obesity can be considered a chronic low-grade inflammatory state characterized by infiltration of macrophages (17, 30, 56). Accumulated evidence strongly suggests that low-grade chronic inflammation plays a crucial role in the development of obesity-related insulin resistance (18). In rodents, diets enriched with n-3 polyunsaturated fatty acids (PUFAs) have been demonstrated to reduce the development of diet-induced obesity (1, 6, 24, 60–62, 73), adipose tissue inflammation (31, 58, 69), and the development of insulin resistance (35, 36, 54, 55, 62, 66). However, we demonstrated recently that sucrose counteracts the anti-inflammatory effect of fish oil in adipose tissue and increases obesity development in mice (40). Mice fed a fish oil-enriched diet in combination with sucrose had markedly higher feed efficiency and required less than 50% of the calories to achieve the same weight gain as mice fed a fish oil-enriched diet in combination with protein (40).

A major difference between proteins and sucrose is the ability of sucrose to elicit a rise in blood glucose and stimulate insulin secretion. Insulin is a powerful anabolic hormone that stimulates adipocyte differentiation and adipose tissue expansion (41), and activation of insulin signaling is crucial for the development of obesity (9). Moreover, increased insulin signaling by transgenic expression of insulin receptor substrate-1 is sufficient to induce obesity (52). Thus, it is possible that increased insulin signaling and glucose uptake in adipose tissue in sucrose-fed mice may override the anti-inflammatory and antiobesity effects of fish oil.

The glucose moiety of sucrose is responsible for the rise in blood insulin upon intake of sucrose because fructose, unlike glucose, is unable to stimulate insulin secretion (16). This in part relates to the very low levels of Slc2a5 [solute carrier family 2 (facilitated glucose transporter), member 5, GLUT5] in pancreatic β-cells (64). Furthermore, fructose does not stimulate the release of gastric inhibitory peptide, which stimulates insulin secretion indirectly (27, 67). Thus, the ability of sucrose to counteract the beneficial effects of fish oil seems to relate to a glucose-dependent stimulation of insulin secretion. The fructose moiety of sucrose may further modulate the effect of fish oil on the development of obesity. Thus, the increased consumption of fructose over the past decades has been linked to development of metabolic disorders (59), and fructose is routinely used to induce glucose intolerance in rats (70).

Because different types of starch differ in their ability to increase postprandial blood glucose and insulin secretion, different types of starch may also modulate the effect of fish oil. The glycemic index (GI) is a measurement of the ability of different types of carbohydrate-based foods to raise blood glucose levels within 2 h (34). The interest in low-GI diets as a tool in weight management is increasing. Although reviews and meta-analyses conclude that such diets may be effective, their efficiency in terms of lasting weight reduction is still a matter of debate (3, 5, 20, 38, 68). Different types of starches with different GI are known to induce different responses in plasma glucose and insulin in rodents (57). However, it is unknown whether different types of starches modulate the effects of fish oil-enriched diets. Here, we have performed systematic analyses to investigate the influence of different carbohydrate sources on the reported antiobesity and anti-inflammatory effect of n-3 PUFAs in mice. By using isocaloric high-fat diets enriched with n-3 PUFAs, we show that the amount of sucrose dose-dependently increased energy efficiency and adiposity. Moreover, we show that nutritional and pharmacological control of insulin secretion plays a pivotal role in determining the obesogenic effect of high-fish oil/high-carbohydrate diets.

MATERIALS AND METHODS

Mice and diets.

