Thrombospondin Modulates OLv~3 Function through Integrin-associated Protein
Ai-Guo Gao, Frederik P. Lindberg, Julie M. Dimitry, Eric J. Brown, and William A. Frazier
Departments of Biochemistry and Molecular Biophysics and Medicine and Infectious Disease, Washington University School of
Medicine, St. Louis, Missouri 63110
Abstract. Integrin-associated protein (IAP) is a recep-
tor for the carboxyl-terminal "cell-binding domain"
(CBD) of thrombospondin 1 (TS1). lAP associates with
av[33integrin and mAbs against lAP inhibit certain in-
tegrin functions. Here we examine the effects of the
TS1 CBD and 4N1K (KRFYVVMWKK), a cell-bind-
ing peptide derived from it, on the adhesion and
spreading on vitronectin (VN) of C32 human mela-
noma cells which express lAP, 0tv133, and avl35.Cells ad-
here to VN at low surface densities via etv[35and spread
very slowly while adhesion to higher density VN in-
volves both etv135and Otv[33and results in rapid spread-
ing. Spreading of the cells, but not adhesion, on sparse
VN coatings is markedly enhanced by the presence of
soluble TS1, the recombinant CBD and 4N1K, but not
the "mutant" peptide 4NGG, KRFYGGMWKK,
which fails to bind IAP. This enhanced spreading is
completely blocked by mAb LM609 against 0tv133 and
the anti-lAP mAb B6H12. Correlated with this en-
hanced spreading is increased tyrosine phosphorylation
of focal adhesion kinase (FAK), paxillin, and a protein
of ca. 90 kD. The enhanced spreading induced by TS1
and 4N1K and the constitutive spreading on higher
density VN are both blocked by calphostin C (100 nM),
wortmannin (10 nM), and tyrosine kinase inhibitors. In
contrast, pertussis toxin specifically blocks only the TS1
stimulated spreading on low density VN, indicating that
lAP exerts its effects on signal transduction via a het-
erotrimeric Gi protein acting upstream of a common
cell spreading pathway which includes PI-3 kinase,
PKC, and tyrosine kinases.
T
HE family of thrombospondins has been implicated
in acute regulation of a number of diverse physio-
logical processes such as platelet aggregation, in-
flammation, angiogenesis, cell adhesion and migration,
and other developmental processes (5, 6, 16, 61). The pro-
totypic member of this family is platelet thrombospondin
or thrombospondin-1 (TS1).1 Many cell interaction sites
and peptides have been identified within the modular TS1
structure and as many as a dozen different receptors, in-
cluding several 131and 133integrins have been proposed to
mediate the effects of TS1 on cells (12, 16, 33, 70). Of this
plethora of putative receptors, the direct binding of TS1 to
molecularly characterized receptors has been demon-
strated only for HSPGs (40), sulfatides (49), CD36 (1), and
the integrins Otllb133 and etv133(33). In the case of the two in-
tegrins, binding of native TS1 has been demonstrated for
some cell types, while for other cells, the TS1 RGD se-
Address all correspondence to William A. Frazier, Biochemistry and Mo-
lecular Biophysics, Box 8231, Washington University Medical School, 660
S. Euclid Avenue, St. Louis, MO 63110. Tel.: (314) 362-3348. Fax: (314)
362-7183. E-mail: frazier@biochem.wustl.edu
1. Abbreviations used in this paper: CBD, carboxy terminal domain; EC,
endothelial celI;FAK, focal adhesion kinase; HUVEC, human umbilical
vein endothelial cell; lAP, integrin-associated protein; LM, laminin; TS1,
thrombospondin-1; VN, vitronectin.
quence appears to be cryptic (62). A number of peptide
epitopes within the complex TS1 structure have been iden-
tified with specific functions of the protein. For example,
sequences within the procollagen-like domain, and two of
the type 1 (properdin- or malaria-like) repeats inhibit an-
giogenesis in vivo and endothelial cell (EC) motility in
vitro (65). Also, a peptide from the amino-terminal hep-
arin-binding domain reverses focal adhesions in fibro-
blasts and EC (41, 42) and a tripeptide in one of the type 1
repeats is capable of activating latent TGF13 thus poten-
tially influencing cell behavior via this pathway (53).
In addition to these active sequences in the central pro-
tease-resistant stalk-like region of the TS1 subunit, the
carboxy terminal domain of TS1 interacts with many cell
types and has been designated the cell-binding domain or
CBD. The mAb C6.7 directed to the CBD, has been used
to implicate this domain of TS1 as a stimulator of motility
of a number of cell types (37, 38, 64, 71, 72) and to estab-
lish a role for TS1 in secondary phase or secretion-depen-
dent platelet aggregation (14). To better define the struc-
ture-function relationships of the CBD, we expressed it in
bacteria exclusive of the nearby RGD sequence of TS1
and established that its cell-binding activity was indepen-
dent of the RGD site (29). Using overlapping peptides we
then defined two related active peptide sequences within
the 221 residues of the CBD: RFYVVM (the minimal C4
(c) The Rockefeller University Press, 0021-9525/96/10/533/12 $2.00
The Journal of Cell Biology, Volume 135, Number 2, October 1996 533-544 533
peptide) and IRVVM (the minimal C7 peptide) (27, 28).
The peptides competed with one another in cell binding
assays and both affinity labeled a 52-kD membrane glyco-
protein CBD receptor candidate (17). We have recently
identified this 52-kD protein as the widely distributed inte-
grin-associated protein (lAP or CD47) (7, 18, 35, 36). lAP
is found in association with av133and other integrins, and it
has been implicated in modulating integrin functions re-
quired for Ca++ fluxes (54), phagocytosis (4), and cellular
migration events such as transendothelial (11) and trans-
epithelial (45) migration of PMNs. The function-blocking
anti-lAP mAb B6H12 abolishes chemotaxis of human um-
bilical vein endothelial cells (HUVECs) towards the CBD
and C4 peptides (18) under conditions in which the HU-
VECs employ ~v133integrin for locomotory traction. Fur-
thermore, both TS1 and RFYVVMWK peptide stimulate
a Ca ++ transient in IMR90 fibroblasts that is inhibited by
mAb C6.7 directed against the CBD (in the case of TS1)
and by anti-IAP mAb B6H12 (66). Taken together these
data suggest that the CBD can stimulate a chemotactic re-
sponse by a concerted mechanism in which Ca++ regu-
lated events necessary for cell motility are activated by a
Ca ++ flux and at the same time, the function of ~v133is
modulated.
The most studied example of 133integrin activation is the
platelet integrin eqtb133which is maintained in a state of
low affinity and avidity on circulating platelets, but is rap-
idly activated to bind fibrinogen by thrombin, epineph-
rine, and ADP acting via seven transmembrane-spanning
(7TMS or heptahelical) receptors (56). The signal trans-
duction pathway by which these receptors "costimulate"
fibrinogen binding by aItb133is incompletely understood
and complex, involving proximal heterotrimeric G pro-
teins, lipid mediators derived from arachadonic acid via
both cyclooxygenase and lipoxygenase, which lead to the
activation of PKC and small G proteins of the rho family
(55, 57). Other examples of integrin activation such as
the leukocyte CDlla,b,c/CD18 (132)family also involve
costimulation from 7TMS chemokine receptors acting
through heterotrimeric and rho family G proteins, lipid
mediators, and PKC (60). There is also accumulating evi-
dence that integrin-mediated cell spreading on a variety of
matrix proteins requires signal transduction events with
features common to the pathways of integrin activation
and costimulation, suggesting that some integrins are con-
stitutively activated or can "autoactivate" to cause cell
spreading. Chemotaxis and spreading of cells are closely
allied functions which depend on similar activation states
of the integrin and similar downstream signal transduction
events (9, 15, 25, 55).
Given the previously established role of lAP in etv133in-
tegrin signaling (4, 7, 36, 54, 66, 75), we have investigated
the potential role of the CBD-IAP interaction in modulat-
ing the 133 integrin-mediated adhesion and spreading of
cells on vitronectin-(VN) coated surfaces. We have chosen
a cell type that expresses IAP and the VN receptors etvl33
and av135.Under appropriate conditions of low VN surface
density, the initial adhesion event is mediated by ~v135
while subsequent spreading of the cells requires etv133.
