Anda di halaman 1dari 10

1

IFOM, the FIRC Institute for Molecular Oncology Foundation, at the IFOM-IEO Campus, Via Adamello 16, 20139, Milan, Italy.
2
Department of Medicine, Surgery and Dentistry, Universit degli
Studi di Milano, Via A. Di Rudini 8, 20122, Milan, Italy.
3
Department of Experimental Oncology, European Institute of Oncology, at the IFOM-IEO Campus, Via Adamello 16, 20139, Milan, Italy.
The endocytic matrix
Giorgio

Scita
1,2
& Pier Paolo Di Fiore
1,2,3
Endocytosis has long been thought of as simply a way for cells to internalize nutrients and membrane-
associated molecules. But an explosive growth in knowledge has given a new dimension to our understanding
of this process. It now seems that endocytosis is a master organizer of signalling circuits, with one of its main
roles being the resolution of signals in space and time. Many of the functions of endocytosis that are emerging
from recent research cannot yet be reconciled with the canonical view of intracellular trafficking but, instead,
point to endocytosis being integrated at a deeper level in the cellular master plan (the cellular network of
signalling circuits that lie at the base of the cells make-up). Deconvolution of this level, which we call the
endocytic matrix, might uncover a fundamental aspect of how a cell is built.
Eukaryotic cells use endocytosis to internalize segments of plasma
membrane, cell-surface receptors, and various soluble molecules
(including nutrients) from the extracellular fluid. This is a complex
process, as underscored by the multiple routes by which molecules
can enter a cell through endocytosis. Clathrin-mediated endocytosis
1

(a mode of vesicular transport that is involved in the internalization
and recycling of receptors by endocytic vesicles coated with the pro-
tein clathrin) is the most extensively characterized route, but attention
is increasingly being paid to several mechanisms of non-clathrin-
mediated endocytosis
2
(Fig. 1).
For signal transduction, it is clear that endocytosis is one of the main
ways that signals can be attenuated, through the removal and degra-
dation of signalling receptors (and, in some cases, their ligands) from
the cell surface. Recent studies, however, have uncovered a wealth of
evidence that endocytosis has a much wider impact on signalling, inclu-
ding the finding that signalling pathways and endocytic pathways are
regulated in a reciprocal manner, and the finding that several molecules
have roles in both endocytosis and signalling (see refs 35 for reviews).
The emerging model is that the net biochemical output of signalling
pathways largely depends on topological constraints. These constraints
are imposed by the association of signalling molecules with membranes,
which in turn is regulated by endocytosis and by cycles of endocytosis
and recycling to the plasma membrane (that is, endocytic and exocytic
cycles (EECs)). This set-up allows signals to be decoded by the cell
according to precise kinetics and at spatially defined sites of action. And,
not surprisingly, it translates into endocytosis having a large impact on
almost every cellular process. In addition, evidence is emerging that the
endocytic machinery has molecular functions that are not immediately
reconcilable with membrane trafficking, leading researchers to ques-
tion whether these non-canonical functions are moonlighting jobs
or whether they point to deeper levels of integration of the endocytic
matrix within signalling circuitries and cellular programs.
Here we summarize the current understanding of how endocytosis
is embedded in the cellular master plan, and more specifically its con-
nections to signalling. We review endocytosis at the level of the circuits
involved, highlighting how the integration of endocytosis and signalling
determines the net biochemical output of a cell. Then, we analyse how
endocytosis affects the execution of complex cellular programs. And,
last, we speculate on how endocytosis might have evolved to become a
pervasive component of the cellular master plan.
The circuitry level
Numerous findings support the idea that the integration of signalling
and endocytosis determines the net output of biochemical pathways.
In this section, we discuss the emerging models of how endocytosis
controls signalling at the level of signalling circuits. A vast number of
studies have been published on this topic, so here we provide an over-
view of the basic concepts linking endocytosis and signalling. For an
in-depth analysis of the various issues, see refs 35.
Membranes and signalling effectors
Endocytosis regulates the assembly of signalling platforms at the plasma
membrane by modulating the presence of receptors, their ligands and
downstream effectors at the plasma membrane or at intermediate
stations of the endocytic route. The first consequence of endocyto-
sis is that signalling receptors disappear from the plasma membrane,
thereby limiting the magnitude of signalling from this source (Fig. 2).
Ligand availability can also be controlled by endocytosis, as is the case
for ligands in the DSL (Delta, Serrate and LAG-2) family, which acti-
vate receptors in the NOTCH family (see ref. 6 for a review) (Fig. 2).
The differential distribution of signalling effectors between the plasma
membrane and the endosomal compartment also functions to regulate
signals in time and space.
The integration of different endocytic routes is crucial for determin-
ing the net signalling output. In the case of the epidermal growth fac-
tor receptor (EGFR) and the transforming growth factor- receptor
(TGF-R), clathrin-mediated endocytosis couples the receptors with
recycling (and sustainment of signalling) and non-clathrin-mediated
endocytosis couples the receptors with degradation (and signalling
attenuation)
7,8
(Fig. 2). Notably, in other settings (for example in WNT-
activated pathways), the opposite is true
9
. Thus, although the association
of signalling or attenuation with each route of endocytosis is specific to
the receptor, the relative partitioning of receptors between the two entry
routes generally determines the final net signalling output.
There is an increasing amount of evidence implicating EECs as essen-
tial events in certain types of signalling. The recycling of internalized
receptors to the plasma membrane replenishes the cell surface with
ligand-free receptors. EECs seem, however, to have a much more active
role in signalling. For example, EECs resensitize G-protein-coupled
receptors (GPCRs) that have been rendered signalling impaired dur-
ing internalization (see ref. 3 for a review) (Fig. 2). In addition, EECs
464
INSIGHTREVIEW NATURE|Vol 463|28 January 2010|doi:10.1038/nature08910
464-473 Insight Scita NS.indd 464 21/1/10 11:01:07
20 Macmillan Publishers Limited. All rights reserved 10
E
S
C
R
T
RAB5
RAB11
ARF6
ARF6-dependent
recycling
Late endosome and
multivesicular body
Lysosome
RAB8
RAB7
Clathrin-mediated
endocytosis
Non-clathrin-mediated
endocytosis
RAB4
Fast
recycling
Dynamin
Clathrin
AP-2
Lipid raft
Ligand
MHC class I
molecule
Clathrin-coated pit
B7
Ub
E
S
C
E
SS
E
S
EE
R
T e and
body
Ub
Slow
recycling
Early endosome
RTK
Plasma
membrane
Ligand-
bound RTK
Degradation
restrict signals to limited regions of the plasma membrane during the
execution of polarized functions (see the subsection EECs and polar-
ized functions).
Finally, endosomes are key signalling stations, a concept that is
embodied by the term signalling endosome. Several other types of
intracellular membrane (or endomembrane) are also signalling plat-
forms
10
. Broadly, endosomes have a dual role in signalling
3
: they sustain
signals that originate from the plasma membrane (not shown), and they
generate unique signals that are prohibited at the plasma membrane,
thus contributing to signal diversification and specificity (Fig. 3).
The small volume of an endosome is a necessary feature for signal-
ling, because it favours receptorligand association and sustains receptor
activity
3
. The limited surface area of an endosome also generates the
ideal conditions for coincidence detectors; that is, molecular functions
that require two or more simultaneous, but relatively weak, interactions
11

(Fig. 3). In addition, endosomes are enriched in certain lipids and pro-
teins, such as the lipid phosphatidylinositol-3-phosphate (PtdIns(3)P)
and the lipid-raft adaptor protein p18, providing specific scaffolding
surfaces on which signalling complexes can be assembled (Fig. 3). Other
ideal features of endosomes include rapid microtubule-mediated trans-
port of molecules (which allows the transmission of signals over long
distances, such as from the plasma membrane to the nucleus; see the
next subsection) and acidic pH (which is necessary for a variety of spe-
cific signalling pathways) (Fig. 3).
It is notable that endosomes enriched in particular signalling mol-
ecules might be involved in specific signalling pathways (Fig. 3). This
is the case, for instance, for endosomes marked by the presence of the
protein SARA (SMAD anchor for receptor activation)
1214
, which are
known as SARA endosomes (an operational definition that reflects a
functional state rather than a distinct subpopulation of endosomes),
and for APPL (adaptor protein containing phosphotyrosine-interac-
tion domain, pH domain and leucine-zipper motif) endosomes
1517

