The increased demand for fish meal and fish oil to meet the requirements of expanding aquaculture and other farming industries has led to the search for alternative raw materials. Attention has been focused on plant ingredients and studies conducted in gilthead sea bream (GSB) (Sparus aurata L.) have demonstrated that the combined replacement of fish meal and fish oil (FO) is highly feasible without detrimental effects in growth performance when the theoretical requirements for essential amino acids and fatty acids are met by the diet [1, 2]. However, obvious changes in muscle fatty acid (FA) signatures occur with the replacement of FO by vegetable oils (VO) [3–5] and a wash-out period with fish oil-based diets is needed for the restoration of the desired FA profile having a high concentration of n-3 long-chain polyunsaturated FAs [6].

The effect of the nutritional background on fish health and welfare also merits careful consideration and there is much interest for research in this area in the recent years (reviewed in [7]). Indeed, adequate nutrition is essential to maintain health and to reduce disease susceptibility and pathological changes, and dietary lipids, as other nutritional factors, have specific actions on the immune response [8, 9]. In GSB, it is noteworthy that the redox balance [10] and the cortisol response after stress confinement [11] were altered by the high replacement of FO by VOs. Even lower VO replacement levels (50-60%) increased the cumulative mortality in GSB challenged with Vibrio alginolyticus or increased the intestine expression of TNF-α in Photobacterium damselae subsp. piscicida injected animals (see [9]). Furthermore, fish fed with a diet rich in VO exhibited a worse disease outcome when challenged with the intestinal parasite Enteromyxum leei, consisting in lower biometrical (weight, condition factor, specific growth rate), immunological (haematocrit, complement and lysozyme activity) and antioxidant (hepatic total glutathione) parameters compared to infected fish fed a FO diet [12]. In an effort to understand the possible underlying mechanisms involved in this greater progression of the infection in VO fish, we undertook a series of detailed studies of gut immunology and physiology in fish fed plant- or fish-based diets and challenged with this myxosporean. This intestinal parasite causes severe desquamative enteritis and thus far there are no preventive or curative treatments for this enteromyxosis. However, a few select inflammatory and immune relevant genes have been initially explored [13, 14], and more recently the molecular profiling with a cDNA microarray of the host response to the chronic exposure to E. leei has highlighted a complex interplay of proteases, protease inhibitors, apoptotic factors as well as cell proliferation and antioxidant defence genes [15]. The same array has also proven to be very useful for assessing the time course of stress response after confinement exposure [16], and herein the transcriptome database which served for the construction of these arrays was further enriched by suppression subtractive hybridization (SSH) libraries with lipid-responsive genes. All sequences were de novo assembled and the final annotated sequences were used to construct a customized oligo-array for expression profiling of the GSB intestine using a factorial design (2x2) with diet composition and parasite infection as experimental variables. To pursue this issue, juvenile GSB fed plant protein-based diets with either FO diet or a blend of VOs (66VO diet) as a major dietary lipid source were infected with the intestinal parasite E. leei by water effluent and the intestinal transcriptome of the host was analyzed 102 days post-challenge.

To date, a considerable number of studies have been conducted in fish to identify target genes of relevance for the improvement of production traits of farmed fish species (reviewed in [17–19]). The liver, due to its major role in metabolism, has been the main studied tissue in fish genome-wide analyses with special attention to stress response [16, 20, 21], ecotoxicology [22–24] or nutrigenomics [25–28]. Transcriptome responses involving the fish immune system after bacterial or viral experimental challenges have also been assessed in liver, although the main studied tissue for this purpose is the head kidney because of its central hematopoietic role, equivalent to that of mammalian bone marrow [29–31]. The fish intestine, besides its importance for the absorption of nutrients and osmoregulation, also acts as an immune tissue and constitutes a barrier and first line of defence against certain pathogens and environmental challenges [32, 33], so the gene expression profile of the intestine in nutritional and/or pathogenic trials should be specially considered. However, very few transcriptome-wide studies have paid attention to the response of fish gut, being focused these approaches on salinity adaptation in European eel (Anguilla anguilla) [34], and phosphorus deficiencies and dietary immunostimulants in rainbow trout (Oncorhynchus mykiss) [35, 36] or replacement of fish meal and FO in diets for Atlantic halibut (Hippoglossus hippoglossus), Atlantic salmon (Salmo salar), and Atlantic cod (Gadus morhua) [37–39].

