LI Yangchun, WANG Zhaoping, CUI Yuting, MA Peizhen, ZHANG Xuekai, and FAN Chao

Key Laboratory of Mariculture of Ministry of Education, Ocean University of China, Qingdao 266003, China

Abstract Polyploid breeding is widely used in various marine species. Low salinity treatment is an effective method of inducing triploid of bivalve mollusks. In this study, RNA-seq was performed to determine genes and pathways involved in hyposaline adaption and cell division of Pacific oyster (Crassostrea gigas) zygotes, trying to better understand the possible molecular mechanism of hypo-osmotic induction. A total of 26965 unigenes were generated in the de novo assembly of clean Illumina reads with an average length of 934 bp and N50 of 1721 bp. Of 3024 differentially expressed genes (DEGs), 2501 were up-regulated and 523 were downregulated. GO (Gene Ontology) annotation and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analysis of these DEGs revealed that these DEGs participate a variety of biological processes including osmoregulation, cytoskeleton organization,cell survival and death, and substantially modulate cell proliferation and embryonic development. In summery, RNA-seq methodology was applied for the first time to demonstrate hypotonic-induced transcriptomic alteration in oyster zygotes. Our findings not only interpreted the relatively high mortality of induced larvae, but also provided a valuable reference for further investigations on the mechanism of hyposaline induction, thus should aid to the application of low salinity in triploid induction in large scale aquaculture in future.

Key words RNA-seq; Crassostrea gigas; zygote; triploid induction; low salinity

1 Introduction

As one of the most commercially important shellfish,the Pacific oyster (Crassostrea gigas) usually accounts for a majority in aquaculture worldwide (FAO Yearbook.Fishery and Aquaculture Statistics 2016, http://www.fao.org/fishery/publications/yearbooks/en, accessed on November 5, 2018). Polyploid induction has been proved to be an effective method of generating triploid oysters with the advantages of fast-growth, large size, sterility-induced low mortality and improved meat quality during spawning season (Handet al., 2004; Normandet al., 2009; Piferreret al.,2009; Jeunget al., 2016). Moreover, compared to diploid,triploid oysters are less harmful to native cultivated species due to their partial sterility and thus polyploid induction is proposed as a feasible method of avoiding genetic pollution (Hulata, 2001).

Crossing tetraploid and diploid broodstock is considered as the most effective way toward producing large scale triploid oyster (Wanget al., 2005) while polyploid induction is still necessary for species without commercial tetraploid lines. Various methods have been shown to be effective for polyploid induction in aquatic species, which include chemical treatment with 6-dimethylaminopurine(6-DMAP) (Gerardet al., 1994), cytochalasin B (Nellet al.,1996) and caffeine (Okumuraet al., 2007) and physical treatment through either heat shock (Zhanget al., 2003)or cold shock (Hamasakiet al., 2013). However, most of these methods have significant shortcomings due to their high cost, poor efficiency and high toxicity or carcinogenicity, thus not popularized in large scale aquaculture.Recently, hyper- and hypo-osmotic treatments were demonstrated to be effective polyploid inducers with the advantages of low cost, high convenience, stable triploidization rate and reliable food safety (Maet al., 2018).

Obstructions in energy supply, spindle assembly and constricting ring formation are considered to be probable mechanisms of polyploid induced with chemical and physical methods (Ghoshet al., 1978; Rimeet al., 1989; Zhanget al., 2003; Linet al., 2016). However, the possible mechanism of hyposalinity-induced polyploid remains largely unexplored. As sessile euryhaline species inhabiting intertidal environment, oysters continually experience extreme salinity sometimes its rapid fluctuation because of heavy rains and tidal inundations (Yuanet al., 2000). Therefore, oysters have a relatively intricate response system to cope with sudden ambient stress. Previous studies carried out in larval, juvenile and adult oysters have elicited reactions at both individual and molecular levels such as feeding and respiratory physiology, immune repression and regulation of ion channels and free amino acids (Zurburg and Zwaan, 1981; Saràet al., 2012; Yanet al., 2017). Considering that hyposaline induction of triploid is achieved by interfering cell division after the release of PB1 (the first polar body), a zygote, as a single cell at an early embryonic stage, may exhibit very different responses from an adult organism. Hence, efforts should be undertaken to better understand the underlying molecular changes in oyster zygotes.