In all experiments, male C57BL/6JBomTac mice ∼8 wk of age were obtained from Taconic Europe. The mice were kept at a 12:12-h light-dark cycle at thermoneutrality (28–30°C). After acclimation, the mice were housed individually and fed ad libitum or pair-fed experimental diets prepared by Ssniff Spezialdiäten (Soest, Germany) (Tables 1 and 2). The fish oil was provided by Ssniff, and the fish oil-enriched diets contained 62 ± 3 g/kg n-3 PUFAs [27 ± 3 and 18 ± 3 g/kg eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), respectively]. The diets were kept at −20°C. Fresh water and a fixed amount of feed were provided three times/wk. Body weights and feed leftovers were recorded once/wk. The mice were fed for 4–8 wk as indicated. In separate experiments, mice were treated with nifedipine (N7634; Sigma-Aldrich, St. Louis, MO) or glybenclamide (G0639; Sigma-Aldrich). Nifedipine was included in the diets at a dose (1 g/kg) that was demonstrated earlier to reduce plasma insulin levels in agouti mice (37). The sulfonylurea glybenclamide was used as an insulin secretagogue and was administrated daily by intraperitoneal (ip) injection at a dose of 2 μg/g body wt. Control mice received placebo by daily ip injection. The mice were fed the experimental diets and treated with nifedipine and/or glybenclamide after 1 wk of acclimatization, with free access to a standard low-fat chow diet. Mice fed nifedipine-containing diets or treated with glybenclamide were pair-fed and were terminated after 4 wk of treatments. In experiments where glybenclamide or nifedipine was included, experimental feeding time was reduced to 4 wk to limit possible adverse effects resulting from increased or decreased insulin secretion/signaling.

At the end of all experiments, mice were euthanized in the fed or fasted state by cardiac puncture under anesthesia with isoflurane (Isoba-vet, Schering-Plough, Denmark) using the Univentor 400 Anaesthesia Unit (Univentor), and plasma was prepared from blood. Liver, tibialis anterior muscle, and adipose tissues were dissected out, weighted, freeze-clamped, and frozen at −80°C. The epididymal white adipose tissue (eWAT) depots located around the testis and perirenal and retroperitoneal white adipose tissue (pWAT) depots were selected as representative visceral depots, whereas inguinal white adipose tissue (iWAT) posterior subcutaneous depot was selected to represent subcutaneous adipose tissue (12). Brown interscapular brown adipose tissue (iBAT) was taken from the anterior subcutaneous interscapular region (12). All mice experiments were approved by National Animal Health Authorities (Denmark and Norway). Adverse events were not observed.

Glucose and insulin tolerance tests.

In glucose tolerance tests (GTT), mice were feed-deprived for 6 h before ip injection of 2 g/kg glucose in saline. In insulin tolerance tests (ITT), mice were feed-deprived for 4 h before ip injection of 0.5 U/kg human recombinant insulin (Actrapid; Novo Nordisk, Bagsværd, Denmark) in saline. Blood was collected from the lateral tail vein at indicated time points, and blood glucose was measured with a Bayer Contour glucometer (Bayer).

Energy digestibility and indirect calometry.

Male C57BL/6JBomTac mice (n = 5 for each group) were adapted to experimental diets ad libitum for 4 days. Feces, urine, and feed residues were collected during a 24-h period before respiration measurements. The crude protein (N × 6.25) content in feed, feces, and urine was determined using the Tecator-Kjeltec system 1026 (Foss Tecator, Höganäs, Sweden). The gross energy content of feed and feces was determined using an adiabatic bomb calorimeter (System C700; IKA Analysentechnic, Heitersheim, Germany). An open-air circuit system (Oxymax; Columbus Instruments, Columbus, OH) consisting of two respiration chambers was used to measure 22-h respiratory gaseous exchange.

Plasma analyses.

Insulin, glucose (44), and lipid metabolites (39) were measured in plasma, as described previously. IL-6 (no. KNC0061; Invitrogen), TNFα (no. KMC3011; Invitrogen), adiponectin (no. EZMADP-60K; Millipore), and leptin (RD291001200R; BioVendor, Brno, Czech Republic) were measured using ELISA kits.

Tissue lipid analyses.

Lipid classes (39) and fatty acid composition (2) were analyzed as described previously.

Histology.

Sections of adipose tissue were fixed, dehydrated, embedded in paraffin blocks, cut into 3-μm-thick sections, and stained with eosin and hematoxylin, as described previously (4). Sections were visually examined using an Olympus BX 51 binocular microscope (Olympus, Tokyo, Japan) fitted with an Olympus DP50 3.0 camera.

Quantitative reverse transcriptase PCR.

RNA purification and quantitative reverse transcriptase PCR (qRT-PCR) were performed as described earlier (45). Primers for qRT-PCR were designed using Primer Express 2.0 (Applied Biosystems, Foster City, CA) and are presented in Table 3.

Statistical analyses.

