The NADP+/NADPH assay kit with WST-8, BCA protein assay kit, enhanced ATP assay kit were purchased from Beyotime Institute of Biotechnology (Beyotime®, catalog: S0116, S0179 and S0027, Shanghai, China). The 1 $\times$ PBS, acetyl-CoA carboxylase (ACC), fatty acid synthase (FAS), glutamine synthetase (GS), glutamate synthase (GOGAT), reduced glutathione (GSH), lipase (LPS), isocitrate dehydrogenase (ICDHm), succinate dehydrogenase (SDH), acetyl-CoA, NADP-malate dehydrogenase (NAD-MDH), $\alpha$-ketoglutarate dehydrogenase ($\alpha$-KGDH), pyruvate dehydrogenase (PDH) and ATP citrate lyase (ACL) assay kit were obtained from a reagent manufacturer of Beijing Solarbio Science & Technology Company Ltd. (Solarbio®, Beijing, China), and their catalogs were P1020, BC6025, BC0555, BC0915, BC0075, BC1175, BC2345, BC2165, BC0955, BC0985, BC1055, BC0715, BC0385 and BC4245 respectively. Experimental reagents were purchased from Shanghai Aladdin Bio-Chem Technology Company Ltd. (Aladdin®, Shanghai, China). Purified water was obtained from the Hangzhou Wahaha Group Ltd. (Wahaha®, Hangzhou, Zhejiang, China).

Referring to the previous study [17] and nutritional standard (SCT1025-1998, China) for Nile tilapia formula feed, and all ingredients included ZLM were obtained from the Guangzhou feed market. After pulverizing all the ingredients, water was added until the water content reached 20% and thoroughly mixed. The mixtures were then pressed into particles using an automatic extruder (DGP60-C, Xingtai, China) from Yugong Technology Development Company Ltd., serving as a control for the basal diet without ZLM inclusion. Subsequently, these particles were dried at 65 °C until their water content decreased to less than 10%. The premix of minerals and vitamins used in this study were procured from two feed manufacturers: Guangzhou Southern Biotechnology Company Ltd. (Yulong®, Guangzhou, China) and Chelota Biotechnology Group Company Ltd. (Chelota®, Deyang, China). These premixes complied with executive standards of Guangdong feeding (2020) 01131 (Q/NFSW 19-2021) and Sichuan feed premix (2015) 05005 respectively. Following the aforementioned formula and standard guidelines, soybean meal at levels of 15%, 30%, and 60% was replaced by ZLM in experimental groups. Detailed descriptions mentioned previously [17], a dry extrusion method was carried out and then these feeds were granulated at 85 °C with screw rotary extrusion method with non-adjustable pressure settings. Then, these diet components and nutritional ingredients were shown in Supplementary Tables 1, 2.

According to a previous study [17], 60.00 g Nile tilapia fish with an error range of $\pm$5.35 g were purchased from a tilapia farm in Guangzhou and placed in a 120 L open-circuit water tank equipped with oxygenation. Through microbiological detection, these tilapias should not be infected by specific pathogen (Laboratory animal—Microbiological and parasitical standards and monitoring (GB 14922-2022)) (https://www.cnki.net/). The water temperature was maintained at 25 °C, while the DOC (dissolved oxygen content) was controlled at a level of 8.75 mg/L. Additionally, the pH value of the water remained consistently at 7.0. In order to keep water clean, the system of filtration and circulation was always normally opened. The fish were fed three times a day with a meal amounting to 3% of their body weight, as previously described [17]. After the feeding process, any remaining residue should be eliminated within an hour. In cases where there was a significant amount of feces and minimal or no residual feed, the diet should be repeated once more. Before collection of samples, these fish were kept at the way for 30 days. A total of 144 tilapias were randomly divided into four groups, and each group was evenly divided into three tanks for rearing. After a 30-day feeding period, two fish were randomly selected from each tank, resulting in a total of six tilapia samples per treatment group for subsequent metabolomic analysis. Similarly, six fish per group from these three tanks were prepared and detected by enzymatic testing. Finally, the rest fish were released.

