Experimental Cell Research 274, 169 –177 (2002)

doi:10.1006/excr.2001.5463, available online at http://www.idealibrary.com on

Effects of Collagen IV on Neuroblastoma Cell Matrix-Related Functions

Athina K. Tzinia,* ,1 Paraskevi V. Kitsiou,* Argiris A. Talamagas,* Angelique Georgopoulos,†
and Effie C. Tsilibary*
*Institute of Biology, National Center for Scientific Research “Demokritos,” 15310 Agia Paraskevi, Athens, Greece; and
†Veterans Administration Medical Center, Minneapolis, Minnesota 55455

differentiation [1, 2]. Collagen IV (Col IV) not only

Integrin-mediated interactions with collagen IV and forms the main structural framework of most base-
its domains were examined in a human neuroblastoma ment membranes, but also has the ability to promote
cell line (SK-N-SH). By adhesion assays we demon- cell adhesion of various cell types [3, 4], and it is
strated that neuroblastoma cells bound to solid-phase
important for axonal regeneration [5]. Col IV is a trim-
intact collagen IV and synthetic cell-binding peptide
mer composed of three monomeric chains with the
HEP-III, derived from the collagenous part of the mol-
ecule, but not to the main noncollagenous NC1 domain
usual composition (␣1) 2(␣2) [6], and four isoform
or to the synthetic cell-binding peptide HEP-I, derived chains have been described [7]. Intact Col IV, its main
from this domain. Monoclonal antibodies against ␤1, noncollagenous NC1 domain, and synthetic peptides
␣3, and ␣v␤3 integrins resulted in inhibition of cell representing sequences from the (␣1) NC1 and (␣1)
binding to collagenous substrates by 95, 30, and 35%, collagenous domains (HEP-I and HEP-III, respec-
respectively. By flow cytometry and immunoblotting it tively) have been shown to interact with several cell
was shown that culture of SK-N-SH cells on collagen IV types mainly via integrin receptors [3, 8, 9]. Integrins
resulted in alteration in the expression of major neu- are the major family of cell-surface receptors, ex-
roblastoma cell integrins. Binding to collagen IV in- pressed by all cell types. These transmembrane pro-
duced the expression and activation of matrix metallo- teins consist of ␣ and ␤ heterodimers. It is considered
proteinases A and B (MMP-2, MMP-9), with a that each integrin has distinct ligand binding capacity
concomitant increase at the protein level of tissue-spe- and most known integrins recognize various extracel-
cific inhibitors of metalloproteinases (TIMP-1, TIMP-2). lular matrix proteins, such as fibronectin, laminin, and
Finally, the expression of MMP-2 was significantly up- collagens [10 –12]. In the nervous system, integrins are
regulated by anti-␣3␤1 antibodies, whereas ligation of the major family of adhesion receptors, which mediate
anti-␣v␤3 antibodies resulted in a modest down-regula-
the differentiating effects of ECM components of neu-
tion of MMP-2. Our results indicate that the presence of
ronal cells during development [13]. These integrin-
collagen IV modulates the expression of integrins, which
are used for binding to this glycoprotein, and MMP-2
mediated interactions are intimately involved in the
secreted by SK-N-SH cells. © 2002 Elsevier Science (USA) regulation of cell behavior and gene expression, for
Key Words: collagen type IV; adhesion; neuroblas- example the expression of matrix metalloproteinases
toma cells; integrins; MMP-2. (MMPs) [14, 15].
MMPs are a family of zinc-dependant proteolytic
enzymes involved in the remodeling of the ECM in a
INTRODUCTION variety of physiological and pathological processes, in-
creasingly being implicated in the pathogenesis of sev-
Extracellular matrix (ECM), 2 including basement eral diseases of the central nervous system (CNS) such
membranes, is a complex network of interacting mole- as multiple sclerosis and Alzheimer’s disease [16]. Ac-
cules such as collagens, fibronectin, and laminin. This cording to their substrate specificity they have been
matrix, in addition to acting as a scaffold that stabi- classified to gelatinases, stromelysines, and collag-
lizes the physical structure of tissues, regulates vital enases [17, 18]. The 92-kDa gelatinase B (MMP-9) and
cell processes, including cell migration, growth, and the 72-kDa gelatinase A (MMP-2) efficiently degrade
native collagens such as types IV and V, fibronectin,
1
To whom correspondence and reprint requests should be ad- entactin, and elastin [19, 20]. The proteolytic activity of
dressed. Fax: 01-6511767. E-mail: atzin@mail.demokritos.gr. MMP-9 and MMP-2 is in part regulated by the action of
2
Abbreviations used: MMPs, matrix metalloproteinases; TIMPs,
tissue inhibitors of metalloproteinases; ECM, extracellular matrix;
tissue inhibitors of metalloproteinases TIMP-1 and
PBS, phosphate-buffered saline; FACS, fluorescence-activated cell TIMP-2, respectively [21]. One of the mechanisms in-
sorting. volved in the regulation of the expression of MMPs is
169 0014-4827/02 $35.00
© 2002 Elsevier Science (USA)
All rights reserved.
170 TZINIA ET AL.

