Mitotic spindle association of TACC3 requires Aurora-A-dependent stabilization of a cryptic α-helix

Abstract

Aurora-A regulates the recruitment of TACC3 to the mitotic spindle through a phospho-dependent interaction with clathrin heavy chain (CHC). Here, we describe the structural basis of these interactions, mediated by three motifs in a disordered region of TACC3. A hydrophobic docking motif binds to a previously uncharacterized pocket on Aurora-A that is blocked in most kinases. Abrogation of the docking motif causes a delay in late mitosis, consistent with the cellular distribution of Aurora-A complexes. Phosphorylation of Ser558 engages a conformational switch in a second motif from a disordered state, needed to bind the kinase active site, into a helical conformation. The helix extends into a third, adjacent motif that is recognized by a helical-repeat region of CHC, not a recognized phospho-reader domain. This potentially widespread mechanism of phospho-recognition provides greater flexibility to tune the molecular details of the interaction than canonical recognition motifs that are dominated by phosphate binding.

Keywords: disorder–order transition; intrinsically disordered protein; phosphorylation; protein kinase; protein–protein interaction.

Figure 1. Three interaction sites in the crystal structure of Aurora‐A/TACC3

1. Schematic illustration showing the domain structures and interacting regions of Aurora‐A, TPX2, CHC and TACC3. TPX2 1–43 stabilizes the active conformation of Aurora‐A through interactions with the αC‐helix and the phosphate‐bearing activation loop (left). TACC3 F525 is crucial for the interaction with Aurora‐A (i). Phosphorylation of TACC3 S558 by Aurora‐A (ii) is required for the subsequent interaction with the CHC ankle region (iii).

2. View of the Aurora‐A/TACC3 structure centred on the TACC3 molecule (orange surface), which interacts with three molecules of Aurora‐A (blue, light blue, teal cartoon).

3. View of the Aurora‐A/TACC3 structure centred on one Aurora‐A molecule (teal cartoon), which interacts with three molecules of TACC3 (orange, red, brown).

4. Diagram showing which TACC3 residues interact with Aurora‐A at each of the three sites (teal, blue and light blue boxes), with key interactions marked as “+” and secondary structure of TACC3 indicated. Backbone chemical shift mapping is of inactive Aurora‐A upon addition of TACC3^N‐ACID (solid line; Burgess et al, 2015).

Figure EV1. Structural analysis & size‐exclusion chromatography of the Aurora‐A/TACC3 complex

1. Size‐exclusion chromatographs of Aurora‐A and TACC3 alone and in complex on a Superdex 200 16/600 column (GE Healthcare).

2. Electron density map of the Aurora‐A/TACC3 complex. Aurora‐A is coloured cyan. TACC3 is coloured orange. The 2mFo‐DFc electron density map is shown as a wire mesh contoured at 1.0 σ.

3. Side chain HC chemical shift perturbations observed for ¹³C¹⁵N TACC3 519–540 on the addition of Aurora‐A^KD D274N.

4. Small section showing mainly methyl group resonances from a ¹³C‐HSQC spectrum of TACC3 519–540. The spectrum for TACC3 519–540 is shown in blue. The spectrum collected on the addition of Aurora‐A^KD D274N is coloured red.

Figure 2. TACC3 523–534 forms a docking interaction on the N‐lobe of Aurora‐A

1. Magnified view of the site 1 interaction.

2. Schematic showing the key features of the TACC3/Aurora‐A interface.

3. FP assays of binding between site 1 of TACC3 and Aurora‐A. The sequence of TACC3 present in the complex crystal structure is shown (top) with the residues of the site 1 FP peptide coloured orange and mutated residues identified by coloured circles corresponding to the curves shown in the FP assay (below). Data represent mean of three experiments ± SD. The binding affinity between WT Aurora‐A and FAM‐TACC3 522–536 was 20 ± 1 μM. L532A and F525A mutations resulted in reductions in affinity to 251 ± 14 μM and 2.3 ± 0.4 mM, respectively.

4. FP assays between FAM‐labelled TACC3 site 1 peptide and Aurora‐A mutants (as labelled). Data represent mean of three experiments ± SD. The binding affinity between WT Aurora‐A and FAM‐TACC3 522–536 was 20 ± 1 μM. The affinities between FAM‐TACC3 522–536 and the Aurora‐A point mutants were all reduced compared to the WT. The E134A mutant had an affinity of 98 ± 5 μM; R137A, 149 ± 7 μM; L149A, 79 ± 4 μM; R151A, 771 ± 61 μM; and I158A 163 ± 8 μM for TACC3.

