Research Article Decontaminated fishmeal and fish oil from the Baltic Sea are promising feed sources for Arctic char (Salvelinus alpinus L.)—studies of flesh lipid quality and metabolic profile Ken Cheng 1 , Liane Wagner 1 , Ali A. Moazzami 2 , Pedro G omez-Requeni 1 , AnnaLotta Schiller Vestergren 1 , Eva Br € ann € as 3 , Jana Pickova 1,4 and Sofia Trattner 1 1 Department of Food Science, Uppsala BioCenter, Swedish University of Agricultural Sciences, Uppsala, Sweden 2 Department of Chemistry and Biotechnology, Uppsala BioCenter, Swedish University of Agricultural Sciences, Uppsala, Sweden 3 Department of Wildlife, Fish and Environmental Studies, Swedish University of Agricultural Sciences, Umeå, Sweden 4 Faculty of Fisheries and Protection of Waters, University of South Bohemia in Ceske Budejovice, CENAKVA, Vodnany, Czech Republic The Baltic Sea is one of the world’s most pollution-threatened brackish environments and limited direct consumption of fatty fish from the Baltic Sea is recommended. The use of decontaminated Baltic Sea fish raw materials as fish feed could be a strategy to recycle Baltic Sea nutrients back into food chain, while relieving pressure on aqua-feed in the growing aquaculture industry. In this study, defatted fishmeal and semi-purified fish oil from the Baltic Sea were used in fish feeds for Arctic char (Salvelinus alpinus L.). The effects of the Baltic Sea-sourced fish feeds on flesh lipid quality and fish metabolomics, compared with a standard commercial feed as a control, were determined. 1 H NMR-based metabolomics studies indicated disturbances in energy metabolism and hepatic toxicity in fish fed both crude fishmeal and crude fish oil, associated with up-regulation (IGF-I, GHR-I, PPARa, PPARb1A) and down-regulation (SREBP-1 and FAS) of hepatic genes expression. The content of n-3 long chain polyunsaturated fatty acids was not affected by the decontamination process. Thus, this short-term study demonstrates that decontaminating Baltic Sea-sourced fishmeal and fish oil reduces adverse effects in Arctic char. Practical applications: Decontaminated fish materials from the Baltic Sea were shown to be promising feed ingredients for Arctic char (Salvelinus alpinus L.) compared with untreated Baltic Sea-sourced fish feed, which induced changes in fish physiology associated with energy metabolism and hepatotoxicity. Baltic Sea-sourced fish materials containing high levels of long chain polyunsaturated fatty acids are valuable feed ingredients. Keywords: Fatty acids / GHR-I / 1 H NMR metabolomics / IGF / SREBP-1 Received: May 7, 2015 / Revised: September 3, 2015 / Accepted: September 8, 2015 DOI: 10.1002/ejlt.201500247 : Supporting information available online http://dx.doi.org/10.1002/ejlt.201500247 Correspondence to: Ken Cheng, Department of Food Science, Swedish University of Agricultural Sciences, P.O. Box 7051, 75007 Uppsala, Sweden E-mail: [email protected]Fax: þ46 1867 2995 Abbreviations: CFM, crude fishmeal; CFO, crude fish oil; CV, cross- validation; DFM, defatted fishmeal; DHA, docosahexaenoic acid; DL-PCB, dioxin-like PCB; EF1AA, elongation factor 1AA; EPA, eicosapentaenoic acid; FAME, fatty acid methyl esters; FAS, fatty acid synthase; GHR-I, growth hormone receptor I; GSH, glutathione; IGF, insulin-like growth factor; LC- PUFA, long chain polyunsaturated fatty acids; NDL-PCB, non-dioxin-like PCB; OPLS-DA, orthogonal partial least squares-discriminant analysis; PAH, polycyclic aromatic hydrocarbons; PBDE, polybrominated diphenyl ethers; PCA, principal component analysis; PCB, polychlorinated biphenyls; PL, phospholipid; POP, persistent organic pollutants; PPARs, peroxisome proliferator-activated receptors; SPFO, semi-purified fish oil; SREBP-1, sterol regulatory element binding protein-1; TAG, triacylglycerol; TCA, tricarboxylic acid; TEQ, toxic equivalents; VIP, variable importance for projection Eur. J. Lipid Sci. Technol. 2015, 117, 0000–0000 1 ß 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.ejlst.com
Transcript
Research Article
Decontaminated fishmeal and fish oil from theBaltic Sea are promising feed sources for Arctic char(Salvelinus alpinus L.)—studies of flesh lipid qualityand metabolic profile
Ken Cheng1, Liane Wagner1, Ali A. Moazzami2, Pedro G�omez-Requeni1,AnnaLotta Schiller Vestergren1, Eva Br€ann€as3, Jana Pickova1,4 and Sofia Trattner1
1 DepartmentofFoodScience,UppsalaBioCenter,SwedishUniversity ofAgriculturalSciences,Uppsala,Sweden2 Department of Chemistry and Biotechnology, Uppsala BioCenter, Swedish University of Agricultural Sciences,Uppsala, Sweden
3 Department of Wildlife, Fish and Environmental Studies, Swedish University of Agricultural Sciences, Umeå,Sweden
4 Faculty of Fisheries and Protection of Waters, University of South Bohemia in Ceske Budejovice, CENAKVA,Vodnany, Czech Republic
The Baltic Sea is one of the world’s most pollution-threatened brackish environments and limited directconsumption of fatty fish from the Baltic Sea is recommended. The use of decontaminated Baltic Sea fishraw materials as fish feed could be a strategy to recycle Baltic Sea nutrients back into food chain, whilerelieving pressure on aqua-feed in the growing aquaculture industry. In this study, defatted fishmeal andsemi-purified fish oil from the Baltic Sea were used in fish feeds for Arctic char (Salvelinus alpinus L.).The effects of the Baltic Sea-sourced fish feeds on flesh lipid quality and fish metabolomics, comparedwith a standard commercial feed as a control, were determined. 1H NMR-based metabolomics studiesindicated disturbances in energy metabolism and hepatic toxicity in fish fed both crude fishmeal andcrude fish oil, associated with up-regulation (IGF-I, GHR-I, PPARa, PPARb1A) and down-regulation(SREBP-1 and FAS) of hepatic genes expression. The content of n-3 long chain polyunsaturated fattyacids was not affected by the decontamination process. Thus, this short-term study demonstrates thatdecontaminating Baltic Sea-sourced fishmeal and fish oil reduces adverse effects in Arctic char.
Practical applications:Decontaminated fishmaterials from the Baltic Sea were shown to be promisingfeed ingredients for Arctic char (Salvelinus alpinus L.) compared with untreated Baltic Sea-sourced fishfeed, which induced changes in fish physiology associated with energy metabolism and hepatotoxicity.Baltic Sea-sourced fish materials containing high levels of long chain polyunsaturated fatty acids arevaluable feed ingredients.
