Heterozygous mutation of eEF1A1b resulted in spermatogenesis arrest and infertility in male tilapia, Oreochromis niloticus

Eukaryotic elongation factor 1 alpha (eEF1A) is an essential component of the translational apparatus. In the present study, eEF1A1b was isolated from the Nile tilapia. Real-time PCR and Western blot revealed that eEF1A1b was expressed highly in the testis from 90 dah (days after hatching) onwards. In situ hybridization and immunohistochemistry analyses showed that eEF1A1b was highly expressed in the spermatogonia of the testis. CRISPR/Cas9 mediated mutation of eEF1A1b resulted in spermatogenesis arrest and infertility in the F0 XY fish. Consistently, heterozygous mutation of eEF1A1b (eEF1A1b+/−) resulted in an absence of spermatocytes at 90 dah, very few spermatocytes, spermatids and spermatozoa at 180 dah, and decreased Cyp11b2 and serum 11-ketotestosterone level at both stages. Further examination of the fertilization capacity of the sperm indicated that the eEF1A1b+/− XY fish were infertile due to abnormal spermiogenesis. Transcriptomic analyses of the eEF1A1b+/− testis from 180 dah XY fish revealed that key elements involved in spermatogenesis, steroidogenesis and sperm motility were significantly down-regulated compared with the control XY. Transgenic overexpression of eEF1A1b rescued the spermatogenesis arrest phenotype of the eEF1A1b+/− testis. Taken together, our data suggested that eEF1A1b is crucial for spermatogenesis and male fertility in the Nile tilapia.

gametogenesis via non-canonical function. Oral administration of the eEF1A1 inhibitor gamendazole resulted in infertility in male mice, probably via disruption of actin-filament bundles related to the Sertoli cell-spermatids ectoplasmic specialization junctions 26 . However, there have been no other reports on the role and mechanism of other eEFs, including eEF1A, in gametogenesis.
In addition to somatic cells, eEF1A is also expressed in germ cells. Proteins expressed in germ cells are directly related to gametogenesis. Disruption of genes expressed in germ cells often blocks gametogenesis [27][28][29][30] . Therefore, eEF1A might be directly involved in gametogenesis via protein synthesis in germ cells. Different eEF1A genes are reported to be expressed in the testis and ovary of Xenopus 11,12 . Our gonads transcriptomic data from gonads demonstrated that eEF1A1b and 42Sp50 are XY-enhanced and XX-specific genes in tilapia, respectively 31 . Taken together, we speculate that different elongation factors might be used for protein synthesis during spermatogenesis and oogenesis in lower vertebrates. Species with different eEF1A genes expressed in the testis and ovary would be a good model to study the diverse function of the eEF1A genes in spermatogenesis and oogenesis.
Nile Tilapia (Oreochromis niloticus), is a good model for the study of gene expression and function in fish due to the availability of monosex fish 32 , a short spawning cycle, many gonadal transcriptomes 31 and published genome sequences 33 . In the present study, expression of eEF1A1b was found to been enriched in XY males during tilapia gonadal development. Ontogenetic studies were conducted to determine its expression profile by real-time PCR and Western blot. The cellular localization of eEF1A1b was documented by in situ hybridization (ISH) and immunohistochemistry (IHC). We also mutated eEF1A1b and analyzed the gonadal phenotype, gene expression, serum 11-ketotestosterone (11-KT) level and fertility of the eEF1A1b deficiency and eEF1A1b +/− fish.

Results
Phylogenetic and synteny analyses. As shown in the phylogenetic tree (supplemental Fig. S1), the phylogeny of eEF1A1 was split into two main clades. One clade included the eEF1A1 gene of tetrapods; and the other included the two distinct copies of eEF1A1 from teleosts that we named eEF1A1a and eEF1A1b.
Synteny analyses revealed the organization of the genomic region surrounding eEF1A1 genes in tetrapods and teleosts. eEF1A1a and its upstream gene slc17a5, its downstream gene ddx43 and kcnq5a showed conserved synteny both in teleosts and tetrapods. eEF1A1b and its upstream gene dtd1 showed conserved synteny only in teleosts, and its downstream gene kcnq5b and rims1 showed conserved synteny only in tetrapods. Except kcnq5, no other conserved synteny of gene loci surrounding eEF1A1a and eEF1A1b were found (supplemental Fig. S2).

Tissue distribution.
Analysis by IHC showed that eEF1A1b was highly expressed in the brain, heart, liver, intestine, kidney and testis, weakly expressed in the ovary, but was undetectable in gill, spleen, head kidney and muscle (Supplemental Fig. S3).

Expression profile of eEF1A1b in developing gonads.