(an early-stage precursor of the classic early endosome). These
endosomes are involved in signalling through TGF-R and through
receptor tyrosine kinases (RTKs), respectively. Endosome-specific sig-
nalling also occurs for several other receptor systems, such as GPCRs,
NOTCH-family members, tumour necrosis factor receptor 1 and Toll-
like receptors (see ref. 3 for a review). Thus, the signalling endosome
is a generic concept describing the existence of numerous functional
endosomal states that result in signal-specific platforms. This, in turn,
helps cells to distinguish between signals by attributing them to specific
receptor-activated pathways.
Figure 1 | Endocytic trafficking of signalling receptors.Signalling
receptors (in this example receptor tyrosine kinases (RTKs)) are mainly
internalized through clathrin-mediated endocytosis (left). In this pathway
of endocytosis, ligand binding accelerates the recruitment of receptors to
clathrin (present in clathrin-coated pits) through adaptors, such as AP-2
or -arrestins
1
. Clathrin then polymerizes, and this drives the invagination
of the pit, which is eventually released into the cytoplasm through the
action of the GTPase dynamin
1
. This process seems simple but is clearly
highly complex given that more than 50 different proteins can be found
in clathrin-coated pits. There are many forms of non-clathrin-mediated
endocytosis (right), which, in some cases, depends on plasma-membrane
microdomains enriched in particular lipids (known as lipid rafts). Non-
clathrin-mediated endocytosis is still poorly understood at the molecular
level, and the term encompasses many heterogeneous mechanisms
1,2
.
After internalization, by either clathrin-mediated endocytosis or non-
clathrin-mediated endocytosis, receptors are routed to early endosomes.
Trafficking in the endosomal compartment is controlled by small GTP-
binding proteins of the RAB and ARF (ADP-ribosylation factor) families
72

(some of which are indicated). From the early endosome, cargo is either
recycled to the plasma membrane (green arrows) or degraded (red arrows).
Cargo can be recycled through a fast recycling route (which depends on
RAB4) or a slow recycling route (which depends on the combined action
of RAB8 and RAB11)
72
. In addition, proteins that have been internalized
by non-clathrin-mediated endocytosis, such as major histocompatibility
complex (MHC) class I molecules, can be recycled to the plasma membrane
through ARF6-dependent pathways
73
. Cargo can also be trafficked through
a RAB7-dependent, degradative route, through late endosomes and
multivesicular bodies, and then lysosomes
72
. A crucial signal in this route is
ubiquitylation of the receptors. Ubiquitylated receptors are recognized by
a series of ubiquitin-binding protein complexes: HRSSTAM (also known
as ESCRT-0), and endosomal sorting complex required for transport I
(ESCRT-I), ESCRT-II and ESCRT-III (see ref. 74 for a review).
465
NATURE|Vol 463|28 January 2010 REVIEWINSIGHT
464-473 Insight Scita NS.indd 465 21/1/10 11:01:08
20 Macmillan Publishers Limited. All rights reserved 10
P
P
Lysosome
TGF-B
TGF-CR
TGF-C
EGF
RTK
(EGFR)
EGF
P
NOTCH
Cleavage and
activation
of NOTCH
Early endosome
Early endosome
b
g
f
d
a
e
In cardiomyocytes
Ligand
activation
DSL-family
ligand
To recycling
To lysosomal
degradation
To lysosomal
degradation
c
To clathrin
binding and
endocytosis
Control of
ligand
concentration
To
recycling
(70% of EGFR)
PI(3)K
GRKs
PA
Ras
PLD
GPCR
ligand
PP2A
GRB2
SOS1
complex
C
-
A
r
r
e
s
t
i
n
Clathrin-
coated pit
GPCR
Nucleus
G
s
G
s
G
i
G
i
G
i
(30% of EGFR)
A plea for systems biology
From the molecular data above, it is intuitive that endocytosis confers
spatial and temporal dimensions to signalling. But the real magnitude
and impact of spatiotemporal dynamics can be understood only by
carrying out mathematical modelling (see ref. 18 for a review).
The most relevant aspect of this issue is that molecules cannot travel
far by free diffusion. Signals originating from the plasma membrane
must travel considerable distances, for example to reach the nucleus, and
signal deactivation (such as by dephosphorylation) during cytoplasmic
diffusion can cause precipitous signalling gradients and negligible signal
magnitudes near the target. The initial prediction that this would occur
was shown to be accurate by theoretical modelling
19
and subsequently
by experimental studies of cells (see ref. 20 for an example). Conversely,
endosomes, which are propelled by microtubular motors, provide fast
communication routes for signalling molecules, with diffusion having a
role only in the final travel paths. Although this should be enough for an
average-sized cell to overcome the rate of signal deactivation, larger cells,
such as Xenopus laevis oocytes, might need more help. From theoretical
modelling, it is predicted that, in such cells, regulatory mechanisms are
required for dampening deactivation, a prediction that is being confirmed
by experimental findings
21
. In neurons, for which the signalling distance
can be as much as several centimetres, neurotrophins produced by post-
synaptic cells activate anti-apoptotic signals by engaging pre synaptic TRK
(neurotrophic tyrosine kinase receptor)-family molecules in axonal ter-
mini. The signal must then travel to the neuronal cell body to promote
the transcription of anti-apoptotic genes. A wealth of evidence supports
the prevailing model that predicts the need for retro grade endosomal
transport of activated receptors or of signalling molecules to achieve
this biological effect (see ref. 3 for a review). Yet the average velocity of
molecular motors is not fast enough to account for the experimentally
measured signal propagation time. One mathematical model suggests that
travelling waves of protein activation can carry out the task
22
, although
experimental proof for this is lacking.
Molecular dissection of many of the endocytosis-based signalling
circuits described above has yielded many conflicting results (see ref. 3
for a recent review). Such results are usually attributed to cell-specific
differences but probably betray our lack of understanding at the sys-
tems level. Mathematical simulation has successfully been applied to
explain the complex schemes by which endocytosis contributes to the
control of cell polarity, spatial signal propagation, signal magnitude,
Figure 2 | Endocytosis controls signals at the plasma membrane.There
are several ways in which endocytosis regulates the assembly of signalling
platforms at the plasma membrane. a, The removal of receptors from
the plasma membrane, for example by clathrin-mediated endocytosis,
extinguishes signals that depend on plasma-membrane-specific molecules
3
.
These include signals driven by the following: stimulatory heterotrimeric
G (G
s
) proteins; phosphatidylinositol-4,5-bisphosphate (not shown), which
drives the activity of phospholipase C (PLC) and phosphatidylinositol-3-OH
kinase (PI(3)K); and a circuit that involves an RTK, PLD and phosphatidic acid
(PA), which contributes to the recruitment of the GRB2SOS1 complex and to
the activation of Ras
75
. b, The endocytosis of DSLs (ligands in the Delta, Serrate
and LAG-2 family) is required for activation of, and therefore signalling
through, receptors in the NOTCH family. Endocytosis activates the ligands (by
an ill-defined mechanism) and maintains ligand concentrations at the plasma
membrane
6
. c, The route of receptor entry to the cell affects signalling. For
the epidermal growth factor receptor (EGFR) and the transforming growth
factor- receptor (TGF-R), clathrin-mediated endocytosis preferentially
destines receptors to recycling, and non-clathrin-mediated endocytosis
destines receptors for degradation
7,8
. In other settings (for example in WNT-
activated pathways, not shown), the opposite occurs
9
. dg, Recycling pathways
control signalling. d, EGFRs have different fates when engaged by EGF or
TGF-. The EGFEGFR complex is either recycled to the plasma membrane
or routed to the lysosome (as detailed in c). By contrast, the TGF-EGFR
complex dissociates at the acidic pH of the endosome, and the EGFR is recycled
to the surface. e, G-protein-coupled receptors (GPCRs) are phosphorylated
by GPCR kinases (GRKs). This allows -arrestins to bind to the GPCR,
preventing G
s
protein recruitment and thereby terminating signalling. In
addition, -arrestins bind to clathrin, targeting the receptors for clathrin-
mediated endocytosis. f, In the early endosome, GPCRs are dephosphorylated
by the phosphatase PP2A and are then recycled, returning resensitized GPCRs
to the cell surface
3
. g, In cardiomyocytes, the
2
-adrenergic receptor, which is
a GPCR, can elicit different biochemical and biological responses by coupling
with G
s
proteins or inhibitory G (G
i
) proteins. The switch between these types
of G protein requires endocytic and exocytic cycles (EECs), which probably
work by redirecting the receptors to G
i
-protein-enriched plasma-membrane
microdomains
3
.
466
NATURE|Vol 463|28 January 2010 INSIGHTREVIEW
464-473 Insight Scita NS.indd 466 21/1/10 11:01:09
20 Macmillan Publishers Limited. All rights reserved 10
A
P
P
L
A
P
P
L
R
A
B
5
PP
P
Ub
Deltex
S
M
A
D
2
NOTCH
H-Secretase
To the
nucleus
H
+
H
+
H
+
b
a
c
RAB5
GSK3C
a
b
p
1
8
p14 MP1
PtdIns(3)P
PtdIns(3)P
EEA1
GPCR
a
b
c
A
B
C
S
M
A
D
2
TGFC-R
To the
nucleus
d
MAPK
MAPKK
MAPKKK
C-Arrestin
Sara
SMAD4
To the
nucleus
SARA
RAB5
PtdIns(3)P
EEA1
Ligand
In late
endosomes
MAPK
MAPKK
AKT
signal kinetics, and dynamic synchronization of intracellular signalling
with stimuli received by the cell
18
. So mathematical modelling, coupled
to experimental verification, might be the sole practical solution to the
innumerable conundrums that confound our understanding of the
liaisons between endocytosis and signalling.
The cellular level
A corollary of the many connections between endocytosis and signal-
ling at the circuitry level is that the integration of these two programs is
likely to have a substantial impact on the execution of complex cellular
programs. In this section, we review the cellular and molecular evidence
for this, showing that the functions of endocytosis only partly conform
to those in the canonical view.
EECs and polarized functions
Cells must recognize and process spatial information in order to carry
out polarized functions, such as directed cell migration, cell fate deci-
sions, epithelial cell polarization, growth-cone movement, and tissue
morphogenesis during development (see ref. 23 for a review). All of
these tasks require the spatial restriction of signalling, which is achieved
through asymmetrical distribution (or redistribution) of membranes or
signalling molecules, often mediated by EECs. For instance, a continu-
ous flow of membranes of endocytic origin is essential for the dynamic
changes in cell shape that occur during the directional, chemotactic
migration of the slime mould Dictyostelium discoideum
24
, although
this mechanism may not be used by all motile cells (see ref. 23 for a
review). EECs can also redirect and confine signalling molecules to
specialized areas of the plasma membrane, such as the apical membrane
or basal membrane of polarized epithelial cells
25
, and they can sustain
positive-feedback mechanisms that maintain the asymmetry of crucial
molecules, such as the small GTPase Cdc42, during the formation of
polarized buds in budding yeast
26
.
During the chemotactic migration of cells, whether in two dimensions
or three dimensions, the cells must reorient themselves in the direction
of travel through the polarization of sensors that are present in the plasma
membrane, and they must coordinate the intracellular trafficking of mol-
ecules, the adhesion of the cell to the substrate and the remodelling of the
actin cytoskeleton to generate the propulsive forces (Fig. 4). Again, the role
of EECs is paramount, as has been shown in the border cells of Drosophila
melanogaster. In these cells, when endocytic pathways that depend on
Figure 3 | The signalling endosome.Three situations are depicted, each
showing a different concept in how endosomes generate unique signals
that cannot be generated at the plasma membrane. A, Signals relying
on lipids or proteins that are unique to endosomes. A, a, Locally (and
specifically) produced phosphatidylinositol-3-phosphate (PtdIns(3)P)
allows endosome-specific assembly of signalling complexes
3
, through its
binding to proteins (for example EEA1) that contain protein domains such
as FYVE or PX. A, b, In late endosomes, the endosome-specific lipid-raft
adaptor protein p18 binds to a scaffold that activates mitogen-activated
protein kinases (MAPKs), which consists of a MAPK kinase (MAPKK),
MP1 and p14 (ref. 76), thereby sustaining MAPK signalling. A, c, Some
GPCRs remain associated with -arrestins after being internalized. The
-arrestins stably anchor MAPK, which might bias signalling towards
cytosolic MAPK substrates rather than nuclear MAPK substrates
77
.
A, d, Internalized TGF-R interacts with the PtdIns(3)P-binding protein
SARA (SMAD anchor for receptor activation)
12,13
and phosphorylates the
SARA-associated molecule SMAD2, promoting dissociation of SMAD2
and its interaction with SMAD4. The SMAD2SMAD4 complex then
translocates to the nucleus, where it elicits a transcriptional response.
Another PtdIns(3)P-binding protein, endofin, can interact with TGF-R
and SMAD4 and facilitates the formation of SMAD2SMAD4 complexes
14

