CXCR4/CXCL12 Participate in Extravasation of

Metastasizing Breast Cancer Cells within the Liver in a

Rat Model
Claudia Wendel1., André Hemping-Bovenkerk1., Julia Krasnyanska1, Sören Torge Mees1, Marina
Kochetkova2, Sandra Stoeppeler1, Jörg Haier3*
1 Department of General and Visceral Surgery, University Hospital Muenster, Muenster, Germany, 2 Chemokine Biology Division, School of Molecular and Biomedical
Science, University of Adelaide, Adelaide, Australia, 3 Comprehensive Cancer Center Muenster, University Hospital Muenster, Muenster, Germany

Abstract
Introduction: Organ-specific composition of extracellular matrix proteins (ECM) is a determinant of metastatic host organ
involvement. The chemokine CXCL12 and its receptor CXCR4 play important roles in the colonization of human breast
cancer cells to their metastatic target organs. In this study, we investigated the effects of chemokine stimulation on
adhesion and migration of different human breast cancer cell lines in vivo and in vitro with particular focus on the liver as a
major metastatic site in breast cancer.

Methods: Time lapse microscopy, in vitro adhesion and migration assays were performed under CXCL12 stimulation.
Activation of small GTPases showed chemokine receptor signalling dependence from ECM components. The initial events of
hepatic colonisation of MDA-MB-231 and MDA-MB-468 cells were investigated by intravital microscopy of the liver in a rat
model and under shRNA inhibition of CXCR4.

Results: In vitro, stimulation with CXCL12 induced increased chemotactic cell motility (p,0.05). This effect was dependent
on adhesive substrates (type I collagen, fibronectin and laminin) and induced different responses in small GTPases, such as
RhoA and Rac-1 activation, and changes in cell morphology. In addition, binding to various ECM components caused
redistribution of chemokine receptors at tumour cell surfaces. In vivo, blocking CXCR4 decreased extravasation of highly
metastatic MDA-MB-231 cells (p,0.05), but initial cell adhesion within the liver sinusoids was not affected. In contrast, the
less metastatic MDA-MB-468 cells showed reduced cell adhesion but similar migration within the hepatic microcirculation.
Conclusion: Chemokine-induced extravasation of breast cancer cells along specific ECM components appears to be an
important regulator but not a rate-limiting factor of their metastatic organ colonization.

Citation: Wendel C, Hemping-Bovenkerk A, Krasnyanska J, Mees ST, Kochetkova M, et al. (2012) CXCR4/CXCL12 Participate in Extravasation of Metastasizing
Breast Cancer Cells within the Liver in a Rat Model. PLoS ONE 7(1): e30046. doi:10.1371/journal.pone.0030046
Editor: Syed A. Aziz, Health Canada, Canada
Received August 16, 2011; Accepted December 8, 2011; Published January 13, 2012
Copyright: ß 2012 Wendel et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Funded by the Deutsche Krebshilfe (107010) to JH. The funders had no role in study design, data collection and analysis, decision to publish, or
preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: [email protected]
. These authors contributed equally to this work.

Introduction endothelial cells and their underlying ECM represents an initial

event of metastatic organ colonisation alongside extravasation into
Metastasis is the result of multiple sequential steps and is a the host organ parenchyma [8].
highly organized, non-random, and organ-selective process [1]. Many of these characteristics for metastasis formation are
Tumour cell interactions with endothelium and subendothelial
related to tumour cell adhesion and migration with haptotactic
extracellular matrix (ECM) constitute crucial factors in determin-
guidance. Chemotactic molecules, such as chemokines and their
ing the organ preference of metastasis. The interplay between
receptors, were also shown to play an important role in organ-
malignant tumour cells and their surrounding ECM has been
implicated at nearly every stage of the metastatic process; ranging specific colonization of metastatic tumour cells [9,10,11]. Physi-
from steps that involve the local invasion of tumour cells away ologically, chemokines are active on neutrophils and T-lympho-
from the primary tumour to those that are involved in mediating cytes (-CXC- type), while –CC- type chemokines are active on
extravasation through microvessel-associated basement mem- monocytes and lymphocytes, predominantly mediating stimulation
branes at the site(s) of metastasis formation [2]. Initial arrest and of leukocyte chemotaxis during inflammation [9]. Tumour cell
attachment of circulating tumour cells in the secondary organs are migration and metastasis appear to share many similarities with
believed to be crucial events for haematogenous metastasis, but the leukocyte trafficking. Müller et al. [12] reported that tumour cells
actual processes in in vivo conditions remain a matter of debate express a distinct pattern of functionally active chemokine
[3,4,5,6,7]. Adhesion of circulating tumour cells to microvascular receptors which correlates with their metastatic behaviour.