Referring to previous studies [14, 15, 16] and the national guideline (RB/T 061-2021), tilapias were euthanized by immersing them in an ice bath (mixture of ice slush and water, 5/1 (w/w), maintained at 4 °C) for a duration of 10 min, followed by an additional 20-min immersion after cessation of all movement to ensure death due to hypoxia. Subsequently, the fish were thoroughly wiped with sterilized gauze after rinsing with distilled water. Six livers were excised for preparation, from which approximately 50 mg samples were cut and immediately immersed in 1 mL methanol at –20 °C. The samples were then sonicated using an ultrasonic processor (JY92-IIDN, SCIENTZ, Ningbo, China) at a power of 10 W for 5 min and subsequently centrifuged at 12,000 rpm for 10 min at a temperature of 4 °C. Next, each sample’s supernatant was supplemented with ribitol (A103030, Aladdin®, concentration: 100 µg/mL; used as a quantitative standard internal reference). The supernatants were concentrated into dry extracts using a rotary vacuum centrifuge device (Eppendorf Concentrator plus, Eppendorf Innovation Company, Hamburg, Germany). For gas chromatography-mass spectrometry (GC-MS) detection employing oximated derivatization approach, the dried extracts were dissolved in 100 µL methoxyamine pyridine solution (M300386 and P111513, Aladdin®, concentration: 20 mg/mL) and incubated in an incubator shaker set at 37 °C and 200 rpm for 120 min. Subsequently, the mixture was silylated with 100 µL N,O-Bis(trimethylsilyl)trifluoroacetamide containing 1% trimethylchlorosilane (B118473, Aladdin®), followed by incubation under the same conditions mentioned earlier but for 30 min. Six biological replicates were performed for each experimental group.

According to previous studies [14, 15, 16], a splitless injection of 1 µL derivatized extract was performed using an HP-5MS column (30 m $\times$ 250 µm $\times$ i.d. 0.25 µm, Agilent®, CA, USA) and detected by an Agilent gas chromatography mass spectrometer (7890A GC, 5975C VL MSD, Agilent®). The inlet temperature was maintained at 280 °C throughout the analysis. In the gas chromatography oven, the initial temperature of 85 °C was held for 5 min before being ramped up to 280 °C at a rate of 15 °C/min and held for another 5 min, followed by a further increase to 310 °C at a rate of 20 °C/min. A solvent delay time of 5 min was set after the initial injection prior to data acquisition. High purity helium (99.99%) was used as the carrier gas with a constant flow rate of 1 mL/min. Full scan mode was employed to capture MS ions in the range of m/z from 50–600.

The AMDIS software, version Agilent OpenLAB CDS ChemiStation C.01.01 (Agilent®), was employed for the calibration and deconvolution of the mass spectra (extracted ion chromatogram, XIC). GC-MS peaks with a signal-to-noise ratio (S/N) less than 30 were excluded to avoid false positives [14, 15, 16, 23]. The presence of these questionable artifact peaks was effectively eliminated when compared to the solvent blank control. Metabolite mass spectra were identified by retrieving from the GMD 2011 library (Golm Metabolome Database, Germany) combined with NIST 2011 (National Institute of Standards and Technology, USA), using the following criteria: reverse match value $\ge$800, probability $\ge$60%, and match value $\ge$750 [14, 15, 16, 24]. Ribitol was used as an internal reference standard for calculating relative abundance of all metabolites. Singular value decomposition (SVD) method was utilized to impute non-zero values for missing values and zeroes [14, 15, 16, 25]. Variables in data matrix below a threshold of 50% were removed and replaced by median within their respective groups for handling missing values. Additionally, interquartile range (IQR) filtering approaches improved these datasets [14, 15, 16, 26]. Quantile normalization row-wise procedures along with pareto scaling (mean-centered and divided by the square root of SD of each variable) were employed for combined normalization processing [14, 15, 16, 27]. Data filtration option “none” was applied to variables (metabolites) with less than 5000 counts. Consequently, these metabolome data arrays could be utilized for subsequent analytical calculations.

Biochemical assays were conducted on six liver samples, followed by determination of enzyme activity using the protocols provided in the respective kit manuals. Briefly, 1 g of liver sample was fully rinsed with 1 $\times$ PBS (10 mM, pH 7.2–7.4, Solarbio®), resuspended in 10 mL lysis buffer from these kits, and homogenized by sonication at 340 W and 40 KHz for 5 min in an ice bath. After centrifugation at 12,000 rpm for 10 min, the resulting supernatant containing total crude enzymes was transferred into a new EP tube, and protein content was estimated using a BCA kit (Beyotime®). Subsequently, an aliquot of reaction buffers and crude enzyme proteins (200 µg total crude enzyme) were mixed to obtain a final volume of 200 µL in a microplate well. The microplates were then incubated for 15 min at 25 °C before measuring absorbance values using a microplate reader (Synergy HTX, BioTek Instruments Company Ltd., Winooski, VT, USA). Enzyme activities were evaluated according to the instructions provided with the kits by plotting against their control or standard curve. All determinations were performed in duplicate and values were presented as mean $\pm$ SEM of measurements. Significance tests were conducted using one-way ANOVA and multivariate statistical analysis.

