Anorexia nervosa (AN) is a serious disease characterized by decreased food intake and body weight, associated with excessive physical activity (PA) and a preference for protein-enriched diet (PED) that is known to decrease hunger. AN patients maintain euglycemia and minimal body weight, despite low energy intake and increased PA, suggesting that adaptations should take place at the level of energy metabolism and of endogenous glucose production (EGP). EGP is activated during fasting, with an increased contribution of renal (RGN) and intestinal gluconeogenesis (IGN). Since food restriction (FR) is featured by partial fasting state, we hypothesize a change in EGP characteristic of fasting in FR, i.e. increased RGN and IGN. Since IGN curbs hunger via a signal to the brain, we hypothesize that this could help to endure food restriction in AN. Since PED induces IGN, this could explain why AN patients adopt this kind of regimen. At last, since the portal glucose signal initiated by IGN induces food preference, we hypothesize that increased IGN could enhance the preference for PED in AN patients or FR mice (aim1). PA is matched with fasting and PED under several aspects. We hypothesize that PA should activate IGN, which could enhance the behaviour of PA via IGN action on the reward system (aim 2.) We will then raise the question of the adaptation of energy expenditure and metabolism in FR associated with physical activity, which might explain the plateauing of body weight in this situation (aim 3). Aims 1 to 3 will be assessed using a mouse model of FR/AN recently developed. In parallel, the hypothesis that food preference for PED and PA could be correlated will be assessed in a cohort of patients suffering from AN using dedicated questionnaires (aim 4). At last, the microbiota composition could have a role in the metabolic/behavioral adaptations in AN, and in food restriction and PA. his hypothesis will be assessed in our AN model of mouse and our cohort of patients (aim 5).