Cell, Vol. 87, 601–606, November 15, 1996, Copyright 1996 by Cell Press
Getting Membrane Flow and the
Cytoskeleton to Cooperate
in Moving Cells
Mark S. Bretscher
MRC Laboratory of Molecular Biology
Hills Road
Cambridge CB2 2QH
England
This year sees the 25th anniversary of the discovery by
Taylor and colleagues (1971) of the phenomenon of antigen capping—the remarkable process whereby surface
antigens on a migrating eukaryotic cell, when extensively
crosslinked by antibodies to form patches and made visible by a fluorescent tag, move to the trailing end of the cell.
Although sometimes regarded as an artificially induced
process and therefore of little consequence, this is, I think,
a mistake: cap formation is restricted to motile cells and
can tell us much about how cells migrate. Furthermore,
cell polarity is of increasing importance in understanding
developmental processes. Whether the polarity displayed
in cap formation is related to the polarities exhibited in
developmental processes (such as in some oocytes) remains to be discovered, but any easily observed polar
process in animal cells is currently of interest. The aim of
this review is to compare two different views of the molecular mechanism of cap formation, how the experimental
evidence lines up on each side of the argument and how
these views relate to cell locomotion.
Although the process of cap formation was discovered
by crosslinking antigens on the surfaces of lymphocytes
in suspension, the rearward migration of patches is more
easily identified on substrate attached cells. Because only
those cells that are in a locomotory mode cap surface
antigens—stationary cells do not do so—it has been evident since its discovery that cap formation is closely related to cell locomotion. How they may be connected is
not difficult to see: if the crosslinking molecules happen
to be attached to the substrate, the cell itself will move
forward as crosslinked patches move rearward. In this
case, the crosslinked surface antigens have transiently
become the cell’s feet. (Further advance could obviously
not occur unless the surface antigens were uncrosslinked
and recycled somehow to the front of the cell). A different
way of observing what is presumably the same process
is particle migration, which was discovered a few years
before cap formation (Marcus, 1962; Ingram, 1969; Abercrombie et al., 1970; Bray, 1970): particles attached to the
dorsal surface of a migrating cell move away from the
front of the cell, towards its rear. In this case, the attached
particles—which are usually about 100 nm in size—
become attached to many surface molecules and thus
act as massive crosslinking agents. Particle and patch
migration are thus seen to be aspects of the same underlying mechanism. But what is this?
The movement of patches or particles is a directed
process: it always occurs away from the advancing edge
of the cell, and requires metabolic energy. Any surface
molecules that are extensively crosslinked cap—neighboring untethered molecules do not do so. The rate of
particle, or patch, migration away from the leading edge
Review
of a fibroblast is a few microns per minute: this velocity
decreases towards the rear of the cell so that objects
collect over the nuclear region. And finally, the capping
process itself is completely inhibited in lymphocytes
(where it has been studied most) by preincubating the
cells with a combination of cytochalasin B and colchicine, drugs which are known to interfere with the organization of microfilaments and microtubules respectively.
Preincubation with either drug alone produces, at best,
only a partial inhibition. The fact that both drugs are
required suggests, in the simplest interpretation, that
cap formation can be brought about by some action
involving either the actin or tubulin cytoskeletons independently of the other. It is these crucial facts—the need
for crosslinking and for the cytoskeleton—that provide
a basis for thinking about how capping occurs. And it
is these same facts that have led to very different views
of the process of cap formation and hence of how cells
locomote.
Rakes Versus Flow
For simplicity, I focus on the most plausible cytoskeletal
mechanism—the rake—and the alternative membrane
flow scheme, which I favor (see Figure 1).
The rake scheme outlined here is based on the idea
that certain transmembrane proteins on the cell surface
link the outside world with that part of the actin cytoskeleton adjacent to the plasma membrane (de Petris, 1977),
and that this part of the cytoskeleton is continuously
moving rearward—in the reference frame of the cell.
