Growth status and sex ratios
The body weight and body length of mandarin fish during MT treatment were shown in Table 1. The results showed that the body weight and body length of the MT-treated group were significantly lower than those of control group (P < 0.05) from 20 dpf to 70 dpf. Among the MT treatments, the body weight and body length of M1 group were higher than M2 group with no significant difference (p > 0.05). The body weight and body length of post-treatment were summarized in Fig.1 and Fig.2. At 120 dpf, the body weight and body length of the MT treatments were significantly lower than those of control group (P<0.05). Nevertheless, no significant differences were found for the body weight and body length among the control group and MT treatments at 180 dpf and 240 dpf. There were also no significant differences in body weight and body length between the MT treatment groups at 120 dpf, 180 dpf and 240 dpf.
The survival rates of the MT treatments and control group were not significantly different (Fig.3). The MT-treated sex ratios were 100% males, while the control sex ratios were 51.11% males at 70dpf, and the ratio of females and males in the control group was about 1:1 (Table 2.).
Gonadal morphology and serum steroid hormone
At 70 dpf, the control males had testes which were populated with large volumes of spermatocytes and a few spermatogonia and spermatid,and the MT-treated testes were consisted of large volumes of spermatid and a few spermatocytes and spermatogonia (Fig. 4 a, b, c). At 120 dpf, a few spermatozoa were shown in the control testes (Fig. 4 d), the SZ were strongly basophilic whose nuclei were stained dark blue by hematoxylin. In the seminiferous lobules away from the vas deferens, most of them were filled with spermatid. The maturity of MT-treated testes was significantly higher than that of the control testes, which testes were filled with spermatozoa (Fig. 4 e, f). At 180 dpf, in the seminiferous lobules closed to the vas deferens of the control testes were populated with a few spermatozoa. Moreover, the lobular lumen was reduced and the lobular wall was thickened, and spermatozoa in the lobular lumen were significantly reduced or disappeared. The lobular lumen of MT-treated testes also showed signs of degeneration, but large volumes of spermatozoa were still found in the seminiferous lobules closed to the vas deferens (Fig. 4 h, i). At 240 dpf, the lobular lumen of the MT-treated testes and the control testes were reduced to almost disappear, the lobular wall was thickened, and the lobular interstitial was developed. (Fig. 4 j, k, l). In addition, no ovarian tissue or interstitial gonads were found in MT-treated mandarin fish detected by all tissue sections.
The change in GSI of mandarin fish in control and MT treated groups were shown in Fig. 5. The results showed that GSI of the control and MT treated groups increased gradually from the 70 dpf to 240 dpf, and the highest value were at 240 dpf. GSI of MT treatments were significantly lower than that of the control group (P<0.05) at 70dpf (termination of treatment), however, the value of the control group was significantly lower than that of the MT treatments at 120 dpf (P<0.05). At 180 dpf and 240 dpf, there was no significant difference in GSI between the control and MT treated groups. During the experiment, the difference in GSI of mandarin fish in different concentrations of MT treatment group were not significant.
The changes of serum steroid hormone (T and E2) in mandarin fish from 70 dpf to 240 dpf were shown in Fig. 6 and Fig.7. At termination of treatment (70dpf), the contents of T and E2 of the control group were significantly higher than those in the MT treatment group (P<0.05), and the difference in T and E2 of mandarin fish in different concentrations of MT treatment group were not significant. When the 17α-methyltestosterone treatment was completed, the levels of plasma T and E2 of the MT-treated groups returned to normal, and there was no significant difference with the control group.
Illumina sequencing and unigenes annotation
Gonad cDNA libraries of nine testes and nine ovaries were sequenced. A total of 120,955,742 (97.72%) clean reads were obtained and used for de novo assembly. 60,971,396 and 59,984,346 clean reads were obtained from ovaries and testes, respectively, with a mean sequence length of 150bp (Table 3). After assembling and clustering, a total of 63,632 unigenes ranging from 201 to 21,004 bp with a mean sequence length of 1,182 bp were acquired (Fig. 8).
A total of 33,535 (52.7%) unigenes were annotated from the four databases, among which 33,471 unigenes were annotated from Nr database, 27,685 unigenes were annotated from Swissprot database, 22,501 unigenes were annotated from COG/KOG database and 20,323 unigenes were annotated from KEGG database (Fig. 9).
Analysis of differentially expressed genes
After RPKM standardized calculation, 46,391 DEGs were detected in the gonads of 120 dpf, among which 3,435 genes were highly expressed in ovaries and 42,956 in testes. When RPKM≥2, 811 and 1,530 genes were differentially expressed in ovaries and testes, respectively.
57,171 DEGs were assigned to three GO term categories including biological processes (28,349), cellular components (16,155), and molecular functions (12,667) (Fig. 10). In biological processes, the DEGs of the ovaries and testes were mainly concentrated in cellular process (647 and 5,290 DEGs in ovaries and testes, respectively), metabolic process (547 and 3,850 DEGs in ovaries and testes, respectively) and single-organism process (532 and 4,719 DEGs in ovaries and testes, respectively). In cell components, the DEGs of the ovaries and testes were mainly concentrated in the cell (431 and 2,806 DEGs in ovaries and testes, respectively), cell part (431 and 2806 DEGs in ovaries and testes, respectively), membrane (256 and 2,610 DEGs in ovaries and testes, respectively), membrane part (218 and 2356 DEGs in ovaries and testes, respectively) and organelle (307 and 1,905 DEGs in ovaries and testes, respectively). In molecular functions, the DEGs of the ovaries and testes were mainly concentrated in binding (606 and 4,865 DEGs in ovaries and testes, respectively) and catalytic activity (423 and 3,352 DEGs in ovaries and testes, respectively).
According to KEGG classification of DEGs, 5,378 DEGs were mapped to 260 different pathways. Among these pathways, 36 pathways were the prominent pathways (Pvalue<0.05) which mainly consisted of Neuroactive ligand-receptor interaction (443 Unigene,8.24%), Calcium signaling pathway (409 Unigene,7.61%), and Cell adhesion molecules (CAMs) (250 Unigene,4.65%). Moreover, MAPK signaling pathway (418 Unigene, 7.77%) and Steroid hormone biosynthesis (42 Unigene,0.78%) were the pathways related to reproductive development (Table 4).
Identification of DEGs associated with sex development
By analyzing the transcriptome data of ovaries and testes, the expression levels of 7 genes related to sex determination and differentiation were listed in Table 5. We noticed that dmrt1, amh, sox9, gsdf and dax1 expression in the testes were higher than ovaries (P < 0.05), while foxl2 and cyp19a1a expression in the ovaries were higher than testes (P<0.05).
Real-time PCR verifications
At termination of treatment (70dpf), the relative expression of dmrt1, sox9, foxl2 and cyp19a1a in the gonad were shown in fig.11 and fig.12. The relative expression of dmrt1, sox9 in M0 (♂), M1 and M2 were significantly higher than that in M0 (♀), but the relative expression of foxl2 and cyp19a1a were significantly lower than that in M0 (♀) (p<0.05). In different concentrations of MT treatment groups and M0 (♂), the relative expression of dmrt1, sox9, foxl2 and cyp19a1a were not significantly different (p>0.05).