INTRODUCTION

Research on spermatogenesis in vitro has been ongoing for nearly a century [1, 2]. To the best of our knowledge, however, for mammals, it is not yet possible to make spermatogonia differentiate into haploid cells in vitro [3–5]. After a long nascent period, the research progressed rapidly in the 1960s, mostly owing to extensive studies performed by A. Steinberger and E. Steinberger using an organ culture system [6–8]. They succeeded in inducing gonocytes to differentiate up to the pachytene stage of meiosis using immature rat testes [6]. One of the important aspects of their culture method was the placement of fragments of testis tissue on the surface of the medium so as to allow the easy exchange of oxygen while also allowing them full access to the nutrients in the medium. Trowell [9], who developed this so-called gas-liquid interphase method, applied it to a variety of tissues including the testes, but it is not clear whether it successfully promoted spermatogenesis. In addition to the gas-liquid interphase method, the Steinbergers adopted a new culture medium developed by Eagle [10]. The combination of these innovations contributed to the in vitro differentiation of gonocytes/spermatogonia into meiosis, which was confirmed histologically [6]. However, the above researchers did not observe the completion of meiotic division or the production of spermatids in their experiments. Even with more developed testis fragments from 14-day-old rats, in which meiosis should have started, the progression of meiotic differentiation beyond the pachytene stage did not take place [8]. These results therefore unequivocally demonstrated a limitation of the organ culture method and presented a new challenge to the research community, i.e., to produce haploid spermatids in vitro. In fact, thereafter, several research groups tackled this subject and reportedly succeeded in producing haploid cells in vitro [11–14]. These studies abandoned the organ culture method and adopted the cell culture method instead. However, most of these studies used samples consisting of spermatocytes, and the rest included spermatocytes in their samples, as starting materials of the culture; i.e., meiosis had already commenced at the initiation of culturing. In addition, no protocol for in vitro meiosis has been repeated and verified by other independent research groups. Therefore, the in vitro replication of the whole meiotic process of mammals has not yet been achieved.

After the Steinbergers' study, the organ culture method for spermatogenesis has been used only very limitedly. It has mostly been used as a biological assay of soluble factors to assess their effects on cells of testis tissue, including germ cells and Sertoli cells [15–18]. Regarding the meiotic progression of germ cells in vitro, the organ culture system has not even been reevaluated, leaving the results of the Steinbergers' study untested and unimproved for nearly 40 yr. We thought therefore that it would be worthwhile to verify their results using modern equipment. We especially thought that it would be useful to use transgenic mice that express green fluorescent protein (GFP) specifically at meiosis during spermatogenesis. We also thought that it would be useful to use the new culture media now available for germline stem cells. Using such advanced technologies, we reevaluated the achievements of the Steinbergers.

MATERIALS AND METHODS

RESULTS

In order to confirm the timing of the expression of Acr-GFP and Gsg2-GFP in the testes of pup mice, the testes of the transgenic mice were examined throughout the course of their development. At 3.5 dpp testes contained only gonocytes as germ cells in the center of their seminiferous tubules (Fig. 1A). At 7.5 dpp, the gonocytes had moved to the basement membrane of the tubule to become spermatogonia. Although it was when meiosis was about to commence, no apparent spermatocytes were recognized (Fig. 1B). At 14.5 dpp, the seminiferous tubules had grown in size, spermatocytes had accumulated in them, and meiotic progression was observed (Fig. 1C). Acr-GFP was first expressed around this day (Fig. 1D). As the expression of the Acr-GFP starts in the cytoplasm of midpachytene spermatocytes (stage IV), the expression timing at 14.5 dpp corresponds well with the assumption that preleptotene spermatocytes start meiosis at age around 7.5 dpp because it takes about 7 days from type B spermatogonia to midpachytene spermatocytes [25]. At 19.5 dpp, the germ cells increased in number, and each seminiferous tubule showed spermatogenic wave difference (Fig. 1E). This is when Gsg2-GFP began to be sporadically expressed throughout the testis (Fig. 1F). This 5-day interval between Acr- and Gsg2-GFP may not be enough to complete the rest of meiosis after midpachytene, which indicates that Gsg2 expression is not haploid cell-specific but probably starts during meiotic cell divisions.

FIG. 1.