This rearward movement causes the attached transmembrane proteins—the teeth—to comb the surface
of the cell. Capping is envisaged as occurring when
crosslinked aggregates—or particles—become sufficiently large that they are caught in the teeth of the rake
and hence are moved rearwards towards the back of
the cell. By contrast, uncrosslinked molecules would
remain unaffected by its passage. In this model the
rearward movement of actin filaments is balanced by
filament assembly at the front of the cell and depolymerization towards its rear.
Experimentally, there is no doubt that in fibroblasts
and several other cell types, actin filaments do form at
the leading edge (Wang, 1985; Cao et al., 1993; Forscher
and Smith, 1988), that actin filaments do continuously
migrate rearwards and then disassemble towards the
rear of the cell near the nucleus (Heath and Holifield,
1991). The observed actin flow generates a polarized
actin cycle in which the initial rate of rearward filament
migration is about the same as that of particles attached
to the cell’s dorsal surface. An essential requirement of
this scheme is that, once the teeth of the rake have
reached the sites of actin depolymerization, they find
their ways back to the front of the cell for reuse. This
return could be provided by forward diffusion in the
plasma membrane or by a polarized endocytic cycle.
Other cytoskeletal models have been advanced: these
usually invoke specific recognition between the cytoplasmic domains of just those molecules which have
Cell
602
Figure 1. Two Schemes for Particle Migration on Motile Fibroblasts
In both schemes, the cells are moving to the
left.
(A) Polarized actin cycle in which actin filaments polymerize at the leading edge, pushing the cell’s front forward to extend it. Actin
depolymerization ahead of the cell’s nucleus
generates actin subunits (g-actin) which diffuse to the cell’s front for reuse. Filamentous
actin, moving rearward, pulls tines (small
arrows) and associated particles on the cell
surface towards the cell’s rear. Particles or
patches often migrate with actin, although
this is not always the case.
(B) Polarized endocytic cycle in which exocytosis at the leading edge extends the cell
surface, bringing a fresh source of substratum attachments with it. Endocytosis by
coated pits occurs randomly on the cell surface, generating a lipid flow in the plasma
membrane which decreases with increasing
distance from the leading edge: this is indicated by the decreasing size of the arrows
along the plasmalemma. This flow pushes on
substratum-attached feet, providing a fluid
drive to advance the cell. Particles or patches on the cell surface are swept towards the back of the cell, where rearward movement due to
flow is balanced by random movement due to Brownian motion.
been crosslinked and the cytoskeleton. These models
do not explain how crosslinked lipid molecules or crosslinked glycophosphatidyl-inositol containing proteins
(such as Thy-1), which have no cytoplasmic domains,
can interact with the cytoskeleton and hence be capped.
Nor do most of them include a polarized actin cycle and,
as such, seem less viable.
The membrane flow scheme suggests that capping
surface antigens is effected by a polarized endocytic
cycle (Abercrombie et al., 1970; Bretscher, 1984), in
which endocytosis occurs randomly on the cell surface
and exocytosis at the front of the cell. As the sites of
endocytosis and exocytosis are spatially separated, a
flow of membrane necessarily occurs within the plasma
membrane from the sites of exocytosis (the front of the
cell) towards the sites of endocytosis (towards the rear
of the cell). Since coated pits, which are assumed to be
the principle endocytic agents, internalize domains of
the plasma membrane which are greatly enriched in
some proteins (such as the LDL- and transferrin-receptors) compared to other surface proteins (such as Thy-1),
the actual situation is more complex. The flow of membrane is most easily conceptualized as a rearward lipid
flow (Bretscher, 1976), with some proteins (e.g. LDLreceptors) cycling rapidly and others very slowly. A
plasma membrane protein which does not actively cycle, such as Thy-1, can diffuse about on the cell surface
sufficiently quickly that any consequent drift in its position caused by the lipid flow is small. However, should
such an antigen become extensively crosslinked by external agents, the diffusion coefficient of the resultant
patch would decrease and calculation indicates that the
flow would now sweep it to the rear of the cell (Bretscher,
1976). Capping, then, is seen as a direct consequence
of reducing the diffusion coefficient of an object lying
in the flowing bilayer.