PLoS ONE | www.plosone.org 1 January 2012 | Volume 7 | Issue 1 | e30046

 CXCL12-Induced Tumour Cell Extravasation

Breast cancer is an example for a tumour with an organ-specific Materials and Methods
pattern of distant metastasis formation. It mainly colonizes lung,
liver, lymph nodes and bone marrow, all of which are abundant Reagents
sources of chemokine ligands [12,13]. Overexpression of chemo- Phycoerythrin (PE)-conjugated anti-human CXCR4 antibodies
kines - especially of CXCR4 and CCR7 - was observed in breast were obtained from R&D Systems (Wiesbaden, Germany).
cancer cells and surgical specimens, but chemokine receptors are Unlabelled, function-blocking anti-human CXCR4 antibody was
also highly expressed in other tumour types including cancers of a kind gift from A. Müller, Düsseldorf, Germany. Anti-human
epithelial, mesenchymal and hematopoietic origin [14]. The role RhoA antibody was purchased from Santa Cruz (Santa Cruz, CA,
of CXCR4 in the metastatic cascade of breast cancer and also its USA), anti-human Rac1 from BD Pharmingen (San Jose, CA,
ability to predictpatient survival have been intensively studied USA) and anti-human Cdc42 antibody was obtained from Cell
[15]. Several groups found that CXCR4 and its ligand Signaling (Danvers, MA, USA). For integrin subunits the following
CXCL12 can promote tumour cell migration and invasion antibodies were used: b1, b4, a2, a3 (all from Chemicon,
[9,12,16,17,18,19,20]. For example, CXCL12/SDF-1a induced Hofheim, Germany), a1 (upstate biotechnology, Hamburg,
cellular responses, such as calcium mobilization, actin polymeri- Germany), a5 (Serotec, Eching, Germany) and a6 (gift from J.
zation, and chemotaxis in metastatic cells, whereas non-invasive Eble, Münster, Germany). Alexa Fluor 488 and 546 labelled
cells were unresponsive [21]. In addition, CXCL12 activated secondary antibodies, phalloidin, CalceinAM and Hoechst 33342
multiple signalling pathways downstream of G-proteins in highly were purchased from Molecular Probes/Invitrogen (Karlsruhe,
invasive cells but failed to activate downstream kinase cascades in Germany). Human recombinant chemokine SDF1a/CXCL12
non-invasive cell lines [21]. Since chemotactic tumour cell was obtained from R&D systems (Wiesbaden, Germany).
characteristics are related to cellular interactions with ECM Glutathione-Agarose Rac/Cdc41 PAK-1 PBD beads and Rho
components, the composition of these matrix proteins appears to Rhotekin RBD beads were obtained from Upstate/Millipore (Eching,
Germany). ECM components C I, FN and LN were purchased from
be relevant to the metastatic process [22]. AsECM composition
Sigma-Aldrich (Saint Louis, Missouri, USA). All other chemicals were
differs between organs and tissue types, chemokine activity in
purchased from Sigma or Roth (Karlsruhe, Germany).
breast cancer cells may be dependent on ECMavailability, as is
true for haematogenous cells. In T-lymphocytes, for example,
chemokines presented within a collagen matrix increased depth of Cell lines and culture conditions
migration of infiltrating cells in vitro whereas the presence of Subclones of the MDA-MB-231 cells (originally obtained from
fibronectin within the collagen substrate modulated their adhesive the ATCC) were provided by M. Kochetkowa, Adelaide,
and migratory properties [23]. Australia. These cells were originally derived from a 51 years
Recently, we reported that the hepatic subendothelial space old Caucasian female and are invasive and tumorigenic in nude
of Dissé is constituted of fibronectin (FN), von Willebrandt- mice. MDA-MB-468 cell line was purchased from ATCC
(Manassas, VA) and was originally derived from a pleural effusion
factor, type I (C I) and type IV collagen (C IV) and small
of a 51-year old black female patient with metastatic adenocar-
amounts of laminin-5 (LN). Using different types of inhibitors,
cinoma of the breast. MDA-MB-231 cell line was maintained in
we found that initial adhesion of circulating tumour cells within
RPMI 1640 or DMEM medium (Gibco/Invitrogen, Karlsruhe,
the liver sinusoids was mainly mediated by RGD-dependent
Germany) containing 10% foetal bovine serum (FBS, Gibco)
integrin binding to FN. In contrast, binding to C I or C IV was
without antibiotics. Two clones with shRNA-mediated CXCR4
not involved in initial cell adhesion but C I enabled tumour cell
reduction were also provided by M. Kochetkowa. These clones
extravasation into the liver parenchyma via a2b1-integrins. LN
(MDA-MB-231-19 and MDA-MB-231-27) were obtained using
was not involved in any of these steps [24]. Taken together, CXCR4 shRNA-expressing constructs 5-gatctGGTGGTCTAT-
these results suggest that hepatic colonization is brought about GTTGGCGTCTGttcaagaGACAGACG CCAACATAGACCA-
by organ-specific tumour cell migration and chemotactic stimuli CCtttttta-3 and 5-agcttaaaaaaGGTGGTCTATGTTGGCGTC-
within the liver. Downstream signalling of both integrins and TG tctcttgaacagacgccaacatagaccacca-3 (21-nucleotide CXCR4 at
chemokine receptors involves small Rho-GTPases. Correspond- position 470–490 of human CXCR4 cDNA). [21,27] MDA-MB-
ing to the increased cell motility of cancer cells, it was found that 468 cells were cultured in DMEM medium (Lonza, Verviers,
RhoA, Rac1 and Cdc42 are dramatically overexpressed in Belgium) containing 10% FBS and L-glutamine (Gibco/Invitro-
breast cancer, but also in colon and lung cancer compared to gen, Karlsruhe, Germany) without antibiotics. The cells were
corresponding normal tissues [25]. These GTPases are involved starved overnight in serum-free media before their application in
in the detection of chemotactic gradients and the formation of experiments. After trypsinization, the cells were resuspended in
cell polarity, and they also represent importartant regulators of serum-free adhesion medium (containing 1% bovine serum
actin, modulating the assemby and disassemby of actin albumin) for reconstitution of surface proteins prior to experi-
filaments. Their activation leads to stress fibre formation, mentation.
lamellipodial protrusions, membrane ruffling and directed cell
movement [26]. Transwell assay
We postulated that tumour cell extravasation into host organs is The breast cancer cells were added to FN, LN or C I-coated
a result of the specific availability of chemokine responses within transwell inserts with 8 mm-pore-size (Nunc/Thermo Fisher
the target organ’s microenvironment and the organ’s specific Scientific, Rockford, IL, USA). The cells were suspended into
composition of ECM components. We observed tumour cell arrest the upper chamber at a final concentration of 0,76105 cells/ml in
within the liver intravitally and investigated the adhesive and 500 ml adhesion medium. After 60 min adhesion time, CXCL12
migratory properties of breast cancer cells in response to various in concentrations 25, 50 or 100 ng/ml were added to the lower
ECM components in vitro, thus analysing the role of the chamber. Unstimulated cells served as negative control. For diffuse
CXCL12/CXCR4 axis in the initial steps of the metastatic stimulation CXCL12 at equal concentrations was added in the
cascade of host organ colonization. lower and upper chamber.

PLoS ONE | www.plosone.org 2 January 2012 | Volume 7 | Issue 1 | e30046

 CXCL12-Induced Tumour Cell Extravasation

After 4 or 16 h of incubation, the cells on the upper surface of Fluorescence images were recorded using timer-containing S-
the filter were removed by wiping with Q-tips, and the migrated VHS video system for further analysis.
cells on the lower side were fixed with formalin and stained with
crystal violet and haematoxylin. Cellular transmigration was In Vivo Observation of Metastatic Tumour Cell Adhesion
enumerated in 16 standardized microscopic fields per membrane. and Extravasation
To test the specific chemotactic response of the cells to For intravital observation of adhesive interactions between
CXCL12 the CXCR4 receptor was blocked using a neutralizing circulating tumour cells and the host organ microcirculation,
mouse-anti-human CXCR4 antibody (kind gift of A. Müller, single cell suspensions of CalceinAM fluorescence-labelled tumour
Düsseldorf, Germany). The antibody was added to the upper cells (16106) were injected intra-arterially within 60 sec. Previ-
chamber at concentrations of 5 and 10 mg/ml and migration of ously [6,7], we have shown that the route of cell application (left
the cells was observed for 4 or 16 h. heart, right heart, portal vein) did not influence the adhesive or
migratory behaviour within the liver sinusoids. This technique did
Static adhesion assays not interfere with cardio-circulatory or pulmonary functions of the
Microtiter plates (96 wells, Greiner BioOne, Frickenhausen, animals.
Germany) were coated with C I (10 mg/ml), FN (10 mg/ml), LN Off-line analysis was used to determine tumour cell behaviour
(10 mg/ml) or 1% BSA (negative control). Blocking of nonspecific within the target organs as previously described [24,29]. Various
binding sites was performed with 1% BSA for 30 min. After parameters were used for further investigation and semiquantita-
reconstitution of cell surface proteins in adhesion medium for tive analysis of these interactions. A semiquantitative analysis of
45 min, cells were resuspended in adhesion medium at a final tumour cell adhesion and extravasation was performed throughout
concentration of 16106 cells/ml and seeded to the coated wells. a 30 min observation period, and the numbers of adherent cells
The cells were stimulated with CXCL12 with concentrations of were counted for each of the 5 min intervals. Using a standardized
25, 50 and 100 ng/ml. After 30 or 60 min adhesion time cells procedure, all fields were analysed in each observation period and
were washed, fixed with formalin and then stained with crystal average numbers of adherent cells, migrated cells, and total cells
violet for 15 min. The absorbance was measured at 630nm using a observed were counted. The numbers provided represent the total
spectrophotometer. All experiments were performed in triplicates numbers of cells within 30 microscopic fields for each 5 min
and repeated at least three times. period. Numbers of arrested cells represent the total of adherent
and extravasated cells. Relative migration rates were calculated as
Time lapse microscopy percentages of cells within the host organ parenchyma in relation
The cells were added to ECM-coated culture dishes for 20 min. to the numbers of arrested cells.
After washing adhesion medium containing CXCL12 at different
concentrations was added. The cells were observed for 60 min GTPase activation assay
under a time lapse video microscope (Nikon, Düsseldorf, Non-adherent cells (107) in suspension unstimulated or
Germany). For quantification of cell motility cell tracking was stimulated with CXCL12 (25, 50, 100 ng/ml) for 15 min were
done using software packages CellˆD. washed once with ice cold PBS and lysed with 50mM Tris-HCl
pH 7.4, 150 mM NaCl, 1% NP40, 0.5% deoxycholate, 0.1%
Intravital microscopy SDS, 5mM EDTA and 1 ml inhibitor cocktail (Sigma) per 1 ml
Intravital microscopy was performed as previously described lysis puffer. Alternatively, adherent cells were seeded at different
[6,7]. The adequacy and reliability of this model in the ECM proteins (C I, FN and LN) for 60 min to exclude
investigation of early interactions between circulating tumour cells interference with initial adhesive behaviour and subsequently
and the hepatic microcirculation was confirmed previously [28]. stimulated with CXCL12 in a similar manner. As controls poly-l-
Briefly, Sprague Dawley rats (200 to 250 g) (Charles River, lysine (PLL for non-integrin mediated adhesion) and BSA
Sulzfeld, Germany) were cared for in accordance with standards of (negative control) were used.
the German Council on Animal Care, under an approved Lysates were clarified by centrifugation at 14,000xg for 5 min
protocol of the local animal welfare committee (Landesamt für and stored at 280uC. After protein quantitation and standardi-
Naturschutz, Umweltschutz und Veterinärmedizin: LANUV zation 8 ml Rho Assay Reagent (Rhotekin RBD glutathione
G84/2002). Rats were anesthetized using inhalation of isofluorane agarose beads) or 10 ml Rac/Cdc42 assay reagent (PAK-1 PBD
(Curamed, Karlsruhe, Germany). Permanent catheters were agarose conjugate) were added to 1000 mg total protein and
introduced into the left heart via the carotid artery and the right incubated for 45 min at 4uC. Samples were washed three times
heart via the jugular vein. After a wide median laparotomy was with magnesium-containing lysis puffer (25 mM HEPES pH 7.5,
performed, the left liver lobe was careful mobilized without 150 mM NaCl, 1% Igepal CA-630, 10% Glycerol, 10mM MgCl2,
disturbing hepatic microcirculation. Using a heated operating 1mM EDTA) and 1 ml inhibitor cocktail per 1 ml lysis puffer.
table, animals were fixed under an upright microscope and Agarose beads were resuspended in 4x Laemmli sample puffer and
positioned on their left side. This positioning allowed a partial boiled for 5min. Total Rho, Rac-1 or CDC42, respectively, served
luxation of the mobilized left liver lobe that was placed on a as loading control in each of the experiments. The lysates were
specific holder to investigate its lower surface. During the loaded on 12% polyacrylamide gels, then transferred to PVDF
experiments the liver was continuously irrigated with isotonic membranes and GTPases finally detected with rabbit anti-RhoA
saline solution. antibody (Santa Cruz), mouse anti-Rac antibody (BD Biosciences
An upright epifluorescence microscope (Zeiss, Oberkochen, Pharmingen) or rabbit anti-Cdc42 antibody (Cell Signalling).
Germany) was used with a 20-fold objective that was located over Bands were visualized with enhanced chemiluminescence (Milli-
a glass slip covering the organ surfaces. The microscope was pore, Schwalbach, Germany). Quantitative densitometry analysis
connected with a video enhancer-zoom lens system and a low-light was performed using ImageJ densitometry software (version 1.6,
charge-coupled device video camera (Peiper, Düsseldorf, Ger- NIH, Bethesda, MD) and selected bands were semi-quantified
many) allowing real-time imaging via a separate monitor. based on their optical densities.

