Comparative analysis of the extreme meat color transcriptome reveals key regulatory genes and identifies potential SNPs in Leiqiong cattle Authors: Wenhui Zhang, Zongzhou Chen, Xiaowen Li, Yanyan Zhang, Yanhong Yong, Xiaoxi Liu, Zhichao Yu, Xingbing Ma, A M Abd El-Aty, Xianghong Ju Published: 2026 ABSTRACT AbstractBackgroundLeiqiong cattle, an indigenous beef breed from southern China, possess remarkable adaptability to tropical climates and demonstrate significant market potential owing to their high-quality meat. Meat color serves as a critical quality trait that strongly influences consumer preference and commercial value. Meat color in beef is primarily governed by myoglobin redox chemistry, mitochondrial activity, and post-mortem metabolism-processes modulated by oxidative stress, chaperone networks, and glycolytic flux. While these pathways are recognized in commercial breeds, their genetic architecture in heat-adapted, tropical cattle like Leiqiong remains unknown. Here, we bridge this gap by integrating transcriptomics and targeted genotyping to uncover breed-specific regulatory mechanisms.ResultsTo elucidate the genetic basis of meat color variation, we conducted transcriptome sequencing and comparative analysis between light- and dark-colored meat samples from 8 cattle. We identified 490 differentially expressed genes (328 upregulated and 162 downregulated), among which GSTP1, HSPA2, and LDHD emerged as key candidate genes. Polymorphism screening revealed 25 SNPs across these loci in 100 cattle, with association analysis confirming the significant effects of specific variants (GSTP1: g.45441935G > A, g.4544191937G > A, g.45444424G > A; HSPA2: g.77680461G > A; LDHD: g.2453066T > G) on meat color phenotypes.ConclusionsOur integrated analysis identified GSTP1, HSPA2, and LDHD as key candidate genes and revealed specific SNPs significantly associated with meat color variation in Leiqiong cattle. These findings provide novel insights into the breed-specific genetic mechanisms underlying this trait and supply preliminary molecular markers that could inform future marker-assisted selection strategies aimed at improving meat color.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12864-026-12911-8.Keywords: Leiqiong cattle, Meat color, Transcriptomics, Genetic single nucleotide polymorphisms Background Leiqiong cattle, an indigenous breed from southern China’s tropical regions, exhibit notable adaptability to heat, humidity, and disease, giving them considerable potential for beef production in challenging environments. However, their commercial value is limited by the naturally light color of their meat. Meat color is a direct sensory evaluation index for consumers to choose meat products [1]. In Leiqiong cattle, this light hue is often visually misclassified as pale, soft, and exudative (PSE)-like, leading to discounted value even within its primary regional markets. Consequently, breeding for a moderately darker and more stable red color-while maintaining normal pH-could enhance visual appeal and market competitiveness.Achieving this goal requires a deeper understanding of the transcriptional and genetic basis of meat color variation in this specific breed. Identifying key molecular players and genetic variants is therefore a crucial first step toward enabling marker-assisted selection for improved meat color. Meat color is a heritable, polygenic trait whose variation is influenced by both genetic and environmental factors [2–5]. Crucially, converging evidence from multi-omics studies in various cattle breeds has established a core set of biological pathways underpinning this trait, including oxidative stress response, energy metabolism, and protein homeostasis [6]. For instance, genome-wide association studies in breeds like Nellore have mapped specific genomic regions and candidate genes (e.g., HTRA1, ACADSB) affecting color [7], while proteomic profiles in Ankur cattle have consistently highlighted molecular chaperones (e.g., HSP40, HSP70) and glycolytic enzymes (e.g., LDH, ENO3) as key biomarkers [8]. These findings collectively confirm that meat color regulation is governed by a conserved yet complex molecular network, and that its genetic architecture can be effectively dissected using modern genomics tools. However, the genetic components identified are often breed- or population-specific, reflecting unique selection histories and genetic backgrounds. To date, no such systematic investigation has been conducted in Leiqiong cattle. Consequently, while the general importance of genetics is well-recognized, the specific genes, variants, and regulatory mechanisms responsible for the characteristic light meat color in this adapted indigenous breed remain entirely unknown. Given this gap, an effective strategy is to first identify breed-specific candidate genes through transcriptomic profiling of extreme phenotypes, and