Dietary protein concentration affects intestinal microbiota of adult cats: a study using DGGE and qPCR to evaluate differences in microbial populations in the feline gastrointestinal tract

Authors


K. S. Swanson, Department of Animal Sciences, University of Illinois, 162 Animal Sciences Laboratory, 1207 W. Gregory Drive, Urbana, IL 61801. Tel: 217 333 4189; Fax: 217 333 7861; E-mail: ksswanso@uiuc.edu

Summary

The objective of this study was to identify qualitative and quantitative differences in microbial populations of adult cats fed diets containing different protein concentrations. Following a 4 week baseline period, eight healthy adult domestic short-hair queens (>1-year-old) were randomly allotted to a moderate-protein (MP; n = 4) or high-protein (HP; n = 4) diet for 8 weeks. Fresh faecal samples were collected after baseline and 8 weeks on treatment and stored at −80 °C. Following DNA extraction, samples were analyzed using denaturing gradient gel electrophoresis to distinguish qualitative changes between diets. Quantitative polymerase chain reaction was used to measure E. coli, Bifidobacterium, Clostridium perfringens, and Lactobacillus populations. Compared to baseline, cats fed MP had a bacterial similarity index of 66.7% as opposed to 40.6% similarity for those fed HP, exhibiting marked changes in intestinal bacteria of cats fed HP. Bifidobacterium populations were greater (p < 0.05) in cats fed MP versus HP (9.44 vs. 5.63 CFU/g). Clostridium perfringens populations were greater (p < 0.05) in cats fed HP than MP (12.39 vs. 10.83 CFU/g). In this experiment, a high-protein diet resulted in a dramatic shift in microbial populations. Decreased Bifidobacterium population in cats fed HP may justify prebiotic supplementation for such diets.

Introduction

Intestinal microbiota play an important role in the health of all mammals, but species differences do exist. In ruminants and large herbivores, for example, microbes are crucial in extracting energy from fibrous substrates through anaerobic fermentation. Without it, these animals would be unable to meet their energy requirements. In most non-ruminant omnivores and carnivores, the amount of energy derived from microbial fermentation is low because of the low dietary fiber concentrations and (or) variations in gastrointestinal anatomy and physiology. Despite these differences, microbial populations play an important role in several gastrointestinal functions, including pathogen resistance and the immune system (Cebra, 1999). Because dogs and humans are omnivores consuming plant and animal ingredients, numerous studies have been performed to test the effects of diet on gut microbial populations. Being true carnivores, consuming little vegetable matter, the feline microbiota has often been ignored. Thus, little is known in regards to feline microbiota, the changes that occur from dietary manipulation, and any health-related implications related to diet composition.

Traditional cultivation methods have provided a moderate level of understanding of microbial populations and are the basis of many of today’s practices. However, the culturable fraction of intestinal microbiota is still very low (Zoetendal et al., 2004). Many species cannot be cultured because of unknown growth requirements and death associated with the stress of cultivation. Nucleic acid-based techniques that are culture-independent are now available and are critical in enhancing our understanding of complex microbial ecosystems (Zoetendal et al., 2004). Molecular techniques enable identification of gut microbial species using isolated DNA sequences. Quantitative polymerase chain reaction (qPCR), for example, allows the quantification of microbial populations, while denaturing gradient gel electrophoresis (DGGE) and 16S rDNA sequencing enable the measurement of gut diversity and species identification, respectively. Thus, the objective of this experiment was to use nucleic acid-based techniques to measure quantitative and qualitative changes in faecal microbial populations of cats fed moderate- or high-protein diets.

Materials and methods

Animals and Diet

Eight adult (>1-year-old) female domestic short-hair cats (Liberty Research; Waverly, NY, USA) were used in this experiment. All cats were fed a control diet (Harlan Teklad Global Cat Diet –#2060) for 4 weeks (baseline). The #2060 formula is an extruded diet composed of ground corn, poultry by-product meal, soybean meal, animal fat, corn gluten meal, ground wheat, wheat middlings, fish meal, and dried skim milk. After baseline, queens were randomly allotted to one of two treatment diets formulated and manufactured by Natura Manufacturing (Table 1; Fremont, NE, USA) and fed for 8 weeks. A moderate-protein (MP) diet was formulated to contain approximately 30% crude protein, while a high-protein (HP) diet was formulated to contain approximately 50% crude protein. Table 2 displays the analyzed chemical contents of all diets fed in this experiment. All diets were formulated to meet or exceed all nutrient requirements for growth and reproduction according to the Association of American Feed Control Officials (AAFCO, 2007). Food was offered ad libitum and any food refusals were weighed daily and food intake calculated. Cats were housed in individual cages (1.0 × 0.8 × 0.7 m) in temperature-controlled rooms with a 12-h light: 12-h dark cycle in Edward R. Madigan Laboratory in the University of Illinois, Champaign-Urbana campus. The University of Illinois Institutional Animal Care and Use Committee approved all animal care procedures before initiation of the study.