PLoS ONE | www.plosone.org 3 January 2012 | Volume 7 | Issue 1 | e30046

 CXCL12-Induced Tumour Cell Extravasation

Immunofluorescence staining Chemotactic response to CXCL12 significantly increased 3–4

The cells were added to C I, FN, LN or Poly(L)lysine (PLL)- fold at C I and FN in both cell lines (p,0.001). In contrast, at LN,
coated chamber slides, incubated for 60 min and then stimulated a slight but significant (p,0.05) increase was induced in MDA-
using different chemokine concentrations (25, 50 and 100 ng/ml) MB-231, but not MDA-MB-468 cells. (fig. 1c+d)
for different time intervals (5, 15 and 30 min). Subsequently, cells
were washed and fixed using 4% paraformaldehyde. For In Vitro Tumour Cell Adhesion
intracellular staining, the cells were blocked and permeabilized We analysed whether cell adhesion properties to the proposed
using PBS containing 1% BSA and 0, 1% Triton X100, whereas interacting ECM components could be modified by CXCL12 in
permeabilization was not performed to achieve cell surface vitro. During adhesion to C I, FN or LN for 30 or 60 min, tumour
staining. Cells were incubated for 30 min with anti-human cells were treated with CXCL12 at concentrations of 25, 50 and
CXCR4 antibodies and subsequently with Alexa Fluor conjugated 100 ng/ml or remained untreated. Under untreated conditions,
secondary antibody for additionally 30 min. For actin filament both cell lines showed high adhesion rates at C I and FN, but only
staining phalloidin Alexa Fluor 488 and for nuclear staining moderate (MDA-MB-231, fig. 1e) or weak cell adhesion (MDA-
Hoechst 33342 was used. MB-468, fig. 1f) when LN was present. These differences were
Imaging was performed as combination of 3D-fluorescence expected due to the differing expression of LN-receptors at the
reconstruction and phase contrast microscopy using a Nikon cells’ surfaces (Table 1). The presence of CXCL12 did not
Eclipse TE2000 microscope. For each substrate 20 representative significantly modulate this static cell adhesion at any of the ECM
CXCR4 clusters were evaluated regarding their localization at the components in both cell lines. (fig. 1e+f) In addition, stimulation
cell, their size and shape. The size of CXCR4 clusters was with 25ng/ml CXCL12 for 15 min in nonadherent single cell
calculated using software packages Cell̂D (Olympus, Münster, suspensions of both cell lines did not affect the integrin surface
Germany). expressions. (data not shown)

