Arachidonic Acid Metabolic Rewiring Drives Differential Plant Protein Adaptation in Golden Pompano (Trachinotus ovatus)

TL;DR

Replacing fishmeal with plant protein in golden pompano diets causes growth differences linked to changes in liver metabolism and inflammation.

Contribution

This study reveals how arachidonic acid metabolism and liver changes drive growth variation in fish on plant-based diets.

Findings

Soy protein diets reduced growth and feed efficiency in golden pompano compared to fishmeal diets.

Arachidonic acid metabolism was significantly altered in fish fed plant-based diets.

Growth divergence was linked to liver inflammation and disrupted tight junctions.

Abstract

The replacement of fishmeal with plant protein is widely regarded as a key strategy for sustainable aquaculture. However, carnivorous marine fish often show limited tolerance to fishmeal-free diets. Here, we investigated growth performance, hepatic physiological responses, and molecular mechanisms underlying adaptation to a soy protein concentrate-based diet (SPCD) in golden pompano (Trachinotus ovatus). An 8-week feeding trial was conducted under communal rearing conditions, followed by the phenotypic stratification of SPCD-fed fish into high- and low-growth subgroups. Growth performance, serum biochemical indices, and liver histology were assessed, and integrated transcriptomic and metabolomic analyses were performed on liver tissue. At the population level, the SPCD resulted in reduced growth, a lower feed intake, and decreased feed utilization efficiency compared with a fishmeal-based diet. Notably, marked inter-individual variation was observed: fish fed the SPCD exhibited significantly lower final body weights and a higher FCR compared with the FMD group (p < 0.001), and pronounced growth divergence was observed between the PB and PS subgroups, with a subset of SPCD-fed fish maintaining growth comparable to fishmeal-fed controls, whereas others exhibited severely constrained growth. Divergent phenotypes were associated with distinct hepatic alterations, including aggravated vacuolation, the enrichment of tight junction-related and immune regulatory pathways, and the broad reprogramming of lipid metabolism. Integrated multi-omics analysis identified arachidonic acid metabolism as the most significantly perturbed pathway, characterized by altered membrane phospholipid composition, the upregulation of RARRES3L, increased COX/LOX-derived eicosanoids, and the suppression of the CYP–EET branch. Collectively, these findings indicate that soy protein replacement induces coordinated hepatic structural and metabolic remodeling, with tight junction disruption and arachidonic acid metabolic reprogramming contributing to inflammatory imbalance and divergent growth phenotypes in T. ovatus.

Figures

Full text

1. Introduction

Aquaculture is the most rapidly expanding industry in the global production of animal food, offering sustainable prospects and ensuring food security for an increasing global population [1] (FAO, 2022). This has generated an increase in the demand for fishmeal (FM) used as protein and lipid source in commercial fish feeds, a situation that is both environmentally and economically unsustainable [2]. Consequently, there have been studies that have used plant proteins to replace FM [3]. In the search for viable plant ingredients in fish diets, soybean-derived products have emerged as a primary research subject, owing to their broad accessibility, economic advantage, and generally suitable nutrient content [4]. Soybean meal (SBM) and soy protein concentrate (SPC) have been successfully incorporated into diets for several carnivorous marine fish species including the rainbow trout (Oncorhynchus mykiss) [5], olive flounder (Paralichthys olivaceus) [6], Tiger puffer (Takifugu rubripes) [7], and red seabream (Pagrus major) [8] without significantly affecting fish growth. These findings indicate that the partial replacement of FM with plant proteins is feasible in many cultured species.

The golden pompano, Trachinotus ovatus, is an important marine economic fish; farming output reached a record of 292,263 tons in 2023 in China [9]. As the highest-yielding marine cultured fish in China and a carnivorous species of significant aquaculture value, T. ovatus exhibits a lower capacity for utilizing plant-based proteins compared to omnivorous or herbivorous teleost fish, demonstrating a pronounced dietary reliance on FM for optimal growth performance [10,11,12]. The limited supply of high-quality FM has become a critical constraint in the intensive cultivation of this species. Notably, increasing evidence from other carnivorous fish species suggests substantial intraspecific variation in the ability to tolerate and utilize plant protein; in some studies on European seabass (Dicentrarchus labrax) [13] and large yellow croaker (Larimichthys crocea) [14], it was observed that some populations showed significant differences in tolerance to plant protein, manifested as differences in growth rate, feed conversion ratio, and feeding response latencies. This phenotypic heterogeneity suggests the presence of genetic subgroups with an enhanced ability to utilize plant protein. Some studies on salmonids have suggested there is genetic variability with regard to the ability to utilize plant-based diets [15,16]. A strain of rainbow trout (UI ARS-CX) has been developed through selective breeding, which exhibits exceptionally high growth rates when fed an all-plant protein diet [17]. However, comparable studies focusing on plant protein utilization and adaptive variation in T. ovatus remain extremely limited.

