Actincores of hair-cell stereocilia supportmyosin motility · Proc. Natl. Acad. Sci. USA87(1990) 8629 cores in a matter of seconds. Beads reaching the cores by either method bound

Proc. Nati. Acad. Sci. USAVol. 87, pp. 8627-8631, November 1990Cell Biology

Actin cores of hair-cell stereocilia support myosin motility(adaptation/cytoskeleton/optical tweezers/auditory system/vestibular system)

GORDON M. G. SHEPHERD*t, DAVID P. COREY*t, AND STEVEN M. BLOCKt§*Neuroscience Group, Howard Hughes Medical Institute, and Department of Neurology, Massachusetts General Hospital, Boston, MA 02114; tProgram inNeuroscience, Harvard Medical School, Boston, MA 02115; tRowland Institute for Science, 100 Cambridge Parkway, Cambridge, MA 02142; and§Department of Cellular and Developmental Biology, Harvard University, Cambridge, MA 02138

Communicated by Howard C. Berg, July 18, 1990 (receivedfor review May 29, 1990)

ABSTRACT The actin cores of hair-cell stereocilia weretested as a substrate for the movement of myosin-coated beadsin an in vitro assay. Large numbers of stereocilia from bullfrogsacculi and semicircular canals were isolated by blotting ontocoverglasses and were demembranated to expose the polaractin tracks of their cytoskeletal cores. Silica or polystyrenebeads, coated with thick filaments of chicken skeletal musclemyosin, were added to this core preparation in the presence ofATP. Myosin-coated beads could reach some of the cores bydiffusion alone, but the efficiency and precision of the assaywere improved considerably by the use of "optical tweezers"(a gradient-force optical trap) to deposit the beads directly onthe cores. Beads applied in this fashion bound and movedunidirectionally at 1-2 ,zm/s, escaping the retarding force ofthe trap. Actin filaments within the stereocilia are cross-linkedby fimbrin, but this did not appear to interfere with the motilityof myosin. Beads coated with optic-lobe kinesin were also testedfor movement; these bound and moved unidirectionally at0.1-0.2 ,.m/s when applied to microtubule-based kinociliarycores, but not when applied to actin-based stereociliary cores.Our results are consistent with, and lend support to, a modelfor hair cell adaptation in which a molecular motor such asmyosin maintains tension on the mechanically gated transduc-tion channels. Optical tweezers and video-enhanced differen-tial interference contrast optics provide high efficiency andimproved optical resolution for the in vitro analysis of myosinmotility.

Hair cells, the receptor cells of the vertebrate inner ear,transduce mechanical displacements of their hair bundlesinto electrical signals. The hair bundle ofeach cell consists of30-300 actin-based extensions, called stereocilia, and a singlemicrotubule-based kinocilium. The one or two transductionchannels at the tip of each stereocilium (1, 2) are gated bydisplacements of the bundle (3, 4). Tension stimuli arethought to reach the channels via fine filaments extendingfrom the tip of each stereocilium upward to the side of itstaller, neighboring stereocilium (reviewed in ref. 5). Ultra-structurally, these filaments have been described as "tiplinkages" (6); physiologically, they behave as elastic "gatingsprings" (7).The resting tension of the gating springs, and hence the

channel open probability, is regulated by an adaptationprocess (8-10). A maintained displacement of the hair bundletoward its taller side elicits a depolarizing current thatdeclines, over tens of milliseconds, toward the resting level.Similarly, displacement in the opposite direction elicits arapid hyperpolarizing reduction in the transduction current,followed by a slower recovery to the resting level (9). Thisrecovery suggests that the gating springs are actively reten-sioned. Assad et al. (11) found that voltage changes that alter

calcium entry through transduction channels induce activebundle movements. These movements have a magnitude,direction, and time course similar to adaptation, suggestingthat the two phenomena are generated by the same activeprocess (11, 12). It has been proposed that a myosin-likemolecule, attaching the tip linkage to the side of the tallerstereocilium and moving along the actin core, could serve asthe putative adaptation motor. Such a motor, attempting tomove toward the tip of the stereocilium, would maintaintension in the tip linkage (7, 10).The mechanoenzymatic properties ofdifferent myosins are

