The central nervous system acts as a transducer of stress-induced masculinization through corticotropin-releasing hormone B

As a general trend, the response of the neuroendocrine system to environmental stressors is elevation of hypothalamic corticotropin-releasing hormone (Crh). Crh in turn stimulates the secretion and release of adrenocorticotropic hormone (Acth) from the pituitary gland (Aguilera and Liu, 2012; Kovács, 2013), which regulates cortisol levels through the adrenal gland (Mommsen et al., 1999). This is known as the hypothalamic-pituitary-adrenal (HPA) axis or the hypothalamic-pituitary-interrenal (HPI) axis in tetrapods and fish, respectively. In the latter, two crh ohnologs, named crha and crhb, have been identified (Grone and Maruska, 2015). The expression of crha has been mainly observed in the retina (Grone and Maruska, 2015; Hosono et al., 2015), with weak expression in the brain (about 100 times less than in retina) (Hosono et al., 2015), in fish. By contrast, crhb is mainly expressed in the central nervous system (CNS), i.e. in the preoptic area, the hypothalamus and the caudal neurosecretory system. For this reason, it has been related to the control of Acth in the pituitary gland (Alderman and Bernier, 2009; Bernier et al., 2008; Carpenter et al., 2014; Chen and Fernald, 2008; Grone and Maruska, 2015).

The action of Crh in the pituitary is mediated by the binding and activation of two highly conserved membrane receptors (Crhr1 and Crhr2), which belong to class B of the G protein-coupled receptors (Lovejoy et al., 2014). Although in tetrapods, it has been reported that Crh has higher affinity for Crhr1 (Vaughan et al., 1995), in teleosts both Crh ligands have similar affinity for both Crh receptors (Hosono et al., 2015). Several studies in mammals have also demonstrated the ability of Crh receptor antagonists to block stress responses, such as anxiety or depression (Backström and Winberg, 2013; Grammatopoulos and Chrousos, 2002; Holsboer and Ising, 2008), placing Crh receptors at a crucial point in regulation of the HPA axis.

The molecular and morphological processes of masculinization by stress have been investigated at the local, gonadal level, in several species from nematodes (Chandler et al., 2012, 2008), fish (Hattori et al., 2007; Hayashi et al., 2010; Kitano et al., 2012) and amphibians (Nakamura, 2009), to reptiles (Ge et al., 2018; Mork et al., 2014; Yatsu et al., 2015), but the involvement of the brain in sex-reversal is still under scrutiny. In all these vertebrates, exposure to environmental stressors during early life has several implications in reproduction. For instance, when reptiles and fish are exposed to stress during the crucial period of gonadal differentiation, a strong bias in sex ratios can be induced (Capel, 2017; Fernandino et al., 2013). The downstream factors involved in stress-induced masculinization in fish are well known (Hattori et al., 2009; Hayashi et al., 2010; Mankiewicz et al., 2013; Ribas et al., 2017; Tsalafouta et al., 2014; Yamaguchi et al., 2010). These factors can act by three different mechanisms: (1) inhibition of estrogen synthesis (Kitano et al., 2012; Nozu and Nakamura, 2015), (2) elevation of androgen synthesis (Fernandino et al., 2012; Hattori et al., 2009), and (3) apoptosis or meiotic arrest of germ cells (Yamaguchi and Kitano, 2012; Yamamoto et al., 2013). However, the molecular processes and key players controlling cortisol increase, which regulates these three mechanisms, remain unexplored.

In this study, we provide clear evidence of the role of the CNS in regulation of the HPI axis, shedding light on the triggering mechanism of masculinization induced by environmental factors.

First, we examined the regulation of both crh paralogs under normal and masculinizing temperature. The mRNA levels of crha and crhb were analyzed in medaka embryos at stages 26, 33 and 37, incubated at control (24°C; CT) or high (32°C; HT) temperatures (Fig. 1). No differences were detected for crha between treatments, at any of the developmental stages examined (Fig. 1A). In contrast, we observed high transcript levels of crhb under HT conditions at stage 37, corresponding to the gonadal sex determination period (Fig. 1B). It is noteworthy that the expression levels of both crha and crhb were not affected by the sex genotype (XX versus XY) (Fig. S1).

