A Structural Analysis of the Interaction between ncd Tail and Tubulin Protofilaments

A Structural Analysis of the Interaction between ncdTail and Tubulin Protofilaments

Thomas Wendt1, Arzu Karabay2, Angelika Krebs1, Heinz Gross3

Richard Walker2 and Andreas Hoenger1*

1European Molecular BiologyLaboratory, Meyerhofstrasse 169117 Heidelberg, Germany

2Department of BiologyVirginia Polytechnic Instituteand State UniversityBlacksburg, VA 24061, USA

3Institute of Cell Biology, SwissFederal Institute of Technology8093 Zuerich-HoenggerbergSwitzerland

ncd is a minus-end directed, kinesin-like motor, which binds to micro-tubules with its motor domain and its cargo domain as well. Typical ofretrograde motors, the motor domain of ncd locates to the C-terminalend of the polypeptide chain, and hence, the cargo domain constitutesthe N-terminal region. To date, several studies have investigated the inter-action properties of the motor domain with microtubules, but very fewstructural data are available about the tail itself or its interaction withmicrotubules as cargo. Here, we applied cryo-electron microscopy andhelical 3D image reconstruction to 15 protofilament microtubulesdecorated with an ncd tail fragment (N-terminal residues 83–187, namedNT6). In our study, the ncd tail shows a behaviour resembling filamentousMAPs such as tau protein, exhibiting a highly flexible structure with nolarge globular domains. NT6 binds to four different sites on the outerside of microtubules within the proximity of the kinesin motor-bindingsite. Two of these sites locate within the groove between two neighbouringprotofilaments, and appear as strong binding sites, while the other twosites, located at the outer rim, appear to play a secondary role. In addition,the ncd tail fragment induces the formation of large protofilament sheets,suggesting a tail-induced modification of lateral protofilament contacts.

q 2003 Elsevier Ltd. All rights reserved.

Keywords: kinesin; non-claret disjunctional (ncd); ncd tail domain;microtubules*Corresponding author

Introduction

Microtubules (MTs) interact with molecularmotors of the dynein and kinesin families andmediate cellular processes such as organelle trans-port, chromosome movement, flagellar motion orMT-bundling.1 The molecular architecture ofkinesin motors is composed of a conserved motordomain that contains the ATP-dependent MT-binding site, a central a-helical stalk regionresponsible for dimerization and a tail domainthat is involved in cargo binding and regulation ofmotor-specific functions.2,3 The structures of

kinesin motor domains have been the subject ofnumerous biochemical, biophysical, and structuralinvestigations.3,4 However, the cargo-bindingregions of the various kinesins are less wellexplored. Unlike the conserved motor domains,the cargo-binding domains are highly adapted totheir particular tasks and hence, their structureand functions are very diverse.2,5

Non-claret disjunctional (ncd) is a retrogradekinesin-like motor involved in spindle assemblyand maintenance during meiosis and early mitosisin Drosophila melanogaster.6 – 8 Like all retrogradekinesin-like motors known to date, the ncd-poly-peptide chain contains the cargo-binding tail at itsN-terminal end, and the motor at its C terminus.The motor domain interacts with MTs with a bind-ing geometry very similar to that of conventionalkinesin9,10 but, unlike conventional kinesin, it is arelatively slow, non-processive motor11 that per-forms an asymmetric power stroke.12 ncdinteracts with MTs via its motor domain, and withits tail domain in an ATP-independent manner.13

0022-2836/$ - see front matter q 2003 Elsevier Ltd. All rights reserved.

Present address: A. Karabay, Molecular Biology andGenetics Department, Istanbul Technical University,34469 Maslak-Istanbul, Turkey.

E-mail address of the corresponding author:[email protected]

Abbreviations used: MT, microtubule; cryo-EM, cryo-electron microscopy; ncd, non-claret disjunctional; NT6,ncd tail fragment (residues 83–187).

doi:10.1016/j.jmb.2003.08.051 J. Mol. Biol. (2003) 333, 541–552

Figure 1. (legend opposite).

