A Loss-of-Function Mutation in bco1l Underlies Yellow Coloration in Large Yellow Croaker (Larimichthys crocea)

TL;DR

A genetic mutation in the bco1l gene causes yellow coloration in large yellow croaker fish, which can be used to improve aquaculture breeding.

Contribution

Discovery of a 10 bp deletion in the bco1l gene as the genetic basis for yellow body coloration in large yellow croaker.

Findings

A 10 bp deletion in the bco1l gene causes loss of enzymatic activity for carotenoid cleavage.

The deletion leads to yellow body coloration due to carotenoid accumulation in the skin.

The mutation is rare and provides a validated marker for marker-assisted selection in aquaculture.

Abstract

Body color represents a commercially valuable phenotypic trait in the large yellow croaker (Larimichthys crocea). Here, we identify a 10 bp deletion in the carotenoid-cleaving enzyme gene bco1l that underlies its distinctive yellow body coloration. This loss-of-function mutation abolishes enzymatic activity, thereby impairing the metabolic cleavage of dietary carotenoids and resulting in their selective accumulation in the skin. Our work establishes a causal genetic mechanism for this economically important trait and delivers a robust, functionally validated molecular marker to accelerate marker-assisted selection in aquaculture breeding programs.

Carotenoid-based coloration significantly influences the ornamental appeal and market value of aquatic species. This study identifies the genetic basis of yellow body coloration in the large yellow croaker (Larimichthys crocea), a commercially important aquaculture species in China. Through a genome-wide association study (GWAS), we identified a 10 bp deletion (bco1l-Δ10) in the bco1l gene, a critical enzyme in carotenoid metabolism. This deletion, located in exon 10, causes a frameshift mutation and premature stop codon, resulting in loss of the enzymatic function essential for carotenoid cleavage. Sanger sequencing confirmed that all yellow juveniles were homozygous for the deletion, while none of the wild-type juveniles harbored this mutation. Functional assays demonstrated that the recombinant Bco1l-Δ10 protein failed to catalyze β-carotene cleavage, validating the genetic association with the yellow phenotype. The bco1l-Δ10 mutation was detected only in a limited number of juvenile individuals, indicating it is a rare occurrence in the population. Our findings provide a valuable genetic marker for marker-assisted selection (MAS) to enhance yellow pigmentation in aquaculture breeding programs, offering a potential strategy to improve the market value of L. crocea.

Figures

Full text

1. Introduction

Animal coloration serves as an excellent model system for investigating the genetic mechanisms underlying phenotype variation [1]. Research has shown that pigmentation genes exhibit high functional conservation across vertebrates. However, the underlying genetic mechanisms remain poorly understood, particularly regarding non-melanin coloration such as carotenoid-based coloration, where research progress has lagged. This knowledge gap limits our comprehensive understanding of the evolutionary processes driving body color phenotype diversity [2].

In vertebrates, aquatic animals exhibit remarkable diversity in body coloration. The large yellow croaker (Larimichthys crocea) is a commercial marine fish that is mainly limited to the coastal waters of East Asia [3]. According to the China Fishery Statistical Yearbook (2025), L. crocea production reached 297,683 tons in 2024, establishing it as one of the leading marine cultured fish species in China [4]. As the name suggests, the large yellow croaker has two distinct characteristics: its yellow body coloration and its ability to produce vocalizations. The yellow coloration on the ventral skin of the large yellow croaker is one of the important reasons for its popularity, as well as an important factor determining its economic value [5]. According to aquatic market surveys, consumers tend to prioritize yellow-colored individuals when making purchasing decisions, as this coloration is not only aesthetically pleasing but also widely recognized as a reliable indicator of superior meat quality, higher carotenoid content, and better freshness [6,7]. From an economic perspective, this preference translates into significant market differences: yellow L. crocea typically commands a 15–20% higher price compared to conspecifics with pale or dull body coloration, which directly improves the economic returns of aquaculture operations [7]. However, under practical aquaculture conditions, the yellow coloration on the ventral skin of the large yellow croaker can only be maintained at night or under specific wavelengths of light, which results in economic losses and difficulties in fishing [8].