via integrin-mediated signaling [22, 23]. Regulation of 20. Incubations with peroxidase-conjugated anti-rabbit immuno-
the expression of MMPs, and the expression of their globulins and detection of bound peroxidase activity were carried out
as described in the ECL-blotting detection system (Amersham). Im-
specific inhibitors (TIMPs), represents a mechanism ages of Western blots were analyzed using image-processing soft-
for controlling ECM turnover, in addition to the ex- ware (Bioprofil Vilber Loumart).
pression and secretion of ECM components by the cells. Flow cytometry. Expression of integrin subunits was evaluated
The aim of the present study was to examine the by indirect immunofluorescence staining and flow cytometry. Briefly,
effect of Col IV, an important component for neuronal cells were released from confluent monolayer cultures (either in the
presence or in the absence of Col IV) with trypsin, washed, and
cell differentiation and axonal regeneration, on the
resuspended in FACS buffer (2% fetal calf serum, 0.02% sodium
expression of specific integrin subunits, which are azide in PBS); 1 ⫻ 10 6 cells were incubated with the primary anti-
present on the surface of SK-N-SH cells, and on the body for 45 min on ice and washed twice with FACS buffer. Anti-
expression of proteins involved in the degradative mouse IgG conjugated with fluorescein isothiocyanate was then
pathways of ECM. added in a total volume of 0.1 ml FACS buffer, and cells were
incubated for 45 min on ice. After washing twice with FACS buffer
the cells were fixed with 1% formaldehyde in PBS. Analysis was
MATERIALS AND METHODS performed using CELL QUEST software on a FACScan (Becton–
Dickinson). Positive fluorescence was determined on a four-decade
Cell cultures and treatments. The human neuroblastoma cell line log scale and fluorescence intensity (log F1) was expressed as the
SK-N-SH was cultured in minimum essential medium Eagle with 2 mean channel number of 10,000 cells.
mM L-glutamine adjusted to contain 0.1 mM nonessential amino Cell adhesion assays. Ninety-six-well plates were coated with 50
acids, 1 mM sodium pyruvate, 1.5 g/L sodium bicarbonate, 10% fetal ␮l/well of increasing concentrations of various substrates (0.3–100
bovine serum, and antibiotics. ␮g/ml) in PBS for 24 h at 29°C to complete dryness, followed by a 2-h
To estimate the expression of MMPs and TIMPs in the presence of incubation at 37°C in 0.5 mg/ml bovine serum albumin in PBS to
Col IV, 2 ⫻ 10 6 cells were cultured in 58-cm 2 tissue culture dishes cover uncoated sites on plastic. Neuroblastoma cells were cultured
precoated with 50 ␮g/ml Col IV (⬃1 ␮g/cm 2) for 48 h. The medium until they reached ⬃80% confluency. The cells were metabolically
was then removed, the dishes were washed three times with serum- labeled overnight with 0.15 mCi of [ 35S]methionine per T-25-cm 2
free medium, and cells were incubated for 24 h in 10 ml serum-free flask. Cells were released with trypsin/EDTA, washed, resuspended
medium. At the end of the 24-h incubation period, the conditioned in binding buffer (Dulbecco’s modified Eagle’s medium, 2 mg/ml
media were transferred to 15-ml conical tubes, centrifuged to remove bovine serum albumin, 20 mM Hepes, pH 7.5), and plated at a
cells and cell fragments, supplemented to 1 mM Na 2EDTA and 0.02% density of 5000 cells/well. Adhesion was allowed to occur for 30 – 60
sodium azide, and stored at ⫺20°C for further use. min at 37°C. At the end of the incubation period, nonadherent cells
Antibodies and substrates. The following antibodies were used in were removed by washing with binding buffer, the adhering cells
this study: Rabbit anti-human polyclonal antibodies to the integrin were lysed and transferred to scintillation vials, and radioactivity
subunits ␣2, ␣3, ␣5, ␣v, ␤1, and ␤3 were purchased from Chemicon was measured with a ␤-counter. Percentage adhesion of cells was
International. Monoclonal antibodies against ␣3 (P1B5), ␤1 (P5D2), calculated as (cpm of cells adhered per well/cpm of 5000 cells) ⫻ 100.
␣2 (P1E6), and ␣v␤3 (LM609) integrins were also purchased from For competition assays, cells were preincubated in suspension
Chemicon International. Well-characterized polyclonal antibodies with appropriate concentrations of monoclonal antibodies against
Ab45 and Ab110 [24, 25] against the collagenases MMP-2 and human integrins for 30 min at 4°C and then plated to wells and
MMP-9, respectively, and antibodies against their inhibitors TIMP-1 allowed to adhere for 30 min. Inhibition of adhesion was calculated
and TIMP-2 were kindly provided by Dr. Stetler-Stevenson (NCI, as a percentage of the cell adhesion in the absence of antibodies. Cell
National Institute of Health, Bethesda, MD). adhesion and competition experiments were done in six replicates
Intact Col IV was isolated from the Engelbreth–Holm–Swarm and repeated a minimum of three times.
tumor system using previously described techniques [26]. The car- Zymography. Gelatin zymography was performed as previously
boxyl terminal noncollagenous domain 1 (NC1) was prepared by described [30]. Briefly, aliquots of each sample of conditioned me-
digestion with bacterial collagenase as previously described [27]. dium were subjected to SDS–PAGE under nonreducing conditions in
Peptides derived from the sequence of either NC1 (HEP-I) or triple 10% polyacrylamide gels containing 0.1% gelatin. The volume of
helix-rich domains (HEP-III) were synthesized according to the conditioned medium loaded per lane was adjusted according to the
method of Barany and Merrifield and purified through an HPLC cell number obtained at harvest. After electrophoresis, SDS was
reverse-phase column as described previously [26]. removed from the gel by washing in 50 mM Tris–HCl, pH 7.5, 5 mM
Total protein extraction. Cells grown on culture dishes either CaCl 2, 1 ␮M ZnCl 2, 2.5% Triton X-100, and 0.02% NaN 3 for 3⫻ 30
precoated or not with 50 ␮g/ml Col IV (⬃1 ␮g/cm 2) were collected by min at room temperature. The gel was then incubated in the same
treatment with trypsin/EDTA and lysed in PBS, pH 7.4, containing buffer excluding Triton X-100 for 48 h at 37°C. After staining with
1 mM phenylmethylsulfonyl fluoride, 1 mM N-ethylmalemide, 1% Coomassie brilliant blue R-250 for 3 h and destaining in water,
Triton X-100, 1 mM CaCl 2, and a cocktail of protease inhibitors gelatin-degrading enzymes were identified by their ability to clear
(Sigma P8340) for 30 min at 4°C. Insoluble material was removed by the gelatinous substrate at their respective molecular weights.
centrifugation and the supernatant was stored at ⫺20°C. Protein Incubation of cells with antibodies. Cells were detached from
estimation was done by the method of Bradford (Pierce). confluent monolayer cultures by treatment with trypsin/EDTA, col-
Immunoblotting. Electrophoresis in the presence of SDS was per- lected by centrifugation, resuspended in culture medium (3 ⫻ 10 6
formed according to the method of Laemmli [28] on either 7.5 or 10% cells/ml), preincubated with anti-␣3, -␣2, and -␣v␤3 monoclonal an-
polyacrylamide gels, under reducing or nonreducing conditions, as tibodies at a concentration of 40 ␮g/ml each and anti-␤1 (0.5 ␮g/ml)
indicated in the figure legends. The resolved proteins were subse- for 30 min at 37°C, and then diluted 10-fold and cultured for 24 h on
quently transferred to Hybond–ECL nitrocellulose membrane (Am- 24-well plates precoated with 50 ␮g/ml Col IV. Control cells were
ersham) as described by Towbin et al. [29]. Blots were saturated for cultured in the absence of antibodies. The medium was then re-
2 h at room temperature with 5% nonfat milk in Tris-buffered saline, moved, the plates were washed three times with serum-free medium,
0.1% Tween 20 and incubated overnight at 4°C with the appropriate and cells were incubated for 24 h in serum-free medium in the
dilutions of polyclonal antibodies, in the same buffer without Tween absence or presence of 4 ␮g/ml of each anti-␣3, anti-␣2, and anti-
 NEUROBLASTOMA CELLS–COLLAGEN IV INTERACTIONS 171