Real-time PCR demonstrated that expression of eEF1A1b was low from 10 to 90 dah, increased sharply from 90 dah in testis and slightly decreased from 150 to 180 dah; while no significant changes were found during all stages in ovary (Fig. 1a).
Western blot analyses revealed that eEF1A1b was expressed in both testis and ovary, with significantly higher expression in the testis than in the ovary from 90 dah, peaking at approximately 120 dah, and decreased to a much lower level at 300 dah (Fig. 1b).
Cellular localization of eEF1A1b in gonads. ISH analysis demonstrated that eEF1A1b was expressed in the spermatogonia of the testis, and in the oogonia and phase I oocytes of the ovary with eEF1A1b antisense probe (Fig. 2a, A and B). eEF1A1a, the paralogous gene of eEF1A1b, was expressed in the somatic cells of the testis, and in the oocytes and somatic cell of the ovary (Fig. 2a, C and D). In contrast, signal was detected in neither testis nor ovary with eEF1A1a and eEF1A1b sense probe (Fig. 2a, E-H).
Consistently, IHC analysis revealed that eEF1A1b was expressed exclusively in the spermatogonia of the testis from 5, 30, 90 and 180 dah (Fig. 2b, A-D), and was expressed in the oogonia of the ovary at 5 dah, later in the oogonia and phase I oocytes of the ovary at 30 to 180 dah (Fig. 2b, E-H).
Disruption of eEF1A1b by CRISPR/Cas9 and production of F 1 generation. A guide RNA site containing BsajI adjacent to proto-spacer adjacent motif in the second exon of eEF1A1b was selected for mutation analysis (Fig. 3a). Genomic DNA from 20 pooled injected embryos and from fin of mutants was used as template for PCR amplification and mutation assays. Complete digestion produced fragments of 425 and 102 bp in the control groups whereas an intact DNA fragment was observed in embryos injected with both Cas9 mRNA and target guide RNA (Fig. 3b). The mutation rate in the pooled embryos was approximately 97%. In-frame and frame-shift deletions induced at the target site were confirmed by Sanger sequencing. We screened 40 individual microinjected fish by restriction enzyme digestion at 90 dah, and found 38 to be mutant for eEF1A1b. All mutant fish had mutation efficiency over 90% (Supplemental Tab. S3). Twenty mutant fish were further confirmed by Sanger sequencing. In total, we analyzed 200 sequences and identified 14 mutation types, including 11 types of deletion mutation and 3 types of insertion mutation. The majority (67.5%) were frame-shift mutations, while 32.5% were in-frame mutations (Supplemental Fig. S4). F 1 offspring were obtained by outcross eEF1A1b chimeric XX with XY control fish, and two mutation types (4 and 13 bp deletion) were identified in F 1 offspring by Sanger sequencing (Fig. 3c,d and e).
Effects of eEF1A1b deficiency on spermatogenesis and fertility in F 0 fish. Morphologically, all kinds of germ cells including spermatogonia, spermatocytes, spermatids and spermatozoa were presented in the control testis at 90, 120, 150 and 180 dah (Fig. 4a, A-D). In contrast, in the eEF1A1b deficient testis only spermatogonia were observed at 90 and 120 dah, spermatocytes appeared at 150 dah and spermatids and spermatozoa appeared at 180 dah (Fig. 4a, E-H). Additionally, fewer spermatozoa were observed in the efferent duct of the eEF1A1b deficient testis at 180 dah compared with the control.
By IHC, eEF1A1b was detected in the spermatogonia, and Cyp11b2 was observed in the Leydig cells in the control testis at 180 dah (Fig. 4b, A and B). In contrast, reduced expression of eEF1A1b and Cyp11b2 was observed in the eEF1A1b deficient testis (Fig. 4b, D and E). However, Vasa was detected in the spermatogonia and spermatocytes in the eEF1A1b deficient and control testis (Fig. 4b, C and F).
Western blot analyses revealed that the expression of eEF1A1b, Cyp11b2 and Vasa was significantly reduced in the eEF1A1b deficient testis compared with those of the control (Fig. 4c). Additionally, the size of the eEF1A1b deficient testis was significantly smaller than that of the control testis as demonstrated by gonadal somatic index (GSI, gonad weight/body weight × 100%) (Fig. 4d).
