Leading Edge

Review

Cancer Invasion and the Microenvironment:

Plasticity and Reciprocity
Peter Friedl1,2,3,* and Stephanie Alexander2,3
1Microscopical Imaging of the Cell, Department of Cell Biology, Radboud University Nijmegen Medical Centre,

6500 HC Nijmegen, The Netherlands

2David H. Koch Center for Applied Research of Genitourinary Cancers, Department of Genitourinary Medical Oncology,

The University of Texas MD Anderson Cancer Center, Houston, TX 77030, USA

3Department of Dermatology and Rudolf Virchow Center, DFG Research Center for Experimental Biomedicine,

University of Wuerzburg, D-97080 Wuerzburg, Germany

*Correspondence: p.friedl@ncmls.ru.nl
DOI 10.1016/j.cell.2011.11.016

Cancer invasion is a cell- and tissue-driven process for which the physical, cellular, and molecular
determinants adapt and react throughout the progression of the disease. Cancer invasion is initi-
ated and maintained by signaling pathways that control cytoskeletal dynamics in tumor cells and
the turnover of cell-matrix and cell-cell junctions, followed by cell migration into the adjacent tissue.
Here, we describe the cell-matrix and cell-cell adhesion, protease, and cytokine systems that
underlie tissue invasion by cancer cells. We explain how the reciprocal reprogramming of both
the tumor cells and the surrounding tissue structures not only guides invasion, but also generates
diverse modes of dissemination. The resulting ‘‘plasticity’’ contributes to the generation of diverse
cancer invasion routes and programs, enhanced tumor heterogeneity, and ultimately sustained
metastatic dissemination.

Introduction ular, and even adverse microenvironmental conditions (Friedl

Cancer invasion and metastasis are landmark events that trans- and Wolf, 2010; Sahai, 2007; Sanz-Moreno and Marshall,
form a locally growing tumor into a systemic, metastatic, and 2010).
live-threatening disease. The initial steps of local invasion Plasticity of invasion, together with other hallmarks of
include the activation of signaling pathways that control neoplasia, including cancer cell growth, survival, and genomic
cytoskeletal dynamics in tumor cells and the turnover of cell- instability, lead to morphological, signaling, and genetic differ-
matrix and cell-cell junctions, followed by active tumor cell ences between primary and metastatic lesions within the same
migration into the adjacent tissue (Chambers et al., 2002; Friedl patient (intrapatient heterogeneity), within the same lesion
and Wolf, 2003; Sahai, 2007). Metastasis then occurs when (intratumoral heterogeneity), and across time (Choi et al., 2011;
invading tumor cells engage with blood and lymph vessels, Honeth et al., 2008; Lopes et al., 2009; Shapiro et al., 2011;
penetrate basement membranes and endothelial walls, and Stoecklein et al., 2008; Wang et al., 2009).
disseminate through the vessel lumen to colonize distant organs Such heterogeneous tumor progression is mirrored by an
(Fidler, 2003). Like cells in primary tumors, cells in metastases ‘‘activation’’ response of stromal cells nearby the growing tumor,
also proliferate, invade, and enter blood vessels, leading to including fibroblasts, endothelial cells, and macrophages. Once
secondary metastasis (Kienast et al., 2010; Armstrong et al., ‘‘activated,’’ these cells reorganize the structure and composi-
2011; Hou et al., 2011). tion of the connective tissue by depositing extracellular matrix
In the past few decades, cell and tumor biologists have iden- components (ECM), cytokines, and growth factors (Egeblad
tified the mechanisms of cell migration in normal and malignant et al., 2010; He et al., 2011; Picchio et al., 2008; Shapiro et al.,
cells, including the regulation of cell adhesion and cytoskeletal 2011). By remodeling the tissue structure, releasing growth
dynamics (Lauffenburger and Horwitz, 1996; Ridley et al., factors, and imposing metabolic stress, the reactive tumor
2003; Sanz-Moreno and Marshall, 2010). However, attempts stroma, in turn, influences cancer cell functions, often enhancing
to define the rate-limiting mechanisms that govern invasive tumor growth and invasion and aggravating cancer resistance
and metastatic cancer cell migration, such as a dominant during metabolic challenge and therapy (Alexander and Friedl,
signaling pathway, receptor-ligand interaction, or protease- 2012; Giese et al., 2003; Sansing et al., 2011; Yao et al., 2011).
substrate interaction, have largely failed. Instead, cancer cell Thus, in a reciprocal manner, tumor cells influence the stroma
invasion is now regarded as a heterogeneous and adaptive and vice versa, jointly driving cancer progression (Nelson
process. Indeed, it is this ‘‘plasticity’’ in cell adhesion, cyto- et al., 2008; Xu et al., 2009). Here, we summarize the adhe-
skeletal dynamics, and mechanotransduction that perpetuates sion, protease, and cytokine systems that underlie tissue
migration and dissemination under diverse structural, molec- invasion by cancer cells. We discuss how the reactive tumor

992 Cell 147, November 23, 2011 ª2011 Elsevier Inc.

 Figure 1. Cell Migration: A Multistep Process
In general, cells can migrate individually or collectively as
multicellular groups.
(A) Single-cell migration involves five molecular steps that
change the cell shape, its position, and the tissue structure
through which it migrates.
(B) Collectively migrating cells form two major zones: zone
1, in which a ‘‘leader cell’’ generates a proteolytic micro-
track at the front of the migrating group, and zone 2, in
which the subsequent cells then widen this microtrack to
form a larger macrotrack. (Figure modified from Friedl and
Wolf, 2008.)

Invasive single-cell migration results from five

interdependent molecular steps that change the
cell shape, its position, and the tissue structure
through which it migrates (Friedl and Wolf,
2009; Lauffenburger and Horwitz, 1996; Sheetz
et al., 1999) (Figure 1A). In step 1, the cytoskel-
eton polarizes by actin polymerization and forms
a leading protrusion at the opposite end of
a ‘‘pre-uropod’’ region, which marks the consti-
tutive rear end of the cell (Estecha et al., 2009;
Poincloux et al., 2011). In step 2, the leading
edge protrusion engages with extracellular
substrates, followed by recruitment and adhe-
sion of cell surface receptors that form focalized
clusters and couple extracelluar adhesion to
intracellular mechanosignaling and force gener-
ation (Friedl et al., 1997). In step 3, several
micrometer rearward of the leading edge, cell
surface proteases become engaged with extra-
cellular scaffold proteins and execute locally
controlled proteolysis (Friedl and Wolf, 2009).
This proteolysis modifies the molecular and
mechanical tissue properties and allows space
microenvironment steers plasticity and perpetuates invasion and for the advancing cell body (Friedl et al., 1997). In step 4, the small
metastatic dissemination. GTPase Rho activates myosin II, and contraction mediated by
actomyosin generates tension inside the cell. In step 5, this
Mechanisms of Cell Migration contraction is followed by the gradual turnover of adhesion bonds
Cancer invasion is a cyclic process in which the cell changes at the trailing edge, which slides forward while the leading edge
shape, produces morphological asymmetry, and then translo- protrudes further.
cates the cell body. Depending on the cell type and tissue envi- In most cells, the leading edge protrusion is controlled by the
ronment, cells can migrate in two major ways: individually, when small GTPase Rac or Cdc42, which generate pseudopodia or
cell-cell junctions are absent, or collectively as multicellular filopodia that engage with ECM substrate (Sanz-Moreno and
groups, when cell-cell adhesions are retained (Friedl and Wolf, Marshall, 2010). In some cell types with low Rac activity or in
2010) (Figure 1). The underlying process in both types of migra- poorly adhesive environments, a variation of steps 1 and 2
tion is the dynamics of the cytoskeleton coupling with cell occurs (i.e., the pseudopod protrusion and adhesion steps) in
surface receptors that engage with surrounding tissue struc- which the leading edge generates leading bleb-like or even
tures; thus, the cytoskeleton serves as the cell’s engine, and bleb- and pseudopod-free smooth membrane propulsions.
the cell surface receptors act as its transmission (Ridley et al., These propulsions are stabilized by cortical F-actin and interca-
2003). Cancer cells recapitulate the types and mechanisms of late between extracellular tissue structures (Lorentzen et al.,
migration used by normal, nontumor cells. They activate the 2011; Poincloux et al., 2011). Here, the force is generated near
same machineries for changing shape, generating force, and the rear pole in an actomyosin- and integrin-dependent manner
remodeling ECM (Friedl, 2004) as normal cells, but neoplastic (Poincloux et al., 2011).
cells lack physiological ‘‘stop signals’’ immobilizing and The physicochemical steps in single-cell migration are coor-
anchoring the cells (Cox et al., 2001), which arguably perpetu- dinated within the same cell body and executed in a synchronous,
ates neoplastic cell migration. often pulsatile manner, which allows the cell body to protrude and

Cell 147, November 23, 2011 ª2011 Elsevier Inc. 993

 Figure 2. Modes of Cell Movement Implicated in
Cancer Invasion and Metastasis
Single-cell and collective cell migration can be further
partitioned based on the specific cell-cell junctions, the
contractility of cytoskeleton, and the turnover of cell
attachments to extracellular matrix (ECM). These modes
of migration can be further unstable and change upon
alterations of cell-cell interactions, cell-ECM adhesion,
or cytoskeletal contractility, resulting in intermediate
phenotypes.

Rounded/Amoeboid Migration
Cells migrating with low adhesion force or
high actomyosin-mediated contractility adopt
morphologically spherical shapes. This is re-
ferred to here as amoeboid migration because
the Dictyostelium discoideum amoeba mi-
grates by this mechanism (Friedl et al., 2001).
Amoeboid movement, which uses Rac-depen-
dent filopodia, has small or diffusely organized
adhesion sites that generate weak to negligible
adhesion force toward the substrate (Lämmer-
mann and Sixt, 2009). The second form of
amoeboid movement, which uses Rho-domi-
nated blebbing, lacks defined adhesions and
mediates cell translocation by propulsion using
either blebs or smooth membrane protrusion
at the leading edge and lateral intercalation
(Lorentzen et al., 2011; Paluch et al., 2006;
Poincloux et al., 2011). Amoeboid cells tend
to migrate in the absence of proteolytic
ECM breakdown by adapting their shape to
and squeezing through tissue gaps and
trails (Wolf et al., 2003b). The origin of amoe-
boid tumors is often hematopoietic or neuro-
ectodermal, including leukemias, lymphomas,
and small cell lung carcinoma, but amoeboid
movements are also detected as cell subsets
generate traction in an oscillatory manner (Ridley et al., 2003). in most other tumor types (Madsen and Sahai, 2010; Wolf
If multiple cells originate from the same location, such as a et al., 2003b).
tumor, the ‘‘leader cell’’ forms a proteolytic microtrack of locally Mesenchymal Migration
removed ECM barriers (zone 1). The following cells then widen When cytoskeletal protrusions and adhesion capabilities are
or excavate this microtrack by mechanical force and proteolysis strongly developed, invading cells adopt spindle-shaped, elon-
to form a larger macrotrack (zone 2) (Ilina et al., 2011; Wolf gated morphology with focalized cell-matrix adhesions contain-
et al., 2007) (Figure 1B). In collective migration, protrusion and ing multimolecular integrin clusters and proteolytic
retraction are coordinated in a ‘‘supracellular manner,’’ in which activity toward ECM substrates (Wolf et al., 2007). Focalized pro-
cytoskeletal protrusion and contractility are mechanically teases on the cell’s surface generate small microtracks through
mediated through cell-cell junctions (Hidalgo-Carcedo et al., which subsequent cells can follow (Friedl and Wolf, 2009) (Fig-
2011; Tambe et al., 2011), allowing the cell group to behave as ures 1A and 2). Mesenchymally migrating tumor cells originate
‘‘mega-cell’’ (Figure 1B). from tumors of the connective tissue, including soft tissue
sarcomas. They also originate from all other tumor types after
Patterns and Diversity of Cancer Cell Invasion the tumor cells dedifferentiate and lose cell-cell junctions (Bra-
The five-step model of cell migration is active in many types of bletz et al., 2001; Friedl and Wolf, 2009; Sanz-Moreno et al., 2008).
cell movement for both normal and neoplastic single cells. Oper- Multicellular Streaming
ationally, individual cell and multicellular migration follow the When individual cells move one after each other using the same
paradigm of active cell migration, whereas multicellular growth path within the tissue, it is referred to as ‘‘multicellular
leads to passive cell movement by pushing (Figure 2). streaming.’’ This occurs mainly when individual cells become

