JCS ePress online publication date 4 December 2002
Research Article
377
α5β1 integrin mediates strong tissue cohesion
Elizabeth E. Robinson, Kathleen M. Zazzali, Siobhan A. Corbett and Ramsey A. Foty*
Department of Surgery, University of Medicine and Dentistry-Robert Wood Johnson Medical School, CAB Room 7070, New Brunswick, NJ 08648,
USA
*Author for correspondence (e-mail: [email protected])
Accepted 16 October 2002
Journal of Cell Science 116, 377-386 © 2003 The Company of Biologists Ltd
doi:10.1242/jcs.00231
Summary
Integrins and cadherins are considered to have distinct and
opposing functions. Integrins are traditionally cited for
their role in cell-substratum interactions, whereas
cadherins are thought to mediate strong intercellular
cohesion. Together, these adhesion systems play crucial
roles in a wide variety of cellular and developmental
processes including cell migration, morphology,
differentiation and proliferation. In this manuscript we
present evidence that integrins possess the ability to
mediate strong intercellular cohesion when cells are grown
as 3D aggregates.
Much of the data elucidating the role of integrins as
mediators of cell-extracellular matrix (ECM) interactions
have been generated using conventional cell culture
techniques in which cells are plated onto ECM-coated 2D
surfaces. In vivo, cells are embedded in a 3D meshwork
of ECM proteins. We hypothesized that, within this
meshwork, integrin-ECM interactions may impart
cohesivity to an aggregate of cells by linking adjacent cells
together. To test this hypothesis, we transfected Chinese
hamster ovary (CHO-B2) cells to express α5β1 integrin
and found that these cells formed compact, spherical
aggregates. We measured aggregate cohesivity using tissue
surface tensiometry, a novel technique that quantifies cellcell cohesivity of spheroids under physiological conditions.
Introduction
Cell-cell and cell-extracellular matrix (ECM) interactions are
fundamental regulators of both normal and abnormal
biological processes including morphogenesis (Fujimori et al.,
1990), wound healing (Eckes et al., 2000) and malignant
invasion (Behrens, 1993; Foty et al., 1998; Foty and Steinberg,
1997; Okegawa et al., 2002; Shiozaki et al., 1996; Steinberg
and Foty, 1997; Tlsty, 1998; Zhou et al., 2000). The roles of
cadherins and integrins in these processes are well defined but
have traditionally been considered as having two distinct and
opposing functions. Cadherins regulate strong cell-cell
cohesion, whereas integrins are primarily responsible for cellECM adhesion.
α5β1 integrin binds to fibronectin (FN) and has a well-defined
role in cell adhesion, migration and matrix formation. FN is a
ubiquitous, multifunctional component of the ECM that exists as
a dimer, with the two chains connected by disulfide bonds at the
C-terminus (Hynes, 1992). Structurally, each FN chain contains
a single cell-binding domain including an RGD sequence to
which α5β1 integrin specifically binds (Akiyama, 1996).
We determined that α5β1 integrin is capable of conferring
strong cohesivity (σ=8.22±0.68 dynes/cm) to aggregates of
α5-integrin-transfected cells. This cohesion was found to be
independent of cadherin expression and was significantly
greater than the cohesivity conferred onto CHO-B2 cells
transfected with N-cadherin (σ=3.14±0.20 dynes/cm,
P≤0.0001), a more traditional cell-cell cohesion system.
Fibronectin-null CHO cells that express α5β1 integrin
but do not secrete endogenous fibronectin do not form
aggregates in fibronectin-depleted medium. Addition of
increasing amounts of exogenous dimeric fibronectin to
these cells resulted in a dose-dependent compaction.
However, compaction failed to occur in the presence of
fibronectin monomers. These data indicate that fibronectin
is required for α5β1-mediated compaction and that the
dimeric structure of fibronectin is essential for this process.
Additionally, aggregate formation of the α5 integrin
transfectants was inhibited by an RGD peptide thus
confirming α5β1 integrin specificity. Collectively, these data
confirm our hypothesis that α5β1 integrin acts through
fibronectin to link adjacent cells together, thus promoting
strong intercellular cohesion in 3D cellular aggregates.
Key words: Integrins, Cadherins, Cohesivity, Tissue surface
tensiometry, 3D, Aggregates
Many of the assays measuring the affinity of integrins for
their ligands have relied largely upon either measurement of
the binding kinetics (Goldmann, 2000) or upon assays in which
cells, adherent to a 2D ligand-coated substrate, are subjected
to either centrifugal force (Koo et al., 2002) or to shear stress
(Goldstein and DiMilla, 2002). Although such assays are
useful, they do not take into consideration the 3D nature of
tissues and, as such, do not consider the ECM as a potential
mechanism by which multicellular tissues may be crosslinked.
We reasoned that integrin-mediated adhesion to ECM proteins,
when modeled in a 3D system, could cause cells to aggregate
and compact into spheroids imparting cohesivity to these
aggregates, much as cadherins have been shown to do.
To test this hypothesis, we applied a recently developed
method, tissue surface tensiometry (TST), to measure the
intensity of intercellular cohesion within tissue-like aggregates
under physiological conditions. The biophysical principles
underlying TST have been previously described in detail (Foty
et al., 1994). Briefly, TST measures intercellular binding
energy by compressing spherical cellular aggregates between
378
Journal of Cell Science 116 (2)
parallel plates in a specially designed device. The shape of the
compressed aggregate is recorded, as is the force with which
the aggregate resists the compressive force. These parameters
are then applied to the Young-Laplace equation (Davies and
Rideal, 1963), generating measurements of aggregate surface
tension, or cohesivity. Using TST, we have previously shown
that aggregate cohesivity specifies the spatial positioning of
tissues within multicellular aggregates (Foty et al., 1996),
modulates the rate of tissue spreading on a substratum (Ryan
et al., 2001) and influences the emigration of cells from a tumor
(Foty et al., 1998; Foty and Steinberg, 1997).
We generated a series of cell lines custom-designed to
express α5β1 integrin. We also generated cell lines expressing
high levels of N-cadherin in order to quantitatively compare
the strength of integrin-based cohesivity with that of a more
traditional cell-cell cohesion system. We then prepared
spherical aggregates of these cells and quantified their
cohesivity by TST. We determined that α5β1 integrin mediates
strong cohesivity in 3D tissue aggregates and that this
cohesivity is greater than that conferred by N-cadherin alone.
We also determined that this integrin-mediated cohesivity is
FN dependent and results from a specific interaction between
α5β1 integrin and the RGD site on FN. Structurally, we found
that dimeric FN is required for aggregate formation and
compaction. Therefore, using this 3D culture system and TST,
we were able to quantitatively show that α5β1 integrin is
capable of mediating strong intercellular cohesion. These data
are the first to rigorously quantify the cohesivity imparted to
tissue-like cellular aggregates by the interaction of α5β1
integrin with FN. These data have broad implications in fields
influenced by these processes including malignant invasion,
embryonic development and wound healing.