Accordingly, it is not surprising that ncd wasfound to bundle MTs in vitro.6,14

Here, we present a structural study into the MT-binding properties of the ncd tail. We used a tailconstruct composed of a 105 amino acid residueN-terminal tail fragment13 (NT6: N-terminalresidues 83–187) and applied it to 15 protofilamentMTs, suitable for helical 3D reconstruction (for aprotocol, see Beuron & Hoenger15). From this frag-ment, regions 83–100, and 115–187 are predictedto interact with MTs.13 It has been shown recentlythat NT6 may recognize up to four binding siteson the tubulin dimer16 in the vicinity of acidicclusters at the C terminus and around helix 12.Here, we present a structural investigation intothe MT-binding properties of NT6 by cryo-electronmicroscopy (cryo-EM) and helical 3D reconstruc-tion methods. We observed MT preparations,which often showed a mix of MT bundles andlarge protofilament sheets. These sheets or raftsextend way beyond the width of a regularunfolded tubulin wall.17,18 The distribution ofdensities in our 3D maps that are related to partsof the ncd tail fragment reflect the very flexiblenature of the ncd tail and, hence, while its flexiblenature prevented the reconstruction of the actualshape of NT6, by using statistical differencemapping we confirmed the existence of multiplebinding sites and determined their location on theab-tubulin dimer. The NT6 fragment has twomajor interaction sites that coincide with that ofkinesin motor domains, mostly on b-tubulin, andtwo minor interaction sites at the outer rim of theprotofilament with both a and b-tubulin.

Results

Stabilizing effects of NT6 on microtubulesand protofilaments

Incubation of a solution of pre-formed MTs withthe ncd tail fragment NT6 shifted the equilibriumbetween polymerised and soluble tubulin muchfurther towards polymer formation, indicating astabilizing power of the tail fragment on tubulinpolymers. Two minutes after mixing NT6 with theMT solution at a molar ratio of 7:1 (NT6:ab-tubulindimers) large amounts of protofilament rafts andsmall sheets appear (Figure 1A (negative staining)and C (cryo-EM)). After prolonged incubation, thenumber of stable MTs increased drastically, andthey form large bundles (Figures 1B (negative

staining, 15 minutes) and 2 (freeze-drying/surface-metal shadowing)).14 MT stabilization increasesdramatically at very low concentrations of NT6,indicating that NT6 indeed interacts with them.However, at substoichiometric concentrations, theouter and inner surface of MT walls appear mostlyunchanged as determined with freeze-drying andmetal-shadowed preparations (Figure 2A). At themolar ratio of NT6 to tubulin of 7:1 the protofila-ment structure disappears gradually (Figure 2B).At such saturating conditions the stoichiometry ofbound NT6 to tubulin dimer was in the range offour to five molecules per tubulin dimer (based onmicrotubule pelleting assays/SDS-PAGE; data notshown).

The most striking evidence for microtubulestabilization and bundle-forming activity of NT6was found in frozen-hydrated preparations(Figure 1C), which preserves the specimen undermost physiological conditions. After absorption tothe EM grid, MTs were incubated with the NT6fragment at a molar ratio of 7:1 for about twominutes and then quick-frozen in liquid ethane.The cryo-EM micrograph in Figure 1C illustratesnicely the magnitude of protofilament rafts thatmay form under these conditions and that oftengrow much wider than a single unfolded MT wall(see also Figure 2B). Sheets of that magnitudecannot be observed in preparations of plain MTsdue to their intrinsic instability.19,20 Some areas areordered enough to reveal diffraction spots, corre-sponding to a lateral spacing of parallel proto-filaments of 5.4(^0.14) nm (boxed area) like inregular tubulin sheets.17,18

The 3D reconstruction of microtubulesdecorated with NT6

The 3D reconstructions were performed withfrozen-hydrated NT6-decorated MTs composed of15 protofilaments.15 The ncd tail fragment wasincubated for two minutes with preformed MTs ata molar ratio of 7:1 (NT6 to ab-tubulin dimers)and quick-frozen in liquid ethane. Longerincubation times were avoided because of bundleformation as described above. As a control mapwe calculated a 3D reconstruction of plain MTsfrom 20 datasets, merging with an average phaseresidual of 17.948. The MT–NT6 reconstructionwas calculated from 22 datasets merging with anaveraged phase residual of 21.598. Phase–ampli-tude plots of the strongest layer-lines are displayed

Figure 1. MT-bundling and stabilization effects of NT6. A, Negatively stained images of MTs after two minutes ofincubation with NT6 show mostly parallel arrays of tubulin sheets composed of few protofilaments. B, After prolongedincubation (,15 minutes) parallel MT arrays form. C, Cryo-electron microscopy of NT6–MT complexes after abouttwo minutes of incubation. Large rafts of parallel protofilaments appear that are much wider than just opened MTwalls. The protofilaments in these rafts show the same lateral spacing as found in tubulin sheets, indicating that NT6stabilizes axial and lateral protofilament contacts. The insets show a diffraction pattern and a Fourier-filtered imageof the marked area. The scale bars represent 100 nm.