(not shown). B, The acidic pH of endosomes affects signalling in numerous
ways (see Fig. 2d for another example). B, a, To be activated, NOTCH needs
to be cleaved by the enzyme -secretase, which is present at the plasma
membrane and on endosomes. The peak activity of -secretase is at low
pH
78
, suggesting that endosomal transit is necessary for NOTCH activation.
B, b, This is also supported by findings that the interaction between
NOTCH and its ligands from the DSL family is favoured at low pH
79
and
that endocytosis is necessary for the cleavage of NOTCH
80
. B, c, NOTCH
can also be activated in a ligand-independent manner, by the ubiquitin
protein ligase deltex, which promotes the internalization of NOTCH
and prevents it from being engulfed into the intraluminal vesicles of the
multivesicular body, thus favouring cleavage by -secretase
81
. C, Endosome
plasticity regulates signalling. C, a, In endosomes that contain internalized
EGFR, two related RAB5 effectors are recruited. These effectors, APPL1
(adaptor protein containing phosphotyrosine-interaction domain, pH
domain and leucine-zipper motif 1) and APPL2 (ref. 15), are recruited
through binding to RAB5 and to the EGFR, either directly (as shown)
or indirectly through the protein GIPC (not shown). APPLs lead to the
activation of AKT and to substrate selection by AKT, through activating the
AKTGSK3 axis, which is involved in cell survival
16
. C, b, APPL-enriched
endosomes are early stages of the early endosome, the maturation of which
is controlled by the localized production of different phosphoinositides
17
.
As APPL-enriched endosomes mature and PtdIns(3)P is generated, APPLs
are shed and replaced by PtdIns(3)P-binding proteins, such as EEA1
(ref. 17). MAPKKK, MAPKK kinase.
467
NATURE|Vol 463|28 January 2010 REVIEWINSIGHT
464-473 Insight Scita NS.indd 467 21/1/10 11:01:23
20 Macmillan Publishers Limited. All rights reserved 10
RAB25
RAB5 RAB5
Rac
RAB25
Rac
GTP
Rac
GTP
Rac
GTP
Rac
GTP
RTK (Met)
P
P
HGF
Extracellular
matrix
Focal
adhesion
Lamellipodium
Circular
dorsal
rufes
CAV1
Actin
laments
1
1
1
2
Integrins
2
3
3
TIAM1
Plasma
membrane
Caveola
ARF6
Endosome
Endosome
Endosome
CAV1
RAB5 (which belongs to the RAB family of small GTPases) are disrupted,
cells migrate aberrantly in response to stimulation
27
. There are also similar
circuitries in mammalian cells: the endocytic (RAB5-dependent) traffick-
ing of Rac proteins (small GTPases that relay signals from cell-surface
receptors to the actin cytoskeleton), and their recycling to specific regions
of the plasma membrane, is required for the transduction and spatial reso-
lution of information emanating from motion-inducing stimuli
28
. EECs
also control the trafficking of integrins, which are the best-characterized
cell-surface adhesion receptors and have a crucial role in cell migration.
It has been suggested that the continual internalization and recycling of
integrins between the plasma membrane and the endosomal compart-
ment is essential for controlling cell locomotion (see ref. 23 for a review).
Consistent with this view, it has been found that inhibiting RAB25, which
associates with
5

1
-integrins in endosomes, blocks the recycling of these
integrins to the surface and impairs the formation of cellular protrusions,
thus preventing the cell from migrating in three dimensions
29
(Fig. 4).
Thus, EECs are required across species to resolve signals and to redirect
them in space, preventing the signals from becoming uniformly distributed
and therefore uninformative. Not surprisingly, cancer cells exploit these
mechanisms to gain a selective advantage. For example, metastatic cancer
cells switch between two modes of migration (amoeboid migration and
mesenchymal migration) according to the environmental conditions, and
the switch between these two migratory programs is controlled by the
RAB5Rac circuitry and the RAB25integrin circuitry
28,29
.
Finally, convergent observations support a model (Fig. 4) that explains
how some signalling molecules are recycled to specific regions of the
plasma membrane. In this model, a proportion of the Rac proteins in the
cell activated through a process that depends on clathrin-mediated
endocytosis
28
are recycled to regions of the plasma membrane known
as lipid rafts
30,31
. Integrins act locally to prevent the lipid rafts, which func-
tion as anchor points for Rac, from being internalized by non-clathrin-
mediated endocytosis. This process, in turn, maintains active Rac near
sites of integrin-mediated signalling
30,31
. The crucial recycling route seems
to depend on the GTPase ADP-ribosylation factor 6 (ARF6) and controls
the redelivery to the plasma membrane not only of Rac
28
and integrins
30,31

but also of lipid rafts, ultimately coordinating Rac-mediated signalling
and directional migration of the cell with adhesion-dependent growth
of the cell
32
.
Figure 4 | Cell migration harnesses EECs.Cells extend polarized protrusions,
such as lamellipodia and circular dorsal ruffles, under the control of small
GTPases, such as Rac. Three pathways by which EECs control the migration
of cells are shown. In the first pathway (1), in response to motogenic
stimuli (for example hepatocyte growth factor (HGF)), clathrin-mediated
endocytosis promotes the formation of endosomes containing Rac and its
GEF (guanine-nucleotide exchange factor) TIAM1. This leads Rac being
activated (present in its GTP-bound form) and recycled to specific regions
of the plasma membrane, where circular dorsal ruffles are then formed
28
.
In another pathway (2), lamellipodia (also called pseudopods) depend
on EECs of the
5

1
-integrin, which can enter cells through clathrin-
mediated endocytosis or non-clathrin-mediated endocytosis. This integrin
associates with RAB25 in endosomes and is then recycled to the distal
tips of lamellipodia
29
; it traffics bidirectionally between endosomes and
the plasma membrane within the lamellipodial tips. This promotes the
compartmentalization of a spatially restricted subpopulation of cycling

