The Role of Retinoic Acid (RA) in Spermatogonial Differentiation.

Retinoic acid (RA) directs the sequential, but distinct, programs of spermatogonial differentiation and meiotic differentiation that are both essential for the generation of functional spermatozoa. These processes are functionally and temporally decoupled, as they occur in distinct cell types that arise over a week apart, both in the neonatal and adult testis. However, our understanding is limited in terms of what cellular and molecular changes occur downstream of RA exposure that prepare differentiating spermatogonia for meiotic initiation. In this review, we describe the process of spermatogonial differentiation and summarize the current state of knowledge regarding RA signaling in spermatogonia.

INTRODUCTION

Multicellular organisms contain a wide variety of specialized cell types that originate from less specialized cells by cellular differentiation, which involves a progression of specific changes that prepare them for their ultimate function. Many specialized cells have a finite lifespan and therefore must be periodically replaced by a population of uni- or multipotent adult stem cells. These stem cells balance self-renewal with the production of progenitor cells that proliferate to amplify their numbers before committing to a specific cell fate. As an example, the consistent daily production of 1012 blood cells in the adult mammalian bone marrow is accomplished by a comparatively small population of hematopoietic stem cells (estimates range from approximately 16 800 to 81 000), the progenitors of which follow unique programs of differentiation to become leukocytes, erythrocytes, or megakaryocytes [1-3].

The production of mammalian spermatozoa in the testis is a stem cell-based developmental process. Each adult mouse testis contains approximately 3000 unipotent spermatogonial stem cells (SSCs) that either self-renew or initiate spermatogenesis by producing undifferentiated progenitor spermatogonia that are destined to enter meiosis [4, 5]. This small population of SSCs is responsible for the production of 109 sperm per day throughout the male mouse reproductive lifespan [6]. The decision to remain a stem cell or to proliferate and differentiate is crucial for the reproductive health of the male. In humans, insufficient or excessive differentiation can result in reduced or lost sperm production (nonobstructive oligo- or azoospermia), which are leading causes of male infertility. Strikingly, there appears to be a decrease in overall male reproductive fitness in Western societies over the past several decades, which has been termed testicular dysgenesis syndrome. This syndrome is thought to result from environmental changes and is characterized by a decline in semen quality, increases in hypospadias and cryptorchidism, and an increase in the incidence of testicular cancer [7-9]. Undifferentiated male germ cells (specifically, primordial germ cells, prospermatogonia, and potentially, spermatogonia) that fail to properly differentiate are hypothesized to be the basis for carcinoma in situ, the precursor to most forms of testicular cancer [10-12]. Significant effort has been exerted to understand how the foundational SSC population is maintained, and a number of excellent recent reviews document this progress [13-18]. In contrast, little is known about the cellular changes accompanying spermatogonial differentiation, and the pathways and proteins involved remain poorly defined. The purpose of this review is to provide a developmental perspective on the current state of knowledge about the essential program of spermatogonial differentiation that prepares undifferentiated progenitor spermatogonia for entry into meiosis.

SPERMATOGONIAL BIOLOGY

Following sex determination in the fetal mouse testis, prospermatogonia (also termed gonocytes) proliferate until approximately Embryonic Day (E) 14.5 and then enter a mitotically quiescent state in G0 of the cell cycle until after birth [19, 20]. Prospermatogonia then re-enter the cell cycle at approximately Postnatal Day (P) 1–2 and migrate from the center to the periphery of the testis cords; completion of both tasks is apparently required for their survival [21]. Prospermatogonia become spermatogonia at approximately P3–P4, as they migrate to the periphery of the testis cords and become flanked by somatic Sertoli cells within the testis cord and peritubular myoid cells that surround the outside of the cord (Fig. 1, A and B). The proteins and signaling networks involved in this transition are currently under investigation by several laboratories. It has recently been shown that suppressing NOTCH signaling in Sertoli cells is important for maintaining quiescence in prospermatogonia [22-24]. Also, two reports indicate the requirement for the chromatin-modifying protein Swi-independent 3a (SIN3A) in regulation of mitotic re-entry [25, 26]. Members of the transforming growth factor beta (TGF-β) superfamily, such as the activins, inhibins, and bone morphogenetic proteins (BMPs), have likely roles in the initiation of spermatogenesis, although their requirement in vivo requires further study (for reviewed, see [27]).

