Main

Despite advances in assisted reproductive technologies, infertility is a major health problem worldwide. Approximately 15% of couples are unable to conceive within one year of unprotected intercourse. The fertility potential of a couple is dependent on the coordinated and combined functions of both male and female reproductive systems. Anatomic defects, gametogenesis dysfunction, endocrinopathies, immunologic problems, ejaculatory failure and environmental exposures are significant causes of infertility. Although several infertility disorders are associated with defined genetic syndromes (for example, cystic fibrosis and Turner's Syndrome1,2), almost a quarter of clinical infertility cases of either sex are idiopathic in nature, in part as a result of a poor understanding of the basic mechanisms regulating fertility. It is thought that genetic defects underlie many of these unrecognized pathologies. On the basis of over 200 infertile or subfertile genetic mouse models (see Supplementary Information Table; also see ref. 3), it is not surprising that the diagnosis of idiopathic infertility is common in the clinic4,5.

In this review, we discuss causes of mammalian infertility with an emphasis on the genetic basis of fertility defects in humans and mice. Animal models have defined key signalling pathways and proteins involved in reproductive physiology6. Mouse models have been produced by spontaneous mutations, fortuitous transgene integration, retroviral infection of embryonic stem cells, ethylnitrosurea (ENU) mutagenesis and gene targeting technologies3,7,8. These mutations affect all aspects of reproduction, including ovarian development and function, testis determination, spermatogenesis, sperm function, genital tract development and function, sexual behaviour, fertilization and early embryonic development, and therefore have contributed much to our understanding of infertility. For example, male infertility is observed in the spontaneous mutant models hypogonadal (hpg)9 and testicular feminization (tfm)10 and in models created by transgene integration, such as the kisimo mouse model (which arose by transgene disruption of the Theg gene11) and retroviral disruption of the Bclw12, Mtap13 and Spnr14 genes. These models are improving our knowledge of the genetic basis of mammalian infertility and suggest that in the future, clinical technologies must advance to enable analysis of many more genes when an infertile couple enters the clinic. Currently, karyotype analysis, sequence analysis of the cystic fibrosis transmembrane conductance regulator gene and Y chromosome deletion analysis (for males) are the only genetic tests commonly offered to infertile patients4,5.

Where it all begins

Reproductive development and physiology are evolutionarily conserved processes across eutherian mammalian species and many other vertebrates, including marsupials15, amphibians, reptiles, birds and fish16,17,18,19. Several genes required for vertebrate fertility are also highly conserved in evolution, with orthologues in Drosophila melanogaster (for example, vasa (DDX4), fat facets (DFFRY) and boule (DAZ)20,21,22). Propagation of the vertebrate germline requires development of the gonads, the site of future gamete production. The indifferent gonad forms during foetal development, primordial germ cells enter the gonad primordium and the tissue eventually differentiates along a female (ovarian) or male (testicular) pathway; this differentiation dictates the formation of the secondary sex organs19. Although there may be spatiotemporal variations of these processes in different species (for example, in mice and humans, gonadal sex determination occurs in utero, whereas in marsupial mammals, it occurs after birth), they eventually yield ovaries that produce eggs or testes that generate spermatozoa.

