Introduction

Spotted sea bass (Lateolabrax maculatus), as a eurythermic and euryhaline fish species, is widely distributed along the Chinese coasts, extending south to borders between China and Vietnam and north to the southeastern coast of South Korea (Wang et al. 2016a). With high nutritional values and a pleasant taste (Chen et al. 2019b), L. maculatus has become one of the most popular mariculture fish in China, and the production of this fish has reached 156,000 tons a year (MOA 2018). However, the lack of selective breeding of spotted sea bass leads to the degeneration of genetic characteristics, such as the decline in the growth rate and decreased disease-resistant ability (Wang et al. 2017). In addition, the long-term generation interval (3–4 years) hindered the progress of traditional breeding in this fish species. Marker-assisted selection (MAS), the selection method based on DNA markers that are tightly linked to quantitative trait loci (QTL) for traits of interest, has been suggested to be an accurate and efficient way to improve traits that are difficult to select by traditional breeding programs (Zhu et al. 2019). A number of QTL studies on various economic traits, including growth (Wang et al. 2019a, c), sex determination (Wang et al. 2019b; Zhou et al. 2020), disease resistance (Kong et al. 2019; Wu et al. 2019; Zhang et al. 2020), stress tolerance (Li et al. 2017; Jiang et al. 2019), and body color (Li et al. 2019a), have been successfully conducted for aquaculture fish species; however, little work has been performed for spotted sea bass. In the last 2 years, large-scale genomic resources of spotted sea bass have become available, including whole-genome sequencing data (Shao et al. 2018; Chen et al. 2019a), transcriptome databases (Tian et al. 2019c; Shen et al. 2019; Zhou et al. 2019a; Cai et al. 2020), and studies of functional genes (Wang et al. 2018; Zhang et al. 2019a, b, d, e; Liu et al. 2019; Tian et al. 2019a, b; Li et al. 2019b; Fan et al. 2019a, b; Zhou et al. 2019b). However, the linkage map of spotted sea bass has not been fully developed, and this map is essential for the identification of QTLs of economically important traits (Feng et al. 2018) and further facilitates MAS breeding programs.

Growth is the most economically important trait affecting aquaculture fish and exerts a direct influence on production. Combining the high-density linkage map with related phenotypic data, several growth-related QTLs and genetic markers have been identified in many aquaculture species, such as Asian sea bass (Lates calcarifer) (Wang et al. 2015), common carp (Cyprinus carpio) (Peng et al. 2016), bighead carp (Hypophthalmichthys nobilis) (Fu et al. 2016), crucian carp (Carassius auratus) (Liu et al. 2017), mandarin fish (Siniperca chuatsi) (Sun et al. 2017), pikeperch (Sander lucioperca) (Guo et al. 2018), yellow drum (Nibea albiflora) (Qiu et al. 2018), largemouth bass (Micropterus salmoides) (Dong et al. 2019), snapper (Chrysophrys auratus) (Ashton et al. 2019), channel catfish (Ictalurus punctatus) (Zhang et al. 2019c), Pacific white shrimp (Litopenaeus vannamei) (Huang et al. 2020), and Takifugu (Takifugu bimaculatus) (Shi et al. 2020). In addition, with the availability of whole-genome annotation for many fish species, several candidate genes for growth-related traits have been characterized; for example, in L. calcarifer, six growth-related QTLs were detected, and ACOX1 was considered a vital candidate functional gene (Wang et al. 2015). In C. auratus, eight QTLs were identified for body weight, and five potential genes, including EGF-like domain, immunoglobulin-like, C2H2 zinc finger, TGF-beta, and protein kinase (ATP binding site), were detected in the candidate regions (Liu et al. 2017). A total of six candidate QTLs were presented in I. punctatus for growth-related traits, in which three growth-related genes were identified, including megf9, npffr1, and gas1 (Zhang et al. 2019c). For C. carpio, fourteen QTL regions were detected for body weight, while four QTLs were identified for body length. Important regulators, such as KISS2, IGF1, SMTLB, and NPFFR1, were regarded as candidate genes for both growth traits (Peng et al. 2016). The growth-related genetic markers and candidate genes generated based on QTL mapping provide a useful basis for fish MAS breeding programs. However, further genetic studies of economically important traits of spotted sea bass lag behind.

