In recent years, dietary modulation of the intestinal microbial composition and activity has been the subject of intense research efforts, as evidence has emerged that targeted modulation of colonic bacterial metabolism may lead to health effects( Reference De Preter, Hamer and Windey 1 – Reference Kau, Ahern and Griffin 3 ). Proteins and carbohydrates are two major chemical components of the diet known to affect intestinal microbial metabolism in pigs.
Dietary fermentable carbohydrates (fCHO) include diverse physico-chemical forms such as dietary fibre and resistant starch that escape digestion and absorption and serve as substrates for microbial fermentation yielding lactic acid or SCFA such as acetic, propionic and butyric acids. SCFA are considered to be beneficial to the host intestine as they provide energy for epithelial cells, inhibiting potential pathogen growth, stimulating epithelial proliferation, facilitating tight junction formation and inhibiting inflammation and genotoxicity( Reference De Preter, Hamer and Windey 1 , Reference Blaut and Clavel 4 , Reference Lupton 5 ). Indigestible protein that reaches the large intestine together with endogenous secretions from the proximal parts of the intestine can serve as substrate for microbial fermentation and can thus be called fermentable crude protein (fCP). Microbial breakdown of proteins may yield branched-chain fatty acids, ammonia, biogenic amines, hydrogen sulphide, and phenolic and indolic compounds( Reference Blaut and Clavel 4 , Reference Ball and Aherne 6 – Reference Russell, Gratz and Duncan 8 ). Many of these fermentation products are considered to be toxic and believed to exert negative effects on the intestinal mucosa( Reference Blaut and Clavel 4 , Reference McGarr, Ridlon and Hylemon 9 ) by interfering with inflammatory responses( Reference Maslowski, Vieira and Ng 10 ) or by directly affecting tight junctions, ion transport processes( Reference Bordin, D'Atri and Guillemot 11 – Reference Zeissig, Fromm and Mankertz 13 ) or their interplay( Reference Amasheh, Milatz and Krug 14 ). These responses are known to affect intestinal barrier function and may predispose to inflammatory bowel disease( Reference Capaldo and Nusrat 15 – Reference Prasad, Mingrino and Kaukinen 18 ) and cancer( Reference Lupton 5 ) in humans and to diarrhoea and enteric infection in pigs( Reference Ball and Aherne 6 , Reference Elkouby-Naor and Ben-Yosef 19 – Reference Jeaurond, Rademacher and Pluske 22 ).
Dietary inclusion of fCHO may alleviate the negative effects of protein fermentation. For example, dietary inclusion of fCHO has been shown to effectively modulate intestinal microbial metabolism to increase SCFA content( Reference Hermes, Molist and Ywazaki 20 , Reference Bikker, Dirkzwager and Fledderus 23 ), decrease abundance of protein fermentation products( Reference Jeaurond, Rademacher and Pluske 22 ), and modulate susceptibility to pathogen colonisation and abundance of harmful bacteria in the gut( Reference Pieper, Bindelle and Rossnagel 24 , Reference Metzler-Zebeli, Hooda and Pieper 25 ). However, most of the studies have focused on the luminal environment of the intestine without investigating the effects on the colonic mucosa, particularly on epithelial barrier function. Using the pig model, several studies have reported that there is no change in small intestinal( Reference Bikker, Dirkzwager and Fledderus 23 , Reference Pieper, Bindelle and Rossnagel 24 , Reference Heo, Kim and Hansen 26 ) or colonic morphology( Reference Jeaurond, Rademacher and Pluske 22 ) in response to high-protein diets. Recently, we determined the influence of fCP and fCHO on the large intestinal microbial ecology and the mucosal expression of markers of cell turnover, mucin and inflammatory cytokines( Reference Pieper, Kröger and Richter 27 ). We found that whereas fCHO partially reduced the formation of putatively toxic metabolites such as ammonia and amines and increased clostridial counts, activation of genes associated with immune response, such as IL-6 and IL-1β was increased with fCP irrespective of fCHO inclusion( Reference Pieper, Kröger and Richter 27 ).
The present study aimed to determine the effect of fCP and fCHO and their interaction on the barrier function of proximal and distal colonic mucosae in weaned piglets. We hypothesised that (1) diets high in fCP will alter the functions of colonic mucosa, particularly the barrier function of mucosal epithelium, and (2) this can be reversed by dietary inclusion of fCHO.
Materials and methods
All procedures involving animal handling and treatments were approved by the local state office of occupational health and technical safety (LaGeSo Registration no. 0249/10).
