Activity-based anorexia dysregulates the Irisin-BDNF neurometabolic axis in female rats: a muscle-to brain crosstalk

Background Anorexia nervosa (AN) is a life-threatening psychiatric disorder that mainly affects young females. Despite it shows a clear symptomatology, its aetiology is unknown, and treatments are ineffective. In physiological conditions, physical exercise exerts protective functions on cognitive processes mainly by enhancing BDNF expression in the hippocampus via the peripheral activation of PGC-1α/FNDC5/Irisin pathway from skeletal muscles. Since in AN, physical exercise turns to be pathological gaining compulsive features, our hypothesis is that AN might be driven, at least partially, by a dysfunctional muscle-to-brain crosstalk, which may drive weight loss seeking thus feeding AN maladaptive behaviours. Objective Our goal was to understand whether the exercise-induced hyperactivation of skeletal muscles in AN may alter brain mechanisms through a dysfunctional neurometabolic axis, mediated by the myokine Irisin, as a critical player in the severe enduring of AN phenotype. Methods We took advantage of the activity-based anorexia (ABA) rat model which consists in exposing adolescent female rats to the combination of food restriction and wheel access (from postnatal day (P) 38 to P42). At P35 rats were subdivided in four groups: controls (food ad libitum-sedentary), FR (food restricted-sedentary), EXE (food ad libitum- exercise) and ABA (food restricted-exercise). Animals were sacrificed at P42, at the acute phase, or at P49, after a 7-day recovery period. Trunk blood, soleus muscle, and hippocampus were collected. Gene and protein expression analyses were performed by means of real-time PCR, Western blot, and ELISA assays. Behavioural and molecular data were analysed by two-way ANOVA followed by Bonferroni or Tuckey post hoc tests. Results 24 hours after the AN induction, ABA animals show reduced body weight and exponentially increased wheel activity over days. At P42, exercise induced an increase of PGC1a levels in both EXE and ABA rodents (F (1, 20) = 31.14, p<0.0001) in the soleus muscle, while FNDC5 was increased only in ABA rats (+33% p=0.0130 vs EXE). This effect was paralleled by increased circulating levels of Irisin in ABA rats (exercise: F(1, 19) = 5.232, p=0.0338). However, in the hippocampus, only the EXE group showed an increase in mBDNF (+28% p=0.0034 vs ABA;) and pTrkBY706 (+49% p<0.0001 vs ABA) protein levels, suggesting that the neuroplastic effect of physical activity was lost in ABA rats. Interestingly, after recovery, food restriction reduced PGC1a levels in the soleus (F(1, 20) = 10.79, p=0.0037) and Irisin plasma levels (F(1, 20)=6.736, p=0.0173), while exercise reduced FNDC5 in the soleus (F (1, 20) = 14.23, p=0.0012). In the hippocampus, the ABA-induced reduction in FNDC5/Irisin levels were mirrored by a reduction of mBDNF (F(1, 20)=11.70, p=0.0027) and pTrkBY706 (F(1, 20)=12.71, p=0.0019), suggesting a long-lasting reduced trophic support, mainly induced by the caloric restriction. Conclusion These data suggest that the induction of AN phenotype leads to an altered muscle to brain crosstalk via dysregulated PCG1-α/FNDC5/Irisin/BDNF neurometabolic pathway, even when body weight is restored. This maladaptive plasticity may represent a signal of altered processing of food reward contributing to the severe course of the disease, and to the long-term vulnerability to relapse. Sponsored by: Cariplo foundation 2023-1003