Effects of Brewer’s spent grain and carrot pomace on digestibility, fecal microbiota, and fecal and urinary metabolites in dogs fed low- or high-protein diets

Abstract

Brewer’s spent grain (BSG) and carrot pomace (CAP) were used as fiber sources in low- or high-protein diets in dogs. Ten adult Beagles were involved in 5 feeding periods of 19 d in a cross-over design. Experimental diets contained 7.5% of total dietary fiber (TDF) from BSG or CAP and 20% or 40% of crude protein in dry matter. A diet with 3.5% TDF from both fiber sources and 20% crude protein was used as reference. Fecal dry matter was 27% higher for diets with BSG compared to CAP (P < 0.001). Apparent fecal digestibility of crude protein was 7% to 11% higher in diets with 40% protein concentration (P < 0.001), while apparent digestibility of crude fat was 2% to 3% higher for diets with CAP (P < 0.001). Carrot pomace increased the apparent fecal digestibility of TDF, phosphorus, and magnesium (P < 0.001), while 40% protein diets had a positive impact on TDF and sodium and a negative effect on magnesium apparent fecal digestibility (P < 0.001). Inclusion of CAP increased fecal short-chain fatty acids (P = 0.010), mainly acetate (P = 0.001). i-butyrate (P = 0.001), i-valerate (P = 0.002), biogenic amines (P < 0.001), and ammonium (P < 0.001) increased with higher dietary protein levels. Diet-induced changes in the fecal microbiome were moderate. Relative abundance of Bifidobacteriales was higher for the low-protein diets (P = 0.001). To conclude, BSG and CAP can be used as fiber sources in canine diets and are well tolerated even at higher inclusion rates, the effect on microbial protein fermentation seems to be limited compared to the dietary protein level.

INTRODUCTION

The term “dietary fiber” is described as carbon–hydrogen–oxygen polymers that are resistant to digestion of endogenous enzymes (Jones, 2014). Brewer’s spent grain (BSG) is a by-product from the brewing industry and is rich in non-fermentable dietary fiber (Makowska et al., 2013), protein, carbohydrates, and minerals. Depending on the brewing technology, nutrient composition can vary. Based on own data, BSG contains 258 g crude protein (CP), 133 g crude fiber, and 495 g total dietary fiber (TDF) per kilogram. Today, it is mainly used as feed material for dairy cattle (Mussatto, 2014), but due to its composition, dried BSG could be a promising fiber source for canine diets. Carrot pomace (CAP) is known as a fermentable, pectin-rich fiber that results as a by-product from carrot juice industry (Jafari et al., 2017). It contains 99 g CP, 214 g crude fiber, and 623 g TDF per kilogram (own data), again, composition may vary depending on production technology. Results from an in vitro study with canine gut microbiota revealed high fermentability and production of short-chain fatty acids (SCFAs) from CAP compared to other fruit and vegetable by-products (Swanson et al., 2001). Protein-rich diets are common in dog nutrition. These diets can increase putrefactive metabolites (ammonia, indoles, amines) due to amino acid degradation by the intestinal microbiota (Cummings and Macfarlane, 1991). High amounts of dietary protein, especially when using sources with lower ileal digestibility, can lead to a microbial dysbiosis, which may result in a dysbalanced microbiome, unformed feces, and diarrhea (Hang et al., 2013). For dogs in maintenance metabolism, protein contents of 100 g/kg dry matter are considered sufficient (NRC, 2006). Practical diets are usually significantly higher, especially considering the increasing trend to raw food and low-carbohydrate diets for dogs. Changes in the sense of intestinal microbial dysbiosis were found in diets with 769 and 607 g CP/kg, using a protein source with high collagen concentration (Zentek, 1995; Hang et al., 2013)

Several studies investigated effects of diets with differently fermentable fibers or inulin combined with various protein levels on apparent nutrient digestibility, composition of intestinal microbiota, and microbial metabolites in dogs (Wambacq et al., 2016; Pinna et al., 2018). The major findings of the first study were that a soluble fiber source (sugar beet pulp and guar gum mix vs. cellulose) decreased total apparent protein digestibility by 4.1%, while nitrogen retention was not significantly affected. In the second study, inulin (15 g/kg) decreased fecal apparent protein digestibility by 6% with a diet containing 304 g CP/kg dry matter. Interestingly, inulin increased the protein digestibility with a low-protein diet (229 g/kg dry matter). Soluble fibers have the potential to increase intestinal production of SCFA and lactate and bifidobacteria when added to canine diets. Investigations on BSG and CAP combined with different protein levels on digestive parameters and microbial composition in dogs have not been performed to our best knowledge. Our hypothesis was that non-fermentable and fermentable fiber ingredients will have a different effect on intestinal protein fermentation in diets with low or high protein concentration in dogs. The objective of the study was to compare the effects of BSG and CAP combined with a high and a low dietary protein level on apparent fecal nutrient digestibility, feces quality, fecal microbial metabolites, microbial composition, and metabolites in urine and blood.

MATERIALS AND METHODS

The study was approved by Landesamt für Gesundheit und Soziales Berlin (0228/17).