All data represent means ± SE or means + SE. All data sets were tested for homogeneity of variances using Levene's test. Data were then analyzed using ANOVA post hoc pairwise comparisons. Homogenous data sets were analyzed using a Tukey HSD test. Nonhomogenous data sets from qRT-PCR analyses were log-transformed and retested for homogeneity of variances using Levene's test. Data sets that remained nonhomogenous were analyzed using the Kruskal-Wallis test. The use of a Tukey HSD test following log transformation or a nonparametric test is specifically mentioned in figure and/or table legends. A value of P < 0.05 was considered statistically significant. The Statistica 9.0 software (StatSoft, Tulsa, OK) was used for statistical analyses.

RESULTS

Sucrose dose-dependently counteracts the obesity-protective effect of fish oil.

To investigate whether sucrose dose-dependently counteracted the obesity-protective effect of fish oil, male C57BL/6J mice were fed isocaloric high-fat diets with different sucrose/protein ratios (Table 1) for 6 wk. Earlier studies have demonstrated that a low sucrose/protein ratio slightly reduced feed intake in mice (40). To secure equal energy intake, all groups were pair fed. In this particular case, all groups received the same amount of feed as that consumed by the mice receiving the lowest sucrose/protein ratio. Body weight gain and white adipose tissue (WAT) mass increased in parallel with the increase in the sucrose/protein ratio in the feed (Fig. 1, A, B, and E). The weights of iBAT, liver, and tibialis anterior muscle were not influenced by the different feeds (not shown). We have demonstrated earlier that increasing the amount of sucrose in the diet increased the respiratory exchange ratio (RER) but did not significantly reduce O₂ consumption (40). However, weight loss during 24-h feed deprivation was reduced by increasing the amount of sucrose in the feed (Fig. 1C). Because total energy intake was kept equal, energy efficiency increased dose dependently in response to the increased amount of dietary sucrose (Fig. 1D).

Expression of inflammatory markers such as Serpine1 [serine (or cysteine) peptidase inhibitor, clade E, member 1] or plasminogen activator inhibitor-1 (PAI-1), Ccl2 [CCL2 chemokine (C-C motif) ligand 2, MCP1], and Cd68 (CD68 antigen) was increased in both eWAT and iWAT in the obese mice, suggesting that sucrose attenuated the anti-inflammatory effect of fish oil (Fig. 1F). Expression of Emr1 (epidermal growth factor-like module containing, mucin-like, hormone receptor-like sequence 1, F4/80) was increased significantly in iWAT only (Fig. 1F). The expression of Pparg [peroxisome proliferator-activated receptor-γ (PPARγ2)], Srebf1 [sterol regulatory element-binding transcription factor 1c (SREBP1c)], and Fasn (fatty acid synthase) did not change significantly (Fig. 1F).

In iBAT, Ucp1 (uncoupling protein 1) expression was similar in all groups (Fig. 1G), but the expression of Ucp1 was significantly lower in iWAT in mice fed the high amount of sucrose compared with mice fed the low amount of sucrose (Fig. 1F). This indicates that adipocytes in iWAT from mice receiving a low amount of sucrose and a high amount of protein had a more brownish phenotype.

Sucrose dose-dependently reduced expression of Ppargc1a (peroxisome proliferator-activated receptor-γ, coactivator 1α), Pck1 (phosphoenolpyruvate carboxykinase 1, cytosolic), and Agxt (alanine-glyoxylate aminotransferase) in the liver (Fig. 1H). Expression of the lipogenic gene Fasn was increased, whereas expression of enzymes involved in fatty acid oxidation such as Acox1 (acyl-CoA oxidase 1, palmitoyl) and Cpt1a (carnitine palmitoyltransferase 1a) was unchanged when the sucrose/protein ratio was increased (Fig. 1H).

The glucose moiety of sucrose is responsible for the obesity-promoting effect of sucrose.

