The experimental animals (A. davidianus Blanchard, 1871; Cryptobranchidae, Caudata) were collected from the salamander breeding base at Zhejiang Yongqiang Chinese Giant Salamander Co., Ltd., (Jinhua, Zhejiang Province, China). The salamanders were killed after anaesthesia by MS222 according to Yangtze River Fisheries Research Institute Care Committee (No. 2013001). According to previous study, we found that gonad of A. davidianus began to differentiate on the 98 days after fertilization (Hu et al., 2019a). The developing stage gonads of 62 dpf (days post fertilization), 98 dpf, 130 dpf and 1-, 2-, 3-, 4-, 5-, and 6-years-old (6 years old) giant salamanders in the normal group and sex hormone induction group were collected. Two samples of each tissue were collected, and one was stored in RNAstore Reagent for RNA extraction (Tiangen, DP408); the other part was preserved in 4% paraformaldehyde (pH 7.5) to prepare tissue sections; at the same time, the muscle tissue of each individual above was collected and preserved in absolute ethanol for DNA extraction using the TIANamp Genomic DNA Kit (Tiangen, Beijing, China) including RNase A treatment according the manufacturer’s instructions (Hu et al., 2019b). The testes, ovary, liver, spleen, lung, kidney, stomach, intestine, brain and pituitary tissues were collected from 2 years old A. davidianus, and put into RNAstore Reagent, and then transferred to a refrigerator at −80°C for storage until RNA extraction.

The larvae at 55 dpf were treated by estradiol or high temperature (28°C) respectively until 8 months post-fertilization to obtain the sex reversal individual according the previous study (Hu et al., 2019c). Using tissue sections and microscopes, the physiological sex of A. davidianus samples was identified, and, the genetic sex of used A. davidianus were identified using female-specific markers adf340 developed by our laboratory (Hu et al., 2019b). The PCR system consisted of 12.5 μl 2X Master Mix (TsingKe, TSE003), 100 ng genomic DNA, 0.5 µl of each primer (10 μmol/L), and double distilled water to reach a final volume of 25 μl. The reaction conditions were as follows: 95°C for 5 min; 33 cycles of 94°C for 15 s, 60°C for 15 s, 72°C for 30 s; and 72°C for 5 min. If the sex and genotype were consistent, there was no sex reversal. If the sex and genotype were different, namely, the sex was male, and the genotype was female, the individual was considered a sex reversal male (RM) individual; if the sex was female, and the genotype was male, the individual was considered a sex reversal female (RF) individual.

According to the partial sequences of the fgf9 and rspo1 genes in the giant salamander transcriptome database from our laboratory (Hu et al., 2019c), primers were designed to amplify the 5′ or 3′ UTR (Table 1). The 5′-RACE-Ready cDNA and 3′-RACE-Ready cDNA libraries were constructed according to the SMART™ RACE kit (Clontech, 634923) instructions as templates for RACE. PCR was performed as follows: five cycles of 94°C for 30 s and 72°C for 3 min; five cycles of 94°C for 30 s, 70°C for 30 s, and 72°C for 3 min; and then 27 cycles of 94°C for 30 s, 68°C for 30 s, and 72°C for 3 min. The amplified product was electrophoresed on 1% agarose and the purified fragment was cloned into the pMD18-T vector, transfected into E. coli strain TOP10 (Invitrogen, C404003), and then sequenced.

The sequence was searched in the GenBank database using BLAST (http://www.ncbi.nlm.nih.gov/BLAST/). The protein molecular weight was deduced using the website (http://www.bio-soft.net/sms/index.html). The conserved domain was predicted with online software (http://www.ncbi.nlm.nih.gov/structure). The protein domains of fgf9 and rspo1 were analyzed by the online software SMART (http://smart.embl-heidelberg.de/smart/set_mode.cgi). The phylogenetic tree of fgf9 and rspo1 was constructed based on protein sequence by MEGA 7.0 (Kumar et al., 2016) with the NJ method (neighbor-joining method). Bootstrap values were estimated from 1,000 replications.

To detect fgf9 and rspo1 expression in gonadal tissue, two pairs of primers were designed to obtain a riboprobe (Table 1). A 468 bp cDNA fragment was amplified from the fgf9 ORF domain, and a 533 bp cDNA fragment was amplified from rspo1. Primers were designed to introduce the T7 promoter sequence in the 5′- or 3′- side of the amplicons that will later be used as riboprobes, and PCR products were purified by a QIAquick Gel Extraction Kit (QIAGEN, Cat# 28706). To obtain probes, transcription was performed using the MEGAshortscript T7 High Yield Transcription Kit (Invitrogen, AM1354) according the manufacturer’s instructions In situ hybridization was performed following the method described in previous study (Hu et al., 2016). Sections were deparaffinized, hydrated and treated with proteinase K and then hybridized using sense or antisense DIG-labeled RNA probe at 70°C for 12 h. Hybridization signals were then detected with anti-DIG (Roche, 11093274910) conjugated with POD. DAB (Beyotime, P0202) was used as substrate for POD (Hu et al., 2016).

