Background
Testicular development of the initially bipotential gonad is directed by a signalling cascade that promotes the male pathway, while simultaneously antagonizing female signalling factors. Following the initial differentiation of the supporting cell lineage towards Sertoli cells, paracrine factors and steroid hormones secreted by the fetal Sertoli and Leydig cells contribute to the continued process of testicular development and masculinization of the fetus (reviewed in [
1‐
3]). These processes have not yet been characterized in detail in human fetal development and although information from animal models provides essential insight there are important differences, particularly in relation to germ cell development and regulation of meiosis (reviewed in [
2,
4]). Testosterone produced by the fetal Leydig cells plays a pivotal role as the main driver of fetal masculinization, although increasing evidence suggests that androgen precursors produced in the adrenal, liver and placenta via “backdoor” and 11-oxygenated steroidogenic pathways also contribute to the overall production of androgens in human fetuses as well as in the activation and masculinization of secondary sex characteristics [
5‐
7].
The importance of sufficient androgen exposure during fetal development is evident from the clinical observations that male reproductive disorders included in the testicular dysgenesis syndrome (TDS) are most likely caused by subtle deficiencies in androgen production or action in the testis during fetal life [
8,
9]. TDS is comprised of disorders that manifest either at birth (cryptorchidism, hypospadias) or in young adulthood (low sperm count, testicular germ cell cancer, primary hypogonadism) [
8,
10,
11]. The fetal origin of cryptorchidism and hypospadias is intuitive, but the suggested fetal origin of adult-onset disorders, such as low sperm count, and hypogonadism was initially based on the finding of focal dysgenesis in the testes of the majority of adult men with TDS. The morphological alterations include abnormally shaped seminiferous tubules, Leydig cell nodules and Sertoli cell only (SCO) tubules, in which the Sertoli cells are occasionally visibly in the undifferentiated state [
10‐
13], thereby suggesting abnormal early development of the testis. The resulting slightly impaired function of the somatic niche in the human fetal testis may disrupt the differentiation of germ cells and thus lead to the presence of arrested gonocytes which are the precursor cells of testicular germ cell cancer [
8,
10]. While severe impairment of early testis development can result in overt dysgenesis and ensuing downstream effects as described in patients with differences of sex development [
2,
13], the primary focus of the TDS hypothesis includes men in whom the dysgenetic changes are focal, often in an otherwise largely normal testis [
13].
Normal masculinization of the fetus is dependent on testosterone production in the developing testes, and based on evidence from animal studies, this likely occurs during a specific sensitive period. The masculinization programming window (MPW) was discovered in rats and refers to a window of time during fetal life in which androgen action programs later development of male reproductive organs, including their adult size and function [
14]. Reduced fetal testis testosterone levels during MPW induce focal dysgenesis in rats, followed by relatively normal testis differentiation, resembling TDS in humans [
15]. Importantly, testicular dysgenesis in the rat model was only induced by androgen deficiency that occurs specifically within the MPW from embryonal day 15.5–18.5 [
16]. Based on the results and extrapolations from animal models, it has been proposed that an equivalent human MPW exists most likely in the period between GW 8–14 [
14,
17]. However, the existence of a human MPW and its timing has yet to be identified. Therefore, this study aimed to examine whether an androgen-sensitive window can be identified during early human fetal testis development. Specifically, the study focused on the effects of androgen deficiency induced by treatment with ketoconazole or flutamide on somatic cell function and on germ cell density and maturation during the presumed critical window. Since it is not possible to conduct such a study in humans in vivo, an established and extensively validated ex vivo culture model of human fetal testis was used in this study [
12,
18,
19].
Discussion
The hypothesis that sufficient androgen exposure of the human fetal testis during a presumptive androgen sensitivity period is essential for the programming of male reproductive function throughout life is intriguing. Despite the identification of MPW in rats more than a decade ago [
14] and the suggestion of a human equivalent [
14,
25], experimental evidence for such a presumptive human MPW has not been presented. In the present study, we provide experimental evidence to support the presence of an androgen-sensitive window during human fetal testis development. This was identified after experimentally reducing the androgen production in ex vivo cultured human fetal testis tissue from 1
st and 2
nd trimesters in an established model [
12,
18,
19] and examining the effects on cellular composition, germ cell maturation and testicular function. The ketoconazole-mediated decrease in androgen production resulted in reduced secretion of the Leydig cell factor INSL3 as well as the Sertoli cell factors AMH and Inhibin B in the age groups GW 7–10 and 10–12, but not GW 12–16 and 16–21. Additionally, a reduced density of germ cells (gonocytes) was found in the GW 7–10 age group. These findings were subsequently validated in GW 7–12 human fetal testis ex vivo following the reduction of androgen action via blocking of the androgen receptor by flutamide treatment. The flutamide-mediated reduced androgen exposure also resulted in reduced secretion of INSL3, AMH and Inhibin B as well as a decreased density of pre-spermatogonia. Thus, the overlap in effects observed after ketoconazole and flutamide treatments in human fetal testes suggests that these are indeed the result of experimentally reducing androgen exposure through different modes of action.
