Báo cáo khoa học: Molecular organization and force-generating mechanism of dynein pot

After more than 40 years of investigation since the discovery of dynein, significant breakthroughs have been achieved: the microtubule-binding domain MTBD [68] and motor domains of yeast

Molecular organization and force-generating mechanism

of dynein

Hitoshi Sakakibara1and Kazuhiro Oiwa1,2

1 National Institute of Information and Communications Technology, Kobe, Japan

2 Graduate School of Life Science, University of Hyogo, Japan

Introduction

A high molecular weight ATPase extracted from

Tetra-hymenacilia was the first microtubule-based

force-gen-erating ATPase to be discovered [1] It was named

‘dynein’ after the cgs unit of force, the dyne [2]

Dynein is now known to consist of a functionally

diverse family of proteins, the members of which are

involved in a wide variety of essential cellular

func-tions in various cells There are two major functional

classes of dynein: axonemal and cytoplasmic dyneins

Axonemal dyneins are further classified into two

sub-classes, outer-arm and inner-arm dyneins, based on

their localization in the axoneme, while cytoplasmic

dynein contains two subclasses, dynein-1 and dynein-2

[3] The latter is reported to be involved in intraflagel-lar transport, which is a bidirectional transport of par-ticles along axonemes in cilia and flagella Although discrimination into these classes was originally based

on function and localization, phylogenetic analyses of full-length dynein heavy chain sequences have con-firmed the existence of differences among the various dyneins, and nine classes of dyneins (two cytoplasmic, two outer-arm and five inner-arm) have been identified [4,5]

The axonemal dynein is responsible for generating the force required to drive movement of cilia and fla-gella, while the cytoplasmic dynein is responsible for

Keywords

dynein; intracellular transport; microtubules;

molecular motor; processivity; retrograde

transport; single-molecule nanometry

Correspondence

K Oiwa, National Institute of Information

and Communications Technology, Advanced

ICT Research Center, 588-2 Iwaoka,

Nishi-ku, Kobe 6512492, Japan

Fax: +81 78 969 2119

Tel: +81 78 969 2110

E-mail: oiwa@nict.go.jp

(Received 8 February 2011, revised 20 May

2011, accepted 1 July 2011)

doi:10.1111/j.1742-4658.2011.08253.x

Dynein, which is a minus-end-directed microtubule motor, is crucial to a range of cellular processes The mass of its motor domain is about 10 times that of kinesin, the other microtubule motor Its large size and the diffi-culty of expressing and purifying mutants have hampered progress in dynein research Recently, however, electron microscopy, X-ray crystallog-raphy and single-molecule nanometry have shed light on several key unsolved questions concerning how the dynein molecule is organized, what conformational changes in the molecule accompany ATP hydrolysis, and whether two or three motor domains are coordinated in the movements of dynein This minireview describes our current knowledge of the molecular organization and the force-generating mechanism of dynein, with emphasis

on findings from electron microscopy and single-molecule nanometry

Abbreviations

BFP, blue fluorescent protein; FRET, Fo¨rster resonance energy transfer; GFP, green fluorescent protein; HC, heavy chain; IC, intermediate chain; LC, light chain; LIC, light intermediate chain; MTBD, microtubule binding domain; Tctex1, T-complex testis-specific protein 1.

intracellular transport, in which a wide variety of

car-gos including mRNA, receptor proteins, mitochondria

and several vesicles are transported along microtubule

tracks in cells (reviewed in [6–8]) Surprisingly, recent

studies have indicated that some viruses use the

plasmic dynein for their translocations in the

cyto-plasm following cell entry [9–11] (reviewed in [12]; also

see the minireview in this volume [13]) The

cytoplas-mic dynein also plays important physiological roles in

the maintenance of the Golgi apparatus [14,15], in

endosome recycling, in cytokinesis [15], in chromosome

separation during mitosis, and in the assembly and

maintenance of cilia and flagella [16] The roles of

subunits of cytoplasmic dynein in distinct

membrane-trafficking processes have gradually been revealed

through RNA interference and cell imaging techniques

[15]

Despite their distinct roles in cells, cytoplasmic and

axonemal dyneins, forming large protein complexes,

are constructed along a similar basic plan: the

com-plexes contain heavy chains (HCs), several

intermedi-ate chains (ICs) with WD repeats involved in cargo

attachment (dynein adaptor proteins such as the p150

subunit of dynactin [17] and ZW10 subunit of

Rod-ZW-Zwilch [18] interact with ICs), and at least three

distinct classes of light chains (LCs), namely the highly

conserved LC8, and members of the roadblock⁄ LC7

and T-complex testis-specific protein 1 (Tctex1) protein

families (Table 1; also see the minireview in this

vol-ume [19]) These LCs do not bind directly to the HCs

but are associated with the ICs at the base The LCs

of cytoplasmic dynein work as mediators for

interac-tions with several dynein adaptor proteins such as

nuclear distribution protein E (NudE), NudE-like

(Nudel) and Bicaudal D that bind to LC8 (reviewed in

[20])

