Abstract
Microbial catabolic capacity in digesta from the gastrointestinal tract of pigs fed either dry feed or fermented liquid feed (FLF) was determined with the PhenePlate multisubstrate system. The in vitro technique was modified to analyze the kinetics of substrate catabolism mediated by the standing stock of enzymes (potential rates of fermentation), allowing a quantitative evaluation of the dietary effect on the catabolic capacity of the microbiota. In total, the potential rates of fermentation were significantly reduced in digesta from the large intestine (cecum, P < 0.1; colon, P < 0.01; and rectum, P < 0.0001) of pigs fed FLF compared to pigs fed dry feed. No effect of diet was observed in the stomach (P = 0.71) or the distal part of the small intestine (P = 0.97). The highest rates of fermentation and the most significant effect of diet were observed for readily fermentable carbohydrates like maltose, sucrose, and lactose. Feeding FLF to pigs also led to a reduction in the large intestine of the total counts of anaerobic bacteria in general and lactic acid bacteria specifically, as well as of microbial activity, as determined by the concentration of ATP and short-chain fatty acids. The low-molecular-weight carbohydrates were fermented mainly to lactic acid in the FLF before being fed to the animals. This may have limited microbial nutrient availability in the digesta reaching the large intestine of pigs fed FLF and may have caused the observed reduction in activity and density of the cecal and colonic microbial population. On the other hand, feeding FLF to pigs reduced the viable counts of coliform bacteria (indicator of Escherichia coli and Salmonella spp.) most profoundly in the stomach and the distal part of the small intestine, probably due to the bactericidal effect of lactic acid and low pH. The results presented clearly demonstrate that feeding FLF to pigs had a great impact on the indigenous microbiota, as reflected in bacterial numbers, short-chain fatty acid concentration, and substrate utilization. However, completely different mechanisms may be involved in the proximal and the distal parts of the gastrointestinal tract. The present study illustrates the utility of the PhenePlate system for quantifying the catabolic capacity of the indigenous gastrointestinal tract microbiota.
Ecophysiological characterization of terrestrial and aqueous microbial communities as well as model consortia of bacteria has been performed with carbon utilization tests such as Biolog, based on colorimetric measurements of tetrazolium dye reduction coupled to substrate oxidation (4, 6, 9, 37). Working in a predominantly anaerobic and reduced environment like the gastrointestinal tract, it may be critical, however, to use a system based on a redox indicator, since accumulated reducing equivalents seem to interfere and render false-positive signals (M. Katouli, personal communication; O. Højberg, unpublished data). The PhenePlate system (PhPlate AB, Stockholm, Sweden), on the other hand, is a substrate utilization test system based on detection of the production or breakdown of organic acids by the use of pH indicators (16, 26). The system was developed mainly for phenotyping of bacterial populations based on the analysis of representative isolates (17); however, it has been successfully applied for analyzing bacterial substrate utilization in composite samples from the gastrointestinal tract of pigs (13, 14).
A common approach with multisubstrate utilization systems has been to analyze the patterns obtained after a longer incubation period, e.g., days (13, 14, 37). The biochemical fingerprint obtained this way includes activity mediated by the standing stock of enzymes by the time of sampling as well as the activity of enzymes being activated or synthesized during the incubation period. One of the challenges, however, has been to extract information from the multisubstrate systems to obtain quantitative comparable data related to the real-time conditions of the samples (9). The aim of the present study was therefore to modify the use of the PhenePlate system in order to analyze the kinetics of the initial change in color intensity of the pH indicator and thereby quantify the potential rates of fermentation of the individual substrates as mediated by the standing stock of enzymes.
Low density of Escherichia coli and Salmonella spp. in the gastrointestinal tract of piglets and growing pigs is desired in order to reduce the incidence of postweaning diarrhea as well as the spread of zoonoses to consumers. Feeding liquid feed to piglets is a possible strategy to keep a high and regular feed and water intake in the critical period around weaning (11, 32). Moreover, the use of fermented liquid feed (FLF) has been shown to reduce the number of coliform bacteria (indicator of enterobacteria such as E. coli and Salmonella spp.) and Brachyspira hyodysenteriae in the gastrointestinal tract of piglets and growing pigs (19, 24, 35). The use of FLF has therefore recently gained further interest in pig husbandry as an alternative to the use of in-feed antibiotic growth promoters.
Fermentation of liquid feed leads to the proliferation of lactic acid bacteria and the production of organic acids, especially lactic acid (5, 11). Lactic acid is known to exhibit a strong bactericidal effect towards enterobacteria under low-pH conditions such as those prevailing in the stomach (5, 27, 28, 31, 33). Feeding FLF to pigs is therefore thought to reduce the number of enterobacteria surviving the passage through the stomach and/or proliferating in the small intestine or further down the gastrointestinal tract. Accordingly, the scenario in the proximal part of the gastrointestinal tract seems crucial when considering the antibacterial effect of FLF. It has been suggested, on the other hand, that the amount of fermentable substrate reaching the large intestine can influence expression of Brachyspira hyodysenteriae-mediated swine dysentery (3, 29, 30, 34). The aim of the present study was therefore to analyze microbial response mechanisms in the different parts of the gastrointestinal tract when feeding FLF to pigs.
MATERIALS AND METHODS
Diets.
A standard Danish grower diet was formulated to contain (on an air-dried basis): 10.00% barley; 60.74% wheat; 16.17% dehulled toasted soybean meal; 7.28% wheat bran; 2.74% animal fat; 1.26% calcium carbonate; 0.52% monocalcium phosphate; 0.41% sodium chloride; 0.32% l-lysine; 0.17% methionine-40%; 0.19% threonine-50%; and 0.20% vitamin mixture. The vitamin mixture contained (in international units per gram or milligrams per kilogram): 2,200 IU of vitamin A; 500 IU of vitamin D3; 30,000 mg of vitamin E; 1,100 mg of vitamin K3; 1,100 mg of vitamin B1; 2,000 mg of vitamin B2; 1,650 mg of vitamin B6; 5,500 mg of d-pantothenic acid; 11,000 mg of niacin; 27.5 mg of biotin; 11 mg of vitamin B12; 25,000 mg of FeSO4 · 7H2O; 40,000 mg of ZnO; 13,860 mg of MnO; 10,000 mg of CuSO4 · 5H2O; 99 mg of KI; and 150 mg of Na2SeO3. The feed components were not heat treated.
The FLF was prepared according to the recommendations of Jensen and Mikkelsen (11) by initially mixing dry feed and water in a ratio of 1:2.5 in an 80-liter tank. The mixture was allowed to ferment spontaneously (no addition of starter culture) under agitation at a temperature of 20°C for 4 days prior to the first feeding. During the experimental period, half of the tank content was used at each feeding and replaced by an equal volume of fresh feed-water mixture. The pH of the FLF was measured just before feeding throughout the experimental period and was always between 4.1 and 4.4.
Chemical analysis of feed components.
The chemical composition of dry feed and FLF was analyzed by standard methods for determining the content of energy, ash, fat, crude protein, starch, and specific saccharides (2).
Animals.