1
-integrin within the tip of extending lamellipodia
29
. This trafficking
event is necessary for the extension of the lamellipodium. In the last pathway
shown (3), EECs of membrane regions containing lipid rafts connect Rac-
dependent events and integrin-dependent events. Lipid rafts are internalized
and recycled through caveolae, which contain the protein caveolin 1 (CAV1).
Lipid rafts are binding sites for Rac, thus Rac that has been activated by
clathrin-mediated endocytosis might be recycled specifically to lipid
rafts (dashed arrow). Integrin signalling prevents lipid rafts from being
internalized, by retaining phosphorylated CAV1 in focal adhesions, which
are macromolecular protein assemblies that connect the cytoskeleton to
the extracellular matrix. Thus, when integrins are activated by binding to
the extracellular matrix, binding sites for Rac are available at the plasma
membrane
30,31
. Detachment of integrins from the extracellular matrix causes
signal extinction and allows relocalization of phosphorylated CAV1 to
caveolae. This event is necessary for the internalization of caveolae, an event
that clears lipid rafts (and Rac-binding sites) from the cell surface. Cells that
are deficient in CAV1 consistently show a lack of spatial confinement of Rac
82
.
The GTPase ARF6 has been implicated in the trafficking of Rac, integrins
and lipid rafts
28,3032
. This suggests that ARF6 might be the crucial factor that
regulates the spatial confinement of all of these.
468
NATURE|Vol 463|28 January 2010 INSIGHTREVIEW
464-473 Insight Scita NS.indd 468 21/1/10 11:01:27
20 Macmillan Publishers Limited. All rights reserved 10
N
U
M
B
p53
p
5
3
MDM2
N
U
M
B
p53
MDM2
N
U
M
B
MDM2
Multivesicular
body
Exosome
release
Exosome
uptake
Ceneticreprogrumming
Cuncerousmodicution
oenvironment
a
b
Degruded
p53
1ostemcellute
1oprogenitorute
c
d
Transcriptional
efect
e
RNA
f
Endocvticproteins
orexumpleEPSEPSPepsins
CALMHPCurrestinsundCHC
g
Nuclear
envelope
nnerleuet
onucleur
membrune
proAPEC
proH8ECl
j
Efect on
transcription
Efect on chromatin
remodelling
h
i
CHC
depletion
RAB6A
depletion
Wild type Wild type Dynamin
depletion
ARH
depletion
At mitosis
Chromutid
Kinetochore
Centrosome
Spindle
microtubule
Spindle
microtubule
Ub
Ub
p53
MDM2
a
udeed d
3
te
Ub
UUUUUUUb UUUUUUUUUUUU
Ub
k l m n o p
E
S
C
P
1

Endocytic control of cell division


There is increasing evidence that proteins involved in endocytosis such
as clathrin, dynamin, the endocytic adaptor protein ARH (autosomal
recessive hypercholesterolaemia protein) and RAB6A also have a role
in mitosis. These proteins are found associated with the centrosomes or
the mitotic spindle, cellular structures that are involved in mitosis
3336

(Fig. 5). But do such endocytic proteins have distinct functions in mem-
brane trafficking during interphase and in mitosis? Certainly, endocytic
prote ins present in mitotically relevant structures do not seem to be
connected to membranes, and they interact with binding partners that
are distinct from those involved in trafficking
3336
. In addition, several
studies have concluded that endocytosis ceases during early mitosis (see
Figure 5 | The endocytic machinery in cell-cycle progression, transcription
and mitosis.How the endocytic machinery affects these three interrelated
processes is depicted. First, the tumour-suppressor protein p53 (a master
regulator of cell-cycle checkpoints, apoptosis and the DNA damage response)
is controlled by the endocytic protein NUMB and controls intracellular
traffic. NUMB inhibits the ubiquitylation of p53 by MDM2 (a), thereby
preventing its degradation, leading to increased p53 levels and therefore
p53 activity
56
. In dividing mammary stem cells (b), NUMB partitions to
the daughter cell that adopts the stem-cell fate
57
. One intriguing possibility
is that this might drive high levels of p53 in that daughter stem cell and its
withdrawal into quiescence. Because the MDM2NUMB complex shuttles
in and out of the nucleus (c), it is not clear whether the regulation of p53
by NUMB occurs in the cytosol or in the nucleus. At the transcriptional
level, p53 controls the release of exosomes
58
(d), which are tools for genetic
reprogramming or for controlling the microenvironment after release by
cancer cells
53,55
(e). The endocytic machinery (specifically ESCRT-II) might
further contribute to this process by selecting the RNAs to be included
in exosomes
54
(f). Second, the endocytic machinery affects transcription.
Endocytic proteins shuttle in and out of the nucleus
60
(g), where they
affect gene expression, by binding to the transcription machinery
60
(h) or
to chromatin-remodelling complexes
15,6163
(i). Endocytosis also delivers
cargo to the inner nuclear membrane, by way of a retrograde transport
mechanism
65,66
(j). In this case, two membrane-anchored growth factors,
pro-AREG (precursor of amphiregulin) and pro-HB-EGF (precursor of the
heparin-binding EGF-like factor), are delivered in a signalling-dependent
and endocytosis-dependent manner to the inner nuclear membrane, where
they function as chromatin-remodelling agents (in the case of pro-AREG)
or sequester transcriptional repressors (pro-HB-EGF). Third, the endocytic
machinery affects mitosis. During mitosis, endocytic proteins bind to
components of the chromosome segregation machinery: the kinetochore,
which forms part of the mitotic spindle (k), and the centrosome (n).
The heavy chain of clathrin (CHC) localizes to the kinetochore fibres of
the spindle, and its depletion results in misaligned chromosomes
33
(l).
RAB6A is also recruited to the kinetochores, and it cooperates with the
MAD2-dependent spindle-checkpoint pathway, ensuring that the spindle
microtubules attach to the kinetochores at metaphase (m)
36
. Depletion
of the endocytic protein dynamin causes separation of the centrosome,
indicating that it contributes to centrosome cohesion
34
(o). ARH localizes at
the centrosome during interphase and at the kinetochores and spindle poles
during mitosis. ARH-null fibroblasts have smaller centrosomes than their
wild-type counterparts (p). Because ARH binds to the motor protein dynein,
it could cooperate in the delivery of components to the centrosome
35
.
469
NATURE|Vol 463|28 January 2010 REVIEWINSIGHT
464-473 Insight Scita NS.indd 469 21/1/10 11:01:31
20 Macmillan Publishers Limited. All rights reserved 10
ref. 5 for a review). The situation, however, may be more complex than
this, because clathrin-mediated endocytosis is now known to be active
throughout mitosis (from prophase to anaphase), although recycling
slows down
37
. The net outcome is a decrease in the surface area of the
cell, with ensuing detachment of the cell from the substrate and rounding
up. The recycling pathway recovers at the last stage of mitosis, telophase,
allowing newly formed daughter cells to spread out. Changes in cell shape
and size are important features of mitosis
38
, indicating that endocytosis
and recycling must have a role in the proper execution of this function.
Although the details remain hazy, the endocytic machinery might there-
fore have a role in coordinating mitotic events.
We have a better understanding of how endocytosis regulates
cytokinesis, the division of the cytoplasm. The polarized recycling of
internalized membranes towards the bridge that connects the two cells
being formed is important for abscission, the terminal step of cell divi-
sion
39,40
. Various components of different recycling endosomes (including
RAB11, RAB35 and ARF6) and components of the secretory machinery
(such as soluble N-ethylmaleimide-sensitive-fusion-protein attachment
protein receptors (SNAREs)) cooperate with microtubule-based, plus-
end-directed motor proteins to orient the recycling of endosomes towards
the midbody of the cell. At this location, homotypic vesicle-fusion events
generate distinct membrane subcompartments with defined lipid com-
positions that function as signalling platforms and result in abscission in
animal cells. In plant cells, polarized membrane trafficking and spatially
restricted endosome fusion have long been recognized as central mecha-
nisms in the completion of cell division
40
. Thus, vesicular traffic is used
across species to facilitate rapid closure of the intracellular space during
cytokinesis.
Endocytosis and asymmetrical cell division
Unlike symmetrical cell division, which produces identical daughter cells
(discussed in the previous subsection), asymmetrical cell division gives
rise to two daughter cells with different fates. This process is central to
the maintenance of adult stem-cell compartments, and its subversion is
a fundamental mechanism in the development of cancer (see ref. 41 for
a review). Endocytosis has an important role in several aspects of this
type of cell division.
In D. melanogaster, several molecular determinants of asymmetrical
cell division have been identified
41
. The asymmetrical partitioning of one
such determinant, NUMB, on division of the sensory-organ-precursor
cell confers different destinies on the two daughter cells, known as pIIa
and pIIb. NUMB counteracts the action of the signalling receptor Notch
41
.
Because NUMB is an endocytic protein that can bind to Notch and to the
main endocytic adaptor protein AP-2, models were proposed in which
NUMB promotes preferential internalization and degradation of Notch
(or of other membrane proteins necessary for Notch activity, such as
Sanpodo) in the pIIb cell, in which NUMB and AP-2 are asymmetrically
partitioned
4244
. Moreover, differing recycling and/or degradation fates
of the Notch ligand Delta regulate the abundance of Delta on the plasma
membrane of the pIIb cell
45
(see ref. 41 for a review). This could cre-
ate the asymmetry necessary for unidirectional signalling from the pIIb
(Delta-expressing) cell to the pIIa (Notch-expressing) cell. More recent
evidence, however, suggests a different model. In Caenorhabditis elegans,
the homologue of NUMB inhibits recycling of the Notch homologue
46

