(Received for publication, March 28, 1996, and in revised form, November 4, 1996)
From the Department of Pathology, Division of Neuropathology, University of Alabama at Birmingham, Birmingham, Alabama 35294
Remodeling of the matrix by tumor cells is
necessary for tumor invasion. We have shown previously that malignant
astrocytomas, in contrast to normal astrocytes, synthesize vitronectin
and express integrins v
3 and
v
5. The activity states of these two
integrins are differentially controlled. Thus, we investigated the
regulation of the activity of integrins
v
3 and
v
5
with regard to their role in vitronectin internalization in U-251MG
astrocytoma cell monolayers adherent to fibronectin, collagen, or
laminin in serum-free conditions. Binding of
[125I]vitronectin occurred in a specific, saturable
manner that was partially inhibitable by monoclonal antibodies (mAbs)
specific for integrins
v
3 or
v
5. Specific, lysosomally-mediated
degradation of [125I]vitronectin was detectable at 1 h and increased over the 24-h assay period. The cell substrate affected
the rate of turnover of [125I]vitronectin, which was 3.0 ng/min for cells plated on fibronectin but 0.35 ng/min for cells plated
on collagen. Furthermore, although mAbs specific for either integrin
v
3 or
v
5
inhibited degradation (30%; combined effect 70%) of
[125I]vitronectin by cells plated on fibronectin, only
mAb anti-
v
5 inhibited degradation
(70-90%) by cells plated on collagen or laminin. To determine the
requirement for integrin
5
1 ligation in
order for integrin
v
3 to internalize its
ligand, cells were plated on mAbs anti-integrin
5 or
anti-integrin
3. When plated on mAb
anti-
5, mAbs anti-
v
3 and
anti-
v
5 both inhibited degradation. However, when plated on mAb anti-
3, mAb
anti-
v
3 had no effect whereas mAb
anti-
v
5 inhibited degradation. These data
indicate that a signal from integrin
5
1
is necessary for integrin
v
3 to
internalize vitronectin, whereas integrin
v
5 constitutively internalizes
vitronectin.
We have previously reported that two vitronectin receptors,
integrins v
3 and
v
5, are expressed on malignant
astrocytoma cells both in vivo and in vitro and
that they are markers of the malignant phenotype (1, 2). These two
integrins differ in their ligand specificity and mechanism of
regulation. Integrin
v
5 recognizes only
vitronectin and osteopontin as ligands, whereas integrin
v
3 is a promiscuous receptor that
recognizes vitronectin, fibrinogen, von Willebrand factor, fibronectin,
thrombospondin, thrombin, osteopontin, bone sialoprotein, laminin, and
degraded collagen types I and IV (reviewed in Refs. 3-5; 6-8). The
activity state of integrin
v
5 varies
according to the cell type (2, 9-11), whereas it is thought that
integrin
v
3 is expressed in a
constitutively active state on all nucleated cells (reviewed in ref. 4)
and is regulated by an integrin-associated protein, phosphorylation of
associated proteins, and growth factors (reviewed in Ref. 12; 13-17).
Ligation of integrin
v
3 has been shown to
act as a signal for integrin
5
1,
inhibiting fibronectin internalization by integrin
5
1 (16, 17). There is also indirect
evidence that the expression of integrin
5
1 can modulate the activity of integrin
v
3 (18, 19). Thus, for example, unligated
integrin
5
1 is important for integrin
v
3-mediated motility in Chinese hamster
ovary cells, and absence of
1 integrins in embryonal carcinoma cells reduces integrin
v
3-mediated motility (18, 19).
Previous studies have indicated that in fibroblast monolayers,
v
5 mediates the internalization of
vitronectin in a protein kinase C-dependent manner (20,
21). In granulation tissue, matrix proteins, such as vitronectin, are
removed as part of the tissue remodeling; however, the fate of
extracellular vitronectin in tumors is unclear. Remodeling of the
extracellular matrix is necessary for tumor invasion (22). We have
shown previously that malignant astrocytomas have the potential to
remodel the matrix because, in contrast to normal astrocytes, these
cells synthesize vitronectin (1, 2).