Defects in sexual differentiation pathways can cause infertility in mice and humans of both sexes (Fig. 1)23,24,25. In 1959, through the analysis of human XO (Turner syndrome) females and XXY (Kleinfelter syndrome) males, as well as XO and XX female mice and XY male mice, it was concluded that the Y chromosome was male determining26,27,28. Subsequent chromosomal and genetic studies of humans and mice with sex reversal syndromes and infertility revealed that many XX males have translocations of a small piece of the Y chromosome, so that the sex-determining region Y gene (SRY) results in testis development. Similarly, XY females often have inactivating mutations in the SRY gene, resulting in the development of ovaries29,30. A critical role of Sry in sex determination was confirmed by showing that the expression of an Sry transgene in an XX mouse causes testis formation, and physical and behavioural sex reversal31. Most SRY mutations disrupt the high mobility group (HMG) box of the SRY protein; not surprisingly, this region is highly conserved among different species32. HMG box-containing proteins typically bind and significantly bend DNA and function as transcription factors or facilitators of transcription. Several genes upstream and downstream of SRY in the sex determination pathway are now known (reviewed in ref. 23). For example, XY female sex reversal correlates with a duplication of the human X-linked gene DAX1 (ref. 33) or haploinsufficiency of the autosomal SOX9 gene32,34,35,36. Interestingly, whereas an extra Y chromosome (that is, XYY) has little effect on human male fertility because of the selected loss of the extra Y during spermatogenesis37, Klinefelter (XXY–XXXXY) males account for 10–15% of azoospermic patients38.

Figure 1: Genetic aetiologies of human male infertility.
figure 1

Developmental disorders causing human male infertility result from a failure of gonadal development or testis determination, endocrinopathies, well-known genetic syndromes, and numerical and structural chromosomal abnormalities (translocations, deletions and inversions).

Full size image

From a distance, they will come

In both sexes, the primordial germ cells (PGCs) are defined histologically as alkaline-phosphatase-positive embryonic cells39,40. In the mouse, these cells divide rapidly under the influence of transforming growth factor-β (TGF-β) superfamily signals; knockout models lacking bone morphogenetic protein-4 (BMP-4) or BMP-8b, or the downstream cytoplasmic-to-nuclear relay proteins, SMAD1 and SMAD5, have defects in PGC development41,42,43,44. At mid-gestation, the PGCs begin one of the longest journeys of any mammalian cell, migrating from their origin at the base of the yolk sac, along the hind-gut, to eventually enter the genital ridge. Factors required for this migration in humans are unknown, although chemoattractants and cell adhesion factors have been implicated45. In the mouse, mutations in Kit receptor (KITR) and Kit ligand (KITL) genes block PGC migration, causing infertility, but not altering sexual differentiation46.

Few known human mutations result in a reduction of the PGC or follicle pool, although girls with Turner's syndrome (partial or complete X-chromosome monosomy), have streak (remnant) gonads with no oocytes. Many Turner's syndrome cases with ovarian failure seem to be caused by loss of the short arm of the X-chromosome2. Among the candidate Turner's syndrome ovarian failure genes are ZFX, BMP15, UBE1 and USP9X (ref. 2). An absence of Zfx in mice results in a loss of germ cells and subfertility in both sexes47; loss of BMP15 in sheep causes a block at the primary follicle stage and infertility48. Studies in XO mice suggest that abnormal chromosomal segregation contributes to germ cell problems49, indicating that multiple factors are responsible for these human ovarian abnormalities.

Death in the germline

In females, quiescent primordial follicles (a non-growing oocyte surrounded by squamous granulosa cells) form during prenatal life in humans and post-natally in mice. Recruitment of these follicles during ovarian folliculogenesis permits further growth and development of oocytes. In males, spermatogenesis is characterized by three specific functional phases: proliferation, meiosis and spermiogenesis. The proliferative phase in the testis begins early in embryonic development, and with the exception of a brief period when spermatogonia arrest during late foetal and early postnatal life, they proliferate actively throughout life. Spermatogonial stem cells were one of the first recognized examples of adult stem cells capable of rejuvenating spermatogenesis after toxic insult50,51. In contrast, the formation of primordial follicles in females defines a finite endowment of oocytes. Between the time of ovary development and reproductive sequence, there is a precipitous drop in the number of oocytes. In humans, seven million foetal germ cells at 20 weeks are reduced to two million oocytes at birth, and eventually to 300,000 at puberty52,53. Thus, factors that prenatally and postnatally regulate germ cell survival in the ovary can prolong the reproductive lifespan.

