Influence of macronutrient composition of commercial diets on circulating leptin and adiponectin concentrations in overweight dogs

Abstract Leptin and adiponectin play important roles in obesity‐related inflammation and comorbidities. Previous research suggests that alterations in dietary macronutrient composition can modify circulating leptin and adiponectin concentrations in people, but limited research on this subject has been performed in dogs. This study investigated the effects of commercial high protein (HP), high fat (HF) and high carbohydrate medium protein (HCMP) diets on baseline (T−1) concentrations, post‐prandial peak concentrations and total release in a ten‐hour time span of leptin and adiponectin in dogs, when compared to a maintenance high carbohydrate low protein (HCLP) diet. Thirty‐six overweight dogs were fed the HCLP diet in a one‐week control period, after which the animals were assigned to one of three groups. In three four‐week periods, each group was fed all test diets in a different sequence. At the last day of each period, blood was sampled at one hour before feeding (T−1) and at three (T3), six (T6) and nine (T9) hours after feeding. Feeding caused peak leptin concentrations at T6 and T9 (p < .001). No significant post‐prandial change in adiponectin concentrations was found (p = .056). The HP diet resulted in lower leptin peak concentrations (p = .004) and AUCT−1–T9 (p = .01), but none of the diets influenced baseline leptin concentrations (p = .273). Baseline adiponectin concentrations were lower for the HF diet (p = .018) and HCMP (p < .001), and the HP, HF and HCMP AUCT−1–T9 (p < .001) were lower compared with the HCLP diet. Female dogs had lower adiponectin baseline concentrations (p = .041) and AUCT−1–T9 (p = .023) than male dogs. In conclusion, the HP diet was associated with the lowest post‐prandial peak leptin concentration and the least decrease in adiponectin release, suggesting that a HP diet may improve immune‐metabolic health and post‐prandial satiety in overweight dogs.

Adipokines have physiologic functions in energy homeostasis and immune response regulation, via balanced release of inflammatory and anti-inflammatory factors (German et al., 2009;Radin et al., 2009). In overweight conditions, hypertrophy and hyperplasia of adipose tissue have detrimental effects on this balance. Studies in overweight dogs showed increased release of inflammatory cytokines and decreased release of anti-inflammatory adipokines in comparison with lean dogs (Bastien et al., 2014;Park, Lee, Oh, Seo, & Song, 2014). It is therefore implied that canine obesity is associated with continuous low-grade inflammation, which in turn contributes to the development of comorbidities (Bastien et al., 2014;German et al., 2009).
Leptin has pro-inflammatory and immune-modulating actions, and an increase in fat mass results in increased circulating leptin concentrations, indicating an important role in signalling energy status (Cortese, Terrazzano, & Pelagalli, 2019;Orr & Davy, 2005;Radin et al., 2009). Leptin additionally induces a feeling of satiety after feeding and promotes energy expenditure through fatty acid oxidation and sensitising effects on peripheral insulin receptors (Minokoshi et al., 2002;Orr & Davy, 2005;Radin et al., 2009). However, persistent high leptin concentrations, associated with leptin resistance, promote inflammation and development of comorbidities, and could lead to a lack of satiety after feeding with increased difficulty to induce weight loss (Cortese et al., 2019;Radin et al., 2009). In addition, loss of leptin-induced insulin-sensitising effects assists in the development of insulin resistance and hyperinsulinemia (Jeusette et al., 2005).
Adiponectin, on the other hand, is released from adipocytes during periods of fasting to increase food intake and reduce energy expenditure (Lee & Shao, 2014). Adiponectin increases insulin sensitivity and has an anti-inflammatory effect, which counteracts insulin resistance and low-grade inflammation. Studies in overweight dogs show a reduction in adiponectin release, probably due to inhibition of gene expression by pro-inflammatory cytokines (Park et al., 2014;Radin et al., 2009).
Several studies have described improvement of immune-metabolic health after weight loss in overweight dogs (Bastien et al., 2014;German et al., 2009). However, it is often hard for owners to uphold energy restriction and weight loss is not always achieved (German, Holden, Bissot, Hacket, & Biourge, 2007). In humans, a high-protein diet in combination with exercise and calorie restriction decreased leptin and increased adiponectin concentrations, thus improving insulin sensitivity and lowering inflammation (Ata et al., 2010). Another study in obese and diabetic humans used a low carbohydrate or a low fat, calorie-restricted diet, and decreased circulating leptin and increased circulating adiponectin concentrations (Vetter et al., 2010). Altogether, these studies suggest that diet may play an important role in modulating adipokine release.
Information on the effects of macronutrient composition of diets on leptin and adiponectin concentrations in dogs is limited. One study did not find an effect of protein content of diets on baseline leptin concentrations when combined with neutering (Kawauchi et al., 2017). Another study combined diet with weight loss and revealed lower leptin concentrations with diets containing starches with a low glycemic index and higher adiponectin concentrations in diets with added diacylglycerols, when compared to diets with added triacylglycerols (Mitsuhashi et al., 2010). This study aimed to find out whether commercially available high protein, high fat or high carbohydrate diets can modulate baseline and post-prandial concentrations of leptin and adiponectin in overweight dogs.

