Science and sensibility

In recent years it has become increasingly clear that the regulatory information associated with genes is just as physiologically important an evolutionary malleable as the genes themselves. Well that's all well and good but genomes will still eventually face evolutionary pressures to develop a completely new biochemical trick. But how can an old genome learn new tricks? On the face of it seems a very prickly question. An intellectually lazy person might just give up, call it impossible and accept a grant from the Discovery Institute. Of course, real scientists don't think like that. M Pilar Francino is one such scientist, this month she presented a new model describing the evolution of new information in genomes.

The problem with developing new functions is one of raw materials. Ideally you could mould existing genes to new roles. After all, even with all the limbs, organs and thoughts some organisms go in for life is just chemistry. If you need to recognise, metabolise, cut or otherwise modify a new chemical there is probably already a gene in your arsenal that has evolved to do it to a related one - it may even do the job on the new chemical to a diminished level. The mutations that crop up in each new generation could throw up variants that are better at the new job and evolution could select for those variants. The problem is that there is usually very strong "stabilising pressure" keeping genes more or less as they are - since genes are the product of millions of years of selection any small change is likely to make them worse at their job. For this stabilising pressure to be overcome there would need to be a series of mutational steps from the original function to the addition of a new function in which none of the mutations significantly impaired the original function. Most biologists agree this might happen from time to time but it is not enough to explain the thousands of functional genes found in most genomes.

The classical model for the way genomes add more functions to themselves relies on 'gene duplication' to avoid the effects of stabilising pressure. From time to time genes within genomes and even whole genomes are duplicated. We've seen it happen in labs and we can trace back ancestral duplications by comparing the genomes of related species. The classical model holds that if a functional gene is duplicated the stabilising pressure will act to keep one copy more or less as it is. The other copy of the duplicated gene is free to mutate in a new direction and perhaps acquire a new function. But evolutionary biologists - contra to the bleating of some creationists - don't simply come up with theories, they can test them too

If gene duplication has played an important role in genome evolution then we would expect to find genes within the same genome that share a common ancestor. And of course, this is exactly what we find. One of the best documented cases of this comes in the form of the 'globin' family of genes. The globins bind oxygen molecules, in your body there are five distinct classes of globin all of which serve a slightly different role. When we look at the globin genes in the modern human genome we find them in two clusters. Chromosome 11 has four active globin genes while chromosome 16 has five. When we compare the sequences of the globins we can see that all these genes can be traced back to an ancestral globin that resided in our common ancestor half a billion years ago. That globin gene was duplicated and over time one copy went on to be the progenitor of the cluster we now find on our chromosome 11 and the other founded chromosome 16's cluster.

So, we can see that duplication of genes has furnished genomes with new functions but that's not all that the classical gene duplication model predicts. Remember that I said a new duplicated gene being free of stabilising pressure so able to "mutate in a new direction and perhaps acquire a new function."? Well, some scientists question the chances of a mutation in such 'neutral' conditions throwing up anything but broken genes. You see, perhaps the most powerful aspect of natural selection is that is cumulative. Natural selection acts on genes that are themselves the product of millions of years of evolution. Deleterious mutations are wiped out and very sophisticated functions can be developed because selection acts on a set of already very highly developed genes. But what will happen if deleterious mutations aren't removed?

The best example of the fate of genes freed from selective pressure comes from intracellular parasites, these organisms are cells (by and large bacteria) that live inside other cells. In other words they live in an environment that provides almost all the chemicals they need to live. Consequently most of the parasite's genes are freed from stabilising pressure - there's no need to keep a gene for making glucose if you live in a sea of glucose. Freed from selection the genes of intracellular parasites tend to 'rot'. They pick up damaging mutations that render the gene useless. They are often not even able to be read by the cell's machinery. Some biologists feel that a newly duplicated gene would suffer the same 'mutational load' under neutrality and that this makes the classical gene duplication model an improbable explanation of genome evolution.

Francino's theory, published in Nature Genetics, removes the neutral stage. As I said earlier if a new biochemical niche pops up it is quite likely that there is already a gene that can do the job - just not particularly well. One way to get better at the job would be to take a 'many enzymes make light work' approach and make many copies of the suboptimal gene's product. You could achieve the same effect by making several copies of the suboptimal gene and expressing them at the normal rate, and in fact this very phenomenon has been observed. Most biology undergraduates will be familiar with LacZ- mutant E. coli. LacZ is a gene that plays a key role in metabolising the sugar lactose. LacZ- mutants carry a defective form of this gene that is very much reduced in its ability achieve this task. In 2002 researchers showed that LacZ- mutant E. coli could evolve to effectively metabolise lactose by increasing the number of copies of the defective LacZ genes they had. Francino cites similar exaples in bacteria, insects, yeasts and mammalian cells.