Materials and Methods
Cell lines
Chinese hamster ovary cells (CHO-B2) express β1 integrin but do not
express the α5 subunit (Zhang et al., 1993). Cells were maintained at
37°C in 95% air/5% CO2 in Dulbecco’s Modified Eagle’s Medium
with 10% fetal calf serum, 1 mM sodium pyruvate, 0.1 mM MEM
non-essential amino acids, 2 mM L-glutamine, 100 µg/ml
streptomycin sulfate, 100 units/ml penicillin G sodium and 0.25 µg/ml
amphotericin B (Gibco-BRL, NY). A CHO cell subclone (CHO-α5)
that expresses high levels of α5β1 integrin but does not secrete FN or
express cadherin was also used. These cells were maintained as
described above.
Construction of an α5 integrin cDNA expression vector
A 1.8 kb BamHI-XhoI fragment encoding the N-terminal portion of
the human α5 cDNA in pLJ (kindly provided by Jean Schwarzbauer,
Princeton University) was cloned into pcDNA 3.1(+) (pcDNA 3.1 α5N, Invitrogen, CA). A 2.5 kb fragment containing the C-terminus of
the human α5 cDNA was then ligated into an XhoI digest of pcDNA
3.1 α5-N to yield a complete human α5 cDNA in pcDNA 3.1.
Transfection
Cells were transfected by electroporation in 400 µl of transfection
medium (RPMI, 0.1 mM DTT, 10 mM dextrose) at 200 volts and 960
µF in a 0.4 cm electroporation cuvette using a BioRad Gene Pulser II
apparatus. For expression of human α5 integrin, CHO-B2 cells were
transfected with 20 µg of pcDNA3 plasmid containing the coding
region for human α5 integrin and resistance to G418. For expression
of chicken N-cadherin, CHO-B2 cells were transfected with 20 µg of
the N-cadherin expression plasmid pMiwcN (Fujimori et al., 1990)
(kindly provided by M. Takeichi, Kyoto University) and 5 µg of the
zeocin resistance plasmid pZeoSV (Invitrogen, CA). Cells were
grown to confluence in medium supplemented with either 800 µg/ml
G418 or 500 µg/ml zeocin. Empty vector control cells (CHO-P3;
pcDNA3 only) were generated by the same transfection and selection
processes. α5-integrin-transfected (CHO-A5) and N-cadherintransfected (CHO-Ncad) cells were bulk sorted by FACS.
FACS and flow cytometry
CHO-A5 cells were detached with trypsin-EDTA (Gibco-BRL, NY),
washed three times with ice-cold Hanks’ balanced salt solution
(HBSS) and incubated with an anti-human α5 integrin antibody
(CD49e, PharMingen, CA) at 5 µg/ml on ice for 45 minutes. Cells
were again washed with cold HBSS and incubated on ice for an
additional 45 minutes with a FITC-conjugated goat-anti-mouse
secondary antibody (Zymed, CA). Cells expressing α5 integrin were
FACS sorted (EPICS ALTRA, Beckman Coulter, FL) and expanded.
CHO-Ncad cells were detached from tissue culture plates with
trypsin-calcium (0.1% trypsin/5 mM Ca2+) to preserve cadherin
receptor integrity (Hyafil et al., 1981) and sorted to express levels of
N-cadherin similar to α5 integrin expression by the CHO-A5 cells.
CHO-Ncad cells were incubated with 10 µg/ml anti N-cadherin
primary antibody (NCD2, Zymed, CA), followed by a FITCconjugated goat-anti-rat secondary antibody (Zymed, CA). Cells were
sorted three times to generate pure populations. Receptor expression
was confirmed monthly by flow cytometry.
Western blot analysis
Expression of both α5 integrin and N-cadherin was confirmed by
western blot analysis using standard protocols. Cell lysates were
prepared from near-confluent 10 cm tissue culture plates. Cell
monolayers were washed twice with ice-cold HBSS then lysed by the
addition of 500 µl RIPA lysis buffer (150 mM NaCl, 50 mM TRIS
pH 7.5, 1% NP40, 0.25% DOC) containing a protease inhibitor
cocktail (Calbiochem, CA), EDTA and sodium vanadate. The lysates
were transferred to microcentrifuge tubes, rotated at 4°C for 1 hour,
then passed through a Qia-shredder (Qiagen, CA) and centrifuged at
14,000 g for 15 minutes at 4°C. Cell lysates from 3D aggregates were
prepared in the same manner except that aggregates were disrupted
by sonication in RIPA lysis buffer. Protein concentrations were
determined using the BCA Protein Assay Kit (Pierce, IL). 20 µg of
protein was separated on a 7% SDS-PAGE gel and electroblotted to
nitrocellulose using standard protocols. Blots were blocked overnight
at 4°C in either Membrane Blocking Solution (CHO-Ncad, Zymed,
CA) or 5% nonfat dry milk in TBS-0.2% TWEEN 20 (CHO-A5).
Blots were incubated at room temperature for 1 hour in either an Ncadherin primary antibody (3B9 at 0.5 µg/ml, Zymed, CA) or in an
α5 integrin primary antibody (AB1928 at 1 µl/ml, Chemicon, CA),
washed three times with TBS-0.2% TWEEN 20, followed by an
additional 1 hour incubation in either horseradish-peroxidaseconjugated goat anti-mouse secondary antibody (CHO-Ncad) or goatanti-rabbit secondary antibody (CHO-A5). Blots were developed
using SuperSignal West Pico Chemiluminescent Substrate (Pierce, IL)
and exposed to X-ray film. All blots were then stripped in 62.5 mM
Tris HCl pH 6.8, 2% SDS, and 100 mM 2-mercaptoethanol for 30
minutes at 50°C and re-probed with an anti-actin antibody (Sigma,
MO) to confirm equal lane loading.
Fast aggregation assay
Aggregation assays, performed in the presence and absence of
calcium, measure rapid, calcium-dependent aggregation (Takeichi,
α5β1 integrin mediates strong tissue cohesion
1977). This technique was used to assess cadherin function of the
CHO-B2, CHO-P3, CHO-A5 and CHO-Ncad cell lines. Cells were
detached from near-confluent 10 cm plates by trypsin-calcium,
washed in PBS, and stained with PKH26 (Sigma, MO) red fluorescent
membrane intercalating dye according to the manufacturer’s
instructions. Stained cells were washed three times in HBSS
containing either 2 mM CaCl2 (HBSS + Ca2+) or in Ca2+/Mg2+-free
HBSS (HBSS-Ca2+). Cells were counted and resuspended at 1.0×106
cells/ml of HBSS±Ca2+. 3 ml of each cell suspension was transferred
to individual 10 ml shaker flasks (Belco Glass, NJ) and incubated on
an orbital shaker at 120 rpm for one hour at 37°C. The degree of
aggregation was assessed by fluorescence microscopy.
