ADP-induced rocking of the kinesin motor domin reveled y singlemolecule fluorescence polriztion microscopy Hernndo Sos 1,2, Erwin J.G. Petermn 3,4, W.E. Moerner 3 nd Lwrence S.B. Goldstein 1,5 1 Deprtment of Cellulr nd Moleculr Medicine nd Howrd Hughes Medicl Institute, University of Cliforni Sn Diego, L Joll, Cliforni 92093-0683, USA. 2 Present ddress: Deprtment of Physiology nd Biophysics, Alert Einstein College of Medicine, Bronx, New York 10461, USA. 3 Deprtment of Chemistry, Stnford University, Stnford, Cliforni 94305-5080, USA. 4 Present ddress: Division of Physics nd Astronomy, Vrije Universiteit, Amsterdm, The Netherlnds. 5 Deprtment of Phrmcology, University of Cliforni Sn Diego, L Joll, Cliforni 92093-0683, USA. Kinesin is n ATP-driven moleculr motor protein tht moves processively long microtuules. Despite considerle reserch, the detiled mechnism of kinesin motion remins elusive. We pplied n enhnced suite of singlend multiple-molecule fluorescence polriztion microscopy ssys to report the orienttion nd moility of kinesin molecules ound to microtuules s function of nucleotide stte. In the presence of nlogs of ATP, ADP-Pi or in the sence of nucleotide, the kinesin hed mintins rigid orienttion. In the presence of ADP, the motor domin of kinesin, still ound to the microtuule, dopts previously undescried, highly moile stte. This stte my e generl to the chemomechnicl cycle of motor proteins; in the cse of kinesin, the trnsition from highly moile to rigid stte fter ADP relese my contriute to the genertion of the 8 nm step. Kinesin, n essentil cellulr motor protein, uses the energy of ATP hydrolysis to move mny steps (8 nm displcements etween tuulin inding sites) long microtuule without detchment 1. Recently, vriety of dt led to the proposl tht zipper-like conformtionl chnge in the neck linker region of the kinesin motor domin occurs upon ATP inding 2. Whether this chnge y itself cn generte the oserved 8 nm steps 3 or whether dditionl conformtionl chnges contriute s well re questions of criticl interest. Bulk fluorescence polriztion nisotropy mesurements of kinesin in solution were used to detect chnges in the overll shpe of the molecule 4. However, this method cnnot detect chnges in the reltive orienttion of kinesin with its microtuule trck. Recently, the development of new iophysicl ssys sed on oserving the fluorescence of single smll dye molecule llows mesurements without ensemle verging nd thus cn resolve inhomogeneity due to different sttic nd/or dynmicl sttes 5 7. One of these ssys, fluorescence polriztion spectroscopy of single fluorophores, ws pplied to study DNA conformtion 8, the rottion of F1-ATPse 9 nd the ctomyosin system 10. We developed n enhnced suite of single- nd multiple-molecule fluorescence polriztion microscopy ssys nd hve pplied these to study the kinesin prolem, thus reporting the orienttion nd moility of kinesin molecules ound to microtuules for the first time. Fig. 1 Proe loction nd orienttion. The fluorophore is-((niodocetyl)piperzinyl) sulfonerhodmine (BSR) (Moleculr Proes) ws docked into the rt kinesin structure 28 y plcing its two rective groups t the β-cron positions of mino cids 169 nd 174. The red rrow indictes the orienttion of the BSR trnsition dipole moment. The position of the ADP molecule (green), loop 11 (L-11, ornge) nd BSR molecule (red) re indicted. In this view the neck linker region (lue), N-terminl (yellow) nd C-terminl (lck) regions re on the ck of the molecule. The reltive orienttions of the kinesin motor domin nd the microtuule re ccording to proposed docking model sed on cryo-electron microscopy 13. Although three lterntive microtuule kinesin docking models hve een proposed 15, they differ minly on the zimuthl orienttion (the ngle round n xis centered on the motor nd prllel to the microtuule long xis) of the motor hed. Therefore ll three predict ner perpendiculr orienttion for the proe dipole. However in one model 29 the proe on our leled construct, KMC2 BSR would e fcing the microtuule, possily interfering with inding. This model then seems inconsistent with the oservtion tht KMC2 BSR inds to microtuules nd hve norml microtuule-stimulted ATPse ctivity (see Methods). The docking orienttion shown hs een further supported y using gold lels 2. The ttchment of ifunctionl fluorophore to two residues of the protein structure orients the trnsition dipole so it cn ccurtely mimic the orienttion of the kinesin hed reltive to the microtuule 11. Although the reltive orienttion is sttic nd similr for lmost ll of the possile nucleotide sttes, in the ADP-ound form we discovered previously unidentified stte in which the hed remins ttched to the microtuule ut rocks ck nd forth. Fluorescent proe loction nd orienttion The kinesin construct (KMC2) used in this nlysis is derived from the first 349 residues of the humn kinesin hevy chin, ut with only two cysteines locted in the β5 α3 region defined in the kinesin crystl structure 12. This monomeric construct includes the motor domin nd neck linker ut not the dimeriztion domin. We leled KMC2 with ifunctionl thiol rective rhodmine derivtive, is-((n-iodocetyl)piperzinyl) sulfonerhodmine (BSR). According to the docking model of kinesin motors nd microtuules 13, the kinesin motor domin in the presence of AMP-PNP ( nonhydrolysle nlog of ATP) orients so tht ifunctionl dipole (BSR) ttched t the indicted residues (169 nd 174) plces the trnsition dipole lmost perpendiculr to the microtuule long xis nd nerly prllel to the rdil direction from the microtuule center (Fig. 1). To determine the orienttion of KMC2 BSR while ound to microtuule undles found in cili nd flgell, we used lser light with lternting perpendiculr polriztion xes to excite KMC2 BSR molecules ound to xonemes (microtuule undles found in cili nd flgell) nd imged the emitted fluores- 540 nture structurl iology volume 8 numer 6 june 2001
letters Fig. 2 Fluorescence polriztion mesurements of xonemes decorted with KMC2 BSR., Experimentl set up for polriztion microscopy mesurements using fluorescence microscope (see Methods)., Typicl imges of xonemes decorted with mny KMC2 BSR molecules excited with 0 nd 90 linerly polrized light (rrows) in the presence of AMP- PNP nd ADP. c, PR 0 90 versus xoneme ngle ( ) in the microscope stge plne for KMC2 BSR decorted xonemes in the presence AMP-PNP (n = 98), ADP (n = 119), ADP-AlF 4 (n = 51) nd no-nucleotide (n = 58). Our definition of PR = (I 90 I 0 ) / (I 90 + I 0 ), is equivlent to the reduced liner dichroism 30. One difference etween the PR, s defined, nd other polriztion mesurements, such s fluorescence nisotropy (A) or polriztion (P), is the time scle in which the orienttion signl is verged. In PR the polriztion signl is verged over the length of the mesurement (100 ms minimum in our cse, see Methods), wheres in stedy stte mesurements of P or A, the signl represents n verge over the fluorescence lifetime of the fluorophore (typiclly in ns). The superimposed curves re the est fit of the mximum PR oserved t ech 10 intervl to the PR eqution (see Methods). These curves intersect the verticl xes t the following PR 0 90 vlues: AMP-PNP = 0.53, ADP = 0, ADP-AlF 4 = 0.61 nd no-nucleotide = 0.57. cence (Fig. 2). We first imged xonemes decorted with mny KMC2 BSR molecules (Fig. 2). The imges otined in the presence of AMP-PNP displyed strong nisotropy. When the exciting light ws perpendiculr to the microtuule xes, more light ws emitted s expected, ecuse light perpendiculr to the