5. Kinase assay to monitor autophosphorylation of unphosphorylated Aurora‐A KD and substrate phosphorylation of kinase‐dead Aurora‐A D274N alone and in the presence of TACC3. Reactions were analysed by SDS–PAGE (above) and Western blotting with an αphospho‐T288 Aurora‐A antibody (below).

Figure EV2. Biochemical and structural analysis of the Aurora‐A/TACC3 complex

1. FP assay between FAM‐TACC3 519–570 and Aurora‐A. Reactions were performed in triplicate. Error bars represent standard deviation. The binding affinity between WT Aurora‐A and FAM‐TACC3 519–570 was 3.0 ± 0.2 μM.

2. FP assays between wild‐type FAM‐TACC3 site 1 peptide (residues 522–536) and mutants, and Aurora‐A. Reactions were performed in triplicate. Error bars represent standard deviation. The binding affinity between WT Aurora‐A and FAM‐TACC3 522–536 was 20 ± 1 μM. In contrast, the K _d for the Aurora‐A interaction with a FAM‐TACC3 522–536 peptide incorporating the R526A mutation was 49 ± 2 μM, a modest increase consistent with the minor contribution of the residue to the interface with Aurora‐A. One further peptide, which incorporated phosphorylation of T534, was determined to have a K _d of 14 ± 1 μM. This peptide was based on the location of a sulphate ion, originating from the crystallization solution, at the interface of the Aurora‐A/TACC3 interface in the crystal structure, close to TACC3 T534. The modest increase in affinity suggests that this post‐translational modification is not significant in regulation of the interaction.

3. In vitro radioactive kinase assay to determine the effect of TACC3 and phosphorylated Aurora‐A mutants on Aurora‐A kinase activity. Myelin basic protein (MBP) was used as a generic kinase substrate. Incorporation of radioisotope was measured by scintillation counting. Reactions were performed in duplicate. Error bars represent standard error. Assays were compared to the WT TACC3^N‐ACIDH6c reaction by one‐way ANOVA with Dunnett's post hoc test. *P < 0.1, **P < 0.01 and ****P < 0.0001.

4. Multiple sequence alignment of TACC3 (above) and Aurora‐A (below) orthologs spanning the interaction binding interface. Key binding residues are boxed in red. The TACC3 phosphorylation sites, S552 and S558, are boxed in blue. The TACC3 LL motif required for binding to CHC is boxed in green. The TACC3 α‐helix produced on binding to CHC is identified by a black line.

Figure 3. Structural analysis of the Aurora‐A/TACC3 complex

1. Aurora‐A^M3KD/TACC3^N‐ACID complex, coloured as in Fig 2A. K162 and E181 (blue spheres) do not form a salt bridge, consistent with an inactive state of the kinase.

2. Aurora‐A^KD/TPX2^1‐43 complex. K162 and E181 (blue spheres) form a salt bridge, consistent with an active state of the kinase (PDB 1OL5).

3. Structure of PKA (PDB 1JBP) with the core kinase domain (grey) and N‐ and C‐terminal extensions (magenta). K72 and E91 (blue spheres) form a salt bridge, consistent with an active state of the kinase.

4. Magnified view of the Aurora‐A^M3KD/TACC3^N‐ACID complex, centred on the activation loop (red), on which the activation loop from the Aurora‐A/TPX2^1‐43 structure (pink, PDB 1OL5) is superposed.

5. Magnified view of the Aurora‐A^M3/vNAR‐D01 complex, centred on the activation loop (brown), which adopts a DFG‐up conformation.

6. Schematic illustration of the mechanisms through which TPX2 and TACC3 activate Aurora‐A.

7. Computational model of Aurora‐A in its active conformation (based on PDB 1OL5) in complex with docking and substrate regions of TACC3.

Data information: Black arrows in (A–C) indicate hydrophobic contacts that are in structurally conserved locations.

Figure EV3. Structural analysis of the Aurora‐A/TACC3 complex

1. A

   Alignment of N‐lobe sequences from human kinases Aurora‐A, Plk1, Abl and PKA. Arrows mark residues that line the pocket that binds TACC3 L532. Aurora‐A E181 is in blue.

2. B–E

   Structures of kinase interactions at the N‐lobe pocket. (B) Aurora‐A (cyan)/TACC3 (orange). (C) Plk1 N‐lobe (grey)/cap (beige; PDB 2OU7). (D) Abl N‐lobe (grey)/SH2 (beige; PDB 1OPL). (E) CK2α N‐lobe (grey)/Pc peptide (beige; PDB 4IB5).