Animals, Diets, and Sampling

Ten healthy adult Beagle dogs (6 intact females and 4 chemically neutered males) with a mean age of 4.36 (±0.30) years were housed in indoor pens with daily access to outdoor pens. Five dry diets were formulated as a complete food for adult dogs with BSG (Tremonis GmbH, Dortmund, Germany) or CAP (Ernteband Fruchtsaft GmbH, Winnenden, Germany) as fiber sources. Four diets were formulated to contain 7.5% of TDF. Poultry and greaves meal were used as protein sources and diets contained either 20% or 40% of CP. The diets are further designated as BLP (7.5% TDF BSG, 20% CP), BHP (7.5% TDF BSG, 40% CP), CLP (7.5% TDF CAP, 20% CP), and CHP (7.5% TDF CAP, 40% CP). One diet was formulated with an equal mixture of BSG and CAP, but at lower inclusion level and considered as a reference diet (BCLP; 3.5% TDF from BSG and CAP, 20% CP). Titanium dioxide was added when diets were produced and homogenously mixed with all other ingredients. Detailed information about the diets can be found in Tables 1 and 2. The 5 experimental diets were fed in a randomized cross-over design to 5 groups of 2 beagles each to exclude sequential and time effects. Experimental diets were offered twice daily and food allowances were calculated to maintain body weight according to the recommendations of NRC (2006) for adult dogs. All dogs had free access to fresh water. The duration of each feeding period was 19 d, and samples were collected after an adaptation period of 14 d. Fecal samples were individually pooled over 5 d for each dog and apparent nutrient digestibility as well as fecal dry matter were analyzed. A fresh fecal sample (collected directly after defecation) was taken for bacterial metabolite and microbiome analysis. To avoid usage of metabolic cages, dogs were trained for urinary spot sampling. Urine was collected outside in the morning and in the afternoon. Fasting blood samples were taken once on day 15 of every feeding period. This should avoid postprandial fluctuations as confounding factor, which is particularly significant for urea.

Table 1.

Ingredients of experimental diets

Ingredient, %	Diet1
	BLP	BHP	CLP	CHP	BCLP
Rice flour	61.1	40.0	61.7	40.6	69.1
Poultry meal, low ash	9.50	24.3	8.70	23.9	12.2
Greaves meal	3.20	13.3	6.70	16.8	3.70
Brewer’s spent grain	14.8	15.1	–	–	1.70
Carrot pomace	–	–	10.8	11.0	1.60
Rapeseed oil	6.70	4.40	7.30	4.90	6.90
Bone meal	1.70	0.10	1.90	0.20	1.70
Mineral and vitamin premix2	1.60	1.70	1.60	1.70	1.70
Potassium hydrogen carbonate	1.30	0.90	1.00	0.60	1.20
Titanium dioxide	0.20	0.20	0.20	0.20	0.20

¹BLP = 7.5% TDF BSG, 20% CP; BHP = 7.5% TDF BSG, 40% CP; CLP = 7.5% TDF CAP, 20% CP; CHP = 7.5% TDF CAP, 40% CP; BCLP = 3.5% TDF, 20% CP.

²Per kilogram premix: 600,000 IU vitamin A; 120,000 IU vitamin D3; 8,000 mg vitamin E; 300 mg vitamin K3; 250 mg vitamin B1; 250 mg vitamin B2; 400 mg vitamin B6; 5,000 mg iron [iron-(II)-carbonate]; 1,000 mg copper [copper-(II)-sulfate, pentahydrate]; 5,000 mg zinc (zinc oxide); 2,000 µg vitamin B12; 2,500 mg niacin; 100 mg folic acid; 25,000 µg biotin; 1,000 mg pantothenic acid; 80,000 mg choline chloride; 6,000 mg manganese (manganese oxide); 45 mg iodide (calcium iodide); 35 mg selenium (sodium selenite).

Table 2.

Analyzed chemical composition of experimental diets

Nutrients, g/kg DM	Diet1
	BLP	BHP	CLP	CHP	BCLP
Crude ash	58.7	60.2	56.0	57.9	55.9
Crude fat	103	109	101	105	98.6
Crude protein	213	409	216	399	219
Crude fiber	25.5	25.3	29.7	31.6	9.22
Total dietary fiber	81.3	81.9	75.1	80.2	34.4
Insoluble dietary fiber	63.8	58.6	53.2	55.8	19.7
Soluble dietary fiber	17.4	23.3	21.9	24.4	14.7
Neutral detergent fiber (corrected by residual CP)	70.6	72.2	49.9	54.2	19.3
Acid detergent fiber	36.5	41.5	42.5	41.7	16.3
Calcium	10.1	9.83	9.99	9.84	9.75
Phosphorus	7.23	7.23	6.88	6.75	7.12
Sodium	2.62	3.59	3.22	4.55	3.21
Potassium	6.43	5.99	6.17	5.98	6.18
Magnesium	1.82	1.78	1.82	1.78	1.43

mg/kg DM
Iron	575	630	201	210	204
Manganese	114	127	108	112	113
Copper	19.1	19.1	17.3	23.9	23.3
Zinc	136	156	131	137	143

¹BLP = 7.5% TDF BSG, 20% CP; BHP = 7.5% TDF BSG, 40% CP; CLP = 7.5% TDF CAP, 20% CP; CHP = 7.5% TDF CAP, 40% CP; BCLP = 3.5% TDF, 20% CP.

Fecal Consistency

Fecal scoring was performed once a week according to the system of Middelbos et al. (2007): 1 = hard, dry pellets; 2 = hard-formed, dry stool; remains firm and soft; 3 = soft, firm, and moist stool; 4 = soft, unformed stool; 5 = watery; liquid. In this system, a score in the range of 2 to 3 is considered “ideal.”