Based on the upregulation of crhb at stage 37 in embryos incubated at HT, we analyzed the transcript abundance of other HPI-related genes, such as the Crh receptor genes crhr1 and crhr2; the three urocortins, urocortin [Ucn; also known as sauvagine (Svg) and urotensin 1 (Uts1)], urocortin 2 (Ucn2) and urocortin 3 (Ucn3) (Fig. 2A) (Hosono et al., 2017); and acth (Liu et al., 2003). The expression of both crhr1 and crhr2 (Fig. 2E,F), was upregulated at HT. No significant differences were observed in the transcript abundance of the other Crh-like genes, i.e. uts1, ucn2 and ucn3 (Fig. 2B-D), or acth (Fig. 2G), suggesting that these HPI axis-related genes are not regulated at the transcriptional level during exposure to thermal stress in early development.

To analyze the participation of the HPI axis in temperature-induced masculinization, we disrupted this axis through biallelic mutations (indels in F0 individuals created by injecting cas9 and sgRNA) of crhr1 or/and crhr2 using CRISPR/Cas9 technology. Biallelic mutations of both Crh receptors generated indels in the transmembrane domain resulting in a receptor with a protein segment that fails to anchor into the membrane lipid bilayer, and is therefore unable to activate the intracellular G protein (Grammatopoulos, 2012). Thus, the sgRNAs for crhr1 and crhr2 genes were designed at exons 7 (located in transmembrane helix 3) and 10 (located in transmembrane helix 6; Fig. S2A), respectively. These sgRNAs were synthesized in vitro, and co-injected with nCas9n RNA (cas9) into one-cell-stage embryos. The mutagenesis efficiency for each sgRNA was analyzed by a heteroduplex mobility assay (HMA; Fig. S2B) (Ota et al., 2013), which reached 96.6% (29/30; embryos with biallelic mutations/total of eggs injected) for sgRNA-crhr1 and 100% (30/30) for sgRNA-crhr2 (Fig. S3A,B). Also, biallelic mutants were mated with wild-type fish to generate F1 and then each Crh receptor was sequenced to confirm presence of the indels (Fig. S2C). Data indicate that most cells contained biallelic indels and, consequently, loss of function in crhr1 and crhr2 mutants. None of the embryos analyzed presented indels at the off-target sites for each of the injected sgRNAs of biallelic mutants (Fig. S3A,B).

Furthermore, no alterations in morphology or survival rate were observed in a batch of biallelic mutant animals reared at 24°C (CT) up to 60 days post-hatching (dph) (Fig. S4).

In order to assess the participation of Crh-related genes in the sex reversal of genotypic females to phenotypic males induced by HT, we analyzed the expression of well-known gene markers for gonadal sex differentiation in fish, such as gsdf, sry-box9 type 2α (sox9a2), gonadal aromatase (cyp19a1a, the gene encoding the aromatase enzyme involved in the synthesis of estrogen) and hydroxysteroid 11-beta dehydrogenase 2 (hsd11b2, the gene encoding the 11 beta-hydroxysteroid dehydrogenase, 11β-HSD, which catalyzes the interconversion of cortisol and cortisone, and the synthesis of 11-oxygenated androgens) (Chakraborty et al., 2016; Fernandino et al., 2012; Imai et al., 2015; Kurokawa et al., 2007; Nakamura et al., 2012; Shibata et al., 2010; Zhang et al., 2016; Zhou et al., 2016). Fertilized eggs were co-injected with cas9 RNA and sgRNA for each Crh receptor (cas9+sgRNA-crhr1 or cas9+sgRNA-crhr2) alone or together (cas9+sgRNA-crhr1+-crhr2) and they were then incubated at HT (32°C). Control fertilized eggs were injected only with cas9 and then incubated at CT and HT (cas9-24°C and cas9-32°C, respectively; Fig. 3A). In all treatments, genotypic females (XX) that presented indels were selected for analysis of gene expression at stage 37. As expected, cas9-32°C individuals presented higher levels of gsdf and sox9a2, and lower cyp19a1a expression levels compared with cas9-24°C individuals (Fig. 3B-D), evidencing the molecular mechanism of action of masculinization induced by HT. However, the double biallelic Crh receptor mutants of genotypic females at HT showed a female pattern of lower gsdf and sox9a2 and higher cyp19a1a expression levels, resembling those of the cas9-24°C group (Fig. 3B-D).

