Weight loss and high-protein, high-fiber diet consumption impact blood metabolite profiles, body composition, voluntary physical activity, fecal microbiota, and fecal metabolites of adult dogs

Abstract

Canine obesity is associated with reduced lifespan and metabolic dysfunction, but can be managed by dietary intervention. This study aimed to determine the effects of restricted feeding of a high-protein, high-fiber (HPHF) diet and weight loss on body composition, physical activity, blood metabolites, and fecal microbiota and metabolites of overweight dogs. Twelve spayed female dogs (age: 5.5 ± 1.1 yr; body weight [BW]: 14.8 ± 2.0 kg, body condition score [BCS]: 7.9 ± 0.8) were fed a HPHF diet during a 4-wk baseline phase to maintain BW. After baseline (week 0), dogs were first fed 80% of baseline intake and then adjusted to target 1.5% weekly weight loss for 24 wk. Body composition using dual-energy x-ray absorptiometry and blood samples (weeks 0, 6, 12, 18, and 24), voluntary physical activity (weeks 0, 7, 15, and 23), and fresh fecal samples for microbiota and metabolite analysis (weeks 0, 4, 8, 12, 16, 20, and 24) were measured over time. Microbiota data were analyzed using QIIME 2. All data were analyzed statistically over time using SAS 9.4. After 24 wk, dogs lost 31.2% of initial BW and had 1.43 ± 0.73% weight loss per week. BCS decreased (P < 0.0001) by 2.7 units, fat mass decreased (P < 0.0001) by 3.1 kg, and fat percentage decreased (P < 0.0001) by 11.7% with weight loss. Many serum metabolites and hormones were altered, with triglycerides, leptin, insulin, C-reactive protein, and interleukin-6 decreasing (P < 0.05) with weight loss. Relative abundances of fecal Bifidobacterium, Coriobacteriaceae UCG-002, undefined Muribaculaceae, Allobaculum, Eubacterium, Lachnospira, Negativivibacillus, Ruminococcus gauvreauii group, uncultured Erysipelotrichaceae, and Parasutterella increased (P < 0.05), whereas Prevotellaceae Ga6A1 group, Catenibacterium, Erysipelatoclostridium, Fusobacterium, Holdemanella, Lachnoclostridium, Lactobacillus, Megamonas, Peptoclostridium, Ruminococcus gnavus group, and Streptococcus decreased (P < 0.01) with weight loss. Despite the number of significant changes, a state of dysbiosis was not observed in overweight dogs. Fecal ammonia and secondary bile acids decreased, whereas fecal valerate increased with weight loss. Several correlations between gut microbial taxa and biological parameters were observed. Our results suggest that restricted feeding of a HPHF diet and weight loss promotes fat mass loss, minimizes lean mass loss, reduces inflammatory marker and triglyceride concentrations, and modulates fecal microbiota phylogeny and activity in overweight dogs.

Lay Summary

Canine obesity is associated with reduced lifespan and metabolic dysfunction, but dietary intervention may aid in its management. This study aimed to determine the effects of restricted feeding of a high-protein, high-fiber (HPHF) diet and weight loss on body composition, physical activity, blood metabolites, and fecal bacteria and metabolites of overweight dogs. Twelve overweight dogs were fed a HPHF diet during a 4-wk baseline to maintain body weight and then fed to lose weight for 24 wk. Body composition, blood samples, voluntary physical activity, and fresh fecal samples were measured over time. After 24 wk, dogs lost over 30% of their initial body weight and had 1.4% weight loss per week. As expected, serum triglycerides, leptin, insulin, C-reactive protein, and interleukin-6 decreased with weight loss. The relative abundances of 4 bacterial phyla and over 30 bacterial genera were altered with weight loss. Fecal ammonia and secondary bile acid concentrations decreased, whereas fecal valerate concentrations increased with weight loss. Several correlations between fecal bacteria and physiological parameters were identified. Our results suggest that a HPHF diet and weight loss promote fat mass loss, reduce inflammatory marker and triglyceride concentrations, and modulate fecal bacterial populations and activity in overweight dogs.

This study demonstrates that restricted feeding of a high-protein, high-fiber diet and weight loss promotes fat mass loss, reduces inflammatory marker and triglyceride concentrations, and modulates fecal bacterial populations and activity in overweight dogs.

Introduction

Obesity is a global epidemic disease and its prevalence continues to increase in humans, dogs, and cats. The human-companion animal bond has been researched extensively and shown to improve human health outcomes (e.g., reduction in depression, anxiety, and mortality; Friedman and Krause-Parello, 2018; Miyake et al., 2020). However, this bond and a changing living environment also promote pet obesity as a result of owner misinterpretation of their pet’s body condition score (BCS), inappropriate feeding, insufficient exercise, and gonadectomy. Additional factors that increase pet obesity risk include genetics, age, sex, and health conditions (German, 2006; Laflamme, 2006; Courcier et al., 2010; Linder and Mueller, 2014; Rowe et al., 2017; Simpson et al., 2019).

Obesity is not only a multifactorial disease, but it also increases risk of orthopedic disease, cardiorespiratory disorders, intestinal dysbiosis, gastrointestinal disorders, and cancers (German, 2006; Weiss and Hennet, 2017; Upadhyay et al., 2018; Emerenziani et al., 2019). Because obesity causes alteration of adipokine production, its consequences affect insulin secretion and glucose and lipid homeostasis, contributing to the development of obesity-related metabolic dysfunction and gut dysbiosis (Kahn et al., 2006; Ndumele et al., 2006; Saad et al., 2016). In rodent models and humans, obesity, insulin resistance, and gut microbiota have been linked and thought to be potential pathophysiological causes of one another. Intestinal bacteria have been shown to increase dietary energy extraction and produce metabolites and cytokines that affect host metabolism and consequent weight gain. In some studies, obesity has been shown to diminish gut microbial diversity and alter its composition, alter tight junction protein production, and increase intestinal permeability, leading to lipopolysaccharide translocation and consequent inflammation and insulin resistance (Lee et al., 2020).

Fortunately, the risk of obesity-related morbidities can be reduced by restricted feeding and consequent weight reduction. Generally, successful weight loss comes from a combination of increased physical activity and the restricted feeding of an appropriately formulated diet. High-protein and/or high-fiber diets are typically considered to be the best option when it comes to weight loss programs for dogs and cats (Blanchard et al., 2004; German et al., 2007; Floerchinger et al., 2015; André et al., 2017; Kieler et al., 2017; Pallotto et al., 2017; Salas-Mani et al., 2018; Bermudez Sanchez et al., 2020). Several studies have shown the positive effects of weight loss in dogs, with benefits including improved metabolism, metabolic outcomes, and mobility, decreased disease risk, improved quality of life, and increased life expectancy (German, 2016). Additionally, weight loss reduces serum leptin, insulin, glucose, interleukin-6 (IL-6), and C-reactive protein (CRP) concentrations and increases serum adiponectin concentration and physical activity level (Jeusette et al., 2005; German et al., 2009; Warren et al., 2011; Bastien et al., 2015; Floerchinger et al., 2015; Starr et al., 2019). The relationships between weight loss and gut microbiota have been noted in a few canine studies (Kieler et al., 2017; Salas-mani et al., 2018; Bermudez Sanchez et al., 2020). However, none of those experiments integrated and investigated longitudinal changes in serum metabolites, gut microbiota, and fecal metabolites during weight loss. Additionally, identifying strong correlations among those variables may be key to understanding energy balance and identifying potential remedies of obesity.

The objective of this study was to determine the effects of weight loss on body composition, voluntary physical activity, blood metabolite profiles, serum markers of inflammation, fecal microbiota populations, and fecal metabolites of overweight dogs fed a high-protein, high-fiber (HPHF) diet. We hypothesized that closely monitoring BW and adjusting intake of a HPHF diet would lead to steady weight loss and increased fat loss while maintaining lean mass. Additionally, weight loss was hypothesized to increase voluntary physical activity, reduce blood lipids, and decrease serum inflammatory markers. Based on the results of our previous study testing a similar diet in dogs (Phungviwatnikul et al., 2021), weight loss and consumption of a HPHF diet were expected to beneficially alter the fecal microbiota community (e.g., increased Bifidobacterium, Lactobacillus, Faecalibacterium, Romboutsia, Fusobacterium and decreased Catenibacterium, Streptococcus, and Megamonas) and metabolite concentrations (e.g., increased short-chain fatty acid [SCFA] and total bile acids concentrations and decreased protein catabolites) in overweight dogs.