Animals, diets and sampling
A total of thirty-two weaned piglets (Euroc × Piétrain) were placed in pens balancing for body weight, litter and sex. The piglets were randomly assigned to one of four dietary treatment groups in a 2 × 2 factorial experiment as specified in detail by Pieper et al. ( Reference Pieper, Kröger and Richter 27 ). Piglets in each treatment group were fed one of the four following diets: low-fCP/low-fCHO (14·5 % CP/14·5 % total dietary fibre); low-fCP/high-fCHO (14·8 % CP/16·6 % total dietary fibre); high-fCP/low-fCHO (19·8 % CP/14·5 % total dietary fibre); high-fCP/high-fCHO (20·1 % CP/18·0 % total dietary fibre). The diets were formulated to meet or exceed the nutrient requirements of weanling piglets( 28 ) (Table S1, available online). To increase the flow of indigestible protein and thus fermentable protein into the hindgut, the soyabean meal was autoclaved at 124°C for 20 min for groups being fed the high-fCP diets( Reference Pieper, Kröger and Richter 27 ). The piglets were killed after 20–23 experimental days (48–51 d of age) for sampling of large intestinal tissue. At 20 cm from the caecum, an approximately 20 cm segment was taken to represent the proximal colon. Correspondingly, a 20 cm segment was prepared ending 20 cm before the rectum, and it represented the distal colon. Colonic segments were cut open along the mesenteric border, rinsed with cold saline and transported to the laboratory in oxygenated cold saline solution (4–7°C; 0·9 % NaCl+1 mmol/l CaCl2). Before the experiments, mucosae were stripped of the serosa, and fractions were snap-frozen in liquid N2 and stored at − 80°C for subsequent analysis.
Electrophysiological measurements
The stripped mucosae were mounted into Ussing chambers with an effective area of 0·28 cm2. These were driven by an eight-channel computer-controlled voltage clamp device (Fiebig) as described previously( Reference Kreusel, Fromm and Schulzke 29 ). The bath solution contained the following components: 113·6 mm-NaCl; 2·4 mm-Na2HPO4; 0·6 mm-NaH2PO4; 21 mm-NaHCO3; 5·4 mm-KCl; 1·2 mm-CaCl2; 1 mm-MgSO4; 10 mm-d(+)-glucose; 0·5 mm-β-OH-butyrate; 2·5 mm-glutamine.
Antibiotics (azlocillin (50 mg/l) and tobramycin (4 mg/l)) were used to prevent bacterial growth, and they had no effect on short-circuit current (I SC in μA/cm2). The solution was gassed and mixed using a bubble lift (95 % O2 and 5 % CO2, pH 7·4). The temperature was kept constant at 37 °C. Each side of the tissue samples was perfused with 5 ml of bathing solution, and bovine serum albumin (final concentration 0·01 %) was routinely added and I SC and transmucosal resistance (R t in Ω cm2) were continuously recorded. The resistance of the bathing solution and electrode offsets were determined and subtracted from raw data before each experiment.
Impedance spectroscopy
One-path impedance spectroscopy was carried out as described previously( Reference Günzel, Zakrzewski and Schmid 30 ). The technique differentiates the epithelial (R epi) and subepithelial (R sub) portions of the transmucosal wall resistance (R t) based on the three-parameter model of the colonic wall. In this model, the epithelium is described as an electrical equivalent circuit by a resistor and a capacitor in parallel and the subepithelium by a resistor in series. After application of forty-eight discrete frequencies of an effective sine-wave alternating current of 35 μA/cm2, ranging from 1·3 Hz to 65 kHz, changes in tissue voltage were detected using phase-sensitive amplifiers (1250 frequency response analyser and 1286 electrochemical interface, Solartron Schlumberger). Complex impedance values were calculated and corrected for the resistance of the bath solution and the frequency behaviour of the measuring device. Then, for each tissue sample, the impedance locus was plotted in a Nyquist diagram and a circle segment was fitted by least-squares analysis. Due to the frequency-dependent electrical characteristics of the capacitor, R t was obtained at low frequencies, whereas R sub was obtained at high frequencies. R epi was obtained from R epi= R t− R sub. Impedance spectroscopy was carried out 1 h after mounting the tissue samples.
Activity of the epithelial sodium channel
Specimens of distal colonic mucosae were mounted in Ussing chambers and subsequently incubated with 3 × 10− 9mol/l aldosterone on both sides. The epithelial Na channel (ENaC)-dependent Na flux was determined as the drop in I SC 10 min after addition of amiloride (10− 4mol/l) in μA/cm2 or as the flux of monovalent cations (μmol/h per cm2)(31).
