Microtubule minus-end aster organization is driven by processive HSET-tubulin clusters

Abstract

Higher-order structures of the microtubule (MT) cytoskeleton are comprised of two architectures: bundles and asters. Although both architectures are critical for cellular function, the molecular pathways that drive aster formation are poorly understood. Here, we study aster formation by human minus-end-directed kinesin-14 (HSET/KIFC1). We show that HSET is incapable of forming asters from preformed, nongrowing MTs, but rapidly forms MT asters in the presence of soluble (non-MT) tubulin. HSET binds soluble (non-MT) tubulin via its N-terminal tail domain to form heterogeneous HSET-tubulin clusters containing multiple motors. Cluster formation induces motor processivity and rescues the formation of asters from nongrowing MTs. We then show that excess soluble (non-MT) tubulin stimulates aster formation in HeLa cells overexpressing HSET during mitosis. We propose a model where HSET can toggle between MT bundle and aster formation in a manner governed by the availability of soluble (non-MT) tubulin.

Fig. 1

Full-length HSET organizes growing MTs into asters. a Schematic of HSET truncations purified in this study. HSET contains two MT-binding domains: an ATP-independent globular tail domain located at the N terminus (amino acid 1–138, brown), and an ATP-dependent conserved kinesin motor domain located at the C terminus (aa 305–673, blue). HSET also contains a coiled-coil stalk domain necessary for dimerization (aa 139–304, black). All constructs contained an N-terminal 6× His tag used for affinity purification. b Aster formation of growing MTs by HSET. 20 µM tubulin (10% Alexa594-labeled, magenta) was mixed in assay buffer with the indicated EGFP-HSET truncation (green) and monitored by time-lapse microscopy at 37 °C. With the exception of EGFP-HSETΔTail (20 nM), all HSET constructs were present at 100 nM. c Bundle formation of nongrowing, GMPCPP-stabilized MTs by HSET. Alexa594-labeled GMPCPP-MTs (10% labeled, 1 µM tubulin in polymer form, magenta) were mixed in assay buffer with the indicated EGFP-HSET truncation (green) and monitored by time-lapse microscopy at 37 °C. HSET concentrations are identical to b. For contrast measurements over time, see Supplementary Figure 1d, e. For movies, see Supplementary Movies 1–2. For additional EGFP-HSET images on GMPCPP-MTs, see Fig. 4a. Technical replicates of experiments in b, c were repeated n ≥ 3 times for each condition, and representative images are shown. Scale bars, 50 µm

Fig. 2

Soluble (non-MT) tubulin activates processive motility of HSET on single MTs. a Schematic. EGFP-HSET truncations were diluted in P12 buffer and monitored on GMPCPP-stabilized MTs by time-lapse TIRF. b Representative kymographs for time-lapse TIRF images for the indicated constructs at the following concentrations: EGFP-HSET and EGFP-HSETΔMotor, 50 pM. EGFP-HSETΔTail, 250 pM. Distance is on the x-axis (scale bar, 10 µm), and time is on the y-axis (scale bar, 10 s). c Mean-squared displacement (MSD) analysis of particle motion. The reported diffusion constant D is determined from a linear fit over the first 5 s, with the units nm²/s: EGFP-HSET: D = 6.3 × 10⁴, n = 206; EGFP-HSETΔMotor: D = 9.4 × 10⁴, n = 197; EGFP-HSETΔTail: D = 0.1 × 10⁴, n = 200. Data are presented as the calculated mean MSD (y-axis) from two independent experiments over the indicated time intervals (x-axis) for the indicated n particles ± SEM. d EGFP-HSET in BRB80 + 50 mM KCl was observed in the absence (left) or presence (right) of soluble tubulin and visualized by kymograph (x-scale bar, distance, 10 µm; y-scale bar, time, 1 min). e Quantification of processive (≥5 s) event frequency as a function of [EGFP-HSET] in the presence (dark green) or absence (light green) of 2 µM tubulin. Data are presented as the number of processive events divided by the total observed MT length multiplied by the observation time for two independent experiments ± SD calculated from N ≥ 10 movies for each condition. Boxes represent first and third quartiles, whiskers represent detection limits, and lines represent median (mean overlaid). f Unlabeled HSET was mixed with 10 nM Cy5-tubulin in BRB80 + 50 mM KCl and observed. Velocities and run lengths of moving Cy5-tubulin particles were determined by kymograph and plotted as histograms. Data are reported as the mean velocity and run length values of n particles from CDF fitting ± the 95% CI from bootstrapping from two independent experiments. g 100 nM Cy5-tubulin (magenta) and 1 nM EGFP-HSET (green) were observed near-simultaneously by high-speed TIRF in BRB80 + 50 mM KCl, and visualized by kymograph (x-scale bar, distance, 5 µm; y-scale bar, time, 10 s)