binding amino acid sequence which represents an in-

terruption of the Gly-X-Y sequence of the ␣1 chain of
Col IV from the collagenous part of the ␣1 (IV). In
contrast, within the same time interval the percentage
adhesion to the major noncollagenous NC1 domain and
to the synthetic peptide HEP-I, the sequence of which
was derived from the ␣1 (NC1-IV) domain, was ob-
served to be only 5 and 3%, respectively. This indicates
a preference of SK-N-SH neuroblastoma cells to bind to
the collagenous domain of the molecule rather than to
the noncollagenous NC1 domain. NC1 and HEP-I have
been reported to promote adhesion to other cell types,
such as human mesangial cells, sympathetic neurons,
and aortic endothelial cells [8, 9, 31]. In dose–response
cell adhesion experiments (Fig. 1B), the highest per-
centage adhesion (⬃25% over 30 min) was achieved
when intact Col IV was plated at a concentration of 20
␮g/ml and did not increase further. Results obtained
using the synthetic peptide HEP-III as substrate indi-
cated that, for concentrations of HEP-III ranging from
0.3 to 100 ␮g/ml, maximum adhesion (⬃20%) was ob-
served at 50 ␮g/ml of peptide, following an incubation
FIG. 1. Adhesion of neuroblastoma cells to substrates. (A)
[ 35S]methionine-labeled cells were seeded in 96-well plates (5000 of 30 min.
cells/well) coated with 20 ␮g/ml of either solid phase collagen IV or
NC1 or 50 ␮g/ml peptides HEP-I and HEP-III. Cells were allowed to
adhere for 30 min to 1 h at 37°C. Adherent cells were lysed, and
Integrins Mediating the Adhesion of Neuroblastoma
radioactivity was quantitated as stated under Materials and Meth- Cells to Col IV and Peptide HEP-III
ods. Percentage adhesion of cells was calculated as (cpm of cells
adhered per well/cpm of 5000 cells) ⫻ 100. Nonspecific adhesion of To identify the integrin receptors of neuroblastoma
cells to bovine serum albumin, which was less than 1%, was sub- cells, which are involved in the interaction of these
tracted from adhesion to substrates. (B) In dose–response cell adhe- cells with Col IV and HEP-III, we performed adhesion
sion, cells were allowed to adhere to wells coated with increasing
concentrations of substrates for 30 min. Results are expressed as experiments in the presence of well-characterized
mean ⫾ SD of six replicates, from three experiments. Differences monoclonal antibodies, directed against various inte-
were significant at P ⬍ 0.05. grin subunits expressed by this cell line. A panel of
monoclonal antibodies was used, all of which were pre-
viously reported to inhibit the adhesion of different
␣v␤3 monoclonal antibody (mAb) and 0.05 ␮g/ml anti-␤1 mAb. At the cells types to various substrates [9, 32, 33]. Antibodies
end of the 24-h incubation period, the conditioned media were col- were used at increasing concentrations, ranging from
lected for zymography.
0.3 to 10 ␮g/ml, whereas the selected substrate concen-
Densitometric analysis. Densitometric analysis of images of
Western blots and zymograms was performed using image-process-
trations promoted half-maximal binding of cells. These
ing software (Bioprofil Vilber Loumart). concentrations were 10 ␮g/ml for Col IV and 25 ␮g/ml
Statistical analysis. Mean values were derived from experiments for peptide HEP-III. Cells were preincubated with
performed in triplicate. These values were compared using the Stu- monoclonal antibodies at 4°C for 30 min before plating.
dent t test. A P value of ⬍0.05 was considered statistically signifi- Inhibition of adhesion was determined compared to
cant.
adhesion in the absence of antibodies considered as
100%. From the examined panel of ␣2, ␣3, ␤1, and
RESULTS ␣v␤3 anti-integrin antibodies used, as shown in Fig. 2,
the maximal observed inhibition on both substrates
Neuroblastoma Cell Adhesion to Substrates
was obtained by the use of anti-␤1 integrin antibody.
We examined neuroblastoma cell binding to intact Almost up to 100% inhibition was observed at concen-
Col IV, its noncollagenous domain NC1, and two syn- trations of antibody higher than 1.25 ␮g/ml. At the
thetic peptides, HEP-I and HEP-III, which were previ- same concentration, anti-␣3 antibody resulted in 30
ously reported to promote adhesion of several cell and 35% inhibition of adhesion on Col IV and HEP-III,
types. Figure 1A shows that neuroblastoma cells ad- respectively. Anti-␣v␤3 antibody resulted in inhibition
hered to intact Col IV up to 45% after 1 h and up to 30% of adhesion on both substrates similar to that of
to the synthetic peptide HEP-III, derived from a cell- anti-␣3 (⬃30%), but at eightfold higher concentrations.
172 TZINIA ET AL.

blots (Fig. 4B) confirmed the results obtained by

FACscan analysis.

