Ever since Linda Buck and Richard Axel's 1991 discovery that rat olfactory receptors (ORs) are seven transmembrane molecules that are part of the G-protein coupled receptor family – for which they won the Nobel Prize in 2004 – it has been generally assumed that all ORs show the same conformation.

Over the past few years, work from Leslie Vosshall's laboratory at Rockefeller University has suggested that, unlike vertebrates, insects share a common protein – Or83b – which is necessary for ORs to be integrated into the membrane. Now, in an elegant and thorough paper on Drosophila, Richard Benton and co-workers from Vosshall's group have not only shown how Or83b and OR proteins interact to produce a functional receptor, they have also challenged the idea that insect ORs are G-protein coupled receptors and have effectively turned the receptor molecule inside out. Their findings, if confirmed in other insects, will radically alter our understanding of the evolution and function of olfaction.

In a technical tour de force, the authors make use of the whole toolbox of Drosophila genetics. Nevertheless, the results are presented in a straightforward and clear way, with clear summaries at the end of each experiment that allow even those allergic to molecular genetics to follow the argument and the evidence.

First, Benton and co-workers showed that, in the absence of Or83b, OR molecules are unable to leave the endoplasmic reticulum. They then used the GAL80^ts transgene to control Or83b expression in the growing or adult fly and found that when Or83b expression was `turned on' when the insects were 10 days old, the flies were able to begin expressing ORs in the receptor neuron membrane, and when Or83b was turned off at 3 days old, ORs gradually disappeared from the fly antenna, showing that olfactory receptors are being continually trafficked to the membrane surface throughout life.

To show that Or83b and ORs do in fact associate, the authors then made Or83b and OR transgenes fused to the N- or C-terminal fragments of a Yellow Fusion Protein, knowing that they would get a fluorescent signal if the receptors became associated. Sure enough, the receptors interacted, the molecule flouresced and further studies showed that this heteromer is functional.

A bioinformatics study of the predicted structure of Drosophila ORs led to the biggest surprise: the N-terminal appeared to be intracellular, rather than extracellular as in mice. This meant that the six loops of the seven transmembrane structure changed position compared with the textbooks: the bits that were thought to be outside now seemed to be inside, and vice versa. The authors tested this radical suggestion in a number of ways, including a stunning immuno-electron micrograph of a horizontal section of an antennal sensillum showing 5 nm colloidal gold `staining' of part of the OR sequence that they now predicted to be extracellular: the tiny dots were all on the outside of the dendritic membrane. The sequence and membrane topology of Drosophila ORs are very different to those of mammalian ORs.

Having provocatively demonstrated that Drosophila ORs are no more related to mouse ORs than mouse ORs are to ion channels, the authors then highlight the lack of direct evidence in the literature to demonstrate that insect ORs are in fact G-protein coupled receptors. This, coupled with the suggestion that the orientation of insect ORs is rather different to our previous understanding, will undoubtedly have ruffled some feathers amongst both scientists and textbook publishers. Buck and Axel's model has rightly become a prize-winning icon, but it might not be quite so generalised as we all thought.