Flow Cytometry Expression of CXCR4 chemokine receptors

Cells were fixed with 4% paraformaldehyde and then washed The highly metastatic MDA-MB-231 cells expressed CXCR4 in
and resuspended in PBS containing 0.5% BSA. After this the cells 84.5%, whereas the less metastatic MDA-MB-468 cells showed
were incubated for 45 min with PE-conjugated anti chemokine 78.7% expression. The two clones MDA-MB-231-19 and MDA-
antibodies (R&D systems). After washing the integrin or MB-231-27 had ,40% residual CXCR4 expression as estimated
chemokine receptor surface expression was measured using a by flow cytometry (Table 1, fig. 2).
FC500 flow cytometer (Beckman Coulter, Krefeld, Germany).
After gating of the cell population, the mean fluorescence CXCL12 affects the motility of breast cancer cells
intensities (MFI) of the antibody-stained cells were detected and Our previous observations [24,30] of chemotactic responses and
the relative amounts of positive cells were calculated using the flow ECM-dependence of colon carcinoma cell extravasation into the
cytometer software. liver suggested interactions between ECM composition and
chemotaxis. The results of this study confirmed these observations.
Statistical analysis Thus, we studied the dependence of short term chemokine
Statistical analysis was performed using the SPSS V.14 (SPSS stimulation and cell motility using single cell fluorescence-assisted
Inc., Chicago, IL) statistical program. Data were shown as time lapse microscopy. The cells were observed for 1 h in 30 sec
mean6SD. For comparison of different parameters between the intervals.
treatment groups p-values were calculated according to the For quantification of these effects we initially determined cell
Scheffé-test (ANOVA post-hoc-test) for dependent or independent surface areas, their circumferences, various cell diameters, length
samples as appropriate. For other analyses Student’s t-test has of moving path and the numbers of membrane ruffles over the
been used. Significant differences were accepted for p,0.05. time with or without CXCL12 stimulation. All of these parameters
showed high variability between the cells (data not shown), but
Results length of the motility path was the single parameter with
reproducible effects that was used for further quantitative analysis.
Chemokine-induced cell migration in vitro Using a cell tracking software, time-dependent path lengths of the
To examine chemotactic migratory responses to chemotactic cells were quantified (figure 3a).
stimuli MDA-MB-231 and MDA-MB -468 cells were exposed to If cells remained without chemokine stimulation, both cell lines
different CXCL12 gradients (0, 25, 50 and 100 ng/ml) in showed comparable patterns of basic cell motility, but MDA-MB-
Transwell chambers coated with different ECM components. 231 cells were more motile than MDA-MB-468 cells with longer
Spontaneous migration without chemotactic gradients was found lengths of path during the observation period. In the presence of
in both cell lines at basal levels. Bell-shaped concentration CXCL12, cells plated at C I had a similar increase in motility with
dependence showed maximum effects at 25 ng/ml at C I and a slightly different concentration-dependence in both cell lines
FN (p,0.05). Higher concentrations (100 ng/ml) resulted in (fig. 3b+c). Comparable to transwell migration, only MDA-MB-
migration rates at these ECM components that were comparable 231 cells were responsive to CXCL12 at LN, whereas at FN, only
to unstimulated cells. (fig. 1a). In order to confirm specific MDA-MB-468 cells showed stimulation of motility. The
responses, cells were additionally treated with an inhibitory anti- CXCL12-dependent stimulating effects were not found in
CXCR4 antibody. In unstimulated cells this resulted in slightly CXCR4-knock-down cells (MDA-MB-231-27). (fig. 3d)
increased migration rates suggesting partial agonistic activity of the MDA-MB-231 cells plated at C I developed distinct lamellipo-
antibody. In contrast, in stimulated cells, anti-CXCR4 completely dia. If stimulated with CXCL12, the cells were more motile and
blocked migration and only basal levels of migration were showed directed movement. In the presence of FN, these cells
detected. (p,0.001; fig. 1b) Finally, migration rates were formed smaller lamellipodia but formed protrusions especially
compared in the presence of different ECM components. using 25ng/ml CXCL12. When MDA-MB-231 cells migrated at

PLoS ONE | www.plosone.org 4 January 2012 | Volume 7 | Issue 1 | e30046

 CXCL12-Induced Tumour Cell Extravasation

Figure 1. CXCL12 stimulated cell adhesion and migration. Transwell migration assays were performed for 16 h using different gradients of
CXCL12 (0–100 ng/ml). (a) Bell-shaped response is shown for MDA-MB-231 cells at C I. (b) This response was specific for CXCL12 and could be
reversed by inhibitory anti-CXCR4 antibodies in these cells. A slight agonistic effect of the anti-CXCR4-mAb was observed. (c) In MDA-MB-231 cells
this response (25 ng/ml CXCL12) occurred at all ECM components, but to different extent. (d) In contrast, in MDA-MB-468 cells stimulated migration
(25 ng/ml CXCL12) was found at C I and FN, but not at LN. The values are reported as means 6 SD of three independent experiments (* p,0.001 and
+p,0.05). Cell adhesion: (e) MDA-MB-231 and (f) MDA-MB-468 cells were plated at C I (&), FN (;), LN (%). Binding to BSA coated surfaces was used
as background control. Cells were treated with 25, 50 or 100 ng/ml CXCL12 or left untreated (0 ng/ml). After 30 min adhesion time absorbance was
measured as optical density (OD) at 630 nm. CXCL12 did not significantly modify cell adhesion of both cell lines, but adhesion properties were
depended from the adhesive substrates with high adhesion rates at C I and FN compared to low or absent adhesion at LN.
doi:10.1371/journal.pone.0030046.g001

LN, numerous filopodia and spindle-like cell forms were observed CXCL12 induces tumour cell migration into the liver
(see Video S1). parenchyma
Low metastatic MDA-MB-468 cells developed small filopodia The in vitro results suggested that chemotactic CXCL12/
(microspikes) if seeded at C I or FN. Lamellipodia were not formed CXCR4-signaling is a crucial determinant of site-specific tumour
at C I, but cells appeared stretched when FN was present. In the cell extravasation into the liver that was further analysed using the
presence of high concentrations of CXCL12 (100 ng/ml), the cells in vivo model for hepatic tumour cell colonization. Breast cancer
partially started lamellipodia formation. In contrast to MDA-MB- cells showed specific adhesive and migratory properties as
231 cells, MDA-MB-468 cells did not spread at LN and formed previously described for other tumour entities [6,7,23,26].
only short filopodia (see Video S2). Differentiation between adherent and extravasated cells enabled

PLoS ONE | www.plosone.org 5 January 2012 | Volume 7 | Issue 1 | e30046

 CXCL12-Induced Tumour Cell Extravasation

Table 1. Cell surface expression of chemokines receptors.

MDA-MB 231 wt MDA-MB 231-19 MDA-MB 231-27 MDA-MB 468

neg 0.9 0.7 0.5 0.3

IgG control 3.5 4.2 2.7 14.9
CCR 3 72.6 76.1 75.2 94.2
CCR 7 18.3 26.7 37.4 70.6
CXCR 1 82.7 79.8 81.9 94.2
CXCR 2 5.2 6.6 7.5 56.7
CCR 1 1.6 1.8 1.0 2.0
CCR 2 1.9 1.9 1.3 2.1
CCR 6 33.9 35.1 33.7 89.4
CXCR 4 84.5 39.8 41.5 78.7
CXCR 5 78.7 75.1 77.9 91.3
CCR 5 40.1 58.6 63.1 88.2
CXCR 3 85.4 86.4 80.2 93.2

Cell surface expression was analysed by flow cytometry. shRNA transfection resulted in a reduction of CXCR4 expression in MDA-MB231-27 cells. Negative and IgG
controls are given as examples for the controls that have been used in each measurement. IgG controls were subtype-specific.
doi:10.1371/journal.pone.0030046.t001

semiquantitative analysis. All arrested cells did not completely arrested within the hepatic sinusoids compared to less metastatic
occlude the vessel lumen, suggesting specific adhesive interactions MDA-MB-468 cells (MDA-MB-231: 67–85 cells/interval; MDA-
with the sinusoidal vessel wall. (fig. 4a) Using highly metastatic MB-468: 37–43 cells/interval). In addition, MDA-MB-231 cells
MDA-MB-231 cells, significantly (p,0.05) higher numbers of cells extravasated at similar relative migration rates into the liver

Figure 2. Flow cytometry analysis of breast cancer cells. The cell surface expression of CXCR4 (3) was found in different breast cancer cell
lines MDA-MB-231 (a) and MDA-MB-468 (b). Downregulation of CXCR4 expression in clones MDA-MB-231 Cl 19 (c) and MDA-MB-231 Cl 27 (d). (isotype
control IgG2b: NNNN; negative control:&).
doi:10.1371/journal.pone.0030046.g002