The yellow color in the ventral skin results from the coexistence of light-reflecting iridophores and light-absorbing xanthophores [9,10]. The chemical composition of xanthosomes is pteridines and/or carotenoids [11,12]. Carotenoids are lipophilic pigments synthesized by plants, algae and some micro-organisms. They are involved in various biological processes, such as embryonic development [13], immunity [14], reproduction, and cell growth and differentiation [15], and are known as antioxidants [16]. Carotenoids cannot be biosynthesized by themselves but are obtained directly from the diet [17]. Many studies have proven that the addition of carotenoids to the diet can affect body coloration in many teleost species [6,18,19,20,21]. Dietary addition of astaxanthin can make the skin of the large yellow croaker more intensely yellow [22] and can enhance the red color of clownfish [23] as well as the skin of Australian snapper (Pagrus auratus). However, improving body color through dietary supplementation with carotenoids has disadvantages, such as high cost and unstable effects. Optimizing body coloration at the genetic level is another effective approach. For animals, carotenoid coloration primarily relies on the following three physiological steps: absorbing, transporting, and then metabolizing to apocarotenoids or depositing in the target tissues [24].

Variation in carotenoid metabolism has been linked to the function of carotenoid oxygenase. The animal carotenoid oxygenase gene was first identified in Drosophila melanogaster [25,26]. Subsequently, two carotenoid oxygenase genes were characterized in mammals—β-carotene 15,15′ oxygenase-1 (bco1) and β-carotene 90,10′ oxygenase-2 (bco2)—which cleave carotenoids symmetrically and asymmetrically, respectively [27]. In teleosts, genome sequence analyses have revealed five carotenoid-cleaving genes: two bco1 genes (bco1 and bco1-like) and three bco2 genes (bco2a, bco2b and bco2-like) [28]. As important enzymes that execute the first step in the degradation of carotenoids absorbed from the diet, mutations influencing either transcript levels or the protein-coding sequence of these enzymes lead to the accumulation of intact carotenoids in body tissues [29,30,31,32,33].

Recently, juvenile L. crocea individuals with distinctly yellow body coloration have been observed in hatchery populations, contrasting markedly with the commonly observed transparent phenotypes. This observation suggests that body color variation in L. crocea may not be solely attributable to dietary carotenoid intake but may also be influenced by genetic factors. To investigate the genetic basis of this variation, we collected juveniles representing distinct pigmentation phenotypes and conducted whole-genome resequencing followed by a genome-wide association study (GWAS) to identify genomic loci and candidate genes associated with pigmentation traits. In addition, the function of candidate genes responsible for the different color phenotypes was further examined using an E. coli strain that was specifically engineered to synthesize β-carotene [26].

2. Materials and Methods

2.1. Ethics Statement

All sampling procedures involving animal handling and treatment in this study were approved by the Institutional Animal Care and Use Committee of Xiamen University (IACUC: XMULAC20230011) and the Ethics Committee of Jimei University (protocol code 2021[5], approved on 22 January 2021).

2.2. Sample Collection

On 23 January 2022, juvenile fish were collected from a nursery tank (60 m × 10 m × 2.5 m) in He-Bao-Tian Village, Wenzhou City, Zhejiang Province, China. Thirty-eight individuals were randomly selected, including 18 yellow-colored and 20 transparent fish. Muscle tissue was collected and stored at ~20 °C until DNA extraction.

2.3. DNA Isolation, Whole-Genome Resequencing and Genotyping

Genomic DNA was isolated from muscle tissue following a standard phenol/chloroform extraction method [34]. We evaluated DNA purity and concentration using a Nanodrop-2000 spectrophotometer (Thermo Scientific, Waltham, MA, USA) and a Qubit^®^ 3.0 Fluorometer equipped with the Qubit^®^ DNA Assay Kit (Invitrogen, Carlsbad, CA, USA). All DNA samples met the stringent integrity and concentration requirements and were subsequently submitted to Novogene (Beijing, China) for Illumina library construction and sequencing. The genotyping pipeline was modified from our prior work [35]. Briefly, clean reads were aligned to the L. crocea T2T reference genome [36] via the BWA MEM algorithm (v0.7.17) [37], and the alignment results were sorted using Samtools (v1.19.2) [38]. Variant calling and base quality control were executed by GATK (v4.19.0) [39] with default parameters. We filtered the resulting SNPs to remove those with a missing rate < 20% using bcftools (v1.19) [38] and subsequently performed genotype imputation with Beagle (v5.4) [40]. Final quality filtering was conducted using PLINK (v1.9) [41] based on a minimum allele frequency (MAF) > 0.05 and a Hardy–Weinberg equilibrium (HWE) p-value ≤ 1 × 10^−5^.