Permeability to macromolecules
Macromolecule permeability to horseradish peroxidase (HRP) and fluorescein isothiocyanate (FITC)–dextran 4000 (TdB Consultancy) from the mucosal to the serosal compartment was determined during a 2 h period in Ussing chambers. Briefly, 2 mg/ml of both tracer molecules were added to the mucosal compartment and serosal HRP peroxidase activity was measured using the QuantaBlu Kit (Pierce/Thermo Scientific). In serosal samples from the same flux experiments, HRP (44 kDa) was analysed after electrophoretic separation followed by Western blotting using an antibody specific for HRP (Table 1). The transmucosal passage of HRP was further characterised by quantifying the HRP stainings of mucosal specimens (described in the ‘Staining of tissues and imaging’ section). FITC–dextran fluorescence was measured using a plate reader (TECAN 200M, Tecan). For HRP active enzyme and FITC–dextran, apparent permeabilities were calculated from (dQ/dt)/(A× C o), where dQ/dt is the cumulative amount of tracer compound appearing in the receiver compartment v. time, A is the area tissue and C o is the initial concentration of the tracer in the donor compartment.
Antibodies used for Western blotting and immunofluorescence
IF, immunofluorescence; WB, Western blotting; HRP, horseradish peroxidase; POD, peroxidase.
Dilution potentials
Following flux assays, dilution potential measurements were of changes in permeabilities to cations and anions. Dilution potentials were measured using a modified bath solution on either the apical or the basolateral side of the epithelium. In the modified bath solution, 50 % of NaCl was iso-osmotically replaced by mannitol for the determination of Na+ and Cl– permeabilities. Ion permeabilities were calculated by means of the Goldman–Hodgkin–Katz equation, as described by Günzel et al. ( Reference Günzel, Stuiver and Kausalya 32 ).
Surface biotinylation
After completion of macromolecular fluxes and dilution potential measurements in the Ussing chambers, Sulfo-N-hydroxysuccinimide (NHS)-biotin (443 Da, final concentration 5 mmol/l; Thermo Scientific) was added to the mucosal compartment (10 min; 37°C) to assess permeation into the subepithelial compartment. The tissue samples were then briefly rinsed and fixed with 2 % paraformaldehyde (1 h; room temperature) for further analysis.
Staining of tissues and imaging
Mucosal specimens from the Ussing chamber experiments were fixed with paraformaldehyde (2 %) for 2 h followed by glycine treatment (25 mmol/l, 15 min at room temperature) and stepwise dehydration (10 % sucrose and 20 % sucrose for 1 h, 4°C each, followed by dehydration with 30 % sucrose overnight at 4°C). The tissue samples were embedded in Tissue Tek (Sakura Finetek), and blocks were stored at − 80°C. Cryosections (six per piglet and segment, >50 μm apart, 10 μm) were prepared using a Leica CM1900 cryostat (Leica Microsystems). For staining, sections were permeabilised with 0·025 % Triton X-100 (10 min; room temperature). For all the subsequent washing steps and for dilution of antibodies, tissue samples were blocked in PBS containing 6 % (v/v) goat serum and 1 % BSA for 60 min at room temperature. Immunostaining for HRP was carried out using Cy3–anti-HRP antibodies (Table 1; 120 min at room temperature); and epithelial cells were contrasted using Alexa 647–anti-E-cadherin (Table 1). Labelling of biotin (for NHS-biotinylated structures) was achieved using Dy549-labelled streptavidin (1:200; Dyomics), and 647-anti-E-cadherin was used for counterstaining. 4′,6-diamidin-2-phenylindol (DAPI) staining (diluted 1:2500 in PBS; room temperature for 10 min) was carried out shortly before mounting the sections using ProTaqs MountFluor (Biocyc GmbH). Fluorescence images were obtained using a confocal laser scanning microscope (LSM 510 META, Carl Zeiss MicroImaging GmbH). Relative quantification of subepithelial HRP or NHS-biotin signals was carried out using the ImageJ software package( 33 ). Analysis was carried out for one representative slice (out of six – described above) per segment and piglet. Briefly, images were despeckled and 5 × 5 mean-filtered, and ImageJ's wand tool was used to encircle the subepithelial area (E-cadherin-negative area). DAPI-negative areas were generally excluded. In the remaining area, representing the subepithelial compartment, mean intensity for HRP signal (or biotin signal) was calculated from integrated density/measured area.
