ACMG used throughout the text) (Richards et al., 2015; Tavtigian et al., 2020) as described in Supplementary Table S4. A preliminary assessment of the gene-disease relationship was performed for HIPK4 and male infertility according to ClinGen (Strande et al., 2017).

Results

Biallelic variants in DNAH17 and HIPK4 represent distinct causes of infertility within the family

To evaluate and compare phenotypes and genotypes among four brothers with infertility (M865, M1344, M1670, M1611), their parents, and an unaffected brother (M1688, Fig. 1), we performed an exome-based family analysis. Repeated semen analyses showed varying degrees of quantitatively and qualitatively impaired spermatogenesis in the affected brothers resulting in different combinations of oligo-, astheno-, and terato-zoospermia (Human Phenotype Ontology [HPO] terms: HP:0000798, HP:0012207, HP:0012864). The three younger brothers (M1344, M1670, M1611) constantly displayed teratozoospermia with 100% sperm head defects. In contrast, the older brother M865 had 1–4% of sperm with normal morphology in more than 10 semen analyses (Fig. 1, Supplementary File S1). Clinical history and examination, testicular ultrasound, as well as hormonal analysis (Supplementary File S1) were unremarkable except for left-sided ligation of varicocele in M865. M865 and M1344 underwent successful ICSI and fathered three children and one child, respectively, whereas their normozoospermic brother (M1688) fathered three children without intervention. The percentage of regions of homozygosity throughout the exomes (Supplementary Fig. S1) indicated a distant familial relation of the parents.

Figure 1.

Pedigree, segregation of HIPK4 and DNAH17 variants, and semen phenotypes. (A) Filled symbols depict men who are infertile. The coloured dots represent a specific variant in the monoallelic state as indicated by the Sanger traces of the parents. For instance, both parents are heterozygous for c.1A>G in HIPK4 (blue circle), leading to heterozygosity (one blue circle) or homozygosity (two blue circles) in three of their sons. In DNAH17, one paternal (c.7752+2T>A p.?) and two maternal (c.1076_1077dup p.(Lys360*) and c.5932G>A p.(Glu1978Lys)) variants are highlighted in red, blue-green, and yellow, respectively. (B) The results of the first two complete semen analyses performed at the CeRA according to the WHO guidelines are shown underneath the respective brother. M1344 had a morphology assessment only once (first and third analyses shown). Due to high variability of human semen parameters, all available semen analyses are shown in Supplementary File S1. Reference values according to WHO (2021): concentration ≥16 million/ml; total sperm count ≥39 million; motility: a + b ≥30%, a + b + c ≥42%; morphology: ≥4% normal. Values outside the reference ranges are highlighted in bold. y, years; d, days; Mill, million; ml, millilitre; na, not available.

The exome-based analysis of 32 valid disease genes for qualitative spermatogenic failure and male infertility revealed no shared genetic cause among all four affected men. However, three variants in the autosomal recessive MMAF gene DNAH17 (NM_173628.4) were identified in different combinations (Fig. 1, Supplementary Fig. S2). M865 is compound heterozygous for two loss-of-function (LoF) variants in DNAH17. The variant c.1076_1077dup leads to a premature stop codon in exon 8 of 81 (p.(Lys360*)), while the second variant, c.7752+2T>A p.?, disrupts the canonical splice donor site after exon 49 (SpliceAI score for donor loss: 1.0; Jaganathan et al., 2019). Both variants are ultra-rare (gnomAD: 0.0001773 and 0.00003706) and were listed in ClinVar as pathogenic (Variation ID: 3388882) and likely pathogenic (Variation ID: 2582766), respectively. M1344 and M1611 are also heterozygous for c.7752+2T>A p.? but in a compound heterozygous state with the hitherto undescribed DNAH17 missense variant c.5932G>A p.(Glu1978Lys). The residue is located within the AAA1 ATPase domain, is conserved among mammals (Supplementary Fig. S3), and has a CADD score of 27.6 (version GRCh37-v1.6). Still, the impact of this amino acid substitution on the protein function remains unclear. In M1670, however, c.5932G>A p.(Glu1978Lys) was the only DNAH17 variant, identified in a heterozygous state. This means that irrespective of the missense variant’s significance, DNAH17 variants cannot explain the infertility of all affected brothers.

