We found that the macronutrient composition of a diet greatly affected the plasma metabolite profile of C57BL/6J mice when administered ad libitum. A ketogenic diet administered ad libitum promoted high glucose levels, weight gain, and a hyperlipidemic profile in mice, whereas fish-oil supplementation decreased bodyweight, glucose levels, and yielded a normolipidemic profile. Administration of these diets under CR led to weight loss, increased insulin sensitivity, decreased glucose levels, increased ketone levels, and promoted a normolipidemic lipid profile. When administered under CR, the influence of the composition of the diet on the metabolic profiles of the mice was minimized, yielding similar metabolic profiles across all groups. Thus, a restriction of total energy intake can correct weight gain or a hyperlipidemic profile from these dietary therapies in B6 mice.
We confirmed previous studies that the KD-UR does not lead to long-term weight loss in mice when they are fasted prior to diet initiation [13, 18, 76]. Our work supports the notion that “a calorie is not a calorie”, as the KD-UR group gained approximately 10% body weight despite consuming 10% less in metabolizable calorie intake, compared to the SD-UR. However, our results challenge the metabolic advantage theory of the KD-UR, which states that isocaloric diets low in carbohydrates lead to greater weight loss compared to isocaloric diets of different composition . This was unsurprising, given that dietary fat is metabolized more efficiently than are dietary carbohydrates and proteins . The seeming discrepancy between metabolizable calorie intake and body weight is most likely due to differences in energy utilization between the different dietary groups, including factors such as obligatory thermogenesis, microbial fermentation, diet induced thermogenesis, and basal metabolism (including energy expenditure) .
In addition to the increased body weight under the KD-UR, the plasma profile of the KD-UR group suggests that the diet is unhealthy [80–83]. The plasma profile of the KD-UR mice is similar to a previously published report utilizing a high-fat, high-sucrose diet in the same strain of mice . The atherogenic lipids triglycerides, cholesterol, and cholesteryl esters were highest in KD-UR group. A KD-UR diet also elevated these lipids in children with epilepsy . The presence of high circulating glucose levels in the KD-UR group, which has also been reported in other studies of rodents on high-fat, low-carbohydrate diets [84, 85], could further increase the risk of atherogenesis, as glucose can contribute to endothelial dysfunction [86, 87]. Glucose can be raised in high-fat diets devoid of carbohydrates with excess calorie intake through gluconeogenesis, which occurs indirectly through oxidation of fatty acids to acetyl-CoA . The KD-UR group also had the highest levels of PC, LPC, PE, and SM. It has been suggested that a PC to free cholesterol ratio (mol/mol) of less than 1 is a risk factor for atherosclerosis . The calculated PC/cholesterol ratios in all of the groups (both restricted and unrestricted) were similar, and ranged from 1.5-1.75 (data not shown). LPC levels have been shown to be positively associated with obesity . PE levels are suggested to be positively associated with atherosclerotic complications . SM in plasma is associated with an increased risk of coronary artery disease [92, 93]. Our data are consistent with the observations for LPC, PE, and SM. These findings, together with our observations in B6 mice, indicate that consumption of the KD in unrestricted amounts can have adverse effects on health-related biomarkers.
Administration of the KD in restricted amounts mitigated the hyperlipidemic plasma profile of the B6 mice. In addition to lowering triglycerides, cholesterol, phospholipids (PC, LPC, and PE), and glucose, the KD-R also lowered fasting insulin levels. Reduced levels of circulating insulin is antiatherogenic . Plasma ketone levels were also highest in the KD-R group. Ketones are therapeutic against a variety of neurological diseases and cancer [1, 48, 95–97]. It is well known that ketones can replace glucose as an energy metabolite and can protect the brain from hypoglycemia [16, 98, 99]. Our previous findings, in conjunction with the results in the present study, suggest that the therapeutic efficacy of the KD for epilepsy and brain cancer will likely be best when the diet is administered in restricted than unrestricted amounts [13, 18, 20].
While the KD-UR increased body weight and was associated with a hyperlipidemic plasma profile, we were surprised to find that the FO-UR diet caused changes in body weight and plasma metabolites that were similar to those seen in the CR groups. We cannot completely rule out that these results are a consequence of decreased food palatability in the FO-UR group, as the FO-UR group had reduced calorie intake compared to the SD-UR. The FO-UR group, however, consumed a similar number of calories as the KD-UR group after the fast (16.9 ± 2.3 (FO-UR) calories versus 17.0 ± 0.3 (KD-UR) calories) and throughout the study, and did not have observable aversion to the food. We hypothesized that increased levels of the adipokine adiponectin might be responsible for the increased insulin sensitivity and decreased body weight in the FO-UR group despite feeding ad libitum, since supplementation of fish-oil is associated with increased levels of adiponectin [100, 101]. Interestingly, we found that adiponectin was not elevated in the FO-UR group, suggesting that adiponectin by itself was not responsible for the increase in insulin sensitivity or the CR-like profile of the FO-UR mice. Omega-3 fatty acids and their metabolites can produce changes in the activity of peroxisome proliferator-activated receptor (PPAR) α and γ, along with directly binding to and modulating the activity of sterol receptor element binding proteins (SREBP), which contribute to modulating lipid and glucose metabolism [102, 103]. Omega-3 fatty acids and their metabolites (epoxide derivatives, docosanoids, resolvins, and neuroprotectins) reduce triglyceride levels, adiposity, and inflammation while improving insulin sensitivity in rodents and humans [104, 105]. Thus, the observed plasma profile of the FO-UR mice that mimicked the CR groups were likely due to a multitude of mechanisms regulated by omega-3 fatty acids that are still not fully elucidated.
All of the dietary groups were combined in a post-hoc analysis to examine the relationship among body weight, dietary intake, and plasma profile. Our findings indicated that blood glucose was the strongest predictor of body weight in mice and was highly correlated with body weight. This observation is consistent with previous studies in humans, though the human studies did not review dietary intake [106, 107]. We found that intake of specific macronutrients was not predictive of final bodyweight. This observation questions the utility of advocating for diets devoid in specific macronutrients to control body weight [7, 108]. In terms of predicting glucose levels, total calorie intake was only moderately predictive suggesting that multiple factors modulate blood glucose levels in mice. Our findings suggest that restriction of total calorie intake will be more effective for reducing glucose levels and body weight in mice than will be the restriction of any particular macronutrient under ad libitum feeding. The reduction of total calories was strongly predictive of elevated ketone levels. Reduced carbohydrate intake contributed moderately to increased ketone levels. From a physiological perspective, given that the body has a large reservoir of lipid stores and a small reservoir of glycogen stores, it should not be surprising that restriction of energy intake and carbohydrates will quickly diminish glycogen stores, leading to beta-oxidation of lipids for energy . Dietary strategies that promote beta-oxidation of lipids, through either restriction of overall energy intake or severe carbohydrate restriction, would promote ketosis. Since circulating glucose strongly predicted plasma triglyceride levels, dietary strategies that lower blood glucose levels should have significant effects on promoting cardiovascular and overall health [110, 111].