then screen for functional genetic variants within them. Single nucleotide polymorphisms (SNPs) serve as ideal markers for this purpose. As a fundamental source of phenotypic diversity, SNPs occur frequently in the genome [9] and can directly influence gene function and protein structure when located in coding or regulatory regions [10, 11]. Owing to their stability, abundance, and ease of genotyping, SNPs have become the marker of choice in modern livestock breeding programs [12–15]. In recent years, numerous trait-associated SNPs have been identified and successfully integrated into selection schemes to improve production efficiency [16]. Therefore, identifying and validating breed-specific SNPs associated with meat color constitutes a logical and necessary step in developing molecular tools for Leiqiong cattle breeding. To this end, we first characterize the transcriptomic landscape underlying extreme meat color phenotypes in Leiqiong cattle. From this analysis, we will identified key differentially expressed genes linked to meat color biology. Subsequently, we screened for SNPs within these top candidate genes and evaluated their association with quantitative meat color traits. Methods Animals One hundred 2-year-old healthy Leiqiong cattle were selected from Yangkan Town, Suixi County, Zhanjiang City, Guangdong Province, under the same feeding conditions. The experimental procedures and sample collection were conducted with the informed consent of the farm management. After 8 h of fasting and water fasting, the slaughter process was carried out in a commercial abattoir using a penetrating captive bolt stunning method followed by exsanguination, in strict accordance with Chinese animal welfare regulations and standard ethical practices to minimize suffering. Samples were collected from the Longissimus thoracis muscle at the level of the 12th–13th rib interface, 10 g was frozen immediately, and the remaining tissue was used for other assays. All animals were slaughtered routinely as implemented in the normal farm production cycle. The animals were not slaughtered specifically for this study, and the research did not include any direct physical contact with the animals. The animal procedures followed the OIE Terrestrial Animal Health Code. All the samples were collected postmortem from standard slaughter processes. Color measurements Among these 100 animals, meat color was measured for all individuals at 1 h and 24 h post-mortem to characterize the population’s color distribution via a colorimeter (CR-10 Plus, KONICA MINOLTA, Japan) and expressed in CIE L*, a* and b* units (Table S1). To obtain a representative color value for each muscle sample, five measurements were taken at random locations on the cut surface, avoiding visible connective tissue and fat. These technical replicates were averaged to yield a single a, b, and L* value for each sample/time point for subsequent statistical analysis. From this population, two extreme groups were selected for transcriptomic analysis: four individuals with the darkest meat color (DC group, a* > 9.8) and four with the lightest meat color (LC group, a* < 7.8). Muscle samples for RNA extraction were collected only from these eight extreme individuals. Subsequently, all 100 animals were genotyped for polymorphisms in the candidate genes identified from the RNA-seq analysis. These thresholds were based on the upper and lower quartiles of the a* value distribution from our initial 100-animal survey, ensuring a distinct phenotypic contrast for the extreme group design. The hue angle (H) and chroma (saturation index) (C) were calculated according to the following formulas: C=(a*2+b*2)0.5 and H= arctg(b*/a*). RNA-seq For transcriptomic analysis, samples of the Longissimus thoracis (LT) muscle were collected within 1 h post-mortem from the region between the 12th and 13th ribs. Approximately 100 mg of tissue was immediately snap-frozen in liquid nitrogen and stored at − 80 °C until RNA extraction. The remaining muscle tissue was used for subsequent phenotypic and DNA analyses. Leiqiong cattle muscle samples were collected and used for RNA extraction (TRIzol, TAKARA, Cat. No. 9108). Total RNA integrity was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies), and only samples with an RNA Integrity Number (RIN) > 7.5 were used for library construction. RNA concentration and purity were confirmed using a NanoDrop One spectrophotometer (Thermo Fisher Scientific), with A260/A280 ratios between 1.8 and 2.0. The muscle mRNA of Leiqiong cattle was enriched by magnetic beads with oligo (dT). Fragmentation buffer was added to randomly interrupt the mRNA of Leiqiong cattle. Using the mRNA of Leiqiong cattle as a template, the first cDNA strand was synthesized by random hexamers, and then the second cDNA strand was synthesized by adding buffer, dNTPs, RNase H and DNA polymerase I. The