Fig. 3

Soluble (non-MT) tubulin binding to N-terminal HSET tail domain induces HSET-tubulin clustering to activate long-range unidirectional motion. a Co-immunopreciptation of purified HSET with soluble tubulin. The indicated EGFP-HSET truncation (50 nM concentration) was incubated with soluble tubulin (250 nM concentration) and the common EGFP-tag was used for immunoprecipitation. Inputs were loaded at 15% of total protein. A representative blot from N = 2 independent experiments is shown. b Fluorescence intensity analysis of EGFP-HSET diluted to single-molecule levels and adhered to a glass cover slip (light green) compared to the first frame of moving EGFP-HSET after the addition of tubulin (dark green, concentrations indicated). The fluorescence intensity of individual EGFP-HSET particles was determined by Gaussian fit, and intensity distributions were plotted as histograms for the population, where normalized count is the observed value in the bin divided by the total number of particles. Data are reported as the arithmetic mean ± SD for the indicated n particles from 2 independent experiments, where N ≥ 4 movies for each condition. c, d Fluorescence intensity analysis of Cy5-tubulin diluted to single-molecule levels and adhered to a glass cover slip (light purple) compared to the first frame of moving Cy5-tubulin (10 nM) in the presence of 10 nM unlabeled HSET (dark purple). The fluorescence intensity of individual Cy5-tubulin particles was determined by Gaussian fit, and intensity distributions were plotted as histograms for the population. Data are reported as the arithmetic mean ± SD for the indicated n particles from 2 independent experiments where N = 6 movies for moving Cy5-tubulin. For d, fluorescence intensities were determined in the presence of the indicated guanine nucleotide, where N ≥ 2 movies for each condition. All experiments were performed in BRB80 + 50 mM KCl with 0.5 mg/mL casein

Fig. 4

Soluble (non-MT) tubulin promotes the ability of HSET to drive aster self-organization of GMPCPP-MTs independent of MT polymerization. a EGFP-HSET-driven self-organization of GMPCPP-stabilized MTs with increasing tubulin concentration. Alexa594-labeled GMPCPP-MTs (10% labeled, 1 µM tubulin in polymer form, magenta) were mixed in assay buffer with EGFP-HSET (100 nM, green) and monitored by time-lapse microscopy at 37 °C. Unlabeled tubulin was added to the reaction at the indicated concentration. Technical replicates were repeated N ≥ 3 times for each condition, and representative images are shown. Scale bar, 50 µm. b EGFP-HSET-driven self-organization of GMPCPP-stabilized MTs in the absence of MT polymerization. Experiments were performed identically to a but in the absence of taxol and the presence of saturating colchicine and GDP to prevent polymerization. Technical replicates were repeated N ≥ 2 times for each condition, and representative images are shown. Scale bar, 50 µm

