Understanding the proliferation and differentiation processes of teleost female germline stem cells (GSCs) is of great importance in developmental biology, transgenic fish production and species conservation (Okutsu et al., 2006; Wong et al., 2013; Iwasaki-Takahashi et al., 2020). So far, numerous studies have contributed to understanding these processes (Okutsu et al., 2006; Spradling et al., 2011; Lacerda et al., 2014). However, in spite of these efforts, there remains a need for innovative tools to study the cellular and molecular mechanisms for female GSC development more accurately and effectively. Generally, the controlled environment of in vitro model system enables researchers to exactly evaluate cellular characteristics at cellular and molecular levels and is able to provide a level of precision and experimental control that are sometimes unattainable even in vivo (Xu et al., 2021). However, traditional two-dimensional (2D) culture methods may have a considerable limitation to study female GSC development. Precious communications between different types of gonadal cells by specific tissue structures and cellular composition are critical for the maintenance, proliferation and differentiation of GSCs (Reuter et al., 2014; Alves-Lopes et al., 2017), but reproducing such a system under 2D culture is very hard, which restricts its application to the study of female GSC development. In contrast, ovarian organoids can be a good candidate to overcome this problem since they are able to retain ovarian structures and functions in vitro. So far, several studies have tried to culture ovarian organoids in mammals. In mice, it has been reported that ovarian organoids were made by being induced in ultra-low attachment 96-well plates (ULA) and subsequently cultured on an air-liquid interface (Luo et al., 2021). Using these ovarian organoids, trans-differentiation of mouse spermatogonia to fertilizable oocytes was demonstrated (Luo et al., 2021). In humans, it has been reported that ovarian cancer organoids were cultured by modified Matrigel bilayer culture method for the screening of drug sensitivity (Maru et al., 2019). These indicate that ovarian organoids can be cultured at the current technology level, which imply that this technology also can be applied to teleost so as to develop in vitro model for the study of female GSC biology. However, there has been no report regarding ovarian organoid culture in teleost to our best knowledge raising the necessity of establishing stable ovarian organoid culture system.

As a first step toward establishing such a system, here, we attempted to establish ovarian cell aggregate culture in model fish, marine medaka (Oryzias dancena) and estimated the feasibility of developing ovarian organoid culture system. In order to do that, we first investigated the optimal cell populations that are effective in forming ovarian cell aggregates through the separation of the dispersed ovarian cells by Percoll density gradient centrifugation followed by the suspension culture of different cell populations. Then, the effects of several media supplements on the culture of ovarian cell aggregates were evaluated. Finally, we examined if these ovarian cell aggregates were able to retain the functional aspects of ovaries regarding germ cell maintenance and 17β-estradiol synthesis after culture.

Fish

Marine medaka (O. dancena) were bred in the Laboratory of Cell Biotechnology, Pukyong National University (Busan, Republic of Korea). They were fed four times a day with artificial diet for olive flounder larvae (EWHA, Busan, Republic of Korea) and one time at last with brine shrimp larvae (Artemia nauplius). Water salinity and temperature were adjusted to 5 psu and 26℃, respectively. They were kept under 14 h light and 10 h dark condition during experiments. The females of 3.2 to 3.5 cm in total length were separated from males a day before the experiments. All procedures dealing with animals complied with the guidelines provided by Pukyong National University, and the Institutional Animal Care and Use Committee (IACUC) of Pukyong National University approved our research proposal (approval No. PKNUIACUC-2021-31).

