Association of a new 99-bp indel of the CEL gene promoter region with phenotypic traits in chickens

Carboxyl ester lipase (CEL) encodes a cholesterol ester hydrolase that is secreted into the duodenum as a component of pancreatic juice. The objective of this study was to characterize the CEL gene, investigate the association between the CEL promoter variants and chicken phenotypic traits, and explore the CEL gene regulatory mechanism. An insertion/deletion (indel) caused by a 99-bp insertion fragment was shown for the first time in the chicken CEL promoter, and large differences in allelic frequency were found among commercial breeds, indigenous and feral birds. Association analysis demonstrated that this indel site had significant effects on shank length, shank girth, chest breadth at 8 weeks (p < 0.01), evisceration weight, sebum weight, breast muscle weight, and leg weight (p < 0.05). Tissue expression profiles showed extremely high levels of the CEL gene in pancreatic tissue. Moreover, the expression levels of the genes APOB, MTTP, APOV1 and SREBF1, which are involved in lipid transport, were significantly reduced by adding a 4% oxidized soybean oil diet treatment at the individual level and transfecting the embryonic primary hepatocytes with a CEL-overexpression vector. Interestingly, the results showed that the expression level of the II homozygous genotype was significantly higher than that of the ID and DD genotypes, while individuals with DD genotypes had higher phenotypic values. Therefore, these data suggested that the CEL gene might affect body growth by participating in hepatic lipoprotein metabolism and that the 99-bp indel polymorphism could be a potentially useful genetic marker for improving the economically important traits of chickens.


Results
Conserved synteny analysis and cloning of the chicken CEL gene. The CEL gene is located on chromosome 17, contains 11 exons and 10 introns and encodes 556 amino acids (NP_001013015.1) in chickens. We performed conserved synteny analysis on different species, demonstrating that CEL in chicken and other species is positioned at the same locus of a conserved genomic region containing common genes including TSC1, GFI1B, GTF3C5, RGL4, and GBGT1 (Fig. 1). The full-length CDS region was obtained using Lushi blue-shell (LS) chicken pancreas cDNA as a template ( Supplementary Fig. S1). The results of DNA sequence analysis showed that the number of A bases was 419 (25.07%); G, 426 (25.49%); C, 454 (27.17%); T, 372 (22.26%); A + T, 791 (47.34%); and G + C, 880 (52.66%) and that the molecular weight of the protein was 61.1391785 kDa. Bioinformatics analysis results revealed that there is a Pfam Abhydrolase_3 functional domain at amino acids 121 to 230. There is a signal peptide and a possible transmembrane domain from amino acids 5-27 in the CEL protein sequence with the N-terminus located inside the cell and the C-terminus outside. Moreover, there are three potential N-linked glycosylation (Asn-X-Ser/Thr) sites at amino acid positions 207, 270 and 381 in the extracellular domain ( Supplementary Fig. S2). These results illustrate the conservation of CEL gene sequences in different species.

Identification of a novel CEL 99-bp indel polymorphism.
A new 99-bp indel ( NC _0 06 10 4. 4: g. 72 44 45 9i ns CG TGGGATGAGTCGCA C TC TTGTGGTCTCTGCACTGCTCTCAGGCTACGTTTCTCACAGCTGGT TCTGTTCCAGTCCACAGCCTCCAGCTCTCCTCCTG7244460) polymorphism of the CEL gene was revealed by whole genome resequencing of five Xichuan black-bone (XC) individuals. The indel variants, with the respective alleles being I and D, were genotyped by 2.0% agarose gel electrophoresis in different chicken breeds. As illustrated in Fig. 2 (cropped from Supplementary Fig. S3), three genotypes were observed, including genotype II, genotype ID and genotype DD. Allele I is 246-bp, while the length of allele D is 345 bp (Supplementary Fig. S4). Compared with published chicken genome sequences, the PCR product sequencing result suggests that a 99-bp insertion is located in a region 4310 bp upstream of the CEL gene (Fig. 3). This polymorphic sequence has been deposited in the Database of Genomic Variants (DGVa accession number: estd240).