When each biallelic Crh receptor mutant of XX embryos was analyzed, the gene expression pattern showed an intermediate phenotype, with high gsdf, sox9a2 and cyp19a1a (Fig. 3B-D). Here, it is necessary to take into account that in the biallelic mutant of each Crh receptor, the crhr paralog is fully active. We also analyzed the expression pattern of the androgen-related gene hsd11b2, which did not show differences between treatments (Fig. 3F).

Besides the expression of testis and ovary-related gene markers, we also analyzed gonadal morphology of XX biallelic Crh receptor mutants that were incubated at CT and HT until hatching, and thereafter at 26°C (breeding temperature) for 20 dph, when the gonad could be morphologically well differentiated. Firstly, we analyzed the impact of biallelic mutation on the sex reversal of XX individuals incubated at 24°C injected with cas9, biallelic crhr1, crhr2 and double Crh receptor mutants, in which no sex reversal individuals were found (Fig. 4A,B). In XX juveniles injected with cas9 (control) and incubated at HT until hatching, 68.8% exhibited sex reversal, as evidenced by testis morphology (Fig. 4A,C). The double biallelic Crh receptor mutants showed a wide-ranging insensitivity to HT-induced female-to-male sex reversal, with nearly all XX individuals presenting normal ovary morphology (Fig. 4A,B). Moreover, 19% of the biallelic crhr1 mutants exhibited sex reversal (Fig. 4A,C). However, 9.25% were presumptive intersex individuals, i.e. animals with ovaries containing spermatocytes (ova-testis; Fig. 4D). Finally, in XX biallelic crhr2 mutant juveniles 35% presented sex reversal and 10% had intersex gonads (Fig. 4A,C,D).

As we previously did not observe a correlation between the upregulation of crhb and the abundance of acth transcripts (Figs 1 and 2), we measured Acth-immunoreactive (Acth-ir) cells using immunofluorescence in the pituitary of genotypic female embryos at stage 39, with or without functional receptors, incubated at CT or HT (Fig. 5A). Firstly, we observed differences in the fluorescence intensity of the Acth-ir cells in XX embryos incubated at control and high temperature (Fig. 5B,C,H), suggesting that thermal stress induces Acth release. Moreover, we measured Acth-ir in biallelic Crh receptor mutants and observed higher fluorescence intensity compared with cas9 embryos incubated at HT (Fig. 5C-F, Fig. S5), resembling the XX cas9 control (24°C) embryos (Fig. 5B, Fig. S5). These results show that the biallelic mutation of Crh receptors in XX embryos causes accumulation of Acth in pituitary cells, indicating that both Crh receptors are involved in Acth release in the stress response induced by high temperature.

To corroborate that biallelic mutations of Crh receptors do disrupt the HPI axis, the level of cortisol and the mRNA expression of P450 11-beta (cyp11b), an enzyme expressed by the interrenal gland and involved in cortisol synthesis (Montero et al., 2015), were measured in all treatment groups. We observed in both biallelic crhr1 and crhr2 mutants an increase of cortisol levels at the end of the gonadal sex determination period, whereas the levels of cortisol in the double biallelic Crh receptor mutants were completely suppressed (Fig. 5H). Additionally, cyp11b was upregulated at HT but downregulated in the double biallelic Crh receptor mutants to low transcriptional levels similar to those observed at the control temperature (Fig. 3E), showing that the pathway involved in the synthesis of cortisol is transcriptionally active.

In view of the fact that the entire HPI axis seems to be functional during the crucial period of gonadal fate and the biallelic mutations in Crh receptors inhibited masculinization of genotypic females incubated at HT, we decided to test whether the addition of cortisol could rescue the absence of sex reversal in the mutants. Therefore, we performed an experiment in which all embryos were maintained in an embryo medium with or without cortisol (5 µM) from fertilization to 5 dph (Fig. 6A) (Hayashi et al., 2010). The double biallelic Crh receptor mutants exhibited transcription consistent with the phenotype of XX at HT, with a low transcript abundance of gsdf (Fig. 6B), i.e. a typical XX-24°C gsdf expression pattern.