PLoS ONE | www.plosone.org 6 January 2012 | Volume 7 | Issue 1 | e30046

 CXCL12-Induced Tumour Cell Extravasation

Figure 3. CXCL12 increases cell motility dependent from the ECM. (a) Cells that were attached to different ECM components were observed
for one hour using time-lapse video microscopy. The length of their moving path was determined using a cell tracking software (Cell ˆD H). (b) MDA-
MB-231 and (c) MDA-MB-468 cells showed different patterns of chemokine stimulated motility (% unstimulated; & 25ng/ml; ///50ng/ml; ; 100ng/ml
CXCL12). Highly metastatic MDA-MB-231 cells were more motile at CI and LN (*p,0.001) with a concentration-dependent stimulation (+p,0.05) of
their motility that was not observed at FN. In contrast, low-metastatic MDA-MB-468 cells were less motile at C I and LN, but showed a concentration-
dependent stimulation at FN. (d) In CXCR4-knock-down cells (MDA-MB-231-27) these CXCL12-dependent stimulating effects were not found. The
results are shown as mean 6 SD for three independent experiments. For statistical analysis ECM dependent migration was compared for all three
ECM components and concentration dependence of CXCL12 was compared to unstimulated cells. Significances for ECM dependent cell motility are
not marked for improved readability.
doi:10.1371/journal.pone.0030046.g003

parenchyma. (MDA-MB-231: 19–25%; MDA-MB-468: 20–23%). study were present in similar amounts in both cell lines. The major
(fig. 4b+c) difference between the cell lines was observed for ligands of LN
Since CXCL12 is specifically expressed within the sinusoidal (a3, a6 and b4) that were detected in both MDA-MD-231 and
structures [28] we investigated whether this chemokine and its MDA-MB-468 cells, but were less present in the latter. Integrin
receptor CXCR4 are involved in this behaviour. To validate the receptors for C I or FN binding were available in both cell lines.
CXCL12-induced breast cancer cell migration in vivo MDA-MB- (Table 2) Since both cell lines showed similarly high CXCR4
231 cells with shRNA-mediated reduction of CXCR4 (MDA-MB- expression levels, functional responsiveness of this receptor was
231-19 and MDA-MB-231-27) were compared with wild-type tested.
cells. In both clones this inhibition of CXCR4 receptor expression
did not affect adhesion of tumour cells within liver sinusoids Signalling responsiveness of CXCR4
(fig. 4d) but significantly (*p,0.001) decreased relative migration We investigated whether binding of the ligand CXCL12 could
rates (fig. 4e). activate CXCR4 signalling and whether these responses were be
This observation raised the question whether the rapid modulated by ECM binding. CXCR4 responsiveness was tested
extravasation is mediated by chemotactic responsiveness of breast using GTPases RhoA, Rac1 and Cdc42 as markers for its
cancer cells within the hepatic micro-environment or by different downstream signalling. As mentioned above, FN and C I, but not
integrin expression at their surfaces. C IV or LN were found to mediate cell adhesion and migration
within the liver [24]. Therefore, we specifically focussed on C I
Expression of integrins and FN, respectively. Since LN is not involved in these processes
In order to exclude differences in the cell surface expression of and was found only in small quantities within the liver, we used
both cell lines, flow cytometric analysis of CXCR4 and integrin this ECM component as control.
expression was performed. The cell surface expression of most Initially, CXCL12 treatment was performed in single cell
integrin subunits relevant to the ECM components used in this suspensions to show integrin and adhesion independent activation

PLoS ONE | www.plosone.org 7 January 2012 | Volume 7 | Issue 1 | e30046

 CXCL12-Induced Tumour Cell Extravasation

Figure 4. In vivo migration of breast cancer cells within the liver sinusoids. Single cell suspensions of fluorescence-labelled tumour cells
were injected into Spague-Dawley rats. (a) Example of an adherent cells (white arrow) and a cell starting to migrate (black arrow) into the liver
parenchyma. Magnifications show location of the cells in relation to marked sinusoid-parenchyma borders. Intravital microscopy was done to analyse
adhesion and migration properties of MDA-MB-231 (b) and MDA-MB-468 cells (c). Thirty microscopic fields were analysed in 5min observation periods
for semiquantitative analysis of adherent (x) and migrated cells (%). The total numbers of arrested cells (m) were calculated. CXCR4 inhibition
decreased tumour cell migration in vivo. The MDA-MB-231 cells were transduced with shRNA to inhibit CXCR4 expression. Two clones 19 (e)
and 27 (x) were tested for in vivo adhesion (d) and migration (e) in the rat liver and compared with untreated cells (&). Transfected cells showed
significantly (*p,0.001) decreased relative migration rates into the liver parenchyma but cell adhesion within the hepatic microcirculation was only
slightly influenced by CXCR4 reduction. Relative migration rates were based on the number of arrested cells and were calculated as described [26].
doi:10.1371/journal.pone.0030046.g004

of the GTPases. Basal levels of Rho-GTP, Rac-GTP and Cdc42- showed bell-shaped concentration dependence with a maximum at
GTP in untreated cells were detected in both cell lines. Treatment 25 ng/ml. Rac1 activation was increased up to two-fold at higher
with different concentrations of CXCL12 resulted in increased CXCL12 concentrations of 50 or 100 ng/ml in both cell lines,
activation of RhoA in MDA-MB-231 (+40%) and MDA-MB-468 whereas relevant CXCL12-induced Cdc42 activation was not
(+60%) cells compared to untreated cells. (fig. 5a+b) Both cell lines observed. (fig. 5a+b) This activation reached a maximum at