Population genetic analysis of the indel in 10 chicken populations. Table 1 shows the distribution of genotypic and allelic frequencies of 99-bp indels in feral, local and commercial chicken populations, as well as their genetic diversity, which will help clarify when this insertion mutation occurred during chicken domestication. Large differences in allelic frequency were found across feral birds and indigenous and commercial breeds. For the Chinese indigenous breeds, which are characterized by slow growth and a history of undergone nonsystematic selection, we found that the frequency of the insertion allele was approximately 35.20% to 46.20%. In contract, in highly directionally selected commercial chicken breeds we found low insertion allele frequencies (3.00% to 9.90%). The results of a χ2 test showed that the 99-bp indel was in Hardy-Weinberg equilibrium (HWE) (p > 0.05) in the tested chicken breeds. In addition, the difference in the distribution of www.nature.com/scientificreports www.nature.com/scientificreports/   genotypes among commercial breeds was significant (p < 0.01), but nearly no significant difference was observed among the indigenous breeds (Supplementary Table S1). The values of He and Ne were 0 to 0.497 and 1 to 1.988, respectively. According to the classification of PIC, the indigenous chicken breeds exhibit moderate polymorphism (0.250 < PIC < 0.500), whereas commercial chickens exhibit low polymorphism (PIC < 0.250). Pairwise Fst values also indicated a moderate degree of genetic differentiation between commercial and native breeds (0.05 < Fst < 0.26). A higher differentiation value (more than 0.25) was found between Changshun blue-shell (CS) chickens and Ross 308, suggesting that the insertion mutation was highly selected in modern meat-type chickens. Among the commercial breeds, there was little differentiation, and Fst values among indigenous chickens were less than 0.02, indicating no obvious differentiation ( Table 2). These results suggest that this insertional mutation locus may undergo low selection pressure during the evolutionary history of the indigenous breeds and is more strongly selected in recent artificially selected commercial chickens.
A 99-bp insertion is correlated with CEL expression. Tissue expression profile analysis showed extremely high levels of the CEL gene in pancreas tissues, whereas it displayed hardly any expression in other tissues including the heart, liver, spleen, lung, kidney, duodenum, ovary, abdominal fat, and breast muscle (Fig. 4a, cropped from Supplementary Fig. S5). Additionally, tissue expression profiles of four species also showed extremely high CEL gene expression in pancreatic tissue ( Supplementary Fig. S6). During chicken growth, we noticed that expression of the CEL gene first increased and then decreased in pancreas tissue of Arbor Acres (AA) broilers and LS chickens. The expression of the CEL gene was highest at 3 weeks of age and lowest at 0 weeks of age (p = 0.002); Due to the long growth cycle of local chickens, we observed the trend of CEL gene expression from 0 to 30 weeks of age: the highest expression level was at 14 weeks of age, and the lowest was at 0 weeks ( Fig. 4b, p = 0.025). Next, we detected CEL expression of different genotypes and found that the II genotype showed significantly higher expression than the DD group (p = 0.003, Fig. 4c). Interestingly, we also found that CEL expression in the II genotype (low-weight line, n = 4) was also much higher than that in the DD genotype (high-weight line, n = 4) (Fig. 4d, p = 0.039). Therefore, we suspected that the CEL gene may be involved in the rapid growth of chickens and that this insertion mutation contributes to increased expression of CEL in chicken pancreas tissues.
