Introduction

Type 2 diabetes mellitus (T2DM) is a chronic metabolic disorder characterized by hyperglycemia resulting from insulin resistance and impaired insulin secretion1. Recent data from the National Diabetes Statistics Report indicate that 37.3 million Americans suffer from T2DM2. In addition, the economic cost of diabetes and prediabetes was estimated to reach $322 billion in the U.S in 2012. Accumulating data from several large, placebo-controlled studies suggests that sodium-glucose transporter 2 (SGLT-2) inhibitors and glucagon-like peptide 1 (GLP-1) receptor agonists offer therapeutic benefits in the management of cardiovascular diseases, regardless of the patient's diabetic status3,4,5,6,7,8,9,10,11. In addition to their effects on glucose excretion, SGLT-2-inhibitors positively impact systemic metabolism by reducing inflammation and oxidative stress, shifting metabolism towards ketone body production, promoting autophagy, and suppressing glycation end-product signaling9. These findings align with the American Heart Association's recent definition of cardiovascular-kidney-metabolic health12, a concept which represents the interplay between metabolic disorders like T2DM, chronic kidney disease, and the cardiovascular system (Fig. 1).

Fig. 1
figure 1

Molecular bases for the interrelationship between cardiovascular, renal, and metabolic diseases. Adjusted and simplified from Kadowaki et al. 120.

Full size image

The pleiotropic effects of SGLT-2 inhibitors and GLP-1 agonists provide an opportunity to target several cardiorenal metabolic disorders. This can be achieved experimentally using a disease model that replicates key features of metabolic syndrome (MetS), a cluster of risk factors that include obesity, dyslipidemia, hypertension, and insulin resistance. Collectively, these factors increase the risk of developing cardiorenal diseases, metabolic dysfunction-associated steatohepatitis and T2DM9,13,14. Implementing such a model would enable future mechanistic studies to explore the metabolic effects of novel antidiabetic therapy beyond glycemic control.

Rodent models, especially mice, are commonly used in translational research due to their cost-effectiveness, ethical considerations, and ease of genetic manipulation. However, despite their widespread use, murine models often fail to accurately mimic human diseases, limiting their translational value15. In a study published in the Proceedings of the National Academy of Sciences, Seok et al.16 found that mouse models poorly replicate human inflammatory responses from various causes, reporting a poor correlation (R2 between 0.0 and 0.1) between significantly altered genes in humans and their murine equivalents. Additionally, the high failure rate of drug discovery and development highlights the urgent need for more accurate animal models in preclinical research17. The FDA Modernization Act 2.0 further emphasizes this need by restricting the use of non-informative animal models, such as rodents, in experimental drug testing prior to clinical studies18. As a result, there is an increasing incentive to use more relevant animal models, such as dogs, which often better represent chronic diseases than rodents. Dogs have a larger body size, longer lifespan, and physiological similarities to humans, and they naturally develop diseases analogous to human conditions, including cardiorenal, metabolic, inflammatory, and neurodegenerative diseases15,19,20,21,22,23,24.

Despite presenting some unique features in their lipid metabolism, such as low activity of cholesteryl ester transfer protein and an abundance of high-density lipoprotein (HDL)-bound plasma cholesterol (contributing to an athero-protective profile)25,124, dogs have been extensively used for studying metabolic obesity, diabetes, dyslipidemia, and their response to therapeutic interventions26. As such, a comprehensive analysis published in the Journal of Lipid Research identified dogs as the second most translatable model (after non-human primates) for studying dyslipidemia and related pharmacological agents in humans, among 24 animal species, including rodents, rabbits and pigs27.

In return, from the perspective of developing veterinary pharmaceuticals for spontaneous diseases in animals, the data generated in dogs from this study could guide the development of multiple therapeutic applications for SGLT-2 inhibitors and GLP-1 agonists in veterinary medicine, such as diabetes, congestive heart failure, or chronic kidney disease under the One Health paradigm28,29,30,31.

Previous studies have shown that consistent overfeeding of dogs with a high-calorie Western diet (WD) can lead to MetS, regardless of the diet’s composition32,33. However, these investigations in dogs have primarily focused on the metabolic issues associated with obesity32,34,35,36,37. In both dogs and humans, obesity triggers an adipose tissue redistribution that results in an increase in visceral (vs. subcutaneous) fat. This shift in fat deposition is independently associated with the onset of MetS. A study by Romero-Corral et al.38 analyzing data from more than 6000 subjects showed that normal weight obesity, defined as the combination of normal body mass index (BMI) and high body fat content, was associated with a high prevalence of cardiometabolic dysregulation and cardiovascular risk factors. These findings were later supported by Shi et al.39 in an analysis including 12,047 adults where normal-weight adults with MetS had the highest hazard ratio for mortality (1.70 [1.16–2.51]), compared with obese patients with MetS (1.30 [1.07–1.60]) and obese patients without MetS (1.08 [0.76–1.54]). Additionally, most clinical studies demonstrating therapeutic benefits from SGLT-2 inhibitors on cardiorenal outcome measures include a majority of non-obese patients4,7,40,41,42. A recent study by Adamson et al., published in the European Journal of Heart Failure, has unequivocally established that the efficacy of dapagliflozin for heart failure patients with reduced ejection fraction remains consistent regardless of their BMI43. These findings collectively support the rationale for establishing a metabolic dysfunction model that is independent of obesity to study the pharmacodynamic effects of drug candidates targeting the cardiovascular-kidney-metabolic health.

In a preliminary study conducted by our consortium, dogs isocalorically fed a WD for a few weeks presented with elevated fasting bile acids, cholesterol, triglycerides, and blood pressure compared to control44. These findings suggest that short-term feeding with a WD can induce a clinical response that mimics MetS in healthy dogs. The aim of this research was to characterize the metabolic and molecular signatures associated with a high-fat, high-monosaccharide, and low-fiber isocaloric WD after ten weeks in dogs. Once established, this preclinical model can be used to assess the therapeutic benefits of novel antidiabetic therapy in the context of obesity-independent MetS and pave the way for translational studies that could benefit both human and veterinary medicine.

Methods

Animals

Experimental procedures were approved by the Institutional Animal Care and Use Committee at Iowa State University (Protocol Number: 21-164). All methods were performed in accordance with the relevant guidelines and regulations at Iowa State University. The authors complied with ARRIVE guidelines in the completion of this study. The study population consisted of 18 neutered young adult laboratory Beagles (9 males and 9 females, aged 12–15 months) weighing between 7.5 and 11.5 kg. They originated from a colony that had been previously acclimatized to handling and blood sampling procedures. Prior to inclusion, each dog was assessed for its general health and condition with a physical examination and had received appropriate vaccinations and deworming treatments. Normal cardiovascular structure and function were confirmed through an echocardiogram performed prior to acclimation to the study facility. No dog had received topical or systemic medications within the 28 days preceding inclusion. Each animal was assigned a unique 4-letter ear tattoo for identification purposes. Throughout the in-life phase, daily evaluations of the animals’ general health and behavior were conducted by the study veterinarian (Dr. Agnes Bourgois-Mochel). Body weight and body condition scores (BCS, recorded as “underweight” vs. “ideal” vs. “overweight”) were recorded weekly using the body condition system developed at the Nestle Purina Pet Care Center45. Detailed information on the canine body condition system can be found at: https://www.purinainstitute.com/centresquare/nutritional-and-clinical-assessments/canine-body-condition-system-sheet. All observations, including any adverse events and study interventions were systematically recorded in the raw data file.

