Abstract

   Siniperca chuatsi is currently one of the most important economic
   farmed freshwater fish in China. The aim of this study was to evaluate
   the metabolic profile of recirculating ponds aquaculture system
   (RAS)-farmed S. chuatsi. Gas Chromatography-Mass Spectrophotometry
   (GC-MS) metabolomic platform was used to comprehensively analyze the
   effects of recirculating ponds aquaculture system (RAS) on the Mandarin
   fish S. chuatsi metabolism. Database searching and statistical analysis
   revealed that there were altogether 335 metabolites quantified
   (similarity > 0) and 205 metabolites were identified by mass spectrum
   matching with a spectral similarity > 700. Among the 335 metabolites
   quantified, 33 metabolites were significantly different (VIP > 1 and
   p < 0.05) between RAS and pond groups. In these thirty-three
   metabolites, taurine, 1-Hexadecanol, Shikimic Acid, Alloxanoic Acid and
   Acetaminophen were higher in the pond group, while 28 metabolites were
   increased notably in the RAS group. The biosynthesis of unsaturated
   fatty acids, lysosome, tryptophan metabolism were recommended as the
   KEGG pathway maps for S. chuatsi farmed in RAS. RAS can provide
   comprehensive benefits to the effects of Siniperca chuatsi metabolism,
   which suggest RAS is an efficient, economic, and environmentally
   friendly farming system compared to pond system.

   Subject terms: Metabolomics, Metabolism

Introduction

   The Chinese mandarin fish Siniperca chuatsi (Basilewsky) is a
   freshwater fish with high economic value and is endemic to East Asia,
   specially distributed in the Yangtze River drainage in China^[30]1. The
   resources of wild S. chuatsi have declined dramatically because of
   water pollution, damming and over-fishing in recent years^[31]2. At
   present, with the social and economic development, the popularity of
   the fish has increased, the catches from the wild could not meet the
   demand, and the resulting market price has created much interest in the
   aquaculture of S. chuatsi. In order to adapt to market demand, the
   stocking and culture techniques were developed for the S. chuatsi. The
   mandarin fish is widely cultured throughout the country, and is also
   important in stocking fisheries in lakes and reservoirs^[32]3. However,
   outbreaks of the diseases caused by viruses, bacteria and parasites
   have brought severe economic losses to the S. chuatsi breeding
   industry^[33]4,[34]5. Due to water pollution, low survival rate and the
   nutrients issue of a serious waste of water, traditional ponds of S.
   chuatsi at this stage have encountered many difficulties, which have
   largely hindered its commercial exploitation. Recirculating ponds
   aquaculture system (RAS) is a new ecological ponds aquaculture guided
   by the idea of cyclic economy^[35]6. RAS have gained increasing
   interest in recent years as a means to intensify fish production while
   at the same time reduce water and land usage, minimizing the adverse
   environmental impact^[36]7,[37]8. The pond healthy breeding technology
   demonstration and promotion of Siniperca chuatsi was carried out to
   meet the needs of high-quality aquatic products and promote the
   aquaculture industry in restructuring.

   Metabolomics is an “omics” technique that is situated downstream of
   proteomics, transcriptomics and genomics^[38]9. Metabolomics is defined
   as the quantitative measurement of the dynamic multiparametric
   metabolic response of living systems to pathophysiological stimuli or
   genetic modification^[39]10, which has been proposed as a powerful tool
   for exploring the complex relationship between nutrition and health in
   nutrition research^[40]11–[41]13. Gas chromatography-mass spectrometry
   (GC-MS) has many merits, for example powerful resolving capability and
   high reproducibility, that render it extensively useful in the field of
   metabolomic profiling^[42]14. Serum metabonomic profiles of RAS- and
   pond-cultured S. chuatsi were detected using GC-MS techniques employed
   to globally characterize changes. So far, this is the first time that
   GC-MS techniques were applied to identify the metabolic profile of the
   RAS- and pond-cultured S. chuatsi.

