The AB strain zebrafish and inhibin null mutant (inha−/−) used in this study were maintained in the ZebTEC Multilinking Rack zebrafish system (Tecniplast, Buguggiate, Italy) at 28°C with a lighting scheme of 14-h (8:00 am–10:00 pm) light and 10-h dark. The inha mutant (umo19 with ZFIN) was recently created in our laboratory (Lu et al., 2020). The fish were handled according to the guidelines and protocols approved by the Research Ethics Panel of the University of Macau.

The egf and egfra mutants were generated by CRISPR/Cas9 gene editing method as described in our previous report (Lau et al., 2016). Briefly, the target sites were designed using the ZiFiT Targeter (version 4.2, http://zifit.partners.org/ZiFiT/Disclaimer.aspx), which identified target sequences in exon 4 and exon 6 of egf, and exon 7 of egfra genes, respectively. A target site in egfrb gene, a truncated form of EGFR-like molecule, was also designed based on an EST sequence as mRNA information was not found in the GeneBank. The target oligonucleotides were synthesized and cloned into the sgRNA expression vector pDR274 (Addgene #42250). The DraI-digested pDR274 sgRNA constructs were used as template to transcribe sgRNA by MAXIscript T7 kit (Thermo Fisher Scientific, Waltham, MA, USA). The Cas9 mRNA was transcribed from pCS2-nCas9n plasmid (Addgene #47929, Watertown, MA, USA) by mMESSAGE mMACHINE SP6 kit (Thermo Fisher Scientific). A mixture of 100 pg of sgRNA and 400 pg of Cas9 mRNA were co-injected into the one-cell stage embryos. All embryos were maintained in 28°C fish water. Genotyping was performed on DNA extracted from each zebrafish embryo or tail fin cut (Zhang et al., 2015b).

Genomic DNA was extracted by incubating the embryo or tail fin cut in 50 mM NaOH at 95°C for 12 min. After cooling to room temperature, one-tenth volume of 1 M Tris (pH 8.0) was added to neutralize the solution. Then the solution was centrifuged at 1,500 rpm for 5 min. DNA in the supernatant was used as the template for PCR. The primers for PCR were listed in Supplementary Table S1. Real-time qPCR was performed using SsoFast EvaGreen Supermix on C1000 Thermal Cycler CFX96 Real-time PCR Detection System (Bio-Rad, Hercules, CA, USA). HRMA was performed at the end of reaction with Precision Melt Analysis software (Bio-Rad) to analyze the difference of the melt curves.

The PCR products from HRMA were used for HMA. They were separated on 20% nondenaturing polyacrylamide gel (20% acrylamide, 1 × TBE, 0.01% ammonium persulfate, 0.04% TEMED) at a constant voltage of 150 V for 4 h. The gel was stained with GelRed for 10 min and imaged on the ChemiDoc MP Imaging System (Bio-Rad). The mobility of heteroduplexes was slower than homoduplexes. To distinguish homozygous mutant (−/−) from wildtype (WT, +/+) in F2 generation, WT genomic DNA was added to each sample. The homozygous mutant (−/−) produces hybrid (+/−) after spiking with WT DNA, while WT (+/+) remains the same (+/+).

The F0 fish carrying mutations was crossed with WT fish to obtain the F1 generation (+/−). Different mutation patterns were identified by HRMA and HMA. The PCR products containing target sites from each pattern were cloned into pMD18-T vector (TaKaRa, Shiga, Japan) and transformed into DH5α competent E. coli cells. The plasmids in monoclonal bacteria were purified for DNA sequencing. The male and female fish with the same frame shift mutations were crossed to produce homozygous F2 mutants (−/−).

Growth rate was compared between homozygous mutant fish and their heterozygous siblings. The body weight (BW) and standard body length (BL) of males were measured every 10 days from 50 to 120 dpf, the period when zebrafish displays the highest growth rate. We chose males for growth analysis because their somatic growth was less influenced by gonadal size (n = 30). All data were expressed as mean ± SEM. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s comparison test (p < 0.05).

