Direct production of XYDMY− sex reversal female medaka (Oryzias latipes) by embryo microinjection of TALENs

Medaka is an ideal model for sex determination and sex reversal, such as XY phenotypically female patients in humans. Here, we assembled improved TALENs targeting the DMY gene and generated XYDMY− mutants to investigate gonadal dysgenesis in medaka. DMY-TALENs resulted in indel mutations at the targeted loci (46.8%). DMY-nanos3UTR-TALENs induced mutations were passed through the germline to F1 generation with efficiencies of up to 91.7%. XYDMY− mutants developed into females, laid eggs, and stably passed the YDMY− chromosome to next generation. RNA-seq generated 157 million raw reads from WT male (WT_M_TE), WT female (WT_F_OV) and XYDMY− female medaka (TA_F_OV) gonad libraries. Differential expression analysis identified 144 up- and 293 down-regulated genes in TA_F_OV compared with WT_F_OV, 387 up- and 338 down-regulated genes in TA_F_OV compared with WT_M_TE. According to genes annotation and functional prediction, such as Wnt1 and PRCK, it revealed that incomplete ovarian function and reduced fertility of XYDMY− mutant is closely related to the wnt signaling pathway. Our results provided the transcriptional profiles of XYDMY− mutants, revealed the mechanism between sex reversal and DMY in medaka, and suggested that XYDMY− medaka was a novel mutant that is useful for investigating gonadal dysgenesis in phenotypic female patients with the 46, XY karyotype.


DMY-and DMY-nanos3UTR-TALENs effectively induced DMY gene disruption in Medaka.
To improve the germline integration efficiency, we incorporated the zebrafish nanos-3′ UTR into the TALEN construct (Fig. 1A). Potential TALEN target sites were scanned and designed in the exon of the DMY/DMRT1bY gene (ENSORLT000000025382 and ENSORLT000000025383) (Fig. 1B). We generated TALEN constructs using our previously published method 29 . The mixture containing a pair of TALEN mRNAs was microinjected into one-cell stage embryos of medaka ( Fig. 2A and File S1). 72 hours after microinjection, ten injected embryos were randomly pooled for extracting genomic DNA. As illustrated in Fig. 2B, primers DMY F and DMY R bridge both the effector binding element (EBE) regions, while primers DMY F1 and DMY R link the spacer region and the downstream EBE. If primer DMY F and DMY R generated a 396 bp fragment, while primer DMY F1 and DMY R failed to generate the 167 bp fragment, the result suggested that the targeted gene was disrupted by the TALENs (Fig. 2C). Sequenced PCR positive clones had mutated sequences in the spacer (Fig. 2D). Both DMY-TALENs and DMY-Nanos-3UTR TALENs were effective at disrupting the targeted genes in medaka embryos. Various concentrations (200, 400, 600 and 800 pg) of the TALENs mRNA were microinjected (Fig. 2E,F, File S1 and S2). The targeting efficiency of the DMY-TALENs was good, with a higher TALEN-induced mutation ratio of 42.07% and lower levels of dead and deformed embryos when 600 pg mRNA was microinjected (Fig. 2G,I). Similarly, the targeting efficiency of DMY-nanos-3UTR TALENs was about 41.96% (Fig. 2H,I). Thus, 600 pg was determined as the appropriate concentration in medaka. These results indicated that the TALEN activity was dose-dependent, and high-dose microinjection of TALEN mRNA might cause nonspecific and toxic defects in medaka embryos ( Fig. 2E-I, File S1 and S2).