In vitro, vitronectin exists in two conformers, known as native and altered, both of which promote cell attachment (23); however, only altered vitronectin is internalized by fibroblasts (24). Vitronectin purified by heparin-Sepharose chromatography after urea denaturation is referred to as altered or denatured vitronectin (23). Vitronectin is an adhesive glycoprotein that promotes cell adhesion and migration through its Arg-Gly-Asp (RGD) peptide cell adhesion domain (23). Altered vitronectin, in contrast to native vitronectin, is a multimeric protein and possesses a functional heparin-binding domain through which it regulates the activity of several other proteins, such as the thrombin-anti-thrombin III complex and the C5b-9 terminal complement complex (23).
We have investigated the fate of extracellular vitronectin when
astrocytoma cells are adherent to protein constituents of the
pial-glial and endothelial cell basement membranes because these
membranes are extensively remodeled in this tumor and matrix protein
internalization is thought to occur as part of matrix remodeling (22).
These studies are of physiologic significance, as malignant astrocytoma
cells are known to invade throughout the brain, in part, by adhesion
and migration on pial-glial and endothelial cell basement membranes
(22). We demonstrate that integrin
v
3-mediated internalization of altered
vitronectin requires ligation of integrin
5
1.
Heparin, chondroitin sulfate A,
chloroquine, and mouse IgG were purchased from Sigma.
Peptides were synthesized by the University of California at San Diego
Peptide Synthesis Facility. Neutralizing mAbs1 anti-integrin
v
3 (LM609), anti-integrin
v
5 (P3G2), and anti-integrin
1 (P4C10) were generous gifts from Dr. David A. Cheresh
(Scripps Research Institute, La Jolla, CA). Neutralizing mAbs
anti-integrin
3,
5, and
6
were purchased from Chemicon (Temecula, CA). Affinity-isolated rabbit
anti-integrin
5
1 antiserum, neutralizing
mAbs anti-integrin
2, and anti-integrin
3
ascites were purchased from Life Technologies, Inc. The IgG fraction of
mAbs anti-integrin
2, anti-integrin
3,
and anti-integrin
5
1 were purified as
described (1). Fluorescein isothiocyanate-conjugated goat anti-rabbit
and goat anti-mouse IgG antibodies were purchased ( Jackson Laboratory, West Grove, PA).
Altered vitronectin was purified from human plasma as described (25) and migrated as a 75/65 kDa doublet on disulfide-reduced 10% SDS-PAGE and as a high molecular weight aggregate in nonreduced SDS-PAGE, as expected. Vitronectin was iodinated by the iodogen method (average specific activity of 10 µCi/µg) and free iodine was removed by gel filtration, as described (26, 27). The migration of iodinated vitronectin on 12% disulfide-reduced SDS-PAGE was identical to that of the cold protein. Native vitronectin was a gift from Dr. Deanne Mosher (University of Wisconsin, Madison, WI) and migrated on nonreduced SDS-PAGE as a doublet at 75/65 kDa.
Degradation and Binding AssaysHuman malignant astrocytoma
cells (U-251MG) from the ATCC were cultured in complete medium, as
described (1), and were Mycoplasma-free. For the vitronectin
degradation and binding experiments, 12-well culture plates were coated
with 10 µg/ml fibronectin (Boehringer Mannheim) collagen type I (ICN
Biomedicals, Costa Mesa, CA), or laminin (Collaborative Biomedical
Products, Bedford, MA) at 37 °C for 6 h or overnight, washed,
and then blocked for 1 h at 25 °C with 1% bovine serum albumin
after heat denaturation (98 °C for 10 min). Cells (105
per well) aliquoted onto fibronectin, collagen, or laminin-coated wells
attached and spread, and their morphology was similar. All experiments
were performed with cells that had been plated for 1 h as well as
overnight in serum-free Dulbecco's modified Eagle's medium/1% bovine
serum albumin. 125I-Labeled altered vitronectin (600 ng/ml,
1 × 106 cpm) in 0.2% bovine serum albumin was added
to the wells, followed by incubation at 37 °C and timed removal of
medium aliquots, as described (20, 21, 24, 28).
