In vitro fermentation characteristics of dietary fibers using fecal inoculum from dogs consuming commercial or grain kefir

Abstract

Traditional grain kefir is produced from the fermentation of milk with yeast- and bacteria-containing cultures. To maintain consistency and adhere to food safety guidelines, commercial kefir products are based on starter bacterial cultures. Bacterial profiles of starter vs. grain kefirs differ, and their influence on health effects is unknown. Our objectives were to determine the in vitro fermentation characteristics of common dietary fibers using fecal inoculum from dogs supplemented with kefir or kefir bacterial culture as inoculum. Healthy adult dogs were allotted to one of 3 treatments and supplemented for 14 d (n = 4/treatment): 1) 2% reduced-fat milk treated with lactase (CNTL), 2) starter kefir (S-Kefir), or 3) grain kefir (G-Kefir). After 14 d, fresh fecal samples were collected and frozen in a 20% glycerol solution. For the in vitro experiment, fecal samples were thawed, diluted in an anaerobic diluting solution, and used to inoculate tubes containing semi-defined medium and either cellulose (CEL), pectin (PC), beet pulp (BP), or chicory pulp (CP). Tubes were incubated for 0, 6, 12, or 18 h, with short-chain fatty acids (SCFA), pH, and microbiota measured at each time point. A second in vitro experiment was conducted using similar methods and measurements but with S-Kefir and G-Kefir as inoculum sources. Effects of treatment (inoculum), time, and treatment*time interactions within the fiber source were analyzed statistically using Mixed Models and repeated measures, with P < 0.05 being significant. Using fecal inoculum, BP and PC were rapidly fermented, leading to large pH reductions, SCFA increases, and microbiota shifts. pH change was of greater (P < 0.05) magnitude (PC) and higher (P < 0.05) kinetic rate (CP) when using feces from dogs fed S-Kefir or G-Kefir than controls. Butyrate increases were greater (P < 0.05) in tubes inoculated with G-Kefir feces than in S-Kefir or control feces. When PC and BP were fermented, tubes with S-Kefir feces had greater (P < 0.05) acetate, propionate, and total SCFA increases than G-Kefir or control feces. Fermentations were slower when using kefir cultures as inoculum, but some differences were noted. Bacterial beta diversity and relative abundances shifted over time within each substrate and were unique to the inoculum source. Our data suggest that the activity of kefir bacterial populations differs and that kefir consumption changes the abundance and activity of the fecal microbiota of dogs, justifying in vivo investigation.

Healthy adult dogs were supplemented with milk, starter kefir, or grain kefir for 14 d, with fresh fecal samples being collected and used for an in vitro fermentation experiment. The results of the in vitro experiment demonstrated that kefir consumption changed the abundance and activity of the fecal microbiota of dogs.

Introduction

The market size for fermented dairy beverages, including kefir products, is expected to increase by $456 million from 2021 to 2025, with a compound annual growth rate of 4.37% in 2021 (González‐Orozco et al., 2023). Given their success in the human market, the companion animal supplement market has recently included fermented food products such as kefir. Traditionally, kefir is generated from the fermentation of milk with kefir grains (Rosa et al., 2017). A range of purported benefits has been attributed to grain-fermented kefir, which includes the presence of kefiran, the exopolysaccharide produced by lactic acid bacteria (e.g., Lactobacillus kefiranofaciens) in kefir (Georgalaki et al., 2021). Kefiran acts as a biofilm that may serve as a protective barrier for probiotic strains being delivered to the host gastrointestinal tract as well as antibacterial properties that target the cytoplasmic membrane of sensitive pathogens and other bacteria (Simova et al., 2002; Guzel-Seydim et al., 2011).

While grain-fermented kefirs may provide health benefits, it is unknown if similar benefits may hold for commercialized starter kefir products. Starter kefir (S-Kefir) production is quite different than grain kefir (G-Kefir) due to the exclusion of kefir grains. Rather, its production involves the inoculation of dairy milk with defined microbial cultures commonly derived from yogurt suppliers (Nejati et al., 2022). The rational use of defined microbial starters to ferment milk instead of kefir grains is due to concerns of product consistency, microbial consistency, presence or absence of yeast, and consumer palatability (Hansen, 2002). Due to stark differences in processing and omission of grain fermentation, S-Kefir is more similar to drinkable yogurt than to G-Kefir. Not surprisingly, these differences in inoculum lead to commercial products having different food matrix properties and microbial profiles (Nejati et al., 2022; Metras et al., 2023).

The microbial profiles of G-Kefir have been shown to consist of bacterial species such as L. kefiranofaciens, Lentilactobacillus kefiri, Lentilactobacillus parakefiri and fungal species such as Kluyveromyces lactis, Kluyveromyces marxinus, Saccharomyces cerevisiae, Saccharomyces boulardii, and Saccharomyces unisporus (Gut et al., 2019; Slattery et al., 2019). 16S rRNA sequencing has been used to characterize kefir grains, demonstrating that high variability in microbial profile exists across products (Sindi et al., 2020). While the taxa listed above are often predominant, many other bacterial species have been detected (e.g., Lactiplantibacillus plantarum, Lactobacillus delbrueckii, Lacticaseibacillus casei, Limosilactobacillus reuteri, Lactobacillus buchneri groups, Leuconostoc mesenteroides, Lactococcus lactis, Gluconobacter frateurii, Acetobacter orientalis, Acetobacter lovaniensis) (Korsak et al., 2015; Wang et al., 2020; Georgalaki et al., 2021; Youn et al., 2022).

Little research has been conducted on companion animal kefir products and how they may impact the health of dogs. In one study, our laboratory characterized 6 S-Kefir products, demonstrating that the predominant microbial species included Bacillus coagulans, Lacticaseibacillus paracasei, Lactobacillus acidophilus, L. delbrueckii, Streptococcus thermophilus, Streptococcus salivarius, and L. lactis and that label inaccuracies regarding microbial abundance and identity were frequent (Metras et al., 2020). Other researchers reported that traditionally made G-Kefir brewed daily with 200 mL given to dogs once per day for 14 d modulated the gut microbiota, which included an increase in lactic acid bacteria (Kim et al., 2019). In another study, however, dogs supplemented with a daily dose of >10⁹ active fluorescent units of L. kefiri for 30 d were reported to have no changes to fecal IgA concentrations or microbiota populations (Gaspardo et al., 2020). Most recently, our laboratory investigated the effects of supplementing a G-Kefir or S-Kefir vs. a lactase-treated milk control in healthy adult dogs (Metras et al., 2023). In that study, no significant physiological changes were detected, but fecal Lactococcus relative abundance was greater (P < 0.05) in dogs supplemented with G-Kefir.