| Animals
As overweight dogs have different adipokine levels when compared to lean dogs (Park et al., 2014), 36 mostly overweight, but otherwise healthy experimental Beagle dogs (body condition score (BCS) range 5-8/9 and age range 1-12 years), housed in the research kennel of the Faculty of Veterinary Medicine at Utrecht University, were included (Table 1). Dogs had 3-4 hr of voluntary exercise per day and had voluntary outdoor access. All dogs were intact, and at the trial start, three dogs had a BCS of 5/9. The other dogs had become overweight due to previous excess feeding of a maintenance diet, before the start of the trial. Body weight (kg) and BCS assessment and physical examinations were performed to ensure the health status of each dog prior to the start of the trial. Body condition was determined using a 9-point scale, as validated by Laflamme (Laflamme, 1997). Estimations of the daily energy requirement (DER) were made using the energy intake, BCS and weight of each dog prior to the study. The protocol and study design were approved by the Animal Ethics Committee at Utrecht University (registered under number AVD1080020184847) and the Royal Canin Ethics Committee.

| Trial design and management
To avoid confounding by gender, stratified randomisation based on gender was used to assign each dog in one of three groups (Table 1). After a control period of one week, in which a high in carbohydrates and low in protein (HCLP) maintenance diet was fed, the groups were fed one of the following dry diets: (a) a highprotein diet (HP); (b) a high-fat diet (HF); and (c) a high carbohydrate and medium protein diet (HCMP), when compared to the maintenance diet (Table 2). After a four-week period, the diets were changed without run-in period, with all groups having the three diets in a different sequence (Table 1). At the last day of each four-week period, 2 ml blood was collected at one hour before feeding after an overnight fast (T −1 ) and at three (T 3 ), six (T 6 ) and nine hours (T 9 ) after feeding to determine baseline concentrations and post-prandial kinetics for both leptin and adiponectin, the latter consisting of the post-prandial peak concentrations and the total release via area under the curve calculations for ten-hour release. Each diet was fed isocaloric to avoid changes in body weight and body condition. Portion size of each diet was based on DER estimations and was adjusted to keep the animals overweight.
The animals were fed once a day in the morning with ad libitum availability of water. All dogs ate their food within 30 min, and food bowls were removed one hour after feeding.

| Monitoring
Every week, physical examinations of all dogs were carried out to monitor health status and detect early signs of gastrointestinal upset following dietary change. Median BCS (range) Start trial 6 (6-7) 6 (5-8) 6.5 (5-7) End trial 6 (6-7) 6 (5-7)* 6 (5-7) Body weight (kg) Moisture (g/100 g) 8 9.5 9.5 9.5 g/MJ ME g/100 g DM g/MJ ME g/100 g DM g/MJ ME g/100 g DM g/MJ ME g/100 g DM Additionally, body weight and BCS were assessed every week to determine preservation of body weight and body condition during the trial and to increase or decrease the dietary quantity accordingly.
Examinations and BCS assessments were performed by the same investigator. Dogs with severe gastrointestinal disease or weight change of more than 10% of their starting body weight were excluded from the trial.

