Thrombospondin-1 Peptides Signaling by Thrombospondin-1

␤ 1 Integrin- and Proteoglycan-Mediated Stimulation of T Lymphoma Cell Adhesion and Mitogen-Activated Protein Kinase Signaling by Thrombospondin-1 and Thrombospondin-1 Peptides

Katherine E. Wilson,1 Zhuqing Li, Murat Kara, Kevin L. Gardner, and David D. Roberts2

Cell-cell and cell-matrix interactions play important regulatory roles in lymphocyte homeostasis. Thrombospondin-1 (TSP1) is a matricellular protein that differentially promotes the adhesion of resting and activated T cells. In this work, we show that adhesion ␤ of Jurkat T cells on substrates coated with TSP1 or TSP1-derived peptides is mediated by 1 integrins, CD47, and heparan sulfate proteoglycans. Interactions with TSP1 or TSP1 peptides stimulated CD3-induced Ras activation and tyrosine phosphorylation of several T cell proteins. The signals from TSP1 and its derived peptides differentially synergized with activation of the TCR to induce phosphorylation of linker for activation of T cells (LAT) and extracellular signal-regulated kinase (ERK) 1/2, c-Jun N-terminal kinase, and p38 kinases. The phosphorylation of ERK in the presence of full-length TSP1 was transient and dependent ␤ on a 1 integrin receptor. Interestingly, peptides derived from the type 1 repeats of TSP1 and a CD47-binding peptide from the carboxyl-terminal domain of TSP1 also stimulated mitogen-activated protein (MAP) kinase phosphorylation. Moreover, the TSP1 heparin-binding peptide synergized with Ab-ligated TCR to transduce signals to the nucleus, detected by activation of AP-1- and Elk-dependent transcription. This TSP1 peptide-dependent activation of AP-1 was inhibited by both heparin and the MAP/ERK kinase inhibitor PD98059, providing a functional link between adhesion molecule interaction and nuclear transactivation events via the MAP kinase pathways. These ﬁndings have implications for the role of extracellular TSP1 and TSP1 fragments in the regulation of T cell function during hemostasis, wound repair, and other inﬂammatory responses. The Journal of Immunology, 1999, 163: 3621–3628.

hrombospondins are a family of matricellular proteins expression may modulate local immune responses. Although the that have diverse effects on cell adhesion, motility, pro- tumor-suppressive activity of TSP1 has been primarily associated liferation, and survival (1–6). Thrombospondin-1 with its anti-angiogenic activity, immune modulation may also T 3 (TSP1), the first identified member of this family, is highly ex- contribute to the effects of TSP1 expression on tumor growth in pressed during wound repair and inflammatory responses (re- several animal models (reviewed in Ref. 6). viewed in Refs. 7 and 8). Expression of the THBS1 gene encoding TSP1 is a large multifunctional protein composed of three iden- TSP1 is induced by several growth factors, including platelet-de- tical subunits (7, 16, 17). The diverse effects of TSP1 on cell be- ␤ rived growth factor and TGF- 1. TSP1 is also a major component havior have been associated with several functional sequences (re- ␣ of platelet -granules and is released following platelet activation viewed in Refs. 6 and 7). The amino-terminal domain contains a at sites of injury. Elevated TSP1 levels at these sites can alter high affinity heparin-binding site and a binding site for the low functions of several cell types including endothelial cells, mono- density lipoprotein receptor-related protein that mediates internal- cytes (9, 10), macrophages (11), and NK cells (12). In addition to ization of TSP1 in some cells. The central stalk region of TSP1 its direct actions through binding to TSP1 receptors on target cells, contains anti-angiogenic sequences, sequences that bind to heparin TSP1 can alter cell behavior through activating latent TGF-␤ (13, (18), CD36 (19, 20), and fibrinogen (21), and an RGD sequence 14). These in vitro activities of TSP1 combined with the observed recognized by ␤ integrins (22). Peptide sequences within the C- inflammatory disease in thbs1-null mice (15) suggested that TSP1 3 terminal domain bind to the integrin-associated glycoprotein CD47 ␤ (23), which has been shown to modulate aV 3 integrin function (3) and T cell activation (24, 25). Laboratory of Pathology, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892 T lymphocytes can interact with TSP1 via both integrin-depen- Received for publication May 7, 1999. Accepted for publication July 16, 1999. dent and integrin-independent pathways. CD47, which is a recep- The costs of publication of this article were defrayed in part by the payment of page tor for the C-terminal cell-binding domain of TSP1 (23), provides charges. This article must therefore be hereby marked advertisement in accordance costimulatory signals to T lymphocytes, probably via an integrin- with 18 U.S.C. Section 1734 solely to indicate this fact. independent pathway (24, 25). However, it has not been estab- 1 Current address: Dr. Wilson, Molecular Medicine Unit, St. James’s University Hos- lished that TSP1 binding to CD47 can elicit such a costimulatory pital Leeds, U.K. ␤ 2 signal. 1 integrins mediate adhesion of peripheral T lymphocytes Address correspondence and reprint requests to Dr. David D. Roberts, Building 10, ϩ Room 2A33, 10 Center Drive, MSC 1500, National Institutes of Health, Bethesda, on TSP1 (26). Activation-dependent adhesion of peripheral CD4 MD 20892-1500. E-mail address: [email protected] T cells to TSP1 was inhibited by function-blocking Abs to the 3 Abbreviations used in this paper: TSP1, thrombospondin-1; CAT, chloramphenicol ␣ ␤ ␣ ␤ ␤ 4 1 and 5 1 integrins. 1 integrin interactions with other extra- acetyltransferase; ERK, extracellular signal-regulated kinase; JNK, c-Jun N-terminal kinase; LAT, linker for activation of T cells; MAP, mitogen-activated protein; MEK, cellular matrix ligands can modulate the recruitment and activation MAP/ERK kinase; RBD, Ras-binding domain. of T lymphocytes (27–30; reviewed in Ref. 31), suggesting that

