DNA sequence preference for de novo centromere formation on a Caenorhabditis elegans artificial chromosome
Centromeric DNA sequences vary in different species, but share common characteristics, like high AT-content, repetitiveness, and low, but not no, transcriptional activity. Yet, neocentromeres can be found on non-centromeric, ectopic sequences, suggesting that centromeres can be established and maintained epigenetically. In contrast, canonical centromeric DNA sequences are more competent in de novo centromere formation on artificial chromosomes (ACs).
To determine if specific DNA sequence features are preferred for new centromere formation, we injected different DNA sequences into the gonad of a holocentric model organism, Caenorhabditis elegans, to form ACs in embryos, and monitored mitotic AC segregation. We demonstrated that AT-rich sequences, but not repetitive sequences, accelerated de novo centromere formation on ACs. We also injected fragmented Saccharomyces cerevisiae genomic DNA to construct a less repetitive, more complex AC that can propagate through generations.
By whole-genome sequencing and de novo assembly of AC sequences, we deduced that this AC was formed through non-homologous end joining. By CENP-AHCP-3 chromatin immunoprecipitation followed by sequencing (ChIP-seq), we found that CENP-AHCP-3 domain width on both the AC and endogenous chromosomes is positively correlated with AT-content.
Besides, CENP-AHCP-3 binds to unexpressed gene loci or non-genic regions on the AC, consistent with the organization of endogenous holocentromeres.