Fertility test showed that the eEF1A1b deficient XY fish were infertile (supplemental Tab. S4). When examined under microscope, the sperm of the eEF1A1b deficient fish were abnormal. The sperm from the control fish were normal with a flagella length of 16.2 ± 2.4 μ m, and displayed strong motility. In contrast, the sperm from eEF1A1b deficient fish were characterized by short or absent flagella. The short-flagella sperms with a flagella length of 7.55 ± 3.16 μ m, displayed weak motility, and the flagella-less sperms were stuck together, showing no motility ( Effects of eEF1A1b deficiency on ovarian development in F 0 fish. In contrast to the significant phenotypes of eEF1A1b deficiency in the testis, the ovary of most mutant fish showed normal development at 120 and 180 dah. Histological analysis demonstrated that folliculogenesis was normal in the ovary of the eEF1A1b deficient XX fish (Supplemental Fig. S6).
Effects of eEF1A1b heterozygous mutation on spermatogenesis and fertility in F 1 fish. The heterozygous mutation of eEF1A1b (eEF1A1b +/− ) XY fish with a 4 bp deletion were employed for phenotype analyses. Consistent with the phenotype of F 0 generation, eEF1A1b +/− XY fish also displayed spermatogenesis arrest. Histologically, spermatogonia, spermatocytes and spermatids, were present at eEF1A1b +/+ testis, whereas only spermatogonia were observed in eEF1A1b +/− XY testis at 90 dah (Fig. 5a). By IHC, eEF1A1b and Oct 4 (spermatogonia maker), Phospho-histone h3 (Ph3, spermatocytes maker), Cyp11b2 were observed in the spermatogonia, spermatocytes and Leydig cells, respectively, of the eEF1A1b +/+ testis. In contrast, Oct 4 and reduced expression of eEF1A1b were observed in the spermatogonia, while no Ph3 and Cyp11b2 expression was detected in the eEF1A1b +/− testis. These results indicated that there were only spermatogonia, but no spermatocytes in the eEF1A1b +/− testis (Fig. 5a). In addition, the eEF1A1b, cyp11b2 and vasa mRNA levels were found to be significantly down-regulated in the eEF1A1b +/− testis by real-time PCR (Fig. 5b). Further, significantly lower serum 11-KT level was detected in the eEF1A1b +/− XY fish compared with that of the eEF1A1b +/+ XY fish at 90 dah (Fig. 5c). Furthermore, the size of the eEF1A1b +/− testis was significantly smaller than that of the eEF1A1b +/+ testis as reflected by the GSI. By histology, abundant spermatocytes, spermatids and spermatozoa were observed in the eEF1A1b +/+ testis at 120, 150 and 180 dah (Fig. 6, A-F). In contrast, in the testis of eEF1A1b +/− XY fish, just a few spermatocytes were observed at 120 dah, and a markedly reduced number of spermatocytes, spermatids and spermatozoa were observed at 150 and 180 dah (Fig. 6, G-L).
Fertility test showed that similar to the chimeric XY fish, eEF1A1b +/− XY fish were infertile (Fig. 7A). Morphological and biochemical analyses of sperm were performed to analyze the reason for male infertility in eEF1A1b +/− XY fish. The sperm of eEF1A1b +/+ XY fish were characterized by normal flagella, while the sperm of eEF1A1b +/− XY fish were flagella-less. Therefore, the ratio of sperm with morphological abnormalities to the total sperm was significantly higher in eEF1A1b +/− XY fish than that in the eEF1A1b +/+ XY fish (Fig. 7, B-D). Consistently, the activity of sperm flagella was significantly lower in eEF1A1b +/− XY fish than that in eEF1A1b +/+ XY fish (Fig. 7E). Consequently, the sperm forward motility was lost in the eEF1A1b +/− XY fish (Fig. 7, F-H).
Gonadal transcriptome analysis of eEF1A1b +/− XY fish. Transcriptome sequencing of the testes from the eEF1A1b +/− XY fish at 180 dah yielded a total of 40,039,839 reads. The eEF1A1b +/− XY transcriptome data were analyzed using the transcriptome data from testis of 180 dah XY fish obtained earlier by our group 31 as a control. The total gene counts of the eEF1A1b +/− XY fish (22,284) were slightly more than those of the eEF1A1b +/+ XY fish (21,365). Compared with the control testis, 1,522 genes were down regulated, 1,891 genes were up regulated, 11,554 genes were not differentially expressed and 7,317 genes were at background expression level in the eEF1A1b +/− XY testis (Fig. 8a).
Transgene mediated rescue of spermatogenesis in eEF1A1b +/− XY fish. To rescue the spermatogenesis arrest phenotype of eEF1A1b +/− XY fish, transgenic overexpression of eEF1A1b was conducted in the eEF1A1b +/− XY fish using a CMV promoter-derived expression vector. A specific band of 720 bp was amplified from the genomic DNA of eEF1A1b transgenic (eEF1A1b +/−/TG ) fish (Fig. 9a). GFP signals were detected in the eEF1A1b +/−/TG testis when examined at 90 dah (Fig. 9b, A and B).
