Infertility is a growing health concern involving over 50–70 million childbearing age couples worldwide, with the male factor contributing in approximately 50% of cases (Fainberg and Kashanian, 2019). Non-obstructive azoospermia (NOA) and severe oligozoospermia (SO) due to spermatogenic failure (SPGF) represent the most severe phenotypes of male factor infertility. However, the causes of such conditions are poorly understood and the aetiology of most affected men is usually defined as being idiopathic (Cannarella et al., 2019; Agarwal et al., 2021). Recent advances clearly point to idiopathic male infertility as a complex trait, in which the combined effect of polymorphic risk variants (causing subtle changes in gene expression) may increase the susceptibility of an individual to suffer from this disorder (Singh and Jaiswal, 2011; Cervan-Martin et al., 2020b; Salas-Huetos and Aston, 2021).

In this regard, common variants in the human genome may increase the susceptibility to develop male infertility by altering key events during spermatogenesis, in which DNA integrity is crucial (Neto et al., 2016). Having such in mind, it is important to stress that alterations in the DNA structure frequently occur in meiosis prophase I during the transition from primary to secondary spermatocytes, due to the generation of double-strand breaks (DSBs) for homologous chromosome pairing and crossing over (CO) (Cannarella et al., 2020). Consequently, DNA repair mechanisms are essential to produce fit mature sperm cells. In fact, the assessment of DNA integrity is a relevant marker to determine the sperm quality in assisted reproductive techniques (ART) (Leduc et al., 2008; Gonzalez-Marin et al., 2012). Thus, while the generation of genetic variability through changes in DNA is the basis for evolution, it is strictly controlled at the individual level (Capilla et al., 2016; Alves et al., 2017).

Consistent with the above, the spermatogenic process shows a very complex molecular and cellular control, involving over 2,000 genes, among which 900 are solely expressed in the male germline (Chalmel et al., 2012; Cervan-Martin et al., 2020b). The so-called Testis expressed (TEX) gene family plays a crucial role in spermatogenesis and some of its members have been strongly associated with male infertility (Cannarella et al., 2019; Bellil et al., 2021). Specifically, TEX15, which is predominantly expressed in spermatogonial and primary spermatocytes, encodes a 2,789 amino acid protein necessary for meiotic recombination and DSB repair in the germ line (Yang et al., 2008). Additionally, TEX15 also has a significant epigenetic function in spermatogenesis by interacting with several PIWI like RNA-mediated gene silencing proteins (namely PIWIL2, PIWIL4), in order to maintain spermatogonial stem cell integrity through the silencing of transposable elements (Schopp et al., 2020; Yang et al., 2020). Indeed, inactivation of Tex15 expression in mutant male mice causes sterility by meiotic arrest and impaired activity of RAD51 and DMC1 recombinases (Yang et al., 2008; Chen et al., 2016).

In this context, high-penetrance pathogenic variants in TEX15 have been widely linked to human male infertility due to spermatogenic impairment in different studies. For instance, non-sense mutations producing truncated TEX15 proteins have been reported to lead to cryptozoospermia (where spermatozoa are apparently absent in fresh semen samples but are recovered in centrifuged pellets) and SPGF in unrelated families (Okutman et al., 2015; Colombo et al., 2017; Wang et al., 2018; Cannarella et al., 2021; Tian et al., 2021). Moreover, mutations in this gene have been associated with NOA due to Sertoli cell-only (SCO) phenotype (characterised by a complete lack of germ cells in the seminiferous tubules) (Araujo et al., 2020).

On the other hand, case-control genetic association studies have attempted to assess the involvement of TEX15 polymorphic variation, such as single nucleotide polymorphisms (SNPs), in the development of SPGF resulting in NOA or SO. The first approach was conducted in a cohort of European ancestry and identified two TEX15 SNPs putatively associated with SPGF (Aston et al., 2010). However, such associations were not replicated in a subsequent study carried out in an independent South-Eastern European population (Plaseski et al., 2012). Similarly, two novel TEX15 variants have been recently proposed to be involved in the genetic predisposition to SPGF in the Iranian population (Ghadirkhomi et al., 2022). Finally, although TEX15 common variation was initially associated with SPGF risk in Han Chinese (Ruan et al., 2012), another report could not confirm this observation in this ethnicity (Zhang et al., 2015). Therefore, no conclusive results are currently available on the possible role of the common TEX15 genetic variation in SPGF development.

In view of the above, we aimed to clarify the involvement of TEX15 common variation in SPGF predisposition in a European genetic background, by analysing a large and phenotypically well-characterised study cohort.

