foxl3, a sexual switch in germ cells, initiates two independent molecular pathways for commitment to oogenesis in medaka

Significance

In teleost fish, unlike mammals, retinoic acid (RA) does not exert a conspicuous feminization effect of germ cells. Instead, in the teleost fish medaka, foxl3 serves as a germ cell-autonomous factor acting at the very beginning of the sexual fate decision. This study shows that foxl3 initiates oogenesis through two genetically independent pathways, meiosis and follicular development. One pathway, involving rec8a, drives female-specific regulation of meiosis. The other, involving fbxo47, both directs follicular development and suppresses spermatogenesis commitment. The involvement of two evolutionarily conserved factors in oogenesis suggests the existence of an RA-independent pathway that promotes oogenesis, common in vertebrates. In addition, our findings reveal the timing of the establishment of germ cell feminization.

Abstract

Germ cells have the ability to differentiate into eggs and sperm and must determine their sexual fate. In vertebrates, the mechanism of commitment to oogenesis following the sexual fate decision in germ cells remains unknown. Forkhead-box protein L3 (foxl3) is a switch gene involved in the germline sexual fate decision in the teleost fish medaka (Oryzias latipes). Here, we show that foxl3 organizes two independent pathways of oogenesis regulated by REC8 meiotic recombination protein a (rec8a), a cohesin component, and F-box protein (FBP) 47 (fbxo47), a subunit of E3 ubiquitin ligase. In mutants of either gene, germ cells failed to undergo oogenesis but developed normally into sperm in testes. Disruption of rec8a resulted in arrest at a meiotic pachytenelike stage specifically in females, revealing a sexual difference in meiotic progression. Analyses of fbxo47 mutants showed that this gene regulates transcription factors that facilitate folliculogenesis: LIM homeobox 8 (lhx8b), factor in the germline α (figla), and newborn ovary homeobox (nobox). Interestingly, we found that the fbxo47 pathway ensures that germ cells do not deviate from an oogenic pathway until they reach diplotene stage. The mutant phenotypes together with the timing of their expression imply that germline feminization is established during early meiotic prophase I.

Germ cells, which have the potential to differentiate into eggs and sperm, must determine their sexual fate. In the teleost fish medaka (Oryzias latipes), the onset of the germline sexual fate decision is triggered by foxl3 (forkhead box protein L3) (1). During medaka gonadal development, foxl3 expression is maintained only in female germ cells, and the loss-of-function mutant of foxl3 results in production of functional sperm in the ovary, indicating that foxl3 is required for suppression of spermatogenesis in female germ cells (1). However, the specific role of foxl3 in germline feminization (process of commitment to oogenesis) has not been clarified.

To elucidate the molecular pathways organized by foxl3, we previously identified rec8a (REC8 meiotic recombination protein a) and fbxo47 (F-box protein 47) as direct targets of FOXL3 (2). rec8a and fbxo47 transcripts are detected in mitotic germ cells expressing foxl3 in wild-type ovaries but not in foxl3^−/− ovaries and are indicated to act downstream of foxl3 during germline sex determination (2). rec8a is one of two Rec8 orthologs resulting from the teleost-specific whole-genome duplication (3). REC8 is a meiosis-specific kleisin subunit of the cohesin complex, which plays crucial roles in various meiosis-associated chromosomal events, including chromosomal axis formation and homolog association (4). Rec8-null mutant mice of both sexes are sterile and fail to complete meiotic prophase I (5, 6). On the other hand, the F-box domain-containing protein FBXO47 has been implicated to act as a subunit of Skp1-cullin1-FBP (SCF)-type E3 ubiquitin ligases (7). Notably, germ cells in the fbxo47 mutant exhibit defects in meiotic recombination in both mice and Caenorhabditis elegans (8–10). To date, however, neither protein has been implicated in germline feminization.

During oogenesis in mice, several transcription factors facilitate folliculogenesis: lhx8 (LIM homeobox 8) and figla (factor in the germline alpha) regulate differentiation of primordial follicles (11–14), and nobox (newborn ovary homeobox), which acts downstream of Lhx8, is indispensable for the formation of secondary follicles (12, 15). It remains unknown how these transcription factors cooperate after sexually indifferent germ cells commit to oogenesis.

