The advantages and disadvantages of horizontal gene transfer and the emergence of the first species

Abstract

Background: Horizontal Gene Transfer (HGT) is beneficial to a cell if the acquired gene confers a useful function, but is detrimental if the gene has no function, if it is incompatible with existing genes, or if it is a selfishly replicating mobile element. If the balance of these effects is beneficial on average, we would expect cells to evolve high rates of acceptance of horizontally transferred genes, whereas if it is detrimental, cells should reduce the rate of HGT as far as possible. It has been proposed that the rate of HGT was very high in the early stages of prokaryotic evolution, and hence there were no separate lineages of organisms. Only when the HGT rate began to fall, would lineages begin to emerge with their own distinct sets of genes. Evolution would then become more tree-like. This phenomenon has been called the Darwinian Threshold.

Results: We study a model for genome evolution that incorporates both beneficial and detrimental effects of HGT. We show that if rate of gene loss during genome replication is high, as was probably the case in the earliest genomes before the time of the last universal common ancestor, then a high rate of HGT is favourable. HGT leads to the rapid spread of new genes and allows the build-up of larger, fitter genomes than could be achieved by purely vertical inheritance. In contrast, if the gene loss rate is lower, as in modern prokaryotes, then HGT is, on average, unfavourable.

Conclusions: Modern cells should therefore evolve to reduce HGT if they can, although the prevalence of independently replicating mobile elements and viruses may mean that cells cannot avoid HGT in practice. In the model, natural selection leads to gradual improvement of the replication accuracy and gradual decrease in the optimal rate of HGT. By clustering genomes based on gene content, we show that there are no separate lineages of organisms when the rate of HGT is high; however, as the rate of HGT decreases, a tree-like structure emerges with well-defined lineages. The model therefore passes through a Darwinian Threshold.

Figure 1

Mean fitness of the population versus HGT rate, h, for three different rates of gene deletion, v.

Figure 2

Mean values of properties of the population as a function of h for the case of high gene deletion rate, v = 0.01. $\overline{n}$, number of genes per individual; $\overline{n_{types}}$, number of different types of gene per individual; $\overline{n_{pan}}$, number of different types of gene in the whole population; p_useful, probability that a horizontally transferred gene is useful to the receiving organism.

Figure 3

Variation of mean deletion rate, $\overline{v}$, and mean HGT rate, $\overline{h}$, as a function of time in simulations in which both quantities are heritable. Error bars show standard deviations over five runs.

Figure 4

Clustering individuals according to similarity of genomes.