Experimentally, exocytosis of cycling membrane does
occur at the leading edge of migrating fibroblasts
(Bretscher and Thomson, 1983; Ekblom et al., 1983; Hopkins et al., 1994) and coated pits are spread randomly on
the surface of these cells. This polarized endocytic cycle
therefore leads to a rearward lipid flow in the cell’s plasma
membrane, whose rate decreases from the front of the
cell towards its rear. Such a decreasing flow rate would
be expected to result in a decreasing rate of rearward
particle migration as the object moved away from the
front of the cell, as is observed (Harris and Dunn, 1972).
Furthermore, it is possible to estimate how fast rearward
migration should be for a particle near the leading edge
of a fibroblast, given the known rate of endocytosis by
coated pits. Calculation and observation agree within a
factor of about two. (Coated pits internalize about 2% of
the cell surface per minute; if added uniformly to the front
of a 100 m long triangular flat cell, this would produce a
frontal extension of 1 m/minute. Particles or Con A
patches move rearwards on migrating fibroblasts at
about 2 m/minute [Abercrombie et al., 1970,1972]).
In many respects, the predicted behaviors of patches
as they cap, whether viewed by a rake or by a flow
mechanism, are quite similar. Both leave uncrosslinked
molecules more or less randomly placed; as the aggregates become larger, so the effects of the teeth of the
rake, or the flow, predominate. Although these schemes
place quite different emphases on the role of the cytoskeleton, in neither of them is it clear how those drugs
which disrupt the actin or tubulin cytoskeletons actually
have their effects. Before considering this, the relationship of each scheme to how cells locomote is presented.
Cell Locomotion
These two different models of how capping occurs have
given rise to two views for how cells migrate. One class
of models proposes that extension of the cell’s leading
edge is effected by the cytoskeleton, although several
Review
603
different schemes—invoking different cytoskeletal elements—exist (see, for example, Condeelis, 1993; Mitchison and Cramer, 1996). In some, actin polymerization
itself provides the force to push the front of the cell
forward in much the same way that listeria (Tilney and
Portnoy, 1989) and vaccinia virus (Cudmore et al., 1995)
are propelled in the cell’s cytoplasm. In others, the force
is provided by actin filaments in conjunction with singleheaded myosins. Traction is presumably required by a
migrating cell in vivo as it pushes its way through a
matrix or between other cells. The force for this needs
to be applied to those cell surface receptors bound to
the extracellular framework—the feet of the cell, which
are usually integrins. This could be provided by the rearwardly moving cytoskeleton: if the cell’s feet, while
bound by the substratum, attach to the actin cytoskeleton (either directly or indirectly), a force would be transmitted through them to push the cell forward. In this
case, the feet could be the same molecules which constitute the tines of the rake which effect capping. Like
the tines, a mechanism would be required to bring them
from the rear of the cell up to the front for reuse and
this transport could occur either by diffusion or by the
endocytic cycle.
The alternative viewpoint sees locomotion from a
membrane perspective: endocytosed membrane is reinserted at the leading edge of the cell, extending the front
of the cell forward. The flow of lipid generated—which
causes particles to migrate rearward—pushes on those
feet attached to the substratum. These feet are, in effect,
patches of integrins crosslinked by the substratum: the
force provided by the lipid flow, which would normally
cap them, pushes against them to move the cell forward—a novel form of fluid drive. In addition, feet released from the substratum away from the front of the
cell could be endocytosed and included in that membrane transported through the cytoplasm and added
to the front of the cell (Bretscher, 1984; Lawson and
Maxfield, 1995). This would allow the extension of the
leading edge to make new attachments to the substratum for the cell to advance.
How can these two cycles—a polarized endocytic cycle and a polarized actin cycle—be distinguished? An
examination of the effects of drugs or of specific mutations might help.