Interaction of ncd Tail with Microtubules 543


in Figure 3 and illustrate the quality of the datasets,which reach a resolution of approximately 2.5 nm.For comparison, layer-line data from averagedMT–kinesin motor head complexes are shown inFigure 3 on the right. Layer-line 17 (Bessel order22) corresponds to the axial 8 nm repeat of eachab-tubulin dimer, and layer-line 34 (Bessel order24) represents the 4 nm a–b–a–b repeat. Layer-line 1 (Bessel order 15) corresponds to the axialsupertwist of the entire lattice, and layer-line 35(Bessel order 11) represents a convolution of thissupertwist with the axial 4 nm repeat. The moststriking differences are found within the equator,and in particular on layer-line 17. The amplitudeof this layer-line is very strong on kinesin-decorated MTs (Figure 3 right) due to the definedmass of one kinesin motor head per ab-tubulindimer. On non-decorated MTs, however, the ampli-tude of layer-line 17 is almost invisible due to thestrong structural similarity of a- and b-tubulin(Figure 3, left) which makes no clear 8 nm feature.On NT6-decorated MTs, this layer-line showsgood phases, but its amplitude is still rather weak,indicating the highly flexible nature of NT6 andits more random MT-binding properties comparedto a kinesin motor domain. Nevertheless, mostparts of the mass attributable to NT6 locates onMTs with an 8 nm repeat, interacting mostly withb-tubulin, but some binding to both tubulinmonomers occurs (Figures 4 and 5).

Figure 4 shows a composition of (infinitely thin)cross-sections perpendicular to the MT axis(Figure 4A, B, D and E) and projections summingthe densities along the axis of a 40 nm long tube(Figure 4C and F), corresponding to a stretch ofabout five ab-tubulin dimers. In both images, theminus end is towards the observer. Most of thedensity attributable to NT6 appears as a con-tinuous low-density shell of protein above the MTsurface, clearly visible in sections (Figure 4D andE; note the increase of green and yellow areaswhen compared to Figure 4A) as well as in axialprojections (compare Figure 4F with C). Thedensity levels marked with green and yellow inFigure 4A and B, and D and E extend radiallymuch further out on NT6-decorated maps. Thislow-density coat around the microtubule reflectsthe flexible nature of most parts of the NT6 frag-ment and illustrates the randomly distributedmass of NT6 around the tube.

Despite the flexible appearance of NT6 on theMT surface described above, some parts of NT6interact with MTs at several distinct sites on theab-tubulin dimer, numbered 1–4 in Figures 4 and5. The strongest interaction sites were identified

by Student’s t-test analysis, comparing 3D mapsof NT6-decorated MTs with that of native ones.This allowed locating areas within the averagedmaps, which are clearly different betweenaverages. Areas that differ between the maps witha statistical certainty of .99% are marked in redin Figures 4 and 5. They locate in the grooveformed by adjacent protofilaments (Figures 4Gand 5A, C and D, marked as position 1 and 2) andcoincide with the kinesin-binding site, mostly onb-tubulin (Figure 5B and D). Cross-sections(Figure 4A and B, and D and E) reveal that theareas with the most significant density differences(Figure 4D, red contour lines) locate to the leftside of b-tubulin (when viewed from the minus-end) right there where undecorated maps exhibita very low-density area, which fills with massupon decoration. Other differences are visible butstatistically less significant due to a higher variancein these regions.

In addition to the strong interaction sites labelled1 and 2, two additional sites (numbered 3 and 4 inFigures 4 and 5), appear on NT6-decorated maps.These sites are clearly visible as protrusions fromthe outer protofilament rim, repeating approxi-mately every 4 nm (Figure 4, red asterisk, Figure 5,yellow rings). In the absence of NT6, the outer sur-face of protofilaments is quite smooth (Figures 4Aand C, and 5B) The possible reasons why bindingsites 3 and 4 do not appear at a confidence level of.99% in t-tests are discussed below.

