Aeromonas salmonicida Binds Differentially to Mucins Isolated from Skin and Intestinal Regions of Atlantic Salmon in an N-Acetylneuraminic Acid-Dependent Manner

Abstract

Aeromonas salmonicida subsp. salmonicida infection, also known as furunculosis disease, is associated with high morbidity and mortality in salmonid aquaculture. The first line of defense the pathogen encounters is the mucus layer, which is predominantly comprised of secreted mucins. Here we isolated and characterized mucins from the skin and intestinal tract of healthy Atlantic salmon and studied how A. salmonicida bound to them. The mucins from the skin, pyloric ceca, and proximal and distal intestine mainly consisted of mucins soluble in chaotropic agents. The mucin density and mucin glycan chain length from the skin were lower than were seen with mucin from the intestinal tract. A. salmonicida bound to the mucins isolated from the intestinal tract to a greater extent than to the skin mucins. The mucins from the intestinal regions had higher levels of sialylation than the skin mucins. Desialylating intestinal mucins decreased A. salmonicida binding, whereas desialylation of skin mucins resulted in complete loss of binding. In line with this, A. salmonicida also bound better to mammalian mucins with high levels of sialylation, and N-acetylneuraminic acid appeared to be the sialic acid whose presence was imperative for binding. Thus, sialylated structures are important for A. salmonicida binding, suggesting a pivotal role for sialylation in mucosal defense. The marked differences in sialylation as well as A. salmonicida binding between the skin and intestinal tract suggest interorgan differences in the host-pathogen interaction and in the mucin defense against A. salmonicida.

INTRODUCTION

Fish is a uniquely healthy food; e.g., it has been proposed to decrease cardiovascular risk (1, 2). Natural fish stocks are overexploited in a nonsustainable way. Aquaculture is the fastest-growing food-producing sector worldwide, providing approximately 1.5 million tons of Atlantic salmon, Salmo salar L., each year (3), thus exceeding the contribution of fishing to the total global salmon market. Aquacultured fish may be subjected to more low-intensity, long-term stress than fish in the wild, which in turn may make them more susceptible to infections (4–7). Historically, furunculosis disease caused by Aeromonas salmonicida subsp. salmonicida has been recognized as one of the most destructive infections in salmonid aquaculture. Vaccinations are usually successful, but the protection is not absolute, and vaccinations often lead to reduced welfare through injuries and behavioral changes. Pathological changes after intraperitoneal vaccinations include peritonitis (8), soft tissue adhesions (9, 10), and deposition of melanin (11) but also skeletal deformities (12, 13). Further, the behavior of the fish is affected; intraperitoneal vaccination, using oil-based adjuvants, has been shown to reduce food intake and interest in feed, as well as swimming activity (8). Thus, even though intraperitoneal vaccinations are relatively effective, available vaccines are costly and impose severe welfare problems on the farmed fish.

For a pathogen such as A. salmonicida to infect the host, innate, primary barriers must first be passed. The primary barriers constitute the surfaces of the fish, i.e., the border between the environment and the internal milieu. The intestine of Atlantic salmon and rainbow trout, Oncorhynchus mykiss, has been demonstrated to be a translocation route, but translocation has also been suggested to occur across both the skin and the gills (14–19). The epithelial surfaces of fish are covered by a mucus layer, which is the first barrier the pathogen encounters. In fish as in mammals, it has been shown that the mucus layer is continuously secreted and thereby washes away trapped particles. The main components of this mucus layer are secreted mucins (20, 21). In mammals, each mucin can carry on the order of 100 different carbohydrate structures, which provides a vast array of potential binding sites for microbes (20). Furthermore, mucins bind a range of bacteria and constitute an important part of the mucosal defense against infection (20). However, it has also been suggested that binding to mucins could provide the pathogen with an adhesion site and thereby instead promote infection (22), although no conclusive evidence for this exists. The mucous niche is very unstable, and pathogen binding to mucins may instead serve as a decoy for the more intimate adherence that can occur between the pathogen and, e.g., glycolipids of the cell membrane. Indeed, mucin binding to the human gastric pathogen Helicobacter pylori acts as a decoy and prevents prolonged adherence (23). Furthermore, in the rhesus monkey model of H. pylori infection, monkeys with mucins that bind H. pylori more effectively have a lower H. pylori density in their stomachs, indicating that mucin binding to H. pylori aids in removing the bacteria from the gastric niche (24). Consistent with these results, exposure of the intestinal tract of the common carp, Cyprinus carpio, to bacterial lipopolysaccharide (LPS) increased mucus secretion and changed the glycosylation pattern and thus the bacterial adherence properties, indicating a potential increase in pathogen removal (25).

In humans and mice, a range of mucins designated MUC1 to MUC20, with different distributions throughout the body, have been identified (20). Five orthologues of these have been identified in the genome of the pufferfish, Fugu rubripes (26), and 13 orthologues have been identified in the genome of the zebrafish, Danio rerio (27). Next-generation sequencing of salmon skin has revealed partial mucin sequences with homology to the human MUC2 and MUC5 mucins (28). However, knowledge on the composition and function of the fish mucins in pathogen defense is scarce.

The main objectives of the present study were to isolate and characterize Atlantic salmon intestinal tract and skin mucins and to study how A. salmonicida binds to these mucins.

MATERIALS AND METHODS

Fish and sampling procedure.

Atlantic salmon parr (Långhult lax, Långhult, Sweden) were transported to the Department of Biology and Environmental Sciences and kept in 500-liter tanks. The fish were held in recirculating 10°C fresh water, supplemented with 10% salt water (yielding salinity of 2‰ to 3‰), at a flow rate of 8.5 liters/min. The fish were exposed to a simulated natural photoperiod and hand-fed ad libitum once daily with a commercial dry pellet (Nutra Olympic; Skretting Averøy Ltd., Stavanger, Norway) (3 mm in diameter).

