Reconstructing the evolutionary history of transposable elements

Abstract

The impact of transposable elements (TEs) on genome structure, plasticity, and evolution is still not well understood. The recent availability of complete genome sequences makes it possible to get new insights on the evolutionary dynamics of TEs from the phylogenetic analysis of their multiple copies in a wide range of species. However, this source of information is not always fully exploited. Here, we show how the history of transposition activity may be qualitatively and quantitatively reconstructed by considering the distribution of transposition events in the phylogenetic tree, along with the tree topology. Using statistical models developed to infer speciation and extinction rates in species phylogenies, we demonstrate that it is possible to estimate the past transposition rate of a TE family, as well as how this rate varies with time. This methodological framework may not only facilitate the interpretation of genomic data, but also serve as a basis to develop new theoretical and statistical models.

Fig. 1.—

Single simulation of the temporal dynamics of a TE family with a constant transposition rate ( per copy and per time step), and no deletion (“pure birth” model). X axes are oriented from past to present in reconstructed dynamics (A, B, C) ( corresponds to the start of the transposition history, each bar stands for four successive generations). With a constant transposition rate per copy (dashed line on A), the number of copies increases exponentially. This increase is reflected by the log-linear pattern of the LTT plot (C), which can be used as a basis for reconstructing the dynamics of the TE family.

Fig. 2.—

Simulated LTT plots in four scenarios. Each line is the average over 1,000 replicates. The pure birth model corresponds to a transposition-only model; the birth–death model features both transpositions and deletions; and the increasing and decreasing birth models represent linear changes in the transposition rate (see Materials and Methods for details). Different transposition dynamics generate different LTT profiles, illustrating how the branching pattern from phylogenetic trees can be used to estimate the transposition history.

Fig. 3.—

ML reconstructed phylogenies for the four Fot subfamilies. Trees were rooted with the other subfamilies. Ultrametric trees were obtained through the “pathd8” algorithm (see “Materials and Methods”). Asterisks (*) denote nodes that are supported by bootstrap scores ≥50.

Fig. 4.—

Lineage-through-divergence plots for the four Fot subfamilies. The dashed line illustrates the expectation for a “pure birth” model (constant transposition, no deletions).

Fig. 5.—

Illustration of the estimated ML exponential dynamics (dots), and the corresponding 95% support intervals from 100 bootstrapped trees.