microtuule will e ligned with the fluorophore dipole (Fig. 1). To quntify this polriztion nisotropy we clculted rtio PR 0 90 = (I 90 I 0 ) / (I 90 + I 0 ), where I 90 nd I 0 re the totl light emitted with excittion light polrized t 90 or 0, respectively, in the plne of the microscope stge. PR 0 90 ws mesured for mny KMC2 BSR decorted xonemes in rndom orienttions in the presence of different nucleotide species (Fig 2c). Ech mesurement corresponds to the dded fluorescence signl of >1,000 KMC2 BSR molecules. For ech nucleotide stte we lso clculted the expected PR 0 90 for n ensemle of fluorophores with cylindricl symmetry out the xoneme nd given ngle with the microtuule long xis, xil ngle (superimposed curves), chosen to fit the mximum PR 0 90 oserved. Vrious nonidel effects, such s locl distortions of the decorted microtuule structure, would lower the oserved PR 0 90 vlues. Therefore, we concentrted on the highest oserved vlues within ech 10 rnge. In the cse of AMP-PNP, the clculted curve corresponds to dipole xil ngle of 74 (see Methods) from the microtuule xis. This vlue is within the rnge expected from the docking of kinesin motors to microtuules 13 (Fig. 1). In the sence of dded nucleotides (pyrse present) or in the presence of ADP-AlF 4 (thought to mimic the ADP-Pi stte), only smll ngulr chnges were detected. The fitted curves (Fig. 2c) correspond to dipole xil ngle of 74, 76 nd 77 for the AMP-PNP, no-nucleotide nd ADP-AlF 4 sttes, respectively (ssuming the sme mount of rpid proe c wole for these three nucleotide sttes). Consistent with cryo-electron microscopy dt, these results indicte similr configurtion for the microtuule-ttched kinesin motor in these three nucleotide sttes 2,14,15. The ADP-ound stte of kinesin In contrst to the AMP-PNP, ADP-AlF 4 nd no-nucleotide sttes, in the presence of ADP ll PR 0 90 re close to 0 (Fig. 2c). This lck of polriztion nisotropy cn hve three cuses: (i) the xil ngle of fluorescence dipoles hs shifted to 54.7, which leds to cncelltion of the nisotropy from the cylindricl verge round the xoneme; (ii) the dipoles re in rndom ut fixed orienttions; or (iii) the dipoles re highly moile on the time scle of the mesurement. Mesurements on single kinesin motors ound to xonemes llowed us to differentite etween these possiilities. In contrst to the ner uniform fluorescence emission long the xoneme seen t higher concentrtions (Fig. 2), reduced kinesin concentrtions in the ssy displyed line of individul fluorescent spots (Fig. 3). Time courses of the fluorescence intensities from these spots (Fig. 3) reveled constnt intensity vlues with discrete photoleching events, typicl of singlemolecule recordings 16. To differentite whether the low PR 0 90 vlue for prticulr molecule ws cused y proe moility or y unfvorle orienttion tht is, 45 from either polriztion xis we dded two dditionl excittion polriztion xes t 45 nd 135. These conditions llowed us to clculte n dditionl PR mesure, PR 45 135. For n immoile dipole the reltionship etween PR 0 90 nd PR 45 135 is fixed regrdless of the dipole orienttion ecuse of the constnt ngle (45 ) etween the two pirs of xes. An immoile dipole will lwys stisfy the reltionship r = (PR 0 902 + PR 45 1352 ) 1/2 = 1. A more moile dipole nture structurl iology volume 8 numer 6 june 2001 541
