A framework for high-resolution phenotyping of candidate male infertility mutants: from human to mouse

Abstract

Male infertility is a heterogeneous condition of largely unknown etiology that affects at least 7% of men worldwide. Classical genetic approaches and emerging next generation sequencing studies support genetic variants as a frequent cause of male infertility. Meanwhile, the barriers to transmission of this disease mean that most individual genetic cases will be rare, but because of the large percentage of the genome required for spermatogenesis, the number of distinct causal mutations is potentially large. Identifying bona fide causes of male infertility thus requires advanced filtering techniques to select for high probability candidates, including the ability to test causality in animal models. The mouse remains the gold standard for defining the genotype-phenotype connection in male fertility. Here, we present a best practice guide consisting of a) major points to consider when interpreting next generation sequencing data performed on infertile men, and, b) a systematic strategy to categorize infertility types and how they relate to human male infertility. Phenotyping infertility in mice can involve investigating the function of multiple cell types across the testis and epididymis, as well as sperm function. These findings will feed into the diagnosis and treatment of male infertility as well as male health broadly.

Introduction

Male infertility affects approximately 7% of men in developed nations and represents a major public health concern (Krausz and Riera-Escamilla 2018). While genetic causes are already known to cause a significant percentage of cases (McLachlan and O’Bryan 2010; Oud et al. 2019), many additional genetic causes likely exist. Family-based studies into genetic causes of infertility have historically been challenging, due to the self-limiting nature of the phenotype, the difficulty of enrolling extended families, and, until recently, the inability to screen entire genomes for very rare and patient-specific mutations. In addition, it is known from human and animal studies that at least ~15,000 (human, 77% of all genes) to ~19,000 (mouse) genes are expressed during spermatogenesis, of which at least 2,300 are highly enriched within the meiotic and haploid cell compartment (Schultz et al. 2003; Djureinovic et al. 2014, Soumillon et al. 2013). Thus, there are many opportunities for genetic changes to impact male fertility. At a practical level, these two factors mean that within human populations there are likely to be many genetic causes of male infertility, consistent with clinical data, but each individual cause will be rare. Therefore, the validation of bona fide causes of human male infertility will require the sequencing of potentially tens of thousands of infertile men in order to obtain a close to complete picture of all the genetic variants that will impact human male fertility. The validation of variants using cell and animal models will be especially important wherein specific defects are identified in a small number of patients (potentially an n=1 patient). Within this context, the mouse remains the model of choice to model / test the requirement of genes for human male fertility. Mice easy to house and maintain, highly fecund, and share 99% of their gene content with humans (Rosenthal and Brown 2007). They are also highly comparable in terms of germ cell developmental chronology and function. Thus, the ability to recapitulate human mutations in mice is extremely valuable. While not the focus of the current review, mouse models are also an excellent ‘tool’ to define the precise function of gene products in relation to physiology and cell biology as a whole.

Fortunately, as a consequence of the reduced cost of next generation sequencing techniques and advances in genome editing, the process of mapping disease causing mutations and validation in mice is accelerating. Several large scale infertile male sequencing projects have been initiated and global consortia, including the International Male Infertility Genomics Consortium and the Genetics of Male Infertility Initiative (GEMINI), have been established to provide a strong foundation for claims of causality in human male infertility (www.imigc.org, www.gemini.conradlab.org). Within this review we focus on two important elements of this process: 1) how to interpret and filter human whole exome sequencing data in the context of male infertility; and 2) how to test / validate these genetic variants in mouse models. The aim is to define a pipeline for discovery of novel disease causing variants that can be used broadly across the spectrum of male infertility, regardless of whether infertility occurs in isolation in otherwise healthy men / mice or is part of a syndrome affecting multiple organs.

1. Male fertility

Male fertility relies upon the production of spermatozoa via spermatogenesis in the testes, their subsequent maturation in the epididymides (Figure 1) and functional activation (or capacitation) in the female reproductive tract. This review will briefly touch on these processes in order to familiarize lay readers with the complex processes required for development of a functional spermatozoon and highlight the many steps where infertility can arise.

Figure 1. Sperm production and epididymal maturation.

A. Spermatogenesis. The five major germ cell types within the testis are shown in a tubule cross section, where they develop in intimate association with Sertoli cells. Spermatogonia committed to spermatogenesis undergo mitotic divisions to form spermatocytes, which in turn undergo meiosis to generate round spermatids. Throughout the spermiogenesis period, spermatids become dramatically remodelled from a round to an elongated shape and eventually are released from the Sertoli cell as sperm in a process known as spermiation. On the right hand side, the typical phenotypes expected for knocking out genes with clear roles in each respective germ cell type are shown. Azoo = azoospermia; terato = teratozoospermia; astheno = asthenozoospermia; oligo = oligozoospermoa; OAT = oligoasthenoteratozoospermia. Basement membrane – dark purple; Sertoli cell nucleus – light purple; peritubular myoid cells – green. B. Sperm are generated within the seminiferous tubules of the testis and then travel to the epididymis, wherein they undergo a complex process of epididymal maturation. Sperm enter the epididymis through initial section of the epididymis and travel through the caput and corpus, the major sites of epididymal maturation. They are then concentrated within the cauda, where they are stored in a quiescent state prior to ejaculation. However, they will only achieve full functional capacity after undergoing the process of capacitation whilst in the female reproductive tract.

Spermatogenesis occurs within the seminiferous tubules of the adult testes (Figure 1). It involves three broad phases; mitosis, meiosis and a dramatic morphological remodeling of germ cells via spermiogenesis, followed by their release as spermatozoa into the lumen of the seminiferous tubules within the testes then ultimately the epididymis. Spermatogenesis is supported by two major cell types, the Sertoli and Leydig cell populations, although we note the presence of multiple other cell types including peritubular myoid cells and immune cells. Sertoli cells are considered nurse cells and prime determinants of male infertility. They are a remarkable cell type capable of supporting the synchronous development of up to five cytologically different germ cell types within their depth. Sertoli cell number is set in the pre-pubertal period and their number defines the upper limit of sperm production (reviewed by Sharpe et al. 2003). They are also the prime mediator of hormone action on germ cells development, including follicle stimulating hormone and androgens (reviewed in Dimitriadis et al. 2015). Leydig cells are present within the interstitium between tubules where they are the major source of androgens (in response to luteinizing hormone), which are essential for spermatogenesis (reviewed by Tremblay 2015). Within this review we focus on the consequences of mutations in germ cell expressed genes. It is, however, essential to recognise the extensive amount of cross-talk between germ and somatic cells.

Within the seminiferous tubules, spermatogonial stem cells reside on the basement membrane, where they provide a continual source of germ cells for entry into spermatogenesis (reviewed by Oatley and Brinster 2012). These cells undergo division to form daughter cells committed to spermatogenesis, and others which will replenish the stem cell population. The committed population enters spermatogenesis where they undergo a number of mitotic divisions, known as the transit amplifying step, to form primary spermatocytes. The number of these mitotic divisions is dependent on the species, as is the size of the spermatogonial stem cell population (see Fayomi et al. 2018).

Immediately prior to meiosis, pre-leptotene spermatocytes duplicate their DNA and pass through the blood-testis barrier, behind which they are protected from the immune system by tight junctions existing between Sertoli cells (reviewed by Mruk and Cheng 2015). This sequestration, however, results in additional dependence on Sertoli cells for nutritional support (Smith and Walker 2014). Primary spermatocytes then undergo two meiotic divisions (I and II) to generate secondary spermatocytes and then haploid round spermatids (Kerr et al. 2006). During prophase prior to the first division (meiosis I), homologous chromosomes participate in crossing over and DNA recombination to increase genetic diversity. In turn, spermatocytes have high DNA repair activity to regulate the levels of fragmentation that occur. In the second meiotic division, secondary spermatocytes undergo the final cell divison process of spermatogenesis in order to generate haploid round spermatids.

During the third phase of spermatogenesis, known as spermiogenesis, haploid round spermatids are dramatically remodeled from a spherical shape into highly polarized spermatozoa. This involves the sequential replacement of histones with transition proteins then protamines to facilitate super-condensation of the sperm DNA; the formation of the acrosome, which is necessary for digestion of the oocyte zona pellucida; and formation of the sperm tail to provide motility (reviewed by O’Donnell 2014). Finally, spermatids are released into the lumen of the seminiferous tubules through of a process known as spermiation (reviewed by O’Donnell et al. 2011), and travel onward into the epididymis via the rete testis for further maturation.

Sperm are immature when they enter the epididymis and have no capacity for motility or fertilization. During their transit through the initial segment, and caput, corpus and cauda regions of the epididymis, they are bathed in a variety of secretions from epithelial cells, including exosomes (reviewed by Cornwall 2009), that leads to the exchange (both an uptake and loss) of a massive number of proteins and non-coding RNAs, which confer the potential for fertility (Reilly et al. 2016; Nixon et al. 2019). Sperm are then stored in a quiescent state within the cauda epididymis prior to ejaculation.

Sperm leave the epididymis with the capacity for motility but only become motile upon dilution (Morton et al. 1978). They must, however, undergo one final maturation event in the female reproductive tract to achieve fertilization. This step is known as capacitation, and involves changes in sperm membrane fluidity and modification of the suite of receptors present on the sperm surface (reviewed by Aitken and Nixon 2013). Although a distinct process from capacitation, sperm also acquire hyperactivated motility, where they move with high amplitude tail beating and an asymmetrical waveform. This allows sperm bound to the epithelium in the isthmus (oviduct) to dislodge and travel to the site of fertilization (Demott and Suarez 1992), and then aids sperm in penetrating a protective layer of cumulus cells and the outermost layer of the oocyte, the zona pellucida (Stauss et al. 1995). Capacitation ultimately culminates with the ability of sperm to undergo the acrosome reaction to digest the zona pellucida and initiate fertilization, through binding with a receptive oocyte and transferring the paternal genome.

As a reflection of the number of genes expressed during spermatogenesis, there are many steps in sperm production, sperm maturation and fertilization that can break down, resulting in infertility. It is our aim to explain how to identify pathogenic genetic variants within infertile men that may perturb these processes, and then assess how modeling these mutations in mice can validate causality and shed light on the mechanisms leading to human male infertility.