The fertility of different genotypes was assessed by natural mating with WT partners. Individuals that failed to spawn after at least 10 trials were considered infertile. Once fertilized embryos were obtained, genotyping was performed on sampled embryos by HMA to confirm the genotypes of the parents.

Fecundity of egf mutant females was assessed at 4 and 7 months post-fertilization (mpf). All females were separated from males for 1 week before fecundity test. Then, each egf ^−/− female fish was mated with two WT males in the spawning box (Tecniplast). Five females were examined in each test. Total egg number was counted including both fertilized and unfertilized ones. The fertilization rate was the ratio of fertilized eggs in the total. The fecundity tests were repeated five times at 4-day interval. Age-matched WT females were used as the control. The data were expressed as mean ± SEM. Statistical significance was determined using independent sample t-test (p < 0.05).

All fish used for histological examination were genotyped first on DNA from the tail fin cut. Sibling WT (+/+) and/or heterozygous (+/−) fish were used as the controls. The fish was anesthetized with MS-222 (tricaine methanesulphonate, 250 mg/L; Sigma-Aldrich, St. Louis, MO, United States) and BW and BL were recorded. The body and gonad were photographed before fixing with Bouin’s solution. The samples were fixed for at least 24 h, washed with 50% ethanol, dehydrated, and embedded in paraffin. The samples were sectioned at 5 µm and stained with hematoxylin and eosin (HE) for microscopic examination.

The body weight (W_b) and gonad weight (W_g) of each fish were measured at 4 mpf for homozygous mutant and their heterozygous siblings. The GSI was calculated as follows: GSI (%) = 100 * W_g/W_b. More than 10 fish were examined for each genotype (n = 11–19). Statistical significance was determined using independent sample t-test (p < 0.01).

Tissue samples or different stage follicles (primary growth, PG; previtellogenic, PV; early vitellogenic, EV; mid-vitellogenic, MV; full-grown, FG) were collected and homogenized in 500 μl Trizol Reagent (Thermo Fisher Scientific) according to our previous reports (Wang and Ge, 2004b; Zhou et al., 2011). Total RNA was extracted from each sample according to the protocol of the manufacturer and treated with DNase I (Invitrogen, Carlsbad, CA, United States) to ensure no genomic DNA contamination. Reverse transcription was carried out at 37°C for 60 min in 10 µl reaction buffer containing 1 μg total RNA, 5 μM oligo dT primer, 0.75 mM deoxynucleotide triphosphate mixture, and 200 U M-MLV reverse transcriptase (Thermo Fisher Scientific). The reaction was inactivated by heating at 70°C for 15 min. Quantitative PCR (qPCR) reactions were performed on the CFX96 Real-Time PCR Systems using SsoFast EvaGreen Supermix (Bio-Rad). The expression levels of target genes during folliculogenesis were first normalized to that of the housekeeping gene ef1a and then expressed as the fold change relative to that at the PG stage. The primers used for cDNA amplification were designed using Primer Premier 6 (Supplementary Table S1). Each experiment was performed at least twice in triplicate and all values were expressed as mean ± SEM. The qPCR data were analyzed by ANOVA followed by Tukey’s comparison test (p < 0.05).

To compare ovarian transcriptomes between egfra ^+/− and egfra ^−/−, the ovaries of 45-dpf females were selected for RNA-seq analysis. Three fish with BL at 1.8 cm and BW at 100 mg were sampled for each genotype (n =3 biological replicates). The fish were anesthetized and the whole ovary from each fish was homogenized for RNA extraction. RNA-seq and data analysis were performed by Novogene Bioinformatics Technology (Tianjin, China). The sequencing of each sample generated approximately 4G raw data, which have been submitted to the NCBI SRA database with accession numbers SRR12432918–12432923. TopHat2 algorithm was chosen to map reads to the zebrafish genome, and FPKM (Fragments Per Kilobase Million) was used to normalize the expression data followed by expression level estimation and differential expression analysis by HTSeq and DESeq software (padj <0.05). Gene ontology (GO) analysis was performed on genes that showed twice higher or lower expression in the ovary of egfra−/−fish than that of egfra+/−fish.