Materials and Methods

All procedures were approved by the University of Illinois Institutional Animal Care and Use Committee prior to experimentation (IACUC #18268).

Experimental design

Twelve overweight adult spayed female beagles (age: 5.5 ± 1.1 yr; BW: 14.8 ± 2.0 kg; BCS: 7.9 ± 0.75) were used in a longitudinal weight loss study. The experiment consisted of 28 wk, with a 4-wk baseline phase, followed by a 24-wk weight loss phase. All dogs were considered healthy except for being overweight. Complete blood count and serum biochemistry panel were within the reference range. Dogs had not received any medications that could affect blood parameters and gut microbiota for at least 4 wk before and during the experiment. The maintenance energy requirements (MER) of all dogs were obtained during a 4-wk baseline phase, where animals were fed the experimental diet.

The diet was formulated to meet all Association of American Feed Control Officials (AAFCO, 2018) nutrient recommendations for adult dogs at maintenance (Table 1). Several dietary fiber and prebiotic sources and functional ingredients were used in the diet. Inclusion of barley, beet pulp, cellulose, psyllium husk, scFOS, and brown flax seed has been shown to provide beneficial effects on the host, with this mixture providing a balance of insoluble and soluble dietary fibers in the diet. Additionally, the inclusion rate of scFOS and high concentrations of barley and beet pulp in diets were reported in previous literature and were well-tolerated by dogs (Respondek et al., 2007; de Godoy, 2011; Donadelli and Aldrich, 2019). Furthermore, several functional ingredients were included in the diet. L-carnitine aids in long-chain fatty acid transport and energy production, especially during weight loss, and has been shown to improve energy expenditure of dogs (Varney et al., 2020). The concentration of L-carnitine inclusion followed AAFCO (2018) recommendations. Fish oil was included in the diet for its anti-inflammatory and blood lipid-lowering properties in dogs (Adler et al., 2018; de Godoy et al., 2018). Green tea extract was included in the diet because it has been shown to improve insulin sensitivity and lipid profile of obese dogs (Serisier et al., 2008). Chromium, added in the form of chromium picolinate, is an essential trace mineral that is required for the cellular uptake of glucose (Schachter et al., 2001). Lastly, vitamin C and vitamin E are important in metabolic function and have antioxidant properties (de Oliverira El-Warrak et al., 2012; Gordon et al., 2020). The concentrations of fish oil, green tea extract, chromium picolinate, vitamin C, and vitamin E were based on the published literature.

Table 1.

Ingredient and analyzed chemical composition of the high-protein, high-fiber experimental diet

Ingredient	% as-is	Analyzed composition	%, DM
Poultry meal	35.00	Dry matter (DM), %	92.47
Soy protein concentrate	22.00	Organic matter	89.65
Barley	14.00	Ash	10.32
Beet pulp	10.00	Crude protein	42.04
Brewer’s rice	4.075	Acid hydrolyzed fat	12.14
Chicken fat	3.00	Crude fiber	3.90
Liquid palatant	3.00	Total dietary fiber	26.81
Cellulose	2.00	Insoluble fiber	15.00
Fish oil	2.00	Soluble fiber	11.81
Psyllium husk	1.00	Nitrogen-free extract1	8.69
Short-chain fructooligosacharides2	1.00	Metabolizable energy (ME)1, kcal/g	2.81
Brown flax seed	0.50	Gross energy, kcal/g1	4.49
Powder palatant	0.50	Macronutrients on energy basis (% ME)
Sodium chloride	0.50	Protein	52.41
Potassium chloride	0.45	Fat	36.76
Vitamin premix3	0.18	Carbohydrate	10.83
Mineral premix4	0.18
L-carnitine, 50%	0.15
Choline chloride	0.13
Green tea extract	0.10
Natural antioxidant5	0.10
Vitamin C	0.06
Vitamin E	0.05
Chromium methionine6	0.025

Nitrogen-free extract = 100 − (ash + crude protein + acid hydrolyzed fat + total dietary fiber); metabolizable energy = 8.5 kcal ME/g fat + 3.5 kcal ME/g protein + 3.5 kcal ME/g nitrogen-free extract; gross energy was measured by bomb calorimetry.

Short-chain fructooligosaccharides: Fortifeed scFOS prebiotic fiber, Ingredion Inc., Westchester, IL, USA.

Provided per kg diet: vitamin A, 5.28 mg; vitamin D3, 0.04 mg; vitamin E, 120.00 mg; vitamin K, 0.88 mg; thiamin, 4.40 mg; riboflavin, 5.72 mg; pantothenic acid, 22.00 mg; niacin, 39.60 mg; pyridoxine, 3.52 mg; biotin, 0.13 mg; folic acid, 0.44 mg; vitamin B12, 0.11 mg.

Provided per kg diet: Mn (as MnSO4), 66.00 mg; Fe (as FeSO4), 120 mg; Cu (as CuSO4), 18.00 mg; Co (as CoSO4), 1.20 mg; Zn (as ZnSO4), 240 mg; iodine (as KI), 180 mg; Se (as Na2SeO3), 0.24 mg.

Natural antioxidant: Naturox liquid antioxidant, blend of vegetable oil, natural mixed tocopherols, lecithin, and rosemary extract.

Chromium methionine: Microplex, a national feed ingredient for animals that contains organic chromium, Zinpro Corporation, Eden Prairie, MN, USA.

After the baseline phase (week 0), dogs were fed at a rate to lose approximately 1.5% BW per week (range from 1% to 2% for dogs) as recommended by the American Animal Hospital Association—weight management guidelines for dogs and cats (Brooks et al., 2014). To achieve weight loss, dogs initially received 80% of the food required to maintain BW during the baseline phase and then energy intake was adjusted weekly based on the level of weight loss. Dogs were weighed and BCS (9-point scale; Laflamme, 1997) was evaluated weekly, with all being performed in the morning before feeding. Once dogs met their target weight, food intake (FI) was adjusted to maintain BW.

Dogs were housed individually in pens (1.22 m wide × 1.85 m long) in a temperature-controlled room under a 12 h light:12 h dark cycle in the Veterinary Medicine Basic Sciences Building at the University of Illinois. Dogs had free access to fresh water and were fed twice daily (9:00 a.m. and 5:00 p.m.) throughout the study. Food offerings and refusals were measured daily to calculate intake. Dogs were allowed outside of their pens a couple days a week for socialization with other dogs in compatible groups and humans, except on collection days. Pens were cleaned daily and dogs were bathed every 2 wk.

Chemical analysis of diets

The experimental diet was subsampled and ground through a 2-mm screen using a Wiley mill (model 4, Thomas Scientific, Swedesboro, NJ). The sample was analyzed according to procedures of the Association of Official Analytical Chemists (AOAC) for dry matter (DM; 105 °C) and ash (organic matter [OM] was calculated from ash) (AOAC, 2006; methods 934.01, 942.05). Crude protein content was calculated from Leco total N values (TruMac N, Leco Corporation, St. Joseph, MI; AOAC, 2006). Total lipid content (acid-hydrolyzed fat) of the sample was determined according to the methods of the American Association of Cereal Chemists (AACC, 1983) and Budde (1952). Gross energy of the diet was measured using an oxygen bomb calorimeter (model 6200, Parr Instruments, Moline, IL). Total dietary fiber (TDF) content was determined according to Prosky et al. (1985).

Body composition

Body composition was evaluated by dual-energy x-ray absorptiometry (DEXA: Hologic X-ray Bone Densitometer QDR 4500 Elite Acclaim Series) at the University of Illinois Veterinary Teaching Hospital at baseline (week 0) and after 6, 12, 18, and 24 wk of the weight loss. To perform DEXA scans, dogs were sedated by an intramuscular injection of Dexdomitor (0.02 mg/kg) and Torbugesic (0.2 mg/kg), and positioned in sternal recumbency. The four legs, trunk, and head of each dog were scanned individually, and measurements of fat, lean, and bone mineral content were taken in each body region. Body fat percentage was calculated for each part and the entire body. After the measurement, an intramuscular injection of the reversal agent for dexmedetomidine, atipamezole (0.2 mg/kg BW), was given.