Table I. TSP1 peptides and analogs

Peptide Amino Acid Sequence Activities

246 KRFKQDGGWSHWSPWSS Binds heparin, promotes cell adhesion, inhibits cell proliferation & angiogenesis (18, 45) 388 KRFKQDGGASHASPASS Inactive control for 246 7N3 FIRVVMYEGKK CD47 ligand; activates integrins (23) 604 FIRGGMYEGKK Inactive control for 7N3 Hep-I ELTGAARKGSRRLVKGPD Disrupts focal adhesions (2) Mal II SPWSSCSVTCGDGVITRIR CD36 ligand; inhibits angiogenesis (43, 56) 500 NGVQYRNC-am CD36 ligand; inhibits angiogenesis (43, 56)

␤ TSP1 binding to 1 integrins on T cells may also have functional To investigate the role of sulfated glycoconjugates in adhesion, cells consequences. were grown in the presence of chlorate to inhibit sulfation (38). Cells for Although we have previously demonstrated that TSP1 can dif- these experiments were grown in Hams-F12 medium containing 10% dialyzed FCS, 20 mM HEPES, and 2 mM glutamine for 48 h before being ferentially promote adhesion of resting and activated T lympho- transferred to the same media containing 2% dialyzed FCS with or without cytes and of naive and memory T cells (26), the effects of TSP1 on sodium chlorate for 24 h. To perform the adhesion assay, the cells were function of T cells have not been examined. To begin to define the resuspended in Hams-F12 medium containing 0.1% BSA and the relevant molecular responses of T cells to TSP1, we have examined the concentration of chlorate, and adhesion assays were performed as above. adhesive and signaling responses of T cells to intact TSP1 and Western blot analysis several defined functional sequences from TSP1. We demonstrate ␤ Bacteriological polystyrene 35-mm petri dishes (Falcon 1008) were coated in this study that several cell surface receptors, including 1 inte- with the described combinations of Abs and TSP1/peptides overnight at grins, CD47, and heparan sulfate proteoglycans, can transduce sig- 4°C. Cells were added to the plates (5 ϫ 105 cells in 0.1% BSA in RPMI) nals following interaction with TSP1 or specific TSP1 peptides. with soluble peptides or TSP1, where applicable. Cells were incubated at Binding to these receptors modulates the activation of p21 Ras and 37°C for 5 min to 4 h, the suspended cells were centrifuged for 5 min at ϫ ␮ phosphorylation of several MAP kinase pathways, and induces 600 g, and 50 l of RIPA buffer containing protease inhibitors (50 mM Tris-HCl, pH 7.4, 0.15 M NaCl, 1% (w/v) Nonidet P-40, 0.5% (w/v) so- AP-1-dependent transcription activity. dium deoxycholate, 1 mM EGTA, 1 mM Na3VO4, and protease inhibitors at 10 ␮g/ml antipain, pepstatin A, chymostatin, leupeptin, soybean trypsin Materials and Methods inhibitor, aprotinin, and 1 mM phenylmethane sulfonyl fluoride) was added Proteins and peptides to the plate and cell pellet. These two fractions were then combined and incubated on ice for 30 min. The cell lysates were centrifuged at 14,000 ϫ Human TSP1 was isolated from the supernatant of thrombin-stimulated g for 30 min at 4°C. The proteins were separated by SDS-PAGE on a human platelets by gelatin and heparin affinity, followed by gel filtration 4–15% gradient polyacrylamide gel and transferred to nitrocellulose mem- chromatography (32). Purified TSP1 was stored in aliquots at Ϫ70°C. Syn- brane by semidry electroblotting. thetic peptides derived from several functional domains of TSP1 were pre- Tyrosine-phosphorylated proteins were detected using the HRP-conju- pared as previously described (2, 18, 23) and are summarized in Table I. gated anti-phosphotyrosine Ab RC20 (Transduction Laboratories, Lexing- ton, KY). The phosphorylated forms of the MAP kinases ERK, stress- Antibodies activated protein/JNK, and p38 were