Here, we show that foxl3 orchestrates two independent pathways for oogenesis: one leading to meiosis by rec8a and the other to follicle development by fbxo47. In addition, we show that the fbxo47-downstream pathway ensures that germ cells do not deviate from the oogenesis pathway during early meiotic prophase I; otherwise, the sexual fate of the germ cells would change to spermatogenesis. Thus, we propose that the diplotene stage of meiotic prophase I is the point of no return for establishing germline feminization.

Results

rec8a Is Required for Female Meiotic Progression.

To investigate the function of rec8a in germline feminization, we generated rec8a mutants by CRISPR/Cas9 (SI Appendix, Fig. S1). rec8a^−/− XY males developed normal testes with mature sperm (SI Appendix, Fig. S2 A and B). Although the hatching rate of eggs fertilized with rec8a^−/− XY sperm was slightly lower than that of eggs with wild-type sperm, rec8a^−/− XY sperm were fertile (SI Appendix, Fig. S2 C and D). On the other hand, rec8a^−/− XX females were sterile, and all of the mutant ovaries lacked follicles (Fig. 1 A and B). Because we could not find germ cells at the diplotene stage in the mutant ovaries, meiosis was likely arrested before the diplotene stage (Fig. 1 A and B, Insets).

Fig. 1.

rec8a is required for female meiosis. (A and B) Periodic acid–Schiff (PAS) staining of 2.5-mo-old wild-type (A, n = 3) and rec8a^∆22−/− (B, n = 2) XX gonads. Magnified images of zygotene (A′ and B′), pachytene (A′′ and B′′), and diplotene (A′′′) oocytes, indicated by arrowheads, are shown in the Insets. Bouquet chromosomal structure and paired bivalents were observed in zygotene and pachytene oocytes, respectively. Asterisks in A and B indicate the ovarian cavity, a typical ovarian structure of medaka. (Scale bars: 100 μm [A and B], 10 μm (A′, A′′ A′′′, B′, and B′′). (C–H) Immunohistochemistry (IHC) of 10-dph rec8a^∆22+/− (C–E, n = 5) and rec8a^∆22−/− (F–H, n = 4) XX ovaries. In the Top, samples were stained for a synapsis marker SYCP1 (green) and a germ cell marker OLVAS (blue) with DAPI staining (gray). Gray scale images of SYCP1 signals are shown in the Bottom. Germ cells indicated by dotted lines represent zygotene (C and F), pachytene (D), diplotene (E), and pachytenelike stages (G and H) characterized by cellular sizes and chromosomal structures. (Scale bar: 10 μm.) (I and J) Immunostaining of XX adult ovaries of wild-type (I, n = 3) and rec8a^∆8−/− fish (J, n = 3) for cleaved CC-3 (green) and a female mitotic germ cell marker FOXL3 (blue) with DAPI staining (gray). Arrowheads in J indicate CC-3-positive germ cells with DAPI-dense DNA fragments. (Scale bar: 50 μm.) (K) Percentage of CC-3-positive germ cells in pachytenelike oocytes (length of long axis: 15–25 µm, SI Appendix, Table S1) in wild-type and rec8a^−/− XX ovaries. Values are means ± SD from three ovaries. P value was calculated by two-tailed Student’s t test. Exact values are shown in SI Appendix, Table S2.

We next examined the synaptonemal complex (SC) formation in rec8a^−/− ovaries. In the control (rec8a^+/−) ovaries, synapsis marker SYCP1 (a central element of SC) localized between homologous chromosomes of zygotene to pachytene oocytes, and these signals disappeared in diplotene oocytes (Fig. 1 C–E). In rec8a^−/− ovaries, although SYCP1 localized normally to the forming SC axes of zygotene chromosomes, the signal became punctate at a pachytenelike stage (Fig. 1 F–H). These results suggest that SC axes prematurely disassembled in the rec8a^−/− oocytes and that rec8a is required for the stability of the SC axis during the pachytene stage.