Association analysis of the indel with chicken economic traits. Further association analysis was used to investigate the functional effects of the 99-bp insertion polymorphism in the CEL gene on chicken phenotypic traits. In individuals aged 4 to 12 weeks, the 99-bp indel showed significant associations with shank length (SL), shank girth (SG), chest breadth (CB) at 8 weeks, and breastbone length (BBL) at 4 weeks (p = 0.005; p = 0.007; p = 0.007; and p = 0.032, respectively). Individuals with ID genotypes had a longer BBL at 4 weeks than the II birds. At 8 weeks old, the SL, SG, and CB of birds with the DD genotype were significantly larger than those of birds with the II genotype (Table 3). Carcass traits directly affect economic value, and we found that the effect of the indel on carcass traits, including semi-evisceration weight (EW), subcutaneous fat weight (SFW), breast muscle weight (BMW) and leg weight (LW), was significant (Table 4) (p < 0.05) and that the phenotypic values of DD genotype individuals were the highest. However, the difference between ID and II genotypes were not significant (p > 0.05). Furthermore, no association was found between the indel locus and body weight (BW) traits of chickens at ages of 0 and 12 weeks (p > 0.05, Supplementary Table S2), and the BW of homozygous DD genotypes was greater than that of homozygous II genotypes. These data implied that this insertion mutation could be associated with growth performance.

Association analysis of the indel with chicken biochemical variables. Animal blood function is
closely related to the body's metabolism, nutritional status and disease. When animals are in different growth states, the physical and chemical indicators of blood may change accordingly; thus, it is necessary to investigate the impact of the indel mutation on chicken biochemical variables in blood. Association analysis showed that the indel polymorphism was significantly associated with alanine transaminase (ALT), gamma-glutamyl transpeptidase (GGT) and albumin (ALB) (p < 0.05). The levels of total cholesterol (TC), triglyceride (TG), HDL and LDL of II genotype birds were higher than those with the DD genotype but no association was found between the indel locus and those traits (Table 5). Collectively, we suggested that this insertion mutation may lead to blood metabolic disorders that affect the growth of chickens.  Table 2. Pairwise fixation index (Fst) of the CEL gene between indigenous and commercial chickens. Note: KF = Kauai feral chicken, TiC = Tibetan chicken, XC = Xichuan black-bone chicken, LS = Lushi blue-eggshell chicken, CS = Changshun blue-eggshell chicken, DX = Dongxiang blue-eggshell chicken, HB = Hy-Line Brown, AA = Arbor Acres. All Fst statistical values were tested by GENEPOP 4.5 with Bonferroni correction. The cutoff value was set at p < 0.002 at a 5% confidence interval.

Effects of CEL on cholesterol transport in vivo and in vitro.
Compared with the control diet, the addition of 4% oxidized soybean oil to the diet showed that the treatment significantly increased the expression of the CEL gene (Fig. 5a). Interestingly, the expression levels of the genes APOB, MTTP, APOV1 and SREBF1, which are involved in lipid transport were significantly reduced (Fig. 5b). Furthermore, in vitro treatment results showed that the expression level of CEL significantly increased in chicken primary hepatocytes after transfection with pcDNA3.1-CEL-EGFP for 12 h compared with the control condition (Fig. 6a). Additionally, the expression levels of APOB, MTTP, APOV1 and SREBF1 significantly increased in the CEL-overexpressing group (Fig. 6b). This result suggests that the CEL gene may play a role in hepatic lipoprotein metabolism and regulation of growth by participating in lipid transport.

Discussion
As important economic traits, growth and body weight performance have become highly important factors in molecular breeding. Because of their simple and convenient characteristics, indel molecular markers have been used widely to improve the economic traits of livestock and poultry. In this study, an indel caused by a 99 bp insertion fragment was shown for the first time in the promoter region of the CEL gene. Indel mutations in the promoter region have been found in livestock, suggesting their key role in animal growth and carcass traits 25,26 .
Allelic frequency is used to show the diversity of genes in a population with potential genetic drift, and the novel introduction of mutations 27 . In this study, the genotype frequencies of the CEL gene were consistent with the HWE (p > 0.05) and exhibited low to moderate polymorphism. However, the II genotype is rarely seen and the D allele is the highly predominant allele in commercial breeds, which may be due to the high degree of selection for the CEL gene. Artificial selection has produced large phenotypic changes during the domestication of different breeds. Our results indicate that the chicken CEL gene has abundant indel polymorphisms in different chicken breeds, and further correlation with growth traits can be analyzed.