identified using phospho-specific mAbs (New England Biolabs, Beverly, MA), followed by HRP-conjugated Anti-human CD3 mAb (PharMingen, San Diego, CA; clone HIT3a) was anti-rabbit Ab. Proteins were visualized by chemoluminescent detection. ␮ ␤ immobilized on plastic at 0.5 g/ml. The function-blocking 1 integrin Ab, mAb13 (provided by Dr. Ken Yamada National Institute of Dental and Activated Ras interaction assay Craniofacial Research, Bethesda, MD) (33), was used in solution at 10 ␮ ␤ An activated Ras interaction assay was used to measure Ras activation by g/ml, and the 1 integrin-activating Ab TS2/16 (34) was purified from the hybridoma (American Type Culture Collection, Manassas, VA) and used TSP1 (39). In brief, Escherichia coli transformed with the pGEX-RBD in solution at 20 ␮g/ml. Anti-CD47 mAb (clone CIKm1; ICN Pharmaceu- plasmid was induced to produce the rGST-RBD fusion protein, which con- ticals, Costa Mesa, CA) was immobilized on plastic at 0.5 ␮g/ml. Phospho- tains a Ras-binding domain comprising amino acids 1–149 of c-Raf-1 (39). specific Abs to MAP kinases were obtained from New England Biolabs. The recombinant fusion protein was affinity purified using glutathione- ␮ Rabbit anti-LAT (35) was provided by Dr. Larry Samelson, National Can- Sepharose 4B beads. Mouse anti-human CD3 mAb (0.5 g/ml, clone cer Institute (Bethesda, MD). HIT3a; PharMingen) was coated on six-well plates (Nalge Nune, Naper- ville, IL) overnight in PBS at 4°C. Approximately 1.5 ϫ 106 Jurkat cells Cell culture suspended in RPMI 1640 with 0.1% BSA were added into each well. TSP1 and its peptides were added at the indicated concentrations, and the cells o The human Jurkat T-lymphoma cell line (36) was routinely cultured in were incubated in 5% CO2 at 37 C for 15 min. The cells were collected by RPMI 1640 supplemented with 10% FCS, 2 mM L-glutamine, 100 IU/ml centrifugation and washed with PBS three times. penicillin, and 100 ␮g/ml streptomycin at 37°C in an atmosphere of 5% The cells were lysed by addition of magnesium lysis buffer (39). Cell CO2. All media and culture components were from Biofluids lysates were normalized based on protein level determined by BCA assay (Rockville, MD). (Pierce, Rockford, IL). Equal amounts of cell protein were incubated with purified GST-RBD immobilized on glutathione beads prepared according Adhesion assays to the manufacturer’s instruction. After extensive washing, bound proteins Purified TSP1 and peptides, diluted in Dulbecco’s PBS, were coated onto were eluted by boiling in SDS-PAGE sample buffer for 5 min and sepa- 96-well flat-bottom plates (Immulon 2) overnight at 4°C. After aspiration rated on 4–12% SDS-PAGE gradient gel. Western blotting analysis was of the buffer, nonspecific adherence to plastic was blocked by incubation performed as described above, and Ras activity was detected by probing with Dulbecco’s PBS containing 1% BSA for 30 min at 37°C. Cells were with an anti-Ras mAb (clone18; Transduction Laboratories) and chemolu- washed in serum-free media and resuspended at 5 ϫ 105 cells/ml in RPMI minescent detection (Pierce). containing 0.1% BSA. Aliquots of cells (100 ␮l) were added to each well Reporter assays alone or with Abs at the indicated concentrations. The plate was incubated at 37°C for 15 min, before nonadherent cells were removed by washing. Jurkat cells were cultured in RPMI 1640 medium with 10% FCS at 37°C The number of adherent cells was quantified using the previously described for 16–18 h. Cells were then transiently transfected with 5 ␮g of indicated colorimetric hexosaminidase assay (37). CAT reporter constructs DNA by electroporation (40). For the Elk reporter The Journal of Immunology 3623