Using WPS Office for Windows (Beijing Kingsoft Office Software, Beijing, China), data manipulations and transformations were performed. The differentiation of metabolites among these groups was assessed using analysis of variance (ANOVA) with a significance level set at $\alpha$ = 0.01 in SPSS 23.0 (IBM®, Armonk, NY, USA). Prior to analysis, the metabolomic datasets were normalized by subtracting their median and scaling them by the quartile range in these samples. Subsequently, a multivariate statistical analysis was conducted using an online website MetaboAnalyst 6.0 [28]. The correlation among the experimental groups was investigated through partial least squares-discriminant analysis (PLS-DA) and principal component analysis (PCA). A cut-off value greater than one was determined based on the variable importance in the projection (VIP) value obtained from PLS-DA analysis. Metabolomic biomarkers were then screened and presented in a scatter plot.

Based on previous researches [14, 15, 16], metabolomic pathway enrichment analyses included hypergeometric test (over representation analysis, ORA) and pathway network topology assay (relative-betweeness centrality). Using an online tool MetaboAnalyst 6.0 (https://www.metaboanalyst.ca/) with a Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway library (https://www.genome.jp/kegg/, version Oct 2019), Danio rerio (KEGG No.7955) was selected as the model organism since it’s currently the only zebrafish species available as a fish model organism. Default parameters were used for other analyses before executing enrichment analyses. All significant metabolites were used to define a reference metabolome which underwent hypergeometric testing to calculate -log (p) values and impact values reflecting each metabolic pathway’s influence. Only pathways with p 0.05 were considered significant and retained for further investigation. The resulting histograms were analyzed and plotted using GraphPad Prism software version 7.0 (GraphPad Software, Inc., San Diego, CA, USA).

The values were recorded as means $\pm$ SEM (n = 12), and statistical significance was determined using the nonparametric Kruskal-Wallis one-way analysis with Dunn’s multiple comparison post hoc test for data processing, where p 0.05 (*) and p 0.01 (**) indicated statistical significance.

Six dorsal muscle samples were subjected to sensory evaluation tests. The procedures and objectives of the sensory evaluation tests were explained to all recruited volunteers, who then provided informed consent. Following the approach described by Pu et al. [29], a ranking test was conducted with 20 trained volunteers (equally divided between males and females, aged 18–21 years, non-smokers and non-drinkers, without taste/odor disorders). The volunteers underwent a three-week training period. Taste solutions containing sweetness, sourness, saltiness, umami, bitterness, and fishiness standards were prepared and diluted in doubling dilutions up to 1/16 concentration. Specifically for the fishiness sensory test, participants were trained in sniffing techniques. The fishiness standard solution used was trimethylamine hydrochloride (TMA) at a concentration of 20 g/L (T818827; Shanghai Macklin Biochemical Technology Co., Ltd., Shanghai, China). According to the Chinese standard GB/T16291.1-2012 on sensory analysis methodology (https://www.cnki.net/) as well as previous research [17], these dorsal muscle samples were added into test bottles and randomly assigned to three groups for subsequent sensory analysis at one-hour intervals.

Furthermore, as depicted in Fig. 1, the meat sensory evaluation test was conducted on tilapia fed with a basal diet supplemented with ZLM at levels of 15%, 30%, and 60%. The control group, consisting of fish fed only the basal diet, exhibited predominantly sour and bitter taste attributes. Conversely, the experimental groups receiving supplementation of ZLM at levels of 15%, 30%, and 60% displayed enhanced sweetness and umami characteristics. Notably, as the level of ZLM supplementation increased, both umami perception and sweetness intensity were significantly augmented (p 0.05). These findings aligned consistently with alterations observed in flavor metabolites accumulation following incremental increases in ZLM supplementation from 15% to 60%. Consequently, it could be inferred that dietary inclusion of ZLM had potential to enhance the palatability of tilapia meat.

Fig. 1.

Sensory evaluation. The sensory perception of tilapia meat, when fed with 15%, 30%, and 60% Zophobas atratus larval meal (ZLM) feed in comparison to the basal diets, was evaluated.

