Fruit Flies Smell Vibrations?

You probably recall our conversation on Friday concerning a recent paper which claims that fruit flies (Drosophila melanogaster) are capable of distinguishing between deuterated and undeuterated compounds.

Franco, M. I. et al., (2011), Molecular vibration-sensing component in Drosophila melanogaster olfaction, www.pnas.org/cgi/doi/10.1073/pnas.1012293108

This seemed like a tall order at the time, and still does, but after reading the paper I have to conclude that the authors were exceedingly thorough in ruling out as many experimental variables as possible: it definitely appears that the effect is completely real, and vibration sensing is a proposal which seems to be in keeping with the experimental results.

The paper is cogent and laid out in a thoroughly logical manner, systematically eliminating possible sources of bias. All the relevant data are given too.

Before we even get started, the authors point out that deuteration doesn't appreciably change bond length, stiffness or angle.

Potential bias #1 - humans have become accustomed to different scents throughout their life.
Solution -- use fruit flies: they have such short lives that you can ensure that they haven't become accustomed to a smell, and they are generally more sensitive to odotopes (odourant molecules) than humans to boot. Other studies have shown that humans can't perceive a difference between deuterated compounds and their regular isotopes, but the authors believe that (a) humans can't smell well and (b) we probably could tell the difference if the odourants were purified well.

The basic idea of the rest of experiment is simple: run two fragrances down either arm of a t-junction `maze' and see which direction the flies pick most often.

Potential bias #2 -- Some visual, auditory or other cues may lead to flies picking one arm over the other, even in the absence of olfactory stimuli.
Solution -- run a control: one with wildtype flies and no odours, and another with mutant fruit flies which cannot smell at all. It turns out that no significant differences are seen in either of these cases.

Big question #1 -- do flies show spontaneous (naive) responses to odourants?
Answer -- yes: no surprises here really.

Big Question #2 -- can flies distinguish between different isotopes of acetophenone (apparently a common perfume base)?
Answer -- yes! The hydrogen-bearing version is attractive, and as deuterium is added (3, 5 or 8 deuterium substitutions) the acetophenone becomes less attractive and then more and more repulsive. The same goes for octanol and benzaldehyde.

Big Question #3 -- can we teach flies to recognise a deuterated compound?
Answer: yes -- this was done by foot-shock, which apparently is a well-tested method of conditioning fruit flies. If conditioned to deuterated compounds the flies consistently chose that arm of the maze, even if it went against their naive reaction.

Potential bias #3 -- maybe it depends on the type of fruitfly chosen.
Solution -- try a different wildtype strain... which gives the same results.

Potential bias #4 -- perhaps the flies are only recognising and responding to different impurities left over from synthesis and purification of the different deuterated compounds.
Solution -- try conditioning them with one compound (deuterated or undeuterated), but then test them with a different type of compound that is also either deuterated or undeuterated (respectively). Amazingly, the flies can still differentiate the deuterated compound from its regular counterpart.
Potential bias #5 -- could there be residual odours?
Solution -- completely replace all of the tubing after every trial.

Big Question #4 -- if the mechanism underlying this discrimination is based on molecular vibrations we should be able to train fruit flies using deuterated/undeuterated compounds, but then test them using completely unrelated compounds that happen to have similar IR spectra in the region of 2200cm^-1 (C-D stretch vibration). Do we see similar effects?
Answer -- yes! It turns out that the CN triple bond has a similar vibrational frequency to C-D, so this was used to test against. Flies again could be conditioned with deuterated/undeuterated compounds, then select completely unrelated molecules with similar regions of the vibrational spectrum.

I have to admit that it sounds pretty conclusive: the flies are, in some way, sensing molecular vibrations, or something very closely related to them. The only really obvious way in which they could be wrong is if the whole thing were fabricated (I'm not suggesting it is though)!

The authors go on to outline the very core ideas behind inelastic electron tunneling spectroscopy, which they believe is a reasonable model for olfaction. An electron acceptor sits near a donor molecule, but electron transfer only occurs when a suitable intermediary is able to accept an amount of energy, becoming vibrationally excited. Perhaps this is how smell works: after all, we see light because the absorption of a few quanta of light are enough to cause a conformational change in a protein. Why shouldn't smell have a related mechanism?

I can see two obvious ways forward: the first is to selectively knock out each of the 62 different olfactory sensors that drosophila is endowed with, and test which is most important in the apparent discrimination of deuteration. With this determined, electrophysiology would be a must, probably along with gene expression analysis.

The second is to selectively deuterate different portions of materials, to see if the effect depends on the vibrational mode or just the frequency. This could involve some intense chemistry! Finally, perhaps some more dramatic investigations could be carried out: how about if tritium is used?

I'm interested to hear your thoughts!