Correlation analysis statistical results showed that there were significant effects of the indel on SL, SG, and CW at 8 weeks and on BBL at 4 weeks and that the DD genotype has a higher phenotypic value. Notably, the body shape of chickens is related to bone and muscle development. The direct manifestation of body shape and bone is SL, which is generally used to measure and express bone development. In chickens, the speed of growth is highly correlated with carcass traits and would have a greater impact on economic value 28 . The data associated with the carcass traits indicated that compared with that of the II and ID genotypes, the phenotypic values of the DD genotype were higher in BMW, LW, LMW, CW, SFW, EW, and SEW. Although this mutation had no significant effect on body weight, we still observed that the DD genotype had greater phenotypic values than the II genotype group. This analysis illustrated that the DD genotype is the dominant genotype and that the indel polymorphism can be used as a potentially useful DNA marker. www.nature.com/scientificreports www.nature.com/scientificreports/ At present, the protein coding region remains the main focus of human disease research. However, the contribution to variation in the disease does occur in the noncoding region of the genome 29,30 . For example, promoter-specific and enhancer-specific mutations have a clear causal relationship with many Mendelian diseases 31 . Low-density lipoprotein receptor (LDLR) promoter mutations have been shown to cause familial hypercholesterolemia (FH) 32 . In addition, mutations in the core promoter region of the telomerase reverse transcriptase (TERT) gene result in uncontrolled cell proliferation and malignant transformation of hepatocellular carcinoma 33 . In fact, the regulatory mechanism of gene promoter region variation may regulate gene expression by altering transcription factor binding site or promoter activity changes 34 . Studies have shown that the CEL gene is highly expressed in the exocrine pancreas with the help of pancreatic transcription factor 1 in humans 5 , mainly due to the presence of the classical TATA and CCAAT elements associated with tissue-specific expression of the gene in the 5'-flanking sequence of the CEL gene 35,36 . The tissue expression profile of this study showed that CEL is also   www.nature.com/scientificreports www.nature.com/scientificreports/ highly expressed in the pancreas, and we speculate that this is related to the regulatory elements present in the flanking sequences. In addition, CEL is also present in the liver and blood vessel walls of mammals 37 , and the levels of CEL mRNA expression in other tissues including the liver were quite low in the current study. The digestion of fat requires the joint action of bile acids, lipases and colipases; however, the physiological functions of fat digestion in chicks have not yet matured and continue to develop for a few weeks after emergence. The results of this test also show that expression of the CEL gene has a low-high-low change trend in AA broiler chickens and local chickens in the pancreatic gland tissues, which was related to the growth rhythm of chickens. Taken together, this evidence suggests that CEL may affect body growth by regulating the body's lipid metabolism-related pathways.