Discussion

ncd binds to MTs with its motor domain and, inan ATP-independent manner, through its taildomain.13 In this way ncd organizes MTs duringspindle formation in meiosis and early mitosis.6 – 8

Unlike its globular motor domain,12,21,22 most partsof the tail domain consist of random coil. Here,we investigated the MT-binding properties of atail fragment named NT6 corresponding to theN-terminal residues 83–187.13

The full ncd molecule and even the isolated NT6fragment alone show a strong MT and proto-filament bundling activity (see Figures 1 and 2).14

This could either be achieved by binding acrossprotofilaments and entire MTs through its twopotential binding sites mapped on the fragment(residues 83–100 and 115–187),13,14 or by changingthe surface charge distribution in a way thatfavours lateral contacts. However, the bundlingeffects observed with NT6 seem to follow twodifferent patterns, and their mechanisms may not

Figure 2. Unidirectional surface shadowing with Ta/W reveals the surface features of NT6–MT complexes. A, Thepresence of NT6 is confirmed through the obvious stabilization of MTs and tubulin sheets, but the outer surfacefeatures at substoichiometric concentrations are similar to an undecorated microtubule. The molecule itself becomesvisible only through the loss of the outer microtubule surface features at higher concentrations (B). The scale barrepresents 100 nm. Arrows indicate the shadowing direction.


Figure 3. Phase–amplitude plots of selected layer-lines from the 3D reconstructions of undecorated MTs, NT6-decorated, and kinesin-decorated MTs. Continuous lines areamplitudes, dotted lines show phases. Layer-line 17 (Bessel order 22) represents the 8 nm tubulin dimer repeat and layer-line 34 represents the 4 nm a–b–a–b repeat. Layer-line 1 (Bessel order 15) corresponds to the axial protofilament supertwist. The equators are shown at the bottom.

Figure 4. Cross-sections and axialprojections from cryo-EM data andhelical 3D reconstructions revealthe additional mass of NT6 inNT6–MT complexes. A, B, D and E,Axial cross-section through a mapfrom undecorated microtubules(A and B) and NT6–MT complexes(D and E). Red contour lines in Dmark the location of mass differ-ences related to the presence ofNT6 with a significance of .99%.This location marks binding sites 1and 2 (G, see also Figure 5), whichshow a strong affinity for NT6. Theweaker sites 3 and 4 are markedwith a red asterisk. The mostobvious increases in density aremarked in green and yellow, andisolated for clarity in B and E. Cand F, Axial projections along a40 nm long tube of (C) a nativemicrotubule or (F) NT6–MT com-plexes. NT6 forms a continuouslayer of low density around thetube, indicating the flexible natureof the NT6 fragment. Our datasuggest that NT6, which containstwo potential MT-binding regions,may well be capable of bindingacross adjacent protofilaments asoutlined in G. Red lines sketch therandom distribution of NT6molecules. The microtubule minus-end is towards the observer.



548 Interaction of ncd Tail with Microtubules

be related: one pattern is bundling of intact MTs(Figure 1B),14 and the second effect is the formationof protofilament rafts or sheets, shown in Figure 1Cand D. MT bundling appears to involve mainly theoutermost protofilament rim (e.g. the tubulin Ctermini), while sheet formation may be caused bya modification of lateral tubulin contacts throughthe presence of NT6 in the lower areas of thegroove between protofilaments (see Figures 4Dand 5A). The protofilament rafts observed inFigure 1C exhibit a regular 5.4 nm lateral spacingbetween parallel protofilaments. This is identicalwith the spacing found in MTs or opened MTwalls,17 and indicates that the lateral protofilamentcontacts are still intact. This indicates that NT6stabilizes MTs and protofilaments axially andlaterally. This resembles our observations on tauprotein (results submitted elsewhere) but differsfrom the effects of kinesin motor domains, whichappear to stabilize mainly the protofilament axis,23

but not lateral contacts.Sequence analysis of the 105 residue NT6 tail

fragment predicted mostly random coil. Secondarystructure predictions using circular dichroismspectroscopy showed a content of about 30%a-helices and 14% b-sheet (data not shown),indicating a highly flexible structure. Hence, themethods used for electron microscopy 3D recon-struction of macromolecular assemblies arereaching their limits in this case, and the resultshave to be interpreted accordingly. Helical 3Dimage analysis averages image elements accordingto a given helical symmetry (that of the underlyingtubulin structure in this case) and hence, 3D mapsof complexes between NT6 and MTs cannot revealthe full shape of NT6, which is anyway most likelynot defined. Unfortunately, today there is still nomethod that would allow reconstructing thesekinds of small, flexible structures without theneed for image averaging. Currently, the resolutionof cryo-tomography is not good enough to visual-ize these dynamic complexes. Hence, although the3D map presented here represents an averageover many different NT6 configurations andlocations, the resulting 3D reconstructions allowmapping qualitatively the main MT-binding sites

and reveal a distribution range of unbound partsof NT6 on the outer surface of the MT (seeFigure 4, green and yellow areas in A, B, D and E).

