A Whiff of Ether

My first experience with research — and ether — was in a laboratory I will not name at the University of Chicago. I can say that I learned a lot from both experiences; what I can’t say is that I learned what I was supposed to.

Herr Doctor Professor looked and sounded the part of the biochemistry professor. But the cutting edge of biochem research had, even in the 1980s, passed his little laboratory by; the short-lived undergrad project he assigned me was to try to isolate a bacterial metabolic product, starting with a French Press (though unlike the pic in the Wikipedia entry; this one was enormous, made of steel, and used incredibly high pressures to squeeze the guts out of bacterial cells). This was classic stuff, in the sense that much the same work was going on in biochem labs 40 years earlier.

Somewhere in the procedure was an ether extraction — mixing the water containing the bacteria guts with ether, an organic solvent known for its ability to knock people cold. The idea was that our molecule might be happier in ether than water, and so might be enriched when the two solvents separated.

What nobody told me at the time is that you need to do ether extractions in a fume hood, not out in the open. I’d come to dinner at the dorm every night stoned out of my gourd — but it isn’t a high that I’d recommend to anybody, it’s a headachy, nasty affair that nobody in his right mind would seek out. I may have engaged in conversation at these dinners — my dorm mates in the immortal Fishbein House would know better than me, I barely remember.

Still, these ether-induced trances were nothing compared to the other-worldly head of steam I'd work up in physical chemistry class. I’d stumble in to the dining hall, head filled with Eigenfunctions and atomic orbital maps, literally incapable of (and probably not fit for) human interaction, at least until I got past my entree and the dancing symbols began to fade.

All of which by way of saying, though I have in the past been prepared to and even capable of performing, to a fashion, in that world [1], I’ve never pretended to be a physical chemist or biophysicist. I fully expect to get some stuff wrong in this blog site; occasionally I expect I’ll get something egregiously wrong. But there’s no wrong quite so wrong as the wrong I’m likely to be when I trespass in the land of biophysics.

Having said that, I feel compelled to talk about 1993 report by Doron Lancet and folks at Tel Aviv University that I just discovered. I was digging around for information on binding constants in olfactory receptors, and stumbled across an amazing little article that I’m surprised hasn’t been more heavily referenced. It provides that rare commodity in biology: A theoretical result that may matter.

What Lancet and buds did was ask a simple question: Based on what we know about the olfactory system’s mission and the nature of the molecules it recognizes, can we predict how many receptor types you’d need to do the job? (See Broken Bottle Fight for a discussion of how many they actually have.)

Conceptually, it’s pretty simple. Most enzymes and receptors in the body have been honed by evolution to interact very specifically and tightly to a biological target molecule — for an enzyme, it’s the substrate; for a receptor, the ligand. But for three systems in the body — the olfactory system, the immune system, and the liver’s system for disarming toxins — that kind of 2.7-plus-billion-year prep isn’t possible. All of these systems need to be able to respond to something that’s absolutely brand-spanking new, and respond more or less the moment it heaves into view.
So you need a system containing a randomly generated variety of receptors (or a variety of enzyme binding sites, for the liver’s cytochrome P-450 system) that sit there, waiting for the target that they just happen to recognize.

Here’s where I’m speaking in metaphor not as a didactic tool, but because it’s more or less my own level of understanding. Your mileage may vary, may cause cancer in high doses, don’t stand on top rung of ladder.

The Tel Aviv team concentrated on iodovanillin, a chemical derivative of the molecule that gives vanilla its flavor and odor, as a molecule pretty typical of both the odorants that olfactory receptors have to recognize and the haptens to which the immune system needs to respond. The idea was to estimate how many randomly generated receptors you’d need for a system to recognize iodovanillin about as well as the sense of smell recognizes it.

A random population of receptors should recognize a ligand roughly as the right half of a kind of bell curve [2]. Most of your random receptors will hardly recognize the ligand at all, but a few curve-breakers will latch onto it much better than the average bear:

It’s the right-hand part of that curve that's interesting: The receptors in that tail are the few that have a strong enough recognition for the new ligand that they can trigger a biological response. And making that right-hand tail nice and fat depends on how hard the ligand is to recognize and how many receptor types you have. The harder the ligand is to recognize, the more random receptors you’ll need to have, by pure chance, one or more that can do the job.

Lancet and crew came at the question in two ways, both of which essentially measured how hard iodovanillin is to recognize. Note that I’m not sure whether iodovanillin actually has a smell — again, the idea was to get an estimate for a typical odorant or hapten, and iodovanillin fit the bill.

For the first calculation, they used a random mix of cow antibodies (which they got from blood from a random cow) to see just how well they stuck to this molecule; in the second, they used data from experiments to measure human olfactory sensitivity.

The two methods basically agreed — 300 to 1,000 receptors from the cow data, 500 from the human data. More interesting, those numbers are pretty close to the known number of genes producing olfactory receptors in vertebrates (again, Broken Bottle Fight goes into this in more detail).

Even neater: in terms of number of olfactory receptor types present in each species, the lower bound of that range is where humans seem to be, and the higher bound is close to where animals normally though of as being able to smell very acutely, such as dogs and mice, sit. Not coincidentally, maybe, the theory would predict that animals with more receptor variety would by definition have more potential for developing an acute sense of smell.

Though we may be over-mining the result at this point, and the number of human receptors isn’t completely nailed down, it is interesting that the number of receptors the second determination projects as necessary to really get a good snootful of iodovanillin is more than humans have at our disposal. So maybe our dogs enjoy vanilla more than we can.

Unless I’ve got it wrong. Who knows how much damage all that ether — let alone the physical chemistry — did?

[1] Straight Bs, if you must know, though it was a spectacular pattern: I’d have an A through the quarter, and then crash and burn on the damned final exam — see an early credit of mine in Nature magazine, though I’m not sure whether you’ll be able to download the text.
[2] Actually, it’s technically a binomial distribution, which isn’t exactly the same thing, but the idea is similar — a few members of the population are going to be curve-breakers that are very good or very bad at the activity in question, but most will be so-so.