Microtubule end conversion mediated by motors and diffusing proteins with no intrinsic microtubule end-binding activity

Abstract

Accurate chromosome segregation relies on microtubule end conversion, the ill-understood ability of kinetochores to transit from lateral microtubule attachment to durable association with dynamic microtubule plus-ends. The molecular requirements for this conversion and the underlying biophysical mechanisms are elusive. We reconstituted end conversion in vitro using two kinetochore components: the plus end-directed kinesin CENP-E and microtubule-binding Ndc80 complex, combined on the surface of a microbead. The primary role of CENP-E is to ensure close proximity between Ndc80 complexes and the microtubule plus-end, whereas Ndc80 complexes provide lasting microtubule association by diffusing on the microtubule wall near its tip. Together, these proteins mediate robust plus-end coupling during several rounds of microtubule dynamics, in the absence of any specialized tip-binding or regulatory proteins. Using a Brownian dynamics model, we show that end conversion is an emergent property of multimolecular ensembles of microtubule wall-binding proteins with finely tuned force-dependent motility characteristics.

Fig. 1

Microtubule (MT) wall-and-tip interactions with a molecular lawn of CENP-E motors and Ndc80 complexes. a Experiment with taxol-stabilized MT immobilized on a coverslip and a bead with randomly conjugated Ndc80 and CENP-E molecules. b Velocity of bead transport along the MT walls (means ± SEM; red curve plotted using left axis) and attachment at the MT end (blue, right axis) versus the ratio of Ndc80 and CENP-E concentrations used for bead coating. Beads were scored as pausing at the MT tip if they remained attached for at least 2 s. Source data are provided as a Source Data file. c MT gliding on a coverslip coated with protein mixtures. Top: schematic of coverslip conjugation using tag-binding antibodies, ensuring that MT-binding domains are not sterically inhibited. Bottom: gliding velocity on coverslips with CENP-E motor and either Ndc80-GFP or CENP-E Tail-GFP versus the protein coating density, as determined by green fluorescent protein (GFP) fluorescence. Points are means ± SEM for average velocities from N = 3 independent trials, examining total n MTs: for CENP-E n = 84; for Ndc80+CENP-E n = 522, for Ndc80+CENP-E Tail n = 305. Source data are provided as a Source Data file. d Example of a bead walking to the end of the dynamic MT at the velocity of CENP-E motor, continuing in the same direction at the velocity of MT polymerization, and then moving backward when the MT disassembles (MT tip tracking). Left: time-lapse images acquired with differential interference contrast. Right: kymograph of the entire trajectory

Fig. 2

Microtubule (MT) wall-to-end transition by CENP-E paired with Ndc80 complex. a Schematics of the MT wall-to-end transition assay and a representative imaging field with GMPCPP-stabilized MTs (red) and coverslip-immobilized beads coated with green fluorescent protein (GFP)-labeled Ndc80 protein (green). Bar, 3 µm. b Selected images showing motions of MTs on immobilized beads coated with the indicated proteins. Numbers are time (min) from the start of observation. Arrows show direction of MT gliding. Bar, 3 µm. c Quantifications for the wall-to-end transition assay using a mixture of CENP-E motors and different Ndc80 proteins. Columns are means ± SD for results from N independent trials, which are shown with gray dots. For CENP-E only, CENP-E paired with either Ndc80 Broccoli, Ndc80 Bonsai, Δ1–80 Ndc80 Bonsai, or K166D Ndc80 Bonsai, N = 4, 7, 2, 3, and 2, respectively. Total number of examined MTs in all trials is indicated below each column. Source data are provided as a Source Data file. A successful MT end-retention event was counted if the trailing MT end was coupled to the bead for longer than 4 s. All Ndc80 proteins showed high percent of end-retention events but their durations were dramatically different. d Kaplan–Meier survival plot for MT end-retention time based on N independent trials examining end-attachment for n MTs: CENP-E motor: N = 4, n = 23; Ndc80+CENP-E, N = 7, n = 111

Fig. 3

Mathematical model of the molecular ensemble of motors and diffusing microtubule-associated proteins (MAPs). a Multiple MAPs (red) and motors (blue) are randomly distributed on the surface, forming a molecular lawn. Stabilized microtubule (MT) moves under force from kinesin motors in the presence of thermal noise. b, c Summary of kinetic transitions. Molecules bind stochastically to the 4 nm sites on the MT wall, and their unbinding is increased by force. Stepping of the motors and diffusional steps of the MAPs are also force dependent. The motor dissociates from the MT end and the MT wall at the same rate. The MAP molecule can dissociate from the MT end fully or continue to diffuse on the MT wall

