Effects of the total replacement of fish-based diet with plant-based diet on the hepatic transcriptome of two European sea bass (Dicentrarchus labrax) half-sibfamilies showing different growth rates with the plant-based diet

Diets and fish

Daily growth coefficient (% day): DGC = 100 × (final individual weight (g) ^1/3 - initial individual weight (g) ^1/3)/days (According to Cho [29]).

Feed efficiency (FE): [final weight (g) - initial weight (g)]/feed ration (g) (According to Carter et al. [30])

At the end of the growth trial, liver, muscle and plasma were sampled from 15 fishes per half-sibfamily and per dietary treatment. Muscular LC-PUFA profiles and real-time PCR investigations were performed on these 15 sampled individuals per group. Microarray analysis of hepatic RNA and investigation of immune parameters in plasma were investigated on 5 to 8 sampled individuals per group in order to respect similar sex ratio, sampling time and RNA quality (RNA Integrity Number as determined by Bioanalyser (Agilent) > 9).

Lipid extraction and fatty acid analysis

One gram of white muscle was dissected from 15 fish per dietary treatment and immediately frozen in liquid nitrogen. Total lipids were extracted according to the method of Folch et al. [31], by Accelerated Solvent Extraction 200 (ASE, Dionex) with dichloromethane/methanol (2:1) containing 0.01% butylatedhydroxutoluene (BHT) as antioxidant. Lipids were extracted at 100 bars, 100°C, with a 5 min precallingphase, 2 min static phase, and 60% flush for 60 sec (3 cycles). The separation of neutral and polar lipids was performed according to the procedure described by Juaneda and Roquelin [32]. The total lipid (TL) extracts were fractionated on silica cartridges (Sep-Pack, Waters^®), neutral lipids (NL) were eluted with chloroform and polar lipids (PL) with methanol. Fatty acid methyl esters (FAME) were prepared from total lipids by acid-catalysed trans-esterification. FAME were quantified by gas-liquid chromatography (Clarus 500, Perkin Elmer) with a BPX70 column of 30 m length and 0.22 mm I.D. Hydrogen was used as carrier gas and temperature programming was from 50°C to 180°C at 20°C/min and then to 220°C at 3°C/min. Individual methyl esters were identified by comparison with known standards. The fatty acid analysis was performed on one sample per fish.

RNA extraction and real-time quantitative PCR analysis

Total mRNA of liver was extracted using Trizol reagent (Invitrogen, USA) and quantified by measuring absorbance at 260 nm in a spectrophotometer (Nanodrop Labtech, France). The reverse transcription was performed using the QuantiTect^® Reverse Transcription kit (QIAGEN), including a genomic DNA elimination reaction. Reactions were carried out in a volume of 20 μl, containing 1 μg of total RNA, 1 unit of Quantiscript Reverse Transcriptase, 4 μl of Quantiscript RT buffer (5 ×) and 1 μM primer mix.

Seven genes involved in metabolic and/or immune pathways of interest (fads2, hmgcr, fabp7, angptl3, cxcl10, gck and lpl), whose the oligonucleotides were spotted on the chip, were analysed by real-time PCR in order to validate the gene expression patterns obtained through the microarray approach. The relative mRNA levels were automatically normalized with housekeeping elongation factor 1 (ef1) gene expression and measured by Bio-Rad IQ5 software using ΔΔCt method (Ct for Cycle threshold):

Gene Normalized Expression in sample 1 = (Gene Relative Quantity in sample 1)/(EF1 Relative Quantity in sample 1) with Relative Quantity in sample 1 for a gene i = E_{(gene i)}^{Ct (sample 2) - Ct (sample 1)}

Ef1 was chosen to provide an internal control for real-time PCR, since contrary to 18S rRNA and actin initially tested, we did not observe any significant difference (t-student test, p > 0. 1) between Ct values for Ef1 between the dietary groups (See additional file 1: Comparison of Ct values for ef1 gene between the dietary groups). Its stability was also assessed by a low coefficient of variation over all the samples (CV < 5%).

Plasma lysozyme concentration

Plasma lysozyme activity was determined at ambient temperature using a turbidimetric assay [33], adapted to microtitration plates. Briefly, a bacterial suspension of Micrococcus lysodeikticus (Sigma) was prepared at a concentration of 1.25 g.l^-1 in a 0.05 M sodium-phosphate buffer, pH 6.2. Fifty microlitres of the samples were placed in 96-well microtitration plates. The reaction was initiated in a Labsystems-iEMS analyser, by addition of 160 μl well^-1M. Lysodeikticus suspension using an automatic dispenser. The optic density (OD) reading was taken at a wavelength of 450 nm every 15 s for 3 min, the plate being shaken before each reading. Lysozyme values for samples were converted to mg.ml^-1, using a reference curve established with hen egg white lysozyme (Sigma).