Fig. 3 Fluorescence polriztion from single molecules., Imges of xonemes sprsely decorted with KMC2 BSR in the presence of AMP- PNP nd ADP., Fluorescence intensity (ritrry units) records from single molecules. These records correspond to the time course of the emitted intensity of fluorescent spot like the ones present in () for ech polriztion direction of the exciting light. The intensities (ritrry units) of the mesured molecules hve single pek distriution (AMP- PNP: men = 4.7, s.d. = 2.5, n = 161; ADP: men = 3.6, s.d. = 2.3, n = 117). The intensity records show single photoleching event nd tht their vlues hve single pek distriution, indicting tht they come from single molecules. c, Distriution of r vlues for single fluorescent molecules in the presence of AMP-PNP nd ADP. The r vlue mesures the immoility of the individul fluorophore. on the other hnd will hve smller vlues for oth PR mesures, resulting in r < 1; thus, r my e interpreted s mesure of immoility. We used this property to compre the mount of moility in the presence of AMP-PNP nd ADP. Histogrms of the distriution of r vlues (Fig. 3c) show tht the distriution hs much lower r vlues (men = 0.38, n = 117) in the presence of ADP thn with AMP-PNP (men = 0.68, n = 161). This result indictes tht in the presence of ADP the proe hs ecome more moile tht is, dynmiclly disordered on the 100 ms time scle of the mesurement. Therefore, the loss of polriztion nisotropy of the xonemes decorted with mny KMC2 BSR molecules results from the proes ecoming highly moile nd not from sttic disorder or shifting the xil ngle closer to 54.7. Even though the kinesin ADP stte is considered to e the one with the wekest ttchment to microtuules 17, we elieve tht our mesured kinesin molecules re ttched to the xoneml microtuules ecuse: (i) they come from ligned spots (Fig. 3), nd (ii) their r distriution is different from tht for kinesin molecules stuck to the glss (men = 0.52, n = 144 nd men = 0.51, n = 41 in the presence of ADP nd AMP- PNP, respectively). The dynmiclly disordered ADP stte of KMC2 BSR could result from either higher moility of the kinesin loop where the proe is ttched or the whole kinesin motor ecoming moile while still ttched to the microtuule y flexile link. Plcing proes t different loctions could help to distinguish etween these two lterntives. However, in ll kinesin tomic structures solved, the region where we ttched the proe is highly ordered with reltively low B-fctors 18 despite ADP eing in the ctive site. These results rgue ginst the locliztion of the ADP-induced incresed moility to the loop where we ttched the proe. Repeted inding nd dissocition of KMC2 BSR to the xonemes is lso unlikely to cuse the oserved moility. A KMC2 BSR molecule relesed from the xoneme for ny significnt frction of the mesurement time would diffuse wy from the mesured region. Thus, we fvor the interprettion tht the whole motor domin in the presence of ADP ecomes moile while still ttched to the microtuule. Our results for the ADP stte disgree with previous cryo-electron microscopy dt interpreted s evidence for similr configurtions for the microtuule ttched motor domin in severl nucleotide sttes including ADP 2,14,15. However, disordered stte, like the one we detect, would e missed y verging methods tht rely on mny molecules hving equivlent configurtions. The flexile, moile link etween motor nd trck Wht prt of the motor domin might medite this moile link? Ovious cndidtes re the flexile loops present t the c kinesin microtuule interfce. The C-terminl end of tuulin is ner the kinesin inding site nd is disordered in the crystl structure 19. This prt of the tuulin molecule hs een proposed to medite the link etween prticulr type of kinesin motor (KIF1A) nd microtuules, llowing the KIF1A motor to move y ised, one-dimensionl diffusion mechnism 20. In keeping with the proposl tht specific elements of KIF1A re responsile for this prticulr mechnism, our single-molecule records did not show one-dimensionl diffusion of KMC2 BSR during the moile ADP stte. Another disordered element tht could form the flexile link etween the microtuule nd kinesin is loop 11 on the kinesin motor domin 12. Locted t the