Large palindromes on the primate X Chromosome are preserved by natural selection

Abstract

Mammalian sex chromosomes carry large palindromes that harbor protein-coding gene families with testis-biased expression. However, there are few known examples of sex-chromosome palindromes conserved between species. We identified 26 palindromes on the human X Chromosome, constituting more than 2% of its sequence, and characterized orthologous palindromes in the chimpanzee and the rhesus macaque using a clone-based sequencing approach that incorporates full-length nanopore reads. Many of these palindromes are missing or misassembled in the current reference assemblies of these species' genomes. We find that 12 human X palindromes have been conserved for at least 25 million years, with orthologs in both chimpanzee and rhesus macaque. Insertions and deletions between species are significantly depleted within the X palindromes' protein-coding genes compared to their noncoding sequence, demonstrating that natural selection has preserved these gene families. The spacers that separate the left and right arms of palindromes are a site of localized structural instability, with seven of 12 conserved palindromes showing no spacer orthology between human and rhesus macaque. Analysis of the 1000 Genomes Project data set revealed that human X-palindrome spacers are enriched for deletions relative to arms and flanking sequence, including a common spacer deletion that affects 13% of human X Chromosomes. This work reveals an abundance of conserved palindromes on primate X Chromosomes and suggests that protein-coding gene families in palindromes (most of which remain poorly characterized) promote X-palindrome survival in the face of ongoing structural instability.

Figure 1.

Overview of human X-Chromosome palindromes. (A) Schematic of a palindrome. (B) Schematic of a triangular dot plot. Dots are placed at a 90° angle between identical k-mers, or “words,” within a DNA sequence. Palindromes appear as vertical lines. (w) Word size used to construct the dot plot. (C) Triangular dot plot for human X palindrome P3, including annotated protein-coding genes.

Figure 2.

Improvements to prior reference assemblies for chimpanzee and rhesus macaque. (A) Sequencing approach. Top bar: Locations of 26 human X palindromes (blue bands). Gray band shows centromere location. Bottom: Expansion of a single region containing a human X palindrome (solid black box). One or more clones were selected to span orthologous regions in chimpanzee and rhesus macaque (dashed black boxes). Tree shows estimated divergence times from TimeTree (Kumar et al. 2017). (B) Full-length nanopore reads supporting the structure of a single finished chimpanzee clone (CH251-385I8). (C,D) Two primate X palindromes resolved using SHIMS 3.0 that were missing or misassembled in existing X-Chromosome assemblies. (C) Triangular dot plots from the region orthologous to human P9 in chimpanzee assemblies Pan_tro_3.0 (Kuderna et al. 2017), Clint_PTRv2 (Kronenberg et al. 2018), and SHIMS 3.0. (D) Triangular dot plots from the region orthologous to human P8 in rhesus macaque assemblies Mmul_8.0.1 (Zimin et al. 2014), Mmul_10 (Warren et al. 2020), and SHIMS 3.0.

Figure 3.

Conservation of X-Chromosome palindromes across primates. (A) Conservation status of 26 human palindromes in chimpanzee and rhesus macaque. (B) Triangular dot plots for a palindrome (P26) conserved between human, chimpanzee, and rhesus macaque. Images are to scale. (C) Arm-to-arm divergence within species (above blue arrows) versus between species (left of blue arrows). Values are the average percent divergence across 12 palindromes shared by human, chimpanzee, and macaque. Divergence times estimated using TimeTree (Kumar et al. 2017).

Figure 4.

Structural changes between orthologous X-Chromosome palindromes are concentrated around the center of symmetry. (A) Square dot plots comparing the center of the palindrome, including the spacer and 10 kb of inner arm sequence on each side, between the indicated species. (Orthologous spacer) >20% of the spacer from one species aligned to the spacer from the other, in either the same orientation (“Human configuration”) or opposite orientation (“Inversion”); (Non-orthologous) <20% of the spacer from one species aligned to the spacer from the other. Values show the number of orthologous palindromes shared by human, chimpanzee, and macaque in each category. (B) Average fraction of sequence that could be aligned between species. (*) P < 0.05, (**) P < 0.01, Mann–Whitney U test. (C) Sizes of human spacers, binned according to the species between which they are conserved.

Figure 5.

Natural selection has preserved X-palindrome gene families. Results based on 12 palindromes conserved between human, chimpanzee, and rhesus macaque. (A) Square dot plot comparing structure of P25 between human and rhesus macaque. Indels highlighted in gray; nearly all fall between protein-coding genes. (B) Fraction of bases within indels for protein-coding gene sequence versus all other sequence. Results are the average of all pairwise species comparisons. Indels were defined as uninterrupted stretches of at least 1 kb in one species without orthologous sequence in the other species. (**) P < 0.01, (***) P < 0.001, Mann–Whitney U test.

Figure 6.

Deletions are enriched in human X-Chromosome spacers. (A) Normalized coverage depths for P2 spacer. Red arrow indicates four X Chromosomes with depth near 0. (B) Coverage depths across P2 and flanking sequence for two individuals with reference structure (HG02398, NA20897) and four with spacer deletions (NA21117, NA20905, HG04015, HG02687). (C) Square dot plot comparing palindrome centers (spacer + 10 kb inner arm on each side) for P17 reference structure and P17 deletion. (D) Frequency of P17 spacer deletions across five superpopulations from 1000 Genomes. (EUR) European, (AFR) African, (AMR) Admixed Americas, (EAS) East Asian, (SAS) South Asian. (E,F) Frequency of deletions detected in palindrome spacers compared to palindrome arms and flanking sequence. Size-matched regions from palindrome arms and single-copy sequence were selected at random; results from 100 iterations are shown.