I. INTRODUCTION

In mammals, gametogenesis ultimately requires that diploid germ cells undergo the process of reduction division known as meiosis to form functional gametes. However, oogenesis and spermatogenesis occur at very different times during development and achieve different endpoints. In females, this process is initiated in the fetus well before birth, with the goal of forming a finite number of stored gametes that are used periodically over a defined reproductive lifetime. In males, meiosis is not initiated until postnatal life at the onset of puberty, and the goal is to form the millions of gametes required for male fertility (37).

The generation of sperm via spermatogenesis is a continuous process throughout the reproductive lifetime or season of animals. The end products (sperm) are expelled (spermiation) from the organ, and the next generation of sperm begins to develop from spermatogenic stem cells. Therefore, to maintain the continuum of sperm production, the initiation of spermatogenesis and spermiation must be coordinated. In most mammals the time required to generate spermatozoa from spermatogenic stem cells is ∼30-40 days (19).

The necessity for continual production of a large number of mobile gametes imposes a number of requirements on spermatogenesis. First, a thriving stem cell population is necessary throughout the reproductive lifetime of the organism. Second, to produce enough gametes to ensure fertilization, a major expansion of progenitor cells is required. Third, the need for morphological transformation of sperm and the acquisition of mobility requires the expression of genes unique to spermiogenesis. Fourth, a high level of organization and control is required to ensure the continuous availability of spermatozoa.

The spermatogonial stem cell population (SSCs) must be able both to self-renew to maintain stem cell populations and to generate progenitor cells that proceed through spermatogenesis to form sperm. The determination and fate of the SSC population is determined by complex interactions between the germ cells, the testicular somatic cells, and a number of growth factors. Failure of the SSC population to function properly, in either self-renewal or the generation of progenitor cells, results ultimately in the failure of spermatogenesis.

Differentiating spermatogonia, spermatocytes, and spermatids develop from stem spermatogonia through a well-defined progression of mitotic expansions, meiotic reduction divisions, and morphological transformations. Growth factors and hormones tightly regulate many of these crucial steps leading to the successful production of spermatozoa. Because of recent breakthroughs in the understanding of these early events, this review focuses on the commitment of male germ cells to meiosis. The case will be made that this commitment occurs when undifferentiated A spermatogonia undergo an irreversible transition to differentiating A1 spermatogonia (A to A1 transition). This A to A1 transition generates the germ cell component of the complex architecture of the testis and ensures constant generation of gametes. Consideration of the commitment to meiosis requires an understanding of this complex architecture and the basis for the necessity of asynchronous spermatogenesis. Most of the available information on this process has been obtained from the mouse, so the following discussion focuses on mouse spermatogenesis.

The key to the commitment to meiosis is the action of RA on cells competent to the spermatogenic lineage. Since retinoids are common therapeutics for treatment of cancer and acne (14, 104), it is important to understand their role in the normal testicular architecture and function.

II. FORMATION OF THE TESTES, ESTABLISHMENT OF THE STEM CELL POOL, AND DEVELOPMENT OF UNDIFFERENTIATED SPERMATOGONIA

The formation of the germ and somatic cell lineages of the testis occur independently. During mammalian murine embryonic development, the bipotential gonad can be first seen as a thickening of the ventral side of the mesonephros at embryonic day E10. Over the course of the next 2 days, these gonads acquire their sex specificity, driven primarily by the expression of Sry only within cells destined to become the Sertoli cells of the testis. The absence of Sry expression in females leads to a cascade of gene expression events that drives the formation of granulosa cells in the developing ovary. Primordial germ cells (PGCs) first appear within the epiblast at E6, undergo both passive and active cell migration before finally arriving at and infiltrating the developing gonadal ridge at E11.0 (80). Once the PGCs interact with embryonic Sertoli cells, myoid cells and Leydig cells by E12.5, sex determination is complete and the embryonic testis is formed. (20, 21). At this point, the PGCs are found in close association with Sertoli cells and together they form seminiferous cords that will ultimately become the seminiferous tubules. The PGCs within an embryonic testis undergo a period of mitotic proliferation within the cords and are then known as prospermatogonia or gonocytes. The term prospermatogonia is a more specific and descriptive designation (79). After expansion of the population, the prospermatogonia enter a quiescent nonproliferative phase until around the time of birth in the rodent. The prospermatogonia are located initially towards the center of the seminiferous cords but ultimately migrate to the periphery where several important transitions occur, resulting in the appearance of the morphologically distinct spermatogonia within the first few days after birth. Some of the prospermatogonia find the appropriate conditions (the stem cell niche) to begin to function as spermatogenic stem cells designated As spermatogonia or SSCs (for a review, see Ref. 90).