994 Cell 147, November 23, 2011 ª2011 Elsevier Inc.

chemotactically attracted by a particular source or jointly follow determine the morphology and mechanism by which normal
microtracks that are often present in peripheral connective and neoplastic cells move through tissues.
tissue (Kulesa and Gammill, 2010). In neoplasia, multicellular
streaming is often seen as chain- or swarm-like (i.e., diffuse) Physical and Molecular Determinants of Invasion
tissue infiltration of many tumor cells in hematologic and solid The molecular mechanisms underlying each migration mode
tumors (Kedrin et al., 2008). depend on a set of connected mechanical and signaling path-
Collective Invasion ways, which vary in their coordination and strength depending
Collective invasion requires cell-cell adhesion and multicellular on the particular migration mode. Such variations include the
coordination to occur simultaneously with migration, which organization of the cytoskeleton; the capability to remodel tissue
results in multicellular groups and strands originating at the inter- structures; and the type, strength, and turnover of cell-substrate
face between tumor and stroma (Friedl et al., 1995; Ilina and adhesions, cell-cell adhesions, and intercellular communication.
Friedl, 2009). Collective invasion may adopt different morphol- Stromal signals modulate all of these pathways through chemo-
ogies, which depend on the cell type, the number of jointly kines, cytokines, and growth factors (Friedl and Wolf, 2010).
moving cells, and the tissue structure being invaded. For Different types of adhesion systems contribute directly or indi-
instance, groups of cells can form small clusters, solid strands, rectly to mechanocoupling between the actin cytoskeleton and
or files; if epithelial polarity is retained during migration, these extracellular ECM (Figure 3A) or cell surface scaffolds (Fig-
structures can even form an inner lumen (Friedl and Gilmour, ure 3B). Together, these adhesion systems govern migration.
2009). In most cases of collective cancer invasion, one or several ECM Receptors
leader cells with mesenchymal characteristics form the tip of Integrins are heterodimeric surface receptors composed of
multicellular strands and generate forward traction and pericellu- a and b chains. Together, these chains mediate adhesion and
lar proteolysis toward the tissue structure (Gaggioli et al., 2007; mechanotransduction to extracellular ligands, including a2b1
Khalil and Friedl, 2010). In a second type of collective invasion, integrin predominantly binding to fibrillar collagen; aVb3, aVb1,
a blunt bud-like tip protrudes along tissue space consisting of and a5b1 interacting with fibronectin; and a3b1 and a6b1
multiple cells that variably change position, lacking defined engaging with laminin (Hynes, 2002). After associating with
leader cells (Ewald et al., 2008); this type of invasion occurs ligands, the cytoplasmic tails of integrins connect to cytoskeletal
preferentially in soft tissues and cells of strong epithelial adaptor proteins, including talin, paxillin, and kindlin and the me-
polarity. Collective migration is prevalent in morphogenesis chanosensing modulators vinculin and p130Cas (Geiger et al.,
during development and recapitulated in most epithelial and 2009; Grashoff et al., 2010). Adaptor and mechanosensing mod-
mesenchymal tumor types (Friedl and Gilmour, 2009; Friedl ulator proteins engage with the actin cytoskeleton and trigger
et al., 1995). signaling to protein kinases, including the focal adhesion kinase
Expansive Growth (FAK) and Src (Geiger et al., 2009; Hodivala-Dilke et al., 1999;
Some surrounding tissues impose little to no physical confine- Hynes, 2002). Downstream integrin effectors further include
ment on proliferating tumor cells and thus do not hinder the the small GTPases Rac and Rho, which reinforce cell protrusion
expansion of a cancerous lesion. When tumor cells grow into and rear contraction (Ridley et al., 2003). In addition to contact
these tissues, the increase in volume leads to multicellular to ECM substrate, integrin engagement with extracellular
outward pushing with intact cell-cell junctions and no signs of ligands is also activated by inside-out signaling through Rac,
active migration (Iguchi et al., 2008; Ishizaki et al., 2001). Eventu- the Ras-related GTPase Rap1, and talin (Lee et al., 2009; Ridley
ally, this expansive growth without migration results in spherical et al., 2003).
lesions within a ‘‘capsule’’ of ECM, formed by aligned collagen CD44 and its alternatively spliced variants bind to hyaluronic
fibers in circular orientation (Ishizaki et al., 2001). Expansive acid (i.e., a high-molecular weight glycosaminoglycan abun-
growth may displace cells by volume expansion and pushing dantly present in all connective tissues) and, with low affinity,
when migration activity is absent or, if coupled with migration, to heparan sulfate, collagen, and fibronectin (Zöller, 2011).
contributes to and enhances collective invasion (Ilina et al., CD44 connects to the actin cytoskeleton by the adaptor proteins
2011) (B. Weigelin and P.F., unpublished data). ezrin, radixin, and moesin (ERM) and ankyrin and mediates intra-
Although these migration modes can be classified as morpho- cellular signaling through Src kinase and small Rho GTPases,
logic and mechanistic entities for experimental and conceptual including RhoA (Zöller, 2011). CD44 and its splice variants also
purposes, cells often display features from multiple modes in bind to chemokines and growth factors, and they enhance
three-dimensional (3D) tissues. This includes intermediate or signaling through cis interactions with growth factor receptors,
transition states in which cells may change their molecular including the hepatocyte growth factor receptor (c-Met), fibro-
profiles and switch migration mode (e.g., from proteolytic to non- blast growth factor (FGFR-1), the epidermal growth factor recep-
proteolytic migration or single-cell to collective migration) (Friedl tors (EGFR), and its variants ERBB2–4. Thus, CD44 delivers joint
and Wolf, 2010; Wolf et al., 2003a). For example, when individual ECM and growth factor signaling to invading cells (Couchman,
cells become attracted by the same chemotactic source, they 2010; Zöller, 2011). CD44 also serves as a coreceptor for other
may first undergo multicellular streaming with short-lived cell- adhesion receptors, including integrins and podoplanin (Zöller,
cell junctions that briefly form and resolve again; when cell-cell 2011). Podoplanin is a cell surface mucin that connects to the
adhesion molecules are then upregulated, the cells may join actin cytoskeleton through ezrin. Podoplanin signals to enhance
each other and convert to a collective migration mode (Kulesa RhoA activity, which strongly increases cell invasion (Martı́n-Vil-
and Gammill, 2010). Thus, diverse molecular programs jointly lar et al., 2006; Wicki et al., 2006). Given its extensive crosstalk

Cell 147, November 23, 2011 ª2011 Elsevier Inc. 995

Figure 3. Molecular Determinants of Cell Migration
Simplified view of molecules mediating adhesion and migration signaling.
(A) Cell surface receptors and adaptors that mediate the dynamic interface between the actin cytoskeleton and promigratory signaling and the extracellular matrix
(ECM).
(B) Cell surface proteins that mediate and regulate interactions between cells. Similar adhesion mechanisms may mediate homotypic cell-cell cohesion during
collective invasion and transient and more dynamic heterophilic interaction to resident tissue cells encountered during tissue invasion.
(C) Protease systems upregulated in cancer progression, invasion, and metastasis.
(D) Receptors for chemokines, cytokines, and growth factors, which sense soluble, ECM-, or proteoglycan-bound factors and interaction partners. Green
symbols represent selected intracellular adapters to the actin cytoskeleton, as specified below the drawing (A and B); shaded labels represent major signaling
molecules regulating actin organization and cell migration.

with other receptor/ligand systems, it is unclear whether CD44 et al., 2006). DDRs support E- and N-cadherin-mediated cell-
serves as bona fide adhesion receptor that mediates adhesion cell adhesion (Eswaramoorthy et al., 2010; Shintani et al.,
and mechanotransduction in the absence of other pathways or 2008), and they increase proteolytic cell functions via matrix met-
whether its primary role is to provide cosignaling (Maaser alloproteinases (MMPs), including MMP-1, MMP-2, MMP-9, and
et al., 1999). MMP-10 (Ruiz and Jarai, 2011). The signaling activity of DDRs
Similar to CD44, membrane-bound proteoglycans, such as after they bind to the ECM is well established; however, whether
syndecans, glypicans, and neuropillin, interact through their DDRs connect to the actin cytoskeleton and directly contribute
sugar moieties weakly with ECM components, including hyalur- to mechanotransduction is unknown.
onic acid, fibronectin, collagen, or elastin. These interactions Integrins are thus the main adhesion and mechanotrans-
enhance adhesion in cooperation with integrins, and together duction system for interstitial migration, with modulatory input
with integrins and growth factor receptors, the proteoglycans and crosstalk to alternative cell-ECM and growth factor sig-
deliver signals via PKC and Src (Couchman, 2010; Theocharis naling systems through CD44, cell surface proteoglycans, and
et al., 2010). Syndecans-2 and -4 engage with ezrin or a-actinin, DDRs.
respectively, and they couple to the actin cytoskeleton. Cell-Cell Adhesion Receptors
However, their direct contributions to adhesion and migration Receptors that transmit cell-cell adhesion forces toward the
are still unclear (Couchman, 2010). actin cytoskeleton provide cooperation between tumor cells
The discoidin domain receptors DDR1 and DDR2 interact during collective invasion (Giampieri et al., 2009; Ilina and Friedl,
selectively with fibrillar collagen and transmit signaling via 2009). These receptors also support single-cell and collective
STAT5, NFkB, and p38 MAPK/ERK or the Src-related kinases movement along the surfaces of other tissue-resident cells
Syk, Shc, and Src, respectively (Neuhaus et al., 2011; Vogel encountered during the migration process (Figure 3B).

996 Cell 147, November 23, 2011 ª2011 Elsevier Inc.

 Members of the cadherin family of adhesion receptors mediate pressed by the tumor cells and mediate intravascular migration
homotypic interactions between cells of the same type and and adhesion arrest of circulating tumor cells (Hynes, 2002; Sto-
heterotypic interactions between different cell types. These letov et al., 2010). Integrins may further engage with ECM
interactions include stable cell-cell adhesion through adherens proteins tethered and immobilized on encountered cell surfaces
junctions (Harris and Tepass, 2010), dynamic adhesion via the (e.g., fibronectin and laminin) and mediate cell-cell adhesion
transient co-engagement of small GTPases Rac1 and RhoA, between tumor cells (Casey et al., 2001).
and dynamic junctional remodeling by cytoskeletal dynamics Besides mechanocoupling, CAMs enhance the signaling
(Kardash et al., 2011). In both stable and dynamic cell-cell of integrins and growth factor receptors (e.g., EGFR and
adhesion, cadherins engage with cytoskeletal adaptor and sig- FGFR) through ERK, ILK, or Src (Kiefel et al., 2011; Zecchini
naling proteins, including a-catenin, b-catenin, and p120-cate- et al., 2011). Their contributions to homotypic interaction
nin, which connect to the actin and microtubule cytoskeleton between tumor cells and heterotypic interactions between
(Berx and van Roy, 2009; Harris and Tepass, 2010; Reynolds, tumor and stromal cells make CAMs versatile mechanotrans-
2010). Depending on the type of tumor, different sets of cadherins duction and signaling devices in both single-cell and collective
are expressed and involved in cell-cell interaction, including E-, invasion.
N-, and P-cadherins, cadherin-11, and cadherin-13 (Berx and Several other receptor families contribute to cell-cell contacts
van Roy, 2009). In polarized resting epithelium, E-cadherin and multicellular coordination. These include connexins that
suppresses migration signaling by inhibiting Rac1 (Kitt and form gap junctions (Li et al., 2008), as well as ephrins and Eph
Nelson, 2011) and further maintains cell-cell cohesion, polarity receptors. Ephrins and Eph receptors weaken homotypic and
between the basal and luminal layer of an epithelium, and epithe- heterotypic binding by engaging with alternative sets of ephrins
lial stability. In contrast, in activated and neoplastic epithelium, expressed by neighboring tumor and stromal cells (Astin et al.,
E-cadherin and other cadherins jointly coordinate collective 2010), thereby contributing to tumor cell guidance and migration
movements (Friedl and Gilmour, 2009). In activated epithelial in a tissue context-dependent manner.
cells, cosignaling of E-cadherin and integrins, together with Overall, tumor cells engage in a variety of overlapping and
downstream Src activation, enhances actin dynamics and acto- synergistic cell-matrix and cell-cell adhesion systems that
myosin contractility, leading to both single-cell and collective balance cell-cell cohesion within the tumor and cohesion toward
migration (Geisbrecht and Montell, 2002; Kardash et al., 2010; stromal interfaces.
Martinez-Rico et al., 2010). When co-engaged with DDR1, E- Protease Systems
cadherin signaling limits actomyosin contractility along cell-cell In both tumor and stromal cells, multiple protease systems are
junctions, which stabilizes cell-cell junctions and supports upregulated with overlapping substrate specificities. These
collective invasion (Hidalgo-Carcedo et al., 2011). Compared to systems include MMPs, ADAMs, cathepsins, the serine protease
E-cadherin, N-cadherin and cadherin-7 mediate weaker adhe- urokinase plasminogen activator (uPA), and its receptor uPAR
sion strengths (Chu et al., 2006) and are associated with further (Mason and Joyce, 2011; Rizki et al., 2008) (Figure 3C). Upregu-
increased motility in cancer. This enhanced motility is most lated proteases contribute to tumor invasion and progression
likely due to N-cadherin and cadherin-7’s co-engagement with through at least three distinct mechanisms (Egeblad and Werb,
growth factor receptors, including FGFR or PDGFR, which 2002; Wolf and Friedl, 2011).
enhances downstream signaling through MAPK and PI3K (Berx First, cell surface proteases, notably membrane-type (MT)
and van Roy, 2009). Thus, cadherins show duality in delivering MMPs and ADAMs (a disintegrin and metalloproteinases),
both migration-inhibiting and migration-promoting signaling in a execute contact-dependent proteolysis of structural ECM
context-dependent manner (Martinez-Rico et al., 2010). proteins, including fibrillar and nonfibrillar collagens, fibronectin,
The immunoglobulin superfamily of cell adhesion molecules and laminins, as well as ECM-tethered, matricellular proteins
(CAM) mediates homophilic cell-cell interactions in neoplastic (e.g., tenascin and glypican) (Sabeh et al., 2004, 2009; Wolf
cells through the direct or indirect coupling to the actin cytoskel- et al., 2007). Proteolytic ECM degradation has a dual function:
eton via actin-binding adaptor proteins a-actinin, ankyrin, and (1) it generates biologically active epitopes of ECM components
ezrin (Gavert et al., 2010; Maness and Schachner, 2007). This with adhesion- or migration-promoting effects (Kenny et al.,
family of adhesion molecules includes L1CAM, EpCAM, NCAMs, 2008), and (2) it structurally remodels tissue to form de novo
or ALCAM. gaps and trails bordered by multifiber ECM bundles (Gaggioli
L1CAM is upregulated in the leading front of collectively et al., 2007; Sabeh et al., 2009; Wolf et al., 2007).
invading epithelial tumors that display a stabilized mesenchymal Second, proteases that are secreted and tethered to the cell
phenotype with high invasion capability (Bergmann et al., 2010; surface, notably MMPs and ADAMs, enzymatically process
Hung et al., 2010). This is consistent with a role for L1CAM in other proteases and cell surface receptors, including adhesion
leader-cell function and partial EMT during collective invasion and growth factor receptors (Overall and Blobel, 2007). This
(Gavert et al., 2011). Likewise, ALCAM is upregulated in cell- controls the activation and turnover of these receptors and
cell junctions of collectively invading epithelial cancer associated thus accounts for adaptive changes of receptor availability on
with increased metastasis (van den Brand et al., 2010). both tumor and stromal cells and interstitial protease content.
Similar to leukocytes, tumor cells develop heterophilic cell-cell Finally, secreted proteases, particularly MMPs and plasmin,
interactions with endothelial cells and platelets that express regulate the repertoire of available extracellular growth factors
ICAM-1, VCAM-1, or PECAM-1. These interactions occur by enzymatic activation, inactivation, or degradation (Dean et al.,
through b1 and candidate integrins b2, aVb3, and a4b7 ex- 2008; Mu et al., 2002; Sounni et al., 2010). MMPs and ADAMs