Five fish (length, 31.06 ± 0.49 cm; weight, 280.70 ± 12.78 g; means ± standard errors of the means [SEM]) were randomly netted, anesthetized in metomidate (12.5 mg/liter), and killed with a sharp blow to the head. Mucus from the skin was sampled by gentle scraping of the skin of the entire fish using two microscopy glass slides. The fish was then opened longitudinally and the intestine, from the last pyloric cecum to the anus, was quickly dissected. The intestine was cut open along the mesenteric border, the proximal region was separated from the distal at the ileorectal valve, and the mucus and mucosa were scraped off using microscopy slides. The pyloric ceca were dissected using small scissors and placed in liquid nitrogen for pulverization using a mortar and pestle. All samples were placed in 10 mM sodium dihydrogen phosphate containing 0.1 mM phenylmethylsulfonyl fluoride (PMSF; pH 6.5) (sampling buffer). Atlantic salmon from another cohort (10.1 ± 0.27 cm in length and 11 ± 1 g in weight; mean ± SEM; n = 9) were used for quantification of goblet cells in the two intestinal regions. Fish were sampled as described above, and proximal and distal intestine samples were fixed in methacarn (methanol-chloroform-glacial acetic acid, 60:30:10 [vol/vol/vol]) overnight at 4°C and placed in 70% ethanol (at 4°C) until being embedded in histowax.

Isolation and purification of mucins.

The scrapings and pulverized tissues in sampling buffer were placed into five sample volumes of extraction buffer (6 M guanidine hydrochloride [GuHCl], 5 mM EDTA, 10 mM sodium phosphate buffer [pH 6.5], 0.1 mM PMSF), dispersed using a Dounce homogenizer (four strokes with a loose pestle), and stirred slowly at 4°C overnight. The insoluble material was removed by centrifugation at 23,000 × g for 50 min at 4°C (Beckman JA-30 rotor), and the pellet was reextracted twice with 10 ml extraction buffer. The supernatants from these three extractions were pooled and contained the GuHCl-soluble mucins (referred to here as soluble mucins). The “insoluble mucins” were brought into solution from the final extraction residue (pellet) by reduction with 10 ml of reduction buffer (6 M GuHCl, 10 mM 1,4-dithiothreitol [DTT], 5 mM EDTA, 0.1 M Tris-HCl buffer, pH 8.0) for 5 h at 37°C followed by alkylation with 25 mM iodoacetamide overnight in the dark at 23 to 24°C and then centrifuged as described above. The insoluble and soluble samples were dialyzed twice against 10 volumes of extraction buffer, and then the reaction volume was adjusted to reach 26 ml by the addition of extraction buffer. CsCl was added by gentle stirring and the samples were transferred to Quick Seal ultracentrifuge tubes (Beckman Coulter). The tubes were filled with 10 mM NaH₂PO₄ to give a starting density of 1.35 g/ml and subjected to density gradient centrifugation at 40,000 × g for 90 h at 15°C. The fractions were collected from the bottom of the tubes with a fraction collector equipped with a drop counter. Density gradient fractions of purified mucin were analyzed for carbohydrates as periodate-oxidizable structures in a microtiter-based assay: fractions diluted 1:100, 1:500, and 1:1,000 in 4 M GuHCl were coated into 96-well plates (PolySorp; Nunc A/S, Roskilde, Denmark) and incubated overnight at 4°C. The rest of the assay was carried out at 23 to 24°C. After washes were performed three times with washing solution (5 mM Tris-HCl, 0.15 M NaCl, 0.005% Tween 20, 0.01% NaN₃, pH 7.75), the carbohydrates were oxidized by adding 25 mM sodium metaperiodate–0.1 M sodium acetate buffer (pH 5.5) for 20 min. The plates were washed again, and the wells were blocked with Delfia blocking solution (50 mM Tris-HCl, 0.15 M NaCl, 90 μM CaCl₂, 4 μM EDTA, 0.01% NaN₃, 0.1% bovine serum albumin [BSA], pH 7.75) for 1 h. After further washing steps, the samples were incubated for 1 h with 2.5 mM biotin hydrazide–0.1 M sodium acetate buffer (pH 5.5), followed by another washing step. Europium-labeled streptavidin was diluted 1:1,000 in Delfia assay buffer (50 mM Tris-HCl, 0.15 M NaCl, 20 mM diethylenetriaminepentaacetic acid [DTPA], 0.01% Tween 20, 0.02% NaN₃, 1.5% BSA, pH 7.75) and was added to the wells. After 1 h of incubation, the plates were washed six times and then incubated with Delfia enhancement solution (0.05 M NaOH, 0.1 M phthalate, 0.1% Triton X-100, 50 mM TOPO, 15 mM biotin nitrilotriacetic acid [b-NTA]) for 5 min on an orbital shaker. Fluorescence (excitation gamma [λ_ex], 340 nm; emission gamma [λ_em], 615 nm) was measured using a Wallac 1420 Victor2 plate reader with the europium label protocol (PerkinElmer, Waltham, MA). Density measurements were performed using a Carlsberg pipette as a pycnometer: 300 μl of sample was sucked into a pipette and weighed, and density was calculated as g/ml. DNA content was calculated from UV light absorbance at 260 nm.

Preparation of mucin samples.

Gradient fractions containing mucins were pooled to obtain one sample for each gradient (Fig. 1). Mucin concentrations in pooled samples were determined by serial dilution of the samples as well as by the use of a standard curve of a fusion protein of the mucin MUC1, 16TR, and IgG2a Fc (29), starting at a concentration of 20 mg/ml, using seven 1:2 serial dilutions and detection of carbohydrates as periodate-oxidizable structures in a microtiter-based assay as described above. The mucin concentrations were calculated from the standard curve. This method was chosen since we know from previous experiments that all mucins do not resolubilize after freeze-drying and that determining concentrations by freeze-drying can therefore result in large errors as well as in the selective removal of mucin species. Although the method employed does not result in an exact measure of concentration, it can be used to ensure that the mucins are at the same concentration for comparative assays, and since bacterium-mucin interactions largely occur via the mucin glycans (20), setting the concentration on the basis of the glycan content is most appropriate.