FIG. 1

Germ cell differentiation in the neonatal testis. A) At P1, prospermatogonia are located in the center of the testis cords with adjacent Sertoli cells and peritubular myoid (PM) cells surrounding the tubules. B and C) By P4, spermatogonia have become either Aundiff (GFRA1+, in green) or Adiff (STRA8+, in red). In C, the testis cords are outlined by F-actin staining with phalloidin (in blue). Bar = 10 μm.

This initial neonatal spermatogonial population is heterogeneous, and both undifferentiated (Aundiff) and differentiating (Adiff) spermatogonia are detectable as early as P3–P4 (Fig. 1C) [28-31]. The origin of this heterogeneity is currently undefined, although it is apparent that spermatogonia are capable of differentially responding to extrinsic signals from somatic cells, which will be discussed in greater detail below. A small percentage of the Aundiff population contains the future SSC pool, which functions to support steady-state spermatogenesis throughout the remainder of the male reproductive lifespan. The rest of the surviving neonatal spermatogonia become progenitor or differentiating spermatogonia that will enter meiosis beginning at approximately P10 to give rise to the first fertilizable sperm that are seen around P35. It is presumed that the first differentiating spermatogonia arise directly from prospermatogonia without first forming an SSC [29, 30, 32]. This transition (prospermatogonia to type A spermatogonia) marks the “initiation of spermatogenesis” in the mouse.

During steady-state spermatogenesis, the products of SSC divisions either maintain the stem cell pool (self-renewal division) or generate progenitor spermatogonia that will proliferate and differentiate to eventually enter meiosis. Daughter cells of an SSC division that are destined to differentiate retain a relatively large (∼1 μm diameter) tubular connection, termed an intercellular or cytoplasmic bridge, that results from incomplete cytokinesis [33, 34]. The function of these bridges is unclear. However, they are highly conserved through evolution and allow sharing of molecules and even organelles such as mitochondria between cells within a syncytium (for review, see [35]). This likely aids in the synchronization of subsequent divisions, and it may also provide for the sharing of essential X-linked gene products between adjacent X- and Y-bearing haploid postmeiotic spermatids later during spermiogenesis. Single spermatogonia are termed Asingle (As), whereas those connected by an intercellular bridge are termed Apaired (Apr). As progenitor spermatogonia undergo transit-amplifying divisions, they form progressively longer interconnected chains as Aaligned (Aal) spermatogonia. Although As, Apr, and Aal spermatogonia are all classified as undifferentiated, evidence supports a diminution of stem cell capacity that accompanies increased chain length (for review, see [17]). However, evidence indicates that some Apr spermatogonia, termed false pairs, represent SSCs that have divided but not moved away from one another (for review, see [13]).

The commitment to enter meiosis is made with the transition of Aundiff into Adiff spermatogonia. This transition requires retinoic acid (RA), which will be discussed in detail below. The first differentiating spermatogonia are termed type A1, which undergo five subsequent divisions to form A2, A3, A4, intermediate (In), and finally, type B spermatogonia before becoming preleptotene spermatocytes that are in the first phase of meiosis I. During differentiation, the cell-cycle duration decreases, and a significant amount of germ cell loss occurs, such that only an estimated 39% of the expected numbers of preleptotene spermatocytes are formed [36-38]. It is important to note that male germ cells must complete this prolonged stepwise differentiation process that takes approximately 1 wk in order to gain competence to enter and successfully complete meiosis. In the fetal testis, prospermatogonia respond to precocious RA exposure by inappropriately expressing meiotic markers before rapidly dying by apoptosis [39-41]. In the neonatal and adult testis, Aundiff spermatogonia exposed to exogenous RA cannot be hastened to enter meiosis without progressing through these steps [42-46].