Received: May 7, 2015 / Revised: September 3, 2015 / Accepted: September 8, 2015
DOI: 10.1002/ejlt.201500247
:Supporting information available online http://dx.doi.org/10.1002/ejlt.201500247
Correspondence to: Ken Cheng, Department of Food Science, SwedishUniversity of Agricultural Sciences, P.O. Box7051, 75007Uppsala, SwedenE-mail: [email protected]: þ46 1867 2995
Since the 1960s, the Baltic Sea has been one of the world’smost threatened marine environments due to extensiveindustrialization and urbanization of coastal areas. Organiccontaminants in the marine environment can accumulate infish fat and jeopardize fish health and human wellbeing afterconsumption. Some marine pollutants, such as persistentorganic pollutants (POP) and heavy metals, have beenshown to provoke metabolic disorder in fish and to generatesymptoms associated with diabetes in mammals consumingpolluted fish [1]. Therefore, even though the level of POP hasdeclined significantly in recent decades, direct consumptionof fatty Baltic fish is not recommended based on a riskassessment [2].
The flesh of fatty fish is considered a good source of n-3long chain polyunsaturated fatty acids (LC-PUFA), mainlyeicosapentaenoic acid (EPA; 20:5n-3) and docosahexaenoicacid (DHA, 22:6n-3), in human food [3]. These FA havebeen shown to have a positive impact on human health, suchas lowering the risk of cardiovascular and inflammatorydisease [4].
With the decline of capture fisheries production, aqua-culture is considered an important way of supplying fish tosatisfy the increasing market demand for fish products [5].Fish raw materials, in the form of fishmeal and fish oil, aretraditionally used in aqua-feed for fish farming. However,with growing aquaculture production, the availability of fishraw materials for fish feed has become limited relative todemand over the past 20 years [6]. To solve this problem,research efforts have focused on the use of alternativefeedstuffs for fish, such as vegetable ingredients. However,fish fed higher amounts of vegetable ingredients have areduced content of n-3 LC-PUFA, lowering the nutritionalquality of the fish flesh [3, 7]. An alternative strategy is to usefish raw materials of no/low interest for human food, whichcould possibly enter the food chain after certain treatments[8, 9]. Olli et al. [10] showed that Atlantic salmon (Salmosalar L.) fed decontaminated fish oil containing less POP hadbetter growth performance, feed conversion ratio and filletquality than those fed the unprocessed fish oil. Thus,following a process of decontamination, pelagic fatty fishfrom the Baltic Sea might be a valuable source of fishmealand fish oil for animal feeds. To the best of our knowledge,the effects of re-introducing decontaminated Baltic-sourcedfish from which pollutants have been removed back intothe food chain, in the form of fish feed, have not beeninvestigated previously.
Despite the complexity of biological responses, metab-olomics methodology can provide insights into changes inorganism metabolites and can help reveal the mechanisms ofphysiological disruption and toxicological response [11]. The1H NMR-based metabolomics approach has been provento be powerful and reliable technology for assessment ofbiological toxicity in fish [12]. Furthermore, fish liver is a
principal site of detoxification, lipid synthesis, and deposi-tion, related to energy homeostasis. Environmental stresscould affect the expression of key hepatic genes involved inlipid and energy metabolic regulation, such as fatty acidsynthase (FAS) and growth hormone receptor/insulin-likegrowth factors (GHR/IGF) axis [13, 14]. Many of theseeffects are reported to be caused by interference withreceptors regulating cellular signalling pathways, such asperoxisome proliferator-activated receptors (PPARs), andsterol-regulator element-binding protein-1 (SREBP-1) [15,16]. A recent study [12] showed disruption of lipidmetabolism in mice liver, with altered genes and serummetabolites, induced by chemical wastewater exposure.Thus, it is important to evaluate changes in metabolismand key gene expression in fish arising from dietarycomponents, in order to assure fish health and welfare.
Arctic char (Salvelinus alpinus L.) is a predatory coldwater salmonid species with a high content of n-3 LC-PUFA and a good potential market in the northerncircumpolar region [17]. The aim of this study was toevaluate use of decontaminated Baltic Sea-sourced fishmaterials in feed for Arctic char, compared with a standardcommercial fish feed and an untreated Baltic Sea-sourcedfish feed. The analysis focused on metabolic responses in theliver and muscle using a 1H NMR-based metabolomicsapproach and on variations in hepatic gene expressioninvolved in lipid and energy metabolism. Lipid quality infish flesh was also examined.
2 Materials and methods
2.1 Experimental design
Arctic char with an average initial weight of 131.3� 12.19 g(S.D.) were individually tagged with PIT-tags (Passiveintegrated Transponders, Biomark HPT12) and randomlydivided into five groups (10 fish per group). There was nosignificant difference in initial weight between groups(P¼ 0.786). The fish were reared at Aquaculture CentreNorth in central Sweden and kept in a flow-through systemat 10°C for 11 weeks (June–August). The experimentfollowed the guidelines of the Animal Care and UseCommittee at the Swedish University of AgricultureSciences.
One commercial feed purchased from Skretting Com-pany (Stavanger, Norway) was used as the control feed andfour experimental feeds from TripleNine (Esbjerg, Den-mark) were formulated with Baltic Sea-sourced fish rawmaterials (Table 1). Defatted fishmeal (DFM) was producedby TripleNine by organic solvent extraction of lipids fromcrude fishmeal (CFM) to avoid fat-soluble pollutants, whilesemi-purified fish oil (SPFO) was produced by purification ofcrude fish oil (CFO) with activated carbon adsorption(TripleNine) [18]. Four experimental feeds were made by
2 K. Cheng et al. Eur. J. Lipid Sci. Technol. 2015, 117, 0000–0000
arranging fishmeal and fish oil in pairs: three decontaminateddiets, DFM with SPFO (DFMþSPFO), DFM with CFO(DFMþCFO), and CFM with SPFO (CFMþSPFO); andone untreated diet, CFM with CFO (CFMþCFO).
The concentrations of selected organic pollutants (non-dioxin-like polychlorinated biphenyls, NDL-PCB; dioxin-like polychlorinated biphenyls,DL-PCB; polycyclic aromatichydrocarbons, PAH; polybrominated diphenyl ethers, PBDE)in untreated/decontaminated fishmeal and fish oil were
measured by an authority research centre in Czech Republic(Institute of Chemical Technology, Prague, SupplementaryTable S1). The total concentrations of pollutants measuredin the four experimental feeds were calculated (Table 1). Thelipid content and FA composition of the five feeds wereanalyzed (Table 2).
At the end of the feeding trial, final fish body weight andlength were measured. Fish were anaesthetized by MS 222and killed by a blow to the head. Tissue samples (muscle and
Table 1. Dietary formulation (g/kg) and pollutant content of the four experimental diets
liver) were dissected, frozen immediately in liquid nitrogenand stored at �80°C until analysis.