It has been proposed that some of this pool of prospermatogonia in mice can initially directly form differentiating spermatogonia to constitute what has been termed a “first wave” of germ cell development (87). The term wave in this context may be confused with the “spermatogenic wave” discussed later in this review, and the term first round is more appropriate to refer to this process. The first round of spermatogenesis in mice can be defined as the first cohort of germ cells that progress to spermatozoa. The first round of spermatogonia develop from a unique neurogenin 3 (NGN3) negative pool of prospermatogonia that transition directly into A1 cells while the NGN3 positive spermatogonia remain undifferentiated and comprise the pool of progenitor cells and possibly the stem cells. The NGN3 negative A1 spermatogonia have been first observed at 2 dpp, and they appear to differentiate directly into A3 and A4 spermatogonia and into type B spermatogonia by postnatal day 5 (32, 129). Subsequent rounds of germ cell development arise from stem cell mitoses and progenitor cell expansion. Only a small subset of As cells function as spermatogenic stem cells in the adult, and the development of markers to identify that subset is the focus of many investigators (91). One of the most promising markers to date is ID4 that is expressed in adult testes on a small transplantable subset of As cells in the mouse and has also been detected in human testicular cells (13, 92, 101). The pool of undifferentiated spermatogonia consists of single cells and chains of interconnected cells that arise as a result of incomplete randomly timed cytokinesis (for a review, see Ref. 27). These syncytia can exist as chains of 2, 4, 8, or 16 cells and are termed A paired (Apr) or A aligned (Aal) spermatogonia (25, 30, 56). While all of these cells are considered as “undifferentiated” spermatogonia, the Apr and Aal cells are on the pathway towards germ cell differentiation and are considered transit amplifying progenitor cells (27). The term transient amplifying progenitor cells has also been used (90). Most of the Aal syncytia transition without cell division into A1 differentiating spermatogonia (Figure 1). This transition from Aal to A1 spermatogonia is followed by five synchronized cell divisions to form A2, A3, A4, In, and B spermatogonia, and a final mitosis results in the formation of preleptotene spermatocytes (26). The transition of Aal to A1 represents an irreversible dramatic change in both the morphology and the mitotic behavior of the spermatogonia. This transition is controlled by at least one extrinsic factor retinoic acid (RA) and multiple intrinsic factors (72). In the absence of these factors, undifferentiated spermatogonia do not progress developmentally. Therefore, the A to A1 transition signifies the commitment to meiosis and leads 35 days later to spermiation of haploid gametes. Androgen receptor gene deletions have shown that testosterone is required for early elongating spermatids, terminal differentiation, and release of spermatids (55).

III. ESTABLISHMENT AND MAINTENANCE OF THE CYCLE OF THE SEMINIFEROUS EPITHELIUM AND THE SPERMATOGENIC WAVE

Histological cross sections of the murine testis reveal a seemingly complex interaction between multiple germ cell types. There are recurring well-defined germ cells in different states of differentiation that are always found together. It was first realized in the 1890s that these recurring “cellular associations” can be interpreted to define a cyclic process known as the cycle of the seminiferous epithelium (for a review, see Ref. 70). The recurring cellular associations were first termed the “spermatogenic cycle” in the late 1890s while the current term “cycle of the seminiferous epithelium” was used by LeBlond and Clermont in 1952 (70). The cycle of the seminiferous epithelium was defined as a “series of changes occurring in a given area of the seminiferous epithelium between two successive appearances of the same cellular association.” In 1972, Clermont (19) stated that for a given species each step of spermatogenesis “has a constant duration; thus, germ cell differentiation unfolds as if regulated by a rigid time-scaled program.” The “cycle” as defined above represents the recurring cellular associations at one point within the seminiferous epithelium over time. A single stage of the cycle actually occupies a linear “segment” rather than a point along the seminiferous tubule, representing an area containing similar cellular associations (Figures 2 and 3). Even though one segment may constitute a single stage, the segments representing adjacent stages are arranged in a linear sequence along the length of the tubule. This spatial arrangement of the stages along the length of the tubule constitutes the “wave of the seminiferous epithelium”. Clermont defined the wave of the seminiferous epithelium “as a complete series of the successive cell associations found along a seminiferous tubule.” He continued to define the length of the wave as the distance between two successive, identical cell associations and stated that “the sequence of pictures along a wave is similar to the sequence of events taking place in one given area during a cycle of the seminiferous epithelium”. (19, 70). Both the “cycle” and the “wave” define the timing of the commitment to meiosis, and the key to understanding them is to understand the control of the A to A1 spermatogonial transition. This transition is staggered along the tubule in a progressive manner, and as a result, it defines the initiation of the cycle and the spermatogenic wave. The outcome of this tightly regulated transition is that both the entry into the cycle and the release of sperm occur continuously along the tubules, the hallmark of mammalian male fertility.

Murine male reproduction requires the rapid and efficient production of gametes, so while the development of A1 spermatogonia to the point of spermiation takes 35 days, the initiation of the process occurs much more frequently, every 8.6 days (19). The net result is that at any point along the seminiferous tubule there are multiple developing germ cell types present. In Figure 2, the development of the germ cell component of the seminiferous epithelium of the murine testis is illustrated. This simplified diagram shows that the periodic and frequent initiation of the A to A1 transition of spermatogonia results in a buildup of the germ cell component with the more advanced cells being moved towards the tubule lumen. The end result after 35 days is a germ cell layer consisting of four distinct differentiating germ cell types with spermiation occurring into the lumen of the tubule and initiation of the next round of spermatogenesis at the basal aspect of the tubule. It can easily be seen from Figure 2 that the linear development of germ cells over 35 days also results in the same grouping of germ cells every 8.6 days. Over the intervening 8.6 days between spermiation/initiation events, each layer of cells undergoes a series of developmental steps, and the result reveals the recurring cell associations. In Figure 3 linear germ cell development is shown in more detail, with each 8.6-day window shown as a blue rectangle. The same color codes for germ cell types are shown for one 8.6-day period. The red arrows indicate the point in the 8.6 days where the A to A1 transition occurs. The cycle of the seminiferous epithelium in the first round is shortened by ∼2.5 days from the normal 8.6 days in testes of adult mice. The length of the cycle in immature rodents appears to be generally shorter than that of adult animals (64). It is likely that these examples of accelerated development are confined to the biological advantage of rapid gamete development in short-lived species such as the mouse.