The spermatogonia proliferation rate, the highest in the body, is well regulated; thus, it is not surprising that genes involved in growth (for example, Kit, Csf and Bmp8b) and apoptosis are also required for normal spermatogonia (Fig. 2). This stage of spermatogenesis is also noteworthy for its inefficiency; in rats, 75% of spermatogonia do not survive to become mature sperm54. A balance of anti-apoptotic members of the BCL2 family (that is, BCL2, BCL6, BCLX, and BCLW) and the pro-apoptotic BAX protein is extremely important in the regulation of germ cell survival prenatally and postnatally in both sexes, and in response to toxins in the ovary55,56,57. It is possible that defects in this delicate balance of cell proliferation and cell death contribute to the clinical pathology of hypospermatogenesis (all cellular elements of the testis are present, but at low cellularity). In the mouse, an absence of BCLX results in a complete loss of germ cells before birth in both males and females; furthermore, a lack of BCLW results in a partial reduction of PGCs in females, whereas an absence of BCL2 results in only decreased oocyte numbers postnatally12,58,59. Absence of either BAX or BCLW causes male infertility, and absence of BCL6 causes male subfertility, again suggesting that a balance of apoptotic/anti-apoptotic factors is necessary for normal spermatogenesis. In contrast, a prolonged female reproductive lifespan occurs in the absence of BAX in mice, consistent with its pro-apoptotic role60. Polycyclic aromatic hydrocarbons in cigarette smoke and air pollution bind to the aryl hydrocarbon receptor to stimulate transcriptional activation of Bax, thereby enhancing apoptosis and oocyte loss61. In the future, factors that inhibit the BAX pathways or stimulate the anti-apoptotic pathways could prolong the reproductive lifespans of women.

Figure 2: Genes involved in the regulation of male reproduction in the mouse.
figure 2

Spermatogenesis requires a complex interaction of the various cellular compartments of the testis (seminiferous epithelium containing spermatogenic cells, Sertoli cells and peritubular myoid cells, the interstitial cell compartment containing the steroidogenic Leydig cells, macrophages, and other interstitial cells, and the vasculature). Targeted mutation of the genes shown affects specific testicular cell types and reproductive function, resulting in male infertility or subfertility in the mouse (detailed in the Supplementary Information Table).

Full size image

Meiosis, recombination, the integrity of the genome and more death

Meiosis is a process of cell division that is unique to germ cells and is required for the production of healthy haploid gametes (reviewed in greater detail in ref. 62). This process is evolutionarily important for both the integrity and diversity of species, as recombination of homologous chromosomes occurs during prophase of the first meiotic division, helping to orient chromosomes on the meiotic spindle, as well as introducing genetic variability. Male spermatogenesis is initiated postnatally (in mice at postnatal day 7) and is a continuous process producing spermatozoa. Proliferating spermatogonia differentiate and enter meiosis as spermatocytes. In contrast, oogenesis is initiated prenatally (in mice at embryonic day 13), arrests initially at the diplotene stage of meiotic prophase, resumes with the preovulatory luteinizing hormone (LH) surge and arrests again after the first polar body is released before fertilization.

Despite sexual dimorphism in meiosis, many regulators of the process are common to the germ cells of both sexes. In the absence of these proteins, prophase arrest and accompanying germ cell death occur in male and female germ cells. Infertility in both sexes is observed in knockout mice lacking the recombination and DNA damage/mismatch repair proteins, SPO11, DMC1, ataxia telangiectasia (ATM), MSH4, MLH1, and MSH5 (refs 63–74). Mutations in ATM and Fanconi anaemia (FA) complementation-group-protein genes result in fertility defects in humans and mice of both sexes (Figs 1,2). ATM is involved in DNA metabolism and cell cycle checkpoint control75, whereas FA is a hereditary chromosomal instability syndrome76. FA men are hypogonadal, oligospermic and rarely fertile; FA women can experience premature ovarian failure in their 20s. Several FA mouse models have been created and display reduced fertility77. Thus, similar mechanisms for germline monitoring are conserved in mammals and in both sexes.

