Discussion
Our original study describes some effects of systemic administration of anesthetic drugs on hypothalamic arcute nucleus peptide mRNA expression, in association with modified feeding behavior. However, hypnotic-induced effects depend on the drug tested. Some results of this study deserve comments. The lack of randomization and the potential for bias must be taken into account. The choice of hypnotic doses used in our work was based on the doses known to induce anesthetic effects in mice [
15]. All anesthetics used induced similar hypnotic effects on mice with comparable recovery time, likely after 30 min.
We observed increased appetite in mice following midazolam administration. Benzodiazepines, to which midazolam belongs, are drugs currently used for anxiolytic or sedative effects and have been shown to stimulate appetite in both rats and humans independently from anxiolytic effect [
9,
10,
12]. They exert their action through potentiation of the activity of the γ-aminobutyric acid type A receptor (GABA
AR) [
16], which is known to promote feeding behavior. Indeed, Stratford et al. showed that micro-infusion of muscimol (a GABA
AR agonist) near lateral ventricules and into the nucleus accumbens shell, induced strong feeding behavior in rats [
17]. In addition, it has been demonstrated that administration of flumazenil, a specific antagonist of the benzodiazepine site on the GABA
AR, abolishes hyperphagic response to intracerebroventricular administration of midazolam [
10]. This suggests that the orexigenic effect of midazolam can be mediated by the GABA
AR. Accordingly, hypothalamic POMC and NPY neurons are both able to synthesize GABA and GABA
ARs, which appear to be present in both neuronal populations [
18‐
20]. Our study showed that midazolam inversely modifies peptide mRNA expression with a rise in NPY and a decrease in POMC expression. These data are in agreement with a previous report suggesting that benzodiazepine drugs inhibit POMC gene expression in the rat ARC [
21,
22].
Propofol also exerts anesthesia through stimulation of the GABA
AR [
16]. Furthermore, it has also been shown to stimulate appetite and feelings of hunger in the post-operative recovery period in humans [
7,
8,
11,
23]. Our results are in agreement with this observation because propofol administration markedly favored appetite in mice. This finding could be related to a decrease in anorexigenic POMC mRNA level in ARC. However, no effect was observed on NPY mRNA level, as previously observed in a clinical study by Grouzmann et al. (2000) in which propofol administration did not change NPY mRNA level in the cerebroventricular fluid during craniotomy [
7] but induced effective appetite in patients. Therefore, it is possible to hypothesize that the orexigenic effect of propofol could be related to inhibition of POMC hypothalamic anorexic pathway rather than NPY orexigenic stimulation.
Isoflurane is a soluble volatile GABAergic anesthetic currently used in general anesthesia [
24]. Currently no study to evaluate the effect of this drug on food intake can be found. Nevertheless, data comparing propofol with volatile anesthetics have highlighted altered food intake by volatile anesthetics compared with anesthesia using propofol [
7]. Our study shows that isoflurane did not significantly stimulate food intake and further induced a rise in mRNA level encoding POMC, known to be anorexigenic. Currently, no data could explain this result. The anesthetic effect of several drugs are mediated by GABA
AR exhibiting different subunit compositions depending on neuronal population or brain area [
16]. The hypothalamic neuronal population are equiped by different subtypes of GABA
AR [
24]. This suggests the anesthetic GABA
AR agonist could modulate food intake depending on subunits composition in some specific neuronal populations of ARC.
Thiopental also exerts anesthesia through the GABA
AR [
16]. Thiopental is known to induce an increase in food intake in rats at similar dosages to those in our study, but to a lesser extent than benzodiazepines [
12]. These results are in agreement with our data and may be related to a stimulation of the orexigenic NPY pathway.