The accurate identification of spermatogonial subtypes at the histological level takes a considerable amount of experience, and it relies on characteristic differences in nuclear shape and diameter as well as heterochromatin appearance in paraffin-embedded testis sections carefully prepared with certain fixatives (e.g., 5% glutaraldehyde or Bouin solution). In the adult testis, identification is aided by the fact that germ cells are present in defined stages of the seminiferous epithelium [29, 47]. Neonatal and juvenile testes lack clearly defined epithelial stages, although some have proposed that staging is possible beginning with the appearance of preleptotene spermatocytes at approximately P8. The absence of defined stages makes it difficult to impossible to discriminate reliably between all spermatogonial subtypes at the light microscopic level based on morphology alone. This is because differences in morphology (especially of chromatin) are both subtle and quite variable [28, 48]. The topological arrangement of As, Apr, and Aal spermatogonia can be visualized in whole mounts of seminiferous cords or tubules, but the germ cells must be labeled either by transgenic expression of a fluorescent reporter such as GFP or by using indirect immunofluorescent antibody staining, as described below.

Another useful tool for spermatogonial identification is labeling with antibodies against specific protein markers that have been linked to cell fate or function (Fig. 2). A number of characteristic proteins have been identified that are detectable primarily in Aundiff spermatogonia in the adult testis, including ID4, GFRA1, RET, ZBTB16/PLZF, CDH1 and PAX7 [17] as well as NANOS2 and NANOS3 (which are indirectly labeled as FLAG-tagged transgenes) [49, 50]. In contrast, KIT and STRA8 are the only two markers currently detectable in differentiating, but not undifferentiated, spermatogonia [51-53]. Although STRA8 is generally considered to be a nuclear protein, it has also been detected in the cytoplasm [54, 55]. Its function is currently unknown, but its determination should help clarify this observation. For Adiff subtypes (A1, A2, A3, or A4), no protein markers are currently identified that allow them to be distinguished from one another without knowing the stage of the adult seminiferous tubule within which they reside. Some protein markers have increased levels in differentiating spermatogonia, including RHOX13, SOHLH1, and SOHLH2 [42, 56–60]. Whereas the above markers are restricted to either undifferentiated or differentiating spermatogonia in the juvenile and adult testis, this is largely not true in the neonatal testis. We recently reported significant overlap, in that markers for Aundiff colocalized with the spermatogonial differentiation marker KIT (as well as STRA8) through approximately P10, with GFRA1 and RET being notable exceptions [44]. This suggests that the differentiating program is overlaid on the undifferentiated state in prospermatogonia and spermatogonia in the neonatal testis. Whereas these fates appear more clearly established in the juvenile and adult, transient overlap of some undifferentiated and differentiating markers also occurs in adult spermatogonia.

FIG. 2

Selected spermatogonial protein fate markers. The estimated detection level for each marker is listed by a plus sign (+, ++, or +++). Levels are subjective, and comparisons are only valid for tissues prepared, incubated, and stained similarly. The minus sign (−) indicates that the listed protein is undetectable but not necessarily absent from the listed cell type. *Inferred (largely studies using fluorescent reporter constructs in transgenic mice).

Three types of spermatogonia are present in the human testis—namely, types Adark, Apale, and B. Prospermatogonia transition into both Adark and Apale spermatogonia by 2–3 mo after birth. The first differentiating type B spermatogonia are visible by 4–5 yr of age but only represent approximately 10% of the spermatogonial population by age 10 [61]. It is currently held that Adark spermatogonia are non- or slow-cycling and represent the “reserve” stem cell population that can be activated following damage to the germ cell population. In contrast, Apale spermatogonia are analogous to the Aundiff population in rodents and can be present as single cells or longer chains of interconnected cells (for review, see [62]).

EXTRINSIC SIGNALS DIRECT SPERMATOGONIAL FATE

Tissue homeostasis is maintained in many epithelial tissues that have a high rate of turnover through a delicate balance between stem cell self-renewal and the production of progenitors that proliferate and differentiate. In the testis, distinct signals received by developing male germ cells direct this balance. Numerous studies have revealed that the undifferentiated state is maintained by the binding of the ligands provided by Sertoli and/or peritubular myoid cells to their cognate receptors on spermatogonia both in vivo and in vitro. While numerous additional signaling pathways surely await discovery, the best described ligand/receptor pairs currently include glial cell-derived neurotrophic factor (GDNF) to GFRA1/RET [63-70], chemokine (C-X-C motif) ligand 12 (CXCL12) to CXCR4 [71], and fibroblast growth factors 2 and 8 (FGF2 and FGF8, respectively) to FGFR [72-77].