Accordingly, the objectives for this study are (1) constructing the first high-density linkage map for L. maculatus, (2) performing fine QTL mapping for growth-related traits, including body weight and body length, and (3) identifying a list of candidate SNP markers and genes associated with growth-related traits. Our results provide a valuable resource for elucidating the genetic mechanisms of growth and accelerating the genetic breeding of L. maculatus.

Materials and Methods

Ethics Statement

All experimental procedures involving the fish were conducted in accordance with approved guidelines of the respective Animal Research and Ethics Committees of Ocean University of China (Permit Number: 20141201). The present study did not include endangered or protected species.

Resource Family and DNA Extractions

L. maculatus adults were collected from Qingdao, Shandong, China, and considered candidate broodstock for establishing mapping families. A total of 7 candidate full-sib families were produced in November 2016 through mating 7 sires and 7 dams by artificial fertilization. F1 larval fish of those families were raised at Shuangying Aquatic Seedling Co., Ltd., Lijin, Shandong, China. After 1 year of cultivation, one family exhibiting high within-family phenotypic variation in growth was selected for linkage map and QTL analyses. In total, 333 1-year-old progenies were chosen for linkage map construction. The body weights and body lengths of these fish were measured and recorded individually for subsequent growth-related QTL analyses (Supplementary Table S1). A piece of pectoral fin of parents and progenies was separately sampled and stored in 95% ethanol for DNA extractions. Following a traditional phenol-chloroform protocol (Zhan et al. 2007), genomic DNA of each sample was obtained using the CTAB method and purified with RNA digestion (TaKaRa). The concentration and quality of genomic DNA were determined by a Biodropsis BD-1000 nucleic acid analyzer (OSTC, Beijing) and electrophoresis in 1% agarose gel. Finally, high-quality gDNA was prepared for further sequencing.

2b-RAD Library Construction and Sequencing

2b-RAD libraries were prepared for two parents and 333 progenies following the protocol (Wang et al. 2016b). Briefly, the genomic DNA from each individual was first digested with BsaXI (New England Biolabs); then, the digestion products were ligated to adaptors, and the ligated fragments were amplified via polymerase chain reaction (PCR). Finally, the purified amplification products were further digested to generate cohesive ends and then ligated in a predefined order to produce five concatenated tags for paired-end sequencing. Pooled sequencing was carried out on an Illumina HiSeq Xten platform at Qingdao Oebiotech Co. Ltd.

Data Filtering and Genotyping

After sequencing, these raw paired-end reads were first filtered to remove reads that had N bases (ambiguous bases) greater than 8% of the total bases, and reads over 15% of the length were less than Q30. The paired-end reads were merged by Pear software (Version 0.9.6). The merged reads were processed using a custom Perl script to trim adaptor sequences and the terminal 3-bp bases. The remaining five-tag reads were divided into single-tag reads using a Perl script and then assigned to each individual via tag position and length. Finally, clean reads with a length of 27 bp were used for genotyping using the RADtyping software package (Fu et al. 2013). Raw SNP data were filtered as follows. (1) The SNP sites with one or four base types were removed. (2) The sites with minor allele frequency < 0.05 were discarded. (3) The sites with less than 80% of progeny samples being genotyped were filtered. (4) The tags containing more than one SNP were discarded. (5) SNPs with nn × np (heterozygous in female parent), lm × ll (heterozygous in male parent, and hk × hk (heterozygous in both parents) segregation patterns were retained. The SNP annotations were performed using SnpEff software (version 4.1) (Cingolani et al. 2012) with the genome file of L. maculatus (ASM402866v1).

Linkage Map Construction

The Mendelian segregation pattern of each SNP was examined using χ2 tests. Both female and male linkage maps were constructed under a logarithm of odds (LOD) threshold of 5.0 using JoinMap 4.1 (Van Ooijen 2006). Recombination rates were calculated by the regression mapping algorithm and were converted into map distances in centimorgans (cM) using the Kosambi mapping function. The sex-averaged map was produced by integrating the two sex-specific maps using MergeMap software (http://www.mergemap.org/). All genetic linkage maps were drawn via MapChart v 2.3 software (Voorrips 2002).