Previous research indicated that CEL is involved in cholesterol esterification, the hydrolysis of dietary fat, and the uptake of fat-soluble vitamins in the duodenum 8 . However, the role of CEL in intestinal cholesterol absorption remains controversial. Cholesterol absorption was shown to be reduced by 80% when pancreas-and bile-diverted rats were supplied with bile and pancreatic juice from which CEL had been removed compared with rats given bile and pancreatic juice containing normal amounts of the enzyme 38 . These results were supported by other studies showing that CEL-specific inhibitors inhibited cholesterol absorption in rats and dogs and reduced serum cholesterol levels in cholesterol-fed rats 39 . In contract, Watt and Simmonds showed that absorption of intestinal cholesterol does not require pancreatic juice, as CEL may remain on the intestinal microvilli after pancreas metastasis 40 . Since cholesterol in the diet and cholesterol in the bile are mainly unesterified and CEL promotes only the absorption of esterified cholesterol by cells 41,42 , the importance of CEL in intestinal cholesterol absorption is not obvious and lack of CEL also does not affect the rate of cholesterol transport from the intestinal lumen to the lymph 10 . Our correlation analysis of biochemical variables in blood showed that TC birds with II and DD genotypes did not show statistically significant differences. In addition to its presence in the digestive tract, Camulli et al. first determined that the cytosolic cholesterol esterase present in the liver is CEL, which is identical to the   www.nature.com/scientificreports www.nature.com/scientificreports/ pancreatic enzyme [42][43][44] , and that CEL is present in specific endosomal compartments of hepatocytes 45 . Winkler et al. 46 showed that hepatocytes can directly secrete hepatogenic CEL. Therefore, the CEL found in the endosomal compartment of the liver may be a recaptured enzyme secreted by hepatocytes, which may play a role in the hepatic secretory capture pathway in which lipoproteins are metabolized 47 . Moreover, the addition of CEL to the culture medium promotes selective uptake of cholesterol esters in HDL by hepatocytes, suggesting a role for CEL in hepatic lipoprotein metabolism 48 . Our results also confirmed that overexpression of CEL in liver primary cells significantly reduced the expression levels of the APOB, MTTP, APOV1 and SREBF1 genes. These results together confirm that CEL may affect the growth traits of chickens by participating in the regulation of hepatic lipoprotein metabolism, which may help understand the mechanism of CEL gene regulation in chickens.
In summary, through the identification of a new 99-bp indel of the CEL gene promoter region, and analysis of the association of this site with economic traits of chickens, we found significant associations of the indel with chicken growth and development traits. Population genetic analysis revealed that the D alleles are significantly different in the chicken domestication process of different populations, tending to fix in commercial broilers. Further gene expression analysis showed that the gene was highly expressed in pancreatic tissue and participated in the regulation of hepatic lipoprotein metabolism. Collectively, these results will help us understand the CEL regulatory mechanisms of chickens and suggest a molecular marker that may help improve genetic selection, thereby improving poultry performance.

Materials and Methods
Ethics statement. All animal experiments were performed according to the Regulations of the Chinese National Research Council (1994) and were approved by the Henan Agricultural University Institutional Animal Care and Use Committee (Permit Number: 11-0085). Chickens were given ad libitum access to feed and water, and all efforts were made to minimize suffering during the experiment 49 . Animals and sample preparation. Blood samples were obtained from 2060 individuals including an F 2 resource population (F 2 , n = 794), four local Chinese chicken breeds (Xichuan black-bone chicken (n = 185, XC), LS chicken (n = 190), CS chicken (n = 144) and Dongxiang blue-shell chicken (n = 172, DX)) and three commercial breeds (Hy-Line Brown hens (n = 217, HB), Ross 308 (n = 166), AA broilers (n = 192)). The F 2 resource population was constructed via reciprocal crosses between Gushi (GS) chickens and Anka broilers the resulting growth and carcass trait data for 804 F 2 individuals were recorded using measurement methods detailed in a previous study 50 . We used the growth, carcass and blood metabolic traits of 794 chickens for association analysis. Genomic DNA was isolated from blood samples by the phenol-chloroform method using a commercial kit (Tiangen, Beijing, China), and all DNA samples were diluted to 50 ng/μl and stored at −20 °C for subsequent experiments. In addition, Kauai feral chickens (KF) and Tibetan chickens (TiC) were genotyped using genome resequencing datasets from the NCBI Sequence ReadArchive (SRA) database.