Cell 147, November 23, 2011 ª2011 Elsevier Inc. 997

can release ECM-bound factors, which then form diffusing cell and multicellular tumor invasion are guided and supported
gradients toward neighbor cells (Shiao and Coussens, 2010). by pre-existing structures and interfaces present in every tissue
Thus, the proinvasive tumor microenvironment dominated by (Alexander et al., 2008; Condeelis and Segall, 2003; Grytsenko
proteases consists of both structural ECM remodeling sup- et al., 2011; Schedin and Keely, 2011) (Figure 4). Conceptually,
ported by pericellular proteolysis and deregulated proteolytic tissue structures that guide invasion can be categorized as
processing of chemokines, growth factors, and their receptors, ‘‘2D’’ and ‘‘3D’’ depending on whether cells adhere to a substrate
which impacts both tumor and stromal cells. on one or several sides. 2D surfaces, with even or irregular
Chemokines, Growth Factors, and Their Receptors conformation, form nearly barrier-free track-like gaps and trails
The transition from a fixed, tissue-anchored state to a mobile that typically contain interstitial fluid and glycosaminoglycans.
state is often induced by extracellular chemokines, cytokines, In vivo, most 2D surfaces are encountered in a 3D context,
and growth factors released by tumor cells themselves or acti- such as a second opposing surface, or a nearby 3D scaffold;
vated stromal cells. These factors engage redundant and nonre- therefore, with the notable exception of adherence to the wall
dundant intracellular signaling networks in both tumor and of a larger vessel, cell invasion is, in most cases, constitutively
stromal cells (Figure 3D). Invasion-promoting chemokines three dimensional.
include CXCL12, CXCL10, CCL21, or CCL25. They mediate Inner-body surfaces are always covered with an epithelium or
and perpetuate invasive migration of tumor cells in the primary endothelium layer, and thus, interacting cells depend upon cell-
tumor and likely during metastatic dissemination (Allinen et al., cell rather than cell-matrix interactions. 2D cell surfaces include:
2004; Zlotnik et al., 2011). Migration-promoting signals induced the peritoneum covering all internal organs; the pleura covering
by chemokines and their receptors CXCR4, CXCR3, and CCR9 the lungs and thorax wall; the ventricles of the brain; and inner
are mainly mediated by JAK/PI3K/JNK, PI3K, Src-family kinase surfaces of larger blood and lymph vessels (Figure 4A). Viewed
Syk, and the small GTPases Rac1, RhoA, and Rap1 (El Haibi from a cell mechanics angle, cell surfaces allow for highly effec-
et al., 2010; Lee et al., 2009; Tybulewicz and Henderson, tive, almost barrier-free dissemination of tumor cells. This is
2009). Besides their control on the cell cycle and cell survival, observed during peritoneal or pleural carcinosis in which tumor
many growth factors, including HGF, EGF, FGF, and TGFb, share and other cells readily spread centimeters, likely by both active
signaling through ERK, JNK, Src, mTOR, and PI3K pathways migration and migration-independent passive drift (Zecchini
toward Rac and Cdc42 activation and enhanced cytoskeletal et al., 2011).
dynamics (Massagué, 2008; Shapiro et al., 2011; Trusolino When viewed at microscopic resolution, connective tissue is
et al., 2010). Because of their pleiotropic effects, promigratory not a uniform, homogeneous meshwork of ECM, but rather, it
conditioning of the tumor-associated tissue increases: (1) the is composed of nonrandom structures, including discontinuities
invasion and dissemination of tumor cells; (2) the motility and formed by surface-like gaps and tracks. The anatomic function
activity of stromal cells, including fibroblasts and macrophages; of these gaps and tracks is likely transportation of tissue fluids,
(3) the recruitment and transendothelial migration of circulating tissue elasticity, and mechanical sliding of tissue components
leukocytes and precursor cells into the tumor stroma; and (4) relative to each other. 3D tracks with bordering 2D interfaces
the mobilization of bone marrow-derived cells into the circu- are formed by larger anatomic structures covered by a basement
lation through systemic growth factor effects in other organs, membrane, including small blood vessels, myofibers, nerve
including the bone marrow (Orimo et al., 2005; Padua and Mas- tracks, and adipocytes (Figure 4B). Similar longitudinal tracks
sagué, 2009; Roussos et al., 2011; Zlotnik et al., 2011). are formed by bundled 3D collagen fibers (Figure 4C). These
Thus, multiple overlapping adhesion and signaling networks ‘‘inner surfaces’’ likely correspond anatomically to narrow clefts
cooperate toward molecular and structural reorganization of (‘‘shrinkage artifacts’’) that are abundantly present in virtually
contacted tissues and support tumor cell invasion and meta- every tissue after fixation and, when reconstructed three
static dissemination. dimensionally, display a 3D track system (O. Ilina and P.F.,
unpublished data) along and between fibrillar interstitial tissue
Heterogeneity of Invasion Routes structures. In cancer lesions and tumor xenografts monitored
In vivo, cancer invasion and metastatic dissemination depend by 3D intravital microscopy, these interfaces are often used by
upon two interconnected complementary cell escape strategies. invading cells with little sign of structural alteration or degen-
The first and simplest strategy is the movement of cells along eration (Alexander et al., 2008; Condeelis and Segall, 2003).
pre-existing tissue structures in which the available space Other tissue-specific guidance structures preferentially used
matches or exceeds the volume of the cell or cell group. The by metastasizing cells are bone cavities, which are covered by
second strategy results from proteolytic breakdown of tissue a monolayer of lining cells, and the perivascular tracks in brain
structures to generate de novo space required for invasion vessels formed between glial cells and the basement membrane
(Wolf et al., 2009; Wolf and Friedl, 2011). of vascular smooth muscle cells (Figure 4D). Another special
Guidance Structures in Tissues case of barrier-free dissemination is the lumen of small vessels,
Recently, two-dimensional (2D) and 3D microscopy have map- which provide a tube-like track for rapid intravascular dissemina-
ped the structural organization of tissues during cell invasion, tion of cancer cells through capillaries in peripheral tissue and
and intravital microscopy has been used to examine experi- liver sinusoids (Tsuji et al., 2006) (Figure 4A).
mental tumors in vivo (Pittet and Weissleder, 2011 [this issue of Lastly, 3D scaffolds composed of randomly organized fibrin
Cell]). These approaches, combined with histopathological and collagen fibrils provide a combination of 1D (the string-like
analysis of human tumors, strongly suggest that both single- linear fiber) and 3D scaffold with pores of complex geometry

998 Cell 147, November 23, 2011 ª2011 Elsevier Inc.

Figure 4. Anatomic Tissue Structures Guiding Cancer Invasion
(A) Epithelial and endothelial surfaces devoid of ECM.
(B) Basement membranes interfacing with the ECM between cells and tissues.
(C) Collagen-rich interstitial scaffolds of compact or loose structure and organization.
(D) Complex interfaces composed of both cell surfaces and ECM scaffolds. Solid multimeric scaffold structures interface with tissue pores and track-like gaps
(cyan).

(Doyle et al., 2009; Wolf et al., 2009) (Figure 4C). Such ECM networks are similar to in vitro-reconstituted 3D collagen
networks are predominant in loose connective tissue, such as matrices or basement membrane equivalents frequently used
the dermis of young mice, provisional tissue such as a fibrin for cell invasion research.
clot after tissue wounding, and as largely ECM-free tracks Invading cells are thus required to accommodate diverse
(formed by astrocytes and neuronal fibers) the white matter of geometries and molecular ligand systems for adhesion and
the brain (Grytsenko et al., 2011; Wolf et al., 2009). These migration in vivo. Whereas in vitro-reconstituted ECM models,

Cell 147, November 23, 2011 ª2011 Elsevier Inc. 999

due to their cell-independent polymerization process, mimic substrate enhances and reinforces the clustering of integrins
random ECM structures, complex cell and tissue engineering and the secondary formation of focal adhesions and cytoskeletal
is needed to recapitulate the multicomponent complexity of 2D linkages through the adaptor proteins p130Cas and vinculin
interface-based track geometries of guiding scaffolds in vivo (Grashoff et al., 2010; Sawada et al., 2006). This augments
(Ilina et al., 2011). both cell contractility and mesenchymal functions (Levental
Molecular Guidance Cues et al., 2009).
Physical space is likely translated into directed cell polarity and Consequently, the upregulation of peri-tumor collagen
cytoskeletal dynamics through receptor-mediated molecular production favors cancer cell invasion and metastasis in breast
recognition of the adjacent scaffold structure. Invading cancer and other cancer models (Goetz et al., 2011; Levental et al.,
cells often simultaneously integrate signals from: (1) ECM mole- 2009; Paszek et al., 2005). Besides modulating cell invasion,
cules, including collagens, laminins, fibronectin, and elastin; (2) the physical tissue properties determine the fate of normal cells
cell surfaces, including cadherins, CAMs, and proteoglycans; with consequences for cell growth and differentiation (Discher
and (3) gradients of promigratory factors, i.e., chemotactic et al., 2005). As a central downstream signaling pathway that
(soluble factors) and haptotactic gradients (ECM-bound factors). connects mechanotransduction to gene expression, cell prolifer-
Molecular sensing through adhesion and chemotactic receptors ation in response to substrate stiffness is regulated by Yap1, a
directs tumor cell migration to mediate chemotaxis and hapto- transcription factor downstream of the hippo pathway (Dupont
taxis (i.e., directional moltility along a gradient of cell adhesion et al., 2011). Yap1 engagement supports epithelial stem cell
sites or substrate-bound chemoattractants) jointly with physical growth and hyperproliferation, which is counteracted by
contact guidance, which is often tissue context dependent. signaling through a-catenin (Schlegelmilch et al., 2011). Thus,
Besides the physical scaffold structure, the ECM guidance of Yap1 represents an important mechanosensitive candidate
invading cells is mediated by covalently and noncovalently effector for neoplastic progression in cells with deregulated cad-
associated accessory components deposited by stromal cells. herin/catenin axis.
Guidance by ECM Guidance along Cell Membranes
The molecular and physical characteristics of the ECM strongly Besides cell-ECM interactions mediating cell migration, an
contribute to cell adhesion, migration, and cell fate decisions understudied but emerging mechanism is the guidance by cell-
with consequences for cancer cell invasion and dissemination. cell junctions. Besides the epithelium covering inner-body cavi-
The quantitatively most abundant and important component of ties and endothelium forming the lumen of blood and lymph
connective tissue is collagen type I, which serves as structural vessels, intermittent and likely discontinuous cell scaffolds are
frame for cells and other scaffold proteins (Grytsenko et al., abundant in most tissues, including fibroblast networks, macro-
2011; Wolf et al., 2009). In the activated tumor stroma, the phages, and epithelial structures (Figure 4). In developing zebra-
density of collagen fibers is often increased (i.e., desmoplasia), fish, primordial amoeboid germ cells migrating individually
and hyaluronan, proteoglycans, and glycoproteins (e.g., fibrin, through a cell-rich tissue scaffold employ E-cadherin and Rho-
fibronectin, and vitronectin) become upregulated. Together mediated actomyosin contraction for migration (Kardash et al.,
these molecules decorate and ‘‘functionalize’’ the collagen scaf- 2011). The small group of germ cells of the developing
fold. In vitro, mesenchymal fibroblasts and cancer cells migrate Drosophila ovary, called border cells, provides an example of
along 2D or through 3D collagen using a2b1, a1b1, or a11b1 in- collective migration mediated by E-cadherin. These cells are
tegrins and a5b1, aVb3, and aVb5 for migration along or through connected by E-cadherin, guided by EGF, and depend upon
fibrin or fibronectin scaffolds (Even-Ram and Yamada, 2005; E-cadherin for migration (Geisbrecht and Montell, 2002). Cell-
Maaser et al., 1999). cell junctions thus represent an alternative mechanotransduc-
Integrins also mediate the migration of normal and cancer cells tion mechanism for migration, when cell-matrix adhesions are
along structural components of basement membranes by downregulated or absent. However, though effective in morpho-
engaging with collagen type IV (a1b1, a2b1), laminins (a3b1, genesis, the role of cadherin-based cell-cell interactions in tumor
a6b1), fibrillin (a5b1, aVb3, aVb6), perlecan, and versican (b1) cell invasion is still unclear.
(Hynes, 2002). Consequently, perivascular invasion of glioma Secreted Guidance Molecules
cells and perineural invasion of pancreatic cancer cells are linked Many chemokines and growth factors contain one or several
to the function of laminin-binding b1 integrins (Piao et al., 2009; ECM-binding domains, which immobilize the factors in tissues,
Ryschich et al., 2009), but the mechanisms of other basement thereby forming a stable promigratory scaffold. Chemokines
membrane-dependent routes, including peri- and intramuscular and growth factors contain binding sites to heparan sulfate
guidance and adipose tissue invasion, are unknown. side chains (Lortat-Jacob et al., 2002) that are present in intersti-
Most ECM proteins undergo enzymatic postprocessing by tial and cell surface proteoglycans and heparin (Hynes, 2009).
cell-derived proteases (Figure 3C) or crosslinking proteins. After functionalization, a scaffold contains both adhesion sites
Fibrillar collagens become glycosylated and crosslinked by lysyl (for integrins and other receptors) and immobilized migration-
oxydases (LOX) and lysyl hydroxylases, which increases inducing signal (via CCR or GFR) on the same geometric struc-
collagen stiffness and resistance to assault by pH changes and ture, such as a fibril or basement membrane, to support cell
proteolytic degradation (Levental et al., 2009; Paszek et al., protrusions and adhesion in close vicinity along the same
2005). Normal and neoplastic cells sense differences in physical substrate. In 3D invasion models of branching morphogenesis,
ECM properties and migrate preferentially toward regions of which support collective sprouting of epithelial ducts of the
increased stiffness, termed durotaxis (Lo et al., 2000). Stiff mammary or salivary gland, immobilization of FGF10 to heparan