Table 1.   Ingredient composition of treatment diets fed in experiment
Ingredient nameModerate proteinHigh protein
  1. *Trouw Nutrition (Highland, IL, USA).

Potato product32.883.19
Chicken meal, low ash15.9048.96
Chicken meal14.9415.83
Chicken fat12.798.57
Potato starch7.507.50
Dried egg product5.005.00
Fish meal5.005.00
Beet pulp3.003.00
Digest1.001.00
Premium cat vitamin premix*0.650.65
Potassium chloride0.390.10
Fish oil0.340.66
Salt0.250.25
Premium cat mineral premix*0.150.15
Dried chicory root0.100.10
dl-Methionine0.070.00
Natural antioxidants, dry0.030.05
Table 2.   Analyzed chemical composition of diets fed in experiment
ItemControlModerate-proteinHigh-protein
Dry matter (DM), %92.2794.2294.83
Organic matter (DM), %91.5891.2889.19
Crude protein (DM), %37.5734.3452.88
Acid hydrolyzed fat (DM), %15.5119.2323.50
Total dietary fiber (DM), %8.646.882.01
Gross energy, kcal/g5.075.215.54

Chemical analyses

Diet samples were ground with dry ice through a 2-mm screen in a Wiley mill (model 4, Thomas Scientific, Swedesboro, NJ, USA) in preparation for chemical analyses. Diets were analyzed for dry matter (DM) and organic matter (OM) according to AOAC (1984). Crude protein was determined according to AOAC (1995) using a Leco Nitrogen/Protein Determinator (model FP-2000, Leco Corporation, St. Joseph, MI, USA). Fat concentrations were determined by acid hydrolysis (AACC, 1983) followed by ether extraction (Budde, 1952). Total dietary fiber (TDF) was determined according to Prosky et al. (1992). Gross energy was determined by use of a bomb calorimeter (Model 1261, Parr Instrument, Moline, IL, USA).

Faecal collection and bacterial genomic DNA extraction

Fresh faecal samples were collected after baseline and 8 week treatment periods. Samples were stored at −80 °C until DNA extraction was performed. Bacterial genomic DNA was extracted and isolated from faecal samples (50 mg) using the QIAamp DNA Stool Mini-Kit (Qiagen, Valencia, CA, USA). Isolated DNA concentration (ng/μl) and purity (260 nm/280 nm; ratio of 1.8 to 2.2 was acceptable) were measured using a ND-1000 spectrophotometer (Nanodrop Technologies, Wilmington, DE, USA). Template DNA was diluted to 10 ng/μl and stored at −20 °C until further analysis.

Quantitative polymerase chain reaction analysis

Escherichia coli, Bifidobacterium genus, Lactobacillus genus, and Clostridium perfringens were quantified via qPCR using specific primers (Table 3). Amplification was performed in a set of triplicate reactions for each bacterial group within each sample according to the procedures of Deplancke et al. (2002). For amplification, 10 μl final volume containing 2X SYBR Green PCR Master Mix (Applied BioSystems, Foster City, CA, USA), 15 pmol of each primer and 10 ng of template DNA were used. Pure cultures of each bacterium were utilized to create a five-fold dilution series (10 × 100–10 × 105) in triplicate from target species. DNA from each serial dilution was extracted using a QIAamp DNA Stool Mini-kit (Qiagen) and amplified along with faecal DNA samples using an ABI PRISM 7900HT Sequence Detection System (Applied BioSystems). The colony forming units (CFU) of each standard curve serial dilution was determined by plating the E. coli grown on Luria-Bertani Medium [10 g/l tryptose, 5 g/l yeast extract, 5 g/l NaCl (ph=7)], Lactobacillus genus on Difco Lactobacilli MRS broth (Becton Dickenson Company, Sparks, MD, USA), and the C. perfringens and Bifidobacterium genus on Difco Reinforced Clostridial Medium (Becton Dickenson Company). Cycle threshold (Ct) values were plotted against standard curves for quantification (CFU/g faeces) of the target bacterial DNA from faecal samples.