Since little attention has been paid to osmotic stress management in zygotes, identifying candidate genes implicated in hypo-osmotic induction is imperative to disclose the underlying molecular basis. The transcriptome, a set of RNA molecules directly reflecting gene expression, is powerful to monitor the dynamic alteration in gene expression and reveal functional genomic elements (Menget al., 2013). With superior time and cost-efficiency, RNAseq is frequently used to quantitatively analyze transcript abundance (Chenet al., 2016). In recent years, transcriptomic analysis of marine animals under hyper- and hyposalinity has identified numerous genes involved in various biological processes and putative modulations conferring tolerance to osmotic stress (Lvet al., 2013; Menget al.,2013; Zhanget al., 2018). Therefore, RNA-seq was pursued to provide valuable information for unraveling the molecular mechanism underlying osmotic induction of polyploidy oyster zygotes.

In this study, high throughput Illumina RNA-seq technology was used for the first time to investigate low salinity-induced transcriptomic change inC. gigaszygotes.We utilized the treatment condition of salinity 8 for a duration of 15 min, which has been proved to be effective.Such treatment yields a triploidization rate as high as 91.0%(Konget al., 2011). In our study, we presented insights into candidate genes and pathways involved in osmoregulation, cell division and embryonic development, which are important for interpreting the relatively high mortality of induced larvae and for future research on the mechanism of hypotonic triploid induction.

2 Materials and Methods

2.1 Experimental Animals, Treating Conditions and Ploidy Determination

Two-year-old oyster individuals, 40 in total, shell length 8 – 15 cm, were collected from Yantai, Shandong, China. Natural sea water (22℃, salinity 30) was filtered to eliminate interfering organisms. Sperms and eggs were stripped from parental oysters and the quality of gametes was determined by microscopic examination. Females with oval/dense eggs and males with high-motility sperms were selected as candidate parents. Twelve individuals (nine females and three males) were selected and separated randomly into three biological replicates, one male × three females each. Eggs of females were mixed and fertilized with sperms from male. Half of the pooled zygotes were then treated for triploid induction and the other half was used as control. Gamete handling and fertilization were carried out according to Zhanget al. (1998) with slight modifications. Briefly, in each group, oocytes from three females were released by the gonad and then mixed. The pooled eggs were filtered through a 80 μm sieve net to remove large debris, then through 20 μm membrane to retain the oocytes, and finally immersed in fresh sea water for 40 – 50 min for further maturation. Supernatant containing immature eggs was discarded and sea water was added to reach an egg density of 5000 – 7000 mL?1. Sperms were collected and activated for 10 min before fertilization.

The inducing conditions were described early. When about 40% – 50% first polar body appeared, the fertilized eggs were placed at salinity 8 and temperatures ranging from 20 to 22℃ for 15 min (Konget al., 2011). Fertilized eggs treated in normal sea water (salinity 30, 20 – 22℃) for 15 min were collected as control.

After induction, half of the zygotes were collected and immediately transferred to liquid nitrogen. The rest were transferred into 5 L seawater and incubated at 20 – 22℃ for about 24 h. The newly hatched D-larvae were collected for ploidy determination on a Flow Cytometry (FC500 MPL,Beckman Coulter, USA) after staining with DAPI (Sigma,USA) (Normandet al., 2009). D-larvae of the control group were set as diploid standard for ploidy determination.

2.2 RNA Extraction, Library Construction and Transcriptome Sequencing

Total RNA was isolated using Trizol buffer. Samples each replicates were grounded to a fine powder in liquid nitrogen separately. Extracted RNA was then treated with RNase-free DNase followed by 1% agarose gel electrophoresis to verify the integrity (Yanet al., 2017). Qualified total RNA run on a denaturing gel will have two sharp rRNA bands (28S and 18S), of which the 28S rRNA band should be approximately twice as intense as the 18S rRNA. Purity and concentration of the extracted RNA was performed on Nanodrop 2000 spectrophotometer (Thermo, USA) and a 2100 Bioanalyzer (Agilent Technologies,USA), respectively (Caoet al., 2018).

An equal amount of RNA (1.5 μg) each sample was utilized for cDNA library construction using a TransCript?cDNA sample prep kit (TransGen Biotech, China). Pairend RNA sequencing was performed on the Illumina HiSeqTM 2500 (2×150 bp read length, San Diego, CA, USA)platform by Gene Denovo Biotechnology Co., Ltd. (Guangzhou, China).