Spreading of unstimulated cells occurs only slowly, but
addition of TS1, its recombinant CBD or the RFYVVM
containing CBD peptides all greatly accelerate spreading
with the concomitant tyrosine phosphorylation of focal
adhesion kinase (FAK) and associated proteins. This lAP-
dependent response to the CBD requires a pertussis toxin-
sensitive, heterotrimeric G protein. In contrast, constitu-
tive spreading on high density VN surfaces is unaffected
by pertussis toxin while both TSl-stimulated and constitu-
tive spreading are blocked by inhibitors of PI-3 kinase,
PKC, and tyrosine kinases. Thus, ligation of lAP by the
TS1 CBD stimulates a Gi-like protein, which then appears
to lead to the activation of PI-3 kinase, protein kinase C,
and tyrosine kinases, elements common to the activation
of integrin-dependent cell spreading in many systems (2,
25, 34, 44, 55, 67, 69).
Materials and Methods
Reagentsand Antibodies
Human vitronectin, fibronectin, and collagens I and IV were obtained
from Collaborative Biomedical Products (Bedford, MA). Fibrinogen was
from Sigma Chem. Co. (St. Louis, MO). The rCBD was expressed as a
Hisr-tagged protein in E. coli using the pQE30 vector (Qiagen, Chatsworth,
CA) and purified by Ni-NTA chromatography (Dimitry, J., A. Jurkovich
and W. Frazier, manuscript in preparation). Monoclonal antibodies
LM609 (anfi-av133) and PIF6 (anti-C~v135)were the generous gifts of Dr.
D. Cheresh (Scripps Research Institute). B6H12 and 2D3, anti-human
IAP mAbs, have been previously described (7, 35, 36). mAbs 4G10 (anti-
phosphotyrosine) and 2D7 (anti-p125FAK,for immunoprecipitation) were
from Upstate Biotechnology Inc. (Lake Placid, NY). Western blotting
mAb against FAK, phosphotyrosine (PY20), and paxillin were the prod-
ucts of Transduction Laboratories (Lexington, KY). All protein kinase in-
hibitors were purchased from CalBiochem (La Jolla, CA). Rainbow pro-
tein molecular weight markers and ECL Western blotting detection kit
were from Amersham Corp. (Arlington Heights, IL). Anti-mouse IgG-
agarose and other reagents were from Sigma. All peptides used were syn-
thesized by the Protein and Nucleic Acid Chemistry Laboratory of Wash-
ington University School of Medicine as described previously (17, 28).
Peptides were evaluated by mass spectrometry before and after purifica-
tion on high performance liquid chromatography. Human platelet TS1
was purified as described (50). The amino acid sequences of the TS1 peptides
used in this study are 4NK, KRFYVVMWKQVTQSKKY (native sequence
1016-1029); 4N1K, KRFYVVMWKK; 4N7G, KRFYVVMGKK; 4NGG,
KRFYGGMWKK; Hep III, TRDLASIARLRIAKGVNDNF (170-190);
MAL III, SPWDIASVTAGGGVQKRSR (481-499). The synthetic pep-
tides GRGDSP and GRGESP were from GIBCO-BRL (Gaithersburg, MD).
Cell Culture,Adhesion, and Spreading
C32 human melanoma cells (ATCC CRL 1585) we're cultured in Gibco's
Minimum Essential Medium (MEM) supplemented with 10% FCS. Mu-
rine 3T3 fibroblasts (ATCC CRL 1658) were grown in DMEM plus 10%
calf serum. Cell adhesion assays were performed as described previously
(28). Briefly, 96-well plates were coated with substrates overnight at 4C.
C32 cells were harvested with 0.04% EDTA in PBS and washed twice
with PBS. Cells were resuspended in Hepes-buffered saline (HBS; 20 mM
Hepes/NaOH, pH 7.4, 0.15 M NaC1, 1 mM CaC12, and 1 mM MgC12) al-
lowed to adhere to the immobilized substrates for 90 min at 37C in the
presence of indicated additions in solution. After washing, the adherent
cells were lysed in 1% Triton X-100 and quantified by endogenous phos-
phatase activity with p-nitrophenyl phosphate as a substrate (46). For cell
spreading experiments, glass chamber slides (LAbTek) or 12-well tissue
culture plates were coated with VN overnight at 4C. C32 cells were har-
vested for cell adhesion experiments and resuspended in HBS containing
calcium and magnesium, both at 1 mM. Indicated additions of stimulators
and inhibitors were also added to cell suspensions at this time. Cell
spreading at 37C was followed by washing, fixing, and staining with a
LeukoStat stain kit (Fisher Scientific, Pittsburgh, PA). For the inhibition
of cell attachment or spreading, C32 cells were pre-incubated with anti-
bodies or inhibitors for 15 rain at 0C before the assay. The required light
activation of calphostin C (8, 19, 26) was also performed during this pre-
incubation time.
The Joumal of Cell Biology, Volume 135, 1996 534
FACSAnalysis of the Expressionof Cell
Surface Receptors
The expression of integrins and IAP was detected by fluorescent flow cy-
tometry. C32 cells were harvested and washed as described above and re-
suspended in C32 cell culture medium (with 10% FCS). The mAbs LM609
vs ctvp3, P1F6 vs ctv135,or 2D3 vs IAP were added to aliquots of cell suspen-
sions and incubated for 60 min at 4C with rocking. After several washes
in PBS, the aliquots were stained with FITC-labeled anti--mouse second-
ary antibody (Pierce, Rockford, IL) for another 60 min in the cell culture
medium, washed again with PBS, and then analyzed by flow cytometry.
The second antibody and the mouse IgG, instead of primary antibody,
were both used as negative controls and mAb W6/32 against HLA was the
positive control.
lmmunoprecipitation Studies
C32 cell spreading was carried out as described above. At indicated times,
cells were lysed by addition of 6x precooled modified RIPA lysisbuffer. 1x
RIPA lysis buffer is 50 mM Tris/HCl, pH 7.4, 0.15 M NaC1, 1% (wt/vol)
NP-40, 0.5 % (wt/vol) sodium deoxycholate, 1 mM EGTA, ] mM Na3VO4,
and protease inhibitors consisting of 10 I~g/ml each of antipain, pepstatin
A, chymostatin, leupeptin, soybean trypsin inhibitor, aprotinin, and 1 mM
phenylmethylsulfonylfluoride. Cell lysis was completed by 30 min rocking
at 4C followed by microcentrifugation at top speed (13,000 rpm) for 30
min. The soluble material from equal numbers of cells (or equal amounts
of protein) was incubated with the specified monoclonal antibody for 3 h
at 4C and immunoprecipitated with anti-mouse IgG-agarose. The precip-
itates were extensively washed with lx RIPA buffer, dissolved, and
boiled in a small volume of SDS-sample buffer.
Identification of ImmunoprecipitatedProteinsby
Western Blot Analysis
Proteins were separated by SDS-PAGE on 4-12% or 8% precast Tris-
Glycine gels (NOVEX, San Diego, CA) and transferred to nitrocellulose
membranes. Blots were blocked with 3% BSA plus 3% dried milk in TBS-T
(0.1% Tween 20, 50 mM Tris/HC1, pH 7.4, 0.15 M NaCl) overnight and
probed with the indicated mAb at 1 tLg/ml in TBS-T plus 10% blocking
buffer for 1-2 h, washed, and incubated with a 1:10,000 dilution of horse-
radish peroxidase-conjugated goat anti-mouse IgG (Pierce). Detection
was by chemiluminescence. For identification of coimmunoprecipitated
protein(s), the same blot was stripped at 68C (2x, 20 rain) in stripping
buffer (62.5 mM Tris/HCl, pH 6.8, 2% SDS, 100 mM p-ME). The mem-
brane was reprobed with the indicated mAb and processed as described
above.
Results
Spreadingof C32 Cellson VN
The spreading of C32 cells was examined on surfaces
coated with a concentration range of VN (0.1-10 ixg/ml).
Below ca. 3 i~g/ml VN, spreading was very slow, requiring
1.5-2 h at 37C to develop fully. At 5-10 ~g/ml spreading was
relatively rapid, being complete within 30 min (not shown).