When the germline 'proofreading' system goes awry in an otherwise 'normal' individual, there are major consequences. Despite a normal somatic karyotype, sperm collected from oligospermic men exhibit an increased frequency of chromosomal abnormalities78,79. Aneuploidy is the most common genetic abnormality in humans80, and the common trisomies (for example, trisomy 21 (Down's syndrome and trisomy 18)) arise primarily in the children of ageing women through non-disjunction defects during the first meiotic division81. These findings are exemplified in mice lacking synaptonemal complex protein 3 (SYCP3), which functions in synapsis (pairing) of the homologous chromosomes during meiosis. Sycp3 knockout males are infertile; females are subfertile, exhibiting loss of aneuploid embryos82,83. Interestingly, germline deletions resulting in Duchenne muscular dystrophy (DMD) more often arise during oogenesis, whereas DMD point mutations result more commonly from spermatogenic failure84. This suggests that some proofreading mechanisms during male and female gametogenesis may differ (see also ref. 80).

Hormones take control

After sexual maturity, all stages of spermatogenesis (male) and folliculogenesis (female) are observed, the end result in each case being gamete production. Hypothalamic pituitary control of gonadal somatic cells is critical for fertility in all mammals and in both sexes (Fig. 3; reviewed in refs 85–88). Gonadotropin releasing hormone (GnRH) from the hypothalamus regulates the pituitary gonadotrope production of follicle stimulating hormone (FSH) and LH, α–β heterodimers that share a common α subunit with placental human chorionic gonadotropin (hCG) and pituitary thyroid stimulating hormone (TSH). Spontaneous deletion of the hypothalamic GnRH-encoding sequences in mice (that is, hpg), mutation of the human GnRH processing enzyme gene (PC1), disruption of developmental migration of the GnRH neurons in human Kallman Syndrome, or mutation of the GnRH receptor gene (expressed in gonadotropes), results in hypogonadotropic hypogonadism (HHG) and infertility (Fig. 1). Loss-of-function mutations in the pituitary-expressed FSHβ genes and gonadal-expressed FSH receptor genes decrease testis size and spermatozoa counts in men and male mice, and cause a block in folliculogenesis and infertility in women and female mice. This emphasizes the conservation and importance of these signalling pathways. Similarly, pituitary gland development and downstream steroidogenic pathways are conserved in humans and mice and are critical for fertility in both sexes. For example, a homozygous Prop1 missense mutation causes multiple pituitary defects in the Ames dwarf mouse, including defects in gonadotrope differentiation and infertility in all female and most male mice. Similarly, human PROP1 mutations cause combined pituitary hormone deficiency, including HHG and infertility (Fig. 1).

Figure 3: Spermatogenic failure in the human.
figure 3

Spermatogenesis in the human is characterized by six stages that are present in a mosaic fashion in the seminiferous tubule. a, Normal human spermatogenesis with Sertoli cells (SC) spermatogonia (Sg) towards the basal portion of the tubule, spermatocytes (Scyte) and maturing spermatids (Stid) located towards the lumen of the tubule. The tubules are surrounded by peritubular myoid cells (PT). The interstitial area contains the steroidogenic Leydig cells that secrete testosterone. b, An example of the most severe testicular pathology, with a total absence of germ cells and a Sertoli-cell-only pathology. Mild thickening of the peritubular layer is also observed (arrows; peritubular fibrosis). c, A late maturation arrest. The most mature cell type present is the round spermatid. d, An example of hypospermatogenesis, where all cell types are present in the testis, but with a low level of cellularity within the seminiferous tubule.