The process of nutrient digestion and absorption in fish involves several key organs: the stomach, intestines, and liver. Initially, proteins are broken down into peptides within the stomach. These peptides are then further digested into amino acids as they move through the intestines. Lipids and carbohydrates, on the other hand, are absorbed mainly as fatty acids and simple sugars. Once absorbed, they pass through the epithelial cells lining the intestines and are transported to the liver, where they undergo further metabolic processing [18,19]. However, plant-based feeds frequently contain antinutritional factors, such as protease inhibitors, phytates, lectins, and soy antigens, which may impair liver function by interfering with normal digestive and metabolic processes [20]. Therefore, the efficiency of nutrient utilization may be compromised. For instance, after red seabream were fed with soy-based feed, hepatocyte atrophy was observed [21]. Replacing FM with a blend of plant proteins in yellow catfish has been associated with increased hepatic vacuolization, reduced antioxidant capacity, and elevated inflammatory responses [22]. Grouper that are provided with high amounts of plant-based protein in their diet show signs of liver vacuolization, accompanied by a decrease in lipid droplet accumulation [23]. Furthermore, substituting FM with a combination of plant proteins in the feed has been shown to negatively impact on liver health. This substitution leads to an increase in the expression of inflammation-related genes [24]. These findings underscore the importance of maintaining liver health for the successful adaptation of carnivorous fish, including T. ovatus, to FM-free or low-FM diets.

Transcriptome and metabolome analyses can provide many useful insights in fish physiological and nutritional studies [25,26,27]. Xun et al. [28] showed that dietary sodium acetate affects digestion and absorption, lipid metabolism pathways, and the intestinal immune system in T. ovatus. In the nutritional metabolism experiment conducted by Yoshinaga et al. [21] on P. major fed with soybean meal-based diets, the results of metabolomic and transcriptomic analyses indicated changes in various metabolic processes, such as glutathione metabolism and glycine metabolism, during the latter stage of the trial in the high-plant-protein group. Such multi-omics approaches provide critical insights into the mechanisms underlying dietary adaptation in marine fish.

This study aimed to elucidate the molecular mechanisms underlying adaptation to FM-free diets in T. ovatus by selecting individuals with markedly divergent growth phenotypes under communal rearing conditions. Integrated transcriptomic and metabolomic analyses were performed on liver tissue, followed by a combined analysis of differentially expressed genes (DEGs) and differentially accumulated metabolites (DAMs). These findings advance our understanding of the molecular mechanisms underlying adaptation to fishmeal-free diets in T. ovatus and may facilitate the development of strains with improved performance under such dietary conditions.

2. Results

2.1. Growth Performance and Nutrient Utilization

After the 8-week feeding trial, the growth performance and feed utilization parameters of T. ovatus fed the experimental diets are summarized in Table 1. No significant difference was observed in the survival rate between the two dietary groups (FMD: 98.8%; SPCD: 99%). However, fish fed the SPCD exhibited a significantly lower final body weight compared to those fed the fishmeal-based diet (p < 0.001). The total feed intake for the SPCD group was lower than that of the FMD group, while the feed conversion ratio (FCR) was markedly higher in the SPCD group, indicating reduced feed utilization efficiency under the FM-free dietary condition.

To further characterize phenotypic variation within the SPCD group, individuals were stratified by body weight, as described in the Methods. Growth performance indices of FMD-Ref (selected 20 fish from FMD with body weights close to the population mean) and the SPCD subgroups (PB and PS) (Table 2). The PS subgroup exhibited significantly lower final body weight, weight gain, and weight gain rate (WGR) compared with both FMD-Ref and PB (one-way ANOVA with Tukey’s test, p < 0.001), whereas no differences were detected between FMD-Ref and PB (p > 0.05). HSI ranked as PB > PS > FMD-Ref. VSI ranked as PB ≈ PS > FMD-Ref (Tukey’s test, p < 0.05), with no difference between PB and PS (p > 0.05).