To explore the possibility that abnormal oocytes were eliminated by apoptosis, we immunostained for activated caspase-3 (cleaved caspase-3 [CC-3]) in wild-type and rec8a^−/− ovaries. We first measured the length of the oocyte long axis (diameter) at zygotene, pachytene, and diplotene stages in wild-type ovaries and found that the diameter of pachytene oocytes ranged from 15 to 25 µm (SI Appendix, Table S1). Immunostaining of rec8a^−/− ovaries revealed that most CC-3-positive germ cells with fragmented DNA had a pachytenelike diameter (Fig. 1 I and J and SI Appendix, Table S2). These CC-3-positive pachytenelike oocytes (long axis: 15–25 µm) accumulated at higher levels in rec8a^−/− than in wild-type ovaries (Fig. 1K). This observation, combined with the mutant phenotype, suggests that aberrant rec8a^−/−oocytes were removed by apoptosis.

fbxo47 Is Involved in Suppression of Spermatogenesis in Female Germ Cells.

To examine the function of fbxo47, we developed a germline-targeted mutant of this gene (SI Appendix, Fig. S3). fbxo47-mutant XX ovaries exhibited a defect in folliculogenesis (Fig. 2A), whereas mutant XY testes did not have any remarkable phenotype (Fig. 2B). In the mutant ovaries, we observed germ cells with abnormally condensed chromosomes (Fig. 2 A, Inset arrowheads). Whole-mount immunostaining, combined with 3D reconstruction of DAPI signals, revealed that fbxo47-mutant germ cells contained U-shaped chromosomes (SI Appendix, Fig. S4 A and B). The U-shaped chromosomes were positive for phosphorylated histone H3, which is present in mitotic germ cells in wild-type ovaries (SI Appendix, Fig. S4 C–E). Chromosome spreads revealed that the chromosomes in mitotically dividing fbxo47-mutant nuclei (2n = 48) were highly condensed (SI Appendix, Fig. S4G), consistent with the images in SI Appendix, Fig. S4 B and B′. By contrast, control chromosomes at the same stage were well extended (SI Appendix, Fig. S4F). These results suggest that cell division was impaired or delayed at metaphase in fbxo47-mutant ovaries and that fbxo47 is associated with regulation of mitotic chromosomal structure.

Fig. 2.

fbxo47 is involved in suppression of spermatogenesis in female germ cells. (A and B) PAS staining of 4-mo-old fbxo47-mutant XX (A, n = 2/3) and XY (B, n = 2) gonads. Inset in A is a magnified image of the region indicated by the white square. Arrowheads and arrows indicate germ cells with abnormally condensed chromosomes and spermlike cells, respectively. (Scale bars: 200 μm.) (C) Giemsa staining of meiotic chromosomes of wild-type XX (Top), wild-type XY (Middle), and fbxo47-mutant XX (Bottom) germ cells. The meiotic stage is shown above each panel. spermatocyte (SPC). (Scale bar: 10 μm.) (D–G) In situ hybridization for shippo1b, a marker of secondary SPCs (second. SPC) and spermatids (ST), using wild-type testes (D, n = 3), fbxo47-mutant ovaries (E, n = 3), fbxo47^∆4+/− 10-dph XX gonads (F, n = 2), and fbxo47^∆4−/− 10-dph XX gonads (G, n = 4). Arrowheads indicate shippo1b-expressing cells (blue). Germ cell cysts (D) or 10-dph gonads (F and G) are outlined by dotted lines. (Scale bar: 20 μm.) (H and I) Immunostaining of 10-dph foxl3^∆17−/− XX gonads (H, n = 6) and fbxo47-induced foxl3^∆17−/− XX gonads (I, n = 5) for OLVAS (red), EGFP (green), and DAPI (gray). Yellow dotted lines indicate secondary SPCs and STs. All EGFP+ germ cells observed in fbxo47-induced gonads were mitotic (arrowhead in I), suggesting that ectopic fbxo47 expression suppressed progression of spermatogenesis in foxl3^∆17−/− XX gonads. In I, two images taken from the anterior and posterior regions of the same gonad were merged to produce a single entire image. (Scale bar: 20 μm.)