1000 Cell 147, November 23, 2011 ª2011 Elsevier Inc.

Figure 5. Plasticity of Cell-Matrix Interaction, Invasion, and Tissue Remodeling
(A) Migrating cells transition from an initial nondestructive dissemination to migration that involves small- and large-scale tissue remodeling. The pre-existing
space available to invading cells governs the caliber of individual and multicellular invasion and becomes iteratively widened by pericellular proteolysis.
(B) Epithelial-to-mesenchymal transition of a stable epithelium after downregulation of cell-cell junctions and facilitated single-cell detachment.
(C and D) Invasion programs display plasticity, or adaptability, including transition from collective cell migration to individual cell migration (C) and mesenchymal-
to-amoeboid transition (D). Key regulators of these transitions altered in expression or function are indicated.

sulfate maximizes duct elongation and growth by engaging recruited into the tumor and contribute to the formation of tumor
FGFR2 signaling; in contrast, a diffusion-only variant of FGF10 blood vessels (Orimo et al., 2005), and osteopontin activates
lacks the proinvasive function and supports only growth (Makar- bone marrow-derived cells that integrate into the tumor and
enkova et al., 2009). Likewise, the chemokine CXCL12/SDF-1 accelerate tumor outgrowth (McAllister et al., 2008).
and EGF interact with interstitial heparan sulfate and form stable Thus, multiple and partially overlapping mechanisms
gradients that guide migrating tumor cells and passenger leuko- contribute to the mechanical and molecular guidance of tumor
cytes (Allinen et al., 2004; Netelenbos et al., 2002; Wyckoff et al., cells, but their crosstalk and hierarchy still remain unknown.
2004).
TGFb is a master inducer of mesenchymal invasion and stem Plasticity of Invasion and Metastasis Programs
cell functions in cancer cells. It is immobilized to the ECM via Together, the different modes of cancer cell invasion, the recep-
fibronectin and fibrillin by latent TGFb-binding proteins and tors and cytoskeletal regulators available for cell-cell and cell-
becomes released through limited proteolysis mediated by matrix adhesion, the divergent degree of ECM remodeling capa-
MMPs or furin that are activated and released by activated bility, and the range of invasion-guiding molecular and physical
stromal cells (Mu et al., 2002). Alternatively, integrins bind to tissue environments provide a multiscale framework of combina-
latent TGFb-binding protein complexes and pull to induce a torial possibilities or states that allow cancer invasion to be
conformational change that is required to make ECM-bound a plastic and adaptive process (Friedl, 2004; Friedl and Wolf,
TGFb accessible for its receptors (Wipff et al., 2007). Osteopontin 2010). Consequently, with altered tissue composition and condi-
is a secreted cytokine-like proteoglycan that binds to CD44 and tioning by released factors, tumor cells undergo changes in
integrins. Osteopontin is upregulated in many tumor types and signaling and function that lead to secondary effects in the
strongly supports invasive cell guidance (Bellahcène et al., 2008). invaded tissue and, in turn, the tumor cells themselves.
In addition to these examples, gene expression profiling and Plasticity of Tissue Structures
proteomics have revealed abundant sets of soluble factors and Interstitial cancer cell invasion occurs in different phases that
ECM proteins upregulated in the microenvironment of tumors, can be labeled operationally as an initial, nondestructive guid-
indicative of complex signaling pathways induced in both tumor ance phase, followed by a phase of tissue remodeling. In a step-
and stromal cells (Allinen et al., 2004; Wang et al., 2004). Besides wise manner, invasive migration leads to the production of
their local peri-tumor function, many deposited factors diffuse pores, tunnels, and lagunae, which guide and can be populated
into blood and lymphatic fluid, which unfolds systemic, by mobile tumor cells (Alexander et al., 2008; Condeelis and
hormone-like functions. CXCL12/SDF-1 and osteopontin are Segall, 2003) (S.A., unpublished data) (Figure 5A). With upregu-
important examples of these soluble factors. Tumor-derived lated MMPs, most notably MT1-MMP/MMP14, pericellular
CXCL12/SDF-1 mobilizes bone marrow stem cells that become proteolysis executed by tumor cells themselves or fibroblasts

Cell 147, November 23, 2011 ª2011 Elsevier Inc. 1001

generates micro- and macrotracks bordered by condensed During EMT, upstream signals through growth factors of the
ordered collagen bundles, which strongly support both single- tumor stroma, including Wnt, TGFb, FGF, and EGF, lead to the
cell and collective invasion (Friedl et al., 1997; Gaggioli et al., activation of transcriptional repressors, including ZEB1, Twist,
2007; Goetz et al., 2011; Wolf et al., 2007) (Figure 5A). These and Snail1 and 2, which directly and indirectly inhibit E-cadherin
de novo tracks guide tumor cells and, with pressure exerted by transcription (Spaderna et al., 2008; Yang et al., 2004). For
the invading cells, become gradually widened until the tissue example, in breast cancer cells undergoing EMT in response to
space consumed by invading cell masses matches the regres- MMP-3 (which cleaves cell surface E-cadherin and thus
sion of the ECM (Gaggioli et al., 2007; Ilina et al., 2011; Wolf weakens cell-cell junctions), the onset of migration depends on
and Friedl, 2011). Such trails, often filled by tumor cells, are Rac activation and cell-derived production of reactive oxygen
abundant in most interstitial collagen-rich tissues, including des- species (ROS), which in turn upregulate Snail (Radisky et al.,
moplastic stroma (Levental et al., 2009; Paszek et al., 2005). 2005). With E-cadherin expression diminished, adherens junc-
Thus, despite its increased absolute collagen density, the tions and the signaling thereof are weakened or replaced by
signals, gaps, and trails present in desmoplastic stroma enhance less stringent cell-cell adhesions through N-cadherin or
cancer invasion and progression rather than acting as barrier. L1CAM (Gavert et al., 2007; Yano et al., 2004). This results in
By direct and indirect mechanisms, desmoplastic tissue the disturbance of apicobasal polarity and cell anchoring to the
remodeling is a strong mediator of neoplastic progression, inva- basement membrane, which, in turn, allows the cells to acquire
sion, and metastasis (Egeblad et al., 2010). As key mediators of a mobile mesenchymal phenotype (Thiery et al., 2009). The
desmoplasia, resident fibroblasts and immigrated fibroblast EMT program further favors a stem cell-like phenotype that
precursor cells receive activation signals through growth factors, invades, disseminates, and is able to establish distant metas-
including TGFb, IL-1, and PDGF. They then develop into cancer- tases (Mani et al., 2008).
associated myofibroblasts (CAF) that deposit, remodel, and The induction of EMT with downregulation of E-cadherin
contract fibrillar collagen (De Wever et al., 2008; Egeblad et al., expression is likely tunable, dependent on whether complete
2010). As a physiological process of the connective tissue reor- or partial EMT signaling is present. As consequence, EMT can
ganization, the postpartum involution of the mammary gland be complete with loss of E-cadherin and the typical EMT
leads to the deposition of fibrillar collagen and collagen-induced signaling and protein expression profile. However, EMT may
release of proinflammatory COX-2 (Lyons et al., 2011). This be partial with different levels of E-cadherin expression retained,
reorganization is sufficient to impose growth, invasion as multi- and even EMT-like dissemination without EMT-associated gene
cellular strands and metastasis programs in otherwise benign expression patterns may develop (Christiansen and Rajase-
or less aggressive breast tumor lesions (Lyons et al., 2011). A karan, 2006; Gavert et al., 2011; Pàez-Ribes et al., 2009; Wicki
similar progression of breast cancer is triggered by activated et al., 2006). In epithelial cancer lesions, EMT is detected in
fibroblasts that reorganize and condense the breast stroma to a few often cohesive cells located at the leading edge, as well
aligned, bundled collagen tracks that condition breast cancer as small cohesive groups and individual cells scattered and
invasion and metastasis (Goetz et al., 2011). moving independently without connection to the main tumor
In bone metastases, bone resorption is executed by osteo- (Brabletz et al., 2001; Gavert et al., 2007). Thus, besides repre-
clasts in which RANKL (receptor activator of NF-kB ligand) is senting a program for complete loss of cell-cell junctions, EMT
activated by a TGF-b and MMP13-dependent mechanism; further may contribute to collective cell functions, including
this allows the growing tumor to expand into de novo space, collective invasion. This is consistent with the prominent collec-
which eventually results in local bone destruction (Nannuru tive invasion of primary mesenchymal tumors and melanoma
et al., 2010; Nannuru and Singh, 2010). In all cases, pre-existing (Alexander et al., 2008; Hegerfeldt et al., 2002).
tissue space is first filled by invading cells without apparent EMT is also thought to represent a program transiently
degradation, and then, with increasing cell density and the upre- controlled by the microenvironment, which locally downregu-
gulation of MMPs and other proteases, the tissue is degraded lates epithelial characteristics and facilitates cell escape from
and reorganized. As an outcome, cancer invasion leads to the primary tumor. However, with local upstream signaling lost,
secondary loss of tissue integrity and function, including tissue cells undergo mesenchymal-to-epithelial reversion after meta-
necrosis, ulceration, and vessel rupture. Accordingly, the struc- static seeding in the secondary organ (Spaderna et al., 2006;
tures detected by histology represent statically looking snap- Thiery et al., 2009). Thus, EMT-dependent invasion and metas-
shots of an otherwise dynamic and plastic process by which tasis programs are strongly responsive to microenvironmental
the growing and invading tumor replaces and eventually changes and adaptive in their signaling program and associated
destroys interstitial tissue. invasion dynamics.
Plasticity of Cell-Cell Junctions: Plasticity of Cell-Matrix Interactions
The Epithelial-to-Mesenchymal Transition and Cytoskeletal Dynamics
A central molecular program enhancing tumor cell invasion in The executive mechanotransducing mechanisms of cell migra-
response to environmental triggers is the epithelial-to-mesen- tion are plastic and allow the rapid adaptation to environmental
chymal transition (EMT). EMT initiates or augments invasive changes and challenges; these adaptations often result in transi-
functions by enhancing Rac-dependent mesenchymal migra- tions between different modes of migration (Friedl and Wolf,
tion. It also contributes to cell growth, cell survival, and the ree- 2010; Sanz-Moreno and Marshall, 2010). Such plasticity likely
mergence of stem cell characteristics (Thiery et al., 2009) originates in response to tissue microregions and responses to
(Figure 5B). therapeutic challenge. The natural regulation of gene expression