The effect of soluble TS1, its recombinant CBD and pep-
tides on C32 cell spreading on VN was tested. As seen in
Fig. 1, all three agents dramatically accelerated spreading
on suboptimal I Ixg/ml VN-coated surfaces. Fig. 1 A shows
the complete lack of spreading observed on VN (1 ixg/ml)
at 30 min, while Fig. 1, B-D show fully spread, polygonal
cells at 30 min in the presence of TS1, the rCBD and 4NIK
in solution. This cell spreading data is quantified in Fig. 2.
When coated on the surface in the absence of VN, TS1
supports only very weak spreading of these cells, while the
rCBD and 4NIK (and related peptides from the CBD)
support no spreading at all when immobilized (not shown).
A VN-coated surface is required for spreading to take
place since no spreading occurs on BSA-coated surfaces
with or without TS1, the rCBD, and its peptides (not
shown). To determine if TS1 and its fragments could stim-
ulate cell spreading on other matrix proteins besides VN,
C32 melanoma cells were plated on concentration gradi-
ents of fibronectin, collagens I and IV, gelatin, fibrinogen,
and human and mouse laminin (LM) (as well as VN as a
positive control). Concentrations used for coating were
determined in preliminary experiments and covered a
range from no visible spreading to fully spread cells during
a ]-h incubation. The C32 cells spread well on the higher
densities of all these proteins except fibrinogen. The addi-
tion of 50 IxM 4N1K or 50 ixg/ml TS1 did not stimulate cell
spreading at any density of fibronectin, collagens, or fi-
brinogen. However, 4N1K and TS1 did stimulate cell
spreading on lower densities of gelatin and LM (and VN).
Attachment to gelatin, like VN, was inhibited by blockade
Figure1. TS1, rCBD, and 4NIK all accelerate C32 cell spreading on immobilized vitronectin. C32 cells were allowed to spread on I ~g/
ml VN-precoated glass slides for 30 min at 37C, followed by fixing and staining. The following were included in solution: (A), no addi-
tion; (B), TS1 at 50 txg/ml;(C), recombinant CBD at 50 ~g/ml; (D), 50 I~M4N1K peptide. BSA or peptides from other domains of TS1 did
not affect cell spreading (not shown).
Gao et al. Thrombospondin Modulates Via lntegrin-associated Protein 535
100
80
W
"0
8
~. 40
n
20
Figure 2. The effect of TS1, rCBD, and peptides on C32 cell
spreading. C32 cell spreading assays were performed as in Fig. 1.
After washing, fixing, and staining, cells with a diameter at least
three times larger than that of round unspread cells were counted
as spread cells. Results are expressed as the mean percentage
(_+SD) of spread cells compared to total attached cells in six high
power fields from three separate experiments. The peptide se-
quences are described in Materials and Methods and all peptides
were used at 50 IxM in solution. Peptides 4NGG (KRFYGGM-
WKK) and 4N7G (KRFYVVMGKK) are "mutants" of 4N1K
while Mal III and Hep III are cell-binding peptides from two
other domains of TS1.
of ~vl35and spreading was inhibited by anti-etvl33 (LM609)
and anti-IAP (B6H12) (not shown). Attachment to LM
was blocked by an anti-131 mAb, not by LM609 or P1F6,
and stimulated spreading on LM was blocked by B6H12
but not by LM609 (not shown) suggesting that ligation of
LM609 is not required for the action of IAP. To determine
if TS1 and its CBD fragments could stimulate spreading of
other cell types, murine NIH3T3 cells were tested for at-
tachment and spreading on VN and the other matrix pro-
teins above. As with the human C32 cells, TS1 and 4N1K
stimulated spreading on low densities of VN, gelatin, and
LM, but not other matrix proteins (not shown).
The possibility that contaminants in the platelet TS1
preparation were responsible for the stimulation of cell
spreading is ruled out by the potent stimulatory activity of
the rCBD produced in bacteria and purified on Ni-NTA
resin (as opposed to heparin-agarose for TS1) and the ac-
tivity of the synthetic peptide 4NIK. Other versions of the
CBD peptide containing the RFYVVM sequence in the
context of more TS1 native sequence (17, 28) were all ac-
tive in stimulating cell spreading, while peptides from
other domains of TS1, including other cell-binding pep-
tides such as Mal III and Hep III, were without effect (Fig. 2,
see Materials and Methods for peptide sequences). The
other VVM containing peptide from the CBD, FIRVV-
MYEGKK (17, 28), was also active in stimulating cell
spreading.
Our previous studies of cell attachment to the CBD pep-
tides indicated that the VVM sequence is important for
recognition of IAP (17, 18, 28). Thus "mutant" 4N1K pep-
tides were synthesized in which residues between the KR
and KK termini of 4NIK were replaced with G, and one in
which the VV sequence was replaced with GG (4NGG).
These peptides were tested in a C32 cell adhesion assay
which reflects IAP-dependent binding of cells to the im-
mobilized peptide (18), and in the cell spreading assay
(Fig. 2). Peptides 4N3G (KRFGVVMWKK) and 4N7G
(KRFYVVMGKK) had partial activity in both assays
while peptide 4NGG (KRFYGGMWKK) was completely
inactive in cell spreading (Fig. 2) and in binding cells (not
shown). These results are important since the mutant pep-
tides retain the precise charge distribution of the parent
peptide.
In summary, these data are important in several re-
spects: they demonstrate that the accelerated spreading is
not due to adsorption of the TS1, the rCBD, or peptides to
the surface in a nonspecific manner, since the rCBD and
peptides support no spreading at all when directly coated
on the surface, and only spreading on VN, LM, and gelatin
exhibits the effect (and all surfaces are blocked with BSA).
Second, only spreading on certain proteins, and thus medi-
ated via certain integrins, is affected by the ligation of IAP
m
.1 1 FL1
lAP
"'I '
O~V~3
I
1
~'"'i~b ' ' ~b'~
~
=
av[15
...............
t.~i"~o~ i~b ' ' ~b'~i
Figure 3. Expression of lAP and integrins Ot.v~3 and av135on C32 cells. C32 ceils were stained with mAbs 2D3 (anti-lAP, left), LM609
(anti-integrin tXv133,middle), and PIF6 (anti-integrin tXv~35,right) followed by fluorescein-conjugated goat anti-mouse IgG. The controls
were stained with mouse IgG as the primary antibody, the distribution of which is plotted with each indicated mAb for comparison.
The JournalofCell Biology,Volume 135, 1996 536
Figure4. Characterization of C32 cell attachment to vitronectin. Plastic
wells were coated with VN at the indicated concentrations overnight at
4C. C32 ceils were preincubated for 15 min with the indicated mAb or
peptide, and then transferred to VN-coated wells for the cell attachment
assay as in Materials and Methods. The mAbs were each present at 50
~g/ml and the peptides GRGDSP or GRGESP were at 50 IxM.TS1 (50
ixg/ml) or 4N1K peptide (50 I~M) were added right before plating 5 x
104 C32 cells per well. After washing, attached cells were quantified
with endogenous phosphatase activity. (A) Effect of TS1, mAbs, and
peptides on C32 cell attachment to VN coated at 1 p.g/ml. (B) At higher
surface densities of VN both mAbs LM609 and P1F6 are required to
block cell adhesion. (C) 4N1K peptide does not augment av133-depen-
dent cell adhesion when ~v135is blocked with mAb PIF6.
with the TS1 rCBD or peptides. Third, the effect is seen in
another species of cell which is not transformed. Fourth,
the VV sequence of 4N1K is critical for both binding to
IAP and stimulation of cell spreading.
Adhesion of (?32 Cells
We next examined the receptors used by the melanoma
cells for interaction with surfaces coated with VN over a
range of concentrations (0.1-10 ~g/ml). Flow cytometry
was used to determine the levels of IAP and two common
VN receptors etv133and ctv135,expressed on the surface of
C32 cells. As seen in Fig. 3, IAP, O~v[33,and t~vl35are all rel-
atively abundant. Antibodies to these receptors were used
to determine which participated in cell adhesion to VN-
coated surfaces. Fig. 4 A shows that an mAb vs Ctv135
(P1F6) effectively blocked adhesion to VN coated at a low
density (1 Ixg/ml) while anti-ctv133mAb (LM609) had no ef-
fect. As expected, RGD (but not RGE) peptide also
blocked adhesion to VN. Thus, Ctvl33is expressed on C32
cells (Fig. 3), but did not mediate binding at the low con-
centration of VN used. We then examined the attachment
of cells to higher surface densities of VN, and found that as
the amount of VN adsorbed on the surface is increased, a
larger fraction of adhesion is mediated by av133as judged
by the fact that inhibition requires mAbs against both Otvl33
and etvl35(Fig. 4 B).