The study included 10 (five pairs of littermates) Danish Landrace × Yorkshire pigs separated into two groups of five animals (one littermate per group). The animals were shifted to the experimental diets at the age of 10 weeks (live body weight, approximately 30 kg) and housed individually with free access to water. The pigs were fed restrictively twice a day (7:30 and 15:30 hours) following a standard feeding norm. The animals always cleaned up the troughs, and there was no feed left over at any point. The experimental period started after 1 week of adaptation to the diets, and the animals were slaughtered at the age of 18 to 19 weeks (live body weight of approximately 60 kg).
Sample handling.
During the experimental period, rectal samples of feces were taken once a week and immediately transferred to the laboratory for further processing under anoxic conditions (see below). On the day of slaughter, the animals were fed at 6 a.m. and sacrificed by stunning 3 h later. The gastrointestinal tract was immediately removed from the body cavity and divided into eight segments: stomach, three equal-length segments of the small intestine, the cecum, and three equal-length segments of the colon (including the rectum). The digesta was immediately removed from the segments for further processing under anoxic conditions. The majority of the response parameters were analyzed only in digesta from the stomach, the distal segment of the small intestine (Si3), the cecum, and the midsegment of the colon (Co2). Based on experience in our lab, these four segments provide a satisfactory and representative picture of the gastrointestinal tract compartments for comparison of different dietary treatments.
pH and dry matter.
The pH of digesta and FLF was measured by direct insertion of a pH electrode (Radiometer Analytical S.A., Villeurbanne Cedex, France) into the matrices. Dry-matter content of the samples was determined by freeze-drying.
Potential rates of fermentation and biochemical fingerprints (PhenePlate technique).
The Ph-48 general 96-well microplate contains two identical sets of 48 freeze-dried substrates, including a range of low-molecular-weight carbohydrates (mono-, di-, and trisaccharides) as well as a number of carbohydrate derivatives (sugar alcohols, sugar acids, and glucosides), organic acids, urea, ornithine, and two blank control wells. The substrates of the 48 wells were (in order from 1 to 48) mannonic acid lactone; l-arabinose; d-xylose; galactose; maltose; cellobiose; trehalose; palatinose; sucrose; lactose; melibiose; lactulose; gentobiose; melezitose; raffinose; inosine; adonitol; inositol; d-arabitol; glycerol; maltitol; sorbitol; dulcitol; control (pH 7.4); sorbose; deoxyglucose; deoxyribose; rhamnose; d-fucose; l-fucose; tagatose; amygdalin; arbutin; β-methylglucoside; 5-ketogluconate; gluconate; melbionate; galacturonic lactone; salicine; control (pH 5.5); citrate; fumarate; malinate; malonate; pyruvate; l-tartrate; urea; and ornithine.
Production (wells 1 to 39) and degradation (wells 40 to 48) of acids as well as hydrolysis of urea to ammonia were visualized by the pH indicator bromomethyl blue. The color changed gradually from dark blue (alkaline; pH > 8) over green to yellow (acidic; pH < 5.5) as the pH decreased or vice versa. The intensity of the blue color was registered on an EL808iu Ultra microplate reader (Bio-Tek Instruments Inc., Winooski, Vt.) as absorbance at 620 nm (A620). Within nearly two pH units (6 < pH < 8), the A620 was shown to be inversely proportional to the proton concentration (Fig. 1). In practice, a stock solution (0.11%) of bromothymol blue (Merck 1.03026) was made in 100 mM NaOH, and the A620 was checked according to the PhenePlate system manual (PhenePlate AB, Stockholm, Sweden). The test buffer was made up of Bacto proteose peptone (Difco 0120-17-6), 1 g; 0.2 M phosphate (NaH2PO4 + Na2HPO4) buffer (pH 7.5), 20 ml; bromothymol blue stock solution, 90 ml; and H2O (Elga Ltd., Bucks, England), 800 ml. The pH was adjusted to 8.0, and the volume was finally adjusted to 1 liter. The solution was checked according to the manual and boiled (20 min) to remove oxygen.
FIG. 1.
Readings of color intensity as absorbance at 620 nm (A) and pH (B) in bromothymol blue test buffer amended with increasing amounts of HCl. The linear part of the plot in panel A illustrates the proportionality between color intensity and proton concentration over a range representing approximately 1.5 pH units (6.5 < pH < 8.0). The results are presented as means ± standard deviation (n = 3).
Aliquots of 9 ml and 90 ml were transferred under N2 flow into 15-ml Hungate tubes and 120-ml infusion bottles, respectively. The tubes and bottles were sealed with butyl rubber stoppers and autoclaved (121°C, 15 min). Digesta samples (approximately 10 g) were transferred under N2 flow to the infusion bottles containing 90 ml of test buffer. Nitrogen was used instead of CO2 to avoid acidification of the buffer. The suspensions were mixed, poured into N2-flushed stomacher bags, and homogenized for 2 min. Samples (1 ml) were transferred to the Hungate tubes with N2-flushed syringes. All tubes and bottles were weighed during the process to obtain precise dilution factors. The Hungate tubes were vortexed, and 1-ml samples were transferred to a new set of tubes. In an N2-flushed glove box, the diluted samples were dispersed into Ph-48 plates, adding 150 μl to each well. To avoid penetration of oxygen and evaporation of the suspension during incubation, the wells were sealed by overlaying each well with 100 μl of sterile, degassed paraffin oil. The plates were then incubated under atmospheric air at 37°C, and the A620 was read on the plate reader every 30 min for the first 4 h and then again after 7 to 8 h, 24 h, and 48 h.
The potential rates of fermentation were determined as the acid production or consumption mediated by the standing stock of enzymes in the digesta by the time of sampling. The potential rates of fermentation were thus calculated from the initial linear change in proton concentration measured as the A620 change in the individual wells of the Ph-48 plates. The absorbance data were normalized to the individual amount of sample and converted into proton (H+) concentration with the slope of the plot in Fig. 1A.
The biochemical fingerprint, on the other hand, was determined from the substrate utilization pattern obtained after prolonged incubation (7 to 8 h, 24 h, and 48 h) of the microplates with the PhenePlate system software as described by Katouli et al. (14, 15). Compared to the potential rate of fermentation, the biochemical fingerprint therefore also included enzyme activity caused by de novo enzyme activation or synthesis and microbial growth. In short, the absorbance values were multiplied by 10, yielding scores ranging from 0 to 30 for each sample. After the final reading, the mean value of the three readings was calculated, resulting in 48 quantitative values (including two negative controls) ranging from 0 to 30 for each sample (the biochemical fingerprint [26]). The biochemical fingerprints were then compared in pairs; similarity among them was calculated as similarity coefficient, which was clustered according to the unweighted pair group method with arithmetic averages (UPGMA) to yield a dendrogram. In the dendrogram, a horizontal line represents each sample. Different samples are connected with vertical lines at the similarity level they showed to each other, and thus, the more to the right this line is, the more similar are the samples. A high similarity coefficient (maximum = 1) between two samples means that they have similar metabolic fingerprints (14).
Growth of bacteria.