1002 Cell 147, November 23, 2011 ª2011 Elsevier Inc.

and signal states in tumor cells by the microenvironment thus lead to plasticity of invasion and metastasis. Mutation or loss
accounts for the often heterogeneous invasion pattern in of p53 leads to enhanced integrin turnover and recycling, which
progressing tumor lesions. In addition, diversity of persisting converts cells to the amoeboid dissemination mode and
invasion is caused by rewiring of signaling networks and differ- strongly enhances invasion/metastasis (Gadea et al., 2007;
ential cell survival during tissue damage and therapy (Alexander Muller et al., 2009).
and Friedl, 2012). Thus, in context, with gain or loss of cell-cell junctions and
In collectively invading tumors, cell-cell coordination and adaptive signaling control through Rac and Rho, cancer invasion
signaling are mediated by either E-cadherin expressed at levels programs are plastic and responsive to microenvironmental
that do not confound the migration process or alternative cad- signals and molecular interference, which secures migration
herins, including N- or VE-cadherin (Yano et al., 2004) (Fig- under challenging conditions (Madsen and Sahai, 2010). These
ure 5C). Mechanisms of collective invasion with expressed basic conversion mechanisms have been established for
E-cadherin in cell-cell junctions include the upregulation of: in vitro conditions using cell lines, and their relevance for tumor
L1CAM, which strongly promotes migration (Gavert et al., lesions in vivo await confirmation by using 3D histopathology
2011; Shtutman et al., 2006); the guanin nucleotide exchange combined with intravital imaging (Pittet and Weissleder, 2011).
factor Tiam-1, which activates Rac1 but maintains adherens Likewise, how invasion plasticity is connected with or distinct
and tight junctions (Mertens et al., 2005; Walch et al., 2008); or from EMT programs remains to be shown in vitro and in vivo.
podoplanin, which increases RhoA activity in the presence of This will identify EMT-dependent and -independent routes and
E-cadherin-based adhesions (Wicki et al., 2006). niches of natural and therapy-induced plasticity of invasion
Similar to the EMT program, the transition from collective cell and their contribution to metastatic dissemination (Christiansen
migration to individual cell migration (i.e., the collection-to- and Rajasekaran, 2006).
individual transition) is triggered by local Rac1 engagement,
allowing for ectopic tip cell behavior, substrate engagement, Plasticity and Reciprocity—A Model
and eventually, cell detachment (which is facilitated by down- The mechanisms of spatiotemporal plasticity (i.e., to change
regulation of cadherin-based cell-cell adhesion) (Figure 5C). phenotype and function) and reciprocity (i.e., to do this by pro-
Environmental stimuli can favor single-cell detachment from cessing signals received from the environment) are fundamental
tumors, partly through EMT and partly in the absence of EMT to the step-wise changes in both tumor cells and the microenvi-
(Bertout et al., 2008; Pennacchietti et al., 2003). These stimuli ronment, a process that receives further drift with cells moving
include TGFb, EGF, and other growth factors, but also metabolic from one environment to another. The concept of dynamic reci-
stress, such as acidification of the stroma causing a shift in tumor procity for cells engaging with and thereby altering the ECM was
metabolism (i.e., the Warburg effect) and hypoxia with reactive originally coined by Paul Bornstein for cell-matrix interactions in
HIF-1a signaling (Bertout et al., 2008; Pennacchietti et al., 2003). wound healing (Bornstein et al., 1982) and was further developed
Alternatively, amoeboid dissemination may originate from by Mina Bissell and coworkers for epithelial morphogenesis
collective invasion when cell-cell junctions are abandoned and (Bissell et al., 1982) and cancer (Nelson et al., 2008; Xu et al.,
release the cells toward a single-cell migration program of low 2009). Accordingly, plasticity and reciprocity account for the
integrin-mediated adhesion and high Rho-mediated cortical morphologic and functional inter- and intralesion heterogeneity
actomyosin contractility (Hegerfeldt et al., 2002; Sanz-Moreno driven by complementary mechanisms, including genomic insta-
et al., 2008). In breast cancer lesions, EGF secreted predomi- bility as well as epigenetic, signaling, and functional adaption to
nantly by activated macrophages activates and guides tumor cope with altering environmental conditions (Figure 6A). Such
cells that have detached from the epithelial main mass by amoe- ‘‘fate-changing’’ events that trigger significant adaptation in
boid dissemination (Wyckoff et al., 2004). Thus, the type of tumor cells occur in response to metabolic changes in the micro-
migration maintained after detachment from the multicellular environment, including hypoxia and severe metabolic stress
state depends upon the governance of adhesion strength, cyto- (Bertout et al., 2008), as well as chronic growth factor stimulation
skeletal protrusions and contractility, and the competence to and inflammation (Allinen et al., 2004; Polyak et al., 2009)
remodel the ECM. (Figure 6B).
Mesenchymal invasion may undergo secondary conversion A critical common mediator of plasticity and reciprocity is the
to amoeboid, rounded migration by diverse mechanisms, such change of cell position. Cell invasion provides access to different
as a decrease in Rac activity and concomitant activation of physical and molecular structures, including the local tumor stroma
Rho-mediated actomyosin contractility (Sahai and Marshall, and secondary organs after metastatic colonization, and thus
2003). The therapeutic inhibition of MMPs can also trigger refines signaling input (Figures 6B and 6C). Consequently, cancer
amoeboid migration because pericellular proteolysis by MMPs invasion and metastasis are both cause and consequence of
prompts conversion to nonproteolytic amoeboid cell deforma- plasticity and reciprocity. Over time, the changes driving adaptive
tion to bypass narrow ECM barriers (Wolf et al., 2003a) reprogramming of tumor cells and the reactive tumor stroma thus
(Figure 5D). Likewise, amoeboid movement in otherwise mesen- lead to a kinetic, ever-changing coevolution of the tumor with its
chymal cells is induced by inhibiting chemokine-meditated environment (Hanahan and Weinberg, 2011; Polyak et al., 2009).
Rac activation (Gérard et al., 2007), activating Rho by inhibiting Related tissue remodeling processes, such as morphogenesis
p190RhoGAP (Nimnual et al., 2003), or engaging EphA2 (an and wound healing, follow well-defined programs with rate-limiting
indirect Rho activator) (Parri et al., 2009). Lastly, gene mutations steps and end points (e.g., for limb or organ formation or closure of
that impact integrin availability or the Rac/Rho balance may a tissue defect). In contrast, cancer is more flexible in time, space,

Cell 147, November 23, 2011 ª2011 Elsevier Inc. 1003

 signaling programs, and genetics and thus represents a perpetu-
ating process without a clear end point, illustrated here as open-
ended spirals (Figure 6).

Concluding Remarks
Well-defined experimental conditions in vitro have allowed the
precise delineation of receptor-ligand interactions and their
basic involvement in invasive migration, but their complexity
and synergistic availability in vivo make it challenging to identify
the dominant and compensation mechanisms that maintain and
rescue metastatic dissemination (Friedl and Wolf, 2003; Madsen
and Sahai, 2010; Sanz-Moreno and Marshall, 2010). Although
the intracellular machinery that generates force via actomyosin
is well-defined, the range of molecular and physical adhesion
and transmission modes at the cell and tissue level support the
adaptation of cell migration. This adaptation is similar to and
likely intertwined with the compensation and plasticity of focal
adhesion and other signaling networks during cell invasion
(Zaidel-Bar et al., 2007). Consequently, because most pathways
of adhesion, proteases, and chemokine/growth factors exhibit
ample overlap and redundancy, the natural or therapeutically
induced loss of one mechanistic pathway may lead to a drift in
signaling and mechanotransducing effector networks. This
may trigger alternative mechanisms of invasion and dissemina-
tion instead of inhibiting function.
Recent studies identified two unexpected examples for alter-
native migration modes in 3D environments: the interstitial mi-
gration of leukocytes independently of integrin (Friedl and Wei-
gelin, 2008; Lämmermann et al., 2008; Lämmermann and Sixt,
2009) and the propulsive cell migration of normal and neoplastic
cells by blebs (Fackler and Grosse, 2008; Lorentzen et al., 2011;
Poincloux et al., 2011). The contribution of these types of cell
movement to metastatic cancer invasion in vivo awaits determi-
nation. In addition to active actomyosin-driven migration, other
mechanisms of cell transport include passive drift along tissue
Figure 6. Reciprocity, Plasticity, and Evolution of Tumor Cell Inva- structures and cell pushing by expansive growth. Both of these
sion and Metastasis alternative mechanisms still have not been formally integrated
(A) Reciprocal crosstalk between tumor cells and the stroma (i.e., stromal cells into the spectrum of cell translocation principles.
together with ECM and released factors) results in evolutionary plasticity of
both tumor cells and the tissue environment. ‘‘Reciprocity’’ results from the
The invaded tissue is often regarded as a homogeneous and
bidirectional communication between stromal (S) and tumor (T) compart- passive scaffold that the tumor cells modify in a unidirectional
ments, which is transmitted by mediators (M) released by both compartments manner, and indeed, this notion has been strengthened by
in a reciprocal manner. Stromal alterations include cell-derived physico-
most 2D and 3D in vitro models (Friedl et al., 1997; Wolf et al.,
chemical changes of the microenvironment, such as deposited ECM
components, ECM degradation and remodeling, change of ECM stiffness and 2009). The concept of plasticity and reciprocity of cancer inva-
porosity, and released cytokines and growth factors. Plasticity of the cell sion, however, describes invasion as a reciprocal process gov-
phenotype and function consists of: changes in the activation, migration, and erned by multiple sets of overlapping, redundant, and poten-
differentiation state of the cell; metabolic switches; and epigenetic alterations
that may further prompt secondary genomic instability. Consequently, with tially, ever-changing active and passive mechanisms of
each cycle of interactive engagement with the stroma, the cell state diverges molecular mechanotransduction. This adaptability renders cell
from its origin, leading to progression of the tumor or the metastasis (indicated invasion and metastasis as a robust perpetuating process, the
as spiral).
(B) Branching and altered direction of reciprocal plasticity in the course of
targeting of which—if ever possible—will require understanding
cancer progression. Direction-changing dichotomy is reached by a change in the hierarchy of stringent control points. To this end, systems
the position of the tumor cell, which results in a different tissue location and biology and mathematical modeling approaches are required
change in environmental input (upper spirals); likewise, step-wise bifurcation to classify rate-limiting nodes and modifiers of molecular mecha-
of reciprocal evolution may be induced by changes of the local tissue condi-
tions, including altered composition of infiltrate cells during inflammatory and notransduction for each migration mode and tissue-context.
metabolic stress, insufficient perfusion resulting in hypoxia, and tissue repair
programs induced by spontaneous or therapy-induced (tumor) necrosis (lower
spirals). inputs synergize to drive local plasticity of tumor cells and the tumor stroma,
(C) Second- and third-order reciprocity. Reciprocal plasticity can evolve during resulting in the evolution of tumor subregions with diverse progression and
metastatic progression to generate second- and third-order reciprocity. In the adaptation capabilities. As a consequence, reciprocity and plasticity may
course of metastasis, tissue-specific reciprocity and microenvironmental impose parallel or divergent evolution of cell clones and populations.

1004 Cell 147, November 23, 2011 ª2011 Elsevier Inc.

ACKNOWLEDGMENTS Christiansen, J.J., and Rajasekaran, A.K. (2006). Reassessing epithelial to
mesenchymal transition as a prerequisite for carcinoma invasion and metas-
We gratefully acknowledge Bettina Weigelin, Antoine Khalil, and Vaishnavi tasis. Cancer Res. 66, 8319–8326.
Narasimhan for helpful comments and corrections on the manuscript. Chu, Y.S., Eder, O., Thomas, W.A., Simcha, I., Pincet, F., Ben-Ze’ev, A., Perez,
The work of the laboratory is supported by grants from the Netherlands’ E., Thiery, J.P., and Dufour, S. (2006). Prototypical type I E-cadherin and type II
Science Foundation (NWO-VICI 918.11.626), the Dutch Cancer Foundation cadherin-7 mediate very distinct adhesiveness through their extracellular
(KWF 2008-4031), the European Union (ENCITE HEALTH TH-15-2008- domains. J. Biol. Chem. 281, 2901–2910.
208142; FP7-T3Net-237946; FP7-PEOPLE-2010-IEF-276443), and the Deut-
Condeelis, J., and Segall, J.E. (2003). Intravital imaging of cell movement in
sche Forschungsgemeinschaft (SPP1190/FR1155/8-3).
tumours. Nat. Rev. Cancer 3, 921–930.
Couchman, J.R. (2010). Transmembrane signaling proteoglycans. Annu. Rev.
REFERENCES Cell Dev. Biol. 26, 89–114.
Cox, E.A., Sastry, S.K., and Huttenlocher, A. (2001). Integrin-mediated adhe-
Alexander, S., and Friedl, P. (2012). Cancer invasion and resistance - intercon-
sion regulates cell polarity and membrane protrusion through the Rho family
nected processes of disease progression and therapy failure. Trends Mol.
of GTPases. Mol. Biol. Cell 12, 265–277.
Med. 10.1016/j.molmed.2011.11.003.
De Wever, O., Demetter, P., Mareel, M., and Bracke, M. (2008). Stromal myo-
Alexander, S., Koehl, G.E., Hirschberg, M., Geissler, E.K., and Friedl, P. (2008).
fibroblasts are drivers of invasive cancer growth. Int. J. Cancer 123, 2229–
Dynamic imaging of cancer growth and invasion: a modified skin-fold chamber
2238.
model. Histochem. Cell Biol. 130, 1147–1154.
Dean, R.A., Cox, J.H., Bellac, C.L., Doucet, A., Starr, A.E., and Overall, C.M.
Allinen, M., Beroukhim, R., Cai, L., Brennan, C., Lahti-Domenici, J., Huang,
(2008). Macrophage-specific metalloelastase (MMP-12) truncates and inacti-
H., Porter, D., Hu, M., Chin, L., Richardson, A., et al. (2004). Molecular char-
vates ELR+ CXC chemokines and generates CCL2, -7, -8, and -13 antago-
acterization of the tumor microenvironment in breast cancer. Cancer Cell 6,
nists: potential role of the macrophage in terminating polymorphonuclear
17–32.
leukocyte influx. Blood 112, 3455–3464.
Armstrong, A.J., Marengo, M.S., Oltean, S., Kemeny, G., Bitting, R.L., Turnbull,
Discher, D.E., Janmey, P., and Wang, Y.L. (2005). Tissue cells feel and respond
J.D., Herold, C.I., Marcom, P.K., George, D.J., and Garcia-Blanco, M.A.
to the stiffness of their substrate. Science 310, 1139–1143.
(2011). Circulating tumor cells from patients with advanced prostate and
breast cancer display both epithelial and mesenchymal markers. Mol. Cancer Doyle, A.D., Wang, F.W., Matsumoto, K., and Yamada, K.M. (2009). One-
Res. 9, 997–1007. dimensional topography underlies three-dimensional fibrillar cell migration.
J. Cell Biol. 184, 481–490.
Astin, J.W., Batson, J., Kadir, S., Charlet, J., Persad, R.A., Gillatt, D., Oxley,
J.D., and Nobes, C.D. (2010). Competition amongst Eph receptors regulates Dupont, S., Morsut, L., Aragona, M., Enzo, E., Giulitti, S., Cordenonsi, M., Zan-
contact inhibition of locomotion and invasiveness in prostate cancer cells. conato, F., Le Digabel, J., Forcato, M., Bicciato, S., et al. (2011). Role of YAP/
Nat. Cell Biol. 12, 1194–1204. TAZ in mechanotransduction. Nature 474, 179–183.

Bellahcène, A., Castronovo, V., Ogbureke, K.U., Fisher, L.W., and Fedarko, Egeblad, M., Rasch, M.G., and Weaver, V.M. (2010). Dynamic interplay
N.S. (2008). Small integrin-binding ligand N-linked glycoproteins (SIBLINGs): between the collagen scaffold and tumor evolution. Curr. Opin. Cell Biol. 22,
multifunctional proteins in cancer. Nat. Rev. Cancer 8, 212–226. 697–706.

