We have characterized a human homologue of anillin, a Drosophila actin binding protein. Like Drosophila anillin, the human protein localizes to the nucleus during interphase, the cortex following nuclear envelope breakdown, and the cleavage furrow during cytokinesis. Anillin also localizes to ectopic cleavage furrows generated between two spindles in fused PtK1 cells. Microinjection of antianillin antibodies slows cleavage, leading to furrow regression and the generation of multinucleate cells. GFP fusions that contain the COOH-terminal 197 amino acids of anillin, which includes a pleckstrin homology (PH) domain, form ectopic cortical foci during interphase. The septin Hcdc10 localizes to these ectopic foci, whereas myosin II and actin do not, suggesting that anillin interacts with the septins at the cortex. Robust cleavage furrow localization requires both this COOH-terminal domain and additional NH2-terminal sequences corresponding to an actin binding domain defined by in vitro cosedimentation assays. Endogenous anillin and Hcdc10 colocalize to punctate foci associated with actin cables throughout mitosis and the accumulation of both proteins at the cell equator requires filamentous actin. These results indicate that anillin is a conserved cleavage furrow component important for cytokinesis. Interactions with at least two other furrow proteins, actin and the septins, likely contribute to anillin function.

CENP-E is a kinesin-like protein that when depleted from mammalian kinetochores leads to mitotic arrest with a mixture of aligned and unaligned chromosomes. In the present study, we used immunofluorescence, video, and electron microscopy to demonstrate that depletion of CENP-E from kinetochores via antibody microinjection reduces kinetochore microtubule binding by 23% at aligned chromosomes, and severely reduces microtubule binding at unaligned chromosomes. Disruption of CENP-E function also reduces tension across the centromere, increases the incidence of spindle pole fragmentation, and results in monooriented chromosomes approaching abnormally close to the spindle pole. Nevertheless, chromosomes show typical patterns of congression, fast poleward motion, and oscillatory motions. Furthermore, kinetochores of aligned and unaligned chromosomes exhibit normal patterns of checkpoint protein localization. These data are explained by a model in which redundant mechanisms enable kinetochore microtubule binding and checkpoint monitoring in the absence of CENP-E at kinetochores, but where reduced microtubule-binding efficiency, exacerbated by poor positioning at the spindle poles, results in chronically monooriented chromosomes and mitotic arrest. Chromosome position within the spindle appears to be a critical determinant of CENP-E function at kinetochores.