Housing

The study animals were acclimatized to the facility for one week before the start of the study. Housing conditions were strictly in accordance with the requirements set by the United States Department of Agriculture. Each dog was housed in a 16-square foot kennel (dimensions: 4ʹ × 4ʹ) with an interconnecting door, allowing for the co-housing of two animals. However, individual separation was implemented during specific periods, such as feeding times, or when necessary for specific interventions or observations.

The lighting schedule was kept from 6 a.m. to 6 p.m. The ambient temperature within the housing facilities was consistently set to 70 F (21.1 °C), with continuous monitoring. Throughout the study, the recorded temperature varied minimally, with the range extending from 67 F (19.4 °C) to 72 F (22.2 °C). Relative humidity was also closely monitored, with values fluctuating between 34 and 45%. The dogs were provided with unrestricted access to tap water, delivered via individual nipple water feeders.

Experimental design

To replicate the dietary intake of an average American diet, dogs were fed a high-fat, high-monosaccharide, low-fiber WD adjusted from parameters of the National Health and Nutrition Examination Survey (NHANES 2015–2016: Males and Females over 20 years)46 for ten weeks. Dogs were fed isocalorically based on individually calculated metabolizable energy needs. Blood samples were collected at baseline (BAS1) when dogs were fed their regular diet, and then again after ten weeks of WD feeding (BAS2).

Diet composition for BAS1 measurements

Dogs were fed a daily diet of Royal Canin® Beagle Adult dry food (12% fat content), once in the morning, around 9 a.m. The portion size for each dog was individually calculated based on weight and resting energy requirements (RER) calculated by multiplying the animal’s body weight (BW) in kilograms raised to the 0.75 power by 70: RER = 70 × (BW)0.75 (Eq 0.1). Any leftover was weighed and recorded in the raw data file. Nutritional information on the Royal Canin® Beagle Adult dry food is presented in Table 1.

Table 1 Nutritional information of the Royal Canin® Beagle adult diet.
Full size table

Diet composition for BAS2 measurements

Western diets were formulated to model the average intake of American subjects over 20 years from the NHANES and were fed to meet the energy and National Research Council47 requirements for adult dogs. Diets were home cooked and prepared in batch (N = 18 dogs) on a weekly basis, then stored frozen. Ingredients were purchased from the same supplier and brand to ensure consistency. Food was offered once daily in the morning, around 9 a.m. In cases where the provided meal was not entirely consumed, the remaining portion was carefully weighed and documented in the raw data file. The exact composition of the WD can be found in Table 2.

Table 2 Nutritional characteristics (2A) and composition (2B) of the western diet.
Full size table

Sample collection

  • Blood samples were drawn using a jugular catheter whenever possible, or from the saphenous or cephalic veins with single use needles. Blood samples were collected at baseline (BAS1) when dogs were fed their regular diet, and then again after ten weeks of WD feeding (BAS2) for measurement of:

    1. o

      Complete blood count (CBC) (plasma, K3 EDTA, Iowa State University College of Veterinary Medicine);

    2. o

      Standard chemistry panel, including alanine aminotransferase (ALT), alkaline phosphatase (ALP), albumin, total protein, triglycerides, total cholesterol, blood urea nitrogen (BUN), serum creatinine, serum bicarbonates, calcium, phosphorus, chloride, sodium and potassium (serum, plain tube, Iowa State University College of Veterinary Medicine);

    3. o

      Fasting blood glucose (serum, plain tube, Iowa State University College of Veterinary Medicine);

    4. o

      Glucagon (serum, plain tube, Cornell University College of Veterinary Medicine. EMD Millipore’s Glucagon Radioimmunoassay (RIA) Kit GL-32 K);

    5. o

      Lipid profiling: HDL and Low-Density Lipoprotein (LDL) cholesterol (serum, plain tube, Texas A&M College of Veterinary Medicine);

    6. o

      Renin-angiotensin aldosterone system (RAAS) biomarkers (serum, plain tube, Attoquant Diagnostics, Vienna);

    7. o

      N-terminal prohormone of brain natriuretic peptide (NT-proBNP) (plasma, K3 EDTA, IDEXX Laboratories, Maine. Cardiopet ProBNP Test-Canine);

    8. o

      Oxidative stress biomarkers (serum, plain tube, University of Murcia Facultad de Veterinaria);

    9. o

      Metabolomics, including (1) General Metabolism; (2) Complex Lipids and (3) Biogenic Amines (serum, plain tube, University of California Davis Genome Center).

  • Serum insulin concentrations were measured on a separate day through an oral glucose tolerance test (plain tube, Michigan State University Veterinary Diagnostic Laboratory. EMD Millipore’s Insulin Radioimmunoassay (RIA) Kit HI-14K). The dogs received an oral dose of 5 g of dextrose per kg in the form of a solution containing 1 g of dextrose powder per mL of water. Blood samples were collected at 30, 60, 90, 120, and 180 min after oral administration of the glucose solution.

  • Voided urine and fecal samples were collected at BAS1 and BAS2 for the purpose of conducting metabolomic analyses, including (1) General Metabolism; (2) Complex Lipids and (3) Biogenic Amines (University of California Davis Genome Center).

  • Serum, plasma, urine, and fecal samples were divided into separate aliquots and stored at − 80 °C prior to analysis.

  • BP was measured at BAS1 and BAS2 by a certified cardiologist using a Doppler device, following standard procedures from the American College of Veterinary Internal Medicine (ACVIM), as outlined in consensus panel guidelines48. As Doppler-derived single measurements of blood pressure are an estimate of systolic blood pressure (SBP)125, the abbreviation SBP will be used throughout this manuscript. To avoid any potential disruptions or bias in the recordings, these measurements were consistently taken before any blood was collected during each study period. To follow the consensus panel guidelines for assessing hypertension48 and ensure accuracy, five consecutive and consistent SBP measurements were obtained from each subject. These values were then averaged to calculate an individual estimate of SBP.

Specific analytical methods

Lipoprotein profiling

Lipoprotein profiling was carried out using the continuous lipoprotein density profiling (CLPDP) method, adhering to procedures detailed in prior literature49,50. Briefly, a solution of 0.18 M NaBiEDTA (Tokyo Chemical Industry) measuring 1280 µL was combined with 10 µL of both serum and NBD C6-ceramide (Cayman Chemical Company). Subsequently, 1150 µL of the resultant blend was allocated to a polycarbonate centrifuge container (Beckman Coulter). The samples underwent centrifugation for 6 h at 4 °C and 867,747 g using an Optima MAX-LP ultracentrifuge (Beckman Coulter) equipped with a fixed-angle rotor (MLA-130; Beckman Coulter). Immediately post-centrifugation, the samples were imaged using a fluorescence imaging system comprising a digital camera (Quantifire XI; Optronics) and a constant metal halide light source (Dolan-Jenner Industries).

The images obtained were transformed into density profiles via software analysis (OriginPro7.5; OriginLab). Lipoprotein profiles were produced by plotting the average intensity of fluorescence on the y-axis, while the actual centrifuge tube coordinates (mm) served as the x-axis. A unique numbering system was established for the statistical examination. The area under the curve (AUC) of the total fluorescence trace and each segment were used to determine the total lipoprotein intensity and fractional intensities, respectively. AUCs were then calculated for LDLs and HDLs based on their density intervals. Individual AUC values were finally normalized using the total AUC and expressed as percentage, as presented by Minamoto et al.50.