PLoS ONE | www.plosone.org 8 January 2012 | Volume 7 | Issue 1 | e30046

 CXCL12-Induced Tumour Cell Extravasation

Table 2. Expression of integrin subunits at cell surfaces. components compared to PLL-mediated cell attachment
(* p,0.001; +p,0.05; fig. 6e). Corresponding to the lower level
of integrin ligand expression for cell adhesion at LN, spreading at
Integrin subunits MDA-MB-231 MDA-MB-468 Ligands this adhesive substrate and formation of CXCR4 clusters was less
pronounced than for FN and C I.
Neg. 0.5 0.3
Using 3D-reconstruction, we found that larger clusters were
IgG control 1.6 0.9 located above the leading edges slightly apart from the adherent
a1 93.7 96.0 C I, LN cell regions, whereas small dot-like clusters were located directly at
a2 24.9 98.6 C I, LN the leading edges (fig. 5f). The presence of CXCL12 did not result
a3 99.8 96.5 LN 5+10 in differing sizes or localizations of these clusters compared to
unstimulated cells in both cell lines. In contrast, using PLL for
a5 87.9 53.7 FN
non-integrin-mediated cell attachment, CXCR4 was detected
a6 81,4 33,8 LN 1+5 without signs of special distribution (fig. 6d). Co-localization of
av 97.2 99.7 FN CXCR4 clusters with focal adhesion kinase was used to stain
b1 98,6 99,2 C I, FN, LN redistribution towards integrin-mediated focal adhesions, but these
b4 68,7 48.5 LN were not detected in any setting (not shown). The clusters started
to disappear 30 min after the beginning of chemokine stimulation
Cell surface expression was analysed by flow cytometry. The major difference in a comparable manner for both C I and FN. (fig. 7)
between the cells was observed for integrin ligands of LN. Negative and IgG
controls are given as examples for the controls that have been used in each
measurement. IgG controls were subtype-specific. Discussion
doi:10.1371/journal.pone.0030046.t002
The interaction of tumour cells with ECM components is an
important step in organ-specific metastasis formation. In order to
15 min after stimulation and subsequently dropped to basal levels
invade potential metastatic target organs, tumour cells are usually
after 30–45 min. (data not shown)
required to establish specific adhesion to vascular endothelial cells
For analysis of integrin-dependent activation of the Rho
and/or the organ’s ECM. Since metastatic colonization of the liver
GTPases, tumour cells were plated at C I, FN or LN. MDA-
is one of the most important life-threatening events in breast cancer,
MB-231 cells (fig. 5c) showed Rho, Rac and Cdc42 activation if
we investigated potential mechanisms that might mediate this
cells were plated at C I and FN, but lesser activation at LN. Using
process. Müller et al. [12] showed that chemokines and their
MDA-MB-468 cells, (fig. 5d) the GTPases showed less activation
receptors can promote colonization of breast cancer cells into their
in the presence of C I and FN compared to MDA-MB-231 cells.
targets, such as lymph nodes, lung, liver and bone marrow, all of
The most important difference however, was the almost complete
these being sites where chemokine ligands are mainly expressed. For
lack of GTPase activation after seeding MDA-MB-468 cells at LN.
example, Kupffer cells and sinusoidal endothelial cells can secrete
As expected, BSA induced no activation of GTPases in both cell
CXCL12 that is the ligand for CXCR4 [28]. Breast cancer cells
lines. PLL served as an integrin-independent control. Optical
specifically express functionally active CXCR4 and CCR7, which
densities were standardized against the Total-GTPase counter-
can trigger actin polymerization, pseudopodia formation and
parts and these relations were confirmed in all experiments as
directional movements [12]. In this study, we investigated the role
shown for one example in fig. 5e.
of CXCR4/CXCL12 in breast cancer cell migration and adhesion
In order to assess a potential synergism between integrin- with a focus on its dependence on liver ECM components that may
mediated and chemokine-induced activation of the GTPases, provide an organ-specific microenvironment for rapid tumour cell
ECM-seeded cells were stimulated with different concentrations of extravasation into this major metastatic organ.
CXCL12. This resulted in comparable GTPase activation In our study, breast cancer cells with differing metastatic
between unstimulated and stimulated cells. Only FN induced a potential demonstrated different cell adhesion to ECM compo-
slight synergistic increase of Rho in MDA-MB-468, but not MDA- nents in vitro and slightly different migrative properties. Both
MB-231cells (fig. 5f). investigated breast cancer cell lines expressed functionally active
CXCR4 at their surfaces and their in vivo potential for
ECM induced reorganization of CXCR4 cell surface chemotactic extravasation rates into the liver parenchyma was
distribution comparable. Two clones of MDA-MB-231 cells were transduced
Lateral redistribution of cell surface receptors, such as integrins with retroviral shRNA [27] to inhibit CXCR4 expression. This
and growth factor receptors, is known as a mechanism for their resulted in decreased extravasation in vivo, but did not affect cell
activation and can create synergistic functional effects. Since an adhesion within the sinusoids. However, only wild-type controls,
interaction between different ECM components and CXCR4 was but not shRNA-control transductions, were available for this
postulated, we further investigated the distribution of this receptor study. Although cell viability was not altered in the transduced
on the surface of adherent cells using immunohistochemical clones, some toxic effects cannot be completely ruled out. These
staining. After seeding at ECM components, CXCR4 was intravital results were confirmed by in vitro behaviour in our study
detected in clusters in both cell lines. Initially, small clusters were and by Chen et al. reporting the inhibition of breast cancer
found in cells in the presence of all ECM components including invasion/migration by down-regulation of CXCR4 [18]. In
PLL without preference of certain localizations. (not shown) contrast, the adhesive properties of the cells were not influenced
During cell spreading on ECM components, CXCR4 clusters by chemokine stimulation, results that are similar to those found
were relocated to lamellipodia (predominantly seen in FN) or by Fernandis et al. [19] for MDA-MD-231 cells. Overexpression
pseudopodia (predominantly seen in C I) within 5–30 min. In of CXCR4 was not used since both cell lines have high
contrast, in cells adherent to PLL, only small CXCR4 clusters percentages of CXCR4-positive cells. In addition, the receptor
were observed without changes over time. (fig. 6a–d) The size of availability at the cell surfaces is strictly regulated and therefore,
the clusters significantly increased in cells spread at ECM overexpression of CXCR4 would be unlikely to result in increased

PLoS ONE | www.plosone.org 9 January 2012 | Volume 7 | Issue 1 | e30046

 CXCL12-Induced Tumour Cell Extravasation

Figure 5. CXCL12 and ECM-induced activation of small Rho GTPases. Cells were treated with different concentrations of CXCL12 in single
cell suspensions (a+b) or plated at various adhesive substrates (c+d). Level of activation of RhoA, Rac1 and Cdc42 assessed by pull down assays of the
GTP-bound forms were determined in MDA-MB-231 (a+c) and MDA-MB-468 cells (b+d). As example for total GTPase detection independent from
phosphorylation status in each experiment Total-Rho is shown for (a) and (b). These loading controls were used for standardization of optical
densities (e+f). RhoA and Rac1 activation increased in a dose-dependent manner, but Cdc42 remained at baseline levels. If MDA MB 231 (c) and MDA
MB 468 (d) cells were plated at C I, FN, LN, PLL and BSA differences in ECM dependent activation of GTPases were observed. Comparing relative
optical densities these matrix dependent differences were confirmed (e). Synergistic activation between chemokine stimulation and ECM binding,
such as demonstrated for Rho (f), were not found.
doi:10.1371/journal.pone.0030046.g005

cell surface availability. These findings suggest that CXCL12/ similar CXCR4 surface levels, they did not respond to CXCL12
CXCR4-mediated chemotactic extravasation of breast cancer cells due to lack of GTPase activation. Although MDA-MB-231 and
is required for metastatic colonization of the liver. Since less MDA-MB-468 cells had comparable basic in vitro cell motility at
metastatic cells showed less initial tumour cell arrest within the all ECM components in our study, their response to CXCL12
liver, these initial adhesions - but not subsequent extravasation - stimulation differed between the ECM components. MDA-MB-
appear to be rate-limiting for hepatic metastasis formation. 468 cells mainly responded to CXCL12 stimulation when plated
In our current study, we observed a dose- and matrix-dependent at FN. In contrast, in MDA-MB-231 cells’ chemotactic response in
CXCL12-induced migration of breast cancer cells. The bell- the presence of C I and LN were found to be more pronounced
shaped response of the migration rates to different concentrations than in MDA-MB-468 cells. The slight increase in cell migration
of CXCL12 has also been shown by others [9,18]. Decreased under the influence of anti-CXCR4 is likely caused by a partial
migration at higher concentrations may be caused by saturation of agonistic effect of this antibody as has been described for many
CXCR4 ligand binding at cell surfaces and rapid internalization. function-blocking antibodies. Since C I appears to be the main
This appears to result in subsequent lack of GTPase activation. ‘‘road’’ for tumour cell extravasation into the liver parenchyma,
Using breast cancer cells in suspension, Holland et al. [21] found the enhanced chemotactic motility at C I might be an important
that only highly invasive cells were able to form G protein abc determinant for liver colonization. It can be speculated that MDA-
heterotrimers with CXCR4. Although non-invasive cells expressed MB-231 cells, which were also very motile when LN was present,

PLoS ONE | www.plosone.org 10 January 2012 | Volume 7 | Issue 1 | e30046

 CXCL12-Induced Tumour Cell Extravasation

Figure 6. Clustering of CXCR4 at tumour cell surfaces. CXCR4 was visualized in wild-type MDA-MB-231 cells using fluorescence-labelled
antibodies (red). Nuclei were counterstained (blue). Scale 50 mm. Figures were taken 15 min after CXCL12 stimulation and merged with phase-
contrast for cellular location of the CXCR4. In both cell lines similar distributions of CXCR4 were observed. MDA-MB-231 cells at C I (a+b; as integrin
dependent) and PLL (c+d; as integrin independent control) are shown as examples. Cell spreading at ECM components induced formation of large
CXCR4 clusters at lamellipodia and pseudopodia (a), whereas in cells adherent at PLL only small dot-like CXCR4 clusters were observed (c). Presence
of 25 ng/ml CXCL12 (C I: b, PLL: d) did not result in alterations of the size or localization of these clusters. (e) The sizes of 20 clusters in at least 10 cells
per experiment were evaluated for each adhesive substrate. Areas (mm2) of these clusters (&) and their circumference (mm, %) were significantly
higher in cells that were spread in an integrin-dependent manner at C I, FN or LN compared to unspread cells at PLL (* p,0.001; +p,0.05). (f) Larger
clusters (c) and dot-like CXCR4 structures (R) were found with a preference of the dot-like structures in the direct neighbourhood of leading
adhesive boundaries.
doi:10.1371/journal.pone.0030046.g006