intrinsic and essentially independent of the type of actinfilament on which they move in in vitro experiments (13);actins, which are highly conserved, are equally effectivesubstrates for all myosins. Within a cell, however, differentgroups of actin filaments can specialize in particular struc-tural and motile functions by interacting with one or more ofa variety of actin-binding proteins. Little is known of theeffects of these proteins on actomyosin activity in nonmusclecells. Stereociliary actin filaments are extensively cross-linked by fimbrin (14-18), an actin-bundling protein firstidentified in microvilli and microspikes (19-21). It was notclear whether this would inhibit the myosin movement pro-posed to mediate adaptation.To address this question, we modified existing in vitro

assays for myosin (22, 23), incorporating differential inter-ference contrast (DIC) optics, "optical tweezers" (24-27),and stereociliary cores as the actin substrate. Actin coreswere obtained by demembranation of hair bundles (28) iso-lated by the "bundle blot" procedure (17), preparations thathave been studied ultrastructurally and biochemically. Op-tical tweezers use an infrared laser beam directed through themicroscope to capture and manipulate objects by means ofradiation pressure-in this case, silica beads coated withmyosin thick filaments. We found that stereociliary actincores were a competent substrate for myosin-based motility,whereas kinociliary cores, containing microtubules, were acompetent substrate for kinesin-based motility. Thus, a my-osin-like motor remains an attractive mechanism for adap-tation in hair cells.

MATERIALS AND METHODSBlotting and Demembranation of Stereocilia. Bullfrogs

(Rana catesbeiana) were purchased from Ming's Market(Boston). They were pithed and decapitated, and their sacculiand semicircular canals were removed. Dissection was per-formed in a cold saline solution (100 ,M Ca2+/120 mMNaCl/2 mM KCI/3 mM dextrose/5 mM Hepes, pH 7.2) withprotease inhibitors (0.15 AM aprotinin/20 AM leupeptin/0.15 ,uM pepstatin/0.15 mM phenylmethylsulfonyl fluoride;Boehringer Mannheim). The apical surfaces of the sensoryepithelia were exposed by mechanical removal of the oto-

Abbreviation: DIC, differential interference contrast.

8627

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

9, 2

020

8628 Cell Biology: Shepherd et al.

lithic membrane or cupula. Stereocilia were isolated byblotting lightly onto coverglasses, which had been cleanedwith ethanolic KOH and treated for 1-2 min with poly-L-lysine or poly-L-ornithine (0.1-1.0 mg/ml in H20; 150-350kDa; Sigma).

Isolated stereocilia were demembranated in a 1% solutionof Triton X-100 in motility buffer (50 mM KCI/25 mMimidazole/4 mM MgCl2/1 mM EGTA, pH 7.4) for 5 min andthen washed in motility buffer. Residual adhesive sites on theglass were blocked with bovine serum albumin (0.5 mg/ml inmotility buffer; ICN). For fluorescence microscopy, filamen-tous actin was labeled with Bodipy phallacidin (50 units/ml;Molecular Probes) as described in ref. 17.Bead Preparation. Chicken skeletal myosin was prepared

according to ref. 29 and stored at -20'C in 0.5 M KCl/10 mMimidazole, pH 7.0/10 mM EDTA/0.1 mM dithiothreitol/50% (vol/vol) glycerol. All other myosin solutions contained1 mM dithiothreitol. Thick filaments were formed in low ionicstrength (25 or 50 mM KCl/10 mM imidazole, pH 7.2) andadsorbed onto Covaspheres (1.0-,um diameter; Duke Scien-tific, Palo Alto, CA) or silica beads (0.4-,um-diameter silica,prepared by Howard Berg by the procedure given in ref. 30).The ability of each myosin preparation to support movementwas confirmed with the Nitella assay (22, 23). For assays withoptical tweezers and stereociliary cores, it was important forbeads to be monodisperse and not clustered in the largeaggregates typically observed in the Nitella assay. This wasachieved by using a 1:10 dilution of silica beads incubatedwith a 1:100 dilution of thick filaments for 10-30 min on ice,with gentle agitation. Bead solutions were triturated beforeand after dilution. The myosin-coated beads were then di-luted 1:4 in motility buffer with 1 mM ATP and 0.5 M sucrosefor use in the assay.