Ovary of female at 1 year old were dissected and washed with the PBS containing 1% Penicillin-Streptomycin (P/S) (Gibco, Thermo Fisher Scientific), then the gonads were cut into small pieces and placed on a 70 μm nylon mesh (Biologix,15-1070) and gently pushed through the mesh with constantly dripping of cold DMEM/F-12 medium (Gibco, 11320082) containing 10% FBS (BioInd, 18455), 1% heparin (Gibco, A16198.03) and Penicillin-Streptomycin (P/S) (Gibco, 15140122). The collected cells were washed using DMEM medium and then centrifuged at 500 × g for 10 min at 4°C. The supernatant was discarded and the cells were collected. The cells were divided into six groups and seed in a 6-well plate (Corning) with the density of 1 × 10⁶ cells/well. Three were set as control and the other three for FH535 treatment. The FH535 (MCE,108409-83-2) was added to the treated groups with the final concentration of 20 uM and incubate at 28°C for 3 days. The cells were collected and RNA was extracted for qRT-PCR.

According to fgf9 and rspo1 sequence, three siRNA at different sites were designed and synthesized for each gene (GenePharma, Shanghai, China, Table1). The siRNA marked by FAM was used as the negative control. The primordial gonadal cells were prepared according to the above description. The siRNA was transfected into the gonadal cells according to the manufacture’s instruction of Lipofectamine TM3000 (Thermo Fisher Scientific). Then the cells were incubated in opti-MEM medium (Gibco, Thermo Fisher Scientific) at 37°C for 6 h. After that, opti-MEM medium was removed and 2 ml of DMEM medium containing 10% FBS and P/S were added and incubated at 28°C for 48 h. The signal of green fluorescence was detected by OLYMPUS IX73. The RNA was extracted according the TRIzol method for qRT-PCR. To further detect the function of fgf9 and rspo1, the siRNA of fgf9 and rspo1 with the best inhibition effect were injection into the salamander after the sex was identified by the sex specific primers according to previous study (Hu et al., 2019b; 2019d). Sex related gene expression profile were detected in ovary and testis after rspo1/siRNA and fgf9/siRNA treatment for 72 h, respectively.

Total RNA was isolated from tissues using Trizol reagent (Invitrogen, 15596018). The cDNA was synthesized using PrimeScript™ RT reagent Kit following manufacturer’s instructions (Takara, RR047A). The expression of fgf9 and rspo1 genes in various tissues and developing gonad in A. davidianus was analyzed by quantitative real-time PCR (qRT-PCR). qRT-PCR was conducted using 2 × T5 Fast qPCR Mix (SYBR Green) (TsingKe, TSE202) following manufacturer’s instructions independently in triplicate on the QuantStudio 5 Real-Time PCR System (Applied Biosystems, California, United States). β-Actin was used as the internal control gene, cDNA was used as a template, tissue from three individuals was used, and each reaction was repeated three times. qRT-PCR was performed as follows: 95°C for 30 s; forty cycles of 95°C for 5 s, 60°C for 30 s, and 72°C for 30 s; and then 72°C for 5 min. The difference in expression was analyzed by one-way ANOVA followed by Duncan multiple comparison tests using SPSS 22.0 (IBM, New York, NY, United States). Significance was set at p < 0.05. Bars with different letters differ significantly (p < 0.05).

With RT-PCR and RACE amplification, full-length fgf9 and rspo1 cDNA were cloned from A. davidianus gonad at 1 year old (GenBank Accession No. MW685518 and MW685519). The full length of fgf9 was 1489 bp with a 5′UTR of 612 bp and a 3′UTR of 247 bp, and the full length of rspo1 was 1708 bp with a 5′UTR of 430 bp, an ORF of 786 bp, and a 3′UTR of 492 bp. The ORF of fgf9 encoded a protein of 209 aa, and that of rspo1 encoded a protein of 261 aa, with molecular weights of 23.35 and 28.76 kDa, respectively. A comparison of the selected amino acid sequences of amphibian and mammalian fgf9 showed that this gene is conserved with more than 85% identity (Supplementary Figure S1A). Based on the alignment of the amino acid sequence, a phylogenetic tree was constructed by the neighbour-joining method. The tree of fgf9 showed that A. davidianus is more similar to Cynops orientalis with 93% identity than to Xenopus laevis or Xenopus tropicalis (Supplementary Figures S1A,B). The phylogenetic tree of rspo1 showed that A. davidianus is more similar to C. orientalis with 85% identify than to Xenopus laevis or Xenopus tropicalis (Supplementary Figure S2B).