Apparent Fecal Nutrient Digestibility

Apparent fecal digestibility of nutrients and minerals was determined using titanium dioxide (TiO₂) as indigestible marker. Nutrients (CP, crude fat, dry matter, crude fiber, crude ash), organic matter (OM), and minerals were analyzed in feed and feces according to official method VDLUFA (Naumann and Bassler, 1976). Total dietary fiber, insoluble dietary fiber, and soluble dietary fiber were analyzed according to the protocol K-TDFR-100A/K-TDFR-200A 08/16 Megazyme (Ireland). Titanium dioxide was determined photometrically at 410 nm by an Ultrospec 2100 pro photometer (Amersham Pharmacia Biotech Inc., Piscataway, NJ). The apparent fecal digestibility (AD) was calculated with a standard formula:

$${\displaystyle \begin{array}{ll}{\displaystyle \mathrm{A}\mathrm{D}\left(\mathrm{\%}\right)}& =100-[(\mathrm{\%}\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{o}\mathrm{r}\mathrm{i}\mathrm{n}\mathrm{f}\mathrm{e}\mathrm{e}\mathrm{d}/\mathrm{\%}\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{o}\mathrm{r}\mathrm{i}\mathrm{n}\mathrm{f}\mathrm{e}\mathrm{c}\mathrm{e}\mathrm{s})\\ & \times (\mathrm{\%}\mathrm{n}\mathrm{u}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{f}\mathrm{e}\mathrm{c}\mathrm{e}\mathrm{s}/\mathrm{\%}\mathrm{n}\mathrm{u}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{f}\mathrm{e}\mathrm{e}\mathrm{d})\times 100].\end{array}}$$

Determination of Microbial Metabolites in Fecal Samples

Fecal samples for the analysis of microbial metabolites were immediately stored after sampling at −20 °C until analysis. Short-chain fatty acids were analyzed using a gas chromatograph (Agilent Technologies, Santa Clara, CA) with a flame ionization detector on a polyethylene glycol column (AGILENT 19095N-123, HP-INNOWAX), hydrogen was used as carrier gas at a flow rate of 30 mL/min.

The fecal concentration of d- and l-lactate was determined by HPLC (AGILENT 1100, Agilent Technologies) with a pre-column (Phenomenex C18 4.0 mm × 2.0 mm, Phenomenex Inc., Torrance, CA) and an analytical column [Phenomenex Chirex 3126 (D)-penicillamine 150 × 4.6 mm, Phenomenex Inc.] and UV detection at 253 nm.

The analysis of ammonia was performed by using the Berthelot reaction in microtitration plates and extinction was measured at 620 nm in a Tecan microplate reader (Infinite M200 PRO, Tecan Group Ltd., Männedorf, Switzerland). Phenol and indole were analyzed by gas chromatography (Agilent Technologies, 6890 N) with a flame ionization detector as described before (Pieper et al., 2014). Biogenic amine concentrations in feces were quantified as described previously (Pieper et al., 2012) using ion-exchange chromatography on a Biochrom 20 Amino Acid Analyzer (Biochrom 20Plus, Onken Laborservice GmbH, Gründau, Germany) with postcolumn ninhydrin derivatization. For the measurement of the fecal pH value, a Seven-Multi pH meter with a Micro electrode (InLab, Mettler-Toledo GmbH, Gießen, Germany) was used.

Determination of Fecal Microbiota

For the analysis of fecal microbiota, fresh samples were stored at −80 °C immediately after sampling. DNA extraction was performed using the commercial Quiagen PowerSoil DNA kit (MO BIO Laboratories Inc., Carlsbad, CA) according to manufacturer instructions except for the amount of feces (200 mg were used instead of 250 mg). Thereafter, the extracted DNA samples were sequenced by LGC Genomics GmbH (Berlin, Deutschland) on a MiSeq device (San Diego, CA) (LGC Inc., 2015). Data were analyzed by “MG RAST” (http://metagenomics.anl.gov/), where genus specific 16S sequences were assigned.

Determination of Metabolites in Urinary and Blood Samples

On day 15 of every feeding period, blood samples were collected following an overnight fast and immediately distributed into vacutainers containing lithium heparin or into monovettes (Sarstedt AG & Co., Nümbrecht, Germany). Samples were centrifuged at 2,000 × g for 10 min at 4 °C, supernatants were collected, and were analyzed for urea and creatinine using an automatic method (Konelab 60 i Thermo Fisher Scientific, Thermo Electron GmbH, Dreieich, Germany) at the Clinic for Small Animals, Faculty of Veterinary Medicine, Freie Universität Berlin. Urea and creatinine were analyzed in urinary samples with high-performance liquid chromatography (AGILENT 1100, Agilent Technologies). The preparation and analysis followed the protocol described by Passlack et al. (2014). Determination of urinary indican was based on the method of Curzon and Walsh (1962). In brief, a test kit (QuantiChrom Indican Assay Kit, BioAssay Systems, Hayward, CA) was used and indican concentration was measured colorimetrically at 480 nm with a photometer (Ultrospec 3300 pro, Amersham Biosciences, Little Chalfont, UK). Urinary phenols (p-cresol, 4-ethylphenol) and indoles (3-methylindole, 7-methylindole, 2-methylindole, 2.3-methylindole) were quantified by gas chromatography (GC 6890 N, Agilent Technologies). In short, 5 mL of urine were adjusted to pH 5 with formic acid, diluted with 5-methylindole (internal standard), lyophilized, and mixed with 500 mL methanol before analysis. Urine osmolality was determined by freezing point depression (osmometer OM 806, Vogel Medizinische Technik & Elektronik, Gießen, Germany) and urinary pH was determined using the same pH electrode as described for fecal samples.