Ovarian cell isolation and separation

To isolate O. dancena ovarian cells, adult females were anesthetized with 4℃ waters in 5 min, and exposed to 70% ethanol (Duksan, Asan, South Korea) for 10 s to prevent microbial contamination (Graziano et al., 2013). Then, the ovaries were surgically extracted and washed three times with Dulbecco’s phosphate-buffered saline (DPBS; Gibco, Grand Island, NY, USA) containing 1% P/S (Gibco). The ovaries were placed onto 35 mm Petri dishes (SPL Life Science, Pocheon, Republic of Korea) and were dissected by a surgical blade (No. 20; FEATHER Safety Razor, Osaka, Japan). After that, they were digested by 2 mL of 0.05% (v/v) trypsin-EDTA (Gibco) for one hour with pipetting every 10 min and fully-digested cell suspension was filtered through 40 µm cell strainer (Falcon, Durham, NC, USA). Then, 2 mL DMEM containing 10% fetal bovine serum (FBS; Gibco) and 1% P/S was added to cell suspension for trypsin inactivation and after centrifugation (400 g, 5 min), cell pellet was resuspended with 300 µL DMEM containing 10% FBS and 1% P/S. For ovarian cell separation, Percoll density gradient centrifugation was used. The cell suspension was loaded onto the top of a discontinuous 5-step Percoll (Sigma-Aldrich, St. Louis, MO, USA) solution consisting of 1 mL each of 20%, 30%, 40%, 50%, and 60% in 15 mL conical tubes (Falcon) and centrifuged at 800 g for 30 min. In this study, two types of cell populations were used for the culture of ovarian cell aggregates as follows; (1) the cells from 20-40% layers and (2) the cells from top-50% layers (Fig. 1A).

Figure 1.Evaluation for the formation of ovarian cell aggregates according to cultured ovarian cell populations. (A) Schematic representation of experimental procedures. Two different cell populations from 20 to 40% layers and top to 50% layers after Percoll density gradient centrifugation were compared for ovarian cell aggregate formation after culture for 7 days. (B) Pictures showing two different forms of ovarian cell aggregates. Aggregates were formed in two different ways; large single aggregate and small multiple aggregates. Scale bar = 100 μm. (C) Frequencies of replicates in which different aggregate forms appeared. Ten independent experiments were performed.

Culture of ovarian cell aggregates

For the culture of O. dancena ovarian cell aggregates, modified ESM2 (mESM2) were used as culture media (Choi et al., 2023). The formula of mESM2 was as follows; Dulbecco’s Modified Eagle Medium (DMEM; Gibco) supplemented with 15% FBS, 20 mM hydroxyethyl piperazine ethane sulfonic acid (HEPES; Thermo Fisher Scientific, Waltham, MA, USA), 1 mM non-essential amino acid (Gibco), 1 mM sodium pyruvate (Gibco), 2 nM sodium selenite (Sigma-Aldrich), 100 µM beta-mercaptoethanol (Gibco), 10 ng/mL human basic fibroblast growth factor (bFGF; Gibco), 1% fish serum (FS) from rainbow trout (Caisson Laboratories, Inc., North Logan, UT, USA), 50 µg/mL O. dancena embryo extract (EE), and 1% P/S. EE was added to culture media due to its growth promoting activity (Collodi and Barnes, 1990) and its extraction was carried out according to the method of our previous study (Ryu and Gong, 2017). To induce the formation of ovarian cell aggregates, 1.25 × 10⁶ cells from each of two different cell populations were suspended with 150 µL culture media and seeded into ultra-low attachment 96 well plate (ULA) (Fig. 1A). The culture of ovarian cells was performed for 7 days unless otherwise stated at 28℃ in an incubator with air atmosphere and half of media was changed every three days. In case of culturing more than 7 days, entire (150 µL) culture media was changed every three days. For testing the effects of growth factors on nanos2 and scp3 expression, 10 ng/mL bFGF and/or 10 ng/mL glial cell-derived neurotrophic factor (GDNF; PeproTech, Hamburg, Germany) were added to culture media without FS, bFGF, and EE during culture periods. For evaluating the optimal concentration of human follicle stimulating hormone (hFSH; Thermo Fisher Scientific) on cyp19a1 expression, 50 ng/mL, 100 ng/mL, and 150 ng/mL hFSH dissolved in distilled water were added to culture media and cultured for 7 days with the replacement of half of media every three days. Concentration of hFSH was determined based on a previous study that used FSH to induce E₂ synthesis in Russian sturgeon (Yom-Din et al., 2016). At day 7, ovarian cell aggregates were subjected to histology and cyp19a1 expression analyses.