2. Identification of genetic variants causing male infertility using DNA sequencing

Next generation sequencing techniques are now emerging as an informative approach to screen for disease causing variants in infertile men. The evidence required to implicate a genetic variant as a cause of disease can be complex and multidimensional, and guidelines for interpretation have been forged by clinical and preclinical scientists (MacArthur et al. 2014, Richards et al. 2015). We strongly encourage that interested scientists read these references for a complete discussion. These guidelines specifically relate to interpretation of variation in protein coding genes, which are the variants detected by whole exome sequencing. While the exome represents but 2% of the genome, any variants detected in protein coding genes can be assessed for direct structural and/or functional effects on protein function. Using this technology, novel male infertility variants are now being identified and published (reviewed by Oud et al. 2019). However, identifying rare genetic variants in an exome screen is a complex process, requiring a stringent filtering strategy, and, in many cases, additional experimental work to confirm causality.

After sequencing has been performed and genotypes have been generated by an appropriate and well-established bioinformatics pipeline, variants are computationally filtered and annotated to create a shorter and more manageable list for analysis. We note that the optimal analysis approach will depend on the disease model being investigated. Below, we focus on the identification of large-effect, so-called “Mendelian” disease mutations that produce recessive, dominant, and sex-linked inheritance patterns. These so-called Mendelian “disease models” describe how the genotype of a monogenic disease variant (e.g. heterozygous, homozygous or hemizygous) relates to a phenotype and thus disease clustering in families. More complex disease models, such as oligogenic (large effect mutations in multiple genes) or polygenic (e.g. where a large number of small effect variants produce a continuous disease liabilitiy in the population) are not considered here. As discussed below, it is important to note that the optimal analysis strategy for variant identification will depend on the true disease model. In a family study, a specific disease model can often be identified solely based on the segregation of the trait among relatives, which, in turn, can focus the genetic analysis on variants of a specific ploidy. In the absence of relatives, which is often the case for human infertility, all basic disease models should be considered. We note, however, that the majority of variants causing male infertility identified to date appear to be autosomal recessive or sex-linked (reviewed by Oud et al. 2019), but recognize that this may be a consequence of study design, rather than a reflection of the real situation. The disease model is critical for defining what type of variants to screen for: recessive diseases are typically be driven by loss-of-function mutations on both alleles of a gene, while dominant disorders are driven by heterozygous mutations that may be gain-of-function, or in the case of haploinsufficient genes, loss-of-function. Since recessive diseases may be caused by two different variants in the same gene (“compound heterozygotes”) it is important that the analysis be able to identify pairs of heterozygous variants that are potentially pathogenic, as we discuss below (Wilkie 1994; Bamshad et al. 2019)

Due to the strong purifying selection against infertility causing mutations, the frequency of each causal variant is expected to be very low. Large databases of exome data from population-based sequencing studies, such as gnomAD and the UK Biobank, are the key resources for estimating variant frequencies. These datasets are meant to reflect a relatively random sample of the population, including some that are infertile, thus the simple presence of a candidate genotype in a population database is usually too strict a criterion for its removal from a list of candidate pathogenic variants. Instead, it is recommended to filter out variants with population allele frequency that are incompatible with the known prevalence of male infertility and the presumed mode of inheritance. The key intuition is that, based on the observed prevalence and architecture of the trait, there are population allele frequency ranges that are impossible to observe, and variants observed in such frequency ranges can be readily filtered from the data as unlikely causes of disease (Amendola et al. 2016; Bamshad et al. 2011; Whiffin et al. 2017). For a complex disease such as male infertility, with extensive and yet unmapped loci and allelic heterogeneity, this can be challenging to predict precisely. For instance, state-of-the art methods for implementing this approach can use information on disease model, the number of genes involved in normal tissue function, the number of risk alleles, and the penetrance of the risk alleles; all values which one may confidently estimate for well-studied single-locus disorders such as cystic fibrosis. However, these are much trickier to calculate for male infertility, which is in fact multi-factorial in origin and likely to be caused by Mendelian mutations in likely thousands of genes. For simple illustration, if one assumes that the prevalence of male infertility is 1%, that 500 genes each contribute equally to male fertility, and 50% of the burden of male infertility is due to mutations in these genes, it is unlikely that the population prevalence of dominant mutations in one of these genes is greater than 0.001% (0.5% prevalence of genetic infertility / 500 loci = 0.001%). In the case of autosomal recessive variants, since heterozygous carriers are unaffected, the population frequency filter used could be higher, in the range of 0.5%-1%. In practice, we recommend the more lenient filter of 0.5%-1% allele frequency when the disease model and gene of interest is unknown.

Following on from this stage of filtering, it is common to predict the molecular consequences of the variant on protein function, using algorithms such as PolyPhen and SIFT, other known disease associations for each gene (from human or model organism studies), and expression patterns.

Even after filtering variants on frequency and predicted function, there will typically be tens to hundreds of variants left in contention as potentially causal for each case of male infertility. Here, prioritization of the remaining variants is essential and statistical genetics can be a powerful tool. In studies where DNA is available from relatives, a formal test of linkage or association between the variant and the trait can provide an objective statistic for prioritization of candidate genes (Lee et al. 2014; Hu et al. 2014). Access to parental DNA also provides the unique opportunity to identify de novo mutations that may explain cases of disease. In this scenario, filtering the variant list to just those that are de novo will dramatically shrink the number of candidates to ~1-2 variants per case (in the exome). These variants can then be investigated as an independent list alongside inherited mutations if the researcher is interested in understand the roles of classical inherited mutations versus those that form de novo. Another, highly relevant use of parental genetic data is to phase the alleles of heterozygous variants into maternal and paternal origin. This can be a very powerful method to determine whether two heterozygous variants in the same gene are in cis or in trans. In most investigations of recessive forms of disease one will want to disregard genes where the only damaging variants are observed in cis. Parental data has the power to phase all variants in the proband, while alternative approaches to gaining this information entail labor-intensive and low-throughput experiments that may not be feasible in some contexts.

The availability of parental DNA remains unlikely in most situations of male infertility wherein patients always present as adults. As such, most cases can be considered an n=1 case of a rare disease with the potential for a distinct genetic cause. The genome-wide association studies (GWAS) are the standard method for identified genetic causes of disease when dealing with a collection of unrelated cases, and have been very successful at mapping genetic variants associated with human traits. The typical GWAS statistical tests attempt to identify individual genetic variants with large frequency differences between cases and controls, a strategy that is poorly suited for male infertility as this disease is driven by a large number of individually rare mutations. It is likely that a standard GWAS approach would require >10,000 cases of male infertility and >10,000 controls to achieve statistical power to detect causal variants (Risch and Merikangas 1996), would miss most variants due to lack of power, and may generate false positives due to poor control of population stratification (Mathieson and McVean 2012). Recently, we proposed a way to perform formal statistic inference on variants sampled from n=1 cases of disease. This method uses data stored in large population sequencing databases to learn the patterns of genetic variation that exist in a random population sample (Wilfert et al. 2016). This modeling process can then be used to evaluate the probability of sampling a genotype, or set of genotypes, from the proband (infertile man). We refer to this probability as the “population sampling probability” or PSAP, which represents a formal measurement of unlikelihood for each genotype. PSAP probabilities can be calculated for five standard disease models (autosomal dominant, single variant autosomal recessive, compound heterozygous recessive, X-linked and Y-linked) and have been used successfully as a rational basis for prioritizing genotypes after filtering (Kasak et al. 2018; van der Bijl et al. 2019; Bustamante-Marin et al. 2019).

There are a number of other tools that allow for disease model-specific analysis of case genomes from unrelated cases. “AlleleFrequencyApp” allows users to identify highly-specific allele frequency filters that are based on disease model, disease prevalence, the presumed extent of allelic heterogeneity, genetic heterogeneity and penetrance (Whiffin et al. 2017; cardiodb.org/allelefrequencyapp). Another well regarded tool, pVAAST, uses hypothesis testing, instead of filtering, to identify rare variants potentially involved in disease (Hu et al. 2014; hufflab.org/software/pvaast). pVAAST uses a composite likelihood ratio test that can combine evidence for variant pathogenicity derived from allele frequency information, linkage analysis (if families are available), and functional effect predictions, with three disease models supported. We direct readers to excellent a recent review on the specific subject of identification of rare disease mutations for a more thorough discussion of methods (Carss et al. 2019).

The demographic background of the proband must be also considered when using methods that rely on population-scale data for filtering or prioritization. Probands from populations with poor representation in population databases should be handled with caution, as variants that are globally rare can be present at appreciable frequencies in isolated populations. One can attempt to mitigate this situation is by filtering against the maximum allele frequency observed in any population and / or using population-specific models (as is implemented in PSAP for some populations).

It should also be remembered that probands from families with high rates of consanguinity will be at greater risk for recessive forms of infertility, but will also have a larger number of rare homozygous variants that are unrelated to their fertility status, which must be filtered out. Whether or not the proband is from a consanguineous background, it is helpful to identify and annotate large runs of homozygosity (ROH) in each patient genome, which typically correspond to a pair of haplotypes inherited identical-by-descent (Ceballos et al. 2019). Rare pathogenic homozygous genotype calls are more likely to be real (as opposed to a technical artifact) if they are located in an ROH. In rare cases, a pathogenic, causal variant may appear as homozygous if it is in trans with a deletion of the same locus. Thus, one should investigate this possibility if a promising homozygous variant is not in an apparent ROH.

The problem of annotating ROHs from genetic data is well studied – in the context of Mendelian disease mapping. The primary challenge is determining at what length a run of homozygous genotypes becomes statistically interesting, and not just a chance observation. Methods for annotating large ROHs often take into consideration the background distribution of haplotype similarity in a population, in order to quantitatively identify regions of excess of homozygosity in a single sample (Browning and Browning, 2013; Gusev et al. 2009; Magi et al. 2014; Narasimhan et al. 2016). Important considerations for such analyses are that: a) an appropriate reference population is available for the proband when using population-based methods for ROH detection and b) the statistical method being used is well suited to the input datatype. Early methods for ROH detection were based on evenly-spaced genome-wide genetic maps derived from SNP arrays, while the most common data type for mapping Mendelian disease, at the moment, is exome sequencing. The H3M2 algorithm was devised specifically to map ROHs from exome data. It also appears to be more accurate than older array-based methods (Magi et al. 2014) and performs well in our hands in the study of male infertility.