2.2. Serum Biochemical Indices and Histological Observation of Liver

Serum biochemical indices are summarized in Table 3. Compared with the FMD group, PB and PS showed markedly lower serum CHO and higher TG, with PS exhibiting the highest TG level. Serum TP was comparable between FMD-Ref and PB but was reduced in PS. For enzyme activities, GPT was elevated in PB and PS relative to FMD-Ref, although substantial inter-individual variability was observed in the PB subgroup. GOT and ALP exhibited group-wise differences with considerable dispersion.

Representative H&E-stained liver sections showed distinct morphological features among groups (Figure 1). In the FMD-Ref group, hepatocytes displayed a normal morphology with evenly distributed lipid droplets. The PB subgroup shows more prominent vacuolation, with larger and more numerous fat droplets in hepatocytes. The PS subgroup exhibits a significant reduction in hepatocyte size, with cells more loosely arranged and larger intercellular spaces. The fat droplets are smaller, widely dispersed, and show a certain degree of vacuolation, accompanied by strong eosinophilic staining.

2.3. RNA-Seq Differential Expression Analysis

RNA-seq generated an average of 46.65 million raw reads and 45.58 million clean reads per sample (Table S1). The FASTQ files have been deposited in the NCBI Sequence Read Archive (SRA) under the accession number PRJNA1399997. The principal component analysis (PCA) of six samples from the two groups showed that PC1 accounted for 35.13% of the total variance, with most PB samples distributed on the negative side of PC1 and most PS samples on the positive side, indicating a clear separation between the two groups (Figure 2A). Differential expression analysis further identified a total of 390 DEGs, including 160 upregulated and 230 downregulated genes in the PB group relative to the PS group (Figure 2B).

2.4. GO and KEGG Enrichment Analysis of DEGs

Functional enrichment analyses were performed to explore the biological significance of the DEGs identified between the PB and PS groups. KEGG pathway enrichment analysis revealed that DEGs were significantly enriched in multiple pathways. Among these, the tight junction pathway exhibited the highest gene ratio and the largest number of enriched genes, representing one of the most significantly affected pathways. In addition, DEGs were significantly enriched in pathways related to circadian rhythm, sphingolipid metabolism, complement and coagulation cascades, and adipocytokine signaling. Several pathways associated with energy metabolism, including pyruvate metabolism and the citrate cycle (tricarboxylic acid cycle), were also significantly enriched. Furthermore, enrichment was observed in mineral absorption and proximal tubule bicarbonate reclamation pathways (Figure 3A). Gene Ontology (GO) enrichment analysis showed that, within the Biological Process category, DEGs were mainly associated with immune cell migration and carbohydrate and lipid metabolic processes. In the Cellular Component category, DEGs were predominantly enriched in plasma membrane-related structures, particularly the apical plasma membrane. In the Molecular Function category, DEGs were primarily enriched in receptor ligand activity, chemokine-related functions, and hormone activity (Figure 3B).

2.5. Metabolomic Profiling and Differential Analysis

PCA was performed on the differential metabolites. In the negative ion mode, PLS-DA analysis showed a clear separation between the PB and PS groups, with PC1 and PC2 explaining 27.56% and 18.80% of the variance, respectively (Figure 4A). Differential analysis identified 117 significantly altered metabolites, including 25 upregulated and 92 downregulated metabolites, as illustrated in the volcano plot (Figure 4B). In the positive ion mode, PLS-DA analysis also demonstrated a clear separation between the PB and PS groups, with PC1 and PC2 explaining 19.46% and 17.62% of the variance, respectively (Figure 4C). Differential analysis identified 158 significantly altered metabolites, including 61 upregulated and 97 downregulated metabolites, as shown in the volcano plot (Figure 4D).

2.6. KEGG Enrichment Analysis of Differential Metabolites

KEGG functional classification analysis showed that, in the positive ion mode, the differential metabolites between the PB and PS groups were mainly assigned to metabolism-related categories, with major contributions from carbohydrate metabolism, lipid metabolism, and amino acid metabolism (Figure 5A). Consistently, in the negative ion mode, differential metabolites were also predominantly classified into metabolism (including carbohydrate, lipid, and amino acid metabolism), with additional distribution in categories related to energy metabolism, membrane transport, and signal transduction (Figure 5C). KEGG pathway enrichment analysis further identified multiple significantly enriched pathways. In the positive ion mode, these included fatty acid biosynthesis, cholesterol metabolism, primary bile acid biosynthesis, steroid biosynthesis, the pentose phosphate pathway, and carbohydrate digestion and absorption (Figure 5B). In the negative ion mode, significantly enriched pathways included bile secretion, starch and sucrose metabolism, fructose and mannose metabolism, glycine, serine and threonine metabolism, phenylalanine metabolism, and ATP-binding cassette (ABC) transporters (Figure 5D).