Thus, we performed an exome-wide analysis for variants that are shared among all four affected brothers or among the three without biallelic LoF variants in DNAH17 (Supplementary Fig. S2). This approach also revealed no shared homozygous, (compound) heterozygous, or hemizygous variants among all four affected brothers (Supplementary Table S5). But, the brothers without biallelic LoF variants in DNAH17 (M1344, M1670, M1611) are homozygous for a variant identified in HIPK4 (NM_144685.5). The identified HIPK4 variant c.1A>G disrupts the canonical start codon, which may lead to the use of an alternative downstream start codon or even loss of protein translation. In the fertile brother (M1688), the variant is present in a heterozygous state, fitting with the autosomal recessive inheritance mode for HIPK4-associated infertility observed in male mice (Crapster et al., 2020; Liu et al., 2022).

Based on these findings, we first analyzed the flagellar ultrastructure associated with the identified variants in DNAH17. Next, we investigated the relevance of the HIPK4 variant using RNA sequencing, in silico analyses, and in vitro analyses, and we characterized the sperm morphology associated with HIPK4 deficiency.

DNAH17 variants cause an ODA defect and MMAF in M865

We performed TEM on the sperm obtained from M1688 (fertile) and M865, M1344, and M1670 (infertile) to investigate possible ultrastructural alterations associated with the variants identified in DNAH17. Axonemal cross sections of all three brothers with the DNAH17 missense variant in a heterozygous (M1688 and M1670) or a compound heterozygous state (M1344) exhibited ODAs in flagellar cross sections (Fig. 2A). In contrast, ODAs were consistently lacking in flagellar cross sections of M865, who is heterozygous for two LoF variants in DNAH17 (Fig. 2B). Additionally, we frequently observed axonemal disruption, such as missing or supernumerary microtubule doublets in the sperm of M865. IF staining of the sperm nuclei and flagella revealed short, bent, or coiled flagella in M865 compared to those of the control and of the fertile brother (Fig. 2C). The observed loss of ODAs and axonemal disruption in sperm of M865 further supports the pathogenicity of the DNAH17 LoF variants.

Figure 2.

Transmission electron microscopy and immunofluorescence staining confirming morphological and ultrastructural abnormalities of the sperm flagella in M865 with causal DNAH17 variants. (A) Left: Schematic representation of regular sperm flagellar cross sections at the proximal and the distal part of the principal piece as seen in transmission electron microscopy (TEM). The axoneme consists of nine microtubule doublets and a central pair of microtubules. Radial spokes point to the centre, while outer dynein arms (ODAs, red) and inner dynein arms (not shown) enable flagellar beating. In the proximal principal piece, outer dense fibres (ODFs) and the fibrous sheath (FS) surround the axoneme. Along the principal piece, the ODFs become smaller and fewer until they finally disappear in the distal part. In the midpiece (not shown schematically), the mitochondrial sheath (MIT) replaces the fibrous sheath. Right: TEM images of axonemal cross sections of sperm from M1688, M1344, and M1670. Arrows point to examples of ODAs. Note that the upper panel demonstrates flagellar cross sections in sperm of M1688 and M1670, which derive from the midpiece as indicated by surrounding MIT, while both pictures of the sperm of M1344 represent the principal piece. Scale bar, 200 nm. (B) Left: Schematic representation of a sperm with axonemal disruption associated with DNAH17 deficiency. Right: TEM images of sperm from M865 clearly show an absence of ODAs as well as axonemal disruption, including missing or supernumerary and misplaced microtubule doublets. (C) Immunofluorescence staining of the nucleus (Hoechst, blue) and the flagellum (acetylated tubulin, green) of control sperm and sperm from M1688 and M865. Scale bar, 10 µm. Schematic representations in (A) and (B) were created in BioRender. Friedrich, C. (2026) (https://BioRender.com/frrmysv). ODFs, outer dense fibres; ODA, outer dynein arms; FS, fibrous sheath; MIT, mitochondrial sheath, acet., acetylated.