The consistent differences in effects between the examined age groups following ketoconazole-mediated reduction in androgen production were an important finding indicating (1) the presence of an androgen-sensitive period during human fetal testis development and (2) that the human androgen-sensitive window of testis development lies between GW 7–14 (including the 2-week culture period). Both notions are in accordance with previous suggestions that a human equivalent of the rat MPW exists. Extrapolations from the timing of the MPW in rats to the human fetal developmental timeline suggest that it lies between GW 8–14 [
14,
25]. This androgen-sensitive window during human testis development may similarly program testicular function later in fetal and/or postnatal life, although it was not possible to examine this in the ex vivo culture model of fetal testis used in this study.
Ketoconazole inhibits the activity of several CYP enzymes involved in steroidogenesis resulting in a decreased production of androgens [
26,
27]. Thus, the reduced androgen secretion reported in all examined age groups in this study was in accordance with the expected effect of ketoconazole. Additionally, the ketoconazole-mediated effects found in the age groups GW 7–10 and 10–12 in the present study were overall in accordance with the results previously reported. Using similar types of ex vivo culture approaches for human fetal testis, several studies have examined the effects of ketoconazole [
28‐
30]. In a study examining the effects of ketoconazole (10
−5 M) on human fetal testis from GW 7–12, a reduced production of testosterone, INSL3 and AMH as well as disruption of cord structures was reported [
28]. The 10-fold higher dose of ketoconazole may explain the effect on cord structure since the same group in a subsequent study determined EC
50 for ketoconazole-mediated reduction in testosterone production to be ~10
−6 M and found no effects on cord structure, apoptosis or expression of CYP11A1 following treatment with this dose [
29]. In the present study, no effects on cord structure, expression of examined somatic cell lineage markers or apparent changes in the density of proliferating (BrdU
+) or apoptotic (cPARP
+) cells were evident, indicating that the 10
–6 M treatment dose did not promote cytotoxic effects. This is also in line with results from ex vivo culture of the human fetal testis (GW 7–12) treated with a 4-fold higher dose than the present study where no effect on cord structure or cytotoxicity was reported [
29]. Although the selected 10
–6 M ketoconazole dose used in the present study resulted in significantly reduced androgen production in all four age groups, a more pronounced effect was observed in the GW 7–10 and GW 10–12 groups compared to GW 13–16 and GW 17–21 groups. This may be the result of the higher level of testosterone production expected in GW 12–16 fetal testis [
31]. Thus, it is not possible to exclude that a higher dose of ketoconazole would have been needed to ensure a similar level of reduction in androgen biosynthesis in the 2
nd-trimester ex vivo cultured samples. However, the limited access to human fetal testis tissue did not allow for experiments with several different doses of ketoconazole in the present study and thus a single dose was used for all age groups since it provided the best opportunity for direct comparison between groups. The flutamide-mediated effects reported here cannot be directly compared to previous ex vivo culture studies on human fetal testes. However, in vitro studies have reported IC
50 values ranging from 100 nM to 10 µM in androgen receptor assays with several studies reporting IC
50 values around 10
−6 M [
32‐
34]. Additionally, the effects of 10
−7 M flutamide were recently reported in a bovine in vitro granulosa cell culture model [
35].