Neuronal Processes
The advance of a neuronal growth cone at the tip of
an extending axon is similar to that of a fibroblast. In
particular, particles attached to their dorsal surfaces
migrate rearwards, suggesting that they use the same
basic mechanisms for advancing (Bray, 1970). The effects of drugs which disturb the cytoskeleton have been
most carefully examined in neurons. In the prolonged
presence of concentrations of cytochalasin B which appear to eliminate microfilaments, nerve cells can still
extend long processes (Marsh and Letourneau, 1984).
They are, however, abnormal: the growth cone has lost
its shape, being blunt-ended and packed with small
vesicles, the processes themselves do not extend in a
linear fashion as do normal neurites but tend to grow
in curves, and they only form on a substratum treated to
make it extremely adherent. Nevertheless, the “neurites”
can extend hundreds of microns. The authors suggest
that the actin network in normal growth cones enhances
the attachments they make with the substratum, and
assists filopodia and lamellipodia to form. This experiment shows that an actin-based system is not an absolute requirement for neuronal growth. The observations
that the neuronal ending is packed with vesicles and that
the existing microtubules and neurofilaments in these
“neurites” only extend into the bases of these bluntended growth cones (as they do in normal growth
cones), suggest that extension in this case occurs by
the deposition of membrane material at the distal tip of
the “neurite.”
The suggestion (Marsh and Letourneau, 1984) that a
role for the actin network at the leading edge is to enhance normal attachments is intriguing, especially as it
is consistent with observations of blebbing on migrating
cells. When blebs of membrane appear on the surface
of a migrating cell, they usually do so at the leading
edge. However, these balloons of apparently structureless membrane do not generally lead to new cellular
attachments which advance the cell; instead, they move
rearwards (like a particle) on the cell’s dorsal surface
and disappear as they get resorbed into the cell (Harris,
1973; Trinkaus, 1984). Blebs which appear at the cell’s
leading edge thus seem to be membrane which has
exocytosed there but, for unknown reasons, lacks both
the normal lamellar structure provided (presumably) by
the actin network and its adhesive qualities: this results
in a cell extension which has lost its grip.
Drugs which disrupt microtubules also permit some
form of neurite growth (Lamoureux et al., 1990), although
the extent of this seems to be rather modest. These
findings and those of Marsh and Letourneau (1984) have
been extended to analyze the roles of each filamentous
network separately in axonal transport, using mitochondria as markers (Morris and Hollenbeck, 1995). The results dramatically show that microfilaments and microtubules can independently advance mitochondria in
both directions—towards the cell body or the distal tip.
In the absence of both networks (but in the presence
of neurofilaments), no mitochrondrial movement occurs.
The fact that motors exist which can transport an organelle—the mitochondrion in this case—along microfilaments and microtubules suggests that the transcellular
movement of vesicles in general may also be assisted
by either set of filaments, although the efficiency with
which they do so may vary with the kind of vesicle and
the cell type. However, if endocytosed membrane also
makes use of both filamentous systems for transport to
the leading edge in migrating cells, it could provide the
basis for an explanation—in a flow scheme—for why
cap formation is eliminated in cells preincubated with
drugs which disrupt both filamentous systems, but is
comparatively insensitive in cells treated with a drug
which disrupts only one of them (de Petris, 1978).
A detailed analysis of the effects of cytochalasin B on
Aplysia growth cones shows that, within seconds of
drug addition, actin filament assembly at the leading
edge ceases (Forscher and Smith, 1988; Lin et al., 1996).
Surprisingly, the flow of existing actin filaments away
from the leading edge continues. This demonstrates that
Cell
604
actin polymerization at the leading edge does not itself
drive the rearward flow of filamentous actin and adjacent
particles on the surface of the growth cone (Forscher
and Smith, 1988). The results do, however, indicate that
filopodial growth depends on microfilament assembly.