Hematoxylin and eosin (H&E) staining

For H&E staining, the cultured ovarian cell aggregates were moved in 35 mm petri-dishes containing 2 mL DPBS for washing. Then, the aggregates were fixed in 15 mL conical tubes using 1 mL Bouin’s solution (CliniSciences, Montrouge, France) at 4℃ for overnight. Bouin’s solution were gently removed from the fixed aggregates, and they were dehydrated in discontinuous concentration of ethanol as follows; 70% for 1 h, 80% for 1 h, 90% for 1 h, 95% for 1 h, 99% for 1 h and 99% for 1 h again. Subsequently, samples were embedded in Paraplast® (Leica Biosystems, St. Louis, USA) and sectioned by 5 μm. Aggregates sections were immersed in Xylene (Duksan) for 2 min two times, and rehydrated by reverse sequences of dehydrate protocol for 30 s each. Finally, rehydrated sections were stained by hematoxylin (Leica Biosystems) for 2 min, and eosin (Biognost, Zagreb, Croatia) for 1 min.

Reverse transcription polymerase chain reaction (RT-PCR) and quantitative RT-PCR (qRT-PCR)

For RT-PCR analysis, total RNA was extracted from ovarian cell aggregates by RNeasy Micro Kit (Qiagen, Hilden, Germany), and the RNA was treated with DNase I (Sigma-Aldrich) according to manufacturer’s instructions. Subsequently, cDNA was synthesized by GoScript^TM reverse transcription system (Promega, Madison, WI, USA) according to manufacturer’s instructions. RT-PCR condition was as follows: initial denaturation (94℃ for 3 min), 35 cycles of amplification (denaturation: 94℃ for 30 s; annealing: 60℃ for 30 s; and elongation 72℃ for 30 s). Then, the PCR products were visualized by electrophoresis on 1.2% agarose gel (Lonza, Rockland, ME, USA). For qRT-PCR, cDNA was prepared in the same manner as mentioned above. qRT-PCR condition was as follows; initial denaturation (94℃ for 3 min), 40 cycles of amplification (denaturation: 94℃ for 30 s; annealing: 60℃ for 30 s; and elongation 72℃ for 30 s). The relative mRNA expressions of the genes tested were calculated by 2^-ΔΔCt method, where Ct = threshold cycle for target amplification, ΔCt = Ct_{target gene} – Ct_{internal reference} (18s rRNA), and ΔΔCt = ΔCt_sample - ΔCt_calibrator (Livak and Schmittgen, 2001). Primer sequences used in this study were presented in Table 1.

Table 1. Primer sequences used in this study for RT-PCR and qRT-PCR.

Genes	Primer sequences (5’ > 3’)	Product size (bp)	Accession number
nanos2	Forward: AAACTACACCTGTCCCATCTG	111	XM_036217407.1
	Reverse: AACTTGTAGGAGGGCAGCATC
scp3	Forward: CAGCTGCTAGCTTTGAGGAA	224	XM_024295185.2
	Reverse: CTGAGAGAACTGCTGCATTG
fshr	Forward: GATCTTCTCCTCACTCGCCG	462	XM_024285506.2
	Reverse: TTAAACAAGCCAAAGCGGGC
lhr	Forward: TCATCCTCAATGTTGCCGCT	407	XM_024297985.2
	Reverse: CTGGTTGGTCACTTTGTGCG
foxl2	Forward: CTGATCTGGTTTGCGCGATG	963	XM_024258143.2
	Reverse: TTTACGCAGACGGAAAAGCTTAAATA
cyp19a1	Forward: ACCTCGCGTTTTGGCAGCAAACA	90	XM_02496015.2
	Reverse: TTTCCACAGCGCCACGTTGTTGT
18s rRNA	Forward: TCCAGCTCCAATAGCGATTCACC	253	HM347347.1
	Reverse: AGAACCGGAGTCCTATTCCA