and this is also the case in mammalian cells
47
. Thus, asymmetrical parti-
tioning of NUMB could skew the fate of Notch towards degradation, by
suppressing its recycling. Regardless of the exact molecular mechanism,
endocytosis seems to be pivotal for cells to acquire different fates.
There is evidence that endocytosis has a role not only after asymmet-
rical cell division but also during this type of cell division. The asym-
metrical distribution of endosomes during asymmetrical cell division
might contribute to an unequal distribution of signalling molecules,
thereby imparting different fates. Asymmetrical partitioning of endo-
somes has been observed at the first cleavage of the C. elegans embryo
48

and during asymmetrical cell division of mammalian haematopoietic
stem cells (HSCs)
49
. SARA endosomes might be crucial in this process. In
D. melanogaster sensory-organ-precursor cells, both Notch and Delta are
trafficked to SARA endosomes, which in turn are directionally trans-
ported to the nascent pIIa cell
50
. This is functionally important, because
mistargeting of SARA endosomes to the pIIb cell causes ectopic activation
of Notch in that cell
50
. How this can be integrated into current models of
NOTCH and Delta recycling and degradation remains to be established.
Endocytosis and SARA endosomes may also be pivotal in the mainten-
ance of the mammalian HSC niche, on interaction between HSCs and
osteoblasts. In this case, a specialized membrane domain of the HSC is
trans-endocytosed by the osteoblast and trafficked to SARA endosomes,
where it remains (without being degraded) and triggers signalling path-
ways that lead to attenuation of the SMAD2 and SMAD3 pathway and to
production of chemokines that promote HSC homing
51
.
In summary, although a coherent picture is far from evident, there is
increasing evidence that endocytosis and recycling are central to asym-
metrical cell division during and after the mitotic event.
Endocytosis and genetic reprogramming
The example of HSCosteoblast communication introduces a fascinating
aspect of how the control of endocytosis (and of EECs) affects homeo-
stasis the regulation of cellcell communication through exosomes.
Exosomes are 40100-nm vesicles that are present as intraluminal vesicles
in multivesicular bodies and released extracellularly when these bodies
fuse with the plasma membrane. Exosomes can then be captured by the
surrounding cells and endocytosed. Physiologically, exosomes mediate
several protein-linked functions, such as the loss of the transferrin recep-
tor during reticulocyte maturation, the release of decoy receptors, and
antigen presentation and related aspects of the control of the immune
response
52
.
One surprising aspect of exosome-mediated communication is the abil-
ity of the donor cell to genetically reprogram the recipient cell. Indeed,
exosomes can deliver genetic material. Mast cells, for example, release
exosomes that contain more than 1,000 messenger RNA species and more
than 100 types of microRNA, both of which can genetically reprogram a
cell when taken up
53
. The presence of microRNAs can result in extensive
reprogramming, given the broad regulatory capacity of these molecules.
In addition, only a specific subset of the mRNAs being transcribed in a
cell is found in exosomes, favouring the idea that this process is selective
53
.
Notably, subunits of the ESCRT-II complex bind to specific mRNAs
54
.
Thus, it is possible that the endocytic machinery in the donor cell is not
only involved in producing exosomes but also in actively sorting RNAs
into these vesicles (Fig. 5).
These concepts have important implications for how we understand
cancer. Glioblastoma cell lines release exosomes that can deliver RNAs,
angiogenic proteins and even oncoproteins to the surrounding normal
cells, thus promoting tumour growth
55
. Furthermore, cancer cells seem
to release more exosomes than their normal counterparts, suggesting that
the exosome cycle can be hijacked by mutated cancer proteins to induce
genetic reprogramming of adjacent cells, in much the same way as patho-
gens such as HIV-1 and prions ensure their release from cells
52
.
Endocytosis and p53
NUMB is not only a cell-fate determinant that antagonizes Notch. It also
regulates the levels of the tumour-suppressor protein p53, by inhibiting
its ubiquitylation and degradation. Perturbing the cellular concentra-
tions of NUMB alters p53-mediated responses, including the DNA dam-
age response, apoptosis and the activation of cell-cycle checkpoints
56
.
These observations suggest that there is a level of endocytic control over
the functions of p53, and this control affects the mode of cell division,
whether asymmetrical or symmetrical. Adult stem cells divide asym-
metrically to yield a stem cell and a progenitor cell. Kinetically, the stem
cell is quiescent (but capable of self-renewal), whereas the progenitor cell
actively proliferates and, eventually, differentiates. In mammary stem
cells, NUMB partitions asymmetrically into the daughter stem cell. Del-
etion of the gene encoding p53 (Tp53) in mice skews cell division from
an asymmetrical mode to a symmetrical mode, with both daughter cells
acquiring a proliferative fate
57
. It would be intriguing if inducing the pro-
duction of large amounts of p53, as a consequence of NUMB segregation
470
NATURE|Vol 463|28 January 2010 INSIGHTREVIEW
464-473 Insight Scita NS.indd 470 21/1/10 11:01:35
20 Macmillan Publishers Limited. All rights reserved 10
to one of the daughter cells, led the cell to become quiescent and adopt
a stem-cell fate. Conversely, a lack of NUMB and p53 in the progenitor
cell would determine its proliferative fate (Fig. 5).
The emerging role of p53 in endomembrane-related functions exempli-
fies how connections between the intracellular trafficking machinery and
apparently unrelated molecules point to endocytosis and trafficking being
integrated at a deep level of the cellular master plan. Recently, p53 was
shown to be a regulator of the endosomal compartment
58
and of recycling
pathways
59
. In terms of regulation of the endosomal compartment by p53,
two p53-regulated genes, TSAP6 (also known as STEAP3) and CHMP4C
(the product of CHMP4C being a subunit of the ESCRT-III complex), can
increase the rate of exosome production
58
. In addition, p53 controls the
gene encoding caveolin 1, a major component of non-clathrin-mediated
endocytosis. Activation of TP53 results in the simultaneous disappearance
of both EGFR and caveolin 1 from the plasma membrane, suggesting that
p53 transcriptionally controls some forms of endocytosis
58
. Considering
the role of the endocytic machinery (through NUMB) in controlling p53
abundance, a feedback loop from endocytosis to p53 and back to endo-
cytosis is conceivable (Fig. 5). In terms of regulation of recycling path-
ways by p53, mutant forms of p53 were recently shown not only to lose
tumour-suppressor activity but also to act as gain-of-function mutants,
by increasing the recycling of integrins and EGFR in a way that depends
on the recycling protein RCP (RAB-coupling protein)
59
. This results in
alterations to AKT signalling that contribute to increased invasion and
metastasis. Although the mechanism is not entirely clear, it presumably
involves inhibiting the transcriptional activity of p63 (a transcription fac-
tor and p53-family member).
The above functions rely on the known role of p53 in regulating tran-
scription. However, mounting evidence suggests that p53 also has a cyto-
plasmic role in association with membranes. Transcriptionally impaired
TP53 mutants, expressed by cancer cells, can still function at the mito-
chondrion (causing apoptosis) or as negative regulators of autophago-
cytosis. In addition, evidence that p53 binds to clathrin has been linked to
the direct regulation of transcription by clathrin (see the next subsection),
but it might also imply that p53 has cytosolic functions. Thus, the con-
nection between endocytic pathways and p53 is likely to be an important
area of future investigation.
Endocytosis and transcription
Superficially, endocytosis (which occurs entirely in the cytoplasm) and
transcription (which occurs entirely in the nucleus) seem to be unrelated.
Yet there is growing support for the idea that transcription is directly con-
trolled by endocytic proteins. Various clathrin adaptors and endosomal
proteins translocate to the nucleus, through several mechanisms, and
regulate transcription (see ref. 60 for a review). The regulation of tran-
scription by endocytic proteins occurs at several levels (Fig. 5): remodel-
ling of chromatin, regulation of transcription initiation and delivery of
transcriptionally relevant cargo.
Chromatin remodelling is regulated by APPL1 and APPL2 (ref. 15)
and by ESCRT-III proteins
61
, which bind to chromatin-remodelling
complexes.
Transcription initiation is regulated by many endocytic proteins that
function as co-regulators of transcription by binding known transcription
factors
60
. This interaction affects transcription-factor activity (for exam-
ple in the case of TSG101) or stability (for example HIP1). One exam-
ple of a more complex mechanism is that of the heavy chain of clathrin
(CHC): CHC can be found in the nucleus, where it specifically enhances
p53-dependent transactivation by binding to the p53-responsive promoter
and stabilizing the interaction between p53 and the histone acetyltrans-
ferase p300 (refs 62 and 63). Finally, -arrestins, which are endocytic adap-
tor proteins, also function as nuclear messengers, regulating transcription
in several ways (see ref. 60 for a review). Interestingly, about one-third of
the -arrestin interactome consists of nucleic-acid-binding proteins
64
.
There is also evidence that two membrane-anchored growth factors,
precursors of amphiregulin (pro-AREG; also known as pro-AR) and
heparin-binding epidermal-growth-factor-like factor (pro-HB-EGF),
are delivered in a signalling-dependent and endocytosis-dependent
manner to the inner leaflet of the nuclear membrane, through a retro-
grade transport pathway. When present in the inner leaflet, pro-AREG
and pro-HB-EGF can function as chromatin-remodelling agents or can
sequester transcriptional repressors, respectively
65,66
.
Again, the outstanding question is whether the regulation of nuclear
events represents a freelance function of endocytic proteins (that is, this
function is carried out by some endocytic proteins in addition to, and
independently of, their role in endocytosis) or an institutional trafficking
duty of endocytic proteins (that is, this nuclear regulation is an integral
part of the endocytic matrix). In the latter case, endocytic regulation of
transcription would be a powerful channel through which to transfer
extracellular information to the nucleus. In some instances, however, it
seems probable that the proteins operate on a freelance basis. For exam-
ple, the endocytic and nuclear functions of HIP1 seem to be mutually
exclusive
67
and the transcriptional activity of CHC does not require its
trimerization domain, which is, by contrast, indispensable for its endo-
cytic coat-protein function
63
. In other cases, endocytosis and transcription
seem to be more deeply intertwined
60
. This is the case for -arrestins and
APPL, which travel through the endocytic routes as bona fide trafficking
molecules and eventually translocate to the nucleus to regulate transcrip-
tion. Last, pro-AREG and pro-HB-EGF are clear-cut examples that the
endocytic process is involved in the delivery of transcriptionally relevant
cargo to the nucleus.
Perspectives
Many of the connections between endocytosis and signalling that have
been uncovered in the past decade were unexpected, but these connec-
tions can be accommodated within the standard view of endocytosis. In
this view, endocytic routes contribute to signalling by internalizing mol-
ecules and consigning them to various fates. Some of the newly uncovered
functions of endocytosis are, however, not immediately reconcilable with
this view, for example the control exerted by endocytic proteins over cell-
cycle progression, mitosis and transcription. We propose three hypotheses
that integrate the canonical and non-canonical functions of endocyto-
sis in a unified framework. These hypotheses might also correspond to
increasing levels of complexity in how these functions are integrated.
The moonlighting hypothesis
In this scheme, some endocytic proteins have dual functions, one in
membrane trafficking and one in signalling (including those aspects of
signalling associated with nuclear activities). Endocytosis as a process
would not be involved in non-canonical functions, but individual endo-
cytic proteins would be. There still might be some level of integration
between canonical and non-canonical functions, if only because different