Statistical Analysis

Data analyses were performed using IBM SPSS statistics 25.0 (SPSS Inc., Chicago, IL). Data were tested for normal distribution with Shapiro–Wilk test and are presented as arithmetic mean with standard error of mean (SEM). Differences between dietary treatments were tested by 2-factorial analysis of variance (ANOVA), considering protein and fiber as independent factors. The Tukey test was used to identify differences between groups. The reference diet BCLP was individually compared with the other 4 diets based on a 2-sided t-test. Fecal scores were compared by logistic regression. Bacterial ecological indices, richness, and evenness were calculated and visualized using RStudio (RStudio Inc., Boston, MA). The level of significance was set to α = 0.05.

RESULTS

Body Weight and Nutrient Intake

The body weight of the dogs was constant during the whole experiment (Table 3). The palatability of the diets was high and feed refusal was not observed. The intakes of dry matter and nutrients were comparable between all dietary treatments and covered the recommended allowances for maintenance (NRC, 2006) (Table 3). Dogs were healthy during the whole study time, and no signs of diet-related disorders were noted.

Table 3.

Body weight and intake of dry matter and nutrients of dogs fed the diets BLP, BHP, CLP, CHP, and BCLP¹; n = 10

	Diet1					SEM
	BLP	BHP	CLP	CHP	BCLP
Body weight, kg	13.7	13.7	13.7	13.7	13.7	0.3
Intake/kg BW/d
Dry matter, g	17.4	17.2	17.7	18.1	17.1	0.3
Crude fat, g	1.79	1.87	1.79	1.90	1.68	0.03
Crude fiber, g	0.44	0.43	0.53	0.57	0.16	0.02
Crude protein, g	3.70	7.02	3.83	7.21	3.74	0.25
Calcium, g	0.175	0.169	0.177	0.178	0.166	0.175
Phosphorus, g	0.126	0.124	0.122	0.122	0.122	0.126
Sodium, g	0.045	0.062	0.057	0.082	0.055	0.045
Potassium, g	0.112	0.103	0.109	0.108	0.106	0.112
Magnesium, g	0.032	0.031	0.032	0.032	0.024	0.001
Iron, mg	9.98	10.82	3.57	3.79	3.48	0.49
Manganese, mg	1.98	2.18	1.92	2.02	1.93	0.03
Copper, mg	0.33	0.33	0.31	0.43	0.40	0.01
Zinc, mg	2.36	2.68	2.32	2.47	2.44	0.04

¹BLP = 7.5% TDF BSG, 20% CP; BHP = 7.5% TDF BSG, 40% CP; CLP = 7.5% TDF CAP, 20% CP; CHP = 7.5% TDF CAP, 40% CP; BCLP = 3.5% TDF, 20% CP.

Fecal Consistency, Fecal Dry Matter, and Fecal pH

No differences in fecal quality were observed in dogs fed the experimental diets with a median fecal score of 2.2 (P = 0.129; data not shown). The fecal dry matter was higher for dogs after feeding BSG (P < 0.001). Fecal pH was not affected by fiber sources or protein levels (Table 4).

Table 4.

Fecal dry matter, pH, and apparent fecal digestibilities of nutrients of dogs fed the diets BLP, BHP, CLP, CHP, and BCLP; n = 10

	Diet1					SEM	P-value
	BLP	BHP	CLP	CHP	BCLP		Protein	Fiber	Protein × fiber
Fecal dry matter, %	31.8b	33.0b	25.3a#	25.8a#	30.7	0.58	0.360	<0.001	0.657
Fecal pH	7.04	7.03	7.02	6.64	6.92	0.05	0.091	0.143	0.098
Apparent digestibility, %
Organic matter	82.7ab#	82.5a#	85.7c#	84.3bc#	90.3	0.45	0.028	<0.001	0.113
Crude protein	75.7a#	84.9c#	77.8b#	83.6c	81.8	0.60	<0.001	0.411	0.003
Crude fat	92.4a#	94.7ab	96.3b	96.4b	95.5	0.39	0.078	<0.001	0.100
Crude fiber	13.3	16.0#	13.9	16.9#	8.96	1.13	0.228	0.751	0.976
Total dietary fiber	−2.36a	2.01a	10.4a	25.0b#	−1.9	2.03	0.001	<0.001	0.056
Crude ash	32.7#	30.8#	34.7	34.6	37.1	0.93	0.646	0.191	0.667
Calcium	6.38	5.12	5.21	3.61	5.89	0.61	0.262	0.292	0.896
Phosphorus	28.5a	28.7a	34.6ab	36.6b	33.1	0.86	0.519	<0.001	0.588
Sodium	80.4ab#	86.1b#	75.8a#	86.2b	90.8	1.06	<0.001	0.193	0.185
Potassium	89.2#	91.1#	90.1#	91.2	93.7	0.45	0.142	0.622	0.648
Magnesium	20.7b	10.4c#	32.6a#	25.6ab	19.4	1.43	<0.001	<0.001	0.442
Iron	12.9	12.4	−3.04	11.6	11.1	2.45	0.205	0.135	0.175
Manganese	7.19	17.6	9.40	9.50	10.6	1.27	0.078	0.316	0.083
Copper	26.9ab#	19.9b#	23.8b#	40.6a	43.0	2.04	0.223	0.031	0.004
Zinc	4.73	1.07#	2.65#	−2.50#	10.9	1.21	0.082	0.258	0.764

¹BLP = 7.5% TDF BSG, 20% CP; BHP = 7.5% TDF BSG, 40% CP; CLP = 7.5% TDF CAP, 20% CP; CHP = 7.5% TDF CAP, 40% CP; BCLP = 3.5% TDF, 20% CP.

^a–cDifferent superscripts within a row indicate significant differences between the diets BLP, BHP, CLP, and CHP based on the 2-factorial ANOVA.

^#Significant differences of these diets to diet BCLP (P ≤ 0.05).

