Journal of Structural Biology 184 (2013) 335–344

Contents lists available at ScienceDirect

Journal of Structural Biology

journal homepage: www.elsevier .com/locate /y jsbi

Common mechanistic themes for the powerstroke of kinesin-14 motors

1047-8477/$ - see front matter � 2013 Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.jsb.2013.09.020

⇑ Corresponding author.E-mail address: hoenger@colorado.edu (A. Hoenger).

Miguel A. Gonzalez a, Julia Cope a, Katherine C. Rank b, Chun Ju Chen c, Peter Tittmann d, Ivan Rayment b,Susan P. Gilbert c, Andreas Hoenger a,⇑a Department of Molecular, Cellular, and Developmental Biology, University of Colorado, Boulder, CO 80309-0347, USAb Department of Biochemistry, University of Wisconsin, Madison, WI 53706, USAc Department of Biology and the Center for Biotechnology & Interdisciplinary Studies, Rensselaer Polytechnic Institute, Troy, NY 12180, USAd EMEZ, Swiss Federal Institute of Technology, Hoenggerberg, 8093 Zuerich, Switzerland

a r t i c l e i n f o

Article history:Received 18 December 2012Received in revised form 19 September2013Accepted 25 September 2013Available online 4 October 2013

Keywords:Kinesin-14Kar3Cik1MicrotubulesHelical 3-D analysisMolecular dockingCryo-electron microscopy

a b s t r a c t

Kar3Cik1 is a heterodimeric kinesin-14 from Saccharomyces cerevisiae involved in spindle formationduring mitosis and karyogamy in mating cells. Kar3 represents a canonical kinesin motor domain thatinteracts with microtubules under the control of ATP-hydrolysis. In vivo, the localization and functionof Kar3 is differentially regulated by its interacting stoichiometrically with either Cik1 or Vik1, twoclosely related motor homology domains that lack the nucleotide-binding site. Indeed, Vik1 structurallyresembles the core of a kinesin head. Despite being closely related, Kar3Cik1 and Kar3Vik1 are eachresponsible for a distinct set of functions in vivo and also display different biochemical behaviorin vitro. To determine a structural basis for their distinct functional abilities, we used cryo-electronmicroscopy and helical reconstruction to investigate the 3-D structure of Kar3Cik1 complexed to micro-tubules in various nucleotide states and compared our 3-D data of Kar3Cik1 with that of Kar3Vik1 and thehomodimeric kinesin-14 Ncd from Drosophila melanogaster. Due to the lack of an X-ray crystal structureof the Cik1 motor homology domain, we predicted the structure of this Cik1 domain based on sequencesimilarity to its relatives Vik1, Kar3 and Ncd. By molecular docking into our 3-D maps, we produced adetailed near-atomic model of Kar3Cik1 complexed to microtubules in two distinct nucleotide states, anucleotide-free state and an ATP-bound state. Our data show that despite their functional differences,heterodimeric Kar3Cik1 and Kar3Vik1 and homodimeric Ncd, all share striking structural similaritiesat distinct nucleotide states indicating a common mechanistic theme within the kinesin-14 family.

� 2013 Elsevier Inc. All rights reserved.

1. Introduction

Microtubule-based molecular motors of the kinesin superfamilyare classified into fourteen different classes (Lawrence et al., 2004).Here we focus on the members of the kinesin-14 family, which,according to current knowledge are the only kinesins that areminus-end directed and form their motor domain at the C-terminalend of the polypeptide chain (Ncd: McDonald and Goldstein, 1990;Kar3Vik1: Endow et al., 1994; XCTK2: Walczak et al., 1997; HSET:Mountain et al., 1999). All of the kinesin-14’s that have been stud-ied so far were found to be non-processive retrograde directed mo-tors (Walker et al., 1990; Case et al., 1997; de Castro et al., 2000;Foster and Gilbert, 2000; Fink et al., 2009). Kinesin-14s contributeto an inward directed force towards the minus-end of the microtu-bules (Saunders and Hoyt, 1992; Sharp et al., 1999). The mostextensively studied kinesin-14 is Ncd from Drosophila melanogaster(McDonald et al., 1990; Endow et al., 1990, for structure–function

investigations see: Hoenger et al., 1995; Sosa et al., 1997; Sablinet al., 1998; Wendt et al., 2002; Endres et al., 2006; Fink et al.,2009). Ncd is required for proper chromosome distribution in mei-osis and early mitosis and has been implicated in having roles inbipolar spindle assembly and its maintenance as well as spindlepole formation (Endow et al., 1994; Oladipo et al., 2007). Ncd usesits C-terminal motor domains and N-terminal non-motor microtu-bule binding domains (Karabay and Walker, 1999; Wendt et al.,2003) to crosslink antiparallel microtubules within the spindlemidzone and parallel microtubules at the spindle poles.

Of the six kinesins found in Saccharomyces cerevisiae, one ofthem, Kar3, belongs to the kinesin-14 family. Kar3, like all otherkinesin-14s contains a C-terminal motor-head domain. In combi-nation with an N-terminal non-motor microtubule binding site itcrosslinks microtubules during karyogamy and mitosis (Meluhand Rose, 1990; Rose, 1996). Kar3 is essential for yeast nuclear fu-sion during mating, but it is apparently not essential for mitoticgrowth (Meluh and Rose, 1990; reviewed in: Winey and Bloom,2012). The structure of the monomeric Kar3 motor domain hasbeen solved to atomic detail by X-ray crystallography (Gulick

336 M.A. Gonzalez et al. / Journal of Structural Biology 184 (2013) 335–344

et al., 1998) and shares most structural elements with other kine-sin motor domains.

Kar3 function in vivo is regulated by its capacity to selectivelyheterodimerize with either of two motor-homology domainscalled Vik1 and Cik1 (Manning et al., 1999). While Vik1 and Cik1have a relatively high level of sequence similarity (24% identity,37% homology: Manning et al., 1999), their localization duringthe yeast life cycle and genetic analysis demonstrated that Kar3-Vik1 and Kar3Cik1 exhibit remarkably different properties andare responsible for carrying out distinct sets of functions (Manninget al., 1999). Kar3Vik1 is only expressed during mitosis where itlocalizes predominantly to the poles of the mitotic spindle andlikely contributes to spindle stabilization by crosslinking andfocusing the minus-ends of parallel microtubules as proposed forother kinein-14 motors (Manning et al., 1999; Allingham et al.,2007). An EM structure (Cope et al., 2010, 2013), as well as anX-ray crystal structure (Rank et al., 2012) of a truncated heterodi-meric Kar3Vik1 were published recently. In contrast to Kar3Vik1,Kar3Cik1 crosslinks anti-parallel interpolar MTs in the overlapzone, providing stabilization and controlling the spindle mid-region geometry (Gardner et al., 2008). Upon pheromone treatment,Kar3Cik1 localizes to cytoplasmic microtubules (Page et al., 1994)to facilitate karyogamy by shortening microtubules, pulling thetwo nuclei together for nuclear fusion (Page and Snyder, 1992;Page et al., 1994; Maddox et al., 2003). In support of this, Kar3Cik1has been shown to have robust depolymerizing activity in vitropredominantly from microtubule plus-ends (Sproul et al., 2005).Conversely, Kar3Vik1 only gradually and non-specifically depoly-merizes MTs from both the plus and minus ends in vitro, similarto the weak depolymerizing activity reported for Ncd (Sproulet al., 2005; Allingham et al., 2007). Though Cik1 is not essentialfor mitotic growth (Page et al., 1994; Manning et al., 1999), Kar3-Cik1 localizes predominantly to the interpolar MTs at the midzoneof the mitotic spindle where it is thought to crosslink antiparallelmicrotubules and provide an inward force during spindle assembly(Meluh and Rose, 1990; Page et al., 1994; Gardner et al., 2008).