2.3 De novo Assembly, Functional Annotation and Differentially Expressed Genes Analysis

Raw reads from RNA-seq were filtered to remove lowquality ones, of them the number of bases with Q ≤ 20 accounts for more than 40% of the total read length. Additionally, reads with more than 10% poly-N, residual rRNA(sequences that match to the ribosomal RNA database),and adapter sequences (adapters that are ligated to short fragments of DNA for Illumina libraries construction) were also be filtered before assembly. Clean reads were thende novoassembled using Trinity assembly program (Zhanget al.,2017), and the generated sequences were matched against NCBI protein non redundant (nr), Swiss-Prot, KOG (eu-Karyotic Orthologous Groups) and KEGG (Kyoto Encyclopedia of Genes and Genomes) databases with a threshold ofE-value ＜ 1e-5 (Zhanget al., 2017). Blast2go software was employed for Gene Ontology (GO) enrichment,and KEGG analysis was performed for pathway analysis.

All sequences that matched a single reference gene were clustered and only the longest one each cluster was considered as the ‘unigene’ for subsequent annotation. The abundance of unigenes was calculated with RPKM (reads per kb per million reads) method (Wanget al., 2018), and log2(Fold change) ＞ 1.28 or ＜ 0.83 andP-value ＜ 0.05 (Menget al., 2013; Zhanget al., 2016) were set as the thresholds of differentially expressed genes (DEGs). DEGs were thereafter annotated against GO and KEGG databases in functional analysis.

2.4 Real-Time Quantitative PCR Analysis

qRT-PCR was used to validate 15 randomly selected DEGs (Lvet al., 2013; Menget al., 2013; Zhanget al.,2016). One μg of total RNA (isolated from the samples used in RNA-seq) was transcribed into cDNA using Fast-Quantity RT Kit (TIANGEN, Beijing, China) according to the manufacturer’s instructions. Elongation factor 1 alpha (EF1α) gene was used as internal control (Yanet al.,2017). Details of cDNA-specific primers are listed in Table 1. qRT-PCR was conducted on an ABI StepOnePlus?(Applied Biosystems, USA) with the following reaction conditions: 95℃ for 90 s followed by 40 cycles of 95℃for 5 s, 60℃ for 15 s and 72℃ for 20 s. Relative expression of unigenes was calculated with the 2???CTmethod (Livak and Schmittgen, 2001) and Log2qRT-PCR (control/treatment) each genes was calculated for comparison with the RNA-seq results.

Table 1 Primers of qRT-PCR

3 Results

3.1 Ploidy Determination

The average hatching rates of control and treated groups were 86.00% ± 2.85% and 48.10% ± 12.11% (Mean ± SE,the same below); the average triploid rate of the three replicates was 71.55% ± 11.88%. Fig.1 shows the FCM results of one replicate. There were 13894 particles examined, of them 10527 (75.77%) possessed 1.5 times fluorescence intensity compared to the diploid and were determined as triploid (3N), indicating that the segregation of the chromosomes of most zygotes was significantly disturbed and the transcriptomic analysis should reflect the possible molecular changes during hypotonic induction.

Fig.1 Ploidy determination. The fluorescence intensity of the control group was used to determine the position of 2N particles. FL3, detector for DAPI staining; Number,number of detected pellets; HP X-CV, coefficient of variation which reflects the concentration of peaks.

3.1 Illumina Sequencing and de novo Assembly

In total, 326 million raw reads were obtained using 150 bp pair-end Illumina sequencing, of them 312 million clean reads (95.72%) were obtained (as shown in Table 2).The mapping ratio of clean reads (upon elimination of reads mapping to rRNA) was over 94.78% (Table 2) while the correlation coefficients between biological replicates measured by Pearson analysis were greater than 0.88. A total of 26965 unigenes were obtained with an average length of 934 bp and N50 of 1721 bp. Sequences and raw reads were deposited to the NCBI Sequence Read Archive(SRA) database under the accession number PRJNA50 7322.

3.2 Annotation of Unigenes and Analysis of Differentially Expressed Genes

A total of 17473 unigenes (64.80%) were annotated against at least one database atE-value ＜ 1e?5.Crassostrea gigaswas the most similar species to the assembled zygotes genome (89.0%), followed byLingula anatina(0.76%)andAplysia californica(0.69%).