Recent studies in which (single-headed) myosins are
inhibited (Lin et al., 1996) suggest that the rearward
flow of microfilaments and associated particle migration
depend, at least in part, on myosin. Although these experiments do not indicate how the myosin acts, the
authors suggest that myosin in the growth cone could
be fixed to a frame (fixed with respect to the substratum)
and which could directly effect the rearward flow of
microfilaments and thereby movement of actin-linked
particles. However, other possibilities are not ruled out.
For example, endocytic vesicles associated with more
stable actin filaments could be cotransported by myosin
to the leading edge and thereby provide the putative
rearward flow of surface membrane needed, in a flow
scheme, to move the particle.
Recently, Dai and Sheetz (1995) have reexamined the
behavior of particles attached to the axons of growing
neurites: previous evidence indicated that such particles
remain stationary (Bray, 1970). By contrast, Dai and
Sheetz find that particles (attached through appropriate
antibodies to the axonal surface) move quite rapidly
along the axon, from the growth cone towards the cell
body. They propose that this movement is driven by an
underlying endocytic cycle, in which endocytosis in the
cell body is approximately balanced by exocytosis in the
growth cone. The authors report differences in surface
tension between different positions along an axon
which, they suggest, could support this flow. The particles, then, move up the axon with the surface membrane—or lipid—flow in the same way proposed for particles attached to the dorsal surfaces of fibroblasts
(Bretscher, 1984). Although the actual site of exocytosis
within the growth cone is not known, parallels with fibroblasts would imply that it occurs at the leading edge.
In summary, actin polymerization occurs right at the
leading edge of a growth cone, where it may help advance the leading edge (as described for fibroblasts).
Its presence there, however, is not an absolute requirement for neurite extension; long-term treatment of neurons with cytochalasin B allows some neurite growth,
and the results suggest that filamentous actin in a normal growth cone may enhance its attachment to the
substratum. The rearward flow of actin and of surfaceattached particles appear not to be driven by actin polymerization at the leading edge, but the involvement of
myosin in both processes is likely. Particles attached to
the surface of some axons migrate towards the cell
body; this movement seems to be driven by a polarized
endocytic cycle. Axonal transport of mitochondria—and
hence possibly other membraneous organelles—can be
effected by both microtubules and microfilaments. This
raises the possibility —pure speculation at this stage—
that the reason cap formation in lymphocytes is eliminated after the cells have been incubated with drugs
which inactivate both cytoskeletal systems is because
each system contributes, in parallel, to the intracellular
transport of endocytosed membrane.
Genetic Evidence
Genetics is usually the salvation of biologists trying to
solve complex problems: cell migration is something of
an exception. Two rather unsatisfactory “results” exist.
First, deletion of conventional myosin from Dictyostelium might be expected either to eliminate locomotion
(if sliding filaments are directly involved in lamellar extension), or to have no effect: in fact, the result is in
between (Wessels et al., 1988). Furthermore, although
rearward particle movement is unaffected in this mutant
(Jay and Elson, 1992), cap formation—always regarded
as closely related to particle movement—is reported to
be inhibited (Pasternak et al., 1989), an unhappy result.
Many other mutations affecting components of the
Dictyostelium cytoskeleton—and combinations of them—
do not seriously impair locomotion. Second, any process that eliminates endocytosis should, in flow models,
halt locomotion. The shibire-ts1 mutation, a dynamin
mutant originally discovered in Drosophila, reversibly
blocks endocytosis by coated pits and thereby paralyzes flies above 308C. This mutation seemed ideal for
investigating the role of endocytosis in cell locomotion.
It has been transplanted into mammalian tissue culture
cells, where receptor-mediated endocytosis becomes
reduced above the restrictive temperature (Damke et
al., 1995), but its effect on locomotion has not been
reported. However, any effect it has may only be small
because, when these cells are shifted to the restrictive
temperature, fluid phase endocytosis is initially only partially inhibited and fully recovers within an hour. Other
kinds of endocytosis seem to exist, and the value of
shibire in this context is thus greatly diminished unless
motile cells can be found whose only endocytic route
is via clathrin-coated pits. It seems evident that many
basic cell biological processes are too vital to be left in
the hands of a single gene, making a genetic approach
less straightforward.