Each dynein HC typically has a molecular mass of

500–540 kDa, consisting of approximately 4500 amino

acid residues (Fig 1A) It contains a fundamental

motor domain in the C-terminal 380 kDa fragment

[50–53] (in budding yeast,  314 kDa), incorporating

sites for both ATP hydrolysis and microtubule

bind-ing, and a tail domain in the N-terminal, which

medi-ates dimerization of the HCs and also provides a

scaffold for ICs and light intermediate chains (LICs)

(Fig 1B) While cytoplasmic dynein has identical HCs

that form homodimers, axonemal dynein is organized

with a few distinct HCs that form heterotrimers,

hete-rodimers or monomers together with ICs, LICs and

LCs The number of HCs in axonemal dyneins

depends on the species of origin: outer-arm dyneins

from most sources consist of two distinct HCs

[39,40,44,54,55], whereas those from Tetrahymena and

Chlamydomonas [36,38] each contain three distinct HCs Inner-arm dyneins contain one or two HCs [56– 58] and at least seven subspecies were identified in Chlamydomonas axonemes and termed a, b, c, d, e, f (or known as I1) and g [21,27,31–33] Studies on flagel-lar mutants of Chlamydomonas have revealed that inner-arm dyneins are responsible for determining the size and shape of the flagellar bend [42,59] Phenotypic data demonstrate that dynein I1⁄ f may play key roles

in flagellar beating, and phylogenetic analysis shows that dynein I1⁄ f is highly conserved This dynein I1 ⁄ f

is composed of two distinct HCs, 1a [31] and 1b [32], and three ICs, IC140, IC138 and IC97, and members

of LCs related to those of the outer arms: Tctex1, Tctex2, roadblock⁄ LC7 and LC8 [34,60] The known ICs and LCs in I1⁄ f dynein are not directly associated with the motor domains This is in contrast to the LC1 subunit of the Chlamydomonas outer dynein arm that interacts with the c HC motor domain [61,62] The inner-arm dyneins, except I1⁄ f, consist of mono-meric HCs, each of which associates with one actin molecule and either the Ca2+-binding protein centrin (dynein b, e and g) [21,29] or a dimer of the essential

LC termed p28 (dynein a, c and d) [24,26] It is sug-gested that actin plays a role in the proper assembly of dynein subunits or attachment of the assembled com-plex onto the doublet microtubules [63]; the function

of this actin subunit remains unknown Among these monomeric dyneins, dynein c has been intensively studied by electron microscopy [64] and single-mole-cule nanometry [65] because of its mono-disperse prop-erty in solution and non-labile motility

The bulkiness of the molecule and consequent diffi-culties in expressing and purifying mutants in large quantity have hampered the progress in structural and mechanistic studies on dyneins However, recent success in expressing active cytoplasmic dyneins in Dictyostelium discoideum [52], yeast [66] or insect cells [67] have ushered in a new era of dynein research After more than 40 years of investigation since the discovery of dynein, significant breakthroughs have been achieved: the microtubule-binding domain (MTBD) [68] and motor domains of yeast [69] and Dictyostelium cytoplasmic dyneins [70] have been crys-tallized, thus enabling new insights and research direction As well as X-ray crystallography, single-molecule measurements and advanced electron micro-copy, combined with protein engineering of dyneins, have now shed light on key unsolved questions con-cerning the organization of the molecule, the confor-mational changes accompanying ATP hydrolysis, and coordination among multiple motor domains during their motions

Heavy chain

Ortholog in

Intermediate chains

Light chains

Adaptor proteins

MT movements [21–23,25,28,41]

Cytoplasmic dynein

IC LIC

Dynactin Lis1 NudE

Outer-arm dynein

IC1 IC2

DC1 DC2 DC3 ODA5 Lis1 ODA7

Inner-arm dynein

[31,42,43] DHC10

IC140 IC138 IC97 FAP120

Tctex1 Tctex2b LC7a LC7b LC8

DHC3 DHC4 DHC11 DHC3 DHC4 DHC11

Molecular organization of dyneins

AAA modules

Dynein is a member of the AAA+ ATPase

superfam-ily (AAA: ATPases associated with diverse cellular

activities), whose members mostly function as

hexa-meric rings [71,72] However, it is quite unusual that

six non-identical AAA modules (AAA1–AAA6) are

linked in tandem in a single polypeptide (Fig 1A)