Plasma alternative complement pathway activity

Determination of the alternative pathway of plasma complement activity was carried out at 4°C through a haemolytic assay with rabbitred blood cells (RRC, Biomérieux, Craponne, France) as described by Yano [34], adapted to microtitration plates. Sea bass samples, diluted to 1/64 in EGTA-Mg-GVB buffer to avoid natural haemolytic activity, were added in increasing amounts, from 10 to 100 μl well^-1. Wells were filled with EGTA-Mg-GVB buffer to a final volume of 100 μl. Finally, 50 μl of 2% RRC (Biomérieux) suspension was added to each of the wells. Control values of 0% and 100% haemolysis were obtained using 100 μl of EGTA-Mg-GVB buffer and 100 μl of non-decomplemented trout haemolytic serum at 1/50 in ultrapure water, respectively. The samples were then incubated for 1 h at 20°C. The microplates were centrifuged (400 g, 5 min, 4°C) and 75 μl of supernatant from each well was then transferred into another 96-well microplate with 75 μl of phosphate buffered saline (PBS, Biomérieux). The absorbance (A₅₄₀) was read in a Labsystems-iEMS analyser and the number of ACH₅₀ units per ml of plasma was determined by reference to the 50% haemolysis.

Dicentrarchus labrax oligonucleotide microarray

Gene expression profiles were investigated using the Agilent-019810 Dicentrarchus labrax oligo microarray (GEO accession: GPL9663). This platform represents 19, 035 unique transcripts of the European sea bass. Two non-overlapping probes were designed for each transcript for a total 38, 070 oligonucleotide probes (60 mers) synthesized onto the array (for details see [35]). All sequences (DLPD) are publicly available in a dedicated database [35], together with associated annotations, GO entries and putative homologous genes in fish model species.

Microarrays were synthesized in situ using Agilent non-contact ink-jet technology with a 4 × 44 K format, and included default positive and negative controls. Microarray analysis was based on a single color (Cy3) design. A mixture of 10 different viral polyadenilated RNAs (Agilent Spike-In Mix) was also added to each RNA sample to monitor labelling and hybridization quality as well as microarray analysis work-flow.

Sample labelling and hybridization were performed as reported in Ferraresso et al. [28]. Briefly, for each sample, 200 ng of total RNA were linearly amplified and labelled with Cy3-dCTP according to the Agilent One-Color Microarray-Based Gene Expression Analysis protocol. A mixture of 10 different viral polyadenilated RNAs (Agilent Spike-In Mix) was also added to each RNA sample to monitor labelling and hybridization quality as well as microarray analysis work-flow. After fragmentation, a total of 1, 650 ng of labelled cRNA were dispensed in the gasket slide and assembled to the microarray slide (each slide containing four arrays). Slides were incubated for 17 h at 65°C in an Agilent Hybridization Oven and washed following manufacturer's instructions.

Data acquisition and normalization

Hybridized slides were scanned at 5 μm resolution using an Agilent G2565BA DNA microarray scanner. Default settings were modified to scan the same slide twice at two different sensitivity levels (XDR Hi 100% and XDR Lo 10%). The two linked images generated were analyzed together; data were extracted and background subtracted using the standard procedures contained in the Agilent Feature Extraction (FE) Software version 9.5.1. Spike-in probe intensities were used to assess the performance of the normalization procedure for each dataset. Data normalization was performed using R statistical software http://www.r-project.org; microarray data were normalized across all arrays using the cyclic loess approach [36]. Fold changes (FC) were calculated for each gene by finding the average value for each group (dietary or sibfamily group). Raw and normalized fluorescence data of the present microarray experiment have been deposited in the GEO database under accession number (under submission: NCBI tracking system #16023742).