kinesin microtuule interfce 13, this loop is disordered in the kinesin crystl structure ut is ordered in the minus-end directed motor, ncd 21. We propose tht these flexile loops in kinesin nd tuulin in the ADP stte medite some of the contcts etween kinesin nd the microtuule. The loop flexiility might llow the whole motor to move ck nd forth etween different configurtions. Evidence for flexile link etween motor nd its trck is lso present in the ctomyosin system 22,23, suggesting common occurrence in the chemomechnicl cycle of motor proteins. This flexile interction my llow the motor to initite contct with its trck even when the motor is in n unfvorle position for exmple, due to the inding of the prtner motor hed in the kinesin dimer (Fig. 4, sttes 2 nd 3) or to position mismtches in the muscle ctomyosin lttice. Also, trnsition etween this stte nd stereo specific ttchment my produce forwrd motion (Fig. 4, stte 3 4) in ddition to the one ssocited with the movement of the neck linker (kinesin) (Fig. 4) or lever rm (myosin). There is evidence for suelementl steps from micromechnicl experiments 24,25, nd recent models consider the presence of two steps in the 542 nture structurl iology volume 8 numer 6 june 2001
letters kinesin chemomechnicl cycle 26. We propose tht for kinesin, the structurl difference oserved etween ADP nd other nucleotide sttes identifies sustep in the trnsloction cycle. Methods Proteins nd leling. We mde two-cysteine kinesin construct, KMC2, y modifying pet23 plsmid contining the coding sequence of the first 349 mino cids of the humn kinesin hevy chin gene (KIF5B) with His 6 -tg t the protein C-terminl end. Of the nine nturlly occurring Cys residues, eight were replced y Al residues, leving Cys 174. An extr Cys ws introduced to replce Thr 169. This construct ws expressed in Escherichi coli BL21 cells, nd the KMC2 protein ws purified y Ni-NTA grose chromtogrphy. KMC2 ws leled with the ifunctionl thiol rective fluorophore is-((n-iodocetyl)piperzinyl) sulfonerhodmine (BSR) (Moleculr Proes). Leling ws done overnight t 4 C in 80 mm 1,4-Piperzinediethnesulfonic cid (PIPES), ph 6.8, 2 mm MgCl 2, 1 mm EGTA, 50 µm ATP nd 1:1 molr rtio KMC2:BSR. The rection ws stopped y dding 1 mm dithiothreitol (DTT), nd the unrected dye ws removed y pssing the leled protein through deslting column. We estimted tht 60 75% of the protein ws leled with BSR. The KMC2 BSR showed norml microtuule-ctivted ATPse ctivity (V mx = 55 ATP s 1 hed 1, K m MT = 0.5 µm in 12 mm PIPES, ph 6.8, 2 mm MgCl 2, 1 mm EGTA, 0.5 mm ATP). We verified y mss spectrometry tht BSR rected with KMC2 to yield one fluorophore per KMC2 molecule. Overnight digestion of KMC2 BSR with endoproteinse Lys-C, followed y liquid chromtogrphy mss spectroscopy (LC-MS), showed product with mss corresponding to peptide with the two Cys residues linked to the BSR proe (predicted mss: 2,884.148 D, experimentl mss: 2,884.15 D). Digestion with trypsin (for which there is clevge site t residue Arg etween the cysteines) resulted in product with mss corresponding to the two cysteine-contining peptides hydrolyzed t the middle Arg residue (expected mss increment of 18 D for hydrolysis) ut still joined y the proe crosslink (predicted mss: 2,902.15 D, experimentl mss: 2,902.45 D). These protese digestion/lc-ms Fig. 4 Proposed structurl sttes of kinesin., Conformtions of monomeric kinesin ound to microtuules in the different nucleotide sttes. In the ATP (T) nd ADP-Pi (D-P) sttes the motor hed is rigidly ttched to the microtuule; fter Pi relese the ttchment to the microtuule ecomes flexile in the ADP (D) stte. After ADP relese the motor enters the no-nucleotide stte (no letter), nd the microtuule ttchment ecomes rigid gin. The proe position