RAAS fingerprinting

Determination of angiotensin and aldosterone analytes from canine serum was derived as previously published by our consortium51,52,53,54. Briefly, serum samples were analyzed to determine the equilibrium concentrations of Angiotensin I (Ang I (1–10)), Angiotensin II (Ang II (1–8)), Angiotensin III (Ang III (2–8)), Angiotensin IV (Ang IV (3–8)), Angiotensin 1–7 (Ang1–7), Angiotensin 1–5 (Ang1–5), and aldosterone using validated Liquid Chromatography-Tandem Mass Spectrometry (LC–MS/MS) assays at a commercial laboratory (Attoquant Diagnostics, Vienna, Austria)55. Following ex vivo equilibration, each sample was spiked with a stable isotope-labeled internal standard for each angiotensin peptide and a deuterated internal standard for aldosterone (aldosterone D4). The analytes were then extracted using C18-based solid-phase extraction. The extracted samples underwent mass spectrometry analysis using a reversed-analytical column, which was operated in tandem with a XEVO TQ-S triple quadrupole mass spectrometer in multiple reaction monitoring mode. Internal standards were used to ensure analyte recovery throughout the sample preparation process for each sample. The concentrations of the analytes were calculated from the chromatograms where the integrated signals exceeded a signal-to-noise ratio of 10, considering the corresponding response factors derived from appropriate calibration curves in the serum matrix. The lower limit of quantification (LLOQ) was established at 3.0 pM, 2.0 pM, 3.0 pM, 2.0 pM, 2.5 pM, 2.0 pM and 13.9 pM for Ang I (1–10)), Ang II (1–8), Ang1–7, Ang1–5, Ang III (2–8), Ang IV (3–8) and aldosterone, respectively. Markers for renin (PRA–S) and angiotensin-converting enzyme (ACE–S) based on angiotensin were obtained from Ang II (1–8) and Ang I (1–10) concentrations by calculating their sum and ratio, respectively56. Renin-independent alternative RAAS activation (ALT–S) was calculated using the formula [(Ang 1–7 + Ang 1–5) / (Ang I + Ang II + Ang 1–7 + Ang 1–5)]57.

Oxidative stress markers

The development and validation of analytical techniques for assessing oxidative stress markers adhered to protocols outlined in previous studies58. The following provides an abridged overview of the specific procedures used in evaluating antioxidant and oxidant statuses.

Antioxidant status

  • The Cupric Reducing Antioxidant Capacity (CUPRAC) assay, initially described by Campos et al.59, is based on the conversion of Cu2+ to Cu+ through the action of non-enzymatic antioxidants in the serum sample. Quantification of CUPRAC followed the protocol previously validated for canine serum60, with results reported in mmol/L.

  • The Ferric Reducing Ability of Plasma (FRAP) assay relies on the conversion of ferric-tripyridyltriazine (Fe3+-TPTZ) to its ferrous form61. Quantification of FRAP was performed as described in previous studies60,61. Results are expressed in mmol/L.

  • Measurements of Trolox Equivalent Antioxidant Capacity (TEAC) followed the procedures outlined by Arnao et al.62 later adapted to canine serum samples by Rubio et al.60. The assay involves the generation of ABTS radicals and their subsequent reduction by non-enzymatic antioxidants in the serum specimen62, with results presented in mmol/L.

  • Total thiol (µmol/L) determination was based on the reaction between sample thiols and DTNB126,63.

  • The evaluation of Paraoxonase type 1 (PON-1) was based on the conversion of phenylacetate to phenol, following the same methods used for canine serum by Tvarijonaviciute et al.64. Results are expressed as IU/mL.

  • Quantification of Glutathione Peroxidase (GPx) activity was performed using a commercial assay kit according to the manufacturer's instructions (RANDOX Glutathione Peroxidase (Ransel) Kit RS504), as described in previous studies65,66. Results are reported in IU/ml units.

Oxidant status

  • The Total Oxidant Status (TOS) was determined following Erel’s method67, which had previously been applied to dog serum60. Results are expressed in µmol/L.

  • The Peroxide-Activity (POX-Act) assay involved the detection of total peroxides through a peroxide-peroxidase reaction using tetramethylbenzidine as the chromogenic substrate68. Results are expressed in µmol/L.

  • The Derivatives-Reactive Oxygen Metabolites (d-ROMs) assay used an acidic medium to react with the sample in the presence of DEPPD, as per the method previously established by Alberti et al.69. Results are reported in Carratelli Units (U.CARR).

  • Determination of Advanced Oxidation Protein Products (AOPP) was based on oxidized albumin and di-tyrosine containing cross-linked proteins, as described in previous studies70,71. Results are expressed in µmol/L.

Serum/urine/fecal metabolomics

General metabolism

Samples were extracted using the extraction procedure by Matyash et al.72, which includes MTBE, MeOH, and H2O. The organic (upper) phase was dried down and submitted for resuspension and injection onto the LC, while the aqueous (bottom) phase was dried down and submitted for derivatization for GC. Samples were shaken at 30 °C for 1.5 h. Then, 91 μL of MSTFA + FAMEs were added to each sample, and tubes were shaken at 37 °C for 0.5 h to complete the derivatization. Samples were then vialed, capped, and injected onto the instrument. A 7890A GC coupled with a LECO time of flight mass spectrometer (TOFMS) was used for the procedure. Then, 0.5 μL of the derivatized sample was injected using a splitless method onto a RESTEK RTX-5SIL MS column with an Intergra-Guard at 275 °C with a helium flow of 1 mL/min. The GC oven was set to hold at 50 °C for 1 min, then ramped up to 20 °C/min to 330 °C and held for 5 min. The transfer line was set to 280 °C, while the EI ion source was set to 250 °C. The mass spectrometry parameters collected data from 85 to 500 m/z at an acquisition rate of 17 spectra/second. All compounds detected were tentatively identified to the Metabolomics Standards Initiative (MSI) Level 2 with a spectral library match score of 800 or higher73.

Complex lipids

Samples were extracted using the extraction procedure by Matyash et al.72, which includes MTBE, MeOH, and H2O. The organic (upper) phase was dried down and resuspended for injection onto the LC, while the aqueous (bottom) phase was dried down and submitted for derivatization for GC. The samples were then resuspended with 110 μL of a solution of 9:1 methanol:toluene and 50 ng/mL CUDA. Samples were then shaken for 20 s, sonicated for 5 min at room temperature, and centrifuged for 2 min at 16100 rcf. Thirty-three μL of samples were aliquoted into a vial with a 50 μL glass insert for positive and negative mode lipidomics. The samples were then loaded onto an Agilent 1290 Infinity LC stack. The positive mode was run on an Agilent 6546 with a scan range of m/z 120–1200 Da and an acquisition speed of 2 spectra/s. Positive mode had between 0.5 and 2 μL injected onto an Acquity Premier BEH C18 1.7 μm, 2.1 × 50 mm column. The gradient used was 0 min 15% (B), 0.75 min 30% (B), 0.98 min 48% (B), 4.00 min 82% (B), 4.13–4.50 min 99% (B), 4.58–5.50 min 15% (B) with a flow rate of 0.8 mL/min. Another aliquot was run in negative mode on an Agilent 1290 Infinity LC stack and injected onto the same column, with the same gradient, using an Agilent 6550 QTOF mass spectrometer. The acquisition rate was two spectra per second with a scan range of m/z 60–1200 Da. The mass resolution for the Agilent 6530 is 10,000 for ESI ( +) and 20,000 for ESI (-) for the Agilent 6550.