Measurement of viability and size of ovarian cell aggregates

To measure the viability of the cells comprising ovarian cell aggregates, the aggregates cultured for 7 days were moved to 1.5 mL tube (Corning, NY, USA) and dissociated by 0.05% trypsin-EDTA for 30 min with pipetting every 10 min. Subsequently, trypsin-EDTA was inactivated by treating one volume of DMEM containing 10% (v/v) FBS and 1% (v/v) P/S, and centrifuged at 400 g for 5 min. Cell pellets were resuspended by 500 μL DPBS, and their viabilities were investigated by trypan blue (Gibco) assay. After staining with 0.4% (w/v) trypan blue, stained and non-stained cells were regarded as dead and live cells, respectively, and the viability was presented as the percentage of live cells. Localization of live and dead cells in ovarian cell aggregates was identified using Live/Dead Viability/Cytotoxicity kit (Molecular Probes, Eugene, OR, USA) according to manufacturer’s instructions. After staining with fluorescent dyes indicating live and dead cells (2 μM calcein AM for live cells and 4 μM ethidium homodimer-1 for dead cells), ovarian cell aggregates were observed under a fluorescent microscope (Olympus, Hamburg, Germany). For negative control, ovarian cell aggregates were immersed in 70% ethanol to induce the death of all the cells comprising aggregates. For size measurement, ovarian cell aggregates cultured for 7 days were collected from ULA and washed with 2 mL DPBS in 35 mm Petri dishes. After transferring them into 96-well plates (Thermo Fisher Scientific), pictures of 20 aggregates were taken for size measurement. The size was measured using TSView7 software, Version 7 (Fuzhou Xintu Photonics CO., Ltd., Fuzhou, Fujian, China) and defined as the mean of width and height of aggregate on the picture.

Measurement of 17β-estradiol (E₂) concentration

For measurement of E₂ concentration, 150 μL culture media were collected at day 7, 10, 13, 16, 19 and 22 during culturing ovarian cell aggregates and stored in a deep freezer adjusted to -70℃ until E₂ concentration was measured. E₂ concentration was measured by Estradiol ELISA Kit (Cayman Chemicals Company, MI, USA) following manufacturer’s instructions. Fresh mESM2 was used as the blank control.

Statistical analysis

SPSS program (SPSS Inc., Chicago, IL, USA) was used to analyze numerical data. The data were analyzed by one-way analysis of variance (ANOVA) or t-test. When ANOVA detected a significant main effect, treatments were analyzed subsequently by Duncan’s method. Significant differences were determined when p values were less than 0.05.

Investigation of the optimal cell populations for ovarian cell aggregate formation

To find the optimal cell populations to be able to form the desirable ovarian cell aggregates, whole ovarian cells derived from enzymatic digestion of O. dancena ovaries were separated by Percoll density gradient centrifugation using a discontinuous 5-step Percoll solution. After that, two different cell populations from 20 to 40% layers and top to 50% layers were harvested and cultured separately in ULA. Then, ovarian cell aggregate formation and its morphology were observed after culture for 7 days (Fig. 1A). As the results, it was observed that aggregates were formed in two different ways. One way was that the cells formed a large single aggregate and the other way was that the cells formed small multiple aggregates (Fig. 1B). As shown in Fig. 1C, when the cells derived from 20 to 40% layers were cultured, 90% of attempts (9 out of 10) formed small multiple aggregates and the other 10% formed large single aggregate. In contrast to this, 100% of attempts (9 out of 9) formed large single aggregate when the cells from top to 50% layers were cultured. Next, the differences between ovarian cell aggregates derived from two different cell populations were analyzed by histology and gene expression. In histological observation, the presence of pre-vitellogenic oocytes was confirmed in ovarian cell aggregates derived from top to 50% layers, but not in those from 20 to 40% layers (Fig. 2A). RT-PCR results showed that ovarian cell aggregates regardless of its original cell population expressed not only nanos2 and scp3 specific for germline cells but also fshr and foxl2 specific for ovarian somatic cells. However, lhr gene was expressed only in the ovarian cell aggregates derived from top to 50% layers and not in those from 20-40% layers (Fig. 2B). Based on these results, ovarian cell aggregates derived from top to 50% layers were used for further experiments.