Currently, the field relies on the use of fecal microbiomics to assess the effects of gut-modifying substances such as dietary fibers, prebiotics, probiotics, and fermented foods. Because many of the fermentation end-products [e.g., short-chain fatty acids (SCFA)] are rapidly absorbed by the host, fecal samples are unable to accurately assess microbial activity. In vitro fermentation assays allow for the measurement of microbial abundances and metabolite concentrations over time, thus are a useful way to assess the activity of gut microbiota (Payne et al., 2012; de Godoy et al., 2015; Traughber et al., 2020). Therefore, the objectives of the current study were to determine the in vitro fermentation characteristics of common dietary fibers using 1) fecal inoculum from dogs supplemented with S-Kefir or G-Kefir; and 2) a S-Kefir bacterial culture or G-Kefir bacterial culture as inoculum. We hypothesized that differences in fecal microbial populations would impact dietary fiber fermentation, as evidenced by greater decreases in pH and greater production of SCFA in tubes inoculated by feces of dogs fed kefir. We also hypothesized that fiber fermentation by kefir beverages would occur at a slower rate than fecal sample fermentations, with changes in pH and SCFA production differing between tubes inoculated with S-Kefir and G-Kefir.

Materials and Methods

Study #1: in vitro experiment using fecal inoculum

All procedures were approved by the University of Illinois Institutional Animal Care and Use Committee prior to experimentation. All methods were performed in accordance with the United States Public Health Service Policy on Humane Care and Use of Laboratory Animals.

Substrates and analysis

Test fibers included beet pulp (BP; Archer Daniels Midland Co., Decatur, IL), chicory pulp (CP; BioMatrix International, Princeton, MN), pectin (PC; Modernist Pantry LLC; Eliot, ME), and cellulose (CEL; Solka Floc International Fiber Corporation; North Tonawanda, NY). Samples were analyzed for dry matter and ash according to AOAC (2006; methods 934.01 and 942.05), with organic matter calculated.

Animals, experimental design, and fecal sample collection

Twelve healthy female adult beagle dogs (age: 5.67 ± 1.72 yr; body weight: 7.27 ± 1.15 kg) were used in a completely randomized design. All dogs were housed individually in pens (1.0 m wide by 1.8 m long) in an environmentally controlled facility at the University of Illinois Urbana-Champaign. Dogs had free access to fresh water at all times. All dogs were fed a commercial diet that was formulated to meet all essential nutrients recommended by the Association of American Feed Control Officials (AAFCO, 2022) for adult dogs (Best Dog 21/12, Coker Feed Mill, Inc., Goldsboro, NC). Dogs were then randomly assigned to one of the following treatments and fed for 2 wk (n = 4/treatment): 2% milk treated with lactase (CNTL; 2% Reduced-Fat Milk, Great Value Farm, Fort Wayne, IN; 60 mL/d); S-Kefir (Champions Choice Kefir; Champions Choice, Millersburg, IN, 60 mL/d); and G-Kefir (Kefir Garden Grains, Hamilton Ontario, Canada; 60 mL/d). S-Kefir was used prior to its stated 3-mo expiration date and G-Kefir was brewed fresh daily for use throughout the 14-d intervention. On the basis of the maintenance energy requirement for adult dogs and information from previous feeding records, an amount of food to maintain body weight was offered and intake was measured twice daily (8:00 a.m.; 4:00 p.m.).

After 14 d of treatment, fresh fecal samples were collected within 15 min of defecation. For each dog, a 10 g aliquot of feces was collected into a 50 mL conical tube and mixed with 10 mL of 20% glycerol solution according to Cammarota et al. (2017). Samples were immediately placed on dry ice and then stored at −80 °C until the in vitro fermentation assay.

In vitro fermentation procedures

On the day of the in vitro experiment, fecal samples were carefully thawed and heated to 39 °C using a water bath, pooled by treatment, and then diluted 1:4 (wt/vol) in anaerobic diluting solution and blended for 15 s in a Waring blender (Waring Products, New Hartford, CT). Blended, diluted feces were filtered through 4 layers of cheesecloth and sealed in 125 mL serum bottles under a stream of CO₂ to minimize exposure to oxygen. Sample and blank tubes were then aseptically inoculated with diluted feces and added to the medium (Table S1) as described by Bourquin et al. (1993). Four mL of diluted feces was used to inoculate tubes containing 26 mL of semi-defined medium and one of the following fiber sources (300 mg/tube); CEL as the negative control, PC as the positive control, BP, or CP. Triplicate tubes of each fibrous substrate were incubated at 39 °C for 0, 6, 12, or 18 h, with periodic mixing. At each time point, the incubation was stopped, and samples were processed immediately. For each time point, the pH of tube contents was measured using a pH meter. Samples to be analyzed for SCFA (2 mL) were mixed with 0.5 mL of 25% metaphosphoric acid and detected according to Erwin et al. (1961) using a Hewlett-Packard (Avondale, PA) Model 5890A gas chromatograph equipped with a flame ionization detector on a column (1.8 m × 4 mm i.d.) packed with GP 10% SP-1200/1% H₃PO₄ on 80/100 Chromosorb W-AW (Supelco, Bellefonte, PA). The carrier gas was nitrogen, with a flow rate of 75 mL/min. The oven, injection port, and detector port temperatures were 125, 175, and 180 °C, respectively. Aliquots for microbial analyses were collected into sterile cryogenic vials and placed on dry ice until being transferred to a −80 °C freezer where they were stored until analysis. Data were corrected by blank tube (inocula and media, but no fiber source) and baseline (0 h) sample production. Negative values were expressed as 0.0 and subsequent acetate, propionate, and butyrate production values from each set of triplicate tubes were averaged.

Microbial DNA extraction

Genomic DNA was extracted from fermentation media using the Mo-Bio PowerSoil Kits (MO-BIO Laboratories, Inc., Carlsbad, CA) according to the manufacturer’s specifications. The quality of DNA was assessed via gel electrophoresis (E-Gel EX Gel 1%; Invitrogen, Carlsbad, CA) and the concentration of extracted DNA was quantified using a Qubit 3.0 Fluorometer (Life Technologies, Grand Island, NY) with a Qubit dsDNA BR assay kit (Invitrogen, Carlsbad, CA). 16S rRNA gene amplicons were generated using a Fluidigm Access Array (Fluidigm Corporation, South San Francisco, CA) in combination with Roche High Fidelity Fast Start Kit (Roche, Indianapolis, IN). The primers 515F (5′-GTGCCAGCMGCCGCGGTAA-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) that target a 252 bp-fragment of the V4 region of the 16S rRNA gene were used for amplification (primers synthesized by IDT Corp., Coralville, IA) (Caporaso et al., 2012). The CS1 forward tags and CS2 reverse tags were added according to the Fluidigm protocol. The quality of the amplicons was assessed using a Fragment Analyzer (Advanced Analytics, Ames, IA) to confirm amplicon regions and sizes. A DNA pool was generated by combining equimolar amounts of the amplicons from each sample. The pooled samples were then size selected on a 1-2% agarose E-Gel (Life Technologies, Grand Island, NY) and extracted using a Qiagen gel purification kit (Qiagen, Valencia, CA). Cleaned size-selected pooled products were run on an Agilent Bioanalyzer to confirm the appropriate profile and average size. Illumina sequencing was performed on a MiSeq using v3 reagents (Illumina Inc., San Diego, CA) at the Roy J. Carver Biotechnology Center at the University of Illinois.

