| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Cell Migration Laboratory, Department of Dermatology, University of Würzburg, Würzburg, Germany
Correspondence: Peter Friedl, Cell Migration Laboratory, Department of Dermatology, University of Würzburg, Josef-Schneider-Str. 2, 97080 Würzburg, Germany. E-mail: peter.fr{at}mail.uni-wuerzburg.de
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONTHE LIFE CYCLE OF...PARADIGM OF AMOEBOID MOTION,...CYTOSKELETAL ORGANIZATIONSIGNALING CONTROL OF ACTIN...ADHESION AND DETACHMENT...GENERAL LOCOMOTOR...STRUCTURE AND FUNCTION OF...SIGNALING CONTROL OF...FUNCTION OF ADHESION RECEPTORS...DIRECTIONAL GUIDANCE OF...CONCLUSIONSREFERENCES |
|---|
Key Words: T lymphocytes • collagen matrix • cytoskeletal dynamics • migration strategies
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONTHE LIFE CYCLE OF...PARADIGM OF AMOEBOID MOTION,...CYTOSKELETAL ORGANIZATIONSIGNALING CONTROL OF ACTIN...ADHESION AND DETACHMENT...GENERAL LOCOMOTOR...STRUCTURE AND FUNCTION OF...SIGNALING CONTROL OF...FUNCTION OF ADHESION RECEPTORS...DIRECTIONAL GUIDANCE OF...CONCLUSIONSREFERENCES |
|---|
Such a broad spectrum of migratory capabilities through or across biophysically and biochemically very different substrates implies a great degree of flexibility and adaptability of leukocyte migration and positioning strategies. Such diversity may set leukocytes apart from other more specialized migratory cell types. As examples, fibroblasts, keratinocytes, and neurons migrate only under circumstances highly restricted in location and time, i.e., on morphogenesis or wound healing within a specific tissue context.
Cell migration within the tissues is a complex mechanochemical process that requires the integration of key events in signaling, cytoskeletal, membrane, and adhesion systems [reviewed in ref. 10 11 12 ]. Based on cell type-specific morphological and functional criteria of migration within an extracellular matrix (ECM) environment, i.e., polarization and shape change, migration velocity, cytoskeletal organization, and integrin and protease expression and function, at least three migratory prototypes can be classified: (1) amoeboid crawling can be distinguished from (2) fibroblast-like, mesenchymal migration and (3) collective cell movement, as seen in multicellular strands, sheets, or clusters [12 14 ].
The concept of amoeboid motion is most clearly established by studies using the single-cell stage of the lower eukaryotic ameba Dictyostelium discoideum [15 ]. Amoeboid movement results from alternating cycles of morphological expansion and contraction driven by cytoskeletal dynamics, shape change, and low adhesivity. These migration characteristics allow ameba to rapidly adapt to a given environment, develop high migration velocities, and contact other cells in a dynamic yet reversible manner. The Dictyostelium paradigm of movement has important implications for the understanding of cell migration strategies in higher eukaryotes, most notably for neutrophils, lymphocytes, and some tumor cells [15 , 16 ].
Fibroblast-like mesenchymal migration, as detected in fibroblasts, myoblasts, neural crest cells, and many cells from solid tumors, depends on relatively slow, adhesion receptor- and integrin-dependent cell-substrate interactions that fulfill the criteria of adhesion-mediated ("haptokinetic") migration [11 , 17 , 18 ]. These cells express high levels of integrins and matrix-degrading proteases and, while migrating, can create substantial remodeling of the ECM [19 20 21 ; reviewed in ref. 12 ].
Collective cell movement, in extension of haptokinetic migration of single cells, is dependent on highly adhesive substrate interactions and remodeling of the extracellular matrix. Multicellular migration requires the maintenance of stringent cell-cell adhesion and communication mechanisms [13 , 22 23 ; Y. Hegerfeldt, M. Tusch, E. B. Brocker, P. Friedl, unpublished results], as detected in migrating cell clusters from epidermal keratinocyte sheets, angiogenic sprouts, and tumors.
Because form follows function and vice versa, the shapes adopted by single cells or cells within tissue frequently correspond to functional specificities of those cells. We here summarize cellular and molecular strategies involved in amoeboid cell movement and use the paradigm of amoeboid crawling for understanding the locomotor mechanisms of leukocytes, particularly by T lymphocytes and neutrophils migrating on 2-D and within 3-D tissue substrates.
|
THE LIFE CYCLE OF D. DISCOIDEUM |
|---|
|
TOPABSTRACTINTRODUCTIONTHE LIFE CYCLE OF...PARADIGM OF AMOEBOID MOTION,...CYTOSKELETAL ORGANIZATIONSIGNALING CONTROL OF ACTIN...ADHESION AND DETACHMENT...GENERAL LOCOMOTOR...STRUCTURE AND FUNCTION OF...SIGNALING CONTROL OF...FUNCTION OF ADHESION RECEPTORS...DIRECTIONAL GUIDANCE OF...CONCLUSIONSREFERENCES |
|---|
|
PARADIGM OF AMOEBOID MOTION, INCLUDING SHAPE CHANGE AND MIGRATION DYNAMICS |
|---|
|
TOPABSTRACTINTRODUCTIONTHE LIFE CYCLE OF...PARADIGM OF AMOEBOID MOTION,...CYTOSKELETAL ORGANIZATIONSIGNALING CONTROL OF ACTIN...ADHESION AND DETACHMENT...GENERAL LOCOMOTOR...STRUCTURE AND FUNCTION OF...SIGNALING CONTROL OF...FUNCTION OF ADHESION RECEPTORS...DIRECTIONAL GUIDANCE OF...CONCLUSIONSREFERENCES |
|---|
An important aspect of cell movement is how biochemical signals are integrated into characteristic and cell type-specific morphological dynamics. In resting Dictyostelium, the size of the cell body is 8–12 µm and the shape is spherical. After polarization and on crawling, the length axis averages 12–16 µm and can extend to 30 µm and more. In principle, the amoeboid morphology consists of an elliptoid core body with a broader leading edge and a narrowing trailing edge, termed the uropod. The overall morphology is highly reactive to external stimuli and can fundamentally change within seconds. Shortly before and on crawling, multiple pseudopods form around the leading edge and lateral portions of the cell body, while the uropod appears to remain a largely passive contractile zone. While the cell body advances, the bottom surface has been shown to maintain knobby foot-like actin-rich interaction zones ("eupodia") of 1 µm in diameter towards the underlying substrate [29 ], which, together with lateral adhesion sites, are thought to act as anchoring points. The average velocity varies from 4 to 12 µm/min, whereas accelerations in pseudopod extension result in peak velocities up to 25 µm/min [30 ]. If external chemotactic triggers are lacking, autochthonous pseudopodal oscillations and shape changes lead to short-lived forward movement alternating with rapid turns. Consequently, the directional persistence of nonchemotactically crawling ameba on nonaligned substrates is low, in agreement with the concept of persistent random walk [31 , 32 ].