Apparent Fecal Nutrient Digestibility

The apparent fecal digestibility of OM was highest when dogs received diets with CAP (P < 0.001; Table 4), and particularly the reference diet BCLP with a lower fiber inclusion rate. The apparent fecal digestibility of CP was higher for the high-protein diets (P < 0.001), whereas fiber sources had no effect. The apparent fecal digestibility of crude fat (P < 0.001; Table 4) was affected by the fiber sources and was higher for the diets with CAP than with diets containing BSG. The apparent fecal digestibility of TDF was higher with inclusion of CAB than with BSG and with the high-protein diets compared to the low-protein diets (Table 4).

Mineral apparent fecal digestibilities are summarized in Table 4. Calcium was not affected, whereas the apparent fecal digestibility of phosphorus increased (P < 0.001) when dogs were fed diets containing CAP compared to BSG. The apparent fecal digestibility of sodium (P < 0.001) was affected by the protein level, high-protein diets resulting in higher sodium digestibility. Diets with CAP and low dietary protein level showed the highest apparent fecal digestibility for magnesium (P < 0.001). Dietary protein and fiber did not affect the apparent fecal digestibility of potassium, iron, manganese, and zinc. For copper, the apparent fecal digestibility increased when dogs were fed diet CHP compared to CLP, while no influence of protein level was found for the diets with BSG indicating an interaction of the main factors (P = 0.004).

Fecal Microbial Metabolites

The fecal concentrations of d- and l-lactate were reduced when the high-protein diets were fed (P = 0.023 resp. 0.042) (Table 5). The total fecal short-chain acid (P = 0.010) and acetate (P = 0.001) concentrations were higher with CAP as fiber source, whereas concentrations of i-butyrate (P = 0.001), i-valerate (P = 0.002), and ammonium (P < 0.001) increased with the high-protein diets. Fecal concentrations of total biogenic amines (P < 0.001), including putrescine (P = 0.002) and cadaverine (P < 0.001), increased with higher dietary protein concentrations (Table 5), whereas spermidine and spermine concentrations showed an interaction of protein and fiber (P = 0.035 and 0.045) (Table 5). Fecal phenols and indoles did not show significant changes depending on the diets. Diet BCLP was in all measured fecal traits comparable with the other diets with minor effects on the fecal metabolites.

Table 5.

Fecal concentration of microbial metabolites of dogs fed the diets BLP, BHP, CLP, CHP, and BCLP; n = 10

Item (μmol/g)	Diet1					SEM	P-value
	BLP	BHP	CLP	CHP	BCLP		Protein	Fiber	Protein × fiber
D-Lactate	1.26ab	0.23a	0.98b	0.59a	0.92	0.14	0.023	0.891	0.298
L-Lactate	1.60	0.48	1.16	0.73	1.29	0.18	0.042	0.793	0.358
SCFA2	73.9a	80.4ab	93.9ab	111b	91.1	4.44	0.215	0.010	0.576
Acetate	37.8a	43.6a	55.1ab	66.4b#	44.6	2.75	0.131	0.001	0.619
Propionate	18.7	20.5	21.8	23.9	18.6	1.10	0.460	0.207	0.956
i-Butyrate	1.72a	2.47b	1.50a	2.08ab	1.99	0.09	0.001	0.107	0.650
n-Butyrate	11.9	10.2#	12.3	15.0	20.1	1.28	0.834	0.269	0.348
i-Valerate	2.18ab	2.91b	1.70a#	2.63ab	2.58	0.12	0.002	0.139	0.684
n-Valerate	1.63	0.84#	1.55	0.99	3.34	0.30	0.002	0.139	0.684
BCFA2	4.32	4.71	4.11	3.96	4.67	0.21	0.805	0.329	0.569
Biogenic amines	1.64a	2.41b	1.63a	2.52b	1.94	0.10	<0.001	0.800	0.764
Putrescine	0.68ab	1.10ab#	0.64a	1.21b#	0.68	0.07	0.002	0.827	0.628
Histamine	0.01a#	0.02a	0.02ab	0.04b	0.03	0.00	0.024	0.011	0.171
Cadaverine	0.04a#	0.22a	0.16a	0.44b#	0.16	0.03	<0.001	0.005	0.390
Spermidine	0.55a	0.90b#	0.53a	0.58a	0.58	0.03	0.006	0.020	0.035
Spermine	0.35b	0.16a#	0.27ab#	0.23ab#	0.49	0.02	0.003	0.882	0.045
Ammonium	21.8a	39.5b	19.8a	30.8ab	27.7	1.68	<0.001	0.133	0.243
Phenol	9.03	36.4	10.3	14.6	34.6	2.37	0.080	0.250	0.193
Indole	630	827	717	579	702	5.10	0.782	0.448	0.122
p-Cresol	3.51	12.4	2.5	n.d.3	n.d.	0.20	–	–	–
4-Ethylphenol	38.0	35.7	28.3	38.2	40.8	0.34	0.523	0.543	0.303
7- Methylindole	196	83.1	61.5	96.1	226	3.45	0.443	0.234	0.154

¹BLP = 7.5% TDF BSG, 20% CP; BHP = 7.5% TDF BSG, 40% CP; CLP = 7.5% TDF CAP, 20% CP; CHP = 7.5% TDF CAP, 40% CP; BCLP = 3.5% TDF, 20% CP.

²BCFA = branched chain fatty acids; SCFA = short-chain fatty acids.

³n.d. = not detected.

^{a, b}Different superscripts within a row indicate significant differences between the diets BLP, BHP, CLP, and CHP based on the 2-factorial ANOVA.

^#Significant differences of these diets to diet BCLP (P ≤ 0.05).

