Immune and proteomic responses to the soybean meal diet in skin and intestine mucus of Atlantic salmon ( Salmo salar L.)

The main objective of this study was to increase the knowledge about the mucosal immunity of Salmo salar , using soybean meal- induced enteritis as a model of inflammation. A control fish meal (FM) and a diet containing 20% soybean meal (SBM) were fed to salmon for seven weeks in seawater. There was no growth difference between groups. However, histology of distal intestine (DI) showed a mild inflammation in the fish fed SBM. Proteomic results revealed differences between the diets. Among the proteins detected uniquely in DI mucus of SBM group, complement C5, Galectin and Glutathione synthetase are involved in innate and adaptive immunity, inflammation, redox signalling and detoxification of xenobiotics in mammals, and similar roles are hypothesized in salmon. Adenylosuccinate synthetase and putative aminopeptidase were uniquely detected in the skin mucus of SBM group. Trypsin enzymatic activity was significantly decreased in the DI of SBM group. Significantly higher production of immunoglobulin M and Mucin- like protein in DI mucus in SBM group was observed, while an increase in immunoglobulin D and lysozyme but decrease in chymotrypsin was detected in the skin mucus of the same group. We propose mucosal immunoglobulins as diagnostic biomarkers for assessment of novel feed ingredients and aquafeeds.


| INTRODUC TI ON
Atlantic salmon (Salmo salar L.) production is under constant health management pressure, which is critical for further sustainable growth of the aquaculture industry (Lekang et al., 2016). Increased incidence of infectious diseases causes substantial economic losses; therefore, there is a need for better preventive measures (Gudding & Van Muiswinkel, 2013;Murray & Peeler, 2005). Fish are continually interacting with aquatic microbiota that affects mucosal surfaces of gills, skin, and intestinal mucus, which are part of the first lines of defence (Salinas, 2015). Fish are also exposed to many different stressors, including mechanical, environmental and nutritional, that together can damage fish mucosal barriers and facilitate pathogen entry into the host (Cabillon & Lazado, 2019). Nutritional stressors arise from various alternative feed ingredients, which may contain antinutritional factors, imbalanced amino acid profile, indigestible sugars and other chemicals that can cause reduced feed intake, reduced nutrient digestibility and adversely affect growth performance and health (Francis et al., 2001). Soybean meal (SBM) is the major alternative to fish meal (FM), due to its high protein content and its favourable amino acid profile. However, the use of SBM in diets for salmon has disadvantages due to its high content of fibre and antinutritive factors, such as trypsin and protease inhibitors, lectin, antigen proteins and alkaloids, which affect digestive and nutrient absorption processes (Chikwati et al., 2013). The gut pathology known as SBM-induced enteritis (SBMIE) is a well-described inflammatory response in salmonids (Baeverfjord & Krogdahl, 1996;Krogdahl et al., 2010), but the specific antigens inducing this pathology are still unidentified and the mechanisms underlying SBMIE are not fully understood. In this regard, saponins have been pointed out as the main compound associated with gut inflammation in SBM fed fish (Knudsen et al., 2007;Krogdahl et al., 2015). However, recent studies demonstrated that not only saponins but also the combination of soyasaponins and other antinutritional factors present in other plant ingredients such as pea protein concentrate could induce enteritis Kortner et al., 2012). Even when pea or bean protein concentrate are used alone, without SBM, it is possible to observe symptoms of intestinal enteritis such as seen in SMBIE (De Santis et al., 2015;Penn et al., 2011). There are numerous approaches that evaluated the effects of SBM in salmonids; for example, growth performance, nutrient digestibility, enzyme activity, gut transcriptome, gut microbiota and immunity, histology and histopathology (Bakke-McKellep et al., 2007;Heikkinen et al., 2006;Marjara et al., 2012;Merrifield et al., 2011;Zhou et al., 2018).
Nevertheless, studies that evaluated the effect of SBM on the responses of skin and intestinal mucus on the protein level are scarce.
Most of the studies evaluated the changes and the impact of SBM on the transcriptome level (Król et al., 2016), which also provides valuable evidence. Still, the translation from gene to protein is a complex process, and the conclusions based on the transcriptome level may vary enormously from the proteome data (Manzoni et al., 2016). The uncovering of functional proteomics changes of intestinal and skin mucus in FM and SBM fed fish will add significant insight into the understanding of the mechanisms of SBMIE. The technique also provides opportunities to search for biomarkers to be used as preventive methods against diseases in the aquaculture industry.
The main objective of this study was to increase the knowledge about the mucosal immunity of Salmo salar using SBMIE as a model of inflammation. The gained knowledge and standardized phenotypical tools will help us to assess the impact of other novel feed ingredients that are constantly emerging on the market and to reveal the mechanism of dietary immunomodulation.