In the natural environment, Dictyostelium cells migrate on or within three-dimensional (3D) complex substrates such as soil particles, fragmented leaves, and debris of very different physicochemical properties. The cells are able to move on humid as well as on dry substrates. Consequently, amoeboid migration must be a very robust process that is resistant to many adverse events. On the other hand, conventional laboratory substrates represent relatively smooth surfaces, such as agar, nitrocellulose filters, or cover slips, lacking biophysical constraints. As discussed below, these substrates might not mimic the natural environment closely enough, recently prompting investigations on reconstructed soil substrate [33 , 34 ].
|
CYTOSKELETAL ORGANIZATION |
|---|
|
TOPABSTRACTINTRODUCTIONTHE LIFE CYCLE OF...PARADIGM OF AMOEBOID MOTION,...CYTOSKELETAL ORGANIZATIONSIGNALING CONTROL OF ACTIN...ADHESION AND DETACHMENT...GENERAL LOCOMOTOR...STRUCTURE AND FUNCTION OF...SIGNALING CONTROL OF...FUNCTION OF ADHESION RECEPTORS...DIRECTIONAL GUIDANCE OF...CONCLUSIONSREFERENCES |
|---|
Mechanisms of cell motility
Three different yet putatively synergistic mechanisms have been
proposed and further studied that might contribute to cell motility:
actin polymerization to filaments, myosin-mediated stiffening and
contraction, and osmotic or hydrostatic force.
Actin polymerization
Actin polymerization occurs closely juxtaposed to the inner
leaflet of the plasma membrane, possibly favored by random membrane
fluctuations to be filled by expanding actin filaments. Actin, the main
component of the microfilamentous cytoskeleton, exists as soluble,
monomeric globular actin (G-actin) which homoaggregates to form
filamentous actin polymers (F-actin). Actin polymerization is a
directional process. Rapid aggregation of monomers occurs at one side
of the filament (barbed end), while depolymerization is present at the
other end (pointed end). Growing actin filaments elongate, branch, and
form a viscous actin-rich gel ("gel" phase). This actin gel is
characterized by a certain degree of stiffness and rigidity, allowing
rapid membrane protrusion and pseudopod growth. From scanning electron
microscopy studies detecting labeled actin monomers newly incorporated
into the filaments, it appears that most of these barbed ends are
directed towards the plasma membrane and, while growing, might generate
an outward movement of the membrane [27
].
Myosin sliding
The myosin sliding concept predicts that, also in nonmuscle cells,
actin filaments can be cross-linked and contracted through myosin,
generating both cortical actin stabilization and cell contraction.
Myosin II (conventional myosin) forms bipolar filaments that cross-link
actin filaments and stiffen cell protrusions [38
].
Myosin II was further shown to mediate posterior contraction enabling
the rear of the cell to detach from the substrate and move in concert
with the leading edge [39
]. Although myosin II-deficient
cells are principally able to generate pseudopods and undergo
locomotion on surfaces, the migration velocity as well as mechanical
force generated at the leading edge of myosin II-deficient cells is
reduced as a consequence of reduced cortical stiffness, delayed
posterior release of adhesive bonds, and uropod retraction
[40
, 41
].
Cortical expansion
The cortical expansion model predicts that local osmotic pressure
favors focal cytoplasmic swelling and pseudopod extension
[35
]. Osmotic pressure might be generated by ion influx
at the leading edge mediated by ion channels [42
] or the
cytoplasmic release of osmotically active molecules. In addition, some
hydrostatic pressure might be generated by cell contraction or uropod
retraction that pumps cytoplasmic fluid into regions of increased
elasticity. Net pressure of the aqueous phase could then lead to
localized swelling and passive outward extension of a nascent
pseudopod.
Role of cytoplasmic proteins in actin function
It is quite likely, that the above concepts synergize and can be
integrated into one common scheme. In Dictyostelium, the
molecular control of actin filament assembly and disassembly is
provided by an ever increasing number of known cytoplasmic proteins,
which bind to actin and control actin function. The following
paragraphs highlight selected proteins, most of which share structural
and functional homology to mammalian cells.
Capping proteins
Capping proteins bind to, protect, and sequester actin filaments.
At the onset of filament growth, capping proteins dissociate from
preexisting filaments and increase the number of barbed ends
[43
]. Capping proteins appear to deliver actin monomers
to privileged actin nucleation centers, e.g., at the leading edge
[44
], thereby putatively confining the addition of actin
monomers to growing filaments at special sites in time and space
[45
]. On the other hand, while reversibly capping older
filaments, capping proteins also terminate polymerization. Capping
proteins include gelsolin, protovillin, profilin, and Cap32/34. Actin
uncapping and initial polymerization are rapid processes. After
chemotactic stimulation, increased barbed ends are observed in
Dicytostelium with 1-s delay [46
], indicating
that uncapping is a rapid process that initiates filament growth and
remodeling. The importance of capping proteins in actin dynamics and
remodeling is underlined by deletion mutants that develop exaggerated
levels of polymerized actin while motility is impaired, putatively as a
consequence of decreased actin turnover [47
].