QTL Analyses

Combining phenotype data with the high-density linkage map, QTL mapping for growth-related traits was performed with MapQTL version 6.0 (Van Ooijen and Kyazma 2009). The LOD scores of each LG were first analyzed using the interval mapping method, and then, permutation tests were performed to confirm the significant thresholds at linkage group-wide and genome-wide with 1000 replicates at a confidence interval of 95%. According to the reference genome of L. maculatus (ASM402866v1), potential candidate genes were extracted within the ± 50-kb genome regions surrounding these significant SNPs.

Results

2b-RAD Sequencing

A total of 2.26 billion raw reads were produced containing the female parent (9.20 million reads), the male parent (9.20 million reads), and their 333 progenies (averaged 6.74 million reads per progeny). After filtering, 2 billion clean reads were adopted for further analysis. The averaged sequencing depths were 27.42-fold for parents and 20.80-fold for progenies. The averaged mapping rates for all samples were more than 88% (Table 1). The raw read data were submitted to the NCBI Sequence Read Archive (SRA) with Accession Number PRJNA580292.

Table 1 Summary of data filtering results of Lateolabrax maculatus
Full size table

SNP Filtering and Linkage Map Construction

A total of 7,334,449 SNP markers were initially obtained from two parents and 333 progenies. After step-by-step filtering, finally, only 11,360 SNP markers were used to construct the linkage map (Supplementary Table S2). Among these markers, three segregation types were classified, including nn × np, lm × ll, and hk × hk. These two sex-specific maps were primarily constructed and then integrated to produce the sex-averaged linkage map (Fig. 1; Tables 2 and 3). A total of 24 LGs were detected, which was consistent with the haploid chromosome number of L. maculatus (Sha et al. 2003). The total length of the female-specific linkage map was 1929.90 cM with an average marker interval of 0.60 cM. The male-specific map length was 1433.61 cM, and its average marker interval was 0.54 cM. The average ratio of female-to-male genetic length was 1.35:1.

Fig. 1
figure 1

Genetic length and marker distribution of 24 LGs in the linkage maps of Lateolabrax maculatus. a–c The male-specific map, the female-specific map, and the sex-averaged map, respectively. The scaleplate on the left indicates genetic distance (cM as unit). The warmer colors indicate a higher density of markers

Full size image
Table 2 Summary of sex-specific linkage maps of Lateolabrax maculatus
Full size table
Table 3 Summary of the sex-averaged linkage map of Lateolabrax maculatus
Full size table

After integration, the sex-averaged map comprised 6883 markers spanning 2189.96 cM. The average marker interval was 0.33 cM, which ranged from 0.23 cM (LG10 and LG12) to 0.50 cM (LG20) (Table 3). The details about the sex-averaged map and sex-specific maps are presented in Supplementary Table S3.

QTL Mapping of Body Weight and Body Length

A total of 24 QTLs comprising 318 significant SNPs for growth traits were identified on LG1, LG3, LG5, LG10, LG11, LG15, LG20, and LG21 (Fig. 2; Table 4). The phenotypic variance explained (PVE) values varied from 5.1 to 8.6%, with LOD scores ranging from 3.75 to 6.47. For body weight, 13 QTL regions were detected with 5.6 to 8.6% PVE, and qBW-11.1 was identified as only one genome-wide significant QTL (Fig. 3; Table 4). For body length, 11 QTL regions were mapped to 8 different LGs, with PVE ranging from 5.1 to 6.2% (Table 4). These significant QTLs were further amplified to obtain candidate genes for growth traits of L. maculatus (Fig. 3; Supplementary Table S4).

Fig. 2
figure 2

QTL mapping of growth traits of body weight and body length in Lateolabrax maculatus. At the threshold of p < 0.05 (linkage group-wide), 13 QTLs were identified for body weight, and 11 QTLs were characterized for body length. At the threshold of p < 0.01 (linkage group-wide), 5 QTLs were identified for body weight distributed on LG1, LG10, and LG11, and 4 QTLs for body length were solely identified on LG20. Only one QTL on LG11 for body weight reached the genome-wide significance of p < 0.05

Full size image
Table 4 Summary statistics of the candidate QTLs for growth traits in Lateolabrax maculatus
Full size table
Fig. 3
figure 3

Regional amplification of candidate regions for body weight and body length on LG1 (a), LG10 (b), LG11 (c), and LG20 (d). Genes are extracted in the ± 50-kb genome regions surrounding the significant SNP. Peak marker positions within each candidate region are noted with red arrows. The red dashed line indicates linkage group-wide significance at the 1% level. The black dashed line represents linkage group-wide significance at the 5% level. The blue line indicates genome-wide significance at the 5% level