Voluntary physical activity

At baseline (week 0), weeks 7, 15, and 23, accelerometers (Actical devices: Mini Mitter, Bend, OR) were used to measure voluntary physical activity. During activity monitoring periods, Actical devices were attached to collars worn around the neck for six consecutive days. Mean activity was presented in activity counts per epoch (epoch length = 0.25 min), with light hour (0700–1900 h) and dark hour (1900–0700 h) also being measured.

Complete blood count, serum chemistry profile, blood hormones, and inflammatory markers

Fasted (at least 10 h) blood samples were collected via jugular or cephalic puncture at baseline (week 0) and after 6, 12, 18, and 24 wk of weight loss. Samples were immediately transferred to appropriate vacutainer tubes. Sterilized glass serum tubes (#366430 BD Vacutainer, Becton Dickinson, Franklin Lakes, NJ) were used for serum chemistry profile, hormones (leptin, insulin), and inflammatory marker (CRP, IL-6) analyses. Plastic whole blood tubes with K₂EDTA additive (#365974 BD Microtainer, Becton Dickinson, Franklin Lakes, NJ) were used for complete blood count. Tubes were centrifuged at 2,000 × g for 15 min at 4 °C for serum collection. Serum chemistry profile and complete blood count were analyzed using a Hitachi 911 clinical chemistry analyzer (Roche Diagnostics, Indianapolis, IN) at the University of Illinois Veterinary Medicine Diagnostics Laboratory.

The concentrations of serum leptin, insulin, CRP, and IL-6 were measured using commercial ELISA kits (leptin: #EZCL-31K, MilliporeSigma, Burlington, MA; insulin: #10-1203-01, Mercodia, Winston Salem, NC; CRP: #ab157698, Abcam, Cambridge, MA; IL-6: #ab193686, Abcam, Cambridge, MA).

Fecal sample collection

Fresh fecal samples (within 15 min of defecation) were collected at baseline (week 0) and after 4, 8, 12, 16, 20, and 24 wk of weight loss for measurement of fecal scores, pH, DM content, microbiota populations, and fermentative metabolite concentrations. Fecal pH was measured immediately using an Accumet AP1001 Portable pH Meter Kit (Fisher Scientific, Waltham, MA) and then feces were aliquoted for other measures, including SCFA and protein fermentative products (ammonia and branched-chain fatty acids [BCFA]). One fecal aliquot (~5 g/dog) was collected and placed in 2 N hydrochloric acid in a 1:1 (weight: weight) ratio and stored at −20 °C for SCFA, BCFA, and ammonia analyses. An additional aliquot was collected for DM determination. Finally, 4 aliquots of fresh feces were collected in sterile cryogenic vials (Nalgene, Rochester, NY), frozen on dry ice, and stored at −80 °C for microbiota and bile acid analyses.

Fecal scores

All fecal samples during the collection phase were scored according to the following scale 1 to 7 (Greco, 2015): 1 = very hard and dry, often expelled as individual pellets, no residue left on ground when picked up; 2 = firm but not hard, segmented in appearance, little or no residue on ground when picked up; 3 = log-shape, moist surface, leaves residue on ground but hold form when picked up; 4 = very moist, soggy, log-shaped, leaves residue and loses form when picked up; 5 = very moist but has a distinct shape, piles rather than distinct logs, leaves residue and loses form when picked up; 6 = has texture but no defined shape, present as piles or spots, leaves residue when picked up; 7 = watery, no texture, flat puddles.

Fecal chemical analyses

Fecal samples were analyzed according to procedures of the Association of Official Analytical Chemists (AOAC, 1975) for DM using a 105 °C oven. Fecal SCFA and BCFA concentrations were determined by gas chromatography according to Erwin et al. (1961) using a gas chromatograph (Hewlett-Packard 5890A series II, Palo Alto, CA) and a glass column (180 cm × 4 mm i.d.) packed with 10% SP-1200/1% H₃PO₄ on 80/100+ mesh Chromosorb WAW (Supelco Inc., Bellefonte, PA). Nitrogen was the carrier with a flow rate of 75 mL/min. Oven, detector, and injector temperatures were 125, 175, and 180 °C, respectively. Fecal ammonia concentrations were determined according to the method of Chaney and Marbach (1962).

The protocol for quantifying bile acids was adapted and modified from the methods previously described by Batta et al. (2002). Briefly, an aliquot of 10-15 mg lyophilized stool was added to 200 µL of 1-butanol containing internal standards (cholic acid-d₄ and lithocholic acid-d₄) followed by adding 20 µL of hydrochloric acid. Samples were incubated for 4 h at 65 °C. Following incubation, samples were completely evaporated at 65 °C under nitrogen gas, 200-µL trimethylsilylation derivatization agent was added, and samples were incubated for 30 min. The samples were then evaporated under nitrogen gas and resuspended in 200-µL hexane, vortexed, and centrifuged at 4 °C for 10 min at 3,000 × g. The supernatant was then analyzed by gas chromatography and mass spectrometry according to methods described by Blake et al. (2019). Cholic acid (CA), chenodeoxycholic acid (CDCA), lithocholic acid (LCA), deoxycholic acid (DCA), and ursodeoxycholic acid (UDCA) were measured.

Fecal microbiota populations

Total DNA from fecal samples was extracted using DNeasy PowerLyzer PowerSoil Kit (Qiagen, Carlsbad, CA) with bead beating using a vortex adaptor. The concentration of extracted DNA was quantified using a Qubit 3.0 Fluorometer (Life Technologies, Carlsbad, CA). 16S rRNA gene amplicons of the V4 region were generated using a Fluidigm Access Array (Fluidigm Corporation, South San Francisco, CA) in combination with a Roche High Fidelity Fast Start Kit (Roche, Indianapolis, IN). The primers 515F (5ʹ-GTGYCAGCMGCCGCGGTAA-3ʹ) and 806R (5ʹ-GGACTACNVGGGTWTCTAAT-3ʹ) that target a 252 bp-fragment of that region were used for amplification (primers synthesized by IDT Corp., Coralville, IA; Caporaso et al., 2012). CS1 forward tag and CS2 reverse tag 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 2% agarose E-gel (Life Technologies, Carlsbad, CA) and extracted using a Qiagen gel purification kit (Qiagen, Carlsbad, CA). Cleaned size-selected pooled products were run on an Agilent 2100 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 W. M. Keck Center for Biotechnology at the University of Illinois.

Fecal microbiota bioinformatics and statistical analyses

Forward reads were trimmed using the FASTX-Toolkit (version 0.0.14) and QIIME 2.2019.4 (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 were demultiplexed. Data were then denoised and assembled into amplicon sequence variants (ASV) using DADA2 (Callahan et al., 2016). The SILVA 132 database (Quast et al., 2013) was used to assign taxonomy. An even sampling depth (23,385 sequences per sample) was used to assess 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 analyses

FI, CI, BW, BCS, body composition, serum chemistry, complete blood cell count, blood hormones, inflammatory markers, fecal characteristics, fecal metabolites, and fecal microbiota data were analyzed using the linear Mixed Models procedure of SAS (version 9.4; SAS Institute, Cary, NC). Data were analyzed using repeated measures analysis, with differences due to time being the focus. All results are presented as least squares means ± standard error of the mean (SEM). A P < 0.05 was considered significant and a P < 0.10 was considered a trend. The R programming language (RStudio version 1.1.463) was used to calculate the Spearman correlation coefficients between key biological parameters (BW, BCS, FI, CI, fecal bile acids [CA, CDCA, LCA, DCA, UDCA, primary bile acid, secondary bile acid, total bile acid], fecal metabolites [SCFA, BCFA, ammonia], serum metabolites [cholesterol, triglycerides], serum hormones [leptin, insulin], serum inflammatory markers [CRP, IL-6], and total fat mass) and gut microbiota, with P < 0.05 considered significant, and the results displayed using a heat map.