Finally, XX biallelic crhr1 mutants, treated with or without cortisol at HT, presented high levels of gsdf, similar to those of control XX cas9-injected larvae (Fig. 6B), suggesting that the XX biallelic crhr1 mutation is not sufficient to induce a female (low) pattern of gsdf. These results are in agreement with the high level of cortisol observed at stage 39 (Fig. 5H). However, XX biallelic crhr2 mutant larvae reared at HT maintained low transcript abundance of gsdf (Fig. 6B), a typical female-like expression pattern. Most importantly, XX biallelic crhr2 mutants reared with 5 µM cortisol at HT showed a male-like (high) gsdf expression pattern, similar to that of XX cas9-injected XX fish (Fig. 6B). To better understand the compensatory molecular mechanism, we analyzed the transcript abundance of crhr2 and crhr1 in the XX biallelic crhr1 and crhr2 mutants, respectively. We observed upregulation of crhr2 in the XX biallelic crhr1 mutants (Fig. 6C), but not of crhr1 in XX biallelic crhr2 mutant larvae (Fig. 6D), suggesting a molecular compensatory mechanism.

Environmental factors that act during the crucial period of fish gonadal development are able to alter sex ratios, especially toward males (Fernandino et al., 2013; Ospina-Álvarez and Piferrer, 2008). Despite established genotypic sex-determining mechanisms with known sex-determining genes, many fish species produce male-skewed sex ratios when environmental temperatures are elevated during early development. However, whether this phenomenon has any adaptive value or not is unknown for the vast majority of species. Although understanding of this mechanism is of great interest for basic biology and from the perspective of global climate change, the pathways that mediate environmental cues and gonadal fate, and the involvement of extra-gonadal organs in this process, are still unknown. Our results demonstrate for the first time the fundamental role of the CNS as the transducer in a form of environmental sex determination (ESD), through regulation of the HPI axis.

In the current work we demonstrated that, during the gonadal sex determination period, the HPI axis is active. Moreover, we proved that out of two crh paralogs, only crhb was upregulated at high masculinizing temperatures along embryonic development. In medaka, the crha gene was previously misidentified as a new member of Crh family and named as telocortin (tcn). The expression of crha has been mainly observed in the retina, with a weak expression in the brain (Hosono et al., 2015). In another teleost, Astatotilapia burtoni, the presence of crha has been related to the mediation of social information or stress responses in the visual system, facilitating signal processing before it even reaches the brain (Grone and Maruska, 2015). These previous results are in concordance with our observation that crha transcription does not seem to be induced by environmental stressors, such as high temperature. Besides crha and crhb, other members of the Crh gene family are present in the medaka genome, such as the uts1, ucn2 and ucn3 (Hosono et al., 2017). Known as urocortins, these genes code for neuropeptides that share structural similarity with Crh proteins and can act as additional endogenous ligands for Crh receptors. In mice, they have been involved in stress responses and also anxiety (Bale and Vale, 2004; Sztainberg and Chen, 2012). Nevertheless, none of the urocortins was upregulated during the sex determination period at high, masculinization temperature.

The high expression of crhb we observed at HT is in agreement with results of other well-known stress responses, including its role in regulating the release of glucocorticoids (Alderman and Bernier, 2009; Carpenter et al., 2014; Chen and Fernald, 2008; Grone and Maruska, 2015). Moreover, a similar pattern was obtained for the Crh receptors (crhr1 and crhr2), which are crucial for activation of the HPI axis (Lovejoy and de Lannoy, 2013) during the gonadal sex determination period. In medaka, the first peak of cortisol occurs in 2 dph larvae, when animals are reared at normal breeding temperatures (Trayer et al., 2013). However, Hayashi et al. (2010) and our study showed an early rise in cortisol in embryos reared at HT, at the time of the gonadal sex determination period, evidencing an earlier activation of mechanisms involved in the surge of cortisol levels. In this regard, the high expression of crhb and the receptors crhr1 and crhr2 in our study is consistent with the timing of cortisol increase. Notably, overlap between the timing of early activation of the HPI axis and the gonadal sex determination period is crucial for understanding how high levels of cortisol are triggered and are related to male-skewed sex ratio (Hattori et al., 2009; Hayashi et al., 2010). Subsequently, in order to validate our hypothesis, we disrupted the HPI axis with biallelic mutation of both Crh receptors. These mutants were characterized by a lack of cortisol response at HT, downregulation of testicular gene markers, and concomitant inhibition of sex reversal (masculinization in XX) induced by stress. Thus, double biallelic Crh receptor mutants phenocopied the previous results on the inhibition of cortisol synthesis, with the absence of sex reversal in genotypic females (Hayashi et al., 2010). These observations demonstrate for the first time participation of the brain in stress-induced masculinization.