2.2 Lipid analysis
White muscle samples (1 g) of ten fish per group weretaken from the dorsal part under the dorsal fin withoutbones or skin. Lipids in tissues and feeds were extractedwith hexane:isopropanol (3:2, v/v, 15mL) according to themethod described by Mr�az and Pickova [19].
The FA composition in white muscle, liver, and feedwas measured as FA methyl esters (FAME). For hepaticFA analysis, total lipids were first separated into phospholipid(PL) and triacylglycerol (TAG) by thin layer chromatographyusing 20�20cm2 TLC silica gel plates (Merck, Darmstadt,Germany). Boron trifluoride and dry methanol were used formethylation and 15-methylheptadecanoate was added asinternal standard [19]. The GC-FID method and identifica-tion of FA by comparison of retention time with the standardFA mixture GLC-68A (Nu-check-Prep, Inc., Elysian,Minnesota, USA) as described previously [20] were applied.
2.3 1H NMR-based metabolomics analysis
2.3.1 Sample preparation for metabolomics study
Fish liver and white muscle samples (100mg) were extractedusing methanol:chloroform (2:1, v/v, 3mL) as describedpreviously [21]. Samples were separated and collected fromthe aqueous and chloroform phases.
For the aqueous phase, the collected samples were driedusing an evacuated centrifuge (Savant, SVC 100H, TechtumInstrument AB, Umeå, Sweden) and re-dissolved with240mL Millipore water and 280mL sodium phosphatebuffer (0.25mol/L, pH 7.0) [19]. Nanosep centrifugalfilters (3KDa, Pall Life Science, Port Washington, USA)were used for removal of residual proteins in the polar phase[22]. D2O (50mL) and sodium-3-(trimethylsilyl)-2,2,3,3-tetradeuteriopropionate (30mL, 0.3mmol/L, CambridgeIsotope Laboratories, Andover, USA) as internal standardwere added to 520mL filtrate. Finally, 600mL samples wereanalyzed by Bruker NMR spectrometer at 600MHz usingzgesgp pulse sequence (Bruker Spectrospin Ltd., BioSpin,Karlsruhe, Germany) at 25°C with 128 scans and 65,536data points over a spectrum width of 17,942.58Hz.Acquisition time was 1.8 s and relaxation delay 4.0 s.
For the chloroform phase, the collected samples weredried under nitrogen and re-dissolved in 600mL CDCl3(99.96 atom% D, Larodan, Malm€o, Sweden). The NMRspectrometer was the same as used for the polar phase.1H NMR spectra were obtained using zg30 pulse sequence(Bruker Spectrospin Ltd., BioSpin, Karlsruhe, Germany) at20°C with 128 scans and 65,536 data points over a spectrumwidth of 12 019.23Hz. Acquisition time was 2.7 s andrelaxation delay 3.0 s.
2.3.2 Processing and profiling of NMR spectra
All spectra were processed using Bruker Topspin 3.1 asdescribed previously [19]. For the polar phase, 45 metab-olites for muscle and 49 metabolites for liver were identified.The concentrations of these metabolites were calculatedfrom spectra using NMR suite 6.1 profiler (ChenomX Inc.,Edmonton, AB, Canada) after accounting for overlappingsignals and expressed in mmol per gram tissue. For thechloroform phase, NMR spectra were integrated using Amix3.7.3 (Bruker Biospin GmbH, Rheinstetten, Germany)into 0.01 ppm integral regions (buckets) between 0.3 and5.6 ppm. Each spectrum region was scaled to the sum oftotal intensity. The results were expressed as intensity,without unit. NMR Suite 6.1 library (ChenomX Inc.), theHuman Metabolome Database (www.hmdb.ca) and pre-vious literature [21–25] were used for identification of the1H NMR signals.
2.4 Gene expression analysis
Liver samples were first homogenized inQIAzol for 2� 5min(Qiagen, Valencia, CA, USA), followed by chloroformprecipitation. Total RNA in the collected water phase wasisolated using RNeasy Mini kit (Qiagen). The concentrationof purified RNA was normalised to 250 ng/mL. Reversetranscription was performed twice using a High CapacityRNA-to-cDNAKit (Life Technologies, Carlsbad, CA,USA)and a TATAA Grandscript cDNA Synthesis Kit (TATAABiocenter AB, Gothenburg, Sweden). All protocols wereaccording to the manufacturer’s instructions.
Primer design for IGF-I and IGF-II was performed withPrimer-BLAST (Supplementary Table S2). The other assays(GHR-I, PPARa, PPARb1A, PPARg, SREBP-1, and FAS)were designed based on available salmon sequences [20,26–30]. The efficiency and linearity of primers wereevaluated and melting curves were always checked. Elonga-tion factor 1AA (EF1AA) was selected as the reference gene.Primers were obtained from Invitrogen (Life Technologies,Foster City, CA, USA).
The evaluation of SREBP-1 and FAS expression wasperformed using the SYBR green technology with FastSYBR
1
Green Master Mix on a StepOnePlusTM RT-PCRSystem (Life Technologies, Foster City, CA, USA),according to the manufacturer’s instructions [27]. Theremaining genes were evaluated using TATAA SYBR
1
GrandMasterMix #TA01 (TATAABiocenter, Gothenburg,Sweden), 2mL cDNA and 0.20mM of each forward andreverse primer, on an M�3000P instrument (AgilentTechnologies, Palo Alto, CA, USA). The samples wereactivated at 95°C for 1min, followed by 45 cycles ofamplification (95°C for 5 s, 60°C for 10 s, and 72°C for 15 s).All samples were run simultaneously for each gene intriplicate, with a non-template control on each plate. Therelative expression was calculated by comparing the DCT
4 K. Cheng et al. Eur. J. Lipid Sci. Technol. 2015, 117, 0000–0000
values between experimental group and control group usingthe term 2�DDCT, reported as fold change [27].
2.5 Data analysis
The Statistical Analysis System (SAS 9.3, SAS Institute,Cary, NC, USA) was used for lipid and gene expression dataanalysis. Percentage data were first square-root-arcsinetransformed. Distribution of normality (Anderson–Darlingtest) and homoscedasticity (Bartlett’s test or Levene’s test)were checked. If the data failed the tests, the initial data werelog-transformed and retested. General Linear Model wasused for statistical comparison. If the data did not satisfy thetest of distribution or homoscedasticity, Mann–Whitneytest was applied. Tukey’s test was used as a post-hoc test(P< 0.05).