To investigate whether the obesity-promoting effect of sucrose fed in combination with fish oil depended on the glucose or fructose moiety of sucrose, we prepared diets where fish oil was combined with sucrose, glucose, or fructose (Table 2). Male C57BL/6J mice were fed isocaloric high-fat diets with different carbohydrate sources ad libitum. After 8 wk, mice receiving fish oil in combination with fructose had gained less weight than mice receiving fish oil in combination with sucrose or glucose (Fig. 2A). Energy intake was not significantly different (Fig. 2B), and hence, energy efficiency was significantly lower in mice fed the fructose-supplemented diets (Fig. 2B). Calculation of digestibility demonstrated a minor but not significantly reduced digestibility of protein and fat in fructose-fed mice (Fig. 2C). Indirect calorimetric measurements were performed in a second set of mice. These measurements revealed that both O₂ consumption and CO₂ production were similar in all groups in both the fasted and the fed state (Fig. 2D). Still, the mice receiving fructose had ∼50% less white adipose tissue mass, iWAT, pWAT, and eWAT than mice receiving sucrose or glucose combined with a tendency toward a slight decrease in the weight of the tibialis anterior muscle (Fig. 2E).

We have previously provided evidence that the different obesogenic effect of corn oil fed in combination with either protein or sucrose is related to the effect of the macronutrient composition on hormonal status (44). Because sucrose and glucose, unlike fructose, stimulate insulin secretion, we measured plasma levels of glucose and insulin in the fed state. As expected, plasma glucose and insulin were lower in fructose-fed mice (Fig. 2F). Similar to mice fed a high-fat diet in combination with proteins, expression of Ppargc1a and Pck1 in the liver was increased (Table 4). However, unlike mice fed a high-protein diet, expression of genes involved in amino acid degradation, such as Pck1 and Agxt, was not increased in fructose-fed mice (Table 4). Plasma levels of 2-hydroxybutyrate, a marker for fatty acid β-oxidation, were similar in all groups (Fig. 2F), but the expression of genes involved in hepatic fatty acid oxidation, such as Acox1 and Cpt1a, was reduced in livers from fructose-fed mice (Table 4). Hepatic expression of lipogenic genes was higher in fructose-fed mice (Table 4). Importantly, however, excess lipid accumulation was not seen in liver or tibialis anterior muscle (Fig. 3A).

Expression of genes involved in lipid uptake and triacylglycerol uptake was not significantly different in WAT (Fig. 3C). The finding that Ucp1 expression was strongly induced in iWAT (Fig. 3C) but not iBAT (Fig. 3B) in fructose-fed mice indicates that energy, as observed in protein-fed mice, may be dissipated in the form of heat. Indeed, a higher expression of markers for brown adipocytes such as Ppargc1a and Cyt COXII (cytochrome c oxidase subunit II) suggests a higher number of brown-like adipocytes in iWAT in fructose-fed mice (Fig. 3C).

Circulating levels of leptin were positively correlated with adipose tissue mass and are known to stimulate the production of TNFα and IL-6 (56). As expected, expression of Lep (leptin) decreased in WAT, and circulating levels of leptin were lower in mice fed fructose than in mice fed sucrose (Fig. 3, C and D). However, neither expression nor circulating levels of TNFα and IL-6 were significantly different between the two groups of mice (Fig. 3, C and D). Surprisingly, circulating levels of adiponectin, an anti-inflammatory adipokine inversely correlated with the state of obesity (56), were also lower in fructose- than in sucrose-fed mice (Fig. 3D). WAT expressions of Ccl2 involved in monocyte recruitment and macrophage marker genes Cd68 and Emr1 were significantly higher in mice fed fish oil in combination with glucose or sucrose than fructose (Fig. 3C). Thus, the glucose moiety of sucrose appears to be responsible for the ability of sucrose to attenuate the anti-inflammatory effect of fish oil.

Because fructose feeding is frequently used to induce glucose intolerance in rats (32, 70) and infiltrating macrophages are causally linked to the development of glucose intolerance (10, 28), a GTT was performed in mice fed the sucrose-, glucose-, and fructose-based diets. Of note, during GTT, blood glucose levels reached higher concentrations in glucose compared with fructose-fed mice, and the area under the curve was significantly higher in glucose than in both sucrose- and fructose-fed mice, although fasting glucose levels in the glucose-fed mice did not differ from that of fructose-fed mice (Fig. 3, E and F).

High-GI starch increases the obesogenic potential of fish oil.