Results

Food intake, caloric intake, BW, and BCS

At baseline, average MER was 89.09 ± 19.5 kcal × BW kg^0.75, average BW was 14.8 ± 2.0 kg, and average BCS was 7.9 ± 0.75. BCS decreased by 2.7 units (P < 0.0001) over 24 wk of restricted feeding and weight loss. Nine out of 12 dogs achieved an ideal BCS (BCS 5/9) by week 24, with 3 remaining overweight dogs having a BCS of 5.5 to 6. During weight loss, the average MER (weeks 1 to 24) of all dogs was reduced to 70.51 ± 7.66 kcal × BW kg^0.75, with MER being ~60-62 kcal × BW kg^0.75 from weeks 14 to 24 when a consistent level of weight loss was reached. After the 24-wk weight loss phase, dogs lost 31.2% of initial BW (P < 0.0001), with 1.43 ± 0.73% weight loss per week. Dogs consumed an average of 457.5 ± 61.4 kcal/d throughout the entire weight loss phase (weeks 1 to 24), but settled on an approximate caloric intake of 360–400 kcal/d from weeks 14 to 24 when a consistent level of weight loss was reached (Figure 1). Energy intake was reduced by 30.96% by week 24 compared to baseline (P < 0.0001). Once dogs reached their target BW and had BCS 5, which started at week 14 in some dogs, they received an amount of food to maintain BW once again.

Figure 1.

Caloric intake (kcal/d) and body weight (kg) data of overweight adult female dogs fed a high-protein, high-fiber diet during weight loss. Data are presented as least square means ± SEM.

Body composition

After 24 wk of restricted feeding and weight loss, lean mass, fat mass, and fat percentage were reduced (P < 0.0001) by 1.27 kg, 3.12 kg, and 11.72%, respectively, whereas lean mass percentage was increased (P < 0.0001) by 11.33% (Table 2). Additionally, bone mineral content was decreased (P < 0.01) by 0.01 kg. The alterations of lean mass and fat mass per kg BW loss were consistent throughout the experiment, with an average reduction of 0.29 kg in lean mass and 0.71 kg in fat mass.

Table 2.

Body composition of overweight adult female dogs fed a high-protein, high-fiber diet during weight loss

Variables	wk 0	wk 6	wk 12	wk 18	wk 24	SEM	P-values
Body composition
Total body mass, kg	14.28a	12.19b	11.86c	10.54d	9.85e	0.451	<0.0001
Total lean mass, kg	7.67a	7.42ab	7.06b	6.57c	6.40c	0.231	<0.0001
Total fat mass, kg	6.36a	5.54b	4.56c	3.74d	3.24e	0.264	<0.0001
Total bone mineral content, kg	0.24a	0.24b	0.24c	0.23d	0.23e	0.007	0.0013
Fat mass percentage, %	44.42a	41.83b	38.20c	35.22d	32.70e	1.026	<0.0001
Lean mass percentage, %	53.88e	56.37d	59.79c	62.60b	65.19a	0.989	<0.0001
Fat mass loss per kg BW	–	0.81	0.77	0.72	0.73	0.033	0.1307
Lean mass loss per kg BW	–	0.25	0.24	0.29	0.27	0.035	0.4479

Mean values within the same row with unlike superscript letters differ significantly (P < 0.05).

Voluntary physical activity

Average daily physical activity and activity during the light cycle were affected by time (Table 3). Total activity was highest at week 23, with a level that was higher (P < 0.05) than that measured at week 15, but not different than baseline or week 7. Light cycle activity was highest at week 23, with a level that was higher (P < 0.05) than levels measured at weeks 7 and 15, but not baseline. Light:dark ratio of activity counts was highest at week 23, which tended to be higher (P = 0.0514) than that measured at weeks 7 and 15, but not baseline.

Table 3.

Voluntary physical activity (counts/epoch) of overweight adult female dogs fed a high-protein, high-fiber diet during weight loss

Variables1	wk 0	wk 7	wk 15	wk 23	SEM	P-values
Daily activity	29.13ab	25.58ab	22.92b	32.21a	1.950	0.0091
12-hour of light activity	44.77ab	37.28b	34.46b	50.72a	3.144	0.0027
12-hour of dark activity	13.49	13.89	11.39	13.70	1.329	0.2842
Light-to-darkness ratio of activity counts	3.59x,y	2.94y	3.02y	4.10x	0.345	0.0514

Mean activity was represented as activity counts per epoch (epoch duration, 15 s).

Mean values within the same row with unlike superscript letters differ significantly (P < 0.05).

Mean values within the same row with unlike superscript letters tend to differ significantly (P < 0.10).

Serum metabolites, complete blood count, and blood hormones and inflammatory markers

Serum creatinine, blood urea nitrogen (BUN), total bilirubin, and chloride concentrations were increased (P < 0.01), serum triglycerides, creatine phosphokinase, calcium, alkaline phosphatase, and corticosteroid isoenzyme of alkaline phosphatase were reduced (P < 0.01), and serum cholesterol tended to be decreased (P = 0.0787) with restricted feeding and weight loss (Table 4). Other serum metabolites (total protein, albumin, globulin, albumin:globulin ratio, gamma-glutamyltransferase, phosphorus, sodium, potassium, sodium:potassium ratio) were affected by time, but were not consistently altered with restricted feeding and weight loss. Total white blood cells, neutrophils, lymphocytes, mean cell volume, and mean corpuscular hemoglobin concentrations were decreased (P < 0.01) with restricted feeding and weight loss (Table 5). All blood hormone and inflammatory cytokine concentrations decreased (P < 0.05) with restricted feeding and weight loss (Table 6). After 24 wk, serum leptin (P < 0.0001), insulin (P < 0.0001), CRP (P < 0.05), and IL-6 (P < 0.05) concentrations were decreased by 68.5%, 61.7%, 57.4%, and 37.5%, respectively.

Table 4.

Serum metabolites of overweight adult female dogs fed a high-protein, high-fiber diet during weight loss

Variables	wk 0	wk 6	wk 12	wk 18	wk 24	Reference1	SEM	P-values
Serum chemistry panel
Creatinine, mg/dL	0.51b	0.52b	0.55ab	0.57a	0.57a	0.5-1.5	0.025	0.0022
BUN2, mg/dL	11.25c	11.33c	12.17bc	13.25ab	14.33a	6-30	0.778	<0.0001
Total protein, g/dL	5.89a	5.94a	5.93a	5.62b	5.86a	5.1-7.0	0.076	<0.0001
Albumin, g/dL	3.30x	3.18z	3.29xy	3.21xyz	3.19yz	2.5-3.8	0.083	0.0577
Globulin, U/L	2.59b	2.76a	2.64b	2.41c	2.67ab	2.7-4.4	0.056	<0.0001
Albumin:globulin ratio	1.28ab	1.17c	1.24bc	1.35a	1.21bc	0.6-1.1	0.048	<0.0001
Total ALP2, U/L	46.08a	34.83b	32.58bc	27.92d	28.08cd	7-92	7.579	<0.0001
CALP2, U/L	22.33a	15.83b	15.00b	13.83b	12.25b	0-40	7.407	<0.0001
ALT2, U/L	28.50	31.00	26.25	25.00	28.83	8-65	4.219	0.4584
GGT2, U/L	2.08c	2.67bc	3.33ab	3.67a	2.58bc	0-7	0.316	<0.0001
Total bilirubin, mg/dL	0.14c	0.18b	0.19ab	0.22ab	0.23a	0.1-0.3	0.013	<0.0001
CPK2, U/L	138.42a	138.33ab	146.08a	136.92ab	119.00b	26-310	22.861	0.0086
Total cholesterol, mg/dL	189.25xy	190.17xy	191.08x	178.83xy	177.50y	129-297	15.839	0.0787
Triglycerides, mg/dL	59.58a	54.75ab	50.75ab	48.50b	46.25b	32-154	4.187	0.0011
Calcium, mg/dL	10.09a	10.10a	10.09a	9.75b	9.76b	1.6-11.4	0.067	<0.0001
Phosphorus, mg/dL	3.36x	3.38x	2.99xy	2.93y	3.07xy	2.7-5.2	0.184	0.0590
Sodium, mmol/L	146.42a	145.08b	145.25b	145.83ab	145.83ab	141-152	0.274	0.0018
Potassium, mmol/L	4.50ab	4.36bc	4.48abc	4.34c	4.53a	3.9-5.5	0.045	0.0014
Sodium:potassium ratio	32.50bc	33.42ab	32.50bc	33.83a	32.17c	28-36	0.357	0.0014
Chloride, mmol/L	109.83b	110.58ab	110.75ab	111.33a	111.67a	107-118	0.583	0.0030
Glucose, mg/dL	87.92	89.42	90.67	86.58	87.17	68-126	2.747	0.1034

University of Illinois Veterinary Diagnostic Laboratory Reference Ranges.

