Structural basis of protein condensation on microtubules underlying branching microtubule nucleation

Targeting protein for Xklp2 (TPX2) is a key factor that stimulates branching microtubule nucleation during cell division. Upon binding to microtubules (MTs), TPX2 forms condensates via liquid-liquid phase separation, which facilitates recruitment of microtubule nucleation factors and tubulin. We report the structure of the TPX2 C-terminal minimal active domain (TPX2α5-α7) on the microtubule lattice determined by magic-angle-spinning NMR. We demonstrate that TPX2α5-α7 forms a co-condensate with soluble tubulin on microtubules and binds to MTs between two adjacent protofilaments and at the intersection of four tubulin heterodimers. These interactions stabilize the microtubules and promote the recruitment of tubulin. Our results reveal that TPX2α5-α7 is disordered in solution and adopts a folded structure on MTs, indicating that TPX2α5-α7 undergoes structural changes from unfolded to folded states upon binding to microtubules. The aromatic residues form dense interactions in the core, which stabilize folding of TPX2α5-α7 on microtubules. This work informs on how the phase-separated TPX2α5-α7 behaves on microtubules and represents an atomic-level structural characterization of a protein that is involved in a condensate on cytoskeletal filaments.


Supplementary Figures
Supplementary Figure 1 | Line broadenings for TPX2 a5-a7 condensates in solution compared to the monodispersed solution. a Representative 1D 1 H traces and b 2D 1 H-15 N HSQC solution NMR spectra of free U-[ 13 C, 15 N]-TPX2 a5-a7 in monodispersed solution (black) and condensates (orange). 1 H and 15 N line broadenings are observed for most detected signals of TPX2 a5-a7 condensates in solution. The 1D traces were extracted from the 2D HSQC spectra at 15 N shifts indicated by dashed lines in b.

Supplementary Figure 2 | MAS NMR spectra of U-[ 13
C, 15 N]-TPX2 a5-a7 /MT assemblies and site-specific resonance assignments. a Mapping of assigned residues (cyan) onto the primary sequence of TPX2 a5-a7 (residues 477-716). The functional regions and regions of interests are indicated. The Phe residues where the single-site mutations severely affect TPX2 activity in stimulating branching microtubule nucleation and those that have no effect are marked by red and green dots, respectively. b 2D 13 C-13 C homonuclear correlated spectra of fully protonated U-[ 13 C, 15 N]-TPX2 a5-a7 /MT assemblies, acquired with 50 ms CORD mixing. Representative resonance assignments are labeled. The spectrum was processed with 90-(gray) and 60-degree (cyan) shifted sine-bell apodization. The spectrum was acquired at a magnetic field of 14.1 T using the sample of 3.8 mg TPX2 a5-a7 /MT assemblies in an 1.3 mm MAS rotor; the MAS frequency was 14 kHz.
The spectra were processed with 60º (gray) and 45º (teal) sine-bell shifted apodization for resolution enhancement. The contact times for 1 H-13 C and 13 C-1 H CP magnetization transfers were 0.8 ms and 0.85 ms, respectively. b Comparison of the 1D 1 H spectra of protonated (cyan) and deuterated (black) TPX2 a5-a7 /MTs. The exchangeable amide sites in deuterated TPX2 a5-a7 were 100% back-substituted with protons. For deuterated TPX2 a5-a7 /MTs, the 1 H signals primarily arise from the fully protonated microtubules, the back-exchanged amide protons and residual sidechain protons in TPX2 a5-a7 protein. The intensities of amide and sidechain protons decrease by 35% and 55% for the deuterated sample compared to the protonated sample, after a correction of signal-noise ratios based on the B0 field dependence. The 1D 1 H spectra were acquired with 16 transients. c 1D 1 H one pulse and h(C)H CPMAS spectra of deuterated TPX2 a5-a7 /MTs, acquired with 4 and 144 transients, respectively. d, e 1D and 2D 13 C-detected MAS NMR spectra of deuterated TPX2 a5-a7 /MTs. The 1D 13 C direct polarization and 1 H-13 C CPMAS spectra (d) were collected with 2048 and 8096 transients, respectively. The 2D direct polarization 13 C-13 C spectrum (e) were acquired with 2.4 ms RFDR mixing. The spectra of deuterated and protonated TPX2 a5-a7 /MTs were acquired at 20.0 T and 14.1 T, respectively. The 1D 1 H spectra of the deuterated sample were acquired at an ultrafast MAS frequency of 100 kHz using a 0.7 mm HCND MAS probe. All the other spectra were acquired at a fast MAS frequency of 60 kHz using a 1.3 mm HCN probe.