To further investigate whether other types of carbohydrates, such as starch with different capacity to stimulate insulin secretion, are able to modulate the antiobesogenic potential of fish oil, we prepared isocaloric diets (Table 2) where fish oil was combined with high or low GI starches (100% amylopectin vs. 60% amylose/40% amylopectin) previously demonstrated to induce the expected differences in postprandial blood glucose levels in a meal tolerance test when combined with a high-fat diet (13, 57). Male C57BL/6J mice were fed the isocaloric high-fat diets with different carbohydrate sources ad libitum for 8 wk. In agreement with earlier findings (57, 65), where mice were fed high- and low-GI starches in low-fat diets, body weights were similar, but adipose tissue mass was lower in mice receiving fish oil in combination with low- compared with high-GI starches (Fig. 4, A and C). Energy intake was also similar, and no significant difference in energy efficiency was found (Fig. 4A). However, mice fed fish oil in combination with the low-GI starch had significantly lower levels of plasma insulin and less WAT (Fig. 4, B and C). The reduced adiposity observed in mice receiving fish oil in combination with the low-GI starch was not due to decreased digestibility, because the digestibility of the diet containing the low-GI starch was slightly higher than the digestibility of the diet containing the high-GI starch (Fig. 4D). In the fed state, mice fed fish oil in combination with the low-GI starch consumed more O₂ and produced more CO₂ than mice fed the high-GI starch (Fig. 4F). RER was not significantly different (Fig. 4F), and expression of genes involved in fatty acid oxidation [Acox1 and Hmgcs2 (3-hydroxy-3-methylglutaryl-coenzyme A synthase 2)], gluconeogenesis (Pck1), and amino acid degradation and urea synthesis [Agxt, Got1 (glutamate oxaloacetate transaminase 1, soluble) and Cps1 (carbamoyl-phosphate synthetase 1)] in the liver was similar in the two groups of mice (Table 5). However, expression of the lipogenic genes Scd1 (stearoyl-coenzyme A desaturase 1) and Fasn was reduced in mice fed low-GI starch (Table 5). Expression of Ucp1 was similar in iBAT, but expression in iWAT in mice fed the low-GI starch was increased (Fig. 4, E and G). Expression of genes involved in lipid uptake and triacylglycerol syntheses was not changed significantly (Fig. 4G).

As expected, expression of Lep (leptin) increased in WAT, and circulating levels of leptin were higher in mice fed the high-GI starch than in mice fed the low-GI starch (Fig. 4, G and H). Expression levels of Il6 was higher in eWAT in mice fed high-GI starch than in mice fed low-Gi starch (Fig. 4G), but circulating levels of IL-6 were similar (Fig. 4H). Neither expression levels nor circulating levels of TNFα were significantly different in the two groups of mice (Fig. 4, G and H). Moreover, both expression and circulating levels of the anti-inflammatory adipokine adiponectin were similar (Fig. 4, G and H). However, WAT expressions of the macrophage marker genes Cd68 and Emr1 were significantly higher in mice fed fish oil in combination with high-GI starch than fructose low-GI starch (Fig. 3C), suggesting that infiltration of macrophages is related to the state of obesity. Altogether, results from the experiments combining fish oil with sucrose, glucose, or fructose as well as high- or low-GI starches suggested that stimulation of insulin secretion decreased the antiobesogenic effect of a diet enriched with dietary fish oil.

The obesity-promoting effect of sucrose in combination with fish oil is associated with increased insulin secretion.