nd orienttion is indicted y the red rrow; α nd β tuulin re represented respectively in drk nd light lue., Hypotheticl sequence of stepping events of dimeric kinesin motor. Without endorsing ny prticulr kinetic scheme, this model is presented only to show how the structurl sttes we mesured my contriute to the kinesin trnsloction mechnism. A previously proposed conformtionl chnge tht moves the nonttched hed forwrd fter ATP inding 2 is included in the model (stte 1 to 2). In sttes 2 to 3 the moile ADP stte helps the leding hed (yellow) to find tuulin inding site in the forwrd direction. From stte 3 to 4 the leding hed (yellow) releses ADP, enters the rigid stte, pulls the triling hed (gry) nd moves the linkge etween the heds forwrd. From stte 4 to 5 the triling hed (gry) with ADP detches from the microtuule. experiments confirmed tht ech of the two functionl groups of BSR ws linked to Cys residue on KMC2. Axonemes were prepred from se urchin sperm 27. Fluorescence polriztion microscopy. Axonemes were mixed with KMC2 BSR molecules, plced etween two coverslips nd oserved (within 20 30 min period). The experimentl solution is 12 mm PIPES, ph 6.8, 2 mm MgCl 2, 1 mm EGTA, 10 mm glucose, 0.1% (v/v) β-mercptoethnol, 0.1 mg ml 1 ctlse, 0.03 mg ml 1 glucose oxidse nd 7.5 mg ml 1 BSA. Nucleotides were dded depending on the experimentl conditions to e tested (AMP- PNP: 2 mm AMP-PNP; ADP: 2mM ADP; ADP AIF 4 : 4mM ADP, 2 mm AlCl 3, 10 mm KF; no-nucleotide: 5 units ml 1 pyrse with no dded nucleotide). KMC2 BSR protein concentrtion ws 75 nm in the multiple-molecules experiments (Fig. 2) nd 0.75 nm in the single-molecule ones (Fig. 3). The xonemes decorted with KMC2 BSR protein were imged y wide field epifluorescence (ojective: Nikon 100, 1.4 NA, PlnApo, oil immersion) (Fig. 2). Lser light excittion (λ = 532 nm) ws linerly polrized in different trnsverse directions in the plne of the microscope stge with n electro-optic modultor (EOM). The emitted light ws imged vi dichroic mirror (Chrom, DR540LP or DR545LP), ndpss filter (Chrom, 570DF40), 532 nm notch filter (Kiser, supernotch 532) nd 4 rely lens on n intensified, frme-trnsfer CCD cmer (I-Pentmx, Roper Scientific). After reflecting from the dichroic mirror, the intensity polriztion rtios were 45:1, 18:1, 390:1 nd 33:1 for the 0, 45, 90 nd 135 cses, respectively. When four polriztion xes were used, λ / 4 plte (Tower Opticl) ws plced fter the EOM, nd two identicl dichroic mirrors with plnes of reflection perpendiculr to ech other were used to prtilly compenste phse distortions introduced y single dichroic (R. Hochstrsser, pers. comm.). Sequences of imges were recorded with the intensified CCD cmer synchronized to the EOM to collect imges with lternte polriztion directions every 100 ms. To clculte the PR the emitted intensity for ech polriztion excittion xis ws verged first over the time period efore leching (verge = 5 s), nd then the PR ws clculted from the intensity verges. Fluorophore xil ngle. We derived n expression to clculte the PR of n ensemle of fluorophores ound to n xoneme with cylindricl symmetry. For this we ssumed tht the totl light emitted y ech fluorophore is proportionl to the light sored y the fluorophore (I e I cos 2 θ, where θ is the ngle the sorption dipole mkes with the excittion light polriztion xis). We then integrted for ll fluorophores within solid cone of semi ngle Γ, which represents the moility of the fluorophore with respect to the protein to which it is ttched, nd round the surfce of cone of semi ngle β, the verge xil ngle the dipole mkes with the xoneme long xis, to ccount for the cylindricl symmetry of inding sites on the xoneme. The resulting eqution reltes the PR to the ngle ω of the xoneme with the x-xis nture structurl iology volume 8 numer 6 june 2001 543