Kinesin-coated beads were similarly prepared by coating0.4-,um-diameter silica beads with squid optic-lobe kinesin(31). Kinesin was prepared as described in ref. 32.Chamber Construction. The assay chamber was made from

a 0.5-mm-thick stainless steel slide with a 16-mm-diameterhole in its center. A circular coverglass carrying demem-branated cores was affixed with Apiezon M grease (VWRScientific) to the underside of the slide and centered over thehole, with care being taken to keep the experimental prepa-ration immersed in buffer. A Teflon spacer ring (16-mm i.d.;1.0-mm thick) was mounted with grease to the top of the slide,over the hole (giving a chamber vol of 300 ul), and motilitybuffer with 1 mM ATP was added. To assay motility, 1-3 ,uof the bead preparation was introduced into the chamber,which was then sealed by mounting a coverglass to the top.

Optical Tweezers. A single-beam gradient-force optical trap(optical tweezers) was created by passing the beam from acontinuous-wave Nd:YAG laser (output power, 1 W; TEMOOmode; wavelength, 1064 nm; C-95 YAGMAX, CVI/Laser,Albuquerque, NM) through the objective of an invertedmicroscope. Salient features of the design may be found inref. 27. The x-y-z position of the trapping zone in themicroscope was controlled by means of external optics; thediameter of the trapping zone was fixed at -1 ,m, corre-sponding to the diameter of a diffraction-limited spot. Thesystem was equipped with a variable attenuator and shutterto adjust the trapping force and to release the trap, respec-tively. Trapping force was calibrated against Stokes' drag(26). Trap alignment and position were monitored by imagingthe laser spot, arising from partial reflection of the light at thecoverglass-water interface, on a video camera sensitive tothe near infrared (extended-red Ultracon tube, series 75;Dage-MTI, Michigan City, IN). The same video camera wasused to observe the microscope image at visible wavelengths.

Imaging. The specimen chamber was viewed with anoil-immersion condenser [numerical aperture (n.a.), 1.4] andoil-immersion objective (ICS 100x Plan-Neofluar; n.a., 1.3;

Zeiss) on an inverted microscope equipped with DIC optics(Axiovert 35; Zeiss); this same objective served to form theoptical trap. Images were taken by video camera and storedwith a videocassette recorder (AG-6300; Panasonic, VHSformat). Bead speeds were measured by using a semiauto-matic computer-based system that acquires bead position andvideo frame count (33). For high-resolution tracking of beadmovements, video images were first transferred to an opticalmemory disk recorder (TQ-2028F; Panasonic), and thenanalyzed by software that performs centroid analysis of thebead image on a frame-by-frame basis using a cross-correlation algorithm [Image-1, Universal Imaging (Media,PA) and custom software; for methods, see ref. 34]. Beadcoordinates were fit to a linear trajectory that minimized theleast-squares perpendicular distance of each point to the line.Centroid analysis gives subpixel resolution, corresponding topositional errors of -20 nm for the bead sizes and magnifi-cations used in this study.

RESULTSActin Cores. To monitor the core preparation procedure,

blotted and demembranated stereocilia were labeled with afluorescent actin probe (Fig. 1). They appeared as hundredsof long straight rods over an area of 0.5-1 mm2. Treatment ofthe coverglass with poly-L-ornithine or poly-L-lysine in-creased the yield of the isolation technique manyfold overblotting onto untreated glass. Contamination by cuticularplates and zonulae adherentes, the other actin-containingorganelles at the hair-cell apical surface, was minimal. Whenthe same fields were imaged with DIC optics, the sterociliawere seen to be straight rods =0.4 ,um wide and 5-100 ,umlong. Often, stereociliary polarity could be discerned by thebasal tapering and blunt tip, or by the geometry of a blottedbundle, in which tips tended to stay together and basestended to splay apart. Kinocilia were identified by theircharacteristic curved appearance, smaller diameter, and of-ten the presence of the kinociliary bulb (Fig. 1c).Myosin Motility. Myosin-coated beads at low density,

diffusing through the medium, took many minutes to reachthe cores. In contrast, micromanipulation of individual beadswith optical tweezers enabled direct placement on selected

I

FIG. 1. Actin cores from saccular stereocilia (a and b) andsemicircular canal stereocilia (c and d) were viewed with DIC optics(a and c) and with fluorescence optics (b and d). The fluorescent labelwas Bodipy phallacidin. For many stereocilia, orientation was dis-cerned from the basal tapering (arrows in a and b) and blunt tips.Stereocilia (sc) and bulb-bearing kinocilia (kc) were easily distin-guished by morphological features with DIC optics. Kinocilia werenot labeled by the fluorescent actin probe. (Bar = 5 gm.)