By IHC, Vasa and Ph3 were observed in the spermatogonia/spermatocytes and spermatocytes, and Cyp11b2 were detected in the Leydig cells of the eEF1A1b +/−/TG testis (Fig. 9b, C-E). In contrast, reduced Vasa, and no Ph3 and Cyp11b2 were observed in the eEF1A1b +/− testis (Fig. 9b, F-H).
By real-time PCR, the expression of eEF1A1b, cyp11b2, vasa, suz12 and usp26 in the eEF1A1b +/−/TG testis showed no significant differences compared with the eEF1A1b +/+ testis, but significantly higher than that of the eEF1A1b +/− testis (Fig. 9c). Consistently, the serum 11-KT level and GSI of the eEF1A1b +/− XY fish showed no significant differences compared with the eEF1A1b +/+ testis, but significantly higher than those of the eEF1A1b +/− testis ( Fig. 9d and e)

Discussion
Evolutionary implication of eEF1A1b. In the present study, eEF1A1a and eEF1A1b genes were successfully identified from the Nile tilapia, zebrafish, medaka and fugu. Phylogenetic analyses of eEF1A1s from vertebrates demonstrated that eEF1A1a and eEF1A1b arose from the teleost genome duplication 34 . eEF1A1a has retained the ancestral functions, while eEF1A1b has evolved novel function/expression patterns. Moreover,  conserved synteny of eEF1A1a with kcnq5a and eEF1A1b with kcnq5b was observed in teleosts. Taken together, these results suggested that eEF1A1a and eEF1A1b were derived from the teleost-specific genome duplication.
The expression pattern and functional relevance of eEF1A1b in tilapia. Spatial-temporal gene expression patterns are important aspects of gene function analysis. Our previous data showed that eEF1A1a displayed no gender difference in expression, while eEF1A1b was a male-biased gene in tilapia gonad 31 . In the present study, both real-time PCR and Western blot analyses demonstrated that eEF1A1b showed sexually dimorphic expression from 70 dah, and was highly expressed in the testis from 90 dah onwards, correlated with the initiation of spermatogenesis and the appearance of spermatocytes. The expression of eEF1A1b reached its highest levels with the increase of the spermatogonia at 120 dah. Further, eEF1A1b was expressed exclusively in spermatogonia, while eEF1A1a was found to be mostly expressed in somatic cells of the testis. In the ovary, the expression of eEF1A1b was low and constant. eEF1A1 is a housekeeping gene required for survival of all eukaryotic cells 35,36 , and therefore, at least one eEF1A gene must be expressed in the germ cells and somatic cells of the gonad in vertebrates. In mammals, eEF1A1 is a sex unbiased gene, and is expressed in the germ cell and somatic cell of the gonad [13][14][15][16] . In Xenopus, eEF-1S and eEF-1O are expressed in somatic cells and germ cells respectively, and 42sp50 is detected only in oocytes 10,11 . In tilapia, eEF1A1a was expressed in the somatic cells of the testis and ovary, while eEF1A1b and 42Sp50 were expressed in the spermatogonia and oogonia/oocytes, respectively. Taken together, different elongation factor isoforms are used for male and female germ cells, as well as germ cells and somatic cells in lower vertebrates, while only one (eEF1A1) is used in both male and female germ cells and somatic cells of gonad in mammals. In the present study, eEF1A1b deficiency resulted in spermatogenesis arrest and consequent infertility in F 0 XY fish. This was reflected by defective of meiosis, reduced GSI and abnormal spermiogenesis. Heterozygous mutants of eEF1A1b displayed a phenotype similar to that of the F 0 XY fish. The eEF1A1b +/− testis showed no and very few spermatocytes at 90 and 120 dah, respectively. Spermatocytes, spermatids and spermatozoa were remarkably reduced at 150 and 180 dah. Further, most sperm from the eEF1A1b +/− XY fish were flagella-less, and therefore, incapable of fertilization. Anatomical examination of eEF1A1b +/− XY fish showed mostly normal organs and tissues, including the heart and brain, the organ with high eEF1A1b abundance (supplemental Fig. S7). Thus, our results indicated the impairment of testis development, and failure of fertilization in mutant fishes were probably not the result of defects in these tissues, but a specific effect of the eEF1A1b deficiency in testis. Both the eEF1A1b deficient and eEF1A1b +/− XY fish were infertile, clearly indicating that eEF1A1b is critical for spermatogenesis and fertility in male tilapia.