cDNA of Leiqiong cattle was purified by AMPure XP beads. The purified Leiqiong cattle double-stranded cDNA was subjected to terminal repair, a tail was added, the sequencing adapter was connected, and the fragment size was subsequently selected with AMPure XP beads. Finally, a cDNA library was obtained via PCR enrichment. After the completion of library construction, the effective concentration of the libraries (the effective concentration of the libraries > 2 nM) was accurately quantified via Q‒PCR to ensure the quality of the libraries. After passing the library inspection, different libraries were pooled according to the target downstream data volume and sequenced on the Illumina platform. Raw sequencing reads were quality-filtered to obtain clean data. Clean reads were then aligned to the bovine reference genome (UMD3.1) using HISAT2 (version 2.2.1) with default parameters. Gene-level read counts were generated using featureCounts (version 2.0.1). Differential expression analysis was performed using the R package edgeR (version 3.40.2). Genes with an absolute log2 fold change (|log2FC|) ≥ 1.5 and a false discovery rate (FDR) < 0.05, adjusted using the Benjamini–Hochberg method, were considered significantly differentially expressed. All the identified differential genes (DEGs) in the light-colored muscles of Leiqiong cattle were subjected to gene function annotation analysis on the basis of the KOG, COG, GO, and KEGG databases and the BMK Cloud (www.biocloud.net). The sample size for RNA sequencing (n = 4 per extreme group) was determined based on the extreme phenotype design strategy, which is well-established for initial discovery-phase studies of complex traits in livestock. This approach maximizes the contrast in gene expression between phenotypic extremes, thereby increasing the probability of detecting differentially expressed genes even with limited biological replicates. We acknowledge that this sample size may limit the detection of genes with subtle expression changes. Quantitative real-time PCR (qPCR) analysis Total RNA was isolated from Leiqiong cattle muscle samples via TRIzol reagent (TaKaRa, Cat. NO. 9108) in accordance with the manufacturer’s instructions. Total RNA (500 ng) was used to synthesize cDNA via a PrimeScriptTM RT Reagent Kit with gDNA Eraser (TaKaRa, Cat. No. RR047A). The transcript levels of different genes were evaluated via qRT‒PCR via the Applied Biosystems 7300 Plus Real-Time PCR System with Hieff-qPCR SYBR Green Master Mix (Yeasen, Cat. No. 11203ES03). The primers used for qRT‒PCR are listed in Table S2. DNA isolation and genotyping Polymorphism analysis of the Gstp1, Hspa2 and Ldhd genes. DNA was isolated from the muscle tissues of 100 Lei qiong by using a TIANamp Genomic DNA Kit (Tiangen, DP304) according to the manufacturer’s instructions. All primers used are listed in Table S3. SNPs were identified through Sanger sequencing of PCR-amplified genomic regions spanning all exons, introns, and UTRs of the three candidate genes (GSTP1, HSPA2, and LDHD) from 100 Leiqiong cattle. Variant calling was performed by aligning chromatograms to the reference UMD3.1 bovine genome using SnapGene, followed by manual inspection to confirm polymorphic sites. This approach was chosen due to its high accuracy for targeted genotyping in non-model populations. PCR was performed in a 25µL reaction volume containing approximately 50 ng of genomic DNA, 0.2µM of each primer, and 12.5µL of 2×Taq PCR MasterMix (Vazyme, P112). The thermocycling conditions were: initial denaturation at 95℃for 5 min; followed by 35 cycles of 95 °C for 30 s, primer-specific annealing temperature (see Table S3) for 30 s, and 72℃ for 1 min per kb of expected product size; with a final extension at 72℃for 5 min. PCR products were purified and subjected to bidirectional sequencing on an ABI 3730xl DNA Analyzer (Applied Biosystems). SNP calling and genotype assignment were performed by visual inspection of sequencing chromatograms and alignment of sequences using SnapGene software (version 6.0.2). Statistical analysis All measurements were acquired using independent samples rather than collected with repeated measurements. GraphPad Prism V.8 (GraphPad Software; San Diego, California) was used for data analysis, and the data were shown as means ± SD, unless stated otherwise. Two-tailed student’s t-test was used to conduct statistical analysis, unless stated otherwise. A p value < 0.05 was regarded as statistically significant. The association between each SNP genotype and the binary meat color phenotype was assessed as follows: Genotype Frequency Comparison: Within the DC group and the LC group, the frequency of each genotype (e.g., AA, AG, GG) was calculated as: Results Phenotypic divergence in meat color between extreme groups To screen genes related to meat color, the cattle used for RNA-seq were characterized by an extreme phenotype