The functional differences between Kar3Vik1 and Kar3Cik1 de-spite being closely related, prompted us to investigate the microtu-bule-bound structure of Kar3Cik1, particularly in comparison withKar3Vik1 in an attempt to obtain insight into how these two com-plexes are uniquely suited to carry out their respective functions.Here we have used electron microscopy to study the conformationof a heterodimeric Kar3Cik1 motor domain construct in the pres-ence of ADP and AMPPNP (mimicking an ATP state), as well as afterapyrase treatment to generate the nucleotide-free state. We alsoreport an intermediate resolution 3-D structure of Kar3Cik1 com-plexed to microtubules in the nucleotide-free and ATP states. Our3-D structures were obtained by cryo-electron microscopy (cryo-EM) and 3-D image reconstruction based on the helical symmetryof microtubules composed of 15 protofilaments (Arnal et al., 1996;reviewed in Hoenger and Gross, 2008). Our results show that Kar3-Cik1’s coiled-coil stalk rotates�65� upon uptake of AMPPNP repre-senting the powerstroke that Kar3Cik1 uses to move toward themicrotubule minus-end. By comparing the structures of Kar3Cik1,Kar3Vik1, and homodimeric Ncd (Hirose et al., 1996; Sosa et al.,1997; Wendt et al., 2002) we found that all these kinesin-14 mo-tors, whether heterodimeric or homodimeric in nature, show al-most identical 3-D microtubule-bound conformations in both thenucleotide-free and ATP-states. Hence, despite some differencesbetween the densities of the Vik1 and Cik1 domains, both com-plexes show a common theme for a powerstroke mechanism,highly reminiscent of that described for Ncd (Wendt et al., 2002;Endres et al., 2006). However, while movements of the structuralelements of Kar3Cik1 and Kar3Vik1 are similar, here we also pro-vide evidence that their microtubule-binding patterns differ in thatKar3Vik1 binds to microtubules in a highly cooperative manner

while Kar3Cik1 binds stochastically. While the basis for coopera-tive binding by Kar3Vik1 is not yet known, this difference may par-tially explain how Kar3Cik1 has adapted to its role as amicrotubule depolymerizing factor where other motors in closeproximity may be a hindrance for MT shortening while Kar3Vik1works in concert by recruiting additional motors to the spindlepoles to facilitate spindle assembly and stabilization.

2. Results

2.1. High-affinity Kar3Cik1 motor-microtubule binding conformations

We have analyzed the 3-D microtubule-binding conformationof Kar3Cik1 by cryo-EM, helical 3-D image reconstruction, andmolecular docking (see Figs. 1, 2 and 4). Micrographs of frozen-hydrated motor-microtubule complexes were screened for fullydecorated 15 protofilament microtubules that were then recon-structed in 3-D by the PHOELIX/SUPRIM image-processing package(Whittaker et al., 1995; Schroeter and Bretaudiere, 1996). Polymer-ized, taxol-stabilized microtubules were mixed at a molar ratio of1:6 ab-tubulin to Kar3Cik1 heterodimers and incubated witheither ADP (see Fig. 3), apyrase to generate the nucleotide-freestate, or AMPPNP, a non-hydrolyzable ATP analog to generate anATP-bound state (Fig. 1A and B). Excess ADP (1 mM) generates aweak microtubule affinity in Kar3Cik1, which at our experimentalconditions produced only sparsely decorated microtubules unsuit-able for helical 3-D analysis. Instead, the ADP state was furtheranalyzed by high-resolution surface metal shadowing (see Fig. 3)discussed below. All other nucleotide states produced high-affinitybinding. Nucleotide-free conditions, or the presence of excessAMPPNP (1 mM) revealed fully decorated microtubules with abinding stoichiometry of one Kar3Cik1 heterodimer on each tubu-lin dimer. The degree of decoration was estimated from diffractionpatterns and the power of the characteristic 8 nm and 4 nm layer-lines (Fig. 1C–E). On raw-data images as shown in Fig. 1A and B,structural differences between the two nucleotide states are essen-tially invisible due to the inherently low signal to noise ratio ofcryo-micrographs (Fig. 1A and B), though the micrographs can beenhanced with Fourier-filtering to reveal distinct conformationalchanges (left panels in Fig. 1C–E). After 3-D reconstruction andaveraging, density differences appear clearly on 5 nm thick slicesthrough the reconstructed 3-D density maps (see Fig. 1G and H,and Fig. 6) as well as on surface rendered volumes (iso-surfacemaps: Figs. 2, 4 and 7).

The iso-surface representations of our 3-D reconstructions en-close roughly 90–95% of the actual protein mass, to cancel out ex-cess noise-related densities. Iso-surfaces do not reveal any innerdensity variations as shown in cross-sections through the maps(Figs. 1F–H and 6), but they provide a comprehensive overviewof the entire structure and nicely illustrate large conformationalchanges. In both, the nucleotide-free and AMPPNP states we findone domain of the Kar3Cik1 heterodimer to contact the microtu-bule surface, while the second domain protrudes outwards withouttouching the microtubule surface. The shapes of the domains andtheir 3-D configuration are very similar to Kar3Vik1 (inner densityin Fig. 7) and only the plus-end oriented region of the Cik1 densityshows some variation to Vik1 (yellow volume in Fig. 7). In addition,the predicted high-resolution structure of the Cik1 motor homol-ogy domain (Fig. 4E) strongly resembles that of Vik1 (Fig. 4D;Allingham et al., 2007; Rank et al., 2012) while both are quitedifferent from the Kar3 motor domain (Fig. 4C; Gulick et al.,1998; Rank et al., 2012). In our previous Kar3Vik1-microtubulereconstructions the two domains were unambiguously identifiedby site-directed maleimide-gold labeling (Cope et al., 2013). Hence,based on structural similarities, especially in the Kar3 domain we

Fig. 1. 3-D helical reconstruction of decorated and undecorated microtubules. Microtubule/Kar3Cik1 solutions were incubated with apyrase (A) or AMPPNP (B) to generatethe nucleotide-free and ATP states respectively. In both states, Kar3Cik1 heterodimers can be seen decorating the microtubules although structural differences between thestates cannot be detected. (C) 3D helical reconstruction of a plain microtubule. The same result was obtained by incubating Kar3Cik1 motors with ADP prior to addition tomicrotubules, but where no decoration was detected. A Fourier transform reveals only the 4 nm layer line, which corresponds to the tubulin repeats, illustrating only sparsedecoration with motor complexes. (D) Reconstruction of a Kar3Cik1 decorated microtubule in the nucleotide-free state. Two distinct densities are visible at each motorsubunit corresponding to Kar3 and Cik1. The Fourier transform reveals a strong 4 nm layer line (tubulin repeats) as well as an 8 nm layer line that corresponds to the motorrepeat every 8 nm along the microtubule. (E) Reconstruction of the ATP state using the non-hydrolyzable ATP analog AMPPNP. The microtubule is fully decorated withKar3Cik1 as in (D) illustrated by the strong 4 and 8 nm layer lines. Corresponding cross-sectional views of the reconstructed microtubules are shown in (F) undecoratedmicrotubule, (G) nucleotide-free state, and (H) AMPPNP state. On the right half of each image, contour lines are shown depicting densities observed from tubulin andKar3Cik1. Circles corresponding to the position of tubulin (blue), Kar3 (purple), and Cik1 (orange) are shown as well as red circles identifying differences in density betweenthe nucleotide-free and AMPPNP states.

M.A. Gonzalez et al. / Journal of Structural Biology 184 (2013) 335–344 337

can safely state that analogous to Kar3Vik1 it is again the Kar3 mo-tor domain of Kar3Cik1 that maintains contact with the microtu-bule surface while the Cik1 motor homology domain protrudesoutwards (Fig. 1G and H, Figs. 2 and 4).