To determine if TS1 or 4N1K could increase the appar-
ent affinity or avidity of etv133, adhesion to VN was per-
formed in the presence of TS1 and the peptide. Addition
of TS1 or 4N1K peptide to the soluble phase had no effect
on the number of cells adhering to VN (Fig. 4 A), even
though TS1 and an active CBD cell-binding peptide 4N1K,
bound cells quite well when they were coated directly on
the plastic, while several control TS1 peptides and BSA
did not bind cells to an appreciable extent (not shown).
Two anti-lAP mAbs (B6H12 and 2D3) also had no effect
on cell attachment to VN (Fig. 4 A), consistent with the
lack of effect of peptide 4N1K, an lAP ligand, on the num-
ber of cells adhering. If etv133were activated by 4N1K, the
proportion of adhesion mediated by av133would be greater
in the presence of the peptide. However, the degree of in-
hibition by LM609 was not increased at any VN surface
density tested (not shown). Also, when av135was blocked
by P1F6 mAb, the addition of 4N1K did not increase cell
adhesion to VN, indicating a lack of effect on adhesion via
Gao et al. Thrombospondin Modulates Via lntegrin-associated Protein 537
Figure5. Effect of mAbs and kinase inhibitors on C32 cell spreading on vitronectin. After 15 min pre-incubation with mAbs or inhibi-
tors (see Materials and Methods for details), cell spreading was allowed to proceed for 30 min on 1 t~g/mlVN-coated glass slides. (A) 50
p,M soluble 4N1K stimulated C32 cell spreading as before; (B) mAb LM609 at 20 ~g/ml completely abolished spreading stimulated by
4N1K. mAb 2D3, a nonfunction-blocking mAb against IAP, had little effect at 100 txg/ml(C) while B6H12 (D), an anti-lAP function-
blocking mAb, completely prevented spreading at 100 ixg/ml.The PKC inhibitor calphostin c (at 100 nM) (E) and wortmannin (at 10
nM) (F), a PI-3 kinase inhibitor, both abolished spreading stimulated by 4N1K as well as the spreading which occurred on higher density
VN without 4N1K (not shown).
Otv[~3 or any other receptor (Fig. 4 C). Thus, it appears that
TS1 and 4NIK peptide do not increase the affinity/avidity
of av[~3even though TS1 and its CBD derived fragments
have dramatic effects on cell spreading on VN (Fig. 1). This
suggests that effects on cell spreading are mediated by stim-
ulation of an IAP-dependent signaling pathway by TS1.
Characterization of TSI-stimulated Spreading
We first determined the effect of mAbs on the 4NIK stim-
ulated spreading of C32 cells. Fig. 5 A shows the stimula-
tion (at 30 min) by 50 IxM 4NIK of cells attached to 1 p.g/ml
VN. Figs. 5 B and 6 show that mAb LM609 directed
against the VN receptor Otv[33,results in complete inhibi-
tion of cell spreading, even though, as seen in Fig. 4 A, ini-
tial attachment of the cells to VN is mediated solely by
etv[35.Thus, it appears that the stimulation of spreading is
mediated by etv133.At higher coating concentrations of VN,
where spreading is more rapid and stimulation by TS1 is
not as easy to discern, spreading is also blocked by anti-
body to Otv~3 and anti-O~vl35has no effect (not shown). This
suggests that higher surface densities of VN alleviate the
need for an activating stimulus, perhaps by cross-linking
the integrin (55). To determine if av135might become acti-
vated and contribute to cell spreading, cells were allowed
to attach to VN for 15 min in the absence of an mAb or pep-
tide, which they cannot do in the presence of mAb P1F6
(Fig. 4 B). Then, either mAb LM609 or PIF6 was added
along with a stimulatory peptide (4N1K). After an addi-
tional 30 min, cells in control mAb or P1F6 had spread
equally well while mAb LM609 completely blocked
spreading as before (Fig. 6). That the stimulation of
spreading by TS1 requires IAP is seen in Fig. 5, C and D.
The function-blocking anti-lAP mAb B6H12 inhibits the
stimulated spreading while the nonfunction blocking anti-
lAP mAb 2D3 has no effect. These results are quantified
in Fig. 6. TS1, the rCBD and 4N1K all stimulate cell
spreading on VN to the same extent: ~80% of total cells
spread in 30 min vs ca. 15% spread with no stimulation. As
seen in Fig. 6, both mAbs LM609 and B6H12 block the
The Journal of Cell Biology, Volume 135, 1996 538
100
80
w
6o
"O
II1
i~ 40
20
od co o~ co
o o ,,
Figure6. Quantitative analysis of C32 cell spreading on immobi-
lized vitronectin. C32 cell spreading assays were performed as in
Fig. 5 and quantified as Fig. 2. "*" indicates spreading in the ab-
sence of 4NIK (negative control). 50 p.M 4NIK peptide was
present in all other cases plus: mouse IgG, 100 Ixg/ml;LM609,
20 Ixg/ml;B6H12, 100 ~g/ml; 2D3, 100 ixg/ml.In separate experi-
ments, C32 cells were allowed to attach to VN for 15 min in the
absence of a mAb or 4N1K peptide. Then, either mAb LM609 or
P1F6 was added along with peptide 4N1K followed by an addi-
tional 30 min spreading incubation. Data from these experiments
are indicated with "**"
stimulation of spreading to about the same extent, nearly
the unstimulated level. The same results were obtained
when spreading was stimulated by TS1 at 50 ~g/ml (not
shown). These data clearly indicate that both av~3 and
IAP are required for the stimulation of cell spreading on
VN by the CBD of TS1. One possibility is that IAP liga-
tion causes an increase in the affinity or avidity (by cluster-
ing) of Otvl~3.If this occurs, then 4N1K and TS1 should en-
hance the adhesion of cells to low density immobilized VN.
As seen in Fig. 4, A and C, this does not occur. Over a wide
concentration range of VN coating, 4N1K peptide never
increases adhesion, only spreading. These results further
suggest that etvl~
3 can form a signal transducing complex
with IAP, as observed in other systems (4, 7, 36, 54, 66,
75), which can lead to the stimulation of cell spreading.
Effect of TS1 and 4N1K on Tyrosine Phosphorylation
Integrin-mediated spreading of many cell types is accom-
panied by the phosphorylation on tyrosine residues of a
set of cellular proteins involved in focal adhesion assembly
and turnover (9, 21, 23, 25, 39, 52). These include focal ad-
hesion kinase (pp125, FAK) and paxillin (pp60-70 triplet).
We thus wanted to determine if the stimulated cell spread-
ing observed with TS1 and its CBD led to tyrosine phos-
phorylation of any such proteins. To do this, C32 cells
were allowed to spread on VN-coated surfaces (1 ~g/ml)
in the absence (HBS) or presence of TS1 or 4N1K for var-
ious times when the cells were rapidly lysed, immunopre-
cipitated with an anti-phosphotyrosine (PY) mAb, the
precipitated proteins run on SDS-PAGE and Western
blotted with the anti-PY mAb 4G10. As seen in Fig. 7 A,
TS1 and 4N1K stimulate the tyrosine phosphorylation of
several proteins with Mr of ca. 125 kD (a), ca. 90 kD (b),
and a cluster at 60-70 kD (c) running just above the mAb
heavy chain. In an attempt to identify these phosphory-
lated proteins, specific antibodies to known signal trans-
duction components were employed. Fig. 7 B shows the
results of an experiment in which cell lysates were treated
like those in Fig. 7 A except that the precipitating antibody
was anti-FAK mAb 2D7 (51). Here it is seen that phos-
phorylated FAK detected with mAb 4G10, appears much
earlier in the time course and accumulates to a greater ex-
tent in cells spreading on VN in the presence of either
4NIK or TS1. In addition, other phosphoproteins are seen
presumably as a result of their association with FAK, with
Mr of ca. 140, 95, and 90 kD. This latter 90-kD protein is of
interest because it becomes phosphorylated earlier than
FAK itself, and in the TS1- and 4N1K-treated cells it ei-
ther dissociates from FAK or becomes dephosphorylated
(or both) by the last time point. When C32 cells are main-
tained in suspension so that focal adhesion formation is
prevented, the addition of 4N1K causes rapid and intense
tyrosine phosphorylation of this 90-kD protein, but phos-
phorylation of FAK and other proteins is not observed.