Results

Metabolomic profiling through GC- MS

   A total of 775 valid peaks were identified in the serum. 335 peaks were
   retained after filtering and de-noising, and most peaks were identified
   and attributed to endogenous metabolites (similarity > 0)
   (Table [43]S1). Among these, 205 metabolites based on mass spectrum
   matching were identified (spectral similarity value > 700), which
   indicated high credibility (Table [44]S1). The specific variable
   quantities were indicated in the RAS compared with the pond by
   fold-change (FC) value. The metabolites of RAS- and pond- farmed S.
   chuatsi could be visually divided into down-regulation (FC value < 1)
   and up-regulation (FC value > 1). As shown in Fig. [45]1, metabolites
   down-regulation was more common than up-regulation at RAS- and pond-
   farmed S. chuatsi groups. Compared with the RAS group, only 3
   metabolites were upregulated in the pond group. 33 metabolites were
   significantly different (p < 0.05 and VIP > 1) between pond and RAS
   groups in the total 335 metabolites quantified (Table [46]1), which
   consisted of 19 metabolites with similarity > 700 and 14 metabolites
   with similarity <700. In the 19 metabolites with similarity >700,
   inositol-4-monophosphate, mannonic acid, linoleic acid, isohexonic
   acid, 2-hydroxybutanoic acid, 2-deoxytetronic acid, uric acid,
   2-hydroxyhexanoic acid, 1-monopalmitin, behenic acid, isoheptadecanoic
   acid, diglycerol, mannose-6-phosphate, 2-ketoisovaleric acid, lauric
   acid, arachidic acid and xanthurenic acid were higher in the RAS group,
   while taurine and 1-hexadecanol were higher in the pond group.

Figure 1.

   Figure 1
   [47]Open in a new tab

   Scatter plots of metabolomics between the RAS group and pond group. The
   horizontal and vertical axes indicate expression levels of the
   metabolites in the two groups. The red and blue dots represent up- and
   down-regulated metabolites with significant differential metabolites
   (−log^10 (p value) and log^2 (FC), P value < 0.05) in Siniperca
   chuatsi, respectively. The gray dots represent the metabolites without
   significant differential metabolites.

Table 1.

   Identification of significantly different metabolites in serum between
   the RAS and Pond groups.
   Metabolite name Similarity Average RT (min) VIP p-value log[2](FC) RAS
   group Pond group
   Inositol-4-Monophosphate 956 30.91 2.077 0.001 −0.776 0.2269 0.1325
   Mannonic Acid 918 25.40 1.613 0.031 −0.444 0.1943 0.1429
   Linoleic Acid 910 28.81 1.725 0.025 −1.002 0.0883 0.0441
   Taurine 890 18.80 1.700 0.023 2.359 0.0956 0.4903
   Isohexonic Acid 875 25.80 1.639 0.049 −0.600 0.0042 0.0028
   2-Hydroxybutanoic Acid 869 7.89 2.006 0.001 −0.865 0.0342 0.0188
   2-Deoxytetronic Acid 867 12.06 1.956 0.012 −0.976 0.0053 0.0027
   Uric Acid 845 27.15 1.678 0.028 −0.706 0.0014 0.0009
   2-Hydroxyhexanoic Acid 841 9.76 2.035 0.001 −1.282 0.0141 0.0058
   1-Monopalmitin 830 32.47 1.674 0.027 −0.851 0.0867 0.0481
   Behenic Acid 795 32.85 1.940 0.002 −0.892 0.0097 0.0053
   Isoheptadecanoic Acid 787 27.45 1.763 0.018 −0.925 0.0157 0.0083
   Diglycerol 773 20.65 1.331 0.045 −0.916 0.0367 0.0195
   Mannose-6-Phosphate 773 30.27 1.982 0.003 −0.758 0.0235 0.0139
   2-Ketoisovaleric Acid 769 7.58 1.839 0.026 −1.340 0.0023 0.0009
   Lauric Acid 762 18.62 1.646 0.019 −0.201 0.0026 0.0023
   Arachidic Acid 758 31.27 1.695 0.017 −1.041 0.0172 0.0084
   1-Hexadecanol 752 24.93 2.019 0.001 0.462 0.0095 0.0131
   Xanthurenic Acid 745 29.3 1.931 0.022 −1.418 0.0043 0.0016
   Shikimic Acid 688 21.65 1.441 0.050 5.448 0.0022 0.0963
   1-Monostearin 680 33.89 1.625 0.018 −1.251 0.0140 0.0059
   Monomyristin 679 30.81 1.728 0.027 −0.994 0.0038 0.0019
   Vanillic Acid 671 21.06 1.890 0.005 −1.246 0.2311 0.0974
   Alloxanoic Acid 664 29.12 1.576 0.017 4.139 0.0012 0.0220
   Trisaccharide 656 35.35 1.841 0.022 −0.911 0.0065 0.0034
   Urocanic Acid 637 25.85 1.780 0.018 −0.709 0.0035 0.0022
   Spermine 616 33.04 1.838 0.020 −0.560 0.0060 0.0041
   Glycyl Tyrosine 610 32.88 1.575 0.030 −0.923 0.0020 0.0010
   Glutamyl-Valine 608 26.71 1.725 0.006 −1.210 0.0065 0.0028
   Indole-3-Acetate 578 25.2 1.739 0.017 −1.010 0.0098 0.0049
   Glucose D7 Labeled 538 23.82 1.710 0.011 −0.405 0.0030 0.0023
   Acetaminophen 493 17.85 1.793 0.005 0.569 0.0031 0.0047
   Tyramine 397 24.6 1.561 0.041 −0.256 0.0013 0.0011
   [48]Open in a new tab