HIPK4 is expressed during later stages of spermatogenesis

Data from bulk RNA sequencing of testicular tissue with spermatogenic arrest at different stages indicated that HIPK4 is neither transcribed in somatic cells nor in early stages of human spermatogenesis, but it is almost exclusively expressed by samples containing germ cells from the spermatid stage onward (Fig. 3A). Higher granularity was achieved by single-cell RNA sequencing analysis of testicular tissues with complete spermatogenesis. This revealed that HIPK4 expression starts in spermatocytes at the end of meiosis 1 (Fig. 3B and C). Based on its expression profile, HIPK4 is a suitable candidate gene for male infertility because it may cause disturbances in later stages of spermatogenesis.

Figure 3.

HIPK4 expression in human testicular tissue. (A) HIPK4 expression in bulk RNAseq data (Siebert-Kuss et al., 2022) from samples without germ cells (Sertoli cell-only), samples showing maturation arrest at the spermatogonial, spermatocyte, or spermatid stage, and samples showing full spermatogenesis (Normal). Expression is shown as normalized counts. (B) Dot plot of HIPK4 expression data derived from single-cell RNA sequencing (Di Persio et al., 2021) in normal testicular tissue. The size of the circle represents the percentage of cells expressing HIPK4, while the intensity of the colour shows the average HIPK4 expression in each cell type. (C) Feature plot of testicular cells expressing HIPK4 based on the single-cell RNA sequencing data shown in (B). The level of expression is represented by the intensity of the colour. SCO, Sertoli cell-only; Spg, spermatogonia; Spc, spermatocytes; Spd, spermatids; LCs, Leydig cells; SCs, Sertoli cells; PTMs, peritubular myoid cells; ECs, endothelial cells; Diff-Spg, differentiating spermatogonia; Undiff-Spg, undifferentiated spermatogonia; UMAP: Uniform Manifold Approximation and Projection.

HIPK4 variant leads to a less stable, truncated protein lacking functionally relevant residues

The variant HIPK4 c.1A>G affects the canonical start codon in transcript NM_144685.5. However, further downstream ATG-codons are located either in-frame or out-of-frame within the first exon of HIPK4 (Fig. 4A). To analyze the translation initiation in vitro, we transfected HEK293T cells with a shortened construct containing WT or mutant HIPK4 exon 1 (Fig. 4B). Given the difference in protein sizes, in the mutant sample protein translation must have initiated at the ATG codon, corresponding to p. Met36 in the WT. Taken together, these data demonstrated that HIPK4 c.1A>G disrupts the canonical start codon and promotes aberrant translation initiation in vitro, leading to p.(Met1_Glu35del).

Figure 4.