Bergmann, F., Wandschneider, F., Sipos, B., Moldenhauer, G., Schniewind, Egeblad, M., and Werb, Z. (2002). New functions for the matrix metalloprotei-
B., Welsch, T., Schirrmacher, P., Klöppel, G., Altevogt, P., Schäfer, H., and Se- nases in cancer progression. Nat. Rev. Cancer 2, 161–174.
bens Müerköster, S. (2010). Elevated L1CAM expression in precursor lesions El Haibi, C.P., Sharma, P.K., Singh, R., Johnson, P.R., Suttles, J., Singh, S.,
and primary and metastastic tissues of pancreatic ductal adenocarcinoma. and Lillard, J.W., Jr. (2010). PI3Kp110-, Src-, FAK-dependent and DOCK2-
Oncol. Rep. 24, 909–915. independent migration and invasion of CXCL13-stimulated prostate cancer
Bertout, J.A., Patel, S.A., and Simon, M.C. (2008). The impact of O2 availability cells. Mol. Cancer 9, 85.
on human cancer. Nat. Rev. Cancer 8, 967–975. Estecha, A., Sánchez-Martı́n, L., Puig-Kröger, A., Bartolomé, R.A., Teixidó, J.,
Berx, G., and van Roy, F. (2009). Involvement of members of the cadherin Samaniego, R., and Sánchez-Mateos, P. (2009). Moesin orchestrates cortical
superfamily in cancer. Cold Spring Harb. Perspect. Biol. 1, a003129. polarity of melanoma tumour cells to initiate 3D invasion. J. Cell Sci. 122, 3492–
3501.
Bissell, M.J., Hall, H.G., and Parry, G. (1982). How does the extracellular matrix
direct gene expression? J. Theor. Biol. 99, 31–68. Eswaramoorthy, R., Wang, C.K., Chen, W.C., Tang, M.J., Ho, M.L., Hwang,
C.C., Wang, H.M., and Wang, C.Z. (2010). DDR1 regulates the stabilization
Bornstein, P., McPherson, J., and Sage, H. (1982). Synthesis and secretion of
of cell surface E-cadherin and E-cadherin-mediated cell aggregation. J. Cell.
structural macromolecules by endothelial cells in culture. In Pathobiology of
Physiol. 224, 387–397.
the Endothelial Cell, P&S Biomedical Sciences Symposia, Volume 6, H. Nossel
and H. Vogel, eds. (New York: Academic Press), pp. 215–228. Even-Ram, S., and Yamada, K.M. (2005). Cell migration in 3D matrix. Curr.
Opin. Cell Biol. 17, 524–532.
Brabletz, T., Jung, A., Reu, S., Porzner, M., Hlubek, F., Kunz-Schughart, L.A.,
Knuechel, R., and Kirchner, T. (2001). Variable beta-catenin expression in Ewald, A.J., Brenot, A., Duong, M., Chan, B.S., and Werb, Z. (2008). Collective
colorectal cancers indicates tumor progression driven by the tumor environ- epithelial migration and cell rearrangements drive mammary branching
ment. Proc. Natl. Acad. Sci. USA 98, 10356–10361. morphogenesis. Dev. Cell 14, 570–581.

Casey, R.C., Burleson, K.M., Skubitz, K.M., Pambuccian, S.E., Oegema, T.R., Fackler, O.T., and Grosse, R. (2008). Cell motility through plasma membrane
Jr., Ruff, L.E., and Skubitz, A.P. (2001). Beta 1-integrins regulate the formation blebbing. J. Cell Biol. 181, 879–884.
and adhesion of ovarian carcinoma multicellular spheroids. Am. J. Pathol. 159, Fidler, I.J. (2003). The pathogenesis of cancer metastasis: the ‘seed and soil’
2071–2080. hypothesis revisited. Nat. Rev. Cancer 3, 453–458.
Chambers, A.F., Groom, A.C., and MacDonald, I.C. (2002). Dissemination and Friedl, P. (2004). Prespecification and plasticity: shifting mechanisms of cell
growth of cancer cells in metastatic sites. Nat. Rev. Cancer 2, 563–572. migration. Curr. Opin. Cell Biol. 16, 14–23.
Choi, Y.P., Shim, H.S., Gao, M.Q., Kang, S., and Cho, N.H. (2011). Molecular Friedl, P., Borgmann, S., and Bröcker, E.B. (2001). Amoeboid leukocyte crawl-
portraits of intratumoral heterogeneity in human ovarian cancer. Cancer Lett. ing through extracellular matrix: lessons from the Dictyostelium paradigm of
307, 62–71. cell movement. J. Leukoc. Biol. 70, 491–509.

Cell 147, November 23, 2011 ª2011 Elsevier Inc. 1005

Friedl, P., and Gilmour, D. (2009). Collective cell migration in morphogenesis, Harris, T.J., and Tepass, U. (2010). Adherens junctions: from molecules to
regeneration and cancer. Nat. Rev. Mol. Cell Biol. 10, 445–457. morphogenesis. Nat. Rev. Mol. Cell Biol. 11, 502–514.
Friedl, P., Maaser, K., Klein, C.E., Niggemann, B., Krohne, G., and Zänker, K.S. He, K., Xu, T., and Goldkorn, A. (2011). Cancer cells cyclically lose and regain
(1997). Migration of highly aggressive MV3 melanoma cells in 3-dimensional drug-resistant highly tumorigenic features characteristic of a cancer stem-like
collagen lattices results in local matrix reorganization and shedding of alpha2 phenotype. Mol. Cancer Ther. 10, 938–948.
and beta1 integrins and CD44. Cancer Res. 57, 2061–2070.
Hegerfeldt, Y., Tusch, M., Bröcker, E.B., and Friedl, P. (2002). Collective cell
Friedl, P., Noble, P.B., Walton, P.A., Laird, D.W., Chauvin, P.J., Tabah, R.J., movement in primary melanoma explants: plasticity of cell-cell interaction,
Black, M., and Zänker, K.S. (1995). Migration of coordinated cell clusters in beta1-integrin function, and migration strategies. Cancer Res. 62, 2125–2130.
mesenchymal and epithelial cancer explants in vitro. Cancer Res. 55, 4557–
Hidalgo-Carcedo, C., Hooper, S., Chaudhry, S.I., Williamson, P., Harrington,
4560.
K., Leitinger, B., and Sahai, E. (2011). Collective cell migration requires
Friedl, P., and Weigelin, B. (2008). Interstitial leukocyte migration and immune suppression of actomyosin at cell-cell contacts mediated by DDR1 and the
function. Nat. Immunol. 9, 960–969. cell polarity regulators Par3 and Par6. Nat. Cell Biol. 13, 49–58.
Friedl, P., and Wolf, K. (2003). Tumour-cell invasion and migration: diversity
Hodivala-Dilke, K.M., McHugh, K.P., Tsakiris, D.A., Rayburn, H., Crowley, D.,
and escape mechanisms. Nat. Rev. Cancer 3, 362–374.
Ullman-Culleré, M., Ross, F.P., Coller, B.S., Teitelbaum, S., and Hynes, R.O.
Friedl, P., and Wolf, K. (2008). Tube travel: the role of proteases in individual (1999). Beta3-integrin-deficient mice are a model for Glanzmann thrombas-
and collective cancer cell invasion. Cancer Res. 68, 7247–7249. thenia showing placental defects and reduced survival. J. Clin. Invest. 103,
Friedl, P., and Wolf, K. (2009). Proteolytic interstitial cell migration: a five-step 229–238.
process. Cancer Metastasis Rev. 28, 129–135. Honeth, G., Bendahl, P.O., Ringnér, M., Saal, L.H., Gruvberger-Saal, S.K.,
Friedl, P., and Wolf, K. (2010). Plasticity of cell migration: a multiscale tuning Lövgren, K., Grabau, D., Fernö, M., Borg, A., and Hegardt, C. (2008). The
model. J. Cell Biol. 188, 11–19. CD44+/CD24- phenotype is enriched in basal-like breast tumors. Breast
Gadea, G., de Toledo, M., Anguille, C., and Roux, P. (2007). Loss of p53 Cancer Res. 10, R53.
promotes RhoA-ROCK-dependent cell migration and invasion in 3D matrices. Hou, J.M., Krebs, M., Ward, T., Sloane, R., Priest, L., Hughes, A., Clack, G.,
J. Cell Biol. 178, 23–30. Ranson, M., Blackhall, F., and Dive, C. (2011). Circulating tumor cells as
Gaggioli, C., Hooper, S., Hidalgo-Carcedo, C., Grosse, R., Marshall, J.F., Har- a window on metastasis biology in lung cancer. Am. J. Pathol. 178, 989–996.
rington, K., and Sahai, E. (2007). Fibroblast-led collective invasion of carci- Hung, S.C., Wu, I.H., Hsue, S.S., Liao, C.H., Wang, H.C., Chuang, P.H., Sung,
noma cells with differing roles for RhoGTPases in leading and following cells. S.Y., and Hsieh, C.L. (2010). Targeting L1 cell adhesion molecule using lenti-
Nat. Cell Biol. 9, 1392–1400. virus-mediated short hairpin RNA interference reverses aggressiveness of
Gavert, N., Ben-Shmuel, A., Lemmon, V., Brabletz, T., and Ben-Ze’ev, A. oral squamous cell carcinoma. Mol. Pharm. 7, 2312–2323.
(2010). Nuclear factor-kappaB signaling and ezrin are essential for L1-medi- Hynes, R.O. (2002). Integrins: bidirectional, allosteric signaling machines. Cell
ated metastasis of colon cancer cells. J. Cell Sci. 123, 2135–2143. 110, 673–687.
Gavert, N., Sheffer, M., Raveh, S., Spaderna, S., Shtutman, M., Brabletz, T.,
Hynes, R.O. (2009). The extracellular matrix: not just pretty fibrils. Science 326,
Barany, F., Paty, P., Notterman, D., Domany, E., and Ben-Ze’ev, A. (2007).
1216–1219.
Expression of L1-CAM and ADAM10 in human colon cancer cells induces
metastasis. Cancer Res. 67, 7703–7712. Iguchi, T., Aishima, S., Taketomi, A., Nishihara, Y., Fujita, N., Sanefuji, K., Mae-
hara, Y., and Tsuneyoshi, M. (2008). Extracapsular penetration is a new prog-
Gavert, N., Vivanti, A., Hazin, J., Brabletz, T., and Ben-Ze’ev, A. (2011). L1-
nostic factor in human hepatocellular carcinoma. Am. J. Surg. Pathol. 32,
mediated colon cancer cell metastasis does not require changes in EMT and
1675–1682.
cancer stem cell markers. Mol. Cancer Res. 9, 14–24.
Ilina, O., Bakker, G.J., Vasaturo, A., Hofmann, R.M., and Friedl, P. (2011). Two-
Geiger, B., Spatz, J.P., and Bershadsky, A.D. (2009). Environmental sensing
photon laser-generated microtracks in 3D collagen lattices: principles of
through focal adhesions. Nat. Rev. Mol. Cell Biol. 10, 21–33.
MMP-dependent and -independent collective cancer cell invasion. Phys.
Geisbrecht, E.R., and Montell, D.J. (2002). Myosin VI is required for E-cad- Biol. 8, 015010.
herin-mediated border cell migration. Nat. Cell Biol. 4, 616–620.
Ilina, O., and Friedl, P. (2009). Mechanisms of collective cell migration at
Gérard, A., Mertens, A.E., van der Kammen, R.A., and Collard, J.G. (2007). The
a glance. J. Cell Sci. 122, 3203–3208.
Par polarity complex regulates Rap1- and chemokine-induced T cell polariza-
tion. J. Cell Biol. 176, 863–875. Ishizaki, M., Ashida, K., Higashi, T., Nakatsukasa, H., Kaneyoshi, T., Fujiwara,
K., Nouso, K., Kobayashi, Y., Uemura, M., Nakamura, S., and Tsuji, T. (2001).
Giampieri, S., Manning, C., Hooper, S., Jones, L., Hill, C.S., and Sahai, E.
The formation of capsule and septum in human hepatocellular carcinoma.
(2009). Localized and reversible TGFbeta signalling switches breast cancer
Virchows Arch. 438, 574–580.
cells from cohesive to single cell motility. Nat. Cell Biol. 11, 1287–1296.
Kardash, E., Reichman-Fried, M., Maitre, J.L., Boldajipour, B., Papusheva, E.,
Giese, A., Bjerkvig, R., Berens, M.E., and Westphal, M. (2003). Cost of migra-
Messerschmidt, E.M., Heisenberg, C.P., and Raz, E. (2010). A role for Rho
tion: invasion of malignant gliomas and implications for treatment. J. Clin.
GTPases and cell-cell adhesion in single-cell motility in vivo. Nat. Cell Biol.
Oncol. 21, 1624–1636.
12, 47–53.
Goetz, J.G., Minguet, S., Navarro-Lérida, I., Lazcano, J.J., Samaniego, R.,
Calvo, E., Tello, M., Osteso-Ibáñez, T., Pellinen, T., Echarri, A., et al. (2011). Kedrin, D., Gligorijevic, B., Wyckoff, J., Verkhusha, V.V., Condeelis, J., Segall,
Biomechanical remodeling of the microenvironment by stromal caveolin-1 J.E., and van Rheenen, J. (2008). Intravital imaging of metastatic behavior
favors tumor invasion and metastasis. Cell 146, 148–163. through a mammary imaging window. Nat. Methods 5, 1019–1021.