Principal component analysis

   To reduce the complexity of the datasets, PCA was applied. The PCA
   score plot of QC (Quality control), RAS and Pond groups from Siniperca
   chuatsi serum metabolites was shown in Fig. [49]2A. PCA score plot
   indicated a clear difference in all metabolites between the pond and
   RAS groups (Fig. [50]2B). The R2X value of the PCA model was 0.255 in
   the Siniperca chuatsi serum. All samples from RAS and Pond groups fell
   outside the Hotelling’s T2 tolerance ellipse with 95% confidence, which
   indicated that no outlier was observed among the samples analysed. The
   validation plot for the PLS-DA model obviously revealed that the
   permutation tests of the serum were valid (R2X = 0.244, R2Y = 0.095)
   (Fig. [51]2C), which was the satisfactory effectiveness of the model.
   All the samples between the RAS and Pond groups were within the 95%
   Hotelling’s T2 ellipse based on the score plots of OPLS-DA model
   (Fig. [52]2D). As shown in Table [53]1, a total of 33 differential
   metabolites were identified between the RAS and pond groups (VIP > 1,
   p < 0.05), which was made of four amino acids and their derivatives,
   three saccharides, eleven organic acids, nine fatty acids and six other
   metabolites. Further, 33 identified metabolites revealed the existence
   of distinct differences between the RAS and pond groups by hierarchical
   cluster analysis (Fig. [54]3). Based on the clustering result of
   metabolites, serine, glycine, threonine, proline, asparagine, valine,
   glutamic acid, propanoic acid, lysine, hexadecanoic acid, octadecanoic
   acid and ornithine were abundant in each sample of the RAS groups,
   alloxanoic acid, 1-hexadecanol, acetaminophen, taurine, shikimic acid
   were abundant in each sample of the pond group. The KEGG pathway
   analysis was performed by MetaboAnalyst 3.0. Functional pathway
   analysis facilitating further biological interpretation revealed the
   most relevant pathways such as Biosynthesis of unsaturated fatty acids,
   Lysosome, Tryptophan metabolism, Taurine and hypotaurine metabolism,
   Neuroactive ligand-receptor interaction, Linoleic acid metabolism,
   Pantothenate and CoA biosynthesis; Phenylalanine, tyrosine and
   tryptophan biosynthesis; Valine, leucine and isoleucine biosynthesis;
   beta-Alanine metabolism, Propanoate metabolism, Glutathione metabolism,
   isoleucine degradation, leucine and Valine; Histidine metabolism,
   Mannose and fructose metabolism, Primary bile acid biosynthesis, Fatty
   acid biosynthesis, Fatty acid metabolism, Tyrosine metabolism,
   Nucleotide sugar and amino sugar metabolism, Proline and arginine
   metabolism, ABC transporters, Purine metabolism, Biosynthesis of
   secondary metabolites (Table [55]S2). KEGG pathway mapper indicated
   that ten metabolic signaling pathways were significantly different
   among two groups (p < 0.05), including Biosynthesis of unsaturated
   fatty acids, Lysosome, Tryptophan metabolism, Taurine and hypotaurine
   metabolism, Neuroactive ligand-receptor interaction, Linoleic acid
   metabolism, Pantothenate and CoA biosynthesis, Tryptophan and tyrosine
   biosynthesis, Leucine, Phenylalanine, Isoleucine and Valine
   biosynthesis and beta-Alanine metabolism (Table [56]S2). The three
   metabolic pathways of Biosynthesis of unsaturated fatty acids, Lysosome
   and Tryptophan metabolism were significant differences between the RAS
   and pond groups (Fig. [57]4).