Statistical analysis

All the results are presented as mean values with standard deviations (SD). Daily Growth Coefficient (DGC) was studied using a model accounting for diet as a fixed effect and tank, sire, dam, sire*diet and dam*diet as random effects, using SAS-GLM. Effects of diet and half-sibfamily factors on biometry (hepatosomatic index: HSI and viscerosomatic index: VSI), fatty acid composition, gene expression (qPCR), plasma lysozyme concentration and alternative complement pathway activity were tested by two-way ANOVAs (P < 0.05) using Statistica biosoft 8.0. The microarray data were analysed by two-way ANOVA using Tmev (Tigr MultiExperiment Viewer) statistical software, and gene expression was considered significantly different when P < 0.01. Significant enrichment of GO biological process categories were tested for using EASE software (version 2.0) with P < 0.05 [35].

Growth and biometry

Lipid and fatty acid compositions of the fillet

The flesh lipid composition was significantly affected by dietary treatment (See additional file 2: Fatty acid composition in muscle). Feeding VD significantly increased the percentage of saturated lipids in both the neutral lipids and phospholipids. The α-linolenic acid (18:3n-3) and linoleic acid (18:2n-6) contents were respectively 10-fold and 3-fold higher, in both lipid classes, when fish were fed VD. In addition, AA, EPA and DHA contents were around 2.5-fold lower in flesh of fish fed VD. The Σn-3 PUFA/Σn-6 PUFA ratio decreased in both the neutral lipid (2-fold) and phospholipid (4-fold) fractions in the flesh of European sea bass fed VD.

Microarray gene expression profiling

In contrast, the main biological processes enriched associated with the genes that were lower-expressed in fish fed VD were particularly related to cellular process (136 genes), cell communication (65 genes) and cell proliferation (32 genes) (See additional file 3: Significantly enriched biological processes associated with genes regulated by diet and half-sibfamily factors). In addition, some genes involved in the immune function were less expressed in fish fed VD (Table 5).

Immune parameters

Lysozyme activity was significantly lower (0.7-fold; P < 0.05) in fish fed VD than in fish fed FD, while the alternative complement activity was 1.5-times higher (P < 0.05) (Figure 1). There was no effect of the half-sibfamily factor on these activities.

LC-PUFA metabolism

Metabolism-related biological processes constitute the largest group among the GO terms associated with the genes regulated in VD-fed fish. Among these, genes involved in lipid metabolism and, particularly, in LC-PUFA biosynthesis, were found to be up-regulated in fish fed VD. Not surprisingly, delta-6-desaturase (fads2), steroyl-CoA desaturase 9 (scd9) and NADH-cytochrome b5 reductase (cyb5r2), involved in long chain fatty acid desaturation and/or elongation, were up-regulated in VD-fed fish. It is indeed established that in most fish species that the LC-PUFA biosynthetic pathway is positively regulated in response to the use of a diet poor in LC-PUFA but rich in PUFA, although this regulation depends on fish species, degree of fish oil substitution, nature of the vegetable oil, and environmental parameters (e.g., salinity) [41].

The positive regulation of the LC-PUFA biosynthetic pathway is in agreement with results obtained by transcriptomic approaches in salmonids fed vegetable oil [39, 40]. Altogether, the expression data obtained in marine fish species and salmonids indicate that the different capacity of marine and fresh water species to grow on a LC-PUFA-deprived diet does not seem to be due to a different transcriptional regulation of key genes involved in lipid synthesis, such as fads2 or scd9. Indeed, the level of induction of fads2 expression found in this experiment (5-fold induction) is of similar amplitude to that observed in the liver and intestine of Atlantic salmon [42].

The stimulation of this biosynthetic pathway in fish fed a diet poor in LC-PUFAs can be explained by the fact that LC-PUFAs play several key physiological roles in vertebrates, particularly in fish. For example, fish are poikilothermic organisms and therefore need a high degree of unsaturation of LC-PUFA included in membrane phospholipids to maintain phospholipid bilayer fluidity at reduced temperature [43]. LC-PUFA, especially arachidonic acid (ARA) and eicosapentaenoic acid (EPA), are also precursors of eicosanoids, which are involved in pro and anti-inflammatory pathways. This hypothesis is reinforced by our data, indicating a stimulation of genes involved in phospholipids biosynthesis (agpat3, pisd and pigf) when fish were fed the VD.

Despite this stimulation of LC-PUFA and the phospholipid biosynthesis pathway at the transcriptional level, our investigation of fatty acid profiles indicated that the amounts of LC-PUFA, particularly eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), were still considerably lower in the flesh of fish from both half-sibfamilies fed VD in comparison with fish fed FD. This finding is in agreement with those previously obtained [13], which revealed that the stimulation of fads2 expression in fish fed a vegetable diet was not associated with an induction of its enzymatic activity, suggesting a post-transcriptional regulation of fads2 expression. Such EPA and DHA deficiency can notably explain the growth deficiency observed in fish fed VD, as well as effects observed on immune function (discussed below).