125I-Labeled altered vitronectin degradation was
quantitated as the radioactivity in a 50-µl aliquot of medium that
was soluble in 20% trichloroacetic acid. Background radioactivity
obtained from substrate-coated wells not containing cells that were
incubated in parallel to the experimental wells was subtracted from the total trichloroacetic acid-soluble radioactivity. In experiments done
in the presence of heparin, heparin was preincubated with the monolayer
prior to [125I]vitronectin addition. For degradation
experiments in which wells were coated with mAbs prior to the plating
of the cells, wells were coated for 1 h at 25 °C with goat
anti-mouse IgG (20 µg/ml), washed, blocked with 1% heat-denatured
bovine serum albumin for 1 h at 37 °C, washed, incubated with
10 µg/ml mAbs anti-integrin 3 or
5 IgG
(Chemicon, Temecula, CA) for 1 h at 25 °C, as described (29,
30). After the wells were washed, cells were plated for 1 h.
Medium aliquots at various time points were subjected to 12%
disulfide-reduced SDS-PAGE and autoradiography to confirm the absence
of proteolytic degradation of [125I]vitronectin in the
media. To measure bound [125I]vitronectin, labeled medium
was removed, the wells were washed with phosphate-buffered saline, and
the cells were solubilized in 1 N NaOH and scintillation
counted. Nonspecific cell binding obtained from wells containing 100 µg/ml unlabeled vitronectin was subtracted from all determinations.
Binding assay data represent the mean ± S.E. from triplicate
wells of one experiment. Binding assay experiments were performed three
times with similar results. Degradation assay data represent the
mean ± S.E. from three or four separate experiments, where in
each experiment the condition was assayed as a single well or in
duplicate.
U-251MG cells were plated onto fibronectin, collagen, or laminin-coated flasks in serum-free media, and after a 1 h or overnight incubation harvested with buffered EDTA, reacted with primary antibody, followed by goat anti-mouse or anti-rabbit fluorescein isothiocyanate-conjugated secondary antibody, and then subjected to FACS analysis, as described (27).
Cell Adhesion Assays96-well plates were coated overnight at 25 °C with 10 µg/ml fibronectin, collagen, or laminin in phosphate-buffered saline. Cells were harvested with buffered EDTA, suspended in adhesion assay buffer, and incubated with 50 µg/ml of the indicated antibody for 30 min (1, 29). Subsequently, cells were plated onto coated wells (40,000 cells/well), allowed to attach for 45 min at 37 °C, washed, crystal violet stained, and quantitated by spectrophotometric absorbance at 570 nm, as described (1, 29).
We
initially characterized the binding of altered vitronectin to a human
astrocytoma cell monolayer. 125I-Labeled altered
vitronectin (600 ng/ml, 1 × 106 cpm) was added to a
subconfluent monolayer of serum-starved U-251MG cells that had been
plated on fibronectin. Labeled vitronectin was found to bind maximally
to the cell monolayer after 3-6 h, at which time approximately 7% (42 ng) of the added [125I]vitronectin was specifically bound
(Fig. 1A). Cold vitronectin (100 µg/ml)
inhibited binding by 85% (data not shown), indicating that binding was
specific. To determine whether integrins
v
3 or
v
5
mediated vitronectin binding, as we have reported their expression on
these cells (1, 2), mAbs anti-
v
3 and
anti-
v
5 were preincubated with the
monolayer, and each inhibited binding by 30% at 6 h. Studies were
also done in the presence of heparin to eliminate a heparan sulfate
proteoglycan-mediated event, and heparin (100 µg/ml) significantly
inhibited [125I]vitronectin binding to the cell monolayer
(70% at 6 h), in contrast to chondroitin sulfate. Vitronectin
binding was identical whether cells were allowed to attach for 1 h
or overnight. [125I]Vitronectin binding was determined at
the end of all subsequent degradation assays and was nearly identical
to the 6 h time point in Fig. 1A.