It is apparent that each 8.6-day period results in a series of repeating cellular associations. These repeating associations are readily seen when each blue rectangle in Figure 3A is examined. It is these repeating cellular associations that give rise to the 12 stages of the cycle of the seminiferous epithelium of the mouse, shown in Figure 3B. It is important to note that the development of the germ cells is a continuum and the number of identifiable stages in the cycle for any given species is arbitrary and determined solely by the morphology of the developing germ cell types and the ability of the morphologist to distinguish between them. The cycle of the seminiferous epithelium is found in similar forms in the testes of all mammalian species.

The cycle and the wave differ somewhat in humans and great apes from most mammals in that a cross-section of a seminiferous tubule will contain more than one stage of the cycle. This was originally explained to be the result of a helical orientation of several spirals of spermatogenic waves (105). This type of organization has been referred to as multistage as opposed to the single stage found in most other mammals (74, 125). The notion of whether the spiral or helical arrangement of the stages form a wave has also been challenged, so ambiguities remain (59). The duration of spermatogenesis in humans is reportedly 74 days where one cycle of the seminiferous epithelium is 16 days. Recently, the number of discernible stages has been increased from 6 to 12 (85). It was suggested by Amann in 2008 that the information on the cycle in humans needs to be reinvestigated, but little new information is available (1).

The timing of the cycle is species specific, but the function is similar for all species: to ensure nonsynchronous and continuous spermatogenesis as opposed to synchronous and pulsatile release of sperm. The constant association of particular cell types in stages suggests that the developmental sequence of events occurring in spermatogonia, spermatocytes, and spermatids are all linked with one another during spermatogenesis. If the cycle from many different mammalian species is examined, several commonalities can be seen. First, the A to A1 transition, the onset of the transition of round haploid cells to highly elongated, motile cells, and the process of spermiation are linked to the same or closely aligned stages, and four rounds of the cycle are required for the development of sperm from the onset of spermatogonial differentiation. The mechanism by which this coordination could occur is not well understood, and some have speculated that biological clocks might be involved in regulating developmental decisions in each of the different cell types. However, it has been clearly shown that there are no internal clocks or circadian rhythms present in the testis (84) so the search for a mechanism has continued. Clermont, in his classic review, speculated that “some other factor(s) must be involved to account for the simultaneous entrance in spermatogenesis of spermatogonial stem cells” (19).

IV. EXTRINSIC FACTOR CONTROLLING THE A TO A1 SPERMATOGONIAL TRANSITION

The notion that the A to A1 transition marks the commitment to meiosis is supported by many observations. Prior to the A to A1 transition, the undifferentiated A spermatogonia existing as chains of 2, 4, 8, or 16 cells undergo random proliferation in stages X-II of the cycle of the seminiferous epithelium. Once the A to A1 transition occurs, the developmental response appears to be rigidly timed and irreversible (26). The pattern of spermatogonial proliferation changes as the division of the syncytial chains of A1 to B spermatogonia are highly coordinated to specific stages of the cycle (stages IX to VI) (26). Clermont (19) first deduced that “for a given species, each step of spermatogenesis has a constant duration; thus germ cell differentiation unfolds as if regulated by a rigid time-scaled program.” While mutants or conditions exist in mice that result in a block at one or more developmental steps between the A to A1 transition and spermiation (28), there are no reported circumstances where the timing of the cycle is altered or where the differentiation of A1 spermatogonia is reversed. As explained above, the terms undifferentiated and differentiating spermatogonia are not descriptive of the true nature of these cells. The population of “undifferentiated” spermatogonia, including the stem cells, is, in fact, differentiated enough to be committed to the spermatogenic lineage (27). However, there is no progression beyond this basal level of differentiation, and the Aal cells can arrest for some time unless acted upon by extrinsic and intrinsic signals (27–29, 36, 106). Once the A spermatogonia (Apr to A16 and possibly As) receive these signals at stage VIII of the cycle of the seminiferous epithelium, there is a rapid transition to a unidirectional differentiation pathway committing the cells to meiosis (27).

The extrinsic factor that signals the mammalian A to A1 spermatogonial transition and meiotic initiation is vitamin A in the form of RA (40, 72). It has been known since 1925 that vitamin A is required for normal spermatogenesis (for recent reviews of this subject, see Refs. 53, 54). Vitamin A-deficient (VAD) mouse testes contain only As, Apr, and Aal spermatogonia and Sertoli cells within the seminiferous epithelium (41, 81, 82, 119, 120), so it is a deficiency of RA that prevents the A to A1 transition from occurring (131). If undifferentiated A spermatogonia are stimulated with RA, several gene products are induced, marking the appearance of A1 spermatogonia. Several additional spermatogenic processes, including blood-testis barrier remodeling and spermiation (15, 17, 18, 43), are also regulated by RA. The well-characterized responses to RA all occur around stages VIII of the cycle of the seminiferous epithelium. As a result, it was proposed that a pulse of RA at high enough concentrations to trigger these responses was generated at stages VIII–IX of the cycle of the seminiferous epithelium. As the stages changed along the length of the seminiferous tubules (spermatogenic wave), the pulse of RA would follow a similar course (Figure 4) (53). RA pulses or gradients have been proposed to control a number of developmental processes, but detection and measurement of these pulses have always been limited by the tissue availability and low concentration and brief half-life of RA (98). Recently, a number of advances have allowed for detection and measurement of the RA pulse in the seminiferous tubules (50).