Proc. Natl. Acad. Sci. USA 87 (1990)

Dow

nloa

ded

by g

uest

on

Aug

ust 1

9, 2

020

Proc. Natl. Acad. Sci. USA 87 (1990) 8629

cores in a matter of seconds. Beads reaching the cores byeither method bound and moved for distances of 1-10 ,gm.Fig. 2a shows a bead being applied to a semicircular canalcore by optical tweezers, then binding, escaping the lasertrap, and moving several micrometers along the core. Thisescape was anticipated; our optical tweezers produced410 pN of force on a 0.4-,tm-diameter silica bead, while athick filament propelling a bead may have 10-100 activemyosin heads, each producing 1-10 pN of force (35).With optical tweezers, the substrate specificity of mecha-

noenzyme-coated beads was readily tested. Kinesin-coatedbeads, deposited by optical tweezers on kinociliary cores,moved at rates of 0.1-0.2 ,tm/s (Fig. 2b), similar to ratesobtained with the same kinesin preparation on sea urchin

axonemes (26). Kinesin-coated beads moved only on kinocil-iary cores, and myosin-coated beads moved only on stereo-ciliary cores.Bead movement could not be attributed to any stereociliary

mechanoenzymes. Beads that moved did so at rates charac-teristic of chicken skeletal muscle myosin, while beadscoated either with kinesin or with small amounts of myosin,at concentrations below the threshold for movement, failed topick up endogenous motors and move on stereociliary cores.Bead movements were unidirectional, toward what were

identified as the tips of the stereocilia. Fig. 3 shows threebeads that moved along stereociliary tracks, joining previ-ously accumulated beads near the top end of the bundle.Beads that moved to the ends of cores always remainedbound, not releasing from the actin. The reason for thisbinding was unclear. Possibly, it involved other structures atthe tips of stereocilia.The speed of myosin-coated beads moving on cores aver-

aged 1-2 ,m/s, the same range as speeds obtained with thesame myosin preparation in Nitella assays (data not shown).Speeds were the same whether beads reached the cores bydiffusion alone or were delivered by optical tweezers, sug-gesting that the laser light did nothing to impair myosinmotility. Once moving, a motile bead did not appear to behalted or even slowed by optical tweezers. Some beadsexhibited smooth, uniform movement, while others slowedor stopped briefly.The fine structure of the motion was analyzed by frame-

by-frame centroid tracking of bead position. The myosin-coated bead of Fig. 2a moved in a nearly straight trajectoryalong the core (Fig. 4a). A plot of the component of thismotion parallel to the trajectory against time showed varia-tion in the speed of the bead (Fig. 4b). It paused briefly, ranfor a stretch with some variation in speed (peak speed,1.5 ,um/s), then paused and started again. A plot of themotion perpendicular to the trajectory against time showedonly minimal side-to-side deviations (10-50 nm), until thefinal 0.5 s, when the direction changed slightly (Fig. 4c).

FiG. 2. (a) A myosin-coated bead (arrow) was deposited byoptical tweezers on a stereociliary core. Successive images (top tobottom) show the bead moving out ofthe trap (seen as the bright spotin the first three images) and along the core for several micrometers,at a speed of -1.5 ,um/s. The bead eventually stopped near the finalposition shown. (b) A kinesin-coated bead (arrow) was deposited bythe optical tweezers on a kinociliary core. Successive images (top tobottom) show the bead moving along the core for a distance of severalmicrometers, at a speed of -0.1 ,Am/s. The trap can be seen pickingup another bead (bright spot) in the final two images.

FIG. 3. Three myosin-coated beads (arrows) moved from thebases (a) to the tips (b) of saccular stereociliary cores,joining clustersof beads already present at the tips.

Cell Biology: Shepherd et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

9, 2

020

8630 Cell Biology: Shepherd et al.

E

.C-a)0Ea)0c

0

-aFucua.

