Physiological and ultrastructural analysis of elongating mitotic spindles reactivated in vitro

We have developed a simple procedure for isolating mitotic spindles from the diatom Stephanopyxis turris and have shown that they undergo anaphase spindle elongation in vitro upon addition of ATP. The isolated central spindle is a barrel-shaped structure with a prominent zone of microtubule overlap. After ATP addition greater than 75% of the spindle population undergoes distinct structural rearrangements: the spindles on average are longer and the two half-spindles are separated by a distinct gap traversed by only a small number of microtubules, the phase-dense material in the overlap zone is gone, and the peripheral microtubule arrays have depolymerized. At the ultrastructural level, we examined serial cross-sections of spindles after 1-, 5-, and 10-min incubations in reactivation medium. Microtubule depolymerization distal to the poles is confirmed by the increased number of incomplete, i.e., c-microtubule profiles specifically located in the region of overlap. After 10 min we see areas of reduced microtubule number which correspond to the gaps seen in the light microscope and an overall reduction in the number of half-spindle microtubules to about one-third the original number. The changes in spindle structure are highly specific for ATP, are dose-dependent, and do not occur with nonhydrolyzable nucleotide analogues. Spindle elongation and gap formation are blocked by 10 microM vanadate, equimolar mixtures of ATP and AMPPNP, and by sulfhydryl reagents. This process is not affected by nocodazole, erythro-9-[3-(2-hydroxynonyl)]adenine, cytochalasin D, and phalloidin. In the presence of taxol, the extent of spindle elongation is increased; however, distinct gaps still form between the two half-spindles. These results show that the response of isolated spindles to ATP is a complex process consisting of several discrete steps including initiation events, spindle elongation mechanochemistry, controlled central spindle microtubule plus-end depolymerization, and loss of peripheral microtubules. They also show that the microtubule overlap zone is an important site of ATP action and suggest that spindle elongation in vitro is best explained by a mechanism of microtubule-microtubule sliding. Spindle elongation in vitro cannot be accounted for by cytoplasmic forces pulling on the poles or by microtubule polymerization.