Fecal Microbiota

The abundance of bacterial orders in the feces of dogs was similar between the dietary treatments. The relative abundance of Selenomonadales and Coriobacteriales increased when diets with BSG were fed (P = 0.006 and 0.024), the order Bifidobacteriales increased with the low-protein diets (P = 0.001). Dogs fed BCLP had a comparable relative abundance bacterial orders compared to the other treatments (Table 6). The relative abundances of microbes at the phylum and genus level is listed in Supplementary Data. Dietary effects on bacterial diversity were identified for protein and dietary fiber on richness (P = 0.006 and 0.018) and for protein on evenness (P = 0.041), but were overall moderate (Table 6).

Table 6.

Relative abundance of bacterial orders, richness, and evenness in the feces of dogs fed diets BLP, BHP, CLP, CHP, and BCLP; n = 10

Bacterial order,%	Diet1					SEM	P-value
	BLP	BHP	CLP	CHP	BCLP		Protein	Fiber	Protein × fiber
Clostridiales	32.8	35.8	33.5	44.4	31.6	2.21	0.164	0.345	0.431
Bacteroidales	26.8	25.4	28.2	25.6	21.5	1.65	0.580	0.823	0.877
Fusobacteriales	20.5	21.5	24.4	21.0	22.0	1.46	0.735	0.633	0.545
Selenomonadales	5.27	5.48	3.30	3.47	3.84	0.32	0.780	0.006	0.978
Bifidobacteriales	2.28	0.48	2.33	0.26#	3.43	0.34	0.001	0.875	0.814
Erysipelotrichales	0.11	0.13	0.30	0.04	2.99	0.55	0.271	0.648	0.204
Lactobacillales	0.38	1.93	1.53	0.42	1.35	0.40	0.823	0.850	0.169
Coriobacteriales	1.58	1.68	0.59	0.49	0.87	0.20	0.996	0.024	0.823
Aeromonadales	0.41	0.35	0.21	0.33	0.31	0.05	0.807	0.333	0.399
Enterobacteriales	n.d.2	n.d.	n.d.	n.d.	0.01	–	–	–	–
Methanobacteriales	0.03	0.03	0.27	0.05	0.06	0.03	0.089	0.056	0.112
Campylobacterales	0.01	0.01	0.04	0.04	0.01	0.01	0.994	0.168	0.897
Actinomycetales	0.01	0.01#	0.01	0.01#	0.03	0.00	0.142	0.631	0.433
Mycoplasmatales	0.11	0.01	n.d.	n.d.	n.d.	0.02	–	–	–
Bacillales	n.d.	n.d.	0.01	n.d.	n.d.	0.00	–	–	–
unclassified Betaproteobacteria	n.d.	0.05	n.d.	0.06	n.d.	0.02	–	–	–
Richness	55.4ab	63.9b#	51.3a	56.5ab	56.4	1.14	0.006	0.018	0.483
Evenness	0.66	0.68	0.66	0.68	0.69	0.005	0.041	0.782	0.468

¹BLP = 7.5% TDF BSG, 20% CP; BHP = 7.5% TDF BSG, 40% CP; CLP = 7.5% TDF CAP, 20% CP; CHP = 7.5% TDF CAP, 40% CP; BCLP = 3.5% TDF, 20% CP.

²n.d. = not detected.

^#Significant differences of these diets to diet BCLP (P ≤ 0.05).

Metabolites in Blood and Urine

Urinary osmolality (P < 0.001; Table 7) was higher for diets rich in protein, independent of the fiber source. Urinary concentration of urea was influenced by protein (P = 0.001) and fiber (P = 0.002) with a significant interaction between both factors (P = 0.029) (Table 7). The other urinary metabolites were not affected. Dogs had higher blood levels of urea (P < 0.001; Table 8) when high-protein diets were fed. There was a trend for creatinine to increase by dietary protein level (P = 0.055)

Table 7.

Urinary parameters of dogs fed BLP, BHP, CLP, CHP, and BCLP; n = 10

Item	Diet1					SEM	P-value
	BLP	BHP	CLP	CHP	BCLP		Protein	Fiber	Protein × fiber
pH	7.10	7.36#	7.11#	6.77#	7.64	0.12	0.877	0.275	0.271
Osmolality, mOsm/L	608ab	838bc#	408a	1018c#	375	54.2	<0.001	0.922	0.068
µmol/mL
Urea	0.38ab	0.59ab#	0.25a	0.76b#	0.23	2.92	0.001	0.002	0.029
Creatinine	4.66	6.51#	3.84	8.55#	3.24	56.4	0.826	0.548	0.976
Indican	0.05.	0.06	0.03	0.08	0.04	1.67	0.142	0.162	0.068

nmol/mL
Phenol	32.7	33.5	25.4	37.6#	20.9	2.43	0.264	0.786	0.326
Indole	15.4	12.5	9.41	14.7	11.6	1.37	0.706	0.545	0.189
4-Ethylphenol	5.24	4.47	4.19	4.39	5.47	0.34	0.730	0.489	0.557
3-Methylindole	15.9	8.65	6.74	7.96	5.24	1.39	0.369	0.146	0.210
7-Methylindole	9.50	8.46	5.10	5.54	8.45	0.98	0.886	0.087	0.725

¹BLP = 7.5% TDF BSG, 20% CP; BHP = 7.5% TDF BSG, 40% CP; CLP = 7.5% TDF CAP, 20% CP; CHP = 7.5 % TDF CAP, 40% CP; BCLP = 3.5% TDF, 20% CP.

^a–cDifferent superscripts within a row indicate significant differences between the diets BLP, BHP, CLP, and CHP based on the 2-factorial ANOVA.