Digesta samples (approximately 10 g) were immediately transferred to infusion bottles containing 90 ml of a prereduced salt broth (7). All bottles were weighed to obtain precise dilution factors. The suspensions were transferred to sterile, CO2-flushed stomacher bags and homogenized for 2 min in a stomacher blender (Seward Medical, London, United Kingdom). Serial dilutions were made anaerobically in prereduced salt medium by the technique of Miller and Wolin (25). Total counts of anaerobic bacteria were determined on rumen fluid-glucose-cellobiose agar with the roll tube technique (8, 25) and incubation at 37°C for 7 days. Lactic acid bacteria were determined by spread plating on MRS (de Man, Rogosa and Sharp) agar (Merck 1.10660) and incubation in an anaerobic jar (Anaerocult A; Merck 1.13829) at 37°C for 2 days. Coliform bacteria were counted on MacConkey (Merck 5465) agar after aerobic incubation at 37°C for 1 day.
ATP, lactic acid, and short-chain fatty acids.
Digesta samples (approximately 5 g) were extracted with 10 ml of 2 M cold perchloric acid in 10 mM EDTA and stored at −80°C until further analysis. The content of ATP in the digesta was quantified by the luciferin-luciferase method according to Jensen and Jørgensen (10). The concentrations of lactic acid and short-chain fatty acids were determined as described by Jensen et al. (12).
Statistical analysis.
The effect of diet on parameters measured in each segment of the gastrointestinal tract and on potential rate of fermentation of each substrate in feces measured at weeks 1, 3, 5, and 7 was subjected to analysis of variance, the GLM procedure of the SAS package (32), with diet and litter as the main effects. The effect of diet on parameters measured in feces over 7 weeks was analyzed with the MIXED procedure of the SAS package (20) according to the model Yijl = μ + αi + βj + (αβ)j + Uil + ɛijl, where Y is the observed response, μ is the overall mean, αi is the effect of diet (i = 1 or 2), βj is the effect of week (j = 1,… , 7), (αβ)ij is the interaction between the effects of diet and week, Uil is the random effect of diet × litter (pig) (l = 1,…, 5), ≈N (0,σ2U), and ɛ is the residual error, ≈N (0,σ2ɛ). Note that the random effect of diet × litter was imposed to account for repeated measurements being made on the same experimental unit (pig). The interaction between diet and week was not significant (P > 0.05) in any case, so the interaction was eliminated and the analysis was repeated.
Statistically treated data are presented as least square means (LSMeans) and standard error of the mean (SEM). Except for those from feed components, all data in the present study are related to the wet weight of the samples.
RESULTS
Chemical composition of dry feed and FLF.
The content of energy, ash, fat, protein, and starch was identical in dry feed and FLF (Table 1). In the dry feed, the content of low-molecular-weight sugars (mono-, di-, tri-, and tetrasaccharides) made up 3.6% of the dry matter, whereas these sugars had almost disappeared in the FLF, where they constituted only 0.1% of dry matter (Table 1). Sucrose alone constituted 2.2% of the dry feed but could not be detected in the FLF, illustrating that the fermentation process actually depleted the feed for the readily fermentable carbohydrates. The content of lactose, maltose, maltotriose, melibiose, and verbascose was below the detection limits of 0.04%, 0.04%, 0.2%, 0.2%, and 0.5%, respectively, in the dry feed as well as in the FLF, and these saccharides are not listed in Table 1. The feed content of nonstarch polysaccharides was only slightly reduced by the fermentation process (Table 1).
TABLE 1.
Content of selected analyzed components in feed
| Component | Contenta
|
|
|---|---|---|
| Dry feed | FLF | |
| Energy (MJ kg−1) | 18.70 | 19.00 |
| Ash | 5.80 | 5.20 |
| Fat | 6.60 | 6.60 |
| Protein | 16.90 | 16.90 |
| Carbohydrates | ||
| Low-molecular-weight free saccharides | ||
| Fructose (monosaccharide) | 0.08 | 0.05 |
| Glucose (monosaccheride) | 0.04 | <0.02 |
| Sucrose (disaccharide) | 2.20 | <0.02 |
| Raffinose (trisaccharide) | 0.36 | <0.20 |
| Stachyose (tetrasaccharide) | 0.86 | <0.20 |
| Total | 3.54 | <0.49 |
| Nonstarch polysaccharidesb | ||
| Mannose | 0.30 | 0.30 |
| Galactose | 0.90 | 0.90 |
| Arabinose | 2.60 | 2.30 |
| Glucose | 4.00 | 3.60 |
| Xylose | 3.90 | 3.60 |
| Uronic acid | 0.90 | 1.00 |
| Total | 12.60 | 11.70 |
| Starch | 46.60 | 47.50 |
Energy is given as per kilogram dry weight. All other components are listed as percent of dry weight.
Content of nonstarch polysaccharides is listed as free saccharides liberated by hydrolyzation.
Potential rates of fermentation.
A linear relationship between color intensity (A620) and proton concentration was observed by titrating the test buffer with 0.1 mM HCl (Fig. 1). Therefore, by reading A620 values of the Ph-48 plates every 30 min during the first 4 h of incubation, substrate catabolism in the individual wells of the pH plates could be quantified as change in absorbance (Fig. 2) and finally expressed as change in proton concentration. The initial linear slopes obtained in the individual wells were further considered proportional to the size of the substrate-specific enzyme pools (enzyme concentration) at the time of sampling and were used as a quantitative estimate of the potential rates of fermentation (Fig. 2). We observed minimal interference or carryover from nonspecific substrate left in the diluted samples; however, the results were always corrected by subtracting the blank values obtained in the control wells.
FIG. 2.
Readings of absorbance (A620) from two wells (A5 and E5) of a Ph-48 plate. Both wells contained maltose as the substrate and were inoculated (time zero) with rectal samples from pigs fed dry feed (•) and FLF (○), respectively. The initial linear slopes, being proportional to the change in proton concentration (see Fig. 1), were normalized to the amount of sample and used for estimating the potential rates of fermentation.
For the rectal samples, the potential rate of fermentation varied considerably for the 46 substrates in the test and was close to the detection limit for 20 of the substrates included (data not shown). The highest rates were consistently observed for the disaccharides maltose, cellobiose, sucrose, lactose, melibiose, and gentobiose and the trisaccharide raffinose (Table 2). Moreover, the rates were consistently lower in samples from the pigs fed FLF, except for the hydrolysis of urea to ammonia. Furthermore, we observed a general decrease in the potential rates of fermentation for both groups of animals during the 7 weeks of the experiment (Table 2). The decrease was most dramatic the first 3 weeks of the experiment and most pronounced for the animals fed FLF (Table 2).
TABLE 2.