Induction of the Expression of MMPs and TIMPs

To determine whether the interaction of neuroblas-
toma cells with Col IV induced the expression of pro-
teins involved in the degradative pathway, we per-
formed gelatin zymography and Western blotting on
conditioned media from cultures of cells grown either
on plastic or on immobilized Col IV substrate. Gelatin
zymography is a sensitive technique for the direct de-
FIG. 2. mAb inhibition of neuroblastoma cell adhesion to sub- tection of any MMPs that can degrade gelatin. Cells
strates. [ 35S]methionine-labeled cells (5000 cells/well) were coincu- cultured on plastic synthesized and secreted high lev-
bated in suspension with the indicated concentrations of integrin els of 72/68-kDa MMP-2 (Fig. 5, lane 1), but very low
mAbs anti-␣3 (P1B5), anti-␣v␤3 (Lm609), and anti-␤1 (P5D2) at 4°C
levels of MMP-9, detected only after loading 10⫻ con-
for 30 min before plating. The cells were then allowed to adhere on
wells precoated with Col IV (10 ␮g/ml) or HEP-III (25 ␮g/ml) for 30 centrated amounts of conditioned medium for zymog-
min at 37°C. At the end of the incubation period, adherent cells were raphy (data not shown). When cells were allowed to
lysed, and radioactivity was quantitated. Bound counts were ex- grow on Col IV for 48 h, the levels of 72/68-kDa MMP-2
pressed as a percentage of the counts bound in the absence of mAb, increased by 2-fold, and the 92/88-kDa MMP-9 was
to give percentage adhesion. Results are expressed as mean ⫾ SD of
six replicates, from three experiments. Differences were significant
also detected in both the pro-MMP and the active
at P ⬍ 0.05. forms, albeit in smaller amounts compared to MMP-2
(Fig. 5, lane 2). To confirm the above results, 10-fold-
concentrated conditioned media were blotted with spe-
Anti-␣2 antibodies had no effect on the adhesion on cific antibodies against MMP-2 and MMP-9 proteins.
either substrate in all concentrations used. Media assessed by Western blotting showed that
MMP-9 was secreted by cells grown on plastic but at
Integrins Expressed in the Presence of Collagen IV very low levels compared to the levels of MMP-2 (Fig.
6A, lanes 1 and 3, respectively). In addition, the acti-
Integrin expression in neuroblastoma cells, in the
vated form of both proteins was present when the cells
presence or absence of Col IV, was examined by flow
were grown on Col IV (Fig. 6A, lanes 2 and 4).
cytometry and immunoblotting. Initially, fluorescence-
Since TIMPs regulate in part the biological activity
activated cell sorting was performed in nonpermeabi-
of MMPs, we studied whether the expression of
lized cell populations to verify the cell surface avail-
TIMP-1 and TIMP-2 was modulated by Col IV. The
ability of specific integrin subunits expressed by SK-
results obtained from Western analysis demonstrated
N-SH neuroblastoma cells in the absence or presence of
an increase in the protein level of TIMP-1 and TIMP-2
Col IV, by the use of specific anti-integrin monoclonal
by 25 and 35%, respectively, expressed by cells grown
antibodies. Cells were allowed to grow for 48 h on
on Col IV (Fig. 6A, lanes 6 and 8), compared to the
either plastic or culture dishes precoated with 50 ␮g/ml
protein levels expressed by cells grown on plastic (Fig.
Col IV. From the panel of integrin subunits examined
6A, lanes 5 and 7). Increased expression of MMPs in
(␣2, ␣3, ␣5, ␣v, ␤1, and ␤3), 96% of cells grown on
parallel with an increase in the protein levels of TIMPs
plastic expressed ␤1 integrin subunit at high density
could represent a regulatory mechanism of the cells to
on cell surface. Forty-five and thirty percent of cells
maintain homeostasis of the ECM environment.
expressed the ␣3 and ␣2 subunits, respectively, while
the percentage of ␣v␤3-positive cells was about 20%
Effect of Anti-integrin Antibodies on MMP-2
(Fig. 3A). In the presence of Col IV as substrate, the
Expression
levels of ␣3 and ␤1 integrin were decreased by 29 and
55%, respectively, whereas the levels of ␣2 and ␣v␤3 Since ␣3␤1 and ␣v␤3 integrins were shown to medi-
were not significantly changed (Fig. 3B). Under both ate the adhesion of neuroblastoma cells to Col IV, and
experimental conditions SK-N-SH cells did not express integrins were reported to be involved in signal trans-
␣5 integrin subunits. duction regulating MMP transcription and secretion
To confirm the above results at the protein level, [14, 15, 34], we examined the possible roles of ␣2, ␣3␤1,
immunoblotting was performed on lysates from cells and ␣v␤3 integrins in the secretion of MMP-2, the
cultured either on plastic (Fig. 4A, lanes 1, 3, 5, 7, and predominant collagenase expressed by these cells. For
9) or on 50 ␮g/ml Col IV (Fig. 4A, lanes 2, 4, 6, 8, and this purpose, cells were cultured on Col IV for 48 h at
10) and probed with the use of specific anti-integrin 37°C, in the absence or presence of 4 ␮g/ml of each
polyclonal antibodies. Densitometric analysis of the anti-␣3, -␣v␤3, and -␣2 monoclonal antibodies, which
 NEUROBLASTOMA CELLS–COLLAGEN IV INTERACTIONS 173