In GSB, the host response to chronic exposure to E. leei was previously analyzed in both intestine and head kidney by transcriptome profiling of few candidate genes [14], and more recently [15] by means of a cDNA microarray that was previously proven successful to assess the response of the liver of fish undergoing confinement exposure [16]. The oligo-microarray designed and used in the current work was developed from these previous annotated nucleotide sequences updated with lipid-responsive genes by means of a SSH approach in order to have a wide range of genes potentially regulated by nutritional deficiencies in essential fatty acids. In this scenario, it must be noted that this is the first attempt to analyze the combined effect of diet and parasite infection on the fish intestinal transcriptome. The detected massive transcriptome changes can be mainly attributed to the progression of the infection, but not to the dietary treatment, since no differences between diet groups were found in the intestine transcriptome of control animals (not exposed to the parasite) as revealed by the one-way ANOVA and principal components analysis. Furthermore, both control diet groups showed similar growth performance with no evidences of hepatic and intestinal histopathology, probably due to the adequate supplementation of diets with soy-lecithin as an extra-source of dietary phospholipids [2, 3]. By contrast, in Atlantic salmon and Atlantic cod, the adverse effects of dietary soybean meal leading to intestinal disorders (inflammation and lipid accumulation) were also accompanied by changes in the intestine transcriptome [38, 39].

In our experimental model, intestine transcriptomic differences between diets only became evident when animals were infected with the parasite, as the number of differentially expressed genes and the degree of fold-change variations were much higher in animals fed the 66VO diet than in those fed the FO diet, which paralleled the increased symptoms of enteromyxosis (lower growth, condition factor and haematocrit in combination with higher anorexia, and intensity and extension of the infection) in parasitized fish of this diet group [12]. Furthermore, the gene expression profile determined by k-means clustering also emphasizes the importance of the progression of the infection, with a stronger effect in terms of fold-change variation in infected animals fed the 66VO diet, and even more in early infected fish. This reinforces the idea that the microarray used in the present study constitutes an excellent diagnostic tool to address changes associated with the action of the pathogen.

When analyzing the biological functions of the genes differentially expressed upon infection, it was evident that the infection produced a detrimental effect on many pathways related to growth and normal metabolism, as genes related to protein synthesis, protein degradation, small molecule biochemistry, lipid metabolism, cancer and carbohydrate metabolism were down-regulated. By contrast, other biological functions, related to infection mechanism, infectious disease and immune response were over-represented by up-regulated genes. It is difficult to compare the current results with those obtained in other fish-pathogen models, since the host-pathogen interactions are different and the times post-infection at which the samples are analyzed differ. However, in most cases up-regulation of different immune genes occurred and the down-regulation of many other genes was also evident, as already shown [31].

In order to define the response to the parasite, special attention was paid to differentially expressed genes (up- and down-regulated genes) of biological processes statistically enriched in the infected groups, and the reliability of the results was validated by qPCR of selected key genes representative of the four clusters. The enzyme arginase-1 was representative of the strongly up-regulated genes of cluster 1 in infected fish. Enhanced expression of this gene was also detected in the head kidney of Atlantic salmon and common carp (Cyprinus carpio) challenged with the bacterium Aeromonas salmonicida and the protozoan parasite Trypanosoma carasii, respectively [30, 40]. The so-called “alternatively” activated macrophages play important roles in the clearance of pathogens (as reviewed in [41]), and their enhanced arginase activity allows them to produce ornithine, a precursor of hydroxyproline and polyamines. A first enzyme step is that of ornithine decarboxylase (ODC), though ornithine can also be produced from glutamate via ornithine aminotransferase (OAT). Interestingly, in our experimental model both ODC and OAT were up-regulated and belong to cluster 2, which confirms and extend the idea of an enhanced production of polyamines in the fish challenged with E. leei.