Multivariate data analysis was performed for metabolo-mics data using SIMCA-P (13.0, UMETRICS, Umeå,Sweden). All variables were Pareto-scaled for all datasets.Principal component analysis (PCA) was used to overviewthe data and to search for outliers using Hotelling’s T2 (95%CI) and DModX (95%CI). Orthogonal partial least squares-discriminant analysis (OPLS-DA), which is a regressionmodel, was used to classify different groups. The significanceof the OPLS-DA model was tested using cross-validation(CV) ANOVA (P< 0.05). Variable importance for
projection (VIP) values and their CI from OPLS-DAmodelswere used to identify discriminative metabolites and spectralbuckets between treatments (VIP> 1 and VIP-CI>0). Thedata were further analyzed using SAS, analogously to lipidand gene expression analysis. Tukey’s test was used as a post-hoc test (P< 0.05) and Bonferroni correction (ab¼ 0.0011for 49 metabolites in liver, ab¼ 0.0010 for 45 metabolites inmuscle) was applied to account for multiple testing [21].
3 Results
3.1 Lipid composition in fish muscle and liver
Nomortality occurred during the feeding trial. There was nodifference in final body weight (237.9� 5.34 g, P¼ 0.715)between the groups. The mean weight of the commercial dietgroup was 232.3� 6.32 g. Regarding lipid content, therewere no differences between groups for either white muscleor liver (Table 3 and Supplementary Table S3a).
For the FA profile in white muscle (Table 3), the mainsignificant differences observed were between the control andthe experimental groups. Compared with the experimentalgroups, the control group had a higher content of 18:2n-6and
Pn-6 and a lower content of 20:4n-3 and
PSAFA.
Between the experimental groups, fish fed the diets
Table 3. Total lipid content (%) and fatty acid composition (% of total identified) in white muscle of Arctic char (mean�SEM; n¼ 10)
abcDifferent superscript letters indicate significant differences between the groups (Tukey’s test, P< 0.05).P-value analyzed with ANOVA or Mann–Whitney test (Tukey’s test).
Eur. J. Lipid Sci. Technol. 2015, 117, 0000–0000 Evaluation of decontaminated fish feed using metabolomic 5
containing CFM had more EPA. There was a higherconcentration of
PSAFA and a lower concentration ofP
n-6 in the muscle of fish fed CFMþCFO.The hepatic FA composition was analyzed separately in
PLandTAGfractions (SupplementaryTableS3).As foundformuscle, themain differences were between the control and theexperimental groups.Therewas no significant difference inFAcomposition between the experimental groups.
3.2 1H NMR metabolomics analysis of liver
In order to identify metabolic changes in fish liver, theabsolute concentrations of 49 metabolites were quantifiedthrough a profiling approach from 1H NMR spectra. Twooutliers were observed in the DFMþSPFO and DFMþCFO groups by PCA model and excluded from the dataset.The PCA model established with five groups explained79.4% of variation using four principal components. Themodel parameters were: R2XPC1¼ 35.5%, R2XPC2¼ 25.4%,Q2¼ 45.1%. Overall, there was a separation betweenCFMþCFO and the other four groups. Ten OPLS-DAmodels were further checked by comparing them in pairsand four of these had significant P-value in CV-ANOVA(P< 0.05, Fig. 1). These four OPLS-DA models werebetween CFMþCFO and the other four groups showingclear separation. The model parameters and the P-values ofCV-ANOVA are shown in Fig. 1a–d.
For identification of the metabolites in liver causing thedifferences between the groups, the S-plot, VIP-plot andloading column plot of OPLS-DA were used. The concen-trations of metabolites were further investigated usingANOVA or Mann–Whitney test. The metabolomics analysisof aqueous liver extracts showed an effect on severalmetabolites (Table 4). Fish fed CFMþCFO had loweramounts of glutathione (GSH), but higher amounts ofcholate, choline, glycine, isoleucine, leucine, methionine,phenylalanine, tyrosine, and valine in liver, compared withthe other groups. Moreover, the control group had a higherlevel of asparagine.
The NMR-based metabolomics data analysis of thechloroform phase in liver was analogous to that performedwith the aqueous extracts, but using the NMR spectral datain the form of integrated 0.01 ppm buckets. In the PCAmodel, two outliers from DFMþSPFO and CFMþCFOwere observed and excluded. The first and second compo-nent explained 38.4 and 29.8% of spectral variation in thePCAmodel, respectively, and the total amount ofX variationexplained in the PCA model was 89%. OPLS-DA modelswere further checked in pairs and only the models createdwith control versus CFMþSPFO and with control versusCFMþCFOwere significant models. TheOPLS-DAmodelcreated with control versus CFMþSPFO was explained byone predictive component (R2X¼ 44.8%, R2Y¼ 86.7%,Q2Y¼ 79.6%) with CV-ANOVA P-value¼0.0188. TheOPLS-DA model created with control versus CFMþCFO
was explained by one predictive (48.5%) and one ortho-gonal (20.5%) component (R2X¼ 69%, R2Y¼ 98.7%,Q2Y¼ 91.2%) with CV-ANOVA P-value¼ 0.0434.
Discriminative spectral buckets (VIP> 1 and VIP-CI> 0) in OPLS-DAmodels were further analyzed by ANOVA.The control group showed higher levels of signals corre-sponding to all FA except EPA and DHA, all FA exceptP
n-3 and unsaturated FA compared with the other fourexperimental groups (Supplementary Table S4).
3.3 1H NMR metabolomics analysis of white muscle
To gain further insights into the effects of the untreated dieton muscle metabolism, the muscle samples from CFMþSPFO and CFMþCFO were investigated using 1H NMR-basedmetabolomics analysis. The absolute concentrations of45 metabolites were quantified and compared by PCAmodel. One outlier from CFMþCFO was detected andexcluded from the dataset. The first and second componentexplained 36.5 and 14% of variation in the PCA model,respectively. The plot did not show a clear separationbetween the two groups. Further investigations with theOPLS-DA model showed a separation along the x-axisbetween CFMþSPFO (left side, Fig. 1e) and CFMþCFO(right side, Fig. 2e), using one predictive and one orthogonalcompound. The model parameters and the P-value of CV-ANOVA are shown in Fig. 1e.
The concentrations of 45 metabolites were furtheranalyzed using ANOVA or Mann–Whitney test. TheNMR-based analysis of the aqueous white muscle samplesshowed lower amounts of formate, glucose and taurine andhigher levels of pyruvate and ADP in CFMþCFO,compared with fish fed CFMþSPFO (Table 5).
The NMR-based spectral data on chloroform extracts ofwhitemuscle were compared by PCA in the formof integrated0.01ppm buckets. The PCA model explained 97.6% of thevariation, with 48% explained by the first component and18.9%bythesecondcomponent.Therewasnocleardifferencebetween the two groups in the PCA score plot. An OPLS-DAmodel using one predictive and one orthogonal componentwas not significant (CV-ANOVA P-value¼ 0.431). In furtherANOVA analysis, no differences in lipid profile were foundbetween CFMþCFO and CFMþSPFO based on metab-olomics analysis of chloroform extracts.