To determine the characteristics of vitronectin internalization, medium
aliquots over time were subjected to trichloroacetic acid
precipitations and the degradation of vitronectin was calculated. Degradation was detected as early as 1 h and the amount of
vitronectin degraded continued to increase over the 24-h assay period
(Fig. 1B). At 24 h, approximately 30% (187 ng) of the
added [125I]vitronectin was degraded. Degradation was
specific and required integrin-mediated internalization because either
cold vitronectin (Fig. 1B) or an RGD-containing peptide
(GRGDSP, 200 µM) (data not shown) markedly inhibited
[125I]vitronectin degradation. The random hexapeptide
(SPGDRG, 200 µM) failed to inhibit degradation.
Chloroquine, a lysosomal inhibitor, greatly inhibited degradation
throughout the 24-h assay period (Fig. 1B), indicating that
degradation of the [125I]vitronectin occurred through a
lysosomal pathway. 125I-Labeled native vitronectin was not
degraded to any significant extent (data not shown), as has been
reported for fibroblasts (24). To determine whether integrin
v
3 or
v
5
mediated vitronectin internalization, the astrocytoma cell monolayer
was incubated (30 min) with neutralizing mAbs specific for integrin
v
3 or
v
5
prior to the addition of [125I]vitronectin (Fig.
1C). mAb anti-
v
3 or
v
5 inhibited vitronectin degradation to a
similar extent (30%) throughout the 12-h assay period. When added
together, the antibodies markedly inhibited (70%) vitronectin
degradation throughout the time course, suggesting that both integrins
v
3 and
v
5
mediate vitronectin internalization by astrocytoma cells plated on
fibronectin. Studies were done in the presence of heparin to eliminate
a heparan sulfate proteoglycan-mediated event, and heparin (100 µg/ml) inhibited degradation after 6 h by 40% (Fig.
1D), consistent with the results of other investigators (24).
To determine the integrin receptor(s)
mediating U-251MG cell adhesion to fibronectin, cell adhesion assays
were performed as described (1, 29). Neutralizing mAb anti-integrin
1 completely blocked astrocytoma cell attachment to a
purified fibronectin substrate, and neutralizing mAb anti-integrin
5 and polyclonal anti-integrin
5
1 IgG inhibited attachment to
fibronectin by 80 and 70%, respectively, indicating that integrin
5
1 is the major receptor mediating
fibronectin attachment of these cells (Fig.
2A). In contrast, neutralizing mAbs
anti-integrin
v
3 or anti-integrin
v
5 (data not shown) failed to inhibit
cell adhesion to fibronectin. As controls, we determined the integrin
receptor(s) mediating U-251MG astrocytoma cell attachment to collagen
type I and laminin. Similar to the results of the fibronectin adhesion assay, neutralizing mAb anti-integrin
1 completely
inhibited cell adhesion to collagen (Fig. 2B). Neutralizing
mAb anti-integrin
3 inhibited adhesion by 50%, and
neutralizing mAb anti-integrin
2 as well as the
combination of mAbs anti-integrin
2 and anti-integrin
3 completely inhibited adhesion. However, mAb
anti-integrin
5 had no effect on cell attachment to
collagen, indicating that
1 integrins, but not integrin
5
1, mediate astrocytoma cell adhesion to
collagen. Also, mAb anti-integrin
v
3
failed to inhibit cell attachment to collagen. Similar to fibronectin-
and collagen-adherent cells, mAb anti-integrin
v
3 did not inhibit cell attachment to
laminin (Fig. 2C). mAbs anti-integrin
2,
anti-integrin
3, and anti-integrin
5 also
failed to inhibit cell attachment to laminin. In contrast, mAbs
anti-integrin
1 and anti-integrin
6
inhibited cell attachment to laminin by 70 and 60%, respectively (Fig.
2C). Integrins
2
1,
3
1,
5
1, and
6
1 were expressed on U-251MG cells
assayed by FACS analysis (data not shown), consistent with the reported
results of other investigators (31).
