GSK3-mediated CLASP2 phosphorylation modulates kinetochore dynamics

Abstract

Error-free chromosome segregation requires dynamic control of microtubule attachment to kinetochores, but how kinetochore-microtubule interactions are spatially and temporally controlled during mitosis remains incompletely understood. In addition to the NDC80 microtubule-binding complex, other proteins with demonstrated microtubule-binding activities localize to kinetochores. One such protein is the cytoplasmic linker-associated protein 2 (CLASP2). Here, we show that global GSK3-mediated phosphorylation of the longest isoform, CLASP2α, largely abolishes CLASP2α-microtubule association in metaphase. However, it does not directly control localization of CLASP2α to kinetochores. Using dominant phosphorylation-site variants, we find that CLASP2α phosphorylation weakens kinetochore-microtubule interactions as evidenced by decreased tension between sister kinetochores. Expression of CLASP2α phosphorylation-site mutants also resulted in increased chromosome segregation defects, indicating that GSK3-mediated control of CLASP2α-microtubule interactions contributes to correct chromosome dynamics. Because of global inhibition of CLASP2α-microtubule interactions, we propose a model in which only kinetochore-bound CLASP2α is dephosphorylated, locally engaging its microtubule-binding activity.

Keywords: CLASP2; GSK3; Kinetochore; Microtubule; Mitosis; Phosphorylation.

Fig. 1.

Positively charged surface loops in TOG3 are necessary for CLASP2 binding along MTs. (A) Domain structure of human CLASP2 isoforms and truncated CLASP2(497-1515) depicting the three TOG domains, the KT-binding domain (KT) and intrinsically disordered multisite phosphorylation regions. NCBI reference sequences: CLASP2α NP_055912; CLASP2γ NP_001193973. (B) Immunoblot of endogenous CLASP2α in lysate from cells treated with a GSK3 inhibitor (20 µM SB216763), that expressed constitutively active GSK3β(S9A) (GSK3β) or that had been arrested in mitosis with an Eg5/KIF11 inhibitor (5 µM STLC). For full-length CLASP2α, the GSK3-mediated gel shift is difficult to discern, and the migration of the different phosphorylated species is indicated. The faint band slightly below CLASP2α is likely to be unspecific. (C) Structures of the TOG3 domain from mouse CLASP2 (PDB ID: 3WOZ), which is identical to the human protein except this structure is missing arginine 931, showing the location of surface lysine residues in loops 2 and 3. (D) Images of HeLa cells expressing the non-phosphorylatable CLASP2(9×S/A) variant of EGFP–CLASP2(497-1515), which binds strongly along the MT lattice with additional amino acid substitutions as indicated in the charged surface loops. Mutation of the positively charged residues in either loop abolished binding along MTs but had no effect on MT plus-end association. WT, wild type.

Fig. 2.

Mitotic phosphorylation inhibits CLASP2α binding to MT ends. (A) EGFP–CLASP2α localization in different cell cycle phases in HaCaT cells that stably expressed histone-H2B–mCherry. (B) EGFP–CLASP2α phosphorylation-site mutants in cells arrested in metaphase with 10 µM MG132. (C) Analysis of EGFP–CLASP2α binding to growing MT ends during different phases of the cell cycle. (D) Comparison of binding to growing MT ends of the indicated CLASP2 phosphorylation-site mutations in metaphase-arrested spindles. Each dot represents the average of three measurements per cell. 8×S/D, eight serine phosphorylation sites mutated to aspartic acid residues in the CLASP2 constructs indicated; 9×S/A, nine serine phosphorylation sites mutated to alanine in the CLASP2 constructs indicated; Ana, anaphase; Inter, interphase; Meta, metaphase; Pro, prophase; Prometa, prometaphase; Telo, telophase; WT, wild type.

Fig. 3.

CLASP2α binding to KTs is not directly controlled by GSK3-mediated phosphorylation. (A) Metaphase-arrested HaCaT cells expressing the indicated EGFP–CLASP2α phosphorylation-site mutants (top) and, in addition, treated with 1 μM nocodazole (bottom). Insets: KT pairs at higher magnification. (B) Relative enrichment of EGFP–CLASP2α on KTs compared to the signal in the cytoplasm. Each dot represents the average of three KT measurements per cell. Noco., nocodazole. (C) Time-lapse sequences of sister KT oscillations in metaphase cells expressing the indicated constructs illustrating EGFP–CLASP2α(9×S/A) (9×S/A) enrichment on the anti-poleward-moving sister KT. (D) Example intensity profiles across sister KT pairs showing double Gaussian fits (solid line) and illustrating how distances between sister KTs (ΔKT) and KT-associated fluorescence intensity was calculated. (E) The EGFP–CLASP2α fluorescence ratio between sister KTs. n=number of KT pairs analyzed. Outliers are shown as individual data points. 8×S/D, EGFP–CLASP2α(8×S/D); WT, wild type.

Fig. 4.

GSK3-mediated CLASP phosphorylation weakens KT–MT interactions. (A) Analysis of ΔKT in cells expressing the indicated EGFP–CLASP2α constructs, or treated with 80 nM nocodazole (Noco.) or 1 μM taxol. n=number of KT pairs analyzed. Outliers are shown as individual data points. (B) ΔKT distributions plotted as histograms and overlaid with Gaussian fits (solid lines) illustrating differences in distribution means and widths. (C) Representative time-lapse data of CENPA–mCherry-labeled metaphase KT dynamics in cells expressing the indicated EGFP–CLASP2α constructs. Shown are color-coded maximum intensity projections to follow KT pairs over time (min:s). Solid and open arrowheads indicate sister KTs of specific pairs showing unusual dynamics in the cells expressing EGFP–CLASP2α phosphorylation-site mutants. (D) ΔKT and alignment relative to the spindle axis over time of the KT pairs indicated in C. 8×S/D, EGFP–CLASP2α(8×S/D); 9×S/A, EGFP–CLASP2α(9×S/A); WT, wild type.

Fig. 5.

Dynamic CLASP2 phosphoregulation promotes error-free chromosome segregation. (A) Representative images of anaphase and telophase cells expressing EGFP–CLASP2α GSK3 phosphorylation-site mutants with chromosome segregation defects. Bar graph shows the proportion of cells in or post anaphase with lagging chromosomes or chromatin bridges. (B) Mitotic forms in MG132-arrested control cells and cells in which CLASP2 had been depleted using shRNA. The graph shows the proportion of cells with normally aligned metaphase plates in cells in which the knockdown phenotype was rescued by expression of the indicated EGFP–CLASP2α phosphorylation-site mutants. In both A and B, different symbols show mean results from three independent experiments, and the bar graphs show the overall mean of these experiments. n=total number of mitotic cells analyzed. 8×S/D, EGFP–CLASP2α(8×S/D); 9×S/A, EGFP–CLASP2α(9×S/A); WT, wild type.

Fig. 6.

Model of KT-associated CLASP2α phosphoregulation. We propose a hypothetical model in which CLASP2 at the KT is locally activated by dephosphorylation, which is consistent with the observed localization and changes in KT dynamics when CLASP2 GSK3 phosphorylation-site mutants are expressed.