Concerning the genetic aspect, the comparable overall expression pattern of genes involved in LC-PUFA synthesis in the liver, associated with similar LC-PUFA profile in muscle, in both half-sibfamilies suggests that the differing capacities of these European sea bass half-sibfamilies to grow on a vegetable diet are not due to differing capacities to synthesize LC-PUFA.

Sterol metabolism

The present microarray data also indicate an increase in expression levels of genes involved in sterol metabolism in VD-fed fish. Among these, isopentenyl-diphosphate delta-isomerase 1 (idi1), lanosterol 14-alpha demethylase (cyp51a1), farnesyl pyrophosphate synthase (fdps), c-4 methylsterol oxidase (sc4mol) and 3-hydroxy-3-methylglutaryl-coenzyme A reductase (hmgcr) genes [44–49] are known to be implicated in the cholesterol metabolic pathway (Figure 2). More particularly, HMGCR, a transmembrane glycoprotein involved in the rate-limiting step of sterol biosynthesis, is increased, as shown in mammals [50]. The stimulation of cholesterol biosynthesis in fish fed VD could be related to the difference in sterol composition between diets. Indeed, while the fish diet is rich in cholesterol, the vegetable diet used in this experiment contains exclusively plant sterols, which have been shown to affect membrane properties by decreasing permeability and fluidity, and modifying phospholipid order in mammals [51, 52]. As a consequence, the increase in cholesterol biosynthesis could be a metabolic response to its deficiency in the diet, as well as a way to restore membrane properties by incorporation of endogenous cholesterol. Since we did not measure the cholesterol content in the liver, flesh or blood, it is not possible for us to assess the capacity of European sea bass to compensate for possible dietary deficiencies in cholesterol through a regulation of its biosynthesis. Moreover, similarly to the LC-PUFA pathway, no significant difference of cholesterol biosynthetic regulation was observed between the half-sibfamilies.

Interestingly, several VD-stimulated genes (fads2, hmgcr, idi1, cyp51a1, fdps and sc4mol) involved in the lipogenic pathway (LC-PUFA and sterol pathways) are known to be molecular targets of sterol regulatory element binding proteins (SREBPs), which are key regulators of fatty acid and cholesterol synthesis [53, 54]. Recent data indicating an up-regulation of the srebp-1 gene expression in European sea bass fed a vegetable diet [13] could thus be due to such stimulations.

Lipid and sterol transport

The present microarray data indicate that the stimulation of genes involved in fatty acid and cholesterol synthesis in VD-fed fish was associated with an over-expression of genes involved in their transport, such as apolipoproteins APOA1 and APOB100, which are the major protein constituents of high and low density lipoprotein (HDL, LDL), respectively (Figure 2). The LDL, including APOB100, are involved in the transport of cholesterol and lipids from the liver to other tissues. Thus, up-regulation of apob100 combined with the induction of the expression of lipoprotein lipase (lpl), a key enzyme involved in the hydrolysis of triglyceride, suggests an increase in lipid transport and metabolism from the liver to tissues in fish fed VD. The decrease in angiopoietin-related protein 3 (angptl3) that we observed in fish fed VD reinforces this idea since ANGPTL3 suppresses LDL clearance via the inhibition of LPL activity [55]. In parallel, we observed that the reverse transport of cholesterol not used by tissues via HDL to the liver was also stimulated in fish fed VD. Indeed, APOA1, which participates in the transport of cholesterol to the liver by promoting cholesterol efflux from tissues and by acting as a cofactor for the lecithin cholesterol acyltransferase (lcat), exhibited higher transcript levels in fish fed VD. Altogether these results reveal that another major response in the liver of European sea bass fed a vegetable diet is the stimulation of cholesterol synthesis and transport, irrespective of the half-sibfamily considered.