Histological (A–C, E) and stereomicroscopic (D, F) views of two lines of transgenic mouse testes, Acr-GFP (C, D) and Gsg2-GFP (A, B, E, F), from the neonatal to the adolescent period are shown. A) At 3.5 dpp, germ cells observed were only gonocytes (arrows). B) At 7.5 dpp, the gonocytes had disappeared after migrating to the basement membrane to become spermatogonia. C) At 14.5 dpp, meiotic figures of spermatocytes were seen. D) At 14.5 dpp, Acr-GFP was first expressed. E) At 19.5 dpp, many spermatocytes had accumulated in the tubules. F) After 19.5 dpp, Gsg2-GFP expression began in several parts of the seminiferous tubules. Bar = 30 μm (A–C, E), 200 μm (D, F).

For organ culture, we adopted the gas-liquid interphase method originally developed by Trowell [9]. For stands to place the testis tissue on, we tested several materials, including paper filters, nitrocellulose membranes, agar gel, and agarose gel [26]. A preliminary study showed that the testis tissues showed the strongest acrosin- and haspin-GFP expression when they were placed on agar or agarose gel (data not shown). So, we adopted agarose gel of 1.5% (w/v) thereafter in all experiments. The amount of medium was adjusted so that its surface level was just below the upper surface of the agarose gel (Fig. 2A). When medium covered the agarose gel entirely, thereby submerging the tissues, the tissues were compacted and became necrotic and no Acr- or Gsg2-GFP was ever expressed (data not shown).

FIG. 2.

Organ culture of testis tissues of Gsg2-GFP and Acr-GFP transgenic mice. A) Schematic presentation of the culture method is shown. AG, agarose gel; TF, testis tissue fragment; M, medium. B, C, D) Gsg2-GFP mouse testis tissue, 11.5 dpp, histology of starting material (B), sample taken on Day 19 (C), and stereomicroscopic view showing GFP emission on Day 19 (D). E, F) 8.5 dpp Acr-GFP testes tissue fragments cultured for 21 days. Bright field picture (E) and GFP emission picture (F). Bar = 30 μm (B, C), 500 μm (D–F).

Every sample expressed no GFP at the initiation of the culture. Both Acr-GFP and Gsg2-GFP showed their expression in vitro mostly at the same time as they did in vivo. The 11.5 dpp Gsg2-GFP testis (Fig. 2B), for example, expressed GFP at 8–10 days in culture, which corresponds to 19.5–21.5 dpp, its original expression time in vivo, and the expression continued for about 15–20 days (Fig. 2, C and D). Generally, Acr-GFP was first expressed around the age corresponding to 14–18 dpp (Fig. 2, E and F), while Gsg2-GFP was first expressed around the age corresponding to 19–25 dpp.

Based on these findings on Acr-GFP and Gsg2-GFP, we tested the effects of several culture condition parameters. Among them, the temperature had quite an effect on testis organ culture. We compared 32, 34, and 37°C for the expression of Gsg2-GFP. As was reported in the literature, 37°C was detrimental to the tissue, and no GFP expression was observed. The temperature of 34°C was better than 32°C; the difference was significant when testes of younger mice, ∼7.5–10.5 dpp, were used (Table 1).

TABLE 1.

Haspin-GFP expression experiment, temperature comparison.

We also used and evaluated three different culture media, α-MEM, DMEM, and StemPro-34 SFM. Beforehand, we tested the effects of adding FBS to these media and found that it had a profound effect. Without FBS, the tissue architecture was poorly maintained, and GFP expression was rarely observed (data not shown). Although chemically defined medium is ideal for analyzing the effects of ingredients of culture media, we added 10% of FBS to every medium tested thereafter. Among the above three media, α-MEM showed the best results, although no statistically significant differences were recognized (Table 2).

TABLE 2.

Haspin-GFP expression experiment, medium comparison.

In order to make this system more quantitative as an assay system, the extent of expression was graded according to the number of tubules expressing GFP in a piece of tissue. They were divided into four grades: grade 0 was no GFP expression, and grades 1, 2, and 3 represented 1–5, 6–15, and over 15 tubules showing GFP expression, respectively (Fig. 3A). Based on this grading, the effects of lower oxygen and anti-RA molecule were tested to confirm the validity of semiquantitative analysis in this organ culture system. For low-oxygen experiments, N₂ gas was infused into the incubator to lower the oxygen concentration to 10%. Under the lower-oxygen condition, Gsg2-GFP was expressed, but its grade was lower than that in the control, which was performed under regular air of 20% oxygen (Fig. 3B). This result confirmed that a lower oxygen concentration was disadvantageous for spermatogenesis. To see the effect of RA on spermatogenesis in this organ culture system, an RAR antagonist, LE540, was added to the medium. Even though LE540 did not completely inhibit meiosis, Acr-GFP expression was greatly reduced (Fig. 3C). These results demonstrated that our culture system is able to measure the effects of different environmental conditions on spermatogenesis.