[125I]Vitronectin Degradation by a Human Malignant Astrocytoma Cell Monolayer Plated on Collagen
Because integrin
5
1 has been reported to collaborate with
integrin
v
3 in cell attachment and
migration (18, 19), we investigated whether the absence of integrin
5
1 ligation affected vitronectin
internalization by integrin
v
3. Similar
to the results described for fibronectin-adherent cells, it was found
that on adherence of U-251MG cells to collagen, 6-7% of the added
[125I]vitronectin bound specifically to the cell
monolayer and that this binding was inhibited by heparin and mAbs
anti-integrin
v
3 and anti-integrin
v
5 (data not shown). The pattern, but not the rate, of [125I]vitronectin degradation on collagen
was similar to that observed in cells plated on fibronectin.
Degradation of [125I]vitronectin by cells plated on
collagen was specific and integrin-dependent because it was
inhibited by cold vitronectin or an RGD-containing peptide (data not
shown). Chloroquine inhibited vitronectin degradation (Fig.
3A), indicating that, similar to cells plated
on fibronectin, vitronectin degradation in cells plated on collagen
occurred through a lysosomal pathway. [125I]Vitronectin
degradation was identical whether cells were plated on collagen for
1 h or overnight. To determine whether integrin
v
3 or
v
5
mediated vitronectin internalization, studies with mAbs
anti-
v
3 and
anti-
v
5 were performed. In contrast to
cells plated on fibronectin (Fig. 1C), mAb
anti-
v
3 had no effect, whereas mAb
anti-
v
5 inhibited degradation by 70%
(Fig. 3B). Furthermore, when cells were treated with both
antibodies, there was no increased inhibition over that obtained with
mAb anti-
v
5 alone. These results indicate
that on collagen, astrocytoma cell integrin
v
3 is not capable of mediating
vitronectin internalization.
[125I]Vitronectin Binding and Degradation by a Human Malignant Astrocytoma Cell Monolayer Plated on Laminin
In order
to demonstrate that ligation of integrin
3
1 or
2
1 by
collagen-adherent cells does not negatively regulate the activity state
of integrin
v
3, binding and degradation
studies were performed on laminin-adherent cells. As shown in Fig.
4A, [125I]vitronectin bound
maximally to a laminin-adherent cell monolayer after 6 h, at which
time approximately 4% of the added label was specifically bound. Cold
vitronectin (100 µg/ml) inhibited binding by 80% (data not shown),
indicating that binding was specific. Similar to fibronectin-adherent
cells, mAbs anti-integrin
v
3 and
anti-integrin
v
5 inhibited binding to
laminin by 40 and 50%, respectively, at 24 h, and heparin (100 µg/ml) also inhibited binding. Similar to the results seen on
fibronectin and collagen, [125I]vitronectin degradation
by laminin-adherent cells was detected within 1 h and the amount
of vitronectin degraded continued to increase over the 24 h assay
period (Fig. 4B). Cold vitronectin markedly inhibited
[125I]vitronectin degradation, indicating this is a
specific process. Degradation by laminin-adherent cells was also
inhibited by chloroquine and an RGD-containing peptide (GRGDSP, 200 µM) (data not shown), indicating that degradation was
lysosomal and integrin-mediated, respectively. A random hexapeptide
(SPGDRG, 200 µM) failed to inhibit degradation. To
determine the integrin(s) mediating vitronectin internalization by
laminin-adherent cells, studies with mAbs anti-integrin
v
3 and anti-integrin
v
5 were performed. Similar to
collagen-adherent cells, mAb anti-integrin
v
3 failed to inhibit
[125I]vitronectin degradation by laminin-adherent cells.
mAb anti-integrin
v
5 or the combination
of mAbs anti-integrin
v
3 and
anti-integrin
v
5 nearly abolished
degradation, indicating that on laminin-adherent cells, integrin
v
5 mediates the major portion of
vitronectin internalization.
Expression of Integrins
To determine
whether the inability of integrin v
3 to
mediate vitronectin internalization on collagen or laminin was due to
differential expression of integrins
v
3
and
v
5 on interaction with the two
substrates, FACS analysis was performed. Approximately 40% of the
cells expressed integrin
v
3 and greater
than 90% of the cells expressed integrin
v
5 with a similar mean fluorescent intensity, whether cells were plated on fibronectin (Fig.