To highlight ZLM flavor effects of sensory evaluation test, then these flavor metabolites in tilapia livers were detected by GC-MS, and their abundances were calculated after fed for 30 days. Compared with basal diets, the fold change of these flavor metabolites’ GC-MS abundance was shown that with increasing ZLM (Fig. 2). In detail, flavor nucleotides guanosine, adenosine-5^′-monophosphate, adenosine, inosine-5^′-monophosphate, uridine, uridine-5^′-monophosphate and xanthine increased, but inosine and hypoxanthine decreased (Fig. 2A). As shown in Fig. 2B, these meaty flavor metabolites of 2-methyl-3-furanthiol and nonanal were also measured with increased abundances following with an increasing ZLM supplement. Here, flavor amino acids of alanine, aspartic acid, glycine, proline, glutamic acid, threonine and serine increased, but these off-flavor amino acids of arginine, histidine, tyrosine and valine decreased (Fig. 2C). Flavor fatty acids and their derivatives of linoleic acid (9,12-Octadecadienoic acid), linolenic acid (9,12,15-Octadecatrienoic acid), oleic acid (9-Octadecenoic acid), 2-ketoglutaric acid, arachidonic acid, citric acid, fumaric acid, heptadecanoic acid, hexadecanoic acid, hexanoic acid, malic acid, octadecanoic acid, oxalacetic acid, pentadecanoic acid, pyruvic acid, succinic acid, tetradecanoic acid were accumulated as shown in Fig. 3. Here, there was no difference in 2-ketoglutaric acid and tetradecanoic acid. In addition, it was notable that, these fishiness metabolites 2-Methylisoborneol, geosmin and trimetlylamine were also determined with decreased abundances after ZLM supplement increased. Thereby, in the biosynthesis of tilapia livers, these positive metabolomic effects of ZLM were to accumulate these flavor metabolites and down-regulate off-flavor metabolites.

Fig. 2.

The flavor and off-flavor metabolites. Taking into account the variation range of the abundance of these metabolites and the coordination of the histogram display, we took their fold change of gas chromatography-mass spectrometry (GC-MS) abundance compared with basal diets groups. With an increase in ZLM supplement, flavor nucleotides guanosine, adenosine, adenosine-5^′-monophosphate, xanthine, inosine-5^′-monophosphate, uridine, uridine-5^′-monophosphate, inosine and hypoxanthine significantly increased (A). We also detected meaty flavor metabolites of 2-methyl-3-furanthiol and nonanal, and their abundance also increased with increased ZLM supplement. Moreover, fishiness metabolites 2-Methylisoborneol, geosmin and trimetlylamine were detected and down-regulated in 15%, 30% and 60% ZLM diet groups (B). And flavor amino acids alanine, aspartic acid, glutamic acid, glycine, proline, serine, threonine, arginine, histidine, tyrosine, valine (C) also increased. ** p 0.01.

Fig. 3.

The flavor fatty acids and their derivatives. Compared with control groups, with an increase in ZLM supplement, flavor fatty acids and their derivatives of arachidonic acid, linoleic acid (9,12-Octadecadienoic acid), linolenic acid (9,12,15-Octadecatrienoic acid), oleic acid (9-Octadecenoic acid), citric acid, fumaric acid, heptadecanoic acid, hexadecanoic acid, hexanoic acid, malic acid, octadecanoic acid, oxalacetic acid, pentadecanoic acid, pyruvic acid, succinic acid were significantly accumulated. Here, there was no difference in 2-ketoglutaric acid and tetradecanoic acid. ** p 0.01.