Electron microscope observations show that these six

AAA modules form a ring-shaped head domain

approximately 13 nm in diameter with a complex

mor-phology [50,64,73–75] (Fig 1B) Like other AAA

hexa-mers the ring has two different faces, suggesting that

the head is not a simple planar ring [64]

The first four AAA modules (AAA1–AAA4),

thought to bind nucleotide, contain a highly conserved

Walker A motif (GXXXXGKT, a so-called P-loop)

and a Walker B motif (DEXX) [76–79] In contrast, the sequences of the two AAA modules (AAA5 and AAA6) most proximal to the C-terminal have highly degraded Walker motifs A principal site of ATP hydrolysis has been mapped to the Walker A and B motifs of AAA1

by vanadate-mediated photocleavage [80] of the HC of axonemal dynein Further strong support for a func-tional role of the Walker motif of AAA1 is provided by molecular dissection of cytoplasmic dyneins in which mutation of the Walker A motif eliminates their motor activities in vivo [66,81] and in vitro [82]

It seems likely that the additional Walker motifs (in AAA2–AAA4) act in a regulatory manner by binding either ADP or ATP In cytoplasmic dyneins, the AT-Pase site in AAA3 plays important roles in motility, since mutations in AAA3 ATP binding and hydrolysis produce severe impairment in dynein motility [82,83] Comparable mutations of the ATPase sites in AAA2 and AAA4 have more subtle effects on motility In

C reinhardtii axonemal dynein c

Linker #1 #2 #3 #4 Stalk #5 #6 C-domain Tail

Motor domain Head ring

H1

H2 H3

H4

H5 H6

CC1

CC2 AAA1

AAA2

Linker

AAA3

AAA4

AAA5

AAA6 Buttress

Stalk

AAA2 AAA3 AAA5

AAA6 AAA1

Linker MTBD

Buttress

AAA4

C-domain Stalk

A

B

Fig 1 Overview of the molecular

organiza-tion of dynein (A) Linear map of the HC of

Chlamydomonas axonemal dynein c

(BAE19786, Chlamydomonas reinhardtii),

showing the domain structure: tail, linker,

AAA modules and MTBD Amino acid

num-bers are shown at the bottom (B) A

sche-matic drawing of the budding yeast dynein

HC in apo state The six AAA modules are

arranged in a ring and the C-terminal domain

is on the ring Each module is composed of

the N-terminal large domain and the

C-termi-nal small domain Dynein has two distinct

faces The linker face (a) corresponding to

the face seen in the left view [64] and the

C-terminal face (b) corresponding to that

seen in the right view [64] (C) Crystal

struc-ture of the cytoplasmic dynein AAA

mod-ules (reproduced with permission from

Carter et al [69]) The six individual AAA

modules are highlighted in color (D) The

atomic model of the distal stalk and MTBD

of cytoplasmic dynein in the weakly binding

state (PDB accession code 3ERR; MMDB

ID 68163 [68]) The figure was prepared

using Cn3D provided by the National Center

for Biotechnology Information (http://

www.ncbi.nlm.nih.gov/Structure/CN3D/

cn3d.shtml).

some axonemal dyneins, the presence of ADP is

known to be essential for motility in vitro [22,84], and

in others ADP increases the gliding velocity of

micro-tubules driven by dyneins, indicating that ADP binds

to at least one of these AAA modules [22,28] A

hypo-thetical atomic structure produced by homology

modelling of the dynein AAA modules suggests that

the nucleotide-binding Walker A motifs lie close to the

interface between adjacent modules [85] Interactions

between adjacent AAA modules through their

nucleo-tide pockets supports the idea that they may act in

concert to produce a functional motor

Recently, a crystal structure of the truncated motor

domain of the yeast cytoplasmic dynein HC (about

300 kDa) without nucleotide with 6 A˚ resolution has

been reported [69] Although 6 A˚ resolution is not high

enough to resolve side-chains of amino acids, the

crys-tal shows virtually all of the helices and b sheets

(Fig 1C) On the basis of features in the crystal

struc-ture, together with information from previous electron

microscopy studies as described above, the six AAA

modules, the mechanical element (termed the linker:

see below) and the base of the coiled-coil stalk were

assigned to the head ring An individual AAA module

is composed of an N-terminal large domain with an

a⁄ b Rossmann fold and a C-terminal a-helical domain

(small domain) (Fig 1B) These AAA modules are

arranged asymmetrically in the motor domain; they

are oriented at different angles and have different

packing between adjacent AAA modules (Fig 1B, C)