cellular processes would be competing for the same hardware.
The autogenous hypothesis
One model of the origin of endomembranes (the autogenous model)
is that they evolved from the inward budding of the plasma membrane
and its subsequent topological separation (see ref. 68 for a review). In
particular, according to this model, the nucleus is thought to have evolved
when newly generated endomembranes surrounded chromatin. Indeed,
the nuclear-pore complex and vesicle-coat complexes have been shown
to be evolutionarily related
68
. In this model, a primordial endocytic or
secretory compartment must have pre-dated the origin of the nucleus.
Eukaryotic cells would thus have evolved as a consequence of the acquisi-
tion of a novel cellular property, the capacity to carry out endocytosis
68
,
putting this process at the centre of the eukaryotic cell master plan. As a
consequence, several functions must have co-evolved with endocytosis.
For example, the evolutionary development of endocytosis must have
co-evolved with that of the cytoskeleton, because membrane dynam-
ics requires cytoskeletal scaffolds and molecular motors
68
. This could
explain the many connections between endocytosis and signalling that
lead to polarized cytoskeletal dynamics. Finally, if endocytosis led to the
formation of the nuclear membrane, the existence of shared machinery
between endocytic and nuclear processes would be less surprising than
it at first seems.
471
NATURE|Vol 463|28 January 2010 REVIEWINSIGHT
464-473 Insight Scita NS.indd 471 21/1/10 11:01:35
20 Macmillan Publishers Limited. All rights reserved 10
The Roman-road-system hypothesis
The Romans first built roads to accelerate the movement of their armies
but then discovered that roads supported other activities (such as com-
merce and communication). So they optimized the roads for non-military
uses. By analogy, regardless of how and why the membrane trafficking
system evolved, its end point is a powerful intracellular communication
infrastructure. Molecules that were not originally involved in endo cytosis
might have learned to exploit this infrastructure, for instance to move
rapidly and precisely throughout the cell or to remain physically seg-
regated and inhibited (or regulated) until the time is right to carry out
their function.
There is evidence to support this hypothesis (see ref. 4 for a review). The
endocytic function of clathrin became increasingly important as eukaryo-
tes evolved, suggesting that clathrin was participating to a greater extent
in endocytic events. In addition, it has been suggested that the primordial
function of dynamin was related to regulating mitochondrial inheritance,
and that during evolution some dynamins were then recruited to the
endocytic pathway to carry out vesicle fission. Recently, putative endocytic
functions have been attributed to known tumour-suppressor proteins,
such as NF2 (also known as merlin), VHL and p53, which might further
corroborate the idea that growth regulators are recruited to the endocytic
pathway. Perhaps the best example is that of pathogens that evolved to
hijack the endomembrane system to facilitate their life cycles.
Different passengers can be envisaged on these endocytic routes: com-
muters, hitch-hikers, hijackers and ticket holders. Commuters are the
regular passengers (cargo and associated machinery) for which the sys-
tem was initially designed. Hitch-hikers are molecules that parasitize the
system (that is, they hitch a free ride) for a purpose not associated with
endocytosis, without altering the functioning of the system. Hijackers are
hitch-hikers that sidetrack the system for their own purposes, causing it
to malfunction, for example pathogens and, probably, cancer proteins.
Ticket holders are hitch-hikers that have evolved to pay the fare, by acquir-
ing a new endocytosis-associated role (and therefore contributing to the
functioning of the endocytic system), while retaining their original role.
Their new endocytic function might be unrelated to their original role to
the extent that they seem to be moonlighting, thus bringing us back to the
first proposed hypothesis.
Future challenges
The evidence that we have reviewed here clearly indicates that endocytosis
and signalling are two sides of the same coin and should be conceptualized
as a single cellular process that is central to the eukaryotic cellular master
plan. Unravelling the logic of the endocytic matrix therefore seems to be
indispensable to any attempt to reverse engineer the cellular master plan
in order to understand how a cell is built. Such reverse engineering will
require the convergence of high-resolution mechanistic approaches and
comprehensive high-throughput approaches.
At the mechanistic level, we need to build a reference map of the
endocytic matrix, through uncovering the molecular underpinning of
the various endocytic phases at the level of both the core machinery and
the accessory proteins. One way to tackle this issue would be to recon-
stitute individual steps of the endocytic process in vitro, from the initial
deformation of the plasma membrane and budding of vesicles to the
maturation of distinct endomembranes with their unique protein and
lipid repertoires (see ref. 69 for an example). At the same time, single-
molecule imaging could be used to add spatial and temporal aspects of
the endocytic process to the map. Bottom-up mathematical modelling at
each step would provide information about the necessary kinetic aspects
of the process, through efforts to incorporate membrane constraints and
dynamics into models of signal transduction. In addition, probabilistic
modelling would define (and predict) the impact of single-cell hetero-
geneity on various endocytic steps
70
.
Essential as this knowledge might be, complete understanding will be
obtained only by integrating an additional level of complexity: information
from omics approaches and top-down modelling. The impact of endo-
cytosis and trafficking on many cellular and organismal systems seems
too vast to be decoded solely through classical high-resolution studies
and will probably require systematic strategies. The field is starting
to move in this direction. At present, functional genomics efforts are
mainly directed at understanding how the perturbation of genes affects
endocytosis and intracellular traffic (see ref. 71 for an example). Research-
ers now need to start devising systematic strategies to study the impact
of the endocytic machinery on non-endocytic processes (the functional
map) and to unravel the endocytic interactome (the interactome map).
By overlaying the functional map, the interactome map and the reference
map, an initial picture of the inner workings of the endocytic matrix will
be obtained.
Finally, scientists have traditionally devoted considerably more energy
to understanding how things are than to how things came to be the way
they are. Re-evolving an endomembrane system in vivo, starting from
prokaryotes, is a formidable task, but if it is successful, it will enormously
improve understanding of the master plan of eukaryotic cells. In this
context, a closer collaboration between cell biologists and developmental
biologists might prove decisive.
The deconvolution of the endocytic matrix, a term we have coined
here, holds promise not only for increasing basic knowledge but also for
improving human health. Cancer and immune, genetic and neurological
disorders are all influenced in some way by the subversion of endocytosis.
For instance, unexpected links have emerged between endocytosis, asym-
metrical cell division, stem cells and cancer. The modulation of endocyto-
sis also has important implications for drug delivery and for determining
drug efficacy. Thus, increasing the understanding of endocytosis at the
molecular, cellular and organismal levels will be important not only for
cell physiology but also for the ability to fight diseases.
1. Doherty, G. J. & McMahon, H. T. Mechanisms of endocytosis. Annu. Rev. Biochem. 78,
857902 (2009).
2. Mayor, S. & Pagano, R. E. Pathways of clathrin-independent endocytosis. Nature Rev. Mol.
Cell Biol. 8, 603612 (2007).
3. Sorkin, A. & von Zastrow, M. Endocytosis and signalling: intertwining molecular networks.
Nature Rev. Mol. Cell Biol. 10, 609622 (2009).
4. Lanzetti, L. & Di Fiore, P. P. Endocytosis and cancer: an insider network with dangerous
liaisons. Traffic 9, 20112021 (2008).
5. Mills, I. G. The interplay between clathrin-coated vesicles and cell signalling. Semin. Cell Dev.
Biol. 18, 459470 (2007).
6. Fortini, M. E. Notch signaling: the core pathway and its posttranslational regulation. Dev. Cell
16, 633647 (2009).
7. Di Guglielmo, G. M., Le Roy, C., Goodfellow, A. F. & Wrana, J. L. Distinct endocytic pathways
regulate TGF- receptor signalling and turnover. Nature Cell Biol. 5, 410421 (2003).
8. Sigismund, S. et al. Clathrin-mediated internalization is essential for sustained EGFR
signaling but dispensable for degradation. Dev. Cell 15, 209219 (2008).
9. Yamamoto, H., Sakane, H., Michiue, T. & Kikuchi, A. Wnt3a and Dkk1 regulate distinct
internalization pathways of LRP6 to tune the activation of -catenin signaling. Dev. Cell 15,
3748 (2008).
10. Fehrenbacher, N., Bar-Sagi, D. & Philips, M. Ras/MAPK signaling from endomembranes.
Mol. Oncol. 3, 297307 (2009).
11. Pawson, T. Dynamic control of signaling by modular adaptor proteins. Curr. Opin. Cell Biol. 19,
112116 (2007).
12. Tsukazaki, T., Chiang, T. A., Davison, A. F., Attisano, L. & Wrana, J. L. SARA, a FYVE domain
protein that recruits Smad2 to the TGF receptor. Cell 95, 779791 (1998).
13. Hayes, S., Chawla, A. & Corvera, S. TGF receptor internalization into EEA1-enriched early
endosomes: role in signaling to Smad2. J. Cell Biol. 158, 12391249 (2002).
14. Chen, Y. G., Wang, Z., Ma, J., Zhang, L. & Lu, Z. Endofin, a FYVE domain protein, interacts
with Smad4 and facilitates transforming growth factor- signaling. J. Biol. Chem. 282,
96889695 (2007).
15. Miaczynska, M. et al. APPL proteins link Rab5 to nuclear signal transduction via an
endosomal compartment. Cell 116, 445456 (2004).
16. Schenck, A. et al. The endosomal protein Appl1 mediates Akt substrate specificity and cell
survival in vertebrate development. Cell 133, 486497 (2008).
In zebrafish, Appl1 was shown to spatially restrict the activity of the signalling receptor Akt
in endosomes, resulting in Akt substrates being biased towards Gsk3 and increased cell
survival during animal development.
17. Zoncu, R. et al. A phosphoinositide switch controls the maturation and signaling properties
of APPL endosomes. Cell 136, 11101121 (2009).
Signalling through EGFR is controlled by the diversity and plasticity of the early endosome,
which undergoes discrete maturation steps through a phosphoinositide switch from
an early APPL endosome (where the induction of mitosis by EGFR is enhanced) to an
endosome rich in PtdIns(3)P and EEA1.
18. Birtwistle, M. R. & Kholodenko, B. N. Endocytosis and signalling: a meeting with
mathematics. Mol. Oncol. 3, 308320 (2009).
19. Kholodenko, B. N. Four-dimensional organization of protein kinase signaling cascades: the
roles of diffusion, endocytosis and molecular motors. J. Exp. Biol. 206, 20732082
(2003).
20. Maeder, C. I. et al. Spatial regulation of Fus3 MAP kinase activity through a reaction-
diffusion mechanism in yeast pheromone signalling. Nature Cell Biol. 9, 13191326 (2007).
21. Perlson, E. et al. Vimentin-dependent spatial translocation of an activated MAP kinase in
injured nerve. Neuron 45, 715726 (2005).
472
NATURE|Vol 463|28 January 2010 INSIGHTREVIEW
464-473 Insight Scita NS.indd 472 21/1/10 11:01:35
20 Macmillan Publishers Limited. All rights reserved 10
22. Markevich, N. I., Tsyganov, M. A., Hoek, J. B. & Kholodenko, B. N. Long-range signaling by
phosphoprotein waves arising from bistability in protein kinase cascades. Mol. Syst. Biol. 2,
61 (2006).
23. Disanza, A., Frittoli, E., Palamidessi, A. & Scita, G. Endocytosis and spatial restriction of cell
signaling. Mol. Oncol. 3, 280296 (2009).
24. Traynor, D. & Kay, R. R. Possible roles of the endocytic cycle in cell motility. J. Cell Sci. 120,
23182327 (2007).
25. Bryant, D. M. & Mostov, K. E. From cells to organs: building polarized tissue. Nature Rev. Mol.
Cell Biol. 9, 887901 (2008).
26. Altschuler, S. J., Angenent, S. B., Wang, Y. & Wu, L. F. On the spontaneous emergence of cell
polarity. Nature 454, 88889 (2008).
27. Jekely, G., Sung, H. H., Luque, C. M. & Rorth, P. Regulators of endocytosis maintain localized
receptor tyrosine kinase signaling in guided migration. Dev. Cell 9, 197207 (2005).
28. Palamidessi, A. et al. Endocytic trafficking of Rac is required for its activation and for the
spatial restriction of signaling in cell migration. Cell 134, 135147 (2008).
29. Caswell, P. T. et al. Rab25 associates with
5