Of all eEF1A isoforms, only eEF1A1b was expressed in the spermatogonia in tilapia. Consequently, mutation of eEF1A1b caused spermatogenesis arrest and infertility. It is well documented that eEF1A is indispensable for protein synthesis. Recently, a report showed that genetic ablation of eEF4, which is a key quality-control factor in translation, causes testis-specific dysfunction in oxidative phosphorylation, leading to male infertility in mice 20 . Besides elongation factors, disruption of other genes expressed in the spermatogonia also blocks spermatogenesis 27,29,37 . Consistently, in the present study, the mRNA level of suz12a, aspm, and usp26, which were expressed in spermatogonia and critical for spermatogenesis 28,[37][38][39] , were significantly down regulated in the eEF1A1b +/− testis. It is reasonable to believe that they were also down regulated in protein levels although we did not check the protein levels of these genes due to lack of specific antibodies. On the other hand, eEF1A1 might also be involved in spermatogenesis via other cellular processes 26 . However, the non-canonical function of eEF1A1b in spermatogenesis of tilapia remains to be elucidated.
In addition, suppression of steroidogenesis was observed in eEF1A1b mutant XY fish, as reflected by the decrease of Leydig cells, down-regulation of the expression of Sf-1 and the steroidogenic enzyme StarI, StarII, Cyp11a1, Cyp11b2 and 11β -HSD and a decrease in serum 11-KT level. It is well known that 11-KT is essential for spermatogenesis in fish 40 , and therefore, spermatogenesis arrest and infertility in eEF1A1b deficient and eEF1A1b +/− XY fish might be the resulted of low levels of 11-KT. Nevertheless, it is an indirect effect of the eEF1A1b mutation as this gene is not expressed in the Leydig cells of tilapia. These results suggested that spermatogonia-derived signals might affect the Leydig cells directly or indirectly. It is widely accepted that, during the development of testis, male germ cells are influenced by signals from the surrounding somatic cells, but not vice versa 41 . But this conventional perspective was challenged recently. For example, mutation dominant-white spotting locus, which is expressed in spermatogonia of testis 42 , resulted in absence of germ cells, decreased Leydig cell population and disordered expression profiles of Leydig cells key markers (cyp11a1, hsd3β and insl3) in mice 43 . Moreover, overexpression of figla resulted in up-regulation of 11-KT in male tilapia 44 . Taken together, we speculate that the Leydig cells in the testis might be influenced by signals from the male germ cells. However, the mechanisms involved remain to be elucidated.
The phenotype of eEF1A1b +/− testis was rescued by eEF1A1b overexpression as reflected by the normal testis size and appearance of all types of male germ cells, including spermatocytes and spermatids, which were absent in the eEF1A1b +/− testis. Moreover, eEF1A1b transgene also rescued the expression of those genes involved in the spermatogenesis process and steroidogenesis altered in the eEF1A1b +/− testis. These results validated that the spermatogenesis arrest phenotype is exclusively caused by eEF1A1b haploinsufficiency rather than off-target effects.
Previous reports showed that asthenozoospermia is a common cause of human male infertility, diagnosed by shortened sperm flagella, reduced sperm concentration and motility 45 similar to the infertile phenotype of eEF1A1b mutant XY fish. The exact cause of asthenozoospermia is unclear so far, and mutation of eEF1A1 might be lethal in mammals. Mutation of eEF1A1b in tilapia could be used as a model for investigating the possible involvement of eEF1A1 in asthenozoospermia.
In summary, phylogenetic and syntenic analyses revealed that eEF1A1b is a duplicated copy of eEF1A1. In the testis it is exclusively expressed in the spermatogonia. Loss of function and transgenic overexpression studies indicated that the spermatogenesis arrest and infertility phenotypes observed in the eEF1A1b +/− XY fish was indeed caused by eEF1A1b haploinsufficiency. Taken together, our results suggested that eEF1A1b is essential for spermatogenesis and male fertility in the Nile tilapia. In addition, mutation of eEF1A1b in tilapia may provide an excellent model for studying human asthenozoospermia. The fertilization rate of eEF1A1b +/− XY fish was significantly lower than that of eEF1A1b +/+ XY fish (A). Sperm of eEF1A1b +/+ XY fish were with normal flagella, while most sperm of eEF1A1b +/− XY fish were flagella-less, and the ratio of sperm with morphological abnormalities to the total sperm was higher in eEF1A1b +/− XY fish than in eEF1A1b +/+ XY fish (B-D). The beat frequency of sperm flagella was higher in eEF1A1b +/+ XY fish than in eEF1A1b +/− XY fish (E). And the ratio of forward motility sperm was higher in eEF1A1b +/+ XY fish than in eEF1A1b +/− XY fish (F-H). The red line indicates the trajectories of sperm. Results were expressed as mean ± SD. Different letters indicate statistical differences at P < 0.05 as determined by one-way ANOVA followed by post hoc test.