FIG. 3. Fluorescence intensity distribution of integrin subunits on SK-N-SH cells. (A) Integrin expression on neuroblastoma cells grown
on plastic was analyzed using anti-␤1, -␣3, -␣2, and -␣v␤3 integrin monoclonal antibodies at saturating concentrations (10 ␮g/ml).
Fluorescein activity was examined using fluorescein isothiocyanate-anti-human IgG as second antibody. (B) Alterations in integrin expres-
sion of neuroblastoma cells, grown on Col IV (50 ␮g/ml), expressed as mean of fluorescence. Results are expressed as mean ⫾ SD from three
experiments.

were used as integrin ligands [22, 35]. Anti-␤1 mAb mAb against ␣2 subunit failed to induce any significant
was used at a concentration of 0.05 ␮g/ml, since at this changes in MMP-2 secretion (Fig. 7, lane 4).
antibody concentration cell adhesion to Col IV was not
significantly affected. Conditioned media were col- DISCUSSION
lected for zymography. As illustrated in Fig. 7, treat-
ment of cells with a combination of mAbs against ␣3 ECM influences cellular functions including adhe-
and ␤1 integrin subunits enhanced the secretion of sion, migration, differentiation, and gene expression.
MMP-2 by ⬃300% (Fig. 7, lane 2), whereas treatment Collagen IV has a pivotal role in ECM interaction with
of cells with anti-␣v␤3 mAb down-regulated the secre- cells, in addition to providing mechanical stability by
tion of MMP-2 by 30% (Fig. 7, lane 3) in comparison to network formation and incorporation of other matrix
untreated cells (Fig. 7, lane 1). Treatment of cells with components. In neuronal cells, Col IV plays an impor-
174 TZINIA ET AL.

X-Y motif in the ␣1 chain of Col IV [9]. We documented

herein that SK-N-SH neuroblastoma cells used intact
Col IV and HEP-III for binding, in a concentration- and
time-dependent manner. Larger peptide concentra-
tions were required for adhesion compared to Col IV,
apparently because of lack of conformation of this pep-
tide and because intact Col IV contains more than one
cell-binding site [9].
Domain NC1 and peptide HEP-I did not significantly
sustain adhesion of SK-N-SH cells. Thus, these cells
apparently do not use this domain for binding to Col
IV, in contrast to other neuronal cells [31]. We conclude
that SK-N-SH cell–Col IV interactions are mediated by
the major collagenous part of this glycoprotein.
We next examined which integrin subunits of neuro-
blastoma cells mediate the binding to Col IV and HEP-
III. Competition experiments were performed, in which
adhesion to Col IV was examined in the presence or
absence of monoclonal antibodies to different integrin
subunits. These experiments demonstrated that bind-
ing to both Col IV and HEP-III was mainly mediated by
the ␤1 integrin subunit (␤1 is apparently the key sub-
FIG. 4. Alterations in integrin protein levels due to Col IV. (A)
Western blot analysis of integrin subunits in neuroblastoma cells
unit interacting with different ␣ subunits). Even small
cultured either on plastic or on 50 ␮g/ml Col IV. Total protein (20 ␮g) concentrations of anti-␤1 antibody (0.3 ␮g/ml) were
extracted from cells cultured on both conditions was analyzed on able to inhibit the adhesion of neuroblastoma cells on
7.5% SDS–PAGE under nonreducing conditions and immunoblotted Col IV by almost 50%. Based on inhibition experi-
with the appropriate dilution of polyclonal antibodies against inte- ments, the ␣3␤1 integrin is a major integrin mediating
grin subunits. (B) Quantification of the protein content of each inte-
grin subunit by scanning densitometry. Each bar represents the
binding to Col IV, whereas the ␣2 integrin subunit was
mean ⫾ SD of three independent experiments. Differences were not apparently involved in this binding. Additionally,
significant at P ⬍ 0.05. the ␣v␤3 receptor mediates neuroblastoma–Col IV in-
teractions, albeit to a lesser extent, since the maximal
observed inhibition in the presence of anti-␣v␤3 anti-
tant role for axonal regeneration [5]. In the present body was ⬃30%. These results suggest that both ␣3␤1
study, we addressed additional effects of Col IV on and ␣v␤3 represent two of the integrin receptors me-
neuroblastoma cell functions. The aim was to deter- diating SK-N-SH–Col IV interactions. Furthermore,
mine the domains of Col IV involved in neuroblastoma the obtained results indicate that the sequence con-
cell binding, the surface receptor(s) participating in the tained in the synthetic peptide HEP-III represents a
binding, and the modulation of gene expression follow- major recognition site for ␣3␤1 binding of neuroblas-
ing interactions of SK-N-SH cells with Col IV. Col IV is toma cells to Col IV. The ␣v␤3 integrin recognized this
an ECM component essential during the differentia- sequence minimally, based on inhibition experiments,
tion of neuronal cells [13, 36]. To unravel the role of Col since the concentration of antibody used to inhibit cell
IV, adhesion experiments were performed using as
substrates intact Col IV, the main noncollagenous NC1
domain of Col IV, and the synthetic peptides HEP-I
and HEP-III derived from the ␣1 (NC1) and ␣1 (IV),
respectively. Other neuroblastoma cell lines were pre-
viously shown to bind to intact Col IV [37]. The NC1
domain of Col IV and peptides HEP-I and HEP-III
were additionally reported to promote adhesion to dif- FIG. 5. Alterations in gelatinolytic activity of MMP-2 and
ferent cell types [8, 9, 38]. Peptide HEP-I originates MMP-9 secreted by neuroblastoma cells, in the presence of Col IV.
from the ␣1 (noncollagenous NC1) chain of Col IV and Aliquots of serum-free media conditioned by cells cultured on plastic
represents a major recognition site, which binds to (lane 1) or on 50 ␮g/ml Col IV (lane 2) were analyzed on 10%
adjacent Col IV molecules during the process of poly- polyacrylamide gel containing 1 mg/ml gelatin under nonreducing
conditions. The volume of conditioned medium loaded per lane was
merization to a network. In addition, this peptide pro- adjusted according to the cell number obtained upon harvest. Gela-
motes cell adhesion [8]. Peptide HEP-III contains a tinolytic activity was detected as clear bands following incubation in
cell-binding sequence from an interruption of the Gly- enzyme buffer and staining with Coomassie brilliant blue.
 NEUROBLASTOMA CELLS–COLLAGEN IV INTERACTIONS 175