It has been shown that NT6 contains two poten-tial sites that may bind to MTs,13,14 and these twoNT6 domains may bind to up to four differentsites on the ab-tubulin dimer.16 Here, we couldindeed identify four different tubulin bindingsites, which, however, seem to interact with NT6with either various affinities or structural vari-ability (see Figures 4 and 5). Here, we labelledthese sites 1–4 (Figures 4G, and 5A, C and D). Thedistribution of these sites is quite different fromtau,24,25 which appears to recognize mostly theouter rim of tubulin protofilaments, but not thesurface of the inner grooves between proto-filaments as shown here for NT6 (Figures 4 and 5).Binding sites 1 and 2 coincide with the bindingsite of kinesin (and ncd) motor domains, and maytake advantage of the same acidic environmentaround helix 12 of b-tubulin. Both sites appear tomediate a relatively strong interaction. Althoughthe NT6-related mass that locates at binding sites1 and 2 may be less obvious on the 3D-maps thansites 3 and 4 (Figure 5A, yellow rings), cross-sections clearly reveal a strong increase of mass atthe locations shown in Figures 4D, and 5A and B(red contours, 99% significance; compare alsoFigure 4B with D). Binding sites 3 and 4 locate tothe outer rim of the protofilaments and maypossibly interact with the C-terminal ends of botha and b-tubulin. We conclude from the continuouslow-density shell formed by NT6 around the entireMT (compare Figure 4D–F to A–C)) that at leastsome of the molecules bind to two sites acrossprotofilaments by taking advantage of their twopotential MT-binding regions on the fragment thatmay interact independently with two adjacenttubulin dimers.

It may be puzzling to the reader that t-testanalysis reveals only binding sites 1 and 2 with acertainty of .99%, while sites 3 and 4, althoughnicely visible on surface-rendered 3D maps(Figure 5A yellow rings), do not show up beyonda confidence level of 98% (Figure 5D). However,given the flexible nature of the NT6 fragment, we

Figure 5. Cryo-EM 3D-reconstructions and statistical difference mapping between native MTs and NT6–MTcomplexes. A, The 3D reconstruction of NT6–MT complexes overlaid with the t-test volume, indicating a differencein mass with a certainty of .99%. This analysis reveals two potential NT6 binding sites marked 1 and 2. In addition,two more sites can be identified marked 3 and 4, which form distinct protrusions on the outer rim of both a- andb-tubulin (yellow rings). However, this position may show a larger structural flexibility that lowers the significancelevel in difference calculations. These peaks show up only at a confidence level of ,98% (wire frames in D). B, The3D map of native microtubules overlaid with the t-test difference density and the binding site of kinesin motordomains (boxed area). C, A model of NT6 binding to tubulin. Binding sites 1 and 2 show a stable configuration withinthe groove formed by two adjacent protofilaments, while sites 3 and 4 are more exposed to the outer periphery.Thereby flexibility increases and so does the variance, which reduces the significance of density differences amongnon-decorated and NT6-decorated maps. D, The location of the binding sites with respect to the tubulin dimer andmay explain the different stabilities found between sites 1 and 2 and sites 3 and 4. Binding sites 1 and 2 locate at theshallow shoulder of tubulin to the left of H12, while binding sites 3 and 4 may be interacting with the tubulinC termini. The microtubule minus-end is at the bottom.


