Orexins, feeding and the big picture.

The number of hypothalamic neuropeptides that have been implicated in the regulation of feeding has grown steadily over recent years (Inui, 1999). Many of these reduce food intake but some increase it. These are neuropeptide Y, Agouti-related protein (a natural antagonist of the melanocortin-4 receptor), melanin-concentrating hormone, galanin and the orexins. Neuropeptide Y and Agouti-related protein are more powerful stimulants of feeding than melaninconcentrating hormone, galanin or the orexins, but this does not preclude a role for the weaker stimulants: strong arguments can be made for all of them (Arch et al. 1999; Kalra et al. 1999). None of these peptides is exclusively involved in the regulation of feeding, however. Now 2 years after the first description of the orexins (Sakurai et al. 1998), significant new information allows us to review the role of the orexins and their receptors in feeding and place this in the context of their wider role. Orexins-A and -B were identified in extracts of rat brain and bovine hypothalamus by their ability to evoke transient elevations of intracellular Ca concentration in cells expressing an orphan 7-transmembrane, G-protein-coupled receptor, now known as the orexin-1 or OX1-receptor. Subsequently, a second (OX2) receptor was identified by its homology (64 % at the amino acid level) to the OX1receptor (Sakurai et al. 1998). Both functional and binding studies show that the OX1-receptor has greater affinity (about 10-fold) for orexin-A than -B, whilst the OX2receptor has similar affinity for the two peptides (Smart et al. 1999, 2000). Orexin-A is thirty-three amino acids long and orexin-B twenty-eight amino acids long. They are produced by processing of the same prepro-orexin peptide. Hypocretins were discovered independently and are sometimes said to be synonymous with orexins, but hypocretin-1 was originally deduced to have five more N-terminal amino acids than orexin-A and both hypocretins were shown in a figure of the original paper (Sakurai et al. 1998) as having C-terminal glycines. These are removed (as predicted in the text) leaving C-terminal amides in the mature peptides (de Lecea et al. 1998). Researchers should be aware that commercial hypocretins corresponding to those shown in the figure of de Lecea et al. (1998) are far less potent than orexins (Smart et al. 2000). The orexins were given their name because intracerebroventricular injection increased food intake in rats (Sakurai et al. 1998). A number of studies have reproduced these findings, but some find that orexin-B has little or no effect, or even that it may inhibit feeding (Haynes et al. 1999). The greater effect of orexin-A compared with -B appears to implicate the OX1rather than the OX2-receptor in the regulation of feeding, but it is equally possible that orexin-B has less effect because it is more rapidly cleared from the area of injection: peripherally administered orexin-B is known to be metabolised faster than orexin-A (Kastin & Akerstrom, 1999). Moreover if the problem in eliciting a feeding effect with orexin-B is simply that it is 10-fold less potent at the OX1-receptor, then it should be possible to obtain an effect by increasing the dose of the peptide. This is not what is found (Haynes et al. 1999). Orexin-A and -B are clearly inadequate tools with which to dissect the roles of OX1and OX2-receptors in vivo. Fortunately, there are now two tools which demonstrate a role for the OX1-receptor in mediating the feeding effect of orexin-A. SB-334867-A is an OX1-receptor antagonist that is about 30-fold selective relative to OX2-receptor antagonism and has very little affinity for a wide range of other receptors (Arch, 2000); there is also an OX1-receptor antibody (Smith et al. 2000). The antagonist has been shown to inhibit orexin-A-driven feeding when given intraperitoneally at doses of 3±30 mg/kg (Arch, 2000; Rodgers et al. 2000), suggesting that the OX1-receptor is at least partly responsible for the orexigenic effect of orexinA. It would, however, be unwise to exclude a role of the OX2-receptor without having investigated the effects of an antagonist of this receptor. Suppression of feeding in response to leptin is blocked by antagonists of the melanocortin-4, glucagon-like peptide-1 and corticotrophin-releasing-hormone receptors (Cone, 1999): all three of these pathways must be operational for leptin to have any effect. Similarly, enhancement of a complex behaviour like feeding may depend upon stimulation of both orexin receptors. It is one thing to demonstrate that a peptide can affect feeding when injected into the brain and quite another to show that it normally plays a role in the regulation of feeding. In support of orexins playing a physiological role, both the OX1-receptor antagonist and the antibody, as well as an antibody to orexin-A (Yamada et al. 2000) have all been reported to inhibit natural feeding. The sceptic might argue that disruption of the orexin system is merely disrupting behaviour in general: the rat might be driven to indulge in other behaviours that preclude feeding. However, it is notable that the normal behavioural satiety sequence (eating, grooming, resting) is preserved in rats treated with the antagonist; what happens is that the transition points between behaviours are advanced. Conversely, low doses of the agonist delay the transition point so that more time is spent in feeding (Rodgers et al. 2000). It can still be argued that the effect of orexins on feeding is secondary to an effect on another behaviour. For example, the primary effect may be to stimulate activity and reduce 401 British Journal of Nutrition (2000), 84, 401±403