Table 3.   Primers used for quantitative polymerase chain reaction (qPCR) analysis
Target speciesPrimerSequence (5′-3′)Reference
Bifidobacterium genusg-Bifid-FCTCCTGGAAACGGGTGGMatsuki et al., 2002
g-Bifid-RGGTGTTCTTCCCGATATCTACA
Clostridium perfringensCP1AAAGATGGCATCATCATTCAACWang et al., 1994
CP2TACCGTCATTATCTTCCCCAAA
Lactobacillus genusLab-0159GGAAACAG(A/G)TGCTAATACCGCollier et al., 2003
Univ-0515ATCGTATTACCGCGGCTGCTGGCA
Escherichia coliE. coli FGTTAATACCTTTGCTCATTGAMatsuki et al., 2002
E. coli RACCAGGGTATCTAATCCTGTT

PCR Amplification of 16S rDNA V3 region

Template DNA was amplified using PCR involving primers (25 pmol of each per reaction mixture) targeting the conserved regions flanking the variable region 3 (V3) of the 16S rDNA gene as described by Muyzer et al. (1996) and Simpson et al. (1999). ‘Eubacterial’ primer 341F (5′-CCTACGGGAGGCAGCAG-3′) and ‘universal’ primer 543R (5′-ATTACCGCGGCTGCTGG-3′), the number corresponding to E.coli 16S rDNA (Muyzer et al., 1996) were obtained from Operon Biotechnologies (Huntsville, AL, USA). To provide a more stable melting behaviour of fragments in DGGE, a GC-rich sequence (GC-clamp) was attached to the 5′- end of 341F (5′-CGCCCGCCGCGCGCGGCGGGCGGGGCGGGGGGCACGGGGGG-3′). Primers 341F and 543R amplify PCR products containing approximately 200 base pairs. Parameters for PCR amplification were as follows: (i) denaturation at 94 °C for 5 min; (ii) denaturation at 94 °C for 30 s; (iii) annealing at 65 °C for 30 s; −0.5 °C/cycle (touchdown); (iv) extension at 72 °C for 30 s; (v) repeat steps (ii) to (iv) for 20 cycles; (vi) denaturation at 94 °C for 30 s; (vii) annealing at 55 °C for 30 s; (viii) extension at 72 °C for 30 s: (ix) repeat steps (vi) to (viii) for 10 cycles; (x) extension at 72 °C for 10 min; (xi) 4 °C final.

Removal of Single-stranded DNA (ssDNA) from PCR product

Single-stranded DNA may remain in samples following PCR and can present problems for image analysis. Thus, mung bean nuclease (MBN; Stratagene, La Jolla, CA, USA) treatment was performed on PCR products to eliminate ssDNA as described by Simpson et al. (1999). The reaction mixture contained 10 μl PCR product, 1.5 μl 10X Tris-acetate EDTA (TAE) buffer, 1.0 μl MBN (diluted 1:20 in sterile deionized water), and 2.5 μl sterile deionized water. The reaction mixture was then incubated at 37 °C for 10 min. After incubation, 15 μl of 2× DGGE loading dye (0.05% bromophenol blue, 0.05% xylene cyanol, 70% glycerol, sterile deionized water) was added to stop the reaction. Samples were refrigerated until DGGE analysis (<30 day storage).

Denaturing gradient gel electrophoresis

Mung bean nuclease-treated amplicons were separated on polyacrylamide gels [non-deionized 40% acrylamide/bis stock solution 37.5:1 (Bio-Rad Laboratories, Hercules, CA, USA)] on a 29% to 48% gradient. A 100% denaturing solution contains 40% (vol/vol) formamide and 7 m urea. Gradient was formed using a Bio-Rad Gradient Former Model 475 (Bio-Rad Laboratories) and gels were polymerized onto a gel support film (FMC, Rockland, ME, USA). Aliquots of 12 μl of MBN-treated samples and 3 μl of 2× DGGE loading dye were mixed and loaded into sample wells. Electrophoresis was performed using a D-Code Universal Mutation Detection System (Bio-Rad Laboratories) using 0.5× TAE run buffer at 60 °C for 10 min at 50 V and then at 60 °C for 3 h at 150 V. Reference ladders were used to standardize band migration and gel curvature among different gels according to Simpson et al. (1999) and were comprised of cloned double stranded DNA amplicons from bacteria that are known to span the length of the gel. Bands were fixed in the gel using a 10% ethanol-0.5% acetic acid solution, silver stained, developed, and scanned using a Bio-Rad GS-800 Scanner.

