Maintenance, repair, and regeneration of many mammalian organs depend on adult stem cells. Stem cells proliferate and differentiate to replace mature functional cells within tissues that have either high turnover such as blood, testis, and epithelia (intestine, skin, and respiratory tract), or tissues that have low turnover but a high regenerative potential upon disease or injury such as liver, pancreas, skeletal muscle, and bone (1). Precise regulation of adult stem cell fate is therefore critical for the support of tissue homeostasis, and stem cell maintenance must involve a fine balance between genetic and epigenetic mechanisms, external factors from the microenvironment and systemic support, and multiple signaling pathways elicited by paracrine and juxtacrine factors.

Over the years, evidence has accumulated showing that stem cell self-renewal depends on the constituents of their microenvironment called the niche (2, 3) and that in turn stem cells influence their own environment (4–6). The constituents of the niche can be classified into adjacent supporting cells such as fibroblasts, tissue macrophages, glial cells (brain), osteoblasts (bone marrow), Sertoli cells (testis) and myofibroblasts (gut), together with paracrine and juxtacrine factors secreted by these supporting cells, and the extracellular matrix. Once they leave the niche, stem cells become progenitor cells that are less plastic and differentiate at the expense of their immortality. Over the last 15 years, critical cellular and molecular components of the specialized niche microenvironment have begun to be unveiled in several tissues. Advanced techniques in lineage-tracing, endogenous cell and gene/protein deletions in animal models, and high-resolution microscopy have significantly improved our understanding of the molecular and cellular intricacies that maintain and integrate the many activities required to sustain tissue homeostasis.

In the mammalian testis, the male germline produces a life-long supply of haploid spermatozoa through the highly regulated and coordinated process of spermatogenesis. This process starts with the self-renewal of a small pool of diploid stem cells called spermatogonial stem cells (SSCs or A_single spermatogonia), which can self-renew to maintain the pool or give rise to more mature germ cells called A_paired and A_aligned spermatogonia. Collectively, A_single, A_paired and A_aligned spermatogonia are called undifferentiated spermatogonia (7). These cells further differentiate into differentiating spermatogonia and spermatocytes that undergo meiosis, producing haploid spermatids that will mature into spermatozoa. The longevity and the high output of sperm cell production relies therefore primarily on the proper maintenance of the pool of SSCs and their proliferation. Like other types of stem cells, SSCs rely on their micro-environment to sustain their growth and to initiate differentiation that signals their release from the basal part of the seminiferous epithelium and exit from the niche.

SSCs reside on the basement membrane that supports the seminiferous epithelium. They are in intimate physical contact with highly specialized somatic niche cells, the Sertoli cells, which directly provide soluble growth factors and membrane-bound signals to the germ cells (8). Other niche cell types have been recently investigated, including peritubular myoid cells, interstitial cells (macrophages and Leydig cells), and endothelial cells from the vascular network, which all produce critical growth factors ( Figure 1 ) (9–15). Because of their direct physical association with germ cells, their secretion of growth factors and basement membrane components, and their architectural support of the seminiferous epithelium, Sertoli cells are considered the most important contributor of the testicular niche, and the regulation of their molecular communications with SSCs and more mature premeiotic germ cells will be the subject of this review.

GDNF is a member of the transforming growth factor beta (TGF-b) superfamily that binds to the GFRA1/RET receptor complex at the surface of SSCs, A_paired and some A_aligned spermatogonia (75, 77). Meng and colleagues were first to demonstrate that GDNF haploinsufficiency in mice induced fertility defects after birth (75). The mice were fertile but exhibited increased numbers of seminiferous tubules lacking spermatogonia as they aged. In addition, transgenic animals overexpressing Gdnf accumulated undifferentiated spermatogonia. In 2006, Naughton and colleagues disrupted the expression of Ret and Gfra1 at the surface of SSCs, which resulted in their loss and led to the definitive proof of the critical function of this receptor-ligand interaction (78). Together with FGF2 and LIF, GDNF is critical for the self-renewal of SSCs in short- and long-term cultures (66). Because of its importance for spermatogenesis, efforts were made to understand the temporal regulation of its expression. Low levels of GDNF and RET are already present in the fetal gonad (76, 110). Since pro-spermatogonia do not proliferate until after birth, GDNF is therefore only necessary for their maintenance, highlighting the importance of its dosage (98). GDNF expression then increases until it reaches a peak at days 3-7 after birth (110, 135, 136). One interesting feature of GDNF expression in the adult is its cyclic pattern throughout the stages of the seminiferous epithelium. Cyclical production of soluble factors according to stages was demonstrated earlier by Johnston and colleagues using transillumination-assisted microdissection and microarray analysis (137). In the rat, GDNF expression is highest at stages XIII-I, and lowest at stage VII of the seminiferous epithelium (138), while in the mouse its expression is highest at stages IX-I and lowest at stages V-VIII when most cells are quiescent and the majority of A_aligned spermatogonia transition to the differentiating A1-A4 cells (85, 98, 139). When GDNF was ectopically overexpressed by Sertoli cells in Stages V-VIII, the number of GFRA1+/LIN28^- germ cells, a subtype of A_s spermatogonia with enhanced self-renewal capacity, was increased (97, 98).