Reminiscent of dimeric Ncd (Wendt et al., 2002; Endres et al.,2006) and Kar3Vik1 (Fig. 7; Rank et al., 2012; Cope et al., 2013),a most striking conformational change occurs upon AMPPNP up-take into the empty nucleotide pocket of the Kar3 motor domain.During the transition from the nucleotide-free to AMPPNP-boundstate, the overall microtubule-binding geometry of Kar3 remainsunchanged. However, the Cik1 head domain rotates substantially(Fig. 2B and C). This rotation consists of a 65-degree swing of Kar3-Cik1’s truncated heterodimeric stalk (Fig. 2B, C, E and F) that is heldtogether by an engineered GCN4 dimerization motif (see: Ranket al., 2012). The stalk region that stabilizes the heterodimer canbe seen as a relatively weak, but nevertheless well defined elon-gated density which is pointing toward the plus end of the micro-tubule. The cross-sectional views shown in Figs. 1 and 6 allow us tostudy in great detail the inner density distributions in the 3-D dataand how Kar3Cik1 interacts with the tubulin subunits of the micro-tubule. The binding geometry of Kar3 motor head to tubulin is verysimilar to monomeric kinesins constructs studied with this method(e.g. see Sosa et al., 1997 (Ncd) Hoenger et al., 2000 (kinesin-1);Hirose et al., 2006 (Kar3)). The Cik1 motor homology domain

adopts a very similar conformation as the second motor heads indimeric Ncd constructs (Sosa et al., 1997; Wendt et al., 2002; End-res et al., 2006) and Kar3Vik1 constructs (Cope et al., 2010, 2013;Rank et al., 2012). It is tethered to the Kar3 motor domain, pro-trudes outwards and does not touch the microtubule surface.

2.2. Low-affinity Kar3Cik1-microtubule binding patterns

As for most kinesin motor domains, ADP-bound Kar3 motor do-mains maintain a low microtubule-affinity. Complete microtubuledecoration would be required for a quantitative analysis by helicalaveraging but this is often very difficult to achieve. 3-D image anal-ysis of frozen-hydrated complexes are usually not conclusive andtomographic reconstructions, unless volume averaging could beapplied are typically too noisy (see Cope et al., 2010) for quantita-tive image analysis. Nevertheless, ADP reveals sparse decoration ofKar3Cik1 complexes that is easily detectable with high-resolution,unidirectional surface metal shadowing (Fig. 3A). The power of thismethod is its focus on surface structures through an almost com-plete elimination of inner density contributions, which dominatecryo-EM projections (e.g. see Fig. 1A and B, and Fig. 6). Fig. 3 com-pares the surface-binding patterns of ADP–Kar3Cik1 (Fig. 3A) withthat of ADP-Ncd (Fig. 3B), which both show striking similarities toADP-Kar3Vik1 (Rank et al., 2012). In all three cases it appears as if

Fig. 2. Isosurface rendering of density maps obtained by helical reconstruction. (A) Isosurface of a naked 15-protofilament microtubule showing the orientation of tubulin(turquoise). (B) The nucleotide-free Kar3Cik1 is in complex with the microtubule. It appears that Kar3 (grey/black rings) is in contact with the microtubule while Cik1(orange/red rings) is oriented away from the microtubule. The coiled coil stalk (light blue/blue rings) connecting Kar3 and Cik1 is pointed toward the microtubule plus end.(C) In the ATP state Kar3Cik1 is also in complex with the microtubule with Kar3 in contact with the microtubule and Cik1 oriented away. Interestingly, there is a change inKar3Cik1’s conformation showing a �65� rotation of the stalk that causes it to point toward the minus end of the microtubule (black arrows). Other subtle differences can beseen when comparing the orange globular region corresponding to Cik1 (red rings). A cross-sectional view of (D) a naked microtubule, (E) the nucleotide-free, and (F) ATPstates shows how Kar3Cik1 (grey and orange) is in contact with tubulin (turquoise).

Fig. 3. High-resolution surface metal shadowing of kinesin motors. Kinesin motors bound to ADP were incubated with microtubules at sub-stoichiometric levels andshadowed with tantalum/tungsten. Dimeric motor constructs are identified by a yellow arrow pairs in (A) Kar3Cik1, (B) dimeric Ncd, and (C) dimeric Eg5 (a kinesin-5). Thekinesin-14 motors appear to bind sparsely along the microtubules and span adjacent protofilaments (red relative to yellow arrows in A and B) while dimeric kinesin-5sappear to bind to adjacent tubulin subunits along the same protofilament (red relative to yellow arrows in C).

338 M.A. Gonzalez et al. / Journal of Structural Biology 184 (2013) 335–344

Fig. 4. Docking of near-atomic structures into cryo-EM derived isosurfaces. (A) In the nucleotide-free state, the structure of tubulin (turquoise, PDB: 1JFF) is shown and theKar3 structure (grey, PDB: 3KAR) is positioned in the globular region in contact with b-tubulin. Kar3’s helix a-4 (red) is at the interface of where Kar3 contacts themicrotubule. As there is currently no known structure for Cik1, a predicted structure was obtained using Phyre 2 (Kelley and Sternberg, 2009). This predicted Cik1 structure(orange) is positioned in the globular region oriented away from the microtubule. The structure of the GCN4 leucine zipper (light blue, PDB: 2ZTA) is docked into the stalkregion. In the nucleotide-free state the stalk is pointing toward the plus end of the microtubule. (B) In the ATP state, there is a conformational change in Kar3Cik1 that leads toa �65� rotation resulting in the stalk pointing toward the minus end of the microtubule. Phyre2 used structural information from (C) Kar3, (D) Vik1, and other proteins togenerate the predicted model of (E) Cik1.

M.A. Gonzalez et al. / Journal of Structural Biology 184 (2013) 335–344 339

all of these kinesin-14 dimers adopt a dimeric binding configura-tion that spans two adjacent protofilaments. While this configura-tion is not always obvious, it becomes clear that this bindingconfiguration is very different from motors such as dimeric kine-sin-1 (Hoenger et al., 2000) or kinesin-5 (Fig. 3C; see Krzysiaket al., 2006) that bind predominantly with both motors to twoadjacent tubulin dimers along the same protofilament (Fig. 3C).Within the same context we observed very little microtubule

Fig. 5. Kar3Vik1/Kar3Cik1 binding cooperativity. Kar3Vik1 appears to bind more cooperacompletely decorated, partially decorated, and undecorated can be seen in the same istochastic pattern, illustrated by scattered motor binding with numerous gaps appearpatterns is most easily visualized along the sidewalls of the microtubule projections (in

binding affinity of both monomeric Cik1 (this work) and Vik1 con-structs (see Cope et al., 2013)

2.3. Prediction of the Cik1 domain fold

Currently there are two independent atomic X-ray structuresavailable for Vik1 from crystals made of a monomeric construct(Allingham et al., 2007) as well as the heterodimeric complex of

tively than Kar3Cik1. (A) Kar3Vik1 exhibits high cooperativity. Microtubules that aremage area (see also: Cope et al., 2010). (B) In contrast, Kar3Cik1 exhibits a moreing in between the motors. The difference in the Kar3Vik1 and Kar3Cik1 bindingsets in A and B).