3. F

   Alignment of primary sequences from human Aurora‐A, Aurora‐B and Aurora‐C covering the binding site of TACC3^N‐ACID on Aurora‐A.

4. G

   Crystal structure of Xenopus laevis Aurora‐B (grey)/INCENP (yellow) complex (PDB 2BFX), with TACC3 (orange) superposed.

Figure 4. Defects in spindle localization and late mitotic progression in CRISPR/Cas9‐engineered HeLa cells with mutations in TACC3

1. A

   PLA showing Aurora‐A/TACC3 (top panel) and Aurora‐A/TPX2 (bottom panel) complexes during mitosis in HeLa cells. Cell cycle stages are indicated above cell images.

2. B

   Dot plot indicates volume of TACC3 protein at spindle pole regions of WT and F525A HeLa cells in telophase. Volumes were measured for 90 spindle pole regions per cell line (45 telophase cells). Representative images are shown in Fig EV4C.

3. C, D

   PLA between Aurora‐A and TACC3 in WT, ΔTACC and F525A HeLa cells. Representative metaphase (C), and anaphase and telophase cells (D) are shown in panels on the left. Box and whisker plots on the right represent the number of PLA dots counted per cell in either metaphase or anaphase/telophase cells, respectively. 30 cells were analysed for each cell cycle stage for each cell line.

4. E

   Duration of anaphase to chromosome decondensation, measured by time‐lapse microscopy in WT, ΔTACC and F525A HeLa cells. Number of cells analysed is indicated on graph (n).

Data information: Cell images include DNA stained with DAPI (blue). Scale bars, 10 μm. In the dot plots (B, E), horizontal lines represent the mean and error bars correspond to standard deviation. P‐values were obtained from Mann–Whitney: ****P < 0.0001. For the box and whisker plots (C, D), the middle horizontal line marks the median and the box represents the 25^th and 75^th percentile. The whiskers extend from the minimum to the maximum values. P‐values were obtained from Student's t‐test, unpaired, two‐tailed: ****P < 0.0001.

Figure EV4. Analysis of mitotic timings in CRISPR‐Cas9 edited cells

1. A

   WT HeLa cells were subjected to a double‐thymidine block, released and fixed with ice‐cold methanol at time points indicated. Cells were processed for PLA using Aurora‐A and TACC3 primary antibodies. FACS profiles show cell cycle distribution, and images below are representative examples of cells from the indicated time point.

2. B

   WT HeLa cells were fixed and processed for PLA with Aurora‐A and TACC3 antibodies (left) or TACC3 antibody alone (right). Representative images are shown.

3. C

   WT, F525A and ΔTACC‐HeLa cells were immunostained with TACC3 (Abcam, 1:200, rabbit) antibody followed by incubation with goat anti‐rabbit Alexa Fluor 488 (1:200). Note the different pattern of localization of TACC3 in F525A cells compared to WT, both in metaphase and telophase.

4. D, E

   WT HeLa cells were immunostained for (D) Aurora‐A (Sigma, 1:500, mouse) or (E) TPX2 (Novus biologicals, 1:250, rabbit) (green).

5. F

   Duration of telophase to chromosome decondensation, measured in WT, F525A and ΔTACC‐HeLa cells by time‐lapse microscopy. Number of cells analysed is indicated on graph (n).

6. G

   Duration of anaphase to telophase measured in WT, F525A and ΔTACC‐HeLa cells by time‐lapse microscopy. Number of cells analysed is indicated on graph (n).

7. H

   Duration of NEBD to anaphase, measured in TACC3 variants by time‐lapse microscopy. Number of cells analysed is indicated on graph (n).

Data information: All scale bars, 10 μm. To assess cell cycle phase, cells in (C–E) were co‐stained for either α‐ or γ‐tubulin (Sigma, 1:500, mouse; Sigma, 1:500, rabbit, respectively) (red), and DNA stained with DAPI (blue). Where γ‐tubulin co‐stain was used instead of α‐tubulin, this was due to the α‐tubulin rabbit antibody showing weak spindle staining. In the dot plots (F–H), horizontal lines represent the mean and the error bars correspond to standard deviation. P‐values were obtained from Mann–Whitney: ****P < 0.0001; n.s., not significant.