Severing proteins
Severing proteins bind to and intercalate between actin filaments
causing filament breakage or actin monomer dissociation at the contact
point. It is thought that repetitive filament severing keeps growing
actin filaments at a certain length for network formation and also
increases the number of freely accessible barbed ends for filament
growth [summarized in ref. 48
]. However, if G-actin levels are
rate-limiting, severing will lead to depolymerization, whereas at high
G-actin concentrations, nucleation and filament growth are supported
[48
]. Important severing proteins are members of the
cofilin family (previously called actin-depolymerizing factors). In
vitro and in vivo analyses have shown that cofilin severs filamentous
actin thereby generating free barbed ends for G-actin binding and
nucleation. Cofilin increases actin turnover rates [49
],
and, consequently, overexpression of cofilin leads to enhanced
ruffling, pseudopod formation, and cell movement [50
,
51
]. It was proposed that cofilin could recycle actin
monomers to regions of nucleation by depolymerizing old filaments,
thereby contributing to the remodeling of the actin cytoskeleton
[44
, 52
]. Other proteins exerting severing
function are gelsolin, severin, and villin [53
].
Branching and cross-linking proteins
Branching and cross-linking proteins bind to and intercalate
between individual actin filaments and contribute to the formation of
actin meshwork [54
]. Cross-linking is important for the
generation of a 3-D actin filament structure of increased mechanical
strength and stiffness. Branching proteins include Arp2, Arp3, members
of the filamin family [e.g., the 120-kDa gelation factor actin-binding
protein (ABP)-120], and coronins. The actin-related proteins Arp2 and
Arp3 are part of a seven-protein cross-linking and signaling complex
localized in actin-rich spots within pseudopodia. The Arp2/3 complex
was shown to be essential in actin network formation
[55
]. ABP-120 and coronin become localized in pseudopods
of the leading edge and are thought to contribute to cross-linking and
remodeling of cortical filaments in pseudopod extensions [56–59].
Other branching proteins are ABP-180 and spectrin. Cross-linking
proteins, such as 
-actinin, fascin, cortexillins, and calpactin
cross-link actin filaments to form thicker bundles or meshwork of
higher mechanical strength. Myosin II also cross-links actin filaments
and contributes to cortical stiffness and pushing force in extending
pseudopods [40
, 41
].
Anchoring proteins
Anchoring proteins are candidate proteins connecting growing
filaments to the inner leaflet of the plasma membrane. Anchoring
proteins include Scar1, myosin I, and talin homologues. Scar1 is an
important multifunctional adapter protein related to the
Wiskott-Aldrich syndrome protein (WASP) family in mammals (see below).
Scar1/WASP binds to membrane-inserted phosphoinositides and
membrane-anchored activated G-proteins, then recruiting actin binding
and branching proteins (e.g., Arp2/3) to the membrane
[60
]. Overexpression of a nonfunctional,
dominant-negative Scar1/WASP fragment uncouples the actin cytoskeleton
from the membrane and leads to complete loss of actin nucleation and
lamellipodial extension. Myosin I ("unconventional myosin") is a
candidate protein to anchor actin filaments to the plasma membrane
predominantly at the leading edge. Myosin-I-deficient cells extend an
increased number of pseudopods, indicating a role for myosin I in
focusing and polarizing pseudopodal action [61
].
Talin-like proteins are homologous to mammalian talin (see below), a
protein that cross-links actin to transmembrane adhesion receptors
[62
, 63
]. Talin homologues localize to the
tips of the filopodia, cross-link actin filaments, and are also
involved in actin nucleation and assembly [62
,
63
]. Other anchoring proteins are ponticulin,
hisactophilin, synexins, and intercaptin [64
66
].
There is a consensus that actin polymerization involves simultaneous and synergistic mechanisms including severing of preexisting actin filaments (cofilin), uncapping of preexisting nuclei by dissociation of capping proteins, and de novo nucleation (Arp2/3, Scar1/WASP) [28 , 48 , 55 , 67 , 68 ]. From a viewpoint of physical chemistry, the formation and turnover of actin-rich gels are sufficient to explain membrane protrusion and the generation of a certain degree of mechanical stiffness. In vivo evidence for this concept comes from observations on the intracellular propulsion of Listeria monocytogenes in infected eukaryotic cells. Listeria rods are pushed forward by a filamentous "tail" of endogenous host actin that forms at one end of the bacterium [reviewed in ref. 69 , 70 ]. A minimum of four components required for this actin-driven force generation were recently identified using purified proteins in cell-free extracts: actin, activated Arp2/3 complex, cofilin, and capping protein [71 , 72 ].
Several models have been proposed illustrating how these components might act together. Activated Arp2/3 induces branched nucleation and filament growth. Two actin sites have been discussed as targets of Arp2/3, the sides of older filaments (dendritic nucleation model) [68 , 73 ] and the barbed ends of actin filaments after uncapping (barbed-end nucleation model) [74 , 75 ]. Cofilin has been proposed to depolymerize pointed ends and to further accelerate actin incorporation into barbed ends, increasing the speed of filament growth [51 , 76 ]. The function of capping protein in this context is less clear. Capping protein could contribute to spatially limit and focus the polymerization process (funneling concept), e.g., by masking older filaments while sparing adjacent rapidly growing filaments [67 ]. Based on these concepts together with other in vitro as well as in vivo findings, actin polymerization alone might generate mechanical pushing force for pseudopod growth and protrusion.