BUN, blood urea nitrogen; total ALP, total alkaline phosphatase; CALP, corticosteroid isoenzyme of ALP; ALT, alanine aminotransferase; GGT, gamma-glutamyltransferase; CPK, creatine phosphokinase.

Mean values within the same row with unlike superscript letters differ significantly (P < 0.05).

Mean values within the same row with unlike superscript letters tend to differ significantly (P < 0.10).

Table 5.

Complete blood cell counts of overweight adult female dogs fed a high-protein, high-fiber diet during weight loss

Variables	wk 0	wk 6	wk 12	wk 18	wk 24	Reference1	SEM	P-values
Complete blood cell counts
Total white blood cells, 106/µL	8.01a	7.12a	6.96a	5.61b	5.24b	6-17	0.586	<0.0001
Neutrophils, 103/µL	5.97a	5.23ab	4.98ab	4.11bc	3.62c	3-11.5	0.475	<0.0001
Lymphocytes, 103/µL	1.45a	1.34ab	1.16ab	1.03b	1.14ab	1-4.8	0.146	0.0229
Monocytes, 103/µL	0.42	0.34	0.32	0.30	0.32	0.2-1.4	0.045	0.2830
Eosinophils, 103/µL	0.16	0.20	0.09	0.13	0.15	0.1-1.0	0.033	0.1728
Red blood cells, 106/µL	6.37	6.45	6.69	6.42	5.98	5.5-8.5	0.289	0.2761
Hemoglobin, g/dL	14.73	14.66	15.18	14.61	13.98	12-18	0.657	0.5506
Hematocrit, %	43.78xy	44.22xy	45.67x	43.65y	44.13xy	35-52	1.797	0.0705
MCV2, fl	68.99a	68.90ab	68.64abc	68.27bc	68.07c	60-77	0.829	0.0009
MCH2, pg	23.13a	22.78b	22.71b	22.47b	22.73b	20-25	0.237	0.0001
MCHC2, g/dL	33.53a	33.03c	33.11bc	33.33abc	33.42ab	32-36	0.322	0.0002
Platelet, 103/µL	284.92	282.58	257.08	271.83	285.58	200-700	12.260	0.1489

University of Illinois Veterinary Diagnostic Laboratory Reference Ranges.

MCV, mean cell volume; MCH, mean corpuscular hemoglobin; MCHC, mean corpuscular hemoglobin concentration.

Mean values within the same row with unlike superscript letters differ significantly (P < 0.05).

Mean values within the same row with unlike superscript letters tend to differ significantly (P < 0.10).

Table 6.

Blood hormones and inflammatory markers of overweight adult female dogs fed a high-protein, high-fiber diet during weight loss

Variables	wk 0	wk 6	wk 12	wk 18	wk 24	SEM	P-values
Hormones
Leptin, ng/mL	12.60a	9.47b	6.67c	4.68cd	3.97d	1.456	<0.0001
Insulin, mU/L	19.38a	15.58ab	12.15bc	10.98c	7.43c	1.919	<0.0001
Inflammatory markers
C-reactive protein, ng/mL	3721a	1937ab	2312ab	1743ab	1587b	1243.6	0.0485
Interleukin-6, ng/mL	0.32a	0.26ab	0.25ab	0.24ab	0.20b	0.052	0.0366

Mean values within the same row with unlike superscript letters differ significantly (P < 0.05).

Fecal characteristics, fermentative metabolites, and fecal bile acids

Fecal DM percentage increased (P < 0.05), whereas fecal scores decreased (firmer stool; P < 0.001) with restricted feeding and weight loss (Table 7). Fecal acetate concentrations tended to decrease (P = 0.051) and fecal ammonia concentrations decreased (P < 0.05), whereas fecal valerate concentrations increased (P < 0.01) with restricted feeding and weight loss. Fecal pH and other fecal metabolites were not altered over time. Fecal DCA concentrations decreased (P < 0.05), fecal secondary bile acid concentrations tended to decrease (P = 0.058), and fecal UDCA concentrations increased (P < 0.01) with restricted feeding and weight loss (Table 8). The other bile acid concentrations and percentages were not altered.

Table 7.

Fecal characteristics and fecal metabolites of overweight adult female dogs fed a high-protein, high-fiber diet during weight loss

Variables	wk 0	wk 4	wk 8	wk 12	wk 16	wk 20	wk 24	SEM	P-values
Fecal characteristics
Fecal pH	5.48	5.46	5.50	5.36	5.50	5.57	5.43	0.055	0.1775
Fecal score2	2.83a	2.83a	2.33b	2.25b	2.17b	2.17b	2.00b	0.121	< 0.0001
Fecal dry matter (DM), %	23.42b	24.13ab	24.90ab	26.25a	25.43ab	25.76ab	24.65ab	0.711	0.0497
Fecal metabolites	µmol/g DM
Acetate	630.80x	589.65xy	511.27y	527.88xy	570.10xy	542.65xy	582.87xy	29.113	0.0507
Propionate	221.12	228.47	196.08	191.03	200.61	192.43	223.01	16.253	0.1634
Butyrate	74.48	76.93	80.42	82.25	91.95	90.05	87.87	5.937	0.1528
Total SCFA1	926.39	895.05	787.77	801.16	862.67	825.13	893.76	43.554	0.1523
Isobutyrate	6.27	6.82	6.12	6.52	6.46	6.44	7.06	0.699	0.8473
Isovalerate	7.54	9.08	8.07	7.95	8.20	7.75	7.76	0.984	0.7316
Valerate	0.85c	1.05bc	1.04bc	1.36abc	1.71ab	1.94a	1.53abc	0.214	0.0002
Total BCFA1	14.66	16.95	15.23	15.82	16.38	16.13	16.35	1.764	0.8367
Total VFA1	941.05	912.00	803.00	816.98	879.04	841.25	910.11	44.108	0.1600
Ammonia	147.69a	124.13ab	115.68ab	124.82ab	126.35ab	111.98b	103.39b	10.526	0.0108

Total SCFA: total short-chain fatty acids = acetate + propionate + butyrate; total BCFA: total branched-chain fatty acids = isobutyrate + isovalerate + valerate; total VFA: total volatile fatty acids = acetate + propionate + butyrate + isobutyrate + isovalerate + valerate.Valerate is technically a SCFA, but is derived from amino acid fermentation so it was included in the BCFA calculation.

Fecal score: 1 = very hard and dry, often expelled as individual pellets; 2 = firm but not hard, segmented in appearance; 3 = log-shape, moist surface; 4 = very moist, soggy, log-shaped; 5 = very moist but has a distinct shape, piles rather than distinct logs; 6 = has texture but no defined shape, present as piles or spots; 7 = watery, no texture, flat puddles (Greco, 2015).

Mean values within the same row with unlike superscript letters differ significantly (P < 0.05).

Mean values within the same row with unlike superscript letters tend to differ significantly (P < 0.10).

Table 8.

Fecal bile acids of overweight adult female dogs fed a high-protein, high-fiber diet during weight loss

Variables	wk 0	wk 4	wk 8	wk 12	wk 16	wk 20	wk 24	SEM	P-values
Bile acids, μg/mg
Cholic acid	0.08	0.09	0.06	0.05	0.06	0.06	0.07	0.021	0.9371
Chenodeoxycholic acid	0.22	0.25	0.19	0.14	0.15	0.18	0.19	0.048	0.6941
Lithocholic acid	1.32	1.46	1.33	1.40	1.32	1.28	1.24	0.109	0.7023
Deoxycholic acid	4.14a	4.27ab	3.78ab	4.03ab	3.51ab	3.03ab	2.77b	0.402	0.0141
Ursodeoxycholic acid	0.05b	0.07ab	0.08ab	0.12a	0.10a	0.10a	0.08ab	0.021	0.0032
Total primary bile acid	0.30	0.34	0.25	0.19	0.22	0.24	0.26	0.067	0.8386
Total secondary bile acid	5.51xy	5.79x	5.19xy	5.54xy	4.93xy	4.41xy	4.09y	0.498	0.0578
Total bile acids	5.81	6.13	5.44	5.74	5.15	4.64	4.35	0.529	0.1146
Secondary bile acid1, %	95.00	94.81	95.34	96.44	95.38	94.52	93.93	0.886	0.3602
Primary bile acid1, %	5.00	5.19	4.66	3.56	4.62	5.48	6.07	0.886	0.4074

Primary bile acid (sum of cholic acid and chenodeoxycholic acid) and secondary bile acid (sum of lithocholic acid, deoxycholic acid, and ursodeoxycholic acid) are expressed as a percent of total bile acid measured.