FIG 1.

Purification of skin and intestinal mucins. (A and B) CsCl density gradient of skin (A) and distal intestinal (B) mucins. The gradients were emptied from the bottom of the ultracentrifuge tubes, and the carbohydrate content (europium count/well; an Eu count of 20,000 approximately corresponds to 4 μg glycoprotein/ml), DNA content (ng/μl), and density (g/ml) were determined. The mucin-containing fractions were pooled on the basis of the carbohydrate content (fractions pooled are indicated by the vertical dashed lines). (C) GuHCl-soluble mucins were predominant in samples isolated from Atlantic salmon skin and intestinal tract (n = 1). Abbreviations: Pyloric = pyloric ceca; Proximal = proximal intestine; Distal = distal intestine; SOL, soluble; INS, insoluble.

To estimate the mucin quantity for each body site, the amount of isolated mucin was expressed in units of measured tissue surface area (cm²) for the proximal and distal intestine and in units of estimated body surface area (cm²) for the skin samples (estimated from the body weight using the linear regression formula by Tucker et al. [30]). Measurement of the surface area of pyloric ceca is less feasible, as the high number of ceca makes surface measurement very time-consuming and the lengthy handling time would severely lower the quality of the preparations. Furthermore, to our knowledge, there is no correlation between surface area of pyloric ceca and body weight as proposed for the skin, mainly due to the high individual variation in the numbers of ceca (31). Instead, the mucins of the pyloric ceca are expressed per unit of weight of the organ. It is therefore not possible to compare the mucus quantity found in the skin and pyloric ceca with the content of the proximal and distal intestine.

Mass spectrometric analysis.

Fish mucins (from skin, pyloric ceca, and proximal and distal intestine, n = 5) were dot blotted to polyvinylidene difluoride (PVDF) Immobilon P membranes (Millipore, Billerica, MA), visualized by alcian blue (AB; Sigma-Aldrich) staining, excised, and subjected to reductive β-elimination. In brief, excised dots were incubated with 0.5 M NaBH₄ and 50 mM NaOH for 16 h at 50°C. Reactions were quenched with glacial acetic acid, and the samples were desalted and dried as previously described (32). Released O-glycans were analyzed by liquid chromatography-mass spectrometry (LC-MS) using a column (10 cm by 150 μm inside diameter [I.D.]) prepared in-house that contained 5-μm-diameter porous graphitized carbon (PGC) particles (Thermo Scientific, Waltham, MA). Glycans were eluted using a linear gradient of 0% to 40% acetonitrile–10 mM ammonium bicarbonate over 40 min at a flow rate of 10 μl/min. The eluted O-glycans were detected using a LTQ mass spectrometer (Thermo Scientific) in negative-ion mode with an electrospray voltage of 3.5 kV, capillary voltage of −33.0 V, and capillary temperature of 300°C. Air was used as a sheath gas, and mass ranges were defined in a manner dependent on the specific structure to be analyzed. The data were processed using Xcalibur software (version 2.0.7; Thermo Scientific). The composition of the glycans was confirmed from their tandem MS (MS/MS) spectra. Values corresponding to the weighted average glycan chain length of carbohydrate structures on each mucin sample were calculated using the relative (percent) abundance of each glycan structure.

Aeromonas salmonicida culture conditions and binding to mucins.

A. salmonicida subsp. salmonicida strain VI-88/09/03175 (culture collection, Central Veterinary Laboratory, Oslo, Norway), strain A449 (33), and strain 01-B526 (isolated from brook trout) were cultured in brain heart infusion (BHI) broth at 19°C, and stocks were stored in BHI broth-glycerol (1:1) at −80°C. Binding of A. salmonicida to mucin samples from fish as well as to gastric rhesus monkey mucins and rat intestinal mucins was assessed using a microtiter-based assay. Rhesus monkey and rat mucins were scraped from the surface of the excised mucosal tissues with microscopy slides and purified as previously described (34, 35), in a manner similar to that employed for the fish mucins. Mucins were diluted in 4 M GuHCl–phosphate-buffered saline (PBS) (140 mM NaCl, 2.7 mM KCl, 10 mM phosphate buffer, pH 7.4) and coated on 96-well PolySorp plates overnight at 4°C. The plates were washed 3 times with washing buffer (PBS containing 0.05% Tween 20), and the wells were blocked for 1 h with blocking buffer (0.5% BSA in washing buffer). After the blocking buffer was discarded, bacteria harvested in the log phase of growth were washed gently (two times at 3,000 × g, resulting in a soft, easily resuspendable pellet) and diluted in blocking buffer to an optical density at 600 nm (OD₆₀₀) of 0.1 and then further diluted 1:10 in blocking buffer. The bacterial suspension was added to the plates, which then were incubated on a shaker (120 rpm) at 23 to 24°C for 2 h. The plates were washed 3 times and then incubated for 1 h at room temperature with blocking buffer containing anti-A. salmonicida IgG monoclonal antibody (clone 3B11/G5; Austral Biologicals, San Ramon, CA) diluted 1:5,000. After washing, horseradish peroxidase (HRP)-conjugated anti-mouse IgG (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) diluted 1:10,000 in blocking buffer was added. After further washing steps, tetramethylbenzidine (TMB) substrate (Sigma-Aldrich Co.) was added and the plates were incubated for 20 min. The reaction was stopped with an equivalent volume of 0.5 M H₂SO₄, and the plates were read in a microplate reader at 450 nm after color stabilization. The binding was analyzed at four mucin concentrations (in 2-fold dilution steps), with binding signal results (expressed per unit of carbohydrate signal) that were similar for at least three of the concentrations (i.e., the assay was performed within a linear range), including a glycan value of 20,000 europium counts (approximately corresponding to a 4 μg/ml mucin concentration). For every analysis, parallel microtiter plates were coated for glycan detection analysis to ensure that small differences in binding were not due to coating differences. The binding values shown were normalized for a glycan value of 20,000 europium counts.

