Wed, 25 Jan 2006
A reader, who goes by the name of Omar, wrote to remind me of the "Hox" (short for "homeobox") genes discussed by Richard Dawkins in The Ancestor's Tale. (No "buy this" link; I only do that for books I've actually read and recommend.) These genes are certainly part of the story, just not the part I was wondering about.
The Hox genes seem to be the master controls for notifying developing cells of their body locations. The proteins they manufacture bind with DNA and enable or disable other genes, which in turn manufacture proteins that enable still other genes, and so on. A mutation to the Hox genes, therefore, results in a major change to the animal's body plan. Inserting an additional copy of a Hox gene into an invertebrate can cause its offspring to have duplicated body segements; transposing the order of the genes can mix up the segments. One such mutation, occurring in fruit flies, is called antennapedia, and causes the flies' antennae to be replaced by fully-formed legs!
So it's clear that these genes play an important part in the overall body layout.
But the question I'm most interested in right now is how the small details are implemented. That's why I specifically brought up the example of a ring finger.
Or consider that part of the ring finger turns into a fingernail bed and the rest doesn't. The nail bed is distally located, but the most distal part of the finger nevertheless decides not to be a nail bed. And the ventral part of the finger at the same distance also decides not to be a nail bed.
Meanwhile, the ear is growing into a very complicated but specific shape with a helix and an antihelix and a tragus and an antitragus. How does that happen? How do the growing parts communicate between each other so as to produce that exact shape? (Sometimes, of course, they get confused; look up accessory tragus for example.)
In computer science there are a series of related problems called "firing squad problems". In the basic problem, you have a line of soldiers. You can communicate with the guy at one end, and other than that each soldier can only communicate with the two standing next to him. The idea is to give the soldiers a protocol that allows them to synchronize so that they all fire their guns simultaneously.
It seems to me that the embryonic cells have a much more difficult problem of the same type. Now you need the soldiers to get into an extremely elaborate formation, even though each soldier can only see and talk to the soldiers next to him.
Omar suggested that the Hox genes contain the answer to how the fetal cells "know" whether to be a finger and not a kneecap. But I think that's the wrong way to look at the problem, and one that glosses over the part I find so interesting. No cell "becomes a finger". There is no such thing as a "finger cell". Some cells turn into hair follicles and some turn into bone and some turn into nail bed and some turn into nerves and some turn into oil glands and some turn into fat, and yet you somehow end up with all the cells in the right places turning into the right things so that you have a finger! And the finger has hair on the first knuckle but not the second. How do the cells know which knuckle they are part of? At the end of the finger, the oil glands are in the grooves and not on the ridges. How do the cells know whether they will be at the ridges or the grooves? And the fat pad is on the underside of the distal knuckle and not all spread around. How do the cells know that they are in the middle of the ventral surface of the distal knuckle, but not too close to the surface?
Somehow the fat pad arises in just the right place, and decides to stop growing when it gets big enough. The hair cells arise only on the dorsal side and the oil glands only on the ventral side.
How do they know all these things? How does the cell decide that it's in the right place to differentiate into an oil gland cell? How does the skin decide to grow in that funny pattern of ridges and grooves? And having decided that, how do the skin cells know whether they're positioned at the appropriate place for a ridge or a groove? Is there a master control that tells all the cells everything at once? I bet not; I imagine that the cells conduct chemical arguments with their neighbors about who will do which job.
One example of this kind of communication is phyllotaxis, the way plants decide how to distribute their leaves around the stem. Under certain simple assumptions, there is an optimal way to do this: you want to go around the stem, putting each leaf about 360°/φ farther than the previous one, where φ is ½(1+√5). (More about this in some future post.) And in fact many plants do grow in just this pattern. How does the plant do such an elaborate calculation? It turns out to be simple: Suppose leafing is controlled by the buildup of some chemical, and a leaf comes out when the chemical concentration is high. But when a leaf comes out, it also depletes the concentration of the chemical in its vicinity, so that the next leaf is more likely to come out somewhere else. Then the plant does in fact get leaves with very close to optimal placement. Each leaf, when it comes out, warns the nearby cells not to turn into a leaf themselves---not until the rest of the stem is full, anyway. I imagine that the shape of the ear is constructed through a more complicated control system of the same sort.