3.4 Gene expression
The gene expression results are presented inFig. 2.Expressionof IGF-I,GHR-I, PPARa, andPPARb1A showed statisticallysignificant up-regulation in liver of fish fed CFMþCFO.However, the hepatic mRNA levels of SREBP-1 and FASshowed significant down-regulation in CFMþCFO, com-pared with the other groups. No significant differences wereobserved in expression of IGF-II and PPARg between thetreatments.
6 K. Cheng et al. Eur. J. Lipid Sci. Technol. 2015, 117, 0000–0000
Figure 1. Score plots of orthogonal partial least squares-discriminant analysis (OPLS-DA) models with significant CV-AVONA P-value,created with the data from 1HNMR spectra profiles of aqueous liver and white muscle extracts in fish fed the diets: control ( ), DFMþSPFO( ), DFMþCFO ( ), CFMþSPFO ( ), CFMþCFO (&). OPLS-DA score plots of aqueous liver extracts in fish fed: (a) control versusCFMþCFO, using one predictive (25.1%) and three orthogonal components (57.2%). The model parameter: R2X¼82.3%, R2Y¼97.1%,Q2¼ 88.5%, R2Y¼ 1, P-value of CV-ANOVA¼ 0.0002. (b) DFMþSPFO versus CFMþCFO, using one predictive (25.4%) and threeorthogonal components (57.1%). The model parameter: R2X¼82.5%, R2Y¼ 93.3%, Q2¼ 74.6%, R2Y¼1, P-value of CV-ANOVA¼0.0122. (c) DFMþCFO versus CFMþCFO, using one predictive (29.0%) and three orthogonal components (57.3%). The modelparameter:R2X¼ 86.3%,R2Y¼ 94.3%,Q2¼77.9%,R2Y¼1,P-value of CV-ANOVA¼0.0062. (d) CFMþSPFO versusCFMþCFO, usingone predictive (32.8%) and three orthogonal components (48.2%). The model parameter: R2X¼81.0%, R2Y¼94.2%, Q2¼69.2%,R2Y¼1, P-value of CV-ANOVA¼0.0307. OPLS-DA score plots of aqueous white muscle extracts in fish fed: (e) CFMþSPFO versusCFMþCFO, using one predictive (13.0%) and one orthogonal component (22.0%). The model parameter: R2X¼ 35.1%, R2Y¼87.5%,Q2¼ 58.9%, R2Y¼1, P-value of CV-ANOVA¼ 0.0102.
Eur. J. Lipid Sci. Technol. 2015, 117, 0000–0000 Evaluation of decontaminated fish feed using metabolomic 7
This study evaluated the use of decontaminated Baltic Sea-sourced fish raw materials in feeds for Arctic char, byassessing the hazard to fish as a result of diet.
4.1 Changes associated with energy metabolism
1H NMR-based metabolomics analyses were used toinvestigate the metabolic response in liver and white muscle.Fish fed CFMþCFO showed higher concentrations of
Table 4. Absolute concentrations of metabolites (mmol/g) in aqueous liver extracts of fish found to be discriminative between treatments inmultivariate� and univariate statistical analysis�� (mean�SEM; n¼10)
Metabolites Control DFMþSPFO DFMþCFO CFMþSPFO CFMþCFO P
�In OPLS-DA models with significant P-values of CV-ANOVA (P<0.05), the important metabolites in VIP-plot (VIP> 1 and VIP-CI> 0)were considered discriminative.��The metabolites significant in Tukey’s test and Bonferroni correction (ab¼0.0011) were considered discriminative.P-value calculated via ANOVA or Mann–Whitney test (Tukey’s test).abcDifferent superscript letters indicate significant differences between the groups (Tukey’s test, P< 0.05).
Figure 2. Relative expression of genes shown as fold change in liver of fish fed a commercial standard diet and four experimental dietsformulated with untreated/decontaminated fish materials from the Baltic Sea (n¼ 6). Different superscript letters denote significantdifferences between dietary treatments (P<0.05).
8 K. Cheng et al. Eur. J. Lipid Sci. Technol. 2015, 117, 0000–0000
alanine, b-alanine, glucose, glycine, isoleucine, leucine,methionine, phenylalanine, tyrosine, and valine in aqueousliver extracts. This might suggest perturbation of proteinbiosynthesis and catabolism and of the energy metabolismpathways leading to the tricarboxylic acid (TCA) cycle [31].These results were in agreement withKokushi et al. [32], thatdietary exposure to environmentally polluted oil high in PAHinduces alterations in metabolites around the TCA cycle incarp plasma, including alanine, glucose, leucine, isoleucine,methionine, tyrosine, and valine. Stressor-induced hyper-glycaemia by mediating effects of catecholamine on glyco-genolysis in fish has also been reported [33]. Thus, the higherlevel of glucose in fish liver and the changes in amino acidsmay indicate the presence of stressors in CFMþCFO, givingrise to the energy imbalance.
In addition to the effects of CFMþCFO on livermetabolism, its influences on muscle metabolism were alsostudied, compared with the effects on muscle samples fromCFMþSPFO, which was selected from the decontaminatedgroups (DFMþSPFO, DFMþCFO, and CFMþSPFO).The control group was not included in muscle metabolomicsanalysis due to the large differences in diet formulationbetween the control and the experimental diets. Significantchanges in metabolites associated with energy metabolismwere also observed in white muscle of CFMþCFO. Areduction in glucose and more than four times higher level ofpyruvate in aqueous white muscle extracts of CFMþCFOindicated a metabolic shift in glycolysis from glucose topyruvate, compared with that of CFMþSPFO. Interest-ingly, it has previously been shown that reduced activity ofoxidative enzymes in the TCA pathway and a dramaticincrease in the activity of glycolytic enzymes can contributeto insulin resistance in humans with non-insulin-dependentdiabetes mellitus [34]. Environmental pollutants have beenreported to modify glucose metabolism through changing the
insulin signalling pathway e.g., by reducing insulin receptorsubstrate-1 levels, causing insulin resistance in mice andhuman studies [1, 15, 35]. Correspondingly, the CFMþCFO feed may have induced changes in insulin signallingpathways in fish, similar to the insulin resistance observed inmammals.
In the present study, we found up-regulation in hepaticgene expression of the main regulators in vertebrate somaticgrowth, IGF-I and GHR-I [36]. IGF-I and insulin possessoverlapping functions in metabolomics regulation, promot-ing glucose uptake and synthesis of glycogen, and IGF-I iseven more effective than insulin in fish [14, 37]. On the otherhand, treatment with growth hormone which is functionedwith GHR has been found to result in increased plasmaglucose levels and metabolic changes in rainbow trout(Oncorhynchusmykiss), similar to the effects inmammals [38].Thus, combined with the changes in the concentrations ofmetabolites, the increased expression of IGF-I and GHR-Imight imply that the regulation of glucose metabolism wasaffected in the CFMþCFO group. Besides, it has beenreported that the gene expression of GH in rainbow troutpituitary glands is up-regulated after exposure to endocrinedisrupters, partly through the estrogen and aryl hydrocarbonreceptors [39].