In mammals, vitamin A in the form of retinol originates in the diet or from stored reserves in the liver and can be transported to target tissues bound to serum retinoid binding proteins. Once in the target tissues, retinol can be converted to retinyl esters for storage or oxidized to RA (73). The conversion of retinol to RA is controlled by two sequential oxidative enzymatic steps catalyzed by the retinol or alcohol dehydrogenases and retinaldehyde dehydrogenases (115). It has been suggested that RA may act primarily in a paracrine manner such that one cell stores retinol and oxidizes it to RA that then activates an adjacent target cell (33). Within target cells, RA interacts with heterodimers of the RA receptors (RARs) and the retinoid X receptors (RXRs) and binds RA response elements (RAREs) in target genes, recruiting co-repressors or co-activators to either inhibit or induce transcription (77). Three cytochrome P-450 enzymes, CYP26A1, CYP26B1, and CYP26C1, degrade RA to inactive forms, and a precise balance is generated between the synthesis and degradation of RA (for a review, see Ref. 33).

Most of the known components of the RA synthesis, signaling, and degradation pathways are present in the testes (Figure 5). Retinol is delivered to the testis bound to the retinol binding protein 4 and transthyretin complex (for review, see Ref. 49). The only identified receptor for this complex is Stra6 (63, 130) that is expressed in the testis on Sertoli cells (7, 36). However, it has been shown that the Stra6 receptor is not essential for maintaining vitamin A homeostasis in any tissue other than the eye, and Stra6-deficient mice have normal fertility (7). It appears that at least in early prepubertal stages that retinol dehydrogenase 10 (RDH10) is the primary enzyme responsible for the first oxidation step. While RDH10 is found in both Sertoli and germ cells, the gene knockout of Rdh10 in Sertoli cells alone has the most severe phenotype (117). For the second oxidation step, all three known retinaldehyde dehydrogenases (ALDH1A1, ALDH1A2, and ALDH1A3) are found in the testis. Two of the retinaldehyde dehydrogenases have unique cellular localizations in the testis tissue, suggesting that they may have specific roles in RA formation (3). ALDH1A1 was shown to be in Sertoli cells, while germ cells contained ALDH1A2. Based on the activity and abundance of the enzymes, ALDH1A1 was predicted to be the main source of intratesticular RA.

All three known isoforms (α, β, γ) of both the RARs and RXRs appear to be present in germ cells and/or somatic cells in the testis (126). RARα has been localized primarily to Sertoli cells and some germ cells in the adult mouse testis (121). The male RARα knockout mice are sterile, but some advanced postmeiotic germ cell development does occur (16). The testicular phenotype of the Sertoli cell-specific RARα gene knockout mouse and the deletion of all three RAR isoforms in concert specifically in Sertoli cells recapitulated the gene knockout in the whole animal (121). RARα is clearly a key signaling molecule during testis development with a major function in Sertoli cells. RARγ has been detected in A spermatogonia (121); however, the Rarγ-knockout male mice can be fertile but display an altered spermatogonial differentiation (39, 96). Of the RXRs, only RXRβ has been shown to be critical for normal spermatogenesis (122). Deleting Rxrβ globally results in a delay in spermatid release from the seminiferous epithelium, the accumulation of cholesterol esters, and ultimately testis degeneration. These aberrant phenotypes are similar when the gene is deleted only in the Sertoli cells, suggesting that the infertility seen in the male Rxrβ knockout animals is due to the lack of this receptor in Sertoli cells (122).

The precise degradation of RA also appears to be critical for the regulation of RA concentrations and proper spermatogonial differentiation. CYP26B1 is expressed in the immature Sertoli cells in the embryonic testis, degrading any RA and preventing the germ cells from prematurely entering meiosis (37, 66). In the Cyp26b1 knockout mouse, germ cells in the embryonic testis prematurely enter meiosis and undergo apoptosis (71, 75). In the postnatal mouse testis, the expression of Cyp26b1 was reported by in situ hybridization to be restricted to the peritubular myoid cells (123); however, in a recent study CYP26B1 protein was detected in germ cells in a heterogeneous pattern throughout the neonatal mouse testis (110). When testes were treated with a CYP26 enzyme inhibitor, the number of germ cells expressing STRA8 was increased, as was the expression of other genes associated with spermatogonial differentiation (110). Clearly the CYP26 enzymes play a role in spermatogonial development, but whether they are responsible for the degradation of RA after the pulse will require further investigation. In a recent study, CYP26A1 and CYP26B1 were eliminated singly or combined in either germ cells or Sertoli cells or both (48). The elimination of CYP26B1 from both Sertoli and germ cells produced the only phenotype that was severe enough to cause subfertility. So, in summary, both germ and Sertoli cells have the ability to generate RA, use RA to regulate gene expression, and also to degrade RA so that the signal is removed.

V. SYNCHRONOUS VERSUS ASYNCHRONOUS MEIOSIS

The cycle of the seminiferous epithelium and the wave were first described in detail in the rat testis (70). It can be shown that only ∼7% of the tubule length represents stage VIII, where both spermiation and the A to A1 transition occur (46, 100). So even though <10% of the seminiferous tubule length is releasing spermatozoa at any one point in time, the fact that the A to A1 transition is staggered along the length of the tubule in a progressive manner means that spermatozoa can be released continuously. Because of the cycle and the wave, the release of spermatozoa is constant with time. This organization is termed “asynchronous” spermatogenesis, involves an asynchronous commitment to meiosis, and results in a continuous supply of gametes despite the 35–40 days required in most mammals to produce them (Figure 6A). The alternative of synchronous spermatogenesis that results in pulsatile sperm production is illustrated in Figure 6B. All mammals except for some nonhuman primates and humans examined thus far have both a well-defined cycle and a wave that allow for continuous sperm production during the reproductive lifetime or season of the species. As described previously, the arrangement of the stages to form a wave in humans has been questioned, but since sperm production is constant and continuous, some form of control must exist (59). Since asynchrony is so highly conserved, evolutionarily it follows that it confers a distinct biological advantage. In lower species, such as fish, there may be a requirement for the release of a large number of sperm at the same time since a large number of eggs and germ cell cohorts develop synchronously (95).

