1. Volume 3, Issue 3, Pages 759-768 (March 2013)
Structural Model for Tubulin Recognition and Deformation by Kinesin-13 Microtubule Depolymerases 
Ana B. Asenjo, Chandrima Chatterjee, Dongyan Tan, Vania DePaoli, William J. Rice, Ruben Diaz-Avalos, Mariena Silvestry, Hernando Sosa 
Cell Reports 
Volume 3, Issue 3, Pages (March 2013)
DOI: /j.celrep
Copyright © 2013 The Authors Terms and Conditions

2. Cell Reports 2013 3, 759-768DOI: (10.1016/j.celrep.2013.01.030)
Copyright © 2013 The Authors Terms and Conditions

3. Figure 1 KLP10AHD Binds to the Tubulin Intradimer Interface
(A) Alternative inter- versus intratubulin dimer-binding modes for the KLP10AHD-tubulin complex. Two possibilities for the interdimer case are considered depending on whether the KLP10AHD can bind to a single tubulin subunit at the end of the protofilament. Cyan triangles: KLP10AHD; purple and magenta squares: tubulin.
(B) Electron micrograph of a field with open ring structures formed by incubating free tubulin with KLP10AHD in the presence of AMP-PNP. Scale bars, ∼45 nm.
(C) Electron micrographs showing tubulin ring depolymerization intermediates in the background and at the MT ends formed by incubating MTs with KLP10AHD in the presence of ATP.
(D) Three independent image class averages. The tubulin dimer at the protofilament end is indicated by the red brackets. The inset shows atomic models of tubulin (purple and magenta) and the KLP10HD (cyan) superimposed on one of the average images. Scale bars, ∼8 nm.
See also Figures S1, S2, and Movie S5.
Cell Reports 2013 3, DOI: ( /j.celrep )
Copyright © 2013 The Authors Terms and Conditions

4. Figure 2
Nucleotide Dependence of KLP10AHD Binding to Tubulin Protofilaments
(A, C, and E) Average negatively stained protofilament ring structures formed by incubating free tubulin with KLP10AHD in the presence of AMP-PNP (A), ADP-AlF4− (C), or ADP (E).
(B, D, and F) Images of aligned and averaged asymmetric units (consisting of a tubulin heterodimer and associated KLP10AHD) in (A), (C), and (E), respectively.
(G) Density contour plots of the average KLP10AHD-tubulin complex in the presence of AMP-PNP (red) or ADP-AlF4− (blue).
See also Figures S1 and S2.
Cell Reports 2013 3, DOI: ( /j.celrep )
Copyright © 2013 The Authors Terms and Conditions

5. Figure 3
Cryo-EM Nanometer-Resolution Structure of the KLP10AHD-Tubulin Ring Complex
(A) Isodensity surface representation of the cryo-EM 3D reconstruction electron density map. Surfaces are colored according to position from the helical axis and fitted atomic structures. Yellow, MT; blue, KLP10AHD; purple, outside curved tubulin protofilament.
(B) Detail of the MT surface showing densities contouring the exposed α-helices on the MT surface.
(C) End-on view of the 3D map density (transparent gray) and the fitted atomic structures of tubulin in the MT. Secondary structural elements of the KLP10AHD at the interface with the curved protofilament (Kin-Tub-1 interface) are indicated.
(C–F) Atomic structures are shown in ribbon representation. Yellow, α-tubulin fitted in the MT density; gold, β-tubulin fitted in the MT density; blue, KLP10AHD; magenta, α-tubulin fitted in the curved protofilament; purple, β-tubulin fitted in the curved protofilament. Cryo-EM isodensity surface is displayed as a semitransparent surface in (B) and (C), and as a mesh in (D)–(F).
(D–F) Close-up of the three areas of interaction between the KLP10AHD and tubulin at the Kin-Tub-1 interface. Putative interacting residues between the KLP10AHD and tubulin (<5 Å separation) are colored dark blue in KLP10HD and red in tubulin. Labels indicate the secondary structure elements (H, tubulin helix; S, tubulin sheet; α, kinesin helix) where the putative interacting residues are located. Labels are also color-coded according to location (blue, KLP10AHD; red, α-tubulin; green, β-tubulin).
See also Figures S3, S5, and Movies S1 and S6.
Cell Reports 2013 3, DOI: ( /j.celrep )
Copyright © 2013 The Authors Terms and Conditions