Full size image

Candidate Genes Identified for Growth Traits

According to genome annotations of L. maculatus, 30 growth-related candidate genes were identified within the detected candidate QTL regions (Table 5; Fig. 3). The potential functions of the genes involved cell adhesion (col4a3, col4a4, pcdhgc5, and pcdh10); cell proliferation, differentiation, and migration (fgfr4, fgf12a, fgf18, prkca, smyd5, foxq1, foxf2, foxc1a, hes6, gpc1b, gpc5c, gpr55, rab11fip5, and pax3a); cytoskeleton reorganization (diaph1, mylk4a, and mid1ip1l); calcium channel (cacng1a, cacng4a, and cacng5a); and neuromodulation (cadm1a, gfra4b, zdhhc9, nyap2b, sytl4, and per2).

Table 5 Summary of candidate genes and their potential functions for growth traits in Lateolabrax maculatus
Full size table

Notably, the fgfr4 gene was determined to have the highest significance for growth traits in our study (Fig. 3), which is reported to be necessary for both bone growth and myogenesis (Cinque et al. 2015, 2016; Lagha et al. 2008; Mok et al. 2014; Zhao et al. 2006). The FGF18/FGFR4 signaling pathway is presented in Fig. 4a and was recently demonstrated to play an important role in chondrocyte autophagy during postnatal bone growth in mammals (Cinque et al. 2015, 2016). In addition, the myogenic programs (proliferation and differentiation) were jointly modulated by Pax3, Fgfr4, and Fgf18 expressions, while muscle regeneration was accomplished by activating the MyoD-Tead2-Fgfr4 pathway in mammals (Fig. 4b) (Lagha et al. 2008, Mok et al. 2014, Zhao et al. 2006). The detailed functional mechanism of the genes is described in “Discussion.”

Fig. 4
figure 4

Schematic diagram showing the function of identified candidate genes for growth, including FGFR4, FGF18, and PAX3, in a bone growth and b myogenesis (modified from Cinque et al. (2015) and Lagha et al. (2008)). a During postnatal bone growth, FGF18 induces chondrocyte autophagy through the VPS34-beclin-1 complex via the activation of FGFR4 and JNK kinase. b PAX3 orchestrates muscle stem cells into the myogenic program by direct activation of MYF5 and FGFR4 expressions. FGF18 affects the transcription of MYOD and PAX3 to regulate myogenesis. The MYOD-TEAD2-FGFR4 pathway is important for effective muscle regeneration

Full size image

Discussion

A linkage map is recognized as the basis for genomic studies, exploration of QTLs and further MAS breeding. The construction of the first linkage map for economically important fish was performed in 1998 in rainbow trout (Oncorhynchus mykiss) (Young et al. 1998), and along with sequencing technological development, high-resolution genetic maps containing thousands of SNPs have been constructed in more than 30 aquaculture species to date (Zhu et al. 2019). In the present study investigating spotted sea bass, the first high-density linkage map was constructed by sequencing 333 individuals using 2b-RAD technology. A total of 6883 SNP markers were anchored onto 24 linkage groups with an average marker interval of 0.33 cM. The length of the L. maculatus female map was longer than the male map with a ratio of 1.35:1. In species with the XY sex determination system, the female map is usually longer than the male map (Chistiakov et al. 2006). This phenomenon has been reported in several teleosts with XY sex determination system, such as medaka (Oryzias latipes) (Kondo et al. 2001), C. carpio (Zhang et al. 2011), I. punctatus (Li et al. 2015), and C. auratus (Liu et al. 2017).

Genetic markers can be used to directly select promising breeding populations, and QTL mapping provides an effective approach to identify these molecular markers (Zhu et al. 2019). In this study, 24 growth-related QTLs were identified on eight LGs, including 13 QTLs for body weight and 11 QTLs for body length. Notably, 6 QTL intervals were shared in both growth traits (Fig. 2), indicating that a higher phenotypic correlation occurred between body weight and body length and that one trait selection may result in the improvement of another growth trait in the breeding of spotted sea bass (Qiu et al. 2018). Furthermore, 318 growth-related SNPs and 30 candidate genes were provided for further MAS breeding of L. maculatus. These candidate genes serve essential functions in multiple growth-related procedures, such as cell adhesion, cell proliferation, cytoskeleton reorganization, calcium channels, and neuromodulation.