There is a large gap between AAA1 and AAA2 in

dynein crystallized without nucleotide It is speculated

that if ATP bound to AAA1, it would draw the

adja-cent AAA2 closer and start hydrolysis of ATP [69]

Movement of AAA2 toward AAA1 starts

conforma-tional spread along the AAA modules The fact that

the head domain of negative-stained dynein c with

ADP-Viis reported to be roughly circular whereas that

in the absence of nucleotide has a markedly different

shape [64] may support this speculation The distortion

of the head ring might represent the signal pathway

between the MTBD of the stalk and the principal

ATPase site in AAA1

As this paper was submitted for publication, another

crystal structure of the cytoplasmic dynein of

Dictyos-telium discoideum was reported at 4.5 A˚ resolution

[70] The structure contains the entire 380 kDa motor

domain including a whole stalk structure in the

pres-ence of Mg-ADP (in the post-power stroke state) The

Dictyostelium cytoplasmic dynein has a more

symmet-rical and planar motor domain than the yeast dynein

does, but a larger C-terminal domain, which is

local-ized on the face of the motor ring opposite to where

the linker resides The large gap between AAA1 and AAA2 was observed in Dictyostelium dynein motor domain, but the gap between AAA5 and AAA6 that is evident in the yeast structure was not [70]

Stalk The head ring has two elongated flexible structures called the stalk (about 15 nm long antiparallel coiled coil) and the N-terminal tail (the cargo binding domain, formerly known as the stem) The stalk extends out from the head ring between AAA4 and AAA5 It was predicted that a helix (CC1) coming out

of AAA4 and a helix (CC2) returning back to AAA5 form a coiled-coil stalk [86] X-ray crystallography showed that the stalk does not work as a bridge between AAA4 and AAA5 but is the extension of heli-ces in the small domain of AAA4 [69,70] A MTBD is localized at the tip of the stalk, forming a small globu-lar domain (Fig 1B, D) Although this globuglobu-lar domain at the stalk tip has poor sequence conservation [87], mutagenesis of conserved residues clearly inter-feres with microtubule binding [51] Microtubule bind-ing of the stalk tip was also examined with a recombinant stalk-tip peptide [88] The stalk-tip pep-tide was observed to bind to a microtubule with a peri-odicity of 8 nm and to share the binding region on the microtubule with kinesin [88]

During the mechanochemical cycle of dynein, bind-ing of ATP to the primary ATPase site of AAA1 causes dissociation of dynein from the microtubule, and binding of the MTBD to the microtubule acceler-ates the dissociation of hydrolysis products from the ATPase site [89,90] Because the sites of microtubule binding and primary ATP hydrolysis are spatially seg-regated (about 25 nm), elucidation of the communica-tion pathway between them is an important issue in understanding dynein function, as described in the pre-vious section To investigate the communication mech-anism of the stalk, Gibbons et al [91] designed a series

of fusion constructs in which the MTBD, along with a portion of its predicted coiled-coil stalk, is fused onto

a stable antiparallel coiled-coil base found in the native structure of seryl-tRNA synthetase They attempted to identify the optimal alignment between the hydropho-bic heptad repeats in the two strands of the coiled-coil stalk Alterations in the phase of the heptad repeats in the CC1 changed the affinity of the MTBD to the microtubules Finally, they identified the pattern of two alternative registries (a and b) having high and low microtubule-binding affinity, respectively On the basis of these results, Gibbons et al hypothesized that during the mechanochemical cycle the two strands of

its coiled-coil stalk undergo a small amount of sliding

displacement as a means of communication between

the AAA core of the motor and the MTBD [91] This

hypothesis is further supported by the use of an

expressed dynein motor domain in which the coiled

coil of the stalk was trapped at three specific registries

using oxidation to disulfides of paired cysteine residues

introduced into the two helices [92] Coupling between

ATPase activity and the binding activity to

microtu-bules depend upon the registry of the coiled coil

Carter et al extended the research on the MTBD

and reported the 2.3 A˚ resolution coordinates of the

MBTD in a weakly binding conformation (b registry)

and the distal portion of the coiled-coil stalk of mouse

cytoplasmic dynein [68] (Fig 1D) As predicted, they

confirmed that the stalk is a coiled coil The MTBD

consists of a bundle of six a-helices (H1–H6) and the

interface against microtubules is made up of three

heli-ces called H1, H3 and H6 The coiled coil of the stalk

is not straight but bent near the MTBD by a pair of

staggered highly conserved proline residues, with the

regular packing of hydrophobic residues in the

coiled-coil core being disrupted in the region between the

prolines When the heptad registry resumes after the

prolines, the registry of CC1 has slipped by one

half-heptad relative to that of CC2 The distal portion of

CC2 makes extensive hydrophobic interactions with

H2, H4, H5 and H6, whereas CC1 makes only a few

contacts with H4 before joining directly into H1 It is

suggested that communication along the coiled coil of

the stalk is effected by interstrand sliding and this

asymmetry at the interface between the stalk and the

MTBD plays an important role in the dynein

mecha-nochemical cycle [68]