To investigate whether increased insulin secretion is able to decrease the antiobesogenic effect of fish oil when the carbohydrate load is low, we fed male C57BL/6J mice the high-fish oil 13% sucrose diet (Table 1) or a standard low-fat diet (Table 2) ad libitum. The sulfonylurea glybenclamide was used as an insulin secretagogue (21, 74). Body weight gain (Fig. 5A) and energy efficiency (not shown) were not affected by glybenclamide, but the amount of white adipose tissue mass tended to increase, albeit not significantly (Fig. 5B). Moreover, the mean average diameter of the adipocytes (Fig. 5B) was not increased by glybenclamide in either eWAT (52 ± 10 vs. 45 ± 12 μm) or iWAT (35 ± 6 vs. 38 ± 4 μm). However, expression of Pparg was increased significantly in both eWAT and iWAT (Fig. 5C). Because insulin levels in the fed state tended only to increase, a GTT was performed to validate whether the dose of glybenclamide used was sufficient to increase insulin levels. After 6-h feed deprivation, fasting glucose levels were similar, but as expected, glucose tolerance was improved by daily injections of glybenclamide (Fig. 5, E and F). This was not accompanied by reduced expression of markers for adipose tissue inflammation (Fig. 5C). Actually, expression of Serpine1, Ccl2, and Cd68 was higher in iWAT from glybenclamide-treated mice (Fig. 5C). In accord with unchanged energy efficiency, Ucp1 expression was not changed significantly. However, glybenclamide treatment reduced hepatic expression of Pck1 and genes involved in amino acid degradation and ureagenesis [glutamic pyruvic transaminase, soluble (Gpt1), Got1, Agxt, and Cps1; Fig. 5D]. Together, these results demonstrate that increased insulin secretion associated with glybenclamide treatment is unable to significantly decrease the antiobesogenic effect of fish oil when the carbohydrate load is low.

To investigate whether inhibition of insulin secretion could attenuate the adipogenic effect of sucrose in combination with fish oil, we fed C57BL/6J mice the high-fish oil/high-sucrose diet (Table 2) with or without nifedipine supplementation. Nifedipine was included in the diets at a dose of 1 g/kg, which was demonstrated earlier to reduce plasma insulin levels in agouti mice (37). A third group of mice received a standard low-fat diet (Table 2). To limit possible adverse effects caused by the decreased insulin secretion, feeding time was limited to 4 wk. Inclusion of nifedipine did not influence body weight gain (Fig. 6A) or feed intake (not shown). Consequently, energy efficiency was not different (not shown). As expected, inclusion of nifedipine in the sucrose-enriched high-fish oil diet resulted in lower levels of plasma insulin in the fed state (Fig. 6B). Importantly, the mice receiving the nifedipine-supplemented fish oil and sucrose-enriched diet had lower adipose tissue mass. The adipocytes had a normal appearance, but nifedipine attenuated the increased average diameter of the adipocytes in eWAT induced by the fish oil diet supplemented with sucrose (Fig. 6C). When nifedipine was added to a standard low fat diet, no effect on adipose tissue mass was observed (not shown). In line with unchanged energy efficiency, we did not detect any increased expression of Ucp1 in iWAT in nifedipine-treated mice (Fig. 6D). Nifedipine did not alter plasma levels of free fatty acids, triacylglycerol, glycerol, or 2-hydroxybutyrate in the fed or fasted state (Fig. 7A). Moreover, nifedipine did not increase hepatic expression of Pck1 or genes involved in lipogenesis or amino acid degradation (Fig. 7B). Thus, inhibition of insulin secretion was able to attenuate the obesogenic effect of sucrose in combination with fish oil but did not reduce energy efficiency. Blood glucose levels after 6-h feed deprivation was normalized by nifedipine, but nifedipine did not improve the reduced glucose tolerance observed after feeding fish oil in combination with sucrose (Fig. 6E). However, the ITT indicated that insulin sensitivity was improved (Fig. 6F). Together, our data indicate that inhibition of insulin secretion attenuated the adipogenic effect of sucrose in combination with fish oil. However, inhibition of insulin secretion was insufficient to reduce energy efficiency.

DISCUSSION

We have described previously that sucrose counteracted the anti-inflammatory effect of fish oil in adipose tissue and increased obesity development in mice (40). Here, we show that the glucose moiety of sucrose was responsible for this effect, and this effect of sucrose can be mimicked by high-GI starch. We further show that a high-fat diet combined with high-GI carbohydrates actually promoted obesity even if the diet contained 18 weight% fish oil. Thus, the background diet critically influenced the ability of fish oil to curb obesity development.

