page 765, Ohzeki et al. present evidence for sequence specificity in assembling chromatin at mammalian centromeres. The sequence requirements imply that construction of centromeric chromatin is more than an epigenetic process.
Centromeric chromatin assembles at AT-rich repeats called α satellite sequences, or alphoids. Type I alphoids contain boxes for the binding of CENP-B, one of several conserved protein components of centromeric chromatin. Alphoid sequences are also found in inactivated centromeres, and Y chromosomes form centromeric chromatin, although they lack CENP-B boxes. Thus, centromeres are thought to be inherited as a preassembled complex.
However, assembly of new centromeric chromatin (de novo formation) requires specific sequences, according to the new results. The authors transformed cells with synthetic type I alphoid DNA constructs and looked for the formation of mammalian artificial chromosomes (MACs), which by definition have assembled functional centromeres de novo. Only alphoid constructs with CENP-B boxes elicited efficient MAC formation. Binding of CENP-B to its cognate DNA sequence initiated assembly of other centromere-specific proteins, including CENP-A, -C, and -E. CENP-B was not sufficient, however, as CENP-B boxes in a GC-rich context did not establish MAC formation. Thus, initial CENP-B binding is sequence specific, whereas further chromatin assembly appears to require the AT-rich repeats. This may reflect the preference of the histone-like CENP-A or other centromere components for AT-rich sequences during nucleosome folding. ▪