The ZLM metabolic effects on tilapia were investigated as outlined in Fig. 4. Liver samples were taken from these fish fed by control basal diet and 15%, 30%, and 60% ZLM supplement for 30 days, and prepared for metabolomic determination based on GC-MS approaches. Here, 48 datastes were obtained from 6 biological individuals with 2 technical replicates in these groups, which representative total ion chromatograms (TIC) were listed in Fig. 4A. And there were 301 aligned individual peaks accessed. The correlation coefficient of their 2 technical replicates indicated the reliability of the GC-MS platform. Then, their correlation coefficient between these technical replicates fluctuated at range of 0.9971 to 0.9999, demonstrating the reproducibility and reliability (Fig. 4B). The annotation of NCBI PubChem (https://pubchem.ncbi.nlm.nih.gov/) and KEGG (http://www.kegg.jp/) was utilized to identify six categories among these 124 metabolites, including amino acids, carbohydrates, aliphatic compounds, flavor compounds, nucleotides, and others. After removal of internal standard ribitol and any known artificial peaks, 124 metabolites were identified. Of these, 18.55%, 24.19%, 13.71%, 12.10%, 4.03% and 27.42% were categorized respectively (Fig. 4C). After the removal of the internal standard ribitol and any known artificial peaks, followed by integration of the same compounds, a total of 124 metabolites exhibiting reliable signals were detected in each sample. It was postulated that these experimental groups gave rise to distinct metabolomes, thereby hypothesizing a correlation between metabolomic variations and these constructed metabolomes. The metabolic biomarkers were uncovered using a permutation test coupled with one-way ANOVA to distinguish these groups, and the differential metabolites in the basal diet and 15%, 30%, and 60% ZLM supplementary groups were identified. Thus, 69 metabolites with significant differences were screened (p 0.01). Therefore, the metabolomic profile of tilapia fed by ZLM supplementary and control basal diets was established.

Fig. 4.

The metabolomic profile. (A) Representative total ion chromatography (TIC) from experimental groups. (B) Reproducibility of metabolomic profiling platform used in the discovery phase. Abundance of metabolites quantified in samples over two technical replicates is shown. Pearson correlation coefficient between technical replicates varies between 0.9971 and 0.9999. (C) Categories of the different metabolites. 124 metabolites with different abundance were searched against in Kyoto Encyclopedia of Genes and Genomes (KEGG) for their categories, and the pie chart was generated in WPS Office for Windows (Kingsoft®, Beijing, China).

The heatmap in Fig. 5 illustrated the distribution of the 69 significant metabolites (p 0.01), clearly separating the basal diets and the groups supplemented with 15%, 30%, and 60% ZLM. The distinct differences among these groups indicated variations at the metabolomic level, suggesting that amino acid metabolism, fatty acid metabolism, and energy metabolism play pivotal roles in mediating the response to ZLM supplementation. Furthermore, this response might be responsible for influencing tilapia’s production performance.

Fig. 5.

Heat map and hierarchical clustering analysis. A heat map depicting the differential abundance of 69 metabolites was generated. Hierarchical clustering analysis revealed that the clustering tree exhibited four distinct main branches, corresponding to the four experimental groups. The color scale indicated an increase or decrease in metabolite levels relative to the median, with red representing an increase and blue indicating a decrease, respectively (see color scale).

After comparing with the basal diet groups, an unsupervised PCA combined with a supervised PLS-DA was conducted to discern the most crucial metabolites that differentiate among 15%, 30%, and 60% ZLM groups, aiming to uncover significant distinctions among these samples. As shown in Fig. 6, these 4 experimental groups were definitely distributed in a 3-D model (R2 = 0.99553, Q2 = 0.99328, Accuracy = 1.0). Then, the 5%, 30% and 60% ZLM supplemental groups were distinguished from the control groups fed with basal diets by principle component 1 to 3 (Fig. 6A) and component 1 to 3 (Fig. 6B) respectively. Using a VIP values (variable importance for the projection) plot in PLS-DA, the cut-off values of discriminating variables were set as more than one, and then these metabolomic biomarkers were screened as showed in Fig. 6C. There were 19 biomarkers of histine, geosmin, hypoxanthine, 2-Methylisoborneol, valine, trimetlylamine, arginine, asparagine, glutamine, tyrosine, uric acid, fumaric acid, gamma-aminobutyric acid, inosine-5^′-monophosphate (IMP), maltose, tetradecanoic acid, glutamic acid, cystine, threonine, which VIP values of component 1 to 3 were more than one. After conducting a 2000 permutation analysis to assess the reliability of PLS-DA, the cross-validated and observed coefficients of R2 and Q2 were depicted in Fig. 6D, demonstrating no evidence of overfitting in PLS-DA due to a p-value less than 5 $\times$ 10^-4 obtained from the permutation test.

Fig. 6.