Microbial data analysis

Forward reads were trimmed using the FASTX-Toolkit (version 0.0.13) and QIIME 2 2023.5 (Bolyen et al., 2019) was used to process the resulting sequence data. Briefly, high-quality (quality value ≥ 20) sequence data derived from the sequencing process was demultiplexed. Data were then denoised and assembled into amplicon sequence variants using DADA2 (Callahan et al., 2016). The SILVA 138.2 database (Quast et al., 2013) was used to assign taxonomy. An even sampling depth (sequences per sample) was used for assessing alpha and beta diversity measures. Beta diversity was assessed using weighted and unweighted UniFrac distance (Lozupone and Knight, 2005) measures and presented using principal coordinates analysis (PCoA) plots.

Statistical analysis

Blank-corrected data were analyzed using the Mixed Models procedure in SAS (SAS Institute Inc., version 9.4, Cary, NC), including a 2-way ANOVA and repeated measures. Data normality was analyzed using PROC UNIVARIATE. Differences among substrates, time, and substrate*time interactions were determined using a Fisher-protected least significant difference test with a Tukey adjustment to control for type-1 experiment-wise error. Analyses for comparing microbiota groupings were conducted using R version 4.4.1 (R Core Team, 2024). The following R packages were used: vegan (Oksanen et al., 2024) for PERMANOVA analysis, tidyr (Wickham et al., 2024) for data tidying, devtools (Wickham et al., 2023) for package management, and pairwiseAdonis (Martinez Arbizu, 2020) for pairwise comparisons. Statistical significance was set at P < 0.05.

Study #2: in vitro experiment using S-Kefir or G-Kefir bacterial cultures as inoculum

Substrates, kefir preparation, and analysis

The same test fibers from Study #1 (BP, CP, PC, CEL) were used in this study. The S-Kefir was purchased from a local vendor (Champions Choice Kefir; Champions Choice, Millersburg, IN, 60 mL/d), shipped in liquid form, and refrigerated (4 °C) immediately upon arrival. The S-Kefir was refrigerated for 1 wk prior to use in the in vitro fermentation, which was well before its 3-mo expiration period. The kefir brewing process to prepare the G-Kefir began with the addition of 5% w/v kefir grains to 2% reduced-fat dairy milk (Great Value Farm, Fort Wayne, IN). Kefir grains and kefir mixtures were incubated at 25 °C in a sterilized lightly sealed glass jar for 24 h. The pH of the kefir was measured after 24 h of fermentation using a pH meter equipped with a pH and temperature probe (Orion Triode 3-in-1 pH/Automatic Temperature Compensation Probe, ThermoFisher Scientific, Waltham, MA), with an ideal pH range maintained between pH 4 and 5. After 24 h, grains were removed using a stainless-steel strainer, with the kefir refrigerated at 4 °C until it was used for the in vitro assay the next day.

Kefir bacterial quantification

To begin, kefir products were inverted 10 times for homogenization, with 1 mL of product added to 9 mL of phosphate-buffered saline, and the product-phosphate-buffered saline mixture inverted 10 times again. Serial 10-fold dilutions were prepared with phosphate-buffered saline. A 50-μL volume of 10⁵ and 10⁷ dilutions were enumerated in duplicate via spiral plater (Eddy Jet Spiral Plater, Neutec Group Inc., Farmingdale, NY) onto deMan Rogosa Sharpe media (BD Difco, Franklin Lakes, NJ). deMan Rogosa Sharpe plates were incubated anaerobically at 37 °C for 48 h and aerobically at 30 °C for 48 h to culture lactic acid bacteria. Colony counts were measured via spiral plating software (Colony counter, IUL Flash and Go, Neutec Group Inc.) using composite treatment samples in triplicate. No attempt to distinguish organisms was made, only enumeration. For microbial dilution purposes, G-Kefir underwent sterilization of live microorganisms through high-temperature and pressure autoclave cycles.

In vitro fermentation procedures

On the day of the in vitro experiment, kefir samples were prepared so that both contained an equal microbial density (10⁶ CFU/mL). A total of 100 mL of S-Kefir was combined with an anaerobic diluting solution by blending it for 15 s in a Waring blender (Waring Products, New Hartford, CT) under a stream of CO₂. To achieve the same microbial density, G-Kefir (20 mL of fresh G-Kefir + 80 mL of autoclaved G-Kefir) was diluted 1:40 in an anaerobic dilution solution by blending it for 15 s in a Waring blender (Waring Products) under a stream of CO₂. Blended, diluted kefir was filtered through 4 layers of cheesecloth and sealed in 125 mL serum bottles under CO₂. Appropriate samples and blank tubes were then aseptically inoculated with diluted kefir (4 mL) and added to a semi-defined medium (26 mL) as described by Bourquin et al. (1993). Once inoculated, the in vitro assay, microbial DNA extraction and data analysis, and statistical analysis were conducted as described in Study #1.

Results

Study #1: in vitro experiment using fecal inoculum

Microbial profiles of the baseline microbial communities (CNTL; S-Kefir; G-Kefir) of the feces (Table S2) and fecal inocula (Table 1) differed significantly, with relative abundances of nearly 50 bacterial genera being different (P < 0.05) among fecal inoculum sources. Significant treatment*time interactions were observed for pH change in tubes containing CEL (P = 0.05), PC (P < 0.001), BP (P = 0.004), and CP (P = 0.03) (Table 2). The pH change for CEL fermentations was quite small for all treatments (0.03 to −0.08). pH change for CP fermentations was moderate (−0.18 to −0.34), with the reduction being faster in tubes containing G-Kefir inoculum than in those containing CNTL inoculum. The pH reductions were much larger for BP (−0.30 to −0.85) and PC (−0.4 to −1.49) fermentations, with PC tubes containing S-Kefir and G-Kefir inoculum having greater reductions than CNTL tubes.

Table 1.