2.4. Genome-Wide Association Studies

To assess population structure, we calculated the pairwise kinship matrix using all dataset SNPs via EMMAX [42]. Binary phenotypes (coded as 0/1) were tested under the same model without extra covariates, relying on the kinship matrix to control for individual relatedness. Association analysis was performed employing the mixed linear model within EMMAX [42]. Significance was defined at a threshold of 0.05/N, with N being the specific marker count for the study (12,180,697 SNPs or 2,887,666 indels). We also applied a “suggestive associations” threshold of 1/N to screen for potential [43]. Visualization of the results, including Manhattan and QQ plots, was carried out using the R package qqman (v0.1.9) https://cran.r-project.org/web/packages/qqman/ (accessed on 12 November 2025).

2.5. Candidate Variation Validation by PCR and Sanger Sequencing

To target the deletion mutation within the bco1l gene of L. crocea, PCR amplification was performed using the following primers: 5′-ACTGCGCTGAGTCAACTGAA-3′ (Forward) and 5′-GCTCGAGCGATCTCCTTGAA-3′ (Reverse). Thermal cycling conditions commenced with denaturation at 94 °C for 1 min. This was followed by a 40-cycle touchdown program comprising denaturation (94 °C, 30 s), annealing (30 s), and extension (72 °C, 2 min). The annealing temperature was decreased from 70 °C to 60 °C in 2 °C decrements every second cycle, followed by 30 cycles at a constant 60 °C. The procedure concluded with a final extension step at 72 °C for 1 min, after which the products were subjected to bidirectional Sanger sequencing.

2.6. Real-Time Quantitative Polymerase Chain Reaction

We extracted total RNA from the visceral mass using RNAex Pro Reagent (Accurate Biotechnology, Changsha, China) in accordance with the standard protocol. RNA concentration was measured using a NanoDrop™ One Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). First-strand cDNA was synthesized from 500 ng of total RNA using the Evo M-MLV Plus Kit (Accurate Biotechnology, China). All qPCR assays were performed in 20 μL volumes using SYBR^®^ Green Pro Taq HS (Accurate Biotechnology, China) on the qTOWER 2.2 Real-Time PCR System (Analytik Jena, Jena, Germany). The expression of bco1l and bco1l-Δ10 was normalized to the reference genes β-actin and gapdh using the comparative Ct method. Primer sequences were as follows: bco1l (Forward: 5′-ACAGATCGCCAAGTTTGACC-3′; Reverse: 5′-AATCACTCCGTCGTCTTCC-3′), bco1l-Δ10 (Forward: 5′-TCCATCCAACAGATCGCCTC-3′; Reverse: 5′-AATCACTCCGTCGTCTTCC-3′), β-actin (Forward: 5′-AGCCATCCTTCCTCGGTATG-3′; Reverse: 5′-CTCCAGACAGCACGGTGTT-3′) and gapdh (Forward: 5′-ACATCATCCCCGCCTCTACT-3′; Reverse: 5′-TCGTCGTATTTGGCGGGTT-3′). Data were analyzed using GraphPad Prism version 8, and statistical significance compared to control was determined using Student’s unpaired t-test.

2.7. Construction and Expression of Recombinant Proteins

The open reading frames (ORFs) of Bco1l and Bco1l-△10 were amplified from the cDNA of wild-type and yellow mutant juveniles, respectively. We used specific primers incorporating NcoI and EcoRI restriction sites: Forward 5′-AGGCCATGGCTATGCAGACCATCTTTG-3′ and Reverse 5′-CTCGAATTCTCACACGGCAGGGATGA-3′. The PCR products were initially ligated into the pMD19-T vector (TaKaRa, Dalian, China), followed by double digestion with NcoI and EcoRI. The fragments were subsequently subcloned into the pET-32a (+) expression vector to construct the recombinant plasmids pET-32a (+)-bco1l and pET-32a (+)-bco1l-△10. These constructs were transformed into E. coli XL1-Blue cells, and expression of the recombinant proteins was induced with 0.1 mM IPTG for 24 h at 22 °C.