Western blotting
The expression of tight junction proteins was determined by Western blotting using the membrane extracts of stripped mucosal specimens. Briefly, mucosal specimens were homogenised using Ultra Turrax (IKA) and sonicated in an iced lysate buffer containing 250 mm-sucrose, 10 mm-triethanolamine and protease inhibitor cocktail (Complete Mini EDTA-free; Roche Diagnostics GmbH). After centrifugation at 1000 g for 10 min at 4°C, the supernatant was centrifuged at 20 000 g for 45 min at 4°C. The membrane fraction was obtained by resuspending the resulting pellet in a buffer containing 150 mm-NaCl, 50 mm-Tris–HCL (pH 7·4), 2 mm-EDTA and 1 % nonyl phenoxypolyethoxylethanol by consecutive passages through 26G × 3/8 needles. Protein concentrations were determined by Pierce bicinchoninic acid assay (Thermo Scientific), and equal amounts of proteins (15 μg of each sample) were loaded on gels, separated by PAGE and transferred onto polyscreen polyvinylidene difluoride transfer membranes (PerkinElmer). Blots were incubated with primary antibodies overnight at 4°C (Table 1), followed by incubation with peroxidase-conjugated secondary antibodies (Table 1). Using the Lumi-LightPLUS Western Blotting Kit (Roche Diagnostics GmbH), chemiluminescence signals were analysed using a FX-7 imaging system (Vilber Lourmat Deutschland GmbH) and the AIDA software (Raytest) or ImageJ( 33 ). For a given protein, densitometric values of single samples were normalised on the arithmetic mean of all the samples, thus representing relative abundance between the groups. HRP was analysed the same way using samples from the Ussing chamber experiments.
Statistical analysis
The results are given as means with their standard errors. Data were analysed using generalised linear model procedures in Statistical Package for the Social Sciences (version 18.0, SPSS, Inc.) with fCHO and fCP and their interaction as sources of variation. Values identified using Grubbs' test for outliers were excluded from the analysis. Differences at P <0·05 were considered significant, and P values less than 0·1 were identified as trends. One-way ANOVA with post hoc test for group comparison (Tukey's honestly significant difference) was only carried out when a significant interaction of fCHO and fCP was observed. Otherwise, only the main effects of the two factors were analysed. Pearson's correlation coefficient (r) was used as a measure of linear association between two variables.
Results
All the piglets remained healthy throughout the study. Data on nutrient digestibility, performance, faecal scores, large intestinal microbial ecology, mucosal oxidative stress measurements and gene expression have been published recently( Reference Pieper, Kröger and Richter 27 ).
Barrier function to macromolecules was largely unaffected
The mucosal-to-serosal permeation of macromolecules of different sizes, namely NHS-biotin (443 Da), FITC–dextran (4 kDa) and HRP (44 kDa), is summarised in Table 2. The permeability of the distal colon to HRP was lower (P <0·05) in the high-fCHO diet-fed piglets when measured as an active enzyme, but it was found to be unaffected by treatment when immunoreactive HRP protein was analysed by Western blotting. The abundance of HRP within the colonic mucosa was measured as immunoreactivity in the subepithelial compartment, and it was found to be lower (P <0·01) in the distal colon of the high-fCP diet-fed piglets. The permeability of the proximal colon to FITC–dextran tended to be higher (P =0·09) in the high-fCP diet-fed groups, whereas in the distal colon, a trend (P =0·07) towards a fCP × fCHO interaction was observed, indicating that the high permeabilities observed with the high-fCP/low-fCHO diet were decreased by the inclusion of high fCHO. The permeation of NHS-biotin tended to be increased also in the proximal colon of the high-fCP diet-fed groups (P =0·088), and it was reduced in the distal colon of the high-fCHO diet-fed groups (P <0·05).
Passage of markers of the paracellular (N-hydroxysuccinimide (NHS)-biotin and FITC-dextran 4000) and transcytotic (HRP, horseradish peroxidase) pathways (Mean values with their standard errors, n 6–8 per group)
fCP, fermentable crude protein; fCHO, fermentable carbohydrates; PFD4, permeability to fluorescein isothiocyanate–dextran 4000; PHRP/enzyme, permeability to HRP measured as enzymatic activity in the serosal compartment; serosal HRPIF, immunofluorescence signal of HRP in the submucosa; ALU, arbitrary light units; serosal HRPWB, relative amount of HRP protein in Ussing chamber serosal compartment measured by Western blotting; serosal NHS-biotin, fluorescence signal of biotin in the submucosa.