(defined s the left-right direction in the plne of the microscope smple stge), the dipole xil ngle β nd the moility cone semi ngle Γ: PR 0 90 (Γ,β,ω) = 3cos(2ω) / {1 + 8 / ((3cos 2 β 1) (cosγ + cos 2 Γ))} To estimte the moility cone semi ngle Γ, we used the moility fctor r determined from single-molecule mesurements. To estimte the reltionship etween the r fctor nd the semi ngle Γ we clculted the verge r fctor from simultion of 2,000 rndomly oriented dipoles (uniformly distriuted mong ll possile orienttions in three dimensions) tht were llowed to rotte freely within cone with hlf ngle Γ (ssuming the dipole smples ech position within the cone with equl proility within the mesurement time). The simultions for different vlues of Γ showed tht tht n verge r fctor of 0.7, s in the AMP-PNP cse (Fig. 3c), corresponds to Γ vlue of 32. An r fctor of 0.38 s we found for the ADP stte corresponds to Γ vlue of 53. Acknowledgments We thnk R. Skowicz for xoneme preprtions, dvice nd discussions; R. Dickson, R. Vle nd D. Pierce for their dvice in the initil phses of this project; L. Gross, H. Deng nd L. Siconolfi-Bez for mss spectrometry nlysis; A. Asenjo for iochemicl ssys, nd S. Brsselet nd B. Lounis for helpful discussions nd experimentl suggestions. L.S.B. Goldstein is n investigtor of the Howrd Hughes Medicl Institute. This project ws supported y NSF grnts. Correspondence should e ddressed to H.S. emil: hsos@ecom.yu.edu Received 10 Novemer, 2000; ccepted 27 Mrch, 2001. 1. Goldstein, L.S.B. & Philp, A.V. Annu. Rev. Cell Dev. Biol. 15, 141 183 (1999). 2. Rice, S. et l. Nture 402, 778 784 (1999). 3. Svood, K., Schmidt, C.F., Schnpp, B.J. & Block, S.M. Nture 365, 721 727 (1993). 4. Rosenfeld, S.S., Correi, J.J., Xing, J., Rener, B. & Cheung, H.C. J. Biol. Chem. 271, 30212 30221 (1996). 5. Moerner, W.E. & Orrit, M. Science 283, 1670 1676 (1999). 6. Weiss, S. Science 283, 1676 1683 (1999). 7. Lu, H.P., Xun, L. & Xie, X.S. Nture 282, 1877 1882 (1998). 8. H, T., Lurence, T.A., Cheml, D.S. & Weiss, S. J. Phys. Chem. B 103, 6839 6850 (1999). 9. Adchi, K. et l. Proc. Ntl. Acd. Sci. USA 97, 7243 7247 (2000). 10. Wrshw, D.M. et l. Proc. Nt. Acd. Sci. USA 95, 8034 8039 (1998). 11. Corrie, J.E.T. et l. Nture 400, 425 430 (1999). 12. Kull, F.J., Slin, E.P., Lu, R., Fletterick, R.J. & Vle, R.D. Nture 380, 550 555 (1996). 13. Sos, H. et l. Cell 90, 217 224 (1997). 14. Arnl, I. & Wde, R.H. Structure 6, 33 38 (1998). 15. Hirose, K., Lowe, J., Alonso, M., Cross, R.A. & Amos, L.A. Mol. Biol. Cell 10, 2063 2074 (1999). 16. Funtsu, T., Hrd, Y., Tokung, M., Sito, K. & Yngid, T. Nture 374, 555 559 (1995). 17. M, Y.Z. & Tylor, E.W. J. Biol. Chem. 272, 724 730 (1997). 18. Sck, S., Kull, F.J. & Mndelkow, E. Eur. J. Biochem. 262, 1 11 (1999). 19. Nogles, E., Whittker, M., Millign, R.A. & Downing, K.H. Cell 96, 79 88 (1999). 20. Okd, Y. & Hirokw, N. Science 283, 1152 1157 (1999). 21. Slin, E.P., Kull, F.J., Cooke, R., Vle, R.D. & Fletterick, R.J. Nture 380, 555 559 (1996). 22. Tylor, K.A. et l. Cell, 99, 421 431 (1999). 23. Wlker, M., Zhng, X.Z., Jing, W., Trinick, J. & White, H.D. Proc. Nt. Acd. Sci. USA 96, 465 470 (1999). 24. Coppin, C.M., Finer, J.T., Spudich, J.A. & Vle, R.D. Proc. Nt. Acd. Sci. USA 93, 1913 1917 (1996). 25. Veigel, C. et l. Nture 398, 530 533 (1999). 26. Schnitzer, M.J., Visscher, K. & Block, S.M. Nture Cell Biol. 2, 718 723 (2000). 27. Gions, I.R. & Fronk, E. J. Biol. Chem. 254, 187 196 (1979). 28. Kozielski, F. et l. Cell 91, 985 994 (1997). 29. Kozielski, F., Arnl, I. & Wde, R.H. Curr. Biol. 8, 191 198 (1998). 30. vn Amerongen, H. & Struve, W.S. Methods Enzymol. 246, 259 283 (1995). 544 nture structurl iology volume 8 numer 6 june 2001