Neutral theory of molecular evolution

Readers are recommended to first familiarise themselves with the background leading up to this model. See: The classical and balanced schools regarding genetic polymorphism

The neutral theory of molecular evolution is a model developed in the early 1970s following a series of experiments in the 1960s that suggested that genetic polymorphisms exist in high frequency in populations. Although a previous selectionist model offered by the balanced school proposed that polymorphisms were maintained by selecting for heterozygotes, many geneticists disputed this, saying that heterozygote advantage for all polymorphisms would create a large genetic load, whereby a significant proportion of each new generation (i.e. those who segregated to be homozygous) would have lowered fitness, theoretically creating a cumulative effect that could lead to the extinction of that population over many generations.

This lead to the development of the neutral theory, which views most polymorphisms (i.e. most mutations) as neutral in their effect on the organism. Unlike the previous selectionist theory, the neutral theory suggests that genetic drift is the dominant force acting on mutations, rather than natural selection.

Neutralists accept that there are a small number of deleterious mutations and a very small number of advantageous mutations that both may be acted on, either removed or fixed, by natural selection. However, they insist that most mutations are neutral, and of these the majority are lost randomly (by drift) while a few may stick around long enough to become fixed (again, by drift). In this context, the term fixed simply means that a mutant allele substitutes the original allele(s) in a population.

Kimura (1968) made some simple calculations to back up this ‘neutral’ theory:

µ = the rate of mutation per gene, per generation and N = the population size, then 2N = the number of alleles in a diploid population and 2Nµ = the number of new mutations per generation.

Most of the time, a new genetic mutation is lost by genetic drift. Rarely, however, it may last long enough to be fixed in the population and substitute the original allele for that gene. The probability of the new allele drifting into fixation is 1/2N – you have one allele within a population of 2N alleles that you want fixed.

The rate of substitution (i.e. how frequently a mutant allele successfully replaces the original) is calculated as 2Nµ x 1/2N which is equivalent to just µ, so the rate of neutral molecular evolution = neutral mutation rate. We can say that the rate of molecular evolution is independent of population size (N).

The average time for a neutral mutation to drift to fixation is 4N generations. Therefore, although the rate of neutral mutation is independent of population size, the rate of progression of the mutation throughout the population is proportional to population size.

In short, the neutral theory tells us that polymorphisms in a large population are the consequence of LOTS of neutral mutations arising and slowly drifting through the population at any one time. So, unlike the theory of balancing selection, the presence of these mutations is not statically maintained in heterozygotes, but instead is a ‘snapshot’ of mutations arising and drifting at any one time.

The neutralists do accept that there is a selective pressure, one which generally serves to remove deleterious mutations and has little role in fixing new mutations. The rate of fixation by drift is 1/2N and as long as this is higher than the rate of fixation by selection, then the effects of drift can be considered to outweigh the effects of selection.

In contrast, selectionists would argue that the new mutations are advantageous and are fixed by reproductive fitness, rather than the random process of drift from one generation to the next. Selectionists consider neutral mutations to be rare. Neutralists consider advantageous mutations to be rare.

The neutral theory tells us that the rate of substitution must be inversely proportional to the functional constraint of a gene. In other words, a gene will rarely be substituted for by drift (and thus have a LOW rate of evolution) when its exact or near-exact sequence is critical to its function because neutral mutations of such a gene will be rare and most will be deleterious (and thereby selectively removed from the gene pool).

We can test the neutrality of mutations using the ratio dN/dS.

By sequencing DNA from a variety of species, we can calculate the average synonymous rate of substitution (dS) and the average non-synonymous rate of substitution (dN). For clarification, dS is the frequency that a mutant allele substitutes the original allele when the mutation has not affected amino acid sequence, while dN is the frequency that a mutant allele substitutes the original allele when the mutation has coded for a new amino acid.

ω = dN/dS

We can assume that synonymous mutations, dS, are neutral and thus for most genes dN < dS and ω < 1.
If dN > dS and ω > 1, then changes in amino acid sequence (non-synonymous mutations) are occurring more frequently than silent mutations (synonymous) meaning we have a case of positive selection for changes in the amino acid sequence.

However we should bear in mind that ω is a value for an entire gene, and it is insensitive to higher/lower rates of mutation at sites within the gene. It is possible that only a few parts of the protein are under selection. This could mean that the ω > 1 signal coming from parts of the gene that are under positive selection are swamped by the ω < 1 signal coming from the majority of the gene.

Since the 1970s, the nearly neutral model has become most accepted. This model states that in introns and non-coding DNA, neutral mutations are universal (synonymous mutations); however exons can be subjected to mutations that are weakly deleterious or weakly advantageous (non-synonymous). Essentially, molecular evolution is the product of a mixture of genetic drift on synonymous mutations and weak selection on non-synonymous mutations. This is the currently accpeted model.