| Animals, feeding and sampling
The experiment was carried out with post-smolts Atlantic salmon (Salmo salar L), provided by AquaGen As (Trondheim, Norway).
Before the experiment, fish were kept at the fish laboratory at the Norwegian University of Life Sciences, Ås, Norway. When fish showed the morphological signs of smoltification (silvery colour, less visible parr marks, loose scales, blackfin margins of dorsal, caudal and pectoral fins, and almost transparent fins colour), they were transferred to the seawater (SW) facility at the Norwegian Institute for Water Research (NIVA). Further, fish were randomly distributed into 6 tanks (22 fish per tank), each one with 250 L capacity and a water flow of 8-10 L minutes −1 , resulting in three tanks per diet. The experimental fish had an average body weight of 107 g (n = 132), on the transfer day and a final average body weight of 158 g (n = 132) at the end of the experiment. The water salinity was gradually increased from 5 ppt at the transfer until full salinity (35 ppt) within three weeks. Continuous 24 hours light was provided during the experimental period. The average water temperature during the experiment was 11.5 ˚C, and the oxygen saturation was between 85 and 97%. The two experimental diets fed to fish for seven weeks were a control diet based on high-quality fishmeal (FM) and a challenge diet containing 20% soybean meal (SBM) ( Table 1).
Automatic belt feeders distributed feed 3 hours per day with a feeding level of 2% of the body weight, adjusted by the average feed consumption in each tank over the last seven days with 10% excess per day. The diets were produced by extrusion and subsequent vacuum coating with fish oil at the Centre for Feed Technology, Norwegian University of Life Sciences, Ås, Norway. The uneaten feed was collected during the whole experiment. The recovery values for each feed (pellets) were measured to ensure correct calculations of uneaten /eaten feed (dry matter (DM), g), according to Helland et al. (1996)). After seven weeks in SW, five fish per tank (15 fish per diet) were randomly taken out, anaesthetized (80 mg L −1 tricaine methanesulfonate (MS 222)) and weighed individually. The skin mucus was gently scraped using a cell scraper (VWR) along the lateral line, collected into 2 ml cryotubes, and immediately frozen for further protein extraction. Thereafter, the distal intestine (DI) was eviscerated by sterilized scissors and divided into two pieces. From the first DI piece, the tissue was cut transversally, intestinal content was removed, then rinsed with phosphate-buffered saline (PBS) and intestinal mucus was gently scraped by using a sterile cell scraper, collected into 2-ml cryotubes and immediately frozen for further protein extraction. The other DI piece was placed in 10% formalin for 48 hours at room temperature, and subsequently transferred to 70% alcohol and stored at 4°C until further histology processing.
The total weight of fish biomass was recorded at the transfer day and on the sampling day. The experiment was performed accord-

| Diets and ingredients
The diets and ingredients were ground and analysed for DM, crude protein (CP), crude lipid (CL) and ash (

| Histology
A histological assessment of DI sections was conducted at the histology laboratory, Veterinary Institute, Oslo. DI tissues sections (n = 30; 15 per diet) were routinely dehydrated in ethanol, equilibrated in xylene and embedded in paraffin. Longitudinal cuts of approximately 5 μm were stained with haematoxylin and eosin and examined under a light microscope. Tissue morphology evaluation was done by using a semi-quantitative scoring system as described by Penn et al. (2011). Selected tissue parameters and criteria for scoring were as following: 1) shortening of mucosal fold length, 2) increase in width and cellularity of the submucosa, 3) increase in width and cellularity of the lamina propria, and 4) reduction in enterocyte supranuclear vacuolization. The degree of histo-morphological changes was assessed and assigned to one of five categories, including scores ranging from 0 to 4, where 0 represented normal histology, 1-mild changes, 2-moderate changes, 3-marked changes and 4-severe changes. We had one intestinal section per fish, but the scoring criteria were quantified in the whole section area. The assignment of individual samples to the test diet groups was obtained after the evaluation was completed.