Materials and Methods
Animals. Nile tilapia, Oreochromis niloticus, was kept in recirculating aerated freshwater tanks at 26 °C under a natural photoperiod. All-XX progenies were obtained by crossing a pseudomale (XX male, producing sperm after hormonal sex reversal) with a normal female (XX). All-XY progenies were obtained by crossing a supermale (YY) with a normal female. Animal experiments were conducted in accordance with the regulations of the Guide for Care and Use of Laboratory Animals and were approved by the Committee of Laboratory Animal Experimentation at Southwest University.
Phylogenetic and synteny analysis of eEF1A1b. A phylogenetic tree of eEF1A1s was constructed using tilapia 42Sp50 as the outgroup. Sequences were obtained from the NCBI (http://blast.ncbi.nlm.nih.gov/) and Ensembl (http://www.ensembl.org/index.html) databases. The accession numbers of the sequences used are listed in the Supplemental Tab.S1. Full-length amino acid sequences were aligned using MEGA5.0 46 . The credibility of the branching was tested using bootstrap resampling with 1000 pseudo-replicates.
For syntenic analysis, the location of eEF1A1 (including eEF1A1a and eEF1A1b) and their adjacent genes were determined for human, rat, chicken, green anole, clawed frog, Nile tilapia, medaka, fugu and zebrafish using the Ensembl genome browser (http://www.ensembl.org/index.html).
Tissue distribution of eEF1A1b by IHC. The various tissues obtained from XX and XY tilapia at 180 dah tilapia were fixed in Bouin's solution for 12 hours at room temperature, dehydrated and embedded in paraffin. All tissue blocks were sectioned at 5 μ m and used for IHC analysis, which was performed as described previously 27 . The antibody against eEF1A1b was diluted 1:1000 for use.
Production and characterization of eEF1A1b antibody. The peptide antigen (SGWNGDNMLEPSPNMT) of tilapia eEF1A1b was prepared and injected into rabbits. eEF1A1b antiserum (a-DTc) was collected and affinity purified. To confirm the specificity of the polyclonal antibody, total protein was extracted from XX and XY gonads of 30, 90, 120 and 300 dah tilapia. Western blots were carried out using the purified antibody at dilution of 1:500.
Expression profile of eEF1A1b in tilapia gonad by real-time PCR and Western blot. For ontogenic expression analysis, three parallel gonadal samples were prepared from the fish at 5, 10, 30, 40, 50, 60, 70, 90, 120, 150 and 180 dah. For sampling points from 5 to 60 dah, fish were dissected and viscera were removed under a stereoscopic microscope. RNAlater reagent (Ambion, TX) was poured on the coelomic epithelium to stabilize the RNA in the gonads, and then the gonads were removed using fine forceps. Gonads were pooled in a tube with RNAlater reagent and stored at − 80 °C before RNA extraction.
Total RNAs (2.0 μ g) were extracted and reverse transcribed into cDNA using PrimeScript RT Master Mix Perfect Real Time Kit according to the manufacturer's instructions (Takara, Japan). Real-time PCR was carried out on an ABI-7500 real-time PCR machine according to the protocol of SYBR ® Premix Ex Taq II. The relative abundance of eEF1A1b mRNA transcripts was evaluated using the formula: R = 2 −∆∆Ct 47 . The geometric mean of the copy numbers of the three reference genes (β-actin, gapdh and eEF1A1a) was used to normalize the expression of eEF1A1b. Primer sequences used for real-time PCR are listed in Supplemental Tab. S2. Data were expressed as the mean ± S.D. Statistical analysis was performed using GraphPad Prism4 software (GraphPad Software, USA). Significant differences between groups were tested by one-way ANOVA with post-hoc test. P < 0.05 was considered to be statistically significant.
Total proteins extracted from testes and ovaries of 30, 90, 120 and 300 dah tilapia were separated using 12% SDS-PAGE under reducing condition. Western blot was carried out as reported previously 48 using the purified antibodies at dilution of 1:500. Cellular location of eEF1A1b in gonads analysis by ISH and IHC. ISH was performed using tilapia gonads collected 180 dah from monosex (XX or XY) fish. Gonads were dissected and fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4, 4% PFA) at 4 °C overnight. After fixation, gonads were embedded in paraffin. Cross-sections of 5 μ m were cut with a sliding microtome and adhered to polylysine-treated slides with diethylpyrocarbonate-treated water. Slides were dried at 180 °C for 4 hours to remove the RNase contamination and stored at 4 °C until use. Probes for both sense and antisense digoxigenin-labeled RNA strands were transcribed in vitro from a linearized pGEM-Teasy-eEF1A1b cDNA clone using the RNA labeling kit (Roche, Germany). ISH was performed as described previously 49 .