1
integrin to promote invasive migration in 3D
microenvironments. Dev. Cell 13, 496510 (2007).
RAB25 binds to the promigratory and invasive molecule
5

1
-integrin, promoting rapid,
spatially restricted recycling of the integrin, ultimately resulting in increased cellular
migration and invasion in three-dimensional matrices.
30. del Pozo, M. A. et al. Integrins regulate Rac targeting by internalization of membrane
domains. Science 303, 839842 (2004).
31. del Pozo, M. A. et al. Phospho-caveolin-1 mediates integrin-regulated membrane domain
internalization. Nature Cell Biol. 7, 901908 (2005).
32. Balasubramanian, N., Scott, D. W., Castle, J. D., Casanova, J. E. & Schwartz, M. A. Arf6 and
microtubules in adhesion-dependent trafficking of lipid rafts. Nature Cell Biol. 9, 13811391
(2007).
33. Royle, S. J., Bright, N. A. & Lagnado, L. Clathrin is required for the function of the mitotic
spindle. Nature 434, 11521157 (2005).
34. Thompson, H. M., Cao, H., Chen, J., Euteneuer, U. & McNiven, M. A. Dynamin 2 binds
-tubulin and participates in centrosome cohesion. Nature Cell Biol. 6, 335342 (2004).
35. Lehtonen, S. et al. The endocytic adaptor protein ARH associates with motor and
centrosomal proteins and is involved in centrosome assembly and cytokinesis. Mol. Biol. Cell
19, 29492961 (2008).
36. Miserey-Lenkei, S. et al. A role for the Rab6A GTPase in the inactivation of the
Mad2-spindle checkpoint. EMBO J. 25, 278289 (2006).
37. Boucrot, E. & Kirchhausen, T. Endosomal recycling controls plasma membrane area during
mitosis. Proc. Natl Acad. Sci. USA 104, 79397944 (2007).
38. Meyers, J., Craig, J. & Odde, D. J. Potential for control of signaling pathways via cell size and
shape. Curr. Biol. 16, 16851693 (2006).
39. Schweitzer, J. K., Burke, E. E., Goodson, H. V. & DSouza-Schorey, C. Endocytosis resumes
during late mitosis and is required for cytokinesis. J. Biol. Chem. 280, 4162841635 (2005).
40. Baluska, F., Menzel, D. & Barlow, P. W. Cytokinesis in plant and animal cells: endosomes
shut the door. Dev. Biol. 294, 110 (2006).
41. Furthauer, M. & Gonzalez-Gaitan, M. Endocytosis, asymmetric cell division, stem cells and
cancer: unus pro omnibus, omnes pro uno. Mol. Oncol. 3, 339353 (2009).
42. Santolini, E. et al. Numb is an endocytic protein. J. Cell Biol. 151, 13451352 (2000).
43. Berdnik, D., Torok, T., Gonzalez-Gaitan, M. & Knoblich, J. A. The endocytic protein -Adaptin
is required for Numb-mediated asymmetric cell division in Drosophila. Dev. Cell 3, 221231
(2002).
44. Hutterer, A. & Knoblich, J. A. Numb and -Adaptin regulate Sanpodo endocytosis to specify
cell fate in Drosophila external sensory organs. EMBO Rep. 6, 836842 (2005).
45. Emery, G. et al. Asymmetric Rab11 endosomes regulate Delta recycling and specify cell fate
in the Drosophila nervous system. Cell 122, 763773 (2005).
46. Nilsson, L. et al. Caenorhabditis elegans num-1 negatively regulates endocytic recycling.
Genetics 179, 375387 (2008).
47. McGill, M. A., Dho, S. E., Weinmaster, G. & McGlade, C. J. Numb regulates post-endocytic
trafficking and degradation of Notch1. J. Biol. Chem. 284, 2642726438 (2009).
48. Andrews, R. & Ahringer, J. Asymmetry of early endosome distribution in C. elegans
embryos. PLoS ONE 2, e493 (2007).
49. Beckmann, J., Scheitza, S., Wernet, P., Fischer, J. C. & Giebel, B. Asymmetric cell division
within the human hematopoietic stem and progenitor cell compartment: identification of
asymmetrically segregating proteins. Blood 109, 54945501 (2007).
50. Coumailleau, F., Furthauer, M., Knoblich, J. A. & Gonzalez-Gaitan, M. Directional Delta
and Notch trafficking in Sara endosomes during asymmetric cell division. Nature 458,
10511055 (2009).
During the asymmetrical division of fly sensory-organ-precursor cells, the asymmetrical
partitioning of SARA endosomes containing Notch and its ligand Delta to the pIIa
daughter cell limits Notch signalling in the pIIb cell, thereby maintaining the asymmetrical
configuration of the pIIbpIIa pair.
51. Gillette, J. M., Larochelle, A., Dunbar, C. E. & Lippincott-Schwartz, J. Intercellular transfer to
signalling endosomes regulates an ex vivo bone marrow niche. Nature Cell Biol. 11, 303311
(2009).
A specialized membrane domain of haematopoietic progenitor cells is trans-endocytosed
by osteoblasts and trafficked to SARA endosomes, where it triggers signalling that leads
to attenuation of the SMAD2 and SMAD3 pathway and to expression of chemokines that
promote the homing of haematopoietic progenitor cells.
52. Simons, M. & Raposo, G. Exosomes: vesicular carriers for intercellular communication. Curr.
Opin. Cell Biol. 21, 575581 (2009).
53. Valadi, H. et al. Exosome-mediated transfer of mRNAs and microRNAs is a novel
mechanism of genetic exchange between cells. Nature Cell Biol. 9, 654659 (2007).
A new endocytic-based mechanism of intercellular communication was uncovered, in
which mRNA and microRNA are transported by exosomes, which can be transferred
between the mast cells of different species, regulating gene transcription and protein
expression.
54. Irion, U. & St Johnston, D. bicoid RNA localization requires specific binding of an endosomal
sorting complex. Nature 445, 554558 (2007).
Components of ESCRT-II are found to bind to, and control, the localization of bicoid mRNA,
which encodes a homeodomain-containing transcription factor that organizes anterior
development in the fly, to the anterior of the D. melanogaster egg.
55. Skog, J. et al. Glioblastoma microvesicles transport RNA and proteins that promote tumour
growth and provide diagnostic biomarkers. Nature Cell Biol. 10, 14701476 (2008).
56. Colaluca, I. N. et al. NUMB controls p53 tumour suppressor activity. Nature 451, 7680
(2008).
57. Cicalese, A. et al. The tumor suppressor p53 regulates polarity of self-renewing divisions in
mammary stem cells. Cell 138, 10831095 (2009).
An essential role for p53 in the regulation of cancer stem cells was uncovered in this study.
In the Erbb2-transgenic mouse model, decreased levels of p53 lead to breast cancer stem
cells undergoing symmetrical divisions at a greater frequency, and pharmacological
restoration of p53 levels in these animals results in increased asymmetrical division of
breast cancer stem cells and a reduction in tumour growth.
58. Yu, X., Riley, T. & Levine, A. J. The regulation of the endosomal compartment by p53 the
tumor suppressor gene. FEBS J. 276, 22012212 (2009).
59. Muller, P. A. J. et al. Mutant p53 drives invasion by promoting integrin recycling. Cell 139,
13271341 (2009).
Gain-of-function mutants of p53 increase the endocytic recycling of
5