Functional category classification aligned the unigenes to 25 KOG categories, of them ‘signal transduction mechanisms’ was the most highly represented (16.46%), which was followed by ‘general function prediction only’ (16.00%)and ‘posttranslational modification, protein turnover, chaperones’ (10.61%).

All the unigenes were assigned into three categories based on GO analysis: cellular component (CC), molecular function (MF) and biological process (BP). The most enriched terms are listed in Table 3 and the level 2 GO term enrichments are depicted in Fig.2. The top two enriched GO terms of CC, MF and BP were cell (GO:0005623), cell part (GO: 0044464); binding (GO: 0005488),catalytic activity (GO: 0003824), cellular process (GO:0009987), and metabolic process (GO: 0008152) respectively, which were in accordance with the findings of several previous studies (Zhanget al., 2016; Zhanget al.,2017; Yanet al., 2018; Zhanget al., 2018), indicating that these categories were common annotation terms of marine organisms under ambient stress.

A total of 3024 unigenes differentially expressed between the low salinity and the control groups, of them 2501 were up-regulated and 523 were down-regulated.The top 10 up- and down-regulated DEGs are listed in Table 4. DEGs were enriched in various pathways implicated in nearly all levels of biological regulation including energy metabolism, DNA replication and repair, transcription and translation (Table 5), apoptosis and cell survival (Table 6), cytoskeleton regulation (Table 7) and osmoregulation (seeing details in Discussion).

Table 2 Description of RNA-seq data

Fig.2 Level 2 GO term enrichment of transcripts.

Table 3 The top 5 enriched GO terms of each category

Table 4 The top 10 up- and down-regulated DEGs with annotated symbol and description

Table 5 KEGG pathway analysis of DEGs-partial

(continued)

Table 6 Details of DEGs in the discussion section-partial

3.3 Validation of RNA-seq Data by qRT-PCR

Fifteen DEGs were randomly selected for validation by qRT-PCR analysis. Details of each selected genes is presented in Table 1 and quantitative results are shown in Fig.3.Nearly all the validated DEGs showed consistent trends(up- or down-regulated, except Gga3 which did not change obviously) compared to the RNA-seq data, and the Pearson correlation index was 0.738, further confirming the reliability and accuracy of the present analysis.

Fig.3 qRT-PCR validation of RNA-seq data. Log2 (Fold change) (CK/S) of qRT-PCR = ?△△Cq; Log2 (Fold change) (CK/S)of RNA-seq = Log2 (RPKM_CK/RPKM_S).

4 Discussion

Since triploid induction at low salinity is believed to be a powerful breeding method for bivalve mollusks, we pursued RNA-seq to unravel the molecular pathways and profiled the gene expression of oyster zygotes under hyposalinity. Various biological processes were found to be modulated and genes and pathways involved in osmotic adaption and cell division were identified. These findings provided an important validation for future investigations on the mechanism of hypotonic induction.

4.1 Typical Stress Rresponse

It is a common phenomenon that fluctuation of abiotic factors (including temperature, salinity) and anthropogenic stress (such as electromagnetic radiation, chemical mutagens) always influence basic life activities such as energy metabolism, apoptosis and cell survival (Menget al.,1998; Mininniet al., 2014; Windischet al., 2014). We observed that these pathways were severely affected in low salinity.

4.1.1 Energy metabolism

More energy is required to maintain cellular homeostasis and osmotic balance under both hypo- and hyper-saline situations (Sangiao-Alvarelloset al., 2003; Xuet al.,2015). This has been shown in many marine species includingLateolabrax maculatus(Zhanget al., 2017),Cynoglossus semilaevis(Siet al., 2018),Litopenaeus vannamei(Zhanget al., 2016). Unlike these early studies, we revealed significant down-regulations of nine energy-implicated KEGG pathways (Table 5) and five GO pathways(GO: 0006091-generation of precursor metabolites and energy, GO: 0015992-proton transport, GO: 0046034-ATP metabolic process, GO: 0022900-electron transport chain, GO:0006754-ATP biosynthetic process) in this study, indicating that energy supply in oyster zygotes is suppressed at low salinity. It is possible that a zygote does not have systems as intricate as a multicellular adult animal does thus can be influenced easily by ambient stress. It is very likely that zygotes respond to hypo-osmotic stress in a different manner from adults.