| Sample collection
Blood was sampled by jugular venipuncture and collected in serum tubes with clotting activator. After centrifugation, serum was collected and stored at −20°C until analysis to avoid multiple freezing and thaw cycles.

| Assay validation and sample analysis
Precision was estimated by calculating the inter-and intra-assay coefficients of variations (CV

| Statistical analysis
Statistical analyses were performed by commercial software (IBM SPSS Statistics for Windows, version 25.0. IBM Corporation). Data were tested for normality using Kolmogorov-Smirnov tests and Q-Q plots. AUCs from/T to T 9 (AUC T−1-T9 ) were calculated using the trapezoidal method as an estimation of the total release in a tenhour time span.
Differences in age and weight between groups were compared with an analysis of variance (ANOVA) with Bonferroni test as post hoc test for significant differences. Comparisons between mean body weight in each group and overall body weight before and after the trial were made with a paired t test. Differences in BCS between groups were compared with a Kruskal-Wallis test. Additional comparisons between BCS before and after the trial for each group and overall BCS were made with the Wilcoxon signed-rank test with Bonferroni correction. Serum leptin and adiponectin concentrations did not comply to the assumptions of a repeated measures ANOVA, and the response of serum leptin and adiponectin concentrations to feeding during the control period at/T, T 3 , T 6 and T 9 were compared with Friedman tests and the Wilcoxon signed-rank tests with Bonferroni correction.

| RE SULTS
No dogs showed adverse events following dietary change, and no dogs were excluded from the trial. Gastrointestinal signs, such as diarrhoea or vomiting, were not observed. T 9 samples of two dogs could not be collected during the maintenance period and were omitted from the post-prandial effect analysis and the mixed model of the AUC T−1-T9 of leptin and adiponectin. Age (p = .16), body weight before (p = .43) and after (p = .19) the trial and BCS before (p = .32) and after (p = .32) the trial did not differ significantly between groups (Table 1). Body weight (p = .048) and BCS (p = .046) in group 2 reduced significantly at the end of the trial (p = .048), despite eating the adjusted amounts of food and without showing gastrointestinal signs, but all dogs remained within the 10% limit of their starting body weight.

| Serum leptin concentrations
Twelve samples of 6 dogs fell under the detection limit of the leptin assay and were regarded as 0 ng/ml. Dietary intake increased serum leptin concentrations (n = 34, p < .001) (Figure 1). Post hoc analysis showed that/T differed from T 3 , T 6 and T 9 , and T 3 differed from T 6 and T 9 . T 6 and T 9 did not differ, and T 6 was regarded as the peak leptin concentration in further analyses.
The effects of diet on leptin concentrations are listed in Table 3.
Five outliers were excluded from the analysis of baseline leptin concentrations, 3 from analysis of peak leptin concentrations and leptin AUC T−1-T9 . Baseline leptin concentrations were not affected by diet (p = .273). Diet affected leptin peak (T 6 ) concentrations (p = .002). The HP diet provided the lowest leptin peak concentrations (p = .004), and a lower total leptin release as measured by the AUC T−1-T9 (p = .01), when compared to the HCLP diet. In all groups, the diet that followed the HCLP diet, that is the HF diet in group 1, the HCMP diet in group 2 and the HP diet in group 3, produced lower leptin peak leptin concentrations (p < .001) and AUC T−1-T9 (p < .001), which was revealed as an interaction between diet and group. No other factors or interactions altered leptin concentrations and were omitted from the final model.

