Functional CENP-B boxes are selectively conserved at centromeric regions

Human centromeres are built over large stretches of repetitive, divergent -satellite DNA. Within these sequences lies a conserved, defined 17-bp sequence named the CENP-B box that is bound by the DNA-binding protein CENP-B. Recent studies have proposed that CENP-B box motifs along chromosome arms exist and may represent conserved, ectopic binding sites with functional relevance. Here, we evaluate the genomic distribution, conservation, and binding capacity of different CENP-B box motifs outside canonical centromeres. Analysis of thousands of complete human centromere assemblies reveals exceptional conservation of a unique and canonical CENP-B box motif within centromeres. This is in contrast with the high sequence variability and stochastic occurrence of more degenerate CENP-B box motifs along chromosome arms. Consistently, CENP-B binds only at canonical CENP-B box motifs embedded within -satellite sites, with no evidence of functional binding at any ectopic sites. Together, these results indicate an adaptive selection of canonical CENP-B boxes within centromeric regions, in contrast to random sequence occurrences for ectopic CENP-B box-like motifs.