FIGURE 1. Adhesion of Jurkat T cells to intact TSP1 is mediated ␤ ␮ through a 1 integrin. TSP1 (20 g/ml) was coated onto all wells of a 96-well plate, except control wells that were coated only with BSA. After removing the unbound protein, the plate was blocked with 1% BSA. Jurkat cells were washed and suspended in RPMI containing 0.1% BSA. Before adding the cells to the plate (1 ϫ 105 cells/well), the following reagents ␤ were added at the indicated concentrations: 1 integrin-activating Ab ␮ ␤ ␮ TS2/16 (20 g/ml), 1 integrin function-blocking Ab mAb13 (10 g/ml), heparin (100 ␮g/ml), and anti-CD47 (10 ␮g/ml). Adhesion was quantified after 15 min in quadruplicate wells by a colorimetric hexosaminidase assay, and is presented as a percentage of the adhesion to TSP1 measured in the absence of inhibitors (mean Ϯ SD). FIGURE 2. Jurkat cells adhere to peptides from the type 1 repeats and C-terminal domain of TSP1. A, Adhesion was examined on immobilized assay, equal amounts of a Gal4-Elk construct (a generous gift from Dr. synthetic peptides from different regions of the TSP1 molecule. All pep- Silvio Gutkind, NIDCR, National Institute of Health, Bethesda, MD) and tides in this assay were coated at 100 ␮M, and adhesion was assayed as an E1B-CAT construct containing five copies of the UAS (Gal4 binding described in Fig. 1. B, Peptides 246 (F) and 7N3 (Œ) and their respective site) upstream of the E1B minimal promoter (41) were cotransfected. E ‚ Transfected cells were diluted in RPMI 1640 containing 0.1% BSA to a control peptides 388 ( ) and 605 ( ) were coated onto plastic at the in- final concentration of 2 ϫ 106 cells/well and stimulated with the indicated dicated concentrations. Adhesion was assayed after 15 min in quadrupli- TSP1 peptides in the presence of immobilized mouse anti-human CD3 Ab cate wells and presented as mean Ϯ SD. (1 ␮g/well; clone HIT3a). For the inhibition experiments using heparin and the MEK inhibitor PD98059 (Calbiochem, La Jolla, CA), transfected cells were stimulated with peptide 246 with or without heparin (100 ␮g/ml) or of TSP1 (43). The cells also failed to attach on the peptide Hep-I after pretreatment for 30 min with the MEK inhibitor PD98059 (10 ␮M). from the heparin-binding domain (2). CAT activity in cell extracts was analyzed as previously described (42). The specificities of the two active TSP1 peptides were verified using peptides with amino acid substitutions in essential residues Results (Fig. 2B and Table I). Cell adhesion on both peptides was specific, ␤ 1 integrins mediate adhesion of Jurkat cells to TSP1 in that neither control peptide significantly promoted adhesion. As previously reported for CD4ϩ peripheral T cells (26), Jurkat T Based on time-lapse video, cells on TSP1 and peptide 246 ex- lymphoma cells attached on substrates coated with TSP1 within 5 tended and retracted pseudopodia, but were less motile compared min, but adhesion was transient and the cells began to detach from with those on fibronectin, where the cell bodies also moved (data ␤ ␤ TSP1 after 30 min. Adhesion to TSP1 was inhibited by the 1 not shown). The 1 integrin-blocking mAb13 had no effect on integrin-blocking Ab mAb13, which reduced the level of cells at- adhesion to either peptide 246 or 7N3 (data not shown), confirming ␤ ␤ tached to 28% of control (Fig. 1). Conversely, addition of the 1 that adhesion to these peptides is not 1 integrin dependent. integrin-activating Ab TS2/16 increased the level of adhesion to 163% of control levels on 50 ␮g/ml TSP1 (Fig. 1) and 251% of the Sulfated glycosaminoglycans mediate adhesion to the type 1 control level on 5 ␮g/ml. More than 90% of the input cells attached peptide 246 ␮ ␤ on 50 g/ml TSP1 in the presence of the 1 integrin-activating Ab. To examine the role of sulfated glycosaminoglycans in Jurkat cell Adhesion to TSP1 was not significantly inhibited by heparin or an adhesion to TSP1 peptide 246, cells were cultured in media con- anti-CD47 Ab (Fig. 1), indicating that proteoglycans and CD47 do taining differing concentrations of the 3Ј-phosphoadenosine 5Ј- not play a major role in adhesion of these cells to immobilized phosphosulfate synthase inhibitor chlorate. Chlorate treatment in- intact TSP1. hibited attachment of cells to immobilized peptide 246 (Fig. 3), Adhesion was also examined using synthetic peptides from sev- approaching background adhesion at 40 mM chlorate. Growth with eral regions of TSP1 coated onto plastic (Fig. 2A). Both peptide chlorate had no significant effect on cell adhesion to intact TSP1. 246, a heparin-binding peptide from the type 1 repeats of TSP1 Treatment of the Jurkat cells with heparatinase, but not with chon- (18), and the CD47-binding peptide 7N3 from the C-terminal do- droitinase ABC, also significantly inhibited adhesion to peptide main of TSP1 (23) promoted adhesion of Jurkat cells. Typically, 246 (data not shown). Therefore, both heparan sulfate proteogly- 60–70% of input cells attached on these TSP1 peptides. No ad- cans and CD47 can mediate adhesion of Jurkat cells to synthetic hesion was observed on the CD36-binding peptides Mal II, from peptides from TSP1, but adhesion to the intact protein primarily ␤ the type 1 repeats, and peptide 500, from the procollagen domain depends on a 1 integrin. 3624 T CELLS USE MULTIPLE THROMBOSPONDIN RECEPTORS