Relative abundances (% of sequences) of bacterial phyla and genera of fecal inoculum from dogs fed the CNTL, S-Kefir, or G-Kefir treatments at baseline (0 h) among all fiber tubes

Phyla	Genera	CNTL1	S-Kefir	G-Kefir	SEM2	P-value
Actinobacteriota		4.98c	7.53b	10.00a	0.274	<0.001
	Adlercreutzia	0.12b	0.07c	0.17a	0.009	<0.001
	Atopobiaceae, undefined	0.11b	0.13b	0.24a	0.022	<0.001
	Bifidobacterium	4.02c	5.39b	8.71a	0.223	<0.001
	Collinsella	0.09b	0.24a	0.09b	0.033	<0.001
	Coriobacteriaceae UCG-002, undefined	0.59b	1.61a	0.76b	0.091	<0.001
Bacteroidota		7.09b	8.12ab	8.96a	0.508	0.050
	Alloprevotella	0.12b	0.25a	0.06b	0.021	<0.001
	Bacteroides	2.49a	2.28a	1.22b	0.177	<0.001
	Muribaculaceae, undefined	3.23b	3.30b	6.64a	0.256	<0.001
	Parabacteroides	0.08b	0.16a	0.09b	0.01	<0.001
	Prevotella	0.34b	0.89a	0.48b	0.041	<0.001
	Prevotellaceae Ga6A1 group	0.57b	0.94a	0.44b	0.056	<0.001
	Rikenellaceae, undefined	0.05b	0.30a	0.04b	0.012	<0.001
	Rikenellaceae R9 gut group	0.21a	0.00b	0.00b	0.017	<0.001
Campylobacterota		0.01b	0.06a	0.00b	0.004	<0.001
	Helicobacter	0.01b	0.06a	0.00b	0.004	<0.001
Firmicutes		72.21a	69.21ab	67.10b	1.095	<0.010
	Allobaculum	8.10b	6.26c	18.47a	0.454	<0.001
	Anaerovoracaceae, undefined	0.44a	0.41a	0.19b	0.027	<0.001
	Bacilli, undefined	2.27a	2.34a	1.61b	0.077	<0.001
	Blautia	1.79b	3.81a	1.87b	0.078	<0.001
	Butyricicoccus	0.02b	0.06a	0.06a	0.004	<0.001
	Candidatus Stoquefichus	0.08a	0.04b	0.03b	0.005	<0.001
	Cellulosilyticum	0.04b	0.02b	0.08a	0.007	<0.001
	Clostridium sensu stricto 1	0.55a	0.26b	0.72a	0.074	<0.001
	Dubosiella	2.96b	6.68a	6.45a	0.183	<0.001
	Enterococcus	2.42a	1.64b	0.72c	0.084	<0.001
	Erysipelatoclostridium	0.48a	0.27c	0.36b	0.013	<0.001
	Erysipelotrichaceae, undefined	10.87a	8.48b	8.52b	0.432	<0.001
	Faecalibacterium	0.30a	0.11ab	0.09b	0.013	<0.001
	Faecalibaculum	2.27	2.05	2.3	0.097	0.157
	Fournierella	0.24b	0.31a	0.21b	0.024	<0.001
	Holdemanella	0.14b	0.26a	0.09c	0.013	<0.001
	Lachnospiraceae, undefined	7.24a	6.30b	4.42c	0.183	<0.001
	Lactobacillus	14.90b	18.71a	8.77c	0.538	<0.001
	Lactococcus	0.00b	0.00b	0.47a	0.035	<0.001
	Peptoclostridium	2.84a	1.99b	2.59a	0.159	<0.010
	Peptococcus	0.47b	0.59a	0.34c	0.03	<0.001
	Peptostreptococcus	1.34c	2.60b	3.94a	0.12	<0.001
	Phascolarctobacterium	0.04b	0.24a	0.20a	0.018	<0.001
	Romboutsia	1.49a	0.71b	0.89b	0.104	<0.001
	Ruminococcus gauvreauii group	0.36b	0.51a	0.28b	0.016	<0.001
	Ruminococcus gnavus group	0.12b	0.44a	0.41a	0.016	<0.001
	Ruminococcus torques group	2.18a	1.86b	1.21c	0.052	<0.001
	Streptococcus	5.57a	0.70b	0.26b	0.273	<0.001
	Terrisporobacter	0.09b	0.00c	0.17a	0.009	<0.001
	Turicibacter	2.11a	1.16b	1.15b	0.152	<0.001
Fusobacteriota		12.36	13.05	11.15	0.573	0.072
	Cetobacterium	2.16b	2.76b	4.26a	0.156	<0.001
	Fusobacterium	10.20a	10.28a	6.89b	0.447	<0.001
Proteobacteria		3.35a	2.02b	2.78a	0.193	<0.001
	Anaerobiospirillum	0.04	0.05	0.05	0.006	0.213
	Parasutterella	3.21a	1.88b	2.70a	0.181	<0.001

¹CNTL: Control; S-Kefir: starter kefir; G-Kefir: grain kefir.

²Pooled standard error of the means.

^a,b,cMeans in the same row with different superscript letters differ (P < 0.05).

Table 2.

Change in pH during cellulose, pectin, BP, and CP in vitro fermentation assays using fecal inocula from dogs fed the CNTL, S-Kefir, or G-Kefir treatments

	CNTL1			S-Kefir			G-Kefir			SEM	P-values2
Fiber source	6 h	12 h	18 h	6 h	12 h	18 h	6 h	12 h	18 h		Trt	Time	Trt*Time
Cellulose	0.00	0.02	−0.08	0.03	0.01	−0.01	−0.03	−0.03	0.03	0.027	0.407	0.579	0.050
Pectin	−0.74a	−0.86ab	−1.15c	−0.72a	−1.09c	−1.49d	−0.89b	−1.09c	−1.40d	0.033	<0.001	<0.001	<0.001
Beet pulp	−0.30a	−0.67b	−0.85c	−0.40a	−0.79bc	−0.82c	−0.40a	−0.77bc	−0.75bc	0.025	0.019	<0.001	0.004
Chicory pulp	−0.18a	−0.22ab	−0.30ab	−0.20ab	−0.33ab	−0.30ab	−0.34b	−0.33ab	−0.27ab	0.032	0.020	0.116	0.030

¹CNTL: control; S-Kefir: starter kefir; G-Kefir: grain kefir.

²Data were analyzed using the Mixed Models procedure of SAS, including a 2-way ANOVA and repeated measures.

^a,b,c,dMeans in the same row with different superscript letters differ (P < 0.05).

As expected, SCFA production was highest in PC fermentations (Figure 1). Acetate production was higher (P < 0.0001) in tubes containing S-Kefir inoculum (1,859 μmol/g at 18 h) than in tubes containing CNTL (1,575 umole/g at 18 h) or G-Kefir inoculum (1,452 μmol/g at 18 h). Propionate production was higher (P < 0.0001) in tubes containing S-Kefir inoculum (412 μmol/g at 18 h) than in tubes containing CNTL (297 μmol/g at 18 h) or G-Kefir inoculum (43 μmol/g at 18 h). Significant treatment*time interactions were observed for butyrate and total SCFA production. While butyrate production increased in all tubes, its production increased at a greater (P < 0.0001) rate in tubes containing G-Kefir inoculum (456 μmol/g at 18 h) than in tubes containing CNTL (126 μmol/g at 18 h) or S-Kefir inoculum (359 μmol/g at 18 h). Total SCFA production increased in all tubes but increased at a greater (P < 0.0001) rate in tubes containing S-Kefir inoculum (2,630 μmol/g at 18 h) than in tubes containing CNTL (1,839 μmol/g at 18 h) or G-Kefir inoculum (1,951 μmol/g at 18 h).