FIG. 3.

Semiquantitative evaluation of the extent of GFP expression in seminiferous tubules. A) Testis tissues showing no GFP were scored 0. When the GFP expression extended along the tubule for more than 1 mm, it was counted as GFP positive. The extent of GFP positivity was divided into three grades. Grades 1, 2, and 3 represent 1–5, 6–15, and over 15 GFP-positive tubules, respectively. Bar = 500 μm. B) Effect of low oxygen concentration on Gsg2-GFP expression. The period of Gsg2-GFP expression was almost identical in both groups, but the extent of the expression was higher under normal oxygen concentration. C) Effect of antiretinoic molecule on Acr-GFP expression. LE540, an RAR pan-antagonist, depressed the expression of Acr-GFP throughout the culture period.

The results of 86 experiments using Acr-GFP and Gsg2-GFP mice testis tissues at 34°C in α-MEM supplemented with 10% FBS are summarized in Table 3. In the Acr-GFP mice testis tissues, GFP expression was observed in every case except one, regardless of the age of the mice used. Namely, the progression of spermatogenesis at least up to the midpachytene stage was demonstrated. Also with Gsg2-GFP mice testis, GFP was observed in most cases, but its success rate declined when younger mice were used. Nonetheless, even with mice as young as 4.5–5.5 dpp, the testis tissues showed Gsg2-GFP expression with a frequency of more than 85%. Among Gsg2-GFP expressing samples, the most evidently GFP-expressing 14 samples from independent experiments were examined histologically. One of them was serially sectioned all through the entire tissue. The remaining 13 were sectioned to show their maximum cut surface for observation. It was, however, difficult to morphologically confirm the presence of spermatids in them. The seminiferous tubules showed different extents of growth and contained meiotic cells (Fig. 2C), but the tubular lumen did not fully enlarge, and the tubular diameter did not compare with that in vivo. This finding suggested that meiotic cell division was disrupted by these special constraints. On the other hand, it seemed that thin-sectioned histological examination was not sufficiently sensitive to detect spermatids.

TABLE 3.

Summary of organ culture system, 34°C in α-MEM + 10% FBS.

Then, we adopted another method to detect spermatids among the Gsg2-GFP-positive cells, Hoechst nuclear staining. Hoechst staining of the testis cells of Gsg2-GFP transgenic mice 23, 27, and 29 days old demonstrated that round spermatids can be distinguished by a combination of GFP expression, cell size (around 11 μm in diameter), and the presence of a single dot stain in the center of the nucleus on Hoechst staining (Fig. 4A). Histological examination of a fully matured Gsg2-GFP transgenic mouse testis by serial cryosections showed that round spermatids having acrosomal caps revealed by PAS staining on a section also expressed strong Gsg2-GFP, which was demonstrated on the adjacent section. Those round spermatids were recognized also with a single dot in the nucleus stained by Hoechst (Fig. 4B). This finding supported the validity of our method for identification of round spermatids by the combination of Gsg2-GFP, Hoechst nuclear staining, and cell size. We then applied Hoechst stain to the cultured testis tissue that expressed Gsg2-GFP. The Gsg2-GFP-positive cells differed in their size (some were larger than 11 μm in diameter) and some were multinucleated, which indicated spermatocytes that had finished nuclear division but not yet cytoplasmic division (Fig. 4C). Among such spermatocytes, spermatids were occasionally observed (Fig. 4D). The frequency of spermatids was up to 30% (17/57) among all GFP-positive cells (Table 4). Finally, we injected the nuclei of these round spermatids into mouse oocytes to induce premature chromosome condensation and counted the number of chromosomes (24). In one case out of 19, we found 20 chromosomes, indicating a haploid genome, in samples cultured from 10.5 dpp mouse testes (Fig. 4E). Taken together, these data supported the assumption that the haploid cells had been produced in our organ culture system.

