the loop 5 element structurally and kinetically ... · abstract eg5 is a homotetrameric kinesin-5...

2760 Biophysical Journal Volume 101 December 2011 2760–2769

The Loop 5 Element Structurally and Kinetically Coordinates Dimersof the Human Kinesin-5, Eg5

Joshua S. Waitzman,† Adam G. Larson,‡ Jared C. Cochran,§ Nariman Naber,‡ Roger Cooke,‡ F. Jon Kull,§

Edward Pate,{ and Sarah E. Rice†*†Department of Cell and Molecular Biology, Northwestern University, Chicago, Illinois; ‡Department of Biochemistry and Biophysics,University of California, San Francisco, California; §Department of Chemistry, Dartmouth College, Hanover, New Hampshire; and{Voiland School of Chemical Engineering and Bioengineering, Washington State University, Pullman, Washington

ABSTRACT Eg5 is a homotetrameric kinesin-5 motor protein that generates outward force on the overlapping, antiparallelmicrotubules (MTs) of the mitotic spindle. Upon binding an MT, an Eg5 dimer releases one ADP molecule, undergoes a slow(~0.5 s�1) isomerization, and finally releases a second ADP, adopting a tightly MT-bound, nucleotide-free (APO) conformation.This conformation precedes ATP binding and stepping. Here, we use mutagenesis, steady-state and pre-steady-state kinetics,motility assays, and electron paramagnetic resonance spectroscopy to examine Eg5 monomers and dimers as they bind MTsand initiate stepping. We demonstrate that a critical element of Eg5, loop 5 (L5), accelerates ADP release during the initialMT-binding event. Furthermore, our electron paramagnetic resonance data show that L5 mediates the slow isomerization bypreventing Eg5 dimer heads from binding the MT until they release ADP. Finally, we find that Eg5 having a seven-residue dele-tion within L5 can still hydrolyze ATP and move along MTs, suggesting that L5 is not required to accelerate subsequent steps ofthe motor along the MT. Taken together, these properties of L5 explain the kinetic effects of L5-directed inhibition on Eg5 activityand may direct further interventions targeting Eg5 activity.

INTRODUCTION

The homotetrameric, human kinesin-5 motor Eg5 is re-quired for mitosis. An Eg5 tetramer attaches a pair of motordomains to each of two overlapping, antiparallel microtu-bules (MTs) of the mitotic spindle and generates anoutward-directed force. Because Eg5 plays a critical rolein setting up the mitotic spindle, its mechanism and regula-tion are of considerable interest in cancer therapy. Twoclasses of agents—allosteric and ATP-competitive Eg5inhibitors—are being developed as antimitotics (1,2), andhave already proved to be useful reagents for cell biologicalstudies (3,4).

Like the two heads of the well-studied kinesin-1 motor,the two heads of an Eg5 dimer appear to be enzymaticallycoupled on MTs. However, the mechanochemical propertiesof truncated Eg5 dimers are markedly different from thoseof kinesin-1 dimers. Eg5 dimers in solution appear to havetwo distinct ADP affinities (0.6 mM and 3.3 mM), in nearlystoichiometric amounts (5). After the first Eg5 head ejectsADP and binds to the MT, the Eg5 dimer undergoes aslow (0.5 s�1) isomerization before the second head ejectsADP (6). The resulting state, with both heads nucleotide-free (APO/APO), does not generally occur in the enzymaticcycle of the kinesin-1 dimer (7). From Eg5’s APO/APO

Submitted July 13, 2011, and accepted for publication October 13, 2011.

*Correspondence: [email protected]

This work is dedicated to Jonathan Widom, whose insights and friendship

shaped our scientific and personal lives.

Jared C. Cochran’s present address is Department of Molecular and Cellular

Biochemistry, Indiana University, Bloomington, IN.

Editor: E. Michael Ostap.

� 2011 by the Biophysical Society0006-3495/11/12/2760/10 $2.00

conformation, the rear head binds to and hydrolyzes ATP,which releases it from the MT and allows the dimer totake a forward step. After this initial step, the Eg5 dimersubsequently takes a total of 8–10 fast processive stepsalong the MT, coupled to ATPase activity, at a rate of10–12 s�1 before dissociating (5,8).

Eg5’s low processivity stands in stark contrast tokinesin-1’s ability to take 100 steps or more per processiverun (9). During Eg5’s fast stepping phase, the rate-limitingstep is ATP hydrolysis (5), whereas for kinesin-1 it is phos-phate release (7,10). The lag that precedes the second ADPrelease (referred to here as ‘‘the slow isomerization’’) isparticularly puzzling, as it only occurs upon initial contactwith the MTand not during subsequent steps of a processiverun. Furthermore, this feature of Eg5’s enzymatic cycleappears to be completely distinct from other kinesins,suggesting that a slow transition into an APO/APO confor-mation may tailor Eg5 to its specific mitotic MT cross-link-ing function (8).

Some of Eg5’s unique properties result from structuralcoupling mechanisms within the motor domain that arespecific to the kinesin-5 family. The loop 5 (L5) elementin the Eg5 head is exceptionally long relative to otherkinesin families. L5 is critical for the binding of severalallosteric inhibitors of Eg5 (11–13). Electron paramagneticresonance (EPR) and fluorescence polarization studies onEg5 monomers have determined that L5 is structurallycoupled to both the nucleotide site and the neck-linkerelement that initiates forward motility in kinesin-familymotors (14–16), possibly using its conformational flexibilityto contact the nearby Switch I motif (14,17). It is unknown

doi: 10.1016/j.bpj.2011.10.032

mailto:[email protected]://dx.doi.org/10.1016/j.bpj.2011.10.032http://dx.doi.org/10.1016/j.bpj.2011.10.032

L5 Coordinates Eg5 Dimers 2761

whether L5 serves the same functions or has emergent func-tions in Eg5 dimers or tetramers. In this study, we examinethe role of L5 in coordinating the two heads of the Eg5dimer. We compare wild-type Eg5 monomers and dimerswith mutants having a deletion within L5, using kinetics,EPR spectroscopy, and motility assays.

MATERIALS AND METHODS

Cloning, expression, and protein purification

Details on cloning are in the Supporting Material. All constructs were

verified by DNA sequencing. Growth, expression, and purification of all

Eg5-367 proteins in Escherichia coli were as described previously

(8,15,18,19). The same protocols were followed for Eg5-513 proteins.

Experiments were performed in HEPES buffer (10 mM HEPES, pH 6.8,

2 mM magnesium chloride, 1 mM EGTA, 100 mM sodium chloride,

150 mM sucrose, 0.02% (v/v) TWEEN 20) at 25�C, unless otherwisespecified.

Steady-state ATPase measurements

A malachite green assay was utilized to monitor the ATPase activities of

the wild-type and DL5 motors as described in Leonard et al. (20). After

each time course was completed, the reaction was transferred to a 96-

well microplate for endpoint absorbance reading at 650 nm using an xMark

spectrophotometer (Bio-Rad, Hercules, CA). A phosphate standard curve,

spanning from 3 to 96 mM KH2PO4, was generated for each experiment

and read in parallel.

Pre-steady-state kinetic assays

The kinetics of MANT-ADP release were measured using an SF-2004

stopped-flow instrument (KinTek, Austin, TX) as described previously

(6,21). Nucleotide was first removed from Eg5 by adding 10 mM EDTA

directly to the protein (chelating the Mg2þ that allows kinesin motors tobind nucleotide), concentrating in a spin concentrator 10-fold, and diluting

in EDTA buffer (10 mM HEPES, pH 6.8, 10 mM EDTA, 1 mM EGTA,

100 mM NaCl, 150 mM sucrose). This process was repeated 3–4 times

before buffer-exchanging the protein into HEPES buffer using a P-30

(Bio-Rad, Hercules, CA) spin column. Experiments were performed by

equilibrating Eg5 (1 mM) with a stoichiometric quantity of MANT-ADP

racemate. This Eg5,MANT-ADP complex was rapidly mixed with10 mM MTs plus 1 mM MgATP, and MANT fluorescence was monitored

over time (Ex, 356 nm; Em, >400 nm).

Protein labeling and EPR spectroscopy

Detailed methods for spin labeling, sample preparation, EPR spectroscopy,

and data analysis are in the Supporting Material, and are similar to those

used in Larson et al. (15,22,23).

