Revisiting the question: When is a centromere not a kinetochore?

Centromeres have been the focus of extensive research for almost a century, so it may come as a surprise that a consistent definition and nomenclature for these structures remains elusive. In recent times, centromeric chromatin is most frequently defined by the presence of nucleosomes containing the H3 variant CENP-A and is typically synonymous with the site of the inner-kinetochore. However, crucial mammalian centromere proteins including CENP-B and INCENP have well defined distributions that show very little overlap with CENP-A. Additional protein localisations spanning the primary constriction or forming a band below CENP-A chromatin have been reported. Together, these observations suggest a complex and multi-layered chromatin organisation that is not well served by the canonical dichotomy of 'centromeric' and 'pericentromeric' chromatin. Strikingly, this is not a new observation but was made soon after the discovery of CENP proteins, including in a 1991 publication titled 'When is the centromere not a kinetochore?'. Here we revisit this question, which has become more pertinent following technical innovations in long-read sequencing and super-resolution microscopy. We present a model of centromere organisation for monocentromeres that incorporates additional complexity. We then use this model to reconceptualise diverse centromere forms in other eukaryotes including regional centromeres, holocentromeres and centromeres that lack key proteins including CENP-A. In this way, we hope to move towards a unified understanding of centromeric chromatin.