may use this ECM component to metastasize into target organs carcinoma cells [34]. In the breast cancer cells investigated in our
where LN is expressed in higher amounts, such as in the lung. This study, similar alterations of integrin expression during exposure to
would result in earlier escape from the potentially toxic CXCL12 were not observed.
environment within the microvessels [31]. We found that integrin-mediated cell adhesion at ECM
In cancer cells, different combinations of chemokine receptor components can specifically induce redistribution of CXCR4 at
activation and integrin binding appear to enable different tumour cell surfaces with formation of larger receptor clusters.
migration programs, such as mesenchymal-amoeboid transition Since this was accompanied by intensified formation of lamelli-
or collective-amoeboid movement. This form of rescued migration podia and pseudopodia, it is likely that specific integrin-mediated
has been shown to be closely related to integrin-ECM interactions cell adhesion can induce increased CXCR4 sensing [12] and
[32]. Potential interactions between integrin binding and chemo- subsequent ability for rapid extravasation along ECM compo-
tactic responses were previously reported for different tumour nents. These morphological alterations after CXCR4 activation
entities. For example, integrin-mediated pancreatic cancer cell were strong in the highly metastatic MDA-MB-231 cells, and
migration at LN was found to up-regulate CXCR4 and IL-8 present to a lesser extent in the less metastatic MDA-MB-468 cells.
expression and responsiveness to CXCL12 stimulation [33]. In We therefore assume that interactions between integrins and
addition, CXCL12 induced redistribution of various integrins chemokines during chemotactic tumour cell extravasation are
between cell surface and intracellular compartments in renal more related to outside-in signalling, inducing higher chemotactic

PLoS ONE | www.plosone.org 11 January 2012 | Volume 7 | Issue 1 | e30046

 CXCL12-Induced Tumour Cell Extravasation

can be differently modified by adhesion of various (tumour) cell

types to various ECM-components [34,39,40].
As previously reported, composition of the ECM within the liver
can provide certain structures that may act as guidance for tumour
cell extravasation. These migrative properties can be influenced by
interactions of integrins with different ECM-components including
FN and type IV collagen for initial cell arrest, and C I for
extravasation [6,7,24]. Sheets of FN and very small amounts of LN
within the subendothelial space of Dissé and fibres of C I occurring
between hepatocytes could form a path for adherent cells to migrate
into the liver parenchyma after sufficient stimulation [24]. Although
some differences in the integrin surface expression were found in the
breast cancer cells in this study, they cannot solely explain the
differing migrative properties. Since tumour cell adhesion but not
extravasation correlated with the metastatic potential of the cells,
initial arrest within the liver sinusoids appears to be more important
as rate-limiting factor in breast cancer metastasis. This is in contrast
to other carcinoma entities where rapid extravasation was central to
determine metastatic potential [27,28].
These findings suggest that the combination of chemokine
availability at optimal local concentrations, specific ECM
composition with differential integrin binding of tumour cells
and interaction of chemokine receptors and integrins via GTPase
activation are determinants of the extent and time course of
tumour cell extravasation into potential metastatic target organs.

Conclusions
In summary, our results show that chemokines appear to be
involved in metastatic tumour cell migration and motility in an
organ-specific manner. Using in vitro and intravital observation
Figure 7. Kinetics of CXCR4 cell surface expression. MDA-MB-231 techniques, we were able to demonstrate that CXCR4 plays an
cells were seeded at C I (left column) or FN (right column) and important role in guiding breast cancer cells to target organs, such
stimulated with CXCL12 for up to 30 min. Fixed cells without
as liver, due to integrin-adhesion dependent activation, [41]. This
stimulation (a+b) or with CXCL12 stimulation (25 ng/ml) for 5 min
(c+d), 15 min. (e+f) or 30 min (g+h) were stained for surface expression finding is further supported by the fact that the required cross-
of CXCR4. The number of large CXCR4 clusters was reduced after 30 at signalling between integrins and chemokine receptors for chemo-
cells in a comparable manner for both ECM components. Scale 10 mm. tactic cell motility is ECM-dependent. Availability of chemokine
doi:10.1371/journal.pone.0030046.g007 receptors at tumour cell surfaces, presence of their ligands within
the microenvironment of potential target organs and the suitability
cell motility with subsequent faster extravasation. In addition, of their ECM composition seem to be required for successful
inside-out activity can improve sensing of chemokine gradients. tumour cell extravasation as early steps of metastasis formation.
A number of cross-signalling pathways between CXCR4 and
integrins may be responsible for these ECM-dependent chemokine Supporting Information
effects, such as PI3K-AKT [16,35], FAK-Crk, RAFTK/Pyk2
[19]. Besides kinases, chemokine receptors and integrins can Video S1 Time-lapse microscopy of MDA-MB-231 at LN
activate or modulate small GTPase signalling pathways [36,37]. in the presence of 25 ng/ml CXCL12 for 60 min.
RhoA promotes the contraction and retraction of the cell body at (AVI)
its rear during directed motility. Rac can induce membrane Video S2 Time-lapse microscopy of MDA-MB-468 at LN
protrusion at the front of the cell and Cdc42 regulates direction of in the presence of 25 ng/ml CXCL12 for 60 min.
migration by regulating cell polarity [38]. In our study, their (AVI)
CXCL12-induced activation in cell suspensions and increased
responsiveness of cells adherent to C I corresponded to the motility Acknowledgments
response of the cell lines. Limited GTPase activation in cells in
suspension without adhesive interactions is likely caused by lost cell We thank Ms. K. Hagen and Mrs. F. Spiecker for their technical
polarity that cannot be developed in a proper manner without assistance.
adhesive contacts. The requirement of specific, integrin-mediated
adhesive contacts for sufficient CXCL12 stimulated GTPase Author Contributions
activation in our study appears to promote increased chemotactic Conceived and designed the experiments: STM JH. Performed the
tumour cell motility in various metastatic target organs. This is experiments: CW AHB JK SS. Analyzed the data: STM JH. Contributed
further supported by previous observations that GTPase activation reagents/materials/analysis tools: MK. Wrote the paper: CW JH.

References
1. Yeatman TJ, Nicolson GL. Molecular basis of tumour progression: mechanisms 2. Lester BR, McCarthy JB. Tumour cell adhesion to the extracellular matrix and
of organ-specific tumour metastasis. Semin Surg Oncol 1993 9: 256–263. signal transduction mechanisms implicated in tumour cell motility,