Carbohydrate metabolism

LC-PUFA and cholesterol biosynthesis require reducing power in the form of NADPH. It is well documented in vertebrates, including fish, that NADPH required for malonyl-CoA synthesis is mainly supplied by the dehydrogenases of the pentose phosphate shunt [39, 56, 57]. Interestingly, our transcriptomic data indicate that the use of the VD induced a significant increase in the level of glucose-6-phosphate dehydrogenase (g6pd transcripts). G6PD catalyses NADP+-linked oxidation of D-glucose-6-phosphate and has been shown to be a major contributor of NADPH production for lipogenesis in Atlantic salmon (Salmo salar) [58] and European sea bass [58]. Moreover, our data indicate an increase in the expression of hexose-6-phosphate dehydrogenase (h6pdh) and phosphogluconate dehydrogenase (pgd), enzymes of the pentose phosphate pathway that generate NADPH, in fish fed VD [59, 60] (Figure 2). Once synthesized, the resulting pentose sugar intermediate generated by the pentose phosphate pathway can be reconverted to intermediates of the glycolysis/gluconeogenesis pathway such as glyceraldehyde 3P or fructose 6P. In the liver of fish, it is known that glycolysis provides essential precursors for biosynthesis rather than pyruvate for oxidation [61]. Thus, the stimulation of fructose-1, 6-bisphosphatase 1 (fbp1) and aldolase (aldoa and aldob) expression that we observe in fish fed VD could provide high levels of fructose-6-phosphate from glyceraldehyde 3P, then glucose 6P that serves as substrate for repeated passage in the pentose phosphate shunt (Figure 2).

Protein/amino-acid metabolism and ATP synthesis

Our data revealed over-expression of genes involved in proteolysis and, more particularly, in proteasome activity (psma4, psmb7, psmc2 and psmd4) and ubiquitin activity (ubap1, ube2a, ube2g1 and ube2n) in fish fed VD (Figure 2), which is in total agreement with proteomic data obtained in rainbow trout, indicating a stimulation of proteolysis in fish fed vegetable diets [8]. In our study, the stimulation of proteolysis in the fish fed the vegetable diet was associated with the induction of 18 genes involved in amino acid metabolism and, more importantly, 4 genes involved in glutamine metabolism. In addition, gmps, aadat, got1 and tat genes, which are implicated in transamination, were also stimulated in fish fed VD. The processes related to amino acid metabolism and, especially transamination, are important steps in the synthesis of some non-essential amino acids such as α-ketoglutarate. For example, the synthesis of α-ketoglutarate through transamination reactions could be used in the TCA cycle to provide energy. Interestingly, we found over-expression of genes encoding enzymes involved in the TCA cycle, such as succinate dehydrogenase (sdhb), isocitrate dehydrogenase (idh3g) and malate dehydrogenase (mdh), in fish fed VD. This stimulation of the TCA cycle could be related to the higher levels of ATP required for LC-PUFA and cholesterol biosynthesis in fish fed VD. Since marine fish have a low capacity to digest complex carbohydrates, in contrast to mammals [62, 63], the use of proteins as an essential source of energy can thus explain the stimulation of the amino acid metabolism in fish fed VD. As shown in rainbow trout fed on a vegetable-based diet [64], the lower growth rate in fish fed VD in the present study could be associated with higher proteolytic activity compared with fish fed FD.

Interestingly, while both half-sibfamilies G and g exhibited similar proteolysis regulation, the expression of several genes involved in macromolecule biosynthesis, and particularly in protein biosynthesis (cct2, eif4g1, kiaa2018, mrpl22, mrpl27, mrpl30, mrpl34, mrpl48, mrps14, mrps17, rars, rpl18, rplp1 and rps18), were up-regulated in half-sibfamily G. This result, suggesting a higher protein turnover in half-sibfamily G compared with half-sibfamily g when fish were fed VD, could be related to the higher growth rate observed in half-sibfamily G fed VD. As protein biosynthesis requires energy from ATP hydrolysis, the higher protein biosynthesis in half-sibfamily could be related to a higher activity of mitochondrial ATP production. Accordingly, genes involved in ATP biosynthesis (atp5c1, atp5i and atp5j2) and ATP synthesis-coupled electron transport (ndufb4, ndufb6, ndufs4, ndufs6, ndufv2 and uqcrb) were found up-regulated in half-sibfamily G. The diet × half-sibfamily interaction that we found for the expression of genes involved in aromatic amino acid metabolism reinforces the difference in protein metabolism between the two half-sibfamilies.

Immune function

In the present study, the vegetable diet used, based on linseed oil, was characterised by a very low ARA content and poor levels of n-3 LC-PUFA. Eicosanoids derived from ARA are known to be involved in the proliferation of hepatocytes and immune cells [65–67]. As a consequence, the lower hepatosomatic index (HSI) measured in fish fed VD could be linked to a lower hepatocyte proliferation due to a deficiency in ARA, which was supported by down-expression of 32 genes involved in cell proliferation in this dietary group. Moreover, fatty acid imbalances can induce an immune deficiency in all vertebrates including fish [14, 15, 22]. In particular, the Σ n-3/Σ n-6 fatty acid ratio is considered as a key element regulating immune cell structure, cell signalling, and eicosanoid production.