Figure 1.

Change in acetate (A), propionate (B), butyrate (C), and total short-chain fatty acid (SCFA) concentrations (μmol/g; D) during pectin in vitro fermentation assays using fecal inocula from dogs fed the control (CNTL), starter kefir (S-Kefir), or grain kefir (G-Kefir) treatments. Data were analyzed using the Mixed Models procedure of SAS, including a 2-way ANOVA and repeated measures. ^a,b,c,d,e,fMeans in the same graph with different superscript letters are different (P < 0.05).

Short-chain fatty acid production was slightly lower in BP fermentations, but still increased greatly over time (Figure 2). Acetate production was higher (P < 0.0001) in tubes containing S-Kefir inoculum (1,863 μmol/g at 18 h) than in tubes containing CNTL (1,644 μmol/g at 18 h) or G-Kefir inoculum (1,549 μmol/g at 18 h). Propionate production was higher (P < 0.0001) in tubes containing S-Kefir inoculum (194 μmol/g at 18 h) than in tubes containing G-Kefir inoculum (70 μmol/g at 18 h). Butyrate production was higher (P < 0.0001) in tubes containing S-Kefir (448 μmol/g at 18 h) or G-Kefir inoculum (578 μmol/g at 18 h) than in tubes containing CNTL inoculum (332 μmol/g at 18 h). While total SCFA production increased in all tubes, its production increased at a greater (P < 0.0001) rate in tubes containing S-Kefir inoculum (2,505 μmol/g at 18 h) than in tubes containing CNTL (2,102 μmol/g at 18 h) or G-Kefir inoculum (2,198 μmol/g at 18 h).

Figure 2.

Change in acetate (A), propionate (B), butyrate (C), and total short-chain fatty acid (SCFA) concentrations (μmol/g; D) during BP in vitro fermentation assays using fecal inocula from dogs fed the control (CNTL), starter kefir (S-Kefir), or grain kefir (G-Kefir) treatments. Data were analyzed using the Mixed Models procedure of SAS, including a 2-way ANOVA and repeated measures. ^a,bMeans in the same graph with different superscript letters are different (P < 0.05).

Short-chain fatty acid production was moderate in all CP fermentations when compared with SCFA production in BP (Figure 3). Propionate production was higher (P < 0.0001) in tubes containing CNTL (81 μmol /g at 12 h) or S-Kefir (106 μmol/g at 18 h) inoculum than in tubes containing G-Kefir inoculum (4 μmol/g at 12 h). While acetate and total SCFA production increased in all tubes, their production increased at a greater (P < 0.0001) rate in tubes containing CNTL (acetate: 596 μmol/g at 12 h; total SCFA: 844 μmol/g at 12 h) or S-Kefir inoculum (acetate: 364 μmol/g at 12 h; total SCFA: 746 μmol/g at 18 h) than in tubes containing G-Kefir inoculum (acetate: 119 μmol/g at 12 h; total SCFA: 577 μmol/g at 18 h). Butyrate production also increased in all tubes, but its production increased at a greater (P < 0.0001) rate in tubes containing S-Kefir (407 μmol/g at 18 h) or G-Kefir inoculum (460 μmol/g at 18 h) than in tubes containing CNTL inoculum (404 μmol/g at 18 h). The production of SCFA for CEL fermentations was low and variable for all treatments (Figure S1).

Figure 3.

Change in acetate (A), propionate (B), butyrate (C), and total short-chain fatty acid (SCFA) concentrations (μmol/g; D) during CP in vitro fermentation assays using fecal inocula from dogs fed the control (CNTL), starter kefir (S-Kefir), or grain kefir (G-Kefir) treatments. Data were analyzed using the Mixed Models procedure of SAS, including a 2-way ANOVA and repeated measures. ^a–dMeans in the same graph with different superscript letters are different (P < 0.05).

Bacterial alpha diversity indices of fermentation media fluctuated over time, with significant treatment, time, and treatment*time interactions being observed depending on the fiber source (Figure 4). Alpha diversity fluctuated over time but was higher (P < 0.01) in BP fermentation than in the PC and CP fermentations (Figure 4A and B). For PC, BP, and CP fermentations, alpha diversity was typically higher (P < 0.05) in tubes containing S-Kefir inoculum than in tubes containing CNTL or G-Kefir inoculum. Bacterial beta diversity is represented by PCoA plots of weighted (Figures 5 and 6) and unweighted (Figures S2 and S3) UniFrac distances of fermentation media. When all fibers and time points were considered, both weighted and unweighted plots showed the separation of bacterial populations over time (Figure 5; Figure S2), with the greatest differences noted in PC and BP fermentations.

Figure 4.

Change in bacterial alpha diversity indices (Shannon Index and Observed Features) during in vitro fermentation assays of all fiber sources (A, B), pectin (C, D), BP (E, F), and CP (G, H) using fecal inocula from dogs fed the control (CNTL), starter kefir (S-Kefir), or grain kefir (G-Kefir) treatments. Data were analyzed using the Mixed Models procedure of SAS, including a 2-way ANOVA and repeated measures. ^a,b,cMeans in the same graph with different superscript letters are different (P < 0.05).

Figure 5.

Bacterial beta diversity of all fiber in vitro fermentation assays using fecal inocula from dogs fed the control (CNTL), starter kefir (S-Kefir), or grain kefir (G-Kefir) treatments. The PCoA plot is based on the weighted UniFrac distances of microbial communities, calculated from the 97% operational taxonomic unit abundance matrix using QIIME 2 2023.5. PERMANOVA analysis revealed significant differences between all fibers, treatments, and time points (q < 0.001). All time points were significantly different from one another (q < 0.005) after FDR correction.

Figure 6.

PCoA plots based on weighted UniFrac distances for cellulose (A), pectin (B), BP (C), and CP (D) in vitro fermentation assays using fecal inocula from dogs fed the control (CNTL), starter kefir (S-Kefir), or grain kefir (G-Kefir) treatments.

Significant treatment, time, and treatment*time interactions were observed for several bacterial taxa within all fiber fermentations (Table 3; Figures 7 to 9; Tables S3 to S6; Figure S4). While many bacterial genera were altered, Bacteroides, Prevotella, Faecalibacterium, and Ruminococcus torques group were predominant taxa altered in all fiber fermentations.

Table 3.