Aggregate formation and hanging drop cultures
Cells were removed from near-confluent 10 cm plates with trypsinEDTA, washed, counted and resuspended at a concentration of
2.5×106 cells/ml in complete medium supplemented with 2 mM
CaCl2. 15 and 20 µl aliquots of this suspension were deposited on the
underside of a 10 cm tissue culture dish lid. The lid was then inverted
over 10 ml of 1× phosphate-buffered saline creating hanging drops on
the upper lid. Drops were incubated under tissue culture conditions
for 2-3 days, allowing the cells to coalesce at the base of the droplets
and form sheets. The sheets from hanging drop culture were then
transferred to 10 ml shaker flasks (Belco Glass, NJ) in 3 ml complete
medium and placed on an orbital shaker at 110 rpm for 2-3 days. This
encouraged cell rearrangement and 3D spheroid formation. Spheroids
ranged in size from 540-800 µm in diameter.
In order to determine the molecular mechanisms of α5β1-mediated
aggregate compaction, we performed hanging drop assays in FNdepleted medium. Serum was depleted of FN by incubation with
collagen sepharose beads as previously described (Corbett et al.,
1997). Depletion was confirmed by western blot analysis. Hanging
drops of CHO-A5 cells were cultured in the presence of a 100 µM
cyclic RGD-blocking peptide (FR-1, Calbiochem, CA). Hanging
drops of the CHO-α5 cells were cultured in FN-depleted medium,
with 3-300 µg/ml exogenous rat plasma FN, or in the presence of a
recombinant monomeric FN fragment.
Synthesis of recombinant fibronectin monomer
Construction of recombinant FN monomers was performed in the
baculovirus vector pVL1393 as previously described. Briefly, a
terminal codon was generated by adding an XbaI linker to a StuI site
located at position 6860 in the FN∆III1-7 cDNA, yielding a FN
molecule structurally identical to FN∆III1-7 but lacking the
dimerization site (Corbett and Schwarzbauer, 1999; Sechler et al.,
1996). This fragment was inserted between the BamHI and XbaI sites
in pVL1393. The construct was confirmed by restriction enzyme
digest analysis. Recombinant protein production was performed as
previously described (Sechler et al., 1997).
Tissue surface tensiometry
Aggregate cohesivity was measured by TST as previously described
(Foty et al., 1994; Foty et al., 1996). Aggregates ranging in size from
540-800 µm were transferred to the inner chamber of the tensiometer
and positioned on the lower compression plate (LCP, Fig. 1A). The
inner chamber contained pre-warmed, de-gassed CO2-independent
medium at 37°C. The upper compression plate (UCP), attached to a
nickel-chromium wire, was then positioned above the aggregate and
connected to a Cahn electrobalance. The weight of the UCP was
zeroed to establish a pre-compression UCP weight baseline. In order
to minimize adhesion of cell aggregates to the compression plates,
both the lower and upper plates were pre-coated with poly 2hydroxyethylmethacrylate (poly-HEMA), a polymeric material to
which cells do not adhere (Folkman and Moscona, 1978).
379
Compression was initiated by raising the LCP until the aggregate
became compressed against the UCP. Adjusting the height of the LCP
controlled different degrees of compression. The force with which the
aggregates resisted compression was monitored by the Cahn recording
electrobalance. Aggregate geometry was monitored through a 25×
Nikon dissecting microscope equipped with a CCD video camera and
connected to a Macintosh Power PC computer. Images of aggregates
were captured, digitized and their geometries were analyzed using
NIH Image software. Each aggregate was subjected to two different
degrees of compression, the second greater than the first.
Measurements of aggregate geometry (Fig. 1B) and the force of
resistance to the compressive force were then applied to the YoungLaplace equation (Davies and Rideal, 1963), producing numerical
values of apparent tissue surface tension (σ) (Eqn. 1):
σ=
 1
1 
+


πR32  R1 R2 
F
–1
.
(1)
Because R1, R2 and H can all be directly measured with greater
accuracy than R3, the latter parameter was calculated using Eqn. 2:
R3 = (R1 − R2) +
冪
H2
 .
2
(R2)2 − 
(2)
A true surface tension is one in which the measured σ is invariant
of the applied force, as would be expected of a true liquid surface
tension. Only those measurements of surface tension exhibiting this
behavior were used to calculate aggregate cohesivity.
The two likely material states to be considered as they apply to
tissue aggregates are liquidity and elasticity. The calculated surface
tension of a liquid aggregate, when subjected to two compressions,
the second greater than the first, will remain constant. By contrast, the
calculated surface tension of an elastic aggregate will obey Hooke’s
law and increase in proportion to the applied force. For example, we
have previously shown that when an elastic poly-acrylamide sphere is
subjected to two successive compressions, the calculated surface
tension increased in proportion to the applied force (Foty et al., 1996).
By contrast, several liquid systems have been described in which
surface tension remains constant irrespective of the applied force.
Such examples include compression of aggregates of embryonic
tissues (Davis et al., 1997; Foty et al., 1994; Foty et al., 1996),
HT1080 human fibrosarcoma (Foty et al., 1998), Lewis lung
carcinoma (Foty and Steinberg, 1997) and genetically engineered cells
(Duguay et al., 2002).
In order to confirm the validity of our TST measurements, we
calibrated our tissue surface tensiometer by compressing an air bubble
in culture medium and comparing the calculated surface tension with
that obtained by the de Noüy ring method (Davies and Rideal, 1963).
Surface tension measured by TST and by the de Noüy technique were
essentially identical, 39 dynes/cm and 42 dynes/cm, respectively (Foty
et al., 1994). On the basis of this result, we are confident that the
measured surface tensions of our transfected aggregates represent
absolute values of aggregate cohesivity.
Results
α5β1 integrin and N-cadherin expression by
transfectants
Stable populations of CHO-A5, CHO-Ncad and CHO-P3
(empty vector control) cells were created by bulk sorting and
antibiotic selection. Receptor expression was confirmed by
flow cytometry (Fig. 2A). A significant and comparable
uniform shift in fluorescence intensity represents α5 and Ncadherin expression in the CHO-A5 and CHO-Ncad cells,
380
Journal of Cell Science 116 (2)
A
Fig. 1. (A) Parallel plate compression device. The apparatus (not
drawn to scale) contains inner and outer rectangular Plexiglas
chambers. The outer chamber (OC) is connected to a thermostatted
circulating water pump and serves to regulate the temperature of the
inner chamber (IC). The lower assembly (LA) screws into the base of
the inner chamber. The position of its central core (CC), whose tip is
the lower compression plate (LCP), can be adjusted vertically by a
screw thread to set the distance between the two plates. The upper
compression plate (UCP) is a cylinder about 15 mm long suspended
from the arm of a Cahn recording electrobalance, labeled as B, by a
0.15 mm diameter nickel-chromium wire (NCW). Its position can be
adjusted horizontally to place the UCP directly above the LCP. Both
plates are coated with poly-HEMA before each use. During an
experiment, a spheroidal cell aggregate, labeled as A, is positioned
on the lower plate and raised until it contacts the upper plate.