Fig. 4

Centromere-associated protein E (CENP-E) and Ndc80 coupling to the dynamic microtubule (MT) ends. a Schematics of the dynamic MT end-conversion assay. Fluorescently labeled GMPCPP-stabilized MT seeds glide on beads, then unlabeled soluble tubulin is added to examine its incorporation at the bead-bound MT plus-end. b Selected time-lapse images recorded with Ndc80+CENP-E beads after addition of unlabeled soluble tubulin (6.3 µM). Numbers are time (min) from the start of observation. Arrows show the direction of motion of the bright MT fragment, reporting on the dynamics of the bead-bound MT plus-end. Bar, 3 µm. c Distance from the distal tip of the fluorescent MT fragment to the bead vs. time, showing repeated cycles. d Dynamics parameters for freely growing MTs (N = 4 independent trials) and for MT ends coupled to protein-coated beads (N = 4 for Ndc80+CENP-E; N = 6 for Ndc80 alone), showing means ± SEM for average results from these trials, source data are provided as a Source Data file. Data for isolated mammalian chromosomes (chromosome-coupled end) are from ref. . Statistical differences were evaluated by Kruskal–Wallis analysis of variance (ANOVA); *p < 0.05. e Images as in b but recorded using beads coated with Ndc80 only. f Percent of bead-coupled MT ends that disassembled, and then initiated a new round of MT polymerization without losing their bead attachment. Source data are provided as a Source Data file. g Plot similar to (c) but for a bead coated with Ndc80 in the absence of CENP-E motor. The Ndc80-coated beads can maintain coupling only for one dynamic MT cycle, detaching after MT depolymerization

Fig. 5

Microtubule (MT) wall-to-end transition in molecular systems combining CENP-E with various MT-associated proteins (MAPs). a Selected time-lapse images of stabilized MTs moving over beads coated with CENP-E motor and the indicated MAP. All proteins were conjugated to beads via anti-GFP antibodies to achieve similar brightnesses, ensuring that any differences in MT interactions are not due to differences in the density of the protein coatings. Numbers are time (min). Arrows show the direction of MT gliding. Bar, 3 µm. Arrowhead in the last EB1 panel points to a loss of tip attachment due to the MT end-to-wall transition. b Quantifications as in Fig. 2c, but for beads coated with mixtures containing the CENP-E motor and the indicated MAP. Data are means ± SEM for results from N independent trials, which are shown with gray dots. For CENP-E only, CENP-E paired with either Ndc80 Broccoli, Ska1 complex, CENP-E Tail, EB1, or CLASP2 N = 4, 7, 3, 4, 3, and 4, respectively. Total number of observed events is indicated below each column. Asterisk above a bar (p < 0.05) indicates a significant difference relative to the analogous measurement for Ndc80 beads, as determined by Kruskal–Wallis analysis of variance (ANOVA). Source data are provided as a Source Data file. c Kaplan–Meier plot for MT end-retention time based on the same data sets as in (b). d Kaplan–Meier plot for the predicted MT end-retention time for different MAPs paired with CENP-E motor (n = 32 simulations for each condition). Different MAPs were modeled using the diffusion coefficients and residence times measured for single molecules in vitro (Supplementary Table 2)

Fig. 6

Behavior of dynamic microtubules (MTs) during end-conversion in vitro. a Schematics depicting different features of experimental kymographs. Fluorescent MT segments move away or toward the bead owing to the addition or loss of unlabeled tubulin from the bead-coupled MT plus-end, motor-dependent transport (gliding), and MT diffusion on bead surface. Vertical black line corresponds to a coverslip-immobilized bead (denoted by blue triangle above), which is often visible in the MT channel owing to the bead-attached motionless MTs. When such MTs were lacking, bead position was determined from green fluorescent protein (GFP) channel. The oblique black lines correspond to motions of the brightly labeled GMPCPP-stabilized MT seeds. Color bars on the right provide visual guides for interpretations of these motions. Arrowhead represents MT detachment; arrow represents end-to-wall transition event leading to MT diffusion. b Example kymographs for dynamic MT ends coupled to beads coated with indicated proteins together with CENP-E