HIPK4 c.1A>G variant validation. (A) Left: HIPK4 gene structure with exons in grey, intronic sequence in black, and 5′- and 3′-untranslated regions (UTR) in light grey. The position of additional in-frame (white) or out-of-frame ATG triplets (black, underlined) within the first exon is depicted in a close-up. Right: Linear protein model of HIPK4 containing a kinase domain (box). White pins indicate the position of methionine residues encoded by in-frame AUG. The first 35 amino acids upstream of the second methionine are highlighted in red. (B) Heterologous expression of a construct containing HIPK4 exon 1 (grey) flanked by its endogenous 5′-UTR (light grey) and a C-terminal HA-tag (black) in HEK293T cells to test protein translation of the wild type (WT) versus the mutant (Mut) construct. A representative western-blot image (n = 4) of WT, Mut, and untransfected cells (−) is shown in relation to a protein marker (M). The observed sizes of the protein fragments match the expected molecular weight for the WT (18.2 kDa) and for the mutant (14.3 kDa) sample if translation initiates at the first in-frame AUG codon (p.Met36). (C) Multiple-sequence amino acid alignment of HIPK4 orthologs from different mammalian species. Less than 65% of the sequence of human HIPK4 (ENSG00000160396) matched the 1:1 orthologue of all 14 fish and 20 bird/reptile species listed by Ensembl and vice versa. Drosophila melanogaster and zebrafish do not have a 1:1 ortholog of human HIPK4. Sequences were retrieved from Uniprot. Stars indicate conserved residues. The phosphate-binding loop (p.17–25, green) contains a glycine-rich sequence motif present in all typical protein kinases (Röhm et al., 2021), with φ being a hydrophobic amino acid like phenylalanine in the case of HIPK4. (D) 3D AlphaFold model of HIPK4, adapted in ChimeraX. The N-terminal truncation (red) also encompasses p.17–25 (black) directly interacting with the phosphate of ATP (Uniprot Q8NE63). The functionally indispensable residues p. Lys40 and p. Asp136 (He et al., 2010) are highlighted in black and blue, respectively. (E) Cycloheximide chase experiment of WT and Mut HIPK4 in transiently transfected HEK293T cells. Cells were either untreated or incubated with cycloheximide for 24 h. GAPDH served as internal control of cellular protein expression. The experiment was carried out four times independently. The graph displays normalized HIPK4 levels (HIPK4-HA/GAPDH) after 24 h relative to its amount at 0 h. The proportion of residual protein amounts after 24 h is significantly lower in the mutant sample compared to WT (paired t-test with significance level of P < 0.05). ** P < 0.01. UTR, untranslated region; kDa, kilodalton; M, marker; WT, wild type; Mut, mutant; −, untransfected cells.

The alternative translation initiation site used in the mutant (p.Met36 in the WT) is located within the kinase domain of HIPK4 (Fig. 4A). This truncation-affected domain is conserved between mammalian species, but not in lower eukaryotes, and shares similarities with the catalytic domains of other protein kinases (Fig. 4C). Here, residues p.17–25 form one part of the ATP-binding pocket that is complemented by p. Lys40 (Fig. 4D). To test the stability of the mutant protein, HEK293T cells were transfected with full-length WT and mutant HIPK4 cDNA followed by translation inhibition with cycloheximide to analyze protein degradation within a 24 h period (Fig. 4E; Supplementary Fig. S4). Protein amounts of both the WT and the mutant decreased over time, but the relative amount of HIPK4 present at the end of the experiment (24 h/0 h) was significantly lower in the mutant than in the WT sample (P = 0.008). These data indicate that the N-terminal truncation results in impaired HIPK4 stability in vitro.

Variable sperm head abnormalities are associated with loss of HIPK4 function

Modified Papanicolaou staining of the fertile brother’s (M1688) ejaculate revealed normally shaped sperm, consistent with the diagnostic semen analysis (Fig. 5). In contrast, the sperm of M1670 showed variable morphological abnormalities like tapered, round, small, and amorphous sperm heads, heads with small or completely absent acrosome, as well as headless sperm tails (acephalic sperm or ‘pinheads’). Further, the staining showed abnormalities in the midpiece and flagellum, like bent or enlarged midpieces and sperm tails coiled around the head. IF staining of the sperm from all three brothers homozygous for the HIPK4 variant (M1344, M1670, M1611) was performed to combine the sperm shapes in bright-field microscopy with the signal of its nuclei (Hoechst 33342) and flagella (acetylated alpha-tubulin). Corresponding with the Papanicolaou staining of M1670, the additional staining showed variable sperm head shapes and sizes as well as short and coiled flagella (Fig. 5C).

Figure 5.

The homozygous HIPK4 variant is associated with variable sperm head abnormalities. (A) Overview of sperm smears stained with modified Papanicolaou from the fertile brother (M1688) and one of the brothers affected by infertility (M1670), who are, respectively, heterozygous and homozygous for c.1A>G p.(Met1_Glu35del) in HIPK4. Scale bar, 30 µm. (B) Close-up of one morphologically normal sperm from M1688 and examples of eight different morphological abnormalities of sperm heads, midpieces, and flagella in the semen sample of M1670. These are, from left to right: tapered, round, amorphous sperm head, bent midpiece, amorphous head and enlarged midpiece, small and round head, headless sperm flagella, and short flagella coiled around the head. Scale bar, 10 µm. (C) Immunofluorescence (IF) staining of sperm from all three brothers homozygous for c.1A>G p.(Met1_Glu35del) in HIPK4 (M1344, M1670, M1611) compared to the fertile brother (M1688) who is heterozygous. The nucleus is shown in blue (Hoechst), the flagellum in green (acetylated tubulin). Scale bar, 10 µm.