| Proteomics
Twenty-four randomly selected mucus samples (skin mucus (n = 12); DI mucus (n = 12)) from both dietary groups were thawed on ice and homogenized using beads and ice-cold lysis buffer (Tris 20 mM, NaCl 100 mM, Triton X-100 0.05%, EDTA 5 mM and protease inhibitor cocktail 1×). Then, the homogenate was centrifuged at 12000 g for 25 minutes at 4°C. The supernatant, containing soluble proteins, was then transferred to new tubes on ice. All protein samples were quantified by a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific) following the manufacturer's instructions. For proteomic analysis, 20 µg of total protein in PBS were pH adjusted to 8 by adding ammonium bicarbonate (Sigma-Aldrich, Darmstadt, Germany).
The samples were then digested with 10 µg trypsin (Promega, sequencing grade) overnight at 37°C. The tryptic peptides were analysed using an Ultimate 3000 RSLCnano-UHPLC system con-  version 1.6.0.7 based on MS1 intensity quantification. Proteins were quantified using the MaxLFQ algorithm. The data were searched against the salmon proteome (82390 sequences). Peptide identifications were filtered to achieve a protein false discovery rate (FDR) of 1% using the target-decoy strategy. The analysis was restricted to proteins reproducibly identified in at least three of the six replicates per diet. The mass spectrometry proteomics data have been depos-  Table 2.
Besides, the detection of trypsin and chymotrypsin was performed using the Fish Trypsin (TRY) ELISA kit (Cat no. MBS017140, MyBioSource) and Fish Chymotrypsin ELISA kit (MBS779023, MyBioSource), both according to the supplier instructions.

| Trypsin activity
Preparation of samples (6 fish per diet) and trypsin activity measurements in skin and intestine mucus samples were done by Trypsin Activity Assay kit (Colorimetric), following the manufacturer's instructions (Cat no. ab102531, Abcam).

| Statistical analysis
For group analysis of the growth, histo-morphological, immunological parameters and trypsin activity, statistical software GraphPad v7.03 was used. Shapiro-Wilk test was used to determine the normal distribution of the ELISA and enzyme activity data. From these results, significant differences between FM and SBM groups were established by Student's t-test (two-tailed). In addition, protein raw data were transferred to log normalization. Then, Volcano plot analysis, multivariate statistical analysis and data modelling were performed in R (https://www.r-project.org/). The results showed a Poisson distribution, and the common proteins detected in both diets are presented in the form of a heat map, with levels of protein expression across two dietary groups (FM and SBM). Hierarchical clustering was performed with the hclust function in R package.
UniprotKB database was used for the functional annotation of the proteins. All differences were considered significant when the pvalue was <.05.

| Growth performance
The experiment was successful in terms of SW adaptation, as indicated by the animals' quick return to normal feeding routines after the transfer to the SW. Table 3 shows the growth parameters. No statistical differences were detected in any of the measured parameters between FM and SBM dietary treatments.

| Histology
Histological examination was performed on the DI samples taken at the end of the trial (49 days in SW). The main finding from the gut health assessment of the histology sections was a mild inflammation in the DI mucosa of the fish fed SBM (Figure 1). The inflammation observed correlates with the well-documented SBMIE known to occur in salmonids fed diets containing standard-processed (hexane-extracted) SBM. Four morphological tissue parameters were reported ( Figure 1B) (mucosal fold height, submucosa width and cellularity, lamina propria width and cellularity and supranuclear vacuolization) to be different between diets. All fish fed FM presented normal DI histo-morphology, while fish fed SBM presented altered histo-morphological parameters with mild to moderate inflammation level ( Figure 1B).

Marker
Source Type Dilution References Muc-like proteins Mouse Polyclonal 1:400 Figure S2 TA B L E 2 Primary antibodies for indirect ELISA

| Proteomics
We performed proteomic analysis on mucosal samples isolated from DI and skin after seven weeks in SW. In total, 723 and 1232 protein were detected in the mucosal samples from DI and skin, respectively (Data S1). From the detected proteins, 292 proteins were common in both groups in the DI mucus, while 1075 were detected in both groups in skin mucus. Venn diagrams show the number of common and unique proteins altered by FM and SBM diet (Figure 2a). We detected one unique protein in DI mucus in the FM group and nine unique proteins in the SBM group. In the skin mucus, we detected two unique proteins in each diet. The detected unique proteins are presented in Table 4.
To study the relative expression of the common proteins detected in both diets, we generated volcano plots comparing SBM to the FM diet. We observed one significant protein in the skin mucus of fish fed SBM diet ( Figure 2c to the left), compared with the 30 significantly detected proteins from the DI mucus ( Figure 2c to the right). The pattern of expression of the 30 significant proteins is shown as a heat map in Figure 2b. Among those, 11 proteins were significantly higher in DI mucus of the SBM group compared with 19 proteins in FM group (Data S1). Most of the proteins overexpressed in DI mucus in SBM group presented a catalytic/hydrolase activity involved in the metabolic and cellular process with more than a half that presented an extracellular region. A similar trend was observed in the DI mucus of the FM group, where most of the proteins presented a catalytic activity, involved in the metabolic process.
Interestingly, we detected several peptides belonging to cathepsin proteins in the FM group (protein's id in red, Figure 2b).