600 -

400 F

200

O_-5-500

2500

2000 -

1500 -

1000 _

500

0-

-500 --0.5

100

7_ 50

a)

-50

0 500

0.0 0.5

-0.5 0.0 0.5

FIG. 4. High-resolution centroid

60 consecutive video frames fromthe

shown in Fig. 2a. (a) The bead's pos

was plotted for each frame; thetr

parallel to the axis of the core. (b)'

parallel to the best-fit linear trajecArrows demarcate a segment of relatcomponent of bead position perpend

plotted against time (note the expat

Since the width of the actin

appeared that the myosin thick f

over only one (or at most a few)

the core during travel.Under optimal conditions, be

the tops of bundles, an indicatii

moved. In many experiments, Iobserved. Sometimes beads bou

out moving, an effect that wasthick-filament/bead ratio. Some

cifically, in which case increasihconcentration or incubation tim(

ing of adhesive sites. Doubling

buffer, from 25 to 50 mM KCI,

although motility was observedThe reproducibility of the assa:

ness of the myosin and stere

presence of protease inhibitors

DISCUS

An Actin-Based Motor inStew

dence implicate a myosin-like nMeasured physiologically, the

hyperpolarizing bundle displac4

The retensioning rate can be ac

the attachment point of the tip

of a stereocilium, at a speed of]

the speeds of myosins. The f

transduction channels, determ

bundle stiffness, are in the 1-5

the forces thought to be generated by a single myosin head(35).The polarity of movement is also consistent with the

model. Myosins move unidirectionally on actin filaments,, ,

toward their barbed ends (36). The actin filaments of stere-1000 1500 2000 2500 ocilia are uniformly oriented with their barbed ends at the tipsX (nm) (16,28, 37). A myosin moving on stereociliary actin filaments

would therefore be expected to climb toward the tip, as we

have observed here; this is the direction required for ten-sioning in the hair cell adaptation model.

Microvilli of the intestinal brush border also have a polaractin core, laterally linked to the membrane by a myosin (38,39). This brush border myosin I (38), unlike skeletal musclemyosin and other myosins II, resembles other myosins I(reviewed in ref. 40) in having a short tail. This tail isapparently specialized for functions other than filament for-mation, such as binding to membranes. Brush border myosin, . ,

~ I also binds calmodulin (41) and displays Ca2+/calmodulin-1.0 1.5 2.0 2.5 dependent motility on actin in vitro (38, 39). Although its

Time (s) physiological role is currently unknown, the presence of amyosin in a structure so similar to stereocilia isintriguing., . , . , .However, brush border myosin I itself does not seem to bethe actin-to-membrane linker of stereocilia. Gels of purifiedstereociliary cores lack a band near the expected mass of110 kDa (17). Brush border myosin I requires ATP for elution, , , .from demembranated microvillar cores (42), but the stereo-

1.0 1.5 2.0 2.5 ciliary actin-to-membrane linkers do not (17, 28).Time (s) Evidence for other myosins in stereocilia is equivocal.

Labeling of stereocilia with antibodies to smooth muscletracking of bead position, using myosin has been reported (43) but disputed on the basis of

e myosin-coated bead movement light-piping artifacts (44) and on the failure of labeling byition in the x-y (specimen) plane several anti-myosin antibodies (45). However, the latteraJectory is along a straight line results may simply reflect lack of antibody cross-reactivityThe component of bead positionctory plotted against time. among myosins. Inasmuch as the adaptationmotor may be

tively smooth movement. (c) The present in hair bundles at a concentration of only a few

licular to the same trajectory was molecules per stereocilium, conventional gel and immuno-nded scale). cytochemical methods may be inadequate for detection. It is

interesting to note, however, that calmodulin appears local-substrate was =300 nm, it ized to the tips (17).l1ament(s) remained centered Our results show that stereociliary actin is available as a

of the actin filaments within substrate for myosin motility and thereby provide circum-stantial evidence supporting the proposal that an actin-based

-ads were found primarily at motor is present in stereocilia.on that almost all beads had Fimbrin and Myosin Motility. Some actin-binding proteinshowever, no movement was inhibit actin-myosin interactions. The 55-kDa protein thatind specifically to cores with- bundles actin filaments in the acrosomal process of Limulus

s reduced by optimizing the sperm prevents binding of myosinS-1 fragments to the actinetimes beads bound nonspe- bundle (46). Villin, at certain concentrations, inhibits acto-ng the bovine serum albumin myosin ATPase activity (47). However, other observationsD seemed to give better block- indicate that myosin motility is compatible with the presence:he ionic strength of the assay of some actin-binding proteins. Nitella actin filaments, bun-ielped for some experiments, dled into cables by an uncharacterized protein, supportunder both concentrations. myosin motility (22, 23), as do actin filaments bound by

y was sensitive to the fresh- phalloidin (13). The fimbrin-bundled actin cores of microvilliocilia preparations and the and stereocilia can be decorated by myosinS-1 fragmentsin the dissection buffer. (and by brush border myosinI, in the case of microvilli) but

myosin motility along these cores has not been previously'SION reported (37, 48).