For Western blot analysis, bacterial extract (one OD600) was resolved on 10% SDS-PAGE gels using the Bio-Rad Mini system and transferred to polyvinylidene fluoride membranes (Roche, Basel, Switzerland). Membranes were blocked for 1 h with 5% skim milk in Tris-buffered saline containing 0.1% Tween (TBS-T). They were then incubated with an anti-HisTag antibody (Quiagen, Duesseldorf, Germany; 1:5000 dilution) overnight at 4 °C, followed by an appropriate HRP-conjugated secondary antibody. Protein bands were visualized using an enhanced chemiluminescence (ECL) detection system (Pierce or Pharmacia).

2.8. High-Performance Liquid Chromatography (HPLC) Analysis

We transformed beta-carotene-producing E. coli XL1-Blue cells with CCD expression plasmids. Cultures were grown in LB medium at 37 °C until an OD600, and then the cultures were shifted to 22 °C for 1 h to achieve production of beta-carotene. We induced Bco1l protein expression using 0.1 mM IPTG and 25 mg/L ferrous sulfate, followed by incubation at 22 °C for 24 h. After harvesting cells by centrifugation (4000× g, 15 min, 4 °C), pellets were processed under dim red light (<600 nm). Samples equivalent to 20 OD600 were dissolved in a mixture of 200 µL 2 M hydroxylamine (pH 6.8) and 200 µL methanol and incubated at room temperature for 10 min. Extraction was performed using 400 µL acetone and 500 µL petroleum ether, with phase separation achieved by centrifugation at 4000× g for 30 s. The organic phase was collected, the extraction was repeated, and the combined phases were dried in a SpeedVac (Eppendorf, Hauppauge, NY, USA). Finally, the dried lipids were redissolved in 150 µL hexane/ethyl acetate (90:10 v/v) for HPLC.

Chromatography was performed using an Agilent 1200 HPLC system (Agilent, San Jose, CA, USA) fitted with a Zorbax silica column (4.6 × 150 mm, 5 µm). Samples were eluted isocratically with 10% ethyl acetate in hexane at 1.4 mL/min. Beta-carotene peaks were identified based on elution time and UV/Vis spectra, referencing authentic standards gifted by DSM (Sisseln, Switzerland).

3. Results

3.1. Body Color Phenotypes of Wild-Type and Yellow Juvenile

Juvenile individuals with yellow body color were visually distinct from wild-type counterparts (Figure 1). At this developmental stage, only melanophores were present on the body surface in both groups (black arrows in Figure 1B,D), indicating that the yellow coloration likely originates from underlying muscle tissue rather than pigment cells on the skin (Figure 1D).

3.2. Resequencing and SNPs Detection

After resequencing the genomes of 38 L. crocea individuals, a total of 215 Gb of raw data was generated. These fish were offspring from hundreds of male and female broodstocks that spawned simultaneously via mixed spawning in a single concrete tank. Due to the mixed parentage and the large population size in the nursery tank, the family relationships of the 38 selected fish were not determined. Following quality control, 208 Gb of clean data was obtained. The clean reads were mapped to the reference genome, with mapping rates exceeding 99.5% across all samples. A total of 18,272,039 SNPs and 4,608,469 indels were identified, among which 12,180,697 high-quality SNPs and 2,887,666 high-quality indels were retained after filtering (Figure S1).

3.3. Identification of bco1l Deletion Mutation as Potential Causal Variation

GWAS identified 651 indel variants on chromosome 24 (chr24) that were significantly associated with body coloration. These variants were highly clustered within a defined genomic interval spanning chr24: 15,422,263–18,387,930 (Figure 2A). To evaluate potential systematic bias, a QQ plot was generated. The genomic inflation factor (λ) was 1.06, suggesting that population structure had minimal impact on the results (Figure 2B). Within this candidate region, 71 protein-coding genes were annotated, among which bco1l was identified as a key gene involved in carotenoid metabolism—a critical pathway underlying yellow pigmentation. Notably, a 10 bp deletion (GGTCAAACTT) was detected in the 10th exon of the bco1l gene (Figure S2). This deletion, designated as bco1l-Δ10, introduces a frameshift mutation in exon 10, resulting in a premature stop codon in exon 11 (Figure 2C). To validate this genetic variant’s association with body coloration, we sequenced the 10 bp deletion region across all juvenile individuals. None of the 18 wild-type juveniles were homozygous for the deletion, whereas all 20 yellow juveniles were homozygous for it (Table 1).