It remains to be determined how the mechanical strength generated by
such restricted actin assembly in vitro can be translated to the
biomechanical requirements of entire cells. In vivo generation of actin
filaments must be robust enough for higher-order shape change and
mechanical resistance towards extracellular constraints. It is
therefore likely that the biomechanics in complete cells require
additional action of ABPs with overlapping function. In
Dictyostelium, a single cell crawling across 2-D surfaces is
a robust process that can be quite resistant to genetic ablation of
structural and signaling proteins. A number of Dictyostelium
deletion mutants for single ABPs (e.g., coronin, profilin, myosin IB,
120-kDa gelation factor, cortexillin, severin, cap32/34, -actinin,
or talin) have produced strikingly minor alterations in single-cell
locomotion, whereas other processes such as phagocytosis, cytokinesis,
and morphogenesis have been more frequently impaired
[77
]. One obvious explanation is that deleting a single
or a limited number of proteins might not sufficiently impair the
overall stability of the cytoskeleton, because functional compensation
of the deleted proteins might be provided by remaining ABPs as
a part of a redundant network. More frequently, loss of function in
mutants with deletion of cytoskeletal proteins is obtained at the slug
or spore stage, indicating that cell-cell and cell-substrate
interactions involved in sorting and positioning at 3-D multicellular
stages demand more stringent cytoskeletal action and control than
biomechanically less demanding single-cell crawling
[25
].
A further critical parameter determining the molecular biomechanics of
amoeboid crawling and positioning could come from the physical 3-D
architecture and composition of the extracellular environment. Ponte et
al. [34
] have reported experiments on
Dictyostelium gene deletion mutants (deletions of genes
encoding, e.g., -actinin, synexin, ABP-34 and ABP-120) using
an artificial soil substrate that reconstructs an arguably more complex
and natural in vitro environment. On reconstructed soil, these mutants
display more profound locomotor and morphogenetic defects than were
initially described for the same mutants using conventional laboratory
substrates [34
]. Therefore, it might be useful to
investigate complex cell functions such as migration also in the
context of biomechanically demanding "nature-like" 3-D
environments.
|
SIGNALING CONTROL OF ACTIN POLYMERIZATION |
|---|
|
TOPABSTRACTINTRODUCTIONTHE LIFE CYCLE OF...PARADIGM OF AMOEBOID MOTION,...CYTOSKELETAL ORGANIZATIONSIGNALING CONTROL OF ACTIN...ADHESION AND DETACHMENT...GENERAL LOCOMOTOR...STRUCTURE AND FUNCTION OF...SIGNALING CONTROL OF...FUNCTION OF ADHESION RECEPTORS...DIRECTIONAL GUIDANCE OF...CONCLUSIONSREFERENCES |
|---|
For the integration of an external chemotactic stimulus towards cell
polarization and directional crawling, triggering should occur in a
polarized manner, e.g., by preferential signaling and cytoskeletal
activation in the cell compartment proximal to the highest
chemoattractant concentration [79
]. Preferential
chemoattractant-induced signaling at the leading edge would require the
polar redistribution and/or clustering of chemoattractant receptors or,
alternatively, G-proteins or downstream effectors. Whereas in polarized
Dictyostelium cells, cARs remain uniformly
distributed, the signal transducing G-protein Gß
shows
preferential accumulation at the leading edge [82
].
Polarized G-protein activation is thought to generate downstream
signaling towards different effectors, most notably to
phosphatidyl-inositol-kinases (PIKs) and small GTPases
[79
]. In principle, PIKs and small GTPases synergize for
downstream signaling by recruiting multiple other signaling and
cytoskeletal proteins to the inner leaflet of the plasma membrane,
leading to actin nucleation, polymerization, and turnover.
Accumulating evidence suggests that PIPs form a major structural link
between the plasma membrane and the actin cytoskeleton. PIPs are
generated by PIKs from membrane phospholipids inserted in the inner
leaflet of the plasma membrane. Important PIKs include
phosphatidyl-inositol-3-kinase (PI3K) and type I
phosphatidyl-inositol-5-kinase (PI5KI). PI3K has been shown to be
directly activated by Gß [83
]. Activated PI3K, in
turn, phosphorylates membrane PIPs at the 3-OH position as well as
other effectors at serine/threonine residues [84
,
85
]. The predominant PIP phosphorylation products
phosphatidylinositol-3,4-bisphosphate (PIP2) and
phosphoinosite-3,4,5-monophosphate (PIP3) are thought to
form multimeric aggregates inserted in the inner leaflet of the
membrane bilayer [86
87
88
]. Clustered PIP2
and PIP3 can be directly bound by proteins that contain
either a pleckstrin homology (PH) domain, Src homology (SH) domains
(e.g., SH2, SH3), or other PIP-binding domains [89
].
Several ABPs, including Scar/WASP, gelsolin, phdA (PH domain containing
protein I), and akt/PKB (also known as Rac protein kinase or protein
kinase B), as well as the signaling molecules guanine nucleotide
exchange factors (GEFs) [90
] and phospholipase C
[45
], have been shown either to contain such PIP-binding
domains or to directly bind PIPs [89
]. By interacting
with PIPs, these proteins link the plasma membrane to the cortical
cytoskeleton and cytoplasmic-signaling apparatus. Within seconds after
stimulation by cAMP, akt/PKB [87
], phdA
[85
], Scar1, and other proteins accumulate at the
leading edge, and it has been predicted that this recruitment is PIP
dependent [79
, 85
, 86
,
89
, 91
]. The amount of cortical actin
polymerization is proportional to the generation of PIPs, suggesting
that PIPs favor actin polymerization [92
,
93
]. In reverse, as PIP levels fall, the reaction of
filament growth is reversibly terminated [92
], which is
consistent with findings on Dictyostelium mutants that lack
PI3K or cells treated with PI3K inhibitor. These cells, presumably
because of insufficient PIP generation, fail to recruit pkB, phdA, and
other adapters to the leading edge [85
]. As a
consequence, PI3K-deficient cells fail to rapidly assemble F-actin at
the leading edge and to polarize, produce fewer pseudopods, and do not
undergo chemotaxis normally [85
]. Taken together, local
generation of PIPs and consecutive recruitment of actin binding and
regulatory factors initiate and maintain uncapping, actin anchoring,
and branching for spatially regulated filament growth.