LS chicken is a Chinese indigenous chicken breed well-suited for research involving functional genes in many important biological processes 51,52 , and this is why it was chosen as one of the models to study the regulation of CEL gene expression. Twenty 10-week-old LS chickens were randomly divided into two groups and raised to 14 weeks. Among these, ten were fed a complete diet as a control group, and the other 10, the treatment group were additionally given 4 grams of soybean oil per 100 grams of the complete diet. Ten tissue samples (heart, liver, spleen, lung, kidney, duodenum, ovary, abdominal fat, breast muscle, and pancreas) were then collected from all twenty 14-week-old LS chickens within 10 min after slaughter, immediately snap-frozen in liquid nitrogen and stored at −80 °C pending examination of the tissues for the CEL mRNA. Furthermore, pancreatic tissues of 14-week-old LS chickens from three individuals of each of the II, ID and DD genotypes were used to detect the CEL mRNA expression of different genotypes, and 8 more birds from the Ross 308 and LS breeds (n = 4 per population) were also sacrificed for the same purpose. To further investigate whether the CEL gene is involved in lipid metabolism regulation in chickens, liver tissue was used to detect changes in genes involved in lipid transport. In addition, to investigate the expression patterns of CEL during chicken growth, we collected 4 pancreatic tissue samples from different growth stages of AA broilers and LS chickens and measured the expression of the CEL gene.
Cell culture and treatment. The isolation and culture of chicken embryonic primary hepatocytes was performed using the same method as described previously 49,53 . In brief, chicken embryonic primary hepatocytes were seeded at a density of 1 × 10 5 cells/mL in 6-well plates at 37 °C and 5% CO 2 . Upon reaching 60-80% confluence, the cells were transfected with a CEL-overexpression vector (50 nM) or a negative control (50 nM) using 4 μL of Lipofectamine ® 2000 (Invitrogen, Carlsbad, CA, USA), and the medium was replaced 6 h later. Cells were collected separately from different treatment groups at 12 h, washed in 1 × PBS (Solarbio, Beijing, China) and collected with TRIzol ® reagents (Takara, Kyoto, Japan) and total cellular RNA was extracted to study the regulation of lipid metabolism by the CEL gene. All transfections were carried out in triplicate. Table S3) for cloning the coding region (CDS) of the genes were designed using the chicken CEL gene and mRNA sequences in GenBank (accession numbers NC_006104 and NM_001012997.1, respectively) with Primer 5.0 software. PCR was performed with a 20 µL final reaction volume containing 10 μL PrimeSTAR ® Max DNA Polymerase (Takara, Kyoto, Japan), 1 μL each of forward and reverse primers, and 6 μL RNase-free water. The PCR products were cloned into pMD-18T (Takara, Kyoto, Takara) and sequenced by Sangon Biotech Co. Ltd. (Shanghai, China). The sequence was confirmed by repeating the sequencing three times. Furthermore, to further validate the regulatory role of CEL in cholesterol metabolism in vitro, we constructed the recombinant expression plasmid pcDNA3.1-CEL-EGFP. The recombinant vectors were transfected into chicken primary hepatocytes.

Polymorphism detection and diversity analyses of different breeds. A total of 8 populations were
used to identify the 99-bp indel in the 5' region of the CEL gene and for allele frequency analysis. Specific PCR primers P1 (Supplementary Table S3) were designed to amplify the 246/345 bp fragment of the 5' region of the CEL gene to detect the 99-bp insertion variation (GenBank accession NC_006104). Detailed PCR amplification steps refer to previously published literature 54 . Three genotypes were observed, including one genotype with a 345-bp band, one with 246-bp and 345-bp bands and one with a 246-bp band. In addition, the study used the high-throughput sequencing dataset in the NCBI database (http://www.ncbi.nlm.nih.gov/) to explore CEL genotype information from two other international breeds of chickens. Indel allele sequences (containing a 99 bp insert) were aligned with clean reads using standard nucleotide BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi).