The ovaries of 45-dpf egfra ^+/− and egfra ^−/− females (n = 3; ∼1.8 cm BL and ∼100 mg BW) were dissected out, dispersed and incubated in M199 medium for 30 min. The medium was then replaced with fresh M199 containing 200 nM recombinant human EGF (PeproTech, Rochy Hill, NJ, USA). The medium was removed after 20 min treatment. The treated ovarian fragments and follicles were homogenized with a pestle in cold SDS sample buffer (63 mM Tris-HCl pH 6.8, 10% glycerol, 5% β-mercaptoethanol, 3.5% sodium dodecyl sulfate, 1% w/v SDS). The samples were centrifuged on a microfuge at the highest speed for 5 min. The supernatant was collected and mixed with loading buffer followed by heating at 95°C for 10 min. The samples were separated on 12% polyacrylamide gels and transferred to PVDF membranes. The membranes were blocked with 5% nonfat dry milk (Bio-Rad) in 1× TBST at room temperature for 1 h. After rinsing once with 1× TBST, the membranes were incubated at 4°C overnight in 1× TBST with 2% BSA and primary antibodies for p-Erk1/2 (1:2,000; #4370) and Erk1/2 (1:1,000; #9102) (Cell Signaling Technology, Danvers, MA, United States). The membranes were then washed with 1× TBST three times followed by incubation with HRP-conjugated anti-rabbit IgG (1:5,000; #7074) (Cell Signaling) at room temperature for 1 h. After washing, the membranes were incubated with ECL Western Blotting Substrate (Thermo Scientific Pierce), and images were detected on the ChemiDoc MP Imaging System (Bio-Rad).

The paraffin sections were deparaffinized in xylene and rehydrated in gradient ethanol and water. Antigen retrieval was performed in 10 mM sodium citrate buffer at sub-boiling temperature for 10 min. The endogenous hydrogen peroxidase was inactivated by treatment with 3% hydrogen peroxide (H₂O₂) for 10 min. The sections were washed with 1× PBS three times for 5 min each before blocking for 1 h at room temperature with normal horse serum. Each slide was then incubated at 4°C overnight with 100 μl of p-MAPK3/1 antibody (#4370, Cell Signaling) diluted at 1:100 in blocking solution. The section was then washed with 1× PBS three times for 5 min each before incubation with 100 μl of HRP-linked anti-rabbit IgG antibody (#7074, Cell Signaling) for 30 min at room temperature. After washing with 1× PBS three times for 5 min each, 100 μl of DAB solution was added to each section and incubated for 10 min. The section was washed with tap water for 5 min to stop reaction, dehydrated, and mounted with Permount (Thermo Fisher Scientific).

The mRNA level of target gene was determined by qPCR, normalized to the housekeeping gene ef1a, and expressed as fold change relative to the control group. The expression level of each gene in transcriptome analysis was expressed as reads per kilobase of exon model per million mapped reads (FPKM). All values were expressed as mean ± SEM. Statistical analysis was carried out with Prism 5 (GraphPad, San Diego, CA, USA), and one-way ANOVA was used to analyze gene expression or fish growth rate. The GSI and fertilization rate were analyzed by t-test. All experiments were performed at least twice.

In addition to EGFR (Egfra/egfra) that we characterized previously in zebrafish (Wang and Ge, 2004a), another putative form of EGFR (Egfrb/egfrb) is also present but not well annotated in the zebrafish genome. Homology and phylogenetic analysis showed that egfra and egfrb are related and positioned in the same cluster. Egfra is more closely related to EGFR of other fish whereas Egfrb is remotely positioned in the cluster (Supplementary Figure S1A). Protein sequence alignment showed that Egfrb shared high homology with the extracellular domain of Egfra, but it ended at or before the transmembrane domain, representing a truncated form (Supplementary Figure S1B).