FIGURE 1 Eg5 head showing probe and L5 positions. Eg5 head (gray)

with the neck-linker in its docked conformation (PDB: 3hqd; blue). The

bound ADP, a6 helix, and neck-linker portions of a perpendicular neck-

linker structure are superimposed (PDB: 1ii6; neck-linker in black). Ribose

oxygens on ADP are marked, and V365C is indicated in spacefill on the

neck-linker. L5 (tan) is in sausage-type representation, except for the seven

deleted residues (red) in the DL5 motor protein (TWEEDPL). All fluores-

cent and spin probe structures have been reported elsewhere (6,15,21).

Motility assays

Motility assays were performed essentially as described in Larson et al.

(24), with details provided in the Supporting Material. For motile velocity

measurements, 30-min time-lapse movies were analyzed using ImageJ soft-

ware (http://rsbweb.nih.gov/ij/). The position of the end of each MT was

evaluated for sixty 30-s time increments. Time-averaged velocities were

measured from at least 10 different MTs on each of 2–5 different slides.

RESULTS

Motor proteins and probes used to determinekinetic and structural properties of the Eg5 dimer

To determine how the two heads of the Eg5 dimer coordi-nate with each other, we performed pre-steady-state andsteady-state kinetics, motility assays, and EPR spectroscopyon both monomeric (Eg5-367) and dimeric Eg5 (Eg5-513)motor proteins. These proteins have been kinetically andthermodynamically characterized by our group and others(5,6,13–15,25). We also constructed mutants having a dele-tion of seven residues within L5 (amino acids 126–132:TWEEDPL, referred to as ‘‘DL5’’ here).

For EPR spectroscopy, spin probes were either attachedto the ribose oxygens of ADP (similar to MANT-ADP) tosample the nucleotide pocket of wild-type Eg5 or theywere covalently attached to a specifically introducedcysteine in the neck-linker of cys-light Eg5; this motorhas been described previously and has wild-type activity(15,26). Fig. 1 shows the location of the labeled sites andprobes used in this study.

Both wild-type and DL5 mutants of Eg5 proteinsare active and have rate-limiting ADP release inthe absence of MTs

We first performed ATPase and ADP release experimentson both Eg5-367 and Eg5-367 DL5 in the absence of MTs

Biophysical Journal 101(11) 2760–2769

http://rsbweb.nih.gov/ij/

2762 Waitzman et al.

(i.e., basal conditions) to test the effects of the DL5 mutationon monomer activity. Fluorescence enhancement wasobserved upon MANT-ADP addition to both proteins, con-sistent with MANT-ADP binding to the nucleotide pocket(5). The fluorescence enhancement observed upon MANT-ADP binding to Eg5-367 DL5 was significantly greaterthan that observed for wild-type (10-fold enhancementversus threefold; data not shown). This difference mayreflect a higher MANT-ADP affinity for Eg5-367 DL5 orstructural differences in the nucleotide pocket of wild-typeEg5 versus DL5 mutant.

The basal ATPase rate (0.25 s�1) and basal ADP releaserate (0.27 s�1) measured for Eg5-367 were similar, consis-tent with rate-limiting ADP release. These values are alsoconsistent with those previously reported in Cochran andGilbert (27). The basal ATPase and ADP release rates forEg5-367 DL5 were substantially slower (0.03 s�1 and0.005 s�1, respectively). The faster ATPase activity relativeto ADP release may be due to difficulties measuring suchslow rates in a stopped-flow device, or it may reflect aslightly higher affinity for MANT-ADP than unmodifiedADP; in either case, ADP release appears to be rate-limiting. Thus, the DL5 mutant binds MANT-ADP, is active,and has rate-limiting ADP release in the absence of MTs.However, the DL5 mutation decreases both the basalATPase and ADP release rates of Eg5-367 by at least anorder of magnitude (Fig. 2 A, Table 1).

FIGURE 2 Summary of MANT-ADP release transients for the different

Eg5 motors. (A) The kinetics of MANT-ADP release for Eg5-367 wild-

type and DL5 were measured by first equilibrating Eg5 with MANT-ADP

(1:1) followed by rapid mixing of the Eg5,MANT-ADP complex withMTs plus excess MgATP to effectively prevent rebinding of MANT-ADP

nucleotide to the Eg5 active site. A control transient (squares) with no

Eg5 added to the MANT-ADP provides the fluorescence intensity of

MANT-ADP in solution. Final concentrations in the wild-type (circles)

and DL5 (triangles) experiments are: 1.25 mM Eg5, 1 mM MANT-ADP,

10 mMMTs, 20 mM Taxol, 500 mMATP. Data were fit to a single exponen-

tial rate of 42.3 5 9.9 s�1 for wild-type Eg5-367 and 0.008 5 0.002 forEg5-367DL5. (B) Transients for basal MANT-ADP release from Eg5-367

wild-type (circles) and DL5 (triangles), as well as MT-stimulated

MANT-ADP release from Eg5-367 DL5 (rhombi) are shown on a longer

time domain. Concentrations of reactants are the same as in panel A.

Data were fit to a single exponential rate of 0.27 5 0.02 s�1 for wild-type Eg5-367 and 0.005 5 0.002 s�1 for Eg5-367 DL5. (C) Transientshowing MT-stimulated MANT-ADP release for Eg5-513 DL5. Concentra-

tions of reactants are the same as in panel A. Data were fit to a two-expo-

nential rates of 0.15 5 0.01 s�1 and 0.005 5 0.001 s�1. In panels A andB, transients have been normalized for comparative purposes.

The DL5 mutation severely inhibitsMT-stimulation of initial ADP release butmodestly slows ATPase activity and motility

To test the effects of L5 on Eg5’s initial MT-stimulated ADPrelease event, we measured the rates of MT-stimulatedMANT-ADP release for Eg5-367 and Eg5-367 DL5, inthe presence of 10 mM MTs. Eg5-367 released ADP at42.3 s�1, comparable with previous observations (26). Instriking contrast, Eg5-367 DL5 released ADP ~5000-foldslower, at 0.008 s�1. MTs stimulate Eg5-367 ADP release>200-fold, but do not significantly stimulate Eg5-367DL5 ADP release. Previous work on allosteric inhibitorsof Eg5 that bind to L5 demonstrated that these drug treat-ments allow the Eg5 head to bind to MTs while remainingtightly bound to ADP, and that MT stimulation of ADPrelease is inhibited (18,21,28). This is similar to our findingshere with Eg5-367 DL5 (Table 1, Fig. 2 B).

After the initial ADP release event, the steady-stateATPase activity of Eg5-367 DL5 at 10 mM MTs isonly twofold slower than wild-type Eg5-367 (3.7 s�1 vs.6.3 s�1; see Table 1), and roughly 450-fold faster than theinitial rate of MT-stimulated ADP release under the sameconditions (3.7 s�1 vs. 0.008 s�1). This high steady-stateATPase activity relative to initial ADP release suggests thatthe Eg5-367 DL5 monomer, unlike wild-type Eg5-367,hydrolyzes several molecules of ATP per MT encounter,


with subsequent rounds of ATP hydrolysis occurring morequickly than the initial ADP release event (29). Our exper-iments reveal the same phenomenon with dimeric Eg5-513DL5, which undergoes MT-stimulated ADP release at a

TABLE 1 ADP release, ATPase, and motility rates of Eg5

motor proteins

Enzymatic activity

Eg5-367 Eg5-367 DL5 Eg5-513 DL5

Basal ADP release (s�1) 0.27 5 0.02 0.005 5 0.002 —Basal ATPase (s�1 site�1) 0.25 5 0.02 0.03 5 0.004 —MT-stim ADP

release (s�1)42.3 5 9.9 0.008 5 0.002 0.15 5 0.01

0.005 5 0.001

MT-stim ATPase

(s�1 site�1)6.25 5 0.21 3.69 5 0.44 1.00 5 0.03

Motility rate (nm/s)

Eg5-367 Eg5-367 DL5 Eg5-513 Eg5-513 DL5

20.5 5 2.8 12.1 5 1.1 32.9 5 4.2 2.8 5 0.5

Basal and MT-stimulated ADP release and ATPase rates, as well as sliding

velocities are experimentally determined averages5 the standard deviation

(n¼ 2–9 for individual reactions). Transients for ADP release rate measure-ments are shown in Fig. 2. Microtubule sliding motility rate measurements

are depicted in Movie S1, Movie S2, Movie S3, and Movie S4 in the Sup-

porting Material.


maximal rate of 0.15 s�1, but has a steady-state ATPase rateof 1.0 s�1 (Table 1, Fig. 2 C).