Figure 2.Characterization of ovarian cell aggregates according to cultured ovarian cell populations. Ovarian cell aggregates derived from two ovarian cell populations from top to 50% and 20 to 40% layers after Percoll density gradient centrifugation were subjected to histological and gene expression analyses. (A) Histological observation. Pre-vitellogenic oocytes were observed in ovarian cell aggregates derived from top to 50% layers but not in those from 20 to 40% layers. Intact ovarian tissues were used as a positive control. Scale bar = 100 μm. (B) RT-PCR analysis for several genes expressed in ovary. Expression of genes specific for germline cells (nanos2 and scp3) and somatic cells (fshr, lhr, and foxl2) was evaluated by RT-PCR analysis. Except for lhr gene, all genes were expressed in ovarian cell aggregates regardless of its original cell populations. Expression of lhr gene was detected only in ovarian cell aggregates derived from top to 50% layers.

Effects of media supplements on the culture of ovarian cell aggregates

As part of a study to find the better media, the effects of three major components (FS, bFGF, and EE) in mESM2 on the culture of ovarian cell aggregates were evaluated. In order to do that, the ovarian cells from top to 50% layers after Percoll density gradient centrifugation were cultured in mESM2 and mESM2 without FS, bFGF, and EE for 7 days and the characteristics between ovarian cell aggregates derived from two media groups were compared with each other. As the results, the formation of large single aggregates was observed in all replicates from both groups and no significant difference was observed in morphology between two groups (Fig. 3A). In live and dead staining, most cells comprising ovarian cell aggregates were live as indicated with green fluorescence (live cells) without red fluorescence (dead cells) regardless of media groups (Fig. 3B). Furthermore, quantitative assessment of viability using trypan blue assay showed high viabilities in both groups without significant difference (Fig. 3C; 92.58 ± 2.14% in mESM2 and 92.61 ± 1.82% in mESM2 without FS, bFGF, and EE). For aggregate size, significant difference was not detected between both groups (Fig. 3D; 876.46 ± 88.69 μm in mESM2 and 852.12 ± 194.23 μm in mESM2 without FS, bFGF, and EE). In the evaluation of relative expression level of two genes, nanos2 (specific for oogonia) and scp3 (specific for meiotic germ cells), significant decrease of scp3 expression was observed when FS, bFGF, and EE were removed from mESM2 whereas no difference was observed in nanos2 expression between both groups (Fig. 3E).

Figure 3.Evaluation of the effects of media supplements on the culture of ovarian cell aggregates. Ovarian cell aggregate culture was performed in mESM2 and mESM2 without FS, bFGF, and EE and after ovarian cell aggregate formation, its viability, aggregate size, and relative mRNA level for nanos2 and scp3 were measured. (A) Representative images of ovarian cell aggregates cultured in each media for 7 days. No significant difference was observed in morphology between two groups. Scale bar = 250 μm. (B, C) Viability of ovarian cell aggregates. Live/dead staining (B) and trypan blue assay (C) were performed to visualize and compare the viability of the cells comprising aggregates cultured in each media. No significant difference was observed between two groups. Scale bar = 200 μm. (D) Size of ovarian cell aggregates. No significant difference was observed between two groups. (E) Relative mRNA expression of nanos2 and scp3 in ovarian cell aggregates. Expression of nanos2 did not show significant difference between two groups, but scp3 decreased in the group cultured without FS, bFGF, and EE. All values are expressed as mean ± standard deviation of three independent experiments. Asterisk (*) indicates significant difference (p < 0.05).