Biogenic amines

Sample extraction for biogenic amines consisted of a liquid–liquid extraction method72 with MTBE, methanol, and water, creating a biphasic partition. The polar phase was then dried down to completion and run on a Waters Premier Acquity BEH Amide column. A short 4-min liquid chromatography method was used for the separation of polar metabolites from a starting condition of 100% LCMS H2O with 10 mM ammonium formate and 0.125% formic acid to an end condition of 100% ACN:H2O 95:5 (v/v) with 10 mM ammonium formate and 0.125% formic acid. A Sciex Triple-ToF scanned from 50 to 1500 m/z with MS/MS collection from 40 to 1000, selecting the top five ions per cycle. Data processing was done with MS-Dial using an MZ-RT list for annotations, in addition to a library for MS/MS matching.

Statistics

The sample size for this experiment was established based on preliminary data from a previous study conducted by our group44. In that study, statistically significant differences in BP and total cholesterol were observed in a group of ten dogs receiving an isocaloric WD, with an alpha level of 0.05 and a statistical power of 80%. Study variables were visually inspected for normality, summarized, and displayed as median (interquartile range [IQR]). The area under the curve for insulin time-concentrations in the serum was determined by using the linear trapezoidal method from 0 to 90 min (AUCins(0–90)). Differences between BAS1 and BAS2 were analyzed using non-parametric Wilcoxon signed rank test with continuity correction. P-values < 0.05 were considered statistically significant. The R software version 4.2.2 was used for statistical analyses. (R Core Team (2022). Graphical representation of the data was produced using the ggplot2 package in R version 4.2.2. For metabolomic analyses, the peak tables were uploaded into the Matlab® (R2023b, The Mathworks Inc., Natick, MA) environment and converted to datasets with appropriate class labels. In PLS_Toolbox (Version 9.0; Eigenvector Research, Manson, WA), a Principal Component Analysis (PCA) was performed on the autoscaled data. To determine the variables responsible for differences between groups, a Cluster Resolution Feature Selection (FS-CR) was applied74,75. For each dataset, the FS-CR process of sequential backward elimination and forward selection was repeated 100 times, permuting the subsets of data, and only variables selected 85% of the time were retained to prevent overfitting75. The distance between clusters (cluster resolution) was used to determine which variables contributed to the separation between classes75. A PCA was then performed using the selected variables from the FS-CR, and the variables and their loadings were extracted. PCA score plots were generated to visualize the relationships between samples and classes, with a 95% confidence interval plotted around the mean center of the data.

Results

Physical examination and adverse events

The study veterinarian, along with approved study personnel, conducted weekly physical examinations and reported no notable changes in the dogs’ overall condition, behavior, cardiovascular system, hydration level, respiratory system, or skin appearance throughout the study.

During the transition from their regular diet (Royal Canin® Beagle Adult) in the first baseline phase (BAS1) to the Western diet (BAS2), several dogs experienced one or more episodes of softened stools. These instances were considered “on-serious” digestive adverse events by the study veterinarian and resolved on their own within a few days. No significant adverse effects were reported over the duration of the study.

Body weight

Differences in body weight between diets (− 3.6% based on median relative change from BAS1) were statistically significant but were not considered clinically meaningful by the study veterinarian (BAS1 8.9 [7.8–9.6] kg) vs. BAS2 8.7 [7.4–9.2] kg). Similarly, no notable changes in body condition scores were reported between BAS1 (N = 0, 13 and 5 for “underweight”, “ideal” and “overweight”, respectively) and BAS2 (N = 1, 11 and 6 for “underweight”, “ideal” and “overweight”, respectively).

Complete blood count and chemistry

All hematological parameters were within normal physiological limits, and there were no clinically relevant changes in CBC between BAS1 and BAS2.

No significant changes in liver-related chemical parameters, including ALT, ALP, albumin, and total protein, were observed between BAS1 and BAS2. However, dogs fed a WD for ten weeks had a decrease in serum bicarbonate (− 2.5 [− 4.0 to − 1.0] mEq/L, P < 0.001), phosphorus (− 0.8 [− 1.3 to − 0.5] mg/dL, P < 0.001), and potassium (− 0.5 [− 0.7 to − 0.3] mEq/L, P < 0.001), and an increase in chloride concentrations (+ 1.5 [0.0–3.0] mEq/L, P = 0.001). The diet also induced some borderline statistically significant changes in calcium (P = 0.049) and sodium (P = 0.041) levels at BAS2. Additionally, there was a significant decrease in BUN at BAS2 (− 4.5 [− 5.0 to − 3.0] mg/dL, P < 0.001), along with an increase in serum creatinine (+ 0.1 [0.0–0.2] mg/dL, P = 0.001). These variations, although statistically significant, remained within physiological limits. A summary of the clinical chemistry parameters at BAS1 and BAS2 is presented in Fig. 2.

Fig. 2
figure 2

Temporal changes in standard clinical chemistry parameters after ten weeks of feeding with a high-fat, high-monosaccharide, low-fiber western diet. No notable alterations were observed in liver-related chemical parameters, such as ALT, ALP, albumin, and total protein, when comparing BAS1 to BAS2. Dogs at BAS2 had decreased levels of serum bicarbonates, phosphorus, and potassium, but increased levels of chloride. There was also a reduction in BUN at BAS2, along with an elevation in serum creatinine levels. Box plots represent the 25th, 50th and 75th percentile of the data ± 1.5 IQR (interquartile range). •: 0.01 < P ≤ 0.05; ••: 0.001 < P ≤ 0.01; •••: P ≤ 0.001. Summary box plots produced using the ggplot2 package in R version 4.2.2.

Full size image

Fasting blood glucose, serum glucagon and insulin

The biological effects of the WD on fasting blood glucose, as well as the glucose-regulating hormones glucagon and insulin, are presented in Figs. 3 and 4. Over a span of ten weeks, the WD induced a significant increase in fasting blood glucose concentrations (Fig. 3). This increase approached the upper physiological limit, demonstrating an overall increase of 15.8% relative to baseline (BAS1 88.0 [82.0–91.0] mg/dL vs. BAS2 102.5 [95.0–109.0] mg/dL, P < 0.001). The increase in fasting blood glucose was accompanied by a non-significant trend suggesting a decrease in serum glucagon levels at BAS2 (BAS1 69.3 [64.0–77.2] pg/mL vs. BAS2 61.8 [49.8–64.3] pg/mL, P = 0.055).

Fig. 3
figure 3

Temporal changes in fasting blood glucose and glucagon after ten weeks of feeding with a high-fat, high-monosaccharide, low-fiber western diet. The WD resulted in a significant 16.5% increase in fasting blood glucose, approaching the upper physiological limit. This was accompanied by a trend towards lower serum glucagon levels which did not reach statistical significance. Box plots represent the 25th, 50th and 75th percentile of the data ± 1.5 IQR (interquartile range). •••: P ≤ 0.001. Summary box plots produced using the ggplot2 package in R version 4.2.2.