FIG. 4.

Hoechst staining of Gsg2-GFP-positive cells. Among Gsg2-GFP-positive cells, some had a single nucleus with a prominent heterochromatin dot, a sign of typical round spermatids. Others had multiple nuclei each with multiple heterochromatin dots. A) Testis cells of adult Gsg2-GFP transgenic mouse, 4 wk old, were dissociated and stained with Hoechst. Spermatids (arrow) were recognized as smaller than spermatocytes (arrowhead), around 11 mm in diameter, and showed a prominent single nuclear dot on Hoechst staining (blue). B) Cryosection of adult Gsg2-GFP transgenic mouse testis stained with PAS-hematoxylin showed round spermatids of which nuclei were covered with acrosomal caps stained by PAS. On the adjacent section, Gsg2-GFP expression (green) was confirmed to correspond mostly with those round spermatids. Hoechst stain (gray) demonstrated the nuclear single dot in those round spermatids. White rectangular area is magnified in the left picture. C) 11.5 dpp Gsg2-GFP transgenic mouse testis tissue cultured for 21 days. Some GFP-positive cells were found to be multinucleated (arrow) and some others showed two dots in a single nucleus (arrowhead). D) 6.5 dpp Gsg2-GFP transgenic mouse testis tissues cultured for 29 days expressing GFP were considered to be spermatids as they showed a single nucleus with a single dot on Hoechst staining (arrow). E) Nuclei of presumptive spermatids were injected into mouse oocytes to induce premature chromosome condensation to allowing the counting chromosome number. Bar = 15 μm (A, C, D), 30 μm (B).

TABLE 4.

Summary of Hoechst nuclear staining of haspin-GFP positive cells.^a

DISCUSSION

For decades, it has been a challenge to induce mammalian spermatogenic meiosis in vitro. Although a number of improvements and advances have been reported, the completion of the whole process, from spermatogonia to spermatids, has not succeeded [3–5]. With organ culture method in particular, it has been recognized that spermatogenic meiosis would arrest at the pachytene stage [2, 6]. In the present study, we reevaluated the organ culture method and found that spermatids were produced in that culture method, which was the first such achievement, to our best knowledge.

The basics of the present study depended on the methods developed by the Steinbergers in the 1960s. However, the materials and equipment used were different. We tried to make the culture system as simple as possible. One of our novel methods was that we used agarose gel, instead of stainless steel grids covered with a Millipore filter [27], to place testis tissue fragments on. The culture media used were commercially available, and no modifications were made to them. The exact contents of the media used in this study and their quality and those in the Steinbergers' might have differed. The quality of serum might also have been different. These possible differences, or improvements, may have been a reason for the progression of meiosis up to the production of spermatids. More importantly, however, a monitoring system using Acr- and Gsg2-GFP contributed superbly to our work. Their expression levels were strong enough to allow them to be detected easily under a stereomicroscope without uncovering the lid of the culture plate, which enabled the experiments to continue. Gsg2-GFP was especially useful for detecting the completion of meiosis [28]. On the other hand, histological examination alone seemed insufficient to evaluate the progress of meiosis accurately, and we were not able to find spermatids using histology. Accordingly, it might be possible that the Steinbergers' experiment also produced a few spermatids, which were not possible to identify at the time.

In the present study, we have confirmed several results of previous studies by the Steinbergers. They showed that the progress of spermatogenesis in organ culture samples relates to the developmental stage of those samples. As age increases, the spermatogenesis of their explants is more likely to proceed in culture [7]. However, on the other hand, even using more developed testes, such as of 14-day-old rats, spermatogenesis did not proceed beyond the pachytene stage of meiosis [7]. Our present study used mouse pup testes from 4.5 to 14.5 dpp and showed high rates of Acr- and Gsg2-GFP expression regardless of the age of the pups. Nonetheless, the rate decreased when younger pups were used, and no differentiation beyond round spermatids, namely spermiogenesis, was observed even when older pups were used. It was reported previously that spermatids had disappeared within several days when adult testis tissues were cultured, while spermatocytes remained for 3–4 wk [6]. Therefore, it seems extremely difficult to induce spermiogenesis in organ culture systems.