16S rDNA sequencing

Bands of interest were extracted from the DGGE gel using a modified QIAquick Gel Extraction Kit (Qiagen) for extraction from polyacrylamide gel. DNA was re-amplified and run through DGGE a second time to ensure purity of DNA bands. Sequencing was performed using primers 341F (no GC-clamp) and 534R at the UIUC Core Sequencing Facility in Edward R. Madigan Laboratory. All 16S rDNA sequences were subject to nucleotide basic local alignment search tool (BLAST; blastn) search against Genbank (http://www.ncbi.nlm.nih.gov) .

Statistical analyses

Dice’s similarity index was calculated using Quantity One software (Bio-Rad Laboratories) on pairs of lanes on DGGE gel to compare baseline versus treatment and MP versus HP. Dice’s similarity index was calculated as:

image

where a is the number of common bands, x is the number of bands in set 1, and y is the number of bands in set 2. Quantative polymerase chain reaciton results were analyzed using the mixed models of sas (SAS Institute, Cary, NC). Baseline values were used as a covariate in the model. A probability of p < 0.05 was accepted as being significant.

Results

Quantitative polymerase chain reaction

Baseline E. coli, Bifidobacterium, Lactobacillus, and C. perfringens concentrations were not different among groups (Table 4). Using baseline values as covariates, Bifidobacterium concentrations were greater (p < 0.05) in cats fed MP versus HP diet after the 8-week-treatment period. Conversely, C. perfringens population were greater (p < 0.05) in cats fed HP versus MP diets. Lactobacillus and E. coli concentrations were not affected by diet in this study.

Table 4.   Faecal microbial concentrations (log CFU/g faeces) in cats fed a moderate- or high-protein diet
SpeciesBaseline (n = 8)TreatmentSEM
MP (n = 4)HP (n = 4)
  1. a,bMeans lacking a common superscript in each row are different (p < 0.05)

Escherichia coli9.438.369.590.888
Bifidobacterium genus10.409.44b5.63a0.334
Lactobacillus genus11.3812.7412.150.569
Clostridium perfringens8.8010.83a12.39b0.185

Denaturing gradient gel electrophoresis

Dice’s similarity coefficient is representative of the number of bands and the number of similar bands in a pair of lanes. As the diversity between the pair of lanes increases, the similarity coefficient decreases. Figure 1 is representative of the results observed in this experiment, providing an example of one MP treatment sample and three HP treatment samples. Similarity of baseline to treatment diet was calculated with an average similarity coefficient of 66.67% in cats fed MP compared to 40.55% in cats fed HP. Visual examination indicates several bands that appear (Fig. 1b) or disappear (Fig. 1c) from baseline to treatment in cats fed HP.

Figure 1.

 Representative denaturing gradient gel electrophoresis polyacrylamide gel containing DNA from cats fed a moderate (MP)- or high-protein (HP) diet. L = bacterial standard ladder; B = baseline; T = treatment. Box a: Bands are highly similar between lanes in cat fed MP. Box b: Three distinct DNA bands appeared after HP. Box c: Three distinct DNA bands disappeared after HP.

16S rDNA sequencing

Using the DGGE output, bands of interest (i.e., prominent bands that appeared or disappeared with HP treatment) were subject to 16S rDNA sequencing and BLAST. Phyla and species were identified (when possible) based on the similarity percentage of the 200 base pair template sequence with sequences available in Genbank. Table 5 shows the results of three HP-treated cats after seven bands of similar melting behaviour between cats were extracted and sequenced. The majority of samples that increased after HP consumption were members of α-proteobacteria, δ-proteobacteria, or clostridia. However, HP feeding appeared to result in a marked disappearance of γ-proteobacteria.

Table 5.   Microbial identification of DNA bands undergoing 16S rDNA sequencing and BLAST
BandCatPhyla Genus/species% SimilarityResponse to HP
  1. HP, high protein; BLAST, basic local alignment search tool.