To determine whether the ATPase activity of the DL5proteins could drive MT sliding, we performed slidingmotility assays (Table 1, and seeMovie S1,Movie S2,MovieS3, and Movie S4 in the Supporting Material). In multiplefields of view on several separate slides, the vast majorityofMTsmoved steadily. There was an approximately twofoldreduction in the motile velocity of Eg5-367 DL5 relative toEg5-367 (12.1 nm/s vs. 20.5 nm/s), which mirrors thetwofold reduction in ATPase activity described above. Wealso observed the MT sliding velocities of the Eg5-513proteins. Wild-type Eg5-513 slides MTs at a rate of 33 nm/s,which is reasonably consistent with several previous reports(8,30,31). The motile velocity of Eg5-513 DL5 was reducedto 2.8 nm/s. This velocity is sevenfold less than that of wild-type Eg5-367, which is again comparable to their sixfolddifference in ATPase activity (6.3 s�1 vs. 1.0 s�1).

These results together indicate that despite their extremelyslow basal and MT-stimulated ADP release rates, the DL5mutants are properly folded and active. As the ATPaseactivity and MT sliding velocities of both Eg5-367 DL5and Eg5-513 DL5 motors exhibit substantially less severedefects than their initial MT-stimulated ADP release rates(5000-fold for the Eg5-367 proteins), we conclude that theDL5 mutation has its most profound effect on Eg5’s initial,MT-stimulated ADP release event. Once Eg5 has bound tothe MT and released its initial ADP, the DL5 mutant motorscan still efficiently couple ATPase activity to motility, albeitat a slower rate than wild-type motors.

EPR spectroscopy on Eg5 dimers

Two questions emerge from the above data: First, do Eg5-367 and Eg5-513 have different kinetics because of an

interaction between the two heads of an Eg5 dimer, orbecause of differences in monomer versus dimer neck-linkerdynamics? Second, what, structurally, occurs in the slowisomerization step, and how is L5 involved? We approachedthese questions using EPR spectroscopy. EPR spectra areexquisitely sensitive to the environment of the EPR probe,and the EPR timescale is fast compared to most proteinconformational changes (15,32,33). Thus, an interactionbetween the two heads of an Eg5 dimer could alter eitherthe components of an EPR spectrum, their relative abun-dance, or both, and these changes are readily quantifiableby spectral deconvolution (23).

The Eg5 head interconverts between highand low ADP affinity states in solution

We performed EPR spectroscopy on labeled Eg5 dimers insolution to detect head-head interactions or head-headcommunication. Fig. 3 A shows spectra of 20,30-SLADPbound to both Eg5-367 and Eg5-513. As described previ-ously, Eg5-367 displays two spectral components (labeledC1 and C2 in Fig. 3 A) whereas Eg5-367 DL5 has onlythe C1 inner component (15). The Eg5-513 and Eg5-513DL5 dimer spectra shown here have exactly the samecomponents in the same relative amounts as Eg5-367 andEg5-367 DL5, respectively. These data suggest that dimer-ization does not alter the abundance of the structural statesobserved for 20,30-SLADP in the Eg5 nucleotide pocket,either directly or allosterically. Furthermore, for Eg5-513bound to 20,30-SLADP, the same C1 and C2 componentsare present, with the same relative amounts of spin probein each component, at 0.7 mM, 7 mM, and saturating 20,30-SLADP. This suggests that each nucleotide binding sitein Eg5-513 transitions between the conformations corre-sponding to the C1 and C2 components quickly, relativeto the bound-state lifetime of the 20,30-SLADP probe (seeFig. S6).

To probe the relative populations of 20,30-SLADP probesin the different conformational states observed by EPR, weperformed spectral deconvolutions on the wild-typeEg5-367 and Eg5-513 spectra shown in Fig. 3 A. Deconvo-lutions were performed using a linear least-squares methodwith known basis spectra as inputs (15,23). Both Eg5-367and Eg5-513 spectra could be deconvolved into threecomponents: free probe in solution, the inner C1 component(which resembles the DL5 spectrum), and the L5-depen-dent, outer C2 component. Approximately 70% of 20,30-SLADP probes in both Eg5-367 and Eg5-513 spectra werefound in the inner, C1 component, whereas 30% were foundin the outer, C2 component. Krzysiak et al. (5) similarlyfound that ~60% of Eg5-513 heads partition into a highADP affinity state, whereas the remaining 40% are in alow ADP affinity state. Meanwhile, the DL5 mutationand/or L5-targeted inhibition have been found to increaseADP affinity (18,34). Therefore, the inner, C1 component


FIGURE 3 EPR on the nucleotide pocket of Eg5-513 in solution and bound to MTs. (A) EPR spectra of wild-type (black) and DL5 (red) Eg5-367 and Eg5-

513 bound to 20,30-SLADP in solution. Spectra of wild-type Eg5 reveal two components corresponding to protein-bound probe—an inner mobile component(C1) and an outer, nearly immobilized component (C2). Spectra are scaled so that C1 components have similar magnitudes, and protein-bound components

are expanded (at left) for comparison. C2 is not observed for the Eg5-513 DL5 mutant, as reported previously for Eg5-367 (15). (B) Comparison of MT-bound

Eg5-513 nucleotide probe spectra with monomer spectra. (Black) Spectra of MT-bound pellets of Eg5-513; (red) Eg5-367 solution spectra; and (blue) Eg5-

367 MT-bound pellet spectra. All spectra are scaled so that the outer, Eg5-bound components have similar magnitudes. Eg5-bound components are expanded

(at left) for comparison. (i) Wild-type Eg5 bound to 20,30-SLADP. (ii) DL5 mutants bound to 20,30-SLADP. (iii) STLC-treated, wild-type Eg5 bound to 20,30-SLADP. (iv) Wild-type Eg5 bound to 20,30-SLADP,AlFx. (C) Splittings of EPR spectral components, reported in milliTesla (mT). Splittings are generallyaccurate to within 50.2 mT; see Larson et al. (15) for details. (Short dashed line) No C2 component is present for the sample. For splittings (labeled with

double asterisks), see Fig. S6 in the Supporting Material; note that although the splittings determined by the peak of the spectral components are not signif-

icantly different, the MT-bound spectrum is broadened.


seen in our wild-type and DL5 EPR spectra may correspondto a highADPaffinity state, whereas the outer, C2 componentmay correspond to a low ADP affinity state that is accessibletowild-type, but notDL5motors. BothEg5-367 andEg5-513can access both conformations, and they both access theL5-dependent conformation ~30% of the time.

EPR spectra of the nucleotide site and neck-linerof Eg5-367 and Eg5-513 are identical

We next performed EPR using an MSL probe site-specifi-cally attached to position V365 within the neck-linker totest whether potential interactions between the two headsestablish an asymmetry between the two heads in solution.Fig. S7 shows the spectra of MSL bound to Eg5-513in the ADP and APO nucleotide states in solution. Notably,the same spectral components are seen in both states, in the


same relative amounts, and these are similar to what hasbeen reported previously for Eg5-367 (15). Thus, weobserve no structural effects of dimerization in either thenucleotide pocket or the neck-linker.

L5 prevents Eg5 dimer heads fromsimultaneously binding ADP and the MT

We proceeded to examine how the heads of Eg5-513 arecoupled to each other whereas the motor is bound to MTsusing spin-labeled Eg5 motors pelleted with MTs. RobustEPR signals from pellet samples verified that Eg5-513 andEg5-513 DL5 motor proteins were bound to MTs in allexperimental conditions, but these proteins may have eitherone or both heads bound.

The spectrum of 20,30-SLADP bound to Eg5-367 on MTsshows a distinctly different set of components from the


solution spectrum, as described previously (Fig. 3 B (i), blueversus red spectrum) (15). In general, EPR spectra of 20,30-SLADP bound to kinesin motors, including Eg5, are broad-ened upon binding the MT due to a movement of Switch Itoward the nucleotide pocket (32,33,35). For Eg5-367, thisappears as an outward shift of the C1 component. Thisbroadening effect occurs for 20,30-SLADP bound to Eg5-367, Eg5-367 DL5, STLC-treated Eg5-367, and for theEg5-367,ADP,AlFx complex upon MT binding (Fig. 3 C)(15). Thus, the Eg5-367 EPR spectra can serve as signaturesfor MT-bound versus unbound Eg5 heads.