FIGURE 3. Adhesion of Jurkat T cells to TSP1 peptide 246 is mediated by a sulfated receptor. Jurkat cells, grown in the presence of the indicated concentrations of sodium chlorate to inhibit sulfation as described in Ma- terials and Methods, were assayed for adhesion on plates coated with 5 ␮M peptide 246 (F), 50 ␮g/ml TSP1 (Œ), or BSA control (E). Adhesion is presented as a percentage of that for untreated cells on TSP1 or peptide 246 (mean Ϯ SD, n ϭ 4).

TSP1 and TSP1 peptides induce rapid changes in tyrosine phosphorylation Lysates were prepared from cells incubated for 15 min in the pres- FIGURE 4. TSP1 and peptide 246 stimulate CD3-induced tyrosine ence of TSP1, the heparin-binding TSP1 peptide 246, or the CD47- phosphorylation of LAT. A, Lysates were prepared with 100 ␮l of RIPA binding TSP1 peptide 7N3. TSP1 and the peptides had similar buffer containing protease inhibitors, from 5 ϫ 10 5 cells incubated for 15 activities in solution or immobilized on plastic, so protein or pep- min with TSP1 (50 ␮g/ml) peptide 7N3 (20 ␮M), or peptide 246 (20 ␮M) tides in solution were routinely used to ensure interaction with all added in solution. Anti-CD3 Ab was immobilized onto the plate at 0.5 of the cells. The cell lysates were separated by SDS-PAGE and ␮g/ml, where indicated. Lysates were separated by SDS-PAGE and trans- transferred to nitrocellulose, and phosphorylated proteins were de- ferred to nitrocellulose. The separated proteins were incubated with the tected using the anti-phosphotyrosine Ab RC20. Phosphorylation HRP-conjugated anti-phosphotyrosine Ab RC20 and visualized by che- of several proteins was enhanced in the presence of TSP1 or the moluminescence. Migration of m.w. standards is indicated in the right TSP1 peptides, including 70- and 120-kDa species (Fig. 4A). Anti- margin. B, Cells were treated with TSP1, anti-CD3, or both. Lysates were CD3 Ab also stimulated phosphorylation of these proteins. Treat- blotted with anti-phosphotyrosine (top panel) or immunoprecipitated with anti-LAT and blotted with anti-phosphotyrosine (middle panel) or anti- ment with the peptides (Fig. 4A) or intact TSP1 (Fig. 4B)inthe LAT (lower panel). presence of anti-CD3 enhanced the CD3-dependent phosphorylation of a 36-kDa protein with similar mobility to the recently identified ZAP-70 tyrosine kinase substrate LAT (35). Although the TSP1 peptide 246 directly stimulated phosphorylation of this protein, intact TSP1 did not reproducibly enhance its phosphorylation, except in the presence of anti-CD3 (Fig. 4B). The 36-kDa phosphoprotein was identified as LAT by immuno- precipitation with a LAT Ab and blotting with anti-phosphotyrosine (Fig. 4B). TSP1 treatment alone did not stimulate LAT phosphorylation, but TSP1 markedly enhanced the phosphorylation of LAT in cells treated with anti-CD3. The increased phosphorylation signal was not due to alteration in LAT protein levels, as verified by blotting of the LAT immunoprecipitates with anti-LAT (Fig. 4B).