PIP generation is further controlled by other signaling pathways that include low-molecular-weight GTPases of the Ras and Rac family, src-related tyrosine kinases, and other kinases. Besides initiating actin polymerization, PIPs have been shown to induce downstream signaling towards mitogen-activated protein kinase/extracellular regulated kinase pathways, Ras and Rac pathways, and protein kinase A- and cGMP-dependent effectors [26 , 94 ]. This suggests that the generation of PIPs and membrane-proximal actin polymerization accompany quite different stimuli and signal transduction pathways.
Another important PI3K substrate in Dictyostelium is Akt, the homologue to mammalian protein kinase B (akt/PKB). After phosphorylation, Akt/PKB becomes localized to the leading edge [95 , 96 ]. In turn, Akt/PKB can phosphorylate multiple effector proteins, regulating not only chemotactic response but also endocytosis, gene expression, and apoptosis [84 ]. Akt/PKB-deficient Dictyostelium cells, similar to cells lacking PI3K function, exhibit a polarization defect, aberrant chemotaxis, and reduced aggregation capacity [87 ].
The Ras and Rac family of small GTPases is involved in various pathways controlling the actin cytoskeleton. Dictyostelium cells express >40 different members of the Ras and Rac subfamily of GTPases [for complete list, see ref. 97 ] and the related GEFs. Ras and Rac GTPases interact with multiple cytoskeletal and adapter proteins that control cytoskeletal remodeling and cell motility, including ruffling, polarization, cortical stiffness, and cytokinesis [97 ]. Rac GTPases contain a PH domain for membrane localization and putative PIP binding on chemotactic response [98 , 99 ], although this feature remains to be confirmed for Dictyostelium. More than 20 downstream targets of Rac GTPases have been reported so far, among them Rho-dependent serine/threonine kinase (ROCK), myosin light-chain phosphatase, and gelsolin [100 , 101 ]. Many of the GTPases have been genetically ablated in Dictyostelium resulting in defined loss-of-function phenotypes. As examples, deletion of the small GTPase RasG, Rac1A, or Rac1b leads to severe impairment of cell polarity and single-cell movement [97 , 102 ]. On the other hand, deletion of RasS or its related exchange factor, RasGEFB results in a threefold increase in migration velocity compared with that of wild-type cells [103 ]. How downstream effectors mediate these different effects remains to be determined. Many mutations of GTPases show no apparent defects in single-cell crawling; however, they severely affect the sorting and differentiation process at the multicellular stages [97 ]. It is therefore likely that each Ras and Rac GTPase controls quite specific cytoskeletal events at a given developmental stage.
Besides its occurrence on chemotaxis, pseudopod extension has been suggested to occur as a spontaneous process involving constitutive oscillation of signaling molecules and cyclic baseline actin polymerization and depolymerization. High-resolution morphometric analysis of shape changes demonstrates that Dictyostelium cells are intrinsically vibrating bodies, transited by "self-organized, superpositioned, harmonic modes of rotating oscillatory waves" [104 ]. These morphological oscillations appear to be associated with intrinsic physicochemical oscillations of actin polarization leading to pseudopodal extensions and retractions [104 ]. In other words, the contractile apparatus of Dictyostelium exhibits a certain degree of autochthonous rhythmicity. In the absence of external triggers, this inherent oscillatory apparatus can lead to constitutive, nondirectional crawling that, as mentioned above, follows a model of persistent random walking [31 ].
|
ADHESION AND DETACHMENT MECHANISMS IN METAZOAN CELLS AND COMPARISON TO DICTYOSTELIUM |
|---|
|
TOPABSTRACTINTRODUCTIONTHE LIFE CYCLE OF...PARADIGM OF AMOEBOID MOTION,...CYTOSKELETAL ORGANIZATIONSIGNALING CONTROL OF ACTIN...ADHESION AND DETACHMENT...GENERAL LOCOMOTOR...STRUCTURE AND FUNCTION OF...SIGNALING CONTROL OF...FUNCTION OF ADHESION RECEPTORS...DIRECTIONAL GUIDANCE OF...CONCLUSIONSREFERENCES |
|---|
Haptokinetic migration
In the model of haptokinetic (i.e., adhesion-dependent) migration
of fibroblasts across ligand-coated surfaces [11
,
17
, 109
], migration efficiency follows an
inverse-U function relative to the adhesive force between the cell and
the underlying substrate (Fig. 1A
). Maximum migration rates are achieved at intermediate strength
of adhesion to the substrate [17
], allowing the
formation of new interactions at the leading edge, while preexisting
bonds are sufficiently released at the trailing edge (Fig. 1A
,
). Increasing adhesion, e.g., by increasing ligand density
or by up-regulating adhesion receptor density, leads to locomotor
reduction (Fig. 1A
, 
), as a consequence of impeded or
delayed detachment form the rear [18
, 110
,
111
]. On the other hand, reducing substrate adhesion to a
minimum, e.g., by decreasing ligand density or by adhesion-perturbing
antibodies, causes a loss of contact to the substrate [Fig. 1A
,
). As net interaction and traction forces are further
lowered, partial detachment and rounding up of the cells are
accompanied by nonproductive oscillatory membrane dynamics and impaired
migration (Fig. 1A
, ) [101
,
112
]. Besides fibroblasts and keratinocytes, the
haptokinetic migration model on 2-D substrate has been confirmed for
myoblasts migrating across a ligand-coated surface
[113
], as well as for fibroblast-like cells (e.g.,
melanoma cells) penetrating 3-D collagen matrices [21
].
|
Focal adhesion
Focal adhesions contain adhesion receptors (in mammals, integrins)
and cytoskeletal and signaling molecules in multimolecular complexes of
0.5–>2 µm in diameter [114
]. Integrins bound to
extracellular ligands become linked to the actin cytoskeleton via
several adapter and signaling proteins, such as talin, -actinin,
filamin, focal adhesion kinase, and others [114
,
115
]. Fully matured focal adhesions represent relatively
stable cell-substrate interactions that persist for longer periods.