5A), on collagen (Fig. 5B), or in
10% serum (data not shown). The cell attachment time (1 h or
overnight) did not affect the results of the FACS analysis. When cells
were adherent to laminin, 70% of the cells expressed integrin
v
3 and 80% expressed integrin
v
5 (Fig. 5C). These results
indicate that differential expression of integrin
v
3 cannot account for the inability of
integrin
v
3 to internalize vitronectin
when these cells are plated on collagen or laminin.
Ligation State of Integrin
To definitively
determine whether ligation of integrin 5
1
was required for integrin
v
3 to
internalize vitronectin, the ability of astrocytoma cells plated on
neutralizing mAbs anti-
5 or anti-
3 to
degrade vitronectin was determined (Fig. 6). Attaching and spreading of cells on neutralizing anti-integrin mAbs has been
described previously (30). The adhesion characteristics of the cells
plated on the mAbs were similar. When astrocytoma cells were plated on
mAb anti-
5, mAb anti-integrin
v
3 and
v
5 each inhibited vitronectin degradation by 60% at 24 h (Fig.
6A) and in combination produced 80-100% inhibition; these
findings were similar to those of fibronectin-adherent cells (Fig.
1C). At 24 h, the specific binding of
[125I]vitronectin to the mAb
anti-
5-adherent monolayer was 5 ± 0.48%, and thus
did not significantly differ from specific
[125I]vitronectin binding to the fibronectin-adherent
cell monolayer (Fig. 1A). In contrast, when cells were
plated on mAb anti-
3, only mAb
anti-
v
5 inhibited degradation (70% at
24 h; Fig. 6B). Furthermore, degradation in the
presence of both mAb anti-
v
3 and mAb
anti-
v
5 was similar to that obtained with
anti-
v
5 alone, indicating that integrin
v
3 was unable to mediate vitronectin internalization when the astrocytoma cells were adherent to mAb anti-
3.
Rate of [125I]Vitronectin Degradation in Astrocytoma Cells Adherent to Fibronectin or Collagen
To determine the
turnover rate of vitronectin by malignant astrocytoma cells plated on
fibronectin or collagen, increasing concentrations of cold vitronectin
were added along with [125I]vitronectin to the monolayer.
[125I]Vitronectin degradation by cells adherent to
fibronectin was saturable at 50 µg/ml vitronectin, and the rate of
degradation determined from medium aliquots obtained between 5 and 60 min was approximately 3 ng/min (Fig. 7A). The
rate of vitronectin degradation in fibronectin-adherent cells
preincubated with mAb anti-integrin v
5
was similar to that obtained in the absence of antibody (Fig.
7A). In contrast, the rate of vitronectin degradation in
fibronectin-adherent cells preincubated with mAb anti-integrin
v
3 was significantly lower (0.10 ng/min
(Fig. 7)) and was comparable to the rate of
[125I]vitronectin degradation on collagen (0.35 ng/min
(Fig. 7)). These data suggest that integrin
v
3 internalizes and degrades vitronectin
at a significantly higher rate than integrin
v
5.
In this report, we demonstrate that altered
[125I]vitronectin binds an astrocytoma cell monolayer
adherent to fibronectin, collagen type I, or laminin in serum-free
conditions in a specific and saturable manner. Subsequent to binding,
astrocytoma cells specifically internalize and degrade vitronectin by a
lysosomal pathway. Integrins v
3 and
v
5 both mediate vitronectin
internalization by cells plated on fibronectin; however, in cells
adherent to collagen type I or laminin, integrin
v
3 fails to mediate vitronectin internalization. Other investigators have recently shown that on
melanoma cells, integrins
v
3 and
v
5 both mediate internalization of
adenovirus (32). This internalization was mediated by cells in
suspension, and it was recently suggested that integrin internalization occurs by different pathways depending on whether the cells are adherent or in suspension (32, 33). Our observations cannot be
accounted for by substrate-dependent differential
expression of
v
3 on astrocytoma cells
because FACS analysis demonstrated no decrease in integrin
v
3 expression when the cells were
adherent to collagen or laminin. Also, vitronectin was available on the cell surface of collagen-adherent cells for internalization because [125I]vitronectin binding to astrocytoma cells plated on
collagen was comparable to that of cells plated on fibronectin (6-7%
at 6 h). Less [125I]vitronectin bound to the
laminin-adherent astrocytoma cell monolayer (4% at 6 h), which
probably accounts for the reduced [125I]vitronectin
degradation (13%) seen at 24 h. Integrin
5
1 mediated astrocytoma cell attachment
to fibronectin in large part, whereas it failed to participate in
astrocytoma cell attachment to collagen type I or laminin, suggesting
that ligation of integrin
5
1 is necessary
for integrin
v
3 to internalize
vitronectin.