However, the concept that the sliding of the

stag-gered coiled coils relative to each other within the stalk

achieves two patterns of alternating registries has been

challenged by the crystal structure of dynein [69] Since

the crystal structure suggests that two helices (CC1

and CC2) merge into the well-packed helices of the

AAA4 small domain, it is unlikely that either helix can

move at its base Furthermore, in addition to the stalk

coiled coil, the crystal structure revealed the presence

of a second antiparallel coiled coil that emerges from

the small domain of AAA5 as a long extension of

heli-ces The structure is called buttress [69] or strut [70]

and it extends toward and makes contact with the

stalk (Fig 1B, C) Although the crystal of the yeast

dynein has no MTBD, owing to optimization of

crys-tallization, the interaction between the stalk and the

buttress⁄ strut provides insight for the regulation

mech-anism of MTBD by the AAA modules in the head

ring Through the interaction of the buttress⁄ strut and

the stalk, the buttress⁄ strut might relay rigid body motion between AAA modules into shear motions between the helices of the stalk coiled coil

The crystallographic analysis of the motor domain

of Dictyostelium dynein provides the evidence for the structural information pathway between AAA1 and MTBD since the crystal unit contains a whole stalk and two independent motor domains, which adopt dif-ferent conformations [70] The major structural differ-ence is found in the stalk–buttress⁄ strut structure The stalk of one motor domain is straight up to the tip, while that of the other motor domain is kinked at the region just beneath the contact site with the but-tress⁄ strut This stalk tilting is accompanied by small conformational changes of the strut The kink of the stalk while holding the basal portion in place could induce interstrand sliding Kon et al thus hypothesize that dynein coordinates AAA1 ATPase and MTBD by switching the stalk-strut structure between the straight and kinked conformations [70] This hypothesis implies

a new communication pathway in which the structural information could propagate from AAA1 to MTBD through the C-terminal domain, AAA5 and then buttress⁄ strut [70]

Tail The amino-terminal tail is involved in dimerization⁄ tri-merization of dynein HCs and acts as a scaffold for the assembly of different ICs and LICs to form the dynein complex Since an expressed cytoplasmic dynein with-out a native tail but with a substituted tail shows intact processive movement in vitro, these subunits are not essential for dynein motility in vitro [53] However, they may regulate dynein motility in vivo and recent data indicate that the non-catalytic subunits link dynein to cargos and to several adaptor proteins that regulate dynein function [20] (see the minireviews [13,19]) For example, missense mutations in the tail domain of cyto-plasmic dynein in mice cause neurodegenerative disease The best characterized model of dynein dysfunction is the Legs at odd angles (Loa) mouse [93] This mutation

is thought to affect homodimerization of the dynein HCs and⁄ or the association of the HCs and ICs [94] Single-molecule nanometry on the mutant dynein showed that dynein purified from mutant mice has lower processivity and shows more frequent bidirec-tional motility along a microtubule and greater propen-sity to sidestep to adjacent protofilaments than the wild-type dynein does These results suggest that muta-tion in the tail domain of dynein causes increased flexi-bility of the dynein molecule and diminished gating between the motor domains [94]

While plus-end-directed transport in cells is carried

out by many kinesin family members with a wide

range of tail domains and despite the large repertoire

of cellular functions that dynein is involved in, all

minus-end-directed transports within the cytoplasm are

carried out by a single cytoplasmic dynein The tail

domain is thus important to mediate interaction with

various types of cargo by recruiting specific and

appro-priate adaptor proteins

Linker

The linker is a structure located in the portion of the

tail proximal to AAA1, which serves as a connection

between AAA1 and the main part of the tail (Fig 1B,

C) The existence of the linker was first indicated in

images of negative-stained monomeric axonemal

dynein [64] Although the linker is normally docked

onto the head ring, it is revealed as a relatively large

structure about 2 nm wide and 10 nm long when the

linker is undocked from the head ring [64] The crystal

structure of dynein showed that the linker is composed

of helical bundles and does not sit flat on the head

ring but rather arches over it [69,70] The linker is

composed of four predominantly helical subdomains

(from N-terminus, subdomain 1, 2, 3 and 4) The

C-terminus subdomain 4 interacts with AAA1 and part

of AAA6 and is connected into AAA1 The

N-termi-nal subdomain 1 contacts AAA5 in the yeast motor

domain [69] or AAA4 in the Dictyostelium head ring

[70], and this contact looks tenuous and may break

and dissociate from AAA4 or AAA5 during the

AT-Pase cycle Although the significance of the difference

in the contact point remains unclear, it could represent

conformational changes upon ADP release during

dynein’s ATPase cycle [70]