Biomarkers. Identification of crucial metabolites was performed using unsupervised and supervised pattern recognition analyses, namely principal component analysis (PCA) and partial least squares-discriminant analysis (PLS-DA), to distinguish the groups consisting of 15%, 30%, and 60% ZLM as well as the basal diet groups. Each dot in the plot represented a technological replicate analysis of samples. (A) PCA score plot with PC1, PC2, and PC3 explained 89.2% of the total variance, enabling confident interpretation of variation. (B) PLS-DA score plot with Component 1, Component 2, and Component 3 explained 89.2% of the total variance, allowing confident interpretation of variation. (C) Variable importance in the projection (VIP) plot generated from PLS-DA (Accuracy = 1.0, R2 = 0.99553, Q2 = 0.99328), where each dot represented a candidate biomarker highlighted by its average abundance shown in color key. There were 19 biomarker with VIP values greater than one for components 1 to 3. Here inosine-5^′-monophosphate (IMP) referred to inosine-5^′-monophosphate. (D) In PLS-DA, cross-validated and observed coefficients of R2 and Q2 were obtained through permutation analysis involving 2000 iterations (p 5 $\times$ 10^-4).

Furthermore, this research uncovered these enriched metabolic pathways which were determined among the basal diet and 15%, 30% and 60% ZLM supplemental groups. These enriched pathways were especially important to learn the metabolomic effects response to these control basal diet and 15%, 30% and 60% ZLM supplementary groups. The enrichment assays were performed using MetaboAnalyst 6.0. A total of 15 metabolic pathways exhibited significant enrichment in these groups (p 0.05), as illustrated in Fig. 7. According to their p values significance, 1 to 15 denoted Alanine, aspartate and glutamate metabolism; Aminoacyl-tRNA biosynthesis; Arginine biosynthesis; Purine metabolism; Citrate cycle (TCA cycle); Glyoxylate and dicarboxylate metabolism; D-Glutamine and D-glutamate metabolism; Taurine and hypotaurine metabolism; Butanoate metabolism; Biosynthesis of unsaturated fatty acids; Arginine and proline metabolism; Glutathione metabolism; Pantothenate and CoA biosynthesis; beta-Alanine metabolism; Pyrimidine metabolism, respectively. As shown in Fig. 8, particular interest was these metabolites’ GC-MS abundance (XIC) included in the 15 significant pathways. Here, a trichromatic diagram was used to plotting the variation of integrative analysis of metabolites in significantly enriched pathways. Red and blue represented these max and min abundance values for each row of metabolites, respectively.

Fig. 7.

Pathway enrichment analysis. Pathway enrichment analysis revealed significant differences in the metabolite profiles among the 15%, 30%, and 60% ZLM groups compared to the basal diet control groups (p 0.05). The significance of these differences was ranked based on their p-values, from highest to lowest, which 1 to 15 represented Alanine, aspartate and glutamate metabolism; Aminoacyl-tRNA biosynthesis; Arginine biosynthesis; Purine metabolism; Citrate cycle (TCA cycle); Glyoxylate and dicarboxylate metabolism; D-Glutamine and D-glutamate metabolism; Taurine and hypotaurine metabolism; Butanoate metabolism; Biosynthesis of unsaturated fatty acids; Arginine and proline metabolism; Glutathione metabolism; Pantothenate and CoA biosynthesis; beta-Alanine metabolism; Pyrimidine metabolism, respectively. Significant enriched pathways are selected to plot.

Fig. 8.

Pathways’ metabolites. Integrative analysis of metabolites’ GC-MS abundance (XIC) in significantly enriched pathways was performed, and a chromatic diagram was utilized to differentiate the average abundance of these metabolites. The colors red and blue were assigned to represent the maximum and minimum abundances for each row (metabolite), respectively.

It was hypothesized that ZLM supplement could promote tilapia meat flavor. To verify this hypothesis, tilapias were fed with the basal diet and 15%, 30% and 60% ZLM supplements for 30 days. The findings of recent study suggested that ZLM supplements possessed metabolic effects in a dose-dependent manner.