Surprisingly, fbxo47-mutant ovaries developed spermlike cells with small condensed nuclei in the expanded germinal epithelium (Fig. 2 A, Inset arrows). To examine the process of gametogenesis in detail, we observed the chromosomal morphology of meiotic germ cells in wild-type ovaries, wild-type testes, and fbxo47-mutant ovaries (Fig. 2C). In wild-type ovaries and testes, chromosomes gradually condensed and synapsed by the pachytene stage of meiotic prophase I. After the pachytene stage, gametogenesis proceeds differently between males and females: Oogenesis arrests at the diplotene stage (Fig. 2 C, Top), whereas, in spermatogenesis, the second meiotic division is completed, followed by spermiogenesis (Fig. 2 C, Middle). Intriguingly, the chromosomal morphology in some population of germ cells in fbxo47 mutants was similar to that observed in wild-type spermatogenesis (Fig. 2 C, Bottom, and SI Appendix, Fig. S4H).

To test the hypothesis that fbxo47-mutant germ cells committed to spermatogenesis, we examined the expression of shippo1b, a spermatocyte and spermatid marker, by in situ hybridization. The shippo1b signal was detected in both the control wild-type XY testes and the fbxo47-mutant XX ovaries (Fig. 2 D and E). Interestingly, shippo1b was already expressed in fbxo47^−/− XX gonads at 15 days posthatching (dph), consistent with the time when spermatogenesis initiated in the foxl3^−/− XX gonads (1) (Fig. 2 F and G and SI Appendix, Fig. S5).

To further analyze the function of fbxo47, we introduced a bacterial artificial chromosome (BAC) containing the fbxo47 allele with enhanced green fluorescent protein (EGFP) into foxl3 mutants (SI Appendix, Fig. S6 and Table S3). None of the EGFP-expressing germ cells in foxl3^−/− XX gonads (n = 5) exhibited spermatogenic progression (Fig. 2 H and I). Collectively, these data strongly suggest that fbxo47 is a downstream factor of foxl3 that is essential for suppressing spermatogenesis.

Two Pathways Downstream of foxl3 Contribute to Germline Feminization.

As described above, two foxl3-downstream genes rec8a and fbxo47 are female-specific regulators of meiosis progression and suppression of spermatogenesis, respectively. To examine the relationship between these two genes and the transcription factors involved in early folliculogenesis, we generated medaka mutants of lhx8b, figla, and nobox, and analyzed the phenotypes with regard to the functions of rec8a and fbxo47.

One of the two medaka Lhx8 orthologs lhx8b was up-regulated downstream of foxl3 as determined by RNA sequencing (2). The lhx8b mutant exhibited female-specific defects in gametogenesis: It developed typical ovarian structures that contained no follicles (Fig. 3A and SI Appendix, Fig. S7). Instead, the lhx8b^−/− ovaries contained spermatocyte- and spermlike cells (Fig. 3A, arrows). In addition, shippo1b was expressed in the lhx8b^−/− ovaries (Fig. 3B), suggesting that germ cells undergo spermatogenesis.

Fig. 3.

lhx8b and figla but not nobox are related to suppression of spermatogenesis. (A) PAS staining of lhx8b^∆31−/− XX gonads. Arrows indicate spermlike cells. (Scale bar: 50 μm.) n = 4. (B) shippo1b expression in lhx8b^∆31−/− XX ovaries. Arrowheads indicate shippo1b-expressing germ cells. Bar: 20 μm. n = 3. (C) PAS staining of figla^∆16−/− XX gonads. Arrows indicate spermlike cells. (Scale bar: 50 μm.) n = 3. (D) shippo1b expression in figla^∆16−/− XX ovaries. Arrowheads indicate shippo1b-expressing germ cells. (Scale bar: 20 μm.) n = 5. (E) Gonadal morphology of 4-mo-old nobox^∆5+/+ (Top) and nobox^∆5−/− (Bottom) XX ovaries. (Scale bar: 1 mm.) (F and G) IHC of nobox^∆5+/+ (F, n = 4) and nobox^∆5−/− (G, n = 2) XX ovaries for OLVAS (green), and FOXL3 (red) with DAPI staining (gray). (Scale bar: 50 μm.)