This also suggests that phosphorylation of this 90-kD pro-
tein is an early event following IAP ligation. Preliminary
experiments indicate that this pp90 species is not phos-
phatidylinositol-3 kinase.
Pertussis Toxin Specifically Inhibits Cell Spreading
Stimulated by 4N1K and TS1
To begin defining a mechanism for the stimulation of av133-
dependent cell spreading by ligation of IAP, we tested a
number of well characterized inhibitors of signal transduc-
tion enzymes. Even though IAP can associate with av133,
previous studies have indicated that signal transduction in-
volving PKC is required for modulation of integrin-medi-
ated functions (75). As noted above, PKC activation is
also a common element in integrin-mediated cell spread-
ing (31, 69). We thus tested a number of protein kinase in-
hibitors for their effect on the rapid spreading elicited by
4N1K and TS1. These included six ser/thr kinase inhibitors
of varying specificities (HA1004, chelerythrine, H7, K252b,
Ro-31-8220, and calphostin C, see Materials and Methods
for details). Of these, the specific protein kinase C inhibi-
tor calphostin C completely inhibited the stimulated cell
spreading (Fig. 8) at a concentration of 100 nM which is
only twice the reported ICs0 (26). Chelerythrine, also a
PKC inhibitor, was effective at higher concentrations (Fig. 8).
Moreover, PMA treatment of C32 cells adherent to sparse
VN coatings rapidly induced cell spreading, demonstrating
that direct PKC activation could induce the same behavior
as TS1. However, the PKC inhibitors (above) also blocked
the rapid constitutive spreading on high densities of VN in
the absence of TS1 or its peptides. This along with previ-
ous reports from others (31, 69) indicates that PKC plays a
general role in spreading, and is not specific to the action
of lAP. Wortmannin, genestein, and herbimycin A also
abolished cell spreading at appropriate concentrations
(Fig. 8). However, as with PKC inhibitors (above), none of
these inhibitors were specific for TSl-induced cell spreading.
Gaoetal. Thrombospondin
ModulatesVialntegrin-associatedProtein 539
Figure7. TS1 and peptide 4N1K stim-
ulate protein tyrosine phosphorylation.
C32 cells were allowed to spread on 1
i~g/mlVN-coated 12-weUtissue culture
plates with or without 50 txg/mlTS1 or
50 ~M 4N1K peptide in solution. At
the indicated times (min) cells were
lysed in RIPA buffer (see Materials
and Methods). Soluble material from
equal numbers of cells (equal amounts
of protein) was incubated with anti-
phosphotyrosine mAb PY20 (A) or
anti-p125FA~mAb 2D7 (B) and precip-
itated with anti-mouse IgG agarose.
SDS-PAGE on 4-12% (A) or 8% (B)
acrylamide gels was followed by West-
ern blotting with anti-phosphotyrosine
mAb 4G10 with ECL detection.
The possible involvement of the Gi/Go class of hetero-
trimeric G proteins in the stimulation of cell spreading on
VN was tested by incubating the C32 cells overnight in B.
pertussis toxin (20-300 ng/ml). The cells were then plated
on 1 p~g/mlVN-coated slides in the presence or absence of
TS1 and 4N1K peptide as above. Interestingly, the pertus-
sis toxin pretreatment (all concentrations) strongly inhib-
ited the ability of TS1 and 4N1K to stimulate cell spread-
ing (Fig. 9, A-D, and G). Unlike inhibition of tyrosine
kinases, PKC and PI-3 kinase, the blockade of Gi-like pro-
teins with pertussis toxin did not affect unstimulated, rapid
spreading on high density VN suggesting that the G pro-
tein(s) act more proximally to IAP and upstream of PKC,
tyrosine kinases, and/or PI-3 kinase. The carbohydrate
binding B oligomer portion of pertussis toxin has been re-
ported to be responsible for some effects of the toxin (24).
We thus tested the isolated B oligomer (30-300 ng/ml) and
found that at all concentrations tested, the B oligomer of
pertussis toxin was without effect on cell spreading (Fig. 9 G).
To ensure that the failure of the whole toxin-treated cells
to spread in response to TS1 and 4N1K was due to the spe-
cific blockade of a heterotrimeric Gi protein and not a
general cytotoxic effect, toxin-treated ceils were plated on
VN and stimulated with PMA. Spreading of the cells en-
sued immediately and proceeded to the same extent noted
when C32 cells were treated with PMA alone (Fig. 9, E
and G). In contrast, cells treated with wortmannin could
not be stimulated to spread with short-term PMA treat-
ment (Fig. 9, F and G). Thus, IAP appears to use a Gi pro-
tein pathway to access the common pathway described by
others in which activation of PKC, PI-3 kinase, and ty-
rosine kinases leads to cell spreading mediated by a num-
ber of integrins in several cell types (3, 13, 20, 56).
Discussion
The data presented here provide insight on a novel mecha-
nism by which TS1 can affect integrin-mediated cellular
processes involved in normal development, homeostasis,
and pathogenesis. It is thus appropriate to view TS1 and
other TS family members which contain the sequences
that bind lAP as acute modulators of integrin function. In
fact, this modus operandi fits well with accumulating
knowledge about the expression and biological roles of
TS1. For example, TS1 has an acute function in platelet
aggregation (see below), and it influences the motility of a
number of migratory cell types, particularly during the in-
flammatory response (37, 38, 63, 64, 71, 72). Its expression
in endothelial cells is tightly regulated where it can inhibit
angiogenesis if expressed early in the process, but may be
important in endothelial cell differentiation into capillar-
ies (58, 65). TS1 expression is acutely regulated by cyto-
kines and growth factors in several cell types (6, 10, 32) as
exemplified in vivo during wound healing (47, 48) and in
development (22). This acute regulation of TS1 synthesis
and secretion is characteristic of agents which regulate un-
derlying processes of adhesion and motility.
In this report we first establish that the effect of TS1/IAP
ligation is on the spreading of cells and not on their adhe-
sion. Interestingly, the interaction of the C32 cells with VN
is a two step process at low densities of VN, with etv[35sup-
The Journal of Cell Biology, Volume 135, 1996 540
100'
8O
=
m
o
0 60
|
D. 40
20
Figure 8. Effect of inhibitors on cell spreading stimulated by
4N1K. Peptide 4N1K was present in all cases (except the DMSO
negative control marked with "*"). (Left to right) DMSO, (con-
trol vehicle); wortmannin 10 nM; calphostin C, 100 nM; chel-
erythrine 7 I~M(L) and 21 IxM(H); genistein25 txM(L) and 75 txM
(H); herbimycin 9 I~M(L) and 27 txM (H); HA1004 390 p.M.See
Materials and Methods for protocols.
plying the initial binding interaction and av133leading to
spreading. At high densities of VN, both integrins partici-
pate in cell adhesion as indicated by mAb inhibition pro-
files (Fig. 4). We could find no evidence that %135partici-
pates in spreading in these cells. In FB carcinoma cells,
%[35 can be stimulated by PMA treatment to support cell
spreading on VN (25), but this effect requires transcrip-
tional activation via NF-KB and synthesis of new proteins,
a process that takes several hours (73). The enhancement
of spreading by TS1 reagents occurs very rapidly, within
minutes, and is specific for the native RFYVVM sequence
as shown with control and mutant peptides. The activity of
the recombinant CBD and whole TS1 further indicates
that this effect is a property of the native TS1 molecule.
The stimulation of spreading via lAP is seen with murine
3T3 cells as well as the transformed human melanoma line
C32 employed in most of our experiments. Hence, this
phenomenon is not unique to a single species of cell or to
transformed cells. Spreading on other matrix proteins such
as collagens, fibrinogen, and fibronectin mediated by inte-
grins other than C~vl33is not augmented by TS1 or 4N1K in
either C32 or 3T3 ceils. Interestingly, cell spreading on LM
is stimulated by the TS reagents, an effect currently under
investigation.