13FusobacteriaUndeterminedNA
14UndeterminedUndeterminedNA
17α-ProteobacteriaNovosphingobium sp.93
23UndeterminedUndeterminedNA
24δ-ProteobacteriaHaliangium ochraceum100
27UndeterminedUndeterminedNA
33UndeterminedUndeterminedNA
34α-ProteobacteriaRhizobiales bacterium94
37UndeterminedUndeterminedNA
43UndeterminedUndeterminedNA
44γ-ProteobacteriaSuccinivibrio dextrinosolvens93
47γ-ProteobacteriaSuccinivibrio dextrinosolvens95
53α-ProteobacteriaMethylobacterium89
54γ-ProteobacteriaAnaerobiospirillium94
57γ-ProteobacteriaSuccinovibrio dextrinosolvens87
63FirmicutesMegasphaera elsdenii98
64γ-ProteobacteriaPsychromonas ingrahamii100
67γ-ProteobacteriaSuccinovibrio dextrinosolvens89
73FirmicutesClostridium hiranonis97
74FirmicutesClostridium paraputrificium96
77FirmicutesClostridium sp.97

Discussion

Although feline species are strict carnivores, the gut microbial populations play an important role in digestion through fermentative processes and overall health of the animal. Through in vitro studies using faecal inoculum, microbes inhabiting the feline GI tract have been shown to be very active and produce a similar fermentative profile as other species tested, including dogs, pigs, and humans (Sunvold et al., 1995b). Feline microbes are also able to ferment a variety of fibrous substrates, generating similar concentrations of short-chain fatty acids when compared to dogs (Sunvold et al., 1995a). Although the fermentative capacity of feline microbes appears to be comparable to other species when provided with carbohydrate-based substrates, we were interested in studying the effects of dietary protein concentrations on gut microbiota.

Amstberg et al. (1980) demonstrated that the amount and type of dietary protein impacts C. perfringens concentrations in dog faeces. Protein quality may greatly influence intestinal microbiota by changing the amount of protein that reaches the lower bowel. Lower quality proteins that are poorly digested will provide more protein to microbes inhabiting the lower bowel and increase the occurrence of proteolytic bacteria, some of which may be pathogenic and/or produce putrefactive compounds. Higher protein concentrations reaching the lower gut will increase the production of ammonia, indoles, phenols, and sulphur-containing compounds that are toxic at high levels and are associated with several disease states. Processing method has also been shown to influence faecal microbiota, as canned versus extruded diets have been shown to alter microbial activity in cats (Backus et al., 2002).

Because high-protein diets have been reported to increase Clostridium populations (Zentek et al., 2004), increased C. perfringens was an expected outcome of cats fed HP in the current experiment. Although cats fed HP in the current experiment were healthy, clostridial diseases have been reported in cats. For example, Clostridium sordellii and C. perfringens have been identified as intestinal pathogens in lions (Panthera leo) and cheetahs (Acinonyx jubatus jubatus) respectively (de la Fe et al., 2006; Citino, 1994). Thus, this change in the intestinal microbial ecology may be thought of as a negative response.

In contrast to most clostridial species, Bifidobacterium and Lactobacillus are generally regarded as beneficial microbes because of their ability to exclude harmful bacteria by producing various antimicrobial agents (Rastall, 2004). Additionally, these microbes have a high capacity to ferment dietary fibers and are considered to do so more effectively than pathogenic species (Buddington and Sunvold, 1998). In pets, the greatest focus has been on increasing bifidobacteria and lactobacilli populations through the use of prebiotics and probiotics. In cats, positive effects have been shown by feeding lactosucrose, which increased Bifidobacterium and decreased clostridia concentrations (Terada et al., 1993). Similarly, C. perfringens concentrations were decreased in cats supplemented with fructo-oligosaccharides (Sparkes et al., 1998). Because we hypothesized a shift from carbolytic to proteolytic bacteria in cats fed high protein, the decrease in Bifidobacterium populations in HP-fed cats was not surprising. Given these results, prebiotic supplementation to maintain a high level of the bifidobacteria may be prudent when feeding high protein diets.

While the microbial groups mentioned above are often used as an indirect measure of intestinal health, we were interested in identifying other microbial species affected by dietary manipulation using DGGE. The Dice’s similarity index used to evaluate DGGE gels is a broad measure of microbial diversity, being based on the similarity between lanes of interest (i.e., baseline vs. treatment samples). According to Dice’s similarity scores, DGGE analyses revealed changes in faecal microbiota of cats fed MP and even greater changes in those fed HP. Because computer software is used to detect bands and to determine similar bands, these analyses provide an objective method for measuring diversity of bacterial populations. Similarity coefficients of cats fed HP (40.55%) and MP (66.67%) show a dramatic change in bacterial population in cats fed both diets as compared to baseline. Figure 1a highlights the two lanes (baseline and treatment) of a cat fed MP. Our data report a similarity (72.7%) between baseline and treatment samples in this particular cat. In Fig. 1b,c, however, marked differences between baseline and treatment samples were observed in all cats fed HP. Consistent bands appeared in all three HP-fed cats, as shown in Fig. 1b, the species of which were speculated to be proteolytic in nature. In contrast, Fig. 1c shows the disappearance of several bands.