Conclusions
I have tried here to provide a balanced view of cap
formation and its extension to cell locomotion, as seen
from two different perspectives.
The cytoskeletal model described is based on a polarized actin cycle (Bray and White, 1988). Filament assembly occurs at the leading edge where it seems to
strengthen substrate attachments. Actin may also lead
to filopodial extension and help push the leading edge
forward either directly by polymerization or in conjunction with other microfilament-associated proteins. Filaments adjacent to the plasma membrane flow from the
leading edge towards the rear of the cell or growth cone,
where they disassemble. This cycle is completed by the
transfer of actin from the back to the front for reuse,
presumably by actin subunits diffusing forward through
the cytoplasm. The cell’s feet, or the rake’s tines, poke
through the plasma membrane and, being attached to
the actin network, also migrate rearwards. This could
provide the motive force for capping and cell locomotion. How the actin network is moved backward in this
scheme is unclear: myosin, the most likely motor, would
have to be linked to a structure to provide a buffer
against which to push. Other models for effecting the
Review
605
actin flow can be envisaged which incorporate different
microfilament-associated proteins, many of which are
well-characterized molecules (Bretscher, 1991). This
general scheme, as presented, has no obvious role for
microtubules in cap formation.
I favor a flow mechanism, driven by a polarized endocytic cycle in which both microtubules and microfilaments effect the intracellular transport of endocytosed
membrane, because it appears to explain cap formation
and particle migration naturally. In addition, the experimental evidence for the sites of endocytosis and exocytosis and the speed at which the endocytic cycle
operates seem able to explain the observed phenomena, at least in fibroblasts. I therefore see related processes from that viewpoint, and offer three examples
of this.
First, the rearward movement of filamentous actin in
migrating fibroblasts may be caused by its association
with, or proximity to, a rearwardly flowing membrane.
This could fit with the observation (Forscher and Smith,
1988) that actin polymerization at the front of a growth
cone is not needed for the rearward movement of microfilaments. However, this possibility would require that,
in the few minutes after cytochalasin B addition when
these observations are made, membrane continues to
be transported, and added, to the leading edge in the
apparent absence of adjacent microtubules and microfilaments.
Second, the shape of a migrating cell could depend
on the efficiency with which its attaching feet circulate.
If they are able to detach easily from the substratum
and circulate quickly, feet would mainly exist near the
leading edge of the cell; this is where other rapidly circulating receptors, such as the transferrin receptor, are
found (Ekblom et al, 1983; Bretscher, 1983, 1996; Hopkins, 1985; Hopkins et al, 1994). This might cause the
cell to be snail-shaped (see also Lawson and Maxfield,
1995). Alternatively, if they circulate rather slowly (as
do a 5b1 fibronectin receptors on fibroblasts [Bretscher,
1989]), the feet would be more uniformly distributed
and the cells would be long and stretched out, as are
fibroblasts on a fibronectin substratum.
Third, the speed at which a cell migrates would depend on the rate at which adhesive membrane extends
the cell forward. For a given surface area of the cell,
this would depend on the rate of surface uptake and
the breadth of the leading lamella: the narrower the front,
the more quickly the cell could move.
These views may be wrong, but at least they may help
focus the picture and some of them may be testable.
For example, it is possible that fish scale keratocytes
move too quickly for their advance (about 30 m/minute)
to be explained by a membrane circulation scheme—but
then their rates of membrane circulation have not been
measured. Indeed, it is difficult to measure the absolute
rate of constitutive endocytosis in any cell because part
of the internalized membrane (e.g. LDL-receptors in fibroblasts) may spend so little time—a fraction of a minute—inside the cell before returning to the cell surface.
A method for measuring the rate of area uptake by individual cells over a period of a few seconds is therefore
required. Although no such method exists at present,
that which comes closest to achieving this uses a dye,
such as FM1–43, which rapidly partitions into only the
external leaflet of the lipid bilayer and can therefore be
used to measure surface uptake (see, for example, Smith
and Betz, 1996); it is unclear whether this method could
be made sufficiently sensitive to measure very short
periods of endocytosis during locomotion.