Mean values within the same row with unlike superscript letters differ significantly (P < 0.05).

Mean values within the same row with unlike superscript letters tend to differ significantly (P < 0.10).

Fecal microbiota

Alpha diversity indices, including the Shannon diversity index (Figure 2A), were not affected by restricted feeding and weight loss. Beta diversity, which is represented by PCoA plots of unweighted (Figure 2B) and weighted (Figure 2C) UniFrac distances, revealed that fecal microbial populations tended to shift away from that measured at baseline (week 0) with restricted feeding and weight loss (P < 0.05).

Figure 2.

Fecal microbial communities of overweight adult female dogs fed a high-protein, high-fiber diet during weight loss. (A) Shannon diversity index suggested that species richness was not different (P > 0.05) due to restricted feeding and weight loss. (B) Principal coordinates analysis (PCoA) plots of unweighted UniFrac distances of fecal microbial communities revealed that dogs at week 0 clustered together and separately from dogs at other time points (P < 0.05). (C) PCoA plots of weighted UniFrac distances of fecal microbial communities revealed that dogs at week 0 clustered together and separately from dogs at other time points (P < 0.05).

Four bacterial phyla and over 30 bacterial genera were altered with restricted feeding and weight loss (Supplementary Table S1; Figure 3). Relative abundances of fecal Proteobacteria, Bifidobacterium, Coriobacteriaceae UCG-002, undefined Muribaculaceae, Allobaculum, Eubacterium, Negativibacillus, Ruminococcus gauvreauii group, uncultured Erysipelotrichaceae, and Parasutterella were consistently increased (P < 0.05) with restricted feeding and weight loss. Relative abundances of fecal Prevotellaceae Ga6A1 group, Catenibacterium, Erysipelatoclostridium, Holdemanella, Lachnoclostridium, Lactobacillus, Megamonas, Peptoclostridium, Ruminococcus gnavus group, and Streptococcus were consistently decreased (P < 0.01) with restricted feeding and weight loss. Relative abundances of fecal Actinobacteria, Fusobacteria, Campilobacterota, Adlercreutzia, Slackia, Prevotella, Anaerofilum, Blautia, Clostridium sensu stricto 1, Enterococcus, Faecalibaculum, Faecalitalea, Lachnospira, Helicobacter, and Fusobacterium were different (P < 0.05) and the relative abundances of fecal Parabacteroides (P = 0.0666), Allisonella (P = 0.0982), Candidatus Stoquefichus (P = 0.0665), Epulopiscium (P = 0.0816), uncultured Ruminococcaceae (P = 0.0569), and Sutterella (P = 0.0840) tended to be different with restricted feeding and weight loss, but with an inconsistent pattern.

Figure 3.

Relative abundances of fecal microbiota of overweight adult female dogs fed a high-protein, high-fiber diet during weight loss that were increased (P < 0.05) or decreased (P < 0.05) over time. (A) Relative abundances of fecal Fusobacteria, Fusobacterium, Proteobacteria, and Actinobacteria. (B) Increased relative abundances of fecal Bifidobacterium, Prevotella, Allobaculum, and uncultured Erysipelotrichaceae. (C) Increased relative abundances of fecal undefined Muribaculaceae, Lachnospira, Clostridium sensu stricto 1, and Parasutterella. (D) Decreased relative abundances of fecal Prevotellaceae Ga6A1 group, Catenibacterium, Holdemanella, Ruminococcus gnavus group, and Streptococcus. (E) Decreased relative abundances of fecal Blautia, Lactobacillus, Megamonas, and Peptoclostridium.

Correlations between key biological parameters and gut microbiota

Several significant correlations were observed between gut microbial taxa and key biological parameters including BW, BCS, FI, CI, fecal bile acids (CA, CDCA, LCA, DCA, UDCA, primary bile acids, secondary bile acids, total bile acids), and fecal metabolites (total SCFA, acetate, propionate, butyrate, total BCFA, isobutyrate, isovalerate, valerate, ammonia) (Supplementary Figures S1 and S2). In brief, BW was positively (P < 0.05) correlated with the relative abundance of Enterococcus (r = 0.4443), but negatively (P < 0.05) correlated with the relative abundance of Romboutsia (r = −0.2322), and Sutterella (r = −0.2549). BCS was positively (P < 0.05) correlated with the relative abundance of Enterococcus (r = 0.3346), but negatively (P < 0.05) correlated with the relative abundance of Lachnospira (r = −0.2926). FI and CI were positively (P < 0.05) correlated with the relative abundance of Enterococcus (FI: r = 0.5544; CI: r = 0.5544) and Streptococcus (FI: r = 0.3086; CI: r = 0.3087), but negatively (P < 0.05) correlated with the relative abundance of Bifidobacterium (FI: r = −0.2384; CI: r = −0.2384) and Allobaculum (FI: r = −0.4275; CI: r = −0.4275).

Serum cholesterol concentrations were positively (P < 0.05) correlated with the relative abundance of Parasutterella (r = 0.6303), but negatively (P < 0.05) correlated with the relative abundance of Blautia (r = −0.4125) and Romboutsia (r = −0.2258). Serum IL-6 concentrations were positively (P < 0.05) correlated with the relative abundance of Catenibacterium (r = 0.3388) and Sutterella (r = 0.3999). A negative (P < 0.05) correlation existed between serum CRP concentrations and the relative abundance of Faecalibacterium (r = −0.0058).

Fecal total BCFA concentrations were positively (P < 0.05) correlated with the relative abundance of Proteobacteria (r = 0.4326) and Parasutterella (r = 0.6580), but negatively (P < 0.05) correlated with the relative abundance of Firmicutes (r = −0.3156) and Blautia (r = −0.4812). Fecal butyrate concentrations were positively (P < 0.05) correlated with the relative abundance of Allobaculum (r = 0.3972), whereas fecal acetate concentrations were negatively (P < 0.05) correlated with the relative abundance of Bifidobacterium (r = −0.3367). A positive (P < 0.05) correlation existed between fecal valerate concentrations and the relative abundance of Prevotella (r = 0.3296). The relative abundance of Bifidobacterium was positively correlated with fecal total bile acid (r = 0.2188) and secondary bile acid (r = 0.2420) concentrations, whereas the relative abundance of Lactobacillus was negatively (P < 0.05) correlated with fecal DCA (r = −0.2490) and secondary bile acid (r = −0.2358) concentrations.

Discussion

Obesity is a chronic progressive disease and its prevalence has been increased in humans and pets over the past few decades (AAHA 2003; Courcier et al., 2010; Rowe et al., 2017; Blüher, 2019; Lumbis and de Scally, 2020). In pets, misperceptions of ideal BW and feeding patterns, shared environmental elements, and lifestyles are important contributors (German, 2006; German, 2015; Caballero, 2019). Like humans, dogs have various similarities in regard to contributing factors of obesity and obesity-related comorbidities, including cardiorespiratory disorders, joint diseases, and gastrointestinal disorders. Moreover, obesity is an economic burden on pet owners, with approximately 20% of pet insurance claims filed being related to obesity (Bomberg et al., 2017; Pet Product News, 2021). Therefore, obesity prevention and weight management are critically important for the health and companionship of pets. In both humans and pets, various weight-loss strategies have been used, with increased physical activity and dietary modification serving as the foundation (Kushner, 2018).