Gene and variant assessment according to clinical guidelines

DNAH17 has already reached a definitive level of evidence for its gene-disease relationship with asthenoteratozoospermia/MMAF (Stallmeyer et al., 2025). For HIPK4, we collected and scored available experimental and genetic evidence according to ClinGen. To establish the gene-disease relationship, it is mandatory to identify further independent cases with a corresponding phenotype. So far, no other patient with biallelic HIPK4 variants and teratozoospermia has been reported; a published homozygous HIPK4 missense variant identified in a man with azoospermia was not considered due to uncertain significance and a different phenotype (Alhathal et al., 2020; Supplementary Tables S6 and S7). By analysing exome/genome data of men with infertility and different semen phenotypes, including crypto-/azoospermia as well as oligo-/astheno-/terato-zoospermia in the MERGE cohort, we did not identify another man with biallelic high-impact variants. Therefore, the gene-disease relationship cannot be formally established yet despite a preliminary score of seven points, which corresponds to a moderate level of evidence (Supplementary Table S6). Thus, HIPK4 remains a promising candidate gene for human teratozoospermia with convincing experimental but so far sparse genetic evidence. According to the ACMG guidelines, both DNAH17 LoF variants (c.1076_1077dup p.(Lys360*) and c.7752+2T>A p.?) are classified as pathogenic, and the DNAH17 missense variant (c.5932G>A p.(Glu1978Lys)) was classified as a variant of uncertain significance (VUS). Assuming a valid gene-disease relationship for HIPK4, the variant c.1A>G p.(1Met_35Gludel) in HIPK4 was classified as likely pathogenic (Supplementary Table S7).

Discussion

Within the same family, we discovered biallelic variants in DNAH17 and HIPK4 likely causing qualitatively impaired spermatogenesis in four brothers. While the infertility of M865 is caused by compound heterozygous LoF variants in DNAH17 leading to an ODA defect and MMAF, we present data that the homozygous HIPK4 variant c.1A>G p.(1Met_35Gludel) explains the infertility of his three brothers. Via heterologous expression in HEK293T cells, we demonstrate that the mutant protein is truncated, lacking a part of its kinase domain, and is less stable. Matching the expression of HIPK4 in later stages of spermatogenesis, all three affected brothers presented with teratozoospermia and sperm head defects most frequently accompanied by oligo- and astheno-zoospermia.

The occurrence of two distinct genetic causes leading to an apparently similar condition might be surprising. But given the already considerable heterogeneity of male infertility regarding validated genes (Stallmeyer et al., 2025) and, moreover, the sheer number of genes expressed in the testis (Uhlén et al., 2015), this is not likely a unique phenomenon. Further, consanguinity of the parents increases the probability that (likely) pathogenic variants are homozygous. Thus, exome analysis of affected and unaffected family members, combined with detailed phenotypic characterization, helps to confirm or reject the hypothesis of a shared genetic basis and helps doctors to provide better counselling on the recurrence risk and treatment options.

The phenotype of M865 aligns with previous reports of DNAH17 LoF or missense variants causing asthenozoospermia and MMAF (see Song et al., 2023 for a review). Loss of DNAH17 function in M865 resulted in a consistent absence of ODAs with largely preserved 9 + 2 axonemal organization, while axonemal disruption represents an additional, variable feature. The severely reduced progressive sperm motility observed in M865 and in previously reported patients (Song et al., 2023) is consistent with the functional consequences of ODA loss described by Whitfield et al. (2019).