figla expression in medaka is first detected in germ cells at early meiotic prophase I (16). In medaka figla^−/− ovaries (16), folliculogenesis is disrupted as in mouse Figla-knockout (KO) ovaries (14). Careful observations of medaka figla^−/− ovaries revealed the presence of spermlike cells with small condensed nuclei within the expanded germinal epithelium (Fig. 3C, arrows). In addition, shippo1b was expressed in the figla^−/− ovaries (Fig. 3D). These results suggest that figla^−/− germ cells, such as fbxo47^−/− and lhx8b^−/− germ cells, committed to spermatogenesis.

Next, we examined the nobox mutant phenotype (SI Appendix, Fig. S8 A–C). The nobox^−/− mutants exhibited female-specific infertility and developed small ovaries devoid of large follicles (Fig. 3E). Histological analysis revealed that follicles were normally formed (Fig. 3 F and G) but their growth was arrested at the stage of 300 µm-diameter oocyte in nobox^−/− ovaries (SI Appendix, Fig. S8D). These data suggest that, in medaka, as in mammals, nobox is required for follicle growth.

Finally, we conducted molecular epistasis analysis for rec8a, fbxo47, lhx8b, figla, and nobox using these mutants. In wild-type ovaries, rec8a expressed in mitotic germ cells, fbxo47 expressed in mitotic and early meiotic (leptotene to zygotene) germ cells, lhx8b expressed from leptotene to mainly zygotene oocytes, and figla and nobox expressed from zygotene to mainly diplotene oocytes (Fig. 4 A, E, I, M, and Q). In rec8a^−/− ovaries, all genes examined were expressed (Fig. 4 F, J, N, and R). On the other hand, in fbxo47^−/− ovaries, expression of lhx8b, figla, and nobox was not detected, whereas rec8a was expressed (Fig. 4 B, K, O, and S). In lhx8b^−/− ovaries, expression of figla and nobox was not observed, but rec8a and fbxo47 were expressed (Fig. 4 C, G, P, and T). In figla^−/− ovaries, nobox expression was not observed, but the other genes were expressed (Fig. 4 D, H, L, and U).

Fig. 4.

Two pathways downstream of foxl3 contribute to germline feminization. In situ hybridization for rec8a (A–D), fbxo47 (E–H), lhx8b (I–L), figla (M–P), and nobox (Q–U) using wild-type XX 10 dph (A, E, I, M, and Q, n = 3), rec8a^∆22−/− XX (F, J, N, and R, n = 3), fbxo4^7∆4−/− XX 10 dph (B, K, O, and S, n = 3), lhx8b^∆31−/− XX (C, G, P, and T, n = 3), and figla^∆16−/− XX gonads (D, H, L, and U, n = 3). Arrowheads indicate signal-positive germ cells (blue). The 10-dph gonads of wild type and fbxo47^∆4−/− are outlined by black dotted lines. Blue dotted lines indicate germ cell cysts in 10-dph gonads. (Scale bar: 30 μm.)

To summarize, there are, at least, two genetically independent pathways downstream of foxl3 (Fig. 5): the rec8a-dependent pathway, which regulates progression of meiosis in female medaka, and the fbxo47-dependent pathway, which suppresses germ cell commitment to spermatogenesis and promotes folliculogenesis. Interestingly, ovaries harboring mutations in fbxo47, lhx8b, or figla developed spermatocytes, whereas nobox-mutant ovaries did not. The mutant phenotypes along with the timing of gene expression in germ cells suggest that spermatogenesis is continuously repressed until the diplotene stage in order to establish germline feminization.

Fig. 5.