The present microarray data revealed genes of the immune system, particularly the innate immune response, exhibiting lower expression in fish fed with VD. Interleukin 8 (il8) and C-X-C motif chemokine 10 (cxcl10), which are chemotactic factors for granulocytes and monocytes, respectively, were found to be less expressed in fish fed VD, suggesting a down regulation of the innate defence system and the pro-inflammatory pathway in this group. Deficiency of the inflammatory response was also in agreement with the higher levels of transcripts of fatty acid binding protein 7 (fabp7), whose expression in mammals has been shown to be restricted to the Kupffer cells [68], and the down expression of the C-reactive protein (crp), an acute phase protein synthesised by hepatocytes, in fish fed VD. Decrease in inflammatory response can also be related to the low level of ARA in the fish fed VD, which induces a reduction of prostaglandin synthesis derived from this fatty acid. Our microarray data indeed show that prostaglandin E synthase 2 (ptgs2), involved in the synthesis of pro-inflammatory prostaglandin E2, is down-regulated in fish fed VD, while prostaglandin E synthase 3 (ptgs3), which has anti-inflammatory properties, exhibited higher messenger levels in fish fed VD. This depression of innate immune system, particularly pro-inflammatory activity, could also be partially explained by a defect in membrane properties in fish fed VD, as revealed by the down-regulation of a large number of genes (65) related to cell communication, including factors such as cytokine receptor common subunit gamma (il2rg), receptor-type tyrosine-protein phosphatase F (ptprf) or integrin beta 2 (itgb2), which are cell-surface receptor binding proteins and/or cell adhesion receptors involved in immune response. The depression of the innate immune response in fish fed VD was confirmed by the lower plasmatic lysozyme concentration and lower-expression of lysozyme g (lyg) gene. Surprisingly, the alternative complement pathway activity involved in the innate immune response, which we assessed by analysis of plasma parameters, showed a significantly higher level in fish fed VD. Such an opposite regulation of the immune pathway revealed that different components of the immune systems can be regulated in opposite directions.

Interestingly, processes related to the humoral immune response were also over-represented among the genes up-regulated in half-sibfamily g. Indeed, complement component c2, c3 and c9 genes showed higher expression levels in half-sibfamily g. However, the up-regulation of genes involved in the alternative complement pathway cannot be associated with an increase of the plasma alternative complement pathway activity, probably due to the complexity of factors and regulation levels (transcriptional and post-transcriptional) involved in the regulation of this pathway [69]. Moreover, the higher expression of masp2, tnrfrf14, c2 and c3 genes involved in the inflammatory response might reflect higher inflammatory states in half-sibfamily g, which could be associated with a decrease in growth rate, as demonstrated in chicken [70].

Blood coagulation

Blood coagulation is another process involved in the innate immune system. LC-PUFA and, more specifically, EPA, DHA and ARA are precursors for eicosanoid synthesis involved in the control of the blood coagulation [71–73]. As mentioned above, the use of a diet composed of vegetable protein and oil induces modifications in the membrane phospholipid composition, with possible consequences for eicosanoid production and the blood coagulation process. We found that the use of the vegetable diet induced an increase in the expression of genes involved in the blood coagulation pathway (Figure 3). Among these genes, prothrombin (f7), coagulation factor × (f10), fibrinogen beta chain (fgb), fibrinogen gamma chain (fgg) and coagulation factor VII (f7) were positively involved in the blood coagulation process. On the basis of these results, the use of a VD seems to cause pro-coagulant action by the stimulation of the blood coagulation pathway, which is in agreement with our visual observation of plasma clotting (data not shown). In agreement with these results, Tavares-Dias [23] showed that, contrary to a vegetable diet, dietary enrichment in long-chain n-3 fatty acids has a strong hypocoagulant action. In addition, the regulation of the coagulation pathway is complex and under the control of several negative factors that maintain a physiological homeostasis. The pro-coagulation effect in response to a vegetable diet is notably reinforced by some of these genes [the vitamin K-dependent protein S (pros1), plasminogen (plg) and antithrombin-III (serpinc1)], which also exhibited higher expression in fish fed VD (Figure 3).