Polyamines have a significant effect on the growth of the gastrointestinal mucosa of a variety of organisms including fish [42], and the in vitro immunostimulatory action of putrescine has been observed in head kidney leukocytes of GSB [43]. Therefore, the increased expression of these genes involved in cell proliferation in infected GSB should be interpreted as an evidence of the regenerative action of the intestinal tissue as a response to the damage induced by the parasite invasion. L-arginine is also the substrate for the synthesis of nitric oxide (NO) by the catalytic action of NO synthases. Thus, if L-arginine is mainly used by arginase-1 in ornithine synthesis, the production of NO should be expected to be reduced, and this is exactly what occurred in the serum NO levels of infected fish [12]. NO is an important molecule in regulating immune functions and also has a direct antimicrobial effect [44]. However, NO has double-edge sword effects [45], since an elevated production of NO for long periods of time not only can have the desirable protective action against the pathogen, but also can lead to a higher concentration of NO available to react with O₂, increasing the production of reactive nitrogen species (RNS) and generating indirect toxic effects on the host. Since reactive oxygen species (ROS) were enhanced in infected fish (increased respiratory burst of circulating leukocytes) [12], it is a reasonable conservative strategy to reduce RNS to avoid detrimental effects to the host.

In cluster 2, several molecular chaperones (mitochondrial 10 kDa and 60 kDa heat shock proteins, glucose-regulated protein 75, heat shock 70 kDa protein 4, mitochondrial chaperone BCS1, proteasome assembly chaperone 3) were up-regulated in the infected fish. The role of these life essential proteins is to stabilize unfolded proteins, often coupling ATP binding/hydrolysis to the folding process. Thus, their expression is often increased by cellular stress, as occurs with heat shock proteins of the HSP70 family, which are highly inducible under stress conditions in higher vertebrates [46] and also in fish [47]. Glucose-regulated protein 75, also named mortalin or mitochondrial HSP70, is one of the molecular chaperones representative of this cluster and their enhanced expression was coincident with the up-regulation of several mitochondrial ATP synthases, included in the same cluster. This finding is consistent with previous results analyzing the gene expression pattern of some target growth, redox and immune-relevant genes in the intestine of GSB [14]. The up-regulated expression of mortalin at the mRNA and protein level has also been observed in the liver tissue of GSB during both acute and chronic confinement stress [48], which emphasizes the relevance of this mitochondrial protein encoded by nuclear DNA as a stress biomarker in this fish species. Interestingly, no changes in mortalin expression were observed in the gills of Atlantic salmon infected with the protozoan Neoparamoeba perurans, the causative agent of amoebic gill disease, but resistant animals exposed but not infected with the parasite showed a significant up-regulation of mortalin expression [49], which evidences a complex and perhaps species-specific protective role of this protein in front of different stressors. Cluster 2 also comprised several immune-related genes, and the expression pattern of some of them, including IL6, IL6 receptor and peroxiredoxin 1 (also named natural killer enhancing factor-A), was validated by qPCR. Their up-regulation probably reflects the activation of innate immune response at the local site of infection. Similarly, in other fish-parasite models, up-regulation of different immune genes, mainly chemokines, cytokines, lectins and enzymes of eicosanoid metabolism, have been reported at the local site of infection (gills, skin, cartilage) [50–55].

Cluster 3 grouped genes severely down-regulated by infection. Among them, genes involved in xenobiotic metabolism and detoxification pathways were represented by cytochromes P450 and metallothionein. Since an increasing body of evidence points to the importance of intestine as a xenobiotic-metabolizing tissue [56], this decreased gene expression should be viewed as a counter-regulatory response that maintains the redox balance between the mechanisms bringing about the pathogen elimination and those governing the growth and repair of damaged tissues in a scenario where ROS production is potentiated in response to the parasite [12]. Another group of genes related to lipid metabolism (FABP2, liver X receptor alpha, phospholipase A2) and other established markers of metabolic activity in GSB, such as UCP1 and peroxiredoxin 6 [57, 58], were also down-regulated in cluster 3. Their reduced expression with infection is also suggestive of the loss of intestinal functions. This was further corroborated by pathway analysis of the differentially expressed genes during infection and by the down-regulation in cluster 4 of components of the somatotropic axis (GHR-I, GHR-II, IGF-II, IGFBP4) of importance in intestinal growth and repair [59], that can be considered a prelude to the severe cachectic episodes typically associated with more advanced stages of E. leei infection.