Spermatogonial differentiation requires all-trans retinoic acid (ATRA; referred to as RA), the bioactive oxidative metabolite of vitamin A/retinol [46, 78–81]. Spermatogonia cannot progress past the Aal stage in mice when RA signaling is blocked following prolonged consumption of a vitamin A-deficient (VAD) diet or administration of specific compounds (e.g., bis-[dichloroacetyl]-diamines such as WIN 18 446) that restrict the synthesis of RA from retinal by inhibiting retinaldehyde dehydrogenases [82-84]. Either treatment causes arrested spermatogonial differentiation at the Aal-to-A1 transition that can be reversed by retinoid supplementation, resulting in resumption of spermatogenesis and restoration of fertility [79, 82, 83, 85]. Although the primary role for RA in the testis in directing spermatogonial differentiation is clearly established, the mechanisms activated downstream of RA exposure are largely undefined.

REGULATING SPERMATOGONIAL EXPOSURE TO RA

Cellular exposure to RA is managed at multiple levels by proteins that regulate its synthesis, reception, storage/transport, and degradation [85-87]. Several laboratories are currently focused on understanding how RA is distributed within the testis such that only Adiff spermatogonia respond to this differentiating signal. Two general scenarios can be envisioned. In option 1, all spermatogonia are primed to respond to RA, but the exposure to RA is tightly controlled; in option 2, all spermatogonia are exposed to RA, but only some can respond. Current evidence in the literature suggests that both scenarios are involved (see Fig. 3A). In support of this notion, the postnatal deletion of single, seemingly key molecules involved in RA reception, storage, and degradation in knockout (KO) mouse models has not thus far resulted in phenotypes that fully recapitulate the VAD model's arrested spermatogonial differentiation and infertility [88-96]. This indicates that both exposure and reception are parts of a complementary system with redundant controls built in to ensure spermatogonia respond appropriately to RA.

FIG. 3

Regulating spermatogonial exposure to RA. A) Two options, described in the text, for how spermatogonia become exposed to or avoid RA. In option 1, an SSC contains RARs (yellow Y) and so could presumably differentiate in response to RA (blue triangle). However, expressed CYP26 (orange T) catabolizes RA and prevents binding to RARs. In option 2, an SSC that does not express RARs is pictured. Therefore, although the cell has access to abundant RA, it does not differentiate because it lacks the requisite RARs to transduce the signal. B) Two type A spermatogonia within the same cord adopt different fates based on their response to RA. The upper left spermatogonium remains GFRA1+ (green) and can respond to GDNF but does not respond to RA. This may result from lack of requisite RARs and/or degradation by CYP26 enzymes active in either the spermatogonium itself or in the RA-producing cell (adjacent Sertoli). The lower right spermatogonium responds to RA by becoming STRA8+ (red) and KIT+ (not shown) and differentiating. This may result from gain of RAR expression or the absence of localized CYP26-mediated degradation within the spermatogonium itself or adjacent Sertoli cells.

Results from several reports support a role for regulated RA exposure (option 1 above) in maintaining spermatogonial cell fate. In the fetal testis, quiescent prospermatogonia must be protected from RA exposure or they will begin to differentiate and enter meiosis precociously and, as a result, die by apoptosis [40, 41]. This protection is provided, at least in part, by the RA-degrading action of the cytochrome P450 enzyme CYP26B1 [39, 94]. After birth, a subset of spermatogonia becomes exposed to RA by P3–P4 (as evidenced by their expression of the RA-inducible Stra8 gene) [42, 53, 58, 84, 97]. If CYP26B1-mediated degradation is responsible for protecting a subset of postnatal spermatogonia from RA exposure, this implies that degradation activity is reduced or lost near STRA8+ Adiff spermatogonia, although this has not been shown experimentally. In the adult testis, the majority of Aundiff spermatogonia transition to differentiating KIT+ A1 spermatogonia at stage VIII of the seminiferous epithelial cycle. This coincides with STRA8 induction in Adiff spermatogonia and preleptotene spermatocytes [54, 98], and it was recently shown that a pulse of RA peaks at stage VIII [51]. Therefore, RA levels are clearly modulated during steady-state spermatogenesis in the adult; epithelial stages VII–VIII, which are exposed to the highest levels of RA, contain germ cells undergoing the three processes that are dependent upon RA (spermatogonial differentiation, meiotic initiation, and spermiation) [46, 79, 99, 100].