* Probability of the main effects of fCHO and fCP as well as their interaction (fCHO × fCP).
Barrier function to inorganic ions was unaffected by diets
The basal short-circuit current (I SC) of colonic mucosae was not altered by the different diets (Table 3). R t was unaffected in the proximal colon, but it tended to be higher in the distal colon of the high-fCP diet-fed piglets (P =0·06). The contributions of R epi and R sub to R t also did not differ, and only a tendency for an interaction was observed for R epi in the distal colon (P =0·09). Dilution potentials for NaCl were measured to detect changes in charge selectivity of tight junctions. The ratio of permeability to Na:permeability to chloride (PNa/PCl) tended to be decreased in high-fCP groups (P =0·09) in the proximal colon, but it was not altered in the distal colon.
Electrophysiological parameters (Mean values with their standard errors, n 6–8 per group)
fCP, fermentable crude protein; fCHO, fermentable carbohydrates; ENaC, epithelial Na channel.
* Probability of the main effects of fCHO and fCP as well as their interaction (fCHO × fCP).
† Parameters at 3 h of incubation.
‡ Parameters at 1 h of incubation.
Epithelial sodium channel activity was reduced in the high-fermentable crude protein diet-fed groups
ENaC activity after ex vivo stimulation with nanomolar aldosterone was measured in distal colonic mucosae. The high-fCP diets suppressed inducible ENaC activity by approximately 40 % (P <0·01; Table 3), whereas the inclusion of fCHO had no effect.
High-fermentable crude protein diets altered tight junction protein expression
The high-fCP diets decreased (P< 0·05) the expression of all claudins in the proximal and distal colons, except that of claudin-4, which was unaffected in the proximal colon. The expression of claudin-2 was most affected, demonstrating approximately 68 % reduction in the proximal colon (Table 4). Indeed, there was a tendency towards decreased PNa/PCl in the high-fCP diet-fed groups (P =0·09), which is consistent with a claudin-2-mediated selectivity preference of cations over anions.
Expression of tight junction proteins as determined by Western blotting* (Mean values with their standard errors, n 7–8 per group)
fCP, fermentable crude protein; fCHO, fermentable carbohydrates.
a,b,c,dMean values within a row with unlike superscript letters were significantly different (P <0·05).
* Relative amount of protein.
† Probability of the main effects of fCHO and fCP as well as their interaction (fCHO × fCP).
The effect of feeding fCHO was less consistent than that of feeding fCP; however, the high-fCHO diets also decreased (P <0·01) the expression of claudin-2 in the distal colon. In the proximal colon, a significant (P< 0·01) fCP × fCHO interaction was observed, indicating a reduction of the expression of claudin-1 by fCHO diets when fCP content was low, but not when it was high. In contrast to the fCP diets, the high-fCHO diets increased (P <0·05) the expression of claudin-4 in both the proximal and distal colons. The expression of both claudin-3 (P =0·07) and claudin-4 (P =0·08) in the proximal colon exhibited a trend towards a significant interaction where the high-fCHO diets appeared to increase the expression only when fCP content was low. Similarly, the expression of tricellulin in the proximal colon was markedly increased by the high-fCHO diets in a low-fCP background (P <0·01), but it remained unaltered by fCHO diets with the inclusion of high fCP (fCP × fCHO; P< 0·05).
The abundance of the tight junction protein occludin was examined by analysing the prominent Western blot bands present at higher molecular weights (59, 65 and 69 kDa) and at a low molecular weight (approximately 32 kDa; Table 5 and Supplementary Fig. 1 (available online)). While total occludin expression was unaffected, analysis of the protein abundance of occludin forms indicated significant (P <0·05) interactive effects. In the proximal and distal colons, the levels of the 69 kDa high molecular weight occludin were increased by the high-fCHO diets when fCP content was low, but not when it was high. A similar pattern was observed for the 59 kDa high molecular weight occludin in the distal colon. Interestingly, for the 32 kDa low molecular weight occludin, protein abundance was lowest in the high-fCP/low-fCHO diet-fed group.
Expression of occludin bands as determined by Western blotting* (Mean values with their standard errors, n 7–8 per group)
fCP, fermentable crude protein; fCHO, fermentable carbohydrates; HMW, high molecular weight; LMW, low molecular weight.
a,bMean values within a row with unlike superscript letters were significantly different (P <0·05).
* Relative amount of protein.
† Probability of the main effects of fCHO and fCP as well as their interaction (fCHO × fCP).