PSAP, and some other methods for filtering and prioritization, incorporate the concept of a “disease model”, which is the theoretical inheritance mode of the trait: e.g. autosomal dominant, autosomal recessive, X-linked or Y-linked. As such, one of the first questions a researcher should ask is “what mode of inheritance do I expect in an individual patient?”. Multiple modes may need to be tested before the ‘high confidence’ candidate is identified. PSAP p-values can be compared across disease models, so in practice we evaluate PSAP for each disease model and combine the results for prioritization.

If interesting variants remain after filtering, Sanger sequencing should be performed on each variant of interest prior to experimental follow-up, to exclude the possibility that the genotype identified by next-generation sequencing is a technical artifact. If this is confirmed for novel genes / variants, researchers should plan to investigate the effect of the variant in experimental models, to explore its effect on gene function and / or male fertility in vivo. Finally, if there are no plausible candidates after filtering, one should bear in mind that there is always a possibility that a particular idiopathic infertility case may not have a genetic etiology, or may be contained in intronic sequences. In many exome sequencing clinics for genetic disease the diagnostic yield is less than 50% (Wright et al. 2018). For a discussion on environmental causes of male infertility, readers are referred to Sharpe (2000) and Wong and Cheng (2011).

2.1. Validating variant causality in male infertility - point mutations or full knockouts?

A common consideration when generating mutant cell or animal models for the study of human disease is whether it will be most informative to generate a full gene knockout or to model the exact mutation observed in the patient. As mentioned in the previous section, there are rigorous guidelines for clinical interpretation of genetic variants, and the observation of infertility in a knockout mouse is typically not sufficient to declare causality of a different type of mutation (e.g. a missense change) in a patient. If you are interested in understanding whether a particular genetic variant is likely causing the infertility, then it is best to generate mice recapitulating the patient’s genotype, for example knocking in missense mutations. Mutation(s) predicted to induce loss-of-function, however, e.g. nonsense mediated decay of RNA, lend support to the generation of knockouts. As a rule of thumb, this occurs when mutations induce the formation of premature stop codons upstream of the last 50 bp in the penultimate exon (Lindeboom et al. 2016). Regardless of whether the mutation is a complete or partial loss-of-function the mouse model will also be a valuable research tool to ascertain gene function.

Overall, we suggest modeling the patient’s precise mutation where possible. In (fortunate) cases where the same gene is mutated within multiple infertile men, and these mutations are predicted to be damaging, one should look for functional equivalency among these mutations – for example, do they all affect the same protein domain? This may simplify the process of selecting a mutation to model in the mouse. Although it is not always feasible, the ability to generate point mutant and full knockout models for a gene of interest side-by-side would allow for a more confident assessment of the effect of the mutation in the patient’s infertility and characterization of gene function broadly. For a comprehensive discussion on CRISPR technologies that can be used to generate full knockout, conditional or point mutant mice please see Yang et al. (2014).

Regardless of the approach chosen, when designing the method, it is essential to determine 1) if all, or which, transcript variants will be targeted by the genetic modification (Mohr et al. 2016), and 2) if any known functional domains will be affected. We routinely use the Ensembl genome browser (www.ensembl.org) to compare transcript structures. We note, the transcriptome in the testis is particularly complex and the use of alternative downstream promoters is common (Yue and Ogawa 2018). As such, it is worth taking time to assess the full extent of transcriptional variation in the testis, using actual testis mRNA, rather than relying solely on databases. For a relatively complete list of available point mutant mice and / or sperm, readers are referred to the Mouse Genome Informatics browser (http://www.informatics.jax.org). As not all mouse models are cataloged in the Mouse Genome Informatics browser, it is also worth searching the literature.

3. Phenotyping mouse models

While there are multiple model organisms available to interrogate the effect of mutation(s) on specific gene function, this review focuses on the benefits of the mouse model. An abnormal semen profile in humans is usually described based on the number of sperm present and their morphology and functionality. The most commonly used terms are 1) azoospermia - the absence of sperm in a human ejaculate, which can arise because of germ cell defects, a failure of Sertoli cells to release sperm (spermiation failure) or obstruction downstream in the tract. The latter is most famously caused by mutations in the human CFTR gene which lead to a congenital absence of the vas deferens (Cuppens and Cassiman 2004); 2) oligozoospermia – wherein the ejaculate contains reduced numbers of sperm; 3) teratozoospermia – where sperm are abnormally shaped; and 4) asthenozoospermia – where sperm have compromised motility. In instances where patients contain a combination of these presentations, the suffixes are combined e.g. oligoasthenoteratozoospermia; where patients present with reduced numbers of sperm, and the sperm present are both abnormally shaped and poorly motile. While these terms are arguably human-specific, they should be kept in mind when characterizing mouse phenotypes, as it will assist in attempting to link the pathology seen in humans with that seen in mice.

Prior to describing the pipeline to phenotype male infertility in the mouse, it is worth first reviewing the major categories of pathology that exist in infertile men, and how to determine if they are recapitulated in the mouse. Figure 1 is an overview of expected infertility phenotypes when knocking out genes required in the major germ cell types. Mutations in genes required for spermatogonial function will likely cause an azoospermia-like, or a germ cell arrest, phenotype. A germ cell arrest is where germ cells develop up to a particular point in spermatogenesis then die by either apoptosis, precocious sloughing from the epithelium, or phagocytosis by Sertoli cells.

Mutations in genes essential for meiosis (spermatocytes) may also cause a germ cell arrest phenotype (meiotic arrest) and azoospermia. Whereas an incomplete block may lead to oligozoospermia and / or sperm aneuploidy. Mutations in genes required for spermiogenesis are likely to result in teratozoospermia and / or asthenozoospermia with or without oligozoospermia. Furthermore, mutations in genes required for Sertoli cell function may lead to reduced sperm number or functionality; or cause a breakdown of the blood-testis barrier. The latter may lead to an abnormal immune response and ultimately orchitis (testis inflammation) and / or the presence of antibodies directed against sperm proteins (anti-sperm antibodies) in the circulation.

3.1. Mouse phenotype

Once a mutant mouse model line has been generated, it is important to carefully plan the phenotyping strategy, both to save money and extract the maximal amount of data from each mouse in accordance with ethical principles. If thought to be a recessive mutation, the first step is to breed heterozygous mice to generate homozygous mutant mice. If dominant infertility is suspected, or the mutation is on the X chromosome, e.g. as was the case for haploinsufficiency in Prm1 or Prm2 null mice (Cho et al. 2001), heterozygous mutants should be investigated. If this is the case, the line will need to be maintained (hopefully) through the female germline.

To aid researchers in conducting the most efficient and comprehensive approach to assessing male infertility in the mouse model, and building on our previous review [Borg et al. 2010], we have constructed a flowchart to guide researchers (Figure 2). Researchers should also compare back to the clinical presentation of the original infertile man (or men). Overall, our strategy assays a broad range of male fertility parameters, where even slight perturbations can render males sub-fertile (Schimenti and Handel 2018). Such impairments can be achieved by any of the following: fluctuations in hormone levels, impairments to spermatogenesis including sperm structural defects, epididymal maturation, impaired sperm motility or acrosome function, precocious or absent sperm capacitation, or compromised sperm-egg interaction. Examples of these infertility types have been collated in Table 1.

Figure 2. Male fertility phenotyping pipeline for mutant mouse models.

Suggested techniques for assessing male fertility in mutant mouse models are listed. Each step is detailed in order from 1-7, where all experiments should utilize age matched wildtype males for comparison. First, males should be caged with females to monitor their ability to mate and form a copulatory plug (1). If males successfully plug females, a breeding trial (2) should be conducted to determine time to pregnancy and the number offspring generated per mating (plugging then no offspring should be recorded as 0). We suggest using 5 males independently mated to 2 wildtype females each. Once age appropriate males are ready to phenotype, they should be humanely culled, weighed and assessed for any gross abnormalities (3). Blood should be collected soon after death to maximize yield (4), allowing for assessment of hormonal parameters. Next, the testes and epididymides should be dissected out and weighed (5, 6). One testis and epididymis should be fixed for histology. The second testis can be frozen for daily sperm production calculations (or confirmation of gene knockout / mutation), or can be fixed for electron microscopy. The cauda and vas deferens of the second epididymis should be used for sperm collection and an assessment of sperm motility via computer assisted semen analysis. Any residual sperm can be fixed for an analysis of sperm morphology (light) and structure (electron microscopy), or capacitated for an assessment of the ability of sperm ability to capacitate or achieve in vitro fertilization. Tail or ear clippings should be collected (7) in order to confirm genotype, if required. Note: not all of these techniques can be undertaken with a single mouse; therefore a full phenotypic investigation will require multiple animals of the same genotype. ELISA, enzyme linked immunosorbent assay; MS, mass spectrometry; EM, electron microscopy; PFA, paraformaldehyde, GA, glutaraldehyde; PAS, periodic acid and Schiff’s reagents; HE, hematoxylin and eosin; CASA, computer assisted semen analysis; IVF, in vitro fertilization; ICSI, intracytoplasmic sperm injection.

Table 1. Examples of phenotyping assays that can be used to categorize male infertility sub-types.

All assays can be performed as a function of age to determine age-related changes in fertility.