6. Figure 4 CS-Tubulin Structure
(A) Comparison of CS tubulin and straight tubulin. Left: end-on view (tubulin protofilament luminal side toward the left). Right: back view (tubulin protofilament luminal side toward the observer). Fitting the structure of the α-β straight tubulin heterodimer (PDB: 1JFF) to best match its α-subunit to the cryo-EM density map leaves the β-subunit outside the density and vice versa (not shown). A better fit is achieved when the two tubulin monomers are independently fitted in the cryo-EM density. The resulting model of the tubulin dimer structure has curvature and shear between the α- and β-subunits (CS tubulin). Yellow, straight tubulin; purple, CS tubulin; blue, KLP10AHD. The cryo-EM 3D map is shown as a semitransparent isodensity surface.
(B and C) The KLP10AHD-CS-tubulin complex (B) and kinesin-1-straight tubulin complex (C; PDB: 2P4N). In (B) and (C) the top corresponds to an end-on view of the complex, and the bottom corresponds to a frontal view (tubulin protofilament luminal side away from the observer). Tubulin atomic structure is shown as a space-filling model (α-tubulin in light yellow; β-tubulin in dark yellow) and the kinesin HD is shown in ribbon representation (blue). In the bottom views, the kinesin HD structure is omitted to show kinesin-binding sites on tubulin. Residues in tubulin at ≤5 Å from residues in the bound kinesin HD (putative kinesin-binding site) are shown in red. Three tubulin residues in the putative kinesin-binding region (α-E415, β-P263, and β-T419) are shown in dark blue to highlight the different relative positions of kinesin-binding areas in the CS-tubulin configuration (B) versus straight tubulin (C). The pink-colored residue on the β-tubulin subunit of the KLP10AHD-CS-tubulin complex (B) corresponds to the one that would be interacting with the L2 tip of a kinesin HD bound to the next tubulin dimer along the protofilament. Tubulins are oriented with the plus end at the top.
See also Figures S4, S5, and Movies S2, S3, S4, and S6.
Cell Reports 2013 3, DOI: ( /j.celrep )
Copyright © 2013 The Authors Terms and Conditions

7. Figure 5 KLP10AHD-MT-Binding Configuration
(A) Location and transition dipole orientation of the BSR probe (red arrow) attached to the KLP10AHD (blue). CS tubulin (light pink and purple) is positioned relative to KLP10AHD as in the cryo-EM fitted complex (Figure 3).
(B) FPM images of MTs decorated with BSR-labeled KLP10AHD. The two side-by-side images correspond to the same field but with different polarization directions of the excitation light (green arrows). The MTs are decorated with a control (WT) KLP10A-BSR construct in the top two images and with the KLP10A L2 KVD mutant in the bottom two. Note the higher fluorescence anisotropy in the top images relative to the bottom ones. In the top images, the fluorescence intensity is strongly dependent on the relative orientations of the MT axis and the excitation polarization direction (green arrow), i.e., more intense when it is perpendicular and less intense when it is parallel. This indicates that the fluorescence transition dipole is nearly perpendicular to the MT axis in the example shown in the top images, and is more disordered in the bottom ones (random orientations).
(C) Fluorescence polarization LD0 values (calculated from the ratio between fluorescence intensities with different excitation polarization directions) of KLP10A-BSR constructs in the presence of ADP or absence of nucleotides (NN). The value of the LD0 parameter depends on the average orientation of the BSR probe relative to the MT, with positive values indicating an average perpendicular orientation (90°–54.7°) from the MT axis. Increased orientational disorder from the average orientation has the effect of reducing |LD0| values (Sosa et al., 2010). Error bars represent the SD of the estimated LD0 values. Number of MTs analyzed: 492, 608, 294, 192, 616, and 520 for WT-ADP, WT-NN, D44A,K446A-ADP, D44A,K446A-NN, KVD-ADP, and KVD-NN, respectively.
Cell Reports 2013 3, DOI: ( /j.celrep )
Copyright © 2013 The Authors Terms and Conditions