Successful germline transmission is essential to establish knockout lines. To evaluate the germline transmission efficiency of the TALEN-mediated gene disruption, ten embryos for DMY and DMY-nanos3UTR from each independent cross were individually collected at 3 days post fertilization (dpf), and genomic DNA was extracted from each cross to assess mutagenesis at the TALEN-targeted site (Fig. 2C,D). 9.02% of DMY-TALEN-induced F1 embryos carried mutations; and 37.56% of DMY-nanos3UTR-TALEN-induced F1 embryos carried mutations in the DMY gene ( Fig. 2I and File S2). The higher proportion induced by DMY-nanos3UTR-TALEN indicated that a majority of gametes in the F0 medaka were mutant. These results indicated that there was no significant difference in the targeting efficiency of F0 somatic mutations at the targeted loci (Fig. 2H,I); however, DMY-nanos3UTR-TALEN induced a higher portion of mutations in the germline than DMY-TALENs did.
A novel program to identify potential off-target sites of TALENs. To test for potential nonspecific mutations induced by TALENs, we designed a program to scan the medaka genomic sequence (http://www.ensembl.org/Oryzias_latipes) to identify potential off-target sites potentially targeted by DMY TALENs. Potential off-target sites of DMY-TALENs were searched using the program and 55 candidate sites were identified (File S3). When the spacers were less than 10 bp or more than 24 bp long, the scaffold of TALENs had lower disrupting activity 33,34 . Five of the 55 candidates had spacers less than 10 bp, and 43 of 55 candidates had spacers of more than 24 bp, indicating that it is unlikely that the TALENs could induce mutations at these sites. We analyzed one candidate site (Chr. 3: 36,197,357-36,197,403) that had 7-bp mismatches in the recognition sequences and a 12-bp spacer (File S4). PCR amplified the identified potential off-target regions using genomic DNA from TALEN-injected embryos as template; no mutations were found at these sites by DNA sequencing. This result suggested that the novel program could predict the potential off-target sites of TALENs; and that TALENs have high specificity for their target sequences.
Mating scheme of TALENs-induced DMY-mutants and mutant phenotypes. The mating scheme for the TALENs-induced DMY-mutant lines is shown in Fig. 3A. The F0 generation was produced by microinjecting 600 pg DMY-nanos3UTR-TALENs into one-cell stage embryos of medaka; the mutation rate was 46.8% (15/32) (Fig. 3B). The DMY gene knockout medaka could develop into females. 11-bp deletions (named DMYΔ 11) and 16-bp insertions (named DMY+ 16) were identified and chosen to establish mutant lines (Fig. 3C,D). Notably, during the establishment of the mutant lines, two genotyping alleles of DMY gene fragments were identified in individual TALEN-induced F1 mutations using the Li-con 4300 system (Fig. 4A). This result indicated that Y DMY− Y male mutants were present in the testcross F1 generation. The genetic males (XY) of the DMY gene mutants, DMYΔ 11 and DMY+ 16 mutant types, developed into females in the F2 generation, which was identified using genomic PCR of the DMY gene (Fig. 4B). To confirm un-expression and dysfunction of DMY gene in the XY DMY− F2 generation, using RT-PCR, expression of DMY gene was not identified in the XY DMY− female medaka (both DMYΔ 11 and DMY+ 16) (Fig. 4C). The first morphological sex difference manifested in the gonads is reflected in the number of germ cells 14,35 . The number of germ cells in several DMY mutants identified from wild populations resembled that of the female 13,16,36 . To elucidate sex reversal during the development of XY DMY− mutants, we evaluated the effect of DMY on germ cell number at 5 days after hatching (DAH) in the XY DMY− F3 generation. The XY DMY− mutant fry had more germ cells than that of the WT XY male at 5 DAH (Fig. 4D). This implied sex reversal of XY DMY− mutants took place in early developmental stages, and the increased number of germ cells in the XY DMY− mutants may be due to the disruption of DMY gene expression.