Compression of the aggregate reduces the load measured by the
balance by an amount equal to the force acting upon the cell
aggregate. (B) Diagram of a liquid droplet compressed between two
parallel plates at shape equilibrium. R1 and R2 are the two primary
radii of curvature, at the droplet’s equator and in a plane through its
axis of symmetry, respectively. R3 is the radius of the droplet’s
circular area of contact with either compression plate. H is the
distance between the upper and lower compression plates. Because
R1, R2 and H can all be directly measured with greater accuracy than
R3, the latter parameter was calculated using Eqn. 2.
B
NCW
OC
IC
UCP
A
LCP
LA
CC
used as a comparison for the CHO-A5 aggregates for two
reasons. First, they represent a traditional cell-cell cohesion
system, and second, they formed spheroids and therefore their
cohesivity could be quantified by TST.
B
F
R2
R1
H
R3
respectively. Receptor expression was further confirmed by
western blot analysis. Strong 130 kDa and 120 kDa bands were
detected on the CHO-A5 and CHO-Ncad blots, respectively,
representing the published weights of α5 integrin and Ncadherin (Fig. 2B).
CHO-A5 cells formed compact spheroids
All cell lines were cultured in hanging drops and transferred to
shaker flasks as previously described. CHO-B2 and CHO-P3
cells did not form spherical aggregates but instead remained as
thick, flat sheets (Fig. 3A). CHO-A5 cells, however, formed
compact spheroids (Fig. 3B). CHO-Ncad cells also formed
spherical aggregates (Fig. 3C). CHO-Ncad aggregates were
α5β1 integrin confers stronger aggregate cohesivity than
N-cadherin
When subjected to TST, CHO-Α5 aggregates were found to
have a surface tension of 8.22±0.68 dynes/cm. These data were
generated from 18 aggregates each subjected to two successive
compressions, the second greater than the first (Table 1). By
contrast, CHO-Ncad cells were found to have a surface tension
of 3.14±0.20 dynes/cm (n=20), a value significantly lower
than that measured for CHO-A5 cell aggregates (P≤0.0001,
Student’s unpaired t-test, Table 1). These data suggest that
α5β1 integrin confers stronger cohesivity to 3D tissue
aggregates than does the expression of comparable levels of Ncadherin.
Assessment of liquid behavior of CHO-A5 and CHONcad aggregates
To confirm that our TST measurements represented true
surface tension, we demonstrated that σ was size and force
independent, as would be expected of a true liquid system
(Davies and Rideal, 1963). As surface tension is an inherent
physical property of a liquid, size independence is an absolute
requirement. Air bubbles should therefore have the same
surface tension irrespective of whether they are large or small
if measured under similar experimental conditions. Surface
tension of CHO-Ncad aggregates remained constant over a
threefold range of volumes. A linear regression analysis of σ
versus volume generated a correlation coefficient (r2) of 0.099
for CHO-Ncad aggregates, demonstrating no statistical
correlation between the two parameters. This was confirmed
α5β1 integrin mediates strong tissue cohesion
381
Fig. 2. (A) Flow cytometric analysis of CHO-Ncad and CHO-A5
transfectants. Note the significant shift in mean channel fluorescence
of positive cells, representing approximately a 100-fold increase in
protein expression. (B) Western blot analysis of CHO-Ncad and
CHO-A5 transfectants. 25 µg of protein from cell lysates of CHONcad and CHO-A5 cells were separated by SDS-PAGE, blotted to
PVDF and subjected to immunoblot analysis using appropriate
antibodies. Enhanced chemiluminescence detected a weak 130 kDa
band corresponding approximately to the known average molecular
weight for N-cadherin in the CHO-B2 parent line. A much stronger
130 kDa band is evident in the CHO-Ncad transfectant. α5 integrin
was undetectable in CHO-B2 cells but was strongly expressed by the
CHO-A5 cell line as represented by a 120 kDa band.
by performing a t-test to compare the confidence intervals. We
found that for an n of 20, the calculated r value (0.315) for
CHO-Ncad is below the critical value of r for testing P=0,
(0.378, α=0.05), proving the null hypothesis that the slope of
the linear regression is zero. This indicates that, for CHO-Ncad
aggregates, surface tension is size independent.
A similar analysis was generated for CHO-A5 aggregates
yielding an r2 value of 0.402, suggesting a possible weak
relationship between σ and volume (Fig. 4). A t-test comparing
the confidence intervals in this data set generated an r value
(0.634) greater than the critical value of r (0.400, n=18,
α=0.05), thus confirming a correlation between surface tension
and volume. We have previously shown that aggregate surface
tension can become variable as a function of time in culture
(Foty et al., 1996). To determine if this increased variability
was a possible explanation for the correlation described above,
we analyzed both CHO-Ncad and CHO-A5 TST data as a
function of time in culture. Whereas σ of CHO-Ncad remained
relatively constant from 4-6 days in culture and exhibited
relatively little scatter in the data, σ of CHO-A5 nearly doubled
over the same time period (Fig. 5). Linear regression analysis
of σ versus volume for CHO-A5 aggregates at 5 days in culture
generated an r2 value of 0.164 (Fig. 6). The calculated r value
(0.405, n=9) is below the critical value for r (0.582, α=0.05),
indicating that, for 5 days in culture, surface tension is size
independent. The 6 day data, however, were much more
scattered and yielded an r value of 0.741 (r2=0.550, n=6),
which is greater than the critical value for r (0.729), indicating
size dependence. We conclude from this analysis that, although
the CHO-A5 aggregates behave as liquids, the properties of
these aggregates can, over time, become more variable,
possibly owing to matrix deposition or a transition to elasticity.
In true liquid systems, σ must also be independent of the
applied force. An air bubble, compressed in liquid, should have
the same surface tension irrespective of the degree of
compression. Therefore, the ratio of the applied forces (F2/F1)
will be greater than 1, whereas the ratio of the measured surface
tensions (σ2/σ1) should be equal to 1. As Table 1 demonstrates,
σ of CHO-Ncad and CHO-A5 aggregates is indeed
independent of the applied force as surface tension measured
after two different compressive forces was not significantly
different based on a student’s t-test (CHO-A5, P=0.47; CHONcad, P=0.74; Table 1). Moreover, the ratio of F2/F1 relative
to σ2/σ1 was approximately 1.3:1, indicating that increasing the
Table 1. Aggregate surface tension values for aggregates of CHO-Ncad and CHO-A5
Cell line
CHO-Ncad
CHO-A5
σ1
(Dynes/cm)
σ2
(Dynes/cm)
σ1,2
(Dynes/cm)
σ2/σ1
(Dynes/cm)
F2/F1
3.06±0.21
7.86±0.66
3.20±0.23
8.58±0.72
3.14±0.20
8.22±0.68
1.07±0.03
1.12±0.05
1.33±0.05
1.30±0.04
For each aggregate type, the mean surface tension values measured in the first compression (σ1) are compared with the mean values measured in the second
compression (σ2). A student’s unpaired t-test was applied to evaluate the statistical significance of the differences between these two sets of means. σ1 and σ2
were not found to be statistically different, indicating that the mean surface tension values measured in the two sets of compressions were equal and that the
aggregates behaved as liquids. For both CHO-Ncad and CHO-A5 aggregates, the ratio of F2/F1 was greater than σ2/σ1, further confirming that the aggregates
behaved as liquid systems and not as elastic solids.