Technique / assay	What it assesses	Defects and phenotypes	Method	Reference
Breeding trial	Overall fertility including the presence of mating	Quantitative readout of overall fertility – normal, sub-fertility and sterility	Males are placed with 2 individual female each and the number of pups per plug is recorded. Number of days taken to plug females can also be recorded	Gold et al., 2009
Daily sperm production	Sperm production rate in the testis	Hypospermatogenesis, hyperspermatogenesis, azoospermia	Testes are decapsulated and sonicated in buffer, then sperm heads are counted with a hemocytometer	Lim et al., 2015
Epididymal content	Sperm numbers within the epididymis	Quantification of sperm reservoir in the epididymis. By comparison with DSP, spermiation defects can be quantitated	As above, except epididymides are homogenized	Dunleavy et al., 2017
Electron microscopy of the testis / sperm (or epididymis)	Germ cells, Sertoli cells and other cells within the testis or epididymis ultrastructure	Spermiogenesis defects, teratozoospermia	Glutaraldehyde fixed tissues. This will require training or specialized equipment	Shang et al., 2017
Computer assisted sperm analysis (CASA)	Sperm motility and velocity parameters	Asthenozoospermia. Changes in sperm total and progressive motility, velocity, and other swimming parameters	Sperm are back-flushed into medium and placed in a CASA slide chamber. Motility is analyzed by an appropriate CASA system	Harris et al., 2007
High-speed motility imaging	Sperm waveform parameters	Subtle changes in sperm waveform, power dissipation and periodicity	As above but sperm are placed in a medium that allows head tethering to the slide to measure motility parameters of the tail. Requires custom-made software and microscope setup, and will likely be a collaborative effort	Hansen et al., 2018
Capacitation (measured by tyrosine phosphorylation)	Activation of sperm proteins during capacitation (tyrosine residues phosphorylated)	Precocious, delayed or absent capacitation; normal / abnormal sperm head and tail phosphorylation localization	Sperm are capacitated and fixed, then stained with a phosphotyrosine clone	Cotton et al., 2006
Acrosome reaction	Ability of sperm to undergo acrosomal exocytosis, the acrosome reaction	An inability to manifest an acrosome reaction or precocious, partial acrosome reaction	Sperm are capacitated then treated with progesterone to induce the acrosome reaction. Sperm are fixed and stained with peanut agglutinin lectin	Lim et al., 2019
In vitro fertilization	Fertilizing potential of sperm when placed in close proximity to a collection of oocytes	Defects in fertilization: interaction with the zona pellucida (denuded) or plasma membrane binding (zona free); inability to support early embryo development, e.g. DNA damage in sperm or failure to activate the oocyte	Sperm are capacitated and placed with cumulus-oocyte clutches, denuded or zona free oocytes. After sperm-egg interaction, unbound sperm are washed away and potential zygotes are cultured, or sperm bound to the zona / fused with oocyte are recorded	Taft, 2017; Inoue et al., 2005
Hormone analyses	Circulating levels of reproductive hormones	Perturbed hormonal regulation, including hypogonadotropic and hypergonadotropic hypogonadism	Blood is collected via cardiac puncture and allowed to clot to form the serum. This is taken for mass spectrometry or ELISA	Ren et al., 2019; Rebourcet et al., 2016
Flow cytometry	Unknown cell types, e.g. immune cells; fluorescently labelled germ cells	Immune cell infiltration, changes in fluorescently labelled germ cell numbers	Testes or epididymides are enzymatically digested and cells can be labelled, or sorted as is, using a flow cytometer	Voisin et al., 2018; Pitetti et al. 2013
Anti-sperm antibodies	Presence of auto-antibodies recognizing germ cells and sperm	Breakdown of the blood-testis barrier, immune cell infiltration	Sera is taken from wildtype and mutant males and incubated with 1) testis protein lysates via western blotting or 2) on testis sections via immunofluorescence	Meng et al., 2011

In mutant models unexpected phenotypes may arise. Therefore, it is important to carefully assess mice for abnormalities and weigh each mouse and organ. There is often unexpected value in assessing a broad range of tissues for phenotypes, using relatively quick and cheap methods such as histology, especially for those in which the target gene is highly expressed. If abnormalities are found, advice from ethics committees (or similar) should be sort as to whether the data should be communicated to the treating clinician in the hope that they may help to prevent discomfort in the patient. In all the following assays / experiments, age-matched wildtype littermates should be utilized as controls and assessed in parallel. We strongly advise against using heterozygous males as controls, as many genes are known to have dose effects. The focus of this review will be on mutations that affect the testis, epididymis and sperm structure and function.

3. 2. Fertility assessment

Once mutant males have been produced, they should be aged to at least ten weeks of age before their fertility is tested by natural mating. We choose ten weeks as spermatogenesis and epididymal maturation are well established by this time, and sperm output is thought to be close to maximal. This is also a pragmatic choice to balance housing costs with allowing enough time for a phenotype to appear i.e. multiple rounds of spermatogenesis will have been initiated by this time. We recommend that five males should be investigated and that each male is placed with two wildtype females. We also recommend the age range of males be kept relatively tight to avoid compounding effects of aging by using males at 10-12 weeks of age. Alternatively, and particularly if an effect of age is suspected, fecundity (or individual fertility parameters) can be plotted as a function of age. In this case you may choose to start mating as early as 7-8 weeks in order to assess sperm produced during the first wave of spermatogenesis, and then mate for a continued period, or utilize older males to determine if there is an age-related component to the infertility. We note that spermatogonia produced in the first wave of spermatogenesis are produced though a different developmental pathway than those produced in subsequent waves (Yoshida et al. 2006). As such, if you suspect a spermatogonia phenotype it may be productive to assess fertility from both the first wave of spermatogenesis and subsequent waves separately.

For all breeding strategies, females will ideally be young, morphologically normal and have had at least one litter previously. For consistency, it is best to keep the age of females tightly grouped. Mating usually takes place overnight with one female per night and should be confirmed by the presence of copulatory plugs early the following morning (Figure 2, flowchart step #1), as an indicator that mating actually took place. Plugs should appear within 5 days of co-habitation (see Behringer et al. 2016). It is worth noting that altered hormonal balance can lead to changes in mating that will initially appear as sterility e.g. mutations that affect hormone signaling (Eddy et al. 1996; Gibson et al. 1997; McDevitt et al. 2007; Goulding et al., 2010) and monitoring both plugging and sexual behaviour can assist in identifying these phenotypes. A more detailed description of the effects of hormonal irregularities can be found in section 3.3.9. Mating will also be affected by problems with sex determination or sex reversal (reviewed in Ewen and Koopman 2010; Camerino et al. 2006). If this is suspected, the genotypic and phenotypic sex should be compared, by investigating gonads and expression of key male and female sex genes e.g. as done for Dmrt1 overexpression mice (Zhao et al. 2015). In these infertility cases, it is likely that the human phenotype will form part of a spectrum of disorders of sexual development, which extend from sex reversal to some types of male infertility (see Eggers et al. 2016). During mating trials, the presence of a plug, but the absence of a subsequent litter should be recorded as zero.

Care should also be taken to ensure that animal housing conditions are optimal, as stressed animals will have erratic fertility. Stressful factors include: noise, changes to staff, foreign smells and the close proximity (or smell) of predatory species (Bind et al. 2013). For example, animal houses undergoing construction works that generate extra noise or vibration are stressful for mice and this will affect the likelihood of normal sexual behavior and overall reproductive success (Diercks et al. 2010). Lastly, certain knockouts have been shown to affect plug formation but not mating (Laing et al. 2000; Noda et al. 2019). If this is suspected, we recommend monitoring mating by video. In cases where females are plugged but no pregnancy occurs, or litter size is decreased, further investigation is required to determine the cause of sub-fertility / sterility.

3.3. Male reproductive tract assessment

As indicated above, the age of the male should be considered prior to starting phenotyping for infertility. Any histological samples / biopsies of the infertile man will give an indication of what to expect if the gene of interest is causing their infertility. Mating at 7-8 weeks will allow an assessment of the first wave of spermatogenesis, whereas older ages will allow an assessment of the adult stem cell population. Examples of this include variants that affect spermatogonial stem cell renewal, such as in Etv5 knockout or mutant mice. These males display a single round of spermatogenesis, followed by the age-dependent degeneration of the seminiferous epithelium, then azoospermia (Morrow et al. 2007; Jamsai et al. 2013). Similarly, the analysis of older mice may reveal phenotypes that arise with age or as a consequence of interactions with the environment, including the abnormal activation of the immune system. Given that infertile men are extremely unlikely to present for an infertility assessment at the time of the first round of spermatogenesis, it is strongly advised that mice older than 7-8 weeks of age are assessed. An example of an interaction between genotype and age is the Crisp1/4 double knockout mice wherein sperm dysfunction is compounded by a break down in immune privilege in the epididymis with age (Hu et al. 2018). As a general first principle, however, we recommend initially mating males at 10 weeks of age, followed by tissue collection prior to 12 weeks of age so as to maximise the number of parameters that can be assayed from small numbers of animals.

3.3.1. Assessment of male reproductive organs

After culling and weighing mice (flowchart step #3), blood should be isolated by cardiac puncture (flowchart step #4) and the resulting serum stored for hormone analysis (Musicki et al. 2015; O’Hara et al. 2011; Welsh et al. 2010). Testes and epididymides should be dissected out carefully to ensure the integrity of each tissue (in mice the corpus epididymis is extremely thin) and that fat is removed prior to weighing (flowchart steps #5 and #6). Decreased testis weight is suggestive of spermatogenic impairments which can be confirmed histologically. While processing the testis for histology, it is important to carefully pierce the tunica albuginea (capsule) with a fine needle and immerse it in an excess of Bouin’s solution (5 ml) at room temperature. Bouin’s solution has superior infiltration compared to many other fixatives as it contains picric and acetic acids, minimizing tissue artifacts (Morita, 1971; Howat and Wilson 2014). Davidson’s fixative also results in good tissue preservation, whereas paraformaldehyde penetrates relatively slowly and will leave a gradient of fixation in the testis and results in poor tissue architecture. While Bouin’s solution allows for rapid protein crosslinking (fixing) that is conducive for quality histology, it does have the potential to alter protein confirmation, influencing the success of immunohistochemical techniques. We recommend fixing testes for a maximum of 5 h prior to placing them into 70% ethanol as a compromise between excellent histology and maintaining antigenicity. Shrinkage artifacts between the inner tubules of the testis, resulting in an increased interstitial space or vacuolization within tubules, indicates inadequate fixation or compromised fixative penetration (Hess and Moore 1993). After alcohol processing, each testis can be carefully cut in half transversally with a razor and embedded in paraffin wax using standard methods.