PLoS ONE | www.plosone.org 12 January 2012 | Volume 7 | Issue 1 | e30046

 CXCL12-Induced Tumour Cell Extravasation

invasion and metastasis. Cancer Metastasis Rev. Review 1992 11: 31– 22. Alberts B, Johnson A, Lewis J, Raff M, Roberts K, et al. Molecular Biology of
44. the Cell. Sarah Gibbs. New York, Garland Science, 2002 1090-1092.
3. Mook OR, Van Marle J, Vreeling-Sindelárová H, Jonges R, Frederiks WM, 23. Ivanoff J, Talme T, Sundqvist K-G. The role of chemokines and extracellular
et al. Visualization of early events in tumour formation of eGFP-transfected rat matrix components in the migration of T lymphocytes into three-dimensional
colon cancer cells in liver. Hepatology 2003 38: 295–304. substrata. Immunology 2005 114: 53–62.
4. Steinbauer M, Guba M, Cernaianu G, Köhl G, Cetto M, et al. GFP-transfected 24. Rosenow F, Ossig R, Thormeyer D, Gassmann P, Schlüter K, et al. Integrins as
tumour cells are useful in examining early metastasis in vivo, but immune Antimetastatic Targets of RGD-Independent Snake Venom Components in
reaction precludes long-term tumour development studies in immunocompetent Liver Metastasis. Neoplasia 2008 10: 168–172.
mice. Clin Exp Metastasis 2003 20: 135–141. 25. Fritz G, Just J, Kaina B. Rho-GTPases are overexpressed in human tumours. Int
5. Ding L, Sunamura M, Kodama T, Yamauchi J, Duda DG, et al. In vivo J. Cancer 1999 81: 682–687.
evaluation of the early events associated with liver metastasis of circulating 26. Sahai E, Marshall CJ. RHO-GTPases and cancer. Nat Rev Cancer 2002 2:
cancer cells. Br J Cancer 2001 85: 431–438. 133–142.
6. Enns A, Gassman P, Schlüter K, Korb T, Spiegel HU, et al. Integrins can
27. Akekawatchai C, Holland JD, Kochetkova M, Wallace JC, McColl SR. Trans-
directly mediate metastatic tumour cell adhesion within the liver sinusoids.
activation of CXCR4 by the insulin-like growth factor-1 receptor (IGF-1R) in
J Gastrointest Surg 2004 8: 1049–1059.
human MDA-MB-231 breast cancer epithelial cells. J Biol Chem.2005 280:
7. Enns A, Korb T, Schlüter K, Gassman P, Spiegel HU, et al. Alpha v beta 5-
39701–39708.
integrins mediate early steps of metastasis formation. Eur J Cancer 2005 41:
1065–1072. 28. Haier J, Korb T, Hotz B, Spiegel HU, Senninger N. An intravital model to
8. Gassmann P, Haier J. The tumour cell host organ interface in the early onset of monitor steps of metastatic tumor cell adhesion within the hepatic microcircu-
metastatic organ colonisation. Clin Exp. Metastasis 2008 25: 171–181. lation. J Gastrointest Surg 2003 7: 507–515.
9. Ali S, Lazennec G. Chemokines novel targets for breast cancer metastasis. 29. Schlüter K, Gassmann P, Enns A, Korb T, Hemping-Bovenkerk A, et al. Organ-
Cancer Metastasis Rev 2007 26: 401–420. specific metastatic tumour cell adhesion and extravasation of colon carcinoma
10. Baruch-Ben A. Organ selectivity in metastasis: regulation by chemokines and cells with different metastatic potential. Am J Pathol 2006 169: 1064–1073.
their receptors. Clin Exp Metastasis 2008 25: 345–356. 30. Gassmann P, Haier J, Schlüter K, Domikowsky B, Wendel C, et al. CXCR4
11. Dittmar T, Heyder C, Gloria-Maercker E, Hatzmann W, Zänker KS. Adhesion regulates the early extravasation of metastatic tumour cells in vivo. Neoplasia
molecules and chemokines: the navigation system for circulating tumour (stem) 2009 11: 651–661.
cells to metastasize in an organ-specific manner. Clin Exp Metastasis 2008 25: 31. Qiu H, Orr FW, Jensen D, Wang HH, McIntosh AR, et al. Arrest of B16
11–32. melanoma cells in the mouse pulmonary microcirculation induces endothelial
12. Müller A, Homey B, Soto H, Ge N, Catron D, et al. Involvement of chemokine nitric oxide synthase-dependent nitric oxide release that is cytotoxic to the
receptors in breast cancer metastasis. Nature 2001 410(6824): 24–25. tumour cells. Am J Pathol 2003 162: 403–12.
13. McPherson K, Steel CM, Dixon JM. ABC of breast diseases. Breast cancer- 32. Wolf K, Friedl P. Molecular mechanisms of cancer cell invasion and plasticity.
epidemiology, risk factors, and genetics. BMJ 2000 321: 624–628. Br J Dermatol Review 2006 154: 11–5.
14. Balkwill F. Cancer and the chemokine network. Nat Rev Cancer 2004 4: 33. Grzesiak JJ, Smith KC, Burton DW, Deftos LJ, Bouvet M. Integrin-mediated
540–550. laminin-1 adhesion upregulates CXCR4 and IL-8 expression in pancreatic
15. Mirisola V, Zuccarino A, Bachmeier BE, Sormani MP, Falter J, et al. CXCL12/ cancer cells. Surgery 2006 141: 804–14.
SDF1 expression by breast cancers is an independent prognostic marker of 34. Jones J, Marian D, Weich E, Engl T, Wedel S, et al. CXCR4 chemokine
disease-free and overall survival. Eur J Cancer.2009 45: 2579–2587. receptor engagement modifies integrin dependent adhesion of renal carcinoma
16. Lee B-C, Lee T-H, Avraham S, Avraham HK. Involvement of the chemokine cells. Exp. Cell Res 2007 313: 4051–65.
recptor CXCR4 and its ligand stromal cell-derived factor 1a in breast cancer cell 35. Andrew N, Insall RH. Chemotaxis in shallow gradients is mediated
migration through human brain microvascular endothelial cells. Mol Cancer independently of PtdIns-3-kinase by biased choices between random protru-
Res 2004 2: 327–338. sions. Nat Cell Biol 2007 9: 193–200.
17. Kulbe H, Levinson NR, Balkwill F, Wilson JL. The chemokine network in 36. Tsubouchi A, Sakakura J, Yagi R, Mazaki Y, Schaefer E, et al. Localized
cancer – much more than directing cell movements. Int J Dev Biol 2004 48: suppression of RhoA activity by Tyr31/118-phosphorylated paxillin in cell
489–496. adhesion and migration. J Cell Biol 2002 159: 673–683.
18. Chen Y, Stamatoyannopoulos G, Song C-Z. Down-regulation of CXCR4 by
37. Kirui JK, Xie Y, Wolff DW, Jiang H, Abel PW, et al. Gbetagamma signaling
inducible small interfering RNA inhibits breast cancer cell invasion in vitro.
promotes breast cancer cell migration and invasion. J Pharmacol Exp Ther 2010
Cancer Res 2003 63: 4801–4804.
333: 393–403.
19. Fernandis AZ, Prasad A, Band H, Klösel R, Ganju RK. Regulation of CXCR4-
mediated chemotaxis and chemoinvasion of breast cancer cells. Oncogene 2004 38. Raftopoulou M, Hall A. Rho GTPases lead the way. Developmental Biology
2: 157–167. 2004 265: 23–32.
20. Kang H, Watkins G, Parr C, Douglas-Jones A, Mansel RE, et al. Stromal cell 39. Liu Y, Senger DR. Matrix-specific activation of Src and Rho initiates capillary
derived factor-1: its influence on invasiveness and migration of breast cancer morpho-genesis of endothelial cells. FASEB J 2004 18: 457–468.
cells in vitro, and its association with prognosis and survival in human breast 40. Jianguo G, Yasuhiro S, Noriko S, Kiyotoshi S. Laminin-10/11 and fibronectin
cancer. Breast Cancer Research 2005 7: 402–410. differentially regulate integrindependent Rho and Rac activation via p130Cas-
21. Holland JD, Kochetkova M, Akekawatchai C, Dottore M, Lopez A, et al. CrkII-DOCK180 Pathway. J Biol Chem 2001 276: 27090–27097.
Differential functional activation of Chemokine receptor CXCR4 is mediated by 41. Furusato B, Mohamed A, Uhlén M, Rhim JS. CXCR4 and cancer. Pathol
G proteins in breast cancer cells. Cancer Res 2006 66: 4117–4124. Int.2010 60: 497–505.

PLoS ONE | www.plosone.org 13 January 2012 | Volume 7 | Issue 1 | e30046