Fig. 3. Aggregate formation of CHO-B2, CHOA5 and CHO-Ncad cell lines. Hanging drop
cultures containing 2.5×106 cells/ml were
incubated for 2-3 days then transferred to shaker
flasks and incubated for another 2-3 days. Note
that CHO-B2 (A) aggregates only formed thick,
flat sheets, whereas aggregates of CHO-A5 (B)
and CHO-Ncad (C) formed spheres. Bar,
1.67 mm.
382
Journal of Cell Science 116 (2)
16
16
14
14
10
8
6
r2
= 0.099
σ (dynes/cm)
σ (dynes/cm)
12
r 2 = 0.402
12
10
8
6
4
4
2
2
0
0.05
0.1
0.15
0.2
0.25
r 2 = 0.550
0.3
Volume (mm3)
0
0.05
r 2 = 0.164
0.1
0.15
0.2
0.25
Volume (mm3)
Fig. 4. Linear regression analysis of surface tension versus volume
for CHO-Ncad and CHO-A5 aggregates. Aggregates of CHO-Ncad
(∆) ranging in volume from 0.1 to 0.3mm3 were subjected to TST.
Linear regression analysis of the data generated a correlation
coefficient (r2) of 0.099, indicating that no statistically significant
correlation exists between σ and volume. A similar analysis of CHOA5 (O) aggregates produced an r2 value of 0.402, suggesting a
possible weak relationship between σ and volume.
Fig. 6. Linear regression analysis of surface tension versus volume of
CHO-A5 aggregates after 5 (O) and 6 (∆) days in culture. The
surface tension of CHO-A5 aggregates cultured for 5 days remained
relatively constant over a threefold range in volume. Linear
regression analysis generated a correlation coefficient (r2) value of
0.164, indicating that surface tension at 5 days is size independent.
At 6 days in culture, however, a greater degree of scatter in the data
gave rise to an r2 value of 0.550, indicating size dependence.
force at compression 2 had no effect on surface tension (Table
1). This force independence further establishes our σ values as
representative of true aggregate cohesivity.
CHO-P3 and CHO-A5 cells failed to undergo calciumdependent aggregation (Fig. 7B). As expected, CHO-Ncad
cells aggregated in a calcium-dependent manner. Thus,
upregulation of N-cadherin expression or function was not
responsible for CHO-A5 aggregate formation.
CHO-A5 aggregate cohesion is cadherin independent
Transfection of α5 integrin into chick myoblasts has previously
been shown to upregulate expression of N-cadherin
(Huttenlocher et al., 1998). One possible explanation for the
observed cohesivity of the CHO-A5 aggregates is upregulation
of cadherin expression or function. CHO-B2 cells naturally
express low levels of N-cadherin. No upregulation of Ncadherin was detected in the CHO-Α5 cells by western blot
analysis, irrespective of whether cells were grown as 2D tissue
cultures or as 3D aggregates (Fig. 7A). Moreover, no
upregulation of N-cadherin function was noted as CHO-B2,
12
σ (dynes/cm)
10
8
6
4
2
0
4 days
5 days
6 days
Days in Culture
Fig. 5. Surface tension versus days in culture for CHO-Ncad and
CHO-A5 aggregates. Surface tension of CHO-Ncad aggregates
(white bars) remained relatively constant between 4 and 6 days in
culture. The surface tension of CHO-A5 aggregates (shaded bars),
however, increased between 4 and 6 days in culture. On each day, the
surface tension of the CHO-A5 aggregates was greater than that of
the CHO-Ncad aggregates.
Cohesivity of CHO-A5 aggregates is fibronectin
dependent
FN is the primary ligand of α5β1 integrin. As α5β1 integrin is
not known to form strong homophilic interactions, we
hypothesized that the strong cohesivity observed in the CHOA5 tissue aggregates was due to α5β1 integrin’s interaction
with FN. To test this hypothesis, we generated hanging drop
cultures of CHO-A5 cells in FN-depleted medium. Such
cultures produced small irregularly shaped, loosely associated
clusters of cells (Fig. 8A). The normal physiological range of
plasma FN is 100-1000 µg/ml (Ouaissi et al., 1986; Swisher
and Rannels, 1997). When 300 µg/ml of plasma FN was added
to the FN-depleted medium, the ability of CHO-A5 cells to
form compact spheroids was restored (Fig. 8B). These data
suggest that FN plays a role in α5β1-integrin-mediated
aggregate formation and compaction. As CHO cells are known
to secrete small amounts of FN (Rajaraman et al., 1980), we
postulated that this endogenous FN production could
contribute to the basal level of aggregation of CHO-A5 cells
observed in FN-depleted medium.
We further confirmed the role of FN in α5β1-integrinmediated cohesivity by utilizing the CHO-α5 cell line that
expresses α5β1 integrin but does not produce endogenous FN
(Sechler et al., 1996). Hanging drop cultures of these cells did
not aggregate in the absence of exogenous FN (Fig. 9A)
but progressively formed more compact sheets as FN
concentration was increased from 3-100 µg/ml (Fig. 9B-D).
The addition of 300 µg/ml of FN resulted in the formation of
spheroids (Fig. 9E). Spheroid formation was dependent purely
on integrin expression since this subclone does not express
cadherin and does not aggregate in a calcium-dependent
α5β1 integrin mediates strong tissue cohesion
383
Fig. 7. (Top panel) Western blot analysis of N-cadherin expression of
CHO-B2 and α5 integrin transfected cell lines grown as 2D or 3D
cultures. Cell lysates were prepared from cells grown on tissue
culture plastic and cells grown as 3D spheroids. N-cadherin was
detected by immunoblot analysis. Note the presence of a 130 kDa
band, corresponding to the published molecular weight of Ncadherin. Note also that transfection of CHO cells with α5 integrin
did not result in increased N-cadherin expression irrespective of
whether cells were grown in conventional tissue culture or as
spheroids. (Bottom panel) Assessment of cadherin function by fast
aggregation assay. Cells from near-confluent plates of CHO-B2
(A,B), CHO-P3 (C,D), CHO-A5 (E,F) and CHO-Ncad (G,H) were
detached by trypsin/calcium (0.05% trypsin/2 mM CaCl2) treatment.