It has been suggested that the linker is involved in

generation of force through its interaction with the

head ring [64] Two-dimensional analysis on

negative-stained dynein c described the conformations of the

dynein molecule in two different nucleotide states

which mimic the post- and pre-power stroke

conforma-tions of the motor (Fig 2) In the absence of

nucleo-tide (post-power stroke conformation, state I) the tail

emerges near the base of the stalk In the presence of

ATP and vanadate, which forms a dynein–ADP–Vi

complex that mimics the dynein–ADP–Pi

conforma-tion (pre-power stroke conformaconforma-tion, state II), the tail

emerges further away from the stalk base These

obser-vations were interpreted to originate from the swinging

of the linker relative to the head ring The existence

and the movement of linker have subsequently been

confirmed in cytoplasmic dynein, identified as the

N-terminal region of the motor domain using green fluorescent protein (GFP) and blue fluorescent protein (BFP) tagged constructs by negative stain electron microscopy and Fo¨rster resonance energy transfer (FRET) [75,89] In the absence of nucleotide or in the presence of ADP, GFP inserted at the linker’s N-ter-minus lies close to AAA4 (the crystal structure of the yeast dynein [69] shows the N-terminus lies close to AAA5, as described above), at the base of the stalk, in the so-called un-primed position, whereas in the pres-ence of ATP and vanadate the GFP lies close to AAA2, in the primed position [75]

Dynamic measurements of the linker movement were performed by measuring the FRET between a GFP and a BFP both fused into a dynein construct molecule [89] A series of 380-kDa dynein constructs from Dictyostelium were prepared that had a GFP attached at the N-terminus and a BFP inserted into various sites on the dynein head ring The efficiency of FRET was measured in each construct at various nucleotide states under steady-state conditions The results showed two distinct values: a high FRET effi-ciency and low FRET effieffi-ciency suggesting movement

of the N-terminus relative to the head ring Using mutants that were trapped in specific intermediate states, it was shown that this movement is coupled to ATPase steps [89]

Lever-arm model or winch model

The observations described above pose the model of dynein force generation, which is the most widely cited one: on binding of ATP to AAA1, the orientation of the docked linker on the head ring causes the tail to emerge far from the stalk (state II, top panel in Fig 2A) Upon release of products, the linker orienta-tion on the ring changes, bringing the tail closer to the stalk (state I, bottom panel in Fig 2A) The linker changes its orientation by switching between two differ-ently docked positions on the head ring, thus producing

a rotation of the head ring that causes the stalk to swing In addition to this head ring motion, an ATP-driven alteration in the coiled-coil registry may control the affinity of the dynein HC for the microtubule How-ever, the stalk is not rigid since stiffness of a 15-nm length of coiled-coil peptide clamped at one end can be estimated to be about 0.4 pNÆnm)1 The range of con-formations for an axonemal dynein molecule observed

in negatively-stained samples [64,73] also provides an estimation of the stiffness of the stalk, which is 0.5 pNÆnm)1 in apo molecules and 0.14 pNÆnm)1 in ATP-vanadate molecules [95] These estimates imply that stalk is too flexible to work as a rigid lever [96,97]

Cryo-electron-microscope images of whole dynein

molecules interacting with microtubules [98] have

recently revealed that, even though the tail and linker

shift relative to the head ring and stalk as in isolated

dynein molecules, the stalk orientation on the

microtu-bule remains fixed The observation provides strong

evidence for the concept of dynein as an

ATP-depen-dent winch [96] (Fig 2B)

Furthermore, winch-type motion was also observed

in an axoneme Cryo-electron tomograms of axonemes

in the apo state were compared with those obtained in

the presence of ADP-vanadate [99] Global changes of

dynein arm complexes were shown and several key

changes in dynein structures were found Although the

stalks are not clearly visible in the tomograms, close

examination showed that the stalks typically tilt

towards the proximal end of the axoneme (the base of

the axoneme) in both nucleotide conditions The

dynein head rings were observed to move 8 nm toward

the distal end of the axoneme upon release of the

nucleotide Since the MTBD attached to the adjacent

microtubule, the movement results in dragging the

adjacent microtubule distally and producing the shear [99]