Potential rates of fermentation of selected substrates and hydrolysis of urea in rectal samples from pigs fed dry feed (DF) or FLFa
| Substrate | Potential rate of fermentation (mmol of H+ kg−1 h−1)
|
|||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Week 1
|
Week 3
|
Week 5
|
Week 7
|
|||||||||||||
| DF | FLF | SEM | P | DF | FLFb | SEM | P | DF | FLFb | SEM | P | DF | FLFb | SEM | P | |
| l-Arabinose | 25 | 13 | 5.4 | 19 | 13 | 3.7 | 10 | 3 | 2.9 | 4 | 1 | 2.2 | ||||
| d-Xylose | 11 | 8 | 2.3 | 4 | 2 | 1.2 | 4 | 0 | 1.2 | (∗) | 1 | 1 | 0.7 | |||
| Galactose | 62 | 28 | 6.6 | ∗ | 33 | 20 | 5.7 | 22 | 8 | 1.9 | ∗ | 20 | 6 | 4.5 | (∗) | |
| Maltose | 124 | 73 | 17.0 | (∗) | 66 | 32 | 10.2 | (∗) | 58 | 31 | 3.6 | ∗∗ | 69 | 19 | 11.0 | ∗ |
| Cellobiose | 87 | 51 | 8.8 | ∗ | 55 | 24 | 7.2 | ∗ | 45 | 23 | 2.6 | ∗∗ | 53 | 15 | 10.6 | (∗) |
| Sucrose | 135 | 74 | 18.5 | (∗) | 74 | 39 | 8.3 | ∗ | 72 | 39 | 5.4 | ∗ | 83 | 20 | 11.8 | ∗ |
| Lactose | 99 | 65 | 11.3 | (∗) | 54 | 29 | 5.8 | ∗ | 51 | 29 | 2.2 | ∗∗ | 48 | 20 | 8.2 | (∗) |
| Melibiose | 103 | 44 | 9.6 | ∗ | 54 | 23 | 9.4 | (∗) | 66 | 31 | 4.6 | ∗∗ | 66 | 18 | 8.3 | ∗ |
| Lactulose | 46 | 34 | 5.0 | 26 | 17 | 2.7 | (∗) | 21 | 10 | 1.3 | ∗∗ | 18 | 6 | 4.5 | ||
| Gentobiose | 78 | 41 | 6.5 | ∗ | 34 | 14 | 7.6 | 41 | 11 | 8.0 | (∗) | 50 | 5 | 11.2 | (∗) | |
| Raffinose | 119 | 52 | 12.9 | ∗ | 61 | 30 | 7.9 | (∗) | 66 | 29 | 4.0 | ∗∗ | 72 | 18 | 10.5 | ∗ |
| Tagatose | 8 | 12 | 3.3 | 14 | 4 | 7.5 | 2 | 1 | 0.8 | 2 | 0 | 0.9 | ||||
| Amygdalin | 45 | 31 | 5.3 | 31 | 16 | 4.7 | (∗) | 21 | 11 | 2.1 | ∗ | 22 | 7 | 3.5 | ∗ | |
| Arbutin | 44 | 21 | 5.0 | ∗ | 22 | 11 | 1.7 | ∗ | 25 | 8 | 5.6 | (∗) | 18 | 4 | 4.7 | (∗) |
| β-Methylglucose | 38 | 16 | 3.9 | ∗ | 18 | 5 | 4.0 | (∗) | 20 | 5 | 0.8 | ∗ | 18 | 2 | 4.6 | (∗) |
| Salicine | 50 | 24 | 4.7 | ∗ | 32 | 11 | 6.3 | (∗) | 28 | 10 | 3.8 | ∗ | 29 | 6 | 6.0 | (∗) |
| Urea | 65 | 118 | 31.8 | 68 | 31 | 11.8 | (∗) | 52 | 53 | 12.5 | 48 | 71 | 12.4 | |||
Results are presented as LSMean ± SEM (n = 5). Significance: (∗), P ≤ 0.1; ∗, P ≤ 0.05; ∗∗, P ≤ 0.01.
From week 3 to week 7, n = 4.
In the digesta samples taken at slaughter, the potential rates of fermentation increased from the proximal (stomach and ileum) to the distal (cecum and colon) parts of the gastrointestinal tract, especially for the pigs fed dry feed (Table 3). As observed for the rectal samples, the rates determined in samples from the cecum and colon were consistently lower for the pigs fed FLF compared to the pigs fed dry feed. The only clear exception from this general picture was again the hydrolysis of urea to ammonia.
TABLE 3.
Potential rates of fermentation of selected substrates and hydrolysis of urea in digesta samples from pigs fed dry feed (DF) or FLFa
| Substrate | Potential rate of fermentation (mmol of H+ kg−1 h−1)
|
|||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Stomach
|
Ileum
|
Cecum
|
Colon segment 2
|
|||||||||||||
| DF | FLF | SEM | P | DF | FLF | SEM | P | DF | FLF | SEM | P | DF | FLF | SEM | P | |
| l-Arabinose | 1 | 2 | 0.8 | 2 | 0 | 0.8 | 4 | 0 | 1.9 | 9 | 1 | 2.0 | ∗ | |||
| d-Xylose | 1 | 1 | 0.4 | 2 | 1 | 1.2 | 1 | 0 | 0.2 | 4 | 1 | 1.3 | ||||
| Galactose | 8 | 5 | 1.9 | 2 | 6 | 1.9 | 13 | 6 | 3.0 | 31 | 6 | 3.9 | ∗ | |||
| Maltose | 13 | 12 | 3.2 | 3 | 15 | 5.5 | 29 | 15 | 6.3 | 83 | 16 | 9.0 | ∗∗ | |||
| Cellobiose | 8 | 7 | 1.4 | 3 | 6 | 2.6 | 20 | 12 | 3.3 | 63 | 14 | 7.8 | ∗ | |||
| Sucrose | 15 | 13 | 2.9 | 7 | 15 | 7.2 | 32 | 17 | 6.9 | 98 | 18 | 11.0 | ∗∗ | |||
| Lactose | 3 | 2 | 1.1 | 5 | 4 | 2.3 | 22 | 10 | 3.2 | (∗) | 54 | 13 | 5.9 | ∗∗ | ||
| Melibiose | 3 | 3 | 1.0 | 8 | 4 | 3.0 | 30 | 10 | 4.9 | ∗ | 71 | 11 | 7.4 | ∗∗ | ||
| Lactulose | 2 | 3 | 1.7 | 3 | 2 | 1.0 | 13 | 4 | 2.0 | ∗ | 24 | 6 | 2.6 | ∗∗ | ||
| Gentobiose | 6 | 5 | 1.2 | 3 | 2 | 1.9 | 15 | 6 | 2.5 | (∗) | 59 | 8 | 7.8 | ∗∗ | ||
| Raffinose | 8 | 4 | 1.5 | 4 | 5 | 2.5 | 29 | 10 | 5.3 | (∗) | 84 | 13 | 8.7 | ∗∗ | ||
| Tagatose | 1 | 3 | 0.5 | ∗ | 2 | 2 | 0.9 | 3 | 2 | 0.6 | 4 | 1 | 0.7 | ∗ | ||
| Amygdalin | 5 | 6 | 1.4 | 4 | 4 | 1.4 | 14 | 6 | 2.3 | (∗) | 34 | 8 | 4.9 | ∗ | ||
| Arbutin | 0 | 1 | 0.7 | 0 | 1 | 0.7 | 6 | 1 | 1.5 | (∗) | 32 | 3 | 5.9 | ∗ | ||
| β-Methylglucose | 3 | 3 | 1.3 | 3 | 3 | 1.2 | 8 | 3 | 1.2 | (∗) | 26 | 4 | 3.6 | ∗ | ||
| Salicine | 3 | 4 | 0.8 | 3 | 1 | 1.2 | 10 | 4 | 1.5 | (∗) | 37 | 6 | 4.0 | ∗∗ | ||
| Urea | 2 | 0 | 1.0 | 7 | 6 | 1.5 | 62 | 51 | 6.9 | 61 | 68 | 5.7 | ||||
See Table 2, footnotes a and b.