Cells were stained with the membrane intercalating dye PKH-2 and
resuspended at a concentration of 1×106 cells/ml in 3 ml of either
calcium/magnesium-free HBSS (A,C,E,G) or HBSS with 2 mM Ca2+
(B,D,F,H), transferred to shaking flasks and placed on a gyratory
shaker at 37°C and 120 rpm. Aggregation was monitored 1 hour later
and imaged by fluorescence microscopy. Note that only the CHONcad cell line aggregated in a calcium-dependent manner (G,H).
cohesivity was dependent upon FN. This could be due to two
possible mechanisms: α5β1 integrin specifically binding to FN
at the cell-binding domain or, alternatively, an interaction
between FN and other cell surface receptors, such as
syndecans. To define the mechanism by which FN promotes
aggregate formation, we incubated the CHO-A5 cells in
hanging drop culture with an RGD-blocking peptide. This
peptide competes for the ligand-binding pocket of α5β1
integrin and specifically interferes with its ability to bind FN
(Akiyama, 1996). Hanging drop cultures formed in the
presence of RGD peptide produced only dispersed cell clusters
(Fig. 10). These data support the role of a specific interaction
between α5β1 integrin and FN in mediating cohesivity. No
change was observed with the addition of RGD peptide to
hanging drop cultures of the CHO-P3 cells in FN depleted
medium (data not shown).
manner (E.E.R., S.A.C. and R.A.F., unpublished). Moreover,
de novo N-cadherin expression was not detected in 3D
spheroids (data not shown). These data demonstrate that α5β1integrin-mediated cohesivity of the CHO-A5 aggregates is
dependent upon FN.
CHO-A5 aggregate formation and compaction requires
a specific interaction between α5β1 integrin and
fibronectin
FN is a complex, dimeric ECM protein that binds both cells
and other ECM proteins including heparin, fibrin and collagen.
α5β1 integrin is known to interact specifically with an RGD
site located at type III repeat 10 in the cell-binding domain.
We earlier demonstrated that the α5β1-integrin-mediated
Dimeric fibronectin is required for aggregate formation
and compaction
FN exists as a dimer, each arm containing a single cell-binding
site. The dimeric structure of FN is essential for its assembly
into a fibrillar matrix. We hypothesized that this dimeric
structure was a necessary component in the linkage of adjacent
cells and that monomeric FN would be unable to provide such
a linkage. To test this hypothesis, we used a recombinant FN
monomer that lacks the dimerization site but that contains all
the structural elements necessary for FN matrix assembly.
CHO-α5 cell hanging drops were cultured in FN-depleted
medium with 100 µg/ml of either monomeric recombinant FN
or dimeric rat plasma FN. Aggregates failed to form in the
presence of the recombinant monomer (Fig. 11). These data
demonstrate that the dimeric structure of FN is essential for
α5β1-integrin-mediated aggregate formation and is required to
link adjacent cells together.
Discussion
Cell-cell and cell-substratum adhesions involving cadherins
and integrins have been widely studied and have been
demonstrated to be vital contributors to normal and abnormal
384
Journal of Cell Science 116 (2)
Fig. 8. FN-dependent aggregation of CHO-A5 cells. CHO-A5 cells
secrete low levels of endogenous FN. When cultured in hanging
drops in FN-depleted tissue culture medium, cells formed loose
sheets (A). Addition of 300 µg/ml of exogenous FN resulted in
compact spheroid formation (B). Bar, 1.0 mm.
biological processes such as embryonic development, wound
healing and malignant invasion. Cadherins have long been
considered the primary mediators of tissue cohesivity, whereas
integrins have traditionally been considered as mediators of
cell-substratum interactions. In this manuscript we show that
α5β1 integrin is also capable of conferring strong cohesivity
to 3D cellular aggregates. We demonstrate that interaction of
α5β1 integrin with its ligand, FN, confers greater cohesivity to
Fig. 9. Dose-dependent
aggregation of FN-null CHO-α5
cells. CHO-α5 cells express high
levels of α5 integrin but do not
express N-cadherin or secrete FN.
Cells were cultured in hanging
drops either in the absence of FN
(A) or in 3 (B), 30 (C), 100 (D) or
300 (E) µg/ml rat plasma FN. Note
the dose-dependent aggregation
and compaction of CHO-α5 cells
in response to the addition of
exogenous FN. Bar, 0.5 mm.
Fig. 10. Aggregate formation and compaction of CHO-A5 cells in
the presence of RGD peptide. CHO-A5 cells secrete low levels of
endogenous FN. Culturing these cells in FN-depleted medium
resulted in formation of cellular sheets (A). With the addition of
100 µM RGD peptide CHO-A5 cells failed to form aggregates (B).
Bar, 1.0 mm.
an aggregate of cells than does N-cadherin, a typical cell-cell
adhesion system.
This observation was made possible due, in large part, to
application of TST, a method that measures the strength of
intercellular cohesion of multicellular aggregates under
physiological conditions. The main advantage of TST is that
this method measures intercellular adhesive intensity within
3D aggregates and therefore more accurately mimics in vivo
cellular interactions than do conventional 2D assays. Cells in
tissues establish contacts and assume shapes more reminiscent
of 3D foams than of the typical ‘fried egg’ configurations
adopted when cells are placed on 2D substrate-coated surfaces.
Consequently, TST is able to accurately measure the effect of
both direct intercellular cohesion (as mediated by cadherins),
as well as indirect intercellular contacts mediated by the ECM
in which cells are embedded.
It is known that some integrins, including α1β2, α4β7 and
αMβ2, recognize integral membrane proteins of the IgG
superfamily such as ICAM-1, ICAM-2 and VCAM-1 and are
thus able to mediate direct cell-cell adhesion (Hynes, 1992).
Others, such as integrin αIIbβ3, are almost exclusively
responsible for platelet-platelet interaction through GPIb/V/IX
(Schoenwaelder et al., 2000). Typically, however, such
interactions are not known to contribute to tissue cohesivity.
The more classically defined ECM binding heterodimers, such
as α5β1, function principally to promote the adhesive events
required for cell motility and tension-generated matrix
remodeling. However their contribution to the overall
cohesivity of a tissue has never been rigorously quantified. This
work is the first to rigorously quantify the cohesivity imparted
by the interaction of α5β1 integrin with FN in 3D tissue-like
aggregates.
α5β1 integrin is unique amongst the FN-binding integrins in
that it is the only integrin that naturally assembles FN into
a matrix. Studies have defined many of the structural
requirements for FN matrix assembly. These include FN’s
dimer structure, the N-terminal assembly domain and FNbinding sites in the first two type III repeats (Schwarzbauer,
1991; Sechler et al., 2001). Integrin binding to the RGD
sequence in the cell-binding domain is also an essential
requirement for matrix assembly (Sechler et al., 1996). FNintegrin interactions promote intermolecular association
between the FN dimers, leading to the formation of fibrils. The
fact that FN monomers failed to support aggregate formation
supports the concept that FN matrix assembly may contribute
to aggregate cohesivity by creating a scaffold, or an organized
α5β1 integrin mediates strong tissue cohesion
Fig. 11. Aggregate formation and compaction of CHO-α5 cells in the
presence of recombinant monomeric FN. CHO-α5 cells were
cultured in FN-depleted medium with 100 µg/ml of either
monomeric (A) or dimeric (B) rat plasma FN. Note that aggregates
failed to form in the presence of FN monomers (A). Bar, 1.0 mm.