According to the amino acid sequence of DMY, DMYΔ 11 and DMY+ 16 are frame-shifted mutant alleles that would produce truncated DMY protein caused by a region of altered translation (Fig. 4E). Thus, an error of the coding sequence has occurred in the mutated DMY gene that resulted in the loss function of the DMY gene. The XY DMY− female or Y DMY− Y male (N = 10) from F2 generation that had lost the DMY gene were crossed with the WT to obtain the testcross F3 generation. XX female, XY male, XY DMY− mutant female and Y DMY− Y mutant male were identified in the XY female testcross F3 generation. XY males and XY DMY− mutant females were identified in the Y DMY− Y male testcross F3 generation. This showed that the DMY gene in the Y chromosome of WT medaka rescued the female phenotype of DMY gene disruption in mutated XY female medaka. Unfortunately, an Y DMY− Y DMY− mutant female with a genomic homozygous DMY gene mutation on the Y chromosome was never found.
Mature XY DMY− mutants in the F3 generation were obtained for phenotype identification, histological analyses and fluorescence in situ hybridization (FISH) (Fig. 5). There are significant differences between females and males in the size and shape of dorsal and anal fins 13,14 , which are main part of the secondary morphological sexual characteristics in medaka. The shape of both dorsal and anal fins of XY DMY− sex reversal female were similar to that of the WT XX female, rather than to that of the WT XY male. The size of both dorsal and anal fins of XY DMY− female was significant smaller than those of the WT XY males (Fig. 5A). In addition, ovarian tissue was identified in XY DMY− mutants (Fig. 5B). To confirm the Y chromosome in the TALENs-induced mutants, we analyzed the karyotypes of metaphase cells from the WT male, WT female and XY DMY− female using FISH showing the male specific hybridization signal 12 . Compared with the two spots in females, three hybridization spots for the specific probe were identified in males (Fig. 5C). The additional FISH signal in males is on the Y chromosome. XY DMY− mutants did not express DMY gene, which were different from WT XY individuals (Fig. 4C). The gonadosomatic index (GSI) and the maturation stages of oocytes are commonly used to evaluate the gonad and gonad development 37 . At 90 days after hatching, the GSI of XY DMY− mutated female was 10.9 (N = 10); that of the WT female and WT male were 19.1 and 0.7, respectively (Fig. 5D). During the generation of offspring from the XY DMY− mutated female testcross, we found that it was difficult to obtain a sufficient number of offspring. Therefore, we performed the comparative analysis of the number of matured oocytes between the WT female and XY DMY− mutants (N = 10). During 10 days' embryos collecting, the XY DMY− mutated female produced six eggs per day on average; whereas, the WT XX female produced 20 eggs per day (Fig. 5E). The results from F2 phenotypes revealed that the genetic males (XY) of TALENs-induced DMY gene disruption mutants all developed into females and laid eggs (DMYΔ 11 and DMY+ 16 mutants). Furthermore, histological analyses demonstrated that all XY mutants developed into females; however, there were significant differences in germ cells and gonad (Fig. 5F). The ovary of XY DMY− mutants (12/15) appeared to have fewer oocytes than that of the WT XX female (Fig. 5F). According to the developmental stage of oocytes in medaka, there were equilibrium distributions in each stage of oocytes in the WT XX female ovary, containing cortical alveolar oocytes (CA), vitellogenic oocytes (VO) and mature oocytes (MO). Among 15 XY DMY− mutants (15 mutants were identified in the same genotype, fed in the same tank and sampled at the same time), three did not form mature oocytes, with majority of chromatin nucleolar oocytes (CN), several perinucleolar oocytes (PO) and few CA oocytes in the ovary, and no further developmental stage oocytes (Fig. 5F). Twelve of them successfully formed MOs, however, the size of ovary and total number of each developmental stage oocytes were significant smaller and fewer than that of WT XX female (Fig. 5F). This can also explain it is difficult to collect embryos in XY DMY− mutants, even there is sometimes no embryos could be collection (Fig. 5E). The analysis of gonadal histology, GSI, and the statistics of mature eggs demonstrated significant differences in the gonadal development and maturity between WT females and XY DMY− mutant females.  Transcriptomic analysis of TALENs-induced DMY-mutants and WT. To better explore the transcriptional regulation function of DMY, and identify DMY-related downstream factors that affect the generation, development and maturation of testis or ovary, we used RNA-seq to analyze the transcriptome of WT_M_TE, WT_F_OV and TA_F_OV after 90 days. For both the WT and DMY-mutants, ten samples of gonadal tissues were mixed for library construction. RNA-seq generated 157 million raw reads comprising 10333, 8746 and 8621 transcripts, respectively (Fig. 6A). Using the Ensembl medaka genome database as the reference, 75.8% of the raw reads matched medaka genomic sequences (File S6). Twelve genes were selected randomly from hundreds of different expression transcripts between the gonad of WT and that of DMY mutants. The result of real-time quantitative PCR showed that the trends of these genes expression were the same as in the RNA-seq data (File S7). Thus, the RNA-seq information was accurate and reliable.
