Kinetochore specification and assembly requires the targeted deposition of specialized nucleosomes containing the histone H3 variant CENP-A at centromeres. However, CENP-A is not sufficient to drive full-kinetochore assembly, and it is not clear how centromeric chromatin is established. Here, we identify CENP-W as a component of the DNA-proximal constitutive centromere-associated network (CCAN) of proteins. We demonstrate that CENP-W forms a DNA-binding complex together with the CCAN component CENP-T. This complex directly associates with nucleosomal DNA and with canonical histone H3, but not with CENP-A, in centromeric regions. CENP-T/CENP-W functions upstream of other CCAN components with the exception of CENP-C, an additional putative DNA-binding protein. Our analysis indicates that CENP-T/CENP-W and CENP-C provide distinct pathways to connect the centromere with outer kinetochore assembly. In total, our results suggest that the CENP-T/CENP-W complex is directly involved in establishment of centromere chromatin structure coordinately with CENP-A.

Engineered kinetochores reveal distinct functions of the CCAN in recruiting CENP-A to the centromere and acting as a structural core to directly recruit kinetochore proteins.

SummaryCentromeres are specified by sequence-independent epigenetic mechanisms in most organisms. Rarely, centromere repositioning results in neocentromere formation at ectopic sites. However, the mechanisms governing how and where neocentromeres form are unknown. Here, we established a chromosome-engineering system in chicken DT40 cells that allowed us to efficiently isolate neocentromere-containing chromosomes. Neocentromeres appear to be structurally and functionally equivalent to native centromeres. Chromatin immunoprecipitation sequencing (ChIP-seq) analysis with 18 neocentromeres revealed that the centromere-specific histone H3 variant CENP-A occupies an ∼40 kb region at each neocentromere, which has no preference for specific DNA sequence motifs. Furthermore, we found that neocentromeres were not associated with histone modifications H3K9me3, H3K4me2, and H3K36me3 or with early replication timing. Importantly, low but significant levels of CENP-A are detected around endogenous centromeres, which are capable of seeding neocentromere assembly if the centromere core is removed. In summary, our experimental system provides valuable insights for understanding how neocentromeres form.