The genotyping success rate of the three analysed variants was above 98% and none of them showed a significant deviation from HWE either in cases or controls. Additionally, the MAFs in the control group were concordant with those described for both the Iberian subpopulation and the European super population of the 1KGPh3 project (Auton et al., 2015), with no statistically significant differences observed between the population-representative group and the normozoospermic group in either the allele or the genotype frequencies. Furthermore, our study population had an appropriate overall statistical power to identify genetic associations with moderate to high effects, as detailed in Supplementary Table S2.

Understanding the genetic basis of SPGF is a necessary step to improve the clinical management and genetic counselling of infertile patients (Fainberg and Kashanian, 2019; Sharma et al., 2021). In the present study, we conducted a comprehensive evaluation of the possible involvement of TEX15 genetic variation in the predisposition to SPGF, using a large European cohort that ensured a high statistical power.

Our results clearly suggest that TEX15-rs1362912, or any of its tagged SNPs, is directly involved in the pathological mechanisms underlying mild forms of SPGF, i.e., SO, with little or no contribution to extreme patterns, such as NOA. Taking into account the high relevance of TEX15 in the reproductive function and the insight provided by previously published genetic studies on this gene (Aston et al., 2010; Plaseski et al., 2012; Ruan et al., 2012; Okutman et al., 2015; Zhang et al., 2015; Colombo et al., 2017; Wang et al., 2018; Araujo et al., 2020; Cannarella et al., 2021; Ghadirkhomi et al., 2022), we hypothesise that NOA could be mostly influenced by high-penetrance damaging mutations in this locus rather than by common SNPs, which would be responsible for increasing SO predisposition instead. Consistent with this idea, only 27 SO-associated point mutations are currently annotated in both the “Male Infertility Genomic Consortium (IMIGC) database” and the “Infertility Disease Database (IDDB)”, whereas NOA has 138 entries, amongst which TEX15 is included (Houston et al., 2021; Wu et al., 2021). Moreover, a higher proportion of NOA cases present Mendelian causes of their infertility when compared to SO (e.g., AZF microdeletions and chromosomal abnormalities are three times more prevalent in NOA than in SO) (Krausz and Riera-Escamilla, 2018).

Although our data are consistent with this hypothesis, the causality of the TEX15 common variation in SO predisposition remains controversial. Previous genetic association reports in this locus have not been fully replicated likely due to differences in the genetic architecture of the different study groups or, more probably, to type I or II errors caused by the limitations in the statistical power of the studied cohorts to detect low or modest effects. In this regard, the TEX15 SNP rs323346 was not associated with SO neither in our study nor in an independent population of European ancestry previously assessed by Aston et al. (2010). Inconsistent results were obtained in two additional studies performed in Chinese, as Ruan et al. (2012) described a genetic association between this SNP and SO that was not replicated by Zhang et al. (2015) in the same ethnicity. Moreover, there are striking differences between the reported rs323346*C allele frequencies for the human populations included in the 1KGPh3 project (e.g., EUR = 0.17; EAS = 0.11; SAS = 0.33; AFR = 0.80, Supplementary Figure S2) (Auton et al., 2015), which might indicate population-specific causal variants within this region (Aston et al., 2010; Ruan et al., 2012; Ghadirkhomi et al., 2022). Therefore, subsequent studies in independent study cohorts should be performed to clarify further the possible existence of different SO functionally relevant SNPs in this gene.

Remarkably, animal models clearly support the role of the TEX15 locus in SO development (Yang et al., 2008; Schopp et al., 2020; Yang et al., 2020). In fact, TEX15-defficient murine models show reproductive system abnormalities, including male infertility, altered spermatogenesis, arrest of male meiosis, decreased male germ cell number, DSB repair anomalies, and reduced testis weight (Yang et al., 2008; Carvalho-Silva et al., 2019). This gene is mostly expressed in spermatogonia and spermatocytes, in which it has a relevant regulatory role in the control of gene expression by interacting with a vast number of transcription factors related to the reproductive function, as reported in the Open Targets Platform (Carvalho-Silva et al., 2019).