Theoretically, a number of differentially expressed genes between the testis (WT_M_TE) and ovary (WT_F_OV) (Data B) might confound Data A. More accurately, Data B must be excluded from Data A. The intersection of Data A and Data B identified the real differentially expressed genes between the WT XY male testis and XY DMY− female ovary (Data C, Dataset 2), which were the sum of 309 transcripts, 276 transcripts, 144 transcripts, 293 transcripts, 62 transcripts and 78 transcripts, which are DMY -related or affected genes (Fig. 6F). Relative to WT_F_OV, 144 ovary-specific transcripts were downregulated and 293 ovary-specific transcripts were upregulated in male-to-female reversed gonads (Fig. 6F). 62 transcripts in XY DMY− female ovary were downregulated relative to WT_M_TE and were upregulated relative to WT_F_OV. 78 transcripts in XY DMY− female ovary were upregulated relative to WT_M_TE and were downregulated relative to WT_F_OV. Relative to WT_F_OV and WT_M_TE, 309 transcripts were upregulated in TA_F_OV, and 276 transcripts were downregulated in TA_F_OV ( Fig. 6F and Dataset 2). Seventy-three testis-specific transcripts were identified ( Fig. 6E and File S8). GO analysis and homologous annotation with human genes is a traditional way to further predict the functions of the genes in Data C (Dataset 2). According to the annotation of Blast2GO, genes of Data C were associated with the regulation of ubiquitination and fertilization. Moreover, a large number of genes, such as ENSORLT00000000529, have not been previously reported.
GO analysis is not a straightforward way to predict the relationships among genes. Sry and DMY are transcription factors; theoretically, their downstream genes should have the direct binding regions for Sry or DMY. The predicted binding site of human SRY is shown in Fig. 6G. Human homologous genes of data C were scanned for SRY binding sites (Dataset 3). If SRY binding sites could be found in the human homologous genes, we could speculate that their medaka homologous genes may have the DMY binding sites or be a directly affected gene. There were 9844 unique genes in medaka that were analyzed: 7440 of which had homologous genes in the human genome. There were 4644 human homologous transcripts that may have more than one potential SRY binding site ( Fig. 6H and Dataset 3). A number of potential DMY regulated genes, such as SLC25A38, had a potential SRY binding site in the upstream region of its human homolog; and a novel transcript, ENSORLT00000000529, had six potential SRY binding sites in the upstream region of its human homolog. These genes are significant for investigating the transcriptional function of the DMY gene.