Figure 5. Crystal structure of ClACCp

1. Schematic illustration of CHC (grey) and TACC3 (orange) domain structures.

2. Summary of the binding data between CHC proteins and TACC3^CID peptides. See also panel (G) and Fig EV5A and B.

3. FP assays to measure the binding affinity of CHC 1–574 to FAM‐labelled TACC3^CID phosphorylated (pS558), unphosphorylated (S558) and phosphomimetic (S558E) forms. See also Fig EV5C.

4. Overview of the crystal structure of ClACCp. Note that the linker between CHC aa574 and TACC3 aa550 is disordered.

5. Magnified view of the CHC/TACC3 interaction. The side chains of key interacting residues are shown as sticks.

6. Schematic illustration of the CHC/TACC3 interface.

7. FP binding curves of CHC 1–574 with FAM‐phospho‐TACC3^CID bearing alanine point mutations. Affinities are in parentheses. Data represent mean of three experiments ± SD.

Figure EV5. FP binding data to study the interaction between CHC and TACC3

1. FP assays were performed to measure the binding affinities of four different CHC fragments to FAM‐TACC3 549–570 phosphorylated on S558. The longest fragment (aa1–574) had the tightest binding, whereas the shorter fragments showed progressively weaker binding.

2. FP assays were carried out to measure the binding affinities of CHC 1–574 with different length FAM‐TACC3 peptides phosphorylated on Ser558.

3. FP assays to measure the binding affinity of CHC 1–574 to FAM‐non‐phosphorylated TACC3^CID and phosphomimetic S558E TACC3^CID.

Data information: Reactions were performed in triplicate with error bars representing standard deviation.

Figure 6. Mutations in the TACC3/CHC interface reduce spindle localization of the complex

1. Cartoon representation of the TACC3/CHC interface (orange and grey, respectively). Key CHC residues are shown as spheres.

2. FP binding curves of CHC 1–574 mutants with FAM‐phospho‐TACC3^CID. Affinities are in parentheses. Data represents mean of three experiments ± SD.

3. Representative confocal images of HeLa cells at metaphase expressing the indicated GFP‐TACC3 construct on a background of TACC3 depletion and stained for tubulin and CHC. Below, scatter plot to show mitotic spindle enrichment of TACC3 constructs (red) and endogenous CHC (blue); GFP controls are pale red. ANOVA with Tukey's post hoc test, only comparison of TACC3 constructs with WT is shown. *P < 0.05, ***P < 0.001.

4. Representative confocal images of HeLa cells at metaphase expressing the indicated GFP‐CHC construct on a background of CHC depletion and stained for tubulin and TACC3. Below, scatter plot to show mitotic spindle enrichment of CHC constructs (blue) and endogenous TACC3 (red); GFP controls are pale blue. ANOVA with Tukey's post hoc test, only comparison of CHC constructs with WT is shown. **P < 0.01, ***P < 0.001.

Data information: Each spot represents the average measurement per cell. Bars show mean values ± standard deviation. Scale bar, 10 μm.

Figure 7. Phosphorylation of S558 promotes formation of an α‐helix that binds CHC

1. Structure of unphosphorylated TACC3^N‐ACID extracted from site 2 of the complex with Aurora‐A. Disordered side chains are marked with asterisks.

2. Two views of phosphorylated TACC3^CID extracted from the complex with CHC. Polar interactions are marked with dashed lines.

3. Superposition of the structures from panels (A) and (B) shows the transition in residues 558–563 from extended conformation to α‐helix.

4. Superposition of the ¹³C‐HSQC spectra of unphosphorylated (red) and phosphorylated S558 (blue) TACC3 549–570 peptides. Shown is the region of the Cα‐Hα correlations from which the data for the secondary chemical shifts calculations were derived.

5. Secondary chemical shifts for unphosphorylated (red) and phosphorylated S558 (blue) TACC3 549–570 peptides. Data for αH are shown on the top and αC on the bottom.

6. FP binding curves of CHC 1–574 with WT and mutant FAM‐phospho‐TACC3^CID. Affinities are in parentheses. Data represent mean of three experiments ± SD.

Figure 8. Schematic model of the interactions and conformational changes in TACC3 during its recruitment to the mitotic spindle

(1) TACC3 docks to Aurora‐A through a motif centred on F525 and L532. (2) Activated Aurora‐A phosphorylates TACC3 on S558. (3) Phosphorylated TACC3 has increased helical propensity. (4) The helical conformation of TACC3 docks to CHC via hydrophobic residues L559, Y560 and F563.