Full size image
Fig. 4
figure 4

Serum insulin time-course post-glucose tolerance test. Dogs received an oral dose of 5 g of dextrose per kg in the form of a solution containing 1 g of dextrose powder per mL of water. Blood samples were collected at 30, 60, 90, 120, and 180 min after oral administration of the glucose solution. Compared to BAS1, dogs fed an isocaloric WD for ten weeks showed a trend towards a higher peak concentration of insulin, with an increase of approximately 35% in AUCins(0–90) at BAS2. Time-course data are presented as mean ± S.E. Scatter plots produced using the ggplot2 package in R version 4.2.2.

Full size image

Additionally, the results of the glucose tolerance test suggest a trend towards an elevation in peak insulin concentrations in dogs fed a WD. The median AUCins(0–90) showed a 34.9% increase from 202.7 [103.1–276.8] uIU/mL x min at BAS1 to 232.7 [144.2–276.2] uIU/mL x min at BAS2 (P = 0.39). The average (arithmetic mean) for this increase at BAS2 was 58.2% (Fig. 4).

Blood pressure

Overall, SBP measurements were significantly higher at BAS2 compared with pre-WD readings (BAS1 133.5 [126.0–141.0] mmHg vs. BAS2 143.0 [133.0–152.0] mmHg, P = 0.017) (Fig. 5A).

Fig. 5
figure 5

Temporal changes in systolic blood pressure (A) and NT-proBNP (B) after ten weeks of feeding with a high-fat, high-monosaccharide, low-fiber western diet. (A) Dogs fed a WD for ten weeks had significantly higher blood pressure measurements compared with baseline (BAS1). Measures were taken by a certified cardiologist using a Doppler device. To avoid bias in the recordings, these measurements were consistently taken before any blood was collected during each study period. To follow the ACVIM consensus panel guidelines for assessing hypertension48 and ensure accuracy, five consecutive and consistent SBP measurements were obtained from each subject. These values were then averaged to calculate an individual estimate of SBP. (B) NT-proBNP concentrations significantly increased at BAS2, with two dogs presenting values exceeding 900 pmol/L, a level commonly associated with structural heart disease in canines94,95. Box plots represent the 25th, 50th and 75th percentile of the data ± 1.5 IQR (interquartile range). •: 0.01 < P ≤ 0.05; •••: P ≤ 0.001. Summary box plots produced using the ggplot2 package in R version 4.2.2.

Full size image

NT-proBNP

NT-proBNP levels significantly increased at BAS2, as shown by the change from baseline (BAS1 250.0 [250.0–401.0] pmol/L) to post-WD (BAS2 460.5 [330.0–750.0] pmol/L) (P < 0.001) (Fig. 5B). Notably, two dogs exhibited NT-proBNP concentrations exceeding 900 pmol/L.

Renin-angiotensin aldosterone system (RAAS)

Our analysis revealed a slight downward trend in biomarkers in both the traditional and alternative arms of the RAAS, though this trend was not statistically significant. This included reductions in plasma renin activity (PRA–S), Angiotensin I (Ang I (1–10)), Angiotensin II (Ang II (1–8)), Angiotensin III (Ang III (2–8)), Angiotensin IV (Ang IV (3–8)), Angiotensin 1–7 (Ang1–7), and Angiotensin 1–5 (Ang1–5). Aldosterone data was not available for statistical analysis, as over 45% of the samples had analyte levels below the lower limit of quantification. A comprehensive overview of the RAAS biomarker profile is provided in Table 3.

Table 3 Effect of the western diet model on biomarkers of the Renin-angiotensin aldosterone system (RAAS).
Full size table

Total cholesterol, triglycerides and lipoproteins

Figure 6 summarizes the impact of the WD on total cholesterol, HDL-cholesterol, and LDL-cholesterol. After ten weeks of feeding with the WD, there was a 44.0% increase in total cholesterol concentrations (from BAS1 130.0 [125.0–145.0] mg/dL to BAS2 187.5 [173.0–219.0] mg/dL, P < 0.001), along with a significant reduction in the AUC percentage of HDL-cholesterol (from BAS1 84.2 [80.0–85.6] % to BAS2 81.1 [73.8–83.1] %, P < 0.001) and a 26.8% elevation in the AUC percentage of LDL-cholesterol (from BAS1 14.5 [13.0–17.0] % to BAS2 18.0 [15.5–24.5] %, P < 0.001). Looking at the distribution of the absolute values for each lipoprotein fraction, the absolute HDL-cholesterol concentrations increased from BAS1 (6,977.7 [6,091.7–7,319.3] fluorescence intensity (FI)) to BAS2 (8,029.1 [6,998.5–8,346.4] FI) (P = 0.008). However, this increase was more pronounced for the absolute LDL-cholesterol concentrations (BAS1 1,206.5 [976.8–1,505.8] FI vs. BAS2 1,857.2 [1,293.9–2,467.8] FI, P < 0.001), suggesting that the WD increased total cholesterol and caused a shift from HDL-cholesterol to LDL-cholesterol. The detailed lipoprotein profiles, including levels at both baseline (BAS1) and post-WD feeding (BAS2), along with their statistical significance, are presented in Table 4. Overall, besides the changes in the AUC percentages of total HDL and total LDL-cholesterol, the WD caused a significant decrease in HDL 2a (P < 0.001), which are larger and less dense subfractions of HDL-cholesterol. Additionally, the WD induced a significant increase in LDL 3 and LDL 4 (P < 0.001), which are small and dense subfractions of LDL-cholesterol. Notably, these changes were not accompanied by significant alterations in serum triglyceride concentrations (BAS1 38.0 [33.0–45.0] mg/dL vs. BAS2 37.5 [34.0–48.0] mg/dL, P = 0.54).

Fig. 6
figure 6

Temporal changes in total cholesterol, HLD-cholesterol and LDL-cholesterol after ten weeks of feeding with a high-fat, high-monosaccharide, low-fiber western diet. Circulating levels of cholesterol were significantly increased (+ 44.2%) after ten weeks of feeding with the isocaloric WD. Notably, this change was accompanied by a significant reduction in HDL-cholesterol and a 26.8% elevation in LDL-cholesterol. Box plots represent the 25th, 50th and 75th percentile of the data ± 1.5 IQR (interquartile range). •••: P ≤ 0.001. Summary box plots produced using the ggplot2 package in R version 4.2.2.

Full size image
Table 4 Effect of the western diet model on circulating lipoprotein fractions.
Full size table

Oxidative stress

Antioxidant status

Overall, the effect of the WD on antioxidant markers was mild, with no significant changes in CUPRAC, FRAP, TEAC, and Thiol values. In contrast, PON-1 levels significantly decreased at BAS2 compared to BAS1 (BAS1 4.2 [3.7 to 4.4] IU/mL vs. BAS2 3.8 [3.6–4.0] IU/mL, P = 0.004), and GPx activity increased significantly at BAS2 (BAS1 6460.0 [5448.0–7764.0] U/L vs. BAS2 8,432.0 [6964.0–8852.0] U/L, P < 0.001). These effects are summarized in Fig. 7A.