When we compared TA_F_OV with WT_F_OV, 1163 differentially expressed transcripts (Data D) were found in XY DMY− female (Dataset 4). There were 515 upregulated genes and 647 downregulated genes in WT-F-Ov (Fig. 6B,D). Using Blast2GO, the differentially expressed genes were blasted and annotated on biological process (BP), cell components (CC), and molecular function (MF). The genes of Data D are associated with the wnt receptor signaling pathway (Predicted: syntabbulin-like, axin-2-like), ovarian follicle development (Forkhead box O5, Predicted: beta-arrestin-1-like, adenomatous polyposis coli protein-like, ubiquitin-protein ligase E3A-like, bone morphogenetic protein receptor type-1B-like), and the follicle-stimulating hormone signaling pathway (luteinizing hormone receptor,  lhr, Predicted: beta-arrestin-1-like) (Dataset 4). In terms of the number and repetitions of the genes, the genes of the wnt signaling pathway were the most significant proportion of Data D. PRCKA (CL423) and DKK (Unigene31972) were up-expressed in TA_F_OV; Wnt1 (CL10671), MAPK8 (CL4166), FZD6 (CL9539) and PRCK (CL13707) were down-expressed in TA_F_OV. These genes, especially Wnt1 and PRCK (CL13707), were down-expressed in XY DMY− mutants, compared with WT XX females (Dataset 4). Therefore, we reasonably believed that the wnt signaling pathway is one of the vital pathways on regulating the development and mature of oocytes in XY DMY− mutants, why the ovarian function of XY DMY− mutants are significant lower than that of WT females. Comprehensive Data A, B, C, and D, greatly promote research into the transcriptional function of the DMY gene in medaka, not only in sex reversal, but also in the normal development and maturation of gonads.

Discussion
Nanos3UTR-TALENs effectively generated a targeted gene mutant line in Medaka. Germline transmission of the knockout genotypes is critical to obtain homozygous gene knockout animals. Generally, a 46.8% mutation rate in the F0 is not ideal for gene disruption. Perhaps this low efficiency should be attributed to the DMY gene on the Y chromosome rather than our optimized TALENs. In general, there should be equal numbers of male and female embryos in a generation. This means that the targeting efficiency of DMY-TALENs could not beyond be greater than 50%. Indeed, in our results, the efficiency was never more than 50%, even if the concentration of TALENs was increased to 800pg (Fig. 2I). To improve the germline integration efficiency, we incorporated the zebrafish nanos-3′ UTR into the TALEN construct (Fig. 1A), which was reported to protect mRNA from degradation in primordial germ cells and improve the germ cell targeting efficiency 28,41,42 . Compared with 9.02% DMY-TALEN-induced F1 embryos carrying mutations; there were 37.56% DMY-nanos3UTR-TALEN-induced F1 embryos that carried mutations in the DMY gene ( Fig. 2I and File S2). These results indicated that DMY-nanos3UTR-TALEN induced a high proportion, possible a majority, of mutant gametes in the F0 medaka. Thus, the incorporation of the nanos-3′ UTR into the TALEN construct improved the germ cell targeting efficiency in medaka, permitting the generation of medaka knockout lines.
A novel program to identify potential off-target sites of TALENs. Specificity is essential to establish precisely targeted gene knockout lines. TALENs have become an accepted tool for targeted mutagenesis, but undesired off-targets, in addition to the targeted genomic region, remain an important issue 15,21,[43][44][45][46][47] . Unfortunately, using e-PCR to perform BLAST searches, potential off-target sites were identified in several studies 46 . Using Primer3 and BLAST, potential off-target sites of the DMY gene in medaka were identified (File S5). According to this data, BLAST emphasized sequence similarity, but there were few base pair sites of TALENs that could match to potential off-target sites of the genomes. In fact, not only the similarity of sequence, but also that fact that the 0 position of EBEs must be T is very important to TALENs binding site 48 . Using the novel program, 55 candidates' off-target sites were identified in the medaka genome. We amplified the top predicted potential off-target locus by PCR, sequenced it and found no mutations (File S4). Using our data, our program was more efficient than ePCR to predict TALENs off-target sites. Recently, tools for predicting TALEN off-targets have been developed, such as idTALE (http://idtale.kaust.edu.sa), Paired Target Finder (https://tale-nt.cac.cornell. edu), and TALENoffer 49 . Compared with these, the advantage of our program is its simplicity and reliability. However, a more detailed investigation on the possible off-target effects of TALENs and more accurate program will be needed in the future.