for meat color. According to the CIE records (Table S1), four DC (dark color) individuals (2 female cattle and 2 male cattle) and four LC (light color) individuals (2 female cattle and 2 male cattle) were selected for meat color. The results indicated that the a* value of meat color was significantly different between the LC and DC groups (1 h: mean DC = 11.04, SD-DC = 1.47, mean LC = 6.11, and SD-LC = 1.4, p < 0.05; 24 h: mean DC = 12.74, SD-DC = 1.75, mean LC = 6.86, and SD-LC = 1.64, P < 0.05) (Fig. 1A). Preliminary analysis of phenotypes and transcriptome profiles of diverse meat colors. A Comparison of meat color phenotypes. Letters indicate significant differences between the dark and light color groups. B Principal component analysis by meat color was performed. The proportion of variability explained is shown in parentheses. C All gene expression in each sample. D Volcano plots of DEGs related to meat color. The green dots represent the downregulated genes; the red dots represent the upregulated genes; and the black dots represent the nondifferentially expressed genes. E KOG analysis. F Heatmap of genes related to posttranslational modification, protein turnover, chaperone activity and energy production and conversion. G and H Expression levels of the candidate DEGs Gstp1, Hspa2 and Ldhd via RNA-seq (G) and RT‒PCR (H). (* indicates p < 0.05, ** indicates p < 0.01 and *** indicates p < 0.001) Transcript sequencing analysis To screen important genes related to meat color, we performed transcript sequencing. Principal component analysis (PCA) revealed clear segregation of the samples on the basis of meat color (Fig. 1B). To contextualize our search for meat color-related genes, we first assessed the overall transcriptional similarity between the two extreme phenotype groups. Visual comparison of the global gene expression profiles, as summarized in (Fig. 1C), revealed substantial overlap between the light (LC) and dark (DC) meat groups. This indicates that the stark difference in meat color is not driven by a genome-wide transcriptional shift but rather by more targeted molecular changes. This focused background allowed us to subsequently identify specific, high-confidence differentially expressed genes (DEGs) with greater precision, as detailed in the following analyses. Differential expression analysis revealed 490 DEGs related to meat color. More specifically, compared with those in LC meat, there were 328 upregulated genes and 162 downregulated genes in DC meat. The volcano plots of the DEGs are shown, and the results revealed that the expression patterns of the DEGs were consistent with those of the sample groups (Fig. 1D and Table S4). To investigate the functional implications of the transcriptional changes, all 490 differentially expressed genes (DEGs) were subjected to enrichment analysis. Functional annotation and categorization were performed using the KOG/COG, GO, and KEGG databases, with the complete set of genes detected in our LT muscle transcriptome serving as the background reference. KOG analysis of the DEGs highlighted significant enrichment in categories related to posttranslational modification, protein turnover, chaperone activity, and energy production and conversion (Fig. 1E). To prioritize candidate genes for functional validation and polymorphism screening, we applied a multi-step filtering strategy. From the 490 DEGs, we first focused on those annotated to the KOG categories most significantly enriched in our analysis ( Posttranslational modification, protein turnover, chaperones and Energy production and conversion), as these pathways are strongly implicated in meat color biochemistry. Second, within this subset, we selected genes that exhibited both a high magnitude of differential expression (absolute log2 fold change > 2.0) and high statistical significance (FDR < 0.01). Finally, we considered biological relevance based on prior literature linking gene function to oxidative stress, protein homeostasis, or glycolysis in meat quality. This integrative approach identified Glutathione S-transferase P1 (GSTP1) [17], Heat Shock Protein Family A Member 2 (HSPA2) [18], and Lactate Dehydrogenase D (LDHD) [19, 20]as top-tier candidates (Fig. 1F and G; Table 1). Their differential expression was successfully validated by qRT-PCR (Fig. 1H), confirming the RNA-seq results. Table 1Meat color-related genes identified via integrated analysisGeneBiological functionFunctionReferenceGSTP1Intracellular oxidative processesRegulated muscle oxidative processes on beef color(Falomir-Lockhart et al., 2015) [17]HSPA2Heat shock proteinsRegulated Inosine 5′-monophosphate metabolism(Ma et al., 2015) [18]LDHDglycolysis processRegulated meat color stability(Kim et al., 2006; Kim, Keeton, Smith, Berghman, & Savell, 2009) [19, 20] Meat color-related genes identified via integrated analysis GO analysis was performed (Fig. S1A-C), and the DEGs were classified into three primary categories: biological process (BP), cellular component (CC) and molecular function (MF). A total of 58 subcategories were obtained, including 23 BP terms, 18 CC terms and 17 MF terms. For the biological process category, most genes were annotated into the following categories: “cellular process”, “single-organism process” and “biological regulation”. For the CC category, most genes were annotated as cell, cell part or organelle. For molecular function, most genes were annotated as binding, catalytic activity or molecular transducer activity. To further systematically investigate the potential biological significance of phenotype divergence, the identified DEGs were annotated into various pathways via KEGG bioinformatic databases (Fig. S1D). The results revealed that the enriched pathways included the phagosome, the acin cytoskeleton, apoptosis, cellular senescence, and focal adhesion. Gstp1, Hspa2 and Ldhd gene polymorphism analysis To analyze the gene polymorphisms of the DEGs (Gstp1, Hspa2 and Ldhd), we collected muscle tissues from 100 Leiqiong cattle and amplified the whole exons, introns and UTRs. The results revealed 11 SNPs in Gstp1, 1 SNP in Hspa2 and 13 SNPs in Ldhd (Fig. 2A-C). A total of 25 SNPs were found. Analysis of polymorphisms in the Gstp1, Hspa2 and Ldhd genes. A Representative sequence chromatograms of PCR amplification products at the Gstp1 locus. B Representative sequence chromatograms of PCR products amplified at the Hspa2 locus. C Representative sequence chromatograms of PCR products amplified at the Ldhd locus Association analysis of Gstp1, Hspa2 and Ldhd gene polymorphisms with meat color We collected 100 cattle samples and detected ICE and Gstp1, Hspa2 and Ldhd gene polymorphisms in each type of cattle (Fig. 3A and Table S4-5). The allele frequencies are shown in Table 2. To analyze the effects of different gene types on meat color, we analyzed the ICE score and gene polymorphisms (Fig. 3B-F and Fig. S2-5), which revealed that Gstp1 (g.45441935G > A, g.45441937G > A and g.45444424G > A), Hspa2 (g.77680461G > A) and Ldhd (g.2453066T > G) had significant effects on dark meat color (Fig. 3B-F). On the basis of these findings, when we screened shared cattle, we obtained 4 cattle with all these SNPs. Notably, 2 of these 4 cattle exhibited a dark color (Fig. 4A and B). These results imply that the SNPs we screened are valuable for the conservation of good traits in Leiqiong cattle. Association analysis of Gstp1, Hspa2 and Ldhd gene polymorphisms with meat color. A Distribution of a* values in cattle. B Meat color score of different Gstp1 g.45441935G>A (upper), and Proportion of dark meat color compared with light meat color (down). C Meat color score of different Gstp1 g.45441937G>A (Left), and Proportion of dark meat color compared with light meat color (Right). D Meat color score of different Gstp1 g.45444424G>A (Left), and Proportion of dark meat color compared with light meat color (Right). E Meat color score of Hspa2 g.77680461G>A (left) and the proportion of dark meat color compared with light meat color (right). F Metal color score of Ldhd g.2453066T>G (Left) and the proportion of dark meat color compared with light meat color (right).Distribution of a* values in cattle Table 2Allele frequencies of polymorphisms in the Gstp1, Hspa2, and LDHD genes in 100 Lei qiong cattleGene nameSNP nameAllele frequency (%)Gstp1g.45441933G > CGG (75)GC (21)CC (4)g.45441935G > AGG (17)GA (33)AA (50)g.45441937G > AGG (17)GA (24)AA (56)g. 45,441,998 C > TCC (92)CT (8)TT (0)g. 45442078G > CGG (12)GC (45)CC (43)g.45,442,471 C > GCC (22)CG (37)GG (41)g.45,442,616 C > TCC (2)CT (25)TT (73)g.45,444,249 A > GAA (0)AG (22)GG (78)g.45444424G > AGG (15)GA (49)AA (36)g.45,444,484 A > CAA (66)AC (34)CC (0)g.45444613G > AGG (84)GA (16)AA (0)Hspa2g.76680461G > AGG (4)GA (31)AA (65)Ldhdg.2453066T > GTT (31)TG (44)GG (25)g.2453080T > CTT (61)TC (37)CC (2)g.2453710T > CTT (6)TC (13)CC (81)g.2453823T > CTT (8)TC (25)CC (67)g.2453892T > CTT (8)TC (25)CC (67)g.2454359G > CGG (8)GC (23)CC (69) Allele frequencies of polymorphisms in the Gstp1, Hspa2, and LDHD genes in 100 Lei qiong cattle Screening dark color meat by Gstp1, Hspa2 and Ldhd gene polymorphisms in Lei Qiong cattle. A Venn diagram illustrating the overlap among cattle carrying these specific SNPs (related to Fig. 3). B CIE values of cattle carrying SNPs Discussion Improving meat color, a paramount determinant of consumer preference and market value, is of specific economic importance for indigenous breeds like Leiqiong cattle. While the genetic architecture of this complex trait has been explored in several commercial breeds [21–24], it remains largely uncharted in tropical-adapted breeds shaped by distinct selection pressures. Our study implemented an integrated transcriptomic and candidate gene association strategy