The higher gene expression of PPARa and PPARb1A inCFMþCFO might also indicate that the existing endocrinedisruptors interfered with lipid and carbohydrate metabo-lism, possibly through modulating heterodimer formationwith the retinoid X receptor or through binding to theperoxisome proliferator response elements [16, 40].
4.2 Changes associated with hepatotoxicity
The higher content of choline detected in liver of fishfed CFMþCFO has been associated with contaminant-stimulated oxidative damage and induced disruption of cellmembranes [31]. Increased levels of choline and phospho-choline and decreased levels of phosphatidylcholines havebeen observed in mice exposed to inorganic arsenic,demonstrating increased degradation of membrane PL[31]. However, according to other studies [41, 42], thereduced hepatic expression of SREBP-1 and FAS inCFMþCFO might result in inhibition of synthesis ofphosphatidylcholine from choline, by regulating phospho-choline cytidylyltransferase (CCT) at the transcriptionallevel and modulating the activity of choline kinase (CHK),thereby increasing the content of choline. In the presentstudy, the signals corresponding to phosphatidylcholine inthe NMR spectrum of chloroform liver extracts and theconcentration of phosphocholine in aqueous liver extractsdid not show differences between groups, possibly due to thelow level of contaminants and short exposure period. Thelipid analysis showed no difference in lipid content in liveror FA composition in the PL fraction. Thus, it limited us toclarify that the increased choline in fish liver of CFMþCFO
Table 5. Absolute concentrations of metabolites (mmol/g) withsignificant variations in multivariate and univariate statisticalanalysis in aqueous white muscle extracts of fish fed CFMþSPFOand CFMþCFO (mean�SEM; n¼ 10)
P-value calculated with ANOVA or Mann–Whitney test (Tukey’stest).�Values significant in Tukey’s test, but not significant afterBonferroni correction (ab< 0.0010).
Eur. J. Lipid Sci. Technol. 2015, 117, 0000–0000 Evaluation of decontaminated fish feed using metabolomic 9
was caused by higher degradation of PL or inhibited synthesisof phosphatidylcholine.
Increased choline has previously been observed in analcoholic fatty liver in zebrafish (Danio rerio) [43], anddecreased levels of hepatic SREBP-1 and FAS mRNAand disturbance of lipid metabolism have been seen inyellow catfish (Pelteobagrus fulvidraco) exposed to waterbornechronic copper [13]. Hence, the higher choline content in theliver of Arctic char fed CFMþCFO is associated withdisturbance of the choline metabolism pathway, possiblycausing changes in PL synthesis and liver damage.
Other metabolite related to hepatic toxicity was alsoaltered in our study. Cholate is a main component of bileacids, important metabolic regulators of glucose and lipidmetabolism. Bile acid feedback repression allows the liver toincrease or decrease bile acid synthesis, and thus maintain aconstant bile acid pool [44]. The change in the content ofcholate caused by the presence of xenobiotics in the dietimplies that the feedback regulation of bile acid, mediatedby farnesoid X receptor and CYP7A1, could have beendisturbed [44].
4.3 Other marked changes associated withxenobiotics
GSH is a co-substrate for glutathione S-transferases, whichare involved in the conjugation and elimination of xeno-biotics [31]. Depletion of GSH can be used as an indicator ofadverse health effects caused by xenobiotics [45]. Similarly,taurine with a similar structure to acetyl-cysteine has beenreported to be involved in protective functions againstpathological changes and in heavy metal detoxificationeffects in fish [46, 47]. Thus, the lower contents of GSHand taurine in fish from CFMþCFO might be associatedwith detoxification, further indicating the hazards of theuntreated diet to fish.
4.4 Contamination status in fish feeds
In the present study, the concentrations of organic pollutants(PCB, PAH, PBDE) in the fishmeal and fish oil weremeasured (Table 1 and Supplementary Table S1). Com-pared with the EU regulation for animal feed [48], the levelsof DL-PCB and NDL-PCB in the experimental diets werelow, around 5% of the limit. The differences in PAH andPBDE content between diets were larger than for PCB levels,so the fish metabolic changes were more likely to be due toPAH and PBDE than to PCB. The organic pollutants PAHand PBDE have been shown to affect fish energy metabolismand induce fish hepatic stress [32, 49]. Nevertheless, theamounts of pollutants measured cannot directly explainthe metabolic differences observed between the untreatedgroup (CFMþCFO) and the decontaminated groups(DFMþSPFO, DFMþCFO, and CFMþSPFO). Thedifferences may possibly be attributable to some unidentified
compounds. such as estrogens [24] and phthalate esters [50].These substances, which are suggested to be environmentallipophilic contaminants, could be removed by defattingand/or activated carbon [24, 50]. Synergistic effects ofdifferent compounds could be another explanation.
In the metabolomics and gene expression studiesdescribed above, toxic effects were only observed when bothsources of contaminants (CFM and CFO) were available.None of the diet sources alone can explain the metabolicchanges observed. Therefore, we deduced that when crudefishmeal and crude fish oil sources were used, there wasenough pollutant or pollutants present, alone or in synergywith others, to cause the metabolic effects observed.
4.5 Fish nutritional values
The FA composition in flesh varied in accordance with thecomposition of the diet. In comparison with the experimentaldiet, the commercial diet contained more plant-sourcedmaterials, thus having a higher level of
Pn-6 and lower
level ofP
n-3 in feed and fish flesh (Tables 2 and 3). Therewas no important difference in
Pn-3 composition between
the experimental diets except that EPA was lower inCFMþSPFO. These findings were in agreement withSprague study [9] in which crude and decontaminatednorthern fish oil were fed to Atlantic salmon for 11 weeks,and no differences were seen in flesh EPA and DHA content.Thus after decontamination, Baltic Sea-sourced fish materi-als with high
Pn-3 would be valuable sources of fish feed.
5 Conclusions
Arctic char fed a diet with both CFM and CFO showeddisturbances in their metabolic profile and gene expressionassociated with energy metabolism and hepatic toxicology.These results indicate that compared with the untreateddiets, the use of decontaminated fish materials from theBaltic Sea as feed could reduce the impact on fishmetabolismin the short-term. We found no significant differencesbetween the decontaminated groups (DFMþSPFO, DFMþCFP, andCFMþSPFO) and the control group in terms ofmetabolomics or gene expression. However, further studieswith higher numbers of animals and longer periods ofexposure are needed to confirm the possible hazard to fish,since the existence of a pollutant accumulation thresholdmay induce metabolic disturbance in fish. 1H NMR-basedmetabolomics proved to be a powerful tool for monitoringthe response to pollutants, even in this short-term study witha low level of dietary contaminants.
