Catalase Website IndexJamesTechnoEvolution •Meta Genes •

Evolution - Meta Genes

1.1 Coselection

Can evolution act on genes which do not have a direct effect on the phenotype?

My answer is that yes, in some circumstances it can. As an example, variability can be selected for at the same time as a changed feature. I call this 'coselection'.

Index1.1 Co-Selection

1.1.1 The VL Model.

To show how coselection can happen I offer a simplified model.

We start with a herd of short necked giraffes in an environment where longer necks would be a distinct advantage. We suppose a gene L for neck length and a gene V for variability of neck length. Gene V has an effect on the descendants of the giraffe rather than the giraffe itself, moreover it acts indirectly. V is in fact a gene specific mutagen. Variants of V influence the likelihood of changes in L, but do not materially influence likelihood of changes in other genes. There are three variants or alleles of V, which we'll call V+, V0 and V-.

  • V+ tends to modify L so as to increase length,
  • V0 has negligible effect and
  • V- tends to modify L so as to decrease length.
With this setup, after a few generations the population is going to be enriched for V+ giraffes. Amongst longer necked giraffes, all of which have a modified L, more will have V+ than in the unchanged population.

The population after selection will have more variable neck length than before.

It is interesting that in 'Origin of Species' Darwin remarks (Chapter V) that 'A part developed in any species in an extraordinary degree or manner, in comparison with the same part in allied species, tends to be highly variable'. The arms of ourang-outangs, the valves of pyrogma, the beaks of tumbler pigeons are given as examples.

Index1.1 Co-Selection

1.1.2 Mechanism for V+

At this stage of the argument a possible objection is that a gene such as V and its variants is preposterous. How could one gene 'know' to make directed rather than random changes in another?

The reasonableness of a gene such as V+ rests on known mechanisms in molecular biology.

One way V+ could work depends on L being a promoter, a DNA sequence which controls quantities of a particular protein. Different promoters have different strengths, the strength depends on the DNA sequence. Such things as:
  • The promoter's distance form the start of the gene it promotes,
  • The composition of the DNA (AT rich DNA unwinds easily, GC rich less so),
  • The number of binding sequences,
These each have an effect on promoter strength. V is a gene which affects likely mutations in the promoter L. For example we could have:
  • V+ tends to make the promoter more AT rich,
  • V- tends to make the promoter more GC rich, whilst
  • V0 is neutral.
V is some signal in the DNA sequence which causes a 'hotspot' region for changes to GC or AT. The proximity of the signal to the gene L is the reason why V affects L rather than some other gene.

This kind of model doesn't demand extraordinary new abilities on the part of DNA.

A still more biologically realistic mechanism would involve changes in short range repetitions in DNA near to the promoter through a process known as 'slippage' rather than simple base substitutions - but it isn't necessary to go into details of such a mechanism. For this discussion it's only necessary to show one way that V could act.

Index1.1 Co-Selection

1.1.3 Evolution of V+

A second objection to the gene V is 'How could it arise by chance?'.
It could arise as an indirect result of economisation of energy expenditure in DNA repair. Repair of mutations in coding regions of DNA is more important than repair of junk and non-coding DNA. Genomes that focus their repair work on coding regions would be at a selective advantage. These genomes would go wrong less often. This implies some marker in DNA that modifes local repair properties. As soon as we have a mechanism that allows for differential repair of DNA, genes such as V become possible.

E.Coli does in fact have 'hotspots' for change where a CG pair is up to a hundred times more likely to mutate to an AT pair than is normally the case. The mechanism is related to DNA repair mechanims and depends on modification of cytosine to 5-methylcytosine by a methylase enzyme.

Index1.1 Co-Selection

1.1.4 Two Allele Variant

With genes L and and gene V having alleles V+, V0 and V- we see over time an increase in V+ in the population.

An elaboration of the model is that it still works if instead of three alleles for V we have just two:

  • V0 - As before, has no effect.
  • V* - Increases mutation rate in L, as frequently decreasing length as increasing it.
This is counter intuitive. It seems that any indirect benefit V* offers by increasing length should be offset by a disadvantage from an equally frequent decrease in length. However, provided V* has a moderate or weak effect and only infrequently affects L it will nevertheless come to dominate.

This works because beneficial mutations to L spread. When a V* leads to an increase in neck length, over a number of generations that individual's modified L and with it the V* will spread pretty much throughout the entire population, since the modified L confers a selective advantage. Many copies of V* are gained. However, when an individual's V* gene leads to decrease in length, only one copy of V* is lost from the population. A V*, provided it does not act to modify L too frequently is thus a net asset and gets coselected with beneficial changes in L.

Index1.1 Co-Selection

1.1.5 Evolution of indirectly acting Genes

In either model once the ideal length has been reached, V+ and V* both become a liability and selection will favour a reversion to V0.

Whilst both versions are compatible with 'punctuated evolution' - a burst of rapid change in population interspersed with pewriods of little change - the point that's important for the discussion is that a 'meta gene', a gene which has only an indirect effect on phenotype, can get selected with the gene that it affects. Good 'meta genes' are ones which speed up the process of evolution of the genes which they affect and/or which reduce the cost of bad trials. Bad 'meta genes' do the reverse and should be eliminated quite rapidly, whereas good ones will survive.

Because of coselection, meta genes are subject to evolution, they vary, will be selected for and will be refined.

Index1.1 Co-Selection

1.1.6 Variable Variability for Genome

The examination of the effect of variable variability on a single gene can be extended to a whole genome.

In general a genome in which the mutation rate for all genes is the same and does not change will adapt to a new environment less rapidly than one in which each gene has its own independent variability which has the potential to change. With variable variability evolution can select for gene variability, not just for good genes. This can 'focus' variability onto those genes which are most relevant to matching the environment whilst avoiding excessive mutation rates on genes that are close to optimum. In situations where one aspect of the environment changes, and multiple changes to the same gene are called for, variable variability is a net asset.

Catalase Website IndexJamesTechnoEvolution •Meta Genes •
"Evolution - Meta Genes" page last updated 5-July-2003