1
-integrin and
EGFR in cooperation with p63, promoting invasion and metastasis.
60. Pyrzynska, B., Pilecka, I. & Miaczynska, M. Endocytic proteins in the regulation of nuclear
signaling, transcription and tumorigenesis. Mol. Oncol. 3, 321338 (2009).
61. Stauffer, D. R., Howard, T. L., Nyun, T. & Hollenberg, S. M. CHMP1 is a novel nuclear matrix
protein affecting chromatin structure and cell-cycle progression. J. Cell Sci. 114, 23832393
(2001).
62. Enari, M., Ohmori, K., Kitabayashi, I. & Taya, Y. Requirement of clathrin heavy chain for
p53-mediated transcription. Genes Dev. 20, 10871099 (2006).
63. Ohmori, K. et al. Monomeric but not trimeric clathrin heavy chain regulates p53-mediated
transcription. Oncogene 27, 22152227 (2008).
64. Xiao, K. et al. Functional specialization of -arrestin interactions revealed by proteomic
analysis. Proc. Natl Acad. Sci. USA 104, 1201112016 (2007).
65. Isokane, M. et al. Plasma-membrane-anchored growth factor pro-amphiregulin binds
A-type lamin and regulates global transcription. J. Cell Sci. 121, 36083618 (2008).
66. Hieda, M. et al. Membrane-anchored growth factor, HB-EGF, on the cell surface targeted to
the inner nuclear membrane. J. Cell Biol. 180, 763769 (2008).
67. Mills, I. G. et al. Huntingtin interacting protein 1 modulates the transcriptional activity of
nuclear hormone receptors. J. Cell Biol. 170, 191200 (2005).
68. Jekely, G. Origin of eukaryotic endomembranes: a critical evaluation of different model
scenarios. Adv. Exp. Med. Biol. 607, 3851 (2007).
69. Ohya, T. et al. Reconstitution of Rab- and SNARE-dependent membrane fusion by synthetic
endosomes. Nature 459, 10911097 (2009).
70. Snijder, B. et al. Population context determines cell-to-cell variability in endocytosis and
virus infection. Nature 461, 520523 (2009).
71. Pelkmans, L. et al. Genome-wide analysis of human kinases in clathrin- and caveolae/raft-
mediated endocytosis. Nature 436, 7886 (2005).
72. Stenmark, H. Rab GTPases as coordinators of vesicle traffic. Nature Rev. Mol. Cell Biol. 10,
513525 (2009).
73. Donaldson, J. G. Arfs, phosphoinositides and membrane traffic. Biochem. Soc. Trans. 33,
12761278 (2005).
74. Raiborg, C. & Stenmark, H. The ESCRT machinery in endosomal sorting of ubiquitylated
membrane proteins. Nature 458, 445452 (2009).
75. Zhao, C., Du, G., Skowronek, K., Frohman, M. A. & Bar-Sagi, D. Phospholipase D2-generated
phosphatidic acid couples EGFR stimulation to Ras activation by Sos. Nature Cell Biol. 9,
706712 (2007).
76. Nada, S. et al. The novel lipid raft adaptor p18 controls endosome dynamics by anchoring
the MEKERK pathway to late endosomes. EMBO J. 28, 477489 (2009).
77. DeWire, S. M., Ahn, S., Lefkowitz, R. J. & Shenoy, S. K. -Arrestins and cell signaling. Annu.
Rev. Physiol. 69, 483510 (2007).
78. Pasternak, S. H. et al. Presenilin-1, nicastrin, amyloid precursor protein, and -secretase
activity are co-localized in the lysosomal membrane. J. Biol. Chem. 278, 2668726694
(2003).
79. Pei, Z. & Baker, N. E. Competition between Delta and the Abruptex domain of Notch. BMC
Dev. Biol. 8, 4 (2008).
80. Vaccari, T., Lu, H., Kanwar, R., Fortini, M. E. & Bilder, D. Endosomal entry regulates Notch
receptor activation in Drosophila melanogaster. J. Cell Biol. 180, 755762 (2008).
81. Wilkin, M. et al. Drosophila HOPS and AP-3 complex genes are required for a Deltex-
regulated activation of Notch in the endosomal trafficking pathway. Dev. Cell 15, 762772
(2008).
82. Grande-Garcia, A. et al. Caveolin-1 regulates cell polarization and directional migration
through Src kinase and Rho GTPases. J. Cell Biol. 177, 683694 (2007).
Acknowledgements We apologize to those colleagues whose primary research
papers or important discoveries could not be properly acknowledged because of space
constraints. We thank A. Sorkin and M. von Zastrow for sharing, ahead of publication,
their excellent review on endocytosis and signalling. We also thank P. R. Romano for
critically editing the manuscript. Work in the authors laboratories is supported by
grants from the following: the Italian Association for Cancer Research (AIRC) and
the Italian Ministry of Education, University and Scientific Research (MIUR) (to G.S.
and P.P.D.F.); the Association for International Cancer Research (G.S.); and the Italian
Ministry of Health, the European Community, the European Research Council, the
Cariplo Foundation, the Ferrari Foundation and the Monzino Foundation (P.P.D.F.).
Author Information Reprints and permissions information is available at www.nature.
com/reprints. The authors declare no competing financial interests. Correspondence
should be addressed to P.P.D.F. (pierpaolo.difiore@ifom-ieo-campus.it) or G.S.
(giorgio.scita@ifom-ieo-campus.it).
473
NATURE|Vol 463|28 January 2010 REVIEWINSIGHT
464-473 Insight Scita NS.indd 473 21/1/10 11:01:35
20 Macmillan Publishers Limited. All rights reserved 10

Anda mungkin juga menyukai