340 M.A. Gonzalez et al. / Journal of Structural Biology 184 (2013) 335–344

Kar3Vik1 (Rank et al., 2012). However, so far no such structure isavailable for Cik1, though our EM results strongly suggest thatthe overall globular shape is very similar to that of Vik1. To im-prove the accuracy of our Kar3Cik1 dimer-docking attempts(Fig. 4A and B) and to determine whether Cik1 would adopt similarsecondary structure elements as Vik1 (Fig. 4D) we predicted theCik1 structure (Fig. 4E) with the help of the protein structure pre-diction software Phyre2 (Kelley and Sternberg, 2009). This algo-rithm generates a 3-D tertiary structure model based on proteinsequence, fold recognition, and by comparisons to 3-D structuresof similar proteins. In our attempts Vik1 (25% sequence identityto Cik1) (Allingham et al., 2007; PBD: 2O0A) and Ncd (Sablinet al., 1998; PDB: 2NCD) were the most important contributorsto the predicted Cik1 structure. The predicted model (Fig. 4E) iscomposed of residues Asn344–Asp594, which includes part ofCik1’s neck-helix that dimerizes with Kar3 (Fig. 4C). The resultingstructure shows many similarities with Vik1 (Fig. 4D) and someof the folding patterns observed in kinesin motor domains thathave been discussed in detail by Allingham et al. (2007). Theseare in particular the conserved arrangement of three inner b-sheetsand most of the surrounding helices. Overall the predicted Cik1structure and folding pattern strongly resembles that of Vik1,and at a resolution of �2.5 nm as in the EM maps, the differencesare marginal.

2.4. Docking X-ray crystal structures into the 3-D EM scaffold

Despite numerous attempts, so far no kinesin-motor domainhas been crystalized in complex with the ab-tubulin dimer andsolved to near-atomic resolution. Hence, the structure of themotor-tubulin interface has been modeled by docking the atom-ic-resolution X-ray crystal structures of various kinesin motordomains and the electron crystal structures of the ab-tubulindimer (Nogales et al., 1998; Löwe et. al., 2001; PDB: 1JFF) intocryo-EM maps of motor-microtubule complexes (for monomericKar3 see: Hirose et al., 2006; Kar3Vik1: Rank et al., 2012 andCope et al., in press; Ncd: Sosa et al., 1997; Wendt et al.,2002; Endres et al., 2006).

Fig. 4A and B show the docking of X-ray crystal structures ofmonomeric Kar3 (grey; Gulick et al., 1998; PDB: 3KAR) togetherwith our predicted model of Cik1 (orange in Fig. 4A and B; see alsoFig. 4E). In Fig. 4A, the near-atomic resolution structures of ab-tubulin, Kar3, Cik1 and the stalk including the GCN4 leucine zippersequence used to initialize dimerization (O’Shea et al., 1991; PDB:1ZTA; colored light blue in Fig. 4), were docked into the nucleotide-free state. The refined tubulin dimer structure (Löwe et. al., 2001)was docked into the density corresponding to the microtubule.The structure of Kar3 (Gulick et al., 1998) was docked into theglobular region that is in contact with the microtubule (see alsoHirose et al., 2006). Helix a-4 of Kar3 (shown in red) forms partof the motor’s so-called switch-II region (Vale and Milligan,2000) and was positioned according to Hirose et al. (2006) in con-tact with the highly negatively charged C-terminal helix-12 andtail region of b-tubulin.

In both the nucleotide-free and ATP states, the predicted Cik1structure fits the EM structure with remarkable precision(Fig. 4A and B). As discussed above (Fig. 2), there are significantrearrangements from the nucleotide-free configuration that oc-cur upon AMPPNP uptake. Most importantly, the AMPPNP-boundconformation has the stalk pointing toward the minus end of themicrotubule. This result shows that Kar3Cik1 uses a stalk rota-tion for minus-end directed movement whose angle and timingin the nucleotide-hydrolysis cycle strongly resemble the power-stroke shown for Kar3Vik1 (Rank et al., 2012; Cope et al.,2013) as well as that of homodimeric Ncd (Wendt et al., 2002;Endres et al., 2006). Fig. 4A and B shows stereo pairs of the

two distinct conformations that allow the docked structures tobe viewed in 3-D. The powerstroke that occurs between thenucleotide-free and AMPPNP states (Figs. 2 and 4) implies thatthe Cik1 motor homology domain undergoes a �65 degree rota-tion around a pivot point near the C-terminus of the stalk dimer-ization motif, at the end of the neck. The stalk region has torotate as shown to fit into the density map, but also to avoidsteric clashes between the Kar3 and Cik1 domains.

2.5. Structural Comparisons between the nucleotide-dependent 3-Dconformations of Kar3Cik1 and Kar3Vik1

Vik1 and Cik1 are both motor homology domains in S. cerevi-siae that lack nucleotide-binding sites, but serve to regulateKar3’s localization and function by differentially pairing withKar3 to form a functional kinesin-14 complex. Detailed biochem-ical studies have been carried out on both heterodimeric com-plexes (Allingham et al., 2007; Chen et al., 2012 (both:Kar3Vik1); Chen et al., 2011 (Kar3Cik1)), but to date no directstructural comparisons are available. In our hands, the microtu-bule binding properties of both motor complexes with regard tonucleotide state have been rather similar. Both complexes showweak microtubule binding affinity in the presence of ADP, butvery strong binding affinity in the absence of nucleotide and inthe presence of AMPPNP. In both strong binding affinity condi-tions, Kar3Vik1 binds to microtubules with a striking cooperativebehavior (Fig. 5A; Cope et al., 2010), very much like Ncd (seeFig. 2 in Wendt et al., 2002). However, for Kar3Cik1 this cooper-ative binding pattern is much less pronounced (Fig. 5B). Microtu-bule binding is more stochastic and resembles that ofmonomeric Kar3 (Sproul et al., 2005; Allingham et al., 2007;Cope et al., 2013) or other kinesins such as dimeric kinesin-1(Hoenger et al., 2000) or kinesin-5 (Eg5: Krzysiak et al., 2006;see also Fig. 3C). The cooperative binding of Kar3Vik1 is bestseen on projections of the outer microtubule walls, which aresometimes fully decorated and sometimes completely empty (in-set and arrows in Fig. 5A). Some microtubules appear to be fullydecorated while others are almost entirely free of motors. Incontrast, the outer walls of microtubules decorated with Kar3-Cik1 show scattered densities with numerous empty spots be-tween the motors (inset and arrows in Fig. 5B), and decorationis much less clustered than with Kar3Vik1.

A thorough comparison between the reconstructions of Kar3-Cik1 (this paper) and Kar3Vik1 (Cope et al., 2013) in both thenucleotide-free (Figs. 6 and 7A and B) as well as the AMPPNP states(Figs. 6 and 7C and D) revealed many similarities. For the mostpart, their iso-surfaces are almost identical (Fig. 7) featuring onlyminor differences that are best seen on cross-sections throughthe 3-D densities (Fig. 6). Horizontal comparisons marked with col-ored rings highlight the most notable differences between Kar3-Cik1 (Fig. 6A and C) and Kar3Vik1 (Fig. 6B and D). Verticalcomparisons (white rings) show differences within each complexcaused by the nucleotide state. The side-by-side comparison inFig. 6 reveals some positional shifts within the outermost domainsthat are formed by Cik1 and Vik1 respectively (arrows in Fig. 6).The conformational differences between Kar3Vik1 and Kar3Cik1have been analyzed by statistical difference mapping (Fig. 7) basedon a Student’s t-test (Milligan and Flicker, 1987). The wireframevolume in Fig. 7 represents Kar3Vik1, the red/green/blue densitiesare Kar3Cik1, and the solid yellow density marks the location of adifference volume representing a range of 99% significance. Despitethe apparent low resolution of�2.5 nm in these maps the position-ing of mass centers with respect to each other between two alignedand normalized 3-D datasets can be assessed with sub-nanometerprecision, resulting in difference maps as shown in Fig. 7.

Fig. 6. Structural Comparison of Kar3Cik1 and Kar3Vik1. By comparing the densitymap of a single 3.8 nm slice of (A) Kar3Cik1 and (B) Kar3Vik1 in the nucleotide-freestate, a difference in the density of the motors is detected (green circles). Cik1appears to have less density present than Vik1, which may suggest that Cik1 is moremobile than Vik1. The same observation is seen when comparing (C) Kar3Cik1 and(D) Kar3Vik1 in the ATP state. Cik1 again seems to have less density (and thusperhaps more flexibility) than Vik1 (orange circles). White circles show differencesin Kar3Cik1 and Kar3Vik1 between their own nucleotide-free and ATP states. Sincethese 15-protofilament microtubules exhibit a dominant 2-start short-pitched helix(Bessel order-2), a thin slice (�4 nm) of a half cross-section, as shown here,represents the complete density distribution of a single motor from these maps.

