Transposable element expression is associated with sex chromosome number in humans

Abstract

Why women live longer than men is still an open question in human biology. Sex chromosomes have been proposed to play a role in the observed sex gap in longevity, and the Y male chromosome has been suspected of having a potential toxic genomic impact on male longevity. It has been hypothesized that transposable element (TE) repression declines with age, potentially leading to detrimental effects such as somatic mutations and disrupted gene expression, which may accelerate the aging process. Given that the Y chromosome is rich in repeats, age-related increases in TE expression could be more pronounced in males, likely contributing to their reduced longevity compared to females. In this work, we first studied whether TE expression is associated with the number of sex chromosomes in humans. We analyzed blood transcriptomic data obtained from individuals of different karyotype compositions: 46,XX females (normal female karyotype), 46,XY males (normal male karyotype), as well as males with abnormal karyotypes, such as 47,XXY, and 47,XYY. We found that sex chromosomes might be associated to TE expression, with the presence and number of Y chromosomes particularly associated with a global increase in TE expression. This tendency was also observed across several TE subfamilies. We also tested whether TE expression is higher in older males than in older females using published human blood transcriptomic data from the Genotype-Tissue Expression (GTEx) project. However, we did not find increased TE expression in older males compared to older females probably due to the heterogeneity of the dataset. Our findings suggest an association between sex chromosome content and TE expression and open a new window to study the toxic effect of the Y chromosome in human longevity.

Fig 1. TE expression in the different karyotypes after removing batch effect, considering (A) all TE subfamilies (1,212 TE subfamilies) and (B) TE subfamilies belonging to LTR retrotransposons (582 TE subfamilies), (C) SVA non-LTR retrotransposons (6 TE subfamilies), (D) SINE non-LTR retrotransposons (62 TE subfamilies), (E) LINE non-LTR retrotransposons (175 TE subfamilies), or (F) DNA elements (285 TE subfamilies).

Global TE expression was measured in each karyotype (x-axis) as the proportion of TE read counts among all read counts (TEs and genes) (y-axis). In this calculation, read counts were used after DESeq2 normalization (“normalized counts”) to remove any depth sequencing bias. Each dot represents one individual. Dots are colored according to karyotype (four levels: XX, XY, XXY, XYY). The age of each individual is indicated in a labeled box linked to the corresponding dot. Batch effect was removed before graphical display. P-values from Wilcoxon test comparing pairwise karyotypes: (*) p-value < 0.05 and (**) p-value < 0.01..

Fig 2. Boxplots for the 15 most significantly differentially expressed TE subfamilies according to karyotype using the likelihood-ratio test, after removing batch effect.

Specific TE subfamily expression was estimated using the number of read counts after DESeq2 normalization (“normalized counts”) to remove any depth sequencing bias. Each dot represents one individual. Dots are colored according to karyotype (four levels: XX, XY, XXY, XYY). Batch effect was removed before graphical display. Adjusted p-values from DESeq2: (*) p-value < 0.05, (**) p-value < 0.01, (***) p-value < 0.001. All of these 15 subfamilies had at least one copy on the Y chromosome, except for HERVH48-int, in the TE sequence reference file. All of these 15 subfamilies had at least one copy on the X chromosome in the TE sequence reference file. Underlined bold TE subfamilies are enriched in the Y chromosome (binomial test adjusted p-value < 0.05 and number of observed copies located on Y chromosome> expected). See Fig L in S1 File for the other significantly differentially expressed TE subfamilies according to karyotype. See S5 Data for number of TE subfamilies upregulated and downregulated per TE family or TE superfamily/order/class in all the pairwise karyotype comparisons.

Fig 3. TE expression in the filtered GTEx dataset (no disease group) according to sex (A) regardless of age group or (B) also according to age group.

Global TE expression was measured in each karyotype (x-axis) as the proportion of TE read counts among all read counts (TEs and genes) (y-axis). In this calculation, read counts were used after DESeq2 normalization (“normalized counts”) to remove any depth sequencing bias. Each dot represents one individual. Dots are colored according to sex. Sex variable has two levels: female, male. Age group has five levels: [20-30], ]30-40], ]40-50], ]50-60], and ]60-70]. (A) Proportion of TE read counts among all read counts was modeled using a linear model with sex and age group as independent variables. From this model, the p-value testing the effect of sex on proportion of TE read counts among all read counts adjusted on age group was not significant (p-value = 0.788). (B) A Kruskal-Wallis test comparing proportion of TE read counts among all read counts across age groups was used in females and then in males. This test was not significant in females (p-value = 0.585) as well as in males (p-value = 0.471).

Fig 4. Normalized counts according to sex and age group for the 3 TE subfamilies that are differentially expressed between males and females adjusted on age group in the filtered GTEx dataset (no disease group).

Specific TE subfamily expression was estimated using the number of read counts after DESeq2 normalization (“normalized counts”) to remove any depth sequencing bias. All of these 3 TE subfamilies had at least one copy on the Y chromosome and one copy on the X chromosome, but none are enriched on the Y chromosome. For boxplots of normalized counts according to sex and age group for the significantly differentially expressed TE subfamilies according to age group see Fig P in S1 File.