Some genes stand out of the general trend of higher regulation (either up or down) with longer infection time, as shown in Figure3. This was the case for immunoglobulins, whose production was more up-regulated than expected in early-infected fish. This is consistent with the fact that the specific immune response takes more time to appear in fish than in higher vertebrates [60] and this time is particularly extended in this fish-parasite model. In fact, it has been shown that time of exposure to E. leei is the most determinant factor for the intestine expression of immunoglobulin M, the major component of fish specific humoral response, and this response is also magnified in 66VO fish [61]. In any case, we cannot discard the possible action of another Ig isotype, IgT/IgZ, which seems to act exclusively in mucosal areas, and has been described in very few fish species, with outstanding results in another myxosporean infection [62], but not yet found in GSB nor present in the current microarray. On the other hand, some genes related to complement pathways were strongly down-regulated, which agrees with the previously observed down-regulation in E. leei-exposed GSB, although timing of infection was not considered [15]. These observations reflect the exhaustion of the alternative complement pathway also reported in other Enteromyxum spp. chronic infections [13, 63, 64].

Taking together all these results, it appears that the changes detected in the intestine transcriptome are mostly a consequence rather than a cause of the different disease progression, so the differentially expressed genes can function as diagnostic markers of disease progression but lack a prognostic value to predict in advance the susceptibility of diet groups to an infective scenario. Although we did not find any transcriptomic differences in the intestine of the control groups through this microarray approach, other possibilities rather than a lack of effect of a differential diet on the intestinal gene expression profile must be considered to explain this result: (i) technical limitation of our microarray to detect significant differences in the expression of some genes, (ii) absence of genes with potential prognostic value in our nucleotide database or the resulting microarrays, or (iii) differences between groups could be due to non-transcriptional mediated processes. Further work is under way in order to achieve a more complete picture of the transcriptome of GSB. For instance, next-generation sequencing techniques (454 pyrosequencing) have been applied to normalized libraries from several GSB tissues including intestine, yielding millions of new nucleotide reads that will update and enrich in thousands of genes our GSB nucleotide database. Thus, the potential for the detection of changes in the expression of new genes will be greatly increased in future studies.

Finally, in this and other related works where the fish transcriptome response to an infective pathogen is observed, the transcriptome response is focused exclusively on the host [65–67], but the direct effects of diet composition and nutrients on parasite physiology and survival should also be considered, as it has been in the case of the direct effect of FAs as antimalarial agents where a direct effect on the FA biosynthetic machinery of the parasite Plasmodium falciparum has been shown [68]. It must be noted that most oils of vegetable origin are less susceptible to oxidative degradation than fish oils due to their lower content of very-long-chain n-3 polyunsaturated fatty acids [69], so parasite fatty acid uptake when feeding with FO diet could make it more sensitive to oxidative stress and consequently to the immune host response.

Plant oils in plant protein-based diets did not modify the intestine transcriptome of GSB, although there were significant effects when these fish were exposed to a parasite challenge with the myxosporean E. leei. The detected changes are mostly a consequence rather than a cause of the different disease progression in the two diet groups. The microarray approach as used here constitutes an excellent diagnostic tool to address changes associated with the action of intestinal pathogens, but lacks a prognostic value to predict in advance the susceptibility of different diet groups to this pathogen. Further studies are needed to explore the mechanisms involved in the altered susceptibility of GSB against the parasite E. leei when fed different diets, without excluding direct effects of fish feeds on parasite metabolism.