The winch model explains the result of Carter et al [68], in which cytoplasmic dynein with its stalk coiled coil either lengthened or shortened by seven heptads moves towards the minus end of a microtubule, irre-spective of the length of the stalk This result is remarkable since these stalk length changes would be predicted to rotate the head ring by 180 and reverse the direction of dynein movement according to the lever-arm model To explain the directionality of dynein, it is proposed that the head ring does not elicit

a lever-like rotation of the linker domain perpendicular

to the stalk, but rather, contraction where the force vector of the linker domain’s conformational change is directed parallel to an angled stalk [68] (Fig 2B)

Mechanical properties of dynein

Characterizations of the mechanical properties of dyneins have been carried out using in vitro motility assays, which enable the motility of dyneins along

S

tate II

State I

State II

State I

Fig 2 Proposed mechanisms of dynein’s power stroke (A) Negative-stain electron microscopy followed by single-particle analysis suc-ceeded in capturing two distinct conformations of dynein c molecules isolated from Chlamydomonas flagella in the ADP-V i (state II) and apo (state I) state [64] On ATP binding to AAA1, the tail emerges far from the stalk (state II, top panel) Upon product release, the tail emerges closer to the stalk (state I, bottom panel), suggesting movement of the linker domain (and tail relative to head ring and stalk) This conforma-tional change swings the stalk by  15 nm This model is based upon [64] and [89] (B) The winch model of dynein force generation [68,96,98,99] Product release from AAA1 leads to the contraction of the whole dynein molecule by the movement of the linker As shown in the electron microscope images of microtubules decorated with stalks [98], the stalk points toward the microtubule minus end and connects the head ring to the microtubule The microtubule is thus dragged by the contraction induced by the shift of the head ring relative to the tail.

microtubules to be reconstituted from purified and

characterized component molecules Since experimental

conditions such as temperature, buffer compositions

and ATP concentrations are readily controlled, these

assays permit precise measurements of dyneins’

mechanical properties Several important findings

about dyneins are listed below

l The minimal components of dynein motility are the

380 kDa domain of Dictyostelium dynein (314 kDa

domain of yeast dynein) which contains a linker,

six AAA domains and a stalk [52,53]

l Each of the HCs of axonemal dyneins so far

stud-ied has distinct motile activity [21,22,41]

l Some inner-arm dyneins can generate torques and

rotate a microtubule about its long axis while it is

moved on the dynein-coated surface [21,100]

l Some inner-arm dyneins require a trace amount of

ADP for their microtubule motility [22,84]

l Some inner-arm dyneins, even some single-headed

or heterodimeric motors, show high processivity

in vitro[25,65]

l The three HCs of outer-arm dyneins are likely to

play distinct roles and regulate each other to

achieve coordinated force production [101]

l Dynactin enhances the processivity of cytoplasmic

dynein [102–104] Molecular dissections of dynactin

showed that the mechanism of processivity

enhance-ment is not due to anchoring the motor domain of

dynein to microtubules via dynactin but a

mecha-nism independent of microtubule tethering [104]

Furthermore, in vitro motility assays have paved the

way to single-molecule studies on dynein In recent

decades, the development of a number of technologies,

such as atomic-force microscopy, optical-trap

nanome-try and fluorescence imaging with nanometer precision

have provided tools for studying the dynamics of

sin-gle molecules in situ over time scales from milliseconds

to seconds The single-molecule sensitivities of these

methods permit studies to be made on conformational

changes and functions of dyneins that are masked in

ensemble-averaged experiments Processivity, step size

and dwell-time distributions are among properties that

can be directly measured by single-molecule

tech-niques Our understanding of the functions of dyneins

has benefited considerably from the application of

sin-gle-molecule techniques

However, single-molecule measurements on the force

generation of dyneins have raised some questions The

stall force generated by single dynein molecules varies

from measurement to measurement: for axonemal

dyneins, a value of 1–2 pN in single-headed inner-arm

dynein [65],  6 pN in an inner dynein arm in an

axoneme [105] and 4.7 pN in outer-arm dynein [106];

for cytoplasic dyneins, a value of  1 pN in bovine cytoplasmic dynein [107,108], 3–4 pN [109] and 7–