Full size image

Members of the steroid receptor superfamily and their transcriptional coactivators (for example, AR, ER, PR, RXRβ, SF1, DAX1, and SRC1) are pivotal in the regulation of reproductive function. Disruption of any of the genes involved in androgen biosynthesis, metabolism and action negatively impact male development, spermatogenesis and function. Spontaneous mutations of the X-linked androgen receptor gene in XY mice (that is, tfm (testicular feminization)) and humans result in individuals with abnormal testes, no ductal system and external female genitalia10,89 (http://ww2.mcgill.ca/androgendb). Absence of steroid 5α reductase, which converts testosterone to dihydrotestosterone, results in external female genitalia, prostate absence in XY humans and developmental disruption of the male ductal system (that is, seminal vesicles and prostate)90,91. Similarly, mutations in the orphan nuclear receptors, steroidogenic factor-1 (SF1) and DAX1, have been described in mice and humans; mutations in DAX1 cause almost universal HHG in adult humans (Fig. 1).

Not surprisingly, oestrogen and progesterone are key to early folliculogenesis and corpus luteum maintenance of early pregnancy in the female92. Targeted deletion of oestrogen receptor α in mice revealed that it is also required for male fertility and for male and female sexual behaviour93. Similarly, the progesterone receptor (also required for female fertility) is important in sexual behaviour in the mouse94. In the evaluation of the infertile couple, assessment of circulating hormone levels (FSH, LH, testosterone, prolactin and free testosterone in the male; FSH, LH, oestradiol and progesterone levels in the female) can provide important information concerning the function of the hypothalamic–pituitary–gonadal axis and the presence of endocrinopathies.

Spermatogenesis has many unique players

Spermatogenesis requires not only the appropriate hormonal milieu, but also autocrine, paracrine and juxtacrine signalling between the various testicular compartments. The testis is composed of an interstitial cell compartment with androgen-producing Leydig cells, and the seminiferous tubule containing Sertoli cells, peritubular myoid cells and germ cells. Whereas follicles are recruited each cycle to enter the ovulatory pathway in females, all stages of spermatogenesis are present at any one time in different tubules within the testis. Thus, the wave of spermatogenesis resulting in development of mature sperm is a spatial cycle rather than a temporal one. The importance of growth factors and cytokines, their receptors and signal transduction pathways to gametogenesis cannot be underestimated. For example, deletion of mouse Desert hedgehog (Dhh) affects testicular development, resulting in anastomatic seminiferous tubules and an absence of adult Leydig cells. Similarly, the insulin-like growth factor (Igf1) null male mouse is characterized by vestigal vas deferens, prostate and seminal vesicles, caused by a steroidogenic Leydig cell defect.

Once male germ cells complete meiosis to achieve a haploid chromosomal complement, they are called spermatids. Spermatids undergo a process of cellular differentiation known as spermiogenesis, progressing from round to elongating to elongated spermatids, culminating in the development of spermatozoa. Many male-specific genes are involved in this extensive cellular remodelling and concomitant condensation of the chromatin (for example, Tnp1, Tnp2, Prm1, Prm2, Theg and Hsp60-2; see Fig. 2). In common with many of the Y chromosome genes that encode RNA-binding proteins and are implicated in human infertility, an absence of similar proteins in the mouse, such as STYX, disrupts spermatid development95. After spermiogenesis and release of the spermatozoa from the Sertoli cells into the seminiferous tubule lumen, acquisition of motility occurs during transit through the epididymis and capacitation occurs in the oviduct (fallopian tubes) of the female genital tract. Both of these processes are required for effective penetration of the zona pellucida and egg.

In the evaluation of the infertile male, a semen sample is ordered to determine sperm count, motility and morphology. Some laboratories perform sperm function tests that predict defects in sperm–zona or egg interaction, or in penetration. However, semen analysis is not a definitive test of the fertility potential of an individual unless there are no sperm in the ejaculate. This is also true in mice. For example, FSH-β mutant mice exhibit reduced sperm counts, but fertility is normal96. Conversely, many mouse models are infertile and demonstrate abnormal sperm function, sperm motility (for example, ApoB, CatSper, Dnahc1, Hmgb2 and Ros knockout mice), or morphology (for example, Tnp2, Tnp1 and Sperm1 knockout mice) with no detrimental effect on sperm count. In addition, targeted deletion of the Ace, Adam2, Adam3, calmegin, Pc4, Spam1, Spnr, Trg26, Jdf2 or Mdhc7 genes results in normal sperm count, motility and morphology; however, sperm function is defective (Fig. 2). As a majority of unexplained cases of infertility in human males result from spermatogenic defects (Fig. 3), the homologues of the above-described mouse genes are actively being pursued for their possible roles in human infertility.