^#Significant differences of these diets to diet BCLP (P ≤ 0.05).

Table 8.

Parameters in the blood of dogs fed BLP, BHP, CLP, CHP, and BCLP; n = 10

Item	Diet1					SEM	P-value
	BLP	BHP	CLP	CHP	BCLP		Protein	Fiber	Protein × fiber
Urea, mmol/L	4.62ab	6.96ab#	4.35a	7.23b#	4.39	0.24	<0.001	1.000	0.431
Creatinine, µmol/L	57.0	60.2	56.2	67.3	58.5	1.85	0.055	0.388	0.281

¹BLP = 7.5% TDF BSG, 20% CP; BHP = 7.5% TDF BSG, 40% CP; CLP = 7.5% TDF CAP, 20% CP; CHP = 7.5% TDF CAP, 40% CP; BCLP = 3.5% TDF, 20% CP.

^{a, b}Different superscripts within a row indicate significant differences between the diets BLP, BHP, CLP, and CHP based on the 2-factorial ANOVA.

#Significant differences of these diets to diet BCLP (P ≤ 0.05).

DISCUSSION

The present study provides data of digestive parameters and fecal microbial composition of dogs fed with diets with BSG and CAP combined with different protein concentrations. Common fiber sources used in pet food are cellulose and beet pulp (de Godoy et al., 2013), both differing in solubility and affecting fecal quality. Dietary protein is another important factor for optimal fecal characteristics (Hang et al., 2013). In the present study, fecal quality was not visually different between the different dietary treatments, as indicated by fecal scores, but diets containing BSG increased fecal dry matter compared to CAP, probably owing to a higher content of non-fermentable fiber. Higher water-binding capacity of fermentable fibers (Bednar et al., 2000) and pectins (Stephen and Cummings, 1979) may result in loose feces in dogs.

Fiber digestion was not reflected in the fecal crude fiber digestibility, indicating that crude fiber is not a suitable measurement for describing fiber effects in dogs. Total dietary fiber fecal digestibility was significantly lower when BSG was included as ingredient compared to CAP. This can be explained by the different relation of insoluble and soluble fiber in BSG and CAP and is in agreement with a previous study with lignocellulose and sugar beet pulp (Kröger et al., 2017).

The lowest apparent fecal digestibilities of CP were identified for diets containing 20% of CP irrespective of the dietary fiber source. Apparent fecal protein digestibility is influenced by endogenous nitrogen secretion, which amounts to 54 to 73 mg/kg^0.75/d (data reported by Hendriks et al. 2002). Endogenous nitrogen secretion has a negative effect on the apparent fecal protein digestibility, which is more pronounced, when dietary protein intake is low. This might explain the protein effects observed in this study. Another factor affecting fecal apparent protein digestibility is dietary fiber. The diet BCLP with a lower fiber content had a higher fecal apparent CP digestibility compared to the other experimental diets. A previous study revealed that added fiber progressively decreases apparent fecal digestibilities of dietary protein and fat (Burrows et al., 1982). In line, apparent fecal digestibility of OM was highest after dogs received diet BCLP. This finding might be related to the fact, that apparent fecal digestibility of OM is influenced by the amount of dietary protein and fiber as well as fermentability of the fiber source (Zentek, 1996; Bosch et al., 2009).

Apparent fecal digestibility of crude fat was lower for diets including BSG, contrary to results in previous studies. In other studies, the use of the non-fermentable fiber source cellulose did not affect crude fat digestibility (Kröger et al., 2017) or increased digestibility of crude fat in dogs (Bosch et al., 2009).

In the present study, experimental diets containing BSG exhibited a lower apparent fecal digestibility of phosphorus compared to diets with CAP. To our best knowledge, there are no data in the impact of different fiber sources on phosphorus digestibility in dogs. The explanation for lower digestibility might be related to phytic acid in BSG or a specific effect on intestinal phosphorus absorption or secretion. The apparent fecal digestibility of sodium was higher with the high-protein diets. This might be related not only to the protein concentration but also to other dietary factors, as the low-protein diets contained more carbohydrates which might affect water-binding capacity of the digesta matrix. The apparent fecal digestibility of magnesium was positively influenced by lower dietary protein and CAP as fiber source. The explanation could be a more intensive intestinal fermentation and formation of organic acids, which might have beneficial effects on magnesium solubility in the digesta.

Apparent fecal digestibilities of sodium, potassium, copper, and zinc appeared partially positively affected by diet BCLP, with its lower dietary fiber concentration and the combination of BSG and CAP. These findings are in line with a report that a low dietary protein concentration combined with fermentable fructooligosaccharides increased apparent mineral digestibilities in dogs (Pinna et al., 2018). Interpreting apparent fecal digestibilities of minerals, mineral intake due to water intake needs to be taken into account in general. In this study, however, a potential “water effect” could be avoided by a randomized sequence of the dietary periods for 5 groups of 2 dogs each.

Fecal concentrations of SCFA were higher when CAP was included in diets; however, fecal concentrations maybe misleading as they represent a snapshot and do not reflect production rates. Fermentable fibers are known to stimulate production of SCFAs (Sunvold et al., 1994). Short-chain fatty acids are associated with beneficial effects on colonocytes, host health (Topping and Clifton, 2001), and intestinal motility (Suchodolski, 2011). Fecal concentrations of i-butyrate and i-valerate are considered as markers of branched chain amino acid catabolism and were highest when dogs were fed the high-protein diets, reflecting colonal microbial degradation (Macfarlane and Macfarlane, 2003). Fermentable fiber sources can reduce BCFA concentration in canines (Kröger et al., 2017). In this study, the higher dietary protein level increased the fecal i-butyrate and i- valerate concentration, while both fiber sources had no effect on the branched chain fatty acids. In a previous study, sugar beet pulp, guar gum mix, and cellulose (9.0% to 10.1% TDF DM) did not affect fecal i-butyrate and i-valerate concentration in canine feces, when dogs were fed diets with low (<17% CP) protein levels (Wambacq et al., 2016).