Long-lasting contact leads to integrin-ligand strengthening and
bundling of preexisting actin filaments in parallel to longitudinal
thick actin bundles, termed stress fibers, inserting into these
attachment zones. The formation of fully matured focal
adhesions and stress fibers can be induced by activated
low-molecular-weight GTPase Rho via downstream effector ROCK
[101
, 116
]. Stress fiber formation is
counteracted by actin-severing proteins such as cofilin or gelsolin
[117
], as shown by inhibitory targeting of severing
activity. In fibroblasts, Rac has been shown to mediate cortical actin
turnover and filopodal ruffling via activation of gelsolin, thereby
preventing stress fiber formation [100
].
Focal adhesions and stress fibers appear to maintain a critical basal adhesion strength and traction. In slowly spreading and/or migrating cells, focal adhesions develop within 10 to 20 min, which is a time-consuming process [114 ] resulting in relatively stable and long-lived interactions towards the substrate that are inversely proportional to pseudopodal dynamics [117 ] and migration speed [11 , 18 , 101 ]. This does not exclude that slow turnover of focal adhesions and stress fibers can contribute to slow motion by gradual translocation. On promigratory stimulation, however, fibroblasts and keratinocytes disassemble preformed stress fibers and focal adhesions, decrease adhesiveness, and, consequently, increase the turnover of adhesion receptors for migration [118 ]. In mammals, the control of focal adhesion and stress fiber disassembly has been shown for the Rho family member Rnd1 [112 ], counteracting stress fiber assembly by the Rho and ROCK pathways [101 ]. Rnd1 appears to favor cell detachment from the rear, increasing migration speed; however, if Rnd1 is constitutively active, substrate adhesion becomes disrupted, and the cells round up, detach, and stop migration [112 ].
Focal contact
The focal contact is a smaller, less developed, and more transient
relative to focal adhesions [114
]. Focal contacts
contain smaller clusters of adhesion receptors and a reduced array of
cytoskeletal and signaling elements, which are not linked to
stress fibers but rather to a more diffuse cortical F-actin
[114
]. Focal contacts are considered to represent more
dynamic junctions predominantly under the control of Rac and Cdc42
[101
, 119
].
Highly dynamic surface structures, such as microvilli, filopodia, and pseudopodia, frequently display no apparent structural-compartment formation of proteins in the plasma membrane. Here, F-actin, adhesion receptors, and signaling molecules show a rather diffuse distribution, indicating high molecular dynamics, lateral mobility, and turnover of protein-protein interactions.
Detachment mechanisms
Membrane flow models for locomotion predict that membrane
extension at the leading edge requires the insertion of new membrane
from intracellular sites, such as intracellular vesicles
[120
]. Endosomes containing internalized adhesion
receptor and plasma membrane are formed at the trailing edge and
traffic towards the leading edge for receptor and membrane recycling at
pseudopod protrusions [120
]. For crawling
Dictyostelium cells, endocytosis and turnover rates of
fluorescently tagged surface membranes have been established in the
range of one cell surface equivalent every 4 to 8 min
[121
], indicating high membrane turnover rates.
If high substrate adhesivity counteracts receptor release from the substrate, as seen in migrating fibroblasts and tumor cells, detachment can be further achieved by the deposition ("shedding") of adhesion receptors, such as integrins and CD44 [19 , 110 , 122 ]. One mechanism of receptor shedding is proteolytic cleavage of cytoskeletal anchorage points by the protease calpain as well as the cytoplasmic portion of integrin adhesion receptors . Calpain cleaves the adapter proteins talin and paxillin [125 , 126 ]. By degrading cytoskeleton-membrane junctions, calpain is considered an important regulator of focal adhesion disassembly and cell detachment in fibroblasts [18 ]. Other cells, however, such as crawling leukocytes, appear to detach independently of calpain-mediated cleavage, suggesting that rapidly moving amoeboid leukocytes could use different mechanisms to regulate adhesive release at the cell rear than more adhesive, slowly moving cells [18 ].
Adhesion receptors in Dictyostelium
Although the generation of pseudopodal extension is well
understood in Dictyostelium, candidate adhesion receptors in
regions that contact the underlying substrate remain to be identified
[127
]. So far, neither tyrosine kinase receptors nor
integrin homologues have been identified in Dictyostelium.
Known transmembranous adhesion receptors include gp80/contact site A
(csA) [33
, 128
], gp24/DdCAD-1
[129
], gp150 [130
], and DTFA
(defective in tip formation) [131
]. gp80/csA mediates
EDTA-resistant homotypic end-to-end adhesion in aggregating cells,
which can be blocked by antibody [128
]. In addition,
gp80/csA-defective cells develop increased substrate adhesiveness and
delayed migration (compare state , Fig. 1A
), suggesting a
function of gp80/csA in the regulation of de-adhesion
[33
]. gp24/DdCAD-1 is a secreted nontransmembrane
protein with sequence similarities to mammalian cadherins which binds
to the extracellular surface and mediates calcium-dependent cell-cell
adhesion in early morphogenesis [132
]. gp150 is a
candidate receptor involved in the sorting of prestalk and prespore
cells at the mound stage [130
]. DTFA is involved in
cell-cell interaction at early aggregation as well as a later sorting
stage [131
]. So far, none of these receptors have been
implicated in binding to the underlying substrate and in single-cell
crawling or chemotaxis. Because specific adhesion receptors remain to
be identified in Dictyostelium, it is difficult to relate
the mechanism of amoeboid substrate interactions with principles of
haptokinetic migration to mammalian cells. In nature,
Dictyostelium is confronted with substrates of very diverse,
nonuniform physicochemical properties, such as leaves and soil. It
therefore remains subject to speculation whether engagement of specific
adhesion receptors would be a useful strategy for interaction with such
diverse environmental substrates after all.