Fig. 7

Frequent loss of microtubule (MT) end-coupling by EB1 and CENP-E. Five typical kymographs of dynamic MTs coupled to beads coated with a mixture of EB1 and CENP-E proteins. Shown are enlargements of experimental kymographs illustrating highly complex behavior of EB1-dependent MT end-coupling, which is lost and reestablished frequently. MT elongation continued during MT diffusion on the bead, leading to “gaps” in typical MT polymerization kymograph pattern. Colored bars provide interpretations for the kymographs. For other details, see legend to Fig. 6

Fig. 8

Kinesin-1 paired with Ndc80 in the microtubule (MT) wall-to-end transition assay. a Selected time-lapse images of a stabilized MT moving over a bead coated with Kinesin-1 and Ndc80 proteins in experiment carried out as in Fig. 2a. Numbers are time (s). Arrows show direction of MT gliding. Bar, 3 µm. b MT behavior as a function of the ratio of bead-bound proteins. Points are means ± SEM; curves are exponents (left) and hyperbolic functions illustrating the trends. Source data for (b, c) are provided as a Source Data file. c Average results for beads coated with Ndc80 and motors at ratios from 0.2 to 0.5. Bars show means ± SD for results from N independent trials, which are shown with gray dots. For CENP-E only, Ndc80 paired either with CENP-E, Kinesin-1, or Kinesin-1 at low adenosine triphosphate (ATP), N = 4, 3, 5, and 2, respectively. Numbers under each bar indicate the total number of observed events. Experiments were carried out in motility buffer supplemented with 2 mM ATP, except at low ATP concentration which was 20 µM

Fig. 9

Impact of motor’s force-sensitivity on microtubule (MT) end-retention. a Force-unbinding (top) and force-velocity (bottom) characteristics used in model simulations, see Supplementary Note 1 for details. b Kaplan–Meier plots for the MT end-retention time. Experimental results are as in Fig. 2d but supplemented with measurements for Ndc80+Kinesin-1 at 2 mM adenosine triphosphate (ATP) (N = 7, n = 50) and 20 µM ATP (N = 2, n = 25). Theoretical plot is based on n = 32 simulations for each condition modeled using characteristics in a. c MT end-retention time vs. preceding gliding velocity for individual MTs in experiments with stabilized MTs. Data are based on N = 5, n = 76 for Ndc80+CENP-E; N = 7, n = 50 for Ndc80+Kinesin-1 (2 mM ATP); N = 2, n = 24 for Ndc80+Kinesin-1 in low ATP (20 µM ATP). d Kaplan–Meier plots calculated for molecular patches containing Ndc80 molecules and motor molecules with different force-dependent characteristics. Predictions for Ndc80+CENP-E and Ndc80+Kinesin-1 are the same as in b. Prediction for Kinesin* was calculated using force-dependent unbinding rate of CENP-E and the force-velocity function of Kinesin-1. Prediction for Kinesin** was calculated using the force-dependent unbinding rate of Kinesin-1 and the force-velocity function of CENP-E. e Average duration of MT end-retention vs. average gliding velocity in MT wall-to-end transition assay using CENP-E motor and indicated MT-associated proteins (MAPs). Duration of MT end-attachment is represented by the half-life of an exponential fit to the corresponding survival probability curve in Fig. 5c. Horizontal error bars are same as for MT end-retention time in Fig. 5b. Black line is the linear fit to all points. Pearson's correlation analysis gives R² = 0.91 with 95% confidence, and hence the anti-correlation with gliding velocity is significant

Fig. 10

Key phases of the CENP-E-dependent microtubule (MT) end-conversion in cells. Schematics show multimolecular ensemble of CENP-E kinesins and Ndc80 complexes, forming a molecular lawn that interacts with the MT. a Ndc80 slows down CENP-E kinesin during the plus-end–directed transport. Ndc80 plays an essential role in providing durable and mobile attachment to the end-proximal MT wall. In our in vitro experiments and in silico, these molecular interactions are concentrated at one side of the MT, which forms oblique contact with the molecular lawn. This configuration is also likely to occur transiently at the kinetochores of mitotic cells, as shown in b. However, forces acting on the chromosomes and kinetochore-bound MTs reorient the kinetochore, promoting the classical end-on configuration (c). We propose that in this configuration, kinetochore attachment is mediated by essentially the same molecular interactions with the MT wall, as described in this work