M.A. Gonzalez et al. / Journal of Structural Biology 184 (2013) 335–344 341

3. Discussion

Within the kinesin superfamily (for a unified nomenclature see:Lawrence et al., 2004), kinesin-14 members have distinct struc-tural and functional properties that render them quite differentfrom other kinesin families. These unique properties include theirdirection of movement towards the minus-end of microtubules,and their reversed building plan that places the motor domain atthe C-terminal end of the polypeptide chain (Walker et al., 1990;McDonald et al., 1990; Endow et al., 1994; Mountain et al.,1999). In addition, all kinesin-14’s studied thus far are non-proces-sive motors (Case et al., 1997; de Castro et al., 2000; Foster andGilbert, 2000; Fink et al., 2009), and for all of them the cargo-binding site interacts with a second microtubule, but in anon-motor like fashion and without the control of nucleotide state(Karabay and Walker, 1999; Wendt et al., 2003). Hence, unlikekinesin-1 members or other highly processive kinesins,kinesin-14s would be very ineffective motors for long-distancetransport. Instead, kinesin-14’s act in the densely packed microtu-bule arrangement of a bipolar spindle where they maintain thespindle arrangement and dynamics (Fink et al., 2009; reviewedin Peterman and Scholey, 2009).

The close relationship between the budding yeast (S. cerevisiae)Kar3Cik1 and Kar3Vik1 (recently reviewed in: Winey and Bloom,2012), invites a careful exploration into the structural and func-tional differences between these two related heterodimers. Howare the two alternating configurations able to perform their dis-tinct set of functions and how do they compare to homodimerickinesin-14’s such as Ncd? Here we tried to answer these questionsby investigating in detail the microtubule-binding properties ofKar3Cik1 in the presence of ADP, AMPPNP, and the absence ofnucleotide. Our data presented here are discussed with respect torecent structural studies on Kar3Vik1 (Rank et al., 2012; Copeet al., 2013) as well as to previous EM data on homodimeric Ncd(Sosa et al., 1997; Wendt et al., 2002; Endres et al., 2006) a kine-sin-14 from D. melanogaster (McDonald et al., 1990; Walkeret al., 1990).

Excess ADP significantly lowered the microtubule-bindingaffinity of Kar3Cik1 in our experiments, which is reminiscent ofboth Kar3Vik1 (Rank et al., 2012; Cope et al., 2013) and dimericNcd (Sosa et al., 1997; Wendt et al., 2002). However, microtubulebinding increased greatly in the absence of nucleotides or in thepresence of AMPPNP. At high microtubule-affinity conditions theoverall structural configurations of the three kinesin-14’s domainsare very similar (see Figs. 6 and 7, and for Ncd: Hirose et al., 1996;Sosa et al., 1997; Wendt et al., 2002; Endres et al., 2006), suggest-ing a common mechanism for the kinesin-14 powerstroke withinthis kinesin family. Homodimeric Ncd has been shown to bind withonly one of its motor domains to the microtubule surface while thesecond one protrudes outwards. This matches the domain configu-ration of the heterodimeric complexes where the Kar3 motor do-main maintains contact to the microtubule surface while theCik1 (wireframe in Fig. 7) or Vik1 motor homology domains (solidvolume in Fig. 7) extend away from the microtubule. In the case ofthe Ncd homodimer, it has been argued that having two identicalmotor domains increases the chance of finding a microtubule bind-ing site and that prior to binding both motors may have the sameprobability to become the microtubule-binding domain (Wendtet al., 2002; Liu et al., 2012). On the other hand, Foster et al.(2001) reported about asymmetry in dimeric Ncd in solutionwhere one head held ADP tightly and the other weakly, whichwould enhance its probability for microtubule binding over theother one. In any case, upon binding, the motor adopts an asym-metric configuration where the microtubule-bound domain divertsits task to regular motor function, and the tethered domain adopts

a helper function perhaps aiding the stalk rotation. Kar3Cik1 andKar3Vik1, on the other hand, do not have the advantage of a di-meric motor domain, but instead maintain a distinct variabilityin their function by selectively forming heterodimers dependingon the circumstances (reviewed in: Winey and Bloom, 2012). Theinitial process of finding a microtubule-binding site appears to beaided by the motor-homology domains (Allingham et al., 2007;Duan et al., 2012; Chen et al., 2012), but during the powerstrokeKar3 takes over and remains bound to the microtubule whileCik1 or Vik1 rotate together with the motor’s coiled-coil stalk in re-sponse to nucleotide uptake into Kar3’s active site.

Despite their strong structural similarities upon microtubulebinding (see Figs. 6 and 7), there are small but distinct differencesbetween the 3-D maps of microtubule-bound Kar3Cik1 and

Fig. 7. 3-D structural comparison of Kar3Cik1 and Kar3Vik1. The 3-D electron density map of Kar3Cik1 was docked into the isosurface mesh representation of the 3-Delectron density map of Kar3Vik1. Longitudinal, (A) and cross-sectional (B) views of the nucleotide-free state show that Kar3Cik1 docks into the Kar3Vik1 map extremely welldemonstrating that Kar3Cik1 adopts the same pre-powerstroke configuration as Kar3Vik1. Similarly, longitudinal (C) and cross-sectional (D) views of the AMPPNP stateconfirm that Kar3Cik1 fits excellently into the Kar3Vik1 structure showing a nearly identical post-powerstroke position following uptake of ATP. Difference mapping (yellow)reveals locations of density differences between Kar3Vik1 and Kar3Cik1 with a significance of >95%. Interestingly, these results show that overall, Kar3Cik1 and Kar3Vik1appear structurally identical at this resolution, and it seems that they utilize the same mechanism of movement to perform different functions in the cell. Wire mesh:Kar3Vik1; red/green/blue diffuse density: Kar3Cik1; solid yellow density: difference map at >95% significance.

342 M.A. Gonzalez et al. / Journal of Structural Biology 184 (2013) 335–344

Kar3Vik1 for both the nucleotide-free and ATP binding states (seeFig. 6; circles and arrows). Most of these differences are betweenthe tethered Cik1 (Fig. 7A and C) and Vik1 domains (Fig. 7B andD) while the tubulin and Kar3 domains are essentially identical(Figs. 6 and 7). Given their distinct functional capabilities, this re-sult suggests that subtle structural differences between the twocomplexes, such as the increased flexibility seen in Kar3Cik1 overKar3Vik1 (Fig. 6) may still result in crucial differences in motorfunction. Kinesin heads of various families show very little varia-tions between each other. However, ATP binding sometimes in-duces a slight rotation of the bound kinesin head with respect tothe tubulin protofilament (Kikkawa et al., 2001; Skiniotis et al.,2003; Hirose et al., 2006). Therefore, the variations observed be-tween Cik1 and Vik1 may be indicative of their different cellulartasks.