The properties of the TSl-enhanced spreading on VN
are consistent with previous information about IAP func-
tion obtained with the function-blocking mAb B6H12 in
leukocyte systems where lAP was seen to be required for
signal transduction resulting in effects on the function of
~5131integrin in phagocytosis (4). In both endothelial cells
(54) and fibroblasts (66), IAP inhibition blocks a calcium
transient which may augment PKC activation. B6H12 treat-
ment of either the PMNs or the HUVECs blocks the trans-
migration of PMNs across HUVEC monolayers in vitro
(11). We have shown that the 4N1K peptide and TS1 stim-
ulate HUVEC chemotaxis on gelatin-coated filters, an
etv133-dependent process, and this chemotaxis is blocked by
B6H12 (18). Others have noted the similarities between
integrin-dependent cell migration and cell spreading (15,
55), and our results with the HUVEC chemotaxis system
indicate that the inhibitors which we have found to block
the stimulation of C32 cell spreading also block lAP-depen-
dent HUVEC chemotaxis (Tolsma, S., N. Bouck, M.B. Finn,
and W. Frazier, manuscript in preparation).
While the use of inhibitors cannot provide definitive
proof of the essential role of a given signal transduction
component in a pathway, we have used a panel of inhibi-
tors here as a starting point to begin assessing the types of
systems which TSI-IAP ligation might affect. In fact our
results agree with recent data from other groups who have
found that PKC activation may be a universal intermedi-
ate step in cell spreading (2, 25, 34, 44, 67, 69) whether
"constitutive" or "stimulated" as seen here. Accumulating
data using wortmannin also implicate PI-3 kinase as an im-
portant component of integrin-mediated cell activation
(30, 59, 74). Our data indicate that TS1 and lAP represent
a previously unknown way of modulating these constitu-
tive signaling pathways that lead to cell spreading. More
interesting from a mechanistic standpoint is the finding
that Gi-like proteins are required for the lAP-mediated
stimulation, and that pertussis toxin does not block the
constitutive spreading response as seen on high density
VN. Of all inhibitors used in signal transduction research,
the cholera and pertussis toxins can claim specificity due
to their recognition of their substrates via the enzymatic
subunits of the toxins. We have shown here that pertussis
toxin specifically blocks the TS1/IAP-mediated stimula-
tion of spreading on VN. The carbohydrate binding B oli-
gomer has no effect, and the blockade is bypassed by di-
rect PKC stimulation with phorbol ester ruling out a
general cytotoxic effect (Fig. 9). The constitutive spread-
ing on high density VN is not affected by pertussis toxin,
again ruling out a toxic effect, and, more importantly, indi-
cating that the G protein activated step is proximal to lAP
and upstream of PKC activation. Mansfield et al. (1990)
reported that the migration of leukocytes toward TS1 is
sensitive to pertussis toxin and mAb C6.7 directed to the TS1
CBD (37, 38). Our preliminary data indicate that chemo-
taxis of several cell types, including monocytes, toward
4N1K is pertussis toxin sensitive (W.A. Frazier, unpub-
lished data). This toxin is known to cause the ADP-ribosy-
lation of the et subunit of Gi isoforms 1, 2, 3, and Go, as
well as the Ga subunits of transducins 1 and 2 (43). It is
unlikely (but not impossible) that the transducins or Go
are expressed in C32 cells, thus narrowing the field of
likely candidates to the GOql.3. We are currently investigat-
ing the expression of Get subunits in C32 cells and whether
there is a direct interaction between lAP and a Giprotein.
The signaling pathway proposed here for IAP costimu-
lation of etv133-mediated cell spreading is remarkably simi-
lar to the pathway reported by Shattil and coworkers (57)
to explain the costimulation of t~iib133-mediated platelet
spreading on fibrinogen-coated surfaces. In that system,
unstimulated platelets adhere to fibrinogen but do not
spread in the presence of apyrase, an ADP scavenger, un-
less a costimulatory signal is provided by epinephrine or
Gao et al. Thrombospondin Modulates Via lntegrin-associated Protein 541
100
80
0
Q" 40
20
Figure 9. Effects of pertussis toxin and phorbol ester on C32 cell
spreading on vitronectin. C32 cells were allowed to spread on I ixg/ml
VN in the absence (A, C, and E) or presence (B, D, and F) of 50 ~M
4N1K peptide. The phorbol ester PMA alone at 10 ng/ml (C) caused ex-
tensive spreading of the cells. After overnight treatment with pertussis
toxin (see Materials and Methods), 4N1K was no longer able to stimu-
late cell spreading (D), however, a 10-min pretreatment with 10 ng/ml
PMA could cause spreading of the pertussis toxin-treated cells (E).
PMA (10 min, 10 ng/ml) was not able to reverse the block to spreading
imposed by wortmannin (10 nM) (F). (G) Quantification of effects of
these inhibitors and controls. PT-B refers to the B oligomer of pertussis
toxin present at the same concentration used for pertussis toxin in D
above.
TheJournalofCellBiology,Volume135,1996 542
thrombin binding to a heptahelical receptor. These recep-
tors are coupled to heterotrimeric Gi proteins which lead
to the activation of PKC via lipid mediators including
arachidonate, thromboxane A2 (which itself binds to an-
other heptahelical receptor), and diacylglycerol. In this
context it is noteworthy that the first reported function for
the CBD of TS1 was that of stimulation of secondary
phase or secretion dependent platelet aggregation. This
was based on inhibition by the mAb C6,7 directed against
the CBD (14). Several TS1 receptors have been identified
on platelets and several explanations for the role of TS1 in
platelet aggregation have been proposed (68). None of
these satisfactorily explain the inhibition by the anti-CBD
mAb C6.7. lAP is present on platelets in abundance and
associates with alibi33 (7). Based on the similarities be-
tween the lAP signal transduction pathway in C32 cells
presented here and the costimulatory pathway in platelets
(57), we suggest that a role of TS1 in platelets is that of an
integrin costimulator acting through IAP. Experiments to
test this hypothesis are currently in progress.
In conclusion, we have identified previously unknown
functions for both TS1 and IAP in costimulating c~vl33inte-
grin-mediated cell spreading. The involvement of PKC
and PI-3 kinase in cell spreading (2, 25, 34, 44, 55, 67, 69)
appears to be a general constitutive mechanism. TS1 bind-
ing to lAP can modulate or augment this pathway by acti-
vating a Gi type heterotrimeric G protein. This suggests
that TS1 acting through IAP may have a general role in
modulating cell motility and other matrix interactions
where integrins are the primary mediators. This model for
TS action provides a new perspective on the role of TS1 in
many cases where it is transiently synthesized and/or se-
creted at sites of tissue development or remodeling,
wound healing, inflammation, and angiogenesis.
We thank D. Cheresh for a generous supply of LM609 and P1F6 mAbs,
the Protein and Nucleic Acid Chemistry Laboratory of Washington Uni-
versity (Dr. G. Grant, Director) for synthesis, purification, and verifica-
tion of the peptides used in this study, M. Linder and L. Pike for critical
reading of the manuscript and helpful discussions.
This work was supported by grants from the National Institutes of
Health to W.A. Frazier and E.J. Brown. F.P. Lindberg is an Investigator
of the Arthritis Foundation.
Received for publication 4 March 1996 and in revised form 18 July 1996.
References
1. Asch, A.S. 1993. The role of CD36 as thrombospondin receptor. In Throm-
bospondin. J. Lahav, editor. CRC Press, Boca Raton, FL. 265-275.
2. Auer, K.L., and B.S. Jacobson. 1995. 131integrins signal lipid second mes-
sengers required during cell adhesion. MoL BioL Cell. 6:1305-1313.
3. Bargatze, R.F., and E.C. Butcher. 1993. Rapid G protein-regulated activa-
tion event involved in lymphocyte binding to high endothelial venules. J.
Exp. Med. 178:367-372.
4. Blystone, S.D., F.P. Lindberg, S.E. LaFlamme, and E.J. Brown. 1995. Inte-
grin 133cytoplasmic tail is necessary and sufficient for regulation of c~5131
phagocytosis by o%133and integrin-associated protein. Z Cell BioL 130:
745-754.
5. Bornstein, P. 1995. Diversity of function is inherent in matricellular pro-
teins: an appraisal of thrombospondin 1.J. Cell BioL 130:503-506.
6. Bornstein, P., and E.H. Sage, 1994. Thrombospondins. Methods Enzymol.
245:62-85.
7. Brown, E., L. Hooper, T. Ho, and H. Gresham. 1990. Integrin-associated
protein: a 50-kD plasma membrane antigen physically and functionally
associated with integrins. J. Cell Biol. 111:2785-2794.