3D matrix, which functionally links the cells to each other and
promotes force generation. In the absence of solid structural
support, the net effect of FN-mediated force generation is
tissue compaction and remodeling, processes that are essential
for a variety of biological functions.
The demonstration that the α5β1-integrin–FN interaction
confers strong intercellular cohesivity in 3D cellular aggregates
significantly impacts several areas of interest. The study of
malignant invasion, for example, represents a model system in
which the forces of adhesion and cohesion influence cellular
behavior (Behrens, 1993; Foty et al., 1998; Foty et al., 1994;
Foty and Steinberg, 1997; Shiozaki et al., 1996; Steinberg and
Foty, 1997; Tlsty, 1998; Zhou et al., 2000). Normal E-cadherin
expression and function is considered to be important in
maintaining tumor integrity whereas overexpression of various
integrins has been associated with increased potential for
invasion and migration. α5β1 integrin is often found to be
downregulated in metastatic cancer, and overexpression has
been shown to rescue a transformed phenotype (Giancotti and
Ruoslahti, 1990). These observations diverge from the
commonly held view of integrins as metastasis-promoting
molecules and suggest a possible role for α5β1 integrin as a
potential invasion suppressor molecule, much as has been
reported for E-cadherin. Being able to quantify the contribution
of integrins in regulating tumor cohesivity will provide
information necessary for the development of strategies aimed
at promoting intercellular cohesivity, thus discouraging
dissemination of cancer cells.
The α4, α5 and β1 integrin subunits have been shown to be
essential for embryonic development. The knockout of these
genes in mouse embryos always leads to embryonic lethality
(Beauvais-Jouneau and Thiery, 1997). Davis et al. have shown
that the cohesivity of the germ layers in amphibian gastrulae
correlate perfectly with their spatial position (Davis et al.,
1997). Injection of RGD peptide into amphibian blastulae has
been shown to block gastrulation by disrupting cell interactions
with FN and preventing formation of the meshwork of FN
fibrils involved in migration (Boucaut et al., 1984). If α5β1integrin–FN interaction is indeed capable of conferring
cohesivity onto tissues, particularly very early in development
when embryos are essentially aggregate-like, then it is possible
that α5β1-integrin–FN interactions also have the capacity to
specify the relative cohesivity of cells within the gastrula. FNmediated compaction has also been shown to be important
later in development. For instance, Downie and Newman
demonstrated that a correlation exists between FN secretion
and pre-cartilage mesenchymal condensation during wing and
385
leg bud development in chick embryos (Downie and Newman,
1995), with high FN secretion correlating with compact and
spheroidal condensation. Modulation of tissue cohesivity
through differential cadherin expression during tissue
development provides one possible mechanism of tissue
assembly (Gumbiner, 2000; Steinberg and McNutt, 1999).
Modulation of α5β1 integrin and FN provides yet another
potential cell-cell cohesion mechanism available to the embryo
during the self-assembly process. Accordingly, molecular
pathways regulating either the expression or function of
α5β1 integrin or FN have the capacity to function as
potential morphogens, modulating tissue behavior, perhaps
independently of cadherins.
During wound healing, alterations in integrin expression
coincide with the increase in cell migration required for the
repair process. Cells in the wound milieu alter their integrin
expression to interact with ligands of the newly formed
provisional ECM, whose primary structural components
include fibrinogen and FN. The α5β1-integrin–FN interaction
has also been shown to be vital for retraction of 3D FN-fibrin
clot matrices (Corbett and Schwarzbauer, 1999), a process
crucial to early wound healing and tissue remodeling. It is
thought that engagement of α5β1 integrin with FN generates
the tractional force required for clot retraction. We propose that
α5β1-FN-mediated intercellular cohesivity also contributes to
clot retraction, in much the same way as the apparent
‘retraction’ or compaction observed for CHO-A5 aggregates in
response to increased concentrations of FN (Fig. 9).
The information presented in this manuscript is, to our
knowledge, the first quantitative demonstration that the
interaction between α5β1 integrin and FN can give rise to
strong intercellular cohesivity of 3D aggregates. Our data bring
into question the undisputed role of cadherins as the prime
mediators of tissue cohesivity and provide an alternative or
additional mechanism by which the balance of cohesive and
adhesive forces can be used to alter tissue behavior. Further
understanding of these forces is important, for example, in such
recent developments as the use of ligand-coated microparticles
to compact complex cellular aggregates for tissue engineering
and biomedical applications (Dai et al., 1994; Saltzman and
Olbricht, 2002). The balance between cell-cell cohesion and
cell-substratum adhesion, the regulatory mechanisms involved
in maintaining this balance, and our ability to quantify them
are important for developing unifying principles governing
such processes as embryonic development, wound healing,
malignant invasion, tissue regeneration and tissue engineering.
References
Akiyama, S. K. (1996). Integrins in cell adhesion and signaling. Hum. Cell 9,
181-186.
Beauvais-Jouneau, A. and Thiery, J. P. (1997). Multiple roles for integrins
during development. Biol. Cell 89, 5-11.
Behrens, J. (1993). The role of cell adhesion molecules in cancer invasion and
metastasis. Breast Cancer Res. Treat. 24, 175-184.
Boucaut, J. C., Darribere, T., Poole, T. J., Aoyama, H., Yamada, K. M.
and Thiery, J. P. (1984). Biologically active synthetic peptides as probes
of embryonic development: a competitive peptide inhibitor of fibronectin
function inhibits gastrulation in amphibian embryos and neural crest cell
migration in avian embryos. J. Cell Biol. 99, 1822-1830.
Corbett, S. A. and Schwarzbauer, J. E. (1999). Requirements for
alpha(5)beta(1) integrin-mediated retraction of fibronectin-fibrin matrices.
J. Biol. Chem. 274, 20943-20948.
Corbett, S. A., Lee, L., Wilson, C. L. and Schwarzbauer, J. E. (1997).
386
Journal of Cell Science 116 (2)
Covalent cross-linking of fibronectin to fibrin is required for maximal cell
adhesion to a fibronectin-fibrin matrix. J. Biol. Chem. 272, 24999-25005.
Dai, W., Belt, J. and Saltzman, W. M. (1994). Cell-binding peptides
conjugated to poly(ethylene glycol) promote neural cell aggregation.
Biotechnology (NY) 12, 797-801.
Davies, J. T. and Rideal, E. K. (1963). Interfacial Phenomena. New York:
Academic Press.
Davis, G. S., Phillips, H. M. and Steinberg, M. S. (1997). Germ-layer surface
tensions and “tissue affinities” in Rana pipiens gastrulae: quantitative
measurements. Dev. Biol. 192, 630-644.
Downie, S. A. and Newman, S. A. (1995). Different roles for fibronectin in
the generation of fore and hind limb precartilage condensations. Dev. Biol.