2.7. Combined Analysis of Transcriptomic and Metabolomic Data

Based on the integrative analysis of DEGs and DAMs, multiple metabolic pathways were identified as being concurrently affected at both the transcriptomic and metabolomic levels (Table S2). Among them, the arachidonic acid metabolism pathway showed high enrichment significance in the joint pathway analysis (p = 0.0039), with the number of mapped differential molecules (Hits = 7) markedly exceeding the theoretical expectation (Expected = 2.01), together with a relatively high pathway impact value (Impact = 0.27), indicating the substantial perturbation of this pathway at both the structural and functional levels (Figure 6). In addition, the ABC transporter pathway and the folate transport and metabolism pathway also exhibited high enrichment significance (p < 0.01), suggesting that the dietary treatment influenced transmembrane transport processes and one-carbon metabolism-related functions. However, the pathway impact values for these pathways were low, implying that the observed changes may be concentrated in non-central nodes rather than key hubs of the pathways.

Furthermore, cysteine and methionine metabolism, biotin metabolism, folate-mediated one-carbon pool, histidine metabolism, and arginine biosynthesis were also significantly enriched in the joint analysis (p < 0.05), indicating coordinated alterations in amino acid metabolism and cofactor-related metabolic networks under different dietary treatments.

3. Discussion

3.1. Growth Performance and Physiological Responses to a Plant Protein Diet

At the end of the 8-week feeding trial, fish fed the SPCD exhibited a significantly lower final body weight, reduced feed intake, and poorer feed conversion efficiency compared with those fed the FMD, consistent with previous reports on high-level plant protein substitution [29,30]. Notably, pronounced inter-individual variation in growth performance was observed within the SPCD group. While the PB subgroup achieved a mean body weight comparable to that of FMD-fed fish, indicating the effective utilization of the plant-protein-based diet in a subset of individuals, the PS subgroup showed minimal growth throughout the experimental period, suggesting the maintenance of basal survival rather than somatic growth. Together, these results demonstrate substantial inter-individual divergence in terms of the capacity of T. ovatus to adapt to a fishmeal-free diet. Moreover, the markedly lower feed intake observed in the SPCD group indicates reduced dietary acceptance at the population level, a phenomenon also reported in other marine fish species fed plant-protein-based diets [31].

The alterations in serum biochemical profiles observed in this study are consistent with physiological responses commonly reported under high dietary plant protein inclusion. Previous studies have shown that the partial or total replacement of fishmeal with plant protein sources is often associated with reduced serum total protein and cholesterol levels, together with elevated liver-related enzymes such as GPT (ALT), reflecting altered nutrient utilization and hepatic status [3,32]. Consistently, Torstensen et al. [33] reported significantly decreased plasma cholesterol levels in Atlantic salmon (Salmo salar) fed plant-based diets lacking marine lipid components, while Lim et al. [7] observed changes in serum protein and transaminase activities in marine fish fed soybean meal-based diets. Collectively, these findings indicate that changes in serum biochemical parameters represent a common physiological response to plant protein substitution across fish species.

3.2. Diet-Induced Modulation of Immune Responses and Metabolism in the Liver

GO and KEGG enrichment analyses revealed that a substantial number of DEGs were involved in immune-related pathways and functions, including complement and coagulation cascades, leukocyte migration, monocyte chemotaxis, and other immune-related biological processes. Notably, in the PS group, several immune-associated genes were significantly upregulated in the liver, particularly the IFNG1 gene and FGB gene. Interferon gamma (IFN-γ), encoded by IFNG1, is a Th1-derived cytokine and a pivotal regulator of cell-mediated immune responses in vertebrates. In teleost fish, IFNG1 has been identified as the major IFN-γ isoform associated with antigen-specific cell-mediated immunity (CMI), as demonstrated in ginbuna crucian carp (Carassius auratus langsdorfii) [34]. IFN-γ is known to promote Th1-type immune responses and activate macrophages, thereby playing a central role in immune modulation [35]. The FGB gene encodes the β chain of fibrinogen, a highly pleiotropic protein involved not only in coagulation cascades but also in wound healing, inflammation, and angiogenesis [36]. Similar upregulation of FGB has been reported in D. labrax fed a totally plant-based diet [13], suggesting that dietary protein sources are closely associated with alterations in hepatic immune-related gene expression. Although no overt inflammatory lesions were observed in liver histology, the liver is increasingly recognized as a central immune-regulatory organ in teleosts [37]. Therefore, the upregulation of genes involved in antibacterial defense and immune modulation observed in the PS group may represent an adaptive immune regulatory response rather than overt pathological inflammation. We speculate that high inclusion levels of plant-derived proteins may impose intestinal stress or compromise intestinal barrier integrity, leading to enhanced immune signaling along the gut–liver axis and triggering compensatory immune regulation in the liver.