Our results show that myosin can bind and move alongreoci ia? Several lines of evi- stereociliary actin filaments in the presence of actin-bindingnotor in hair cell adaptation. proteins. The actin cores of saccular stereocilia have arate of retensioning after a relatively simple biochemical composition, consisting pri-ement is relatively constant. marily of actin and fimbrin, with several other proteins,counted for by movement of present in lower abundance (17). We cannot rule out thei

nkage upward, along the side possibility that myosin movement in our assay was along bare[-2 g.m/s (10), comparable to actin filaments that had become partially dissociated from theForces required to open the cores. However, the ionic conditions favored the associationined from measurements of of fimbrin with actin (20, 21, 28), and the cores, in additionpN range (1), comparable to to containing fimbrin when assayed electrophoretically (17),

a

b

I,

r-

Proc. Natl. Acad. Sci. USA 87 (1990)

I

c

Dow

nloa

ded

by g

uest

on

Aug

ust 1

9, 2

020

Proc. Natl. Acad. Sci. USA 87 (1990) 8631

did not have the characteristic, curved morphology of "de-fimbrinated" stereociliary cores (28).A Myosin Assay. In addition to demonstrating the move-

ment of myosin along stereociliary actin cores, these exper-iments illustrate the application of two useful technologies toin vitro motility assays for myosin. Using optical tweezers asa micromanipulator for bead delivery improved the efficiencyof the assay, enabling precise placement (and release) ofbeads onto the substrate. This feature has recently beenexploited in motility assays for kinesin to study movementbased on single kinesin molecules (31). An optically thinpreparation of actin bundles that allowed video-enhancedDIC optics facilitated precise measurements of position:good spatial and temporal resolution are more easily achievedwith refractile beads and DIC optics than with fluorescentactin filaments and low light-level cameras (e.g., see ref. 13).DIC optics were previously used to visualize endogenousmotor activity on Chava actin cables by Kachar (49). Weanticipate that the combination of these technologies, cou-pled with the relatively simple biochemical components ofthis assay, will prove helpful in the further study of actin-based motors.

We thank Bruce Schnapp for advice and help with kinesin assaysand tracking analysis. Dan Kiehart and Tung-Ling Chen generouslyprovided lab space and advice for myosin purification. Howard Bergkindly donated silica beads. We thank Bechara Kacharfor suggestingpoly-L-ornithine and for stimulating our interest in motility assays.This work was supported by grants from the National Institutes ofHealth (NS-22059) and the Office of Naval Research (754-90924) toD.P.C., and by the Rowland Institute for Science (S.M.B.). D.P.C.is an Associate Investigator ofthe Howard Hughes Medical Institute.

1. Howard, J. & Hudspeth, A. J. (1988) Neuron 1, 189-199.2. Hudspeth, A. J. (1982) J. Neurosci. 2, 1-10.3. Corey, D. P. & Hudspeth, A. J. (1979) Biophys. J. 26, 499-506.4. Corey, D. P. & Hudspeth, A. J. (1983b) J. Neurosci. 3, 962-

976.5. Hudspeth, A. J. (1989) Nature (London) 341, 397-404.6. Pickles, J. O., Comis, S. D. & Osborne, M. P. (1984) Hearing

Res. 15, 103-112.7. Howard, J. & Hudspeth, A. J. (1987) Proc. Nati. Acad. Sci.

USA 84, 3064-3068.8. Corey, D. P. & Hudspeth, A. J. (1983a) J. Neurosci. 3, 942-

961.9. Eatock, R., Corey, D. P. & Hudspeth, A. J. (1987)J. Neurosci.

7, 2821-2836.10. Hacohen, N., Assad, J. A., Smith, W. & Corey, D. P. (1989) J.

Neurosci. 9, 3988-3997.11. Assad, J. A., Hacohen, N. & Corey, D. P. (1989) Proc. Nati.

Acad. Sci. USA 86, 2918-2922.12. Assad, J. A. & Corey, D. P. (1990) Soc. Neurosci. Abstr. 16,

1078.13. Kron, S. J. & Spudich, J. A. (1986) Proc. Nati. Acad. Sci. USA

83, 6272-6276.14. Tilney, L. G., DeRosier, D. J. & Mulroy, M. (1980) J. Cell Biol.