According to the results of our in silico functional analyses, amongst the linked polymorphisms of the tagger rs1362912, the non-coding TEX15 variant rs4733201 may be responsible for an important contribution to the final phenotype. This SNP showed strong evidence of functionality, including high scoring by different functional impact prediction methods as well as an overlap with testis-specific active chromatin epigenetic marks. Interestingly, the sequence surrounding this SNP corresponds to a TFBS for different key transcription factors related with spermatogenesis, including EGR1 (Tourtellotte et al., 2000; Man et al., 2014), CTCF (Hernandez-Hernandez et al., 2016; Rivero-Hinojosa et al., 2021), SMC3 (Eijpe et al., 2000; James et al., 2002), YY1 (Wu et al., 2009; Kim et al., 2016), CCNT2 (Teng et al., 2011), E2F6 (Pohlers et al., 2005; Dahlet et al., 2021), and MAX (Maeda et al., 2013), and it is annotated to be transcriptionally active in the testis (Davis et al., 2018). Moreover, this SNP was predicted to modify the activity of E2F1, which has been associated with human SPGF (Jorgez et al., 2015; Rocca et al., 2019).

Similarly, the results of the biological pathway enrichment analysis of the transcription factors with TFBSs influenced by the analysed variants also support their involvement in SPGF, as some of the enriched pathways included “reproductive process”, “pituitary gland development”, “ovarian infertility”, and “male infertility”. To our knowledge, there are no reported studies in which the possible effects of TEX15 on female infertility were evaluated (Yang et al., 2008; Bellil et al., 2021). However, TEX15 is expressed in the female reproductive tract (Supplementary Figure S1), “ovarian infertility” was amongst the most enriched pathways, and there is evidence of the implications in female fertility of some of the transcription factors with binding sites in the genomic positions of some of the SNPs considered in our functional analysis. Specifically, VDR is associated with female infertility (Djurovic et al., 2020), whereas CEBPB has been described as relevant factor for female reproduction for its role in ovarian follicle development (Fan et al., 2009), together with SMAD3 (Tomic et al., 2004; Li et al., 2008) and ESR2 (Khristi et al., 2018; Chakravarthi et al., 2020). Such is consistent with the assumption that the effect of the common variation in the genome on pathological conditions is not limited to altering the expression or function of a single gene or protein, but rather to unbalance complex molecular networks (Cui et al., 2015).

Regarding the possible limitations of this work, it is important note that our exclusion criteria based on genetic abnormalities considered only karyotype alterations and Y chromosome microdeletions. Although both type of alterations account for the majority of known genetic causes of SPGF (Cioppi et al., 2021), screening of high-penetrance point mutations in reported SPGF genes was not performed. As a consequence, some cases of our study cohort could harbour single-gene mutations that may explain their infertility. However, considering the low frequency of the reported monogenic mutations in SPGF (Cioppi et al., 2021), this limitation is not expected to have a relevant impact in the allele frequencies of our study cohort. The possible statistical noise added by such cases would likely cause a subtle increase in the probability of obtaining type II errors (false negative results), but it would hardly affect the consistency of the observed associations.

Overall, the insight gained in this study supports the notion of idiopathic SPGF as a complex trait, where common genetic variation has a determinant role in disease susceptibility and development. In the case of TEX15, it is likely that the selective pressures have prevented the dissemination of high-penetrance deleterious variants leading to NOA in the human populations, considering the crucial role that this gene has in the reproductive success of an individual (Yang et al., 2008; Chen et al., 2016). Under this assumption, it is reasonable to assume that the common TEX15 polymorphisms affecting male infertility were associated with less extreme manifestations of SPGF, such as SO.

Ana Aguiar, (Unidade de Medicina da Reproducao, Hospital de Santa Maria, Centro Hospitalar Lisboa Norte, Lisboa, Portugal); Carlos Calhaz-Jorge, (Unidade de Medicina da Reproducao, Hospital de Santa Maria, Centro Hospitalar Lisboa Norte, Lisboa, Portugal); Joaquim Nunes, (Unidade de Medicina da Reproducao, Hospital de Santa Maria, Centro Hospitalar Lisboa Norte, Lisboa, Portugal); Sandra Sousa (Unidade de Medicina da Reproducao, Hospital de Santa Maria, Centro Hospitalar Lisboa Norte, Lisboa, Portugal), and Sónia Correia (Centro de Medicina Reprodutiva, Maternidade Alfredo da Costa, Centro Hospitalar Lisboa Central, Lisboa, Portugal); Maria Graça Pinto(Centro de Medicina Reprodutiva, Maternidade Alfredo da Costa, Centro Hospitalar Lisboa Central, Lisboa, Portugal). Alberto Pacheco, (IVIRMA Madrid, Spain); Cristina González, (IVIRMA Sevilla, Spain); Susana Gómez, (IVIRMA Lisboa, Portugal); David Amorós, (IVIRMA Barcelona, Spain); Jesús Aguilar, (IVIRMA Vigo, Spain); Fernando Quintana, (IVIRMA Bilbao, Spain).