Summing up the potential rates of fermentation of all substrates included (excluding the hydrolysis of urea to ammonia) clearly illustrates the pattern obtained for the fermented sugars in general (Fig. 3). Throughout the experimental period, a constant lower level (Pdiet < 0.0001) was observed in rectal samples taken from animals fed FLF compared to animals fed dry feed (Fig. 3A). For both groups of animals, however, the results revealed a significant time-dependent (Pweek < 0.001) and parallel (Pdiet×week ≈ 1.0) decrease in the potential rate of fermentation, especially the first 3 to 4 weeks of the experimental period, after which it leveled out. At slaughter, no significant diet-mediated differences were observed in samples from the stomach (P = 0.71) and ileum (P = 0.97) (Fig. 3B). On the other hand, the potential rates of fermentation in the cecum and colon were significantly higher for the pigs fed dry feed compared to the pigs fed FLF (Fig. 3B). Measured in feed samples directly, the potential rate of fermentation was very low in the dry feed itself compared to that of the FLF (Fig. 3B).
FIG. 3.
Potential rate of fermentation summed up for 45 of the substrates analyzed (hydrolysis of urea is not included). The data represent samples taken from the rectum (A) and the gastrointestinal tract (B) of pigs fed dry feed (• and black bars) and FLF (○ and white bars). Data from feed analysis (n = 2) are included in panel B. The other results are presented as LSMeans ± SEM (n = 5; in A, n = 4 from week 3 to week 7 for pigs fed FLF).
Biochemical fingerprints.
The microplates were read after 7 to 8 h, 24 h, and 48 h of incubation for determination of the biochemical fingerprints as described by Katouli et al. (13, 14). The prolonged incubation of the microplates led to enzyme induction and/or bacterial growth for some of the substrates. This was clearly reflected in the change in absorbance in these wells as an initial lag phase followed by a sigmoidal growth curve-like pattern (data not shown). The dendrograms derived from the biochemical fingerprints showed a tendency for clustering of the samples according to diet irrespective of the segment of origin (Fig. 4). They may therefore illustrate development of diet-mediated differences in composition of the indigenous microbiota in the two groups of animals. However, the discrimination between the two groups of animals with this approach is clearly not as unambiguous as observed for the potential rates of fermentation.
FIG. 4.
Dendrograms obtained by clustering similarity coefficients of the biochemical fingerprints (see Materials and Methods). The PhenePlate system software was used to analyze the substrate utilization patterns of samples taken from the stomach (A), ileum Si3 (B), cecum (C), and colon Co2 (D) samples after 7 to 8 h, 24 h, and 48 h of incubation of the Ph-48 plates. The samples were taken from pigs fed dry feed (•; pigs 109, 121, 126, 155, and 163) and FLF (○; pigs 113, 117, 125, 159, and 167). In the dendrograms, a horizontal line represents each sample. Different samples are connected with vertical lines at the similarity level that they showed to each other. A high similarity coefficient (maximum = 1) between two samples indicates similar biochemical fingerprints. An arbitrary level of identity was set to 0.975 (dashed line).
Bacterial counts.
The total counts of anaerobic bacteria in rectal samples obtained from the two groups of animals tended to be lower (Pdiet < 0.1) for the pigs fed FLF. However, this tendency was mainly observed during the last 3 weeks of the experiment (Fig. 5A). For both groups of animals, the total counts of anaerobes decreased during the first 4 to 5 weeks of the experiment, reaching a level approximately 5 (dry feed) to 10 (FLF) times below the starting point. In the stomach, cecum, and colon, the total count of anaerobes reflected the tendency of the rectal samples in the last part of the experimental period, namely, that lower counts were obtained for the pigs fed FLF (Fig. 5B); however, the difference was most significant for the colon samples (P = 0.04).
FIG. 5.
Counts (CFU) of total anaerobes (A, B), lactic acid bacteria (C, D), and coliform bacteria (E, F) in samples from the rectum (left panels) and the gastrointestinal tract (right panels) of pigs fed dry feed (• and black bars) and FLF (○ and white bars). Data from feed analysis (n = 2) are included in right-hand panels. Digesta from the stomach (St), the distal segment of the small intestine (Si3), the cecum (Cae), and the midsegment of the colon (Co2) were tested.The other results are presented as LSMeans ± SEM (n = 5; in the left panel, n = 4 from week 3 to week 7 for pigs fed FLF). The indicated levels of significance in the right panel are (*), P ≤ 0.1; *, P ≤ 0.05; and **, P ≤ 0.01.
The counts of lactic acid bacteria were significantly lower (Pdiet < 0.001) in rectal samples from the pigs fed FLF (Fig. 5C) throughout the experiment. As for total counts, we observed a decrease in the number of lactic acid bacteria in the first 4 weeks, especially for the animals fed FLF. No significant diet-mediated difference could be observed in the counts of lactic acid bacteria in the stomach (P = 0.73) and the small intestine (P = 0.56) between the two groups of animals (Fig. 5D). The counts of lactic acid bacteria tended to be higher only in the cecum (P = 0.13) and colon (P = 0.11) for the pigs fed dry feed. Analyzing the feed directly revealed an almost 10,000-fold difference in the level of lactic acid bacteria between dry feed and FLF (Fig. 5D).
Overall, the number of coliform bacteria in the rectal samples tended to be lower (Pdiet = 0.10) for the animals fed FLF during the 7 weeks of growth (Fig. 5E), however, mainly based on the counts in weeks 1, 3, and 4. The counts of coliforms showed a clear time-dependent decrease (Pweek < 0.001) for both groups of animals. At slaughter, numerically lower counts of coliform bacteria were observed along the gastrointestinal tract for the pigs fed FLF (Fig. 5F), but for this group of bacteria, the most significant differences were obtained in the stomach (P = 0.07) and the small intestine (P = 0.08).
Lactic acid and short-chain fatty acids.
Lactic acid was detected mainly in the stomach and the small intestine, and significantly higher levels were determined in digesta from the animals fed FLF compared to the ones fed dry feed (Fig. 6A). In the cecum and colon, lactic acid could not be detected in samples from pigs fed dry feed, whereas low levels were measured in the pigs fed FLF. The concentration of acetic, propionic, and butyric acids, on the other hand, was low in the proximal part of the gastrointestinal tract and high in the cecum and colon (Fig. 6B). The highest concentrations of these short-chain fatty acids were determined in the digesta of the pigs fed dry feed, but the difference was only significant in samples from the cecum (P = 0.05). Branched short-chain fatty acids (isovaleric and isobutyric acid) were detectable only in the four distal segments (cecum and colon) of the gastrointestinal tract (data not shown). Higher concentrations of the branched short-chain fatty acids were obtained for the dry-fed pigs, but the diet-mediated differences were only significant in the cecum and the first segment of the colon (P < 0.05).
FIG. 6.