TSP1/peptides increase expression of activated p21 Ras Ras activation is an important early signaling event in T cell responses to several external signals. To determine whether Ras par- FIGURE 5. Intact TSP1 and TSP1 peptides synergize with TCR signal- ticipates in the responses of T cells to TSP1, we measured the level ing to activate Ras. Jurkat cells were cultured in RPMI 1640 medium with of activated p21 Ras in Jurkat cells stimulated by TSP1 using a Raf 10% FCS at 37°C for 16–18 h. Cells were then pelleted, resuspended in RPMI 1640 with 0.1% BSA to a ﬁnal concentration of 2 ϫ 106 cells/well, interaction assay (39). As shown in Fig. 5, the level of activated and stimulated with 5–45 ␮g/ml TSP1 or 20 ␮M of the TSP1 peptides 7N3 Ras in Jurkat cells detected by binding to Raf1 was increased signif- and 246 in the presence of immobilized anti-CD3. Treated and untreated icantly and in a dose-dependent manner following stimulation with control cells were lysed and incubated with GST-Raf1-RBD immobilized TSP1 in the presence of anti-CD3. As was observed for induction of on glutathione-Sepharose 4B beads, as described in Materials and Meth- LAT phosphorylation, TSP1 did not activate Ras in the absence of ods. Bound proteins were separated by 4–15% gradient SDS-PAGE gel anti-CD3 (results not shown). The two TSP1 peptides 246 and 7N3 and analyzed by Western blotting using an anti-p21Ras mAb. The Journal of Immunology 3625

FIGURE 6. Intact TSP1 and TSP1 peptides induce ERK phosphorylation. Lysates were prepared as described previously and Western blotted using a phospho-speciﬁc ERK Ab. A, Cells were incubated for 15 min with immobilized TSP1 (50 ␮g/ml), peptides 246 and 7N3 (100 ␮M) alone, or CD3 Ab (0.5 ␮g/ml). B, The indicated concentrations (␮M) of peptides 246 or the control peptide 388 were added in solution to Jurkat cells on plates with immobilized CD3 Ab. ERK phosphorylation was detected in lysates prepared after 15 min. C, Peptide 246 was used to study the time course of ERK phosphorylation. Parallel cultures were incubated on dishes coated with anti-CD3 (ϩ) or uncoated dishes (Ϫ). Peptide 246 was added to the cells in solution (20 ␮M), and lysates were prepared after 15 min, 1 h, 4 h, and 24 h. ERK phosphorylation was quantiﬁed as above.

also increased activation of Ras beyond that induced by treatment of Jurkat cells with suboptimal doses of CD3 Ab alone.