RT-PCR analysis showed that egf was predominantly expressed in the ovary and testis with weak expression in some other tissues such as the brain and liver. However, egfra exhibited ubiquitous expression in various tissues investigated with low expression in the liver and muscle. These agree well with our previous reports (Wang and Ge, 2004a; Tse and Ge, 2010). Interestingly, egfrb also showed distinct tissue distribution with expression detectable only in liver and ovary, but not in other tissues (Figure 1A).

We then analyzed the temporal expression profiles of egf and egfra during folliculogenesis in sexually mature females. Quantitative PCR analysis demonstrated that egf and egfra mRNA levels were relatively low in primary growth (PG, stage I) follicles. When PG follicles were activated or recruited into pre-vitellogenic (PV, stage II) stage, their expression levels were both significantly increased and remained high during vitellogenic growth from PV to full-grown (FG) stage (stage III) (Figures 1B, C). Interestingly, egfra but not egf showed a further dramatic increase in expression in FG follicles prior to final oocyte maturation (Figure 1C). These results were consistent with our previous report using real-time RT-PCR (Tse and Ge, 2010).

We generated loss-of-function mutants by CRISPR/Cas9 method, which targeted the coding region downstream of the translation start codon to knock out the protein by frameshifting indel mutations. To disrupt egf, the target site was located in exon 4 of the gene and a mutant containing a 2-bp deletion was selected for phenotype analysis (egf−/−; ZFIN line number: umo21). This frameshifting mutation introduced a stop codon near the target site that resulted in translation of a truncated fragment. To verify the loss of egf gene, we analyzed the expression of egf transcript in the ovary by using a mutant-specific primer (F2) with its 3′-end located in the deletion region. As shown in Supplementary Figure S2A, the primer pair (F1/R1) that flanks the deletion site could amplify signals in both WT (egf+/+) and mutant fish (egf−/−); however, the primer pair F2/R1 could only detect the signal in WT fish (Supplementary Figure S2A). For egfra gene, the target site was located in exon 7 and a 4-bp deletion mutant was identified (egfra−/−; umo22). The 4-bp deletion caused frameshift and translation of a truncated product. We used similar RT-PCR approach to demonstrate the mutation at the transcript level (Supplementary Figure S2B). We also generated a mutant line for egfrb gene with 11-bp deletion (umo23) (Supplementary Figure S2C).

Considering that EGF is a well-known growth factor involved in cell growth and proliferation, we first compared the growth rate of the homozygous mutant males (egf−/−) with their heterozygous male siblings (egf+/−) from 50 to 120 days post-fertilization (dpf). We chose males for growth analysis because their somatic growth was less influenced by gonadal development. The BW and BL of the mutant fish (egf−/−) were comparable with those of their heterozygous siblings (egf+/−) from 50 to 70 dpf. However, the growth of egf−/− mutant slowed down slightly after 70 dpf with BL being slightly but significantly shorter than that of egf+/− fish from 80 to 110 dpf (Figure 2A) and BW lower than that of the heterozygous siblings from 70 to 120 dpf although statistical significance was only detected at 80 dpf (Figure 2B). Similarly, both BL and BW of the homozygous egfra−/− mutant were slightly but significantly lower than those of heterozygous fish (egfra+/−) from 50 to 80 dpf; however, the difference diminished afterward (Figures 2C, D).

Our previous studies showed that EGF was most abundantly expressed in zebrafish gonads (ovary and testis) (Wang and Ge, 2004a; Tse and Ge, 2010) and that it stimulated oocyte maturation in vitro (Pang and Ge, 2002). The expression pattern was confirmed in this study (Figure 1A). After demonstrating that the loss of egf only had slight effect on somatic growth, we turned our attention to its potential impact on gonadal development and function. The mutant egf−/− fish had normal fertility (fecundity and fertilization rate) at 3 mpf in both females and males compared with that of control fish (egf+/−) (data not shown). Histological examination demonstrated well-developed follicles of all stages in the ovaries of both mutant (egf−/−) and control (egf+/+ and egf+/−) females (Figure 3A) and normal spermatogenesis in the testes of all three genotypes as well (Figure 3B). In agreement with this, GSI was also comparable between egf+/− and egf−/− fish in both sexes (Figure 3C). To confirm these results, we also examined another egf mutant line with 44-bp deletion (Supplementary Figure S3; umo24); it also showed normal folliculogenesis and spermatogenesis (Supplementary Figure S4).