For IHC analysis, the gonads of 5, 30, 90 and 180 dah monosex (XX and XY) fish were dissected, fixed in Bouin's solution for 12 hours at room temperature, dehydrated and embedded in paraffin. All tissue blocks were sectioned at 5 μ m for IHC analysis, which was performed as described previously 27 . The antibody against eEF1A1b was diluted 1:1000 for use. Disruption of eEF1A1b by CRISPR/Cas9. CRISPR/Cas9 was performed to knockout eEF1A1b in tilapia 27 .
The gRNA target site was selected from eEF1A1b sequences corresponding to GGN18NGG on the sense or antisense strand of DNA (http://zifit.partners.org/ZiFiT/). Candidate target sequences were compared with the entire tilapia genome using the Basic Local Alignment Search Tool to avoid cleavage of off-target sites. One-cell stage embryos were divided into two batches, one for microinjection and the other for control. The gRNA and Cas9 mRNA were co-injected into one-cell stage embryos with the optimal concentration of 150 ng/μ l and 300 ng/μ l, respectively. Twenty injected embryos were collected 72 hours after injection. Genomic DNA was extracted from pooled injected embryos and control embryos, and used to access the mutations. DNA fragments spanning the eEF1A1b target site was amplified using gene specific primers (Supplemental Tab. 2). The indels were confirmed by restriction enzyme digestion with BsajI and Sanger sequencing. In addition, the percentage of uncleaved band was measured by quantifying the band intensity using Quantity One Software (Bio-Rad, USA) 50 . The indel (insertion and deletion) frequency was calculated by dividing uncleaved band intensity to the total band intensity from single digestion experiment.
The remaining micro-injected fish were reared until sampling for phenotypic assays. To screen the mutant fish, a piece of tail fin was clipped from each individual, and genomic DNA was extracted as described above. The target genomic locus was amplified using the primers eEF1A1b-cas-F/R. Mutations were assessed by restriction enzyme digestion and Sanger sequencing. The genetic sex of the mutants was determined using a sex-linked DNA marker (marker5) reported previous by our group 32 . The eEF1A1b-deficient fish were used for gonad histology, gene expression (real-time PCR, IHC, and Western blot), serum androgen (11-KT) analyses and fertility test at 90 and 180 dah. eEF1A1b deficiency on spermatogenesis and fertility. Gonads of eEF1A1b deficient fish were sampled at 90, 120, 150 and 180 dah. Samples were fixed in Bouin's solution for 24 hours at room temperature, dehydrated and embedded in paraffin. Tissue blocks were sectioned at 5 μ m and stained with hematoxylin and eosin (H.E) or used for IHC analysis. Vasa (germ cell marker) antibody was donated by Prof. Nagahama, the National Institute of Basic Biology, Okazaki, Japan. The specificity of the antibody has been analyzed previously 51 . eEF1A1b and Cyp11b2 (11β -hydroxylase, the key enzyme for androgen 11-KT synthesis) antibody were prepared by our laboratory. The specificity of Cyp11b2 antibody was checked previously 52 . Antibodies against eEF1A1b, Vasa and Cyp11b2 were diluted 1:1000, 1:1000 and 1:500 for use, respectively. IHC analyses were performed as described above. Photographs were taken under Olympus BX51 light microscope.
Western blots were performed to confirm the expression change of eEF1A1b, Cyp11b2 and Vasa in the eEF1A1b deficient and the control testis at 180 dah.
Serum 11-KT level was measured using the EIA Assay kits (Cayman Chemical Co, USA) following the manufacturers' instructions. Blood samples were collected from the caudal vein of the eEF1A1b deficient and control XY fish at 90 dah and kept at 4 °C overnight. Serum was collected after centrifugation and stored at − 20 °C until use.
Sperm collected from eEF1A1b deficient and the control XY fish at 180 dah were diluted 1:1000 with phosphate buffered saline. Examination of morphology and motility of sperm was performed under Olympus BX51 light microscope.
To test fertility, 1200 mature eggs from an XX mother fish were divided into six groups (each with 200 eggs), and artificial insemination was performed using sperm from eEF1A1b deficient and control XY fish (n = 3). 12 hours later, the number of gastrula-stage embryos was scored under a light microscope to calculate fertilization rate. Hatching rates were calculated at 1 dah (5 days post fertilization). The concentration of sperm from eEF1A1b deficient and control XY fish was calculated using cell count data from flow cytometry (n = 3).