It seems probable that a polarized endocytic cycle
exists in migrating cells whose function is, at least in
part, to transport the cell’s feet from the rear to its front.
Whether this membrane cycle is sufficient to extend a
cell forward and to propel patches of crosslinked antigens rearward, or whether the polarized actin cycle is
the principle driving force for these movements, is the
nub of how a cell puts its best foot forward.
References
Abercrombie, M., Heaysman, J.E. M., and Pegrum, S.M. (1970). The
locomotion of fibroblasts in culture. III. Movement of particles on
the dorsal surfaces of the leading lamella. Exp. Cell Res. 62, 389–398.
Abercrombie, M., Heaysman, J.E. M., and Pegrum, S.M. (1972). Locomotion of fibroblasts in culture. V. Surface marking with concanavalin A. Exp. Cell Res. 73, 536–539.
Bray, D. (1970). Surface movement during growth of single explanted
neurons. Proc. Natl. Acad. Sci. USA 65, 905–910.
Bray, D., and White, J.G. (1988). Cortical flow in animal cells. Science
239, 883–888.
Bretscher, A.P. (1991). Microfilament structure and function in the
cortical cytoplasm. Annu. Rev. Cell Biol. 7, 337–374.
Bretscher, M.S. (1976). Directed lipid flow in cell membranes. Nature
260, 21–22.
Bretscher, M.S. (1983). Distribution of receptors for transferrin and
low density lipoprotein on the surface of giant HeLa cells. Proc.
Natl. Acad. Sci. USA 80, 454–458.
Bretscher, M.S. (1984). Endocytosis: relation to capping and cell
locomotion. Science 224, 681–686.
Bretscher, M.S. (1989). Endocytosis and recycling of the fibronectin
receptor in CHO cells. EMBO J. 8, 1341–1348.
Bretscher, M.S. (1996). Moving membrane up to the front of migrating cells. Cell 85, 465–467.
Bretscher, M.S., and Thomson, J.N. (1983). Distribution of transferrin
receptors and coated pits on giant HeLa cells. EMBO J. 2, 599–603.
Cao, L., Fishkind, D.J., and Wang, Y.-L. (1993). Localization and
dynamics of non-filamentous actin in cultured cells. J. Cell Biol. 123,
173–181.
Condeelis, J. (1993). Life at the leading edge. Annu. Rev. Cell Biol.
9, 411–444.
Cudmore, S., Cossart, P., Griffiths, G., and Way, M. (1995). Actinbased motility of vaccinia virus. Nature 378, 636–638.
Dai, J., and Sheetz, M.P. (1995). Axon membrane flows from the
growth cone to the cell body. Cell 83, 693–701.
Damke, H., Baba, T., van der Bliek, A.M., and Schmid, S.L. (1995).
Clathrin-independent pinocytosis is induced in cells overexpressing
a temperature-sensitive mutant of dynamin. J. Cell Biol. 131, 69–80.
de Petris, S. (1977). Distribution and mobility of plasma membrane
components in lymphocytes. In Dynamic Aspects of Cell Surface
Organisation, G. Poste and G.L. Nicolson, eds. (Amsterdam: North
Holland Press), pp. 643–728.
de Petris, S. (1978). Non-uniform distribution of Concanavalin-A receptors and surface antigens on uropod-forming thymocytes. J. Cell
Biol. 79, 235–251.
Ekblom, P., Thesleff, I., Lehto, V-P., and Virtanen, I. (1983). Distribution of the transferrin receptor in normal human fibroblasts and
fibrosarcoma cells. Int. J. Cancer 31, 111–116.
Forscher, P., and Smith, S.J. (1988). Actions of cytochalasins on
Cell
606
the organization of actin filaments and microtubules in a neuronal
growth cone. J. Cell Biol. 107, 1505–1516.