Chromosomal abnormalities are observed in 5.8% of infertile males97 and more commonly involve sex chromosomes (4.2%), as opposed to autosomes (1.5%; Fig. 1). In addition to SRY, other Y chromosome genes are required for spermatogenesis. This became obvious in the XX Sxr male mouse and the XX Sry transgene-positive male mouse, which are sex-reversed, but display spermatogenesis blocks31. Similarly, a region at Yq11 that is deleted in several infertile men was termed the azoospermia factor (AZF) region98. In general, the reported incidence of deletions in this region in severe oligospermic/azoospermicmen is 10–18% and varies depending on the stringency of diagnostic classification99. This region (now further subdivided into AZFa, AZFb and AZFc), contains several genes involved in spermatogenesis, including deleted in azoospermia (DAZ)100,101. In the mouse, disruption of the testis-expressed Dazla homologue gene on chromosome 17 abrogates gamete production. Other putative evolutionarily conserved spermatogenesis genes have been mapped to Y chromosome regions commonly deleted in infertile men (reviewed in ref. 99). Mutations in the human gene USP9Y (ubiquitin-specific protease 9, Y chromosome or DFFRY), a homologue of the D. melanogaster development Fat facets gene21, cause infertility102. Functional analysis of additional Y chromosome genes in the mouse has been complicated by the presence of multiple copies or X-chromosome homologues, as well as technical difficulties related to the low efficiency of Y chromosome homologous recombination in embryonic stem cells. However, it is expected that the recent and exciting technological breakthroughs achieved by Bishop and colleagues103, who developed a method for successful gene targeting of the Y chromosome in embryonic stem cells, and the development of an embryonic stem cell line104 that will facilitate germ line transmission of Y chromosome targeted genes, will rapidly translate into an enhanced understanding of the role of specific Y chromosome genes in male reproductive function.

No crosstalk in females, no folliculogenesis progression

Although several proteins are involved in ovarian folliculogenesis, meiosis, and oocyte survival, oocyte–somatic cell crosstalk is especially critical for release of a fertilizable egg (Fig. 4 and ref. 105). Without the helix-loop-helix protein factor in the germline α (FIGα), pre-granulosa (somatic) cells fail to form a monolayer around individual primordial oocytes, resulting in rapid germ cell depletion from the neonatal mouse ovary and sexual infantilism106. Similarly, oocyte growth during folliculogenesis is regulated by signalling of granulosa KITL to the oocyte-expressed KITR46. KITL expression is controlled by both hormonal (FSH) and oocyte (growth differentiation factor-9 (GDF-9)) factors107. In the absence of the TGF-β superfamily ligand GDF-9 in mice108 or its close oocyte-specific homologue, BMP-15, in sheep48, an arrest in folliculogenesis at the primary follicle stage is observed (Fig. 5). FSH has no effect on the 'arrested' primary follicles of Gdf9 knockout ovaries108, suggesting that GDF-9 allows the granulosa cells to grow and acquire the competence to respond to FSH. Absence of GDF-9 results in elevated levels of KITL109, which signals back to markedly increase oocyte size108. These findings were confirmed by studies showing that recombinant GDF-9 downregulates levels of Kitl mRNA107. In addition to oocyte factors, FSH96 functions with IGF-1 (by stimulating cyclin D2 and oestrogen synthesis)110,111 to regulate the growth of the follicle through the pre-ovulatory stage (Fig. 5). In pre-ovulatory follicles, LH, in conjunction with the oocyte-secreted proteins GDF-9 and BMP-15, signals to somatic cells to initiate ovulation of a healthy cumulus-oocyte complex (COC). Thus, important crosstalk between somatic cells and oocytes, as well as endocrine signalling, is necessary for normal folliculogenesis and ovulation.