Concentration of lactic acid (A) and selected short-chain fatty acids (acetic, butyric, and propionic acids) (B) in digesta from pigs fed dry feed (•) and FLF (○). Digesta from the stomach (St), three segments of the small intestine (Si1, -2, and -3), the cecum (Cae), and three segments of the colon (Co1, -2, and -3) were tested. The results are presented as LSMeans ± SEM (n = 5), and the indicated levels of significance are (*), P ≤ 0.1; *, P ≤ 0.05; and **, P ≤ 0.01.
pH, ATP, and dry-matter content.
The pHs of the rectal samples showed only minor variation within a range between 5.9 and 6.7; however, significantly lower values (Pdiet < 0.001) were consistently obtained for the animals fed FLF (data not shown). No significant differences in pH could be measured between the two groups of animals along the gastrointestinal tract at slaughter (Table 4). The ATP content of the rectal samples was not determined. Higher concentrations of ATP were measured in digesta from the stomach, cecum, and colon of the pigs fed dry feed, whereas in the ileum, the highest level of ATP was obtained for the pigs fed FLF (Table 4). The difference was only statistically significant (P < 0.01) for the colon samples. The dry-matter content of the rectal samples was 0 to 4% higher for the animals fed dry feed (data not shown), which was not a significant difference (Pdiet = 0.14). No significant differences were observed in the cecum and colon, whereas in the stomach and ileum, the dry-matter content was 6% and 4% higher, respectively, for the animals fed dry feed (Table 4).
TABLE 4.
Effect of diet on pH, ATP, and dry-matter content of digestaa
| Segment | pH
|
ATP (μg g−1)
|
Dry matter (%)
|
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DF | FLF | SEM | P | DF | FLF | SEM | P | DF | FLF | SEM | P | |
| Stomach | 4.1 | 3.9 | 0.13 | 6.1 | 4.0 | 0.94 | 32.1 | 26.1 | 1.09 | ∗ | ||
| Ileum (Si3) | 6.1 | 5.9 | 0.20 | 7.0 | 9.2 | 1.48 | 14.2 | 9.9 | 1.02 | ∗ | ||
| Cecum | 5.4 | 5.3 | 0.12 | 13.4 | 11.2 | 1.47 | 11.4 | 10.9 | 0.27 | |||
| Colon (Co2) | 5.9 | 5.7 | 0.15 | 14.7 | 12.1 | 0.39 | ∗∗ | 18.6 | 17.0 | 0.69 | ||
See Table 2, footnote a. DF, dry feed.
DISCUSSION
With the PhenePlate system and amending the principle of analyzing the kinetics of the catabolism of the individual substrates (6), we obtained quantitative in vitro estimates of substrate fermentation capacity by applying a linear regression model to the initial change in color intensity of the pH indicator. The estimates therefore reflected a net change in proton concentration proportional to the actual substrate turnover in the wells and did not, as such, account for the exact biochemical stoichiometry for the catabolism of the individual substrate. Nevertheless, the estimated levels of proton production, ranging from 1 to 100 mmol of H+ kg−1 h−1 in the colon samples (Table 3), are in good accordance with the levels observed for in vitro production rates of fatty acids (18, 22; N. Canibe and B. B. Jensen, submitted for publication). Likewise, a total production of short-chain fatty acids of 6.5 mmol kg−1 h−1 was reported for slurries of colon digesta amended with d-tagatose (18), resembling the potential rate of fermentation of 4 mmol of H+ kg−1 h−1 for the dry-fed pigs in the present study (Table 3). It therefore seems reasonable to use the multisubstrate system as presented here not only as a fingerprint tool, but also to semiquantify the capacity for catabolizing the individual substrates included.
It should be emphasized that compared to the proton production rates, we may have underestimated the rates of proton consumption in the wells amended with substrates such as citrate and fumarate (initial low pH), since we may have been working below the linear part of the absorbance curve (Fig. 1A). However, at least for the hydrolysis of urea to ammonia, we always observed an immediate and linear increase in coloration.
It has been demonstrated that factors such as inoculum size and dilution rate can have a significant influence on the results obtained with the Biolog system (6). On the other hand, the PhenePlate system has been introduced as not being extremely sensitive to these factors, and successful community analysis of fecal microbiota has been performed on the basis of unbalanced samples taken by the use of rectal swabs (14). To meet the concerns raised by Haack et al. (6) and to obtain comparative and quantitative results, we aimed to use a uniform inoculum size and dilution factors for all the samples and made corrections for any inaccuracies when calculating the potential rates of fermentation. We therefore consider the in vitro rate of substrate catabolism presented (potential rate of fermentation) a valuable quantitative and real-time response parameter suitable for characterizing the catabolic capacity, i.e., the standing stock of enzymes, of the indigenous gastrointestinal microbiota.
In contrast to the potential rates of fermentation, the biochemical fingerprint obtained after prolonged incubation of the PhenePlates may include substrate utilization based on bacterial growth and/or de novo enzyme synthesis and activation during the incubation period and therefore may not reflect the real-time phenotypic conditions. If enzymes are actually induced during incubation of the samples, the biochemical fingerprints are more likely to reflect the genetically conditioned capacity to utilize the tested substrates based on community composition. However, it should be emphasized that the enzyme systems involved in the catabolism of the individual substrates may not all be inducible simply by exposure to the substrates under the incubation conditions provided.
In the present experiment, prolonged incubation did reveal clustering of the biochemical fingerprints according to diet, indicating differences in microbial community composition and/or enzyme expression in the two groups of animals. However, the biochemical fingerprints did not differentiate the two groups of animals as clearly as the potential rates of fermentation, suggesting that feeding FLF to the pigs may have affected bacterial density as well as activity more than community composition. By using more specific substrates than the ones included in the Ph-48 general plate, the fingerprint method may, on the other hand, be modified to target specific groups or microorganisms (36), which could change the outcome. We are presently performing comparative studies involving molecular fingerprinting (terminal-restriction fragment length polymorphism) to evaluate the biochemical fingerprint technique.
The utility of the potential rates of fermentation obtained by the modified PhenePlate method was supported by the fact that they allowed us to quantify diet-mediated changes in the catabolic capacity of the indigenous microbiota that made biological sense. Thus, in contrast to the concerns raised by Haack et al. (6), we observed that the substrate utilization data were supported by observations on the size of specific bacterial groups, e.g., lactic acid bacteria. We therefore believe that the observed differences in potential rates of fermentation represent real differences in the number or types of microorganisms and that these differences further reflect the enzymes expressed by the microbiota by the time of sampling.
Actual microbial activity cannot be extrapolated directly from the potential rates of fermentation presented here, since the concentration of substrate rather than enzyme may be the limiting factor in situ. On the other hand, the digesta content of short-chain fatty acids (acetic, propionic, and butyric acids) and ATP has been used as a measure of microbial activity in the gastrointestinal tract of pigs (10, 12). The digesta content of short-chain fatty acids reflects microbial activity but as a net measure, since these compounds are absorbed over the epithelium (12). The ATP molecule is not absorbed and is readily degraded, and therefore the size of the standing stock should correlate to real-time microbial activity (10). In the present study, the ATP and short-chain fatty acids data supported the results of the potential rates of fermentation, indicating higher microbial activity in the cecum and colon of the pigs fed dry feed. The discrimination between the two groups of animals based on these measures was not as obvious as depicted by the potential rates of fermentation. This discrepancy between actual and potential activity for both groups of animals could reflect the time factor included; substrate availability may eventually become a limiting factor for microbial activity in the large intestine not only of pigs fed FLF but also of pigs fed dry feed.