| Serum adiponectin concentrations
All samples were above the detection limit of the adiponectin assay. In addition, male dogs had significantly higher baseline adiponectin concentrations (p = .041) and AUC T−1-T9 (p = .023) than female dogs (Table 4). Adiponectin AUC T−1-T9 was also influenced by age (p = .082), which increased with 2.86 μg ml −1 10 hr −1 per year of age. No other factors or interactions influenced adiponectin concentrations. and insulin-like growth factor 1 (IGF-1) (Kawauchi et al., 2017). In the present study, the effect of the HP diet on the post-prandial leptin suggests a role of dietary protein in the regulation of long-term satiety (Kawauchi et al., 2017;Orr & Davy, 2005).

| D ISCUSS I ON
Release of leptin and adiponectin from adipocytes is closely related to glucose metabolism and the balance between pro-and anti-inflammatory cytokines. Leptin is released when circulating insulin concentrations and pro-inflammatory cytokines increase in humans, the latter also inducing leptin resistance (Park & Ahima, 2015;Sáinz, Barrenetxe, Moreno-Aliaga, & Martinez, 2015;Seufert, 2004). Lower insulinemic responses and decreases in inflammatory cytokines, as observed in dogs and obese humans when using HP diets (Amini, Maghsoudi, Feizi, Ghiasvand, & Askari, 2016;André et al., 2017), could thus decrease the post-prandial leptin release. The observed effects could also be associated with the relative lack of carbohydrates and dietary fat in the HP diet, as these macronutrients have been associated with leptin resistance (Giugliano, Ceriello, & Esposito, 2006;Koch et al., 2014).

TA B L E 4
Estimated marginal means of adiponectin baseline concentrations (μg/ml) (n = 36) and AUC T−1-T9 (μg ml −1 10 hr −1 ) (n = 34) for each diet and gender as obtained by the mixed model increased with the HP diet. Previously reported increases in circulating adiponectin concentrations after weight loss with HP diets could thus be the result of weight loss, rather than alterations in macronutrient composition (André et al., 2017;Ata et al., 2010;Tvarijonaviciute, Tecles, Martinez-Subiela, & Cerón, 2012). It is, however, possible that the maintenance diet already provided optimal adiponectin concentrations, as feeding HC diets to cats increases circulating adiponectin concentrations (Tan et al., 2011).
As adiponectin has a strong anti-inflammatory effect (Radin et al., 2009), it is possible that the test diets exacerbate low-grade inflammation. However, overweight individuals have lower adiponectin concentrations than lean individuals, and the biologic effects of further decrease are unknown (André et al., 2017;Bastien et al., 2014;Ishioka et al., 2006;.
The post-prandial release of leptin was decreased with the HP diet, despite the fact that most overweight individuals have hyperleptinemia sequential to leptin resistance (Cortese et al., 2019;Park et al., 2014). As dietary intake leads to prolonged periods of high leptin concentrations, with a maximal concentration around six to nine hours after feeding (Ishioka et al., 2005), lower post-prandial leptin release could alleviate signs of hyperleptinemia for a prolonged period, improving post-prandial satiety, insulin sensitivity and immune-metabolic health (Cortese et al., 2019;Radin et al., 2009;. Leptin concentrations in the present study were considerably lower than a previous report in dogs that used the same assay (Park et al., 2014). This discrepancy might originate from the use of solely intact dogs, as opposed to the several neutered dogs in the study of Park et al. (2014), although it was previously suggested neutering status does not affect circulating leptin concentrations (Ishioka et al., 2007). In accordance with preceding studies, baseline leptin concentrations were not affected by diet (Adolphe et al., 2015;Kawauchi et al., 2017), which is likely to be the result of the lack of overall change in body composition during the present study (Ishioka et al., 2007;Park & Ahima, 2015). Leptin concentrations were also not influenced by age and gender, as was previously shown by Ishioka et al. (2007), which makes fasting leptin concentrations a reliable marker of changes in fat mass without being confounded by age, gender and diet when the time of feeding is stated (Ishioka et al., 2005).
From a physiologic perspective, it may be expected that adiponectin concentrations decrease post-prandially to limit its effects on energy uptake and expenditure (Lee & Shao, 2014).
Although in the present study, numerically, the highest concentrations of adiponectin were found before dietary intake and the values decreased after food intake, these differences were not statistically significant, as was also reported by . The use of overweight dogs in this study, in which adiponectin release is already decreased (Ishioka et al., 2006;Tvarijonaviciute et al., 2010), might account for this lack of response to food intake.
In the present study, the concentrations of adiponectin measured with a human adiponectin assay were comparable to a study that validated this assay (Tvarijonaviciute et al., 2010). In contrast to what was observed by Verkest et al. (2011), we found a significant gender effect with regard to adiponectin concentrations, with male dogs having higher adiponectin concentrations than female dogs.
This sex dimorphism disagrees with studies in humans, where testosterone inhibits the release of adiponectin from adipocytes (Xu et al., 2005). Other endocrine influences might contribute to this dimorphism in dogs, as was suggested previously (Verkest et al., 2011).
A limitation of using commercial diets is the variation in ingredients in each diet, which could have interfered with the results.
Considering the effects of different sources of carbohydrates (Carciofi et al., 2008) and proteins (Nuttal, Gannon, Wald, & Ahmed, 1985) on the post-prandial insulin response and sensitivity, it is expected that micronutrient composition also influences the regulation of circulating leptin and adiponectin concentration.
Future studies, preferably with experimental diets, are needed to accurately assess these effects on adipokine release. Additionally, no golden standard of a "normal" diet exists in veterinary nutrition.
In the present study a maintenance diet, high in carbohydrates and low in proteins in comparison with our test diets, was considered closest to a control diet. To exclude the presence of a carry-over effect due to composition, which could explain the interaction that was found between group and diet for leptin concentrations, using a Latin square design with complete randomisation and inclusion of wash-out periods between diets would have been preferred.