8. Figure 6 Structural Model for Kinesin-13-Induced MT Depolymerization
(A) The interface between KLP10AHD and CS tubulin has three major areas of interaction along the tubulin heterodimer that suggest a crossbow-type mechanism for inducing tubulin curvature. In this model, middle interactions push the tubulin intradimer relative to the two extremes, producing tubulin curvature. Bending and shearing forces are represented as red and blue arrows, respectively.
(B) Mechanochemical model for the kinesin-13-tubulin complex. The nucleotide states of kinesin-13 in solution and while interacting with the MT lattice are according to previous work (Friel and Howard, 2011). The kinesin-motor domain binds weakly to the tubulin lattice through interactions involving kinesin L2, α4, and regions outside the motor domain such as the neck (not depicted here). The kinesin-13 HD cannot bind strongly to the MT lattice due to lack of complementarity with the binding site in straight tubulin. On the MT ends, kinesin-13 finds isolated protofilaments where it can induce/stabilize the CS-tubulin structure in the ATP state. Curved protofilaments stabilized by kinesin-13 binding cannot form lateral contacts with other protofilaments and eventually break from the MT ends. Release of ATP hydrolysis products results in dissociation of the kinesin-13 HD complex. Interactions of the kinesin-13 L2 with β-tubulin in the interdimer interface drive binding to a MT end protofilament over free tubulin in solution.
Cell Reports 2013 3, DOI: ( /j.celrep )
Copyright © 2013 The Authors Terms and Conditions

9. Figure S1
Class Averages of Particles Preliminarily Identified as KLP10AHD-Tubulin Complexes, Related to Figures 1 and 2, and Table 1
The image shows class-average images resulting from correspondence analysis and k-means classification (see methods) of the group of particles corresponding to the AMPPNP nucleotide condition and preliminarily identified as KLP10AHD-tubulin ring complexes Class average images of the particles preliminarily identified as “non KLP10AHD-tubulin complexes” groups are shown in Figure S2.
Cell Reports 2013 3, DOI: ( /j.celrep )
Copyright © 2013 The Authors Terms and Conditions

10. Figure S2
Class Averages of Particle Groups Preliminarily Identified as Non-KLP10AHD-Tubulin Complexes, Related to Figures 1 and 2, and Table 1
Four representative examples in each of the ATP analogs conditions tested are shown. In most nucleotide states the particles in these groups produce class averages resembling tubulin protofilaments with no bound KLP10AHDs. However, in the case of AMPPNP many of the class averages (red asterisk) do resemble KLP10AHD tubulin ring complex similar to most class averages shown in Figure S1 albeit shorter (less number of asymmetric units), with less curvature and noisier due to the lesser number of particles in these sub-groups. Final identification of particles as bona fide KLP10AHD-tubulin ring complexes (Table 1) was based on the appearance of the class-average image to which the particle belong and not in the two preliminary groups in which they were first divided.
Cell Reports 2013 3, DOI: ( /j.celrep )
Copyright © 2013 The Authors Terms and Conditions

11. Figure S3 Cryo-EM 3D Reconstruction Resolution, Related to Figure 3
(A) Fourier shell correlation (FSC) curves were calculated from two 3D reconstructions each obtained from one half of the data set (Frank, 2006). The FSC curve (black-line) was obtained in the traditional manner by dividing the data set randomly after alignment. Another FSC∗ curve (blue line) was obtained from two independently refined data sets (each data set corresponding to one half of each filament and using 1/18 Å−1 low-pass filtered 3D maps for alignment). The FSC∗ provides a more stringent resolution criteria as it reduces the effect of spurious correlations affecting FSC curves calculated in the traditional manner (Henderson et al., 2012). The resolution of the 3D reconstruction according to the FSC∗0.143 threshold criteria is estimated at 10.8 Å. FSCCRIT (thin line) corresponds to the 3σ curve.
(B) Lateral projection of the 3D map before imposing helical symmetry.
(C) Power spectrum of the image shown in B. The power spectrum shows layer lines with 1/80 A−1 as expected from labeled MTs. Layer lines are clearly visible up to a resolution of 1/10 Å−1.
(D) Comparison of the experimentally determined 3D density map (left) with a model density low-pass filtered to 11 Å (right). Isocontour surfaces of the density maps are shown as a semitransparent gray mesh with the fitted atomic model inside. Secondary structure elements of KLP10A (L2 and L4 in blue) and MT α and β tubulin (H11 and H12 in yellow and gold) are highlighted. The model density map was calculated from the KLP10A MT ring complex atomic coordinates using the program pdb2mrc (Ludtke et al., 1999)· The map was initially calculated to a resolution of 4 Å and then low-pass filtered with a top-hat function to 11 Å. The similarity between the experimental and the 11 Å resolution model map; the presence of clear layer lines up to 1/10 Å−1 in the experimental 3D map power spectrum and the FSC∗0.143 threshold criteria; all indicate that the experimentally determined cryo-EM 3D map has a spatial resolution of 11-10 Å.
Cell Reports 2013 3, DOI: ( /j.celrep )
Copyright © 2013 The Authors Terms and Conditions