| Immunological markers
The results of ELISA in skin mucus samples (Figure 3) showed that the production of IgD and lysozyme were higher (p < .05) in fish fed the SBM diet compared with fish fed the FM diet. The same trend was observed for IgM and Muc-like proteins on DI mucus samples (SBM group). On the other hand, chymotrypsin in skin mucus samples (SBM group) showed a significant decrease in its production compared with the FM diet. No significant differences in the production of trypsin were observed between the diets.

| Trypsin activity
In DI mucus, the enzymatic activity of trypsin ( Figure 4) showed a significant decrease in the SBM group compared with the FM diet.
In skin mucus, the activity of trypsin did not show significant differences between dietary groups.

| DISCUSS ION
In the aquaculture industry, different factors such as the optimization of environmental parameters and dietary regimens have a stronger impact on the mucosal surfaces of farmed fish compared with their terrestrial agricultural counterparts (Beck & Peatman, 2015). Therefore, mucosal immunity has been proposed as a key component for maintaining optimal fish health during its productive stage (Lazado & Caipang, 2014). The mucosal surface is the physical interface between fish and its environment, acting against external aggressions, such as microbes and stressors, in coordination with the immune system (Salinas et al., 2011).
The main goal of the present study was to detect proteomic and

TA B L E 3
Growth rate, feed intake (g/dry matter (DM)), feed conversion ratio, in Atlantic salmon fed fish meal (FM) and diet containing 20% soybean meal (SBM) diet for 49 days in seawater. Data are presented per dietary group (3 tanks per group) as means ± standard error mean Complement C5 is a key factor in a lytic cascade that can directly ex- presentation since the first steps in antigen degradation require GSH (Ghezzi, 2011;Short et al., 1996). In the mice model that studied acute lung injury induced by lipopolysaccharide (LPS) (Ghezzi, 2011), GSH is shown to be a regulator of the balance between innate immunity (leukocyte infiltration at the site of infection to kill bacteria) and inflammation (leukocyte infiltration to the lung to failure the organ). In Atlantic salmon, changes in plasma and liver GSH were reported to be affected by infectious salmon anaemia (Hjeltnes et al., 1992), with increased levels of plasma GSH, but decreased levels of hepatic GSH in diseased fish compared with their healthy controls.
Among the proteins uniquely detected in the skin mucus of SBM fed salmon, we detected adenylosuccinate synthetase and putative aminopeptidase, which are involved in AMP biosynthesis and proteolysis, respectively. However, we did not find a relationship between SBMIE and protein profiles. Besides, among the common proteins, detected in both groups, just one protein was significantly different and corresponded to an uncharacterized protein with 72% confidence alignment to hydrolase. On the other hand, most of the proteins that were differentially detected in DI mucus of fish fed SBM corresponded to trypsin and chymotrypsin. In a study with humans suffering from Crohn's disease (CD), which presents symptoms similar to enteritis in salmonids, increased faecal and digestive proteases such as trypsin and chymotrypsin have been observed (Midtvedt et al., 2013). The authors suggest that the increased level of digestive protease might be due to the changes in microbiota, where patients with CD presented a reduction in the number of Bacteroides, the main group of bacteria known to be able to inactivate pancreatic trypsin (Midtvedt et al., 2013). Although we did not perform microbiota analysis in this study, a recent study of Booman et al. (2018) showed  (Bryant et al., 2002). In fish, cathepsins are widely distributed in muscle and immunologically important organs, such as head kidney and spleen. This protein may play an important role in the protection of the host against microorganisms through the modulation of innate immunity (Tähtinen et al., 2002;Yamashita & Konagaya, 1990); however, more studies are needed to clarify the cathepsin roles in fish immunity under the effect of different feed ingredients.   (Lazado & Caipang, 2014;Salinas et al., 2011).
Our results showed that the use of a SBM diet was able to induce a response by secretable molecules from both innate and adaptive immunity, since we detected an increase in IgD and lysozyme in the skin and IgM and Muc-like proteins in DI. We think that this may be related to the inflammatory condition produced by the SBM diet, which induces a response that tries to fight against SBM, as well as against infectious agents that can even be opportunistic. This proposal can be supported by the properties of the different biomarkers that we detected. Lysozyme is a protein with lithic activity and can act as an opsonin, promoting the phagocytosis process and contributing to the innate defence against bacterial infection (Esteban & Cerezuela, 2015). While IgD is an immunoglobulin with a higher structural plasticity and can be produced as a transmembrane or secreted protein in a species-specific manner (Chen & Cerutti, 2010).