Genetic sex of the salamander were identified and salamanders without sex reversal were used to analyze the expression profile. Expression levels of the fgf9 and rspo1 genes in various tissues from 1 year old and developing gonad were detected using qRT-PCR. High expression levels of fgf9 were observed in the kidney and testis, followed by the liver, and relatively low expression levels were observed in the spleen and intestine; other tissues showed medium expression levels (Figure 1A). Expression pattern of fgf9 in developing gonad indicated that the expression of fgf9 in testes was significantly higher than that in ovaries (p < 0.05), and mRNA transcription gradually increased from 1 to 6 years of life (Figure 1B). Expression of rspo1 was observed in various tissues and significantly higher expression levels were observed in the ovary than that in other tissues (p < 0.05), medium expression levels were observed in the kidney, and low levels were observed in the remaining tested tissues (Figure 1C). In the developing gonad, rspo1 expression was significantly higher in the ovary than in the testis, with expression levels gradually increased from 1 to 5 years of life and decreasing at 6 years (Figure 1D).

To detect whether fgf9 or rspo1 expressed before gonad differentiation, In situ hybridization was conducted to identify the expression of fgf9 and rspo1 in the early developing gonad which was too small to detect by the qRT-PCR. At 62 dpf before the gonad differentiation, fgf9 and rspo1 positive signal were detected in the undifferentiated gonad of A. davidianus (Figures 2A1,A2,B1,B2). While the sense probe had no signal (Supplementary Figure S3). At 98 dpf when the gonad begin to differentiation, positive signal of fgf9 and rspo1 were also both detected in developing gonad (Figures 2A3,A4,B3,B4). At 130 dpf, expression of fgf9 and rspo1 were also detected in the gonad. When the gonadal differentiation is complete, we observed that fgf9 signal was strongly distributed in spermatogonia, spermatocyte and sertoli cells in testis while very weak signal in granular cell in ovary in the 2 years old A. davidianus (Figures 2A7,A8,A9,A10). For rspo1 gene, strong signal was detected in granular cell in ovary and weak signal in spermatogonia, spermatocyte and sertoli cells in testis (Figures 2B7,B8,B9,B10).

Expression level of fgf9 and rspo1 were detected in the gonad and sex-reversal gonad at 1 year old. Expression of fgf9 was significantly higher in testis than in ovary and highest expression was observed in sex-reversal testis (Figure 3A). Conversely, in the gonads and sex-reversal gonads, the expression level of rspo1 was significantly higher in sex-reversal ovaries than in ovaries, and the rspo1 expression level was significantly higher in ovaries and sex-reversal ovaries than the testes and sex-reversal testes (p < 0.05) (Figure 3B).

To provide further evidence of rspo1 in gonad development and potential antagonistic relationship of fgf9 and rspo1 genes, primary gonadal cell from 1 year old salamander ovary were collected and incubated with or without FH535. After the primary gonadal cells were treated with FH535, expression profile of sex related gene showed that cyp17, dmrt1, sf1, fgf9 expression were upregulated while rspo1, wnt4 and β-catenin expression was down-regulated comparing with the control without FH535 treatment (Figure 4). Additionally, we observed that cyp19a and frizzled2 expression profile was unchanged (Figure 4).

To examine the function of fgf9 and rspo1 genes in the gonad, three siRNA sites were designed according the sequence of fgf9 and rspo1, respectively. The control siRNA site marked by FAM was transfected into the gonadal cells and we found that high transfection efficiency was exhibited in ovary cell with 60% and low in testis cell with about 20% (Figure 5A). Expression level of fgf9 and rspo1 were significantly down-regulated in ovary and testis cells at site siRNA R684 and F890, respectively (Figures 5B,D). Seventy-two hours post siRNA R684 treatment in vitro and vivo, we found that expression of cyp17, foxl2, rspo1 and cyp19a were significantly down regulated while dmrt1 and ftz-f1 expression were not changed (Figures 5C,F). Seventy-two hours after siRNA F890 treatment in testis cells, expression of cyp17, fgf9, dmrt1 and ftz-f1 were significantly down regulated while cyp19a expression was up regulated (Figures 5E,G). Only foxl2 expression was not changed.