We also recommend that one epididymis be dissected out, placed on a flat piece of filter paper, and then fixed in freshly prepared 4% paraformaldehyde overnight. Using the filter paper will help to keep the epididymis flat, thus improving the possibility of obtaining a histology section containing all regions of the epididymis. Alternatively, epididymides can be fixed between biopsy pads in a cassette to help maintain a flat shape. After fixation, epididymides should be placed in 70% ethanol and alcohol processed prior to being embedded flat in paraffin wax, for sectioning using standard methods.

When investigating a new mutant mouse line for the first time, we recommend that the second testis should be dissected out, weighed, then snap frozen in either liquid nitrogen or on a dry ice-ethanol slurry and stored at −80°C. The tissue can be used later to calculate the daily sperm production, as explained below, or for protein / RNA extraction. Depending on the priority of data collection, the second epididymis should be used to harvest sperm for functional analysis (see below) or snap frozen for a calculation of total epididymal sperm content.

3.3.2. Sperm collection and motility analysis

If sperm are generated, epididymides should be collected to allow for an analysis of sperm motility. Sperm should be collected from the cauda and vas deferens using the back-flushing method. This method was first described as retrograde perfusion through the vas deferens to push sperm backwards through a hole in the cauda (Henle and Chambers 1938; Morton and Lardy 1967). However, the protocol has been optimized to flush sperm out of the epididymis using air, allowing sperm to be collected with a micropipette then transferred to an appropriate medium, as described by Baker et al. (2014). The back-flushing method provides highly pure sperm, whereas sperm collection from minced epididymides can result in blood contamination and sperm immobilization (Suarez et al. 1981). Swim-up or swim-out methods bias for motile populations of sperm and are incompatible with mutations that result in immotile sperm.

It should be remembered that mouse sperm are very fragile and should always be kept warm at 37°C, not exposed to UV light or shearing forces, such as vigorous pipetting or high-speed centrifugation, and be kept in appropriate medium – not PBS. Incubation in either BWW (Biggers et al. 1971; Smith et al. 2013), MT6 (Cotton et al. 2006; Oliveira et al. 2009; Mann 1988) or HTF (Jin et al. 2007; Goodson et al. 2011) medium for 15 min at 37°C with 5% atmospheric CO₂ will allow sperm to disperse after back-flushing and will best preserve these cells. Sperm viability may be assessed with the eosin exclusion assay, hypo-osmotic swelling test, or by incubation in propidium iodide such as through the LIVE/DEAD Sperm Viability Kit (Garner and Johnson 1995; Viggiano et al. 1996).

3.3.2.1. Computer assisted semen analysis (CASA)

CASA is a useful method to objectively measure sperm motility and directly compare to human semen analyses (when the correct software parameters are set). Sperm motility can be investigated ‘by eye’ as a fall back, taking note of the percentage of motile and progressively motile cells per genotype. As mouse sperm tails are approximately 100 μm long, deep slide chamber (80 μm) slides are required for motility analyses and they should be used on a pre-warmed stage (Gibbs et al. 2011). CASA will provide measurements for total, progressive and rapid motility alongside multiple movement parameters (Harris et al. 2007; Park et al. 2012). A proportion of at least 70-80% of sperm from healthy control males should be motile in the media listed above, although this likely varies with mouse background. Progressive motility will likely achieve upwards of 40% in controls, although we regularly achieve 50%. Care should be taken to ensure sperm are not sticking to the slides. If this occurs, the slides are likely contaminated or the medium has not been prepared correctly.

Despite CASA still being accepted as the gold standard method to quantitate sperm motility clinically, this method has low magnification and resolution, and cannot detect complex motility defects. Comprehensive motility analysis systems are currently under development and are beginning to allow the detection of more subtle motility defects (Bjorkgren et al. 2016; Hansen et al. 2018; Lim et al. 2019), which are predicted to affect sperm competition within the female reproductive tract and evolutionary processes (Firman and Simmons 2010). How this precisely relates to human fertility is unknown but given complex architecture of the female tract these defects are likely to reflect functional compromise. Reductions in sperm motility likely result from imperfections in sperm architecture, some of which may be visible using electron microscopy, or inadequate energy supply, i.e. via the oxidative phosphorylation and glycolysis pathways, or ion channel function. For a summary of sperm energy requirements readers are referred to Storey (2008).

3.3.3. Histological examination of the testis and epididymis

3.3.3.1. Testicular histology

One of the most productive assessments that can be done if male sub-fertility or sterility is suspected, is an assessment of testis and epididymal histology. This is particularly useful if the biopsy from the original patient is available so as to allow a direct comparison of the genotype-phenotype relationship. In brief, the testis can be separated into two main compartments, the seminiferous tubules, containing the germ and Sertoli cells, and the interstitium, containing Leydig cells, immune cells and blood vessels. Spermatogenesis in the mouse occurs in synchronous waves, meaning a continuous piece of tubule can be divided into ‘stages’, each containing a set of germ cell combinations in intimate association with supporting Sertoli cells (Griswold 2016). While chronology of human spermatogenesis is virtually identical, human spermatogenesis is far less synchronous when looking at tubule cross-sections. As such, newcomers to the field of reproductive biology should ask for assistance in interpreting this difference.

For histological examination, testis sections (5 μm) should be stained with periodic acid Schiff reagent and haematoxylin in order to accurately visualize germ cell nuclei (stained blue), basement membranes and acrosomes (stained pink; Russell et al. 1990). Staging of the seminiferous tubules via testis sections will allow a comparison of germ cell type and abundance between genotypes (see Oakberg 1956; Russell et al. 1990; Hess and de Franca 2008). Researchers should closely examine tubules to ascertain whether all cell types are present and that correct cell types appear at the appropriate timing / location during stages. A comprehensive description of spermatogenesis and possible pathologies is available (Kerr et al. 2006), with a gold standard mammalian spermatogenesis guide published by Russell and colleagues (1990). Testis sections can be stained with various markers to define the origin of infertility phenotypes, identify unknown cell types or for use in co-localization experiments. To facilitate such analyses, we have included a table of indicative markers for key cell types and structures throughout spermatogenesis (Table 2).

Table 2.

Markers of main cell types and key structures present within the testis.

Cell type	Marker	Reference
Sertoli cell	WT1, Wilms tumor 1 protein SOX9, SRY-Box 9	Wang et al, 2013; Zheng et al, 2014 Morais da Silva et al, 1996
Adult Leydig cell Fetal Leydig cell Fetal and adult Leydig cell	HDS3B6, Hydroxysteroid dehydrogenase 3 beta 6 MC2R, Melanocortin-2 receptor HSD3B1, Hydroxysteroid dehydrogenase 3 beta 1 CYP11A1, Cholesterol side-chain cleavage enzyme CYP17A1, Cytochrome P450 family 17 subfamily A member 1	Yokoyama et al, 2019 O’Shaughnessy et al, 2003 Yokoyama et al, 2019 Rebourcet et al, 2019 Rebourcet et al, 2019
Peritubular myoid cell	CNN1, Calponin 1	Rebourcet et al, 2014
Spermatogonial stem cell Spermatogonia: progenitor Spermatogonia: differentiating Spermatogonia (general)	ID4, Inhibitor of DNA binding 4 GFRalpha1, GDNF family receptor alpha-1 SOX3, SRY-Box 3 RARgamma, Retinoic acid receptor gamma c-KIT, Receptor tyrosine kinase GILZ, Glucocorticoid-induced leucine zipper SALL4, Spalt-like 4	Chan et al, 2004 Hara et al., 2014 Suzuki et al, 2012 Ikami et al., 2015 Yoshinaga et al, 1991 La et al, 2018 Chan et al., 2017
Spermatocyte (general)	SYCP3, Synaptonemal complex protein 3	Fine et al, 2019
Spermatids	ACRV1 (SP-10), Acrosome vesicle 1	Reddi et al, 1999
Immune cells	Various, see associated reference	Voisin et al, 2018
Structures	Marker	Reference
Acrosome	PNA, Peanut agglutinin lectin	Virtanen et al, 1984
Manchette	α-tubulin, Alpha-tubulin	Cherry and Hsu, 1984
Sperm tail/axoneme	Acetylated alpha-tubulin	Hermo et al, 1991
Centrioles	Centrin	Manandhar et al, 1999
Fibrous sheath	AKAP4, A-Kinase anchoring protein 4	Carrera et al, 1994; Fang et al, 2019
Outer dense fibers	ODF1, Outer dense fiber 1	Higgy et al, 1994
Mitochondria	MitoTracker CMXRos (live stain or pre-load before fixation) TOM20	Fujihara et al, 2012 Hayashida et al, 2005
Crossing over (meiosis)	MLH1, MutL Homolog 1 MLH3, MutL Homolog 3	Marcon and Moens, 2003 Marcon and Moens, 2003
Sex body (meiosis)	ASY, Asynaptin	Turner et al, 2000
Processes	Stain/Marker	Reference
Fibrosis	Masson’s trichrome	Macmillan, 1953
Apoptosis	Cleaved caspase-3 and −9	O’Bryan et al, 2013

Cleaved caspase-3 and −9 can be used to investigate whether elevated numbers of germ cells are undergoing apoptosis, e.g. as in the azoospermic Rbm5 knockout (O’Bryan et al. 2013). However, it is important to note that during the first wave of spermatogenesis there is a much larger number of apoptotic germ cells. To investigate potential increases in fibrosis in either the testis or epididymis, staining of sections with Masson’s trichrome may also be of value (Macmillan 1953; Steiner and Boggs 1965).