The fecundity of egf-deficient females was tested by crossing with the WT males. As shown in Figure 4, the number of eggs produced in each spawning (fecundity) and the fertilization rate were comparable between the control (egf+/−) and mutant (egf−/−) at 4 mpf (Figures 4A, B). However, the mutant females produced less eggs at 7 mpf in most trials and the fertilization rates decreased significantly in all tests (average 25% vs. 87%) (Figures 4C, D).

The lack of substantial impact of egf mutation on development and reproduction was surprising given that egf is most abundantly expressed in the gonads. Considering that multiple EGF family ligands can bind to the same EGF receptor (EGFR, ERBB-1, or HER1 in humans), we hypothesized that the lost EGF function in the egf mutant could have been compensated by other family members. To address this issue, we went on to delete EGF receptor gene (Egfr/egfr) to disrupt the signaling pathway. There were two paralogs of egfr gene in zebrafish genome (egfra and egfrb), and egfrb is a truncated form that has not been characterized. A mutant line of egfra with 4-bp deletion and a line with 11-bp deletion for egfrb were chosen for phenotype analysis.

The female mutant (egfra−/−) exhibited feminine secondary sexual characteristics at 3 mpf, such as silverish body color; however, it showed slim body shape compared with the controls (egfra+/+ and (egfra+/−). When crossing with WT males, all egfra−/− females tested failed to spawn. Dissection revealed that the mutant ovaries (egfra−/−) were translucent and immature, in contrast to the control ovaries (egfra+/+ and egfra+/−), which were full of normal vitellogenic follicles including FG stage. Histological examination showed that the egfra−/− follicles were primarily arrested at the PG stage with only a few entering very early PV stage with some small cortical alveoli in the cytoplasm (Figure 5A).

By comparison, no obvious defects in fertility were observed in egfra−/− mutant males, which could produce viable offspring when crossing with WT (egfra+/+) and heterozygous (egfra+/−) females. Histological examination showed that the testes were structurally normal in egfra−/− males compared with egfra+/+ and egfra+/− males. The testicular lumens were well developed with similar amounts of spermatozoa in egfra−/− (Figure 5B). Measurement of GSI showed that the ovaries of egfra−/− mutant females were significantly smaller than those of egfra+/- females; however, it was comparable between mutant (egfra−/−) and control (egfra+/−) males (Figure 5C).

Similar to egfra, egfrb was also expressed in the ovary (Figure 1). In contrast to egfra, mutation of egfrb (−11-bp) did not cause any abnormal phenotypes and egfrb ^−/− mutant showed normal folliculogenesis and spermatogenesis in adults (90 dpf) (Supplementary Figure S5).

To further investigate the impact of egfra mutation on folliculogenesis, we performed a time course examination of females at different time points of growth and development at 10-day intervals, starting from 25 dpf when zebrafish gonadal sex differentiation started (Maack and Segner, 2003; Chen and Ge, 2013; Qin et al., 2018). At 25 dpf, the ovaries in both egfra+/− and egfra−/− fish contained cystic chromatin nucleolar (CN) oocytes and emerging perinucleolar (PN) oocytes with thread-like chromatin distributed throughout the karyoplasm, representing presumptive juvenile ovary. The presence of PN follicles indicated normal cyst breakdown or follicle assembly from nested oocytes. At 35 dpf, the oocytes in both egfra+/− and egfra−/− had increased significantly in size and most of them had developed into typical PG follicles at PN stage, which is characterized by multiple nucleoli located at the periphery of the germinal vesicle and a layer of somatic cells surrounding each individual oocyte. At 45 dpf, some of the oocytes in egfra+/−fish significantly increased in size with one or multiple layers of small cortical alveoli appearing at the periphery of ooplasm, marking their entry into the PV stage. A thin layer of somatic follicle cells was formed immediately around oocytes. In contrast, the egfra−/− oocytes were mostly at the PG stage. Cortical alveoli also appeared in some mutant oocytes, but with less number and smaller size. At 55 dpf, the oocytes in control females (egfra+/−) underwent extensive growth with abundant and multiple layers of large cortical alveoli. In sharp contrast, the oocytes in mutant females (egfra−/−) remained arrested at PG stage with only a few entering early PV stage that contained small cortical alveoli (Figure 6).