Effects of bFGF and GDNF on nanos2 and scp3 expression

Expression result of scp3 above indicated that three factors (FS, bFGF, and EE), whether it is a single action or a collaboration, played a key role in germ cell differentiation under this aggregate culture condition. Of three factors, we focused on bFGF based on its well-known roles regarding germ cell development; induction of spermatogonial differentiation (Masaki et al., 2018) and upregulation of scp3 expression (Aflatoonian et al., 2009). Thus, the effects of bFGF and GDNF on the expression level of nanos2 and scp3 genes upon ovarian cell aggregate culture was investigated. In case of GDNF, since it has been known to play the different role with bFGF in germ cell differentiation (Masaki et al., 2018), its effects were evaluated together to better understand the role of bFGF. For this, bFGF and GDNF were added alone or together to culture media without FS, bFGF, and EE and mRNA expression levels of both genes were measured by qRT-PCR. As shown in Fig. 4, it was observed that the expression of both genes significantly increased after aggregate culture for 7 days compared to before culture (1.97 ± 0.24 fold increase in nanos2 and 1.45 ± 0.2 fold increase in scp3) even in the condition without both bFGF and GDNF. However, the sole treatment of bFGF resulted in a higher increase of expression of both genes than non-treatment (5.19 ± 1.11 fold increase in nanos2 and 5.29 ± 0.76 fold increase in scp3 compared to before culture). In case of sole treatment of GDNF, a similar level of nanos2 expression with bFGF sole treatment group was observed but scp3 expression was significantly decreased compared to bFGF sole treatment group. When two factors were treated together, nanos2 expression still retained a similar level with those of sole treatment groups of bFGF and GDNF, while scp3 expression was observed to be higher than that of GDNF sole treatment and to be lower than that of bFGF sole treatment. Collectively, these results indicated that bFGF was a critical factor for scp3 expression in ovarian cell aggregate culture.

Figure 4.Investigation of the effects of bFGF and GDNF on the expression of nanos2 and scp3 in ovarian cell aggregate culture. Two factors were added alone or together to culture media (mESM2) without FF, bFGF, and EE and after culture for 7 days, mRNA expression levels of both genes in ovarian cell aggregates formed were measured by qRT-PCR. Both bFGF and GDNF increased nanos2 mRNA expression. In case of scp3, the highest expression was observed when bFGF was added alone. All values are expressed as mean ± standard deviation of three independent experiments. ^abcdeDifferent letters indicates significant differences (p < 0.05).

Evaluation of germ cell maintenance and E₂ synthesis of ovarian cell aggregates

To know if the ovarian cell aggregates could retain ovarian functions after culture, the effects of hFSH on germ cell maintenance and E₂ synthesis of ovarian cell aggregates were investigated. For this, ovarian cell aggregate culture was performed in culture media (mESM2) with or without hFSH and the aggregates formed were analyzed for histology, cyp19a1 expression, and E₂ synthesis. As the results, it was shown that the aggregates formed after culture for 7 days included a small number of pre-vitellogenic oocytes regardless of hFSH treatment (50 ng/mL) implying that the ovarian cell aggregates cultured in our system basically retained germ cell populations without any specific treatment (Fig. 5A). Treatment of hFSH by concentration during ovarian cell aggregate culture showed that hFSH did not influence significantly to cyp19al expression (Fig. 5B, p = 0.2794). However, considering that 50 ng/mL hFSH treatment group showed 1.86 ± 0.42 fold increase of cyp19al expression compared to non-treatment group, the effect of 50 ng/mL hFSH on E₂ synthesis was investigated. When E₂ concentration was measured in the culture media with or without 50 ng/mL hFSH, in which the ovarian cell aggregates were cultured, it was observed that lower E₂ concentration levels compared to day 7 of culture were retained from day 10 to day 22 of culture and hFSH treatment did not influence to E₂ concentration (Fig. 5C, p = 0.1993, 0.0937, 0.0620, 0.0999, 0.6828, and 0.6774 for day 7, 10, 13, 16, 19 and 22, respectively). Although no responsiveness to hFSH has been observed, these results indicated that ovarian cell aggregates retained minimal potential regarding E₂ synthesis during the culture period.