Yet another notable difference between Kar3Cik1 and Kar3Vik1are their microtubule decoration properties in both high-affinitystates (AMPPNP and absence of nucleotides). Kar3Vik1 shows avery distinct, highly cooperative binding mechanism, very similarto Ncd (see Fig. 2 in Wendt et al., 2002). At sub-stoichiometricincubation conditions (Fig. 5) Kar3Vik1 completely fills individualprotofilaments but may leave most of the microtubule surface free(Fig. 5A). Kar3Cik1 complexes, however, bind in a more stochasticfashion, randomly filling the microtubule surface but leavingnumerous gaps in-between (Fig. 5B). The data shown in Fig. 5 havebeen acquired in the presence of AMPPNP but the differences be-tween Kar3Cik1 and Kar3Vik1 with regard to cooperative bindingare also prevalent in the absence of nucleotide. Hence, thoughsmall, structural and functional differences are apparent and areindicative for different tasks: Kar3Cik1 promotes microtubule-depolymerization (Page and Snyder 1992; Sproul et al., 2005),while Kar3Vik1, like Ncd, maintains the mitotic spindle and addsstability to the system, possibly achieved by the cooperative bind-ing property (Manning et al., 1999; reviewed in: Winey and Bloom,2012). Unlike with other microtubule-depolymerizing kinesins wehave never observed any type of kinesin-tubulin protofilamentcurls as previously found with kinesin-13 (see Moores et al.,2006; Mulder et al., 2009). Curls neither appeared with Kar3Cik1,Kar3Vik1 nor Ncd, confirming earlier data that the depolymerizingmechanism of Kar3Cik1 must be different from kinesin-13 (seeSproul et al., 2005).

4. Conclusions

While we have shown that the 3-D structural configurations ofmicrotubule binding and minus-end directed movement are verysimilar between all three kinesin-14s compared here (and very dif-ferent to other kinesins), the key to the specific cellular functions ofKar3Cik1 versus Kar3Vik1 and Ncd, may be influenced by slightconformational variations, and a difference in cooperative binding.Strong cooperative binding may be disruptive to Kar3Cik1’s func-tion as a microtubule-depolymerizing factor, while the enhancedcooperativity observed with Kar3Vik1 and Ncd could benefit theirmodes of action by rapidly recruiting multiple motors to a partic-ular site of interest, thereby facilitating spindle stabilization andcompensating for the inherent lack of processivity, and overcomea situation where an isolated motor complex would be ratherineffective.

Neither this work nor the one on Kar3Vik1 (Cope et al., 2013)revealed any cryo-EM 3-D data that showed stoichiometric bindingof monomeric Vik1 or Cik1 constructs (not in complex with Kar3)to the microtubule at any nucleotide condition, suggesting no tolittle microtubule binding affinity of these domains. Furthermore,only the shadowed images of kinesin-14’s in the presence of ADPindicate a cross-protofilament binding configuration of Kar3Cik1,Ncd (Fig. 3A and B) and Kar3Vik1 (Rank et al., 2012). These resultsare in stark contrast to the microtubule co-sedimentation studiesthat show that microtubule binding by either Cik1 or Vik1 is quitestrong in the presence of ADP based on the assumption that theKar3�ADP head interacts weakly with the microtubule (Allinghamet al., 2007; Duan et al., 2012; Chen et al., 2012). The reason for thisdiscrepancy most certainly lies in the different experimental de-signs which may favor capture of transient intermediates by solu-tion equilibrium binding studies yet not by cryo-EM. The samediscrepancy between EM imaging and kinetic measurements hasalso been found for the ADP-Ncd microtubule complex, despitethe presence of two regular motor domains in this homodimer(Foster et al., 2001; Liu et al., 2012).

These results suggest that the initial microtubule collision byCik1 or Vik1 under some conditions is transient and is followedimmediately by Kar3�ADP microtubule association resulting inADP release and destabilization of the Cik1 or Vik1 microtubuleinteraction (Allingham et al., 2007; Rank et al. 2012; Duan et al.,

M.A. Gonzalez et al. / Journal of Structural Biology 184 (2013) 335–344 343

2012; Chen et al., 2012). This model explains why Cik1 and Vik1,while showing high microtubule binding affinity by them do notstall the motor. These observations emphasize the need for alloste-ric interactions between the motor domain and Cik1 or Vik1 toovercome their lack of an ATP site that would otherwise regulatethe microtubule binding affinity and on–off rate as seen in regularkinesin motor domains.

5. Material and methods

5.1. Microtubule polymerization

Microtubules were polymerized under conditions that favor theformation of 15-protofilament microtubules as described (Beuronand Hoenger, 2001). Briefly: Polymerized microtubules were madein vitro using 45 lM bovine brain tubulin (Cytoskeleton, Inc., Den-ver, CO), BRB80 (80 mM PIPES, pH 6.8, 1 mM MgCl2, 1 mM EGTA),1 mM GTP, 10 lM paclitaxel (Sigma, St. Louis, MO) and 15% (v/v)DMSO that has been found to enhance the fraction of 15-protofil-ament microtubules. They were incubated at 37 �C for 30 minand stored at room temperature overnight which we found to re-duce open tubulin sheets and increase the amount of stable closedtubes.

5.2. Expression and purification of WT GCN4–Kar3Cik1

GCN4–Kar3Cik1 was expressed and purified similarly to previ-ously reported methods (Sproul et al., 2005; Allingham et al.,2007). GCN4–Kar3Cik1 is a truncation of the full-length Kar3Cik1containing residues Lys353–Lys729 of Kar3, and Asn344–Asp594of Cik1. This construct comprises of the complete C-terminal glob-ular domains of Kar3 and Cik1 plus part of native coiled-coil stalkthrough which Kar3 and Cik1 heterodimerize. Furthermore, to ini-tialize the correct hetero-dimerization between Kar3 and Cik1 aGCN4 leucine zipper sequence (O’Shea et al., 1991) was added tothe N-terminus of the truncated Kar3Cik1 construct, as recentlydescribed for Kar3Vik1 (Rank et al., 2012).

5.3. Vitrification of WT GCN4–Kar3Cik1–MT complexes for cryo-EM

Vitrification of protein complexes and data recording by cryo-EM warrants the best possible structure preservation (e.g. seeDubochet, 2007). Cryo-EM GCN4–Kar3Cik1–MT complexes wereassembled directly on holey carbon C-flat grids (Protochips, Inc.,Raleigh, NC). Polymerized MTs were diluted to 2.25 lM withBRB80. 5 ll of diluted MTs were allowed to adsorb to a holey car-bon grid for 30s and excess liquid was blotted away. 13.55 lMGCN4–Kar3Cik1 in ATPase buffer (for nucleotide-free conditions:20 mM HEPES pH 7.2, 5 mM magnesium acetate, 50 mM potassiumacetate, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM DTT; for AMPPNP con-ditions; BRB80) was added to the MTs for approximately 2 min,blotted with a Whatman #1 filter paper and subsequently plungefrozen in liquid ethane at liquid-nitrogen temperature (Dubochetet al., 1988).

5.4. Unidirectional metal shadowing of Kar3Cik1–MT complexes

Kar3Cik1 at 4.5 lM in ATPase buffer and 1 mM ADP was addedto 3.75 lM MTs and incubated for 90–120 s. Kar3Cik1–MT com-plexes were adsorbed to regular 200-mesh copper EM-grids,coated with a single carbon-film and vitrified by plunging the gridsinto liquid ethane. Frozen grids were either stored under liquidnitrogen or subsequently transferred into the so-called Midilabfreeze-drying/shadowing unit, freeze-dried for approximately2 h at –90C and a pressure of �10�7 bar. Then the grids were

unidirectionally shadowed with a �0.3 nm thick layer of tanta-lum/tungsten, transferred under cryo-vacuum conditions directlyinto the microscope and mounted onto a modified Gatan-626cryo-holder (Hoenger et al., 2000).

5.5. ADP state

GCN4–Kar3Cik1 at a final concentration of 13.55 lM in ATPasebuffer, 5% sucrose and 1 mM ADP was incubated at room temper-ature for 10 min. MTs at a final concentration of 2.25 lM wereadded to the GCN4–Kar3Cik1–ADP solution and incubated for an-other 15 min. GCN4–Kar3Cik1–ADP–MT complexes were appliedto a holey carbon grid and vitrified as described above. ADP mo-tor–MT complexes were also prepared for shadowing experiments,described above.