3.3. Impact of Plant-Based Diet on Liver Lipid and Cholesterol Metabolism

Under SPCD feeding, the HSI% levels in T. ovatus were higher than those in the FMD-Ref group. Similarly, excess lipid droplets were observed in both PB and PS hepatocytes in liver sections. In D. labrax fed a plant-based diet, genes involved in fatty acid and cholesterol synthesis exhibited higher transcript levels, which were associated with the overexpression of genes related to lipid transport [13]. In the PB group, ABCA1b gene was significantly upregulation. This gene is known to play a role in high-density lipoprotein (HDL) synthesis and is closely associated with the efflux of cholesterol. Furthermore, the CYP1B1 gene was also upregulated. CYP1B1, a member of the cytochrome P450 family, is involved in the metabolism of fatty acids and other lipids. According to Bu et al. [38], the upregulation of CYP1B1 impacts a process known as lipophagy, which is critical for the breakdown and utilization of lipid droplets. We speculate that the PB group may alleviate hepatic lipid accumulation induced by plant protein feeding through this mechanism, thereby preventing lipid degeneration. Additionally, in the metabolomics analysis, the level of taurolithocholic acid was found to be significantly upregulated in the PB group. As a bile acid, it can regulate lipid metabolism and, in certain cases, reduce inflammation [39].

3.4. Tight Junction Disruption and Hepatic Vacuolation in Fish Fed a Plant-Protein-Based Diet

Tight junctions are specialized intercellular junctional complexes that play a critical role in maintaining epithelial cell polarity and regulating paracellular permeability [40]. In the liver, they are localized at the bile canalicular domain between adjacent hepatocytes, where they separate bile from the sinusoidal blood compartment and preserve hepatocellular polarity [41]. Hepatic vacuolation is a common histopathological feature in fish exposed to nutritional imbalance or dietary stress and is frequently associated with disturbances in lipid accumulation, glycogen storage, cellular hydration, or organelle integrity. Accordingly, hepatic vacuolation is widely used as an indicator of impaired hepatic metabolic status or dietary maladaptation in aquaculture studies. In the present study, a progressive aggravation of hepatic vacuolation was observed across the FMD-Ref, PB, and PS groups. Similar hepatic histopathological alterations induced by high levels of plant protein substitution have been reported in multiple fish species, including S. salar fed soybean meal-based diets [42], as well as turbot (Scophthalmus maximus L.) [43] and P. olivaceus fed diets containing high proportions of plant-derived proteins [44]. Consistent with these observations, transcriptomic analysis revealed that differentially expressed genes were significantly enriched in the tight junction pathway, suggesting that the replacement of fishmeal with soy protein concentrate is associated with the altered expression of genes involved in cell–cell junction integrity. The disruption of tight junction-associated components in hepatocytes has been suggested to impair bile canalicular integrity and hepatocellular polarity, thereby affecting intracellular trafficking, membrane stability, and osmotic balance. Such alterations are considered important contributors to hepatocellular vacuolation under nutritional or toxic stress. Therefore, enrichment of the tight junction pathway in this study provides a plausible molecular basis for the hepatic vacuolar changes observed in T. ovatus fed a plant-protein-based diet.

3.5. Integrated Remodeling of Arachidonic Acid Metabolism Under Soy Protein Replacement

An integrated transcriptomic–metabolomic analysis revealed a coordinated remodeling of the arachidonic acid (ARA) metabolism pathway under soy protein replacement, involving both upstream membrane lipid remodeling and the downstream divergence of eicosanoid signaling branches.

Metabolomics revealed a significant increase in phosphatidylcholine (PC), whereas the transcriptome showed a marked upregulation of the rarres3l (RARRES3-like) gene. RARRES3 (also known as a member of the PLA/AT (PLAAT/HRASLS) family of membrane-associated enzymes) possesses phospholipase/acyltransferase-related activities and has been implicated in membrane phospholipid remodeling and the mobilization/availability of polyunsaturated fatty acids, including ARA. Therefore, the upregulation of rarres3l can reasonably be inferred to alter PC-associated membrane lipid turnover, thereby influencing substrate supply and flux allocation into the ARA metabolic network [45]. In addition, changes in lipid and choline supply associated with soy protein replacement may reshape the physiological basis of membrane phospholipid metabolism (e.g., PC turnover/remodeling) [46].

