The (r)evolution of SINE versus LINE distributions in primate genomes: sex chromosomes are important

Abstract

The densities of transposable elements (TEs) in the human genome display substantial variation both within individual chromosomes and among chromosome types (autosomes and the two sex chromosomes). Finding an explanation for this variability has been challenging, especially in light of genome landscapes unique to the sex chromosomes. Here, using a multiple regression framework, we investigate primate Alu and L1 densities shaped by regional genome features and location on a particular chromosome type. As a result of our analysis, first, we build statistical models explaining up to 79% and 44% of variation in Alu and L1 element density, respectively. Second, we analyze sex chromosome versus autosome TE densities corrected for regional genomic effects. We discover that sex-chromosome bias in Alu and L1 distributions not only persists after accounting for these effects, but even presents differences in patterns, confirming preferential Alu integration in the male germline, yet likely integration of L1s in both male and female germlines or in early embryogenesis. Additionally, our models reveal that local base composition (measured by GC content and density of L1 target sites) and natural selection (inferred via density of most conserved elements) are significant to predicting densities of L1s. Interestingly, measurements of local double-stranded breaks (a 13-mer associated with genome instability) strongly correlate with densities of Alu elements; little evidence was found for the role of recombination-driven deletion in driving TE distributions over evolutionary time. Thus, Alu and L1 densities have been influenced by the combination of distinct local genome landscapes and the unique evolutionary dynamics of sex chromosomes.

Figure 1.

Relative contribution to variability explained (RCVE) for each genome landscape feature significant to modeling variation in densities of Alus (A) and L1s (B) in 1-Mb windows across the human genome. Results of regressions with either the Alu or L1 density of various evolutionary integration timings as response and genome-wide features as predictors are depicted as bar plots (for details, see Supplemental Table S3). Color-coded areas correspond to the relative share (RCVE) that each predictor contributes to the total variability explained, in the presence of all other predictors (for details, see Methods). Bar plots are proportional to the sum of the RCVEs for each multiple regression model. (HC) Human–chimpanzee branch-specific elements; (HCO) human–chimpanzee–orangutan branch-specific elements.

Figure 2.

Human sex chromosome (X, Y) versus autosome (A) distribution of observed (A) and corrected (B) densities of Alus in 1-Mb windows genome-wide. (A) Observed densities are plotted for Alu elements of various evolutionary ages including human-specific AluYs; human–chimpanzee branch-specific (HC) AluYs; and human–chimpanzee–orangutan branch-specific (HCO) AluYs; AluS and AluJ subfamilies, on the autosomes (white), X (light gray), and Y (dark gray) chromosomes. (B) Residuals from genome-wide multiple regression models represent densities corrected for local variation in genome landscape features. Notches on boxplots indicate the 95% confidence interval of the median.

Figure 3.

Human sex chromosome (X, Y) versus autosome (A) distribution of observed (A) and corrected (B) densities of L1 elements in 1-Mb windows genome-wide. (A) Observed densities are plotted for L1s of various evolutionary ages including human-specific L1s; human–chimpanzee branch-specific (HC) L1PAs; and human–chimpanzee–orangutan branch-specific (HCO) L1PAs, L1PB, and L1M subfamilies, on the autosomes (white), X (light gray), and Y (dark gray) chromosomes. (B) Residuals from genome-wide multiple regression models represent densities corrected for local variation in genome landscape features. Notches on boxplots indicate the 95% confidence interval of the median.

Figure 4.

Sex-chromosome biased patterns of corrected TE densities in primates. Relative densities of Alus (A) and L1s (B), after accounting for regional variation in local genome landscape features, are compared among the X, Y, and autosomes (A) for each branch in the primate phylogeny. Statistically significant differences are indicated with inequality symbols (>), whereas insignificant differences are indicated with equality symbols (≥).