Harris, A.K. (1973). Cell surface movements related to cell locomotion. In Locomotion of Tissue Cells, CIBA Foundation Symposium
Volume 14. (Amsterdam: Associated Scientific Publishers), pp. 3–26.
Harris, A., and Dunn, G. (1972). Centripetal transport of attached
particles on both surfaces of moving fibroblasts. Exp. Cell Res. 73,
519–523.
Heath, J.P., and Holifield, B.F. (1991). Cell locomotion: new research
tests old ideas on membrane and cytoskeletal flow. Cell Motil. Cytoskel. 18, 245–257.
Hopkins, C.R. (1985). The appearance and internalization of transferrin receptors at the margins of spreading human tumor cells. Cell
40, 199–208.
Hopkins, C.R., Gibson, A., Shipman, M., Strickland, D.K., and Trowbridge, I.S. (1994). In migrating fibroblasts, recycling receptors are
concentrated in narrow tubules in the pericentriolar area, and then
routed to the plasma membrane of the leading lamella. J. Cell Biol.
125, 1265–1274.
Ingram, V.M. (1969). A side view of moving fibroblasts. Nature 222,
641–644.
Jay, P.Y., and Elson, E.L. (1992). Surface particle transport mechanism independent of myosin II in Dictyostelium. Nature 356,
438–440.
Lamoureux, P., Steel,V. L., Regal, C., Buxbaum, R.E., and Heidemann, S.R. (1990). Extracellular matrix allows PC12 neurite elongation in the absence of microtubules. J. Cell Biol. 110, 71–79.
Lawson, M.A., and Maxfield, F.R. (1995). Ca21 - and calcineurindependent recycling of an integrin to the front of migrating neutrophils. Nature 377, 75–79.
Lin, C.H., Espreafico, E.M., Mooseker, M.S., and Forscher, P. (1996).
Myosin drives retrograde F-actin flow in neuronal growth cones.
Neuron 16, 769–782.
Marcus, P.I. (1962). Dynamics of surface modification in myxovirusinfected cells. Cold Spring Harbor Symp. Quant. Biol. 27, 351–365.
Marsh, L., and Letourneau, P.C. (1984). Growth of neurites without
filopodial or lamellipodial activity in the presence of cytochalasin B.
J. Cell Biol. 99, 2041–2047.
Mitchison, T.J., and Cramer, L.P. (1996). Actin-based cell motility
and cell locomotion. Cell 84, 371–379.
Morris, R.L., and Hollenbeck, P.J. (1995). Axonal transport of mitochondria along microtubules and F-actin in living vertebrate neurons. J. Cell Biol. 131, 1315–1326.
Pasternak, C., Spudich, J.A., and Elson, E.L. (1989). Capping of
surface receptors and concomitant cortical tension are generated
by conventional myosin. Nature 341, 549–551.
Smith, C.B., and Betz, W.J. (1996). Simultaneous independent measurement of endocytosis and exocytosis. Nature 380, 531–534.
Taylor, R.B., Duffus, W.P. H., Raff, M.C., and de Petris, S. (1971).
Redistribution and pinocytosis of lymphocyte surface immunoglobulin molecules induced by anti-immunoglobulin antibody. Nature
New Biol. 233, 225–229.
Tilney, L.G., and Portnoy, D.A. (1989). Actin filaments and the growth,
movement and spread of the intracellular bacterial parasite, Listeria
monocytogenes. J. Cell Biol. 109, 1597–1608.
Trinkaus, J.P. (1984). Cells into Organs. (New Jersey: Prentiss-Hall,
Inc.), pp. 197–198.
Wang, Y.-L. (1985). Exchange of actin subunits at the leading edge
of living fibroblasts: possible role of treadmilling. J. Cell Biol. 101,
597–602.
Wessels, D., Soll, D.R., Knecht, D., Loomis, W.F., DeLozanne, A., and
Spudich, J.A. (1988). Cell motility and chemotaxis in Dictyostelium
amebae lacking myosin heavy chain. Dev. Biol. 128, 164–177.