In eukaryotic cells, the onset of mitosis involves cyclin molecules which interact with proteins of the cdc2 family to produce active kinases. In vertebrate cells, cyclin A dependent kinases become active in S- and pro-phases, whereas a cyclin B-dependent kinase is mostly active in metaphase. It has recently been shown that, when added to Xenopus egg extracts, bacterially produced A- and B-type cyclins associate predominantly with the same kinase catalytic subunit, namely p34cdc2, and induce its histone H1 kinase activity with different kinetics. Here, we show that in the same cell free system, both the addition of cyclin A and cyclin B changes microtubule behavior. However, the cyclin A-dependent kinase does not induce a dramatic shortening of centrosome-nucleated microtubules whereas the cyclin B-dependent kinase does, as previously reported. Analysis of the parameters of microtubule dynamics by fluorescence video microscopy shows that the dramatic shortening induced by the cyclin B-dependent kinase is correlated with a several fold increase in catastrophe frequency, an effect not observed with the cyclin A-dependent kinase. Using a simple mathematical model, we show how the length distributions of centrosome-nucleated microtubules relate to the four parameters that describe microtubule dynamics. These four parameters define a threshold between unlimited microtubule growth and the establishment of steady-state dynamics, which implies that well defined steady-state length distributions can be produced by regulating precisely the respective values of the dynamical parameters. Moreover, the dynamical model predicts that increasing catastrophe frequency is more efficient than decreasing the rescue frequency to reduce the average steady state length of microtubules. These theoretical results are quantitatively confirmed by the experimental data.
Control of microtubule dynamics by the antagonistic activities of XMAP215 and XKCM1 in Xenopus egg extracts