Figure 4: Female fertility proteins.
figure 4

Knockout mouse models have defined key proteins that function at various stages of follicle formation, folliculogenesis, ovulation, and post-ovulatory events. FIGα is required for primordial follicle formation, and several proteins are needed for oocyte and granulosa cell (GC) growth and differentiation, ovulation, and the integrity of the cumulus oocyte complex (COC) (reviewed in the Supplementary Information Table).

Full size image
Figure 5: Mouse knockout models to study folliculogenesis.
figure 5

a, Targeted mutation of the oocyte-secreted growth factor, Gdf9, results in an early folliculogenesis block, resulting in an ovary with only primordial (P) and primary (1) follicles108. b, Absence of the endocrine hormone, FSH, results in a later block at the secondary (2) to antral follicle transition96. c, These knockout models contrast with wild-type ovaries that contain pre-ovulatory (PO) follicles and corpora lutea (CL). Primary- to preovulatory-stage oocytes are surrounded by a zona pellucida (magenta).

Full size image

Post-fertilization, Mater (maternal antigen that embryos require) and several other genes (including Dnmt1o, Pms2 and Hsf1) (Fig. 4), have been identified by knockout mouse studies as maternal effect (oocyte-synthesized) genes that are essential for development112. The human homologue of Mater has been identified and may be a candidate gene for premature ovarian failure113. Similarly, several uterine proteins are required for implantation (Fig. 4 and ref. 114). Thus, these studies have pinpointed multiple putative diagnostic targets in women who present with infertility.

In women, several syndromes — including ovarian failure and infertility — are attributed to autosomal recessive mutations88,115. Blepharophimosis/ptosis/epicanthus inversus syndrome, the only autosomal dominant disorder associated with premature ovarian failure (POF), is caused by mutations in the forkhead transcription factor gene (FOXL2)116. Expansion of a CGG trinucleotide repeat of the Xq27.3 fragile X mental retardation gene (FMR1) to over 200 repeats is the most common heritable cause of mental retardation. The unstable premutation FMR1 allele (60–199 CGG repeats) causes POF in 21% of heterozygote carriers and increased twin pregnancies117. Furthermore, 2% of sporadic cases and 14% of familial cases of POF are associated with the premutation allele. The pathophysiology of the premutation allele in POF is unknown, but this finding clearly represents a step forward in identifying a genetic locus for POF. To date, all other identified single gene autosomal dominant or recessive mutations with isolated infertility in humans affect steroidogenic or gonadotropin pathways, often in both sexes. However, many candidate genes await analysis in human idiopathic infertility cases.

Descent of the testis and problems with sperm transit

Testis determination and gametogenesis are necessary, but not sufficient, for male fertility, as testicular descent down the inguinal canal into the scrotum, in addition to the development of the genital tract and penis, are also critical. Mutations of the mouse genes Insl3, Great (G-protein coupled receptor that affects testicular descent; a possible relaxin receptor) and Hoxa10 (refs 118–122) result in male infertility secondary to cryptorchidism. The second phase of testicular descent requires androgens and a functional androgen receptor. In humans, cryptorchidism results from anti-Mullerian hormone (AMH) deficiency caused by obstruction of the genital tract.