86, 244-259.15. Flock, A., Bretscher, A. & Weber, K. (1982) Hearing Res. 6,

75-89.

16. Hirokawa, N. & Tilney, L. G. (1982) J. Cell Biol. 95, 249-261.17. Shepherd, G. M. G., Barres, B. A.& Corey, D. P. (1989) Proc.

Natl. Acad. Sci. USA 86, 4973-4977.18. Tilney, M. S., Tilney, L. G., Stephens, R. E., Merte, C.,

Drenckhahn, D., Cotanche, D. A. & Bretscher, A. (1989) J.Cell Biol. 109, 1711-1723.

19. Bretscher, A. & Weber, K. (1980) J. Cell Biol. 86, 335-340.20. Glenney, J. R., Kaulfus, P., Matsudaira, P. & Weber, K. (1981)

J. Biol. Chem. 256, 9283-9288.21. Bretscher, A. (1981) Proc. Natl. Acad. Sci. USA 78, 6849-

6853.22. Sheetz, M. P. & Spudich, J. A. (1983) Nature (London) 303,

31-35.23. Sheetz, M. P., Chasan, R. & Spudich, J. A. (1984) J. Cell Biol.

99, 1867-1871.24. Ashkin, A. & Dziedzic, J. M. (1987) Science 235, 1517-1520.25. Ashkin, A., Dziedzic, J. M. & Yamane, T. (1987) Nature

(London) 330, 769-771.26. Block, S. M., Blair, D. F. & Berg, H. C. (1989) Nature (Lon-

don) 338, 514-518.27. Block, S. M. (1990) in Noninvasive Techniques in Cell Biology,

eds. Grinstein, S. & Foskett, K. (Wiley/Liss, New York), pp.375-401.

28. Neugebauer, D.-C. & Thurm, U. (1984) J. Neurocytol. 13,797-808.

29. Kielley, W. W. & Harrington, W. F. (1960) Biochim. Biophys.Acta 41, 401-421.

30. Stoeber, W., Fink, A. & Bohn, E. (1968) J. Colloid InterfaceSci. 26, 62-69.

31. Block, S. M., Goldstein, L. S. B. & Schnapp, B. J. (1990)Nature (London), in press.

32. Vale, R. D., Reese, T. S. & Sheetz, M. P. (1985) Cell 42,39-50.

33. Sheetz, M. P., Block, S. M. & Spudich, J. A. (1986) MethodsEnzymol. 134, 531-544.

34. Gelles, J., Schnapp, B. J. & Sheetz, M. P. (1987) Nature(London) 331, 450-453.

35. Huxley, A. F. & Simmons, R. M. (1971) Nature (London) 233,533-538.

36. Huxley, H. E. (1969) Science 164, 1356-1366.37. Flock, A. & Cheung, H. C. (1977) J. Cell Biol. 75, 339-343.38. Mooseker, M. S. & Coleman, T. R. (1989) J. Cell Biol. 108,

2395-2400.39. Collins, K., Sellers, J. R. & Matsudaira, P. (1990) J. Cell Biol.

110, 1137-1147.40. Korn, E. D. & Hammer, J. A. (1988) Annu. Rev. Biophys.

Biophys. Chem. 17, 23-45.41. Glenney, J. R. & Weber, K. (1980) J. Biol. Chem. 255, 10551-

10554.42. Matsudaira, P. & Burgess, D. R. (1979) J. Cell Biol. 83,

667-673.43. Macartney, J. C., Comis, S. D. & Pickles, J. 0. (1980) Nature

(London) 288, 491-492.44. Egelman, E. (1981) Nature (London) 294, 674.45. Drenckhahn, D., Kellner, J., Mannherz, H. G., Groschel-

Stewart, U., Kendrick-Jones, J. & Scholey, J. (1982) Nature(London) 300, 531-532.

46. Tilney, L. G. (1975) J. Cell Biol. 64, 289-310.47. Coleman, T. R. & Mooseker, M. S. (1985) J. Cell Biol. 101,

1850-1857.48. Mooseker, M. & Tilney, L. G. (1975) J. Cell Biol. 67, 725-743.49. Kachar, B. (1985) Science 227, 1355-1357.

Cell Biology: Shepherd et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

9, 2

020