Fig. 3 B (i) shows an overlay of Eg5-513 bound to20,30-SLADP and MTs (black) with spectra of Eg5-367bound to 20,30-SLADP in solution (red) and bound to MTs(blue). The spectral components of 20,30-SLADP bound toEg5-513 within these pellets were more similar to thoseobserved for Eg5-367 in solution than to those of Eg5-367bound to MTs. The majority of 20,30-SLADP probes aretherefore likely to be bound to MT-unbound Eg5-513 heads.As the concentrations of MTs in these pelleted samples is60 mM before centrifugation and in the range of severalhundred micromolar afterward, these experiments are notsuited for determining the affinities of ADP-bound headsfor MTs. Rather, these experiments represent an extremepoint: even at several hundred micromolar MTs, we donot observe a significant fraction of 20,30-SLADP-boundEg5 heads bound to MTs. Thus, the vast majority of Eg5-513 heads are either MT-bound or 20,30-SLADP-bound,but not bound to both. Notably, this binary split betweenADP-binding and MT-binding is an emergent property ofEg5-513 that is not present in Eg5-367 (21).

Our previous work showed that the DL5 mutation didnot affect the EPR spectrum of 20,30-SLADP on MT-boundEg5-367 (Fig. 3 B (i and ii)) (15). The EPR spectrum of20,30-SLADP bound to Eg5-513 DL5 on MTs is very similarto that of 20,30-SLADP bound to either Eg5-367 or Eg5-367DL5 on MTs (Fig. 3 B (ii), black versus blue spectrum). Thisspectrum appears shifted outward slightly relative to theEg5-367 DL5 solution spectrum (Fig. 3 B (ii), black versusred spectrum; see also Fig. 3 C, and Fig. S8 in the Support-ing Material). Similarly, STLC treatment of Eg5-367yielded a single spectral component for 20,30-SLADP, andthis component clearly shifted outward when the monomerbound to MTs (Fig. 3 B (iii), red versus blue spectrum) (15).The spectrum of Eg5-513 in the presence of STLC resem-bles the MT-bound spectrum of Eg5-367 in the presenceof STLC. We interpret these data to indicate that disruptionof L5, either by the DL5 mutation or by STLC treatment,enables Eg5-513 to bind to both 20,30-SLADP and MTs.

In our previous work, we observed that 20,30-SLADP,AlFx bound to Eg5-367 in solution yields the same twospectral components as 20,30-SLADP (red spectra, Fig. 3 B(i and iv)) (15). On MTs, however, 20,30-SLADP,AlFxinduces pronounced spectral broadening of both compo-nents. Interestingly, the spectrum of 20,30-SLADP,AlFx

bound to wild-type Eg5-513 resembles the MT-boundEg5-367 spectrum, consistent with dimer heads containingADP,AlFx being bound to MTs. Thus, the inability of anEg5-513 head to bind both ADP and the MT does not applyto the ADP,AlFx state.

In summary, Eg5 dimers differ from monomers in that ahead of an Eg5 dimer that is bound to ADP does not bindto MTs, and vice versa (Fig. 3 B (i)). In a preparation ofADP-Eg5-513 bound to MTs, some dimers will adopt aconformation having an APO MT-bound head and anADP-bound free head, but many dimers may, in fact,discharge both ADPs.

Eg5 dimers bind to MTs with one neck-linkerdocked and one undocked

We then turned our attention to how dimerization affects theconformation of the neck-linker element. In previous work,our lab used MSL attached to position V365 of Eg5-367 toreport on the neck-linker conformational state (Fig. 1 andFig. 4 A). These spectra showed two components in boththe ADP and APO states on MTs, with a significant shiftto the more immobilized component in the APO state rela-tive to ADP (15). Based on previous studies of kinesin-1 andEg5, the immobilized component has been interpreted toreport the docked neck-linker conformation that positionsa partner head over its binding site, one tubulin dimertoward the plus-end of the MT. The mobile componenthas been interpreted to represent the perpendicular confor-mation of the neck-linker that allows the forward headof a pair of heads to assume an extended length and reachits binding site (15,16,36,37).

Spectra of MSL attached to V365C on Eg5-513 boundto MTs revealed the same immobilized component with asplitting of ~6.5 milliTesla (mT) as observed for Eg5-367(Fig. 4, A and C). Spectra taken in the presence of ADP,in the APO state, and in the presence of ADP,AlFx are virtu-ally identical. All of these spectra show roughly half ofprobes in the mobile component and half in the immobilizedcomponent (Fig. 4 C and see Fig. S9). As discussed above,our preparation of ADP-Eg5 dimers bound to MTs mayhave discharged one or both ADPs, and we suspect that asignificant fraction of these motors may be bound to theMT in an APO/APO state.

Several lines of evidence suggest that the vast majority ofEg5-513 dimers in the APO/APO and ADP,AlFx-boundstates have both heads bound to the MT. Our data aboveusing 2030-SLADP,AlFx indicated that the vast majority ofEg5-513 heads in that sample are MT-bound. The APO stateof an Eg5 head, like the ADP,AlFx state, has high MTaffinity, and EPR spectra of MSL bound to the neck-linkerreveal the same spectral components in the same relativeamounts as for ADP,AlFx-bound Eg5.

Unlike wild-type Eg5-513, Eg5-513 DL5 can bind to20,30-SLADP and MTs simultaneously (Fig. 3 B (ii)). As


FIGURE 4 EPR spectra of MSL bound to the neck-linker of Eg5 bound

to MTs. (A) Wild-type Eg5-367 undergoes a significant conformational

restriction from ADP (black)/APO (blue), and remains immobilized inADP,AlFx (purple). In contrast, Eg5-367 DL5 has a highly immobilizedneck-linker throughout its nucleotide cycle (15). (B) Both Eg5-513 and

Eg5-513 DL5 are ~50% immobilized in all three nucleotide states, suggest-

ing that one neck-linker is docked and one is undocked. (C) Splittings of

EPR spectral components in mT and spectral deconvolution results.


shown in Fig. 4 B, the MSL-labeled Eg5-513 DL5 spectraare nearly identical to those taken of Eg5-513 without theDL5 mutation, as ~50% of probes are in the more immobi-lized component and 50% are in the more mobile com-ponent (Fig. 4 C and see Fig. S9). These data imply aconformation for Eg5-513 DL5 like that of wild-type Eg5-513, with one head having a docked neck-linker and theother with its neck-linker undocked. Taken together, ourdata suggest that Eg5 dimers initiate movement by bindingto MTs with both heads; regardless of DL5 mutation status,one neck-linker of Eg5-513 must enter an undocked con-formation. From this conformation, wild-type Eg5 beginsstepping rapidly, whereas the DL5 mutant is limited by itsslower rate of MT-stimulated ADP release.


DISCUSSION

Based on our results, we have three major conclusionsregarding the Eg5 dimer mechanochemical cycle: First,major structural changes resulting from an interaction ofthe two heads of an Eg5 dimer are unlikely to occur in solu-tion. Second, L5 induces a conformation of the head in anEg5 dimer that cannot simultaneously bind to ADP andMTs. This finding provides a structural basis for the slowisomerization step. Third, Eg5 heads release ADP morequickly from the ADP,Pi state that occurs after the APO/APO conformation on MTs than from the ADP state in solu-tion. A model based on these findings is shown in Fig. 5; thedata supporting these conclusions are discussed below.

The nucleotide pockets and neck-linkersof dimeric Eg5 appear to be structurallyequivalent in solution

The structural environments of the nucleotide pocket andthe neck-linker as observed by EPR provide a sensitivereadout of many aspects of motor activity, including thenucleotide state, MT-binding, neck-linker conformation,and even the concentrations of various buffer components(15,32,33,36). Nonetheless, the EPR spectra of 20,30-SLADP bound to the nucleotide pocket of Eg5-513 wereindistinguishable from those of Eg5-367. In both cases,70% of probes were found in a higher mobility componentand 30% in a lower mobility component. Similar spectrataken with DL5 motor proteins or STLC treatment, whichresult in increased ADP affinity, yielded only the highermobility component. Taken together, these data suggestthat each wild-type Eg5 head, regardless of dimerizationstatus, occupies a high ADP affinity state ~70% of thetime and a low ADP affinity state ~30% of the time.A more detailed discussion of these results is given in theSupporting Material, including a comparison to theMANT-ADP titrations described in Krzysiak et al. (5).