Discussion
In the present study, the hypothesis that (1) diets high in fCP will alter the functions of colonic mucosa, particularly the barrier function of mucosal epithelium, and (2) this can be reversed by the inclusion of fCHO was investigated. We found that barrier function to macromolecules was only marginally altered and electrophysiological measurements of epithelial function, short-circuit current and transmucosal resistance were unaffected by the diets. At the same time, colonic epithelial cells reacted strongly to the imposed alteration of luminal milieu as demonstrated by (1) the decreased capacity of ENaC-mediated Na absorption (and probably the coupled water absorption) and (2) the markedly altered expression of tight junction proteins.
A companion study on the same animals demonstrated that the luminal microbial and metabolic milieu was profoundly altered by fCHO/fCP variation( Reference Pieper, Kröger and Richter 27 ), but an overt phenotype with regard to health and performance measurements of pigs was absent. Nevertheless, a host response was clearly observed: the mucosal response to high-fCP diets included changes in the mucosal cytokine mRNA abundance profile: up-regulation of the expression of putative pro-inflammatory cytokines (such as IL-6 and IL-1β) as well as anti-inflammatory cytokines (IL-10 and transforming growth factor-β)( Reference Pieper, Kröger and Richter 27 ) in the absence of overt inflammation.
Barrier function to macromolecules
The passage of macromolecules was studied with tracer molecules spanning three orders of magnitude in molecular weight (approximately 400–40 000 Da), since permeability to bigger solutes is not necessarily predictable by transepithelial resistance( Reference Krug, Amasheh and Richter 34 ), particularly where the transcellular passage route plays a role. While the passage of FITC–dextran and NHS-biotin is considered to occur via the paracellular epithelial pathway, that of HRP, frequently used as a ‘fluid-phase marker’ occurs via the transcellular epithelial passage pathway( Reference von Bonsdorff, Fuller and Simons 35 , Reference Heyman, Ducroc and Desjeux 36 ). To obtain information on the processing of foreign protein cargo during passage, we additionally quantified the expression of HRP protein by Western blotting as well as HRP immunoreactivity within mucosae. The passage of macromolecules was significantly altered only in the distal colon: inclusion of high fCHO reduced the permeation of the 443 Da tracer NHS-biotin into the submucosa and reduced the permeability to enzymatically active HRP. This effect was independent of dietary fCP content and thus unlikely to be mediated by the reduction in the levels of toxic N-containing fermentation products( Reference Pieper, Kröger and Richter 27 ). The only significant effect of fCP on permeability parameters was the reduced HRP immunoreactivity within the subepithelial compartment. Since no differences were observed for intact HRP protein (44 kDa band), these findings point towards diet-dependent handling of foreign proteins, where high fCP probably altered mucosal intracellular protein processing. A change in the paracellular passage of HRP in the high-fCP diet-fed piglets seems unlikely, since permeability to fluorescein isothiocyanate–dextran 4000 was not increased accordingly. Hence, the passage of small- to mid-sized macromolecules appeared to be increased with high-fCP dietary treatments and larger foreign proteins are likely to be processed differently.
Ion selectivity
Ion charge selectivity did not differ significantly among the treatment groups. Compared with the strong effect of claudin-2 overexpression on ion charge selectivity in cell-culture studies( Reference Amasheh, Meiri and Gitter 37 ), a weak correlation was observed for claudin-2 v. PNa/PCl in the present study (r 0·41 in the proximal colon) and PNa/PCl tended to be lower in the high-fCP diet-fed groups (P= 0·089). If claudin-2 is sufficient to confer cation selectivity on intestinal mucosa, a strong correlation of its expression and PNa/PCl can be expected, which was not the case in the present study. The change in ion charge selectivity is usually accompanied by an altered R epi ( Reference Prasad, Mingrino and Kaukinen 18 , Reference Amasheh, Meiri and Gitter 37 ), and it is assumed that the increased abundance of claudin-2 decreases R epi in the intestinal mucosa also( Reference De Preter, Hamer and Windey 1 ). Dietary variation of fCP/fCHO resulting in the altered expression of claudin-2 did not affect transmucosal or epithelial resistance in the present study. In contrast to cell-culture studies, this is probably compensated by the up-regulation of tightening tight junction proteins in vivo ( Reference Yu, Enck and Lencer 38 ).