The first report of synchronized spermatogenesis in mammals occurred in VAD rats that had been treated with an injection of retinol (83). One interpretation of this result is that because of the deficiency, undifferentiated A spermatogonia and stem cells were the only germ cells that remained in the seminiferous tubules. When these deficient animals were then treated with vitamin A, all of the non-stem cell A spermatogonia underwent differentiation and entry into meiosis at the same time. The cycle of the seminiferous epithelium remained the same but there was no spermatogenic wave, and cohorts of germ cells occupying the entire testis entered meiosis and underwent spermiation on every cycle or every 12.6 days (length of cycle of the seminiferous epithelium in rats). The end result was the pulsatile or synchronous production of sperm that lasted for several months (6). Results on rats were soon after extended to mice where even though it was more difficult to produce the VAD model, the results on synchrony were the same (120).

The studies described above demonstrated that normal asynchronous sperm production could be converted to synchronous sperm production using VAD mice or rats supplemented with a bolus of RA. The question that remained was how asynchronous spermatogenesis is set up initially during development. Clues to the developmental onset of asynchronous spermatogenesis were obtained with a transgenic mouse model developed by Janet Rossant expressing β-galactosidase under the control of an RA response element (RARE-hsplacZ mice) (99, 109, 110). In these animals the cells in the neonatal testis undergoing active RA signaling were visualized by β-galactosidase activity and the resulting blue color. β-Galactosidase activity was first detected in spermatogonia of 2 dpp mice and was distributed in patches along the seminiferous tubules (Figure 7). The β-galactosidase activity in these cells colocalized with marker proteins indicating that they were A1 differentiating spermatogonia. These results demonstrated that the action of RA was distributed in a subset of spermatogonia in patches along the tubule and that these distributions lead to nonuniform initiation of the A to A1 spermatogonial transition in the neonatal testis. It has been documented that precocious RA exposure will result in germ cell changes similar to those that occur normally at 3–4 dpp. These changes stimulated by RA included spermatogonia proliferation, maturation of cellular organelles, and expression of markers characteristic of differentiating spermatogonia (12). The interpretation of the RARE-hsplac mice was that these patches of β-galactosidase positive cells represented the visualization of the onset of asynchronous spermatogenesis. Exogeneous RA treatment of testes in culture or injection of RA into neonatal animals eliminated the patches and resulted in consistent blue β-galactosidase-positive spermatogonia along the entire length of the tubule (Figure 7B) (102, 103). When neonates were treated with exogenous RA and then allowed to grow to adults, the spermatogenic cycle was normal, but there was no wave and spermatogenesis was synchronous across the entire testis. These results were extended when mice of different ages starting at 4 dpp were given exogenous RA and the degree of spermatogenic synchrony was assessed when they became adults (24). Interestingly, the level of spermatogenic synchrony was inversely proportional to the presence of preleptotene spermatocytes. If injected at 2 or 4 dpp, the level of synchrony was very high but as the mouse reached the age of 6 or 8 dpp, the level of synchrony was greatly reduced, approaching a normal distribution of stages. These data were consistent with the observation that when RA is injected into adult mice, while some spermatogonia undergo apoptosis, there is no effect on the asynchrony or distribution of stages in the testis of the adult (132). Therefore, in the rodent, the development of asynchrony in spermatogenesis occurs in the neonatal testis before the appearance of preleptotene spermatocytes. The asynchrony is apparent in the β-galactosidase-positive patches that are seen in the seminiferous tubules of the neonatal RARE-hsplacZ mice, and it also becomes fixed in place, i.e., it cannot be modified by exogenous RA once the advance germ cells appear. The mechanism by which this asynchrony appears will require additional study.

VI. THE RETINOIC ACID PULSE

The first suggestion of an RA pulse along the tubules was a result of several observations describing the actions of RA on spermiation and tight junctions at stages VIII–IX of the cycle of the seminiferous epithelium and the localization of mRNAs associated with retinoid storage and RA synthesis and degradation in a stage-enriched manner (112, 121). More recently, the discovery of an RA responsive gene known as Stra8 expressed primarily at stages VIII–IX of the cycle strengthened those observations.

Stra8 or ”stimulated by retinoic acid gene 8“ was first described as one of a group of RA-responsive genes in a cell line (93). Stra8 is a vertebrate-specific, germ cell-specific transcript that is an excellent marker for the action of RA on the germ cells. Stra8 expression can be induced by RA both in vivo and in vitro in the intact mouse testis and in isolated germ cells (131, 132). Stra8 is induced in the presence of RA, and STRA8 protein accumulates in both spermatogonia and preleptotene spermatocytes primarily at stage VIII–IX of the cycle of the seminiferous epithelium (66). In Stra8 gene deletion studies in the male mouse, meiosis is blocked at the leptotene stage, and the A to A1 spermatogonial transition is altered (2, 5). As described above, it is during stage VIII that the A to A1 transition takes place and preleptotene spermatocytes are formed. STRA8 appears to be required for both of these processes to occur normally (106), yet Stra8 induction is not identical to RA deficiency as A to A1 transition can occur when the Stra8 gene is deleted (76). In normal mouse testes, the STRA8 protein is detected in spermatogonia as early as 2 dpp (110). The Stra8 transcript in the mouse encodes a 45-kDa protein that has two domains linked by ∼80 amino acids composed primarily of polyglutamic acid. The two protein domains are well conserved across species, but the linker region is highly variable and is not present in human STRA8. There is some evidence that STRA8 can shuttle between the nucleus and cytoplasm and can bind to DNA (114), but the exact function of this protein in germ cells has never been described. So, the induction of Stra8 serves as an excellent biochemical marker for the A to A1 transition and the action of RA, but very little is known about the function of STRA8 protein.