MAPK3/1 (ERK1/2) is the major pathway in zebrafish ovary that transmits EGF/EGFR signaling (Chung and Ge, 2012). To further confirm the loss of egfra gene in the mutant, we analyzed phosphorylation of MAPK3/1 in the follicles from both control (egfra+/−) and mutant (egfra−/−) at 45 dpf and its response to EGF treatment. The ovaries were dispersed into fragments and incubated with EGF (200 nM) for 20 min followed by Western blot analysis for MAPK3/1 phosphorylation. As expected, the control follicles (egfra+/−) showed a strong response to EGF with increased MAPK3/1 phosphorylation. In contrast, no response was observed in mutant follicles (egfra−/−). Interestingly, despite its lack of response to EGF, the mutant follicles showed higher basal levels of MAPK3/1 phosphorylation, which was similar to the EGF-induced level in the control (Figure 7A). This was further confirmed by immunohistochemical staining for phosphorylated MAPK3/1 in the ovary. Strong basal signals could be detected in sections of the mutant ovary without EGF treatment, but not the control, and the staining was mostly localized to the somatic follicle cells surrounding the oocytes (Figure 7B). Transcriptome (see below) and RT-PCR analyses demonstrated significant upregulation of chemokine ligand genes involved in positive regulation of ERK cascade, including ccl25b, ccl34b.1, ccl34b.8, and ccl35.2 (Figure 7C). The increased expression of these genes might be responsible for the increased basal MAPK3/1 phosphorylation observed in egfra null ovary (egfra−/−).

To understand the molecular mechanisms underlying Egfr regulation of follicle activation, we performed a transcriptome analysis on egfra+/− and egfra−/− follicles at 45 dpf when the first wave of follicle transition from PG to PV occurs. Histological examination showed that most follicles in egfra+/− ovaries were at PG stage with a few entering PV stage (body size: ∼1.90 cm BL; ∼138 mg BW). Similarly, most follicles in egfra−/− ovaries were also at PG stage with some at early PV stage (body size: ∼1.90 cm BL; ∼132 mg BW) (Figure 8A). Transcriptome analysis showed that 5.7% (847/14,830) of the genes detected were differentially expressed between egfra+/− and egfra−/− follicles (Figure 8B). Among the differentially expressed genes, 44 genes were significantly downregulated and 23 were upregulated in egfra−/− ovaries compared with egfra+/− ovaries (padj < 0.05) (Figure 8C ; Supplementary Table S2).

GO analysis revealed that genes involved in transcription regulation, oxidoreductase activity, cytomembrane transport, oxidative phosphorylation, cell adhesion, and hydrolase activity were downregulated in egfra−/− ovary (Figures 8D–I ), whereas those in extracellular space and immune response were upregulated (Figures 8J, K).

Interestingly, several genes of great importance in gonadal development and function showed significant changes in expression. In particular, cyp19a1a (ovarian aromatase) (Figure 8E) and ar (androgen receptor) (Supplementary Table S2) were significantly downregulated, whereas amh (anti-Mullerian hormone) (Figure 8J and Supplementary Table S2), a male-promoting gene, was upregulated in egfra−/− follicles. Aromatase is responsible for producing estrogens, and androgen receptor-mediated signaling plays an important role in stimulating follicle development, including promoting granulosa cell proliferation (Laird et al., 2017; Franks and Hardy, 2018). In addition, several vitellogenin genes were found to express in control ovaries (egfra+/−), including vtg1, vtg2, vtg4, vtg5, vtg6, and vtg7; however, none of these genes showed expression in mutant ovaries (egfra−/−) (Figure 8L).