in Leiqiong cattle. We identified GSTP1, HSPA2, and LDHD as key candidates and reported specific SNPs significantly associated with meat color variation. More importantly, we anchored these molecular players within the enriched biological pathways of oxidative stress response, protein homeostasis, and energy metabolism. This systems-biology perspective allows us to propose a coherent, albeit preliminary, model for meat color regulation in this breed and to thoughtfully contextualize our findings within both their biological implications and inherent limitations. The comparative transcriptome analysis revealed 490 differentially expressed genes (DEGs) between light- and dark-colored meat groups. Enrichment analyses highlighted pathways integral to post-translational modification, chaperone activity, energy metabolism, and cellular stress response (Fig. 1. E). These pathways are mechanistically linked to meat color through their roles in myoglobin stability [25, 26], redox homeostasis [27], and post-mortem glycolysis [28, 29]. Within this established biochemical framework, the roles of our candidate genes gain mechanistic plausibility. GSTP1, a pivotal enzyme in glutathione metabolism, is a direct functional effector within the oxidative stress response pathway [30]. Its elevated expression in darker meat may enhance the scavenging of reactive oxygen species, thereby potentially slowing the oxidation of red oxymyoglobin to brown metmyoglobin—a primary determinant of color stability. The associated SNPs (e.g., GSTP1 g.45441935G > A) could contribute to this phenotype by altering gene expression or enzyme kinetics. Similarly, HSPA2, a molecular chaperone, represents a node within the enriched protein homeostasis network. Its differential expression suggests a role in mitigating protein denaturation and aggregation during post-mortem acidification, processes that influence light scattering and cellular structure, thereby indirectly affecting color [31–33]. The link to LDHD within the energy metabolism pathway is particularly compelling. By influencing D-lactate metabolism and the NADH/NAD+ ratio, LDHD activity could modulate the reducing capacity of the muscle [34, 35]. Since NADH is an essential cofactor for metmyoglobin reductase, genetic variation affecting LDHD (e.g., LDHD g.2453066T > G) presents a logical, though unverified, mechanism to influence the myoglobin redox balance. Thus, our integrated data suggest that genetic variation impacting these three interconnected pathways—antioxidant defense, protein stability, and metabolic flux—collectively shapes the post-mortem muscle environment to produce the observed meat color variation. The significant associations of specific SNPs with meat color are a key finding, yet their interpretation requires nuance. A critical observation is the incomplete penetrance of the combined “favorable” SNP genotype, where not all carrier animals exhibited dark color. This is a classic signature of a highly polygenic trait, where the phenotype results from the aggregate effect of many genetic variants across the genome, often with small individual effects [36–39]. Our candidate SNPs are thus likely contributors within this larger network, not sole determinants. This view is supported by modern beef genomics, which emphasizes the complex genetic networks underlying quality traits [40] and the documented additive or epistatic interactions among multiple loci, such as CAPN1, CAST, and GHR, in shaping meat quality [41–43]. Consequently, the proportion of phenotypic variance explained by our three candidate SNPs is necessarily partial. Therefore, our findings represent initial, partial insights into the genetic basis of meat color. To fully quantify the contribution of these variants and place them within the broader genetic landscape, future work will require validation and modeling in expanded cohorts, integrating genome-wide data to distinguish their effects from the polygenic background. This study has several important limitations that must be acknowledged, as they delineate the scope of our current conclusions and chart a clear path for essential future research. The primary constraints stem from its discovery-phase design. The transcriptomic analysis, while efficient, was based on a modest sample size from extreme phenotypic groups, which may affect the statistical robustness and generalizability of the full list of differentially expressed genes and could overlook variants influencing the continuous spectrum of meat color. Consequently, the associations between the identified SNPs in GSTP1, HSPA2, and LDHD and meat color, though significant within our cohort, require validation in an independent and larger population of Leiqiong cattle to confirm their effect size and breed-wide applicability. Furthermore, our candidate-gene approach, focused on three high-priority targets, did not explore