Change in bacterial phyla and genera relative abundances (% of sequences) from 0h during cellulose, pectin, BP, and CP in vitro fermentation assays using fecal inocula from dogs fed the CNTL, S-Kefir, or G-Kefir treatments

Fiber	Phyla	Genus	CNTL1			S-Kefir			G-Kefir			SEM	P-values2
			6 h	12 h	18 h	6 h	12 h	18 h	6 h	12 h	18 h		Trt	Time	Trt*Time
Cellulose	Bacteroidota	Parabacteroides	0.22a	0.95a	0.93a	0.40a	0.82a	0.52a	0.06ac	0.46ab	0.73a	0.074	<0.001	<0.001	0.002
		Rikenellaceae, undefined	0.13ab	0.36a	0.33ab	0.21a	0.16a	0.02ab	0.15a	0.21a	0.38a	0.047	0.003	0.084	0.001
	Firmicutes	Negativibacillus	−0.01ab	−0.03a	0.06a	0.07a	0.21a	0.38a	−0.04ac	−0.03ab	0.03a	0.037	<0.001	0.001	0.032
		Oscillospiraceae UCG-005, undefined	0.10b	0.19a	0.22a	0.10b	0.10b	0.11b	0.03bc	0.15ab	0.14ab	0.021	0.001	0.001	0.028
		Peptoclostridium	4.11ab	1.16bc	3.50ab	1.80bc	0.69c	3.52ab	6.82ab	5.01ab	3.66b	0.654	<0.001	0.058	0.010
		Peptostreptococcus	−0.10ab	−0.27b	−0.29b	−1.63bc	−0.33b	1.42ab	−1.07b	2.06ab	2.05ab	0.251	<0.001	<0.001	<0.001
	Proteobacteria	Sutterella	1.61a	4.48a	5.87a	2.67ab	4.80a	5.85a	0.21abc	0.63abc	1.94ab	0.226	<0.001	<0.001	<0.001
		Parasutterella	−0.23bc	−0.89c	−1.28c	2.31a	0.27b	−0.66bc	0.01bc	−1.01c	−0.88bc	0.327	<0.001	<0.001	0.034
Pectin	Bacteroidota	Parabacteroides	0.10bc	0.14b	0.25a	0.13ab	0.22ab	0.08bc	0.03c	0.18ab	0.11bc	0.020	0.011	<0.001	<0.001
		Rikenellaceae, undefined	0.01a	0.06a	0.07a	0.02a	0.07a	−0.12b	0.02a	0.08a	0.08a	0.021	0.002	0.004	0.003
	Firmicutes	Enterococcus	4.77b	2.71c	6.99a	1.88c	0.72cd	0.53cd	1.37c	0.89cd	0.62cd	0.402	<0.001	0.001	<0.001
		Lactobacillus	−6.67c	−6.14c	−11.22cd	−3.37c	−10.35cd	−8.02c	16.78a	8.92b	5.11b	1.394	<0.001	<0.001	0.010
		Phascolarctobacterium	−0.01bc	0.07b	0.58a	−0.11c	0.01bc	0.56a	−0.13c	−0.11c	−0.01bc	0.036	<0.001	<0.001	<0.001
		Ruminococcus gnavus group	0.14c	0.70b	0.37bc	−0.06c	0.75b	0.31bc	−0.11c	0.37bc	1.22a	0.079	0.067	<0.001	<0.001
	Fusobacteriota	Fusobacterium	14.81a	9.09b	5.87b	14.90a	14.50a	5.21b	13.07ab	9.76ab	6.37b	1.046	0.095	<0.001	0.036
	Proteobacteria	Sutterella	0.34c	1.36bc	1.88b	0.74c	2.85a	2.25ab	0.05c	0.60c	1.15bc	0.159	<0.001	<0.001	0.001
Beet Pulp	Actinobacteriota	Collinsella	0.06b	0.01b	0.19a	0.05b	0.19a	0.08b	0.06b	0.11ab	0.06b	0.042	0.769	0.295	0.030
	Firmicutes	Anaeroplasma	0.95b	1.31a	−0.02bcde	0.24b	0.28b	0.02bcde	0.67ab	0.17bc	0.05bcd	0.171	0.012	<0.001	0.012
		Blautia	0.05ab	0.00ab	−0.05ab	−0.30b	0.45a	−0.67b	−0.27b	−0.14ab	0.05ab	0.167	0.439	0.058	0.010
		Lachnospiraceae, undefined	−0.65abc	−0.50abc	1.34ab	−0.99bc	2.31a	−0.27b	−0.22ab	1.05ab	0.73b	0.408	0.400	0.001	0.001
		Peptoclostridium	1.80ab	−0.64b	0.79b	0.75b	0.30b	0.00b	1.99ab	0.31b	−0.28b	0.336	0.435	<0.001	0.014
		Streptococcus	5.91a	0.31b	5.62a	1.29b	1.77b	1.50b	0.68b	0.71b	0.66b	0.348	<0.001	<0.001	<0.001
	Fusobacteriota	Fusobacterium	14.67ab	11.13ab	4.14bc	16.67a	8.97bc	9.70b	23.92a	14.51ab	12.49ab	1.196	<0.001	<0.001	0.028
	Bacteroidota	Parabacteroides	0.09b	0.65a	0.39ab	0.19b	0.19b	0.35b	0.03bc	0.17b	0.24b	0.047	<0.001	<0.001	<0.001
	Proteobacteria	Sutterella	0.72bc	1.40b	1.81ab	0.87bc	0.94b	1.95a	0.13d	0.31bcd	0.70bcd	0.097	<0.001	<0.001	0.004
Chicory Pulp	Bacteroidota	Rikenellaceae, undefined	0.10b	0.18ab	0.17ab	0.01bc	0.04b	−0.02bc	0.12b	0.26ab	0.31a	0.036	<0.001	<0.001	0.001
	Firmicutes	Anaeroplasma	0.28a	0.11b	0.01b	0.10b	0.03b	0.01b	0.39a	0.09b	0.05b	0.029	<0.001	<0.001	0.003
		Negativibacillus	−0.05b	−0.03b	−0.02b	−0.06b	0.09a	0.07ab	−0.06b	−0.06b	−0.04b	0.020	<0.001	0.003	0.015
		Ruminococcaceae, undefined	0.13ab	0.25a	0.24a	0.02b	0.26a	0.28a	0.06b	0.33a	0.28a	0.021	0.157	<0.001	0.005

¹CNTL: Control; S-Kefir: starter kefir; G-Kefir: grain kefir.

²Data were analyzed using the Mixed Models procedure of SAS, including a 2-way ANOVA and repeated measures.

^a,b,c,dMeans in the same row with different superscript letters differ (P < 0.05).

Figure 7.

Change in Bacteroides (A), Prevotella (B), Faecalibacterium (C), and Ruminococcus torques group (D) relative abundances (% sequences) during pectin in vitro fermentation assays using fecal inocula from dogs fed the control (CNTL), starter kefir (S-Kefir), or grain kefir (G-Kefir) treatments. Data were analyzed using the Mixed Models procedure of SAS, including a 2-way ANOVA and repeated measures. ^a,b,c,dMeans in the same graph with different superscript letters are different (P < 0.05).

Figure 9.