Evidence supports a requirement for the production of RA by Sertoli cells. Circulating retinol is converted into RA by two successive reactions: retinol to retinal by retinol dehydrogenases, and retinal to RA by retinaldehyde dehydrogenases. The conditional deletion of retinol dehydrogenase 10 (Rdh10) in Sertoli plus germ cells, and, to a lesser extent, in Sertoli cells only, resulted in loss of A1 differentiating spermatogonia [101]. Interestingly, this defect is only manifest during the first wave of spermatogenesis in young mice (age, <7 wk); after that, KO males exhibit normal fertility and testis histology. These results clearly suggest that the first step in the synthesis of RA (retinol to retinal) is performed by another retinol dehydrogenase during steady-state spermatogenesis in the adult. In addition, three retinaldehyde dehydrogenases (Aldh1a1–3, previously termed Raldh1–3) have been conditionally deleted in mouse Sertoli cells [45]. Spermatogonia in these mice fail to differentiate; however, injection of RA or a retinoic acid receptor (RAR) A-selective agonist reinitiates spermatogenesis.

There is also evidence supporting a role for regulated RA reception (option 2 above) in maintaining spermatogonial cell fate. RA is a lipid-soluble molecule that enters the cell to bind with high affinity to its cognate receptors. The RAR has three isotypes (RARA, RARB, and RARG), and each is capable of heterodimerizing with a retinoid X receptor isotype (RXRA, RXRB, or RXRG). The RAR isotypes have distinct expression patterns in the testis: RARB is undetectable, RARA predominates in Sertoli cells, and RARG predominates in Adiff spermatogonia in the neonatal, juvenile, and adult testis [92, 102]. Therefore, based on this expression pattern, it is logical to assume that RA signaling through RARG directs spermatogonial differentiation. Whole-body as well as germ cell and Sertoli cell KO mice have been generated with deletions of each of the RAR isotypes to address their respective roles in the testis. Rarb-null mice are viable and fertile, with no apparent defects in spermatogenesis [103], which is expected based on its apparent lack of expression in the testis. In contrast, both Rara-null and Rarg-null mice exhibit varying defects in spermatogenesis, although loss of either gene singly or in combination does not recapitulate the VAD phenotype of blocked spermatogonial differentiation [88–90, 92, 104, 105]. Although Rara KO mice are infertile [88, 89, 105], deletion of Rarg has no obvious effect on spermatogonial differentiation during the first wave of spermatogenesis, and many tubules are apparently normal until KO mice reach advanced age (∼12 mo) [92]. Unfortunately, the reproductive performance of Rarg germ cell KO males has not been reported; based on the histology images provided in Gely-Pernot et al. [92], it is reasonable to expect that young KO mice are fertile. Therefore, it can be concluded that although RAR isotypes clearly participate in spermatogonial differentiation, none is essential for the process to occur. As mentioned above, seminiferous epithelial stages VII–VIII are exposed to the highest measured RA levels [22, 43, 51], and spermatogonia differentiate during this time. However, stage-VII and -VIII tubules also contain SSCs, which remain undifferentiated and STRA8−/KIT−, implying that they did not receive the RA signal. A recent study employed a transgenic approach to create mice expressing RARG from the Gfra1 promoter in Aundiff spermatogonia; the results indicate that these spermatogonia inappropriately differentiate during stages VII–VIII [102]. These results, taken together with the Rarg KO data above, support the concept that RARG is sufficient, but not required, to direct spermatogonial differentiation.

The testis contains a consistently small SSC population and a system for the proliferation and then differentiation of millions of progenitor spermatogonia required for gamete production. Accordingly, it should not be surprising that the responsiveness of spermatogonia to RA is controlled by redundant mechanisms (expression of RARG in spermatogonia primed to differentiate but not in SSCs, and careful modulation of RA levels by RA-synthesizing and -degrading enzymes) (see Fig. 3B). The evidence that such a robust system is seemingly not reliant upon the function of a single gene product highlights how critical modulating male germ cell RA exposure is for both male reproductive health and success.