Figure 2.

   [58]Figure 2
   [59]Open in a new tab

   Score plot of PCA (principal component analysis) modeling for different
   treatments. (A) The PCA (principal component analysis) score plot of QC
   (Quality control), RAS and Pond groups in Siniperca chuatsi serum
   metabolites. (B) PCA score plots for pairwise comparisons between RAS
   and Pond group. (C) PLS-DA score plots for pairwise comparisons between
   RAS and Pond group. (D) OPLS-DA score plots for pairwise comparisons
   between RAS and Pond group.

Figure 3.

   [60]Figure 3
   [61]Open in a new tab

   Heat map of varied abundance of metabolites. Red and green indicate
   increase and decrease of metabolites relative to the median metabolite
   level, respectively (see color scale).

Figure 4.

   Figure 4
   [62]Open in a new tab

   Kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment
   analysis of Siniperca chuatsi serum differential metabolites in RAS and
   Pond groups. All the matched pathways are displayed in columns, and the
   length represents the fold change of enrichment, the pink dotted line
   represents the 0.01 P-value and the blue dotted line represents the
   0.05 P-value. Total: the total number of metabolite in that set
   pathway; Expected = the expectation value; Hits = the number of
   metabolites in the experimental set matching the pathway set; Holm
   P = the P-value adjusted by Holm-Bonferroni method; FDR = the P-value
   adjusted using False Discovery Rate; Impact = the pathway impact value.

Discussion

   The mandarin fish has a relatively high market value in
   China^[63]3,[64]15, which is widely cultured throughout the country due
   to their rapid growth, large size, high nutritional value and delicious
   flesh and high price^[65]3,[66]15,[67]16. The successful artificial
   reproduction, fry and fingerling rearing technology of S. chuatsi could
   have potential for meeting the demands of S. chuatsi for commercial
   fish culture in China^[68]1,[69]17. Mandarin fish is a key demersal
   piscivore in lakes and river systems, which feed on live shrimps, fish
   and other aquatic animals that are smaller than themselves,
   particularly fish. In the wild the fry of S. chuatsi start feeding,
   they prefer to take live prey fish and not take dead prey fish or
   artificial diets^[70]18–[71]20. Recently, RAS have been well studied,
   and it has been proven major leaps in fish culture to maximize profit
   by increasing production, lowering costs and conserving
   water^[72]21–[73]23. Therefore, in the present study, S. chuatsi were
   farmed in the RAS and pond, respectively, which were fed on live fry of
   cultivated Hypophthalmichthys molitrix and Aristichthys nobilis. In
   this study, the metabolite profiling was investigated using GC/MS to
   determine the metabolic characteristics of RAS and pond-cultured S.
   chuatsi, which could be regarded as the ultimate responses of
   biological systems to the environmental changes^[74]24–[75]26. Up to
   now, many powerful analytical techniques have been used for identifying
   metabolite profiling, such as GC-MS and HPLC-MS (high performance
   liquid chromatography with mass spectrometry)^[76]27–[77]29. GC-MS as a
   mature technology has a lot of advantages, such as good
   reproducibility, high sensitivity, high resolution, with a large
   standard library, relatively low cost and powerful analysis, which can
   be used for analyzing the primary metabolism products, including
   carbohydrates, organic acids, amino acids, and fatty
   acids^[78]30–[79]32. For instance, glucose, oxypurinol, taurine,
   lactate, creatine, glutamate, alanine, sn-glycerol-3-phosphorylcholine,
   glycine, phenol, hypoxanthine, acetic acid, lysine, leucine and valine
   were main hepatopancreas metabolites differing in serum of normal grass
   carp based on ^1H-NMR^[80]33. The metabolic differences were revealed
   in Eriocheir sinensis fed with dietary olive oil or palm oil using
   untargeted GC-MS metabolomics^[81]31. 68 metabolites were identified
   with high credibility and five metabolites were significantly different
   between PO group and OO group^[82]31. The derivatization is required in
   order to achieve prior to the analysis using GC-MS because of most of
   these metabolites are nonvolatile^[83]34.