In fish, IgD has been described as a modulator between the innate and adaptive response, due to V-less region (Ramirez-Gomez et al., 2012). On the other hand, in higher vertebrates, secreted IgD can enhance mucosal homeostasis and immune surveillance by "arming" myeloid effector cells with antibodies against mucosal antigens (Gutzeit et al., 2018). However, this pathway can also cause an overreaction in the immune response by increasing inflammation and tissue damage (Chen & Cerutti, 2010). This is relevant if we relate the increase in IgD to the inflammatory condition caused by SBM in Atlantic salmon.
Continuing with the properties of the biomarkers that we detected in DI mucus, IgM is the most ancient antibody class and has the same function in all gnathostomes. The transmembrane form of IgM defines the B cell lineage, and in teleost fish, a secreted IgM form is produced as a tetramer in a different reduction or oxidation state that seems to modify the binding strength of the antibodies (Flajnik & Kasahara, 2010). It has been reported that the SBM diet causes significantly raised levels of IgM in the mid and DI mucosa (Krogdahl et al., 2000), which may be due to the different antigens present in SBM, such as glycinin and β-conglycinin (Wang et al., 2014) and as well, DI enteritis caused by SBM can lead to increased susceptibility to bacterial infections (Krogdahl et al., 2000). Finally, Muc-like proteins are involved in maintaining homeostasis between the local microbiota and the host. These proteins belong to a heterogeneous family of proteins composed of a long peptide chain with many tandem repeats (Koshio, 2016;Pérez-Sánchez et al., 2013). C5 could be produced, as they are also related to the modulation of the immune response. Finally, regarding the skin mucus, we believe that the production of IgD and lysozyme in fish fed with SBM is interesting, since they could be proposed as non-lethal and less invasive markers for the future investigations with fish, which could also integrate different phenotypic evaluation strategies to understand the interaction between nutrition and immunity at a deeper level.

| CON CLUS ION
In this study, Atlantic salmon mucosal immunity parameters at the protein level were characterized towards a deeper understanding of the already established SBMIE model. Immunological markers IgD, IgM, Muc-like proteins and lysozymes detected at a phenotypical level in the skin and DI mucus demonstrated that both innate and adaptive parameters could be stimulated in SBMIE. Therefore, we propose that the use of phenotypic strategies can help to assess the impact of feed ingredients through potential diagnostic markers in aquaculture when new aquafeeds need to be evaluated. Extensive studies, with increased number of fish, are needed to build proteomic baseline data for healthy and diseased fish which will serve for comparison and correlation with other available diagnostic methods.
New emerging knowledge will possibly help us to create diets that can assure good health and optimal growth.

ACK N OWLED G M ENTS
The authors would like to thank Ricardo Tavares Benicio for his skilful help during the feeding experiment and to the staff of the Fish laboratory at NMBU and NIVA (Solbergstrand) for their support during the experiment. We would also like to thank Elvis M. Chikwati for his assistance with histology samples.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest. All authors read and approved the final manuscript for submission. The content of the manuscript has not been published or submitted for publication elsewhere.

AUTH O R CO NTR I B UTI O N S
BD and MØ designed the feeding experiment. BD coordinated the execution of the experiment as well as sampling. BD and LL planned laboratory analysis. BML performed ELISA and enzyme assays. BD prepared samples for proteomics analysis and LL performed proteomics data analysis and proteomics statistics. BD, BML and LL were involved in manuscript writing and review, data analysis, producing figures and tables for the manuscript as well as statistical analysis and quality checking. MØ was involved in writing and reviewing the manuscript. LM designed antibodies and reviewed the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD017744. Other raw data that support the findings of this study are available from the corresponding author, upon reasonable request.