3.4. The Expression Level of bco1l and bco1l-Δ10 in Visceral Tissues of Juvenile L. crocea

Visceral tissues were collected separately from wild-type and yellow juvenile individuals. Using β-actin as an internal reference gene, the relative expression levels of bco1l mRNA in wild-type juveniles and bco1l-Δ10 mRNA in yellow juveniles were quantified. Results indicated that bco1l-Δ10 was actively transcribed, but its expression level was 27.53-fold lower than that of bco1l in wild-type juveniles (p = 3.77 × 10^−5^; Figure 3).

3.5. Effect on Carotenoid Cleavage Assay of Recombinant Bco1l and Bco1l-Δ10 Protein

This 10 bp deletion in the bco1l gene resulted in substantial changes in the subsequent amino acid sequence, causing defects in the RPE65 domain, a retinol isomerase (Figure 4A). Western blot confirmed that both Bco1l and Bco1l-Δ10 were successfully expressed (Figure 4B). To confirm whether bco1l-Δ10 transcripts may produce non-functional proteins, the catalytic activity of the recombinant bco1l and bco1l-Δ10 was tested in vivo, using an E. coli strain specifically engineered to synthesize β-carotene. Following the induction of gene expression, enzyme activity could be inferred from the color of the bacterial pellet, which would remain yellow in the absence of enzyme activity. Conversely, a color shift from vibrant yellow to almost-white indicated the presence of enzyme activity and the conversion of β-carotene to retinaldehyde. Bacteria expressing Bco1l contained significant amounts of retinoids (retinyl-ester, all-trans-retinal, and all-trans-retinol) in addition to β-carotene, and the color of the bacteria pellets was almost white (Figure 4C). In contrast, the color of bacteria transformed with bco1l-Δ10 alone remained yellow, and no peaks for retinoids were detectable (Figure 4D). Thus, these experiments showed that the croaker Bco1l has 15,15′ dioxygenase activity, while the bco1l-Δ10 variant lost such catalytic activity.

4. Discussion

Carotenoid-based polymorphisms are widespread across populations of birds, fish, and reptiles [44]. Herein, through a genome-wide association study, we identified an indel located in the coding region of the bco1l gene, known to be involved in carotenoid metabolism, as the most likely genetic basis for the intensive yellowness of body color in juveniles of L. crocea. Results from Sanger sequencing confirmed a homozygous 10 bp deletion in the 10th exon of the bco1l gene. Moreover, the recombinant Bco1l-Δ10 protein failed to catalyze beta-carotene.

Carotenoid coloration in animals primarily involves three physiological processes: absorption, transport, and subsequent metabolism to apocarotenoids or deposition in target tissues [24]. Accumulating evidence indicates that carotenoid oxygenases serve as pivotal regulators in carotenoid processing across animal species. Mutations and disruptions in carotenoid oxygenase genes frequently alter tissue carotenoid status and coloration. For instance, in Darwin’s finches, a mutation in bco2 causes beak color to change from pink to yellow [33]. Similarly, mutations in bco2 cause yellow discoloration of white fat in Norwegian sheep and bovine [32]. In Atlantic salmon (Salmo salar), expression levels of bco1l and bco2l genes are associated with fresh color [45]. In the scallop (Patinopecten yessoensis), downregulation of the Pybco-like 1 gene leads to orange adductor muscle with carotenoid deposition [31]. In chickens, polymorphism in the promoter of the bcmo1 gene is strongly related to variations in breast meat color [30], while the yellow skin phenotype is caused by a mutation in the tissue-specific regulatory elements of the bcmo2 gene [43]. In the present study, multiple lines of evidence indicate that the bco1l deletion mutation represents the genetic basis underlying the yellow color of the juvenile body. Although carotenoid content in yellow juveniles could not be determined in the present study due to the limited number of samples, the association between body color variation and differences in carotenoid content has been reported in previous studies [7].