The Steinbergers used Eagle synthetic medium [10] supplemented with sodium pyruvate and several amino acids as a basic medium. They found that the addition of vitamin A, E, and C or glutamine fostered spermatogenetic activity [6]. Glutamine was especially useful for producing more spermatocytes in their organ culture system [6]. αMEM, which was the best medium in the present study, includes 2 mM glutamine and 0.28 mM vitamin C. DMEM also contains 4 mM glutamine. As all media used in this study included 10% FBS, it might not be worthwhile to compare the detailed contents of the media and the supplements added to them with those of the Steinbergers' study. In future studies, we would like to omit FBS and replace it with chemically defined supplements in order to enhance our understanding of the environmental factors influencing the spermatogenesis.

It is known that spermatogenesis does not proceed well at 37°C [29]. The Steinbergers tested different temperatures ranging from 31°C to 37°C in their organ culture experiment and found that 33°C or even lower was better for rat spermatogenesis [6, 8]. In the present experiment, the difference between 37°C and the other two temperatures was drastic, which became obvious due to very poor Gsg2-GFP expression at 37°C. The difference between 32°C and 34°C was rather subtle, but 34°C appeared to promote spermatogenesis more. The mechanism that induces such temperature sensitivity is not fully understood, but several responsible genes have been postulated [30, 31]. Izu et al. [31], for instance, reported that heat shock transcription factor 1 (HSF1) was activated in male germ cells by heat stress and that HSF1 induced apoptosis of spermatocytes at the pachytene stage. They reasoned that HSF1 contributed to the quality control of spermatogenesis. As high testicular temperature is caused by cryptorchidism and varicocele, which are directly involved in male infertility, the mechanical relationship between high temperature and spermatogenic impairment needs to be explored in more detail. Our organ culture method could be useful for such molecular-level studies because it supported spermatogenesis beyond the pachytene stage.

The liquid-gas interphase method allows each tissue fragment easy access to oxygen [32]. Accordingly, Steinberger et al., seeking the beneficial effects of higher O₂ concentrations, tested 10% CO₂ + 90% O₂ in their organ culture experiments and found that, although the penetration of O₂ into tissues was improved as shown by the smaller area of central necrosis in each tissue, the seminiferous tubules were more distorted and the numbers of spermatogonia and spermatocytes had decreased dramatically [8]. As too much oxygen can damage tissues and cells via the production of reactive oxygen species, the optimal oxygen concentration might be lower than 20%. In fact, in the testis, blood vessels are located exclusively between the tubules, and oxygen reaches the lumen of the tubules only by diffusion. It was reported therefore that luminal oxygen partial pressures were extremely low due to the diffusion distance and the high oxygen consumption by the proliferating germ cells [33–36]. Based on such considerations, we used a lower oxygen concentration of 10% on our organ culture experiments and found that it was disadvantageous for spermatogenesis compared to 20%. Further investigation is necessary to find a more suitable oxygen concentration.

The semiquantitative evaluation of spermatogenic progression in our organ culture method demonstrated that disruption of RA action prevented spermatogenesis as it does in vivo. The effect of RA on spermatogenesis has been well known for decades, mainly due to the fact that a vitamin A-deprived diet prevents spermatogenesis in rodents, leaving only primitive spermatogonia in the seminiferous tubules [37]. Recent studies have revealed the mechanism of this phenomenon and found that RA is a pivotal trigger of meiosis [38]. The present method combined with the use of specific inhibitors to particular molecules in our organ culture system might be useful in future studies for elucidating the roles of such molecules in spermatogenesis.

Recently, several independent groups reported that embryonic stem cells can be induced to differentiate into germ cells in vitro, which then differentiate into spermatogenic cells [39–41]. However, the spermatogenesis seen in these studies did not rely on testicular somatic cells of Sertoli or Leydig cells. Therefore, the role of somatic cells in spermatogenesis was not addressed properly. Authentic spermatogenesis depends on Sertoli cells and other somatic cells outside the seminiferous tubules. Sertoli cells above all are indispensible in supporting various aspects of spermatogenesis. Studies of spermatogenesis from ES cells are certainly important and rewarding, but at the same time it is desirable to have an authentic in vitro spermatogenesis system, which is able to dissect the mutual interplay between germ cells and somatic cells.

In conclusion, we have shown that the induction of spermatogenesis up to haploid spermatids in an organ culture is possible. But it is still a rare event, and further improvements of the culture technique are needed in the future.