Fig. 7
figure 7

Temporal changes in antioxidant (A) and oxidant (B) stress markers after ten weeks of feeding with a high-fat, high-monosaccharide, low-fiber western diet. The WD had mild effects on antioxidant markers, with no significant changes in CUPRAC, FRAP, TEAC, and Thiol values. However, PON-1 levels significantly decreased at BAS2. The impact of the WD on oxidative stress parameters was more consistent, with total oxidant status significantly increasing at BAS2. The increase extended to reactive oxygen metabolites (d-ROMs). Conversely, there was a decrease in POX-Act post-WD, but no notable effects on AOPP. Box plots represent the 25th, 50th and 75th percentile of the data ± 1.5 IQR (interquartile range). •: 0.01 < P ≤ 0.05; ••: 0.001 < P ≤ 0.01; •••: P ≤ 0.001. Summary box plots produced using the ggplot2 package in R version 4.2.2.

Full size image

Oxidant status

The impact of the WD on oxidative stress parameters was more consistent, with total oxidant status significantly increasing at BAS2 (BAS1 4.8 [3.9–5.8] µmol/L vs. BAS2 7.0 [4.9–8.7] µmol/L, P = 0.018). This also extended to an increase (trend) in reactive oxygen metabolites (BAS1 21.3 [13.2–28.9] U.CARR vs. BAS2 28.8 [17.9–43.0] U.CARR, P = 0.084). Conversely, there was a decrease in POX-Act post-WD (BAS1 101.8 [79.1–114.0] µmol/L vs. BAS2 92.3 [62.1–94.2] µmol/L, P < 0.001). However, there were no discernible effects on AOPP (Fig. 7B).

Metabolomics

For all three –omics evaluated (General Metabolism, Complex Lipids, and Biogenic Amines), significant differences were observed in the metabolomes between BAS1 and BAS2.

General metabolism

Before feature selection, a clear separation between BAS1 and BAS2 was observed in the PCAs for urine, stool, and serum (Fig. 8A–C). To identify variables responsible for this separation, FS-CR was further employed74,75. FS-CR identified 48 significant metabolites in the urine samples, 37 significant metabolites in stool samples, and 10 in serum samples. The loadings of the selected variables are included in the Supplementary Information (Supplementary Figs. S1–S3, S2–S6 and S7–S9, for General Metabolism, Complex Lipids and Biogenic Amines, respectively). Following feature selection, BAS1 and BAS2 were clearly separated along PC1 for all three sample types, which explained 29.4%, 48.6%, and 82.3% of the total variance for urine, stool, and serum samples, respectively (Fig. 9A–C).

Fig. 8
figure 8

PCA score plots (General Metabolism) of (A) Urine, (B) Stool, and (C) Serum before feature selection using the PLS_Toolbox software (Version 9.0; Eigenvector Research, Manson, WA).

Full size image
Fig. 9
figure 9

PCA score plots (General Metabolism) of (A) Urine, (B) Stool, and (C) Serum after feature selection using the PLS_Toolbox software (Version 9.0; Eigenvector Research, Manson, WA).

Full size image

In urine, 29 metabolites were positively correlated with BAS1, including pipecolinic acid, piperidone, cytosine, and nicotinamide (Supplementary Table 1). Additionally, 19 metabolites were strongly positively correlated with BAS2, including 2,3-dihydroxybutanoic acid (tartaric acid), arabitol, cellobiose, and glycerol (Supplementary Table 1).

In stool, seven metabolites were positively correlated with BAS1, such as cadaverine, trans-4-hydroxyproline, tryptamine, and isopalmitic acid (Supplementary Table 2). Thirty metabolites were strongly positively correlated to BAS2, including fructose, pipecolinic acid, erythrose, and 2-deoxyerythritol (Supplementary Table 2).

In serum, nine of the ten significant metabolites from FS-CR were positively correlated to BAS1, including 3-Amino-2-piperidone and 2-picolinic acid (Supplementary Table 3).

In summary, the clear separation of BAS1 and BAS2 in PCA space before feature selection suggests that the WD is causing significant, measurable changes in the general metabolism of the dogs.

Complex lipids

Prior to feature selection, no separation was observed between BAS1 and BAS2 for complex lipid urine samples (Fig. 10A). However, separation between BAS1 and BAS2 was observed along PC1 and PC2 for stool (Fig. 10B), and along PC1 for serum (Fig. 10C).

Fig. 10
figure 10

PCA score plots (Complex Lipids) of (A) Urine, (B) Stool, and (C) Serum before feature selection using the PLS_Toolbox software (Version 9.0; Eigenvector Research, Manson, WA).

Full size image

With feature selection, a clear separation was achieved between BAS1 and BAS2 along PC1 for all three biospecimens (Fig. 11A–C). It is noteworthy that more than three-quarters of the total variation was explained by PC1 for stool (76.7%) and serum (82.6%) samples. With FS-CR, 36 lipids in urine, 36 in stool, and 30 in serum were selected as significant metabolites describing differences between BAS1 and BAS2.

Fig. 11
figure 11

PCA score plots (Complex Lipids) of (A) Urine, (B) Stool, and (C) Serum after feature selection using the PLS_Toolbox software (Version 9.0; Eigenvector Research, Manson, WA).

Full size image

In urine, 25 lipids were positively correlated with BAS1, and 11 lipids were correlated with BAS2 (Supplementary Table 4).

In stool, 32 lipids were positively correlated with BAS1, including eicosapentaenoic acid and various triglycerides, and four lipids were correlated with BAS2, including margaric acid (Supplementary Table 5).

In serum, 14 lipids were positively correlated with BAS1, including phosphatidylcholine 38:5 and phosphatidylcholine 40:7, and 16 lipids were correlated with BAS2, including sphingomyelin (d36:2) and several phosphatidylcholines (Supplementary Table 6).

Overall, the separation of BAS1 and BAS2 stool and serum in PCA space before feature selection indicates that the WD is altering the lipidome of the dogs. Separation of BAS1 and BAS2 for complex lipids in urine was only observed after feature selection. However, 36 urinary lipids were altered as a result of the WD, suggesting that the urinary lipidome of the dogs is also affected.

Biogenic amines

Prior to feature selection, there was significant overlap between BAS1 and BAS2 for urine (Fig. 12A). However, for stool (17.6%) and serum (11.8%) samples, there was a clear separation along PC2 (Fig. 12B and C). FS-CR identified 90 significant metabolites in urine, 68 significant metabolites in stool, and 26 significant metabolites in serum. After feature selection, BAS1 and BAS2 samples were clearly separated along PC1 for all biospecimens, accounting for approximately half of the total variance in the experimental data (Fig. 13).

Fig. 12
figure 12

PCA score plots (Biogenic Amines) of (A) Urine, (B) Stool, and (C) Serum before feature selection using the PLS_Toolbox software (Version 9.0; Eigenvector Research, Manson, WA).

Full size image
Fig. 13
figure 13

PCA score plots (Biogenic Amines) of (A) Urine, (B) Stool, and (C) Serum after feature selection using the PLS_Toolbox software (Version 9.0; Eigenvector Research, Manson, WA).

Full size image

In urine, 47 metabolites were positively correlated with BAS1, including N-acetylmannosamine, threonic acid, nicotinamide, and dopamine (Supplementary Table 7). Additionally, 43 urinary metabolites were positively correlated with BAS2, including N-methylphenylalanine, tartaric acid, and propoxyphene (Supplementary Table 7).