Grashoff, C., Hoffman, B.D., Brenner, M.D., Zhou, R., Parsons, M., Yang, M.T., Kenny, H.A., Kaur, S., Coussens, L.M., and Lengyel, E. (2008). The initial steps
McLean, M.A., Sligar, S.G., Chen, C.S., Ha, T., and Schwartz, M.A. (2010). of ovarian cancer cell metastasis are mediated by MMP-2 cleavage of vitro-
Measuring mechanical tension across vinculin reveals regulation of focal nectin and fibronectin. J. Clin. Invest. 118, 1367–1379.
adhesion dynamics. Nature 466, 263–266. Khalil, A.A., and Friedl, P. (2010). Determinants of leader cells in collective cell
Grytsenko, P., Ilina, O., and Friedl, P. (2011). Interstitial guidance of cancer migration. Integr. Biol. (Camb) 2, 568–574.
invasion. J. Pathol. Published online October 18, 2011. 10.1002/path.3031. Kiefel, H., Pfeifer, M., Bondong, S., Hazin, J., and Altevogt, P. (2011). Linking
Hanahan, D., and Weinberg, R.A. (2011). Hallmarks of cancer: the next gener- L1CAM-mediated signaling to NF-kB activation. Trends Mol. Med. 17,
ation. Cell 144, 646–674. 178–187.

1006 Cell 147, November 23, 2011 ª2011 Elsevier Inc.

Kienast, Y., von Baumgarten, L., Fuhrmann, M., Klinkert, W.E., Goldbrunner, Martı́n-Villar, E., Megı́as, D., Castel, S., Yurrita, M.M., Vilaró, S., and Quinta-
R., Herms, J., and Winkler, F. (2010). Real-time imaging reveals the single nilla, M. (2006). Podoplanin binds ERM proteins to activate RhoA and promote
steps of brain metastasis formation. Nat. Med. 16, 116–122. epithelial-mesenchymal transition. J. Cell Sci. 119, 4541–4553.
Kitt, K.N., and Nelson, W.J. (2011). Rapid suppression of activated Rac1 by Mason, S.D., and Joyce, J.A. (2011). Proteolytic networks in cancer. Trends
cadherins and nectins during de novo cell-cell adhesion. PLoS ONE 6, e17841. Cell Biol. 21, 228–237.
Kulesa, P.M., and Gammill, L.S. (2010). Neural crest migration: patterns, Massagué, J. (2008). TGFbeta in Cancer. Cell 134, 215–230.
phases and signals. Dev. Biol. 344, 566–568. McAllister, S.S., Gifford, A.M., Greiner, A.L., Kelleher, S.P., Saelzler, M.P.,
Lämmermann, T., Bader, B.L., Monkley, S.J., Worbs, T., Wedlich-Söldner, R., Ince, T.A., Reinhardt, F., Harris, L.N., Hylander, B.L., Repasky, E.A., and Wein-
Hirsch, K., Keller, M., Förster, R., Critchley, D.R., Fässler, R., and Sixt, M. berg, R.A. (2008). Systemic endocrine instigation of indolent tumor growth
(2008). Rapid leukocyte migration by integrin-independent flowing and requires osteopontin. Cell 133, 994–1005.
squeezing. Nature 453, 51–55. Mertens, A.E., Rygiel, T.P., Olivo, C., van der Kammen, R., and Collard, J.G.
Lämmermann, T., and Sixt, M. (2009). Mechanical modes of ‘amoeboid’ cell (2005). The Rac activator Tiam1 controls tight junction biogenesis in keratino-
migration. Curr. Opin. Cell Biol. 21, 636–644. cytes through binding to and activation of the Par polarity complex. J. Cell Biol.
170, 1029–1037.
Lauffenburger, D.A., and Horwitz, A.F. (1996). Cell migration: a physically inte-
grated molecular process. Cell 84, 359–369. Mu, D., Cambier, S., Fjellbirkeland, L., Baron, J.L., Munger, J.S., Kawakatsu,
H., Sheppard, D., Broaddus, V.C., and Nishimura, S.L. (2002). The integrin
Lee, H.S., Lim, C.J., Puzon-McLaughlin, W., Shattil, S.J., and Ginsberg, M.H.
alpha(v)beta8 mediates epithelial homeostasis through MT1-MMP-dependent
(2009). RIAM activates integrins by linking talin to ras GTPase membrane-tar-
activation of TGF-beta1. J. Cell Biol. 157, 493–507.
geting sequences. J. Biol. Chem. 284, 5119–5127.
Muller, P.A., Caswell, P.T., Doyle, B., Iwanicki, M.P., Tan, E.H., Karim, S., Lu-
Levental, K.R., Yu, H., Kass, L., Lakins, J.N., Egeblad, M., Erler, J.T., Fong, kashchuk, N., Gillespie, D.A., Ludwig, R.L., Gosselin, P., et al. (2009). Mutant
S.F., Csiszar, K., Giaccia, A., Weninger, W., et al. (2009). Matrix crosslinking p53 drives invasion by promoting integrin recycling. Cell 139, 1327–1341.
forces tumor progression by enhancing integrin signaling. Cell 139, 891–906.
Nannuru, K.C., Futakuchi, M., Varney, M.L., Vincent, T.M., Marcusson, E.G.,
Li, Z., Zhou, Z., and Donahue, H.J. (2008). Alterations in Cx43 and OB-cadherin and Singh, R.K. (2010). Matrix metalloproteinase (MMP)-13 regulates
affect breast cancer cell metastatic potential. Clin. Exp. Metastasis 25, mammary tumor-induced osteolysis by activating MMP9 and transforming
265–272. growth factor-beta signaling at the tumor-bone interface. Cancer Res. 70,
Lo, C.M., Wang, H.B., Dembo, M., and Wang, Y.L. (2000). Cell movement is 3494–3504.
guided by the rigidity of the substrate. Biophys. J. 79, 144–152. Nannuru, K.C., and Singh, R.K. (2010). Tumor-stromal interactions in bone
Lopes, F.F., da Costa Miguel, M.C., Pereira, A.L., da Cruz, M.C., de Almeida metastasis. Curr. Osteoporos. Rep. 8, 105–113.
Freitas, R., Pinto, L.P., and de Souza, L.B. (2009). Changes in immunoexpres- Nelson, C.M., Khauv, D., Bissell, M.J., and Radisky, D.C. (2008). Change in
sion of E-cadherin and beta-catenin in oral squamous cell carcinoma with and cell shape is required for matrix metalloproteinase-induced epithelial-
without nodal metastasis. Ann. Diagn. Pathol. 13, 22–29. mesenchymal transition of mammary epithelial cells. J. Cell. Biochem. 105,
Lorentzen, A., Bamber, J., Sadok, A., Elson-Schwab, I., and Marshall, C.J. 25–33.
(2011). An ezrin-rich, rigid uropod-like structure directs movement of amoe- Netelenbos, T., Zuijderduijn, S., Van Den Born, J., Kessler, F.L., Zweegman, S.,
boid blebbing cells. J. Cell Sci. 124, 1256–1267. Huijgens, P.C., and Dräger, A.M. (2002). Proteoglycans guide SDF-1-
Lortat-Jacob, H., Grosdidier, A., and Imberty, A. (2002). Structural diversity of induced migration of hematopoietic progenitor cells. J. Leukoc. Biol. 72,
heparan sulfate binding domains in chemokines. Proc. Natl. Acad. Sci. USA 353–362.
99, 1229–1234. Neuhaus, B., Bühren, S., Böck, B., Alves, F., Vogel, W.F., and Kiefer, F. (2011).
Lyons, T.R., O’Brien, J., Borges, V.F., Conklin, M.W., Keely, P.J., Eliceiri, K.W., Migration inhibition of mammary epithelial cells by Syk is blocked in the pres-
Marusyk, A., Tan, A.C., and Schedin, P. (2011). Postpartum mammary gland ence of DDR1 receptors. Cell. Mol. Life Sci. 68, 3757–3770.
involution drives progression of ductal carcinoma in situ through collagen Nimnual, A.S., Taylor, L.J., and Bar-Sagi, D. (2003). Redox-dependent down-
and COX-2. Nat. Med. 17, 1109–1115. regulation of Rho by Rac. Nat. Cell Biol. 5, 236–241.
Maaser, K., Wolf, K., Klein, C.E., Niggemann, B., Zänker, K.S., Bröcker, E.B., Orimo, A., Gupta, P.B., Sgroi, D.C., Arenzana-Seisdedos, F., Delaunay, T.,
and Friedl, P. (1999). Functional hierarchy of simultaneously expressed adhe- Naeem, R., Carey, V.J., Richardson, A.L., and Weinberg, R.A. (2005). Stromal
sion receptors: integrin alpha2beta1 but not CD44 mediates MV3 melanoma fibroblasts present in invasive human breast carcinomas promote tumor
cell migration and matrix reorganization within three-dimensional hyalur- growth and angiogenesis through elevated SDF-1/CXCL12 secretion. Cell
onan-containing collagen matrices. Mol. Biol. Cell 10, 3067–3079. 121, 335–348.
Madsen, C.D., and Sahai, E. (2010). Cancer dissemination–lessons from Overall, C.M., and Blobel, C.P. (2007). In search of partners: linking extracel-
leukocytes. Dev. Cell 19, 13–26. lular proteases to substrates. Nat. Rev. Mol. Cell Biol. 8, 245–257.
Makarenkova, H.P., Hoffman, M.P., Beenken, A., Eliseenkova, A.V., Meech, Padua, D., and Massagué, J. (2009). Roles of TGFbeta in metastasis. Cell Res.
R., Tsau, C., Patel, V.N., Lang, R.A., and Mohammadi, M. (2009). Differential 19, 89–102.
interactions of FGFs with heparan sulfate control gradient formation and Pàez-Ribes, M., Allen, E., Hudock, J., Takeda, T., Okuyama, H., Viñals, F.,
branching morphogenesis. Sci. Signal. 2, ra55. Inoue, M., Bergers, G., Hanahan, D., and Casanovas, O. (2009). Antiangio-
Maness, P.F., and Schachner, M. (2007). Neural recognition molecules of the genic therapy elicits malignant progression of tumors to increased local inva-
immunoglobulin superfamily: signaling transducers of axon guidance and sion and distant metastasis. Cancer Cell 15, 220–231.
neuronal migration. Nat. Neurosci. 10, 19–26. Paluch, E., Sykes, C., Prost, J., and Bornens, M. (2006). Dynamic modes of the
Mani, S.A., Guo, W., Liao, M.J., Eaton, E.N., Ayyanan, A., Zhou, A.Y., Brooks, cortical actomyosin gel during cell locomotion and division. Trends Cell Biol.
M., Reinhard, F., Zhang, C.C., Shipitsin, M., et al. (2008). The epithelial-mesen- 16, 5–10.
chymal transition generates cells with properties of stem cells. Cell 133, Parri, M., Taddei, M.L., Bianchini, F., Calorini, L., and Chiarugi, P. (2009).
704–715. EphA2 reexpression prompts invasion of melanoma cells shifting from mesen-
Martinez-Rico, C., Pincet, F., Thiery, J.P., and Dufour, S. (2010). Integrins stim- chymal to amoeboid-like motility style. Cancer Res. 69, 2072–2081.
ulate E-cadherin-mediated intercellular adhesion by regulating Src-kinase Paszek, M.J., Zahir, N., Johnson, K.R., Lakins, J.N., Rozenberg, G.I., Gefen,
activation and actomyosin contractility. J. Cell Sci. 123, 712–722. A., Reinhart-King, C.A., Margulies, S.S., Dembo, M., Boettiger, D., et al.

Cell 147, November 23, 2011 ª2011 Elsevier Inc. 1007

(2005). Tensional homeostasis and the malignant phenotype. Cancer Cell 8, Sawada, Y., Tamada, M., Dubin-Thaler, B.J., Cherniavskaya, O., Sakai, R., Ta-
241–254. naka, S., and Sheetz, M.P. (2006). Force sensing by mechanical extension of
Pennacchietti, S., Michieli, P., Galluzzo, M., Mazzone, M., Giordano, S., and the Src family kinase substrate p130Cas. Cell 127, 1015–1026.
Comoglio, P.M. (2003). Hypoxia promotes invasive growth by transcriptional Schedin, P., and Keely, P.J. (2011). Mammary gland ECM remodeling, stiff-
activation of the met protooncogene. Cancer Cell 3, 347–361. ness, and mechanosignaling in normal development and tumor progression.
Pittet, M.J., and Weissleder, R. (2011). Intravital imaging. Cell 147, this issue, Cold Spring Harb. Perspect. Biol. 3, a003228.
983–991. Schlegelmilch, K., Mohseni, M., Kirak, O., Pruszak, J., Rodriguez, J.R., Zhou,
Piao, Y., Lu, L., and de Groot, J. (2009). AMPA receptors promote perivascular D., Kreger, B.T., Vasioukhin, V., Avruch, J., Brummelkamp, T.R., and Ca-
glioma invasion via beta1 integrin-dependent adhesion to the extracellular margo, F.D. (2011). Yap1 acts downstream of a-catenin to control epidermal
matrix. Neuro-oncol. 11, 260–273. proliferation. Cell 144, 782–795.

Picchio, M., Beck, R., Haubner, R., Seidl, S., Machulla, H.J., Johnson, T.D., Shapiro, I.M., Cheng, A.W., Flytzanis, N.C., Balsamo, M., Condeelis, J.S., Ok-
Wester, H.J., Reischl, G., Schwaiger, M., and Piert, M. (2008). Intratumoral tay, M.H., Burge, C.B., and Gertler, F.B. (2011). An EMT-driven alternative
spatial distribution of hypoxia and angiogenesis assessed by 18F-FAZA and splicing program occurs in human breast cancer and modulates cellular
125I-Gluco-RGD autoradiography. J. Nucl. Med. 49, 597–605. phenotype. PLoS Genet. 7, e1002218.