Several lines of evidence have suggested that increased insulin secretion plays a crucial role in the obesity-promoting effect of high-fat diets containing sucrose, glucose, or high-GI starch. Dietary sucrose via the glucose moiety, as well as high-GI starch, stimulates secretion of insulin from pancreatic β-cells, and inhibition of insulin secretion by administration of nifedipine attenuated the adipogenic effect of fish oil in combination with sucrose. Insulin is an important driver for adipocyte differentiation (41). Mice lacking insulin receptors in adipose tissue (FIRKO mice), are protected against obesity and remain glucose tolerant on a high-fat diet (9), and white adipose tissue mass is reduced dramatically in Irs1^−/−/Irs2^−/− double-knockout mice (48). Furthermore, weight gain is a well-recognized side effect of type 2 diabetic drugs that increase insulin sensitivity (25, 49). However, an increase in insulin secretion alone was insufficient to promote obesity development because mice receiving glybenclamide in combination with proteins and fish oil did not become obese. This finding is in keeping with the observation that a high-fat diet is unable to increase adipose tissue mass in the absence of carbohydrates (47, 50). Together, these observations suggest that hyperinsulinemia is a contributing factor to the development of obesity, and reducing hyperinsulinemia would possibly counteract obesity development.

The obesity-promoting effect of increased insulin secretion is supported further by our finding that glucose, but not fructose, is the obesity-promoting moiety of sucrose. Unlike sucrose and glucose, fructose does not stimulate insulin secretion from pancreatic β-cells (70), and mice receiving fish oil in combination with fructose had less adipose tissue mass than mice receiving fish oil in combination with glucose or sucrose. However, because the mass of adipose tissues in sucrose-fed mice was as high as in glucose-fed mice, it is not likely that fructose is able to counteract the obesity-promoting effect of glucose. Mice receiving fructose in combination with fish oil were also less glucose intolerant than mice fed fish oil in combination with sucrose or glucose. Fructose is commonly used to induce glucose intolerance in rats, and some mice strains also develop metabolic syndrome in response to high-fructose feeding (53). However, in C57BL/6J mice neither hyperinsulinemia nor hyperglycemia developed in response to high-fructose feeding (53).

Our finding that inclusion of sucrose abolished the antiobesity effect of fish oil seems to be at variance with a recent study from Sato et al. (63), because these authors demonstrated that inclusion of 5 g/kg of the (n-3) PUFA EPA into a high-fat/high-sucrose diet reduced body weight gain in mice. The reason for this discrepancy is not clear, but differences in the dietary compositions may account for the different results. Although the amount of n-3 PUFA used in this study, 6 g/kg n-3 fatty acids, is comparable with the 5 g/kg EPA used by Sato et al. (63), they used purified EPA ethyl ester, whereas the n-3 PUFAs used in our study comprised a mixture (32 ± 3 and 18 ± 3 g/kg EPA and DHA, respectively) in the form of triacylglycerols. This is a significant difference, because it is well established that both the metabolism and the metabolic effects of EPA supplementation differ from those of DHA, although the underlying molecular mechanisms remain to be fully elucidated (7, 15, 26, 29, 43, 46). Furthermore, because the kinetics of absorption of fatty acids from triacylglycerols and ethyl esters in the gut differs (51), we cannot exclude that this difference may contribute to the observed differences between the two studies. It should also be mentioned that the main fat source in the diets used in our study is corn oil, which is rich in n-6 fatty acids, whereas Sato et al. (63) used milk fat containing >60% saturated fat. Moreover, the highest dose of sucrose used in our study was higher than that used by Sato et al. (63). Expression of Ucp1 in iWAT decreased dose-dependently with increasing concentrations of sucrose (Fig. 1F), which would suggest that expression of Ucp1 in iWAT in our experiments was lower than in the experiments of Sato et al. (63). This becomes of particular importance because our experiments, in contrast to those of Sato et al. (63), were conducted at thermoneutrality, where lower or complete lack of Ucp1 expression, particularly in white adipose tissues, promotes adipose tissue mass expansion and obesity, which is not observed at temperatures below thermoneutrality (11, 22, 42).