Our data show that some drugs used during anesthesia exert changes in feeding behavior and this effect is associated with modulation of two interconnected populations of neurons (POMC and NPY) in hypothalamic ARC. One hypothesis that could explain these observations would be a direct hypothalamic stimulation by anesthetic drugs. The very short time between anesthetic administration and observed modulation of mRNA expression in ARC (less than one hour) supports this hypothesis of direct modulation. Thus, the orexignic effect observed in mice could be due, according to the anesthetic used, to stimulation of expression of NPY (by thiopental) or to a decrease in the expression of POMC (by propofol) or both (by midazolam). The absence of effect for halogenated volatile drug isoflurane could be due to the stimulation of expression of POMC in ARC. The discrepancy between halogenated and other drugs could be explained by different affinities for specific subunits composition of hypothalamic GABA
AR. Indeed, the GABA
AR is composed of 5 glycoproteic subunits. Several subtypes of subunits exist resulting in multiple possible compositions, each presenting a specific sensitivity to the different anesthetics. Thus, isoflurane effects are highly dependent from α subunit and this agent has an increased sensitivity to receptors comprising an α5 or α6 subunits; benzodiazepine effects are mainly dependent on α1 and α2 subunits; and intravenous propofol and thiopental effects appear to be primarily subject to the type of β subunit (for a complete review see [
25]). On the other hand, there is heterogeneity in the expression of the GABA
AR subunits within the brain, particularly in the hypothalamic areas including the NPY and POMC neurons of the ARC [
19]. Indeed, POMC neurons contain α 1, α2 and α3 subunits and NPY neurons contain α3 subunits, which could explain the effects of midazolam, while the presence of α5 subunits in the ventromedian nucleus of the hypothalamus, containing orexin receptors with neuronal projections to the ARC, could explain the effects of isoflurane [
19,
26]. Thus, the variable effect of halogenated drugs compared with other anesthetics could be related to this heterogeneity within key areas of food intake regulation. However, to date, no study has evaluated the in vivo binding of hypnotics to the subunits of the GABA
AR in the various areas of the brain to support or refute this hypothesis.
Although the ARC is the main centre for food intake regulation, other brain areas influence this function. Indeed, NPY and POMC neurons present numerous projections in adjacent areas such as the paraventricular nucleus, producing cortisol releasing hormone with anorexigenic proprieties, and the lateral hypothalamus producing orexin A and B with orexigenic proprieties. In particular, it has been demonstrated that activation of GABA
AR by a specific agonist (muscimol) in the accumbens nucleus induced activation of the orexin-producing neurons of the lateral hypothalamus and a decrease in the activity of the POMC neurons in the ARC, resulting in an increase in food intake [
27] . On the other hand, there is a social and environmental component to food intake. Thus, the limbic system and its reward system is involved in the regulation of food intake, in particular related to galanin-producing neurons, which themselves present projections towards the NPY neurons of the ARC. This neuropeptide has orexigenic properties with a stimulation of NPY production. For complete reviews see [
28,
29]. All these data demonstrate the complexity and interconnections between systems, making it difficult to identify a specific target for the effects of anesthetics. However, it appears that the ARC is at the crossroads of these systems and the presence of GABA
ARs make it a potential target for anesthetics. Another hypothesis for the effects observed in our study could be an indirect pathway of anesthetic with modulation of systemic factors involved in feeding behavior, such as ghrelin, insulin or leptin, which could, at a later stage, modulate the neuronal activity in ARC.
Despite these interesting results, transposition to humans, especially in a surgical context, is difficult. Indeed, anesthesia in surgical context is not limited to the use of a single anesthetic but to a combination, inducing complex and combined effects on many physiological variables, and therefore probably on the feeding behavior. In addition, surgical stress and pain are identified as altering the feeling of hunger. If our work does not provide full certainty on the choice of anesthetic, it deepens our understanding of the effects of gabaergic drugs on the complex system of food intake and opens up therapeutic possibilities for post-operative rehabilitation. Thus, we are awaiting the results of a randomized controlled study on the effects of propofol versus sevoflurane on post-operative feeling of hunger, which may provide additional elements of response (NCT02272166).