172, 519-530.
Duguay, D., Foty, R. A. and Steiberg, M. S. (2002). Cadherin mediated cell
adhesion and tissue segregation: qualitative and quantitative determinants.
Dev. Biol. (in press).
Eckes, B., Zigrino, P., Kessler, D., Holtkotter, O., Shephard, P., Mauch, C.
and Krieg, T. (2000). Fibroblast-matrix interactions in wound healing and
fibrosis. Matrix Biol. 19, 325-332.
Folkman, J. and Moscona, A. (1978). Role of cell shape in growth control.
Nature 273, 345-349.
Foty, R. A. and Steinberg, M. S. (1997). Measurement of tumor cell cohesion
and suppression of invasion by E- or P-cadherin. Cancer Res. 57, 50335036.
Foty, R. A., Forgacs, G., Pfleger, C. M. and Steinberg, M. S. (1994). Liquid
properties of embryonic tissues: Measurement of interfacial tensions. Phys.
Rev. Lett. 72, 2298-2301.
Foty, R. A., Pfleger, C. M., Forgacs, G. and Steinberg, M. S. (1996). Surface
tensions of embryonic tissues predict their mutual envelopment behavior.
Development 122, 1611-1620.
Foty, R. A., Corbett, S. A., Schwarzbauer, J. E. and Steinberg, M. S.
(1998). Dexamethasone up-regulates cadherin expression and cohesion of
HT-1080 human fibrosarcoma cells. Cancer Res. 58, 3586-3589.
Fujimori, T., Miyatani, S. and Takeichi, M. (1990). Ectopic expression of
N-cadherin perturbs histogenesis in Xenopus embryos. Development 110,
97-104.
Giancotti, F. G. and Ruoslahti, E. (1990). Elevated levels of the alpha 5 beta
1 fibronectin receptor suppress the transformed phenotype of Chinese
hamster ovary cells. Cell 60, 849-859.
Goldmann, W. H. (2000). Kinetic determination of focal adhesion protein
formation. Biochem. Biophys. Res. Commun. 271, 553-557.
Goldstein, A. S. and DiMilla, P. A. (2002). Effect of adsorbed fibronectin
concentration on cell adhesion and deformation under shear on hydrophobic
surfaces. J. Biomed. Mater. Res. 59, 665-675.
Gumbiner, B. M. (2000). Regulation of cadherin adhesive activity. J. Cell
Biol. 148, 399-404.
Huttenlocher, A., Lakonishok, M., Kinder, M., Wu, S., Truong, T.,
Knudsen, K. A. and Horwitz, A. F. (1998). Integrin and cadherin synergy
regulates contact inhibition of migration and motile activity. J. Cell Biol.
141, 515-526.
Hyafil, F., Babinet, C. and Jacob, F. (1981). Cell-cell interactions in early
embryogenesis: a molecular approach to the role of calcium. Cell 26, 447454.
Hynes, R. O. (1992). Integrins: versatility, modulation, and signaling in cell
adhesion. Cell 69, 11-25.
Koo, L. Y., Irvine, D. J., Mayes, A. M., Lauffenburger, D. A. and Griffith,
L. G. (2002). Co-regulation of cell adhesion by nanoscale RGD organization
and mechanical stimulus. J. Cell Sci. 115, 1423-1433.
Okegawa, T., Li, Y., Pong, R. C. and Hsieh, J. T. (2002). Cell adhesion
proteins as tumor suppressors. J. Urol. 167, 1836-1843.
Ouaissi, M. A., Neyrinck, J. L. and Capron, A. (1986). Development of a
competitive radioimmunoassay for human plasma fibronectin. Int. Arch.
Allergy Appl. Immunol. 81, 75-80.
Rajaraman, R., Sunkara, S. P. and Rao, P. N. (1980). Morphological reverse
transformation of Chinese hamster ovary (CHO) cells and surface
fibronectin. Cell Biol. Int. Rep. 4, 897-906.
Ryan, P. L., Foty, R. A., Kohn, J. and Steinberg, M. S. (2001). Tissue
spreading on implantable substrates is a competitive outcome of cell-cell vs.
cell-substratum adhesivity. Proc. Natl. Acad. Sci. USA 98, 4323-4327.
Saltzman, W. M. and Olbricht, W. L. (2002). Building drug delivery into
tissue engineering. Nat. Rev. Drug Discov. 1, 177-186.
Schoenwaelder, S. M., Yuan, Y. and Jackson, S. P. (2000). Calpain
regulation of integrin alpha IIb beta 3 signaling in human platelets. Platelets
11, 189-198.
Schwarzbauer, J. E. (1991). Identification of the fibronectin sequences
required for assembly of a fibrillar matrix. J. Cell Biol. 113, 1463-1473.
Sechler, J. L., Takada, Y. and Schwarzbauer, J. E. (1996). Altered rate of
fibronectin matrix assembly by deletion of the first type III repeats. J. Cell
Biol. 134, 573-583.
Sechler, J. L., Corbett, S. A. and Schwarzbauer, J. E. (1997). Modulatory
roles for integrin activation and the synergy site of fibronectin during matrix
assembly. Mol. Biol. Cell 8, 2563-2573.
Sechler, J. L., Rao, H., Cumiskey, A. M., Vega-Colon, I., Smith, M. S.,
Murata, T. and Schwarzbauer, J. E. (2001). A novel fibronectin binding
site required for fibronectin fibril growth during matrix assembly. J. Cell
Biol. 154, 1081-1088.
Shiozaki, H., Oka, H., Inoue, M., Tamura, S. and Monden, M. (1996). Ecadherin mediated adhesion system in cancer cells. Cancer 77, 1605-1613.
Steinberg, M. S. and Foty, R. A. (1997). Intercellular adhesions as
determinants of tissue assembly and malignant invasion. J. Cell Physiol.
173, 135-139.
Steinberg, M. S. and McNutt, P. M. (1999). Cadherins and their connections:
adhesion junctions have broader functions. Curr. Opin. Cell Biol. 11, 554560.
Swisher, J. W. and Rannels, D. E. (1997). Assembly of exogenous fibronectin
into type II cell extracellular matrix. Am. J. Physiol. 272, L908-L915.
Takeichi, M. (1977). Functional correlation between cell adhesive properties
and some cell surface proteins. J. Cell Biol. 75, 464-474.
Tlsty, T. D. (1998). Cell-adhesion-dependent influences on genomic instability
and carcinogenesis. Curr. Opin. Cell Biol. 10, 647-653.
Zhang, Z., Morla, A. O., Vuori, K., Bauer, J. S., Juliano, R. L. and
Ruoslahti, E. (1993). The alpha v beta 1 integrin functions as a fibronectin
receptor but does not support fibronectin matrix assembly and cell migration
on fibronectin. J. Cell Biol. 122, 235-242.
Zhou, J., Sargiannidou, I. and Tuszynski, G. P. (2000). The role of adhesive
proteins in the hematogenous spread of cancer. In Vivo 14, 199-208.