Animals and diets

The present study was performed at the WALTHAM^® Centre for Pet Nutrition and all animal care and study procedures were approved by the Animal Welfare and Ethics Review Body. A total of twelve healthy adult Miniature Schnauzer dogs (eight males; four females; mean age: 3·3 (1·6) years; mean body weight: 8·4 (1·2) kg; mean body condition score: D/ideal) were included in the study. A balanced 3 × 3 Latin square design with 21 d trial periods was conducted. Each period consisted of a diet adaptation phase (days 0–17) and a total and fresh (within 15 min of defecation) faecal collection phase (days 18–21). During the adaptation phase, dogs were pair-housed overnight, with free access to indoor and outdoor paddock areas, where they were also group-housed during the day. During the collection phase, dogs were individually housed overnight and group-housed during the day.

All dogs were fed a complete and balanced canned diet twice daily (08:30 and 15:00) to maintain body weight throughout the study. The diet consisted of 92·4 g protein/4184 kJ (1000 kcal) metabolisable energy and 65·0 g fat/4184 kJ (1000 kcal) metabolisable energy. Intake for each dog was recorded at each meal. Prior to receiving their meals, dogs were given 0, 10 or 20 g of biscuit treats per d (split into two doses) that provided approximately 0, 2·5 or 5 g RS/d from the biscuits. The biscuits consisted of 43·2 g protein/4184 kJ (1000 kcal) metabolisable energy and 33·5 g fat/4184 kJ (1000 kcal) metabolisable energy. Dogs were weighed and assessed for body condition score⁽ Reference German, Holden and Moxham ^{11 )} weekly. The chemical composition of the maintenance diet and biscuits can be found in Table 1.

Table 1. Chemical composition of maintenance diet and biscuits

AAFCO, Association of American Feed Control Officials.

*Metabolisable energy (ME)_AAFCO=35·564 kJ ME/g fat + 14·644 kJ ME/g CP + 14·644 kJ ME/g nitrogen-free extract.

Sample collection

During the 4-d collection phase, all faecal output was collected, weighed, scored and frozen at −20°C for further analysis; one fresh faecal sample from each dog per period was also collected. Faecal quality was assessed on each sample according to visual appearance as described by Rolfe et al. ⁽ Reference Rolfe, Adams and Butterwick ^{12 )} with grade 1 representing ‘hard, dry, crumbly faeces’ and grade 5 ‘watery diarrhoea’. To account for all the discernible points in between, half-scores are used, giving a total of nine possible categories, with a score of 1·5–2·5 considered ideal.

The fresh faecal sample was collected within 15 min of defecation on day 1 of the 4-d collection phase and prepared immediately to minimise loss of volatile components. Samples were weighed and pH measured using a Mettler Toledo FG2/EL2pH meter equipped with a Mettler Toledo LE 438 probe (Mettler Toledo) after mixing 2 g of fresh faeces into 18 ml of sterile water. Fresh faecal DM was determined (105°C). Aliquots of faeces for analysis of phenols and indoles were frozen at −20°C immediately after collection. An aliquot (5 g) of faeces was mixed with 5 ml 2 m HCl for ammonia, SCFA and branched-chain fatty acid determination and stored at −20°C until analysed. Aliquots of fresh faeces were transferred to sterile cryogenic vials (Nalgene) and frozen at −80°C until DNA extraction for microbial analysis.

Chemical analyses

Dried diet and faecal samples were ground through a 2-mm screen in a Wiley Mill (intermediate, Thomas Scientific). Samples were analysed according to procedures by the Association of Official Analytical Chemists (AOAC) for DM (105°C), organic matter and ash^{( 13 )}. Crude protein content was calculated from Leco total N values (TruMac^® N, Leco Corporation)^{( 13 )}. Total lipid content (acid hydrolysed fat) of the samples was determined according to the methods of the American Association of Cereal Chemists (AACC)^{( 14 )} and Budde⁽ Reference Budde ^{15 )}. Gross energy of the samples was measured using an oxygen bomb calorimeter (model 1261, Parr Instruments). Dietary fibre concentrations (total dietary fibre, soluble dietary fibre and insoluble dietary fibre) were determined according to the method of Prosky et al. ⁽ Reference Prosky, Asp and Schweizer ^{16 )}.

SCFA and branched-chain fatty acid concentrations were determined by GC according to the method of Erwin et al. ⁽ Reference Erwin, March and Emery ^{17 )} using a Hewlett-Packard 5890A series II gas chromatograph (Palo Alto) and a glass column (180 cm × 4 mm i.d.) packed with 10 % SP™-1200/1 % H₃PO₄ on 80/100+ mesh Chromosorb WAW (Supelco Inc.). Phenol and indole concentrations were determined using GC according to the method of Flickinger et al. ⁽ Reference Flickinger, Schreijen and Patil ^{7 )}. Ammonia concentrations were determined according to the method of Chaney & Marbach⁽ Reference Chaney and Marbach ^{18 )}.