To investigate the effect of desialylation on A. salmonicida binding, 75 μl mucins (at a concentration yielding a glycan count of 20,000 europium counts) were coated as described above for 8 h. The plates were then washed 3 times with washing buffer and blocked for 1 h with blocking buffer. After the blocking buffer was discarded, 75 μl sialidase A enzyme (ProZyme, Hayward, Canada) diluted 1:200 in sterile, 50 mM Na₃PO₄ buffer (pH 6.0) was added and digestion was carried out at 37°C, with continuous orbital shaking at 70 rpm for 13 h. Mucins treated with Na₃PO₄ buffer only were used as mock-treated controls. After digestion, plates were washed 3 times with washing buffer and the wells were blocked for 1 h with blocking buffer. After the blocking step, the binding experiment was carried out as described above.

Goblet cell staining and quantification.

The intestinal tissues were dehydrated through an alcohol gradient and embedded in paraffin. Sections were cut at 3-μm intervals and mounted onto 3-aminopropyltriethoxysilane (APES; Sigma-Aldrich)-coated slides and dried at 37°C for 48 h. Tissue sections were cleared in xylene (twice for 10 min each time), hydrated in 100% ethanol (two times for 10 min each time), 90%, 80%, and 70% ethanol (for 5 min each), and double-distilled water (ddH₂O) (two times for 5 min each time). Goblet cells were stained using alcian blue (1% alcian blue 8GX–3% acetic acid). The slides were then dehydrated in 96% ethanol (two times for 3 min each time), 100% ethanol (two times for 3 min each time), and xylene (two times for 5 min each time) and finally mounted in Pertex (Histolab Products AB, Gothenburg, Sweden). A subset of sections were stained with the dual stain alcian blue/periodic acid-Schiff (PAS) by adding the following step after the alcian blue staining procedure: the sections were immersed in 1% periodic acid–water at room temperature for 10 min, washed in water for 5 min, immersed in Schiff's reagent for 10 min, and rinsed one time in water for 5 min and then three times in 0.5% sodium metabisulfite before a final wash in water was performed. The staining results revealed that the goblet cells were homogenously stained with both stains. Photographs were taken using a NikonDXM1200 digital camera attached to a Nikon Optiphot microscope. Images were imported into BioPix iQ software (BioPix AB, Sweden), which was used to quantify the number of goblet cells in relation to the epithelial area.

Statistical analyses.

Statistical analyses were performed using the Graph Pad Prism 5.0 (GraphPad Software Inc.) software package. One-way analysis of variance (ANOVA) followed by post hoc testing using the Student Newman-Keuls test and paired Student t tests (when appropriate) was used for comparisons between groups. Normality and homoscedasticity of data were investigated using the Kolmogorov-Smirnov test and Bartlett's test, respectively. Data not passing the homoscedasticity test (the A. salmonicida binding data) were subjected to log₁₀ transformation. Due to an apparent larger variation in A. salmonicida binding in the distal intestine, transformation was not sufficient to pass Bartlett's test for this parameter. However, in experiments performed with several treatments and balanced data, heterogeneous variances do not noticeably increase the risk of type I error (36). All aspects considered, we decided that one-way ANOVA followed by post hoc testing using the Student Newman-Keuls test was the most appropriate way to treat the data, since one-way ANOVA is a robust statistical test when sample sizes are similar and since the distribution of data also did not fulfill all obligations for use of the Kruskal-Wallis test with Dunn's multiple-comparison test.

For analyzing correlations, Pearson or Spearman's rank correlation analyses were used. The Spearman's rank correlation was used for correlations pertaining to binding. The level of statistical significance was set at a P value of ≤0.05.

RESULTS

Purification of skin and intestinal mucins.

Due to the highly O-glycosylated nature of mucins (70% to 90% carbohydrate content), they can be separated from other proteins on the basis of density. Extracts from the skin, pyloric ceca, and proximal and distal intestines of 5 Atlantic salmon were subjected to isopycnic density gradient centrifugation to separate the mucins from less-glycosylated proteins and lipids of lower density, as well as from nucleic acids of higher density. Periodate-oxidizable (carbohydrate) material was present as a distinct peak at 1.30 to 1.37 g/ml, representing the mucins. UV-absorbing material was found as a sharp peak in the higher density (1.45 to 1.52 g/ml) of the gradient, demonstrating that the mucins were successfully separated from the DNA (Fig. 1A and B). The mucin-containing gradient fractions were pooled to give one mucin sample for each individual and body site and provided the basis for the remainder of the analyses.

Mucins are large molecules that form complex networks by connecting the mucin subunits via disulfide bonds. The extraction procedure used in the present study first extracts the mucins that are soluble in the presence of a chaotropic agent (GuHCl), here called soluble mucins. The mucins remaining in the pellet are reduced and thereby brought into solution and are here referred to as insoluble mucins. In humans, the MUC2 mucin is mainly insoluble, whereas the MUC5AC mucin is mainly soluble (37). Mucus quantification on the basis of glycan quantity analysis of samples from the first fish revealed that the insoluble mucins in salmon were in the minority; 24% of the mucins in the skin and a lower proportion in the intestinal tract were insoluble. Both pyloric ceca and the proximal intestine contained 3% insoluble mucins, whereas the distal intestine contained 9% (Fig. 1C). Due to the small amount of insoluble mucins obtained from the isolates, the remaining analyses were focused on the soluble mucins. The insoluble mucins from this fish were included for comparison for the analyses shown below only where sufficient material was available to carry out the tests.