FIG. 6. Induction of the expression of MMPs and TIMPs in the presence of Col IV. (A) Serum-free media conditioned by neuroblastoma
cells previously cultured either on plastic or on 50 ␮g/ml Col IV were collected. Aliquots of each sample were concentrated and subjected to
10% SDS–PAGE under reducing conditions. The volume of conditioned medium loaded per lane was adjusted according to the cell number
obtained upon harvest. Electrophoreticaly transferred proteins were immunoblotted using polyclonal antibodies against MMP-2, MMP-9,
TIMP-1, and TIMP-2 (B–C). Western blots were analyzed by densitometry. Each bar represents the mean ⫾ SD of three independent
experiments. Differences were significant at P ⬍ 0.05.

adhesion was eightfold higher than the concentrations rhabdomyosarcoma cells cultured on matrigel, largely
of anti-␣3 and anti-␤1 antibodies used. The ␣3␤1 inte- composed of laminin, induced MMP-2 activation [35],
grin from melanoma cells was previously shown to fibrosarcoma cells promoted MMP-2 production cul-
recognize this peptide [38], whereas in mesangial cells tured on fibronectin [40], the adhesion of macrophages
the same sequence was recognized by ␣2␤1 integrin [9]. to fibronectin induced MMP-9 expression [41], and en-
Apparently then, the sequence represented in HEP-III dothelial cells cultured on three-dimensional Col I pro-
peptide is a major recognition site for more than one duced increased amounts of MMP-2 protein [42]. In the
integrin complex in different cell types. In the case of present study we demonstrated that culturing of neu-
SK-N-SH cells, peptide HEP-III primarily interacted roblastoma cells on intact Col IV, an essential matrix
via the ␣3␤1 integrin. Our experiments cannot exclude, component for the differentiation of neuronal cells [13,
however, the possibility that additional sites in the 36], not only induced the expression of both MMP-2
major, triple-helical domain of Col IV may be used for and, to a lesser extent, MMP-9, but also resulted in the
the binding of SK-N-SH cells via ␣3␤1, ␣v␤3, or addi- secretion of the activated forms of these enzymes.
tional integrin complexes. The major recognition site It has been suggested that various integrins can
for ␣v␤3 on the triple-helical domain of Col IV remains signal increased transcription of MMPs depending on
to be elucidated. the cell type [15, 34, 43]. Based on experiments in
In our experiments, the presence of Col IV resulted which cells were incubated in the presence of antibod-
in down-regulation of ␣3␤1 integrin, a major ligand for ies against the ␣3 and ␤1 integrin subunits combined,
this matrix protein. It was previously reported that used as ligands for integrins, we propose that SK-N-SH
integrin receptors are regulated by matrix compo- neuroblastoma cells mainly use the ␣3␤1 integrin re-
nents. More specifically, in the presence of high con- ceptor to up-regulate, at least in part, the expression of
centrations of laminin, the expression of laminin-bind- MMP-2. Culturing neuroblastoma cells in the presence
ing ␣6␤1 integrin was decreased on the surface of of anti-␣v␤3 antibodies resulted in moderate down-
sensory neurons [39]. regulation of MMP-2. Thus, the two major integrins
Previous studies have shown the ability of ECM pro- that are used for the binding of these cells to Col IV
teins, mainly fibronectin, laminin, and Col I, to induce apparently have opposite effects on the expression of
their own degradation by promoting the production MMP-2. This could be the result of two different sig-
and release of several matrix metalloproteinases by naling mechanisms mediated by two different sites of
different adherent cell types. For instance, human cell binding on Col IV, one of which up-regulates the
176 TZINIA ET AL.

and, moreover, promote the activation of both MMPs.

This could be important for pathological conditions of
the central nervous system, in which MMPs are impli-
cated, including Alzheimer’s disease.

The authors are indebted to Drs. Christina Zioudrou and George

Konidakis from the Medical School of the University of Crete, for
solid-phase synthesis of peptide Hep-III, and to K. Economou and P.
Karamessinis, for expert assistance with image processing and anal-
ysis.