Mutant phenotypes. Gonadal dysgenesis in 46, XY patients was first noted by Drash et al. 4 .
Phenotypic female patient with an XY karyotype were initially reported by Kaplan 5 . Several phenotypic female patients with the XY karyotype were evaluated clinically, cytogenetically, hormonally, endoscopically and histologically 6 . The functional study of gonadal dysgenesis in phenotypic female patients with an XY karyotype has been hindered by a lack of animal models with specific mutations, except for the mouse sry mutant 3,31 . Using TALENs, two different Y-linked genes were efficiently manipulated in mouse embryonic stem cells (mESCs) 31 and an Sry knockout mouse was generated 3 . The mutant mice are almost completely infertile, although the Sry knockout mouse is similar to humans in terms of their physiological phenotype. The phenotype of the DMY knockout medaka is also similar to that human XY female syndrome. In particular, a majority of fertile individuals were found in the population. XY DMY− female medakas could help in studies of the mechanism of human XY female syndrome in genetics and reproductive biology. Benefiting from the number of embryos or offspring from the parents, fish are good for large sample analysis and for investigating individual differences of human XY female syndrome in populations.
DMY, a duplicated copy of DMRT1, is identified as the master male SD gene and shows all features of a SD gene in medaka 12 . In this study, we generated both insertions (+ 16) and deletions (− 11) of the TALEN-induced DMY mutant line, which developed into females with the XY karyotype ( Fig. 5A-C,F). Sex reversal was also reported very early in medaka 50 , which could be induced by steroid hormones or high temperature 51,52 . To evaluate whether the sex reversal observed in XY DMY− female medaka was caused by DMY gene mutation, the best method is rescue of the phenotypes of the XY DMY− female medaka. In the former case, the expected mutant DMY lacks functionally important motifs of DMY; this mutated Y chromosome could pair with the WT X and Y chromosome to generate females and males, respectively. Crossing with the WT medaka (Fig. 3), produced Y DMY− Y male mutants in the next generation, which meant that dysfunction of the DMY gene was the unique difference between XY DMY− female and Y DMY− Y male. This result confirmed that the sex reversal observed in XY DMY− female medakas was caused by a DMY gene mutation.
Adults of two phenotypically female mutant lines were evaluated through mutation confirmation, phenotype diagnostic (Fig. 5A) and histological analysis (Fig. 5B,F). Using genomic PCR (Fig. 4B) and FISH of DMY gene (Fig. 5C) showing the specific hybridization signal of the Y chromosome, the two phenotypically female mutant lines were identified as genetic males with the XY karyotype. RT-PCR analysis (Fig. 4C) and RNA-seq (Dataset 1) confirmed that XY DMY− mutants did not express DMY gene, which were different from WT XY individuals. Gsdf, co-localized with DMY in the same somatic cells in the XY gonads, was expressed exclusively in primordial gonads of only the genetic males 39 . Both the increased number of germ cells (Fig. 4D) and the significantly down-expression of Gsdf gene (Unigene44032) implied sex reversal of XY DMY− mutants took place in early developmental stages, attributed to the disruption of DMY gene. As expected, the TALEN-induced DMY knockout medakas had female external and internal specificities. Unlike the Sry KO mouse, which did not produce any offspring 3 , the majority of TALEN-induced DMY knockout medakas were fertile. Twelve of 15 XY DMY− mutants developed functional ovaries; and the ovaries of three females showed incomplete development. There were significant differences among XY DMY− mutants, WT female and WT male in GSI (Fig. 5D) in terms of their GSI scores. Histological analysis showed that the ovary of XY DMY− mutants displayed a reduced number of oocytes (Fig. 5F). In three of the 15 mutants, no eggs could develop beyond CA oocytes. The other 12 could generate mature eggs, however, the mature oocytes number of XY DMY− mutants are significantly fewer than that of WT XX females. Therefore, the ovary function of XY DMY− mutants was lower than that of WT females, not only in quality, but also in the quantity (Fig. 5F). To investigate whether there is significant difference in the fertilization of eggs and between XY DMY− female and WT female; test crosses were continuously and systemically recorded. During generation of offspring from the XY DMY− female testcrosses, we found that it was always difficult to obtain a sufficient number of offspring. After 10 days of continuous monitoring, 6 eggs matured per day from XY DMY− mutants; 20 eggs matured per day from WT XX females (Fig. 5E). The results from F2 phenotypes revealed that a reduced number of mature oocytes were one of the most plausible explanations for the reduced fertility in the DMY KO medakas. Thus, the XY DMY− mutants could develop functional ovaries; however, the development, maturation and fertilization capacity of their eggs were significantly lower than those of the WT female.

