Kinetochores are structures that assemble on centromeric DNA and mediate the attachment of chromosomes to the microtubules of the mitotic spindle. The protein components of kinetochores are poorly understood, but the simplicity of the S. cerevisiae kinetochore makes it an attractive candidate for molecular dissection. Mutations in genes encoding CBF1 and CBF3, proteins that bind to yeast centromeres, interfere with chromosome segregation in vivo. To determine the roles played by these factors and by various regions of centromeric DNA in kinetochore function, we have developed a method to partially reassemble kinetochores on exogenous centromeric templates in vitro and to visualize the attachment of these reassembled kinetochore complexes to microtubules. In this assay, single reassembled complexes appear to mediate microtubule binding. We find that CBF3 is absolutely essential for this attachment but, contrary to previous reports (Hyman, A. A., K. Middleton, M. Centola, T.J. Mitchison, and J. Carbon. 1992. Microtubule-motor activity of a yeast centromere-binding protein complex. Nature (Lond.). 359:533-536) is not sufficient. Additional cellular factors interact with CBF3 to form active microtubule-binding complexes. This is mediated primarily by the CDEIII region of centromeric DNA but CDEII plays an essential modulatory role. Thus, the attachment of kinetochores to microtubules appears to involve a hierarchy of interactions by factors that assemble on a core complex consisting of DNA-bound CBF3.
Clathrin triskelions form polyhedral cages with hexagonal and pentagonal faces when dialyzed against suitable assembly buffers. However, when the buffer is made 12% saturated in ammonium sulfate and the dialysis is performed at 4 degrees C, clathrin polymerizes into cubes. The cube is constructed from eight triskelions with one at each corner. The edge length of the cube is approximately 45 nm, equivalent to the length of the leg of a triskelion. Thus, each edge of the cube is composed of two antiparallel legs overlapping over their whole length. The interactions between the legs in the cube are a subset of those postulated to occur in cages. Indeed, the cube can be derived from a pentagonal dodecahedron by removing 12 of the 20 triskelions with only slight adjustment of the legs of the remaining triskelions. The cube forms regular arrays and appears to be a favorable species for crystallization of clathrin.