Activity of the epithelial sodium channel
The high-fCP diets reduced ENaC activity, whereas the inclusion of high fCHO had no effect. ENaC activity is strongly influenced by the inflammation status of mucosae, and it was thus a candidate effector in the present study. It has repeatedly been shown that IL-1β regulates ENaC and is even sufficient to down-regulate the function of ENaC in ex vivo as well as in cell-culture studies( Reference Barmeyer, Harren and Schmitz 39 , Reference Schumann, Winter and Wichner 40 ). In IL-2-deficient mice, diarrhoea occurs in the presence of inflammation, but without obvious impairment of the barrier function probably via reduced ENaC activity( Reference Barmeyer, Amasheh and Tavalali 31 , Reference Barmeyer, Harren and Schmitz 39 ), and a similar effect can be observed in ulcerative colitis patients( Reference Schumann, Winter and Wichner 40 ). Additionally, ENaC induction is butyrate dependent( Reference Bergann, Plöger and Fromm 12 ) and regulated by TNF-α and glucocorticoids in vitro ( Reference Bergann, Zeissig and Fromm 41 ). SCFA are influenced by diet and ENaC is the last resort ion regulator located in distal epithelia – thus having a strong impact on diarrhoeal status. Indeed, the high-fCP diet-fed piglets showed mild diarrhoea: the faecal score was higher in the low-fCP/low-fCHO diet-fed group (3·49 (se 0·6)) than in the low-fCP/high-fCHO diet-fed (2·80 (se 0·04)), high-fCP/low-fCHO diet-fed (2·53 (se 0·05)) and high-fCP/high-fCHO diet-fed (2·79 (se 0·05)) groups (P< 0·01). The high-fCP/low-fCHO diet-fed piglets had significantly lower faecal scores than all the other groups (P< 0·01) and exhibited high-fCP-dependent IL-1β mRNA up-regulation in the colonic mucosa( Reference Pieper, Kröger and Richter 27 ).
Interestingly, a cell-culture study on ENaC induction has reported the activating effects of butyrate and propionate, but not of acetate( Reference Zeissig, Fromm and Mankertz 13 ). Induction of a more pronounced shift towards butyrate production in the colon might be an option to stimulate the function of ENaC in the high-fCP diet-fed piglets.
Tight junction proteins
Intestinal well-being, inflammation and tight junctions are closely linked: tight junction proteins regulate paracellular barrier properties in epithelia( Reference Elkouby-Naor and Ben-Yosef 19 , Reference Günzel and Fromm 42 ), are sensitive to the mucosal cytokine profile( Reference Prasad, Mingrino and Kaukinen 18 ), which has been studied extensively in inflammatory bowel disease pathology( Reference Turner 16 , Reference Shen, Weber and Raleigh 43 ), and are sensitive to luminal milieu, e.g. SCFA( Reference Bordin, D'Atri and Guillemot 11 , Reference Plöger, Stumpff and Penner 44 ). Permeability changes are often a result of mucosal inflammation, as it is clear that pro-inflammatory cytokines can directly decrease epithelial barrier function( Reference Capaldo and Nusrat 15 , Reference Schmitz, Fromm and Bentzel 45 ). In the present study, the high-fCP diets decreased the expression of colonic claudin-1, claudin-2, claudin-3 and claudin-4 (only in the distal colon). The profoundly reduced expression of colonic claudin-2 in the high-fCP diet-fed animals is surprising. Claudin-2 is a pore-forming protein, sufficient to decrease R epi and to confer cation selectivity on the paracellular pathway( Reference Amasheh, Meiri and Gitter 37 ). In the present study, a reduction in the expression of claudin-2 of more than 60 % unexpectedly did not result in changes in electrophysiological parameters, probably because it was accompanied by a decrease of tightening tight junction proteins. Knowledge on claudin function mostly stems from overexpression and knockdown approaches of the single proteins( Reference Krug, Amasheh and Richter 34 ); reports on disease-specific claudin expression profiles are inconsistent and many aspects of the claudin family interplay and resulting consequences remain mysterious( Reference Elkouby-Naor and Ben-Yosef 19 , Reference Krug, Amasheh and Richter 34 , Reference Turksen and Troy 46 ). To date, the up-regulation of claudin-2 expression in inflammatory bowel disease has been the only consistent observation, and hence it is considered to be an indicator of inflammation( Reference Turner 16 ). Although the observed down-regulation of claudin-2 expression in the present study might indicate that fCP is anti-inflammatory, the mRNA abundance of pro- and anti-inflammatory cytokines does not support this conclusion. Tricellulin is expressed at tricellular junctions. It regulates the passage of mid-sized macromolecules across epithelia( Reference Krug, Amasheh and Richter 34 ). In contrast to high-fCP diets, inclusion of high fCHO increased its expression in a low-protein background in the proximal colon, but there was no correlation between tricellulin expression and macromolecule passage.