As the monomer and dimer spectra are indistinguishable,we conclude that dimerization is unlikely to exert a signifi-cant effect on nucleotide site and/or neck-linker structure forEg5 in solution. Our data are not definitive in ruling outinteractions between the two Eg5 heads. However, they doargue that the coordination between the nucleotide pocketsand neck-linker elements that enables an Eg5 dimer toinitiate a processive run is established upon binding to theMT rather than in solution before binding to the MT.

L5 induces a conformation of Eg5 that cannotsimultaneously bind ADP and MTs

Our kinetic data show that L5 substantially increases theinitial rate of ADP release of Eg5 monomers and dimers,both in solution and on MTs (Table 1). Meanwhile, ourstructural data using EPR indicate that Eg5-513 heads that

FIGURE 5 Model of Eg5 dimer mechanochemical cycle; bound nucleotides indicated as follows. (Red) Eg5 heads in solution. (Cyan) Eg5 heads MT-

bound. (Purple) Eg5 heads undergoing rapid detachment from a rearward binding site followed by rebinding to a forward binding site. In solution, the

two Eg5 dimer heads are equivalent. The initial MT-stimulated ADP release event is rapid (Step 1). After this step, the second ADP-bound Eg5-513

head cannot bind to MTs, whereas the APO state favors a docked neck-linker conformation on MTs. As a result, the transition into the APO/APO confor-

mation, with one docked and one undocked neck-linker, is very slow (Step 2). Subsequent ATP binding to the rear head is rapid (Step 3). After hydrolysis

(Step 4), the rear head, which may be bound to ADP,Pi or may have just released phosphate, is able to rebind in front of the APO head (Step 5) much morerapidly than in Step 2. This ADP,Pi (or ADP) state may prime the detaching head for fast MT rebinding, enabling subsequent steps on the MT to occur muchfaster than the initial slow isomerization step, particularly for DL5 proteins (cycle to Step 3). The DL5 mutation slows Step 1 and Step 2 several hundredfold,

but affects subsequent steps only ~10-fold. These data suggest that the action of L5 during the initiation steps of the Eg5 cycle (Step 1 and Step 2) may mimic

the ADP,Pi (or ADP) state (Step 5) that accelerates Eg5 activity during subsequent steps.


are bound to 20,30-SLADP do not bind to MTs. This ‘‘either/or’’ binding is L5- and nucleotide-dependent, as bothEg5-513 DL5 heads and STLC-treated heads bound to20,30-SLADP bind to MTs, as do Eg5-513 heads bound to20,30-SLADP,AlFx (Fig. 3). Based on all of these data, wepropose that L5 induces a conformation of the Eg5 headin solution that cannot simultaneously bind to ADP andMTs. Our kinetic data show that this L5-dependent confor-mation of the Eg5 head accelerates MT-stimulated ADPrelease.

The structural basis of the slow isomerizationstep of Eg5-513 is the low MT-stimulationof ADP release from the second head

An Eg5-513 dimer releases one ADP molecule upon MTbinding, then undergoes an ~0.5 s�1 isomerization stepbefore releasing its second ADP (5). Our results show thateven at very high nucleotide and MT concentrations, Eg5dimer heads do not bind to both ADP and MTs at thesame time (Fig. 3). Additionally, wild-type, but not DL5nor STLC-treated Eg5 heads, can access a weak ADPaffinity conformation in solution (Fig. 3). Previous resultssuggest that APO-Eg5 is stable in solution (27). Theseresults together suggest that L5 may weaken both theADP and MT affinity of Eg5 dimer heads.

To span the 8-nm distance between adjacent MT bindingsites, the heads of the Eg5-513 dimer must bind one in frontof the other, with the rear head having a docked neck-linkerwhereas the forward head has an undocked neck-linker. OurEPR data support this mode of binding: in all nucleotidestates we observe approximately half of probes in themobile spectral component and half in the more immobi-lized component (Fig. 4). Thus, the slow isomerizationstep results from a conformational conundrum for the twoEg5 heads. As APO Eg5-367 monomers favor the docked

neck-linker conformation (15,16), the two heads are bothenergetically predisposed to dock their neck-linkers uponreleasing ADP and binding the MT. However, structurally,one of the two heads must bind to the MT with its neck-linker undocked. We conclude that after the first head ofthe Eg5-513 dimer binds MTs and releases ADP, the slowisomerization step is a concerted transition from a state inwhich the second head is ADP-bound and MT-unbound toone in which the dimer has two APO, MT-bound headswith a forward, undocked neck-linker and a rear, dockedneck-linker.

The rate of ADP release differs between initialand subsequent steps of Eg5

The Eg5-513 ATPase rate has been estimated to increase~20-fold after the slow isomerization step, to 10.7 s�1 (5).Our results with both Eg5-367 DL5 and Eg5-513 DL5show that the steady-state ATPase activity of these proteinsis much faster than initial MT-stimulated ADP release(Table 1; 3.69 s�1 vs. 0.008 s�1 for Eg5-367; 1.00 s�1 vs.0.15 s�1 for Eg5-513). These data suggest that Eg5 headshave accelerated ADP release in subsequent steps relativeto the first, initiation step, and these subsequent steps do notdepend heavily on L5. There must be a structural differencebetween the initial MT-stimulated ADP release event, whichis accelerated by L5, and subsequent, much faster ADPrelease events that take place after ATP binding and hydro-lysis on theMT,which can be stimulated independently ofL5.

We propose that L5 induces a conformation of the nucle-otide pocket of the ADP-bound Eg5 head in solution withweak MT affinity and weak ADP affinity. This L5-ADPconformation can stimulate initial ADP release by the firstEg5 head because the first head is conformationally freeto dock its neck-linker upon releasing ADP (Fig. 5, Step1), but cannot accelerate MT-binding by the second head



(Fig. 5, Step 2). After both heads have bound to the MTin their APO states, ATP binds to the nucleotide pocketof the rear head and is hydrolyzed (Fig. 5, Step 3 andStep 4). The rear head transitions into a weakly MT-boundADP,Pi conformation that does not require L5. This allowsthe front (APO) head to dock its neck-linker and detach therear ADP,Pi head from the MT (Fig. 5, Step 5). If Pi releaseby the rear head is followed by sufficiently rapid rebindingof that head to the forward binding site with the neck-linkerundocked, it can bypass the L5-ADP conformation and theslow isomerization step. In subsequent steps, MT-stimulatedADP release is rapid and largely L5-independent, due to thisPi-dependent mechanism (Fig. 5, Step 5/Step 3).

The kinetic and structural features of Eg5-367 DL5 andEg5-513 DL5 are consistent with the above model. Bothproteins have severely reduced rates of initial ADP release.However, subsequent hydrolysis events are relatively unhin-dered by the DL5 mutation, reduced only 2–7-fold. Eg5-513DL5 can bind to the MT with both heads without releasingADP (Fig. 3 B (ii)), then undergo multiple rounds of ATPhydrolysis without releasing from the MT (Table 1). BothEg5-367 DL5 and Eg5-513 DL5 proteins perform remark-ably well in motility assays, where other motors may stim-ulate MT detachment (Table 1, and see Fig. S5).

Previous data support a model in which L5 affects the Eg5nucleotide pocket conformation (14,17,25,38). While wehypothesize that L5 competes with Pi, allowing the ADP-bound motor domain in solution to assume an ADP,Pi-like state that can rapidly discharge ADP, other modelsare tenable. Regardless, we have established that there isa substantial difference between Eg5’s L5-dependent, initialMT-stimulated ADP release event and subsequent ADPrelease events, which are accelerated independently of L5.Additionally, wild-type Eg5 must have a MT dissociationpathway that is not available to the DL5 mutant, as thewild-type motor hydrolyzes fewer ATPs per encounterwith the MT. Future work will be needed to determine thestructural mechanisms that underlie these observations.