Most pet diets designed for weight loss have a low energy density due to substitution of high-calorie ingredients with functional dietary fibers (Blanchard et al., 2004; German et al., 2007; Floerchinger et al., 2015; André et al., 2017; Kieler et al., 2017; Pallotto et al., 2017; Salas-Mani et al., 2018; Bermudez Sanchez et al., 2020). Such diets also have increased concentrations of high-quality proteins and micronutrients to preserve lean muscle mass and avoid nutrient deficiency during reduced intake. Increased dietary fiber and protein may also aid in mitigating hunger during weight loss (Weber et al., 2007; Bosch et al., 2009; Ben-Harchache et al., 2021). Restricted feeding of HPHF diets has successfully been used in pets (Jeusette et al., 2005; German et al., 2009; Warren et al., 2011; Bastien et al., 2015; Floerchinger et al., 2015; Salas-Mani et al., 2018; Starr et al., 2019; Bermudez Sanchez et al., 2020). However, the effects of these diets on metabolic responses, physical activity, fecal microbiota, and fecal metabolites have not been studied at the same time. These data and relationships may provide a greater understanding of the underlying mechanisms and identify new approaches to alleviate obesity and its comorbidities in dogs.

In the current study, restricted feeding and weight loss led to reduced BW, BCS, fat mass, and blood triglycerides, cholesterol, leptin, insulin, CRP, and IL-6 as expected and similar to that reported in dogs previously (German et al., 2009; Rafaj et al., 2017). Even though alpha diversity was not impacted, restricted feeding and weight loss in the current study also shifted fecal microbiota populations, with four bacterial phyla and more than 30 bacterial genera being different over time. Changes in gut microbial diversity, composition, and functionality have been linked to many disease states over the past decade, including neurological disorders, allergic diseases, gastrointestinal diseases, cardiovascular diseases, diabetes, and obesity (Mangiola et al., 2016; Patterson et al., 2016; Meng et al., 2018; Blake et al., 2019; Angelucci et al., 2019; Barcik et al., 2020; Kazemian et al., 2020; Li et al., 2021). Most pertinent here, the gut microbiota’s potential contribution to weight gain, obesity, and metabolic dysfunction has been of interest, with reduced microbial diversity and an increased Firmicutes:Bacteroidetes ratio being reported in obese vs. lean individuals (Ley et al., 2005; Turnbaugh et al., 2009; Handl et al., 2013; Park et al., 2015). Those responses are not always observed, however, including the current study and other recent publications (Schwiertz et al., 2010; Frost et al., 2019; Magne et al., 2020; Phungviwatnikul et al., 2021).

Although most microbiota-obesity knowledge comes from humans and rodent models, a few recent dog studies have focused on this topic. Bermudez Sanchez et al. (2020) reported that weight loss increased alpha diversity (richness and evenness) and the relative abundances of Bacteroidetes and Fusobacteria, but decreased the relative abundance of Firmicutes and the Firmicutes:Bacteroidetes ratio in dogs. In another study, however, the biodiversity and relative abundance of bacterial taxa at the phylum level were not different due to weight loss in dogs (Salas-Mani et al., 2018). In the current study, alpha diversity was not impacted by restricted feeding and weight loss, but many bacterial genera shifted. The increased relative abundance of Allobaculum and decreased relative abundances of Lactobacillus, Megamonas, and Catenibacterium were in agreement with previous canine weight loss studies (Salas-Mani et al., 2018; Bermudez Sanchez et al., 2020).

In most studies, including the current one, it is difficult or impossible to distinguish fecal microbiota changes due to weight loss and consequent metabolic changes from those due to restricted feeding or dietary change because reduced feeding and weight loss occur concurrently. Restricted feeding alone will reduce the substrate load reaching the large intestine, potentially affecting fecal microbial populations. Dietary fiber has a strong impact on the gut microbiota and is one of the most important dietary modifications in weight loss diets. A wide range of dietary fibers exist, with each having unique physicochemical characteristics, effects on host physiology and metabolism, and impacts on the gut microbiota populations. Some soluble fibers such as β-glucans and psyllium husk increase gut luminal viscosity and gut microbial fermentability, altering gastrointestinal transit time and enhancing microbiota-derived metabolite production (Gill et al., 2021). SCFA (i.e., acetate, propionate, and butyrate) are the primary metabolites of fermentation coming from dietary fibers and other non-digestible carbohydrates. These organic acids reduce luminal pH and serve as fuel for colonocytes, promoting gastrointestinal health, have anti-inflammatory and anti-carcinogenic effects, and assist in appetite regulation (van der Hee and Wells, 2021). The amount and type of SCFA produced depends on several factors, including bacterial taxonomic groups present, dietary ingredient type and amounts, and gut transit time (Wong et al., 2006). In the current study, total fecal SCFA concentrations were not altered by restricted feeding. This lack of change may be related to the sample analyzed (e.g., feces), as the majority of SCFA are absorbed by colonocytes leaving little for excretion in feces.

Dietary fibers not only serve as food sources for gut microbes, but also may be a predictor of the structure of the gut microbial population. Several human and animal (rodents and dogs) studies have reported that the consumption of barley β-glucans increases relative abundances of fecal Prevotella and Lactobacillus (Kovatcheva-Datchary et al., 2015; Garcia-Mazcorro et al., 2018; Sandberg et al., 2019; Phungviwatnikul et al., 2021), and is positively associated with fecal valerate concentrations (Tap et al., 2015). Furthermore, dietary barley malt melanoidins increase the relative abundances of Parasutterella, Bifidobacterium, and Lactobacillus (Aljahdali et al., 2020). Psyllium husk supplementation increased the abundance of Lachnospira, which was associated with increased fecal water content (Jalanka et al., 2019). In addition, Mayengbam et al. (2019) observed a negative correlation between Lachnospira relative abundance and BW. Others have shown that the gut microbial population is affected by prebiotic inulin-type fructans (e.g., FOS, oligofructose, and inulin) (Kelly, 2008), with increased relative abundance of Bifidobacterium being the most consistent finding (Barry et al., 2010; Vandeputte et al., 2017; Healey et al., 2018; Bastard et al., 2020). Bifidobacterium populations are thought to be important in maintaining gut health, as it has been linked with IBD remission in humans and dogs (Papa et al., 2012; White et al., 2017). Moreover, previous studies have shown that Bacteroides are more prevalent in individuals consuming high protein and fat concentrations, whereas Prevotella is more common in individuals consuming high fiber and carbohydrate concentrations (Wu et al., 2011; Kovatcheva-Datchary et al., 2015; Moreno-Pérez et al., 2018). Finally, several bacterial strains of the Clostridium, Eubacterium, Ruminococcus, and Bacteroides genera are known to have cellulose-degrading properties (Hamaker and Tuncil, 2014).

High-protein and amino acid-rich diets aid in minimizing lean mass loss and stimulating muscle protein synthesis (Wolfe, 2002). In the current study, although dogs were restricted fed, they consumed approximately 2 times the daily protein requirement (Brooks et al., 2014), had constant rates of fat and lean mass loss, and the proportions of fat mass and lean mass loss were 71% and 29%, which were similar to the findings reported by Pasiakos et al. (2013). These results suggest that this protein intake was appropriate for weight loss. High intake of protein will lead to greater amino acid catabolism, leading to higher ammonia and urea production and higher concentrations in the bloodstream and intestinal lumen. Urea and non-digestible peptides will reach the large intestine and may be catabolized by gut microbes (Shen et al., 2015). The utilization of endogenous and exogenous nitrogenous substances by gut microbiota such as bacteria of the genera Clostridium, Peptostreptococcus, Fusobacterium, Bacteroides, Veillonella, Bifidobacterium, Lactobacillus, Eubacterium, and Peptococcus affects the biosynthesis of microbial protein, fermentation products of amino acids (e.g., ammonia, SCFA, and BCFA), and amino acid homeostasis of the host (Suzuki et al., 1979; Dai et al., 2011). In the current study, serum ammonia concentrations decreased over time, being positively associated with the relative abundance of Lactobacillus and negatively associated with the relative abundance of Bifidobacterium. An enrichment of gut Bifidobacterium has been reported with a high-protein, caloric-restriction treatment in a previous study (Dong et al., 2020). The lower serum ammonia concentrations may have been due to a lower intake of protein during restricted feeding.