Concerning HIPK4, human variants have been previously published in the context of azoospermia (Alhathal et al., 2020; Liu et al., 2022). However, the semen analysis of all three men homozygous for the HIPK4 variant presented here revealed teratozoospermia, which aligns with the published mouse models (Crapster et al., 2020; Liu et al., 2022). Similar to Hipk4^−/− mice, unbiased semen analyses that were performed prior to genetic testing showed 100% sperm head defects in all three brothers affected by the homozygous start-loss variant. Morphological sperm analyses revealed variable head abnormalities instead of monomorphic teratozoospermia. Although we cannot rule out a phenotypic spectrum ranging from severe teratozoospermia (with or without oligo- and astheno-zoospermia) to azoospermia, our data indicate that loss of HIPK4 function is compatible with completion of spermatogenesis in humans and, extrapolating from three affected brothers, suggest that HIPK4 deficiency does not result in a phenotype of azoospermia.

Importantly, we clarified the inheritance mode as autosomal recessive. Unlike Liu et al. (2022), presenting solely heterozygous HIPK4 variants, all three affected men in this family are homozygous for c.1A>G p.(1Met_35Gludel) in HIPK4. As both the father and the fertile brother are heterozygous for the same variant, we provide evidence that HIPK4 is a recessive gene associated with male infertility. This corresponds with preserved fertility in Hipk4^+/− mice producing comparable litter sizes as WT males (Crapster et al., 2020). Taken together, these findings argue against a causal association between the heterozygous variants published by Liu et al. (2022) and the respective patients’ azoospermia. Alhathal et al. (2020), however, identified a homozygous missense variant in HIPK4 (NM_144685.3: c.935C>T p.(Ala312Val)), but no functional validation was performed, leaving the variant’s significance uncertain.

In contrast, we present a truncating variant in HIPK4 (c.1A>G p.(1Met_35Gludel)). The mutant HIPK4 protein lacks one part of the ATP-binding site (p.17–25), a hydrophobic pocket that forms at the interface of two lobes (Röhm et al., 2021; Arter et al., 2022). Further, only four amino acids remain prior to the lysine residue (p.40), which contributes to ATP binding and is essential for kinase function (He et al., 2010). Thus, such an N-terminally truncated protein should severely affect if not completely abolish kinase function via impaired ATP binding. This remains speculative because no assay measuring the kinase activity of the actual mutant was performed, which is a limitation of this study. While we cannot rule out a residual function of the truncated HIPK4 protein, our cycloheximide assay did indicate reduced protein stability compared to the WT protein in vitro, suggesting insufficient function. Although we could not perform a functional analysis of patient-derived cells because of HIPK4 expression in the testis, use of HEK293T cells was suitable to analyze the variant’s effect at the protein level, assuming that translation initiation is conserved between eukaryotic cells (Kozak, 1999).

Serine/threonine kinase 33 (STK33) is another testis-expressed protein kinase reported to be required for male fertility in mice and men. Ma et al. (2021) published a homozygous frameshift variant in STK33 identified in four infertile brothers with asthenozoospermia and MMAF, which aligned with the murine phenotype (Martins et al., 2018). Based on the fact that transcription ceases in elongating spermatids, post-transcriptional and post-translational modifications such as phosphorylation become even more important for regulating the differentiation and maturation processes (Ogurtsov et al., 2008; Baker, 2016). Given the high number of testis-specific protein kinases (Ogurtsov et al., 2008), it is likely that additional kinases beyond HIPK4 and STK33 might be related to male infertility.

Among the three brothers homozygous for the truncating HIPK4 variant (M1344, M1670, M1611), semen parameters varied regarding sperm count and motility. A similar intrafamilial variability of semen parameters was also observed in three infertile brothers homozygous for the same frameshift variant in STK33 (Ma et al., 2021). In such cases, the semen phenotype might be modified by the individual genetic background as well as by environmental factors. As the clinical significance of the DNAH17 missense variant remains unclear, we cannot rule out that the identified DNAH17 variants also contribute to the semen phenotypes of M1344 and M1611. Yet, the fact that sperm motility reached normal values at least once in M1611 and flagellar ultrastructure differed from M865 suggest that the DNAH17 missense variant was not significant in this regard.