An age-dependent change in substrate utilization capacity as well as in bacterial counts in fecal samples from pigs has been reported previously (1, 14). According to Bauer et al. (1), the gastrointestinal microbiota of suckling piglets develops a high capacity for utilizing substrates that are components or breakdown products of sow's milk. Postweaning, this capacity decreases gradually as the microbiota adapts to the components of the creep feed. Interestingly, however, Bauer et al. (1) observed that the microbiota of the suckling piglets had a relatively high capacity to ferment dietary fibers that had not been present in their diet. They suggested that the oligosaccharides present in the sow's milk act as or resemble soluble fiber in the gastrointestinal tract of the piglets. The shifts in substrate adaptation postweaning have also been demonstrated for the gastrointestinal microbiota of human infants (21) and may explain the overall decrease in the potential rate of fermentation that we observed for the rectal samples, especially in the first part of the experimental period independent of diet. This is further supported by the fact that we observed a high fermentation capacity for substrates such as lactose, maltose, and melibiose that were detected neither in the FLF nor in the dry feed.
In the present study, we followed young pigs from the age of 10 weeks to the age of 17 weeks and observed an overall decrease in the capacity to ferment the pure substrates tested, independent of the diet. We did not observe an age-dependent increase in fermentation capacity of any of the substrates tested, which is in good accordance with the observations made by Katouli et al. (14). They found that the ability to ferment 18 of the substrates included in the Ph-48 plate was lost in fecal samples for most of the pigs at an age of 20 weeks, compared to the analysis made at an age of 3 days. Only six substrates showed the opposite pattern and only for 1 to 5 of the 14 pigs examined.
Feeding FLF to pigs has been shown to reduce the count of coliform bacteria along the gastrointestinal tract (11, 23, 35). The effect has been assigned mainly to the bactericidal nature of the lactic acid under low-pH conditions, since feeding FLF to pigs leads to increased levels of lactic acid in the stomach and the small intestine, as also demonstrated in the present study. In spite of the increased concentration of lactic acid in the stomach of pigs fed FLF, the pH of the stomach is not always significantly reduced, probably due to the fact that other factors, e.g., HCl secretion, are involved. In the small intestine, an increase in pH has been connected to the use of FLF, since lactic acid induces secretion of pancreatic juice (11). In the present experiment, the use of FLF led neither to a significant pH reduction in the stomach nor to a pH increase in the small intestine, pointing towards the lactic acid molecule as the most important single factor for reducing the population of coliform bacteria. However, the reduction in coliform bacteria has been generally observed throughout the gastrointestinal tract, whereas the increased level of lactic acid has been restricted to the stomach and the small intestine (35). This could demonstrate that the proximal part of the gastrointestinal tract serves as a barrier to the migration of these bacteria into the large intestine.
Swine dysentery is caused by colonization of the large intestine by the spirochete B. hyodysenteriae. It has been shown that expression of swine dysentery can be reduced by feeding pigs with FLF (19) or diets low in soluble nonstarch polysaccharides, oligosaccharides, and resistant starch (3, 29, 30, 34). The positive dietary effect was suggested to be due to the decreased amount of fermentable substrate reaching the large intestine, thus preventing colonization of B. hyodysenteriae either by direct reduction in the substrate available for the pathogen itself or by inhibition of a synergistic microbiota.
It may be a simplification to consider only the microbial substrate being removed by fermentation in the FLF prior to animal intake. The fermentation of the feed may affect the digestibility of the feed components, and increased digestion and absorption of feed components in the small intestine may cause an additional reduction in readily fermentable substrate reaching the large intestine of pigs fed FLF. Our results certainly indicate that the availability of readily fermentable carbohydrates is more likely to become a limiting factor for activity and growth of the microbiota in the large intestine of pigs fed FLF compared to pigs fed dry feed. The mechanisms behind this and how this in detail influences the interactions between pathogens and zoonoses on one side and the population of commensal microorganisms on the other side need further elucidation.
The modified application of the PhenePlate system as presented here seems a strong tool for quantifying real-time catabolic capacity of the gastrointestinal microbiota. The possibility of including a large range of substrates in one analysis makes it perfect for screening the fates of different feed or food components in the gastrointestinal tract. This may also be an important parameter for evaluating the susceptibility of animals to certain enteric bacterial diseases. The approach may finally be an inspiration for the strongly needed functional analysis of the gastrointestinal tract microbiota (36), as well as other microbial communities as a supplement to the continuously growing trove of taxonomic data on microbial community composition.
Acknowledgments
We thank Jane Rasmussen for skillful technical assistance with the PhenePlate technique.
This study was supported by the Research Secretariat of the Danish Ministry of Food, Agriculture and Fisheries (project number Alt-98-1).
REFERENCES
- 1.Bauer, E., B. A. Williams, C. Voigt, R. Mosenthin, and M. W. A. Verstegen. 2001. Microbial activities of faeces from unweaned and adult pigs, in relation to selected fermentable carbohydrates. Anim. Sci. 73:313-322. [Google Scholar]
- 2.Canibe, N., and K. E. Bach Knudsen. 2001. Degradation and physicochemical changes of barley and pea fibre along the gastrointestinal tract of pigs. J. Sci. Food Agric. 82:27-39. [Google Scholar]
- 3.Durmic, Z., D. W. Pethick, J. R. Pluske, and D. J. Hampson. 1998. Changes in bacterial populations in the colon of pigs fed different sources of dietary fibre, and the development of swine dysentery after experimental infection. J. Appl. Microbiol. 85:574-582. [DOI] [PubMed] [Google Scholar]
- 4.Garland, J. L., and A. L. Mills. 1991. Classification and characterization of heterotrophic microbial communities on the basis of patterns of community-level sole-carbon-source utilization. Appl. Environ. Microbiol. 57:2351-2359. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Geary, T. M., P. H. Brooks, J. D. Beal, and A. Campbell. 1999. Effect on weaner pig performance and diet microbiology of feeding a liquid diet acidified to pH 4 with either lactic acid or through fermentation with Pediococcus acidilactici. J. Sci. Food. Agric. 79:633-640. [Google Scholar]
- 6.Haack, S. K., H. Garchow, M. J. Klug, and L. J. Forney. 1995. Analysis of factors affecting the accuracy, reproducibility, and interpretation of microbial community carbon source utilization pattern. Appl. Environ. Microbiol. 61:1458-1468. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Holdeman, L. V., E. P. Cato, and E. C. Moore. 1977. Anaerobe laboratory manual. Virginia Polytechnic Institute and State University, Blacksburg, Va.
- 8.Holdeman, L. V., and W. E. C. Moore. 1972. Roll tube techniques for anaerobic bacteria. Am. J. Clin. Nutr. 25:1314-1317. [DOI] [PubMed] [Google Scholar]
- 9.Insam, H. 1997. Substrate utilization tests in microbial ecology. J. Microbiol. Methods 30:1-2. [Google Scholar]
- 10.Jensen, B. B., and H. Jørgensen. 1994. Effect of dietary fiber on microbial activity and microbial gas production in various regions of the gastrointestinal tract of pigs. Appl. Environ. Microbiol. 60:1897-1904. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Jensen, B. B., and L. L. Mikkelsen. 1998. Feeding liquid diets to pigs, p. 107-126. In P. C. Garnsworthy and J. Wiseman (ed.), Recent advances in animal nutrition. Nottingham University Press, Nottingham, United Kingdom.