Inhibitors of Eg5 that target L5 and the nucleotidepocket target different aspects of motor activity,and would likely have synergistic effects

Our results and previous kinetic characterizations by othergroups suggest that allosteric inhibitors of Eg5 like STLCor ispinesib function much like the DL5 mutation(18,21,28). These agents should drastically slow initialADP release, but have relatively minor effects on Eg5motility once the motors engage MTs. Our motility assaysof the DL5 mutants indicate that the MT-sliding ability ofallosterically inhibited Eg5 motors may increase whenmany motors are coupled together. As the mitotic spindlecouples the motility of multiple Eg5 tetramers, it is not clearwhether mitosis would be effectively inhibited in the pres-ence of many kinetically slowed but not arrested Eg5motors.


Intriguingly, the L5-targeted allosteric inhibitors have facedsignificant setbacks in clinical trials, largely due to insuffi-cient inhibition of cancer cell division in vivo (39).

ATP-competitive inhibitors halt Eg5’s activity by a dif-ferent mechanism. These inhibitors bind to APO, MT-boundheads, locking them tightly to MTs (2,3). This action wouldbe expected to prevent detachment of the rear head of anEg5 dimer, effectively preventing motors from drivingMT sliding. The independent structural mechanisms ofATP-competitive and allosteric inhibitors suggest that acombination of these two types of agents could providesynergistic effects on motor activity in both cell biologicaland clinical studies.

SUPPORTING MATERIAL

Supporting methods, discussions, four movies, and five figures are available

at http://www.biophysj.org/biophysj/supplemental/S0006-3495(11)01253-7.

The authors acknowledge Rice Lab members, S. Gilbert for the Eg5-367

construct, Park Packing for tubulin materials, C. Sindelar for his spectral

deconvolution algorithm, and C. Felix at Medical College of Wisconsin

for providing EPR facilities.

This work was supported by the National Institutes of Health under grant

GM072656 (S.E.R., A.G.L., and N.N.), AR042895 (R.C. N.N.),

GM077067 (E.P. and N.N.), and GM097079 (J.C.C. and F.J.K.). J.S.W. is

supported by the Myhrvold Family Fellowship from the Fannie and John

Hertz Foundation, the Malkin Scholars Program from the Robert H. Lurie

Comprehensive Cancer Center of Northwestern University, and a Cellular

and Molecular Basis of Disease training grant (GM08061).

REFERENCES

1. Luo, L., J. D. Carson, ., R. A. Copeland. 2008. Conformation-dependent ligand regulation of ATP hydrolysis by human KSP: activa-tion of basal hydrolysis and inhibition of microtubule-stimulatedhydrolysis by a single, small molecule modulator. J. Am. Chem. Soc.130:7584–7591.

2. Luo, L., C. A. Parrish, ., K. R. Auger. 2007. ATP-competitiveinhibitors of the mitotic kinesin KSP that function via an allostericmechanism. Nat. Chem. Biol. 3:722–726.

3. Groen, A. C., D. Needleman,., T. J. Mitchison. 2008. A novel small-molecule inhibitor reveals a possible role of kinesin-5 in anastralspindle-pole assembly. J. Cell Sci. 121:2293–2300.

4. Kapoor, T. M., T. U. Mayer,., T. J. Mitchison. 2000. Probing spindleassembly mechanisms with monastrol, a small molecule inhibitor ofthe mitotic kinesin, Eg5. J. Cell Biol. 150:975–988.

5. Krzysiak, T. C., M. Grabe, and S. P. Gilbert. 2008. Getting in sync withdimeric Eg5. Initiation and regulation of the processive run. J. Biol.Chem. 283:2078–2087.

6. Krzysiak, T. C., and S. P. Gilbert. 2006. Dimeric Eg5 maintainsprocessivity through alternating-site catalysis with rate-limiting ATPhydrolysis. J. Biol. Chem. 281:39444–39454.

7. Ma, Y. Z., and E. W. Taylor. 1997. Interacting head mechanism ofmicrotubule-kinesin ATPase. J. Biol. Chem. 272:724–730.

8. Valentine, M. T., P. M. Fordyce, ., S. M. Block. 2006. Individualdimers of the mitotic kinesin motor Eg5 step processively and supportsubstantial loads in vitro. Nat. Cell Biol. 8:470–476.

9. Vale, R. D., T. Funatsu, ., T. Yanagida. 1996. Direct observation ofsingle kinesin molecules moving along microtubules. Nature.380:451–453.

http://www.biophysj.org/biophysj/supplemental/S0006-3495(11)01253-7


10. Klumpp, L. M., K. M. Brendza, ., S. P. Gilbert. 2004. Microtubule-kinesin interface mutants reveal a site critical for communication.Biochemistry. 43:2792–2803.

11. Yan, Y., V. Sardana, ., L. C. Kuo. 2004. Inhibition of a mitotic motorprotein: where, how, and conformational consequences. J. Mol. Biol.335:547–554.

12. Liu, L., S. Parameswaran, ., E. J. Wojcik. 2011. Loop 5-directedcompounds inhibit chimeric kinesin-5 motors: implications forconserved allosteric mechanisms. J. Biol. Chem. 286:6201–6210.

13. Kim, E. D., R. Buckley, ., S. Kim. 2010. Allosteric drug discrimina-tion is coupled to mechanochemical changes in the kinesin-5 motorcore. J. Biol. Chem. 285:18650–18661.

14. Behnke-Parks, W. M., J. Vendome,., S. S. Rosenfeld. 2011. Loop L5acts as a conformational latch in the mitotic kinesin Eg5. J. Biol. Chem.286:5242–5253.

15. Larson, A. G., N. Naber, ., S. E. Rice. 2010. The conserved L5 loopestablishes the pre-powerstroke conformation of the Kinesin-5 motor,Eg5. Biophys. J. 98:2619–2627.

16. Rosenfeld, S. S., J. Xing, ., P. H. King. 2005. Docking and rolling,a model of how the mitotic motor Eg5 works. J. Biol. Chem.280:35684–35695.

17. Harrington, T. D., N. Naber, ., E. Pate. 2011. Analysis of theinteraction of the Eg5 Loop5 with the nucleotide site. J. Theor. Biol.289:107–115.

18. Maliga, Z., T. M. Kapoor, and T. J. Mitchison. 2002. Evidence thatmonastrol is an allosteric inhibitor of the mitotic kinesin Eg5. Chem.Biol. 9:989–996.

19. Krzysiak, T. C., T. Wendt,., A. Hoenger. 2006. A structural model formonastrol inhibition of dimeric kinesin Eg5. EMBO J. 25:2263–2273.

20. Leonard, M., B. D. Song, ., S. L. Schmid. 2005. Robust colorimetricassays for dynamin’s basal and stimulated GTPase activities. MethodsEnzymol. 404:490–503.

21. Cochran, J. C., C. A. Sontag, ., S. P. Gilbert. 2004. Mechanisticanalysis of the mitotic kinesin Eg5. J. Biol. Chem. 279:38861–38870.

22. Crowder, M. S., and R. Cooke. 1987. Orientation of spin-labeled nucle-otides bound to myosin in glycerinated muscle fibers. Biophys. J.51:323–333.

23. Sindelar, C. V., M. J. Budny,., R. Cooke. 2002. Two conformations inthe human kinesin power stroke defined by x-ray crystallography andEPR spectroscopy. Nat. Struct. Biol. 9:844–848.

24. Larson, A. G., E. C. Landahl, and S. E. Rice. 2009. Mechanism ofcooperative behavior in systems of slow and fast molecular motors.Phys. Chem. Chem. Phys. 11:4890–4898.

25. Parke, C. L., E. J. Wojcik,., D. K. Worthylake. 2010. ATP hydrolysisin Eg5 kinesin involves a catalytic two-water mechanism. J. Biol.Chem. 285:5859–5867.

26. Cochran, J. C., T. C. Krzysiak, and S. P. Gilbert. 2006. Pathway of ATPhydrolysis by monomeric kinesin Eg5. Biochemistry. 45:12334–12344.

27. Cochran, J. C., and S. P. Gilbert. 2005. ATPase mechanism of Eg5 inthe absence of microtubules: insight into microtubule activation andallosteric inhibition by monastrol. Biochemistry. 44:16633–16648.

28. Maliga, Z., J. Xing,., S. S. Rosenfeld. 2006. A pathway of structuralchanges produced by monastrol binding to Eg5. J. Biol. Chem.281:7977–7982.

