doi:10.1038/nindia.2008.203 Published online 20 May 2008
The debate over how humans and animals detect smells is old. The earliest influential writings on this matter were by Roman philosopher Titus Lucretius Carus (c99-55BCE), who suggested that "one odour is more apt, to others (than) another because of differing forms of seeds and pores…". These observations laid the ground for thinking on the chemical conformations of odorants (seeds) and odor-activated channels (pores). Interest in odor perception and its tantalizing link to memory was fueled by perfumers and writers alike. Marcel Proust noted that the "smell and taste of things remain poised as tiny and impalpable drops of their essence in the immense edifice of memory".
In the late sixties, Seymour Benzer of Caltech, showed that mutating genes could be used as a powerful means to dissect complex behavioral patterns. Encouraged by the success of these approaches, Obaid Siddiqi at the Tata Institute of Fundamental Research in Mumbai undertook a neurogenetic approach to study the mechanism of olfaction.
Over the late seventies and eighties, the Siddiqi group designed several tests to measure olfaction in larvae and adults which were elegant in their simplicity, allowing measurement of olfactory behavior in normal flies and the isolation of a number of mutants that have lost their ability to smell (anosmic). Several of these mutations caused partial defects. For example, mutants were defective in their ability to detect aldehydes but showed normal responses to all other odorants. At around the same time, mutants showing defects in ability to smell specific chemicals (partial anosmias) were also being identified by the groups of K. Kikuchi and Y. Fuyama in Japan.
Additional screens performed in the Carlson group in the late eighties using newer behavioral paradigms increased the collection of mutants with partial anosmias. The expectation that these phenotypes result from defects in odorant receptors proved disappointingly unfounded and the focus of defects in most cases was attributable to alterations in the development and function of neuronal circuits underlying smell behavior. Hence these pioneering studies in the Siddiqi lab provided much information about the normal behavior of the fruitfly Drosophila melanogaster and laid the beginnings of studies on the development of circuitry in the early nineties.
The combined bioinformatics and molecular approach used in the Axel lab to identify vertebrate odorant receptor genes (which resulted in a Nobel prize for Buck and Axel in 2004) catalyzed the groups of John Carlson, Andrew Chess and Leslie Vosshall (working in Richard Axel's lab) to independently identify a large family of odorant receptor genes in Drosophila in 2000. The approximately 60 genes that encode odorant receptors are expressed in a stereotypic manner with each of the ~1800 olfactory sensory cells expressing a single odorant receptor gene. Using an elegant genetic strategy, the Carlson laboratory demonstrated that most of the receptors exhibited a broad and overlapping spectra of chemicals to which they respond. The fact that each chemical is detected in a combinatorial manner by multiple receptors explains why mutations in odorant receptor genes are unlikely to result in anosmias.
Work in several laboratories is now focusing on how stimulus detection by the receptor results in neural activity. In the last month, three different papers,,10 present seemingly divergent mechanisms. In vertebrates, odorant receptors couple to olfactory G-proteins termed Golf which upon stimulation, activate adenyl cyclase III to raise cAMP levels in cells and open chloride channels. In invertebrates, results are still controversial and different groups have proposed either cAMP or inositol 1,4,5-trisphosphate (IP3) or other second messengers such as a phospholipid — as the chemical entities that mediate mechanisms that convert information about binding of the chemical to its receptor to neuronal activity.
Confounding the classic models of odor transduction even further, the odorant receptor (Or) in insects is now known to form a complex with a co-receptor Or83b which is essential for its transport to the dendritic end of the sensory neuron. Evidence from the Vosshall group suggests that Or83b as well as several Ors examined exhibit an unusual membrane topology with the proposed G-protein interacting region lying on the extracellular side of the membrane.
Based on this unusual receptor topology, the Touhara group, first in 2005 and in a recent paper published in collaboration with Leslie Vosshall, demonstrated that Or/Or83b heterodimers can act as ligand stimulated non-selective ion-channels when expressed in cultured cells. If this mechanism operates in-vivo, it implies that interaction with odors could directly activate ion-channels resulting in neural activity. How can this observation be reconciled with a requirement for signal amplification and regulation and a large body of previous data favoring a role for both cAMP and IP3 in insect olfactory transduction.
Our recent published data provides compelling genetic and electrophysiological evidence towards the existence of a classical transmembrane receptor – G protein signaling mechanism for detection of multiple odorants, through a phospholipid second messenger which is unlikely to be IP3. We demonstrate that mutations that knock-out the gene for a heterotrimeric G-protein, Gαq, reduce the electrical response of olfactory sensory neurons to odors by about 80%. Some response ability still remains suggesting the possibility of multiple signaling systems within the same cell.
Evidence also exists for cAMP-mediated olfactory transduction in insects. Work from Bill Hansson's group together with the paper discussed above, supported the finding of odorant receptor/Or83b channel formation, but showed that these channels are regulated by cAMP or cGMP, possibly through the activation of Gαs. They propose a model where direct activation of an cascade of events initiated by interaction of odors with the heterodimers acts at high odor concentrations, while G-protein amplification could be a slower response acting at higher sensitivity.
We fit our observations into this model by suggesting that signaling through both Gαq and Gαs operate for high sensitivity response to odorants. We suggest the existence of a signaling complex composed of OR83b and an Or protein together with a Gαs which is linked to adenylyl cyclase and Gαq linked to phospholipase Cβ. The site of interaction between the receptor and G proteins still needs to be elucidated, if indeed the putative G-protein binding site on the receptor is located on the extracellular surface. In this model, odorants binding to their receptors would trigger signaling through the Gαq signaling system involving phosopholipid intermediates leading finally to opening of ion channels. This system closely resembles that used in visual transduction in Drosophila where TRP family channels have been shown to open upon Gαq activation of Phospholipase Cβ. We predict that volatiles at high concentration will directly activate the Or83b/Or channels; this ionotrophic response could lead to adaptation which is seen to high odorant concentrations.
It is unclear whether odor sensing neurons that use Gαq also possess Gαs and the cyclic nucleotide system within the same cell. If investigations prove this to be the case, multiple signaling systems could provide a wider range of odorant sensitivity. Electrophysiological studies from the Carlson laboratory have provided an example of a single broad specificity Or (Or59a) that can excite or inhibit a sensory neuron upon stimulation with two different ligands. This type of phenomenon hints at multiple signaling systems within the same cell.
Insects, more than any other species, demonstrate a range of behavioral programs triggered by chemical stimuli. This requires that detection of chemicals and chemical mixtures must operate at very high sensitivity and with a great deal of flexibility. The presence of multiple mechanisms of transduction of odor stimuli allows integration of odors at the peripheral level itself. This possibility is now open to experimentation in Drosophila where elegant genetic strategies allow the manipulation of gene expression in identified sensory neurons in the behaving animal.