Process of germline feminization in medaka. foxl3-dependent and foxl3-independent molecular pathways contribute to germline feminization. Downstream of foxl3, fbxo47, and rec8a promote folliculogenesis and meiosis, respectively. The fbxo47-downstream pathway, which is also important for oogenesis commitment, keeps germ cells in the oogenic pathway and prevents them from differentiating into spermatogenic cells. atp1a3a is up-regulated in female germ cells independently of foxl3 (2).

Discussion

In general, vertebrate meiosis initiates much earlier in the ovary than in the testis. Oogenesis with meiosis occurs during embryogenesis in many vertebrates, while meiosis in the testis starts after the onset of puberty. Accordingly, entry into meiosis is considered a hallmark of feminization of germ cells. In mammals, bone morphogenetic protein and retinoic acid (RA) signaling induce female-type differentiation (oogenesis) and meiosis, respectively (17–22). By contrast, in medaka, it appears that RA signaling does not play a critical role in meiosis initiation during the sex determination period (23). Furthermore, a RA-responsive gene, stimulated by retinoic acid gene 8, has been lost from most teleost genomes (24).

Identification of foxl3 as a sexual switch in medaka reveals that the process of sex determination in germ cells begins much earlier than a series of meiotic events. In this study, we successfully linked the regulatory mechanisms of oogenesis with germline sex determination where rec8a and fbxo47 are critical effectors that initiate germline feminization downstream of the germline sexual switch foxl3. The pathways controlled by these two genes are independent genetically (Fig. 5).

Interestingly, the phenotype of the rec8a mutant indicates that rec8a is required for progression of oogenesis but not spermatogenesis. This raises the possibility that meiotic progression is regulated differently between the sexes. Consistent with this, sexual dimorphisms in meiosis, such as the number of crossovers per nucleus, axis length of meiotic chromosomes, and recombination rate, have been reported in several animals (25–29). Although the molecular mechanisms responsible for generating such sexual dimorphisms have not been elucidated, functional investigation of rec8a may provide insight into the role of sexual dimorphisms in meiosis.

fbxo47-mutant ovaries exhibited two main phenotypes: abnormal morphology of chromosomes and commitment to spermatogenesis. In C. elegans, disruption of the fbxo47 homolog and progression of meiosis 1 causes defects in meiotic processes, such as homologous pairing and synapsis (9). In rice (Oryza sativa), the F-box proteins MEIOTIC F-BOX and ZYGOTENE1 are involved in double-strand break repair and formation of bouquet structure of zygotene chromosomes, respectively (30, 31). These data may collectively imply that the functions of F-box proteins regulate germ cell chromosomal structure not only in mitosis, but also in meiosis in sexually reproductive organisms.

The precocious spermatogenesis in fbxo47^−/− ovaries partially phenocopied that observed in foxl3^−/− ovaries (1). This demonstrates that fbxo47 is involved in suppression of spermatogenesis downstream of foxl3. Supporting this, germ cells artificially inducing fbxo47 in the foxl3^−/− female (XX) did not show any progress of spermatogenesis. In addition, two fbxo47-downstream genes lhx8b and figla, which are expressed in early meiotic prophase I, are also involved in the process of germline feminization as we found that disruption of either gene induced spermatogenesis in ovaries. By contrast, germ cells in nobox^−/− ovaries did not develop spermatogenic cells. These results suggest that germline sex is fixed before the follicles start to grow at diplotene stage of meiosis I. Thus, although foxl3 turns on the switch of the germline sex determination pathway during the mitotic phase (1), our results suggest that the sex of the germline is established during early meiotic prophase I. We propose that the diplotene stage, just before the follicles form but have not yet grown, is the point of no return for germline sex determination (Fig. 5).

Lastly, it may be worth noting that atp1a3a is up-regulated in female germ cells independently of foxl3 (2). In addition to the two pathways revealed in this study, another molecular pathway might function during the course of germline feminization (Fig. 5).

Materials and Methods

Fish.

The OKcab strain of medaka fish (O. latipes) was used in this study. Fish were maintained in fresh water at 25–28 °C under a regulated photoperiod (14-h light and 10-h dark). All experiments were conducted with the approval of the Nagoya University official ethics committee (Approval no. 9 in the Department of Science, Nagoya University).