4.1.2 Apoptosis and cell survival

As an essential regulatory process in numerous biological events such as cell development, immune reaction,cell growth and death, apoptosis is usually induced in extreme ambient situation including temperature and salinity stresses (Yanet al., 2017; Lyuet al., 2018). We observed that the expression of genes encoding caspases 3 and 8, two pivotal apoptotic enzymes, was significantly upregulated. As an initiator, activation of caspase 8 can in turn activates downstream executioner caspases 3, which plays a central role in the execution-phase of cell apoptosis (Guoet al., 2002; Galluzziet al., 2016). Therefore,up-regulation of caspases 3 and 8 reflected a potential motivation for apoptosis.

Interestingly, augmentation in several anti-apoptosis gene transcripts was observed as well. Baculoviral IAP repeat-containing proteins (BIRCs), the crucial anti-apoptosis proteins, can prevent cell death by 1) direct inhibition of caspases 3, 7 and 9; 2) thorough E3 ubiquitin-protein ligase activity which can ubiquitinate apoptotic promoters such as DIABLO/SMAC (Dubrez-Dalozet al., 2008).Our data showed that the expression of two BIRC genes,BIRC5andBIRC7/8, was up-regulated, which might be antagonistic to over-expressed caspase genes. Similar findings were also reported in previous studies on osmotic stress inL. vannamei(Zhaoet al., 2015),Crassostrea gigas(Zhaoet al., 2012) andC. angulata×C. sikameahybrids (Yanet al., 2017).

PI3K/Akt/mTOR signaling pathway is a well-studied pathway regulating cell proliferation and survival (Jankuet al., 2018). Phosphoinositide-3-kinase (PI3K) regulates the generation of phosphatidylinositol 3,4,5-trisphosphate(PIP3), which is essential for the activation of downstream signaling molecules such as Akt and 3-phosphoinositidedependent protein kinase 1 (PDPK1) (Williamset al., 2000).In the present study, four PI3K-related genes (Table 6) and two PI3K/Akt/mTOR signaling pathway genes including insulin receptor (Zhanget al., 2014) and interleukin 2 receptor (Brennanet al., 1997) showed significantly elevated expression, indicating a positive trend for cell survival.

The ubiquitin proteasome pathway (UPP), a major pathway in the catabolism of proteins, is crucial to almost all cellular processes including cell proliferation, apoptosis and DNA damage repair (Schauberet al., 1998). In the last step of proteasome degradation, E3 ubiquitin-protein ligase tags proteins by transferring ubiquitin from the E2 ubiquitin-conjugating enzyme to targeted substrates for subsequent degradation by the 26s proteasome (Kanayamaet al., 1992; Hershko and Ciechanover, 1998). Data collected in this study showed that a total of 25 E3 ubiquitin-protein ligase genes were markedly affected, and 20 were over-expressed (Table 6). Moreover, a notable upward trend was observed in the 26S proteasome subunit P45 family protein (gene ID: Unigene 0003149), further supporting the enhancement of protein ubiquitination and the promotion of cell survival.

Taken together, the findings in energy metabolism, apoptosis and cell survival revealed a complex regulation of cell death and survival, leading to the uncertainty of cell fate. They might be the reason of the fluctuating mortality rates of hyposaline induced oyster zygotes.

4.2 DNA Repair, Transcription and Translation

Despite being a sophisticated biological process, DNA replication is constantly disturbed by a variety of nonbiological factors (Menget al., 1998; Akchaet al., 2016),resulting in damages such as single strand damage and double strand breaks, which are mainly repaired by base excision repair, nucleotide excision repair, mismatch repair and homologous recombination (Lianget al., 2005;Reardon and Sancar, 2006). In this study, all aforementioned DNA repairing types and three other replicationrelevant pathways including fanconi anemia pathway, pyrimidine and purine metabolism (Table 5) were significantly enhanced. The expression of a total of 22 genes (Table 6) assigned to these pathways was up-regulated, implying that DNA replication and recombination were negatively affected under hypotonic condition. Considering the limitation of repair efficiency, uncorrected DNA defect may lead to cell cycle arrest (Houtgraafet al., 2006)and deleterious mutations (Espinoza and Minami, 2018),which may partially account for lagging development and high mortality in induced zygotes.