Mucin quantification.

The total mucin content per cm² tissue surface was higher in the proximal intestine than in the distal intestine (P < 0.001, n = 4; Fig. 2A). This interpretation of the results was supported by the finding that a higher number of goblet cells per area was present in the proximal intestine than in the distal intestine (P < 0.0001, n = 9; Fig. 2B). Images of periodic acid-Schiff (PAS)/alcian blue (which stains neutral mucins fuchsia and acidic mucins blue) revealed that the intestinal mucins were purple and thus carried a mix of charged (blue) structures and neutral (fuchsia) structures (Fig. 2C). The quantity of skin mucins appeared to be in a range similar to that seen with the distal intestine. However, for three reasons, the skin quantification can only be viewed as a rough estimation: (i) the sampling procedure may be less efficient for the skin than for other organs; (ii) the quantification was performed on the basis of glycan content, and the glycosylation of the skin is different from that of the intestine (see below); and (iii) the surface area used for the calculations was estimated (i.e., not measured as performed for the intestinal parts) from the body weight by a linear regression formula (30). Since measurement of surface area in the pyloric ceca is not feasible due to the high number of outpocketings in the organ (31), the mucins of the pyloric ceca were normalized to the weight of the organ (mean ± SEM = 148.9 ± 34.6 mg/g, n = 4).

FIG 2.

Mucin quantity in skin and intestine. (A) The relative mucin quantity of the proximal intestine was higher than that of distal intestine (t test, P < 0.0001, n = 4). The proximal and distal intestinal mucin quantities were normalized for the measured tissue area, whereas the skin mucin quantity was normalized for the estimated body surface calculated from body weight (30). The results are plotted as means ± SEM. (B) The number of goblet cells per epithelial mm² is higher in the proximal intestine than in the distal intestine (t test, P < 0.0001, n = 9). *** = P < 0.001; horizontal bars denote median values. (C) Images of tissue sections stained with PAS/alcian blue (which stains neutral mucins fuchsia and acidic mucins blue). Abbreviations: Proximal = proximal intestine; Distal = distal intestine. Each data point represents results from a biological replicate.

Density and glycosylation of Atlantic salmon mucins.

The densities of the mucins from the pyloric ceca were similar to the densities of the mucins from the proximal and distal intestine (1.34 to 1.35 g/ml), whereas the density of the skin mucins was lower (1.31 ± 0.01 g/ml, P < 0.001, n = 5; Fig. 3A), suggesting that the gastrointestinal mucins carry more carbohydrates. The densities of the GuHCl insoluble samples were similar to those of their soluble counterparts (Fig. 3A) but were excluded from the statistical comparison. Mass spectrometric analysis of released O-glycans revealed that the predominant glycan structure of the skin mucins had a lower molecular mass (m/z 513; Fig. 3B) than the main structures characteristic of the intestinal-tract mucins (m/z 716; Fig. 3C). The number of carbohydrate residues, derived from the structure composition (e.g., NeuAc₁N₁ in Fig. 3B), demonstrated that skin mucins carry short glycan chains consisting of two to five residues, with the majority of glycans built from only two residues (Fig. 3D). The glycan chains from the intestinal tract had a wider range of lengths, and the predominant structures contained three to five residues (Fig. 3D). Due to these differences, the average number of glycan residues in O-glycans released from the intestinal tract was higher than on the skin mucins (P < 0.001; Fig. 3E). There was a correlation between mucin density and the length of glycan chains expressed as the average number of carbohydrate residues on the mucins (r₁₈ = 0.65, P < 0.01; Fig. 3E). This suggests that the higher density of the mucins from the intestinal tract can be explained at least in part by the presence of longer glycan chains, although it does not exclude the possibility that the number of glycan substitutions could also be higher.

FIG 3.

Glycosylation and density of mucins from the skin and digestive tract. (A) The density of mucins isolated from the skin and the density of mucins isolated from the intestinal tract differed (one-way ANOVA with Newman-Keuls test, n = 5, P < 0.0001). The results from the soluble mucins are plotted as means ± SEM. * = P < 0.05 and *** = P < 0.001. The empty diamonds represent the insoluble mucins. (B and C) Representative LC-MS chromatograms of O-glycans released from mucins of skin (B) and distal intestine (C), in negative-ion mode. Major compositions and mass (m/z) are assigned in the chromatogram. NeuAc, N-acetylneuraminic acid; NeuGc, N-glycolyl neuraminic acid; H, hexose; N, N-acetylhexosamine. The numbers (e.g., NeuAc1N1) denote the numbers of monosaccharide residues in the defined compositions. (D) Size distribution of carbohydrate residues derived from compositions of O-glycans released from skin and intestinal mucins (data represent average results from the five fish). (E) The intestinal mucins have a higher average number of oligosaccharide residues than the skin mucins (one-way ANOVA and Newman-Keuls test, n = 5, P < 0.0001). *** = P < 0.001. (F) The mucin density correlated with the average length of glycan chains present on the protein backbone (Pearson r₁₈ = 0.65, P < 0.01). Abbreviations: Pyloric = pyloric ceca; Proximal = proximal intestine; Distal = distal intestine. The data points in panels A, E, and F represent results from biological replicates.

Binding of A. salmonicida to salmon mucins.