suggest the following explanation. The t-testsobviously reveal the most significant differences inmass from one map to another when theseadditional masses (here parts of the NT6 fragment)are present consistently at a distinct location. If thisis not the case (e.g. when a molecule bindsinfrequently or the binding site is flexible), thevariance in a particular region increases and thestatistical significance drops. However, onaveraged maps these regions may still show avisible gain in mass at certain cut-off levels, whichcorresponds to an average over occupied andempty binding sites, or over a flexible site invarious configurations. In our case here, we con-clude that binding sites 1 and 2 show a high levelof occupancy with a stable configuration, whilesites 3 and 4 exhibit a lower occupancy, a higherflexibility, or both. Given the fact that the tubulinC termini are intrinsically flexible and that NT6 isflexible as well, it seems likely that the lowerstatistical significance for binding sites 3 and 4originates from intrinsic flexibility in the system atthese sites. The model shown in Figure 5C maygive an explanation of how the conical shape ofthe t-test volume may form. At the resolutiongiven here, binding sites 1, 3 and 4 may form par-tially overlapping densities. But the combinationof a strong binding site at position 1 and ratherflexible ones at positions 3 and 4 may result in avariance gradient within a common densitydistribution, as outlined with the pink area inFigure 5C. The variance among individual datasetsincreases towards position 3 and 4, and hence, thesignificance for a density difference at theselocations drops. Consequently, the remaining iso-volume (here shown at a confidence level of 99%)narrows in this drop-like shape towards the outerrim, while NT6-related mass spreads over sites 3and 4 in a flexible fashion.

The flexible nature of the ncd tail contrasts therigid arrangement of the other end of the ncd mol-ecule. At least under some nucleotide conditionsthe ncd stalk forms a tight interaction with themotor head domains,26 a configuration thatappears to play a crucial role in the walkingprocess of ncd.12 The rigid conformation of theneck region under ADP conditions seems requiredto position the ncd head into the correct positionon the microtubule track for a minus-end directedpower stroke, and may even prevent the head andtail of one molecule from binding to the samemicrotubule. Hence, flexibility in the tail regionseems required to increase the chance of keepingcontact with the cargo. Interestingly, ncd headsand tails recognize similar binding regions ontubulin. This may simply be caused by the favour-able acidic environment on b-tubulin in this regionwithout having any regulatory effect on the ncdmotor. However, if the full tail in a completedimeric motor complex shows a preferred bindingconformation that is sensitive to the polarity of themicrotubules (something we could not demon-strate here), this may allow the motor to recognise

whether two adjacent microtubules show the samepolarity.

Materials and Methods

Expression and purification of ncd taildomain fragment

A fusion peptide of the ncd tail fragment13 (NT6:N-terminal residues 83–187) and thioredoxin (Trx) wasexpressed in Escherichia coli (strain BL21 DE3 pLysS)and purified as described.13 The Trx fusion partner wasremoved by cleavage with thrombin. The digestionproduct NT6 (10.3 kDa) was purified on an S-Sepharosecolumn and then dialysed against AB buffer (20 mMPipes (pH 6.9), 1 mM MgCl2, 1 mM EGTA, 1 mM DTT).The final concentration was about 0.64 mg/ml (62 mM)in the Bradford assay.

Decoration of microtubules with ncd tail fragment

Tubulin was purchased from Cytoskeleton Inc.(Denver, CO). MTs were polymerised for 20 minutes at37 8C in BRB80 (80 mM Pipes (pH 6.8), 2 mM MgCl2), ata concentration of 5 mg/ml and in the presence of 10%(v/v) DMSO, 2 mM GTP and 20 mM Taxol. MTs werepreformed in solution and left overnight to stabilise thecycling between polymerisation and depolymerisation.Decoration of polymerised MTs with NT6 constructswas performed in solution at a final tubulin concen-tration of 0.5 mg/ml (,4.5 mM ab-tubulin hetero-dimers). Negatively stained samples were prepared bymixing equal amounts of MTs (at 0.5 mg/ml in BRB80)with NT6 (31 mM) and incubation for two minutes. Themix was applied to glow-discharged, carbon-coatedelectron microscopy grids and stained with 1% (w/v)uranyl acetate. Decoration of specimens for cryo-EMwas performed directly on the grid to avoid bundling oftubules. MTs were adsorbed onto holey carbon grids forone minute (Quantifoil, Jena, Germany), incubated fortwo minutes with the NT solution and quick-frozen inliquid ethane by using a plunger essentially asdescribed.27

Alignment of 3D maps

Here, we shifted the maps of undecorated MTs, NT6-decorated and kinesin-decorated maps to the samephase origin by cross-correlating each individual datasetto a common reference at the beginning of the averagingand refinement cycles. The reference used came from aniteratively refined map of NT6-decorated maps. To thisend, we controlled in particular the phases of layer-line17 (Bessel order 22), which corresponds to the 8 nmaxial repeat. Each individual dataset that did not showa good alignment of phases with the reference at thatlayer-line (i.e. shifted by 1808) was omitted. This pro-cedure confirmed that kinesin and NT6 bind to thesame area on MTs.