According to the World Wildlife Fund (WWF), severalother waters worldwide, such as the Gulf of Mexico, YangtzeRiver inChina,MekongDelta in Vietnam, theGreat Lakes inUS, etc., are facing the same pollution issues as the Baltic Seadue to land-based human activities. Recycling polluted fish
10 K. Cheng et al. Eur. J. Lipid Sci. Technol. 2015, 117, 0000–0000
raw materials after decontamination into the food chain asfish feed could be a practical and efficient future solution tothese issues.
This study was funded by Innovative Practices and Technologiesfor Developing Sustainable Aquaculture in the Baltic Sea Region,AQUABEST of the Baltic Sea Region Programme 2007–2013and CENAKVA CZ.1.05/2.1.00/01.0024. The results of theproject LO1205 were obtained with financial support fromMEYSunder the NPU I programme. The authors would also like to thankthe China Scholarship Council (CSC) for its financial support.
The authors declare no conflict of interest.
References
[1] Ibrahim, M. M., Fjære, E., Lock, E. -J., Naville, D., et al.,Chronic consumption of farmed salmon containing persis-tent organic pollutants causes insulin resistance and obesityin mice. PLoS ONE 2011, 6, 25170.
[2] Allsopp, M., Erry, B., Santillo, D., Johnston, P., POPs in theBaltic: a review of persistent organic pollutants (POPs) in theBaltic Sea. Greenpeace International 2001, ISBN: 90-73361-71-0, 2015-09-12. http://www.greenpeace.org/international/en/publications/reports/pops-in-the-baltic-a-review/
[3] Tocher, D. R., Issues surrounding fish as a source of omega-3 long-chain polyunsaturated fatty acids. Lipid Technol.2009, 21, 13–16.
[4] Simopoulos, A. P., n�3 fatty acids and human health:Defining strategies for public policy. Lipids 2001, 36, S83.
[5] FAO, The state of world fisheries and aquaculture: Oppor-tunities and challenges. Fisheries and Aquaculture TechnicalPaper 2014, http://www.fao.org/3/a-i3720e/index.html
[6] Tacon, A. G. J., Metian, M., Turchini, G. M., De Silva,S. S., Responsible aquaculture and trophic level implicationsto global fish supply. Rev. Fish Sci. 2009, 18, 94–105.
[7] Bell, J. G., McEvoy, J., Tocher, D. R., McGhee, F., et al.,Replacement of fish oil with rapeseed oil in diets of Atlanticsalmon (Salmo salar) affects tissue lipid compositions andhepatocyte fatty acid metabolism. J. Nutr. 2001, 131,1535–1543.
[8] Pratoomyot, J., Bendiksen, E. A., Bell, J. G., Tocher, D. R.,Comparison of effects of vegetable oils blended withsouthern hemisphere fish oil and decontaminated northernhemisphere fish oil on growth performance, composition andgene expression in Atlantic salmon (Salmo salar L.).Aquaculture 2008, 280, 170–178.
[9] Sprague, M., Bendiksen, E. A., Dick, J. R., Strachan, F.,et al., Effects of decontaminated fish oil or a fish andvegetable oil blend on persistent organic pollutant and fattyacid compositions in diet and flesh of Atlantic salmon (Salmosalar). Br. J. Nutr. 2010, 103, 1442–1451.
[10] Olli, J. J., Breivik, H., Mørkøre, T., Ruyter, B., et al.,Removal of persistent organic pollutants from Atlanticsalmon (Salmo salar L.) diets: Influence on growth, feedutilization efficiency and product quality. Aquaculture 2010,310, 145–155.
[11] Lenz, E. M., Wilson, I. D., Analytical strategies inmetabonomics. J. Proteome Res. 2007, 6, 443–458.
[12] Zhang, Y., Deng, Y., Zhao, Y., Ren, H., Using combinedbio-omics methods to evaluate the complicated toxic effectsof mixed chemical wastewater and its treated effluent. J.Hazard Mater. 2014, 272, 52–58.
[13] Chen, Q. -L., Luo, Z., Pan, Y. -X., Zheng, J. -L., et al.,Differential induction of enzymes and genes involved in lipidmetabolism in liver and visceral adipose tissue of juvenileyellow catfish Pelteobagrus fulvidraco exposed to copper.Aquat. Toxicol. 2013, 136–137, 72–78.
[14] Castillo, J., Codina, M., Martinez, M. L., Navarro, I.,Gutierrez, J., Metabolic and mitogenic effects of IGF-I andinsulin onmuscle cells of rainbow trout.Am. J. Physiol Regul.Integr. Comp. Physiol. 2004, 286, 935–941.
[15] Ruzzin, J., Petersen, R., Meugnier, E., Madsen, L., et al.,Persistent organic pollutant exposure leads to insulinresistance syndrome. Environ. Health Perspect. 2010, 118,465–471.
[16] Casals-Casas, C., Feige, J. N., Desvergne, B., Interference ofpollutants with PPARs: Endocrine disruption meets metab-olism. Int. J. Obes. 2008, 32, S53.
[17] Klemetsen, A., Amundsen, P.-A., Dempson, J. B., Jonsson,B., et al., Atlantic salmon Salmo salar L., brown trout Salmotrutta L. and Arctic char Salvelinus alpinus L.: A review ofaspects of their life histories. Ecol. Freshw. Fish. 2003, 12,1–59.
[18] Tacon, A. G. J., State of information on salmon aquaculturefeed and the environment. Rep. WWF Salmon AquacultureDialogue 2005, 1–80.
[19] Mr�az, J., Pickova, J., Differences between lipid content andcomposition of different parts of fillets from crossbredfarmed carp (Cyprinus carpio).Fish Physiol Biochem. 2009, 35,615–623.
[20] Trattner, S., Kamal-Eldin, A., Br€ann€as, E., Moazzami, A.,et al., Sesamin supplementation increases white muscledocosahexaenoic acid (DHA) levels in rainbow trout(Oncorhynchus mykiss) fed high alpha-linolenic acid (ALA)containing vegetable oil: Metabolic actions. Lipids 2008, 43,989–997.
[21] Wagner, L., Trattner, S., Pickova, J., G�omez-Requeni, P.,Moazzami, A. A., 1H NMR-based metabolomics studies onthe effect of sesamin in Atlantic salmon (Salmo salar). FoodChem. 2014, 147, 98–105.
[22] Moazzami, A. A., Bondia-Pons, I., Hanhineva, K., Juntu-nen, K., et al., Metabolomics reveals the metabolic shiftsfollowing an intervention with rye bread in postmenopausalwomen-a randomized control trial. Nutr. J. 2012, 11, 88.
[23] Bankefors, J., Kaszowska, M., Schlechtriem, C., Pickova, J.,et al., A comparison of the metabolic profile on intact tissueand extracts of muscle and liver of juvenile Atlantic salmon(Salmo salar L.) -Application to a short feeding study. FoodChem. 2011, 129, 1397–1405.