Francino has seen that this evolutionary directed gene duplication neatly avoids the problem of neutrality found in the classical model. Following a round of gene duplication selection will, just as it does in the old model, favour those organisms that retain at least one copy of the gene more or less as it is to perform the original role. However, Francino's argues, since the rest of the duplicates are acting together to perform a new role natural selection will weed out any of the massively damaging mutations associated with neutrality before they take hold. At the same time duplicate genes are freed from pressure to perform the original task. This means mutants that are better at the task the duplicates are banding together to perform will be favoured and down through the generations the duplicates will get better and better at it. In fact, eventually they will get so fine tuned to the new role that only one gene will be required. The most effective of the duplicates will be 'fixed' and the others will fall into neutrality and face mutations that render them useless.

Breaking it down to the bare essentials Francino's model involves four stages:

1. A new biochemical niche appears - perhaps a new signalling chemical or DNA sequence that would be advantageous for an organism to recognise
2. This niche is then filled by the duplication of an existing gene that currently performs a chemically related but distinct role.
3. Mutations will then crop up in the duplicates that make some of the duplicate genes better at performing the new biochemical trick
4. Selection will fix the best of the duplicates while others fall by the wayside

Like any good scientific model Francino's one makes testable predictions. If genomes have indeed gone trough this procedure to acquire new functions we would expect to be able to detect the wake of that activity on modern genomes.

Research to be published next month (hat tip: afarnesis) seems to provide evidence for the sort of rapid expansion of genes due to positive selection for duplication predicted in the second phase of Francino's model. A team form the University of Washington investigated the distribution of a group of duplicated genes in humans, other apes and old world monkeys. They where surprised to find a large series of duplications that where shared by all the apes and none of the old world monkeys. When you consider this distribution in the light of each species' evolutionary history (depicted in the diagram below) it becomes clear how this might happen.

Do you see? The large scale duplications must have happened in time between the old world monkey branch and the ape branch split but before the apes diversified into their various species. The vast majority of the gene duplications studied in the ape lineage took place in this narrow window of time. This is exactly the sort of pattern we would expect to see if there had been a strong pressure to produce multiple copies of a suboptimal gene to fill a new niche followed by generation upon generation of fine tuning the duplicated genes to the new role.

At the same time we might expect to see a lot of duplicated genes that had at one stage been part of the team of suboptimal genes performing a new task but since fallen by the wayside and into disrepair. In fact, it has been noted for a long time that gene clusters like the globin ones described earlier are often associated with such "pseudogenes." As well as the nine functional globin genes in the two globin gene clusters there are at least four pseudogenes. Gene clusters that are the result of more recent duplications, like the olfactory receptor clusters, have a higher proportion of pseudogenes.

This post has been among the most technical I've ever tried to write, I hope I have managed to make it intelligible without losing any of the detail. I think Francino's model is important not just because it explains an interesting bit of evolutionary biology but because it shows a little of the way real scientists work on problems. A creationist might say " evolutionary theory can't explain how information is added to a genome" or claim evolution is simply a "series of untestable speculations not supported by empirical evidence" then pat themselves on the back for battling those godless materialists who are trying to steal people's souls. You don't need evidence beyond your own ignorance as long as you're fighting the good fight. Well, scientists don't get to just make stuff up. When faced with a problem like the addition of information to a genome they don't throw their hands up in the hands up in the air and say "God did it." They look at what evidence is available to them and they build a model that best fits the facts. At every step in this post we've seen the way that Francino's model, and the classical one before it, shape up against the facts revealed by scientists. The new model will not be the final word; science will continue to get a clearer idea of how populations, organisms, genes and genomes evolve by developing better and better models and testing them against more and more facts. At the same time creationist's arguments will remain rooted to a bronze age mythology.

Some further reading:
Francino's paper
Lac Z mutants evolving by gene duplication
What happens to the genes of intracelular parasites
Associatoin between olfactory receptor gene clusters and peudogenes