3.3.3.2. Epididymal histology and function

Epididymal sections should be stained with periodic acid Schiff reagent and haematoxylin, or haematoxylin and eosin (Abe et al. 1983; Borg et al. 2010; Cheng et al. 2018). While there will likely be little known about epididymal function in the infertile man / men, investigating this structure in the mouse will allow for an analysis of epididymal maturation. If abnormalities are detected, this information may inform clinical practice. In the mouse epididymis there are at least 10 discrete regions (Johnston et al. 2005; Jelinksy et al. 2007), but it is routinely analysed as four major regions: the initial segment, caput (head), corpus (body) and cauda (tail), which increase in luminal volume:epithelial height ratio proximally to distally (Cornwall 2009). Sperm should be present in each section of the epididymis in increasing concentration from caput to cauda, with the cauda acting as a sperm reservoir. Many pathologies of the epididymis have been described (reviewed by De Grava Kempinas and Klinefelter 2014). Researchers should pay close attention to any abnormalities in the epithelial cells. A clear histological assessment of the major cell types present across the initial segment, caput, corpus and cauda can be informative (see Hermo and Robaire 2002; Breton et al. 2019). Meanwhile, the absence of sperm, or severe reduction in sperm numbers, in the epididymis is strong evidence for defects in spermatogenesis (e.g. germ cell arrest / apoptosis or spermiation failure) or transit between the testis and epididymis (e.g. Suzuki et al. 1996; Bolor et al. 2005; Denison et al. 2011).

Epididymal epithelial cells maintain a physiologically acidic pH within the lumen in order to optimize sperm maturation and storage. However, changes in cell composition or activity of the V-ATPase proton pump, as seen in Foxi1 knockout mice, can result in significantly elevated pH and impaired fertility (Yeung et al. 2004; Blomqvist et al. 2006; Park et al. 2017; Carvajal et al. 2018). Epididymal fluid and pH can be measured via small nicks, taking care to avoid any contaminating blood vessels (Blomqvist et al. 2006). In addition, epithelial cell height can be used a marker of cell maturity within the initial segment e.g. Jun et al. (2014). The sperm of these mice, which contain a Ros1 deletion, as well as those in the Esr1 knockout, display a bend at the level of sperm midpiece, as a consequence of reduced osmolality within the epididymal fluid (Joseph et al. 2010).

A second role of the epithelial cells of the epididymis is to maintain the blood-epididymis barrier through tight junctions between principal cells. This role is complemented by a population of immune cells, including dendritic cells, macrophages, B/T cells and monocytes resident within the epididymal epithelium and interstitium (reviewed by Breton et al. 2019). Interestingly, dendritic cells, are more concentrated within the caput than the cauda, while macrophages in the initial segment have been shown to generate an abundance of intraepithelial projections into the lumen, which diminish within the caput and are absent within the cauda (Voisin et al. 2018; Breton et al. 2019). The changes in immune cell populations throughout the epididymis support an important role in both protection against pathogens and maintaining tolerance against sperm antigens. If an altered immune cell profile is expected or observed, this can be characterized using flow cytometry with a variety of markers, as shown by Voisin and colleagues (2018). If this the case, it may also be worth assessing mouse and human serum for the presence of anti-sperm antibodies using methods such as those described previously (Hubert et al. 2009; Meng et al. 2011). These same assays are equally useful if you suspect that there is a break down in the blood-testis-barrier.

3.3.3.3. Sperm morphology

In regard to sperm cytology, cells can be fixed in a variety of solutions depending on how you ultimately want to stain them, i.e. 4% paraformaldehyde, ethanol, ice-cold methanol or acetone. Sperm morphology can be reliably assessed with any of these fixatives, followed by haematoxylin and eosin staining. Researchers should pay close attention to the four major regions of the spermatozoon; the sperm head, and the tail (flagellum), comprising of the midpiece, principal piece and end piece (Borg et al. 2010). Unlike the rounded human sperm head and high variability in sperm morphology, mouse sperm have a falciform (hooked) sperm head and the vast majority have a highly consistent morphology. Sperm head defects can be further quantitated using fixed and DAPI stained sperm, using an analysis program recently released by Skinner and colleagues (2019). Changes in sperm morphology or motility are suggestive of impaired spermatogenesis and should be further explored with electron microscopy.

3.3.4. Electron microscopy of the testis, epididymis and sperm

As more mice become available, we strongly recommend testes, and potentially the epididymides and sperm, be fixed with glutaraldehyde-based solutions for electron microscopy (EM; Russell and Brinster 1996; Dettin et al. 2003; Feng et al. 2017; Tang et al. 2017). Although it is highly unlikely that EM will be available for human testis biopsies, it may have been performed on sperm in a few specialized clinics. Transmission electron microscopy is extremely valuable for the detection of subtle ultrastructural defects that arise in spermatogenesis and epididymal transit (see Figure 3 for examples of sperm defects identified by EM). Scanning electron microscopy, e.g. of spermatozoa, will highlight surface and shaping defects. Possibly the most important aspect of these methods is ensuring the tissues or cells are well fixed to avoid artifacts. For protocols and a comprehensive review of normal spermatogenesis and abnormal sperm phenotypes please see Russell et al. (1990) and Chemes (2013). With electron microscopy, multiple sperm or germ cells within a sample / section, and across multiple mice, should be investigated to ensure that any defects are recapitulated throughout the population.

Figure 3. Electron microscopy of the testis and sperm from a variety of mouse knockouts to assess for structural abnormalities.

Scanning electron microscopy of (A) wildtype and (B) mutant epididymal spermatozoa. Transmission electron microscopy, as shown in (C) wildtype and (D) Katnal2 Y86C point mutant elongating spermatids, can identify subcellular structural abnormalities. Katnal2 mutants have manchette defects leading to abnormal sperm head shaping (arrows depict normal and abnormal constriction of the manchette). (E) Wildtype axoneme structure (with arrow depicting outer dynein arm of one doublet). (F) Loss of outer dynein arms (arrow) from the axoneme microtubule doublets in a patient with a duplication in the DNAH17 gene (modified with permission from Whitfield et al., 2019). (G) Midpiece-principal piece junction (white arrow) in wildtype sperm tail and (H) interruption to this junction as seen in the sperm of Slirp knockouts (white arrow and black line). (I) Normal centriole positioning and docking in cell wildtype spermatids (inset expanded and docked centriole depicted by arrow). (J) Aberrant centriole duplication in spermatids from Katnal2 knockout mice (inset expanded and three centriole pairs are depicted by arrows). Scale bars shown in each wildtype image are representative for corresponding mutant images in each pair of micrographs.

3.3.5. Calculating daily sperm production and epididymal sperm content

Although sperm extracted from the epididymis using back-flushing provides a relative measure of sperm count, this is not precise as not all sperm can be collected, and superior methods exist. One such method relies on the selective solubilization of germ cells using Triton X-100 and then subsequently counting the remaining sperm / spermatids using a hemocytometer (Cotton et al. 2006). We have recently modified this Triton-X-100 solubilization method by adding in a sonication step to separate the sperm (and spermatid) head and tail, which improves reliability of counting. Calculation of daily sperm production (DSP) and epididymal content (EC) relies on the super-condensation of DNA in the head of elongating spermatids / testicular sperm (DSP) and epididymal sperm (EC), which is resistant to detergent and sonication. By contrast, the nuclei of other cells will lyse under these conditions. The Triton-X-100 method is, however, dependent on sperm having a normal detergent-resistant structure. As such, this may not be the best method in instances where abnormalities in DNA packaging are suspected.

Similarly, EC, or the total number of sperm in the epididymis, can be calculated by homogenizing then sonicating the whole epididymis and counting the number of sperm heads. A disconnect in the relative percentage of sperm produced in the testis compared to their content within the epididymis is indicative of a spermiation failure e.g. Katnal2 knockout males (Dunleavy et al. 2017), or that sperm are being abnormally removed by the epididymis (e.g. Serre and Robaire 1999; Denison et al. 2011; Ramos-Ibeas et al. 2013; Hu et al. 2018). Sperm content within the C57BL/6 mouse epididymis reaches maximal levels around 12-16 weeks of age (Ye et al. 2008).

Reductions in daily sperm production in the testis suggest a spermatogenic defect, such as increases in germ cell arrest, apoptosis or sloughing, reduced Sertoli cell numbers, or hormonal imbalances (Blanchard and Johnson 1997; Watanabe 2005; Juma et al. 2017). By staining the testis and epididymis with appropriate markers (see Table 2) and closely examining histology, these phenotypes will become evident. Questions that you may want to ask include: “does germ cell arrest occur before sperm are produced?”, which will be evident via histology and supported an absence of sperm in the epididymis; “is there an increase in the number of germ cells undergoing apoptosis?”, in which case staining with for cleaved caspase-3 and −9, or using the TUNEL method is useful, e.g. O’Bryan et al. (2013); or “are the correct number of Sertoli cells present, and are they carrying the normal number of germ cells?”, where it will be informative to stain testis sections with germ and Sertoli cell markers (Table 2) for use in stereology to quantify the numbers of each cell type (see Garcia et al. 2014; Rebourcet et al. 2014). Similarly, germ cell makers may be useful to highlight particular germ cells types or structures, e.g. the use of an acetylated alpha-tubulin antibody to stain the axoneme can clearly highlight an absence, or reduction in number, of sperm tails. Similarly, for reduced EC, immune infiltration within the epididymis can be detected using haematoxylin and eosin staining and careful histological examination (see Carvajal et al. 2018; Hu et al. 2018).

3.3.6. Sperm capacitation

As indicated above, sperm capacitation is the final maturation process required to unleash the sperm’s full fertilization potential. While this normally occurs in the female reproductive tract, it can be induced by incubating sperm in chemically defined media. Incubating sperm in simple pro-capacitation factors e.g. bicarbonate and bovine serum albumin for 45-60 minutes in an atmosphere of 5% CO₂ will induce capacitation. To drive capacitation more prominently, other factors can be added: the phosphodiesterase inhibitor pentoxifylline and a cyclic-AMP analogue, or methyl-beta-cyclodextrin alone [Visconti et al. 1995a, 1999]). Establishment of capacitation is traditionally assessed by the presence of tyrosine phosphorylation (Visconti et al. 1995a) and while fertility can be established in the absence of typical tyrosine phosphorylation levels (Alvau et al. 2016) this hallmark is a readout of normal sperm function and, thus, the regulation of signaling pathways. Assessing tyrosine phosphorylation should be done using a validated phosphotyrosine antibody e.g. 4G10 and / or PT-66 (Sati et al. 2014; Visconti et al. 1999; Katen et al. 2016), which are used for sperm staining via immunofluorescence or immunoblotting of sperm protein extracts. Normal tyrosine phosphorylation levels should be measured over a specified time course of 1 hour under capacitation conducive conditions, relative to wildtype sperm (see Visconti et al. 1995a; Cotton et al. 2006).