GENE EXPRESSION CHANGES DURING DIFFERENTIATION

Although it has been known for 90 yr that retinoids are essential for male fertility [78], the molecular and cellular events downstream of RA remain largely undefined. A primary reason for our lack of knowledge about spermatogonial differentiation is that studies using whole-genome approaches have identified few changes in steady-state mRNA levels between Aundiff and Adiff spermatogonia or in whole testes during the early phases of neonatal testis development during which differentiation takes place [98, 106, 107]. Many of the upregulated spermatogonia-expressed genes (at the mRNA level) encode proteins with known roles in meiosis (e.g., REC8, STRA8, and SYCP3) [53, 97, 108–110]. This is not meant to conclude that transcriptional regulation is uninvolved in spermatogonial differentiation. Indeed, the deletion of the transcription factors Sohlh1, Sohlh2, and Sox3 blocks (in Sohlh1 and Sohlh2 KO testes) or impairs (in Sox3 KO testes) spermatogonial differentiation, although the defects in Sox3 germ cell KO mice are more severe during the first wave of spermatogenesis and improve over time as the mice age [56, 111–114]. In addition, testis cords are apparently normal through at least P10, which argues against an absolute requirement for SOX3 during spermatogonial differentiation.

It has become clear through genomic and genetic analyses that genes required to maintain SSCs in vivo and in vitro are lost in response to spermatogonial differentiation signals [14–17, 115]. One example is Zbtb16/Plzf, which encodes a putative transcriptional repressor that is detectable in Aundiff spermatogonia in the adult but is present in all spermatogonia in the neonatal and juvenile testis [44, 116, 117]. ZBTB16 is not absolutely required for spermatogenesis; although adult mutant and KO mice are infertile, with very low numbers of epididymal sperm, their testes contain numerous Sertoli cell-only tubules (lacking germ cells) adjacent to rather normal-appearing tubules [116, 117]. A role in SSC self-renewal is concluded based on the histological phenotype and because mutant spermatogonia are unable to colonize testes of recipient mice lacking germ cells [116, 117].

Without dramatic upregulation of mRNAs during spermatogonial differentiation, scientists have lacked targets in the form of proteins and pathways for focused studies. The lack of dramatic differences in the transcriptome of developing spermatogonia suggests, first, very few changes in transcription or mRNA decay (both contribute to resulting steady-state mRNA levels) or, second, differences in the transcriptomes within spermatogonial subtypes that are not discernible when an entire population of spermatogonia is queried. Results from a recent study suggest that both are viable options. Significant differences exist in the abundance of specific mRNAs within single spermatogonia that correlate with their fate status (expression of Id4-GFP in SSCs) [118]. This work also highlights examples of genes for which the mRNAs are present in Id4-GFP+ SSCs without detectable protein, which indicates posttranscriptional regulation of gene expression.

It has been known for many years that not all mRNAs are translated into protein with similar efficiency—mRNAs can be stored, can be inefficiently or efficiently translated, or can be targeted for degradation [119]. The decision among these biochemical fates provides cells with an important level of control over gene expression that allows rapid and large-scale responses to developmental stimuli. Recent results provide a novel perspective that may advance our understanding of events during spermatogonial differentiation. Others and we have reported that mRNAs encoding essential differentiation factors (e.g., KIT, SOHLH1, and SOHLH2) are present, but repressed, in Aundiff spermatogonia but then initiate translation in response to RA in Adiff without a dramatic increase in mRNA abundance [42, 58, 120, 121]. This suggests that, instead of transcriptional activation of unique genes, efficient translation of a pool of repressed mRNAs is the driving force behind gene expression changes during spermatogonial differentiation. Two well-studied mechanisms for posttranscriptional control of gene expression involve miRNAs and RNA-binding proteins, and evidence is growing that both are involved in spermatogenesis.