To confirm that hypothesis, we performed vitronectin degradation assays
on cells plated on mAb anti-5 or anti-
3.
We found that integrin
v
3 mediated
vitronectin internalization when the cells were adherent to mAb
anti-
5; however, integrin
v
3 failed to mediate internalization when
the cells were adherent to mAb anti-
3, demonstrating
that ligation of integrin
5
1 is necessary for integrin
v
3 to internalize
vitronectin. The degradation assays on purified substrates, taken
together with those on the purified mAbs, provide direct evidence that
ligation of integrin
5
1 positively
regulates the activity state of integrin
v
3. The fact that integrin
v
3 failed to internalize vitronectin when
cells were adherent to laminin and that laminin attachment is mediated
largely by integrin
6
1 without a
detectable contribution from integrin
3
1
indicate that our results are not due to a negative regulation of
integrin
v
3 by integrin
3
1 or collagen. It is unclear why we
observed less inhibition of vitronectin degradation with anti-integrin
mAbs at early time points when the cells were plated on fibronectin or
collagen. This has also been reported by other investigators (20).
However, when the cells were plated on mAb anti-integrin
5 and laminin, complete inhibition of vitronectin degradation was demonstrated at early time points with combined mAbs
anti-integrin
v
3 and
v
5 and mAb anti-integrin
v
5, respectively. Other investigators
have shown that integrin
v
3 ligation and complex formation with integrin-associated protein regulates integrin
5
1-mediated fibronectin phagocytosis in
K562 cells (16, 17). We have yet to determine whether
integrin-associated protein regulates integrin
v
3 internalization by the astrocytoma
cells. Taken together, the results suggest that "cross talk"
between integrins
5
1 and
v
3 can potentially occur in both
directions, depending on the cell type, the ligation status of the
receptor, complex formation with integrin-associated protein, and
probably other conditions.
The rate of vitronectin degradation inhibitable by mAbs anti-integrin
v
3 and
v
5
on astrocytoma cells adherent to fibronectin was approximately 3 ng/min, comparable to that reported in fibroblasts (24) and
approximately 10-fold higher than the rate of vitronectin turnover by
astrocytoma cells plated on collagen and inhibitable by mAb
anti-integrin
v
5. This suggests that
integrin
v
3 internalizes vitronectin at a
significantly higher rate than integrin
v
5 on astrocytoma cells. These data
suggest that integrins
v
3 and
v
5 are regulated differently on
astrocytoma cells, consistent with their different cytoplasmic tail
sequences and the differences in regulation reported by other
investigators on other cell types (4, 9-11, 34, 35). The data also are
consistent with the hypothesis that integrins
v
5 and
v
3
mediate different vitronectin-dependent functions in
malignant astrocytomas in vivo.
In summary, these studies demonstrate that malignant astrocytoma cells
are capable of remodeling their extracellular matrix through the
internalization and lysosomal degradation of altered vitronectin. In
astrocytoma cells this process is constitutively mediated by integrin
v
5, and integrin
v
3 participation in vitronectin
internalization is dependent on ligation of integrin
5
1. These results are physiologically
relevant because
1 integrins are predominantly expressed
by perivascular astrocytoma cells (1), and in these experiments, the
cells were adherent to matrix proteins recognized by
1
integrins. Our data indicate that integrin
v
3 internalization of its ligand(s)
contributes to the remodeling of the matrix by astrocytoma cells and
that integrin
v
3 requires a signal from
integrin
5
1 to mediate this process.
We thank J. Robert Grammer for technical assistance and Dr. John R. Couchman (University of Alabama at Birmingham, Birmingham, AL) for reviewing the manuscript.