A. salmonicida (strain VI-88/09/03175) bound better to mucins from the intestinal tract than to mucins from the skin and the negative control (P < 0.001, n = 5; Fig. 4A). Mucins from the skin, pyloric ceca, and proximal intestine from the five fish bound the pathogen to similar degrees within each region, whereas mucins from the distal intestine seemed to differ more in their A. salmonicida binding ability (Fig. 4A). The results shown in the graphs represent strain VI-88/09/03175, and similar binding patterns were obtained with the A449 and 01-B526 strains (data not shown). Bacterial adhesion often occurs via the activity of lectin-like adhesins, binding to the glycans of the mucins (20). In the binding study performed, the concentration of the mucins was therefore set on the basis of the glycan content. In spite of this, A. salmonicida binding was still higher to the samples with higher density (Spearman r₁₈ = 0.65, P < 0.01; data not shown) and with longer glycan chains (Spearman r₁₈ = 0.67, P < 0.01; Fig. 4C). This indicates that either the length of the glycan chain is important for adhesion or a structure (or a group of structurally related epitopes) that is more abundant in longer glycan chains is a specific target for binding. Sialic acids were highly abundant glycan residues present on the mucins, but the numbers of all types of sialic acids per glycan chain in the samples did not correlate well with A. salmonicida binding (Spearman r₁₈ = 0.41, P = 0.07). However, the proportion of N-acetylneuraminic acid (NeuAc; the most abundant sialic acid present in the samples) correlated with the length of the glycan structures (Pearson r₁₈ = 0.78, P < 0.0001; Fig. 4B) and the average number of NeuAc structures per glycan chain in the samples (calculated from the abundance of non-, mono-, di-, and trisialylated O-glycans) correlated with A. salmonicida binding (Spearman r₁₈ = 0.55, P < 0.05; Fig. 4D). This correlation was higher within the pyloric cecal (Spearman r₃ = 0.9, P = 0.08; Fig. 4D) and proximal intestinal (Spearman r₃ = 1.0, P < 0.05) mucin groups. Together, these results indicate that NeuAc may be important for A. salmonicida binding.

FIG 4.

A. salmonicida adhesion to mucins from the skin and intestinal tract. (A) Adhesion of A. salmonicida to mucins from the different intestinal regions was higher than to the negative-control and skin mucins (P < 0.001, one-way ANOVA with Student-Newman-Keuls test). Different letters denote significant differences, symbols denote values from individual fish (n = 5; each represents the mean of the results from 5 technical replicates), and horizontal bars denote median values. (B) The N-acetylneuraminic acid (NeuAc) abundance correlated with the length of the glycan structures (Pearson r₁₈ = 0.78, P < 0.0001). (C) A. salmonicida adhesion correlated with the average length of the glycan chains (Spearman r₁₈ = 0.67, P < 0.01). (D) The number of NeuAc structures on the mucin glycans correlated with A. salmonicida binding (Spearman r₁₈ = 0.55, P < 0.05). This correlation was higher within the pyloric cecal (Spearman r₃ = 0.9, P = 0.08) and proximal intestinal (Spearman r₃ = 1, P < 0.05) mucin groups. Abbreviations: Pyloric = pyloric ceca; Proximal = proximal intestine; Distal = distal intestine. The data points of this figure represent the results from biological replicates; each data point in the binding analysis represents the mean of the results from five technical replicates.

Binding of A. salmonicida to sialylated and desialylated mucins.

The glycan chains of mucins from all organs studied had a high level of sialylation, and NeuAc was present on more than 50% of the glycans in each sample (Fig. 5A). The intestinal mucin glycans carried more NeuAc (mean_{pyloric cecum}, 78%; mean_{proximal intestine}, 88%; mean_{distal intestine}, 89%) than the skin mucin glycans (mean, 58%; P < 0.01; Fig. 5A). The level of NeuAc (Fig. 5A) had a pattern similar to the A. salmonicida binding pattern (compare with Fig. 4A and D). Highly sialylated rat colonic mucins bound more A. salmonicida than the less sialylated rhesus monkey gastric mucins (Fig. 5B). This supports the idea that adhesion of A. salmonicida to mucins is influenced by their abundance in sialic acid residues.

FIG 5.

Atlantic salmon mucin glycan sialylation and A. salmonicida binding to differentially sialylated mucins. (A) The levels of NeuAc differed between skin, pyloric ceca, proximal intestine, and distal intestine (one-way ANOVA and Newman-Keuls test, n = 5, * = P < 0.05, ** = P < 0.01, *** = P < 0.001). The level of NeuAc is expressed as the percentage of the total glycan quantity. The data points represent the results from biological replicates. Abbreviations: Pyloric = pyloric ceca; Proximal = proximal intestine; Distal = distal intestine. (B) Binding of A. salmonicida was stronger to highly sialylated rat colonic mucin (HS) than to rhesus monkey gastric mucin (LS) with less sialylation. Binding of A. salmonicida to both mucins was statistically significant (one-way ANOVA and Newman-Keuls test, P < 0.0001; *** = P < 0.001). Data points represent results from technical replicates. The results are plotted as means ± SEM. (C) Effects of desialylation on A. salmonicida binding. Four hours of treatment with sialidase A (a neuraminidase) deprived skin mucins of their ability to bind A. salmonicida (P < 0.01). Thirteen hours of sialidase A digestion reduced A. salmonicida binding to mucins from pyloric ceca (results not statistically significant [n.s.]) and from proximal intestine (P < 0.05) and distal intestine (P < 0.05, paired t test). The presented data denote results from biological replicates (n = 5). Abbreviations: Pyloric = pyloric ceca; Proximal = proximal intestine; Distal = distal intestine.

Elimination of terminal sialic acid residues and branched sialic acids with sialidase A (which releases α2,3-, α2,6-, α2,8-, and α2,9-linked NeuAc) from the mucin samples reduced the adhesion ability of A. salmonicida (Fig. 5C). A 4-h-long enzymatic digestion was sufficient to completely eliminate binding of A. salmonicida to the skin mucins (n = 5, P < 0.01; Fig. 5C). Extended (13-h) enzymatic digestion of the samples reduced the level of A. salmonicida binding to pyloric cecal mucins, proximal intestinal mucins, and distal intestinal mucins by 42%, 42%, and 61%, respectively (P < 0.05 for the proximal and distal intestine; Fig. 5C). Mass spectrometric analysis of the sialidase A-treated distal intestinal mucins demonstrated that the treatment was effective; the levels of predominant sialylated structures (m/z 513, 716, 878, and 1,081) decreased by 90%, and the relative levels of neutral structures (m/z 425, 587, and 790) increased.