Regarding downstream ARA pathways, we detected elevated levels of the LOX-branch product 8(S)-HETE and related oxidized lipids (e.g., Trioxilin B3), together with a significant increase in the COX-branch prostaglandin E2 (PGE2), suggesting a flux shift in AA metabolism toward pro-inflammatory branches. This framework is consistent with the canonical view that ARA can be converted by the COX/LOX/CYP enzyme systems into diverse bioactive lipid mediators [47]. Meanwhile, the key PGE2-inactivating enzyme 15-hydroxyprostaglandin dehydrogenase (15-PGDH) gene was downregulated at the transcript level. As 15-PGDH is the core enzyme of the NAD^+^-dependent prostaglandin inactivation pathway, its downregulation is more consistent with a compensatory negative feedback response to enhanced PGE2 production—i.e., when prostaglandin synthesis flux increases, the inactivation system is recruited but may not be sufficient to fully offset the increased production—rather than a simple “enzyme up–substrate down” relationship [48]. In contrast, 11,12-EET, a CYP epoxygenase-derived product often associated with anti-inflammatory or cytoprotective properties, was significantly decreased, further supporting a shift in ARA metabolism away from the CYP–EET branch toward the pro-inflammatory LOX/COX branches [49].

Interestingly, Medagoda and Lee [50] found that higher levels of ARA can inhibit inflammation in the intestines of fish fed a high plant-protein-mixed diet. Moreover, when ARA levels are low, hepatocytes exhibit atrophy, nuclear migration, and vacuolation.

We propose the following model: soy protein replacement may affect membrane phospholipid metabolism (reflected by increased PC) and, in parallel, alter membrane lipid remodeling regulation via rarres3l upregulation, thereby reshaping ARA availability and metabolic flux allocation. This, in turn, promotes pro-inflammatory eicosanoids (e.g., PGE2 and 8(S)-HETE) while weakening the anti-inflammatory EET branch, ultimately contributing to differences in inflammatory tone and energy allocation strategies and facilitating inter-individual divergence in growth phenotypes in T. ovatus.

Despite the integrative insights obtained here, this study has limitations. Although pronounced growth divergence under the fishmeal-free, plant-protein-based diet was evident, our multi-omics analyses were restricted to liver tissue. Therefore, upstream intestinal processes—nutrient digestion and absorption, barrier function (tight junction integrity), and local immune responses—were not directly evaluated, despite their likely roles in shaping hepatic metabolic and immune states via the gut–liver axis.

Future work will extend these analyses to the intestine using the same experimental cohort and preserved samples, integrating intestinal histology, tight junction markers, inflammatory profiling, and gut microbiota characterization with hepatic multi-omics. This gut–liver framework is expected to provide a more complete mechanistic explanation for differential plant protein tolerance and to facilitate the identification of biomarkers and regulatory targets for diet optimization and selective breeding in T. ovatus.

4. Materials and Methods

4.1. Feed Preparation

Two experimental diets were formulated for the feeding trial (Table 4). The FM-based diet (FMD) was prepared as a control diet, and a soy protein concentrate diet (SPCD) primarily consisted of soy protein concentrate supplemented with a small proportion of soybean meal. Both diets were isonitrogenous and isolipidic. All the dry ingredients were ground, passed through a 320 μm sieve, and thoroughly mixed for 30 min according to the formulation. The mixture was then processed into spherical pellets of two specifications (diameters of 2.5 mm and 3 mm, respectively) using a twin screw extruder to accommodate the different growth stages of the fish. Finally, the feed was dried at 45 °C until the moisture content decreased to approximately 10%, and was stored at −20 °C for subsequent use.