8. Bruns, R.F., F.D. Miller, R.L. Merriman, J.J. Howbert, W.F. Heath, E.
Kobayashi, I. Takahashi, T. Tamaoki, and H. Nakano. 1991. Inhibition of
protein kinase C by calphostin C is light-dependent. Biochem. Biophys.
Res. Commun. 176:288-293.
9. Clark, E.A., and J.S. Brugge. 1995. Integrins and signal transduction path-
ways: the road taken. Science (Wash. DC). 268:233-239.
10. Clezardin, P. 1993. Expression of thrombospondin by cells in culture. In
Thrombospondin. J. Lahav, editor. CRC Press, Boca Raton, FL. 41-61.
11. Cooper, D., F.P. Lindberg, G.R. Gamble, E.J. Brown, and M.A. Vadas.
1995. Transendothelial migration of neutrophils involves integrin-associ-
ated protein (CD47). Proc. Natl. Acad. Sci. USA. 92:3978-3982.
12. DeFreitas, M.F., C.K. Yoshida, W.A. Frazier, D.L. Mendrick, R. Kypta,
and L.F. Reichardt. 1995. Integrin ot3131is a thrombospondin-1 receptor
on rat sympathetic neurons mediating neurite outgrowth. Neuron. 15:
333-343.
13. Devi, S., J. Laning, Y. Luo, and M.E. Dorf. 1995. Biologic activities of the
13-chemokine TCA3 on neutrophils and macrophages. J. lmmunol. 154:
5376-5383.
14. Dixit, V.M., D.M. Haverstick, K.M. O'Rourke, S.W. Hennessy, G.A.
Grant, S.A. Santoro, and W.A. Frazier. 1985. A monoclonal antibody
against human thrombospondin inhibits platelet aggregation. Proc. Natl.
Acad. Sci. USA. 82:3472-3476.
15. Felding-Habermann, B., and D.A. Cheresh. 1993. Vitronectin and its re-
ceptors. Curr. Opin. Cell Biol. 5:864-868.
16. Frazier, W.A. 1991. Thrombospondins. Curr. Opin. Cell Biol. 3:792-799.
17. Gao, A. G., and W.A. Frazier. 1994. Identification of a receptor candidate
for the carboxyl-terminal cell binding domain of thrombospondins. J.
Biol. Chem. 269:29650--29657.
18. Gao, A. G., F.P. Lindberg, M.B. Finn, S.D. Blystone, E.J. Brown, and W.A.
Frazier. 1996. Integrin-associated protein is a receptor for the C-terminal
domain of thrombospondin. J. Biol. Chem. 271:21-24.
19. Gopalakrishna, R., Z.H. Chen, and U. Gundimeda. 1992. Irreversible oxi-
dative inactivation of protein kinase C by photosensitive inhibitor
calphostin C. FEBS Lett. 314:149-154.
20. Honda, S., J.J. Campbell, D.P. Andrew, B. Engelhardt, B.A. Butcher, R.A.
Warnock, R.D. Ye, and E.C. Butcher. 1994. Ligand-induced adhesion to
activated endothelium and to vascular cell adhesion molecule-1 in lym-
phocytes transfected with the N-formyl peptide receptor. J. lmmunol.
152:4026-4035.
21. Illc, D., Y. Furuta, S. Kanazawa, N. Takeda, K. Sohue, N. Nakatsuji, S. No-
mura, J. Fujimoto, M. Okada, T. Yamamoto, et al. 1995. Reduced cell
motility and enhanced focal adhesion contact formation in cells from
FAK-deficient mice. Nature (Lond.). 377:539-544.
22. Iruela-Arispe, M.L., D.A. Liska, E.H. Sage, and P. Bornstein. 1993. Differ-
ential expression of thrombospondin 1, 2, and 3 during murine develop-
ment. Dev. Dynam. 197:40-56.
23. Jackson, S.P., S.M. Schoenwaelder, Y. Yuan, I. Rabinowitz, H.H. Salem,
and C.A. Mitchell. 1994. Adhesion receptor activation of phosphatidyl-
inositol 3-kinase. J. BioL Chem. 269:27093-27099.
24. Kindt, R.M., and A.D. Lander. 1995. Pertussis toxin specifically inhibits
growth cone guidance by a mechanism independent of direct G protein
inactivation. Neuron. 15:79-88.
25. Klemke, R.L., M. Yebra, E.M. Bayna, and D.A. Cheresh. 1994. Receptor
tyrosine kinase signaling required for integrin ctv135-directed cell motility
but not adhesion on vitronectin. J. Cell Biol. 127:859-866.
26. Kobayashi, E., H. Nakano, M. Morimoto, and T. Tamaoki. 1989. Calphos-
tin C (UCN-1028C), a novel microbial compound, is a highly potent and
specific inhibitor of protein kinase C. Biochem. Biophys. Res. Commun.
159:548-553.
27. Kosfeld, M.D., and W.A. Frazier. 1992. Identification of active peptide se-
quences in the carboxyl-terminal cell binding domain of human thrombo-
spondin-1. J. Biol. Chem. 267:16230-16236.
28. Kosfeld, M.D., and W.A. Frazier. 1993. Identification of a new cell adhe-
sion motif in two homologous peptides from the COOH-terminal cell
binding domain of human thrombospondin. J. Biol. Chem. 268:8808-
8814.
29. Kosfeld, M.D., T.V. Pavlopoulos, and W.A. Frazier. 1991. Cell attachment
activity of the carboxyl-terminal domain of human thrombospondin ex-
pressed in Escherichia coli. J. Biol. Chem. 266:24257-24259.
30. Kovacsovics, T.J., C. Bachelot, A. Toker, C.J. Vlahos, B. Duckworth, L.C.
Cantley, and J.H. Hartwig. 1995. Phosphoinositide 3-kinase inhibition
spares actin assembly in activating platelets but reverses platelet aggrega-
tion. J. BioL Chem. 270:11358-11366.
31. Kundra, V., J.A. Escobedo, A. Kazlauskas, H.K. Kim, S.G. Rhee. L.T. Wil-
liams, and B.R. Zetter. 1994. Regulation of chemotaxis by the platelet-
derived growth factor receptor-13. Nature (Lond.). 367:474-476.
32. Lahav, J. 1993. The distribution of thrombospondin in vivo. In Thrombo-
spondin. J. Lahav, editor. CRC Press, Boca Raton, FL. 33-39.
33. Lawler, J. 1993. The interaction of thrombospondin with integrins. In
Thrombospondin. J. Lahav, editor. CRC Press, Boca Raton, FL. 275-282.
34. Leavesley, D.I., G.D. Ferguson, E.A. Wayner, and D.A. Cheresh. 1992.
Requirement of the integrin 133 subunit for carcinoma cell spreading or
migration on vitronectin and fibrinogen. J. Cell Biol. 117:1101-1107.
35. Lindberg, F.P., H.D. Gresham, E. Schwarz, and E.J. Brown. 1993. Molecu-
lar cloning of integrin-associated protein: an immunoglobulin family
member with multiple membrane-spanning domains implicated in C~v13~-
dependent ligand binding. J. Cell BioL 123:485-496.
36. Lindberg, F.P., D.M. Lublin, M.J. Telen, R.A. Veile, Y.E. Miller, H. Donis-
Keller, and E.J. Brown. 1994. Rh-related antigen CD47 is the signal-
Gao et al. Thrombospondin Modulates Via Integrin-associated Protein 543
transducer integrin-associated protein. J. Biol. Chem. 269:1567-1570.
37. Mansfield, P.J., L.A. Boxer, and SJ. Suchard. 1990.Thrombospondin stim-
ulates motility of human neutrophils. J. Cell Biol. 111:3077-3086.
38. Mansfield, P.J., and S.J. Suchard. 1994.Thrombospondinpromoteschemo-
taxis and haptotaxis of human peripheral blood monocytes. J. lmmunol.
153:4219-4229.
39. Miyamoto, S., H. Teramoto, O.A. Coso, J.S. Gutkind, P.D. Burbelo, S.K.
Akiyama, and K.M. Yamada. 1995. Integrin function: molecular hierar-
chies of cytoskeletal and signaling molecules. J. Cell Biol. 131:791-805.
40. Mosher, D.F., and X. Sun. 1993. Binding of thrombospondinto cell surface
heparin sulfate proteoglycan. In Thrombospondin. J. Lahav, editor. CRC
Press, Boca Raton, FL. 259-264.