Several mechanisms regulating GDNF expression have been recently proposed. Garcia and colleagues established Sertoli cell-specific gain-of-function and loss-of-function mouse models of NOTCH receptor signaling (80, 100). Constitutive activation of this pathway in Sertoli cells led to a complete lack of germ cells by P2, and infertility. Expression of GDNF by Sertoli cells was significantly downregulated in the perinatal and adult testis and was due to upregulation of Hes/Hey transcription factors, which are canonical NOTCH targets and transcriptional repressors that bind to the GDNF promoter (80, 85). Further, loss-of-function of Rbpj, a mediator of NOTCH, and downregulation of Hes/Hey, led to upregulation of Gdnf expression (80) ( Table 1 ). Importantly, the NOTCH ligand JAG1 was expressed mainly by undifferentiated spermatogonia (85). Consequently, the accumulation of undifferentiated spermatogonia around stage VII might increase NOTCH activity in Sertoli cells through JAG1, triggering the observed increase of Hes/Hey inhibitors at this stage and decrease in GDNF expression, leading to its cyclic expression. Therefore, spermatogonia, when in sufficient numbers, regulate their own homeostasis through downregulation of GDNF (55). These data are consistent with the observation that in wild type mice, the absence of germ cells triggered by busulfan treatment correlated with higher expression of GDNF (85, 135, 140) ( Figure 2A ).

Other interesting mechanisms of GDNF regulation have been recently proposed. Given the fact that retinoic acid (RA) concentration is high when GDNF is low during the cycles of the seminiferous epithelium (141), Saracino and colleagues tested whether RA was a direct inhibitor of GDNF expression (142). Using ex vivo cultured immature testes and staged adult seminiferous tubules, they showed that negative regulation of Gdnf by RA indeed takes place in these models and demonstrated that Gdnf expression is directly regulated by RA through a mechanism involving a RARE-DR5 binding site on the Gdnf promoter. Negative regulation requires retinoic acid receptor (RARα) and induces a strong decrease of histone H4 acetylation levels around the transcription start. Further, because of the existence of a NF-kappaB binding site in the GDNF promoter, the same group investigated how TNF-alpha might influence GDNF expression (99). They demonstrated that in primary Sertoli cells, TNF-alpha induces the expression of the transcriptional repressor Hes1 by a NF-KappaB-dependent mechanism, which in turn downregulates GDNF. Therefore, TNF-alpha and NOTCH signaling may converge to regulate the expression of Hes1 and its target genes, including GDNF ( Figure 2A ).

While GDNF is a critical component of the niche, many in vivo and in vitro experiments demonstrated that other factors are needed to support maintenance and self-renewal of SSCs. Earlier examination of perinatal Sertoli cells demonstrated that they expressed FGF2, and that this expression was stimulated by follicle-stimulating hormone in vitro (FSH) (64). Together with EGF, LIF, and GDNF, fibroblast growth factor (FGF2) has been used to sustain the long-term proliferation of SSCs in culture (66, 143). Further, Takashima and colleagues demonstrated that FGF2 could induce SSC self-renewal alone in culture through activation of the transcription factors ETV5 and BCL6B ( Table 1 ) (37, 38, 60, 62, 63, 67). They also showed that FGF2-depleted mouse testes produced increased levels of GDNF, which correlated with SSCs enrichment. This suggests that a balance or complementation between FGF2 and GDNF exists to maintain the stem cell pool (67). More recently, additional studies comparing the effects of GDNF and FGF2 on the proliferation of undifferentiated spermatogonia demonstrated that while both factors expanded the GFRA1+ population, FGF2 rather expanded a subpopulation of cells expressing RARG, which were therefore more susceptible to differentiate (68). This emphasizes the complex nature of signaling and a growth factor demand that is modulated upon the need to maintain germ cell homeostasis.