In stool, 51 metabolites were positively correlated with BAS1, including O-acetylsalicylic acid, caffeic acid, and 3-pyridinemethanol, while 17 metabolites were correlated with BAS2, including stachydrine and prochlorperazine (Supplementary Table 8).

In serum, 11 metabolites were positively correlated with BAS1, including 4-aminobenzoic acid and L-histidinol, while 15 metabolites were correlated with BAS2, including secnidazole, tartaric acid, and vanillin (Supplementary Table 9).

In summary, the separation of stool and serum samples between BAS1 and BAS2 before feature selection in PCA space suggests that the WD is altering the bioamine profile of the dogs. After feature selection, there was a clear separation of BAS1 and BAS2 for urine samples in PCA space, demonstrating that the WD is also altering the urinary biogenic amines profile of the dogs.

Discussion

The pleiotropic effects of SGLT-2 inhibitors and GLP-1 agonists hold the potential to target cardiorenal, hepatic and metabolic disorders using a disease model that replicates key features of MetS. Previous experiments have primarily focused on obesity-related metabolic dysfunction when examining the effects of WDs in dogs. However, there is a lack of comprehensive studies on the biological and metabolic impacts of WDs independent of obesity. In both dogs and humans, obesity causes a redistribution of adipose tissue, leading to an increase in visceral fat compared to subcutaneous fat. This change in fat distribution is independently associated with the development of MetS. This is significant as multiple reports, including analyses by Romero-Corral et al.38 and Shi et al.39 on large adult populations, have established that individuals with normal weight MetS are at an increased risk of cardiovascular diseases, metabolic dysregulation, and higher mortality rates compared to obese individuals with or without MetS. In addition, most clinical investigations on the effectiveness of SGLT-2 inhibitors for cardiovascular and renal outcomes primarily include non-obese subjects4,7,40,41,42. Importantly, a recent study by Adamson et al.43 demonstrated that in heart failure patients with reduced ejection fraction, the SGLT-2 inhibitor dapagliflozin effectively reduced the composite endpoint of worsening heart failure or cardiovascular death, regardless of the patient's baseline BMI. Collectively, these findings provide a compelling rationale for establishing a novel model of metabolic dysfunction that operates independently of obesity. Once established, this disease model will pave the way for in-depth studies on the pharmacodynamics of therapeutic drugs targeting cardiorenal metabolic health, including SGLT-2 inhibitors and GLP-1 agonists.

Our study maintained isocaloric conditions to isolate the effect of the diet’s composition from obesity as a confounding factor. It builds on preliminary data from Lyu et al.76, which showed a tendency towards elevated glucose concentrations in ten healthy Beagles under an isocaloric high-fat diet for six weeks. To the best of our knowledge, our research represents the first comprehensive characterization of the biological effects of a WD model, independent of obesity. By inducing MetS without causing weight gain, we have successfully developed a non-invasive, inducible, and potentially reversible preclinical model in just a few weeks. For ethical reasons and considerations related to animal welfare, it is important to emphasize that our objective was not to induce clinical symptoms of MetS in our study. Therefore, most of the observed changes reported herein remained within physiological limits. Overall, the WD was well tolerated with no reported adverse events during the study. Minor digestive issues appeared when transitioning from a regular diet to the WD, likely due to the absence of a proper weaning period between diets. However, these issues were resolved within a few days.

Hematological parameters typically remained within normal physiological limits, showing no clinically meaningful changes. The most notable variations were observed in metabolic parameters. Specifically, the WD induced a statistically significant increase in fasting blood glucose levels, approaching the upper physiological limit. This resulted in an overall increase of approximately 20% in blood glucose concentrations compared to baseline. Interestingly, this observation was accompanied by a trend towards an increase in peak insulin concentrations, which is consistent with previous literature in dogs32,36 and indicative of a perturbation in insulin sensitivity that ultimately leads to insulin resistance77. The decrease in circulating glucagon concentrations may be symptomatic of a physiological feedback mechanism to maintain glucose homeostasis in response to increased fasting blood glucose78.

Our dietary intervention also resulted in significant changes to serum chemistry parameters. These fluctuations, although still within physiological limits, demonstrate the ability of our model to greatly influence metabolism and homeostasis. Specifically, we observed a decrease in serum bicarbonate levels, which is in line with low-grade metabolic acidosis79. Our findings in dogs agree with previous data from the human literature. WDs are typically characterized by a high dietary acid load due to large amounts of animal protein and processed foods, which can exceed the kidneys' capacity to excrete the excess acid, leading to metabolic acidosis80. The concurrent increase in serum chloride concentrations reflects a state of hyperchloremic acidosis81. Overall, the association of hyperchloremic metabolic acidosis (with increased bicarbonate loss and a marked reduction in both phosphorus and potassium levels is consistent with some degree of renal tubular acidosis82,83,84. These findings are significant, as a recent meta-analysis, which included data from over 30,000 patients, found an association between metabolic acidosis, low bicarbonate levels, and features of MetS85. Similarly, multiple studies have shown that low levels of potassium86,87 and phosphorus88,89 were associated with an increased risk for MetS in humans.

Consistent with the definition of MetS by the National Heart, Lung, and Blood Institute (NHLBI), our diet induced a significant elevation of SBP by approximately 10 mmHg. The increase in SBP appeared to be independent of any notable changes in renin and angiotensin peptides. Interestingly, SBP was not found to increase in a previous canine study focusing on obesity-related cardiac dysfunction and MetS90, again supporting our rationale for studying the effect of western diets independently of obesity. Our study also found mild increases in NT-proBNP, although mostly within the reference range. We suspect that the increase in circulating natriuretic peptides occurred secondarily to the increase in SBP, as previously reported in the literature91,92, but it could also be indicative of cardiac stress93. Notably, some dogs showed NT-proBNP concentrations exceeding 900 pmol/L, a threshold commonly associated with structural heart disease in canines94,95.

Total cholesterol increased by approximately 45% after ten weeks. Importantly, in line with the definition of MetS, dogs fed the isocaloric WD model experienced a significant reduction in the AUC percentage of total HDL-cholesterol, along with an increase of the AUC percentage of LDL-cholesterol. Interestingly, upon examining the absolute value of each lipoprotein fraction, the absolute (untransformed) concentration of HDL-cholesterol increased after BAS1. However, this increase was more modest compared to that of LDL-cholesterol. This suggests that the WD increased total cholesterol and caused a shift from HDL to LDL-cholesterol, which is also consistent with the development of MetS. In addition to changes in total HDL and LDL-cholesterol, the WD also caused a significant decrease in the larger and less dense subfraction of HDL cholesterol (HDL 2a), with previous studies reporting a decreased risk of cardiovascular disease with higher levels of HDL2 cholesterol subfractions96,97. Consistent with these findings, the WD induced a significant increase in the small and dense subfractions of LDL cholesterol (LDL 3 and LDL 4), which are associated with a higher risk of atherosclerosis and stroke98. These changes occurred independently of any corresponding alterations in serum triglyceride concentrations, which is a variation from our preliminary data44 showing an increase in serum triglycerides in 10 healthy dogs fed with a WD for seven weeks. Although we do not have a formal explanation for the discrepancy observed between the two studies, we hypothesize that differences in the age of the dogs (with dogs from our preliminary study being overall older, with a median age of 3 years) and the significant, although modest, increase in their body weight in Iennarella-Servantez et al.44 could have contributed to the differences between studies. While surprising, this finding is also consistent with earlier research from Lahm Cardoso et al.99 which showed a strong correlation between BCS and triglyceride levels in dogs, with values approaching the upper limit of 200 mg/dL in dogs with a BCS of 8 or above (classified as “overweight” or “obese” in our study).