Loss-of-function mutations in core β-carotene metabolic genes that drive endogenous β-carotene accumulation have been shown to exert multifaceted physiological and applied benefits across domestic livestock, commercially important aquaculture species, and wild and ornamental birds. Specifically, this β-carotene enrichment significantly enhances systemic antioxidant capacity and stress resistance, resulting in a 15–20% increase in culture survival rate in bco2-mutant yesso scallops, as well as improved oxidative stability of skeletal muscle during cold storage in Chinook salmon [46,47]; modulates innate immunity and inhibits inflammatory responses, with enhanced resistance to common pathogenic bacteria observed in bco2-mutant sheep [48]; optimizes lipid metabolism profiles and reduces circulating biomarkers associated with atherosclerosis risk in mutant rabbits [49]; markedly improves reproductive performance, including enhanced semen quality in mutant rabbits and elevated fertilization and hatching rates of eggs in coho salmon [45,49]; and further confers adaptive advantages in wild animal populations. For instance, loss-of-function mutations in bco2a enhance tolerance to nutrient fluctuation and chronic hypoxia in cave-adapted fish (Astyanax mexicanus) [50]. The identification of the bco1l-Δ10 mutation in large yellow croaker provides an effective molecular marker for marker-assisted selection (MAS) breeding in L. crocea. By genotyping broodstock at this deletion locus, it may be possible to obtain offspring with more intense yellow body coloration, and on this basis, to further evaluate potential improvements in immunity, disease resistance, reproductive performance (such as egg quality and hatching success), and larval survival rate, all of which are critical traits that significantly contribute to enhancing aquaculture productivity and profitability.

Although polymorphism of carotenoid oxidase genes has been documented in multiple animal lineages, the processes responsible for generating and maintaining such genetic variation are still unclear [46]. Studies of Darwin’s finches indicated that their nestling beak color polymorphism is controlled by a regulatory mutation modulating bco2 expression. This balanced polymorphism maintains stable frequencies in populations while fluctuating according to dietary carotenoid levels, with further modulation by ecological selection and interspecific gene flow [33]. It has previously been shown that fish have evolved and maintained two bco1 genes prior to the teleost–tetrapod split through a gene duplication event [28]. Therefore, it has been hypothesized that the retention of an extended family of beta-carotene oxygenase in teleost fish is due to better utilization of the diverse carotenoids in an aquatic environment [28]. However, under aquaculture conditions, the amount and composition of carotenoids in formulated feed are different from those in natural bait [51]. Therefore, we hypothesized diet as an important factor of selection pressure to induce bco1l mutation in the croaker. For example, bco1l mutation-induced excess carotenoid accumulation can be metabolized later at a time of lower intake. In addition, egg quality is an important selection trait in the genetic breeding process of the large yellow croaker [3]. The enrichment of carotenoids in egg yolk is one of the important factors determining egg quality in fish [52]. Therefore, the genetic variations that cause the egg to be rich in carotenoids can be retained during the breeding process.

Consistent with our findings, studies in reptiles and birds have found that color polymorphisms can often be explained by a single major effect locus [53,54]. The role of a single gene in driving differences between color morphs is surprising, given that morphs often show correlated differences in other key traits. Indeed, differences in reproduction, predation, and physiology have been associated with carotenoid pigmentation in salmonids [55,56,57]. Therefore, fitness-related differences between morphs suggest that bco1l may have pleiotropic effects on other biological processes in large yellow croakers.

5. Conclusions

This study identifies a homozygous 10 bp deletion in the bco1l gene (bco1l-Δ10) as the primary genetic determinant of the yellow body coloration in large yellow croaker (L. crocea). This mutation disrupts the protein’s ability to catalyze β-carotene, likely promoting carotenoid accumulation. By clarifying the role of bco1l in teleost pigmentation, these findings provide a critical molecular target for precision breeding programs aimed at developing novel strains with dual ornamental and commercial value in L. crocea. While further biochemical validation of pigment deposition is needed, this work establishes a significant genetic framework for understanding color polymorphisms in fish.