8 pN [110] in porcine cytoplasmic dynein and 8 pN in yeast cytoplasmic dynein [111] The variation in force may depend upon the type of dynein used and upon distinct roles that dynein plays in vivo [53] It is also suggested that the geometry of the force measurements may influence the force and the mode of motility [109] Nucleotide concentrations may have an effect on the force generation and modes of movement [106,109] The precise measurements and direct comparison of the force generated by cytoplasmic dyneins are now required since they will provide important information

to reveal the mechanism of cargo transport by a num-ber of or several types of motors mechanically coupled

to each other

Processivity and modes of movement

Cytoplasmic dynein is known as a processive motor that can take micrometer-scale movements along a microtubule without dissociating Through the creation

of a cytoplasmic dynein that can be converted between monomeric and dimeric states by a small molecule, rapamycin, it is demonstrated that processive motion requires the dimerization of two motor domains, although the endogenous dimerization domain (tail) is not required for the processivity [53] In addition, pro-cessivity of cytoplasmic dynein in vitro does not require any of the known dynein-associated subunits [53] despite reports that the dynactin complex enhances the processivity of cytoplasmic dyneins [102–104] The step sizes and modes of movement of cytoplas-mic dyneins are under debate Cytoplascytoplas-mic dynein purified from bovine brain primarily takes large steps (24–32 nm) at low loads, but decreases step size from

32 to 8 nm with increasing load to its stall force [107]

In addition, when multiple dynein molecules interact with a microtubule and contribute to movement, the dynein molecules move predominantly in 8-nm steps [108] In contrast, movement of single cytoplasmic dynein molecules purified from porcine brain [110] and

a functional recombinant dimeric dynein of the bud-ding yeast [53] were analyzed with a high spatial preci-sion tracking technique and were stepwise with a regular 8-nm step size, irrespective of the load

Based upon these findings, Reck-Peterson et al proposed a molecular model (which they called the

‘alternating shuffling model’) to explain how processive motion is achieved by cytoplasmic dynein [53] (Fig 3) The model is conceptually similar to the hand-over-hand model proposed for processive kinesin motility: two dynein heads alternate taking 16-nm steps while

the centroid position of the molecule moves by 8 nm

for each step The large dimensions of the head ring

do not allow the head ring to alternately ‘swing’

forward at each step, but the two head rings always

overlap partially during the stepping motion The

model thus requires coordination between two motor

domains Optical-trap studies of dynein force

produc-tion suggest that this coordinaproduc-tion is carried out by

strain transmitted through the linkage between the two

heads [111] Besides the intramolecular strain, physical

contacts between two head rings during the movement

along a protofilament could mediate head–head

coor-dination [112]

However, given the apparently large size, the

flexi-bility of a dynein molecule and the magnitude of the

suggested power stroke [64] in relation to the tubulin

lattice, the 8-nm step size displayed by dimeric

dyne-ins is surprisingly small This is in contrast to other

linear motors since the step size of a motor, except

myosin VI [113], can be predicted to be proportional

to the length of its power stroke [114,115] One

possible explanation is that when a dimeric dynein moves on a microtubule taking 8-nm steps, the mole-cule could be compact and stiff with its two head rings in close, intimate association as is seen in axo-nemal dyneins in situ [116,117] and in electron micro-scope observations of the phi (F) shaped structure of cytoplasmic dynein in which the tails of the HCs are close to each other and the head rings are partially overlapped [118]

In contrast, in early studies of dynein motility, cyto-plasmic-dynein-coated beads exhibited greater lateral movements among microtubule protofilaments than did kinesin [119] Hence, dynein apparently does not have to walk along a single protofilament Precise measurements showed that fluorescently labeled dynein also displayed lateral stepwise movements, which usually occurred simultaneously with forward stepping This shows that dynein has the reach or flexibility to occasionally land on an adjacent protofilament [53] Furthermore, as stated earlier, the Loa mutant cytoplasmic dynein showed increased frequency of

Time

E

D

ATP

ADP

J

ATP

ADP

Fig 3 Alternating shuffling model for processive movement of a dimeric dynein [53] with some modifications on the basis of the crystal structure [69] and kinetics [90,125] To simplify, we draw all presumed elastic elements in the tail domain as a simple spring that connects two head rings The stalk and the head ring are drawn as a rod and a large circle, respectively The linker is drawn as a yellow curved bar, which has a hinge as a yellow small circle We construct linker motion swinging around the hinges (A) Binding of ATP to the trailing head (red) releases the MTBD from the microtubule and then (B) the head is pulled forward by the strain stored in the connecting spring while the leading head (blue) stays bound to the microtubule (C) The linker is detached from the docking site on AAA5 Upon re-binding of the MTBD to the microtubule (D), the head changes its conformation coupled with product release (E) (F)–(J) The trailing head carries out the same mechanical process as shown in (A)–(E) Two dynein heads alternate taking 16 nm steps, whereas the position of the center of mass

of the molecule moves by 8 nm for each step Note that, due to the large size of the head ring, two heads are partially overlapped during stepping without changing the relation of their lateral positions.