Transcriptomic analysis of TALENs induced DMY-mutants and the WT.
Natural sex reversal has been reported in medaka 50 , and mutation of the DMY gene was identified in several artificial mutants 14 . However, the molecular mechanism of sex reversal, inducing by loss of DMY, has not been resolved. RNA-Seq is a recently developed approach for transcriptome profiling that is based on deep-sequencing 32 . RNA-seq could quantify the changing expression levels of each transcript during development and under different conditions. XY DMY− mutant female from the F3 generation were used for RNA-seq analysis, which minimized the off-target effects of DMY-TALENs. Using the RNA-seq approach, we obtained transcriptome information of XY DMY− mutants, and through a comparative analysis of them, revealed the transcriptional function of the DMY gene in medaka.
Using RNA-seq, 157 million raw reads were generated from WT_M_TE, WT_F_OV and TA_F_OV libraries. However, only 75.8% of the processed reads were mapped to the reference genome of medaka in the Ensemble database (File S6). This indicated that the medaka genome information requires improvement, and transcriptomic sequencing could revise and promote the improvement of medaka genome. In addition, it implied that it would be better to use independent analysis with no reference genome in the analysis of medaka transcriptome, which we used in this study. The problem with the medaka genomic information had little bearing on the fact that medaka is a good model for human diseases. Although only 7481 of 9844 transcripts could match the medaka genome, 7440 transcripts had homologous genes in the human genome (Fig. 6H). This suggested that medaka might be similar to humans in terms of sex determination and regulation; the medaka transcriptome data could play a role in the analysis of human sex reversal patients.
Medaka sex is primarily determined by the presence or absence of DMY gene 12,13 . Several studies show that Dmrt1 16,38 , Gsdf 39 and Sox9b 40 is essential to maintain testis differentiation or regulate testis development. Un-expression of DMY (Uingene20290) and lower expression of Dmrt1a (Unigene42535), Gsdf (Unigene44032) and Sox9b (Unigene18419) in RNA-seq analysis provided the molecular basis to the failure of male sex determination, male-to-female sex reversal, and the failure of testis differentiation or development in XY DMY− mutants (Dataset 1). Among the differentially expressed genes in TA_F_OV compared with WT_F_OV and TA_F_OV compared with WT_M_TE ( Fig. 6F and Dataset 2), we found several potential factors may directly bind SRY and DMY, such as SLC25A38 and ENSORLT00000000529. This validated the transcriptional function of the DMY gene. Theoretically, the DMY gene, a SD gene, might have a relationship with the sperm production and cilia assembly. However, from our comparative analysis, it was apparent that differences in the expressions of genes involved in these processes represented background differences between the WT ovary and WT testis tissue. This is an essential problem in the transcriptional function of DMY, but it beyond the scope of this study. Further, we may investigate developmental stages of gonad or different tissue of the HPG axis to explain the problem.