   Currently, we applied the GC-MS metabolomics approach revealed
   metabolic differences in the RAS- and pond-farmed S. chuatsi fed with
   dietary Hypophthalmichthys molitrix and Aristichthys nobilis because of
   the metabolites in circulation have been commonly employed to analyze
   the physiological status in response to multiple factors, including
   stresses, feed change, disease and so on^[84]35. So far, this is the
   first time to evaluate metabolic changes in the RAS-farmed S. chuatsi.
   We have discriminated the Siniperca chuatsi from RAS and pond in a PCA
   analysis of GC-MS metabolites. PCA is an unsupervised pattern
   recognition method, which was performed to examine the intrinsic
   variation in the dataset^[85]36,[86]37. The greatest metabolic variance
   in our studies were related to fish culture environment and not caused
   from farmed Siniperca chuatsi. Since the S. chuatsi were farmed in
   recirculating ponds aquaculture system and pond, respectively. 33
   metabolites were significantly different (p < 0.05 and VIP > 1) between
   RAS and pond groups in the S. chuatsi. Among all of them, organic acids
   accounted for approximately 60.6%, and most of which were related to
   fatty acid metabolism. These metabolites were comprised of 19
   metabolites with the similarity over 700 and 14 metabolites with
   similarity under700. Some marker metabolites were worth investigating
   in the future. Compared with the RAS group, the levels of taurine,
   1-hexadecanol, shikimic acid, alloxanoic acid and acetaminophen were
   higher in the pond group. In fish, taurine is mainly conjugated with
   choline to produce bovine bile sulfonate in the liver. In contrast to
   fish, bilirubin is conjugated with sugar (mainly glucuronic acid), and
   excreted in the bile in mammals^[87]38. Since taurine is involved in
   various functions, like exogenous substances, cell protection,
   neuro-modulation or neuro-transmission, detoxification of
   endogenous^[88]39. A large amount of taurine accumulated in the fish
   body, but its precise physiological significance is uncertain^[89]38.
   The contents of twenty-eight metabolites increased notably in the RAS
   group, including vanillic acid, inositol-4-monophosphate, mannonic
   acid, linoleic acid, isohexonic acid, 2-hydroxyhexanoic acid,
   2-deoxytetronic acid, uric acid, 2-hydroxybutanoic acid,
   1-monopalmitin, behenic acid, isoheptadecanoic acid, diglycerol,
   mannose-6-phosphate, 2-ketoisovaleric acid, lauric acid, arachidic acid
   and xanthurenic acid, 1-monostearin, monomyristin, trisaccharide,
   urocanic acid, spermine, glycyl tyrosine, glutamyl-valine,
   indole-3-acetate, glucose D7 labeled and tyramine. In this study, RAS
   has a lot of advantages, such as more dissolved oxygen and improving
   lots of environmental factors (TAN, NO^−[2]-N, P and COD), which can
   provide comprehensive benefits to the S. chuatsi. The content of
   vanillic acid is the highest in the RAS group. Vanillic acid belonged
   to cinnamic acid derivative with strong antioxidant and antibacterial
   activity^[90]40. Inositol-phosphates, as a class of signal molecule,
   play an important role in cell signal transduction, and are related to
   a lot of physical activities. The mechanism of inositol phosphate
   signal transduction is closely related to the generation, metabolism
   and biological transformation of inositol phosphate. The nature is
   various chemical reactions (phosphorylation/dephosphorylation,
   interactions with inositol phosphate receptors). Mutual transformations
   of these reactions present a very complex network control
   system^[91]41. As compared to the pond group, Oleic acid
   and1-Monopalmitin in RAS group showed a clear increasing trend, which
   might be originated from the desaturation of saturated fatty
   acids^[92]42. Besides, other kinds of compounds, like amines, esters,
   alcohols and so on, were also identified in the RAS and pond groups,
   which could devote to the formation of the secondary metabolites that
   played a key role in the flavor components^[93]43. Significant changes
   in the levels of the main metabolites might show its key role in a
   number of metabolic pathways for regulating adaptation of the RAS- and
   pond-cultured S. chuatsi. The present study revealed that integration
   of metabolomics could be used to produce complementary data, which
   contribute to a better understand regarding correlation of fermentation
   process. However, there are still few limitations with metabolic
   differences of RAS- and pond-farmed S. chuatsi. The most obvious reason
   was only 335 identified metabolites in the RAS and pond groups, which
   might be attributed to high molecular weight metabolites and some
   volatile metabolites being missed. Thus, more sensitive high-throughput
   and advanced omics technologies should be used to accounts for the
   entire metabolic network. KEGG pathway mapper and metabolite set
   enrichment analysis suggested that these metabolites were related
   mainly to unsaturated fatty acid synthesis, Lysosome and amino acid
   metabolism.

