Kinetochore-generated pushing forces separate centrosomes during bipolar spindle assembly

Abstract

In animal somatic cells, bipolar spindle formation requires separation of the centrosome-based spindle poles. Centrosome separation relies on multiple pathways, including cortical forces and antiparallel microtubule (MT) sliding, which are two activities controlled by the protein kinase aurora A. We previously found that depletion of the human kinetochore protein Mcm21R(CENP-O) results in monopolar spindles, raising the question as to whether kinetochores contribute to centrosome separation. In this study, we demonstrate that kinetochores promote centrosome separation after nuclear envelope breakdown by exerting a pushing force on the kinetochore fibers (k-fibers), which are bundles of MTs that connect kinetochores to centrosomes. This force is based on poleward MT flux, which incorporates new tubulin subunits at the plus ends of k-fibers and requires stable k-fibers to drive centrosomes apart. This kinetochore-dependent force becomes essential for centrosome separation if aurora A is inhibited. We conclude that two mechanisms control centrosome separation during prometaphase: an aurora A-dependent pathway and a kinetochore-dependent pathway that relies on k-fiber-generated pushing forces.

Figure 1.

Mcm21R is required for efficient bipolar spindle formation in the prometaphase pathway. (A) Percentage of cells undergoing monopolar anaphase when centrosomes are either together or apart at the time of NEBD. (B) Successive frames every 3 min from a live cell video of HeLa cells expressing H2B-EGFP/a-tubulin–mRFP (to mark chromosomes and MTs). Images for α-tubulin–mRFP and the composite images for H2B-EGFP (green) and α-tubulin–mRFP (red) of two adjacent cells are shown. The top cell follows the prophase (P) pathway, and the bottom cell follows the prometaphase (PM) pathway. (C) Percentage of mitotic HeLa cells using either the prophase or prometaphase pathways in siControl (n = 73), siMcm21R (n = 73), siRootletin (n = 31), or siMcm21R + siRootletin (n = 42)–treated cells. (D) Mean bipolar spindle assembly time in the prophase or prometaphase pathway in untreated (n = 69) or treated with siControl (n = 73) or siMcm21R (n = 73) RNAs. (E and F) Successive frames every 3 min from a live cell video of siMcm21R RNA–treated cells in the prophase (E) or prometaphase (F) pathways. (G) Percentage of monopolar spindles in fixed cells after treatment with siControl (n = 90), siMcm21R (n = 90), siRootletin (n = 90), or siMcm21R + siRootletin RNAs (n = 90). Asterisks represent statistically significant differences (P < 0.01 by Mann-Whitney test). Error bars represent SEM. Bars, 10 µm.

Figure 2.

Depletion of Mcm21R reduces the number of stable k-fibers. (A) Successive frames every 30 or 40 s after photoactivation of stable PA-GFP–α-tubulin/hTERT-RPE1 cells treated either with siControl, siMcm21R, or siMCAK + siKif2A RNAs. Green arrowheads mark the initial position of a photoactivated mark, and red arrowheads show the final position of the mark. (B) Quantification of poleward MT flux rate in cells treated with siControl (n = 9), siMcm21R (n = 9), or siMCAK + siKif2A (n = 5) RNAs. (C–E) Quantification of stable k-fibers in cold-treated cells (for methodology, see Fig. S2, available at http://www.jcb.org/cgi/content/full/jcb.200809055/DC1). Bar graphs indicate the percentage of kinetochores lacking attached MTs (−), having a weak attachment (+), a medium attachment (++), or a strong attachment (+++) in siControl (n = 354)-, siNuf2R (n = 204)-, and siMcm21R (n = 273)-treated cells with bipolar spindles (C), in siMcm21R (n = 215)- and siEg5 (n = 212)-treated cells with monopolar spindles (D), and in siControl (n = 354)-, siMcm21R (n = 273)-, siMCAK (n = 238)-, and siMcm21R + siMCAK (n = 219)–treated cells with bipolar spindles (E). (F) Representative images from C and E of cold-treated cells stained with α-tubulin antibodies (MTs; green) and CREST antisera (kinetochores; red). Error bars represent SEM. Bars, 10 µm.

Figure 3.

Kinetochores and poleward MT flux are required for timely centrosome separation. (A and B) Mean bipolar spindle formation time in the prometaphase pathway in siControl (n = 38)-, siMcm21R (n = 36)-, siMCAK (n = 23)-, or siMcm21R + siMCAK (n = 31)–treated cells (A) or in siControl (n = 38)-, siNuf2R (n = 40)-, siCENP-E (n = 28)–, siMCAK + siKif2A (n = 38)–, or siCLASP1 + siCLASP2 (n = 28)–treated cells (B). (C) Representative images of siControl- or siMcm21R RNA–treated cells stained with CREST antisera (to mark kinetochores) and EB1 antisera (to mark growing MT plus ends). Asterisks indicate statistically significant differences (P < 0.01 by Mann-Whitney test). Error bars represent SEM. Bar, 10 µm.

Figure 4.

Kinetochore and aurora A activities cooperate to drive centrosome separation. (A) Percentage of cells that fail to establish a bipolar spindle within 15 min after NEBD using either the prophase or prometaphase pathway. Cells were treated with siControl (n = 73), siMcm21R (n = 73), siaurora A + siControl (n = 40), siaurora A + siMcm21R (n = 38), siCENP-E (n = 42), siaurora A + siCENP-E (n = 32), siNuf2R (n = 99), siaurora A + siNuf2R (n = 20), siMCAK + siKif2A (n = 79), siaurora A + siKif2A + siMCAK (n = 52), siControl + ZM1 (n = 33), or siMcm21R + ZM1 (n = 46). (B) Composite images of HeLa H2B-EGFP (green) and α-tubulin–mRFP (red) cells treated with siaurora A + siCENP-E or siaurora A + siKif2A + siMCAK in the prophase or prometaphase pathways. (C) Proposed model for the function of kinetochores in bipolar spindle formation during prometaphase. Just after NEBD, centrosomes are positioned on the same side of the chromatin (blue). Sister kinetochores are in close proximity, face the centrosomes, and are connected to centrosomes via k-fibers, thereby establishing a triangular geometry. A pushing force from kinetochores (F1) based on plus end MT polymerization (poleward MT flux; white arrows) drives centrosomes apart from each other. F1 is combined with additional forces (F2), which include astral MT–cortex interactions (mediated by actin-myosin) and extensile MT sliding (mediated by Eg5), which are two processes under the control of the aurora A kinase activity. The resultant force (F3) has an altered magnitude and directionality, which allows more efficient centrosome separation. Note that the centrosome–kinetochore geometry, and therefore the magnitude and direction of forces, will evolve dynamically as the spindle forms into a structure under force balance. Error bars represent SEM. Bar, 10 µm.