Figure 5.Evaluation of the effect of hFSH on germ cell maintenance and E₂ synthesis of ovarian cell aggregates. Ovarian cell aggregate culture was performed in mESM2 culture media with or without 50 ng/mL and the aggregates formed were analyzed for histology and E₂ synthesis. (A) Histological observation. A small number of pre-vitellogenic oocytes (arrow heads) were observed regardless of hFSH treatment. Scale bar = 200 μm. (B) Relative mRNA expression of cyp19a1 in ovarian cell aggregates by hFSH concentration. No significant difference was observed among treatment groups. (C) Concentration of E₂ secreted during ovarian cell aggregate culture depending on hFSH treatment. In this experiment, ovarian cell aggregate culture was performed until day 22 and E₂ concentrations were measured every 3 days from day 7. No significant difference was detected in E₂ concentration by hFSH treatment and from day 10 to day 22, the E₂ concentration remained lower compared to day 7. All values are expressed as mean ± standard deviation of three independent experiments.

Development of in vitro model to study GSCs has been a very important task in the field of research related to reproduction (Ge et al., 2015). In fish, previous studies have shown in vitro culture of GSCs from several fish species such as zebrafish, medaka, rainbow trout, dogfish, and so on (Xie et al., 2020). However, most were studies of male GSCs and only some of them were for female GSCs. Representatively, Wong et al. (2013) cultured zebrafish female GSCs for more than 6 weeks and reported the production of zebrafish offspring from cultured female GSCs through germ cell transplantation technology. Additionally, it has been reported that medaka female GSCs cultured on polymer-coating dishes during short period (10 days) were able to colonize to recipient gonad after cell transplantation (Jeong et al., 2018). This means that there is a lack of proper in vitro model for the study of female GSCs in current status raising the need for its development. So far, all fish GSC cultures using a 2D culture system, including the above two studies, basically requires germ cell transplantation technology to induce the differentiation of GSCs to the functional gametes (Wong et al., 2013; Iwasaki-Takahashi et al., 2020). These indicate that a 2D culture system cannot reproduce the complex mechanisms equivalent to in vivo involving communication with other cells, which are required for the differentiation of cultured GSCs. Therefore, it might be more reasonable to focus on 3D culture rather than 2D culture in the development of culture techniques for female GSCs. Indeed, in the studies regarding the culture of fish male GSCs, the production of functional sperm through in vitro culture without germ cell transplantation techniques has been accomplished only through three-dimensional (3D) aggregate culture (Higaki et al., 2017; Zhang et al., 2022; Choi et al., 2023). The 3D culture system can mimic the in vivo environment, and ovarian organoids, which can be ultimately developed through 3D culture system, will be an excellent model for the study of female GSCs.

