When a fly's smell is bad, its life stinks.

Is the system by which humans detect odors, olfaction, as deeply conserved through evolution as we've long assumed?  Olfaction presents an important chemosensory challenge to organisms, like us, that evaluate their environment by picking up some subset of its chemical aspects.  A wide diversity of animals smell with a series of more or less randomly varying cell-surface receptors on the olfactory tissue which is exposed to the outside world (e.g., the lining of your nose). 

Human olfactory system; 1: Olfactory bulb 2: Mitral cells 3: Bone 4: Nasal Epithelium 5: Glomerulus 6: Olfactory receptor cells; Wikimedia

Olfactory receptor neuron; Wikimedia

Neurons express olfactory receptor (OR) molecules, which are proteins (and hence coded by specific genes) that have a binding pocket in the part that sticks outside of the olfactory neuron.  There are hundreds of different OR genes in the genome, and they vary in their amino acid content so that the binding pocket of each responds only to specific aspects of an odorant molecule.  A given odorant will be detected by only a subset of ORs; this has been known for a long time, but how it happens is still unknown and debated.  Recent arguments suggest that certain aspects of quantum entanglement (rather than molecular binding) are responsible.

Now, if all ORs were expressed by each olfactory neuron, each cell would respond to every odorant, sending "I detect this!" messages to the brain.  Whether you were smelling lion or chocolate, your brain would get the same message--a 'bell' rung by each neuron.

Instead, through various mechanisms each neuron only expresses one of these many OR gene products.  The mechanism is very specific, though currently basically not understood, but that is not our topic today.  The point is that when chocolate molecules waft through your nose, they 'ding' only some of the neurons.  Lion smell dings others.  Indeed, the wiring to the brain's detection center, the olfactory bulb, sends signals from cells using the same OR to the same places in the brain.  This allows your brain to do the bookkeeping and keep an orderly account of what's out there; as a result you can tell if you are about to become a meal or partake of one. 

Insects have a very different repertoire of OR genes, but have similar one-gene mode of using them.  Well, that at least is the story as it has been believed since the whole single-gene per neuron expression and highly variable, numerous OR genome was discovered--a striking finding for which Axel and Buck deservedly won a Nobel prize in 2004.  But does the system actually work that way?

Flies, the standard laboratory Drosophila species, are easy and quick to work with compared to mammals, and one can do genetic engineering to test ideas like these.  And a recent paper in PLoS One by Tharada et al. reports just that, their test of whether the single-neuron unique-address system actually works in insects as has been thought.

Dorsal view of a cutaway fly head showing the main elements of the olfactory pathway. Odours are sensed by olfactory receptor neurons in the antennae and maxillary palps. These neurons project axons along the antennal nerve to the antennal lobe glomeruli, where they are sorted according to chemosensitivity. From there the information is relayed by projection neurons in the inner and medial antennocerebral tract (iACT and mACT) to the mushroom body and to the lateral horn. Gustatory stimuli are sensed by gustatory receptor neurons in the labellum on the tip of the proboscis, the elongated fly mouthpiece.  Source: Nature Reviews Neuroscience, Keene and Waddell 2007

Basically, they found that flies engineered to co-express more than one OR in a given neuron were less able to detect and find a secondary food source in their experimental chamber, compared to control flies expressing only a single OR per neuron (the normal pattern). The experimental group that were forced to search for secondary food had lower survivorship and were thus less fit.
While any experimental study of this sort is somewhat artificial relative to the real world, and perhaps specialists will raise various methodological questions beyond our ability to judge the study, this does seem to be the first direct evidence that the presumed single-expression neural bookkeeping hypothesis about odorant detection, and its evolutionary basis, are supportable.

This is important first to the degree that it confirms prior ideas, suggesting that we understand at least major elements of this system.  More importantly, perhaps, it further opens doors towards understanding the mechanism by which single gene per cell expression patterns are achieved.  One can't automatically extrapolate from flies to vertebrates, but the similarity in gene families and expression pattern suggest that such extrapolations are  not entirely fanciful.

Or, put another way, when a fly's smell is bad, its life stinks!