Pre-biotic soup

PaleoFreak coments on the level of understanding of natural selection theory. Doing so is like asking for somebody to come and tell you "the theory doesn't fit to me because...". Alas, that's what's happened, and it was about one of the most recurrent, interesting and harsh topics of all -- evolutionary theory may explain life's diversity, but what about its origin itself?

Abiogenesis --the generation of life from non-living matter; i.e 'spontaneous generation' well understood-- is Terra Incognita. There are a few theories around, but all of them include to some extent a key element: the rise of self-replicating self-organized systems. For some theories, the first systems of this kind were metabolic networks; for others there was a RNA world. In fact, it little matters. The thing is that once we have such a system working, we have life and then evolutionary theory starts playing a role.

I say that it doesn't matter because once an evolvable system has risen in the first place, it MUST evolve the features common to all modern life forms; all theories trying to explain the origin of life have to converge to this point.


Therefore, abiogenesis theories main point is to explain how this first 'living' system can appear from physico-chemical reactions alone. In contrast to what happens with the explanations of how modern-day life can evolve from the first living system, abiogenetic theories are dramatically different. Since plain chemical systems do not evolve (there must be heredity for that, and for that there must be self-replicant chemical systems; those aren't just 'plain' chemicals, they are far more complex), we cannot explain how from an abiogenetic scenario we can get into another.

However, all of them have something in common. The raw materials (how do those arise is subject of yet further discussions) are: aminoacids, simple sugars, nucleotides... everything we are taught in our first biochemistry lesson at school. And all of them have the same problem: how the heck can you get stable polymers from that?

Fairly easy, as it may seem. A recent paper (Leman L, Orgel L, Ghadiri MR (2004) Carbonyl Sulfide-Mediated Prebiotic Formation of Peptides. Science 306(5694):283-286) proves that cabonyl sulfide (COS), vulcanic gas, can catalyse peptide formation when pumped through a solution of mixed aminoacids, and doing so at room temperature in a matter of a few hours. The authors have tested a number of different reaction conditions and the results are just amazing.

The zebrafish (Danio rerio)

24h of zebrafish embryonic development (Rolf Karlstrom and Don Kane; Development 123:461, 1966)

From zygote to free-swiming larva in 48 hours

Up there should be a couple of videos showing the embryonic development of zebrafish. If they don't work, follow the links. That's what I'm working with for my Ph.D. thesis. The development of this animal is astonishingly fast. That helps a lot the embryological studies and the fact that it's completely transparent during the first 24 hours (after that, it starts developing the pigment cells that will create the stripe pattern typicall of this fish) makes it perfect for visualization (as you can see in the videos). Another great advantage is the number of genetical tools available. You can microinject the eggs with DNA, RNA (antisense or sense), morpholinos, proteins, antibodies, chemicals, whatever you want in order to alter the genetic profile of the developing embryo and see the effects of the injected substance. The use of morpholinos (chemical analogs of nucleic acids, but more stable; they work by complementary hybridisation with the target mRNA in the embryo, thus blocking its expression into protein) is particularly helpful because it allows you to abolish the expression of the genes you want to in a very specific and stable way (more than antisense RNA) without having to generate transgenic knock-out fish. Could we say any more about this wonderful system? Sure! There are vast collections of mutants (albino lines are quite useful for visualization of processes happening during the second day of development) and the genome is being sequenced (there is a very advanced draft publicly available, coming from shotgun assemblies as well as some full BAC clones sequencing). If that wasn't enough, they breed like flies, so it's possible to perform "proper" statistical tests with the experiments. In conclusion, this fish is the Drosophila of the vertebrates. And they look great in your home aquarium (especially the mutants ;-)

The Fate of the Lone Mutant

... or the hopes of a poor monster.

A sci-fi movie? Nop. When I was studying (ahem) for geneticist, I found a Population Genetics book (I don't recall which one of the many I read) that had a chapter with such a heading. Yes, I know that, the guys of PG are nuts. The story is that under that heading that was exactly what it was described, i.e. which is the evolutionary fate of a mutation recently arisen in a population (that bit about the hopes not, I added that myself; you'll see why later).

A number of years ago, before I started takin PG lessons, I met a guy nick-named Xac Mazo. We talked a lot in some public discussion forums (at terra.es) and one thing led to another and in the end we discovered that both of us had had experimented with simulations of the evolution of populations. I'll let himself to explain, if he wants to, what kind of madness made him do so; my answer is something I read when I was younger.

It was an article in PCmania ('Temas Informágicos') dealing with population dynamics as an example of an ordered deterministic-chaos system. A simple BASIC program allowed me to play with the parameters of the logistic growth equation. Since luck favours the trained mind, it took me a long number of years to finally develop the idea any further.

I did so when I (thanks to Mazo) discovered that you can generate recursive functions in MS Excel. Te idea was to simulate the fate of two sympatric populations that have a difference in reproduction rate. The result, the obvious one.