When monitoring sperm capacitaton by immunofluorescent labeling, positive tyrosine phosphorylation in the sperm tail is recorded as a percentage of the population (Asquith et al. 2004; Sakkas et al. 2003). As the phosphorylation first appears within the principal piece then extends throughout the entirety of the tail, positive staining can be recorded as partial or complete (Asquith et al. 2004, Hu et al. 2018) and staining may also be apparent in the sperm head. If monitoring capacitation by western blotting, band intensities of principal phosphorylated proteins can be measured using densitometry analysis or total phosphorylation can be used, both utilizing the constitutively phosphorylated protein hexokinase (~130 kDa) as a loading control (Visconti et al. 1995b; Cotton et al. 2006; Katen et al. 2016). Alternatively, tyrosine phosphorylation can be investigated under non-capacitating conditions to ascertain precocious sperm capacitation (tyrosine phosphorylation) or acrosome reaction status (e.g. Inoue et al. 2003; Katen et al. 2016).

3.3.6.1. The acrosome reaction

The acrosome reaction normally occurs after capacitation and is triggered by binding of the sperm to the zona pellucida of an oocyte. The acrosome harbors hydrolytic enzymes that are intimately released via the acrosome reaction, required for digestion of the zona pellucida (Batiz et al. 2009; Santi et al. 2010). The acrosome reaction can be induced in a number of ways that vary in their ability to mimic the in vivo situation. Incubation of sperm with solubilized zona pellucida proteins is the gold standard approach, however, this is a challenging method (Leyton et al. 1989). An easier, but less physiological approach is to add progesterone to the capacitation media after 45 min of incubation (Aitken et al. 1995; Nixon et al. 2006; Reid et al. 2012). Another way of driving the acrosome reaction is to artificially force this response with the calcium ionophore A23187 (Tateno et al. 2013). The status of the acrosome can be visualized using lectin peanut agglutinin (PNA) staining (Kallajoki et al. 1986; Tao et al. 1993). Sperm can then be assessed for partial or full completion of the acrosome reaction, where a partial acrosome reaction is defined as when the acrosome is clearly disrupted but remains at least partially attached to the nucleus (see Lim et al. 2019).

3.3.6.2. Sperm hyperactivation

Hyperactivated sperm motility is an enhanced form of sperm motility that is required for fertility and allows sperm to reach and interact with the oocyte. During hyperactivation sperm motility shifts from low to high amplitude waveform and from a symmetrical to an asymmetrical tail beating pattern (Yaganimachi 1970). Hyperactivated motility is critically dependent on calcium influx into the sperm tail via the CatSper ion channel, which is present within the principal piece of the sperm tail (Ren et al. 2001). Although hyperactivation can be detected via CASA this is challenging using standard machine settings (reviewed by Suarez 2008). Efforts are being made to further optimize detection of hyperactivation using CASA systems and this is showing promise with systems such as CASAnova (Goodson et al. 2011). Alternatively, some researchers score hyperactivation by eye (Baker et al. 2006).

Hyperactivation is evident when sperm swim in figure-of-eight patterns in aqueous medium and can only be accurately observed when sperm are placed on a microscope slide without a coverslip. It is important to note that the processes regulating hyperactivation and capacitation can be uncoupled (Neill and Olds-Clarke 1987; Olds-Clarke 1989), and thus one does not always reflect the other. As indicated, defects in hyperactivation may be linked to altered ion channel function (Qi et al. 2007), which can be investigated using electrochemical techniques. Patch-clamping is a widely used technique to investigate ionic currents in somatic cells and in sperm involves recordings via attachment to the cytoplasmic droplet of the sperm (residual cytoplasm, see Kirichok et al. 2006). While this is a difficult method (Ren et al. 2001), patch-clamping has been used to define the function of numerous ion channels that affect sperm function in mice and humans (Kirichok et al., 2006; Qi et al. 2007; Lishko et al. 2013; Navarro et al. 2007). An alternative method to investigate ion channel regulation involves imaging the uptake of certain ions, e.g. calcium with Fluo4 (Rennhack et al. 2018).

3.3.6.3. Sperm-egg interaction, in vitro fertilization and intra-cytoplasmic sperm injection

Focusing on the male, the actual process of fertilization is dependent on sperm reaching the oocyte (progressive and hyperactivated motility), penetrating the zona pellucida (acrosome reaction and hyperactivated motility), binding and fusing with the oolemma (correct receptors present), and transferring an intact paternal genome (protamine-rich, condensed DNA) to the oocyte. In order to assess the effects of gene knockout / mutations on sperm function and thus fertilization, a variety of assays can be undertaken, which will be discussed below. These assays all require sperm to be capacitated to achieve interaction with the oocyte, where non-capacitated (negative) controls should be incorporated for comparison in all assays. In all assays, oocytes should be isolated from superovulated mice using standard injections with PMSG and hCG hormones (Inoue and Okabe 2008). Collectively these assays are the equivalent to the use of assisted reproductive technologies in humans.

3.3.6.3.1. Zona binding and sperm-egg fusion

As the first measure of sperm-egg interaction, capacitated sperm should be trialed for their ability to recognize and bind to the zona pellucidae of oocytes (Shur and Hall 1982). After isolation, oocytes should be denuded (cumulus cells are removed) and used immediately, or placed in a high salt medium (Yanagimachi et al. 1979). A set number of sperm should then be incubated with oocytes and allowed to interact for approximately 15-30 min (Asquith et al. 2004). After washing away any unattached sperm, the average number of sperm heads bound to at least 10 different oocytes per replicate should be counted. Compromised sperm-zona binding will be indicated by a reduced number of attached sperm.

Similarly, the ability of a sperm to bind and fuse its plasma membrane with the oolemma of the oocyte can be tested with a fusion assay (Inoue et al. 2005). To do this, freshly isolated, denuded (cumulus cells removed, hyaluronidase treated) and zona free oocytes (acid Tyrode’s treated or piezo removed) should be pre-loaded with DAPI or Hoechst stain and then incubated with sperm for 30 min. After washing and fixation in glutaraldehyde, the number of oocytes with fused sperm can be detected using a fluorescent microscope (see Inoue and Okabe 2008).

3.3.6.3.2. In vitro fertilization (IVF)

IVF is used to discriminate the true fertilization potential of sperm. IVF with mouse sperm involves incubating capacitated sperm with freshly isolated clutches of cumulus-oocyte complexes in HTF medium for a period of 4 hours. After washing potential zygotes and culturing them overnight in KSOM medium, the number of fertilized oocytes can be detected as 2-cell embryos after 24 hours (detailed protocol in Taft 2017). Extended analyses can assess the developmental timing of more mature embryos in comparison to wildtype (Yassine et al. 2015; reviewed by Kojima et al. 2014). Following development to the blastocyst stage, embryos may be transferred to pseudopregnant females to generate potential offspring (Nagy et al. 2006). Alternatively, embryos are flushed from the oviduct of superovulated females 8 hours post-plugging by a male and cultured in vitro, or collected up to 2 days post-plugging to score the number of 2-cell embryos (Nozawa et al. 2018; Shang et al. 2018).

3.3.6.3.3. Intracytoplasmic sperm injection (ICSI)

ICSI involves the careful injection of a sperm into an oocyte and tests for the ability of the sperm nucleus to generate offspring. It bypasses all of the pre-fertilization selection barriers and it is increasingly used in human medicine. Humans with a defect in this process will be detected as recurrent ICSI failure. The analysis of genetically modified mouse models has been used to inform clinical ICSI practice. One clear example of the value of using ICSI to define a phenotype was the Dpy19l2 knockout mouse, which had a globozoospermia phenotype and a loss of PLCZ1 (an oocyte activation protein). The use of ICSI revealed not only an oocyte activation phenotype but also that defects in pre-implantation embryo development were due to poor sperm DNA compaction and integrity (Yassine et al. 2015). Furthermore, this mouse model recapitulates human male infertility, where mutations in DPY19L2 are a major cause of globozoospermia in men (Harbuz et al. 2011; Koscinski et al. 2011) and PLCZ1 mutation prevents oocyte activation (Escoffier et al. 2016).

When investigating centriole gene mutants with mouse sperm it is important to remember that, in mice but not humans, the centrioles are thought to be dispensable for early embryonic divisions, after which they form de novo (Schatten et al. 1991; Howe and FitzHarris 2013). As such, ICSI in mouse only requires injection of the sperm head.

3.3.7. Sperm structural defects arising during spermiogenesis

Sperm head shaping

It will be clear at a light level if there are severe impacts to sperm shape (teratozoospermia), or rare forms of infertility such as globozoospermia (round-headed sperm; Coutton et al. 2012) or acephaly (decapitated sperm, Chemes et al. 1999). Light and electron microscopy will, however, also reveal abnormalities within the sperm head shape, including defects of the acrosome or axoneme structure (reviewed by Fawcett 1975). Abnormalities in acrosome formation include incorrect localization of the acrosome, its separation from the nuclear membrane, or absence, including globozoospermia (Dooher and Bennett 1977; Lerer-Goldshtein et al. 2010; Pierre et al. 2012). Such defects are likely controlled by perturbed Golgi function or improper development / attachment of the pro-acrosomal vesicles to the acroplaxome (described below). Acephaly is largely the result of impaired bridging between the basal plate and implantation fossa, as seen in Sun5 and Pmfbp1 null mice and humans (Elkhatib et al. 2017; Shang et al. 2017; Zhu et al. 2016; Zhu et al. 2018).