29. Hackney, D. D. 1995. Highly processive microtubule-stimulated ATPhydrolysis by dimeric kinesin head domains. Nature. 377:448–450.

30. Kapitein, L. C., E. J. Peterman, ., C. F. Schmidt. 2005. The bipolarmitotic kinesin Eg5 moves on both microtubules that it crosslinks.Nature. 435:114–118.

31. Kwok, B. H., L. C. Kapitein, ., T. M. Kapoor. 2006. Allosteric inhi-bition of kinesin-5 modulates its processive directional motility. Nat.Chem. Biol. 2:480–485.

32. Naber, N., A. Larson, ., E. Pate. 2011. Multiple conformations of thenucleotide site of Kinesin family motors in the triphosphate state.J. Mol. Biol. 408:628–642.

33. Naber, N., T. J. Minehardt,., E. Pate. 2003. Closing of the nucleotidepocket of kinesin-family motors upon binding to microtubules.Science. 300:798–801.

34. Brier, S., D. Lemaire, ., F. Kozielski. 2006. Molecular dissection ofthe inhibitor binding pocket of mitotic kinesin Eg5 reveals mutantsthat confer resistance to antimitotic agents. J. Mol. Biol. 360:360–376.

35. Sindelar, C. V., and K. H. Downing. 2007. The beginning of kinesin’sforce-generating cycle visualized at 9-Å resolution. J. Cell Biol.177:377–385.

36. Rice, S., A. W. Lin, ., R. D. Vale. 1999. A structural change in thekinesin motor protein that drives motility. Nature. 402:778–784.

37. Rosenfeld, S. S., G. M. Jefferson, and P. H. King. 2001. ATP reorientsthe neck linker of kinesin in two sequential steps. J. Biol. Chem.276:40167–40174.

38. Bodey, A. J., M. Kikkawa, and C. A. Moores. 2009. 9-Ångström struc-ture of a microtubule-bound mitotic motor. J. Mol. Biol. 388:218–224.

39. Kantarjian, H. M., S. Padmanabhan, ., M. Minden. 2011. Phase I/IImulticenter study to assess the safety, tolerability, pharmacokineticsand pharmacodynamics of AZD4877 in patients with refractory acutemyeloid leukemia. Invest. New Drugs. 10.1007/s10637-011-9660-2Epub ahead of print.


The Loop 5 Element Structurally and Kinetically Coordinates Dimers of the Human Kinesin-5, Eg5

Joshua S. Waitzman,† Adam G. Larson,‡ Jared C. Cochran,§ Nariman Naber,‡ Roger Cooke,‡ F. Jon Kull,§ Edward Pate,¶ and Sarah E. Rice† †Department of Cell and Molecular Biology, Northwestern University, Chicago, Illinois; ‡Department of Biochemistry and Biophysics, University of California, San Francisco, California; §Department of Chemistry, Dartmouth College, Hanover, New Hampshire; and ¶

Voiland School of Chemical Engineering and Bioengineering, Washington State University, Pullman, Washington

Supporting Material: Supporting Methods Cloning, Expression, and Purification

The Eg5 monomer construct (Eg5-367, containing the first 367 residues of the motor domain) was described previously (15). An Eg5 dimer construct (Eg5-513, containing the first 513 residues of the motor domain and stalk) was cloned from the MegaMan human transcriptome library (Stratagene, La Jolla, CA) and a C-terminal 6X-histidine tag was introduced using PCR. For cys-light Eg5-513, all native cysteines were replaced with alanine. Removal of native cysteines, and introduction of cysteines into cys-light Eg5 for MSL labeling was performed via Quikchange site-directed mutagenesis (Agilent, Santa Clara, CA). The deletion within L5 (DL5) mutants have a deletion of amino acids 126-132 (TWEEDPL) and were created using Exsite mutagenesis (Stratagene, La Jolla, CA).

Protein labeling

Eg5 motors were concentrated and exchanged into labeling buffer at 4ºC. The protein was labeled at the nucleotide-binding site by addition of 70 µM 2',3'-spin-labeled (SL) ATP to 100µM Eg5. 2',3'-SLATP is readily hydrolyzed by Eg5 to 2',3'-SLADP, and is referred to as 2',3'-SLADP throughout the text. Nucleotide-analog spin probes were synthesized following the protocol described in (22).

Covalent modification of single cysteine-containing Eg5 motor proteins was carried out by addition of 300 µM 4-maleimido-2,2,6,6-tetramethyl-1-piperidinyloxy (MSL, Sigma Aldrich, St. Louis, MO) and 100 µM Eg5 to the labeling buffer. Protein remained for 4 h at 4°C prior to separation from unreacted MSL using a P-30 spin column. Stoichiometries of 90-100% were verified by comparing the amount of spin probe present in the sample by double integration of spectra to the protein concentration determined by Bradford assay, as described previously. In that study, Eg5-367 labeled at 90-100% stoichiometry with MSL retained 80-100% ATPase activity (15). Sample Preparation for EPR spectroscopy

For 2’,3’-SLADP containing samples, a roughly stoichiometric amount of 2',3'-SLADP was added and unreacted spin probe was removed using a P-30 spin column equilibrated with labeling buffer as described previously (18). For experiments with the diphosphate•AlFx species at the active site, 10 mM NaF and 2 mM AlCl3 were added. Nucleotide-free (APO) Eg5 was generated by incubating spin-labeled Eg5 with a solution of 2% apyrase (1 unit/µl; Sigma, St Louis, MO) for 15 minutes. For spectra taken in the presence of S-trityl-L-cysteine (STLC; Sigma, St Louis, MO), 200 µM STLC was added. Eg5 samples in solution were loaded into 25 µL glass capillaries for spectral accumulation.

For the 2',3'-SLADP experiments using known quantities of spin probe in Fig. S6, Eg5 dimers were prepared nucleotide-free, then dialyzed overnight into Labeling buffer (HEPES buffer without TWEEN 20) with no nucleotide added. Residual nucleotide was removed as reported in (22). Briefly, protein was concentrated in a spin concentrator to

~100 µM, then 5 mM EGTA and 5 mM EDTA were added, and protein was incubated at 4°C for 30 minutes before being run through a P-30 spin column (Bio-Rad, Hercules, CA). 2',3'-SLADP was added to sample and spectra were taken immediately. Signal was very weak for the sample containing 0.7 µM 2',3'-SLADP; 3 separate 25 µl capillaries were loaded into the EPR machine and 140 scans were averaged at the highest gain possible (7x105). Similarly, the 7 µM 2',3'-SLADP sample required 40 scans at the same high gain. Other aspects of sample preparation for these experiments were performed as described in Methods.

To generate MT-bound pellet samples, 20 µM spin-labeled Eg5 heads and 60 µM paclitaxel-stabilized, polymerized MTs purified from porcine brain were added to labeling buffer. Nucleotide analog reagents, apyrase, and/or STLC were added for some samples as described above. The MT-Eg5 solution was centrifuged at 100,000xg for 15 minutes in an Airfuge (Beckman, Brea, CA). The supernatant was removed, and the remaining pellet of MTs and bound, spin-labeled motors was transferred onto a quartz flat cell, covered with a coverslip, and sealed with vacuum grease before being placed into the EPR cavity (15).

Importantly for the present work, we verified that a significant fraction of spin-labeled Eg5 motors bound to MTs in all MT-bound pellet samples. After the 100,000xg centrifugation step, significant and highly variable loss of protein occurred while transferring the pellet from centrifuge tube to the flat cell. Nonetheless, double integration of EPR spectra from pellet samples containing MSL-labeled Eg5 and MTs revealed that 35-80% of the MSL-bound Eg5 from the original sample was retained in the flat cell after the sample transfer, while supernatants had very low EPR signal. EPR spectroscopy and data analysis

First-derivative, X-band EPR spectra were accumulated with a Bruker EMX spectrometer (Bruker Instruments, Billerica, MA) as detailed in previous work (15). Deconvolutions to determine ratios of MSL probes in various spectral components were performed using a linear least-squares method employing known basis spectra, as described in (15, 23). All deconvolutions produced χ2 values ≤2% using two basis spectral components.