[24] Samuelsson, L. M., F€orlin, L., Karlsson, G., Adolfsson-Erici, M., Larsson, D. G. J., Using NMR metabolomics toidentify responses of an environmental estrogen in bloodplasma of fish. Aquat. Toxicol. 2006, 78, 341–349.
[25] Kullgren, A., Samuelsson, L. M., Larsson, D. G. J.,Bj€ornsson, B. T., Bergman, E. J., A metabolomics approachto elucidate effects of food deprivation in juvenile rainbowtrout (Oncorhynchus mykiss). Am. J. Physiol. Integr. Comp.Physiol. 2010, 299, R1440–R1448.
[26] G�omez-Requeni, P., Calduch-Giner, J., Vega-Rubin deCelis, S., Medale, F., et al., Regulation of the somatotropic
Eur. J. Lipid Sci. Technol. 2015, 117, 0000–0000 Evaluation of decontaminated fish feed using metabolomic 11
axis by dietary factors in rainbow trout (Oncorhynchusmykiss). Br. J. Nutr. 2005, 94, 353–361.
[27] Schiller Vestergren, A., Wagner, L., Pickova, J., Rosenlund,G., et al., Sesamin modulates gene expression withoutcorresponding effects on fatty acids in Atlantic salmon(Salmo salar). Lipids 2012, 47, 897–911.
[28] Skiba-Cassy, S., Lansard, M., Panserat, S., M�edale, F.,Rainbow trout genetically selected for greater muscle fatcontent display increased activation of liver TOR signalingand lipogenic gene expression. Am. J. Physiol. Regul. Integr.Comp. Physiol. 2009, 297, R1421–R1429.
[29] Plagnes-Juan, E., Lansard, M., Seiliez, I., M�edale, F., et al.,Insulin regulates the expression of several metabolism-relatedgenes in the liver and primary hepatocytes of rainbow trout(Oncorhynchus mykiss). J. Exp. Biol. 2008, 211, 2510–2518.
[30] Olsvik, P. A., Lie, K. K., Jordal, A.-E. O., Nilsen, T. O.,Hordvik, I., Evaluation of potential reference genes in real-time RT-PCR studies of Atlantic salmon. BMC Mol. Biol.2005, 6, 21.
[31] Garc�ıa-Sevillano, M. A., Contreras-Acu~na, M., Garc�ıa-Barrera, T., Navarro, F., G�omez-Ariza, J. L., Metabolomicstudy in plasma, liver and kidney of mice exposed toinorganic arsenic based onmass spectrometry.Anal. Bioanal.Chem. 2014, 406, 1455–1469.
[32] Kokushi, E., Uno, S., Harada, T., Koyama, J., 1H NMR-based metabolomics approach to assess toxicity of bunker aheavy oil to freshwater carp. Cyprinus Carpio. Environ.Toxicol. 2010, 27, 404–414.
[33] Bonga, S. E. W., The stress response in fish. Physiol. Rev.1997, 77, 591–625.
[34] Simoneau, J. -A., Kelley, D. -E., Altered glycolytic andoxidative capacities of skeletal muscle contribute to insulinresistance in NIDDM. J. Appl. Physiol. 1997, 83, 166–171.
[35] Sargis, R. M., Neel, B. A., Brock, C. O., Lin, Y., et al., Thenovel endocrine disruptor tolylfluanid impairs insulin signal-ing in primary rodent and human adipocytes through areduction in insulin receptor substrate-1 levels. Biochim.Biophys. Acta. 2012, 1822, 952–960.
[36] Reinecke, M., Bj€ornsson, B. T., Dickhoff, W. W., McCor-mick, S. D., et al., Growth hormone and insulin-like growthfactors in fish: Where we are and where to go. Gen. Comp.Endocrinol. 2005, 142, 20–24.
[37] Caruso, M. A., Sheridan, M. A., New insights into thesignaling system and function of insulin in fish. Gen. Comp.Endocrinol. 2011, 173, 227–247.
[38] Sangiao-Alvarellos, S., M�ıguez, J.M., Soengas, J. L., Actionsof growth hormone on carbohydrate metabolism andosmoregulation of rainbow trout (Oncorhynchus mykiss).Gen. Comp. Endocrinol. 2005, 141, 214–225.
[39] Elango, A., Shepherd, B., Chen, T. T., Effects of endocrinedisrupters on the expression of growth hormone andprolactin mRNA in the rainbow trout pituitary. Gen. Comp.Endocrinol. 2006, 145, 116–127.
[40] Kersten, S., Desvergne, B., Wahli, W., Roles of PPARs inhealth and disease. Nature 2000, 405, 421–424.
[41] Glunde, K., Bhujwalla, Z. M., Ronen, S. M., Cholinemetabolism in malignant transformation. Nat. Rev. Cancer2011, 11, 835–848.
[42] Ridgway, N. D., Lagace, T. A., Regulation of the CDP-choline pathway by sterol regulatory element bindingproteins involves transcriptional and post-transcriptionalmechanisms. Biochem. J. 2003, 372, 811–819.
[43] Jang, Z. -H., Chung,H. -C., Ahn, Y.G., Kwon, Y. -K., et al.,Metabolic profiling of an alcoholic fatty liver in zebrafish(Danio rerio). Mol. Biosyst. 2012, 8, 2001–2009.
[44] Li, T., Chiang, J. Y. L., Bile Acid signaling in livermetabolism and diseases. J. Lipids 2012, 1–9.
[45] Valavanidis, A., Vlahogianni, T., Dassenakis, M., Scoullos,M., Molecular biomarkers of oxidative stress in aquaticorganisms in relation to toxic environmental pollutants.Ecotoxicol. Environ. Saf. 2006, 64, 178–189.
[46] Timbrell, J. A., Seabra, V., Waterfield, C. J., The in vivo andin vitro protective properties of taurine.Gen. Pharmac. 1995,26, 453–462.
[47] Choi, K. -S., Yoo, I. -S., Shin, K. -O., Chung, K. -H., Effectsof taurine on cadmium exposure in muscle, gill, and bonetissues of Carassius auratus.Nutr. Res. Pract. 2013, 7, 22–25.
[48] European Commission, Commission regulation (EU) No.277/2012 of 28 March 2012 amending Annexes I and II toDirective 2002/32/EC of the European Parliament and of theCouncil as regards maximum levels and action thresholds fordioxins and polychlorinated biphenyls. 2012.
[49] Rochman, C. M., Hoh, E., Kurobe, T., Teh, S. J., Ingestedplastic transfers hazardous chemicals to fish and induceshepatic stress. Sci. Rep. 2013, 3, 3263.
[50] Mohan, S. V., Shailaja, S., Krishna, M. R., Sarma, P. N.,Adsorptive removal of phthalate ester (Di-ethyl phthalate)from aqueous phase by activated carbon: A kinetic study. J.Hazard. Mater. 2007, 146, 278–282.
12 K. Cheng et al. Eur. J. Lipid Sci. Technol. 2015, 117, 0000–0000