5.6. Nucleotide-free state

Incubation of motors with the ATP/ADP hydrolyzing enzymeapyrase (Sigma, St. Louis, MO) generated the nucleotide-free state.GCN4–Kar3Cik1 was diluted to 13.55 lM with ATPase buffer and 1unit of apyrase. The kinesin-apyrase mixture was incubated on icefor 30–45 min and vitrified for cryo-EM imaging with MT on holeycarbon grids as described above.

5.7. ATP state

To generate a stable ATP state, ATP hydrolysis in GCN4–Kar3-Cik1–MT complexes had to be prevented. This is routinely achievedby substitution of ATP with the non-hydrolyzable ATP analog aden-ylyl imidodiphosphate tetralithium salt (AMPPNP: see Shimizuet al., 1993) (Sigma, St. Louis, MO). GCN4–Kar3Cik1 at a final con-centration 13.55 lM in ATPase buffer was incubated with 2.2 mMAMPPNP on ice for 20–30 min and frozen with MT on holey carbongrids as described above.

5.8. Cryo-EM data collection and data processing

Vitrified samples were transferred to a Gatan-626 cryo-holder(Gatan, Inc, Pleasanton, CA). Cryo-EM data was collected on anFEI Tecnai F20 FEG transmission EM (FEI-Company, Hillsboro, OR,and Eindhoven, The Netherlands) operating at 200 kV. Single frameimages were collected at a nominal magnification of 29,000� and adefocus value of –2.5 lm with an electron dose of 20 e�/Å2. Imageswere recorded without binning on a 4 K � 4 K Gatan Ultrascan 895CCD camera (Gatan, Inc, Pleasanton, CA) which, at 29,000� createda pixel size of 3.8 Å with respect to the specimen. Due to the natureof helical 3-D reconstruction tilting the specimen was not neces-sary. Images were screened for 15-protofilament MTs and pro-cessed with helical 3-D reconstruction using the softwarepackages PHOELIX (Whittaker et al., 1995) and SUPRIM (Schroeterand Bretaudiere, 1996). For the docking attempts shown here wewere using Chimera (Pettersen et al., 2004), and we docked thetubulin high-resolution structure according to Nogales et al.(1999).

References

Allingham, J.S., Sproul, L.R., Rayment, I., Gilbert, S.P., 2007. Vik1 modulatesmicrotubule-Kar3 interactions through a motor domain that lacks an activesite. Cell 128, 1161–1172.

Arnal, I., Metoz, F., DeBonis, S., Wade, R.H., 1996. Three-dimensional structure offunctional motor proteins on microtubules. Curr. Biol. 6, 1265–1270.

Beuron, F., Hoenger, A., 2001. Structural analysis of the microtubule-kinesincomplex by cryo-electron microscopy. Methods Mol. Biol. 164, 235–254.

Case, R.B., Pierce, D.W., Hom-Booher, N., Hart, C.L., Vale, R.D., 1997. The directionalpreference of kinesin motors is specified by an element outside of the motorcatalytic domain. Cell 90, 959–966.

344 M.A. Gonzalez et al. / Journal of Structural Biology 184 (2013) 335–344

Chen, C.J., Rayment, I., Gilbert, S.P., 2011. Kinesin Kar3Cik1 ATPase pathway formicrotubule cross-linking. J. Biol. Chem. 286, 29261–29272.

Chen, C.J., Rayment, I., Gilbert, S.P., 2012. The ATPase pathway that drives theKinesin-14 Kar3Vik1 powerstroke. J. Biol. Chem. 287, 36673–36682.

Cope, J., Gilbert, S.P., Rayment, I., Mastronarde, D., Hoenger, A., 2010. Cryo-electrontomography of microtubule-kinesin motor complexes. J. Struct. Biol. 170, 257–265.

Cope, J., Rank, K.C., Gilbert, S., Rayment, I., Hoenger, A., 2013. Kar3Vik1 uses aminus-end directed powerstroke for movement along microtubules. PLoS One8, e53792.

De Castro, M.J., Fondecave, R.M., Clarke, L.A., Schmidt, C.F., Stewart, R.J., 2000.Working strokes by single molecules of the kinesin-related microtubule motorncd. Nat. Cell Biol. 2, 724–729.

Duan, D., Jia, Z., Joshi, M., Brunton, J., Chan, M., et al., 2012. Neck rotation and neckmimic docking in the noncatalytic Kar3-associated protein Vik1. J. Biol. Chem.287, 40292–40301.

Dubochet, J., Adrian, M., Chang, J.J., Homo, J.C., Lepault, J., et al., 1988. Cryo-electronmicroscopy of vitrified specimens. Q. Rev. Biophys. 21, 129–228.

Dubochet, J., 2007. The physics of rapid cooling and its implications forcryoimmobilization of cells. Methods Cell Biol. 79, 7–21.

Endow, S.A., Henikoff, S., Soler-Niedziela, L., 1990. Mediation of meiotic and earlymitotic chromosome segregation in Drosophila by a protein related to kinesin.Nature 345, 81–83.

Endow, S.A., Kang, S.J., Satterwhite, L.L., Rose, M.D., Skeen, V.P., et al., 1994. YeastKar3 is a minus-end microtubule motor protein that destabilizes microtubulespreferentially at the minus ends. EMBO J. 13, 2708–2713.

Endres, N.F., Yoshioka, C., Milligan, R.A., Vale, R.D., 2006. A lever-arm rotation drivesmotility of the minus-end-directed kinesin Ncd. Nature 439, 875–878.

Fink, G., Hajdo, L., Skowronek, K.J., Reuther, C., Kasprzak, A.A., et al., 2009. Themitotic kinesin-14 Ncd drives directional microtubule-microtubule sliding. Nat.Cell Biol. 11, 717–723.

Foster, K.A., Gilbert, S.P., 2000. Kinetic studies of dimeric Ncd: evidence that Ncd isnot processive. Biochemistry 39, 1784–1791.

Foster, K.A., Mackey, A.T., Gilbert, S.P., 2001. A mechanistic model for Ncddirectionality. J. Biol. Chem. 276, 19259–19266.

Gardner, M.K., Haase, J., Mythreye, K., Molk, J.N., Anderson, M., et al., 2008. Themicrotubule-based motor Kar3 and plus end-binding protein Bim1 providestructural support for the anaphase spindle. J. Cell Biol. 180, 91–100.

Gulick, A.M., Song, H., Endow, S.A., Rayment, I., 1998. X-ray crystal structure of theyeast Kar3 motor domain complexed with Mg.ADP to 2.3 A resolution.Biochemistry 37, 1769–1776.

Hirose, K., Lockhart, A., Cross, R.A., Amos, L.A., 1996. Three-dimensionalcryoelectron microscopy of dimeric kinesin and ncd motor domains onmicrotubules. Proc. Natl. Acad. Sci. USA 93, 9539–9544.

Hirose, K., Akimaru, E., Akiba, T., Endow, S.A., Amos, L.A., 2006. Large conformationalchanges in a kinesin motor catalyzed by interaction with microtubules. Mol.Cell 23, 913–923.

Hoenger, A., Sablin, E.P., Vale, R.D., Fletterick, R.J., Milligan, R.A., 1995. Threedimensional structure of a tubulin–motor protein complex. Nature 376, 271–274.

Hoenger, A., Doerhoefer, M., Woehlke, G., Tittmann, P., Gross, H., et al., 2000. Surfacetopography of microtubule walls decorated with monomeric and dimerickinesin constructs. Biol. Chem. 381, 1001–1011.

Hoenger, A., Gross, H., 2008. Structural investigations into microtubule-MAPcomplexes. Methods Cell Biol. 84, 425–444.

Karabay, A., Walker, R.A., 1999. Identification of microtubule binding sites in theNcd tail domain. Biochemistry 38, 1838–1849.

Kelley, L.A., Sternberg, M.J., 2009. Protein structure prediction on the Web: a casestudy using the Phyre server. Nat. Protoc. 4, 363–371.

Kikkawa, M., Sablin, E.P., Okada, Y., Yajima, H., Fletterick, R.J., et al., 2001. Switch-based mechanism of kinesin motors. Nature 411, 439–445.