Motility assays

10 µL of 10 µg/ml anti-6xHis antibody (AbCam, Cambridge, MA, catalog #ab18184) in a solution containing 2 mg/ml bovine serum albumin (BSA) was flowed into the chamber, followed by 3 washes with motility buffer (80 mM PIPES, 2 mM MgCl2, 1 mM EGTA, 0.2 mg/ml BSA, 150 mM sucrose, pH 6.8). 0.5 µM Eg5 was added, followed by 3 more washes. MTs, oxygen scavengers, (Glucose oxidase/catalase, 450mg/ml glucose) and 1mM ATP were added prior to imaging. Supporting Discussion The spectrum of 2',3'-SLADP bound to Eg5-513 in solution does not depend strongly on nucleotide concentration

Importantly, our EPR data show that both Eg5-367 and Eg5-513 can access both structural states reflected by the C1 and C2 components, suggesting that individual Eg5 heads can inter-convert between these states. If the two states inter-convert slowly relative to the rate of 2',3'-SLADP dissociation, we should observe a spectral shift as the concentration of 2',3'-SLADP in the sample is increased. Initial saturation of one component should be followed by an increase in the other component at higher probe concentrations. If, on the other hand, inter-conversion is fast relative to probe dissociation, all spectra should contain the same relative amounts of 2',3'-SLADP in both components. Fig. S6 shows spectra of Eg5-513 with either 0.7 µM or 7 µM 2',3'-SLADP added. Free probe was not removed from these two samples prior to collection of spectra. This is in contrast to the spectrum shown in Fig. 3A, in which 0.5 mM 2’3’-SLADP was added and free probe was removed using a spin column. Importantly, the relative amounts of probe in the C1 and C2 components are insensitive to the concentration of 2',3'-SLADP in the sample; therefore, the rate of inter-conversion between the states represented by these spectral components is significantly faster than the rate of dissociation of 2',3'-SLADP from the Eg5 head. We interpret these results to indicate that both Eg5-367 and Eg5-513 can access both conformations reflected by these spectral components, and they both access the L5-dependent state approximately 30% of the time. A comparison of EPR data using 2',3'-SLADP with MANT-ADP titrations performed by Krzysiak, et al. (5)

Based on the fact that L5 mutants have high ADP affinity and that these mutants lack the C2 component, we hypothesize that the inner, C1 component may correspond to a high ADP affinity state, while the outer, C2 component may correspond to the low ADP affinity state. The 70%/30% ratio of probes in the two components appeared to be independent of the concentration of 2',3'-SLADP, indicating that Eg5 heads may switch between high and low affinity states faster than they dissociate 2',3'-SLADP. In apparent contrast to this idea, Krzysiak, et al. showed in MANT-ADP titrations that Eg5 dimers first populated a high-affinity state with a Kd = 0.6 µM and then a low-affinity state with a Kd = 3.3 µM (5). However, in those MANT-ADP titrations, monomers populated a single, intermediate affinity state (Kd = 1.2 µM). We note that a monomeric Kd of 1.2 µM is roughly equal to an average of the dimer affinities as observed by Krzysiak et al, weighted by the 70%/30% distribution of the high and low-affinity states (0.7*0.6 µM +0.3*3.3 µM = 1.4 µM). Provided that MANT-ADP and 2',3'-SLADP form complexes with Eg5 heads with different bound-state lifetimes, both sets of data are consistent with the idea that single heads can toggle between high and low ADP affinity states, independent of dimerization. Supporting this notion, when we examined the structural environment of the neck-linker in solution using MSL, we found no new spectral components or major changes in the abundance of any existing components in dimer spectra relative to the monomer spectra.

Please see attached movie files for: Movie S1: Eg5-367 WT Movie S2: Eg5-367 DL5 Movie S3: Eg5-513 WT Movie S4: Eg5-513 DL5 Movies S1-S4: Representative movies showing MT sliding by Eg5 motors in this study. Scale bars, 10 µM. Each movie is 30 minutes in length. Density of MTs in these movies is relatively high to demonstrate that a majority of MTs are moving steadily throughout the entire 30-minute timecourse, but only non-overlapping MTs that were visible for at least 15 minutes were used for velocity measurements. Stage drift is visible in these movies, but was corrected for in velocity measurements, shown in Table 1.

Figure S5

Figure S5: Schematic showing DL5 mutant behavior in enzymatic and motility assays. In enzymatic assays of Eg5 DL5 mutants, the steady-state ATPase rate is faster than the initial rate of ADP release, indicating that the motor is accelerated through subsequent, possibly futile rounds of ATP hydrolysis. In motility assays, in contrast, motors’ mechanochemical cycles are coupled through their attachment to the glass surface and can undergo forced detachment.

Figure S6: Deconvolution of EPR spectra of 2',3'-SLADP bound to Eg5 dimers

Figure S6: Top, EPR spectra of 2',3'-SLADP bound to Eg5-513 in solution can be deconvolved into two dominant spectral components. These components correspond to (1) free probe, shown in blue, and (2) 2',3'-SLADP bound to Eg5-513, shown in red. In this bound fraction, approximately 70% of probes are in the C1 component, and 30% in the C2 component. Bottom, EPR spectra of samples containing 0.7 and 7 µM 2',3'-SLADP bound to Eg5-513, as well as at saturating 2',3'-SLADP, all appear to contain the same C1 and C2 components, in similar relative amounts.

Figure S7

Figure S7. EPR spectra of Eg5-513 in solution, site-specifically labeled with MSL at position V365 in the neck-linker. Spectra taken in 2 mM ADP are black, and samples treated with apyrase to induce an APO state are blue. Spectral splittings were 6.35 and 6.4 mT for ADP and APO spectra, respectively.

Figure S8

Figure S8. EPR spectra of 2',3'-SLADP bound to Eg5-367 DL5 in solution and bound to MTs. As described in Fig. 3, the peak splittings of Eg5-367 DL5 in solution and Eg5-367 DL5 on MTs are not significantly different. However, by comparing the spectra directly, it is clear that the spectrum of Eg5-367 DL5 bound to MTs (blue) is broadened relative to the spectrum of Eg5-367 DL5 in solution (red).

Figure S9

Figure S9: Basis spectra and deconvolutions of V365C-MSL bound to wild-type Eg5-513 and Eg5-513 DL5. Basis spectra used to deconvolve complex V365C spectra shown at top; these are the same spectra used as those from previous work (15) where the spectra of Eg5-367-E124C-MSL bound to ADP in solution at 25°C was used as the mobile basis spectrum (blue) and the spectrum of APO V365C-MSL bound to MTs at 2°C was used as the immobilized basis spectrum (red). Notably the same basis spectra could be used to deconvolve both monomer and dimer spectra with very little error. Below, experimentally derived spectra (blue), as well as deconvolutions using the linear least-squares algorithm (purple) and residual (green) are shown.

PMID- 22261065The Loop 5 Element Structurally and Kinetically Coordinates Dimers of the Human Kinesin-5, Eg5IntroductionMaterials and MethodsCloning, expression, and protein purificationSteady-state ATPase measurementsPre-steady-state kinetic assaysProtein labeling and EPR spectroscopyMotility assays

ResultsMotor proteins and probes used to determine kinetic and structural properties of the Eg5 dimerBoth wild-type and DL5 mutants of Eg5 proteins are active and have rate-limiting ADP release in the absence of MTsThe DL5 mutation severely inhibits MT-stimulation of initial ADP release but modestly slows ATPase activity and motilityEPR spectroscopy on Eg5 dimersThe Eg5 head interconverts between high and low ADP affinity states in solutionEPR spectra of the nucleotide site and neck-liner of Eg5-367 and Eg5-513 are identicalL5 prevents Eg5 dimer heads from simultaneously binding ADP and the MTEg5 dimers bind to MTs with one neck-linker docked and one undocked

DiscussionThe nucleotide pockets and neck-linkers of dimeric Eg5 appear to be structurally equivalent in solutionL5 induces a conformation of Eg5 that cannot simultaneously bind ADP and MTsThe structural basis of the slow isomerization step of Eg5-513 is the low MT-stimulation of ADP release from the second headThe rate of ADP release differs between initial and subsequent steps of Eg5Inhibitors of Eg5 that target L5 and the nucleotide pocket target different aspects of motor activity, and would likely hav ...

Supporting MaterialReferences

mmc1- waitzman supplementals

the loop 5 element structurally and kinetically ... · abstract eg5 is a homotetrameric kinesin-5...

Documents