Biogenic amines are related to decarboxylation of proteins in the intestine (Blachier et al., 2007). The fecal concentrations of biogenic amines, putrescine, histamine, cadaverine, and spermidine were higher when dogs were fed diets with 40% CP compared to diets containing 20% CP. Former studies in dogs (Barry et al., 2009; Faber et al., 2011) reported that a moderate variation of dietary protein (30.5% to 32.5% DM) in experimental diets did not affect biogenic amine concentration. There was no main effect of dietary fiber on fecal biogenic amine concentrations. Diet BCLP with its lower dietary fiber concentration showed similar magnitudes of biogenic amines as the other diets with some minor differences in individual amines, allowing the conclusion, that fiber might be of minor importance.

Fecal ammonium, another marker of intestinal microbial protein catabolism, was increased with the high-protein diets, especially when BSG was used as fiber source. This is in line with former studies where higher levels of protein in experimental diets lead to higher fecal ammonia concentrations in dogs in vivo (Nery et al., 2012; Hang et al., 2013) and in vitro (Pinna et al., 2016). In this study, dietary fiber sources had no significant impact on fecal ammonium concentration. In a previous study, a higher ileal concentration of ammonia was observed in dogs fed cellulose compared to pectin (Silvio et al., 2000). This indicates that fiber might affect ammonia concentration not only depending on its own fermentability but probably also due to the impact on digestion processes.

The present study also aimed in determining effects of experimental diets on microbial composition. The adaption period is critical, based on previous studies 14 d were considered to be appropriate for the intestinal microbiota to adapt to new diets (Barry et al., 2009; Beloshapka et al., 2013; Sandri et al., 2017). The relative abundance of Bifidobacteriales was highest after dogs were fed the low-protein diets. This order belongs to predominant bacteria in the canine intestine, as documented previously (Suchodolski et al., 2008; Suchodolski, 2011; Beloshapka et al., 2013). Our findings are in accordance with a previous study where Bifidobacteriales were increased in fecal samples of dogs being fed a commercial diet (30.5% CP DM) compared to a high-protein bones and raw food diet (44.4% CP DM) (Schmidt et al., 2018). It can be concluded, that a lower dietary protein level favors the relative abundance of Bifidobacteriales. A higher abundance of bifidobacteria was also observed in a study after dogs consumed a diet low in protein compared to a protein-rich diet, whereby protein source and level were the main factors (Zentek et al., 2003). Additionally to dietary protein, fermentable fiber can influence intestinal microbiota by increasing fecal bifidobacteria in dogs (Middelbos et al., 2007); however this effect was not obvious in this study. The findings suggest that further research is needed to characterize protein and fiber effects on the fecal microbiota in dogs.

Concentrations of urinary urea were increased with the diets high in protein compared to low-protein diets. Interestingly, urinary levels of urea were more affected by high dietary protein diets in combination with CAP compared to BSG. In a previous study in dogs, a fermentable fiber (apple pectin) was described to increase urinary urea and creatinine levels (Zentek, 1996). On the contrary, the inclusion of dried, grounded carrots in a canine diet leads to lower urea concentrations in the urine in dogs (Zentek and Meyer, 1993). It needs to be taken into account, that dietary protein levels in experimental diets of that study varied between 18% and 22% DM, which is lower than the high-protein diets used in present study. In general, urinary creatinine excretion is considered to be constant in dogs (Braun et al., 2003), although our data indicate a numerical effect by dietary protein level. The urinary concentration of indican was not affected by the experimental treatments. In a previous study, an experimental diet with dried, ground carrots decreased renal indican excretion compared with a control diet, while protein levels of both diets were lower (Zentek and Meyer, 1993). Another study found that mannanoligosaccharides, transgalactooligosaccharides, lactose, and lactulose (1 g/kg BW/d) had no influence on renal indican concentrations in dogs (Zentek et al., 2002). Urinary osmolality was highest when dogs were fed the high-protein diets and highly correlated with urea concentration (r = 0.805; P < 0.001). A previous study could not detect any dietary influence on urinary osmolality in dogs of different breeds (van Vonderen et al., 1997). However, it is not clear which diets or dietary protein level were consumed. The authors considered that reference values for urinary osmolality seem not to be appropriate due to wide individual differences. In the present study, dogs had higher blood levels of urea when the high-protein diets were fed, compared to diets low in protein, in accordance with a previous study in dogs (Davenport et al., 1994).

In conclusion, BSG and CAP can be used as fiber sources in canine diets at higher inclusion rates. Their inclusion had some notable effects on digestive parameters, fecal and urinary metabolites, while effects on the intestinal microbiota were moderate. High-protein diets increased protein metabolites in feces, urine, and blood; however, effects of the selected fiber sources on microbial protein fermentation metabolites seem to be limited.

SUPPLEMENTARY DATA

Supplementary data are available at Journal of Animal Science online.

Figure S1. Distribution of fecal microbiota (genus level) of dogs after being fed experimental diets.

Table S1. Relative abundance of bacterial phyla in the feces of dogs fed diets BLP, BHP, CLP, CHP, and BCLP; n = 10.

Table S2. Relative abundance of bacterial genera in the feces of dogs fed diets BLP, BHP, CLP, CHP, and BCLP; n = 10.