- 12.Jensen, M. T., R. P. Cox, and B. B. Jensen. 1995. Microbial production of skatole in the hindgut of pigs given different diets and its relation to skatole disposition in backfat. Anim. Sci. 61:293-304. [Google Scholar]
- 13.Katouli, M., E. Foo, I. Kühn, and R. Möllby. 1997. Evaluation of the Phene Plate generalized microplate for metabolic fingerprinting and for measuring fermentative capacity of mixed bacterial populations. J. Appl. Microbiol. 82:511-518. [DOI] [PubMed] [Google Scholar]
- 14.Katouli, M., A. Lund, P. Wallgren, I. Kühn, O. Söderlind, and R. Möllby. 1997. Metabolic fingerprinting and fermentative capacity of the intestinal flora of pigs during pre- and post-weaning periods. J. Appl. Microbiol. 83:147-154. [DOI] [PubMed] [Google Scholar]
- 15.Katouli, M., L. Melin, M. Jensen-Waern, P. Walgren, and R. Möllby. 1999. The effect of zinc oxide supplementation on the stability of the intestinal flora with special reference to composition of coliforms in weaned pigs. J. Appl. Microbiol. 87:564-573. [DOI] [PubMed] [Google Scholar]
- 16.Kühn, I. 1985. Biochemical fingerprinting of Escherichia coli: a simple method for epidemiological investigations. J. Microbiol. Methods 3:159-170. [Google Scholar]
- 17.Kühn, I., G. Allestam, T. A. Stenström, and R. Mölby. 1991. Biochemical fingerprint of water coliform bacteria, a new method for measuring phenotypic diversity and for comparing different bacterial populations. Appl. Environ. Microbiol. 57:33171-33177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Lærke, H. N., B. B. Jensen, and S. Højsgaard. 2000. In vitro fermentation pattern of d-tagatose is affected by adaptation of the microbiota from the gastrointestinal tract of pigs. J. Nutr. 130:1772-1779. [DOI] [PubMed] [Google Scholar]
- 19.Lindecrona, R. H., B. B. Jensen, T. K. Leser, and K. Møller. 2000. The influence of diet on the development of swine dysentery, p. 7. Proceedings of the 16th International Pig Veterinary Society Congress, Melbourne, Australia.
- 20.Littell, R. C., G. A. Milliken, W. W. Stroup, and W. D. Russell. 1996. SAS system for mixed models. SAS Institute Inc., Cary, N.C.
- 21.McVeagh, P., and J. B. Miller. 1997. Human milk oligosaccharides: only the breast. J. Paediatr. Child Health 33:281-286. [DOI] [PubMed] [Google Scholar]
- 22.Mikkelsen, L. L. 2001. Ph.D. thesis. The Royal Veterinary and Agricultural University, Frederiksberg, Denmark.
- 23.Mikkelsen, L. L., and B. B. Jensen. 1997. Effect of fermented liquid feed (FLF) on growth performance and microbial activity in the gastrointestinal tract of weaned piglets, p. 637-642. In J. P. Leplace, C. Fevrier, and A. Barbean (ed.), Proceedings of the 7th Symposium on Digestive Physiology in Pigs, Saint-Malo, France, 1997. European Association for Animal Production Publication No. 88.
- 24.Mikkelsen, L. L., and B. B. Jensen. 2000. Effect of fermented liquid feed on the activity and composition of the microbiota in the gut of pigs. Pigs News Information 21:59N-66N.
- 25.Miller, T. L., and M. J. Wolin. 1974. A serum bottle modification of the Hungate technique for cultivating of obligate anaerobes. Appl. Microbiol. 27:985-987. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Möllby, R., I. Kühn, and M. Katouli. 1993. Computerised biochemical fingerprinting—a new tool for typing of bacteria. Rev. Med. Microbiol. 4:231-241. [Google Scholar]
- 27.Naughton, P. J., and B. B. Jensen. 2001. A bioreactor to study survival of Salmonella Typhimurium in pig gut content. Berl. Mün. Tierarztl. Wochenschr. 114:378-381. [PubMed] [Google Scholar]
- 28.Nout, M. J. R., F. M. Rombouts, and A. Havelaar. 1989. Effect of accelerated natural lactic fermentation of infant food ingredients on some pathogenic microorganisms. Int. J. Food Microbiol. 8:351-361. [DOI] [PubMed] [Google Scholar]
- 29.Pluske, J. R., P. M. Siba, D. W. Pethick, Z. Durmic, B. P. Mullan, and D. J. Hampson. 1996. The incidence of swine dysentery in pigs can be reduced by feeding diets that limit the amount of fermentable substrate entering the large intestine. J. Nutr. 126:2920-2933. [DOI] [PubMed] [Google Scholar]
- 30.Pluske, J. R., Z. Durmic, D. W. Pethick, B. P. Mullan, and D. J. Hampson. 1998. Confirmation of the role of rapidly fermentable carbohydrates in theexpression of swine dysentery in pigs after experimental infection. J. Nutr. 128:1737-1744. [DOI] [PubMed] [Google Scholar]
- 31.Russell, J. B., and F. Diez-Gonzales. 1998. The effects of fermentation acids on bacterial growth. Adv. Microb. Physiol. 39:205-234. [DOI] [PubMed] [Google Scholar]
- 32.Russell, P. J., T. M. Geary, P. H. Brooks, and A. Campbell. 1996. Performance, water use and effluent output of weaner pigs fed ad libitum with either dry pellets or liquid feed and the role of microbial activity in the liquid feed. J. Sci. Food Agric. 72:8-16. [Google Scholar]
- 33.SAS Institute. 1989. SAS. user's guide: statistics (version 6, 4th ed.). SAS Institute Inc., Cary, N.C.
- 34.Siba, P. M., D. W. Pethick, and D. J. Hampson. 1996. Pigs experimentally infected with Serpulina hyodysenteriae can be protected from developing swine dysentery by feeding the highly digestible diet. Epidemiol. Infect. 116:207-216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.van Winsen, R. L., B. A. Urlings, L. J. A. Lipman, J. M. A. Snijders, D. Keuzenkamp, J. H. M. Verheijden, and F. van Knapen. 2001. Effect of fermented feed on the microbial population of the gastrointestinal tracts of pigs. Appl. Environ. Microbiol. 67:3071-3076. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Williams, B. A., M. W. A. Verstegen, and S. Tamminga. 2001. Fermentation in the large intestine of single-stomached animals and its relation to animal health. Nutr. Res. Rev. 14:207-227. [DOI] [PubMed] [Google Scholar]
- 37.Zak, J. C., M. R. Willig, D. L. Moorhead, and H. G. Wildman. 1994. Functional diversity of microbial communities: a quantitative approach. Soil Biol. Biochem. 26:1101-1108. [Google Scholar]