The actual measurement and verification of the RA pulse was possible as a result of the use of bis-(dichloroacetyl)-diamines (BDADs) (52, 94). These compounds inhibit spermatogenesis and have been shown to be effective as male contraceptives in many species (31, 45, 107). Whole testes or germ cells isolated from mice at 2 dpp were incubated in culture with a BDAD known as WIN 18,446. When cultured in the presence of WIN 18,446 alone or WIN 18,446 plus retinol, Stra8 expression was dramatically suppressed (52). These results and additional studies have demonstrated that WIN 18,446 inhibited the conversion of retinol to RA by inhibiting the retinaldehyde dehydrogenases responsible for the oxidation of retinaldehyde to RA (49, 52). This compound is especially active in the testis and because of the short half-life of RA, the testis becomes RA deficient very quickly (3). This compound and RA supplementation can be used to efficiently generate mice with induced synchronous spermatogenesis such that the entire testis is comprised of a few closely related stages of the cycle of the seminiferous epithelium (51). WIN 18,446 was given to 2 dpp mice once daily for 7 consecutive days. This treatment blocked the A to A1 transition, and the testes in the resulting 9 dpp mice accumulated A undifferentiated spermatogonia. The mice were then treated with RA, which induced almost all A spermatogonia to differentiate and become A1 spermatogonia nearly simultaneously. These mice, maintained until they were adults, displayed highly synchronized spermatogenesis (51). By collection of testes at appropriate intervals following RA injection, testes synchronized to intervals covering the entire cycle of the seminiferous epithelium were generated and used for analysis of stage-specific spermatogenic events (51). This procedure was used to measure the RA levels in different stages of the cycle by liquid chromatography-tandem mass spectrometry. A distinctive peak of RA was found primarily coincident with stages VIII–IX of the cycle and the expression of STRA8 in spermatogonia and preleptotene spermatocytes (50) (Figure 8). These data supported the current understanding of the in vivo RA response in the seminiferous tubules based on immunohistochemistry of STRA8 expression. It should also be noted that since these synchronized testes contain more than one stage and stage determination is completely qualitative, the actual peak width may differ somewhat from the reported values. It appears as if RA levels begin to rise late in stage VII and remain high in stages VIII and IX (50). These data differ somewhat from the predicted model of Hasegawa and Saga (43) and Sugimoto et al. (112) who presented a model showing high RA levels throughout stage VII. Most importantly, these data verified that processive pulses of RA are generated during spermatogonial differentiation. These pulses are the likely trigger for cyclic spermatogenesis and allow us, for the first time, to understand how the cycle of the seminiferous epithelium is generated and maintained. In addition, this study represented the first direct quantification of a retinoid gradient controlling cellular differentiation in a postnatal tissue.

So, if RA is the key extrinsic factor in the testis controlling the cycle of the seminiferous epithelium, the question becomes which cells, enzymes, and receptors are instrumental in generating, responding to, and degrading the RA pulse. If the paracrine paradigm holds true for the testis, the Sertoli cells would be the RA generating cells and the A spermatogonia would be the RA responsive cells. However, in the murine testis, both the germ cells and the Sertoli cells can store retinol, can oxidize it to RA, and can respond via the RAR/RXR receptors. Immunohistochemistry, mRNA detection via arrays or sequencing, and cell specific gene deletions have resulted in a model where the RA is synthesized by the Sertoli cells and acts in a paracrine manner on germ cells, but only for the first round of spermatogenesis. The model then proposes that RA synthesized in spermatocytes or spermatids then stimulates the subsequent cycles of spermatogenesis in a paracrine manner (24, 53, 54, 77, 112). The RA synthesized by Sertoli cells initiates the first round (97, 117), and because it does so in a patchy, nonuniform manner as described previously, the result is the initiation of spermatogenesis in an asynchronous manner. Any interference in this patchy initiation, such as RA deficiency and regression to A spermatogonia or injection of exogenous RA prior to formation of preleptotene spermatocytes, results in synchronous spermatogenesis in the adult (24, 109).