Gonadal sex determines secondary duct differentiation. In females, the Müllerian duct differentiates into the oviducts, uterus and upper portion of the vagina; in males, the Wolffian duct differentiates into epididymis, vas deferens and seminal vesicles19,23,123. The Müllerian duct regresses in response to prenatal production of testicular AMH, and Wolffian duct development requires testosterone. Differentiation of the prostate and male external genitalia is driven by dihydrotestosterone, a product of the conversion of testosterone by 5-α reductase. Mutations in genes that affect steroidogenesis (for example, P450 aromatase (CYP19) and 5-α reductase) and steroid signalling pathways (for example, oestrogen receptor α (ERα) and androgen receptor) have deleterious effects on genital tract development and function in the male. Thus, it follows that pseudohermaphroditism occurs as a result of defects in genes involved in gonad formation (for example, SF1 and WT1). Mutations of the AMH or AMH receptor genes result in persistence of Müllerian duct syndrome (PMDS), resulting in obstructive azoospermia and fertility defects in men, male dogs and mice (Fig. 1 and ref. 123).

One to two per cent of infertile men present with obstructive azoospermia caused by congenital bilateral absence of the vas deferens (CBAVD), as a result of mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene1,124. These CBAVD patients successfully father offspring because microsurgical epididymal sperm aspiration yields 'normal' sperm for in vitro fertilization (IVF). Male fertility also may be compromised by epididymal, ejaculatory or erectile dysfunction, as well as by other congenital anomalies.

New technologies and perspectives

Genome, gene and cDNA sequences are being deposited into public databases (for example, the National Center for Biotechnology Information (NCBI; http://www.ncbi.nlm.nih.gov) or the Wellcome Trust Sanger Institute (http://www.sanger.ac.uk)) with amazing speed. Furthermore, programs to search these databases, such as BLAST (http://www.ncbi.nlm.nih.gov/blast) and the Unigene database at NCBI, are helping scientists to use this wealth of information. In particular, sequences unique to mammalian germ cells have been identified using an in silico subtraction (electronic database) approach125. For example, GASZ (germ-cell-specific and ankyrin repeat, sterile α-motif and basic leucine-zipper-containing protein) was identified as a novel evolutionarily-conserved germ cell-expressed gene lying adjacent to the CFTR gene in human, chimpanzee, baboon, cow, rat and mouse126. Functional expression and sequence data is also being collated into collections, such as the Ovarian Kaleidoscope database (http://ovary.stanford.edu), the Male Reproductive Genetics database (http://mrg.genetics.washington.edu/home.html) and the GermOnline database (http://germonline.igh.cnrs.fr). With the use of microarrays for expression analysis of reproductive tissues127, these 'bits' of data will increase exponentially. Therefore, there is an urgent need for bioinformatics advances to facilitate compiling and sorting through this wealth of in silico information for future applications in the clinic.

Technological procedures and advances in the clinic are also wrought with some controversies and dilemmas. Of particular importance, the application of assisted reproductive technologies (ART) for severe male and female factor infertility serves to not only overcome sterility, but bypasses natural barriers to the inheritance of defective genes. This results in considerable concern that genetic defects will be transmitted to the next generation. Among these controversial treatments are ICSI (intracytoplasmic sperm injection), cytoplasmic (ooplasmic) transfer, round spermatid nuclear injection (ROSNI or ROSI) and reproductive cloning (described in more detail by Schatten in this issue128).

These technologies have prompted significant debates concerning their morality and safety. However, even a United States government moratorium on human cloning has not deterred renegade scientists overseas from actively engaging in this research, which could result in a potentially disastrous outcome.

Conclusions

To date, diagnosis of infertility in the clinic has been hindered by our relatively poor understanding of the underlying molecular mechanisms. Although investigators have attempted to translate findings in animal models to humans by searching for gene mutations/deletions in idiopathic infertility patients, in general, these investigations have not been fruitful. Given the large number of candidate evolutionarily conserved 'fertility' genes yet to be discovered or identified from mutant mouse studies, and the overall complexity of the reproductive system in general, proper diagnosis and treatment of these patients will await the development of more sophisticated and rapid technologies. Finally, if mutation of a gene in mice or humans results in infertility, the protein product of that gene may be a future target for novel contraceptives that are designed to transiently or permanently cause infertility.