In all vertebrates, Crh regulates the synthesis and release of Acth (Mommsen et al., 1999; Wendelaar Bonga, 1997) through their transmembrane receptors in the pituitary gland (Lovejoy et al., 2014). In the present work, although acth transcript abundance did not show any change during the gonadal sex determination period and under stress conditions, we detected low intensities of Acth-ir in HT embryos, which could be explained by a stimulation of Acth release, or the lack of peptide synthesis (Aguilera and Liu, 2012; Kovács, 2013), induced by thermal stress. Moreover, biallelic mutations of both Crh receptors resulted in more Acth in the pituitary, phenocopying the high fluorescence pattern of the control group. Furthermore, this Acth accumulation and/or the lack of its release are in concordance with the complete suppression of cortisol levels observed in the loss-of-function Crh receptor mutants. Therefore, the strong decrease in cortisol level of the double Crh receptor mutants resembled the absence of a stress response observed without an environmental stressor, with the concomitant absence of female sex reversal. In addition, the double Crh receptor mutant phenocopied previous results observed after inhibition of cortisol synthesis in medaka (Hayashi et al., 2010), with the absence of sex reversal. Such disruption of the HPI axis was demonstrated to be crucial for female-to-male sex reversal in our studies with medaka under high, stressful temperatures.

An in-depth analysis of the molecular responses to loss of function of each Crh receptor under thermal stress showed that, whereas embryos co-injected with cas9+sgRNA-crhr1 or crhr2 presented an early inhibition of gsdf expression, high Acth-ir intensity and body cortisol level were observed at the end of the gonadal sex determination period when only one of the Crh receptor genes was biallelically mutated. Initially, it is necessary to take into account that in each biallelic Crh receptor mutant the paralog is fully active, which would generate a late compensatory effect, explaining the late high levels of cortisol at the end of the gonadal sex determination period, with the concomitant partial rescue of the sex reversal. Nevertheless, the regulatory mechanism of each biallelic Crh receptor mutant differs. In the biallelic crhr1 mutants, a transcriptional compensatory effect was observed with the upregulation of crhr2, which could establish the high Acth-ir intensity and cortisol level. However, in the biallelic crhr2 mutants, the transcription of crhr1 did not change. Therefore, the high Acth-ir intensity in these crhr1 biallelic mutants could be explained by a late inhibition of Acth release, leading to a high level of cortisol. In mammals, both Crh and Crhr1 are associated with the HPA axis at the initial stress response, whereas Crhr2 plays a major role during chronic and later responses to stress (Lovejoy and de Lannoy, 2013). Crhr1 knockout mice showed reduced stress-induced release of Acth and corticosterone, providing evidence that Crhr1 mediates stress-induced hormone activation (Smith et al., 1998; Timpl et al., 1998). On the other hand, Crhr2-deficient mice exhibited hypersensitivity of the HPA axis to stress (Bale et al., 2000), presenting a later decrease of plasma Acth and an increase of corticosterone, suggesting that Crhr2 is also involved in the maintenance of HPA axis drive (Coste et al., 2000). In view of these considerations, our data showing the loss of crhr2 function provide a similar regulatory mechanism to that reported in mice, resulting in a late decrease of Acth release, but with a high level of cortisol. Nevertheless, the loss of crhr1 function in medaka presented a novel molecular compensatory mechanism: upregulation of crhr2, promoting late synthesis of Acth, with a concomitant high level of cortisol. Taken together, these results highlight the importance of the involvement of both Crh receptors in fish masculinization induced by environmental stressors.

Once the HPI axis has translated the stimulus of an environmental stressor, is important to know how cortisol transduces this response to masculinize the gonad. In some fish, including medaka, it has been proposed that gonadal aromatase, an enzyme involved in estradiol synthesis, or other genes related to its regulation, such as FTZ-F1 (the ortholog of mammalian steroidogenic factor1), are inhibited by cortisol (Hayashi et al., 2010; Navarro-Martín et al., 2011; Yamaguchi et al., 2010). Furthermore, in pejerrey (Odontesthes bonariensis) it has been suggested that androgens, synthesized through the action of hsd11b2, are considered as mediators of stress (Fernandino et al., 2012, 2013). Our results confirm that cyp19a1a transcription is suppressed at HT and demonstrate that high transcription levels can be rescued in double biallelic Crh receptor mutants.