Fig. 5

Multiple HSET motors conjugated to quantum dots drive self-assembly of GMPCPP-MTs into asters. a EGFP-HSET or EGFP-HSETΔTail was conjugated to streptavidin-QDots via the N-terminal 6× His-tag and a biotin anti-His antibody at a 3:1 ratio and visualized via TIRF. Representative kymographs of EGFP-HSET-QDots (left, 1 nM EGFP-HSET: 0.33 nM QDot) and EGFP-HSETΔTail-QDots (right, 0.5 nM EGFP-HSETΔTail: 0.17 nM QDot) are shown (x-scale, distance, 5 µm; y-scale, time, 10 s). b–d Velocities (b), run lengths (c), and end dwell times (d) for the indicated constructed conjugated to QDots at a 3:1 ratio (EGFP-HSET, black, EGFP-HSETΔTail, red) were determined by kymograph analysis and plotted as histograms for the population. Data are reported as the mean values (insets) from CDF fitting ± the 95% CI from bootstrapping for the indicated n particles from 2 independent experiments, where N ≥ 4 movies for each condition. Populations for EGFP-HSET-QDots (black, upper) and EGFP-HSETΔTail-QDots (red, lower) are shown. For run length/end dwell times, particles reaching the end of MTs/dissociating immediately (<1 frame) are color-coded on the histograms. e Self-organization of GMPCPP-stabilized MTs by EGFP-HSET-QDots and EGFP-HSETΔTail-QDots. Alexa594-labeled GMPCPP-MTs (10% labeled, 1 µM tubulin in polymeric form, magenta) were mixed in assay buffer with the indicated motor-QDot complexes (21:7 nM motor :QDots, green) and monitored by time-lapse microscopy at 37 °C. The yellow box indicates the field of view depicted in f. Technical replicates were repeated n ≥ 3 times for each condition, and representative images are shown. Scale bar, 50 µm. f Zoomed-in view of the indicated field. The yellow arrow indicates EGFP-HSET-QDots that have accumulated on the minus end of an MT bundle. Time is indicated in min:s. Scale bar, 10 µm

Fig. 6

Increasing the relative level of soluble (non-MT) tubulin promotes HSET-driven aster formation in cells. a Generation of a transgenic HeLa cell line where EGFP-HSET expression is under control of doxycycline. Western blot of whole cell lysates probed with antibodies that recognize EGFP, tubulin, and HSET. Numbers indicate days of induction with doxycycline/DMSO. Green arrow indicates overexpressed EGFP-HSET, and black arrow represents endogenous HSET. b Maximum intensity z-projections of metaphase EGFP-HSET HeLa cells treated with DMSO (top) or doxycycline (bottom) for 3 d, then fixed and stained with antibodies against tubulin (yellow), centrin (magenta) or Hoechst (DNA, blue). Numbered boxes correspond to spectrally unmixed centrin-stained regions shown on right. Y/N indicates the presence of centrin at spindle poles. Scale bar, 5 µm. c EGFP-HSET HeLa cells were pretreated with DMSO (top) or doxycycline (bottom) for 3 d, then treated with 500 nM nocodazole (Noc) for 15 min to increase the relative levels of soluble tubulin. Cells were fixed, stained, and imaged identically to b and maximum intensity z-projections are shown. Scale bar, 5 µm. d Quantification of b, c. After analyzing the tubulin/centrin channels, the fraction of cells containing ≥1 acentrosomal aster was compared between 3 d doxycycline/DMSO cells after treatment with 500 nM nocodazole for 0, 15, and 30 min. Overlaid dots represent the average observed fraction for each independent experiment. ^***p < 0.001 by two-tailed t-test. Data were reported as the average ± SEM of N = 3 independent experiments with n ≥ 60 cells for each condition. e EGFP-HSET HeLa cells were treated with doxycycline for 3 days to induce maximal expression, and 500 nM nocodazole was added to increase relative levels of soluble tubulin. Maximum intensity z-projections were acquired in the EGFP channel every 3 min and a representative time-course is shown for the indicated times. Scale bar, 5 µm

Fig. 7

Proposed model for MT aster formation by HSET. a HSET’s N-terminal tail domain is able to bind either soluble (non-MT) tubulin or MT polymer (brown), whereas HSET’s C-terminal motor domain is only able to bind MT polymer (blue). The relative availability of soluble tubulin toggles HSET’s activity between the following states: b HSET is unbound to soluble (non-MT) tubulin and exists as a dimer. This motor is: (i) nonprocessive on single MTs, and (ii) provides MT–MT sliding forces within bundles of MTs similar to established models for Ncd. c HSET is bound to soluble (non-MT) tubulin and exists in heterogeneous multimotor HSET-tubulin clusters. The precise molecular and geometrical nature of these clusters remain to be determined. These motors: (i) move processively toward the minus ends of single MTs, and (ii) processively transport MTs or MT bundles along existing MT tracks to form MT asters. This ability to toggle between modes relies on the unique ability of the N-terminal tail domain to bind both soluble (non-MT) tubulin and MT polymer