Unidirectional shadowing

NT6 was diluted 1:1 in BRB80 and added topolymerised MTs (at a concentration of 0.5 mg/ml inBRB80) that were left overnight to stabilise. The mixwas incubated for five minutes and then applied directlyto glow-discharged, carbon-coated electron microscopy


grids. The grids were then frozen in liquid nitrogen andtransferred to the so-called MIDILAB28 for shadowingand imaging. Images were recorded by a Gatan-794Multiscan CCD camera (Gatan, Pleasanton, CA, USA),at an electron dose of 500–1000 electrons/nm2.

Cryo-electron microscopy, image processing and3D reconstruction

Cryo-EM was performed on a Philips CM20 micro-scope, using a GATAN-626 cryo-holder. Images wererecorded at 38,000 £ magnification on Kodak SO-163EM film at a defocus of 21.5 mm to 22.0 mm. Micro-graphs of 15 protofilament/two start helical MTs15 weredigitised using a Zeiss-SCAI scanner at a stepsize of21 mm corresponding to 0.553 nm on the original object.Suitable MTs were helically reconstructed by using theprogram suite PHOELIX29 and SUPRIM.30 Statisticaldifference mapping was performed using Student’st-test as described.31 All datasets were truncated to amaximum resolution of 25 A. The 3D maps werevisualized by using VolVis (SUNY Stonybrook, Figure 5).Figures including atomic coordinates were preparedusing Bobscript 2.3.32

Acknowledgements

This work was supported by a grant from theDeutsche Forschungsgemeinschaft (HO-2276/1-2)to A.H. and by the Austrian Academy of Sciences(APART 673) to A.K.

References

1. Goldstein, L. S. & Philp, A. V. (1999). The road lesstraveled: emerging principles of kinesin motor utiliz-ation. Annu. Rev. Cell Dev. Biol. 15, 141–183.

2. Hirokawa, N. (1998). Kinesin and dynein super-family proteins and the mechanism of organelletransport. Science, 279, 519–526.

3. Vale, R. D. & Milligan, R. A. (2000). The way thingsmove: looking under the hood of molecular motorproteins. Science, 288, 88–95.

4. Mandelkow, E. & Hoenger, A. (1999). Structures ofkinesin and kinesin–microtubule interactions. Curr.Opin. Cell Biol. 11, 34–44.

5. Vale, R. D. (2003). The molecular motor toolbox forintracellular transport. Cell, 112, 467–480.

6. McDonald, H. B., Stewart, R. J. & Goldstein, L. S.(1990). The kinesin-like ncd protein of Drosophila is aminus end-directed microtubule motor. Cell, 63,1159–1165.

7. Chandra, R., Salmon, E. D., Erickson, H. P., Lockhart,A. & Endow, S. A. (1993). Structural and functionaldomains of the Drosophila ncd microtubule motorprotein. J. Biol. Chem. 268, 9005–9013.

8. Walker, R. A., Salmon, E. D. & Endow, S. A. (1990).The Drosophila claret segregation protein is a minus-end directed motor molecule. Nature, 347, 780–782.

9. Hoenger, A. & Milligan, R. A. (1997). Motor domainsof kinesin and ncd interact with microtubule proto-filaments with the same binding geometry. J. Mol.Biol. 265, 553–564.

10. Hirose, K., Lockhart, A., Cross, R. A. & Amos, L. A.

(1996). Three-dimensional cryoelectron microscopyof dimeric kinesin and ncd motor domains on micro-tubules. Proc. Natl Acad. Sci. USA, 93, 9539–9544.

11. Foster, K. A. & Gilbert, S. P. (2000). Kinetic studies ofdimeric Ncd: evidence that Ncd is not processive.Biochemistry, 39, 1784–1791.

12. Wendt, T. G., Volkmann, N., Skiniotis, G., Goldie,K. N., Muller, J., Mandelkow, E. & Hoenger, A.(2002). Microscopic evidence for a minus-end-directed power stroke in the kinesin motor ncd.EMBO J. 21, 5969–5978.

13. Karabay, A. & Walker, R. A. (1999). Identification ofmicrotubule binding sites in the Ncd tail domain.Biochemistry, 38, 1838–1849.

14. Karabay, A. & Walker, R. A. (1999). The Ncd taildomain promotes microtubule assembly andstability. Biochem. Biophys. Res. Commun. 258, 39–43.