DISCUSSION

In the present study, we demonstrated that mucins from the skin and intestinal tract of Atlantic salmon have different biochemical characteristics, which in turn affect their binding of the pathogen A. salmonicida. The density of the skin mucins was lower, and skin mucin glycans contained fewer residues (i.e., were shorter and less branched) than intestinal-tract mucin glycans. A. salmonicida bound to the intestinal-tract mucins to a greater extent than to the skin mucins; the binding correlated well with the level of NeuAc in the mucins, which was higher in the intestinal regions than in the skin. Desialylation of mucins from the intestinal tract decreased binding of A. salmonicida, while desialylation of skin mucins completely abolished binding. Consistent with this, A. salmonicida also bound better to mammalian mucins with high levels of sialylation. Thus, sialylated structures, and, in particular, the ones containing NeuAc, are important for A. salmonicida binding.

Important biological roles of fish mucus include osmoregulation, respiration, excretion, and immune function (38). These properties of mucus have mainly been ascribed to the mucin molecules. However, Atlantic salmon mucins had not been isolated previously, and the knowledge of their biochemical characteristics in this species is scarce and mainly based on histological staining and lectin binding techniques (39–42). We used isopycnic density gradient centrifugation to isolate mucins on the basis of their characteristic densities. Historically, mucins have been isolated by density gradient centrifugation or by size exclusion chromatography or by a combination of the two methods to remove mucin fragments and molecules of smaller size than the large mucins. However, over the last decade, several smaller mucins have been identified, and we therefore refrained from the latter step. We found a high proportion of GuHCl-soluble mucins in mucus isolated from all investigated tissues of the Atlantic salmon. Mucins are large molecules that form complex networks by connecting the mucin subunits via disulfide bonds. Although not fully proven in any species, it is reasonable to expect that the insoluble mucins represent larger complexes and that altered secretion and posttranslational modification of mucins can result in changes in solubility (43). The presence of both soluble and insoluble mucins may also suggest the existence of different mucin species; e.g., in humans, intestinally derived MUC2 is mainly insoluble whereas gastric MUC5AC and MUC6 are mostly soluble (44, 45). Transcriptome analyses suggest the existence of some of these mucin forms also in fish; MUC2 and a MUC2-like mucin mRNA have been found in the gastrointestinal tract of sea bass (46), and isotigs that exhibited homology to the mammalian mucins MUC2, MUC5AC and MUC5B were found in Atlantic salmon skin (28).

The quantification of isolated mucins can be a technical challenge. Mucins are notoriously difficult to get back into solution after freeze-drying, resulting in potentially selective loss of material. Therefore, we instead determined the concentration on the basis of carbohydrate content as periodate-oxidizable structures using serial dilution of the samples and a standard curve of a fusion protein containing the MUC1 mucin. Although this is not an exact measure of concentration, it can be used to ensure that the mucins are at the same concentration for comparative assays, and since bacterium-mucin interactions largely occur via the mucin glycans (20), setting the concentration on the basis of the glycan content appears most appropriate. Mucin quantification of samples isolated from the proximal and distal intestines of Atlantic salmon revealed that the proximal intestine contains amounts of mucus severalfold larger than those contained in the distal intestine. Although the quantification is based on the glycan content and does not result in an absolute value, both the densities and the levels of glycosylation of the mucins from these sites were similar, and our histological analysis also showed that the proximal intestine harbors higher mucin and goblet cell content than the distal intestine. An estimation of the skin mucin quantity suggested that the quantity of mucins from skin per surface area was similar to that from the distal intestine. However, there are several aspects that make the comparison with the intestinal sites less certain, and the data presented for the skin are most likely an underestimation. First, in the intestinal sites, mucins were isolated both from secreted mucus and from the mucosal tissue containing nonsecreted mucins within goblet cells; for the skin, only the surface was sampled, which may have left parts of the goblet cells and mucus behind as these can be under and around the scales (47). Sampling of fish was performed by netting, which may also have affected the amount of mucus left on the skin surface. Furthermore, the surface area of the skin was calculated on the basis of the length of the fish according to the formula by Tucker et al. (30), which may not be fully representative for all Atlantic salmon. Lastly, the quantification was performed on the basis of the glycan content and the density and glycosylation of the skin mucins was lower than for the intestinal mucins.

We found that the Atlantic salmon intestinal mucins had a lower density than the mucins previously found in the gills of the rainbow trout (48) and also lower than gastrointestinal-tract mucins from humans and other mammals (45, 49, 50) and that the mucins isolated from the salmon skin had an even lower density. Glycan chains present on the protein backbone of mucins are usually responsible for 70% to 90% of the mucin mass, and we found that mucin glycan length correlated well with mucin density; i.e., mucins with a low number of O-glycan residues had a low density, with a notable difference between the skin and intestinal mucins. In the intestine, the goblet cells were uniformly purple after PAS/alcian blue staining of histology sections, indicating that both neutral and acidic mucin glycans are present in all goblet cells of the intestine. In contrast, the brown trout and rainbow trout gills contain goblet cells that stain either pink or purple (39). Thus, certain cells in the gills make neutral mucins and others make a mixture of neutral and acidic ones. In the present study, mass spectrometry demonstrated that the acidic character of the mucins mainly derives from the presence of NeuAc on the majority of the glycan chains. Interestingly, mucin glycosylation appears to be altered by both salinity and disease state; AB/PAS-stained histological sections demonstrated a switch toward a more acidic mucus composition in salt water than in fresh water, whereas salmonids infected with amoebic gill disease shifted the mucus composition to a more neutral state (39). This indicates that mucus is responsive both to environment and to pathogen exposure, probably also resulting in a change in the capacity of the mucins to bind the pathogens.

