Macronutrient Composition and Genetic Background Determine the Response to a Ketogenic Diet
While standard high fat diets cause hyperphagia and obesity in mice, high fat-low carbohydrate ketogenic diets (KDs) reduce food intake and body weight. Because the basis for this difference is still unclear, we systematically altered the macronutrient content of a standard KD and found that feeding C57BL/6J (B6J) mice a KD with 5% protein resulted in hypophagia, weight loss, and hypoglycemia, whereas the same diet with 10% protein led to increased adiposity and glucose intolerance. However, these effects were strain-dependent as C57BL/6NJ (B6NJ) weighed similar amounts on the two diets leading us to investigate the molecular mechanisms. When fed the KD-5% diet, B6J but not B6NJ mice showed increased levels of two anorexigenic factors, GDF15 and LCN2, and loss of function of either blunted the weight loss of B6J mice fed the diet. B6J mice harbor mutations in Nnt (Nicotinamide nucleotide transhydrogenase) and Nlrp12 (NLR family pyrin domain containing 12), both of which are wildtype in B6NJ mice. B6J mice fed the KD-5% diet showed the RNA signature of oxidative and integrated stress responses (ISR) and restoring NNT function in liver reduced the levels of GDF15. RNA-seq also revealed that B6J but not B6NJ mice had the RNA signature for hepatic inflammation and a knockout of Nlrp12 led B6NJ mice to lose weight on the KD-5% diet with increased levels of LCN2. Suppression of oxidative stress with N-acetylcysteine (NAC) reduced expression of both GDF15 and LCN2 and prevented the weight loss associated with the KD-5% protein diet in B6J mice, whereas inhibition of the integrated stress response with ISRIB only attenuated the GDF15 axis. Collectively, these findings explain why B6J mice lose weight on a ketogenic diet and reveal a critical interplay between macronutrient composition and genetic background leading to increased levels of GDF15 and LCN2 to induce hypophagia. Finally, these data suggest that the response to different diets among humans might be similarly variable based on genetic variation and macronutrient composition, suggesting the possible need for personalized dietary interventions.