Simultaneous interference in transcription and translation can be indicated by the significant up-regulation of numerous genes and pathways involved in gene transcription (Table 5). The CCR4-NOT complex is one of the most important mRNA deadenylases that play vital roles in various transcription-linked processes including many repressive pathways such as miRNA-mediated repression and translation repression during translation initiation (Collart and Panasenko, 2012; Zenget al., 2018). Additionally, the over-expression of anti-proliferation BTG/TOB can mediate the inhibition of cell cycle progression, and is the consequence of its interaction with two components of CCR4-NOT complex, CNOT7 and CNOT8 (Doidgeet al.,2012). In this study, genes encoding BTG/TOB and five components of CCR4-NOT complex including CNOT7 and CNOT8) were dramatically over-expressed, displaying an obstruction in cell proliferation and explaining the impeded development of the treated zygotes at the transcriptional and translational levels.

4.3 Genes Associated with Osmoregulation

As euryhaline osmoconformers, oysters possess special abilities to protect themselves against acute drop in salinity caused by intense rains or huge inflow of freshwater in estuaries. It has been elucidated that the modulation of osmolytes, such as free amino acids (FAA) and inorganic ions, is the main contributor to hypo-osmotic adaption in many marine bivalves (Pierce and Amende, 1981; Zhaoet al., 2012). Investigation inPortunus trituberculatusdemonstrated that low salinity stress induces a reduction of FAA levelsviaaugmented amino acid catabolism, attenuated protein degradation and decrease in synthesis of certain non-essential amino acids (Lvet al., 2013). In the present research, biosynthesis of amino acids was dramatically suppressed, however, no significant change was observed in protein degradation or amino acid metabolism.Several studies have put doubts on whether changes in FAAs were rapid enough to resist sudden and shortterm drop in salinity (Pierce and Amende, 1981; Zhaoet al.,2012). Our results indicated that short-term hyposalinity does not provide enough time for zygotes to elicit effective changes in amino acid concentrations to prevent cell swelling and subsequent death.

Another important osmoregulatory gene,aquaporin 1(AQP1), was markedly over-expressed in this study. AQPs are integral membranes which form water-specific channels that facilitate rapid transport of water across membranes (Agre, 2006). AQP1 can osmotically increase the permeability of plasma membrane for water transport and is highly induced in hyper-osmotic conditions (Umenishi& Schrier, 2003). Contrarily, many previous studies have shown that long-term salinity induced down-regulation of AQP genes (Menget al., 2013; Zhanget al., 2017) was attributed to enhanced protein phosphorylation (Tournaire-Rouxet al., 2003). Considering that this study included a treatment for 15 min under a relatively severe osmotic stress of salinity 8. It is very likely that zygotes would exhibit a different response pattern for AQPs.

4.4 Cytoskeleton Regulation

Cytoskeleton proteins which are frequently affected under low salinity are not only engaged in the structural reorganization during hypo-osmotic cell swelling, but also relevant to cell division given their crucial roles in mitosis and meiosis. Microfilaments, microtubules and spectrins,the major and basic constituents of the cytoskeletal network, are critical for the maintenance of cell shape and subcellular organization, as well as essential for spindle and contractile ring assembly. In the present work, a total of 16 actin-, tubulin-, myosin- and spectrin-related genes were notably modulated. In addition, several important regulators of cell division were also affected after hyposaline induction (Table 7), implying a considerable impact on cytoskeletal network and cell division (Blerkom, 1991;Vonget al., 2005; Betapudi, 2010; Collecet al., 2011; Heet al., 2016; Duan and Sun, 2018). Changes in these genes’expressions reflected the possibility that disturbed cell division might lead to the triploid offspring generation.

Table 7 Cytoskeleton regulation- and cell division-related DEGs

5 Conclusions

In the present study, the expression of a total of 3024 genes was significantly altered after a 15 min induction at salinity 8. These genes function in various biological processes including apoptosis, cell survival, osmoregulation and structural reorganization. Our findings not only provided an insight into the gene expression patterns and response networks under hypotonic conditions in zygotes,but also provided, in part, clues to the relatively lower hatching rate and higher mortality of induced larvae, and paved the way for future studies on the mechanisms of triploid induction with low salinity in aquatic species.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (No. 31172403). We are grateful to Drs. Tingting Wang and Zongwu Zhao for their helpful suggestions and writing assistance.