HIPK4 is primarily expressed in the testis, to a lesser extent in the brain, and at low levels in other tissues (Uhlén et al., 2015; Karlsson et al., 2021). Both the clinical data obtained from our patients and the Hipk4^−/− mouse model do not suggest any clinical consequences apart from male infertility. Likewise, additional phenotypes were neither observed in patients with a homozygous STK33 LoF variant nor in Stk33^−/− mice despite a broader expression profile than expected for a gene causing isolated infertility (Martins et al., 2018; Ma et al., 2021). As HIPK4 is also highly expressed in human oocytes, its role in female (in)fertility is yet to be discovered. Female Hipk4^−/− mice, however, have been found to display normal fertility (Crapster et al., 2020).

Understanding genetic causes of male infertility also enables target identification for contraception. Ku et al. (2024) recently developed a selective STK33 inhibitor as a male contraceptive. In mice, inhibiting Stk33 reversibly affected sperm motility and in higher doses also caused teratozoospermia, reproducing the effect of complete knockout. Likewise, HIPK4 has been proposed as a potential target for contraception (Crapster et al., 2020).

Concerning treatment options, both human and murine data indicate that assisted reproduction can be successful in men with biallelic HIPK4 variants. However, based on reduced oocyte binding of HIPK4-deficient sperm in mice (Crapster et al., 2020), it might be advisable to perform ICSI in men with HIPK4-related infertility.

Based on the genetic evidence in this study and published experimental data, HIPK4 represents a promising candidate gene for male infertility. Alhathal et al. (2020) have already evaluated the gene-disease relationship for HIPK4 and male infertility according to ClinGen. Despite sparse genetic data consisting of a single homozygous missense variant identified in an azoospermic patient and the gene being classified as having uncertain significance, they reported a moderate evidence level. According to Strande et al. (2017), however, occurrence of a single homozygous variant of uncertain significance should be regarded as providing only limited evidence. This is the first report of HIPK4 being associated with human teratozoospermia (regularly accompanied by oligo- or astheno-zoospermia), which diverges from the phenotype described by Alhathal et al. (2020) and needs further validation in future studies.

Conclusion

Based on the results in this study, we propose HIPK4 as candidate gene for human male infertility due to sperm head defects, typically also associated with impaired sperm motility and reduced sperm counts (oligoasthenoteratozoospermia). Importantly, HIPK4 deficiency is compatible with successful medically assisted reproduction via ICSI. Further replication is required to firmly establish the gene-disease relationship and to clarify the phenotypic spectrum. This study also highlights that apparently similar phenotypes of male infertility within a family might have distinct genetic causes.

Supplementary data

Supplementary data are available at Human Reproduction Open online.

Data availability

All data underlying this article not included already will be shared on reasonable request to the corresponding author with the exception of the exome sequencing data because of the probands’ privacy protection of genetic data. All genetic variants in DNAH17 and HIPK4 have been published in ClinVar (SCV007518890, SCV007518891, SCV007518892, and SCV007518893).

Authors’ roles

Study conceptualization: SAK, CF, FT. Data curation: SAK, CR, CK, SL, CF, BS, NN, SDP, SK, HO. Funding acquisition: SAK, ASG, RT-W, NN, JR, HO, SL, CF, FT. Investigation: SAK, CR, IA, RT-W, SDP, NN, SL. Visualization: SAK, CR, SL, ASG. Writing of original draft: SAK, CF. Review and editing: SAK, CR, IA, CK, ASG, JW, RT-W, SDP, NN, JR, HO, SL, SK, BS, CF, FT. All authors revised and approved the final version of the manuscript.

Funding

N.N., J.R., H.O., S.L., C.F., and F.T. were supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) within the Clinical Research Unit ‘Male Germ Cells’ (CRU326, project number 329621271). R.T.W., N.N, J.R., H.O., and F.T. were supported by the Federal Ministry of Research, Technology and Space (BMFTR) as part of the project ReproTrack.MS (grant 01GR2303). S.A.K. was supported by the DFG Clinician Scientist Programme CareerS Münster (project number 493624047). A.S.G. was supported by the Medical Faculty Münster via an Innovative Medical Research (IMF) grant (GA-122104).

Disclosures

The authors declare no conflicts of interest.