REFERENCES

1. McClay, D. R., and Ettersohn, C. A. (1987). Cell adhesion and

morphogenesis. Annu. Rev. Cell Biol. 3, 319 –345.
2. McDonald, J. A. (1988). Extracellular matrix assembly. Annu.
Rev. Cell Biol. 4, 183–207.
3. Herbst, T. J., McCarthy, J. B., Tsilibary, E. C., and Furcht, L. T.
(1988). Differential effects of laminin, intact type IV collagen
and specific domains of type IV collagen on endothelial cell
adhesion and migration. J. Cell Biol. 106, 1365–1373.
4. Perris, R., Syfring, J., Paulsson, M., and Bronner-Fraser, M.
FIG. 7. Effect of anti-integrin antibodies on MMP-2 expression
(1993). Molecular mechanisms of neural crest cell attachment
and secretion. (A) Cells in suspension were preincubated with anti-
and migration on types I and IV collagen. J. Cell Sci. 106,
integrin monoclonal antibodies for 30 min at 37°C as described under
1357–1368.
Materials and Methods (lanes 2– 4) or without antibody addition
(lane 1) and then diluted 10-fold and cultured for 24 h on 24-well 5. Joostin, E. A., Dijkstra, S., Brook, G. A., Veldman, H., and Bar,
plates precoated with 50 ␮g/ml Col IV. The medium was removed, P. R. (2000). Collagen IV deposits do not prevent regrowing
the plates were washed with serum-free medium, and cells were axons from penetrating the lesion site in spinal cord injury.
incubated for 24 h in serum-free medium containing or not anti- J. Neurosci. Res. 62, 689 – 691.
integrin mAbs. Aliquots of each sample of conditioned media were 6. Timpl, R., Oberbaumer, I., von der Mark, H., Bode, W., Wick,
analyzed by gelatin zymography (the volume of conditioned medium G., Weber, S., and Engel, J. (1985). Structure and biology of the
loaded per lane was adjusted according to the cell number obtained globular domain of basement membrane type IV collagen. Ann.
upon harvest). (B) Quantification of the 72/68-kDa form of MMP-2 N. Y. Acad. Sci. 460, 58 –72.
using scanning densitometry. 7. Gunwar, S., Ballester, F., Noelken, M. E., Sado, Y., Ninomiya,
Y., and Hudson, B. G. (1998). Glomerular basement membrane.
Identification of a novel disulfide-cross-linked network of al-
expression of MMP-2 and the other of which has the pha3, alpha4, and alpha5 chains of type IV collagen and its
opposite effect. We conclude that interactions of SK- implications for the pathogenesis of Alport syndrome. J. Biol.
N-SH neuroblastoma cells with Col IV lead to regula- Chem. 273, 8767– 8775.
tion of MMP-2 expression, depending on the site of 8. Tsilibary, E. C., Reger, L. A., Vogel, A. M., Koliakos, G. G.,
Anderson, S. S., Charonis, A. S., Alegre, J. N., and Furcht, L. T.
binding and the integrins that become engaged in each (1990). Identification of a multifunctional, cell-binding peptide
case. The binding event(s) and the ensuing result of Col sequence from the ␣1(NC1) of type IV collagen. J. Cell Biol. 111,
IV–neuroblastoma cell interactions would then be de- 1583–1591.
termined, based on the needs of these cells to degrade 9. Setty, S., Kim, Y., Fieldst, G. B., Clegg, D. O., Wayner, E. A.,
their immediate environment. Furthermore, Col IV- and Tsilibary, E. C. (1998). Interactions of type IV collagen and
induced increase of protein levels of TIMP-1 and its domains with human mesangial cells. J. Biol. Chem. 273,
12244 –12249.
TIMP-2, which are the specific inhibitors for MMP-2
10. Hynes, R. O. (1987). Integrins: A family of cell-surface recep-
and MMP-9, respectively, suggested that this increase tors. Cell 48, 549 –554.
could be, in part, a balancing effect to the increased 11. Ruoslahti, E. (1991). Integrins. J. Clin. Invest. 87, 1–5.
expression of MMPs. 12. Giancotti, F. G., and Ruoslahti, E. (1999). Integrin signaling.
In summary, the data presented in this report sug- Science 285, 1028 –1032.
gest that neuroblastoma cell binding to Col IV is me- 13. Reichardt, L. F., and Tomaselli, K. J. (1991). Extracellular
diated by the main collagenous domain of this glyco- matrix molecules and their receptors: Functions in neural de-
protein and does not involve the main noncollagenous velopment. Annu. Rev. Neurosci. 14, 531–570.
NC1 domain. A recognition site for the ␣3␤1 integrin 14. Larjava, H., Lyons, J. G., Salo, T., Makela, M., Koivisto, L.,
was primarily the sequence of peptide Hep-III in the ␣1 Birkedal-Hansen, H., Akiyama, S. K., Yamada, K., and Heino,
J. (1993). Anti-integrin antibodies induce type-IV collagenase
(IV) chain, although additional sites should exist on expression in keratinocytes. J. Physiol. 157, 190 –200.
Col IV. Col IV–SK-N-SH cell interactions, apparently 15. Seftor, R. E. B., Seftor, E. A., Stetler-Stevenson, W. G., and
mediated by ␣3␤1 and to a lesser extent by ␣v␤3 inte- Hendrix, M. J. C. (1993). The 72kDa type-IV collagenase is
grins, modulate the expression of MMP-2 and MMP-9 modulated via differential expression of ␣v␤3 and ␣5␤1 inte-
 NEUROBLASTOMA CELLS–COLLAGEN IV INTERACTIONS 177