Our results showed that the fgfr4 gene was detected in the candidate region with the highest significance; this gene mediates cell proliferation and cell differentiation and is prominently expressed in avian developing skeletal muscles (Shin and Osborne 2009; Marics et al. 2002). A genome-wide association study (GWAS) also showed that FGFR4 variations may influence human height (Lango Allen et al. 2010). As a ligand of Fgfr4 (Falvella et al. 2009), Fgf18 plays an essential role in chondrogenesis and osteogenesis in mammals (Ohbayashi et al. 2002) and is also regarded as an important regulator in GWAS analysis of growth traits in large yellow croaker (Larimichthys crocea) (Zhou et al. 2019c). In mammals, the FGF18/FGFR4 pathway is identified as a novel effector of FGF signaling via controlling chondrocyte autophagy to maintain postnatal bone growth (Cinque et al. 2015, 2016). Another candidate gene of fgf12a was also identified through QTL analysis. This result suggests that FGF signaling may play an important role in the growth of L. maculatus. Furthermore, Pax3 regulates myogenesis through direct activation of both the Fgfr4 gene and the myogenic determination gene of Myf5 (Lagha et al. 2008); meanwhile, Fgf18 signaling can upregulate another differentiation gene of MyoD and has a potential effect on Pax3 expression (Mok et al. 2014; Mohammed et al. 2017). Moreover, the MyoD gene may activate Fgfr4 expression through the MyoD-Tead2-Fgfr4 pathway to regulate muscle regeneration in mice (Mus musculus) (Zhao et al. 2006). However, the studies of these genes in teleost growth are comparatively limited, with only a few species, such as zebrafish (Danio rerio) (Thisse et al. 1995) and olive flounder (Paralichthys olivaceus) (Jiao et al. 2017), being investigated.

Neuromodulation may be a major factor explaining the higher growth performance of fish MAS breeding (Su et al. 2018). Body weight is determined by a balance between food intake and energy expenditure, which is regulated by multiple neural circuits (Rui 2013). Several neuromodulation genes identified in our study have been reported to exert essential roles in related growth processes. The cadm1a gene regulates M. musculus body weight and energy homeostasis by mediating synaptic assembly, which has also been identified in GWAS analysis of body mass index (BMI) in humans (Locke et al. 2015). The Gfra4 gene is expressed in the neural crest, which is necessary for endocrine cell development in M. musculus (Lindahl et al. 2000). The Zdhhc9 gene plays an important role in dendrite outgrowth and inhibitory synapse formation (Shimell et al. 2019), which is also identified as a vital candidate gene for growth-related QTLs in C. carpio haematopterus (Feng et al. 2018). Additionally, the DIAPH1 protein is expressed in human neuronal precursor cells, and patients with a homozygous loss of the DIAPH1 gene suffer from microcephaly and reduced height and weight (Ercan-Sencicek et al. 2015). The important roles played by neuromodulation in growth control are specifically discussed in the genetic breeding of C. carpio (Su et al. 2018).

Cell adhesion plays important roles in cell-cell communication and the development and maintenance of tissues (Khalili and Ahmad 2015). The candidate genes of col4a3 and col4a4 are detected in the basement membrane of Prochilodus argenteus, which is important for maintaining normal cell adhesion to the extracellular matrix (Thomé et al. 2010). The pcdh10 gene, as a cell adhesion factor, exerts vital effects on paraxial mesoderm development and somitogenesis in D. rerio (Murakami et al. 2006). However, further studies are needed to corroborate the roles of cell adhesion genes in L. maculatus growth regulation. For calcium channels, these candidate gene functions are focused on bone resorption via osteoclasts, which is necessary for skeletal development in teleosts (Witten et al. 2000; Zayzafoon 2006).

Conclusion

In this study, the first high-density linkage map of L. maculatus was constructed, which contained 6883 SNP markers spanning 2189.96 cM. For body weight and body length QTL mapping, 24 significant QTLs, including 318 SNPs and 30 candidate genes, were identified. These genes are involved in cell adhesion, cell proliferation, cytoskeleton reorganization, calcium channels, and neuromodulation. Our results provide a useful framework for determining the genetic basis for growth traits in spotted sea bass and establish a foundation for further molecular-assisted breeding programs for this species.