Again, we hypothesized a shift from carbolytic to proteolytic bacteria in cats fed HP. Because these changes in band intensity were of interest, we continued with 16S rDNA sequencing to identify species affected by HP. At this time, our results are inconclusive as to the nature of the following species and their impact on gut health. However, we hope that by identifying species affected by HP feeding, we may be able to get a better understanding of intestinal microbial shifts. The sequence identification accuracy was limited in this study because of the short sequences (ca. 200 base pairs) isolated from DGGE. For increased accuracy, complete 16S rDNA sequences may be isolated in future studies. Despite these shortcomings, interesting results were obtained. DNA sequencing and BLAST analyses of isolated bands suggested an increased prevalence of several α- and δ-proteobacteria and clostridial species with HP feeding. In contrast, HP feeding appeared to decrease the prevalence of γ-proteobacteria.

In Table 5, species identity of the seven prominent bands isolated from each of three cats is presented. According to 16S rRNA sequence, Novosphingobium sp., Haliangium ochraceum, and Clostridia were increased following HP feeding. Prior to this study, the two former microbes had not been reported to reside in the feline gastrointestinal tract. Novosphingobium sp. has been used in groundwater remediation systems (Tiirola et al., 2002). Novosphingobium taihuense have been found to degrade aromatic compounds and may be important for bio-geochemical cycles (Zhi-Pei et al., 2005). While the species and strains measured in this experiment may be different than those used in water treatment, the increased presence of Novosphingobium sp. in cats fed HP may be because of the high level of aromatic end-products of protein fermentation (e.g., phenols, indoles) that are associated with high-protein diets. This is a seemingly positive effect, as these bacteria may decrease the concentration of these harmful compounds.

Hygrohypnum ochraceum have been commonly studied in highly salinated environments, questioning their occurrence in the feline gut. However, H. ochraceum studied thus far have been shown to possess a high activity of proteases, such as leucine arylamidase, valine arylamidase, trypsin, and chymotrypsin, while having low carbolytic activity (Fudou et al., 2002). Fudou et al. (2002) also reported that H. ochraceum and Hygrohypnum tepidum were both found to effectively decompose yeast cells (Saccharomyces cerevisiae) and other bacterial cells (E. coli). Thus, in addition to protein-based substrates, these bacteria may feed off other bacteria present in the gut.

A number of species appeared to be decreased after HP, including methylotrophic Methylobacterium, a microbe previously found in the oral cavity (Anesti et al., 2005), anaerobic carbohydrate fermentor Anaerobiospirillium (Malnick, 1997), and lactate-degrading Megasphaera elsdenii (Ouwerkerk et al., 2002). Because sequence similarities were all similar to our isolated sequences, there is some uncertainty to which species were changed with HP feeding. The decrease of available nutrients for these microbial populations, as well as the sheer numbers of competing proteolytic bacteria, may be the cause for the decrease in these populations. Psychromonas ingrahamii have been reported to ferment several carbon sources such as lactose, sucrose, maltose, d-galactose, and d-glucose (Auman et al., 2006), so with the increase of protein and decrease of carbohydrates, this microbial population would decrease. Succinivibrio dextrinosolvens were also one of the main members of the γ-proteobacteria that was decreasing after HP treatment. This is expected because of its high amylolytic activity and starch utilization (Cotta, 1988). Further research is needed to study these effects in more depth, but these results are consistent with the carbolytic and proteolytic activity of each of the microbes presented.

To conclude, cats fed high-protein diets had increased C. perfringens and decreased Bifidobacterium faecal populations. Thus, prebiotic supplementation would appear to be beneficial in cats fed high-protein formulas. High-protein feeding also appeared to increase microbial diversity as compared to those fed a moderate protein concentration. Using DGGE and 16S rRNA sequencing, several new species affected by dietary protein concentration were also identified. Although further research evaluating the entire 16S gene sequence is required to increase identification confidence, these results should be used as a platform for such efforts.

Ancillary