Sex Genotyping.

Genotypic sex of all animals was determined by qPCR using TaqMan MGB probes (Thermo Fisher Scientific) that detect the male-determining gene dmy (32, 33) and the autosomal gene cyp19a1 as a control (SI Appendix, Table S4). Genomic DNA was extracted from the larval head or tip of the caudal fin by freezing at −80 °C for 5 min and heating at 95 °C for 5 min. Extracted DNA was diluted and used as a PCR template. qPCR was performed using TaqMan Fast Advanced Master Mix on a StepOnePlus real-time PCR system (Thermo Fisher Scientific). Cycling parameters were as follows: 40 cycles of 95 °C for 3 s and 60 °C for 20 s. PCR primers used for genotyping are shown in SI Appendix, Table S4.

Generation of KO Medaka Lines by CRISPR/Cas9.

CRISPR/Cas9 targeting sites were identified using the CRISPR/Cas9 target online predictor (CCTop; https://crispr.cos.uni-heidelberg.de/index.html). Selected target sites are shown in SI Appendix, Figs. S1, S5, S7, and S8.

To prepare template DNA for synthesis of guide RNAs (gRNAs) by in vitro transcription (IVT), we used two methods. For lhx8b and nobox, the designed oligonucleotides were annealed and cloned into vector DR274 (Addgene no. 42250) as described previously (16). For rec8a and fbxo47, we used a simplified method without cloning (34). Briefly, sense and antisense oligonucleotides were 5′-tagged with a T7 promoter and 15 bp of transactivating CRISPR (tracr) sequence, respectively. T7-promoter-target tracr was amplified by annealing and PCR with the following cycle parameters: 95 °C for 1 min; 32 cycles of 98 °C for 5 s; 55 °C for 15 s; and 68 °C for 1 min. PCR primer sets are shown in SI Appendix, Table S4. The resultant PCR products were used as templates for IVT. Protocols for gRNA synthesis, Cas9 mRNA synthesis, and microinjection of gRNA and Cas9 mRNA into medaka embryos were described previously (16). Each mutant allele was identified in the F1 generation by sequence analysis using the primer sets shown in SI Appendix, Table S4. KO efficiency of mutant lines is shown in SI Appendix, Table S5.

For analysis of the fbxo47 function (Fig. 2), we established a method for inducing mutations efficiently in the germline of the F0 generation. Cas9 fused with nanos3 3′UTR (Cas9-nanos3 3′UTR) mRNA was synthesized from vector pCS2+hSpCas9-nos3 3′UTR, which was generated by insertion of the nanos3 3′UTR sequence (35) into the XbaI site of vector pCS2+hSpCas9 (Addgene no. 51815). Efficiency of germline mutagenesis was evaluated by injecting gRNAs targeting DsRed and EGFP (SI Appendix, Table S4) into olvas-EGFP/sox9b-DeRed transgenic medaka (36, 37) (SI Appendix, Fig. S3). Two fbxo47-targeting gRNAs (SI Appendix, Table S4, 25 ng/µL each) and Cas9-nanos3 3′UTR mRNA (50 ng/µL) were coinjected into medaka embryos. Primers used for vector construction are shown in SI Appendix, Table S4.