Occludin is believed to exert more regulatory functions. It generally occurs as several bands on Western blots( Reference Feldman, Mullin and Ryan 47 ), presumably reflecting the functional modifications of the protein( Reference Feldman, Mullin and Ryan 47 , Reference Cummins 48 ). It is still a matter of debate what the biological consequences of these modifications are, but occludin is regarded as a sensitive indicator of tight junction changes( Reference Feldman, Mullin and Ryan 47 , Reference Blasig, Bellmann and Cording 49 , Reference Raleigh, Boe and Yu 50 ). In contrast to changes in the abundance of claudin, we observed a diet-dependent change in occludin modifications (not in total occludin abundance), consistent with a more regulatory function. An interesting finding is a lower relative abundance of the 32 kDa occludin in the distal colon of piglets fed high-fCP diets with low fCHO content. Bojarski et al. ( Reference Bojarski, Weiske and Schöneberg 51 ) showed that apoptosis-induced processing of occludin yielded 32 kDa fragments. The fCP-induced reduction in the abundance of 32 kDa occludin in the present study was abrogated by co-feeding of high fCHO, which may reflect the differential effects of fibre v. protein fermentation on epithelial apoptosis and thus on epithelial turnover. Most tight junction proteins analysed in the present study were stimulated by the high-fCHO/low-fCP diet in the proximal colon and down-regulated by high fCP in both the segments irrespective of fCHO inclusion. This suggests that high fCHO are effective in a low-protein background, but that they are unable to interfere with the effects of high fCP on tight junctions.
Apparent changes in the expression of mucosal tight junction proteins could be the result of structural changes such as alterations in mucosal surface area. This can be excluded, because the tight junction proteins examined were differentially regulated (e.g. the abundance of claudin-4 fCHO-dependently increased), electrophysiological parameters were unaffected and total colonic occludin expression (sum of band signals) was not affected by the diets.
Generally, down-regulation of epithelial cell adhesion structures including tight junctions is a hallmark of epithelial-to-mesenchymal transition and TGF-β is a master regulator of this process( Reference Nieto 52 ). Of the cytokines analysed in the present study, the expression of TGF-β was most prominently up-regulated( Reference Pieper, Kröger and Richter 27 ) and that of tight junction proteins was down-regulated. Although closely linked, the effects of epithelial-to-mesenchymal transition are different from those of classical intestinal inflammation, where the levels of pore-forming claudins are up-regulated and those of tightening claudins are down-regulated( Reference Krug, Amasheh and Richter 34 ).
Conclusion
A clear influence of fCHO and fCP on colon luminal microbial ecology and an increased expression of pro- and anti-inflammatory marker genes in the colonic epithelium have been observed in the same animals that were examined in the present study( 28 ). However, diet-induced changes in the colonic expression of pro- and anti-inflammatory marker genes apparently did not correlate with functional data, as the results of the present study did not demonstrate the dependence of gut barrier on luminal microbial ecology. Interestingly, the expression of tight junction proteins was substantially altered, but it also did not translate into changes in the intestinal barrier function. Collectively, the results of the present study suggest that despite changes in tissue relative concentrations of a number of tight junction proteins, an efficient epithelial barrier is maintained.
We thus propose a previously unrecognised diet-triggered mucosal adaption of proteins known to maintain paracellular barrier function and epithelial homeostasis upon changes in luminal microbial ecology in the large intestine of pigs. Whether the effects observed in the present study would have long-term influence on resistance to infectious challenges requires further studies.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0007114513003498
Acknowledgements
The authors thank A. Fromm, I. M. Lee and D. Sorgenfrei for their excellent technical assistance and K. Neumann and A. Busjahn for profound support in statistical issues.
The present study received financial support from Evonik Industries (Hanau – Wolfgang, Germany) and the German Research Foundation (DFG) through grant no. SFB852/1 and funding through the Canadian Swine Research and Development Cluster. None of the funders had a role in the design and analysis of the study or in the writing of this article.
The authors' contributions are as follows: J. F. R., R. P., D. G., J. D. S. and A. G. V. K. designed the research; J. F. R., R. P., S. S. Z. and A. G. V. K. conducted the research; J. F. R., R. P. and A. G. V. K. analysed the data; J. F. R., R. P. and A. G. V. K. wrote the paper; J. F. R. had primary responsibility for final content. All the authors read and approved the final version of the manuscript.
None of the authors has any conflicts of interest.