The main differentially expressed genes between WT_F_OV and TA_F_OV, were genes in the wnt receptor signaling pathway (Predicted: syntabbulin-like, axin-2-like) (Dataset 4). In mammals, beta-arrestin-1-like gene regulated IGF-1 affects human reproductive endocrinology. The medaka homolog of beta-arrestin-1-like gene was detected 10 times in the RNA-seq data. The high expression of the beta-arrestin-1-like gene could provide some clues to the mechanism of the degradation of the testis and the development of the ovary in XY DMY mutant females. Lhr and Forkhead box O5 factor were also differentially expressed between TA_F_OV and WT_F_OV. Lhr is the luteinizing hormone receptor, and Forkhead box O5 factor plays an important role in metabolism and cell differentiation. The low expression of lhr and Forkhead box O5 factor might affect the incomplete ovary function of the XY DMY mutant female. The majority of genes in wnt signal pathway, such as Wnt genes, PRCK genes, FZD genes and DKK were differentially expressed between WT_M_TE and TA_F_OV (Dataset 1), it implied that the wnt signaling pathway is the main regulation pathway during male-to-female sex reversal process. In addition, the genes in wnt signaling pathway, especially Wnt1 and PRCK, were up-expressed in XY DMY− mutants compared with WT XY males, and down-expressed compared with WT XX females (Dataset 1 and 4). These results implied that the wnt signaling pathway also contributed primarily to the ovarian development, reduced fertility and ovarian maturation in the XY DMY− mutants. Interestingly, PRCKA and DKK were up-expressed in TA_F_OV, whether compared with WT_M_TE (Dataset 1) or compared with WT_F_OV (Dataset 4). In summary, these results implied that the wnt signaling pathway is the root of sex reversal, the incomplete ovarian function and reduced fertility in the XY DMY− mutants.

Construction of TALENS. TALEN were assembled and transferred into vectors pCS2-KKR and
pCS2-ELD 28 . To improve the germline integration efficiency, the zebrafish nanos-3′ UTR was separately inserted into the 3′ end of the pCS2-ELD/KKR vector to replace the SV40 UTR using NotI and XbaI.
Manipulation of medaka embryos. The final TALEN plasmids were linearized using Not1, and the mMessage mMachine SP6 kit (Ambion) was used to synthesize mRNAs. TALENs mRNAs (half left and half right monomer mRNAs) were microinjected into medaka embryos at the one cell stage.
Detection of mutations in TALEN-targeted medaka embryos. 72 hours after microinjection, TALEN targeted embryos were pooled for genomic DNA extraction (10 embryos for each pool). PCR was performed using primers DMY F and DMY R; PCR products were purified from the agarose gel using a gel extraction kit (QIAGEN). Amplicons harboring the targeted gene fragments were sub-cloned into pMD-18T using TA cloning (Takara), and single colonies were examined by PCR using primers DMY F, DMY F1 and DMY R. The PCR conditions were as follows: 5 min at 95 °C; followed by 30 cycles of 15 sec at 95 °C, 20 sec at 52 °C, 30 sec at 72 °C; and a final step of 5 min at 72 °C. PCR products were electrophoresed in a 2% agarose gel and verified by DNA sequencing.
Off-target analysis. The criteria of the novel program for determining off-target sites were that the 0 position of the EBEs must be T, from 1 to 10 mismatch bases occur in the pairs of EBEs, and the spacer between the two putative EBE regions is less than 100 bp, because it has been suggested that longer spacers interfere with Fok I dimerization.
Founder Screening. TALEN-injected medaka embryos were raised to sexual maturity. F0 DMY mutated females were crossed with wild-type males; and F1 embryos were collected at 72 hours post fertilization. Genomic DNA was extracted from ten randomly pooled embryos to assess mutagenesis at the TALEN-targeted site by PCR and sequencing. F1 embryos were individually collected at 7 days post fertilization (dpf), and genomic DNA was extracted from each individual embryo to assess mutagenesis at the TALEN-targeted site by PCR and sequencing.