GFP-pericentrin particles move toward and assemble onto centrosomes. (A) Time-lapse images of a GFP-pericentrin particle (arrowhead) in an interphase COS-7 cell moving toward a GFP-pericentrin–labeled centrosome (upper right). Time elapsed between images is shown in seconds. (B) Time-lapse images of a GFP-pericentrin particle (arrowhead) moving toward and joining a section of a GFP-pericentrin–labeled centrosome (bottom). Note that magnification is threefold higher than in A. (C) Schematic representation of several GFP-pericentrin particle movements (arrows) and their relationship to the centrosome (c). Time elapsed between points is 1 s. No movement was detected in the presence of nocodazole (N represents a summary of movements recorded in the presence of nocodazole; n = 10). The asterisk shows tracing of a particle imaged in A. (D and E) Immunofluorescence localization of endogenous γ tubulin (E) and a GFP-pericentrin particle (D, left) near the centrosome (C). (F–H) Immunofluorescence localization of endogenous pericentrin (F) and cytoplasmic dynein (G) in pericentriolar satellites surrounding prophase centrosomes (c). (H) Colocalized voxels (see Dictenberg et al., 1998). Note that nearly all pericentrin-staining satellites (F, left side) colocalize with dynein (G). (A–E) COS cells; (F–H) CHO cells. Bars, 1 μm (bar in E for D and E, bar in H for F–H).

Assembly of pericentrin and γ tubulin at the centrosome requires cytoplasmic dynein and dynactin. (A) When dynamitin overexpression is induced by microinjection of cDNA into G1 cell nuclei (see MATERIALS AND METHODS), centrosome levels of pericentrin and γ tubulin are reduced (black bars) compared with untreated control cells (stippled bars). Shown are the mean ± SE of a representative experiment. p values reflect differences between dynamitin-expressing cells (n = 20) and control cells (β galactosidase cDNA injected, n = 10; uninjected, n = 177) from seven independent experiments. Values obtained from β galactosidase cDNA-injected cells did not deviate >5.3% from uninjected cells and were not statistically different from uninjected cells (p < 0.87). Dynamitin and β galactosidase expression were confirmed by immunofluorescence staining and were roughly similar (our unpublished results). (B) When G1 cells are microinjected with ascites fluid containing 70.1 antibody, pericentrin and γ tubulin levels are reduced up to 50-fold (black bars) compared with control uninjected cells (stippled bars). Shown are the mean ± SE of a representative experiment. p values reflect statistical differences between 70.1 injected (n = 14) and control cells (control mouse IgG injected, n = 10; uninjected, n = 90) from six independent experiments. Values obtained from cells injected with control mouse IgG were not statistically different from uninjected cells (p ∼ 0.95) and did not vary >6.5% from uninjected controls. (A and B) COS cells. For A and B, G1 levels of centrosome fluorescence were subtracted. (C) Centrosome assembly of γ tubulin in Xenopus extracts in the absence of nocodazole and in the presence of anti-dynein antibody or control antibody (β-galactosidase). Values shown are ×10⁻⁵. Background centriole levels of γ tubulin (<6.7% of experimental values) were subtracted.

Centrosome assembly of pericentrin and γ tubulin but not centrioles requires microtubules. For the experiments described in A–F, cells were treated as indicated at G1, and assays were performed in the following metaphase (see MATERIALS AND METHODS). (A) Immunofluorescence images showing pericentrin staining at centrosomes in cells incubated for one cell cycle in the presence (+) or absence (−) of nocodazole (noc.) to depolymerize microtubules. Depolymerization of microtubules was confirmed by immunofluorescence staining with anti-α tubulin antibodies. (B) Quantity of centrosome-associated pericentrin and γ tubulin assembled in the absence of microtubules as described in A. An intermediate level of pericentrin is observed when centrosomes are allowed to assemble for half of the cell cycle then exposed to nocodazole for the remaining half (middle bar in first graph). Shown are the mean ± SE of representative experiments; p values (nonparametric Wilcoxon rank sum test) reflect statistical differences between nocodazole-treated (pericentrin, n = 134; γ tubulin, n = 146) and DMSO-treated controls (pericentrin, n = 61; γ tubulin, n = 71) determined from four independent experiments. (C) Centrosome-associated levels of pericentrin in the presence (+) or absence (−) of cytochalasin B. Values represent mean ± SE for cytochalasin-treated (n = 38) and DMSO-treated control cells (n = 19) from two independent experiments. For A–C, G1 levels of centrosome fluorescence were subtracted (∼15% of total; see MATERIALS AND METHODS). (D) Confocal microscope image of α tubulin immunostaining structures representing centrioles in cells treated with nocodazole for one cell cycle. (E) Transmission electron microscopic image of a single cell treated as in D showing four centriole profiles. More than three centrioles were detected in serial sections of seven cells. (F) Quantification of α tubulin immunofluorescence in cells treated as in D (black bar; n = 18) compared with control mitotic cells (stippled bar; n = 10). Values are mean ± SE from three independent experiments. Control cells represented by the stippled bar were briefly treated with nocodazole (10 μg/ml, 60 min, 37°C) to depolymerize cytoplasmic microtubules and thus reveal centriolar α tubulin fluorescence. The values are not significantly different, further demonstrating that the centriole pair duplicated from G1 to M in the absence of cytoplasmic microtubules as in control cells. (A–F) CHO cells. (G) Microtubule-enhanced assembly of γ tubulin in Xenopus extracts. Extracts and sperm nuclei were incubated with or without nocodazole (noc.) for the times indicated (see MATERIALS AND METHODS for details). Each time point represents mean ± SE of at least 13 measurements from one experiment; p values (multiple regression with interaction term) were determined by averaging all values from three independent experiments. Values shown are ×10⁻⁵. Background centriole levels of γ tubulin (<12.2% of experimental values) were subtracted.

Pericentrin associates with microtubules in a dynein- and dynactin-dependent manner. Immunofluorescence staining is shown for microtubules (A) and pericentrin (B) in CHO cells treated with nocodazole for one cell cycle and then released from the drug for 20 min to promote microtubule polymerization. Nearly all noncentrosomal pericentrin colocalizes with microtubules. Bar in B, 1 μm (for A and B). (C) Taxol was used to assemble microtubules in CHO cell lysates (see MATERIALS AND METHODS). Other reagents were added as indicated, microtubules were centrifuged through a 20% sucrose cushion, and proteins in the microtubule pellet were analyzed by Western blot for pericentrin, dynein intermediate chain (IC), or α tubulin (to control for microtubule loading). Unless otherwise indicated, all lysates contained GTP, exogenous tubulin, taxol, and AMP-PNP. In addition, lysates contained nocodazole without taxol (lane 1), no exogenous tubulin (lane 2), the absence (lane 3) or presence (lane 4) of AMP-PNP, control antibody (β galactosidase, lane 5), or 70.1 antibody (lane 6). Lanes 7 and 8, lysates from cells expressing control protein (GFP) or dynamitin, respectively. Similar results were obtained with endogenous or exogenous tubulin (compare lanes 2 and 4).