The genetics of olfaction: In evolution, where there's one, or one thousand....there's more. But how much more? Part II.

Last Friday and today we're discussing the means by which you are able to detect and discriminate among infecting pathogens to which you may be exposed.  We reviewed the immune system and the idea of monoallelic expression, by which only one of the two copies of a gene you inherit is chosen for use in a given immune cell.  In fact, in the immune system, it's really only part of a cluster of gene segments that is used, and only one cluster of the two in your inherited genome.  It's called 'adaptive' immunity (you also have a second, very different, 'innate' immune defense unrelated to our subject here).

We went on to say that odor detection is a somewhat similar challenge:  of the countless odor molecules that you might want or need to be able to detect, your body can't know all in advance.  Since odorants are molecules, perhaps a molecular--a genetic--method of open-ended variation would serve as well as it does in our 'adaptive' immune system. In fact, we have something similar in our odor-detection sense.  But it works very differently from the immune system, using unrelated genes and unrelated (as far as is known) gene-selection mechanisms.

We have about 1,000 (yes, that's thousand, about 5% of our genome) different Olfactory Receptor (OR) genes in our genomes. (We've posted on aspects of this story in past years, e.g., here and here.)  Since we have two copies of all our genes (except for XY genes in males), that means we have a total of nearly 2,000 OR genes.  An OR gene codes for a protein that sits in the surface of olfactory neurons, hanging out in the nasal passage, in contact with the air we breathe, and the odorant molecules it carries.  The OR genes are in clusters, large and small arrays of adjacent, related OR coding genes, scattered across most of our chromosomes.  Each OR gene is the result of a past duplication event, so is closely related to some other OR's often right nearby on the same chromosome; 'closely' means a few mutations different from the 'parent' gene that have arisen as part of the duplication or since that occurred.

OR locations in the genome (red); obtained from http://www.leffingwell.com/olfact2.htm

In some mammals that rely heavily on smell, most of the OR genes, in most clusters, are working.  In primates and humans, we've still got as many identifiable OR genes but more of them are dysfunctional, having been mutated out of usability; the protein code just doesn't generate a functional OR protein.  But for many or most, that means that ORs that may be dysfunctional in one of my copies may be functional in the other, or in other people.

Olfactory neurons develop as part of the nasal cell lining.  They express OR genes on their surface, that dangle into the nasal airway.  But as a future olfactory neuron develops, it picks one OR gene, from one cluster, and of that only from one of the two copies of that cluster, to express.  The other copy is inactivated as are all the other 1,999 OR genes!  But this is not so easy to explain as the comparable monoallelic expression of the antibody genes that we described on Friday.

Source

The two copies of them, at least, are in the corresponding place on the two copies of the respective chromosomes (one inherited from each parent).  That makes them in principle easier to be put into correspondence by which one is chosen and the other inactivated.

In the case of the OR genes, the developing olfactory neuron must first pick one of the many clusters to use, and then use only one OR gene from that cluster, and then inactivate the other genes in all other clusters, both copies, on all other chromosomes!  Pick one gene, and then go find all the others and silence them.

This is a remarkable feat and the very explicable outcome is that each olfactory neuron uses only one OR protein, and can detect only odorants that, wafting by, are recognized by that OR protein and trigger the neuron to send an "Aha!"  message to the brain. The brain collates the messages sent at any given time by all the neurons in both nostrils, and assembles and remembers a catalogue: "These neurons respond together and we'll call that 'lemon';  next time this same collection of neurons fires, I'll know to get out the glass, water and ice cubes--lemonade!"

Because they are being duplicated rapidly and they mutate rapidly, the 2,000 OR genes provide a huge repertoire of potential responders, and their combinatorial signals, which provides the kind of open-ended cataloging of odorants--including mates, predators, food and prey.  This is an extensive form of monoallelic expression that seems obviously to have been useful to our ancestors' survival.

Current status
In 2005 I spent a sabbatical leave at the Sanger Genome Center near Cambridge, England, trying to find DNA sequence motifs that might help account for this very selective monoallelic expression.  I didn't succeed....but neither has anyone else since.  A recent review by Ivan Rodriguez in Cell describes what is known about this system, and an article in the same journal describes one facet of that mechanism.

Rodriguez notes that some cluster-specific regulatory regions have been found on some chromosomes, and there are instances of those having similar sequences; but they are far down the chromosome from the OR genes themselves, they only affect their local cluster, which makes such regulatory sequence elements hard to find on other clusters.   Previous ideas of global all-cluster control sequence signals have not panned out.  But a mechanism has been found that, in part at least, explains how a cell that happens at first to pick a defunct OR gene for use fails to develop further but instead is returned to the OR-picking stage.  It's not known what happens if a cell mistakenly picks more than one OR gene, but such cells are rarely seen so they must degrade somehow, or fail to reach the relevant part of the brain (the olfactory bulb).  The part of the DNA including a chosen cluster region is open for expression, but how the rest of the OR-cluster areas on the other chromosomes are silenced isn't known.

Other related systems have similar one-gene selection. These include the V1R and V2R genes that code for pheromone receptors and two other groups, the FPR and TAAR genes, that are each expressed in a one-only way but in a different part of the nose, called the vomeronasal area. 

An older idea that the cluster-containing parts of chromosomes were located near the outside of the neuron cell's nuclear membrane was shown to be wrong; instead these areas may be clustered toward the center of the nucleus.  This may bring various OR regions closer together so they can be activated or shut down.....but how are these areas shepherded there?  And since an olfactory neuron is also expressed in many hundreds of other genes, how are they kept open for expression?

This is considerable advancement in our understanding of how the chosen OR gene leads to a functional neuron or how a dysfunctional OR choice leads to a re-selection. It explains some aspects of how an OR cluster's genes are opened for expression, and how the coded protein is processed outside the cell's nucleus. But it doesn't answer the $64 question:  how is the choice made in the first place and the unlucky 1,999 lottery losers shut down?  How is this partially related to the neighboring cells' choice, presumably because they are recent cellular descendants of each other in the growing nasal epithelium?  And how is the choice totally independent in other cells?

The OR genes are members of a large, ancient family of genes that code for all sorts of cell-surface receptors that are used in all sorts of other functions, but as far as is known (or at least as far as we're aware) their expression is not monoallelic and they are not used in a choose-one/exclude-the-others mode.  So mystery upon mystery still attends the monoallelic expression, the very nice bookkeeping strategy, by which we can tell when we smell a rat.

What is curious is that the same mechanism does not seem to apply to the various other known forms of monoallelic expression, described on Friday, each of which is very deep in evolutionary terms.  And more curious is the question: how widespread is monoallelic expression, if it exists, in other systems where choosing among many genes is not the 'trick' but in which there may be reasons for only one copy of a gene to be used in a given cell in a given context?  If recent evidence is any guide, we will find many more examples.

This is yet another way in which our rather stereotypical, and one may say superannuated Mendelian concepts are in need of some serious rethinking.