The impacts of different types of diets and nutrients on gut microbiota have been extensively studied, but the relationships of dietary restriction and gut microbiota have not yet been fully described, particularly in dogs. Metabolic effects observed during caloric restriction resemble those found during fasting and re-feeding cycles, wherein the body switches between energy sources (glucose to fatty acids) by increasing fatty acid catabolism from adipose tissue and stimulating downstream β-oxidation, and also increasing skeletal muscle protein breakdown to provide amino acids for gluconeogenesis (Cahill, 2006). This condition influences gut microbial activity and structure and vice versa. The decreased relative abundances of fecal Megamonas, Sutterella, and Streptococcus were consistent with findings in obese individuals after enrolling in a caloric-restricted weight loss program (Pisanu et al., 2020). The reduction in Streptococcus may be deemed as being beneficial, as it has been shown to be present at a higher relative abundance in dogs with chronic enteropathies (Suchodolski et al., 2012). Although calorie-restricted feeding had strong impacts on the gut microbiota, fiber supplementation (10 g inulin + 10 g resistant maltodextrin per day) may ameliorate those effects and increase the abundance of Bifidobacterium and Parabacteroides (Benítez-Páez et al., 2021), which were in line with the findings in the current study.

The effects of restricted feeding are not limited to the gut microbiota, but also impact the metabolism of bile acids and other metabolites. The host and gut microbiota work together to diversify the chemical composition of bile acids. The host synthesizes primary bile acids (i.e., CA and CDCA) from cholesterol in the liver, whereas the gut microbiota promotes deconjugation and biotransformation of primary bile acids to secondary bile acids (i.e., DCA, LCA, and UDCA) (Di Ciaula et al., 2017). von Schwartzenberg et al. (2021) reported that caloric restriction decreases total fecal bile acid, DCA, and LCA concentrations as a result of reduced bile secretion (due to decreased fat intake) and changes in the composition of bile acid-metabolizing taxa. The gut bacteria in the genera Clostridium, Bacteroides, Enterococcus, Bifidobacterium, and Lactobacillus produce bile salt hydrolases, which deconjugate bile acids. Other bacterial taxa, such as Eubacterium and Clostridium species, convert primary bile acids to secondary bile acids via 7α-dehydroxylation (Ridlon et al., 2016; Kriaa et al., 2019). Additionally, Ju et al. (2019) reported that Parasutterella lower fecal CA and DCA concentrations in Parasutterella-colonized mice. Similar findings were noted in the current study, with total fecal secondary bile acid and DCA concentrations being reduced and relative abundance of Parasutterella being increased following decreased food consumption. The results of the current study conflicted with others when it came to the relative abundance of Eubacterium and Clostridium sensu stricto 1, however, with these bacterial taxa being increased in the current study, but decreased in those reported by Pilla et al. (2020) and Li et al. (2021). Additionally, bile acids, particularly conjugated DCA and CDCA, exhibit bacteriostatic and bactericidal activities against lactobacilli strains (Wang et al., 2021). This fact may partially explain the reduction in Lactobacillus in the current study. UDCA not only has several therapeutic effects (e.g., anticholestatic, antiproliferative, antioxidant, and anti-inflammation), but also counteracts DCA and LCA (Winston and Theriot, 2020). The inverse concentration of fecal UDCA and DCA was also noted in the current study. As a result of caloric deprivation, intestinal motility and bile pigments elimination are diminished. The buildup of bilirubin in the intestine results in an enhanced enterohepatic circulation, which increases plasma reflux (Kotal et al., 1996). Furthermore, bilirubin accumulation is toxic to certain gut bacteria such as Streptococcus species (Chen and Yuan, 2020), which may partly explain the shifts of those bacterial taxa in the current study.

Correlation analyses identified a high number of microbe-physiological outcome associations that require further study. A total of 32 gut bacterial taxa were associated with multiple clinical parameters. In the current study, Enterococcus showed direct correlations with BW, BCS, FI, CI, and fat mass, whereas Lachnospira was inversely related to BCS. These findings are in line with that of human surgically induced weight loss patients (Sanminguel et al., 2017). Additionally, Kong et al. (2019) reported negative correlations between Allobaculum and Bifidobacterium and obesity in mice, which supports the negative correlations between those bacteria and CI in the current study. Martínez-Cuesta et al. (2021) reported that Romboutsia was diminished in obese individuals, which agreed with the negative correlation observed between those bacterial taxa and BW in the current study. Moreover, fecal Catenibacterium has been reported to be more abundant in the obese and associated with metabolic syndrome and inflammation in a previous study conducted in children (Gallard-Becerra et al., 2020), which was in line with the positive correlation between this bacterial group and serum IL-6 in the current study. Furthermore, in the current study, the relative abundance of fecal Faecalibacterium was negatively correlated with serum CRP concentrations, supporting its anti-inflammatory nature that has been reported previously (Verhoog et. al., 2019). Chen et al. (2021) reported that Sutterella had a positive correlation with cyclooxygenase-2, which may induce the expression and secretion of IL-6 (Hinson et al., 1996), supporting the positive correlation between Sutterella and IL-6 in the current study.

Several correlations between the gut bacterial taxa and fermentation products were noted, with some relationships likely reflecting the cross-feeding that occurs in the large intestine. Bifidobacterium degrades undigestible carbohydrates and yields acetate, lactate, succinate, and BCFA which may be utilized by butyrate-producing bacteria such as Allobaculum to produce butyrate (Teixeira et al., 2018; Fu et al., 2019). These relationships between Bifidobacterium, Allobaculum, acetate, and butyrate were observed in the current study. Additionally, a positive relationship between Prevotella and fecal valerate was consistent with the results reported by Tap et al. (2015). Negative correlations existed between Lactobacillus, fecal secondary bile acids, and DCA, which may reflect the antimicrobial property of DCA on Lactobacillus (Wang et al., 2021). On the contrary, Bifidobacterium was positively correlated with fecal total bile acid and secondary bile acids, which was similar to the findings reported by Wan et al. (2020) and supporting its role in bile acid conversion (Ridlon et al., 2005).

Although some correlations agreed with the literature, several correlations were inconsistent with previous reports. Hou et al. (2017), for instance, reported that Sutterella was enriched in obese children, but the opposite was observed in the current study. Similarly, although a negative relationship was observed between Parasutterella and serum cholesterol concentrations, positive relationships existed between Blautia, BW, and serum cholesterol, and a positive relationship between Romboutsia and serum cholesterol was reported in obese adult humans (Zeng et al., 2019), the opposite was observed in this study. The reasons for these discrepancies may be due to host species differences, the dietary interventions of each study, design or length of study, or other unknown variables.

Overall, the results in this experiment suggest that obesity is a complex disease that not only affects a multitude of body systems, but also impacts the structure and activity of gut microbiota. Weight loss and restricted feeding with a high-protein, high-fiber diet induce physiological adaptation and modify the gut microbial composition, subsequently improving the biological parameters of overweight dogs. The effects of weight loss and restricted feeding are inseparable in this study, however, so that must be acknowledged. Caloric restriction-induced BW and fat mass loss decreased circulating triglyceride, leptin, insulin, CRP, and IL-6 concentrations, diminished fecal ammonia and secondary bile acids, increased the relative abundances of fecal Bifidobacterium, Allobaculum, Eubacterium, Lachnospira, Ruminococcus gauvreauii group, and Parasutterella, and decreased the relative abundances of fecal Catenibacterium, Lactobacillus, Megamonas, Ruminococcus gnavus group, and Streptococcus. Several strong correlations between physiological variables and gut microbiota were observed and demonstrate that certain bacterial taxa may exacerbate the progression of obesity and comorbidities and vice versa. Bacteria with strong correlations with host metabolism should be studied further, as they may serve as biomarkers in the future. Further research may also identify the impacts of weight reduction independently from restricted feeding, in particular on gut microbiota, and further clarify the independent effects between dietary regimen and weight loss.

Glossary

Abbreviations

AAFCO

Association of American Feed Control Officials

AAHA

American Animal Hospital Association

BCS

body condition score

BUN

blood urea nitrogen

BW

body weight

CI

caloric intake

CRP

C-reactive protein

DEXA

dual-energy x-ray absorptiometry

DHA

docosahexaenoic acid

DM

dry matter

FI

food intake

IL-6

interleukin-6

ME

metabolizable energy

MER

maintenance energy requirement

NRC

National Research Council

OM

organic matter

TDF

total dietary fiber

Conflict of Interest Statement

T.P. is employed by Perfect Companion Group Co., Ltd. All other authors have no conflicts of interest.