Krzysiak, T.C., Wendt, T., Sproul, L.R., Tittmann, P., Gross, H., et al., 2006. A structuralmodel for monastrol inhibition of dimeric kinesin Eg5. EMBO J. 25, 2263–2273.

Lawrence, C.J., Dawe, R.K., Christie, K.R., Cleveland, D.W., Dawson, S.C., et al., 2004. Astandardized kinesin nomenclature. J. Cell Biol. 167, 19–22.

Liu, H.L., Hallen, M.A., Endow, S.A., 2012. Altered nucleotide-microtubule couplingand increased mechanical output by a kinesin mutant. PLoS One. 7 (10), e47148.

Löwe, J., Li, H., Downing, K.H., Nogales, E., 2001. Refined structure of alpha beta-tubulin at 3.5 A resolution. J. Mol. Biol. 313, 1045–1057.

Maddox, P.S., Stemple, J.K., Satterwhite, L., Salmon, E.D., Bloom, K., 2003. The minusend-directed motor Kar3 is required for coupling dynamic microtubule plusends to the cortical shmoo tip in budding yeast. Curr. Biol. 13, 1423–1428.

Manning, B.D., Barrett, J.G., Wallace, J.A., Granok, H., Snyder, M., 1999. Differentialregulation of the Kar3p kinesin-related protein by two associated proteins,Cik1p and Vik1p. J. Cell Biol. 144, 1219–1233.

McDonald, H.B., Goldstein, L.S., 1990. Identification and characterization of a geneencoding a kinesin-like protein in Drosophila. Cell 61, 991–1000.

McDonald, H.B., Stewart, R.J., Goldstein, L.S., 1990. The kinesin-like ncd protein ofDrosophila is a minus end-directed microtubule motor. Cell 63, 1159–1165.

Meluh, P.B., Rose, M.D., 1990. KAR3, a kinesin-related gene required for yeastnuclear fusion. Cell 60, 1029–1041.

Milligan, R.A., Flicker, P.F., 1987. Structural relationships of actin, myosin, andtropomyosin revealed by cryo-electron microscopy. J. Cell Biol. 105, 29–39.

Moores, C.A., Cooper, J., Wagenbach, M., Ovechkina, Y., Wordeman, L., et al., 2006.The role of the kinesin-13 neck in microtubule depolymerization. Cell Cycle 5,1812–1815.

Mountain, V., Simerly, C., Howard, L., Ando, A., Schatten, G., et al., 1999. The kinesin-related protein, HSET, opposes the activity of Eg5 and cross-links microtubulesin the mammalian mitotic spindle. J. Cell Biol. 147, 351–366.

Mulder, A.M., Glavis-Bloom, A., Moores, C.A., Wagenbach, M., Carragher, B., et al.,2009. A new model for binding of kinesin 13 to curved microtubuleprotofilaments. J. Cell Biol. 185, 51–57.

Nogales, E., Wolf, S.G., Downing, K.H., 1998. Structure of the alpha beta tubulindimer by electron crystallography. Nature 391, 199–203.

Nogales, E., Whittaker, M., Milligan, R.A., Downing, K.H., 1999 Jan. High-resolutionmodel of the microtubule. Cell 96, 79–88.

Oladipo, A., Cowan, A., Rodionov, V., 2007. Microtubule motor Ncd induces sliding ofmicrotubules in vivo. Mol. Biol. Cell 18, 3601–3606.

O’Shea, E.K., Klemm, J.D., Kim, P.S., Alber, T., 1991. X-ray structure of the GCN4leucine zipper, a two-stranded, parallel coiled coil. Science 254, 539–544.

Page, B.D., Snyder, M., 1992. CIK1: a developmentally regulated spindle pole body-associated protein important for microtubule functions in Saccharomycescerevisiae. Genes Dev. 6, 1414–1429.

Page, B.D., Satterwhite, L.L., Rose, M.D., Snyder, M., 1994. Localization of the Kar3kinesin heavy chain-related protein requires the Cik1 interacting protein. J. CellBiol. 124, 507–519.

Peterman, E.J., Scholey, J.M., 2009. Mitotic microtubule crosslinkers: insights frommechanistic studies. Curr. Biol. 19, R1089–R1094.

Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S., Greenblatt, D.M., et al., 2004.UCSF Chimera–a visualization system for exploratory research and analysis. J.Comput. Chem. 25, 1605–1612.

Rank, K.C., Chen, J.C., Cope, J., Porche, K., Hoenger, A., et al., 2012. Kar3Vik1, amember of the kinesin-14 superfamily, shows a novel kinesin microtubulebinding patter. J. Cell Biol 197, 957–970.

Rose, M.D., 1996. Nuclear fusion in the yeast Saccharomyces cerevisiae. Annu. Rev.Cell Dev. Biol. 12, 663–695.

Schroeter, J.P., Bretaudiere, J.P., 1996. SUPRIM: easily modified image processingsoftware. J. Struct. Biol. 116, 131–137.

Skiniotis, G., Surrey, T., Altmann, S., Gross, H., Song, Y.H., et al., 2003. Nucleotide-induced conformations in the neck region of dimeric kinesin. EMBO J. 22, 1518–1528.

Sablin, E.P., Case, R.B., Dai, S.C., Hart, C.L., Ruby, A., et al., 1998. Directiondetermination in the minus-end-directed kinesin motor ncd. Nature 395,813–816.

Saunders, W.S., Hoyt, M.A., 1992. Kinesin-related proteins required for structuralintegrity of the mitotic spindle. Cell 70, 451–458.

Sharp, D.J., McDonald, K.L., Brown, H.M., Matthies, H.J., Walczak, C., et al., 1999. Thebipolar kinesin, KLP61F, cross-links microtubules within interpolar microtubulebundles of drosophila embryonic mitotic spindles. J. Cell Biol. 144, 125–138.

Shimizu, T., Toyoshima, Y.Y., Vale, R.D., 1993. Use of ATP analogs in motor assays.Methods Cell Biol. 39, 167–177.

Sosa, H., Dias, D.P., Hoenger, A., Whittaker, A., Wilson-Kubalek, M., 1997. A modelfor the microtubule-Ncd motor protein complex obtained by cryo-electronmicroscopy and image analysis. Cell 90, 217–224.

Sproul, L.R., Anderson, D.J., Mackey, A.T., Saunders, W.S., Gilbert, S.P., 2005. Cik1targets the minus-end kinesin depolymerase kar3 to microtubule plus ends.Curr. Biol. 15, 1420–1427.

Vale, R.D., Milligan, R.A., 2000. The way things move: looking under the hood ofmolecular motor proteins. Science 288, 88–95.

Walker, R.A., Salmon, E.D., Endow, S.A., 1990. The Drosophila claret segregationprotein is a minus-end directed motor molecule. Nature 347, 780–782.

Walczak, C.E., Verma, S., Mitchison, T.J., 1997. XCTK2: a kinesin-related protein thatpromotes mitotic spindle assembly in Xenopus laevis egg extracts. J. Cell Biol.136, 859–870.

Wendt, T.G., Volkmann, N., Goldie, K.N., Müller, J., Mandelkow, E., et al., 2002.Microtubule binding patterns of the reverse kinesin motor Ncd reveal a minus-end directed power stroke. EMBO J. 21, 5969–5978.

Wendt, T., Karabay, A., Walker, R.A., Gross, H., Hoenger, A., 2003. A structuralanalysis of the interaction between Ncd tail and tubulin protofilaments. J. Mol.Biol 333, 541–552.

Whittaker, M., Carragher, B.O., Milligan, R.A., 1995. PHOELIX: a package for semi-automated helical reconstruction. Ultramicroscopy 58, 245–259.

Winey, M., Bloom, K., 2012. Mitotic spindle form and function. Genetics 190, 1197–1224.