Analysis of transposon interruptions suggests selection for L1 elements on the X chromosome

Abstract

It has been hypothesised that the massive accumulation of L1 transposable elements on the X chromosome is due to their function in X inactivation, and that the accumulation of Alu elements near genes is adaptive. We tested the possible selective advantage of these two transposable element (TE) families with a novel method, interruption analysis. In mammalian genomes, a large number of TEs interrupt other TEs due to the high overall abundance and age of repeats, and these interruptions can be used to test whether TEs are selectively neutral. Interruptions of TEs, which are beneficial for the host, are expected to be deleterious and underrepresented compared with neutral ones. We found that L1 elements in the regions of the X chromosome that contain the majority of the inactivated genes are significantly less frequently interrupted than on the autosomes, while L1s near genes that escape inactivation are interrupted with higher frequency, supporting the hypothesis that L1s on the X chromosome play a role in its inactivation. In addition, we show that TEs are less frequently interrupted in introns than in intergenic regions, probably due to selection against the expansion of introns, but the insertion pattern of Alus is comparable to other repeats.

Figure 1. Examples of transposon clusters.

A) An ancient L1MB8 element was interrupted by an Alu and a LTR repeat, which in turn was further interrupted by two Alus. B) Alu repeats interrupted by an L1 and other Alus. The UCSC custom track that visualizes TE clusters is downloadable from. http://www.mssm.edu/labs/warbup01/paper/files.html.

Figure 2. The pattern of interruptions of L1s on the autosomes and X chromosomes.

A) The approximate position of evolutionary strata and the pseudoautosomal regions on the human X chromosome and the homology with the X chromosome of the opossum. The largest and oldest stratum on the human X corresponds to the opossum X. B) The frequency of interruptions (per megabase of L1 sequence) of L1 elements in the different evolutionary strata of the X chromosome, and the median of autosomes (error bars indicate quartiles). Clusters containing at least one L1, and only non-L1 interruptions were analysed separately, as the interruption of an L1 repeat by another L1 may not result in loss of function of the locus. On the oldest evolutionary stratum (S1) L1s are significantly less frequently interrupted than on autosomes. C) The frequency of non-L1 interruptions in introns and intergenic sequences of the autosomes and stratum 1 of the X chromosome. The frequency of interruptions in intergenic sequences is grouped into 5 kb bins, as a function of the distance from the nearest gene. D) The frequency of L1 interruptions by non-L1 repeats on the opossum autosomes and X chromosome. The positions of interruptions are grouped in 300 bp bins across the consensus sequence. There is no significant difference in the frequency of interruptions between the autosomal and X-linked L1s. E) The frequency of interruptions by non-L1 repeats of all human L1 elements on the autosomes and stratum 1 of the X chromosome. There are two clear patterns: L1s on the autosomes are more frequently interrupted than on the S1 of the X chromosome, and the 3′UTRs of the L1 consensus sequence are targeted more frequently than other regions. However, no particular region of the L1 consensus is protected from interruptions. F) The frequency of interruptions by non-L1 repeats into the primate specific L1P clade on the autosomes and the three oldest strata (S1–S3) of the X chromosome. Interruptions are grouped into 300 bp bins on the autosomes and S1, and into 900 bp bins on the S2 and S3, due to their smaller size. Despite the accumulation of L1Ps on the X (which has been interpreted as a signature of their function in X inactivation), with the exception of the S2 region, there is only a minor difference between the frequency of interruptions on the autosomes and the X. Although there is a large difference between the frequency of interruptions in the 5′UTR, ORF and 3′UTR of the L1 consensus sequence, no region of L1Ps is free from interruptions. G) The frequency of interruptions by non-L1 repeats into the mammalian-wide L1M clade of L1s on the autosomes and the three oldest strata (S1–S3) of the X chromosome. Interruptions are grouped into 300 bp bins on the autosomes and S1, and into 900 bp bins on the S2 and S3, due to the smaller number of repeats. Unlike the primate specific L1Ps, L1Ms on the X chromosome are much less interrupted than on the autosomes.

Figure 3. The frequencies of interruptions into L1Ms, within 100 kb of genes, for the genes that escape inactivation on the S1, are inactivated on the S1, and on the autosomes.

L1Ms near genes escaping inactivation are interrupted at a higher frequency than near the inactivated ones, but at a lower rate than on the autosomes.

Figure 4. Correlations between the density of coding sequence in the euchromatic region of chromosomes (coding sequence/non-repetitive sequence), and the frequency of interruptions of the main non-LTR repeats.

The positions of selected chromosomes are indicated on the x-axis on panels A and C. Note the logarithmic X axis. Each cross represents an autosome, the three strata (S1–S3) of the X chromosome that contain the most inactivated genes are shown in red. With the exception of Alus, repeats are interrupted more frequently on gene dense chromosomes, and unlike L1s, other non-LTR repeats do not show a depletion of interruptions in the S1–S3 strata.

Figure 5. Interruption pattern of Alu repeats on autosomes.

A) The distribution of interruptions within the Alu consensus sequence. The vast majority of Alus and L1s insert into the poly-A stretch of the linker region, while the insertions of other repeats are distributed equally across the repeats. B) The distribution of Alu interruptions (standardized with the amount of Alu sequence) in introns and intergenic regions as a function of the distance from the nearest exon or gene, in 5 kb bins. The distribution of Alus interrupted by Alus and other repeats are indicated separately, both for intergenic and intronic repeats. C) The distribution of all TE interruptions in introns and intergenic regions.