Change in Bacteroides (A), Prevotella (B), Faecalibacterium (C), and Ruminococcus torques group (D) relative abundances (% sequences) during CP in vitro fermentation assays using fecal inocula from dogs fed the control (CNTL), starter kefir (S-Kefir), or grain kefir (G-Kefir) treatments. Data were analyzed using the Mixed Models procedure of SAS, including a 2-way ANOVA and repeated measures. ^a,b,c,dMeans in the same graph with different superscript letters are different (P < 0.05).

Figure 8.

Change in Bacteroides (A), Prevotella (B), Faecalibacterium (C), and Ruminococcus torques group (D) relative abundances (% sequences) during BP in vitro fermentation assays using fecal inocula from dogs fed the control (CNTL), starter kefir (S-Kefir), or grain kefir (G-Kefir) treatments. Data were analyzed using the Mixed Models procedure of SAS, including a 2-way ANOVA and repeated measures. ^a,b,c,dMeans in the same graph with different superscript letters are different (P < 0.05).

Study #2: in vitro experiment using S-Kefir or G-Kefir bacterial cultures as inoculum

The microbial density of the S-Kefir (2.21*10⁶ CFU/mL) was much lower than that of the G-Kefir (4.31*10⁹ CFU/mL). The microbial profiles of kefir also differed greatly. S-Kefir was dominated by Lactococcus (98.26%). Streptococcus (1.2%) and 10 other taxa (<0.2%) were present at low levels (Table S7). Lactococcus (72.59%) was also the most predominant genera in G-Kefir, but others including L. kefiranofaciens (17.42%), Lactobacillus (2.41%), Bacillus (2.07%), and Ammoniphilus (1.90%) were also present. Another 15 genera were detected at lower levels (<1.0%).

Significant treatment*time interactions were observed for pH change in tubes containing CEL (P = 0.02), PC (P = 0.05), BP (P < 0.01), and CP (P < 0.01) (Table S8). While the reduction in pH was a bit more rapid in tubes inoculated with S-Kefir for the BP and CP fermentations (at 12 hr), the final pH values were not different between kefir sources for all fibers. Acetate production increased over time in all fermentations (Table S8). For CP fermentations, a treatment*time interaction (P < 0.01) was observed, whereby the increase in acetate production was greater in tubes inoculated with G-Kefir (301 μmol/g at 12 h) than those inoculated with S-Kefir (277 μmol/g at 18 h). Propionate concentrations remained low in all fermentations, slightly fluctuating over time. For CP fermentations, propionate concentrations were higher (P = 0.05) in tubes inoculated with S-Kefir (3.7 μmol/g at 12 h) than those inoculated with G-Kefir (2.5 μmol/g at 12 h). Butyrate production fluctuated over time depending on the fiber source and inoculum source, with significant treatment*time interactions noted in CEL (P < 0.01), BP (P < 0.0001), and CP (P = 0.02) fermentations. For CEL fermentations, butyrate concentration slightly increased over time in tubes inoculated with S-Kefir (3.2 μmol/g at 12 h) but decreased in tubes inoculated with G-Kefir (1.1 μmol/g at 18 h). For BP fermentations, butyrate concentrations slightly increased over time in tubes inoculated with S-Kefir (3.6 μmol/g at 18 h) but fluctuated in tubes inoculated with G-Kefir (18.0 μmol/g at 18 h). Finally, butyrate production was increased in all tubes containing CP, with fermentation patterns differing between inoculum sources. Butyrate increased consistently over time in tubes inoculated with S-Kefir (66 μmol/g at 18 h), which ended with higher concentrations than in tubes inoculated with G-Kefir (16.1 μmol/g at 18 h).

Bacterial alpha diversity indices of fermentation media fluctuated over time, with slight differences being observed due to treatment and fiber (Figure S5). In general, alpha diversity was higher in tubes inoculated with G-Kefir than those inoculated with S-Kefir (P < 0.05; Evenness Index). In regard to the fiber source, alpha diversity was higher in tubes containing CP than in tubes containing the other fibers (P < 0.05; observed features). No differences in bacterial beta diversity, as represented by PCoA plots of unweighted (Figure S6) and weighted (Figure S7) UniFrac distances, were observed. Microbial changes during the fermentation inoculated with kefir are presented in Table S9.

Discussion

The increasing popularity of fermented foods such as kefir raises questions about their effects on host health (Vinderola et al., 2023). Variation in microbial composition is expected among fermented foods because environmental factors largely impact which species proliferate. Kefir derived from grains is defined by the Codex Alimentarius Standard as containing L. kefiri, species of Leuconostoc, Lactococcus, and Acetobacter, as well as yeasts such as K. marxianus and Saccharomyces cerevisiae (Codex Alimentarius Commission, 2018; González‐Orozco et al., 2023). Despite many attempts to replicate grains or their effects on large-scale production, the fermentation of kefir grains is the only way to achieve an authentic kefir product (Rosa et al., 2017). Amongst similar products, a regulatory standard of identity currently exists only for yogurt, which establishes a minimal inclusion of specific microbes such as L. delbrueckii subsp. bulgaricus and S. thermophilus to be labeled as such (21 CFR 131.200). Because kefir does not have an established standard of identity, it allows companies to sell a variety of products without the use of kefir grains, the inclusion of authentic kefir microbes, or a minimal microbe inclusion level. This discrepancy highlights the need for label differentiation between starter and grain kefir, as well as investigations into the impacts of associated microbial communities (Bourrie et al., 2021; Nejati et al., 2022).

The success of the companion animal probiotic supplement industry has led to an influx of fermented foods and beverages into the market. Recent testing has demonstrated a lack of ingredient label integrity and accuracy (Metras et al., 2020). Also, while kefir products for companion animals have been researched (Kim et al., 2019; Gaspardo et al., 2020; Metras et al., 2023), more testing of such products is necessary. Dietary fibers are another way to support gastrointestinal health of companion animals. Chicory and BP, for instance, are commonly used fiber sources in dog foods. Understanding how the live microorganisms present in fermented foods and probiotics interact with these fibers may lead to new combinations that have improved synergism and efficacy.

As researchers strive to gain a better understanding of the canine microbiome and how it impacts host health, a reliance on fecal samples has limited progress. Fecal samples provide SCFA concentrations and microbial composition, but cannot provide insight into microbial activity (e.g., SCFA production). Our previous in vivo canine study did not demonstrate significant changes in SCFA concentrations after kefir consumption, but the results relied on fecal samples (Metras et al., 2023). In this study, we chose to test the effects of kefir using an in vitro fermentation assay that allowed the assessment of microbial activity. In vitro fermentation assays are often used to test the fermentability of individual fiber sources (Payne et al., 2012; de Godoy et al., 2015; Carlson et al., 2017; Traughber et al., 2020), but may also be used to assess the activity of fecal microbiota collected from animals given different treatments (Paßlac et al., 2022; Zhang et al., 2023). Although these methodologies of simulated fermentation cannot exactly capture the internal and external environment, nuance, and complexity of fermentation within an animal, the data generated from them complement in vivo data. The S-Kefir product tested was ordered from the same vendor as in the in vivo study. Moreover, the grains used for G-Kefir production were well preserved and used for daily fermentations following the same protocol as the in vivo study.