15. Beuron, F. & Hoenger, A. (2001). Structural analysisof the microtubule–kinesin complex by cryo-electronmicroscopy. Methods Mol. Biol. 164, 235–254.

16. Karabay, A. & Walker, R. A. (2003). Identification ofNcd tail domain-binding sites on the tubulin dimer.Biochem. Biophys. Res. Commun. 305, 523–528.

17. Crepeau, R. H., McEwen, B., Dykes, G. & Edelstein,S. J. (1977). Structural studies on porcine braintubulin in extended sheets. J. Mol. Biol. 116, 301–315.

18. Hoenger, A., Sablin, E. P., Vale, R. D., Fletterick, R. J.& Milligan, R. A. (1995). Three-dimensional structureof a tubulin–motor-protein complex. Nature, 376,271–274.

19. Hoenger, A., Doerhoefer, M., Woehlke, G., Tittmann,P., Gross, H., Song, Y. H. & Mandelkow, E. (2000).Surface topography of microtubule walls decoratedwith monomeric and dimeric kinesin constructs.Biol. Chem. 381, 1001–1011.

20. Mandelkow, E. M. & Mandelkow, E. (1985).Unstained microtubules studied by cryo-electronmicroscopy. Substructure, supertwist and dis-assembly. J. Mol. Biol. 181, 123–135.

21. Sablin, E. P., Kull, F. J., Cooke, R., Vale, R. D. &Fletterick, R. J. (1996). Crystal structure of the motordomain of the kinesin-related motor ncd. Nature,380, 555–559.

22. Sosa, H., Dias, D. P., Hoenger, A., Whittaker, M.,Wilson-Kubalek, E., Sablin, E. et al. (1997). A modelfor the microtubule–Ncd motor protein complexobtained by cryo-electron microscopy and imageanalysis. Cell, 90, 217–224.

23. Hoenger, A., Thormahlen, M., Diaz-Avalos, R.,Doerhoefer, M., Goldie, K. N., Muller, J. &Mandelkow, E. (2000). A new look at the microtubulebinding patterns of dimeric kinesins. J. Mol. Biol. 297,1087–1103.

24. Kar, S., Fan, J., Smith, M. J., Goedert, M. & Amos,L. A. (2003). Repeat motifs of tau bind to the insidesof microtubules in the absence of taxol. EMBO J. 22,70–77.

25. Al-Bassam, J., Ozer, R. S., Safer, D., Halpain, S. &Milligan, R. A. (2002). MAP2 and tau bind longi-tudinally along the outer ridges of microtubuleprotofilaments. J. Cell Biol. 157, 1187–1196.

26. Sablin, E. P., Case, R. B., Dai, S. C., Hart, C. L., Ruby,A., Vale, R. D. & Fletterick, R. J. (1998). Directiondetermination in the minus-end-directed kinesinmotor ncd. Nature, 395, 813–816.

27. Dubochet, J., Adrian, M., Chang, J. J., Homo, J. C.,Lepault, J., McDowall, A. W. & Schultz, P. (1988).Cryo-electron microscopy of vitrified specimens.Quart. Rev. Biophys. 21, 129–228.


28. Gross, H., Krusche, K. & Tittmann, P. (1990). RecentProgress in High-resolution Shadowing for BiologicalTransmission Electron Microscopy. In ProceedingsXII-th International Congress on ElectronMicroscopy, Seattle, San Francisco Press, SanFrancisco, CA pp. 510–511.

29. Whittaker, M., Carragher, B. O. & Milligan, R. A.(1995). PHOELIX: a package for semi-automatedhelical reconstruction. Ultramicroscopy, 58, 245–259.

30. Schroeter, J. P. & Bretaudiere, J. P. (1996). SUPRIM:

easily modified image processing software. J. Struct.Biol. 116, 131–137.

31. Milligan, R. A. & Flicker, P. F. (1987). Structuralrelationships of actin, myosin and tropomyosinreveals by cryo-electron microscopy. J. Cell Biol. 105,29–39.

32. Esnouf, R. M. (1997). An extensively modifiedversion of MolScript that includes greatly enhancedcoloring capabilities. J. Mol. Graph. Model. 15,132–134.

Edited by W. Baumeister

(Received 12 May 2003; received in revised form 13 August 2003; accepted 19 August 2003)


Documents

A Structural Analysis of the Interaction between ncd Tail and Tubulin Protofilaments