Supplementary Figure 4 | 2D 13
C-detected heteronuclear correlated MAS NMR spectra of fully protonated U-[ 13 C, 15 N]-TPX2 a5-a7 /MT assemblies. a 2D NCA spectrum, which reveals intra-residue correlations between backbone amide N and Ca atoms. b 2D NCACX and c NCOCX spectra, acquired with 50 ms 13 C-13 C mixing using the DARR recoupling scheme. The spectra reveal intra-residue correlations and sequential inter-residue correlations, respectively. Correlations of amide resonances with sidechain 13 C resonances were detected and identified. The spectra were processed with 90°-(gray) and 60°-degree (light orange and cyan) sine-bell shifted apodization.

Supplementary Figure 5 | High-resolution 1
H-detected heteronuclear MAS NMR spectra of deuterated TPX2 a5-a7 in assemblies with MTs. a, b 2D 13 C-1 H HETCOR spectra acquired at ultrafast and fast MAS frequencies. Both spectra reveal signals of backbone amide protons and residual protonation in aliphatic sidechains. There is site-to-site variation in the 1 H line widths in both spectra, indicating the conformational heterogeneity of TPX2 a5-a7 condensates on MTs. The spectra were acquired at MAS rates of 100 kHz (a) and 60 kHz (b) using a 0.7 mm HCND probe and 1.3 mm HCN probe, respectively. The CP contact time for the 1 H-13 C and 13 C-1 H transfers was 2.1 ms. c 2D 13 Ca-1 H correlated spectrum of the 3D hCANH experiment. The first 2D plane was acquired with 320 transients for higher signal-to-noise ratio and reveals a few signals that are absent in the 3D hCANH spectrum. In particular, several correlations of Gly residues were readily detected and identified. d 2D 15 N-1 H HETCOR spectra. The spectra were processed with 30°-(cyan) and 60°-degree (gray) sine-bell shifted apodization. e Representative 2D 15 N-1 H plane of the 3D hCONH spectrum with sequential resonance assignments. All spectra were acquired at a magnetic field of 20.0 T; the MAS frequency was 60 kHz for spectra in b-e. The deuterated U-[ 13 C, 15 N]-TPX2 a5-a7 was 100% 1 H back-exchanged and assembled with unlabeled MTs.

Supplementary Figure 6 | Chemical shift index for secondary structures of TPX2
a5-a7 bound to MTs. The secondary structural elements of TPX2 a5-a7 bound to MTs were derived from MAS NMR secondary chemical shifts of 13 Ca resonances. Positive and negative dD 13 Ca values in a consecutive segment indicate the a-helix (purple) and b-sheet (yellow) structures, respectively. Regions with a-helical structures include the segments D513-E526, F562-E567, A571-E583, Q604-E618, A634-Q641, D644-I661, K670-L680 and R688-D699 (sequentially named helix 1-8, or H1-8). Residues at the edge of helices are defined with moderate confidence. Six segments with b-sheet structures (b1-6) were revealed and include two functional regions, the segment that contain FKARP motif and Trop region. The secondary structures derived from TALOS-N prediction are shown at the top (light purple). Cylinders and arrows represent helices and sheets, respectively. Thresholds at +1 ppm and -1 ppm are marked as dashed lines.

Supplementary Figure 7 | Chemical shift index of TPX2
a5-a7 bound to MTs. Secondary chemical shifts of Dd 13 Ca-Dd 13 Cb, Dd 13 Cb and Dd 13 C' were plotted vs. residue number. The secondary structures derived from dD 13 Ca-dD 13 Cb are consistent with those from dD 13 Ca. The only difference is that residues F562-E565 in g-TuNA b motif show tendency to adopt b-sheet structure instead of a-helix.
Supplementary Figure 8 | TPX2 a5-a7 models generated from Robetta (gray) and AlphaFold 2 (purple) predictions. Both homology-based structure predictions reveal three main helices that were previously known from other algorithms such as Jpred. The AlphaFold 2 prediction indicates an unfolded structure of TPX2 a5-a7 , which is a clear discrepancy as compared to the MAS NMR experimental results. Unassigned residues are shown in light yellow. b, c Conformation of the functional regions and regions of interest in TPX2 a5-a7 bound to MT. TPX2 a5-a7 binds to MTs at the tubulin interdimer interface, with the g-TuNA a and b motifs, the Trop region and FKARP motif positioned facing away from the MT. d Distributions of the hydrophobic (orange) and polar residues (salmon) in the 3D molecular structure of TPX2 a5-a7 . The hydrophobic residues are evenly distributed, and a large portion of the polar residues are surface exposed. e Side view of MTs decorated with TPX2 a5-a7 (teal). The models are generated by molecular docking in ClusPro 2.0 without any experimental restraints. Majority of the residues determined by dREDOR-based experiments (orange surface) constitute the binding interface of TPX2 a5-a7 with MTs. The Eg5 region (magenta) remain exposed, which ensures its accessibility to Eg5. The a and b tubulin in tandem heterodimers are differentially color coded.