Three different results – (1) disruption of the HPI axis, (2) the increase of testicular gene markers with the concomitant decrease of sex reversal of genotypic females, and (3) the rescue of masculinization with cortisol – support the fact that the CNS is involved in sex reversal induced by environmental stressors (as summarized in Fig. 6E), in contrast to genotypic sex determination in which the sexual fate decision begins from the gonad.

Fertilized eggs of Oryzias latipes were incubated in 70 mm Petri dishes with embryo medium (17 mM NaCl, 0.4 mM KCl, 0.27 mM CaCl_2·2H₂O and 0.66 mM MgSO₄; pH 7) at 24°C (CT) or 32°C (HT). Sampling was performed at stages 26, 33, 37 and 39, and at 5, 20 or 60 days post-hatching (dph) (Iwamatsu, 2004). These stages correspond to the end of primordial germ cell (PGC) migration and the formation of the gonadal primordium (stage 26), the beginning of dmy/dmrt1bY transcription in gonadal somatic cells (stage 33), the sexual dimorphism in PGC proliferation (stage 35-37), and to the maximum PGCs proliferation in XX embryos and latest embryo stage of the gonadal sex determination period (stage 39) (Saito et al., 2007). Based on previous work, we know that 5-dph larvae are sensitive to cortisol treatment (Hayashi et al., 2010), that 20-dph fish can easily be assessed for gonadal sex morphology, and that 60-dph animals have grown as adult fish so can be used to assess survival. The strain hi-medaka (ID: MT835) was supplied from the National BioResource Project (NBRP) Medaka (www.shigen.nig.ac.jp/medaka/). All fish were maintained and fed following standard protocols to medaka (Kinoshita et al., 2012). Fish were handled in accordance with the Universities Federation for Animal Welfare Handbook on the Care and Management of Laboratory Animals (www.ufaw.org.uk) and internal institutional regulations.

Total RNA was extracted from individual embryos using the RNAqueous-Micro kit (Ambion by Life Technologies) for stage 26, the Illustra RNAspin Mini for stage 33, and 350 µl of TRIzol Reagent (Life Technologies) for stages 37, 39 and 5 dph, following the manufacturer's instructions. To perform cDNA synthesis, RNA of each individual sample (250 ng) was treated with Deoxyribonuclease I Amplification Grade (Life Technologies) and reverse-transcribed using SuperScript II (Life Technologies) with oligo(dT) following the manufacturer's instructions. Each primer pair was previously validated by analyzing the melting curve, requiring an efficiency of 95-105%, with a slope of around −3.30 and R2 value>0.99. Real-time PCR primers are listed in Table S1. Samples were analyzed with the Step One Plus Real-Time PCR System (Applied Biosystems). The amplification protocol consisted of an initial cycle of 1 min at 95°C, followed by 10 s at 95°C and 30 s at 60°C for a total of 45 cycles. The subsequent quantification method was performed using the 2^−ΔΔCt method (threshold cycle; assets.thermofisher.com/TFS-Assets/LSG/manuals/cms_040980.pdf) and normalized against reference gene values for ribosomal protein L7 (rpl7) (Zhang and Hu, 2007).

Each embryo of stages 33, 37 and 39, and each 5-dph and 20-dph larva was analyzed to determine its genotypic sex. Animals were subjected to DNA analysis for the presence of the dmy/dmrt1bY gene. For this purpose, we collected DNA from each RNA extraction following manufacturer's instructions. PCR analysis was then performed using primers for dmy (Nanda et al., 2002) and the presence of β-actin gene was used as a DNA loading control (Table S1) (Hattori et al., 2007). The PCR products were analyzed on a 1% agarose gel.