Mucin oligosaccharides are central elements in the host-pathogen interaction due to the molecular charge they confer (39, 51), the energy source they provide (52), and their involvement in specific recognition and binding phenomena (20, 37). In the mammalian stomach, mucin binding to a pathogen aids in removing the bacteria from the gastric niche (24, 53). Moreover, mucins may act as binding decoys for the more intimate and disease-causing adhesion to similar glycans on the epithelial cell surface for other pathogens in the gastrointestinal tract. However, in external sites such as the skin and gills where there is constant convection caused by the water washing possible pathogens away, mucins with an adhesion site which can bind pathogens could instead aid the infection, although no conclusive evidence for this exists. In the present study, adhesion of A. salmonicida to skin mucins was lower than to the mucins of the intestinal tract, suggesting a difference between the defense mechanisms inside the intestinal tract and on the outer mucosal surface.

It may be that, although adherence and subsequent dissemination may be beneficial in the fish intestinal tract, the skin might have to be less adherent to ensure that pathogens in the water do not stick to the mucus and develop infection. Although the binding specificity of other salmon-associated bacteria is unknown, the skin has indeed a simpler glycan structure repertoire, different flora, and considerably lower bacterial counts than the intestine (54).

Our results demonstrated a positive relationship between the relative amount of NeuAc residues and the number of residues in the glycan chains of Atlantic salmon mucins. Thus, in contrast to the short mucin glycan structures from the skin, which mainly comprise nonsialylated and monosialylated structures, the glycans of the intestine both are more sialylated per se and have longer chains with more-complex sialylation, with up to three NeuAc residues per structure. The longer glycan chains and the larger amounts of NeuAc in the intestinal samples provide more binding sites on intestinal mucins for A. salmonicida. However, attempts to inhibit mucin binding by pretreating A. salmonicida with soluble NeuAc failed to reduce the binding. This indicates that the host epitope recognized by A. salmonicida is a more complex structure that includes NeuAc rather than consisting of NeuAc alone. A similar requirement of more-complex structures for bacterial binding has been demonstrated previously in Helicobacter pylori binding to gastric mucins; binding was inhibited by a pentasaccharide terminating in Leb, coupled to provide multivalency, whereas shorter structures, or fucose alone, failed to inhibit binding (55). Following sialidase A treatment, we found decreased adhesion of A. salmonicida to Atlantic salmon mucins. Sialidase A releases α(2-3)-, α(2-6)-, α(2-8)-, and α(2-9)-linked NeuAc from complex carbohydrates. Although desialylation totally abolished the binding of A. salmonicida to skin mucins but decreased the binding to the intestinal sites by only 42%, the most likely reasons for the lesser reduction in binding to mucins from the intestinal tract is that the level of binding to skin mucins was lower to start with, and the desialylation of the intestinal mucins was most likely less complete. The skin mucin glycans are shorter (consisting mainly of only one sialic acid attached to GalNAc-O-Ser/Thr [N-acetylgalactosamine-O-serine/threonine]); therefore, the sialidase can remove these terminal sialic acids. In contrast, the glycans from the intestine contain both terminal and internal sialic acids and also contain multisialylated structures; indeed, 10% of the NeuAc-containing structures remained after desialylation of the distal intestinal mucins. It is also possible that desialylation exposes a new glycan epitope that A. salmonicida can bind to which partially counteracts the decrease in binding due to loss of the NeuAc epitope. The marked reduction in bacterial binding ability after sialidase A digestion further supports the idea of the importance of sialic acid residues in mucin glycan structures in mucosal defense against furunculosis disease. From an evolutionary perspective, it is interesting that A. salmonicida appears to selectively bind to NeuAc, which is most common in the intestine, whereas NeuGc or deaminated neuraminic acid (Kdn) is almost solely present in the skin mucins. However, we cannot conclude from the current study whether the salmon may present other sialic acid structures on the skin surface due to selective pressure to bind less A. salmonicida or whether the bacteria may have evolved adhesins recognizing the most common glycan structure present in the salmon. Furunculosis disease causes both internal (e.g., gastroenteritis and internal hemorrhage) and external (e.g., tail rot and skin lesion pathology) disease symptoms, but the pathomechanism and the proportion and severity of lesions in the different sites in Atlantic salmon have not yet been clearly defined. However, furunculosis-resistant salmonids harbor a lower number of living A. salmonicida bacteria in their skin mucus than less-resistant ones (56). In vivo, a range of additional factors, such as the resident commensal flora, mucus turnover and degradation, other components harbored in the mucus, water flow, salinity, and temperature, may also affect A. salmonicida adhesion.

Adhesins conferring binding to sialic acids on mucins are known in some bacterial species, e.g., Helicobacter pylori (37), but there is no information of such in A. salmonicida. The A. salmonicida S-layer and type I pilus system have previously been shown to contribute to A. salmonicida adhesion to the salmon intestine ex vivo (57).

In conclusion, we have shown that purified mucins from the skin, pyloric ceca, and proximal and distal intestines of healthy fish have distinctly different characteristics. The complexity of the glycan epitopes and sialylation seem to be key factors in the host-pathogen interaction between the Atlantic salmon and A. salmonicida. The marked differences between the skin and intestinal tract in NeuAc levels as well as in the levels of A. salmonicida binding suggest interorgan differences in the host-pathogen interaction and in the mucin-based defense against A. salmonicida. The knowledge obtained in this study, together with further research on mucin interactions with pathogens such as A. salmonicida and other aeromonads, may enable the development of alternative prophylactic or intervention strategies to control outbreaks in aquaculture.