The Ndc80 kinetochore complex directly modulates microtubule dynamics

Abstract

The conserved Ndc80 complex is an essential microtubule-binding component of the kinetochore. Recent findings suggest that the Ndc80 complex influences microtubule dynamics at kinetochores in vivo. However, it was unclear if the Ndc80 complex mediates these effects directly, or by affecting other factors localized at the kinetochore. Using a reconstituted system in vitro, we show that the human Ndc80 complex directly stabilizes the tips of disassembling microtubules and promotes rescue (the transition from microtubule shortening to growth). In vivo, an N-terminal domain in the Ndc80 complex is phosphorylated by the Aurora B kinase. Mutations that mimic phosphorylation of the Ndc80 complex prevent stable kinetochore-microtubule attachment, and mutations that block phosphorylation damp kinetochore oscillations. We find that the Ndc80 complex with Aurora B phosphomimetic mutations is defective at promoting microtubule rescue, even when robustly coupled to disassembling microtubule tips. This impaired ability to affect dynamics is not simply because of weakened microtubule binding, as an N-terminally truncated complex with similar binding affinity is able to promote rescue. Taken together, these results suggest that in addition to regulating attachment stability, Aurora B controls microtubule dynamics through phosphorylation of the Ndc80 complex.

Fig. 1.

The human Ndc80 complex binds to and diffuses along the microtubule lattice. (A) Negative-stain electron micrograph of the Ndc80 complex on a taxol-stabilized microtubule. (B) A representative kymograph showing the binding and diffusion of Ndc80 complex (5 pM complex in solution) on taxol-stabilized microtubules. Position along the microtubule is depicted on the vertical axis over time on the horizontal axis. (C) Residence time distributions of GFP-tagged Ndc80 complex on microtubules fit with a single exponential (dashed line) to calculate the off-rate constant, k_off. (D) Mean-squared displacement (MSD) ± SEM vs. time lag. A linear fit to the data (dashed line) was used to determine the diffusion constant, D. (C and D) n = 584. (E) Representative image from the bulk microtubule binding assay with GFP-tagged Ndc80 complex (2 nM) on taxol-stabilized Alexa-568-labeled microtubules (2.5 nM tubulin dimer). Panels show microtubules (a), Ndc80 complex (b), and merge (c). Panel dimensions are 66 by 66 μm. (F) Plot of binding density (v) versus free Ndc80 complex concentration (L). A fit to the Hill model (dashed line) was used to determine the apparent affinity (K_d), Hill coefficient (n_H), and lattice occupancy (i, the number of Ndc80 complexes bound per tubulin dimer). (G) Scatchard plot of the same data shown in F, fit to the McGhee and von Hippel model (dashed line) to calculate the K_d, cooperativity parameter (w), and i. For F and G, n = 8–10 replicates per data point, markers are mean ± SEM, and errors on model fit parameters (K_d, n_H, w, and i) represent SD.

Fig. 2.

The Ndc80 complex slows microtubule disassembly and stabilizes protofilament extensions. Kymographs of disassembling microtubules (red) in the presence of (A) 100 pM or (B) 500 pM GFP-tagged Ndc80 complex (green). Brightness and contrast were adjusted equally in A and B. (C) Mean disassembly speeds ± SEM for microtubules in the presence of increasing concentrations of Ndc80 complex (without Ndc80 complex, n = 80; 100 pM Ndc80 complex, n = 31; 250 pM, n = 29; 500 pM, n = 34). (D) Time-lapse images of a disassembling microtubule (red) in the presence of 500 pM GFP-tagged Ndc80 complex (green) as a curled extension formed at the tip. Inset numbers show elapsed time, in seconds. See Fig. S3B for a gallery of images showing curled extensions. (E) Negative-stain electron micrograph of a disassembling microtubule tip (see SI Materials and Methods) stabilized by the Ndc80 complex. An arrow marks the transition from a closed microtubule to an open sheet. The figure was constructed from three images, the boundaries of which are depicted by dotted white lines.

Fig. 3.

Phosphomimetic mutations in the Ndc80 complex inhibit its ability to promote microtubule rescue. (A) Example traces of position vs. time for beads decorated with Ndc80 complex as they tracked microtubule disassembly against ∼2 pN of applied force. Time t = 0 s (dashed vertical line) marks the onset of tracking, when the disassembling microtubule tip began to drive movement of the bead against the force of the trap. Disassembly-driven movement ended when the bead detached (open circles) or when the microtubule rescued (arrows). Traces are offset vertically for visual clarity. (B) The fraction of beads coated with wild-type or mutant Ndc80 complex capable of tracking against ∼2 pN. From the disassembly-tracking events in B, (C) mean microtubule disassembly speeds ± SEM and (D) rescue rates were measured. Without load and in the absence of bead-bound Ndc80 complex, the disassembly rate was 230 ± 14 nm/s (dashed line in C, n = 26) and the rescue rate was 2 ± 1 h⁻¹ (dashed line in D, n = 3 events in 104 min of disassembly). (E) Rescue rate is plotted against the fraction of beads that tracked disassembly against force. (F) Percentage of microtubules for which a curl (Fig. 2D and Fig. S3B) was observed at either tip during disassembly in the TIRF microscopy assay. The n for each data point in B–D is listed in Table S1. Asterisks indicate that no rescues were observed. Unless otherwise noted, all error bars represent uncertainties from counting statistics.

Fig. 4.

The 9D and ΔN Ndc80 complexes exhibit similar binding behavior on microtubules. (A) Histograms of the residence time for single molecules (5 pM complex in solution) of wild-type (black trace, n = 131), 9D (red trace, n = 497), and ΔN (blue trace, n = 705) Ndc80 complex on taxol-stabilized microtubules. Each histogram was fit by a single exponential (dashed lines) to determine the off-rate constant, k_off. (B) Plots of MSD vs. time lag for binding events in A. The diffusion constant, D, was measured from linear fits to the data (dashed lines). (C and D) Bulk binding assays of 9D (red traces, n = 4–7 replicates per data point) and ΔN (blue traces, n = 6–7 replicates per data point) Ndc80 complex on taxol-stabilized microtubules. Dashed lines show fits of binding data to (C) Hill and (D) McGhee and von Hippel models. Errors on model fit parameters (K_d, n_H, w, and i) represent SD. All markers represent mean ± SEM and all assays were performed in BRB40 buffer (see SI Materials and Methods).