the wider genome, potentially missing additional contributing loci and regulatory landscapes. Crucially, the functional consequences of these SNPs on gene regulation or protein activity, and thus the causal roles of the genes, remain inferential without direct experimental validation. The analysis did not explicitly model potential confounding factors such as detailed pre-slaughter conditions, muscle pH kinetics, or sex, nor did it formally test for genotype-by-sex interactions. We also did not evaluate the local haplotype or linkage disequilibrium structure to assess whether the identified SNPs are causal or are tagging nearby functional variants. Finally, while we estimated the individual contribution of these SNPs, their additive, dominance, and interaction effects within a model accounting for the highly polygenic background were not quantified. To address these limitations, future work should prioritize: (i) validating the core genetic associations in expanded, independent cohorts [43]; (ii) employing functional genomics assays (e.g., reporter assays, enzyme kinetics) to establish molecular causality and determine if the SNPs act as cis-eQTLs [40]; (iii) conducting genome-wide association or genomic selection studies to capture the full polygenic architecture and integrate these markers into a robust prediction framework [44]; and (iv) utilizing mixed linear models that include key environmental and physiological covariates to better isolate the genetic signal [44]. Notwithstanding its limitations, this study provides foundational genetic and mechanistic insights for Leiqiong cattle. It shifts the focus from gene lists to pathway-informed understanding, proposing a testable model centered on oxidative stress and metabolic flux. For breeding, the identified SNPs are preliminary candidate markers. Their practical utility will depend on validation and, more importantly, their integration into a genomic selection framework that accounts for polygenicity, which will be more effective than single-marker selection. For biological understanding, this work highlights specific, interconnected biochemical modules as promising focal points for functional studies. The integrated approach demonstrated here serves as a template for investigating complex traits in other under-characterized, locally adapted livestock breeds, balancing the efficiency of discovery-phase research with a clear-eyed view of the necessary steps toward validation and application. Conclusion This study clarifies the genetic architecture of meat color in Leiqiong cattle by linking transcriptional profiles to genetic variation. We identified crucial biological pathways from transcriptomic data and pinpointed GSTP1, HSPA2, and LDHD as core candidate genes. Furthermore, we discovered specific SNPs within these genes that are robustly associated with phenotypic variation. These findings offer novel insights into the molecular regulation of meat color in this breed and deliver tangible genetic markers for potential use in selective breeding programs. Supplementary Information Supplementary Material 1. Supplementary Material 2. Acknowledgements The authors thank cattle breeders for their collaboration in this study. Authors’ contributions WZ and ZC drafted the original manuscript under supervision of XJ. WZ performed the analyses under supervision of XJ. ZC, YZ and XL performed data curation. WZ and XJ contributed to the funding acquisition. ZC, YZ, XL, YY, XL, ZY, XM and AE contributed to the data collection and conception of this study. All authors read and approved the final manuscript. Funding This work was supported by Guangdong Provincial Department of Agriculture and Rural Affairs, Modern Agricultural Industry Technology System, “Southern Modern Grassland Animal Husbandry (Cattle) Innovation Team“(No.2026CXTD13) and Program for Scientific Research Start-Fund of Guangdong Ocean University(No.101402/R17088). Opening Project of Guangdong Key Laboratory of Animal Breeding and Nutrition: 2021SZ02. Data availability The datasets generated during the current study are available in the Sequence Read Archive (SRA)repository and the BioProject accession: PRJNA1348233. Declarations Ethics approval and consent to participate All animals were privately owned by local farms. The experimental procedures and sample collection were conducted with the informed consent of the farm management. The investigation was approved by the Institutional Animal Care and Use Committee (IACUC) at Guangdong Ocean University (ethics approval number 2021-scuec-133). Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Wenhui Zhang, Zongzhou Chen and Xiaowen Li contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Data Availability Statement The datasets generated during the current study are available in the Sequence Read Archive (SRA)repository and the BioProject accession: PRJNA1348233.