In this study, we found that two different forms of aggregates (large single aggregate and small multiple aggregates) were formed depending on cell population used when our culture conditions for inducing ovarian cell aggregates were applied. When the cells from top to 50% layers were used, all replicates (100%) formed large single aggregates rather than small multiple aggregates. On the other hand, most (90%) of replicates formed small multiple aggregates when the cells from 20 to 40% layers were cultured. Considering that large single aggregate is much closer in morphological aspect to the mouse ovarian organoids presented in previous study (Li et al., 2021), it seems more appropriate to use the cells from top to 50% layers for the formation of O. dancena ovarian cell aggregates. Furthermore, the presence of pre-vitellogenic oocytes was observed only in the ovarian cell aggregates derived from top to 50% layers implying that they were more functional than those from 20 to 40% layers. On the basis of our previous study that enriched female GSCs in medaka (Ryu and Gong, 2020), it can be speculated that the proportion of ovarian somatic cells among all cells was much higher in ovarian cell aggregates derived from top to 50% layers than in those from 20-40% layers. This can also be supported by our result for the expression of lhr specific for gonadal somatic cells. Its expression was detected only in ovarian cell aggregates derived from top to 50% layers. Given the critical role of lhr expression contributing to oocyte maturation (Ogiwara et al., 2013; Kitano et al., 2022), more communication between germline cells and somatic cells within ovarian cell aggregates derived from top to 50% layers might give a little more functionality relative to those from 20 to 40% layers. The detailed mechanisms will have to be verified in further study. Collectively, we concluded that the cells from top to 50% layers after Percoll density gradient centrifugation of ovarian cells was more suitable for the formation of O. dancena ovarian cell aggregates than the cells from 20 to 40% layers.

bFGF and GDNF are well known factors to promote the proliferation of mammalian GSCs including mouse (Mus musculus; Kanatsu-Shinohara et al., 2003), hamster (Mesocricetus auratus; Kanatsu-Shinohara et al., 2008), rat (Rattus norvegicus; Wu et al., 2009), bovine (Bos tarurs; Suyatno et al., 2018), and human (Homo sapiens; Sadri-Ardekani et al., 2009). Moreover, many studies have used bFGF and/or GDNF for culturing spermatogonia from teleost fishes including swamp eel (Monopterus albus; Sun et al., 2022), rohu (Labeo rohita; Panda et al., 2011), zebrafish (Danio rerio; Kawasaki et al., 2012), and dogfish (Scyliorhinus canicular L.; Gautier et al., 2014). In this study, our results demonstrated that sole treatment of bFGF upregulated scp3 gene expression from the cultured ovarian cell aggregates. However, in addition to bFGF, additional treatment of GDNF inhibited the effect of bFGF by decreasing scp3 gene expression. Unlike GDNF that is known to promote GSC proliferation with the suppression of differentiation (Meng et al., 2000; Masaki et al., 2018), it has been known that bFGF modified the functions of mouse germline niche to be more appropriate for spermatogonial differentiation (Masaki et al., 2018) and Aflatoonian et al. (2009) reported that bFGF increased scp3 expression in human embryonic stem cells. Comprehensively, these imply that bFGF exhibited a similar action in teleost, marine medaka, as in mouse and human. Therefore, bFGF would be necessary in O. dancena ovarian cell aggregate culture in the aspect of differentiation.

In the absence of 3D culture technology of fish female GSCs, in this study, we attempted to establish O. dancena ovarian cell aggregate culture as a cornerstone of the development of 3D culture technology. The results showed that the ovarian cell aggregates could be formed well under specific culture conditions using ULA, mESM2, and specific cell populations separated by Percoll density gradient centrifugation. In addition, it has been shown that they were able to maintain minimal ovarian function such as germ cell maintenance and E₂ synthesis for a short period. Future studies should be performed to optimize culture conditions so that the ovarian cell aggregates can maintain the improved ovarian functions for a longer period of time. The results from this study will contribute to the development of stable 3D culture technology of fish female GSCs followed by the establishment of fish ovarian organoids which can be used as a model to study female GSCs as well as a tool for various biotechnology applications.

None.

Conceptualization, SPG; data curation, SPG; formal analysis, SPG; investigation, JHC; methodology, JHC; project administration, JHC; resources, SPG; supervision, SPG; wrighting – original draft, JHC; writing – review & editing, SPG.

This work was supported by the National Research Foundation of Korea (NRF) grant funded from the Korea government (MSIT) (No. 2019R1F1A1058145) and by the National Institute of Fisheries Science, Ministry of Oceans and Fisheries, Korea (R2024019).

Not applicable.

Not applicable.

Not applicable.

Not applicable.

No potential conflict of interest relevant to this article was reported.