The support for this model is based on results from several studies, but it is also possible that there are alternative explanations. Gene deletions specifically in Sertoli cells of a single retinol dehydrogenase gene (Rdh10) (117) and in all three retinaldehyde dehydrogenase genes (Aldh1a1, Aldh1a2, Aldh1a3) (97) resulted initially in a block of spermatogenesis at the A to A1 transition. Both of these studies concluded that RA synthesis in Sertoli cells was necessary for the first round of spermatogenesis. Both studies suggested a paracrine action of RA from Sertoli cells on germ cells to initiate the A to A1 transition (97, 117). However, after a single injection of RA into these mutant mice, spermatogenesis proceeded and continued for several months, indicating that once the germ cells entered into meiosis there was a new source of RA that maintained spermatogenesis. Both Sertoli and germ cells contain retinol dehydrogenases (RDH) to carry out the oxidation of retinol to retinal (117). When Rdh10 was deleted specifically in Sertoli cells, there was a mild germ cell deficiency, but when Rdh10 was deleted in both germ and Sertoli cells simultaneously, the phenotype in juvenile mice resembles that seen with complete RA deficiency, with only undifferentiated spermatogonia present. However, even in these mice, spermatogenesis recovered after 4 wk without administration of exogenous RA. It is possible that some germ cells entered meiosis in the mutants as a result of incomplete gene deletion and, after 4 wk, these advanced germ cells produced sufficient RA to maintain spermatogenesis. Alternatively, there was a developmental shift at 4 wk of age to expression of an additional RDH gene in Sertoli cells. In the triple retinaldehyde dehydrogenase (RALDH) gene deletion study, additional experiments concluded that the source of RA in the recovered mice was likely the pachytene spermatocytes (97). A study that examined the localization of the transcripts encoding the RALDH enzymes also concluded that pachytene spermatocytes and round spermatids were a likely source of RA in the adult testis (112). Clearly, advanced germ cells can alter the RA response that is seen in the first round of spermatogenesis (24, 97). However, if advanced germ cells are the source of RA to drive the next A to A1 transition, it would seem that the preleptotene spermatocytes would be the primary candidates (24). This conclusion is supported by the identification of ALDH1A2 protein within these cells by Western blotting (97). After the first round of spermatogonia undergo the transition, the cycle is reinitiated 8.6 days later in mice, and the most advanced cells at that time are the preleptotene spermatocytes. Whether advanced germ cells are involved in generating a pulse of RA in the adult or not, there are clearly alternative sources of RA that come into play. Whether that RA results from another cell type or induction of new metabolic enzymes as development proceeds remains to be determined.

VII. COMPETENCY OF GERM CELLS TO ENTER MEIOSIS

Cells commit to the germ cell lineage within the epiblast in mammals but maintain some plasticity (69). As development proceeds and the gonad is formed, the plasticity is decreased and commitment to the germ cell lineage is increased. In the postnatal testis, prospermatogonia become spermatogonial stem cells and undifferentiated A spermatogonia that eventually commit to meiosis. In the male it has been noted based on many publications from several laboratories that there are two cell competencies that are affected by RA within stages VII–IX. The A to A1 transition requires the competency of A undifferentiated spermatogonia to respond to RA (commitment to meiosis) as does the B spermatogonia to preleptotene spermatocyte transition (entry into meiosis) (35). So while RA is the extrinsic trigger, intrinsic factors are required that allow the cells to become “differentiation or meiosis competent” (9). Most cell types when treated with RA do not induce Stra8 or other meiotic genes. Some non-germ cell types such as embryonic stem (ES) cells or embryonal carcinoma cells can respond to RA by induction of Stra8 (9) but do not enter into meiosis. To become competent to commit to meiosis, the spermatogonia probably need to express some of the core pluripotency transcription factors such as POU5F1, SOX2, and NANOG and have a more open chromatin structure characteristic of many stem cells (9, 89). The reason why ES cells do not commit to meiosis and spermatogonia do is still an open question. Some meiosis-specific chromatin changes in germ cells have been reported to occur shortly after the formation of the gonad, and access to RA at this time results in premature entry of germ cells into meiosis (8, 71). Most undifferentiated A spermatogonia, whether Apr or Aal, respond to RA with induction of Stra8 and appear competent to commit to meiosis. There is a large pool of these cells (transit amplifying progenitor cells) in the testis, with the exception of stages VIII-IX where they have access to RA and undergo the A to A1 transition. Thus one function of intrinsic factors in the testis is to maintain this pool and perhaps tightly regulate responsiveness to RA. While there is some evidence suggesting that a subset of differentiating spermatogonia (NGN3-positive cells) are the only ones that respond to RA (57), other studies have shown that most undifferentiated spermatogonia respond by undergoing the A to A1 transition (51).

For the normal initiation of spermatogenesis, transcription or growth factors in the postnatal testis must function at four key points: 1) to maintain stem cell self-renewal, 2) to generate and maintain the transit amplifying progenitor pool of spermatogonia, 3) to maintain meiotic competency, and 4) to aid in the A to A1 transition. While the search for and the description of gene networks that regulate these four points are the focus of many on-going studies, a number of key factors that influence the maintenance of the spermatogonial pool have been identified (111). Glial cell-derived neurotrophic factor (GDNF) is a key regulatory factor that is produced and secreted by Sertoli cells. A subset of undifferentiated A spermatogonia express ret proto-oncogene (RET) and GDNF family receptor alpha 1 (GFRα1) that serve as the GDNF receptor complex (11, 44, 47, 58, 61). GDNF is required for the survival and proliferation of undifferentiated A spermatogonia in culture (23, 65, 67, 68, 86, 128), and if GDNF or the receptor complex is knocked out, the A spermatogonia are eventually lost due to a lack of proliferation and the differentiation of these undifferentiated spermatogonia into A1 spermatogonia is blocked (34, 47, 88). GFRα1 is expressed by undifferentiated spermatogonia in the postnatal mouse testis (124). When the expression of GDNF across the cycle was examined, it was found to vary 12- to 14-fold, with the highest levels detected at stages XII–I and the lowest at stages VII–VIII (61). GDNF expression in this cyclical fashion by Sertoli cells may be instrumental in the stage-specific proliferation of undifferentiated spermatogonia, as stages XII–I represent the period of time when the pool of A spermatogonia is proliferating (103).

A number of other factors important for the proliferation and maintenance of undifferentiated spermatogonia have been described, including OCT3/4, NANOG, SOX2, and LIN28, and these are also factors that regulate pluripotency in mammalian stem cells. Some of these factors are expressed in both germ and somatic cells in the testis (127). ZBTB16, a transcriptional repressor expressed in A spermatogonia (10), is coexpressed with OCT4 and regulates the epigenetic state of undifferentiated cells (10, 22). The expression of ZBTB16 is commonly used as a distinctive marker for undifferentiated spermatogonia.