TSP1 and peptides induce ERK phosphorylation ␤ FIGURE 7. 1 integrin and sulfated proteoglycans are required for The ERK signaling pathway is a target of Ras/Raf and is important transducing signals from TSP1 and peptides into the ERK pathway. A, ␤ in mediating T cell activation. To determine whether TSP1 utilizes Control cells and Jurkat cells treated with a 1 integrin function-blocking ␮ this signaling pathway, cell lysates were analyzed by Western blot Ab (mAb13, 10 g/ml) were incubated for 15 min with the indicated com- ␮ for the presence of Y204-phosphorylated ERK1/2. ERK phosphor- binations of soluble TSP1 (50 g/ml) and immobilized anti-CD3 Ab (0.5 ␮g/ml). Lysates were prepared and analyzed by Western blotting, and the ylation was significantly stimulated in the presence of TSP1, pep- level of ERK phosphorylation was quantified by densitometry. B, Jurkat tide 246, or peptide 7N3 in the absence of TCR stimulation (Fig. cells were cultured in sulfate-free medium in the presence or absence of 40 6A). At the concentration used, anti-CD3 alone significantly in- mM chlorate. Lysates were produced from the cells after incubation for 15 creased ERK phosphorylation in some experiments, but the CD3 min with the indicated combinations of peptide 246 in solution (20 ␮M) Ab strongly and reproducibly enhanced the phosphorylation in- and immobilized anti-CD3 Ab (0.5 ␮g/ml). The lysates were analyzed for duced by TSP1 or the TSP1 peptides 246 and 7N3. The enhance- the level of phosphorylated ERK. ment of CD3-stimulated ERK phosphorylation was dose dependent and specific in that a control peptide lacking the essential Trp 7B). Thus, both integrin signaling initiated by TSP1 binding to a ␤ residues (peptide 388) was inactive (Fig. 6B). Stimulation of ERK 1 integrin and interaction of peptide 246 with proteoglycans on phosphorylation by the CD47-binding peptide 7N3 was also con- Jurkat T cells can activate the ERK pathway. firmed to be specific using the control peptide FIRGGMYEGKK Differential role of G protein signaling in peptide 246 and 7N3 (results not shown). responses The phosphorylation of ERK induced by peptide 246 was rapid, consistent with the kinetics of adhesion (Fig. 6C). In some exper- Although responses of melanoma cells to the CD47-binding TSP1 iments, increased phosphorylation could be detected within 5 min. peptide 7N3 are mediated by Gi␣ and inhibited by pertussis toxin However, the phosphorylation response was not sustained and de- (3, 44), adhesive and motility responses of the same cells to TSP1 creased over time, both in the presence (results not shown) and peptide 246 are mostly pertussis toxin resistant (45, 46). The ERK absence of anti-CD3 Ab (Fig. 6C). phosphorylation responses induced by these TSP1 peptides in Jurkat We examined the role of specific TSP1 receptors in transducing cells likewise were differentially sensitive to pertussis toxin (Fig. 8). ␤ signals into the ERK pathway. Addition of the 1 integrin-block- The response to peptide 7N3 was completely reversed by pertussis ing Ab mAb13 (Fig. 7A) inhibited phosphorylation stimulated by toxin, whereas the response to peptide 246 was only partially reduced. TSP1 alone by 44% and the synergy observed between TSP1 and anti-CD3 by 15%, but mildly stimulated both control phosphory- TSP1 and peptides induce phosphorylation of JNK and p38 lation and that stimulated by anti-CD3 alone. Similarly, when cells kinase were prepared with 40 mM chlorate, the synergism between pep- To determine whether MAP kinase signaling stimulated by TSP1 tide 246 and CD3-stimulated phosphorylation was inhibited (Fig. was specific to the ERK pathway, cell lysates were analyzed by 3626 T CELLS USE MULTIPLE THROMBOSPONDIN RECEPTORS