MicroRNAs are short, noncoding RNAs that bind to target mRNAs and repress their translation by inducing cleavage or destabilization or by preventing ribosomal association (for review, see [122]). The biogenesis of requires the ribonuclease Dicer1. Somewhat surprisingly, deletion of Dicer1 in fetal male germ cells does not cause noticeable defects until the pachytene stage of meiosis [123-127], indicating that miRNA function is not essential for spermatogonial development. However, recent studies indicate that specific miRNAs and miRNA clusters (Mir146, Mir221/222, Mirc1, Mirc3, and Mirlet7) do contribute to gene regulation in spermatogonia [121, 128–130]. In vivo, however, they likely play a supplementary or supportive role in the translational control of mRNAs encoding factors that direct spermatogonial development. In contrast to the dispensable function of miRNAs in premeiotic male germ cells, numerous RNA-binding proteins have been shown to be essential for fetal and neonatal mammalian germ cell development and differentiation (e.g., NANOS2, NANOS3, DAZL, TIAR/TIAL1, PIWIL2/MILI, PIWIL4/MIWI2, and DDX4/VASA) [131-135]. The most well characterized RNA-binding protein is NANOS2, which is required for the function and survival of both fetal prospermatogonia and postnatal SSCs [50, 136]. It appears to have multiple roles in suppressing translation of target mRNAs during spermatogenesis, both by promoting their degradation [137] and by preventing association with polyribosomes [138]. A formal link between RA and expression of NANOS2 has not been established in the postnatal testis, but in fetal testes of mice lacking Cyp26b1 (which have higher RA levels), Nanos2 mRNA levels were significantly reduced [139]. This suggests that RA may negatively regulate the expression of Nanos2; indeed, exogenous RA decreases Nanos2 mRNA levels in the neonatal testis (our unpublished data). The loss of NANOS2 expression downstream of RA signaling provides a potential mechanistic explanation for the increased translational efficiency of repressed mRNAs encoding determinants of spermatogonial differentiation such as KIT, SOHLH1, and SOHLH2.

Additional evidence supporting the importance of regulated protein synthesis during spermatogonial development comes from studies of mutant and KO mice. The first is from studies using “juvenile spermatogonial depletion” (jsd) mutant mice, which exhibit a normal first wave of spermatogenesis followed by a complete loss of all germ cells in the adult except for Aundiff spermatogonia [140-143]. The mutated gene responsible for this phenotype was later identified as Utp14b [141, 142], which is an autosomally encoded, processed retroposon copy of the X-linked Utp14a gene required for 18S ribosomal RNA processing during ribosome biogenesis. Interestingly, raising testicular temperature in adult mice restores spermatogenesis, leading to the creation of fertilizable sperm [144-146]. This suggests that this naturally occurring mutation is temperature-sensitive, which may explain why no phenotype is apparent during the first wave of spermatogenesis, much of which occurs at 37°C. The phenotype of adult mice, which lack UTP14B function, suggests that ribosome biogenesis is an important aspect of spermatogonial proliferation and differentiation but is not critical in Aundiff spermatogonia. The second example is provided by mice lacking Eif2s3y, a Y-linked gene that encodes a subunit of the eukaryotic initiation factor complex EIF2, which forms a ternary complex with GTP and Met-tRNA during translation initiation. KO testes only contain GFRA1+ Aundiff spermatogonia, indicating that EIF2S3Y is required for spermatogonial proliferation and differentiation [147-151]. Coincidentally, Eif2s3y is one of the two genes (the other being Sry) on the Y chromosome required for male fertility [152]. What specific role this EIF2 gamma subunit isoform plays during spermatogonial proliferation and differentiation is currently unknown, but it is tempting to speculate that it is required for the program of translational activation of repressed mRNAs, as described above.

A perplexing question is why would undifferentiated male germ cells rely on a system of translational control over a subset of mRNAs, as it seems to be a rather inefficient use of cellular resources? This difficult question has many potential answers, and we may never know which one is the most correct. First, transcription is not a particularly expensive process in terms of cellular ATP output, especially in comparison with protein synthesis and degradation [153, 154]. The simplest explanation may be that prospermatogonia and Aundiff spermatogonia lack the ability to precisely control transcription of certain genes and therefore employ posttranscriptional controls as a means to regulate gene expression. An example is provided by the KIT receptor tyrosine kinase; whereas the mRNA is detectable throughout testis development, the protein is only present in discrete stages (for review, see [52]). Protein is expressed in primordial germ cells and is required for their proper migration to the developing fetal gonad. After their colonization, KIT protein expression is silenced for over a week in prospermatogonia but is required again beginning at P3–P4 in a subset of type A spermatogonia for their differentiation [52].