12. Figure S4 Tubulin Structure Comparison, Related to Figure 4
Three structures including two tubulin heterodimers each are superimposed by aligning the top-most α-tubulin subunit. (α-1). The curved-sheared tubulin structure (CS-tubulin) in the KLP10AHD -tubulin complex as fitted in the cryo-EM density map (Figure 3) is shown in magenta. The structure of straight tubulin (PDB ID: 1JFF; Löwe et al., 2001) as fitted into the MT cryo-EM density is shown in yellow. The structure of curved tubulin in complex with colchicine and stathmin (PDB ID: 1SA0; Ravelli et al., 2004) is shown in cyan. Left view is end-on and right view is a frontal view (from the kinesin HD when bound). Tubulin plus end is at the top.
Cell Reports 2013 3, DOI: ( /j.celrep )
Copyright © 2013 The Authors Terms and Conditions

13. Figure S5
Relative Position of Individual Protein Subunits after Many Independent Rigid-Body Fittings, Related to Figures 3 and 4
Eighteen different models of the kinesin-13 MT ring complex asymmetric unit were obtained by rigid body fitting to 18 different cryoem-3D maps. One fit to the helically averaged map reported in the manuscript (fit-1); One fit to each of 15 different asymmetric units around a MT in a non-helically averaged map (fit 2-16); And one fit to each of two maps obtained from the two halves of independently refined data sets (fit 17-18).
(A) All the models are superimposed after aligning the corresponding α-tubulin subunit in the ring protofilament (MT tubulin not shown).
(B) All the models are superimposed after aligning the corresponding α-tubulin subunit in the MT (ring tubulin and kinesin not shown).
(C) Atomic model of straight tubulin (1JFF). Each model was produced by fitting independently into the cryo-EM densities 5 atomic structures, 2 α-tubulins, 2 β-tubulins and one KLP10AHD. Fitting was done quantitatively using the operation fitmap of the program UCSF-Chimera. We first placed each subunit manually inside the corresponding density in the map and let the program find the best fit which gives the maximum correlation value between the cryo-EM density and an 11Å density calculated from the subunit atomic model (fitmap resolution option). A global search was allowed (fitmap search option) within a radius of 20Å (fitmap radius option) from the initial manual placement of the subunit. We verified that the same final best fit position/orientation was always obtained regardless of the initial manual placement of the subunit.
The relative positions of all the individual subunits in the resulting 18 models were all very similar. They all show that the tubulin in the outside ring has curvature and shear as compared to straight tubulin. Aligning the α-tubulin structure in the ring of each fitted complex (fits 2-18) places the accompanying β-subunit in the ring no further away than 2.3 Å from the β-subunit of the first fit (fit-1). Compare this small deviation with the large difference of 16-18 Å when comparing our reported CS-tubulin with straight or curved tubulin (RMSD between β-tubulins atoms after aligning α-subunits of CS-tubulin versus 1JFF or 3RYI).
Cell Reports 2013 3, DOI: ( /j.celrep )
Copyright © 2013 The Authors Terms and Conditions

14. Figure S6 Shape of Tubulin-Kinesin Interfaces, Related to Discussion
(A) Side view of the KLP10AHD-CS tubulin complex.
(B) Side view of the kinesin-1 straight tubulin complex (PDB: 2P4N; Sindelar and Downing, 2007). α-tubulin in yellow; β-tubulin in orange; kinesin HD in cyan; KLP10AHD elongated loop-2 in red. Note that the binding site of kinesin-13 is less convex than other kinesins, such as kinesin-1, mainly due to the presence of the elongated loop-2 (red) sticking out toward the minus end of α-tubulin.
Cell Reports 2013 3, DOI: ( /j.celrep )
Copyright © 2013 The Authors Terms and Conditions