   In conclusion, this study was conducted to evaluate the use of GC-MS
   metabolomic platform to comprehensively analyze the effects of
   recirculating ponds aquaculture system (RAS) on the S. chuatsi
   metabolism. Database searching and statistical analysis revealed that
   there were altogether 335 metabolites quantified (similarity > 0) and
   205 metabolites were described by mass spectrum matching with the
   similarity over 700. Among the total 335 metabolites quantified, 33
   metabolites were significantly different (p < 0.05and VIP > 1) between
   RAS and pond groups. Biosynthesis of unsaturated fatty acids, lysosome,
   tryptophan metabolism were recommended as the KEGG pathway maps for S.
   chuatsi farmed in RAS. The present study revealed that RAS could
   provide comprehensive benefits to the effects of S. chuatsi metabolism,
   which suggested RAS was an efficient, economic, and environmentally
   friendly farming system compared to pond system.

Methods

Ethics statement

   Investigations and protocols were conducted according to the guiding
   principles for the use and care of laboratory animals and in compliance
   with Anhui Science and Technology University Institute of Animal Care
   and Use Committee. The institutional review board approved this
   procedure. Our study had been submitted to and approved by the Academic
   Ethics Committee of Anhui Science and Technology University. All sample
   collection was undertaken in accordance with relevant Academic Ethics
   Committee of Anhui Science and Technology University guidelines and
   regulations.

Fish housing and feeding

   S. chuatsi juveniles were farmed in the RAS and pond of Chuzhou Nanqiao
   District Yangtze River Aquaculture Breeding Ground (Chuzhou, China)
   from July to September 2018. The fish were cultivated with a 300 m^3
   running water aquaculture pond (30 × 5 × 2 m) and 4200 m^3 pond (60 ×
   35 × 2 m). The culturing density in RAS and pond were 180.6 g /m^3 and
   8.0 g /m^3, respectively. The dissolved oxygen (DO), ammonia nitrogen
   (NH[4]-N), pH, nitrite content (NC), etc., were monitored in two RAS
   and pond of S. chuatsi. The RAS and pond of water quality indicators
   were DO 7.50 ± 0.29 mg/ L, 6.48 ± 0.38 mg/ L, 0.0016 ± 0.0007 mg/ L,
   0.0021 ± 0.0009 mg/ L, pH 7.33 ± 0.41, 7.09 ± 0.51, NC
   0.048 ± 0.014 mg/ L, 0.198 ± 0.065 mg/ L, respectively. The perch were
   provided a live prey fish (Hypophthalmichthys molitrix and Aristichthys
   nobilis) during 3- month experiments. The mandarin fish reached average
   terminal body length of 28.98 ± 0.65 cm and 18.64 ± 0.83 cm, average
   terminal body weight of 426.68 ± 60.85 g and 135.6 ± 27.85 g after 90
   days rearing from average initial body length of 7.12 ± 0.53 cm,
   average initial body weight of 2.58 ± 1.15 g, respectively. Five
   samples of S. chuatsi were caught by brail fishing net in the RAS and
   pond, respectively. All captured fish were transferred to the
   laboratory maintaining a continuous supply of oxygenated water in
   refrigerated van for 1.5 h at 25 °C. Each fish was anesthetized by
   tricaine methanesulfonate (MS-222, 0.2 g/L) for 4 min, and blood was
   obtained by venipuncture from the tail vein of S. chuatsi, and placed
   in serum-separating tubes. The blood of Mandarin fish was allowed to
   clot at room temperature for 30 min. Serum was obtained by
   centrifugation (10 min, 2000 rpm at 4 °C), and stored frozen at −80 °C.