Platelet-derived growth factor (PDGF) is specifically produced by Sertoli cells. In rodents, PDGF is critical for prospermatogonia proliferation after birth (103, 104) and cooperates with estrogen signaling (106). Exposure to xenoestrogens in the environment might disrupt crosstalk between PDGF and estrogen-driven signaling pathways. This could lead to alteration of prospermatogonia behavior and induce preneoplastic states (105). Vascular endothelial growth factor A (VEGFA) family members and their receptors are all produced by germ cells, Sertoli, cells and interstitial cells (117, 118). However, only the pro-angiogenic isoform VEGFA164 promotes SSC self-renewal, as determined by the SSC transplantation assay (119). WNT signaling plays a role in SSC maintenance (50, 144). WNT5A is produced by Sertoli cells but does not induce self-renewal. It rather promotes SSCs survival through a β-catenin (CTNNB1)-independent mechanism that activates mitogen-activated protein kinase 8 (MAPK8 or JNK) (50). Confirming this data, CTNNB1 ablation in germ cells led to spermatogenesis disruption but not to SSC loss (51, 52). Finally, leukemia inhibitory factor (LIF) has been used for decades to maintain undifferentiated embryonic stem cells in vitro, therefore an investigation of its expression in Sertoli cells and its effects on SSCs, at least in vitro, was attempted early on (96). LIF production in Sertoli cells was shown to depend on tumor necrosis factor (TNFα) (96) and is still widely used in cultures of primordial germ cells, pro-spermatogonia, and SSCs of many different species. However, LIF does not induce SSC self-renewal, and is rather used to maintain survival and start long-term SSC cultures (10).

Activation of the KIT tyrosine kinase receptor by its ligand KITL is required for the survival and proliferation of primordial germ cells (PGCs) (90). KIT is downregulated in pro-spermatogonia, which stop proliferating once they enter the fetal gonads. After birth, KIT is re-expressed in differentiating spermatogonia (87, 88), which proliferate under the influence of KIT ligand (KITL) produced by Sertoli cells. Together with retinoic acid (RA), the KIT/KITL system is important for triggering meiotic entry of type B spermatogonia (92, 93), and KITL has been recently used in culture to differentiate rat spermatogonia without serum or somatic cells (95). Because KIT/KITL signaling is important not only for germ cells, but also for haematopoietic stem cell and melanoblasts, mechanisms controlling KIT transcription have been extensively studied. Further, KIT is mutated in about 25% of seminoma (91), and accounts for secondary mutations that confer resistance to drugs in other cancers. Therefore, regulation of its expression and identification of downstream effectors as druggable targets are of particular interest. Earlier studies have demonstrated that KIT expression in undifferentiated spermatogonia is repressed by PLZF (promyelocytic leukemia zinc finger), which is a transcriptional repressor with local and long-range chromatin remodeling activity (107, 108). Further, Dann and colleagues demonstrated that RA triggered spermatogonial differentiation through downregulation of PLZF (145). Thus, one mechanism by which PLZF maintains the pool of spermatogonial stem cells is through a direct repression of Kit transcription. The main mechanism of KIT upregulation involves the helix-loop-helix transcription factors SOHLH1 and SOHLH2 (Spermatogenesis and Oogenesis HLH1/2). Both factors are expressed in differentiating spermatogonia and their deletion leads to the disappearance of KIT-expressing spermatogonia. Further, ChIP-PCR analysis demonstrated that SOHLH1 binds the Kit promoter to activate its transcription (115). While investigations have mostly focused on the regulation of KIT, few studies have explored the regulation of KITL expression in the past 10 years. However, one study by Correia and colleagues demonstrated that 100 nM estrogen induced a decrease in Kit expression while increasing expression of Kitl in adult rat seminiferous tubules cultured ex vivo (89). Altered expression of the KIT/KITL system decreased germ cell proliferation and promoted apoptosis, which is not in accord with the data of previous studies (146).

Rats and mice deprived of dietary retinoic acid (RA) can only produce A_undiff spermatogonia and are sterile (147, 148). Since these earlier studies, it has been well documented that retinoic acid (RA) activity is a major determinant of the transition between undifferentiated and differentiating germ cells, and that RA also drives the meiotic process and spermatid maturation at stage VIII of the seminiferous epithelium (134, 149). It has been proposed that pulses of RA are triggered around this stage by somatic cells and germ cells to allow proper germ cell differentiation and maturation (150). This implies that RA must be degraded during the other stages. Recently, Parekh and colleagues demonstrated an inverse relationship between the expression of cytochrome P450 family 26 subfamily B member 1 (Cyp26b1), an enzyme that degrades RA (54), and NOTCH activity in Sertoli cells (56). They further provided evidence that in the adult testis activated NOTCH signaling in Sertoli cells down-regulates Cyp26b1 expression through the HES/HEY transcriptional repressors that bind to the Cyp26b1 promoter (56). Importantly, expression of these inhibitors is highest at stage VIII of the seminiferous epithelium (85). They also demonstrated that A_aligned spermatogonia, through their expression of the NOTCH receptor JAG1, were activating the NOTCH/HES/HEY axis in Sertoli cells and were responsible for Cyp26b1 down-regulation at stage VIII, allowing RA activity and therefore triggering their own differentiation into A₁ spermatogonia ( Figure 2B ).