Obviously, increased adipose tissue mass is related to energy intake. However, macronutrient composition can influence energy efficiency in such a way that mice consuming the same amount of calories end up with quite different amounts of adipose tissue. Thus, increasing the amount of sucrose in the feed from 13 to 43% led to approximately fivefold higher energy efficiency. In mice fed increasing amounts of protein at the expense of sucrose, we observed a dose-dependent increase in Ucp1 expression in iWAT, suggesting increased uncoupled respiration and dissipation of energy in form of heat. We and others have demonstrated recently that cyclooxygenase-dependent induction of Ucp1 expression in WAT counteracts diet-induced obesity (42, 71). Additionally, we observed a dose-dependent increase in Ppargc1a and Pck1 expression in response to a reduced intake of dietary sucrose, suggesting increased hepatic gluconeogenesis (23). Moreover, reduced sucrose combined with increased protein content in the feed would increase the energy cost associated with amino acid catabolism. Thus, UCP1-dependent uncoupled respiration in iWAT in combination with increased energy cost from gluconeogenesis and ureagenesis may account for the reduced energy efficiency that was observed when sucrose levels in the diet were low.

Obesity is associated with a state of chronic low-grade inflammation, and following the onset of obesity, secretion of proinflammatory adipokines is increased. The anti-inflammatory effect of fish oil in adipose tissue is well described (33, 36, 55, 69). Thus, it was remarkable that mice receiving fish oil in combination with sucrose, glucose, or high-GI starch had increased expression of macrophage marker genes in adipose tissue. In agreement with the positive correlation between adipose tissue mass and circulation levels of leptin, mice fed fish oil in combination with sucrose, glucose, or high-GI starch had increased WAT expression and circulating levels of leptin. Although circulating levels of inflammatory markers such as IL-6 and TNFα were not affected by the dietary carbohydrate source, obese mice had higher expression of macrophage markers in adipose tissue. This is in line with our earlier study where we showed that obesity rather than the n-3:n-6 PUFA ratio in both feed and adipose tissues appeared to determine the expression levels of macrophage markers in adipose tissue (40). Of note, expression of macrophage markers in the obese state was as prominent in subcutaneous iWAT as in eWAT. This is in contrast to the general notion that visceral adipose tissues in obese individuals exhibit a more inflamed state than subcutaneous adipose tissue, which was reviewed recently by Blüher (8). However, a few recent reports have demonstrated that subcutaneous adipose tissues compared with visceral adipose tissue in certain circumstances may express similar or higher levels of inflammatory markers (19, 72), and our observations add to this list.

To summarize, we demonstrate that the background diet exerts a crucial influence on the ability of fish oil to protect against obesity development and adipose tissue inflammation. Thus, it cannot be excluded that several additional beneficial effects of fish oil intake might be diminished or completely abrogated by a simultaneous intake of high-GI carbohydrates. If similar effects are found in humans, this is of great concern because the intake of refined sugars from sources such as soft drinks has increased dramatically during the last several decades (14). Moreover, n-3 supplements are often taken in combination with morning meals containing high-GI carbohydrates such as cereal, bread, and orange juice. Thus, comprehensive studies of the interaction between dietary macronutrients and fish oil in humans seem warranted.

GRANTS

This work was supported by the Danish Natural Science Research Council, the Novo Nordisk Foundation, the Carlsberg Foundation, and NIFES. Part of the work was carried out as a part of the research program of the Danish Obesity Research Centre (DanORC). DanORC is supported by the Danish Council for Strategic Research (Grant no. 2101-06-0005).

DISCLOSURES

The authors have no conflicting interests, financial or otherwise.

AUTHOR CONTRIBUTIONS

Q.H., H.H.L., E.F., L.S.M., L.K.M., R.J., T.M., B.J., and R.K.P. performed the experiments; Q.H., H.H.L., E.F., L.S.M., L.K.M., R.J., T.M., B.J., and R.K.P. analyzed the data; Q.H., H.H.L., E.F., L.S.M., L.K.M., R.J., T.M., B.J., R.K.P., S.B.S., A.C., L.F., B.L., K.K., and L.M. interpreted the results of the experiments; Q.H., H.H.L., E.F., L.S.M., L.K.M., and L.M. prepared the figures; Q.H., H.H.L., E.F., R.K.P., S.B.S., A.C., L.F., B.L., K.K., and L.M. edited and revised the manuscript; Q.H., H.H.L., E.F., L.S.M., L.K.M., R.J., T.M., B.J., R.K.P., S.B.S., A.C., L.F., B.L., K.K., and L.M. approved the final version of the manuscript; B.L., K.K., and L.M. did the conception and design of the research; K.K. and L.M. drafted the manuscript.