Lack of focal adhesions/contacts in Dicytostelium
The focal-adhesion model has been mainly established using cells
from higher eukaryotes, whereas in Dictyostelium amebae,
adhesion receptors that mediate contact to substrates other than cells
are unknown. In migrating Dictyostelium cells, on-off rates
of cell substrate interactions are high (in the range of seconds to
minutes), whereas long-lasting stable attachment to the underlying
substrate is lacking. Cell propulsion and traction appear to result
from highly volatile, low-affinity interactions rather than stable
substrate binding. Consequently, focal adhesion- or focal contact-like
structures are not formed, which is consistent with the concept that
migration dynamics prevent focal adhesion formation and vice versa.
Genetically modified ponticulin-deficient Dictyostelium
cells lack approximately 90% of the normal number of high-affinity
interactions between actin filaments and the plasma membrane, but
migration and phagocytosis exhibit no functional alterations
[65
]. Together, these morphological and functional
characteristics indicate a low degree of cytoskeletal-compartment
formation coupled to low-affinity interactions between cytoskeleton,
plasma membrane, and extracellular substrate. Besides in
Dictyostelium, such diffuse membrane and cytoskeletal
organization appears as a general feature present in fast moving
leukocytes and other cells in the absence of focal contacts and stress
fibers [4
, 14
, 104
,
133
]. One conclusion of the above findings is that the
cellular and molecular migration mechanisms diverge between amoeboid
crawling and more adhesive haptokinetic migration strategies in
fibroblasts and other cells [14
].
|
GENERAL LOCOMOTOR CHARACTERISTICS IN CRAWLING LEUKOCYTES |
|---|
|
TOPABSTRACTINTRODUCTIONTHE LIFE CYCLE OF...PARADIGM OF AMOEBOID MOTION,...CYTOSKELETAL ORGANIZATIONSIGNALING CONTROL OF ACTIN...ADHESION AND DETACHMENT...GENERAL LOCOMOTOR...STRUCTURE AND FUNCTION OF...SIGNALING CONTROL OF...FUNCTION OF ADHESION RECEPTORS...DIRECTIONAL GUIDANCE OF...CONCLUSIONSREFERENCES |
|---|
T cell and neutrophil polarization and migration follow a stereotypic intrinsic program in response to different external stimuli. These include chemokines, chemical or mechanical triggers, increase in temperature, contact with ECM ligands, or physical contact with other cells (e.g., endothelium and antigen-presenting cells) [2 , 6 , 136 , 137 ]. Similar to Dictyostelium, migrating leukocytes adopt a polarized morphology (Fig. 2A ) consisting of a leading edge (Fig. 2A and 2B , asterisk) that generates one or several anterior lamellipodia, an elliptoid cell body containing the nucleus, and a tail-like uropod (Fig. 2A , black arrowhead). Both in vitro and in vivo, this prototypic morphology is conserved among different leukocytes, including T cells [4 ], neutrophils [138 ], and dendritic cells [139 , 140 ]. Leukocytes display an extraordinary deformability suitable for flexible adaptation to biophysical constraints of the substrate, such as pores and spaces [141 ]. While the cytoplasm in locomoting leukocytes is flowing forward, lateral protrusions ("footholds") move backwards relative to the advancing cell body (Fig. 2B) .
|
|
STRUCTURE AND FUNCTION OF THE CYTOSKELETON IN LEUKOCYTES |
|---|
|
TOPABSTRACTINTRODUCTIONTHE LIFE CYCLE OF...PARADIGM OF AMOEBOID MOTION,...CYTOSKELETAL ORGANIZATIONSIGNALING CONTROL OF ACTIN...ADHESION AND DETACHMENT...GENERAL LOCOMOTOR...STRUCTURE AND FUNCTION OF...SIGNALING CONTROL OF...FUNCTION OF ADHESION RECEPTORS...DIRECTIONAL GUIDANCE OF...CONCLUSIONSREFERENCES |
|---|
Capping protein homologues controlling initial filament unmasking
include gelsolin, villin, radixin, capZ, adseverin, and -actinin.
Filament severing is achieved by cofilin, depactin, and actophorin
[151
]. Similar to Dictyostelium, Arp2/3 acts
as a key nucleator of branched actin polymerization in neutrophils
[68
, 152
]. Other homologous branching
proteins, such as filamins (homologues to ABP-120 and ABP-280 in
Dictyostelium) and spectrin also contribute to branching and
formation of a filament meshwork [151
]. Filament
cross-linking to parallel bundles is provided by fimbrin, -actinin,
and calpactin. Anchoring to membrane receptors occurs through filamin,
talin, and -actinin. Filamin, talin, and -actinin have been shown
to directly interact with the cytoplasmic portion of ß1 and ß3
integrins and, to some extent, ß2 integrins, representing major
structural links between the actin cytoskeleton and integrins
[115
, 153
154
155
].
The kinetics and spatial distribution of actin assembly and disassembly are very similar in leukocytes and Dictyostelium. In nonmigrating spherical neutrophils, F-actin comprises approximately 30% of total actin and is diffusely distributed throughout the cytoplasm [156 ]. On chemokine triggering, initial actin polymerization occurs within <15 s [156 ]. After 30 s, the content of F-actin has doubled, and most newly assembled actin is detected in the proximity to the emerging leading lamella [156 , 157 ], whereas the cell body and uropod contain a relatively stable cortical rim of F-actin for polarity and stability [158 , 159 ]. In electron microscopic studies from neutrophil cytoplasmic lysates, individual actin filaments elongate rapidly to reach a maximum length after 15–30 s of 0.5–2 µm [158 , 160 ]. Actin filaments at the leading lamella are rapidly remodeled based on polymerization and depolymerization [158 ]. After removal of chemoattractant, actin depolymerization to basal levels is reached within 3–10 s, preferentially at the leading edge [158 ]. This time scale is consistent with in vitro estimation of actin turnover in the range of 10–30 s [161 ].