Sample preparation

   To evaluate the metabolism status, serum metabonomic profiles of RAS-
   and pond-cultured S. chuatsi were detected using GC-MS techniques
   employed to globally characterize changes in this study. Ten serum
   samples were slowly thawed at room temperature for calibration curves
   and quality control (QC) samples. 20 μL internal standard
   (2-chloro-l-phenylalanine, 0.3 mg/mL) and 600 μL extraction solvent
   with methanol/water (4/1, v/v) were added to each sample. Samples were
   stored at −80 °C for 2 min and then grinded at 60 HZ for 2 min. 120 μL
   of chloroform was added to the samples, then the samples were
   vigorously vortexed and followed by 10 min ultrasound-associated
   extraction at ambient temperature, then stored at 4 °C (10 min). The
   samples were centrifuged at 12000 rpm for 10 min at 4 °C. QC sample was
   prepared by mixing aliquots of the all samples to be a pooled sample.
   An aliquot of the 150 μL supernatant was transferred to a glass
   sampling vial for vacuum-dry at room temperature. And 80 μL of
   methoxylamine hydrochloride (dissolved in pyridine, 15 mg/mL) was
   subsequently added. The resultant mixture was vortexed vigorously for
   2 min and incubated at 37 °C for 90 min. 80 μL of BSTFA (with 1% TMCS)
   and 20 μL n-hexane were added into the mixture, which was vortexed
   vigorously for 2 min and then derivatized at 70 °C for 60 min. The
   samples were allowed to placed at ambient temperature for 30 min before
   GC-MS analysis.

GC–MS analysis

   After the completion of sample pretreatment, the samples were sent to
   Shanghai OE Biotech. Co., Ltd. (Shanghai, China) for GC-MS analysis.
   Data were log2 transformed by using Microsoft Excel, and the resulting
   data matrix was then imported into SIMCA-P software (version 14.0,
   Umetrics, Umea, Sweden). Principle component analysis (PCA) and
   (orthogonal) partial least-squares-discriminant analysis (O)
   PLS-DA^[94]31 were used to evaluate the metabolic difference among RAS
   and pond groups, after mean centering and unit variance scaling. The
   ellipse represents the Hotelling’s T2 with 95% confidence interval of
   the modeled variation^[95]44. The heat map was produced in the R
   environment for statistical computing. The heat map color drawn using R
   with ggplot2 represents the z-score transformed raw data for RAS-farmed
   S. chuatsi metabolites^[96]45. Variable importance in the projection
   (VIP) statistics from OPLS-DA modeling was used to identify the overall
   contribution of each variable. Those variables (p < 0.05, VIP >1.0) are
   considered relevant for group discrimination. Furthermore, the proteins
   were further identified according to Kyoto Encyclopedia of Genes and
   Genomes (KEGG, [97]http://www.genome.jp/kegg/)^[98]32.

Supplementary information

   [99]Table S1.^ (108.4KB, xlsx)
   [100]Table S2.^ (13.7KB, xlsx)

Acknowledgements