grins during human melanoma-cell invasion. Cancer Res. 53, on expression of type IV collagen, MMP-2, MMP-9 and TIMP-1
3411–3415. by human mesangial cells. Cell Adhes. Commun. 4, 89 –101.
16. Yong, V. W., Krekoski, C. A., Forsyth, P. A., Bell, R., and 31. Lein, P. J., Higgins, D., Turner, D. C., Flier, L. A., and Ter-
Edwards, D. R. (1998). Matrix metalloproteinases and diseases ranova, V. P. (1991). The NC1 domain of type IV collagen
of the CNS. Trends Neurosci. 21, 75– 80. promotes axonal growth in sympathetic neurons through inter-
17. Sato, H., Takino, T., Okada, Y., Cao, J., Shinagawa, A., action with the ␣1␤1 integrin. J. Cell Biol. 113, 417– 428.
Yamamoto, E., and Seiki, M. (1994). A matrix metalloprotein- 32. Setty, S., Anderson, S. S., Wayner, E. A., Kim, Y., Clegg, D. O.,
ase expressed on the surface of invasive tumor cells. Nature and Tsilibary, E. C. (1995). Glucose-induced alteration of inte-
370, 61– 65. grin expression and function in cultured human mesangial
cells. Cell Adhes. Commun. 3, 187–200.
18. Birkedal-Hansen, H. (1995). Proteolytic remodeling of extracel-
lular matrix. Curr. Opin. Cell Biol. 7, 728. 33. Chen, Y., Krishnamurti, U., Wayner, E. A., Michael, A. F., and
Charonis, A. S. (1996). Receptors in proximal tubular epithelial
19. Matrisian, L. M. (1992). The matrix-degrading metalloprotein- cells for tubulointerstitial nephritis antigen. Kidney Int. 49,
ases. Bioassays 14, 455– 463. 153–157.
20. Nagase, H., and Woessner, J. F. (1999). Matrix metalloprotein- 34. Langholz, O., Rockel, D., Mauch, C., Kozlowska, E., Bank, I.,
ases. J. Biol. Chem. 274, 21491–21494. Krieg, T., and Eckes, B. (1995). Collagen and collagenase gene
21. Murphy, G., and Knauper, V. (1997). Relating matrix metallo- expression in three-dimensional collagen lattices are differen-
proteinase structure to function: Why the “hemopexin” domain? tially regulated by ␣1␤1 and ␣2␤1 integrins. J. Cell Biol. 131,
Matrix Biol. 15, 511–518. 1903–1915.
22. Sugiura, T., and Berdichevski, F. (1999). Function of ␣3␤1- 35. Kubota, S., Ito, H., Ishibashi, Y., and Seyama, Y. (1997).
tetraspanin protein complexes in tumor cell invasion. Evidence Anti-␣3 integrin antibody induces the activated form of matrix
for the role of the complexes in production of matrix metallo- metalloproeinase-2 (MMP-2) with concomitant stimulation of
proteinase 2 (MMP-2). J. Cell Biol. 146, 1375–1389. invasion through matrigel by human rhabdomyosarcoma cells.
23. Hofman, U. B., Westphal, J. R., Van Kraats, A. A., Ruiter, D. J., Int. J. Cancer 70, 106 –111.
and Van Muijen, G. N. P. (2000). Expression of integrin ␣v␤3 36. Kowtha, V. C., Bryant, H. J., Pancrazio, J. J., and Stenger, D. A.
correlates with activation of membrane-type matrix metallo- (1998). Influence of extracellular matrix proteins on membrane
proteinase-1 (MT1-MMP) and matrix metalloproteinase-2 potentials and excitability in NG108-15 cells. Neurosci. Lett.
(MMP-2) in human melanoma cells in vitro and in vivo. Int. J. 246, 9 –12.
Cancer 87, 12–19. 37. Rozzo, C., Ratti, P., Ponzoni, M., and Cornaglia-Ferraris, P.
24. Hoyhtya, M., Turpeenniemi-Hujanen, T., Stetler-Stevenson, (1993). Modulation of ␣1␤1, ␣2␤1, and ␣3␤1 integrin het-
W., Krutzsch, H., Tryggvason, K., and Liotta, L. A. (1988). erodimers during human neuroblastoma cell differentiation.
Monoclonal antibodies to type IV collagenase recognize a pro- FEBS Lett. 332, 263–267.
tein with limited sequence homology to interstitial collagenase 38. Miles, A. J., Knutson, J. R., Furcht, L. T., McCarthy, J. B., and
and stromelysin. FEBS Lett. 233, 109 –113. Fields, G. B. (1995). A peptide model of basement membrane
collagen alpha 1 (IV) 531-543 binds the alpha 3 beta 1 integrin.
25. Hoyhtya, M., Fridman, R., Komarek, D., Porter-Jordan, K.,
J. Biol. Chem. 270, 29047–29050.
Stetler-Stevenson, W. G., Liotta, L. A., and Liang, C. M. (1994).
Immunohistochemical localization of matrix metalloproteinase 39. Condic, M. L., and Letourneau, P. C. (1997). Ligand-induced
2 and its specific inhibitor TIMP-2 in neoplastic tissues with changes in integrin expression regulate neuronal adhesion and
monoclonal antibodies. Int. J. Cancer 56, 500 –505. neurite outgrowth. Nature 389, 852– 856.
26. Koliakos, G. G., Koliakos-Kouzi, K., Furcht, L. T., Reger, L. A., 40. Stanton, H., Gavrilovic, J., Atkinson, S. J., d’Ortho, M. P.,
and Tsilibary, E. C. (1989). The binding of Heparin to type IV Yamada, K. M., Zardi, L., and Murphy, G. (1998). The activa-
collagen: Domain specificity with identification of peptide se- tion of proMMP-2 (gelatinase A) by HT1080 fibrosarcoma cells
quences from the ␣1(IV) and ␣2(IV) which preferentially bind is promoted by culture on a fibronectin substrate and is con-
heparin. J. Biol. Chem. 264, 2313–2323. comitant with an increase in processing of MT1-MMP (MMP-
14) to a 45kDa form. J. Cell Sci. 111, 2789 –2798.
27. Tsilibary, E. C., and Charonis, A. S. (1986). The role of the main
41. Xie, B., Laouar, A., and Huberman, E. (1998). Fibronectin-
noncollagenous domain (NC1) in type IV collagen self-assem-
mediated cell adhesion is required for induction of 92-kDa type
bly. J. Cell Biol. 103, 2467–2473.
IV collagenase/gelatinase (MMP-9) gene expression during
28. Laemmli, U. K. (1970). Cleavage of structural proteins during macrophage differentiation. J. Biol. Chem. 273, 11576 –11582.
the assembly of the head of bacteriophage. Nature 227, 680 – 42. Haas, T. L., Davis, S. J., and Mardi, A. (1998). Three-dimen-
685. tional type I collagen lattices induce coordinated expression of
29. Towbin, H., Staehelin, T., and Gordon, J. (1979). Electro- matrix metalloproteinases MT1-MMP and MMP-2 in microvas-
phoretic transfer of proteins from polyacrylamide gels to nitro- cular endothelial cells. J. Biol. Chem. 273, 3604 –3610.
cellulose sheets: Procedure and some applications. Proc. Natl. 43. Chintala, S. K., Sawaya, R., Gokaslan, Z. L., and Rao, J. S.
Acad. Sci. USA 76, 4350 – 4354. (1996). Modulation of matrix metalloproteinase-2 and invasion
30. Anderson, S. S., Wu, K., Nagase, H., Stetler-Stevenson, W., in human glioma cells by alpha 3 beta 1 integrin. Cancer Lett.
Kim, Y., and Tsilibary, E. C. (1996). Effect of matrix glycation 103, 201–208.

Received December 3, 2000

Revised version received November 29, 2001
Published online February 12, 2002