4.2. Experimental Fish and Husbandry

The experimental Trachinotus ovatus (golden pompano) were offspring derived from broodstock maintained at the Sanya Tropical Fisheries Research Institute (Lingshui, Hainan, China), South China Sea Fisheries Research Institute, Chinese Academy of Fishery. To simulate the large-scale farming conditions of T. ovatus, the experiment was conducted in the offshore net cages (L 2 m × W 1 m × H 1.5 m) in Xincun Town, Lingshui County, Hainan Province, China. Fish were acclimated in the six net cages and fed FMD for 7 days, Subsequently, three cages were randomly assigned to the SPCD group, the feed was switched from FMD to SPCD, and the feeding trial was started (two diets; 600 fish/diet group; each diet to three cages, with 200 fish per cage; and mean body weight 58.8 ± 0.5 g). The entire experimental trial lasted 8 weeks. Fish were fed to apparent satiation twice daily (08:00 and 17:00), with the daily feeding rate approximately ranging from 2.5 to 3.5% of body weight depending on biomass and feeding response. Seawater was continuously supplied from the offshore culture area of Xincun Bay (Lingshui, Hainan, China). Water temperature, salinity, and pH were monitored daily using a multi-parameter water quality meter (YSI, OH, USA). During the experimental period, the temperature ranged from 28 to 31 °C, the salinity from 28 to 30, and the pH from 7.8 to 8.2, and no abnormal fluctuations were observed. Fish were fasted for 24 h prior to sampling, during which feeding was limited to a single morning meal on the previous day.

4.3. Sample Collection

At the end of the trial, all fish were anesthetized with 0.01% eugenol and weighed. In the SPCD group, individuals were ranked by body weight, and the top 20 and bottom 20 fish were defined as the high-body-weight (PB) and low-body-weight (PS) subgroups, respectively. For individual-level comparisons, a reference subset of 20 fish was selected from the FMD treatment with body weights close to the cage-averaged mean (hereafter referred to as the FMD-Ref group). Subsequently, six fish per group (PB, PS, and FMD-Ref) were randomly chosen from the corresponding 20-fish subsets for serum biochemical assays and liver histology. The liver tissue from PB and PS was further preserved for transcriptomic and metabolomic analyses. The blood was collected from the caudal vein and allowed to stand at 4 °C for serum separation. After centrifugation (3000 rpm, 10 min, and 4 °C), the supernatant was collected as serum and stored at −80 °C until the analysis of serum biochemical indices. Meanwhile, the viscera and liver were excised and weighed to calculate the viscerosomatic index (VSI) and hepatosomatic index (HSI), respectively. The liver was then divided into two portions: one portion was fixed in Davidson’s AFA solution (33% ethanol, 22% formalin, 11.5% acetic acid, and 33.5% H_2_O) for 24 h, transferred to 70% ethanol, and stored until histological processing. Paraffin-embedded sections were prepared and stained with hematoxylin and eosin (H&E) for histopathological observation. The other portion of PB and PS was placed into RNase-free tubes, and stored at −80 °C for transcriptome and metabolomics analyses.

4.4. Growth Performance Evaluation

The growth performance was calculated by the following formulae, following the approach of Zhao et al. [51]:

$$[eqn]$$ $$[eqn]$$ $$[eqn]$$ $$[eqn]$$ $$[eqn]$$

Ni and Nf are the total number of fish at the beginning and end of the feeding trail, respectively, Wi and Wf are the initial and final body weights (g) at the beginning and end of the feeding trial, and Wd is the total body weight of the dead fish during the trial.

4.5. Biochemical Analysis and Histological Observation

Serum biochemical parameters, including γ-glutamyl transferase (GGT), glutamate–pyruvate transaminase (GPT), glutamate–oxaloacetate transaminase (GOT), alkaline phosphatase (ALP), total protein (TP), triglyceride (TG), and total cholesterol (CHO) were determined using an automatic biochemical analyzer (Hitachi 7020, Hitachi Science Systems, Hitachinaka, Japan) with commercial diagnostic kits following the manufacturer’s protocols. All assays were performed in duplicate to ensure analytical accuracy.

For histological analysis, fixed liver tissues were dehydrated through graded ethanol, followed by paraffin infiltration and embedding. Paraffin blocks were cut into 5 μm sections using a rotary microtome (RM2235, Leica Microsystems GmbH, Wetzlar, Germany). Sections were stained with hematoxylin–eosin (H&E) and examined under a light microscope (BZ-53, Olympus, Tokyo, Japan). Hepatocyte cytoplasmic and nuclear diameters were quantified from the images, with 50 measurements collected per fish.