We started simulating evolutionary genetics scenarios, including the fate of the lone mutant. Especially that one, I must say, because the point we were trying to prove in a discussion of that time was what chances had a "hopeful monster" of giving birth a new species. Under "normal" circumstances, the chances were really tiny...

What is needed in order to get the newly arisen allele fixed in a stable frequency (and somehow significant)? First, we start from only an heterozygous individual (every new mutation appears first in an heterozygote by definition). The heterozygote will try to reproduce and for that it needs a couple that, obviously, will be homozygous "normal". Therefore, its descendence would be half heterozygous, half homo-normal. Asuming random mating, until a "critical mass" is reached, the heterozygotes will always mate with homo-normals, so there are never homo-mutants until that point in which it's possible for a heterozygote to mate another heterozygote by random mating.

That means that a number of generations have to pass by since the mutation appears until the first homozygotic mutant individual appears. In the meanwhile, the mutation has to survive in heterozygosity, and that's not easy due to the high death rates. In fact, both the theory and the simulations predict that the fate of a mutation is its desappearence. Well, the classic theory says that always an heterozygous individual must remain, but if we take into account 'real-life' parameters such as finite population size, gene frequencies always rational (x/N, x and N being natural numbers), then sooner or later the mutation will be lost.

However that's not true always. By pure chance alone, it can happen that the cumulative effect of random sampling (unavoidable in finite-size populations) end up leading to the complete fixation of the new allele. This very seldom happens.

Selection can help. In 'the selective brake' I argued that selection indeed does accelerate the fixation of beneficial mutations, and here I will explain that a bit more. I've just mentioned that by sampling error, frequencies may vary randomly (Hardy-Weinberg law, the equilibrium of gene frequencies, is maintained in a stadistic fashion: the number of populations (or repetitions if this is a simulation) in which the mutant becomes fixed is proportional to the mutant gene frequency in the original population (1/2N); that is, we need N repetitions/sample populations for the mutation to become fixed in one). Therefore, it's possible that in a good luck strike the heterozygote's frequency rises high enough to reach the 'critical-mass' level and homozygous mutant individuals start appearing in the population. And then selection can see their mutant phenotype and fight against the rampant randomness.

But no matter how high the favorable selection coefficiente is, the fate of the mutation is still under the power of the blind, sheer, luck, since it has to survive the heterozygous period "dormant". The hopeful monster is against all odds.

But what about if heterozygotes do have a selective advantage? There are 4 possible scenarios.
Let's use w for fitness, and the sub-index will tell which genotype it's referring: (m) homozygous mutant, (h) heterozygote, (n) homozygous normal; w_m>w_n

1. w_m > w_h = w_n
2. w_m > w_h > w_n
3. w_m = w_h > w_n
4. w_m < w_h > w_n

We have commented already the first one, the heterozygote not having any selective advantage, only homo-mutant does. Next case is the one in which the heterozygot has some advantage over the homo-normal, but not as much as the homo-mutant (codominant heredity, incomplete heredity). In case number 3 the heterozygote has the same advantage as the homozygous mutant (dominant heredity). The last one is the most extreme situation, the heterozygote has more selective advantage than either kind of homozygotes (heterosis, also known as heterozygote superiority; most famous example is HbS and falciform anemy).

In theory and in silico, the higher the fitness of the heterozygote, the faster the mutation spreads. From this Mazo concludes that if there isn't any selection favouring the heterozygote (all except the number 1), the mutant doesn't have much of a fate.

But those aren't all the possibilities there are. We are assuming random mating, so what happens if there's inbreeding? The initial heterozygote has mixed descendence, but if they do mate between themselves, since the F2 there can be homozygous mutants showing up the new phenotype. Of course, consanguinity will rise, but then it's a matter of balancing the selection against the genetic load with the advantage of the new mutation. And there are several degrees of inbreeding. In nature there's a tendency to mate in the whereabouts an individual was born, so seldom times there's panmixia (even in those cases with high dispersion of seeds or gametes, the chances of mating close to the parent are quite higher than not doing so). The consequence is that, if the species doesn't have much mobility, then the locality is enriched with the new allele and therefore new homo-mutants can appear more often.

The second way of avoiding panmixia is called 'assortative mating'. The individuals don't choose randomly who to mate with, they have their preferences (even pollen grains have their ways to tell who are they mating with). If we can imagine an individual able to distinguish those who bear the mutant allele (for good or bad) and decide to mate accordingly, then the chances of heterozygotes meeting each other rise and therefore homozygous mutants appear sooner.

I like the later a lot, because even without any selective advantage for the heterozygotes, it's possible to produce two populations living alongside, or even one displacing the other! The combination of non-random mating (inbreeding or assortative mating) and positive selection for the recesive mutant is powerful enough as for beating the effect of random sampleing and achieving that, in the end, the lone mutant be a hopeful one.