The genotypes, allele frequencies and HWE of the 10 breeds were calculated using the SHEsis program (http:// analysis.bio-x.cn). Colony genetic indexes, including heterozygosity (He), effective allele numbers (Ne) and polymorphism information content (PIC), were calculated using Nei's methods with PopGene software (Version 1.3.1) 55 . The paired fixed index (Fst) of feral, native and commercial chickens was tested by GENEPOP 4.5 and Bonferroni correction 56 . Total RNA isolation and cDNA preparation. Total RNA was prepared from the different tissues and cells with TRIzol ® reagent (Takara, Kyoto, Japan) according to the manufacturer's protocol, and RNA quality was assessed using electrophoresis and spectrophotometry. Reverse transcription into cDNA was performed using an RNA reverse transcription kit (Takara, Kyoto, Japan) with 2 μg total RNA in a 20 μl reaction.

Real-time quantitative PCR. Primers spanning exon-exon junctions were synthesized by Sangon Biotech
Company (Shanghai, China) for detection of relative expression levels of genes and the ACTB (β-actin) gene was used as an internal control. Real-time PCR (RT-PCR) was performed using the SYBR Green method and a Roche LightCycler ® 96 instrument. The composition of the reaction and the amplification step are referred to published in the literature and the reactions were performed in triplicate for each sample. The PCR products were identified via P2-F and P2-R sequencing.
Bioinformatics analysis. Database searches were performed at NCBI (http://www.ncbi.nlm.nih.gov/) under default settings. The transcriptome data for tissue expression profiling analysis of the CEL gene of four species were obtained from the NCBI, and Ensembl databases and previously published articles 57 . The DNA, nucleotide, and amino acid sequences were analyzed using BioXM (Version 2.7) and DNAMAN (version 6.0). Genetic conservation was analyzed using Genomicus available at http://www.genomicus.biologie.ens.fr/genomicus-71.01/ cgi-bin/search.pl 49 , which was employed for synteny analysis of genes of interest across multiple species in this study. The secondary structures of the amino acid sequences were predicted at http://smart.embl-heidelberg.de/. The signal peptide and transmembrane helix analyses were performed using online software (http://www.cbs.dtu. dk/ser-vices/SignalP/ and http://www.cbs.dtu.dk/services/TMHMM/, respectively). N-linked glycosylation sites were predicted using NetNGlyc (version 1.0) servers at http://www.cbs.dtu.dk/services/NetN-Glyc/.
Statistical analysis. Statistical analyses of associations between the genotypes and selected traits of the F 2 chickens were performed using SPSS 24.0 (SPSS for Windows, Standard version 24.0; SPSS, USA) according to the following two linear mixed models. Model I was used to evaluate growth traits. Considering the effects of (2020) 10:3215 | https://doi.org/10.1038/s41598-020-60168-2 www.nature.com/scientificreports www.nature.com/scientificreports/ body weight on carcass traits, carcass weight was used as a concomitant variable in model II, which was applied to calculate carcass traits 50 where Y ijklm is the observed value; µ represents the overall population mean; G i is the fixed effect of the genotype (i = II, ID, DD); f l is the fixed effect of the family (l = 7); S j is the fixed effect of sex (j = 2); H k is the fixed effect of the hatch (k = 2); b represents the regression coefficient for carcass weight; W is the average slaughter weight; W ijklm is individual slaughter weight; and e ijklm is the random error. G i , S j , H k are fixed factors, and f l is a random factor. Multiple comparisons were analyzed with the least square means to display significant differences among genotypes with Bonferroni's test. These trait phenotypic values of different genotypes are expressed as mean ± standard error (SE) and significance was determined by a p-value <0.05 58 . Relative gene expression data and significance in different tissues were analyzed using the 2 −ΔΔCT method and one-way analysis of variance (ANOVA) followed by Duncan's test 59 . The graphics were drawn using GraphPad Prism 7 (GraphPad, San Diego, CA, USA). A p-value < 0.05 was considered statistically significant, and a p-value < 0.01 was considered highly significant. These data are expressed as the mean ± standard deviation (SD).

Data availability
All data generated or analyzed during this study are included in this published article (and its Supplementary  Information Files).