Poincloux, R., Collin, O., Lizárraga, F., Romao, M., Debray, M., Piel, M., and Sheetz, M.P., Felsenfeld, D., Galbraith, C.G., and Choquet, D. (1999). Cell
Chavrier, P. (2011). Contractility of the cell rear drives invasion of breast tumor migration as a five-step cycle. Biochem. Soc. Symp. 65, 233–243.
cells in 3D Matrigel. Proc. Natl. Acad. Sci. USA 108, 1943–1948. Shiao, S.L., and Coussens, L.M. (2010). The tumor-immune microenvironment
and response to radiation therapy. J. Mammary Gland Biol. Neoplasia 15,
Polyak, K., Haviv, I., and Campbell, I.G. (2009). Co-evolution of tumor cells and
411–421.
their microenvironment. Trends Genet. 25, 30–38.
Shintani, Y., Fukumoto, Y., Chaika, N., Svoboda, R., Wheelock, M.J., and
Radisky, D.C., Levy, D.D., Littlepage, L.E., Liu, H., Nelson, C.M., Fata, J.E.,
Johnson, K.R. (2008). Collagen I-mediated up-regulation of N-cadherin
Leake, D., Godden, E.L., Albertson, D.G., Nieto, M.A., et al. (2005). Rac1b
requires cooperative signals from integrins and discoidin domain receptor 1.
and reactive oxygen species mediate MMP-3-induced EMT and genomic
J. Cell Biol. 180, 1277–1289.
instability. Nature 436, 123–127.
Shtutman, M., Levina, E., Ohouo, P., Baig, M., and Roninson, I.B. (2006). Cell
Reynolds, A.B. (2010). Exposing p120 catenin’s most intimate affair. Cell 141,
adhesion molecule L1 disrupts E-cadherin-containing adherens junctions and
20–22.
increases scattering and motility of MCF7 breast carcinoma cells. Cancer Res.
Ridley, A.J., Schwartz, M.A., Burridge, K., Firtel, R.A., Ginsberg, M.H., Borisy, 66, 11370–11380.
G., Parsons, J.T., and Horwitz, A.R. (2003). Cell migration: integrating signals
Sounni, N.E., Dehne, K., van Kempen, L., Egeblad, M., Affara, N.I., Cuevas, I.,
from front to back. Science 302, 1704–1709.
Wiesen, J., Junankar, S., Korets, L., Lee, J., et al. (2010). Stromal regulation
Rizki, A., Weaver, V.M., Lee, S.Y., Rozenberg, G.I., Chin, K., Myers, C.A., Bas- of vessel stability by MMP14 and TGFbeta. Dis. Model. Mech. 3, 317–332.
com, J.L., Mott, J.D., Semeiks, J.R., Grate, L.R., et al. (2008). A human breast
Spaderna, S., Schmalhofer, O., Hlubek, F., Berx, G., Eger, A., Merkel, S., Jung,
cell model of preinvasive to invasive transition. Cancer Res. 68, 1378–1387.
A., Kirchner, T., and Brabletz, T. (2006). A transient, EMT-linked loss of base-
Roussos, E.T., Condeelis, J.S., and Patsialou, A. (2011). Chemotaxis in cancer. ment membranes indicates metastasis and poor survival in colorectal cancer.
Nat. Rev. Cancer 11, 573–587. Gastroenterology 131, 830–840.
Ruiz, P.A., and Jarai, G. (2011). Collagen I induces discoidin domain receptor Spaderna, S., Schmalhofer, O., Wahlbuhl, M., Dimmler, A., Bauer, K., Sultan,
(DDR) 1 expression through DDR2 and a JAK2-ERK1/2-mediated mechanism A., Hlubek, F., Jung, A., Strand, D., Eger, A., et al. (2008). The transcriptional
in primary human lung fibroblasts. J. Biol. Chem. 286, 12912–12923. repressor ZEB1 promotes metastasis and loss of cell polarity in cancer.
Ryschich, E., Khamidjanov, A., Kerkadze, V., Büchler, M.W., Zöller, M., and Cancer Res. 68, 537–544.
Schmidt, J. (2009). Promotion of tumor cell migration by extracellular matrix Stoecklein, N.H., Hosch, S.B., Bezler, M., Stern, F., Hartmann, C.H., Vay, C.,
proteins in human pancreatic cancer. Pancreas 38, 804–810. Siegmund, A., Scheunemann, P., Schurr, P., Knoefel, W.T., et al. (2008). Direct
Sabeh, F., Li, X.Y., Saunders, T.L., Rowe, R.G., and Weiss, S.J. (2009). genetic analysis of single disseminated cancer cells for prediction of outcome
Secreted versus membrane-anchored collagenases: relative roles in fibro- and therapy selection in esophageal cancer. Cancer Cell 13, 441–453.
blast-dependent collagenolysis and invasion. J. Biol. Chem. 284, 23001– Stoletov, K., Kato, H., Zardouzian, E., Kelber, J., Yang, J., Shattil, S., and
23011. Klemke, R. (2010). Visualizing extravasation dynamics of metastatic tumor
Sabeh, F., Ota, I., Holmbeck, K., Birkedal-Hansen, H., Soloway, P., Balbin, M., cells. J. Cell Sci. 123, 2332–2341.
Lopez-Otin, C., Shapiro, S., Inada, M., Krane, S., et al. (2004). Tumor cell traffic Tambe, D.T., Hardin, C.C., Angelini, T.E., Rajendran, K., Park, C.Y., Serra-Pic-
through the extracellular matrix is controlled by the membrane-anchored amal, X., Zhou, E.H., Zaman, M.H., Butler, J.P., Weitz, D.A., et al. (2011).
collagenase MT1-MMP. J. Cell Biol. 167, 769–781. Collective cell guidance by cooperative intercellular forces. Nat. Mater. 10,
Sahai, E. (2007). Illuminating the metastatic process. Nat. Rev. Cancer 7, 469–475.
737–749. Theocharis, A.D., Skandalis, S.S., Tzanakakis, G.N., and Karamanos, N.K.
Sahai, E., and Marshall, C.J. (2003). Differing modes of tumour cell invasion (2010). Proteoglycans in health and disease: novel roles for proteoglycans in
have distinct requirements for Rho/ROCK signalling and extracellular proteol- malignancy and their pharmacological targeting. FEBS J. 277, 3904–3923.
ysis. Nat. Cell Biol. 5, 711–719. Thiery, J.P., Acloque, H., Huang, R.Y., and Nieto, M.A. (2009). Epithelial-
Sansing, H.A., Sarkeshik, A., Yates, J.R., Patel, V., Gutkind, J.S., Yamada, mesenchymal transitions in development and disease. Cell 139, 871–890.
K.M., and Berrier, A.L. (2011). Integrin ab1, avb, a6b effectors p130Cas, Src Trusolino, L., Bertotti, A., and Comoglio, P.M. (2010). MET signalling: princi-
and talin regulate carcinoma invasion and chemoresistance. Biochem. Bio- ples and functions in development, organ regeneration and cancer. Nat.
phys. Res. Commun. 406, 171–176. Rev. Mol. Cell Biol. 11, 834–848.
Sanz-Moreno, V., Gadea, G., Ahn, J., Paterson, H., Marra, P., Pinner, S., Sahai, Tsuji, K., Yamauchi, K., Yang, M., Jiang, P., Bouvet, M., Endo, H., Kanai, Y.,
E., and Marshall, C.J. (2008). Rac activation and inactivation control plasticity Yamashita, K., Moossa, A.R., and Hoffman, R.M. (2006). Dual-color imaging
of tumor cell movement. Cell 135, 510–523. of nuclear-cytoplasmic dynamics, viability, and proliferation of cancer cells
Sanz-Moreno, V., and Marshall, C.J. (2010). The plasticity of cytoskeletal in the portal vein area. Cancer Res. 66, 303–306.
dynamics underlying neoplastic cell migration. Curr. Opin. Cell Biol. 22, Tybulewicz, V.L., and Henderson, R.B. (2009). Rho family GTPases and their
690–696. regulators in lymphocytes. Nat. Rev. Immunol. 9, 630–644.

1008 Cell 147, November 23, 2011 ª2011 Elsevier Inc.

van den Brand, M., Takes, R.P., Blokpoel-deRuyter, M., Slootweg, P.J., and Wolf, K., Müller, R., Borgmann, S., Bröcker, E.B., and Friedl, P. (2003b). Amoe-
van Kempen, L.C. (2010). Activated leukocyte cell adhesion molecule expres- boid shape change and contact guidance: T-lymphocyte crawling through
sion predicts lymph node metastasis in oral squamous cell carcinoma. Oral fibrillar collagen is independent of matrix remodeling by MMPs and other
Oncol. 46, 393–398. proteases. Blood 102, 3262–3269.
Vogel, W.F., Abdulhussein, R., and Ford, C.E. (2006). Sensing extracellular Wolf, K., Wu, Y.I., Liu, Y., Geiger, J., Tam, E., Overall, C., Stack, M.S., and
matrix: an update on discoidin domain receptor function. Cell. Signal. 18, Friedl, P. (2007). Multi-step pericellular proteolysis controls the transition
1108–1116. from individual to collective cancer cell invasion. Nat. Cell Biol. 9, 893–904.
Walch, A., Seidl, S., Hermannstädter, C., Rauser, S., Deplazes, J., Langer, R., Wyckoff, J., Wang, W., Lin, E.Y., Wang, Y., Pixley, F., Stanley, E.R., Graf, T.,
von Weyhern, C.H., Sarbia, M., Busch, R., Feith, M., et al. (2008). Combined Pollard, J.W., Segall, J., and Condeelis, J. (2004). A paracrine loop between
analysis of Rac1, IQGAP1, Tiam1 and E-cadherin expression in gastric cancer. tumor cells and macrophages is required for tumor cell migration in mammary
Mod. Pathol. 21, 544–552. tumors. Cancer Res. 64, 7022–7029.
Wang, W., Goswami, S., Lapidus, K., Wells, A.L., Wyckoff, J.B., Sahai, E.,
Xu, R., Boudreau, A., and Bissell, M.J. (2009). Tissue architecture and function:
Singer, R.H., Segall, J.E., and Condeelis, J.S. (2004). Identification and testing
dynamic reciprocity via extra- and intra-cellular matrices. Cancer Metastasis
of a gene expression signature of invasive carcinoma cells within primary
Rev. 28, 167–176.
mammary tumors. Cancer Res. 64, 8585–8594.
Wang, X., Zhang, J., Fan, M., Zhou, Q., Deng, H., Aisharif, M.J., and Chen, X. Yang, J., Mani, S.A., Donaher, J.L., Ramaswamy, S., Itzykson, R.A., Come, C.,
(2009). The expression of E-cadherin at the invasive tumor front of oral squa- Savagner, P., Gitelman, I., Richardson, A., and Weinberg, R.A. (2004). Twist,
mous cell carcinoma: immunohistochemical and RT-PCR analysis with clini- a master regulator of morphogenesis, plays an essential role in tumor metas-
copathological correlation. Oral Surg. Oral Med. Oral Pathol. Oral Radiol. tasis. Cell 117, 927–939.
Endod. 107, 547–554. Yano, H., Mazaki, Y., Kurokawa, K., Hanks, S.K., Matsuda, M., and Sabe, H.
Wicki, A., Lehembre, F., Wick, N., Hantusch, B., Kerjaschki, D., and Christofori, (2004). Roles played by a subset of integrin signaling molecules in cadherin-
G. (2006). Tumor invasion in the absence of epithelial-mesenchymal transition: based cell-cell adhesion. J. Cell Biol. 166, 283–295.
podoplanin-mediated remodeling of the actin cytoskeleton. Cancer Cell 9, Yao, H., Zeng, Z.Z., Fay, K.S., Veine, D.M., Staszewski, E.D., Morgan, M.,
261–272. Wilder-Romans, K., Williams, T.M., Spalding, A.C., Ben-Josef, E., and Livant,
Wipff, P.J., Rifkin, D.B., Meister, J.J., and Hinz, B. (2007). Myofibroblast D.L. (2011). Role of a(5)b(1) Integrin Up-regulation in Radiation-Induced Inva-
contraction activates latent TGF-beta1 from the extracellular matrix. J. Cell sion by Human Pancreatic Cancer Cells. Transl. Oncol. 4, 282–292.
Biol. 179, 1311–1323.
Zaidel-Bar, R., Itzkovitz, S., Ma’ayan, A., Iyengar, R., and Geiger, B. (2007).
Wolf, K., Alexander, S., Schacht, V., Coussens, L.M., von Andrian, U.H., van Functional atlas of the integrin adhesome. Nat. Cell Biol. 9, 858–867.
Rheenen, J., Deryugina, E., and Friedl, P. (2009). Collagen-based cell migra-
tion models in vitro and in vivo. Semin. Cell Dev. Biol. 20, 931–941. Zecchini, S., Bombardelli, L., Decio, A., Bianchi, M., Mazzarol, G., Sanguineti,
F., Aletti, G., Maddaluno, L., Berezin, V., Bock, E., et al. (2011). The adhesion
Wolf, K., and Friedl, P. (2011). Extracellular matrix determinants of proteolytic
molecule NCAM promotes ovarian cancer progression via FGFR signalling.
and non-proteolytic cell migration. Trends Cell Biol. Published online October
EMBO Mol. Med. 3, 480–494.
27, 2011. in press. 10.1016/j.tcb.2011.09.006.
Zlotnik, A., Burkhardt, A.M., and Homey, B. (2011). Homeostatic chemokine
Wolf, K., Mazo, I., Leung, H., Engelke, K., von Andrian, U.H., Deryugina, E.I.,
receptors and organ-specific metastasis. Nat. Rev. Immunol. 11, 597–606.
Strongin, A.Y., Bröcker, E.B., and Friedl, P. (2003a). Compensation mecha-
nism in tumor cell migration: mesenchymal-amoeboid transition after blocking Zöller, M. (2011). CD44: can a cancer-initiating cell profit from an abundantly
of pericellular proteolysis. J. Cell Biol. 160, 267–277. expressed molecule? Nat. Rev. Cancer 11, 254–267.

Cell 147, November 23, 2011 ª2011 Elsevier Inc. 1009