4.6. Transcriptome Analysis

Total RNA was isolated from the livers of T. ovatus in the PB and PS groups using the RNeasy Plus Mini Kit (Qiagen) according to the supplier’s protocol. The RNA yield and purity were assessed by NanoDrop 1000, and RNA integrity was evaluated with an Agilent 2100 Bioanalyzer. The mRNA was enriched, fragmented, and reverse transcribed to cDNA for library construction. The fragmented mRNA was reverse transcribed using random hexamer primers and converted to double-stranded cDNA, which was then purified (QIAquick PCR extraction kit; QIAGEN Inc., Valencia, CA, USA). Size-selected cDNA fragments were recovered by agarose gel electrophoresis and amplified by PCR to generate sequencing libraries. The resulting cDNA underwent end repair, poly(A) tailing, and adaptor ligation, after which fragment size selection/cleanup was performed with DNA Clean Beads. Libraries were further enriched by PCR and purified again using DNA Clean Beads. After construction, library concentration was first estimated with Qubit 3.0, and insert size was verified with the Agilent 2100 Bioanalyzer; effective library concentration was then quantified by ABI StepOnePlus Real-Time PCR. Clean reads were summarized using standard quality indicators (Q20, Q30, and GC content). Differentially expressed genes (DEGs) were identified with DESeq2 (v1.16.1) in R, and functional enrichment (GO and KEGG) was carried out using the clusterProfiler package (R v4.3.2). Analyses were mapped to the T. ovatus reference genome (ENA accession: PRJEB22654).

4.7. Metabolome Analysis

The liver samples from the PB and PS fish were analyzed individually as biological replicates, with each sample subjected to three LC–MS/MS technical replicates. Metabolites were extracted following Xie et al. [24] with minor adjustments and profiled by liquid chromatography coupled to tandem mass spectrometry (LC–MS/MS). After quality filtering, features were annotated using reference databases. Autoscaled data were explored using orthogonal partial least squares discriminant analysis (PLS-DA) in SIMCA-P (v16.0). Differential metabolites were defined by VIP > 1.0 together with p < 0.05. For the two-group comparisons (PB vs. PS), univariate testing was conducted using the nonparametric Mann–Whitney U test. Differential metabolites were subsequently subjected to pathway analysis, and candidate metabolite biomarkers were screened based on these results.

4.8. Integrated Omics

To integrate transcriptomic and metabolomic data and further elucidate the comprehensive regulatory effects of feed treatment on key metabolic pathways, the screened DAMs and DEGs were jointly imported into the MetaboAnalyst 6.0 online software (https://www.metaboanalyst.ca, accessed on 13 February 2025) for joint pathway analysis.

The joint pathway analysis constructs a metabolic pathway network based on the KEGG database, where metabolites and genes are mapped to corresponding nodes within metabolic pathways. Given KEGG annotation availability, zebrafish (Danio rerio) was used as the reference for pathway mapping. During the analysis, all significantly differential pathways were included for enrichment analysis, while other parameters were set to MetaboAnalyst’s default configurations. Through the joint pathway analysis, the coordinated changes in differential metabolites and differentially expressed genes within the same metabolic pathways were systematically evaluated to identify key pathways regulated simultaneously at both transcriptional and metabolic levels.

4.9. Statistical Analysis

Prior to statistical analysis, data normality and homogeneity of variance were assessed using the Shapiro–Wilk test and Levene’s test, respectively. The results are presented as mean ± standard deviation (SD). All statistical procedures were performed in SPSS (version 22.0). Differences between two groups were evaluated using independent-samples t tests, whereas comparisons among multiple groups were assessed by one-way ANOVA followed by Duncan’s multiple range post hoc test. Statistical significance was defined at p < 0.05.

5. Conclusions

In conclusion, our integrative transcriptomic–metabolomic analyses indicate that the divergent growth phenotypes of T. ovatus under a fishmeal-free, soy protein-based diet are closely associated with the coordinated hepatic remodeling of tight junction integrity, immune-related signaling, and lipid metabolic regulation. Specifically, soy protein replacement was accompanied by the significant enrichment of the tight junction pathway and progressive hepatocellular vacuolation, together with a shift in membrane phospholipid turnover (increased phosphatidylcholine and the upregulation of rarres3l) and a reprogramming of arachidonic acid metabolism toward COX/LOX-derived eicosanoids (e.g., elevated PGE2 and 8(S)-HETE) and away from the CYP–EET branch (decreased 11,12-EET), with the upregulation of 15-PGDH suggesting the compensatory control of prostaglandin signaling. Collectively, these findings support a mechanism whereby soy protein-driven membrane lipid remodeling and eicosanoid flux redistribution reshape hepatic homeostasis and inflammatory tone, thereby contributing to inter-individual variation in plant protein tolerance, and providing molecular targets for diet optimization and selective breeding to improve plant protein utilization in T. ovatus.