With an increasing ZLM supplement, these unsaturated fatty acids abundances in tilapia liver also increased. To further estimate ZLM impacts on the biosynthesis of fatty acids, the activity involved critical enzyme of ACL, ACC and FAS was determined after feeding for 30 days. Compared with control groups fed by basal diets, these enzymatic activities were all up-regulated in the experimental groups fed by 15%, 30% and 60% ZLM supplements. With an increasing ZLM, the activities also increased. And in detail as shown in Fig. 9A, the ACL activity increased from 300.15 $\pm$ 39.09 U/mg in the control group, to 490.44 $\pm$ 52.89 U/mg and 547.19 $\pm$ 59.20 U/mg in the groups treated with 15% and 30% ZLM respectively, ultimately reaching a level of 887.16 $\pm$ 58.74 U/mg in the group treated with 60% ZLM; Similarly, ACC activity were up-regulated from 82.41 $\pm$ 1.47 U/mg in the control group, to 93.79 $\pm$ 1.91 U/mg and 105.04 $\pm$ 2.24 U/mg in the groups treated with 15% and 30% ZLM respectively, finally reached 127.41 $\pm$ 2.96 U/mg of 60% ZLM group; FAS activity were promoted from 133.98 $\pm$ 16.79 U/mg in the control group, to 150.95 $\pm$ 14.36 U/mg and 201.86 $\pm$ 10.72 U/mg in the groups treated with 15% and 30% ZLM respectively, finally reached 446.58 $\pm$ 31.50 U/mg of 60% ZLM group. On the contrary, the enzymatic activity of LPS was no significant change with an increasing ZLM supplements. And in detail, it’s 282.36 $\pm$ 28.80 U/mg of control group, and 281.18 $\pm$ 47.48 U/mg, 285.31 $\pm$ 33.60 U/mg and 284.50 $\pm$ 40.97 U/mg of 15%, 30% and 60% ZLM group respectively. Moreover, after the fatty acids biosynthesis were promoted, thus acetyl Co-A contents were then evaluated to confirm its origin, which was either from fatty acid degradation or from energy metabolism (TCA cycle). As highlighted in Fig. 9B, the contents of acetyl Co-A were enhanced from 612.51 $\pm$ 54.77 nMol/mg of control group, to 2411.04 $\pm$ 98.36 nMol/mg and 3621.91 $\pm$ 183.67 nMol/mg of 15% and 30% ZLM group respectively, finally upto 3985.44 $\pm$ 209.79 nMol/mg of 60% ZLM group. LPS activity did not change significantly, but the contents of acetyl Co-A increased observably. These findings suggested that ZLM might boost fatty acids biosynthesis to further improve the accumulation of unsaturated fatty acids. Concurrently, acetyl Co-A originated from TCA cycle of energy metabolism. Thereby, the ZLM impacts were focused on enhancing unsaturated fatty acids biosynthesis in tilapia fed with 15%, 30%, and 60% ZLM supplements.

Fig. 9.

The unsaturated fatty acids biosynthesis. (A) Activity of ACL, ACC, FAS, LPS of livers fed with 15%, 30% and 60% ZLM and basal diet. (B) Content of acetyl Co-A of livers in the presence of 15%, 30% and 60% ZLM compared with basal diet. ** p 0.01. ACL, ATP citrate lyase; ACC, acetyl-CoA carboxylase; FAS, fatty acid synthase; LPS, lipase.

These increasing ZLM supplements could accumulate amino acids in tilapia livers. To evaluate the ZLM impacts on amino acid and protein biosynthesis, GS and GOGAT activities were detected after fed by ZLM and control diets for 30 days. Compared with the control, GS and GOGAT were up-regulated in 15%, 30% and 60% ZLM supplement groups. The activity of GS increased from 2.34 $\pm$ 0.19 U/mg in the control group to 2.91 $\pm$ 0.17 U/mg and 3.77 $\pm$ 0.18 U/mg in the groups fed with 15% and 30% ZLM, respectively, ultimately reaching a level of 5.26 $\pm$ 0.21 U/mg in the 60% ZLM group. Additionally, GOGAT activity increased from 52.24 $\pm$ 8.91 U/mg in the control group to 94.84 $\pm$ 7.86 U/mg and 135.83 $\pm$ 13.88 U/mg in the groups treated with 15 % and 30 % ZLM, respectively, further increasing to 178.43 $\pm$ 8.26 U/mg in the 60 % ZLM group (Fig. 10). These results suggested that ZLM might boost amino acid and protein biosynthesis, thereby improving the amino acids accumulation. Thus, ZLM impacts were highlighted on enhancing amino acid and protein biosynthesis.

Fig. 10.

The amino acids biosynthesis. Activity of GS and GOGAT of livers fed with 15%, 30% and 60% ZLM and basal diet. ** p 0.01. GS, glutamine synthetase; GOGAT, glutamate synthase.

