Wednesday, March 11, 2009
A new paper in a rather specialized scientific journal has exciting news for those of us who eagerly anticipate a revolution in olfactory biotechnology.
Humans have about four hundred different olfactory receptors; other mammals have even more. There are thousands of potentially smellable volatile molecules in the world. Ever since the discovery of the olfactory receptor genes in 1991 the big question has been: which odor molecules are detected by which receptors? Until we can answer that question, we cannot truly say that we understand the biological basis of odor perception.
The standard way to investigate the function of a receptor is to express it in cell culture—that is, one inserts the receptor gene into a yeast cell, for example, and the yeast cell obligingly reads the new DNA, creates the receptor, and places it in the cell membrane all ready to go. Do this for enough different receptors and soon you can see which of them respond to rose alcohol, which to vanillin, etc. When automated on an industrial scale this is called high-throughput screening—a standard procedure in pharmaceutical research.
Olfactory receptors have proven exceptionally difficult to express. We’ve been able to match odors to receptors here and there in only about 50 instances; to make real progress we need to do it thousands of times under standard conditions. Recently Dr. Hiro Matsunami’s lab at Duke University created cell lines and test protocols suitable for high-throughput screening. This week they reported on the first large-scale matching of receptors to odor molecules.
Their results are in the March 3rd issue of Science Signaling. They tested 219 mouse receptors and 245 human receptors against 93 different odorants. Some receptors didn’t respond to any of the odorants; others responded to several. Some odorants activated only a couple of receptors; other activated many. As you can imagine, analyzing data on this scale is a challenge.
Ultimately, the Duke group wants to predict which odor receptors are activated by which molecules. They created a big database of odor molecules and used statistical techniques to extract a few physical properties that predict similar receptor responses. On the other hand they created an olfactory receptor database and statistical means of grouping receptors that respond to similar odors. They can now begin to discern which sort of molecules match up to which sort of receptors. Although their conclusions are abstract and very tentative, the Duke group has taken the first step toward making sense of high-throughput data and answering the Big Question.
On a practical level, their challenge is like the one you face when doing a jigsaw puzzle without the knowing the final picture. First, you dump the pieces on the table and start sorting. Light blue pieces with streaks of white go in one pile—they could be sky with clouds or whitecaps on ocean waves. Dappled brown and green pieces go in another pile—they might be tree leaves or a forest path. Pieces with straight sides go in a third pile—they’re the edges of the picture. In other words, you can start sorting without knowing exactly where the pieces fit in the picture.
Substitute “eigenvalue sum from van der Waals weighted distance matrix” for “light blue pieces with streaks of white” and you get an idea of what the Duke group is up to. The fun comes once we get the puzzle put together—when we can match every molecule to a receptor and vice versa. Then we can design highly specific odor blockers and odor enhancers, and even, perhaps, predict what perfumes a person will like. Stuart Firestein and I outlined some of the possibilities in a Nature Neuroscience paper which you can download here.