Identification of several factors important in the maintenance of meiotic competency and in influencing the A to A1 transition have also been identified. Doublesex-related transcription factor (DMRT1) and Deleted in Azoospermia-Like (Dazl) are intrinsic factors important in the decision of germ cells to remain undifferentiated or to enter the differentiation pathway. DMRT1 is a transcription factor expressed by both germ and Sertoli cells. With the use of conditional gene targeting in the mouse, it was shown that the germ cell-specific loss of DMRT1 in adult spermatogenesis results in the precocious ability of spermatogonia to induce Stra8 and enter meiosis. DMRT1 apparently opposes the action of RA but activates transcription of Sohlh1, another spermatogonial differentiation factor described below (78). The Dazl gene is another important intrinsic factor in mammalian competency for meiosis. There is evidence to suggest that DAZL enables germ cells to respond to RA and opposes the activity of DMRT1 (72). DAZL is a member of a protein family that includes DAZ, DAZL, and BOULE. They are all germ cell-specific RNA-binding proteins that have been implicated in the translational regulation of multiple transcripts. DAZL is present in vertebrates, and recent studies suggest that DAZL can function in translational repression and transport of specific mRNAs (108). While embryonic testicular germ cells can normally express Stra8 when exposed to RA, the germ cells from embryonic Dazl-deficient animals fail to express Stra8 under the same conditions (72).

One mechanism to induce differentiation of spermatogonia would be to inhibit the factors described above that maintain the undifferentiated state. Recent studies have shown that when A spermatogonia undergo RA-induced differentiation to A1 spermatogonia, the expression of LIN28 is suppressed and, as a result, the Mirlet7 family of miRNAs are induced and further downregulate a number of genes associated with undifferentiated A spermatogonia (116). In addition, two related helix-loop-helix transcription factors (SOHLH1 and SOHLH2) are essential for the differentiation of spermatogonia. Loss of SOLHLH1 expression disrupts the differentiation of spermatogonia into spermatocytes (4), and SOHLH2-null mice display a block in spermatogenesis and eventually show degenerating colonies of differentiating spermatogonia and a disrupted appearance of KIT-positive spermatogonia (42, 118). It has recently been proposed that SOHLH1 and SOHLH2 suppress genes involved in SSC maintenance, such as Gfrα1, and induce genes important for spermatogonial differentiation, such as Kit (113). However, both SOHLH1 and SOHLH2 can be found in both KIT-positive and KIT-negative spermatogonia.

In a unique approach, Evans et al. (2014) used RiboTag mice (102) combined with WIN 18,446/RA treatment to synchronize the first round of germ cell development (36). The RiboTag mice allow for the germ cell specific isolation of polysomes from a total testis, and the WIN 18,446/RA treatment protocols allow for testes highly enriched for different germ cell types to be generated (52, 102). The transcriptome of the A spermatogonia (from WIN 18,446 only treated animals) and of the various differentiating spermatogonial populations (from WIN 18,446 plus RA-treated animals following different intervals of recovery) were then examined after collection of the germ cell-specific polysomes. Of 17 known meiotic transcripts investigated, 9 were found to be associated with polysomes in spermatogonia before the appearance of spermatocytes. Many were expressed at low levels in A spermatogonia and at higher levels in A1 spermatogonia. These data suggest that once the germ cells are committed to differentiation, at the A to A₁ transition, they begin to build and load mRNAs encoding meiotic proteins onto ribosomes and support the concept that the irreversible commitment to meiosis occurs at this transition.

While the bulk of the studies on meiosis and development factors in the testis have been done on mice, there is some evidence that similar factors and regulatory patterns exist in the human testis. In a recent study where the expression of several factors including STRA8 and DMRT1 were examined in human testis samples, the authors concluded that the expression pattern of these factors was “largely conserved in the human gonads” (62).

VIII. SUMMARY

It was realized by several microscopists in the 1890s that the organization of the testis produced recurring patterns of cells, and these patterns were repeated throughout the reproductive lifetime of the mammal. However, it was not until the now “classical” manuscript by Leblond and Clermont was published in 1952 that the cycle of the seminiferous epithelium was defined and standardized (70). In their summary of this manuscript, these authors noted that “a topographic description was made of the 14 stages of the cycle” of the rat and suggested that it could serve as the basis for future histological studies of spermatogenesis. In retrospect, their suggestion was understated as the universality and centrality of a similar cycle has been shown for all mammalian spermatogenesis. While learning and identifying the stages of the cycle has captivated and confused students of reproduction for decades, very little attention has been focused on the significance of the cycle and the driving forces that create it. The complex architecture of the testis is generated by the “rigid time-scaled development” (70) of the germ cells. This rigid timing appears to be inherent in the genotype of the germ cells (38). The timing and resulting architecture requires a fixed starting point to establish the cell-cell relationships that have been termed the “cycle of the seminiferous epithelium.” The initiation point is the transition of A undifferentiated spermatogonia into A1 differentiating spermatogonia. This transition sets the rigid developmental timing into place and serves as the germ cell commitment to meiosis. Pulses of RA irreversibly generate the A to A1 transition during stages VII, VIII, and IX of the cycle of the seminiferous epithelium. Because of the rigid timing, the transition and hence the RA pulse is staggered at regular intervals along the seminiferous tubule (spermatogenic wave) to enable the constant release of mature spermatozoa in a process known as asynchronous spermatogenesis. The commitment to meiosis (A to A1 transition) and the release of spermatozoa (spermiation) are both responsive to RA allowing for coordination of spermatogenesis.

GRANTS

This work was supported by National Institutes of Health Grants HD10808 and U54 42454.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author.