Our results on redox status align with previous human studies100,101,102. Specifically, we observed significant increases in TOS and d-ROMs at BAS2. In contrast, the effect on antioxidant markers was more nuanced and generally mild, with levels of CUPRAC, FRAP, TEAC, and Thiol remaining stable at BAS2. This is in line with the variable impact of dietary fat on systemic antioxidative stress markers in dogs. Some studies have shown no effect of carbohydrate and fat concentrations on oxidative stress biomarkers103, while others have reported an increase in antioxidant capacity, but no effect on oxidative stress markers37.

Our study highlights the comprehensive changes in the metabolome induced by the WD, including biological pathways related to general metabolism, complex lipids, and biogenic amines. These observations underscore the potential relevance of this model in studying MetS and its associated health complications. Notably, all the metabolites detected in our study were classified according to MSI Level 2 standards73. The positive correlation of nicotinamide to the baseline diet (BAS1) in both general metabolism (urine) and biogenic amines (urine) suggests that dogs had lower amounts of this essential form of vitamin B3 after ten weeks of feeding with a WD (BAS2) compared to their standard diet. This is consistent with earlier findings from Qu et al.104 reporting perturbations in nicotinamide metabolism in dogs fed a high-fat diet for twelve weeks. Nicotinamide plays a crucial role in various metabolic pathways, particularly in energy production and DNA repair105,106. Similarly, the positive correlation of glycerol to BAS2 in general metabolomics (urine) indicates that glycerol concentrations were increased during feeding with the WD. Glycerol is a key component of triglycerides and is involved in energy metabolism, especially in lipid breakdown and synthesis107. This elevation is likely related to an increased metabolism of triglycerides caused by the WD, indicating a potential shift in lipid metabolism. The positive correlation of tartaric acid (2,3-dihydrobutanoic acid) with BAS2 in multiple classes (general metabolomics in urine, biogenic amines in urine, and biogenic amines in serum) indicates that tartaric acid levels increased during the WD phase. These changes are possibly associated with the increased catabolism of the antioxidant ascorbic acid and accompany variations in oxidative stress markers highlighted above108.

A greater diversity of fatty acids was correlated with BAS1, especially in stool, indicating a wider range of fatty acid profiles in the baseline diet. This diversity is essential for energy production and cell membrane structure109. We suspect that the higher diversity of fatty acids in BAS1 is due in part to the presence of more soluble dietary fiber, which is fermented by intestinal microbiota into short chain fatty acids110. After the WD diet, there was an increase in saturated fatty acids, specifically FA 17:0 in stool lipidomics and PC 18:0. High levels of saturated fatty acids have been linked to negative health outcomes, such as cardiovascular disease111,112. Furthermore, palmitoleic acid, an omega-7 monounsaturated fatty acid, showed a positive correlation with BAS2 in stool general metabolomics. Palmitoleic acid (16:1n7) increases lipolysis, glucose uptake, and glucose utilization for energy production in white adipose cells113,114. The increase in saturated fatty acids, specifically FA 17:0 in BAS2, could suggest a shift in lipid metabolism and/or be a result of the consumption of butter and ruminant fats115. While there is a decrease in fatty acid diversity and an increase in saturated fatty acids associated with BAS2, it remains unclear whether these changes are related to MetS per se, or simply changes in stool composition caused by differences in diet composition.

In both general metabolomics (GC–MS) and LC–MS assays, several unidentified metabolites were detected. For GC–MS, this was due to spectral library matches failing to identify metabolites below the 800 threshold. Advanced data processing techniques, such as Parallel Factor Analysis, could be employed to deconvolve data and obtain cleaner spectra116,117. However, this would require a separate and dedicated study. Additionally, the bioamines assay detected several non-amine compounds due to its ability to detect compounds without an amine group. With LC–MS, the presence of unidentified metabolites could possibly be attributed to biotransformation of known metabolites, which involves the addition or removal of specific chemical moieties such as (de)-glycosylation, (de)-methylation, (de)-amination, and (de)-hydroxylation. These transformations often occur during metabolic processes118. To identify these metabolites, biotransformation analysis techniques and exploration of additional libraries and databases would be necessary. However, this is beyond the scope of this study. Despite the challenges in compound identification, the changes in metabolism observed across the three metabolomic assays are noteworthy. Additionally, several metabolites previously identified as both positively and negatively correlated with MetS in other studies, including nicotinamide, tryptophan metabolites, glycerol, and tartaric acid, were found in the metabolomics analyses presented here104,105,106,108,119. This suggests that the WD is inducing metabolomic changes consistent with MetS in dogs.

This study presents several limitations that are worth mentioning. First, the experiment was limited in size and did not address the potential for reversibility of the model, specifically regarding the metabolic impacts of transitioning back to a standard diet. Due to the non-invasive nature of our model, this study was primarily descriptive and did not provide in-depth mechanistic insights into how WDs affect cardiovascular-kidney-metabolic health. The study also lacks some functional data, such as the time-dependent effects of the WD on intestinal permeability and fecal microbiome composition. This was partly deliberate, as those effects have been extensively characterized previously in the literature (e.g., Ref32). Previous studies published by our consortium in dogs51,52,53,54 were conducted in the context of an activated RAAS. No elevation in PRA-S or angiotensin peptides was noted in the current study, which likely explains why many data points on aldosterone were below the lower limit of quantification, preventing statistical analysis of this biomarker. Additionally, our findings related to serum, urine and fecal metabolomics could be confounded by factors other than changes in diet composition, including differences in processing and ingredient composition between a commercial and a home-cooked diet. To minimize the impact of these potential confounding variables, we would need to conduct an experiment using two homemade diets with similar ingredients, while varying the proportions of fat, carbohydrates, and fiber to simulate the effect of a Western versus a standard diet on variables associated with MetS. At last, determining whether metabolites are changing in response to altered metabolism or simply due to changes in dietary composition using untargeted metabolomics can be challenging. Therefore, mechanism-based hypothesis testing on the metabolome would require a targeted or semi-targeted approach that focuses on specific classes of metabolites.

In summary, our isocaloric WD, designed to mimic the NHANES diet, effectively replicated key characteristics of MetS. These included elevated BP, increased fasting glucose levels, and reduced HDL-cholesterol, all independent of abdominal obesity. Additionally, the WD induced an increase in natriuretic peptide levels, along with a mild state of metabolic acidosis and significant changes in the serum, urine, and fecal metabolome. Our findings underscore the utility of this model for investigating the metabolic effects of novel antidiabetic therapies in the context of obesity-independent MetS, which will be presented in a separate manuscript. Furthermore, this research enables future translational studies that could have potential benefits for both human and veterinary medicine.

Ethical approval

Experimental procedures were approved by the Institutional Animal Care and Use Committee at Iowa State University (Protocol Number: 21-164). All methods were performed in accordance with the relevant guidelines and regulations at Iowa State University. The authors complied with ARRIVE guidelines in the completion of this study.