Centrosome nucleates numerous ephemeral microtubules and only few of them participate in the radial array.

It is generally accepted that long microtubules (MTs) grow from the centrosome with their minus ends anchored there and plus ends directed towards cell membrane. However, recent findings show this scheme to be an oversimplification. To further analyze the relationship between the centrosome and the MT array we undertook a detailed study on the MTs growing from the centrosome after microinjection of Cy3 labeled tubulin and transfection of cells with EB1-GFP. To evaluate MTs around the centrosome two approaches were used: path photobleaching across the centrosome area (Komarova et al., ) and sequential image subtraction analysis (Vorobjev et al., ). We show that about 50% of MTs had been nucleated at the centrosome are short-living: their mean length was 1.8 ± 0.8 µm and their life span - 7 ± 2 s. MTs initiated from the centrosome also rarely reach cell margin, since their elongation was limited and growth after shortening (rescue) was rare. After initial growth all MTs associated with the centrosome converted to pause or shortening. After pause MTs associated with the centrosome mainly depolymerized via the plus end shortening. Stability of the minus ends of cytoplasmic MTs was the same as for centrosomal ones. We conclude that in fibroblasts (1) the default behavior of free MTs in the cell interior is biased dynamic instability (i.e., random walk of the plus ends with significant positive drift); (2) MTs born at the centrosome show "dynamic instability" type behavior with no boundary; and (3) that the extended radial array is formed predominantly by MTs not associated with the centrosome.