To investigate the fate of those follicles arrested at PG stage in egfra−/− females, we continued to monitor the state of ovaries over 1 year at 6, 9, and 12 mpf (Figure 9). At 6 mpf, the ovaries of egfra+/− fish showed active folliculogenesis with all stages of follicles present; in contrast, the egfra−/− ovaries still remained translucent and their follicles remained arrested at PG stage. The cytoplasm of some mutant oocytes exhibited significant shrinkage generating large inter-follicular spaces, which were filled up by masses of stromal cells. At 9 mpf, the mutant ovaries continued to degenerate with much less follicles, and the stromal tissues continued to increase and spread, occupying most inter-follicular spaces. Interestingly, testicular tissues and spermatogenic cells started to arise among the stromal tissues of some females. At 12 mpf, the morphology of egfra-deficient gonads in females looked much like testis, which contained mostly spermatogenic cells undergoing active spermatogenesis with large amount of mature spermatozoa in the lumen. Remnants of follicles could still be seen but with amorphous structures undergoing disintegration. The ovaries of egfra+/− female remained normal with active folliculogenesis (Figure 9).

To further investigate the fertility of the sex-reversed males, we tested these fish by crossing them with WT females. The mutant egfra−/− females displaying feminine secondary sexual characteristics, such as protruding genital papilla and light anal fin, were selected out at 6 mpf and raised separately until 13 mpf when all these fish displayed masculine characteristics, including disappearance of genital papilla and golden coloration of the anal fin (Figure 10A). Dissection and histological examination showed well-developed testis with normal spermatogenesis but no oocytes (Figure 10B) These sexually reversed males were able to spawn naturally with WT females to produce viable offspring (Figure 10C). Genotyping showed that the egfra alleles of all the offspring were heterozygous as expected (Figure 10D).

Our previous studies proposed that EGF promotion of oocyte maturation in zebrafish was likely mediated by activins in follicles (Pang and Ge, 2002), and that EGF significantly stimulated expression of activin subunits (inhbaa, inhbab, and inhbb) (Pang and Ge, 2002; Wang and Ge, 2004a; Tse and Ge, 2010; Chung and Ge, 2012) but suppressed that of activin binding protein follistatin (fsta) (Wang and Ge, 2004a) in cultured follicle cells. Our recent study showed that the loss of inhibin (inha), an antagonist of activin, advanced follicle development by promoting follicle activation or PG–PV transition, leading to precocious puberty (Lu et al., 2020). This is opposite to the disruption of egfra, which blocked PG–PV transition as reported in this study. This raises an interesting question about potential interactions between EGFR signaling and activin-inhibin system in folliculogenesis. To address this issue, we created egfra and inha double mutant (egfra−/−; inha−/−).

Using egfra+/−;inha−/− females as the control, we examined follicle growth and development in the double mutant at different time points of folliculogenesis. At 45 dpf, the leading follicles in the control had entered PV stage with multiple layers of cortical alveoli, marking the beginning of the secondary growth phase. In contrast, most follicles in the double mutant (egfra−/−; inha−/−) were at PG stage with only a few entering early PV stage as shown by the presence of a small number of cortical alveoli. The control ovary (egfra+/−; inha−/−) entered vitellogenic growth at 60 dpf with leading follicles accumulating abundant yolk granules. In contrast, the follicles in the ovary of double mutant remained unchanged without any advancement. Meanwhile, a substantial amount of stromal cells started to accumulate in the inter-follicular spaces, and the amount continued to increase at later stages (75, 90, and 150 dpf). Interestingly, some stromal cells developed concentrically around the places occupied or left by oocytes, suggesting possible origin of these cells from the follicle cells. Also interestingly, starting from 75 dpf, the dorsal part of ovarian lamellae became gradually devoid of follicles and the space was filled with serous fluid. This space continued to expand through 150 dpf with follicles being mostly located on the ventral side. Again, the follicles were arrested at PG stage without any advancement to late PV stage. The serous fluid remained in the ovarian lamellae without leaking into the ovarian cavity, and the follicles appeared to undergo liquefactive necrosis (Figure 11).