KINASE SIGNALING DURING SPERMATOGONIAL DIFFERENTIATION

The best-studied action of RA is genomic, in which RA stimulates transcription by binding RARs on RA response elements (RAREs) in target gene promoters. However, lack of significant changes in steady-state mRNA levels after RA exposure suggests other avenues of RA-based regulation may be utilized in the neonatal testis [98, 107]. Compelling evidence in other systems indicates that RA can also utilize alternative, nongenomic pathways via kinase cascades [155, 156]. For example, RARA is bound to the regulatory subunit (p85) of PI3K in several cell types (SH-SY5Y, NIH3T3, and MEFs). RA addition causes the recruitment of the catalytic subunit (p110) to induce rapid phosphorylation of ERK and AKT [155, 156]. In support of this, our laboratory discovered that RA activates the PI3K/PDK1/AKT/mTORC1 signaling pathway, and this is required for translation of repressed mRNAs such as Kit, Sohlh1, and Sohlh2 [42, 58, 120]. Activation of the PI3K/AKT signaling pathway following binding of the receptor tyrosine kinase KIT by KITL is essential for spermatogonial development [157-160]. It is plausible that a main role for RA in the testis is to initiate the expression of KIT protein in Adiff spermatogonia, which then binds KITL to maintain activated PI3K/AKT signaling.

Signaling through PI3K and AKT transmits growth, proliferation, differentiation, and survival signals [161]. These signals can directly impinge upon protein synthesis regulation through mTORC1 activation. Addition of exogenous RA to P1 mice led to an increase in mTOR phosphorylation (and therefore activation) in prospermatogonia [58]. Among the substrates phosphorylated by mTOR are the EIF4E-binding protein EIF4EBP1, which specifically mediates cap-dependent mRNA translation, and ribosomal protein S6-kinase (S6K), which modifies ribosomes directly in conjunction with their enhanced synthetic activity [162, 163]. Thus, phosphorylation of EIF4EBP1 and S6K provide two direct, but distinct, indicators of translational activity in prospermatogonia, as they differ in response to RA.

The importance of kinase signaling in spermatogonial proliferation and differentiation is highlighted by genetic studies in which members or regulators of this signaling pathway are conditionally inactivated with Ddx4-Cre, which is first expressed at E15 in fetal prospermatogonia [164]. In the first study, deletion of Pdk1 (also Pdpk1, activator of AKT) or Foxo1 (downstream target of AKT) results in reduced KIT expression and impaired spermatogonial differentiation [165]. In another study, mTORC1 is precociously activated in Aundiff spermatogonia by deleting Tsc2, which normally functions to indirectly inhibit mTORC1 activity through RHEB [166]. Results of that study indicate that suppression of mTORC1 is important for SSC maintenance, which implies that mTORC1 activation is required for differentiation. A complementary study from our laboratory supports this, as we found that mTORC1 inhibition by rapamycin blocks spermatogonial differentiation in vivo [120]. Taken together, these results support a model in which RA signals through kinase pathways that in turn activate mTORC1 as a requisite step in spermatogonial differentiation (Fig. 4).

FIG. 4

RA stimulates transcription and PI3K/AKT/mTOR kinase signaling in differentiating spermatogonia. RA transcriptionally activates genes required for meiosis (e.g., Stra8 and Rec8) and enhances the translational efficiency of repressed mRNAs required for spermatogonial differentiation (e.g., Kit, Sohlh1, and Sohlh2) through activation of kinase signaling.

CONCLUSION AND FUTURE DIRECTIONS

Spermatogonia differentiate in response to RA and undergo largely unknown, yet clearly essential, cellular and molecular changes that precede meiosis. A major form of gene regulation during spermatogonial development may be exerted posttranscriptionally downstream of kinase signaling pathway activation. Future work should focus on identifying changes in the proteome and further clarifying the involvement of signaling pathways and the consequences of their actions. These efforts will, in our opinion, prove fruitful in the identification of specific cellular processes during differentiation that prepare spermatogonia for entry into meiosis. In addition, exploring this understudied phase of male germ cell development will have the added benefit of providing critical insights regarding the programs that precede (SSC self-renewal) and follow (meiosis).