Starch

Biscuit and diet subsamples were ground through a 0·5-mm screen in a Wiley Mill (intermediate, Thomas Scientific). The method of Muir & O'Dea⁽ Reference Muir and O'Dea ^{19 ,} Reference Muir and O'Dea ^{20 )} was used to determine the amount of starch digested in the stomach and small intestine by measuring glucose in the supernatant resulting from acid–enzyme digestion of the substrate. Briefly, 0·2 g of each substrate was weighed in duplicate and exposed to pepsin/HCl, amyloglucosidase and α-amylase digestion. Tubes containing reagents but no substrate were run as blanks. All tubes were incubated for 15 h at 37°C and then centrifuged for 15 min. Glucose concentrations in the supernatant were determined by reading the absorbance of individual samples at 450 nm on a DU 640 spectrophotometer (Beckman Instruments) and comparing those values against a glucose standard curve. Digestible starch was determined by subtracting (free glucose × 0·9) from (total glucose/original sample weight) present in the supernatant after 15 h of digestion. Because the measurement of glucose was used to determine starch content, a correction factor of 0·9 for the difference in weight between a free glucose unit and a glucose residue in starch was used. Total starch content of samples was determined using the method of Thivend et al. ⁽ Reference Thivend, Christiane and Guilbot ^{21 )} with amyloglucosidase. RS was calculated by subtracting (digestible starch +(free glucose × 0·9)) from total starch. The released glucose value corresponds to the amount of glucose resulting from hydrolytic starch digestion that is available for absorption in vivo.

Microbial analyses

Faecal DNA was extracted from freshly collected samples that had been stored at −80°C until analysis, using the PowerLyzer™ PowerSoil^® DNA Isolation Kit (MO BIO Laboratories, Inc.) according to the manufacturer's instructions. Extracted DNA was quantified using a Qubit^® 2·0 Fluorometer (Life Technologies™, Invitrogen™). Quantitative PCR was performed using specific primers for Bifidobacterium spp.⁽ Reference Matsuki, Watanabe and Fujimoto ^{22 )}, Lactobacillus spp.⁽ Reference Collier, Smiricky-Tjardes and Albin ^{23 )}, Escherichia coli ⁽ Reference Malinen, Kassinen and Rinttila ^{24 )} and Clostridium perfringens ⁽ Reference Wang, Cao and Franklin ^{25 )}. While Bifidobacterium and Lactobacillus are generally considered to be ‘beneficial’ microbes, E. coli and C. perfringens represent potential pathogens, and are commonly measured in prebiotic studies, thus, were also analysed for the present study. Amplification was performed according to DePlancke et al. ⁽ Reference DePlancke, Vidal and Ganessunker ^{26 )}. Briefly, a 10-μl final volume contained 5 µl of 2 × SYBR Green PCR Master Mix (Applied Biosystems), 0·05 µl of bovine serum albumin, 15 pmol of the forward and reverse primers for the bacterium of interest, and 10 ng of extracted faecal DNA. Standard curves were obtained by harvesting pure cultures of the bacterium of interest in the log growth phase in triplicate, followed by serial dilution. Bacterial DNA was extracted from each dilution using a DNA extraction kit (Qiagen) and amplified with the faecal DNA to create triplicate standard curves (ABI PRISM 7900HT Sequence Detection System, Applied Biosystems). Colony-forming units in each dilution were determined by plating on specific agars; lactobacilli MRS (Difco, BD) for Lactobacillus, reinforced clostridial medium (Bifidobacterium, C. perfringens) and Luria Bertani medium (E. coli). The calculated log colony-forming units per ml of each serial dilution were plotted against the cycle threshold to create a linear equation to calculate colony-forming units per gram of dry faeces. Although the standard curves are meant to represent a group of bacteria, our quantitative PCR assays were based on a single bacterial strain within each group. As operon copy number is different among strains, a potential bias does exist in our assay, to minimise this each dog acted as its own control.

Calculations and statistical analysis

Apparent total tract energy and apparent macronutrient digestibility values were calculated using the following equation: nutrient intake (g DM/d) − nutrient output (g DM/d)/nutrient intake (g DM/d) × 100. Data were analysed using the MIXED procedure of SAS (version 9·3, SAS Institute Inc.). Faecal score data were compared using the GLIMMIX procedure of SAS. The statistical model included period and dog as random effects, whereas treatment was a fixed effect. Data were analysed using the type 3 test of the MIXED procedure. All treatment least-squares means were compared using preplanned contrasts that tested for linear effects of RS supplementation. Means were separated using a protected least-squares difference with a Tukey's adjustment. Data were analysed using the UNIVARIATE procedure to produce a normal probability plot based on residual data and visual inspection of the raw data. A probability of P ≤ 0·05 was accepted as being statistically significant and P ≤ 0·10 accepted as trends.