Supplementary Figure 11 | Intramolecular long-range C-H contacts in TPX2
a5-a7 protein on microtubules. a, b 2D CH HETCOR MAS NMR spectra of deuterated TPX2 a5-a7 /MT assemblies. Intra and inter-residue correlations between amide 1 H and backbone or sidechain 13 C resonances were identified. The spectra were acquired at 20.0 T and a MAS rate of 60 kHz. The CP contact times were 2.7-2.8 ms for a and 2.1 ms for b. c Mapping of inter-residue 13 C-1 H distances on the MAS NMR structure of TPX2 a5-a7 bound to MTs. These medium and long-range restraints in TPX2 a5-a7 typically correspond to interatomic distances of 3-8 Å in the TPX2 a5-a7 structure. The identified correlating residues and atoms are shown in sticks and spheres, respectively.

Supplementary Figure 12 | Dipolar-filtered MAS NMR spectra of deuterated TPX2
a5-a7 /MT assemblies. a 1D 13 C direct polarization (orange), 1 H-13 C (black) and dREDOR-filtered (cyan) CPMAS spectra. The strong signals at 13 C chemical shifts of 66 and 76 ppm arise from solvent. b 1D 1 H-13 C dREDOR-CPMAS spectrum (cyan, S) and the control spectrum with zero-quantum relaxation (black, S0). The spectra were acquired with a dephasing time of 0.43 ms and the same relaxation time, respectively. Intense Ca signals (44-64 ppm), which are present in the control spectrum, are completely dephased with dREDOR dipolar filters. c, d 1D dREDOR-CPMAS spectra with dephasing times ranging from 0.43 ms (6 rotor cycles) to 2.0 ms. The signals of directly 13 C and 15 N-bonded protons were suppressed with dREDOR dephasings of 0.43 ms or longer. The S0 control spectra shown in c were acquired with the same relaxation duration as dREDOR dephasing times of 0.86 ms (black) and 0.57 ms (purple) for S spectra. e Overlay of 2D 1 H-13 C HETCOR spectrum(gray) with dREDOR-CP-based HETCOR spectrum (0.43 ms dephasing, cyan).

Supplementary Figure 13 | The minor binding mode for TPX2
a5-a7 interacting with MTs from the molecular modeling. This binding mode was predicted in a blind docking on polymerized microtubules and b restraints-guided docking on ba tubulin interdimer, with substantial less populations than the predominant binding mode. The minor binding mode is ruled out since it contradicts prior experimental evidence showing that the removal of Eg5 domain does not significantly reduce the binding affinity of TPX2 a5-a7 with MTs. The Eg5 binding domain (magenta) is indicated in dashed squares. Other functional regions and motifs of interests are colored in salmon in a and light magenta in b.

Supplementary Figure 14 | Intermolecular interface for TPX2
a5-a7 -MTs interaction. a Residues of TPX2 a5-a7 and MTs at the binding interfaces (colored in orange and salmon, respectively), derived from dREDOR-based MAS NMR experiments. b Interface residues (salmon) of MTs that likely interact with TPX2 a5-a7 . These interaction hotspots were identified using tentative assignments for 1 H resonances of tubulin based on SHIFTX2 prediction.

Supplementary Figure 15 | Dynamic residues and intramolecular contacts in TPX2
a5-a7 bound to MTs affected by temperature changes. a Mapping of inter-residue contacts (dashed lines) onto TPX2 a5-a7 sequence. These correlations are absent in the MAS NMR spectra acquired at the sample temperature of 15 °C. The Phe residues where single-site mutations severely affect TPX2 activity in stimulating microtubule nucleation and those having no effect function are indicated by brown and cyan dots, respectively. b Comparison of the 1D traces of 1 H-13 C CP-based CORD spectra acquired at 3 °C (cyan) and 15 °C (black). Both spectra were acquired with 200 ms homonuclear CORD mixing. The signal-to-noise ratios were 37.1 and 41.6 for the spectra acquired at 3 °C and 15 °C, respectively. There is no substantial difference in the spectral resolution and sensitivity of the two spectra. c Contact map of the intra-and inter-residue restraints in TPX2 a5-a7 bound to MTs that disappear in the CORD spectra at 15 °C. Contacts involving aromatic residues are color coded in purple; those involving dynamic lysine sidechains are shown in orange; others are shown in black. Functional regions are marked by rectangles in light cyan. Secondary structures of TPX2 a5-a7 derived from secondary chemical shifts are indicated on the side; rectangular bars and arrows represent helices and sheets, respectively.

R123 HN A PF1
* : A and B chain denote to a and b tubulin, respectively. # : PF1 and PF2 denote to neighboring protofilaments of MTs (indicated in Figure S12).