| CON CLUS IONS
This study is the first to show the beneficial effects of a HP diet on leptin concentrations, while causing minimal decrease in adiponectin concentrations in overweight dogs without combining dietary change with weight loss. Lower leptin concentrations suggest improved sensitivity to the hormone, thus increasing post-prandial satiety, improving insulin sensitivity and lowering obesity-related inflammation (Cortese et al., 2019;Sáinz et al., 2015). The fact that only post-prandial concentrations of leptin could be altered, might suggest that its baseline levels are foremost dependent on the fat mass of an individual rather than influenced by dietary macronutrients (Ishioka et al., 2007;Park et al., 2014). Adiponectin concentrations could not be increased, but the HF diet, and the HCMP diet decreased baseline adiponectin concentrations and the HP, HF and HCMP diet decreased post-prandial adiponectin release, which could decrease insulin sensitivity and increase obesity-related inflammation (André et al., 2017;Bastien et al., 2014;. To this end, a HP diet might improve immune-metabolic health by decreasing leptin concentrations even before weight loss is achieved, without decreasing baseline adiponectin concentrations and providing minimal decrease in adiponectin concentrations.

ACK N OWLED G EM ENTS
This study was funded by the Veterinary Biomedicine research programme of Utrecht University. Royal Canin is acknowledged for providing the diets in kind. The authors wish to thank all the animal caretakers and students for their care in managing the dogs and for their invaluable assistance with blood sampling. The analysts are thanked for organising and preparing the serum samples. For his advice on the ELISA assays and performing the sample analyses, Dr. J.
A. Mol is sincerely acknowledged

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest. The diets were kindly provided by Royal Canin, but this company was not involved in study design nor the analysis of the results.

A N I M A L WE LFA R E S TATE M E NT
The authors confirm that the ethical policies of the journal, as noted on the journal's author guidelines page, have been adhered to and the appropriate ethical review committee approval has been received.
The authors confirm that they have followed EU standards for the protection of animals used for scientific purposes. The protocol and study design were approved by the Animal Ethics Committee at Utrecht University (registered under number AVD1080020184847) and the Royal Canin Ethics Committee.