To further uncover the ZLM impacts on tilapia energy metabolism, the ATP and NADH contents were detected as profiled in Fig. 11A,B. The relative fluorescence intensity (RFI) of ATP was measured in this study. Specifically, the ATP levels increased from 59,355.00 $\pm$ 2627.77 RFI/mg in the control group to 77,836.67 $\pm$ 656.99 RFI/mg and 87,255.83 $\pm$ 551.35 RFI/mg in the groups treated with 15% and 30% ZLM respectively, ultimately reaching a value of 97,969.17 $\pm$ 1499.21 RFI/mg in the group treated with 60% ZLM. Additionally, NADH levels increased from 3.45 $\pm$ 0.05 µMol/mg in the control group to 4.14 $\pm$ 0.05 µMol/mg in the 15% and 30% ZLM groups respectively, and finally reached 5.44 $\pm$ 0.09 µMol/mg in the group treated with 60 % ZLM. Generally, the ATP and NADH were closely related to the TCA cycle responses. Then, these enzymes of TCA cycle were further determined. As described in Fig. 11C, compared with control, these enzymatic activities were all promoted in these groups fed by 15%, 30%, and 60% ZLM supplements, details as followed. The $\alpha$-KGDH activities were increased from 144.71 $\pm$ 13.01 U/mg in the control group, to 200.52 $\pm$ 9.71 U/mg and 262.54 $\pm$ 11.99 U/mg in the groups treated with 15% and 30% ZLM respectively, ultimately reaching up to 297.68 $\pm$ 12.19 U/mg of 60% ZLM group; PDH was from 120.32 $\pm$ 15.30 U/mg in the control group, to 150.78 $\pm$ 11.04 U/mg and 187.33 $\pm$ 7.09 U/mg in the groups treated with 15% and 30% ZLM respectively, finally reaching up to 211.70 $\pm$ 14.50 U/mg of 60% ZLM group; SDH was from 249.83 $\pm$ 7.91 U/mg in the control group, to 295.25 $\pm$ 7.30 U/mg and 327.31 $\pm$ 8.19 U/mg in the groups treated with 15% and 30% ZLM respectively, ultimately reaching up to 363.39 $\pm$ 20.79 U/mg of 60% ZLM group; NAD-MDH was from 127.59 $\pm$ 2.78 U/mg in the control group, to 146.98 $\pm$ 4.54 U/mg and 176.70 $\pm$ 2.40 U/mg in the groups treated with 15% and 30% ZLM respectively, finally reaching up to 197.71 $\pm$ 2.89 U/mg of 60% ZLM group; And ICDHm was from 80.01 $\pm$ 14.83 U/mg in the control group, to 99.55 $\pm$ 9.09 U/mg and 112.96 $\pm$ 6.05 U/mg in the groups treated with 15% and 30% ZLM respectively, finally up to 150.58 $\pm$ 18.39 U/mg of 60% ZLM group. These findings suggested that ZLM could boost the TCA cycle of tilapia fed by 15%, 30%, and 60% ZLM supplements.

Fig. 11.

The energy metabolism. Compared to the basal diets, the presence of 15%, 30%, and 60% ZLM feeds resulted in significant alterations in liver ATP activity (A) and NADH content (B), as well as ICDHm, NAD-MDH, SDH, PDH, and $\alpha$-KGDH levels (C). The relative fluorescence intensity (RFI) of probe was employed for ATP detection. ** p 0.01. ICDHm, mitochondrial isocitrate dehydrogenase; NAD-MDH, NADP-malate dehydrogenase; SDH, succinate dehydrogenase; PDH, pyruvate dehydrogenase NADH, nicotinamide adenine dinucleotide; $\alpha$-KGDH, $\alpha$-ketoglutarate dehydrogenase.

Ultimately, using iPath 2.0 analysis as shown in Fig. 12, the activated energy metabolism (TCA cycle) promotes amino acid metabolism to increase the expression of flavor amino acids. TCA also promotes the biosynthesis of polyunsaturated fatty acids, thereby increasing the expression of flavor fatty acids of arachidonic acid, linoleic acid (9,12-Octadecadienoic acid), linolenic acid (9,12,15-Octadecatrienoic acid), and oleic acid (9-Octadecenoic acid). Additionally, TCA also activated nucleotide metabolism (Fig. 12).

Fig. 12.

iPath analysis with iPath 2.0. Compared to the basal diets, iPath analysis of the metabolic profiles resulting from tilapias fed with 15%, 30%, and 60% ZLM feed offered a more comprehensive understanding of the impact of 71 differentially abundant metabolites (including acetyl-CoA and GSH) and 10 significant enzymes (p 0.01). Based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) compound (http://www.kegg.jp/kegg/compound/), metabolic network pathways were further analyzed with iPath2.0 (https://pathways.embl.de/). Red line represented increase in the ZLM group; blue and light blue lines represented decrease in the ZLM group.