CRISPR/Cas9 target sites were designed using the CCTop-CRISPR/Cas9 target online predictor (crispr.cos.uni-heidelberg.de/index.html) (Stemmer et al., 2015), which identified the sequences 5′GG-(N18)-NGG3′ in exon 7 of crhr1 (TTGAGGAACATCATCCACTGG) and exon 10 of crhr2 (GAGGCAGCAAGACGAGTGTGG) (Fig. S2A,C). Each sgRNA was synthesized by cloning the annealed oligonucleotides into the sgRNA expression vector pDR274 (Addgene #42250) (Hwang et al., 2013) followed by in vitro transcription, previously established by Ansai and Kinoshita (2014). Briefly, a pair of oligonucleotides at final concentration of 10 mM each was annealed in 10 ml of annealing buffer (40 mM Tris-HCl pH 8.0, 20 mM MgCl₂ and 50 mM NaCl) by heating to 95°C for 2 min and then cooling slowly to 25°C. Then, the pDR274 vector was digested with BsaI-HF (New England Biolabs), and the annealed oligonucleotides were ligated into the pDR274 vector. The sgRNA expression vectors were digested by DraI, and sgRNAs were synthesized using the MEGAshortscript T7 Transcription Kit (Thermo Fisher Scientific). The synthesized sgRNAs were purified by RNeasy Mini kit purification (Qiagen). These RNA sequences were diluted to 50 ng/µl.

The capped cas9 (nCas9n RNA) was transcribed from pCS2-nCas9n plasmid (Addgene #47929). First, the plasmid was linearized by NotI and capped cas9 was synthesized by mMESSAGE mMACHINE SP6 kit (Life Technologies). The synthesized cas9 was purified by RNeasy Mini kit purification (Qiagen) and RNAs were diluted to 200 ng/µl.

Microinjection was performed into fertilized medaka eggs before the first cleavage as described previously (Kinoshita et al., 2000). For the CRISPR/Cas9 system, 25 ng/µl sgRNA and 100 ng/µl cas9 were co-injected in 4.6 nl of RNA mixture. Embryos injected with cas9 were used as controls. Microinjection was performed with a Nanoject II Auto-Nanoliter Injector (Drummond Scientific) coupled to a stereomicroscope (Olympus).

To analyze the efficiency and specificity of the CRISPR/Cas9 system, 3 days post-fertilization embryos were used (Ansai and Kinoshita, 2014). Genomic DNA was extracted by incubating each medaka embryo in 25 µl of 5 mM NaOH, 0.2 mM EDTA at 95°C for 5 min. After cooling to room temperature (RT) 25 µl of 40 mM Tris-HCl, pH 8.0, was added to the extract. The supernatant was used as template for PCR for HMA. Conventional PCR analysis was performed with genomic DNA using primers listed in Table S1. Electrophoresis was performed on 12% acrylamide gel (Ota et al., 2013); gels were stained with ethidium bromide for 15 min before examination. Multiple heteroduplex bands shown by HMA in PCR amplicons from each injected embryo were quantified as embryos with biallelic mutations, whereas single bands were quantified as no-edited embryos (Fig. S2B). The mutation rate was calculated as the ratio of the number of multiple heteroduplex bands shown in PCR amplicons from each Cas9-sgRNA-injected embryo to the sum of all embryos injected multiplied by 100 (n=30/per sgRNA) (Ansai and Kinoshita, 2014).

Potential off-target sites in the medaka genome for each sgRNAs were searched using a ‘Pattern Match’ tool in New Medaka Map (beta) at the NBRP medaka web site (http://viewer.shigen.info/medakavw/patternmatch) and CCTop-CRISPR/Cas9 target online predictor (Stemmer et al., 2015). All potential off-target sites identified were analyzed by HMA using the primers listed in Table S1.

Finally, the screening of indels was performed in F1 fish. Biallelic mutant adult (F0) medaka were mated with wild-type medaka of the Himedaka strain (WT). Genomic DNA was extracted from each F1 embryo for analysis of mutations by HMA, as described previously (Table S1). Mutant alleles in each embryo were determined by direct sequencing of the crhr1 or crhr2 gene region.

Samples for histological examination of gonadal sex (n=15-25/per group) were taken at 20 dph and analyzed following the criteria reported above. Firstly, the caudal fin was taken for gDNA extraction using conventional saline buffer extraction to determine genotypic sex and for HMA analysis (Aljanabi and Martinez, 1997). The body trunk was fixed in Bouin's solution and processed according to standard protocols for the preparation of Hematoxylin & Eosin-stained histological sections. These preparations were examined under a Nikon ECLIPSE Ni-U microscope (Nikon) and captured using a Digit Sight DS-Fi2 digital camera (Nikon).

