The near-complete genome assembly of allotetraploid Pennisetum purpureum 'Purple' reveals the genetic and epigenetic landscape of centromeres

Abstract

Drastic karyotype changes are a major evolutionary force, potentially involving centromere position, number, distribution, or strength alterations. Yet, the genetic and epigenetic landscape of centromeres, especially in allopolyploid plants during subgenome reshuffling, remains poorly understood. Here, we present a near-complete chromosome-scale genome assembly of the allotetraploid Pennisetum purpureum 'Purple', resolving all 14 centromeres. We find that subgenome-biased expansion of six LTR retrotransposons drives architectural divergence between subgenomes. Centromeric satellite repeats (CentPs) show rapid sequence divergence across subgenomes and chromosomes, with CENH3 preferentially binding conserved higher order repeats. Intriguingly, centromeric retrotransposons in Pennisetum (CRPs) are evolutionarily younger compared to their noncentromeric counterparts, coupled with marked subgenome B-biased amplification. Notably, CRP insertions flanking CentP satellites correlate with elevated satellite DNA polymorphism, supporting a model wherein CentP homogenization processes actively purge retrotransposons from centromeric arrays. Despite rapid sequence diversification of centromeric repeats, the epigenetic landscapes remain evolutionarily conserved in the centromeres of two subgenomes. Additionally, comparative analyses across Pennisetum species demonstrate rapid species- and chromosome-level turnover of CentPs and CRPs. Overall, our study illuminates the genetic and epigenetic plasticity of centromeres in allopolyploids, revealing how centromeric repeats adapt post-subgenome reshuffling.

Figure 1

Synteny and the delimiting of centromeres in P. purpureum ‘Purple’. (A) Collinearity analysis between Purple-CEN and Purple-LZU. The collinearity between the two genome assemblies is shown as gray lines or blocks, and the inversions are shown in orange. The purple triangles indicate the presence of telomere sequence repeats. TE distribution is plotted above or under each chromosome in 100-kb bins, and the gene density in 100-kb windows is shown as green bars. (B) The delimiting of Purple-CEN centromeres. The layers of each chromosome graph indicate (1) the density of read mapping from CENH3 ChIP-seq with sliding windows of 10 kb shown in light blue, respectively; (2) the CentP satellite distribution; and (3) TE distribution, respectively. The dotted frame represents the defined centromere region.

Figure 2

Transposable element (TE) dynamics between two subgenomes of P. purpureum ‘Purple’. (A) Repeat sequence distribution patterns and LTR-RT types. (B) Classification of LTR-RT superfamilies. The subgenome location of Ty1-copia and Ty3-gypsy and the subgenomes are indicated by the outer circle. (C) Estimated times of intact Ty1-copia and Ty3-gypsy insertion in two subgenomes of P. purpureum ‘Purple’. (D) Characterization of SubA’-specific and SubB-specific LTR-RTs in P. purpureum ‘Purple’. The y-axis is the subgenome proportion ratios. It represents the relative enrichment of the corresponding LTR-RT in the corresponding subgenome. The x-axis is the genome proportion for each LTR-RT in the genome of P. purpureum ‘Purple’. (E) Chromosome distribution of subgenome-specific abundant LTR-RTs. (F) FISH mapping of two subgenome-specific abundant LTR-RTs in P. purpureum ‘Purple’. Chromosomes counterstained with DAPI. FISH signals of probes SubA’-Athila and SubB-Retand2 in red and green color, respectively. Scale bar = 5 μm.

Figure 3

Evolution of centromeric satellite repeats in the subgenomes of P. purpureum ‘Purple’. (A) CentP monomer lengths (bp) in the genome of P. purpureum ‘Purple’. (B) Sequence variation of CentP monomer in the genome of P. purpureum ‘Purple’. (C) CentP count in all the chromosomes of the two subgenomes. (D) Sequence variation of CentP monomer in all the chromosomes of the two subgenomes. (E) FISH mapping of six CentP monomers in P. purpureum ‘Purple’. Chromosomes counterstained with DAPI. FISH signals of six CentP probes in red color, and corresponding subgenome-specific abundant LTR-RTs in green color. Scale bar = 5 μm. (F) The lengths (bp) of HOR blocks (monomers) in the genome of P. purpureum ‘Purple’. (G) The distance between CentP HORs (kb) in the genome of P. purpureum ‘Purple’. (H) Chromosome-specific distribution of CentP HORs. (I) Heatmap of a representative region within CEN1A’, shaded according to pairwise variants between CentP. (J and K) The distribution of CentPs and their HORs in CEN7A’ and CEN7B.

Figure 4

Subgenome-biased amplification of CRP in P. purpureum ‘Purple’. (A) CENH3 enrichment level of CentP satellite, centromeric intact CRP and non-centromeric intact CRP elements in P. purpureum ‘Purple’. (B) The enrichment of CEN-CRP and non-CEN-CRP elements in SubA’ and SubB. (C) FISH mapping of two CRPs in P. purpureum ‘Purple’. Chromosomes counterstained with DAPI. FISH signals of two CRP probes in red color, and corresponding subgenome B-specific abundant LTR-RT SubB-Retand1 in green color. Scale bar = 5 μm. (D and E) The distribution of CRPs in CEN1A’ and CEN1B. (F) LTR identity of CEN-CRPs and non-CEN-CRPs in SubA’ and SubB. (G) The insertion time of CEN-CRPs and non-CEN-CRPs in SubA’ and SubB. (H) CentP identity surrounding centromeric CRP sites and in the flanking regions in SubA’ and SubB.

Figure 5

Epigenetic landscapes between two subgenomes in P. purpureum ‘Purple’. (A-C) Metaprofiles of CENH3, H3K4me3, and H3K9me2 ChIP-seq signals around centromeric CentP satellites and CRP retrotransposons, and CENH3 domains. (D) HiFi-derived percentage of DNA methylation in CHG contexts across three types of centromeres. (E-L) Plots of CENH3 ChIP enrichment in homogenized CentP regions (E-H) and in divergent CentP regions (I-L), averaged over windows centered on CentP-138, CentP-148, CentP-156 starts in SubA’ and SubB.

Figure 6

Evolution model of the dynamic centromeres in the genus Pennisetum. The centromeric sequences and structural conservation among Pennisetum species, originating from a shared ancestral foundation, underwent rapid post-speciation divergence accompanied by complete centromere turnover. Substantial differences in centromeric architecture in P. glaucum (2n = 2x = 14, x = 7) and Pennisetum alopecuroides (2n = 2x = 18, x = 9). In the tetraploid P. purpureum ‘Purple’ (2n = 4x = 28, x = 7), the fusion of centromeric repeats from its diploid progenitors has intensified subgenomic divergence of CentPs. The presence of chromosome-specific centromeric satellite clusters suggests a recombination-based homogenization mechanism operating at the chromosomal level. Notably, P. purpureum ‘Purple’ exhibits subgenome B-biased amplification of CRPs. Despite these genomic rearrangements, CENH3 nucleosome positioning remains relatively stable, and epigenetic homeostasis of centromeres is maintained in the tetraploid P. purpureum ‘Purple’.