Production of F1. The eEF1A1b heterozygous F 1 offspring were obtained by mating the F 0 female founder with a wild-type male. F 1 fish were screened by restriction enzyme digestion of PCR amplifications, as described above. Mutate sequences were confirmed by Sanger sequencing.
Sperm of eEF1A1b +/− and eEF1A1b +/+ XY fish was sampled 180 dah to assay sperm quality. Fertility tests were performed as described above. Sperm analysis, including sperm count, morphological abnormality, trajectory, flagellum beat frequency, motility and path velocity, were conducted using Sperm Quality Analyzer (ZKPACS-E) in The Ninth People's Hospital of Chongqing.
Transcriptome analysis. The transcriptomes of 180 dah control testes was sequenced previously by our group 54 . Six testes from eEF1A1b +/− XY fish were used for transcriptome analyses. Total RNA was extracted from testes using the RNeasy Mini Kit (50) (Qiagen) according to the manufacturer's instructions. The extracted RNA was further treated with deoxyribonuclease 1 (ribonuclease free) to eliminate genomic DNA contamination. The oligo (dT) bead-enriched mRNA was disrupted into short fragments (200-700 nt) using fragmentation buffer. These short fragments were used as templates for first-and second-strand cDNA synthesis using a DNA synthesis kit (Invitrogen). A QiaQuick PCR purification kit (Qiagen) was used to purify these fragments, and the elution buffer was used for end repair and addition of the poly (A) tail. Then, these short fragments were ligated to sequencing adapters. After agarose gel electrophoresis, fragments between 320 and 370 nt were cut from the gel for PCR amplification. cDNA libraries were constructed from the two samples and sequenced on an Illumina HiSeq2000 instrument. Clean reads from each library were aligned to the reference genome (Orenil1.0, http:// www.ensembl.org/Oreochromis_niloticus/Info/Index) using Tophat with default parameters, and the reads per kb per million reads (RPKM) method was used to calculated gene expression level 55 . The assembled transcripts were merged with the reference annotation (Oreochromis niloticus. Orenil1.0.78.gtf, downloaded from Ensembl) using Cuffmerge, and differential expression analysis was performed using Cuffdiff. Transcriptomes of control-XY and eEF1A1b +/− XY were used to analyze the genes expression profiles. The threshold for the p-value was determined using false discovery rate (FDR) set at 0.01 31 . In this study, genes with RPKM < 1 in the transcriptome were considered to be background expression and were excluded from further analysis. Gonadal expressed genes were divided into non-differentially expressed genes between control and eEF1A1b +/− XY (NDGs), eEF1A1b +/− XY up-regulated genes (URGs), and eEF1A1b +/− XY down-regulated genes (DRGs). 'FDR > 10 −2 ' and '− 1 < log2 |(control-XY_RPKM/eEF1A1b +/− XY_RPKM)| < 1' were used to identify NDGs, 'FDR > 10 −2 ' and log2 (control-XY_RPKM/eEF1A1b +/− XY_RPKM) > 1 were used to identify DRGs, and 'FDR > 10 −2 ' and (control-XY_RPKM/eEF1A1b +/− XY_RPKM) < − 1 were used to identify (URGs). To study the biological pathways of the up/down-regulated genes involved, we mapped these differentially expressed genes to pathways in the Kyoto Encyclopedia of Genes and Genomes (KEGG) using KOBAS web server (http://kobas.cbi.pku.edu.cn/). Transgene mediated rescue of spermatogenesis in eEF1A1b +/− XY fish. In vivo transgenic overexpression of the eEF1A1b in eEF1A1b +/− XY individuals was performed according to the methods of our previous study 49 . An expression plasmid for eEF1A1b was prepared as follows. Complementary DNA encoding the ORF of eEF1A1b was amplified by PCR with a primer set introducing the BamHI and EcoRI sites (Supplemental Tab. 2). The amplified fragment was digested by BamHI and EcoRI and ligated into the multiple cloning sites downstream of the cytomegalovirus (CMV) sequence of the pIRES-hrGFP-1a vector (Stratagene, La Jolla, CA). The prepared construct was microinjected into eEF1A1b +/− XY fertilized eggs at the one-cell stage. Genomic DNA was extracted from injected fish and used to screen transgenic fish at 90 dah. DNA fragments were amplified using gene specific primers (Supplemental Tab. 2). The testis of the injected fish was examined by monitoring the GFP signal, and later the testis was subjected to both histological and immunohistochemical analyses. Antibodies against Ph3, Vasa and Cyp11b2 were diluted as described above. Serum 11-KT level was measured using the EIA Assay kits as described above.