Similar to neutrophils, in T cells, chemoattractant-induced onset of migration is complete after 1–2 min [136 ]; however, more detailed information on actin filament structure and regulators of polymerization in T cells is sparse. In nonmigrating T cells residing within a 3-D collagen matrix, some focal nucleation of actin appears at contacts to collagen fibers [162 ]. With the onset of T cell locomotion, cortical F-actin is present along the rim of the cell body and is enriched in the uropod [162 ], whereas F-actin distribution is highly variable at the leading edge [4 , 159 , 162 ]. Taken together, in crawling leukocytes polymerized actin forms a highly dynamic and flexible leading edge in conjunction with a relatively stiff polarized core body and uropod.
|
SIGNALING CONTROL OF CYTOSKELETAL DYNAMICS IN LEUKOCYTES |
|---|
|
TOPABSTRACTINTRODUCTIONTHE LIFE CYCLE OF...PARADIGM OF AMOEBOID MOTION,...CYTOSKELETAL ORGANIZATIONSIGNALING CONTROL OF ACTIN...ADHESION AND DETACHMENT...GENERAL LOCOMOTOR...STRUCTURE AND FUNCTION OF...SIGNALING CONTROL OF...FUNCTION OF ADHESION RECEPTORS...DIRECTIONAL GUIDANCE OF...CONCLUSIONSREFERENCES |
|---|
or -
3, akt/PKB, Vav, and WASP
[reviewed in ref. 167
]. The importance of PIKs in actin
polymerization and leukocyte motility has been shown using PI3K
inhibitors. The PI3K inhibitor wortmanin leads to severe impairment of
polarization and chemotaxis in neutrophils [168
,
169
] and T lymphocytes [170
], indicating
that, similar to that in Dictyostelium, PI3K is a key
regulator of actin dynamics and motility in leukocytes. Conversely,
exaggerated PIP formation might overstate action polymerization and
delay turnover. PI5KI overexpression in migrating fibroblasts greatly
delays detachment and migration rates, suggesting that balanced amounts
of PIPs are important for cytoskeletal assembly, force generation, and
filament turnover [117
].
WASP and N-WASP, homologues to Scar1 in Dictyostelium, are
important for the induction of actin dynamics and migration in
leukocytes and other cells [171
]. (Wiskott-Aldrich
syndrome is an inherited disorder in humans, caused by a deficiency of
WASP protein function. The lack of WASP leads to severe defects in
leukocyte and platelet function such as recurrent pyogenic and
opportunistic infections, eczema, and thrombocytopenia. The life span
of affected individuals is reduced.) WASP directly interacts with
Cdc42, as shown by yeast two-hybrid screening, and it further has a PH
domain for PIP binding [172
]. After binding
Cdc42/PIP2 complex at the inner leaflet of the membrane,
WASP becomes activated [173
] and recruits and activates
Arp2/3 to form an actin nucleation complex for branched nucleation and
filament growth [174
]. Besides Cdc42 and
PIP2, other factors have been proposed as candidate
regulators of WASP activity, including profilin, phospholipase C-,
and proteins containing SH2 and SH3 motifs (e.g., Grp2 and src kinases)
[175
].
Similar to those in Dictyostelium, small GTPases are central regulators of actin polymerization and motility in higher eukaryotes. Mammalian Rho GTPases share structural and functional homology to Rac homologues in Dictyostelium [97 ]; however, only three family members, Rac, Cdc42, and Rho, have been identified [98 , 116 ]. Rho GTPases are activated by chemoattractant receptors, integrins, or growth factor receptors [176 , 177 ] as well as by intracellular signaling via src-related tyrosine kinases (Syc), the nuclear exchange factor Vav, or other small GTPases (Rho and Ras pathways) [119 , 178 ]. Rac and Cdc42 synergize for the induction of dynamic cytoskeletal events, i.e., the formation of ruffles and pseudopodia [116 ]. Rac forms a complex with PI5KI that is considered to increase PIP2 levels in the membrane [166 , 179 ]. Downstream effects of Rac to cytoskeletal dynamics and remodeling could be mediated by Rac-induced dissociation of capping proteins (e.g., gelsolin) from actin filaments to initiate filament growth [180 ]. Cdc42 binds PIP2 and WASP, as outlined above for the initiation of actin polymerization [181 ]. This is consistent with the finding, that, in T cell lines, dominant-negative Cdc42 abrogates chemokine-induced directional polarization and chemotaxis [182 ]. Because both contribute to actin filament growth, Rac and Cdc42 are considered candidate initiators and regulators of migration-driving membrane-cytoskeleton junctions of high turnover, such as ruffles, filopodia, and lamellipodia [98 116 ] (In contrast to Rac and Cdc42, Rho is involved in focal adhesion assembly and the maturation of actin filaments into stress fibers, putatively via downstream activation of the serine/threonine kinase ROCK [116 , 119 ]. Constitutively, active Rho exaggerates stress fiber and focal adhesion formation, thereby retarding migration in motile fibroblasts [101 ]. This phenotype is similar to the effect of PI5KI overexpression leading to exaggerated PIP production and actin polymerization [117 ]. Conversely, reduction of Rho or ROCK activity leads to the loss of stress fibers while polarized ruffling remains fully intact, thereby increasing migration rates [101 , 117 ]. This is consistent with the effects of Rho inhibition in immobilized monocytic cells, leading to greater spreading dynamics and membrane oscillations [183 ] and suggesting that
200 s. From frames E to F, pseudopod extension
(white arrow) at a lateral portion of the cell body initiates the
process of circum-migration across a horizontal fiber in the absence of
structural remodeling of the matrix structure. Variations of overall
fiber architecture from frame to frame result from manual focussing of
the microscopic stage in depth while the cell is crawling. Scale bar,
10 µm. _art;1>