The primary SCFA, namely acetate, propionate, and butyrate, have been estimated to provide 60-70% of the energy requirements of colonic epithelial cells (Brahe et al., 2013). These acids represent the main carbon flow from the diet, microbiome, and to the host, with commensal microbes such as Faecalibacterium, Lachnospiraceae, lactic acid bacteria, and Streptococcus known to convert bacterial metabolites such as lactate and acetate into butyrate via the acetyl-CoA pathway (Singh et al., 2023). Although it is not possible to know the exact source of SCFA producers in the current study, many commensal microbes detected such as Ruminococcus torques and Clostridium are known to produce SCFA by converting pyruvate via the acetyl-CoA pathway (Markowiak-Kopeć and Śliżewska, 2020; Deleu et al., 2021; Singh et al., 2023). Total SCFA concentrations differed significantly by treatment*time interaction effects in PC, BP, and CP in vitro fermentations when fecal inoculations were used.

Of the SCFA, butyrate is the preferred energy source of host epithelial cells and is known to enhance intestinal barrier function and support mucosal immunity (Fu et al., 2019). Our results demonstrate that butyrate concentrations were highest in both samples where feces from kefir-fed dogs was used as inoculum to ferment selected dietary fibers, with G-Kefir supplementation producing the highest butyrate output overall. While the fecal microbiota populations of our previous study testing these treatments did not identify many changes (Metras et al., 2023), these data suggest that the activity and/or capabilities of the microbial communities differed under in vitro conditions.

The significant abundance of Faecalibacterium in all fecal treatments fermenting BP and CP is supported by literature demonstrating that this genus can break β-(2,1) glycosidic bonds while being a prolific butyrate producer (Moens et al., 2017). This finding is relevant to canine health, as this taxa along with other butyrate-producing bacteria has been shown to be compromised in dogs with IBD (Suchodolski, 2016; Rhimi et al., 2022). Other butyrate producers (e.g., Prevotella, Ruminococcus, Fusobacterium, Lachnospiraceae) may have also contributed to its production. Even though butyrate producers may receive more attention, it is important to highlight the cross-feeding interactions that occur between butyrate-producing bacteria and the microbial taxa responsible for the production of lactate and acetate that feed those metabolic pathways (Rivière et al., 2016; Singh et al., 2023). Collectively, changes to the in vitro microbiota populations due to kefir supplementation were deemed to be beneficial.

To our knowledge, this is the first study testing the fermentative capacity of the bacteria present in kefir. It was important to conduct both in vitro fermentations (Study #1 and Study #2) to determine whether SCFA production could be attributed only to the microbiota present in fecal samples or also to kefir microbiota as well. The combination of biotics with dietary fibers and prebiotics like chicory or BP are becoming more common in the marketplace. It was for this reason that those particular fibers were used in the in vitro fermentation study. Monoculture experiments are often used to investigate individual bacterial strains, but testing the kefir microbial strains from the complete product allowed us to assess the capabilities of the entire community collectively. Testing the kefir microbiota enabled cross-feeding and community dynamics to play out, allowing the data generated to have greater application to the supplement market and canine health. Kefir microbiota populations shifted over time and were able to ferment the test fibers somewhat, but to a much lower extent than fecal microbial populations. pH was reduced during kefir inocula fermentations, but much lower SCFA production resulted than those inoculated with feces. This was expected to some extent because kefirs do not contain many of the fiber-fermenting commensal bacteria that inhabit the gastrointestinal tract. In addition to resulting in much lower SCFA production, the lack of cross-feeding capabilities among the members of the kefir community was obvious when evaluating SCFA profiles. The primary output of kefir fermentations was acetate, with virtually no accumulation of propionate and a very low production of butyrate. These results suggest that kefir-derived bacteria will likely not produce butyrate in the gastrointestinal tract, but may produce acetate, lactate, or other intermediates that contribute to its production by other members of the gut microbiota population.

There are several limitations to using in vitro fermentation assays as a model for the host gastrointestinal system and for this study in particular. While the assay includes a microbiological medium enriched with vitamins and is conducted in a way that controls temperature and pH within an anaerobic system, it does not correct for gastrointestinal secretions (e.g., mucins, host enzymes, bacteriocins), and SCFA are not removed from the system as they are in vivo. The addition of essential vitamins to the medium, which were included to help microbes initially survive the transition to the in vitro environment, may have impacted outcomes. In addition, the journey that lives microbes undergo throughout the digestive tract involves passage through extreme acidity in gastric digestion and competition with native commensal microbes. These differences may impact how microbiota respond in the system, cross-feed, and consequently how fibers are fermented. While 3-time points were sampled for outcome measures, more frequent sampling may provide a more complete picture of the fermentation kinetics. In regard to the application of data to the canine population, the fecal inoculum was collected from a small sample size (n = 4/treatment) and was used to ferment only 4 fiber sources commonly present in pet foods. A larger study using more animals from more diverse backgrounds and/or the testing of other dietary fibers may have resulted in different outcomes. While blank samples were used in the in vitro study, dogs were either fed kefir or a milk control so a true placebo control diet was not used, potentially affecting the data generated. The preferred substrate of kefir bacteria is lactose, thus using fiber as the sole energy source may explain reduced microbial abundances and SCFA production. Efforts were made to correct for the higher microbial density in G-Kefir by diluting it 1:4 with sterilized G-Kefir product so that upon inoculation, both kefirs contained about 10⁶ CFU/mL in the second fermentation experiment. The focus of these experiments was on changes in microbiota, yet it should be noted many species of yeast are present in G-Kefir and may have contributed to SCFA generation and fiber digestion. Finally, fermented foods and beverages are expected to vary in regard to microbial composition, even among similar batches, so truly encapsulating an all-encompassing comparison of starter and grain kefir will require further research.

Glossary

Abbreviations

BP

beet pulp

CEL

cellulose

CNTL

control

CP

chicory pulp

DM

dry matter

DMB

dry matter basis

G-Kefir

grain kefir

PC

pectin

PCoA

principal coordinates analysis

SCFA

short-chain fatty acids

S-Kefir

starter kefir

Conflict of interest statement

The authors have no conflicts of interest.

Author contributions

Breanna Metras (Conceptualization, Data curation, Formal analysis, Writing—original draft), Patrícia Oba (Data curation, Formal analysis), Dalton Holt (Data curation), Laura Bauer (Conceptualization, Data curation, Supervision, Writing—review & editing), Michael Miller (Conceptualization, Supervision, Writing—review & editing), Ryan Dilger (Conceptualization, Supervision, Writing—review & editing), and Kelly Swanson (Conceptualization, Funding acquisition, Project administration, Supervision, Writing—review & editing)