To induce ectopic expression of fbxo47 in foxl3^∆17−/− gonads (SI Appendix, Fig. S6), the fbxo47 cassette (fbxo47 ORF-fbxo47 3′UTR-BGHpA-CAT) and olvasp-EGFP cassette (olvas promoter-EGFP ORF-olvas 3′UTR-BGHpA-Kmr) were inserted into medaka BAC (clone name: ola1-212j01, purchased from NBRP Medaka [https://shigen.nig.ac.jp/medaka/]) by homologous recombination as previously described (38). fbxo47 ORF-fbxo47 3′UTR was cloned into the EGFP locus of vector hsGBA-NCm by using an In-Fusion HD Cloning Kit (Takara no. 639648). hsGBA-NCm was generated by exchanging the Kmr gene of hsGBA-NKm (38) for the CAT gene of pACYC184 (Addgene no. 37033) by In-Fusion. The resultant vector was used as a PCR template to amplify the fbxo47 cassette, which was recombined into the foxl3 gene locus in the BAC (ola1-212j01) that contains the olvasp-EGFP cassette in the region between foxl3 and tmem201 (1). Primers used for vector construction and fragment amplification are listed in SI Appendix, Table S4. The resultant construct was injected into progeny of foxl3^∆17+/− females crossed with foxl3^∆17−/− males at the one-cell stage (1). Injected larvae were fixed at 10 dph and used for IHC. Injection and sampling data are summarized in SI Appendix, Table S3.

In Situ Hybridization, IHC, and Histology.

Whole-mount in situ hybridization, immunohistochemistry, and PAS staining were performed as previously described (16, 39, 40). complementary DNA (cDNA) clones for fbxo47 (NBRP clone name: olte61d07, RefSeq ID: XM_011489632) and lhx8b (NBRP clone name: olova4m02, RefSeq ID: XM_ 023965531) were purchased from NBRP Medaka. rec8a (RefSeq ID: XM_011485338) and nobox (RefSeq ID: XM_004074351) cDNAs were cloned from meioC^−/− and wild-type ovarian cDNA libraries, respectively. Each cDNA was used as the PCR template for amplification of T7-promoter-tagged DNA, which was then used as an IVT template. PCR primers are shown in SI Appendix, Table S4. Digoxigenin-labeled RNA probes were synthesized as described previously (16).

Antiserum or antibodies for the following proteins were used in this study: medaka OLVAS (40) (1:100; rat), SYCP1 (41) (1:100; mouse), CC-3 (1:100, rabbit, CST no. 9661), FOXL3 (2) (1:200, mouse), and phosphorylated histone H3 (1:100, mouse, Millipore nos. 06–570). Secondary antibodies were Alexa Fluor 488-, 568-, and 647-conjugated goat antibodies (1:100, Thermo Fisher Scientific). IHC images were captured and analyzed on an Olympus FLUOVIEW FV1000.

Preparation of Chromosomal Spreads.

Whole gonads were dissected and cut into 1-mm² fragments and treated with hypotonic solution (1% wt/vol trisodium citrate) for 15 min at room temperature. Tissues were placed on a slide glass, and fixative I (ethanol:acetic acid:H₂O = 3:3:4) was dropped onto the tissues. When the fixative dried out, tissues were minced with fine forceps into pieces as small as possible. Residual fixative was removed, fixative I was added, and the sample was incubated at room temperature until the solution just started to dry. Fixative II (ethanol:acetic acid = 1:1) was added; just before the solution dried out, another fixative II was added. Finally, acetic acid was added, and the sample was dried completely. Slides were washed once with 50% methanol and stained with 4% Giemsa (Millipore no. 109204) in Sorenson’s phosphate buffer (33.36-mM KH₂PO₄, 33.46-mM Na₂HPO₄, pH 6.8) for 15 min. Slides were washed with tap water, dried completely, and enclosed in Eukitt (ORSAtec).

Meiotic stages were determined in reference to the following chromosomal characteristics: At the zygotene stage, chromosomes are huddled together in a small compact mass. By the midpachytene stage, the size of the nucleus becomes larger, and all of the synapsed chromosomes are visible as fine threads within the nucleus. At the diplotene stage, each chromosome becomes distinguishable in the spermatocyte, while lampbrush chromosomes are formed in the oocyte. At metaphase I, one or two chiasmata between fully condensed homologs are observed, whereas such structures are not found at metaphase II. Secondary spermatocyte, round spermatid, and sperm can be distinguished on the basis of nuclear size.

In SI Appendix, Fig. S4, ovaries of adult meioC^−/− XX were used as a control for normal mitotic chromosomes because they lack follicles (16), enabling us to obtain mitotic cells efficiently.

Data Availability.

Data availability is described in the text and/or SI Appendix files.