41. Murphy-Ullrich, J.E., S. Gurusiddappa, W.A. Frazier, and M. Hook. 1993.
Heparin-binding peptides from thrombospondins 1 and 2 contain focal
adhesion-labilizing activity.J. Biol. Chem. 268:26784-26789.
42. Murphy-Ullrich, J.E., and M. Hook. 1989.Thrombospondinmodulates fo-
cal adhesions in endothelial cells.J. Cell Biol. 109:1309-1319.
43. Neer, E.J. 1995. Heterotrimeric G proteins: organizers of transmembrane
signals. Cell. 80:249-257.
44. Newton, A.C. 1995. Protein kinase C: structure, function, and regulation. J.
Biol. Chem. 270:28495-28498.
45. Parkos, C.A., S.P.Colgan,T.W. Liang,A. Nusrat, A.E. Bacarra, D.K. Carnes,
and J.L. Madara. 1996. CD47 mediates post-adhesive events required for
neutrophil migration across polarized intestinal epithelia. J. Cell Biol.
132:437-450.
46. Prater, C.A., J, Plotkin, D. Jaye, and W.A. Frazier. 1991. The properdin-
like type I repeats of human thrombospondin contain a cell attachment
site. J. Cell Biol. 112:1031-1040.
47. Raugi, G.J., J.S. Mullen, D.H. Bark, T. Okada, and M.R. Mayberg. 1990.
Thrombospondin deposition in rat carotid artery injury. Am. J. Pathol.
137:179-185.
48. Raugi, G.J., J.E. Olerud, and A.M. Gown. 1987. Thrombospondin in early
human wound tissue. J. Invest. Dermatol. 89:551-554.
49. Roberts, D.D. 1993. Interactions of thrombospondin with sulfatides and
other sulfated glycoconjugates. In Thrombospondin. J. Lahav, editor.
CRC Press, Boca Raton, FL. 73-90.
50. Santoro, S.A., and W.A. Frazier. 1987. Isolation and characterization of
thrombospondin. Methods. Enzymol. 144:438-446.
51. Schaller, M.D., C.A. Borgman, B.S. Cobb, R.R. Vines, A.B. Reynolds, and
J.T. Parsons. 1992. pp125FAK,a structurally distinctive protein-tyrosine
kinase associated with focal adhesions, Proc. Natl. Acad. Sci. USA. 89:
5192-5196.
52. Schaller, M.D., and J.T. Parsons. 1994. Focal adhesion kinase and associ-
ated proteins. Curr. Opin. Cell Biol. 6:705-710.
53. Schultz-Cherry, S., H. Chen, D.F. Mosher, T.M. Misenheimer, H.C.
Krutzsch, D.D. Roberts, and J.E. Murphy-Ullrich. 1995. Regulation of
transforming growth factor-13activation by discrete sequences of throm-
bospondin 1.J. Biol. Chem. 270:7304--7310.
54. Schwartz, M.A., E.J. Brown, and B. Fazeli. 1993. A 50-kDa integrin-associ-
ated protein is required for integrin-regulated calcium entry in endothe-
lial cells.J. Biol. Chem. 268:19931-19934.
55. Schwartz, M.A., M.D. Schaller, and M.H. Ginsberg. 1995. Integrins: emerg-
ing paradigms of signal transduction. Annu. Rev. Cell Dev. Biol. 11:549-
599.
56. Shattil, S.J., M.H. Ginsberg, and J.S. Brugge. 1994. Adhesive signaling in
platelets. Curr. Opin. Cell Biol. 6:695-704.
57. Shattil, S.J., B. Haimovich, M. Cunningham, L. Lipfert, J.T. Parsons, M.H.
Ginsberg, and J.S. Brugge. 1994. Tyrosine phosphorylation of pp125FAK
in platelets requires coordinated signaling through integrin and agonist
receptors. J. BioL Chem. 269:14738-14745.
58. Sheibani, N., and W.A. Frazier. 1995. Thrombospondin-1 expression in
transformed endothelial cells restores a normal phenotype and sup-
presses their tumorigenesis. Proc. Natl. Aead. Sci. USA. 92:6788-6792.
59. Shimizu, Y., J.L. Mobley, L.D. Finkelstein, and A.S.H. Chan. 1995. A role
for phosphatidylinositol 3-kinase in the regulation of 131integrin activity
by the CD2 antigen. J. Cell Biol. 131:1867-1880.
60. Springer, T.A. 1994.Traffic signals for lymphocyte recirculation and leuko-
cyte emigration: the multistep paradigm. Cell. 76:301-314.
61. Suchard, S.J. 1993. The role of thrombospondin in inflammation. In
Thrombospondin. J. Lahav, editor. CRC Press, Boca Raton, FL. 177-198.
62. Sun, X., K. Skorstengaard, and D.F. Mosher. 1992. Disulfides modulate
RGD-inhibitable cell adhesive activity of thrombospondin. J. Cell Biol.
118:693-701.
63. Taraboletti, G., D. Roberts, L.A. Liotta, and R. Giavazzi. 1990. Platelet
thrombospondin modulates endothelial cell adhesion, motility, and
growth: a potential angiogenesis regulatory factor. J. Cell Biol. 111:765-
772.
64. Taraboletti, G., D.D. Roberts, and L.A. Liotta. 1987. Thrombospondin-
induced tumor cell migration: haptotaxis and chemotaxis are mediated
by different molecular domains. J. Cell Biol. 105:2409--2415.
65. Tolsma, S.S.,O.V. Volpert, D.J. Good, W.A. Frazier, P.J. Polverini, and N.
Bouck. 1993. Peptides derived from two separate domains of the matrix
protein thrombospondin-1 have anti-angiogenic activity.J. Cell Biol. 122:
497-511.
66. Tsao, P.W., and S.A, Mousa. 1995.Thrombospondin mediates calcium mo-
bilization in fibroblasts via its Arg-Gly-Asp and carboxyl-terminal do-
mains. J. Biol. Chem. 270:23747-23753.
67. Vuori, K., and E. Ruoslahti. 1993.Activation of protein kinase C precedes
Cts~1 integrin-mediated cell spreading on flbronectin. J. Biol. Chem. 268:
21459-21462.
68. Walz, D.A., R.S. Zafar, and Z. Zeng. 1993.The role of thrombospondin in
hemostatis. In Thrombospondin. J. Lahav, editor. CRC Press, Boca Ra-
ton, FL. 149-164.
69. Woods, A., and J.R. Couchman. 1992. Protein kinase C involvement in fo-
cal adhesion formation. J. Cell Sci. 101:277-290.
70. Yabkowitz, R., V.M. Dixit, N. Guo, D.D. Roberts, and Y. Shimizu. 1993.
Activated T-cell adhesion to thrombospondin is mediated by the a4131
(VLA-4) and ~.5~1(VLA-5) integrins. J. Immunol. 151:149-158.
71. Yabkowitz, R., P.J. Mansfield, V.M. Dixit, and S.J. Suchard. 1993.Motility
of human carcinoma cells in response to thrombospondin: relationship to
metastatic potential and thrombospondin structural domains. Cancer
Res. 53:378-387.
72. Yabkowitz, R., P.J. Mansfield, U.S. Ryan, and S.J. Suchard. 1993. Throm-
bospondin mediates migration and potentiates platelet-derived growth
factor-dependent migration of calf pulmonary artery smooth muscle
cells.J. Cell. Physiol. 157:24-32.
73. Yebra, M., E.J. Filardo, E.M. Bayna, E. Kawahara, J.C. Becker, and D.A.
Cheresh. 1995. Induction of carcinoma cell migration on vitronectin by
NF-KB-dependent gene expression. Mol. Biol. Cell. 6:841-850.
74. Zell, T., S.W. Hung III, J.L. Mobley, L.D. Finkelstein, and Y. Shimizu.
1996. CD28-mediated up-regulation of 131-integrin adhesion involves
phosphatidylinositol 3-kinase. J. Immunol. 156:883-886.
75. Zhou, M., and E.J. Brown. 1993. Leukocyte response integrin and integrin-
associated protein act as a signal transduction unit in generation of a
phagocyte respiratory burst. J. Exp. Med. 178:1165-1174.
The Journal of Cell Biology, Volume 135, 1996 544