Medaka embryos at stage 39 from different treatment groups were used. All individuals were processed under the same conditions for fixation, washing and incubation with serum and antibody. Stage 39 was chosen for analysis of the release of Acth upon upregulation of crhb, which was detected at stage 37. The tail was used for sex genotying by PCR and HMA analysis and the rest of the body was fixed in Bouin's solution overnight. Sections were then washed with 0.1 M phosphate-buffered saline (PBS pH 7.4) and blocked in 0.1 M PBS containing 0.5% bovine serum albumin (Sigma-Aldrich) for 60 min before overnight incubation with a mixture of primary antibody against ACTH-NIDDK-anti-hATCH-IC-3 (rabbit, 1:250; kindly provided by Dante Paz, Universidad de Buenos Aires) (Garrel et al., 2002) at RT. After incubation, the sections were washed twice in PBS for 10 min each and incubated at RT for 90 min with Alexa Fluor 488-conjugated secondary antibody goat anti-rabbit IgG (ThermoFisher Scientific, A-11008) at a dilution of 1:2000 in PBS. Separate sets of slides were treated only with secondary antibody (negative controls). After incubation, sections were rinsed twice with PBS and mounted with Fluoromount mounting medium (Sigma-Aldrich) containing 4′,6-diamidino-2-phenylindole (DAPI, 5 µg/ml, Life Technologies). Section photographs were taken using a Nikon Eclipse E7000 and Image Pro Plus (Media Cybernetics) with the same capture conditions of exposure and gain for all samples. Finally, images were analyzed and measured for fluorescence using ImageJ (https://imagej.nih.gov/ij/) using fluorescence intensity within the pituitary gland relative to mean background fluorescence.

Enzyme immunoassay (EIA) was performed using the Cortisol Express EIA Kit according to instructions from the manufacturer (Cayman Chemical) and previously used by our group (Fernandino et al., 2012). Briefly, pools of 23-25 embryos of each treatment and from both sexes were frozen at −80°C, homogenized in 0.2 ml of PBS, and used for steroid extraction with 1 ml of diethyl ether. This procedure was repeated two times. After evaporation of the diethyl ether, samples were immediately re-suspended in 2 ml EIA buffer and analyzed in a microplate reader (Rayto Model RT-2100C) following the kit instructions. The recovery rate was estimated by the cold-spike method to be 0.85% and the intra- and inter-assay variation (CV%) ranged from 4 to 13%.

Both cas9+sgRNA-crhr1 and/or +sgRNA-crhr2 co-injected fish were treated with 5 µM of cortisol (18) (11β-11,17,21-trihydroxypregn-4-ene-3,20-dione; Sigma-Aldrich) from fertilization to 5 dph. Briefly, after injection with a mixture of sgRNA (crhr1 and/or crhr2) and cas9, the embryos were placed in a 70 mm Petri dish with embryo rearing medium (25 ml) supplemented with cortisol or vehicle control (with the same volume of stock solvent: 4.53 μl ethanol, 0.018%). The medium was changed every day. Both groups were reared at HT. Finally, gDNA and RNA were extracted, as explained above, from each embryo for genotypic sex and HMA analysis, and RT-qPCR analysis, respectively.

All values are presented as mean±s.e.m. Fold change and statistical analysis of RT-qPCR quantifications were performed using FgStatistics interface (http://sites.google.com/site/fgStatistics/), based on the Relative Expression Software Tool (REST) from Pfaffl et al. (2002). Immunohistochemistry quantification was analyzed by χ²-distribution and statistical analyses were performed using SPSS v20 program, using one-way analysis of variance (ANOVA), followed by Tukey's multiple comparison test. Differences in sex ratio were determined by hypothesis test for differences in two population proportions, Z-test. All differences were considered statistically significant when P<0.05.

We thank Gabriela C. López (INTECH) for helping with histological and immunohistochemical preparations. We also thank Masato Kinoshita (Kyoto University) for teaching and helping with CRISPR/Cas9 technique, Adrián Mutto (Insituto de Investigaciones Biotecnológicas-UNSAM) for helping with microinjections, and Ricardo S. Hattori (Unidade de Pesquisa e Desenvolvimento de Campos do Jordão, APTA/SAA) for helpful advice. We are grateful to NBRP Medaka (https://shigen.nig.ac.jp/medaka/) for providing hi-medaka (Strain ID: MT835).