Shaping of the distal half of the spermatid head during spermiogenesis is largely driven by the manchette, a transient microtubule structure emanating from the perinuclear ring. For a review of its structure and summary of several mouse models containing manchette defects please see Lehti and Sironen (2016). By contrast, the apical half of the sperm head is shaped by actin rings within the enveloping Sertoli cells and the acroplaxome, which is an actin and keratin rich structure that exists between the acrosome and nuclear envelope (Kierszenbaum et al. 2003). Defects in acroplaxome formation will likely affect sperm head shape and globozoospermia as indicated above (Pierre et al. 2012). Of note, the dynamics of manchette movements, acrosome biogenesis and actin ring dynamics vary between human and mouse spermatogenesis, although the principle remains the same. As such, we recommend assistance from an expert in spermatogenesis when attempting to correlate the mouse and human genotype-phenotype relationship in this aspect of germ cell biology.

3.3.7.1. Sperm tail formation

If a mouse model contains immotile sperm, the most obvious structure to first check is the sperm tail, and specifically the axoneme, the engine of the sperm cell. The main structure of the axoneme (Figure 3) is comprised of 9 doublet microtubules that are arranged in a circular pattern and surround a central pair of singlet microtubules, in a “9 + 2” formation (Nagano 1963; Baccetti et al. 1981; Lindemann and Lesich 2010). Nexin protein links the outer microtubules together and maintains the ninefold structure by preventing uncontrolled sliding of microtubules. Additionally, both inner and outer dynein arms exist between the microtubules to modulate microtubule movement. While the inner arms principally control the maximal sliding velocity of the microtubules, the outer arms dictate beat frequency and bending patterns (Brokaw and Kamiya 1987). The axoneme is surrounded by the outer dense fibers and the fibrous sheath in the principal piece, and this sheath is further encased by mitochondria within the midpiece. For a review of sperm tail development and function please see Lehti and Sironen (2017) and Pleuger et al. (in press). The latter review includes examples of where mouse studies have confirmed, and informed, human causes of sperm tail immotility and teratozoospermia. Some examples include Armc2, Cfap43/44, Qrich2 and Dnah1 knockouts (Neesen et al. 2001; Ben Khelifa et al. 2014; Tang et al. 2017; Coutton et al. 2019; Kheraff et al. 2019; Shen et al. 2019).

There is a high degree of structural conservation between the axoneme in motile cilia, including the sperm tail, and the axoneme in primary cilia, which plays important roles in many somatic tissues (reviewed by Ware et al. 2011). Therefore, male infertility often exists as an element of broader ciliopathies including primary ciliary dyskinesia, which specifically affects motile cilia. It is therefore relevant to investigate the histology of other cilia containing tissues in mouse knockouts where the axoneme ultrastructure is clearly affected, as these mice may display other health defects and thus provide valuable information for patients (Sapiro et al. 2002). It will also be beneficial to investigate whether mice present with situs inversus, where the organ symmetry is flipped, as this is another common defect generated by defective cilia function (Chiani et al. 2019).

If the sperm midpiece presents as abnormal, or sperm have sluggish motility, it is worth investigating mitochondrial loading and function, as several gene knockouts can affect this process (Weidemann et al. 2016; Colley et al. 2013). These phenotypes can also result in changes to the length of the mitochondrial sheath (Mundy et al. 1995). Furthermore, data suggest that the organization of the sperm midpiece is regulated by the annulus, which contains septin proteins (Figure 3). The role of these proteins in spermatozoa is to develop a barrier between the midpiece and principal piece (Ihara et al. 2005; Kissel et al. 2005; Lhuillier et al. 2009). Electron microscopy will allow for the mitochondrial sheath length and morphology to be interrogated (e.g. Figure 3). This approach can be supplemented with confocal microscopy using sperm stained for TOM20 or with MitoTracker (Table 2) and localization of the annulus with septin 4 staining (Colley et al. 2013).

3.3.8. Analysing isolated germ cells

While not immediately relevant to the topic of this review, the analysis of specific germ cell populations may allow resarchers to define the mechanism of action underlying protein function and thus inform a clinical diagnosis. Classically, the STA-PUT method allows for reliable enrichment of germ cells from adult testes into fractions (Romrell et al. 1976). Among these fractions are those containing spermatocytes, round spermatids and / or elongating spermatids. Meanwhile, the testes of younger males, aged to postnatal day 6 or 8 allow for isolation of type A, or type A and B spermatogonia, respectively (Bellve et al. 1977). An updated, optimized, and step-by-step protocol has recently been published (Dunleavy et al. 2019). Researchers may also perform RNA or protein extraction on these cells, but should investigate the purity of each fraction first. Alternatively, fluorescence activated cell sorting and magnetic activated cell sorting have shown promise for isolation of germ cells, including spermatogonial stem cell isolation (Alipoor et al. 2009; He et al. 2012; Spiller et al. 2017).

Another notable method to allow an analysis of germ cell protein localization is the ‘squash prep method’. This involves careful isolation of short segments of seminiferous tubules, using transillumination and phase-contrast to stage germ cell development (see Kotaja et al. 2004). The tubules are then lightly squashed on a slide using a coverslip, allowing desired cell types to move out of the tubule fragments. After snap freezing in liquid nitrogen, the coverslip is removed, and cells can be fixed and stained in rapid succession. This method is beneficial as it is fast, however, it will require the accurate staging of tubules, which can be difficult in mouse models wherein spermiogenesis is perturbed. Conveniently, slides can be stored at −80°C, meaning that many slides can be generated from a single mutant or wildtype mouse. An extension of this method, known as ‘dry-downs’, involves teasing cells out of the tubule fragments, which can be fixed overnight in 1% paraformaldehyde and 0.15% Triton-X-100 in a humidified chamber. After washing in PBS and drying, the slides can be used for immunolocalization or can be stored at −80°C (Kotaja et al. 2004).

3.3.8.1. Defective meiosis (spermatocytes)

Meiotic arrest occurs in ~8% of men presenting with azoospermia (McLachlan et al. 2007). The analysis of a mouse model may allow research to precisely define the etiology of the defect. Meiotic spreads are commonly used to investigate meiosis I defects in mouse spermatocytes (Peters et al. 1997; Dia et al. 2017). Simplistically, this protocol involves isolating the seminiferous tubules and gently mincing them a sucrose solution. The tubules are then removed, while the remaining germ cells are spread over a slide by gentle rotation or pipetting. Finally, a drop of these cells is transferred onto a slide containing fixative. Slow drying of these cells in a humidified chamber allows for sufficient chromatin separation to stage meiosis (Dia et al. 2017). Spermatocytes isolated using this protocol can be visualized throughout leptotene, zygotene, pachytene and diplotene stages. Similarly, spermatocytes can be examined for differences in synaptonemal complex formation, DNA repair, and the processes of chromosome recombination and segregation can be investigated (reviewed by Handel and Schimenti 2010). As a recent example, meiotic spreads have been used to identify reduced levels of DNA repair in Spo11 mutant mouse spermatocytes (Carofiglio et al. 2018)

3.3.9. Assessment of male reproductive hormones

One of the most in depth set of knockout mouse models to investigate male fertility revolved around cell and tissue specific knockouts of particular hormone receptors (reviewed by Burns and Matzuk 2002; O’Shaughnessy 2014). These studies have informed not only human fertility, but several other aspects of human health (reviewed by Rana et al. 2014; reviewed by Kumar, 2005). Hormone levels can be measured in human and mouse serum (flowchart step #4) and collecting serum samples at a consistent time of the day will help to normalize any fluctuations induced by circadian rhythm (Lucas and Eleftheriou 1980). Reproductive hormones testosterone (T), follicle stimulating hormone (FSH), luteinizing hormone (LH), activin and inhibin should be assessed to provide a comprehensive view of hormone action. Furthermore, these hormones are commonly assayed in infertile men to deduce endocrinological causes of infertility and are thus largely comparable. Additional hormones including estrogen may also be beneficial to review. Using mass spectrometry to measure testosterone is recommended (Handelsman 2017), then moving to less sensitive methods such as ELISA; whereas LH, FSH, activins and inhibins can be reliably measured via ELISA (McNeilly et al. 2000; Kim et al. 2007; Ren et al. 2019; Wijayarathna et al. 2018; Walton et al. 2016). The accuracy of hormone assessment via ELISA relies heavily on the specificity of the antibodies in the assay and thus it is important to source validated antibodies / assays when conducting these analyses. Alternatively, specialized core facilities for hormone analyses exist across institutions worldwide that may also be of use. Mouse to mouse variation in T levels is high compared to humans, meaning that a large number of individuals will be required to reach statistical significance. Alternatively, it may be beneficial to stimulate maximal T levels in mice by hCG challenge (Lau et al. 1978; Rebourcet et al. 2016; Cobice et al. 2016), which will likely cut down on the number of animals required to reach significance. This method also allows for an assessment of the relative stimulatory effect of LH compared to that occurring in wild type mice. As a secondary measure for T, seminal vesicles can be dissected and weighed (Robson 1951; Jean-Faucher et al. 1978). Recently, a method of spectral imaging has been reported for detection of T specifically within the testis (Cobice et al. 2016), which may be of utility. For a summary of the roles and consequences of altered hormonal regulation on male fertility in the mouse readers are referred to reviews (O’Hara and Smith 2015; Kaprara and Huhtaniemi 2018).

Conclusions

While undoubtedly the majority of genetic causes of human male infertility remain to be determined, the advent of affordable exome sequencing, the formation of consortia focused on addressing this problem and updated methods to efficiently generate mutant mouse models suggests that the field is set to rapidly accelerate. The aim of this review was to assist in 1) the ascertainment of high confidence genetic causes of human male infertility and 2) validation of these variants using mouse models. Careful planning and following our pipeline will assist in maximizing efficiency and sample collection. Combining breeding experiments, histology and functional assays is the best approach to assess whether the mutant mouse recapitulates what is shown in the infertile man / men. It is also hoped that the analysis of such mouse models will assist in defining the mechanism of action of such genes and thus ultimately the development of therapies or, on the other hand, novel contraceptive strategies. Equally, the knowledge generated through the comprehensive phenotyping of animal models will inform the clinical management of the affected patient, and, in the longer run, a realistic prediction of how / if the use of genetically compromised gametes will affect the health of children conceived using assisted reproductive technologies.