The reduced androgen exposure during the presumptive androgen-sensitive window of human fetal testis development identified in this study consistently resulted in reduced secretion of AMH and Inhibin B from the Sertoli cells which may be important for the overall fetal testis development and function. Sertoli cells in the fetal testes are involved in supporting germ cell survival and differentiation as well as promoting differentiation of fetal Leydig cells and support of the precursor cells of adult Leydig cells. AMH secretion is essential for masculinization and is normally high during the fetal and neonatal period, despite the high levels of intratesticular androgens, which in postnatal (peripubertal) testes are thought to suppress the AMH expression [
36]. This has been attributed to the lack of androgen receptor (AR) expression in fetal Sertoli cells [
37,
38], and the data in the current study confirm this pattern. The reduced AMH levels found after the ketoconazole-mediated reduction in androgen production in the present study were in accordance with the effects reported previously in the study by Mazaud-Guittot [
28]. Since no AR expression was found in Sertoli cells in the examined developmental period, it is not clear by which mechanism the reduced androgen exposure resulted in the suppression of AMH, and, conversely, how the hCG-induced increase in androgen levels resulted in elevated levels of AMH. There is a possibility that PTM cells, which are observed within the developing fetal testis from GW 12 [
39] and express AR, could be involved. The PTM cells transmit signals between Leydig cells and Sertoli cells, which may in part explain the different responses observed before/after ~GW 12 following manipulation of androgen production in the present study. Additionally, it cannot be excluded that the reduced androgen levels following ketoconazole and flutamide treatment could have affected the somatic niche in a manner that was not distinguishable based on tissue morphology and expression of somatic cell lineage markers, e.g. reduced the density and/or development of Sertoli cells, or secretion of other factors not measured in the present study. Although the expression pattern of the Sertoli cell marker SOX9 showed no apparent change following any of the treatments, there may be a slight reduction in the density of Sertoli cells following ketoconazole treatment in the GW 7–10 and GW 10–12 groups which could in part explain the observed reduced levels of AMH and Inhibin B. Also, it is not evident from our results whether the reduced levels of INSL3 could be the result of reduced AMH and Inhibin B levels or vice versa. The reduced secretion of INSL3 consistently observed after ketoconazole and flutamide-mediated reduction in androgen exposure during the androgen-sensitive window may suggest an additional direct effect on the fetal Leydig cells that express AR throughout the examined developmental period, but the precise regulation and function of INSL3 in the human fetal testis is not understood in detail. However, in vivo INSL3 together with testosterone promotes testicular descent, and reduced levels of INSL3 in cord blood at birth have been associated with cryptorchidism [
40,
41]. The reduced secretion of AMH and Inhibin B may explain the observed effects on germ cell density, which also decreased after ketoconazole- and flutamide-mediated reduced androgen exposure during the androgen-sensitive period (GW 7–12). Interestingly, the affected germ cell type (gonocytes vs. pre-spermatogonia) was not consistent between treatments, which either may be due to direct differential effects (modes of action) of ketoconazole and flutamide treatment or could be the result of altered secretion of other Sertoli cells factors that were not determined in the present study (DHH, Activin B, FGF9). Regardless, this observation may have implications for the establishment of the spermatogonial stem cell population which is essential for future spermatogenesis and fertility. However, this possible effect remains speculative due to the short-term ex vivo culture approach in the present study.
Overall, the use of an ex vivo tissue culture model of isolated human fetal testes in the present study warrants a cautious interpretation of the reported findings since it does not allow the determination of effects on other organs or the impact of endocrine feedback mechanisms. Therefore, it is not possible to directly translate effects from the ex vivo culture model into an in vivo situation nor to provide insight about the development of male reproductive disorders that manifest later in life. The exclusion of hCG from the basic culture media contrasts with the in vivo situation where hCG is continuously present. However, the addition of hCG to culture media resulted in a lack of significant androgen reduction after ketoconazole treatment and since the main purpose of these experiments was to reduce androgen production, the experimental approach with either ketoconazole or hCG treatment was selected for this study. Of note, the culture media composition has been optimized prior to the establishment of the ex vivo culture set-up and used in previous studies [
12,
18,
19]. Another important limitation of the experimental approach in the present study is that it did not allow the detection of possible transient effects of treatments on hormone levels in culture media, which were pooled for each tissue fragment throughout the culture period due to the small volume. Also, transient effects on specific cell populations might not be detected since the tissue is analysed at the end of the 2-week culture period. Thus, it is not possible to exclude transient effects of ketoconazole or flutamide treatment in the present study. Finally, since experiments included in this study were performed in two different laboratories: Copenhagen (fetuses aged GW 7–12) and Edinburgh (fetuses aged GW 13–22), it is not possible to exclude the possibility that minor differences in handling and culture of tissue could have affected the results—although it is important to emphasize that all experiments were performed in a manner where tissues from each individual fetus were subjected to control and treatment under the same conditions and subsequently analysed as paired samples. Thus, the findings of the present study are in line with previous results from animal models. Our study provides novel insights into the existence and timing of a distinct androgen-sensitive period for programming of somatic cell function in the human fetal testis that coincides with the timing of the MPW identified in the rat, which is also associated with somatic cell dysfunction.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.