Rapid proliferation and nucleolar organizer targeting centromeric retrotransposons in cotton

Abstract

Centromeric chromatin in most eukaryotes is composed of highly repetitive centromeric retrotransposons and satellite repeats that are highly variable even among closely related species. The evolutionary mechanisms that underlie the rapid evolution of centromeric repeats remain unknown. To obtain insight into the evolution of centromeric repeats following polyploidy, we studied a model diploid progenitor (Gossypium raimondii, D-genome) of the allopolyploid (AD-genome) cottons, G. hirsutum and G. barbadense. Sequence analysis of chromatin-immunoprecipitated DNA showed that the G. raimondii centromeric repeats originated from retrotransposon-related sequences. Comparative analysis showed that nine of the 10 analyzed centromeric repeats were absent from the centromeres in the A-genome and related diploid species (B-, F- and G-genomes), indicating that they colonized the centromeres of D-genome lineage after the divergence of the A- and D- ancestral species or that they were ancestrally retained prior to the origin of Gossypium. Notably, six of the nine repeats were present in both the A- and D-subgenomes in tetraploid G. hirsutum, and increased in abundance in both subgenomes. This finding suggests that centromeric repeats may spread and proliferate between genomes subsequent to polyploidization. Two repeats, Gr334 and Gr359 occurred in both the centromeres and nucleolar organizer regions (NORs) in D- and AD-genome species, yet localized to just the NORs in A-, B-, F-, and G-genome species. Contained within is a story of an established centromeric repeat that is eliminated and allopolyploidization provides an opportunity for reinvasion and reestablishment, which broadens our evolutionary understanding behind the cycles of centromeric repeat establishment and targeting.

Centromeric DNA characterization in the model grass Brachypodium distachyon provides insights on the evolution of the genus

“…Each Bd21 centromere, similar to previously studied centromeres (Gong et al ., ; Han et al ., ), contained nucleosome blocks that were depleted of CENH3 (i.e. H3 subdomains; Figure S4); thus, they present the expected pattern of intermingled H3 and CENH3 subdomains (Figure ).…”

“…Cent4 was also small (275 kb), and contained both small H3 and CENH3 subdomains, with a maximal size of 7.6 and 25.6 kb, respectively (Figure ; Table ). Some of the small subdomains were close to the statistical threshold for being annotated as CENH3 and H3 subdomains, however, as described in previous studies using similar strategies (Gong et al ., ; Han et al ., ). This phenomenon is likely to be caused by the nature of the data, which were derived from cells at different stages of the cell cycle.…”

“…In the Bd21 centromeres, numerous unassembled gaps (Figure ; International Brachypodium Initiative, ), which most likely represent repetitive centromeric DNA sequences, might indicate that the current genome assembly has underestimated the repetitive sequence content of the centromeres. To solve this problem, we adopted a method to assess the repetitive DNA composition of the centromeres based on the de novo repetitive sequence annotation (Gong et al ., ; Han et al ., ). After analysis of input data (3.3 million sequence reads; see Experimental procedures), we generated a total of 482 677 clusters, which represented different repeat families of the Bd21 genome and accounted for 86.79% of the genome (Figure S6).…”

“…Thirdly, the Bd21 centromeres also showed a lack of active genes. The incompatibility between active transcription and centromeres has been described for both the point centromeres of budding yeast and the regional centromeres of other eukaryotes (Chlebowicz‐Sledziewska and Sledziewski, ; Nagaki et al ., ; Gong et al ., ; Han et al ., ). It has been suggested that this phenomenon arises from the existence of CENH3, which differs from the canonical H3 and leads to the failure of the histone modification pathway that is specific to H3 nucleosomes (Wang et al ., ).…”

“…Secondly, we found that CENH3 binding is discontinuous and is interrupted by canonical H3 nucleosomes in the Bd21 centromeres. This pattern of interspersed CENH3‐ and H3‐containing nucleosomes is an evolutionarily conserved aspect of the centromere structure, and has previously been found in animals and plants (Blower et al ., ; Nagaki et al ., ; Gong et al ., ; Han et al ., ; Zhu et al ., ). This structure is consistent with the physical models of centromeres, in which blocks of CENH3 nucleosomes are oriented on the exterior, facing the kinetochore, whereas the blocks of H3 nucleosomes are located in the interior (Blower et al ., ).…”