A heparin-binding peptide from TSP1 transduces signals to the nucleus via Elk-1- and AP-1-dependent transcriptional activation Because the TSP1 peptides 246 and 7N3 induced phosphorylation of MAP kinases, we used Elk-1 and AP-1 reporter assays to determine whether these responses to TSP1 altered downstream signaling from the MAP kinase pathways. Treatment with the TSP1 peptide 246 increased both AP-1- and Elk-1-dependent transcription of CAT in Jurkat cells plated on anti-CD3 (Fig. 10, A and B). The CD47-binding peptide 7N3 induced a slight stimulation of both AP-1 and Elk-1 activity, whereas ligation of CD47 by an Ab induced only AP-1 activity (Fig. 10, A and B). The AP-1 response stimulated by TSP1 peptide 246 was mediated by a sulfated surface receptor on the Jurkat cells, because the stimulation of AP-1 FIGURE 8. Differential sensitivity of the ERK responses stimulated by activity by this peptide was reversed in the presence of heparin TSP1 heparin-binding and CD47-binding peptides to pertussis toxin. Jurkat T cells were incubated with 0.5 ␮g/ml pertussis toxin (striped bars) or (Fig. 10C). Heparin treatment alone did not alter AP-1 activity, control medium (solid bars) and then treated with 20 ␮M of the indicated verifying the specificity of the inhibitor. The AP-1 response to the peptides or buffer control. The lysates were analyzed for the level of phos- TSP1 peptide also required the Ras/MEK/ERK pathway, because phorylated ERK. Results show densitometric analysis of a representative treatment of the Jurkat cells with the MEK inhibitor PD98059 experiment. inhibited AP-1-dependent CAT expression (Fig. 10D). Therefore, induction of AP-1 in response to the TSP1 peptide 246 is mediated by binding to cell surface proteoglycans and requires signal transduction through the Ras/MEK/ERK MAP kinase pathway. Western blotting for the presence of phosphorylated JNK (Fig. 9A) and p38 kinase (Fig. 9B) after incubation with TSP1 or peptide 246. Phosphorylation of both kinases was observed when TSP1 or peptide Discussion 246 was present along with anti-CD3 Ab. No signal was produced We have demonstrated that TSP1 and specific TSP1 peptides can with anti-CD3 alone or TSP1 alone. The 246 peptide, however, pro- interact with Jurkat T-lymphoma cells and modulate their behavior duced a weak signal in the absence of anti-CD3 Ab. In the Jurkat cell through binding to at least three receptors. Jurkat T cells rapidly lysates, the predominant form of phosphorylated JNK was the 46-kDa but transiently adhere to immobilized TSP1 and to immobilized isoform; the lower m.w. band corresponds to p38 kinase. Blotting peptides from the type 1 repeats and C-terminal globular domain with Abs to detect the total JNK and p38 kinase in the cell lysates of TSP1. Adhesion to the intact protein appears to be mediated ␤ ␤ confirmed that the differences observed with the phospho-specific Abs primarily by 1 integrins, as the addition of a 1 integrin-enhanc- reflect changes in levels of p38 phosphorylation. ing Ab increases the level of adhesion, and adhesion could be ␤ inhibited by a 1 integrin-blocking Ab. The cell surface receptor CD47 mediates interactions with a TSP1 peptide from the C-terminal domain, and sulfated proteoglycans are receptors for a heparin-binding peptide from the type 1 repeats of TSP1. Binding soluble TSP1 or TSP1 peptides to each of these three receptors induces several signaling responses directly or by costimulating signaling from the TCR and results in altered transcriptional activity. Synergism of intact TSP1 with TCR signaling occurs at a point upstream of Ras and LAT, since TSP1 only activated these in the presence of anti-CD3. However, direct activation of MAP kinase phosphorylation by TSP1 and its peptides suggests that a Ras- and LAT-independent pathway may allow TSP1 and its peptides to directly stimulate MAP kinase signaling. CD47 is present at a high density on peripheral T lymphocytes and can serve as a costimulator of T cell activation (24, 25). Li- gation of CD47 by some Abs enhances the proliferation of T cells by costimulation of the CD3/TCR pathway (24). Because TSP1 is a known CD47 ligand, it was proposed to mediate costimulation through binding to this receptor (24). Although our data demon- strating that a CD47-binding peptide from TSP1 activates the MAP kinase pathway support this hypothesis, we also identified FIGURE 9. TSP1 and TSP1 peptide 246 synergize with anti-CD3 to two additional TSP1 receptors on T cells through which TSP1 or induce phosphorylation of JNK and p38 kinase. Jurkat T cells were incu- TSP1 fragments may activate or costimulate the same signaling bated for 15 min with the indicated combinations of immobilized TSP1 (50 pathways. ␤ integrins play a major role in adhesion of peripheral ␮g/ml), peptide 246 (100 ␮M), and CD3 Ab (0.5 ␮g/ml), and lysates were 1 T lymphocytes (26) and Jurkat T lymphoma cells on immobilized prepared in RIPA buffer containing protease inhibitors. A, The lysates were analyzed by Western blotting using anti-JNK (top lanes) or a phospho- TSP1 and mediate activation of the ERK pathway by intact soluble specific JNK Ab (lower lanes). B, The lysates were analyzed by Western TSP1. A heparin-binding peptide from TSP1 induces phosphory- blotting using a phospho-specific p38 kinase Ab (top lanes) or anti-p38 lation of three MAP kinase pathways and increases Elk1- and AP- (lower lanes). The positive control (ϩ) sample on this blot was a lysate 1-dependent transcriptional activities. Binding to a sulfated cell from anisomycin-treated C-6 glioma cells. surface receptor is required for the activation of ERK signaling, The Journal of Immunology 3627

FIGURE 10. Induction of Elk1 and AP-1 activity in Jurkat cells by TSP1 peptides. A, Jurkat cells were cotransfected with 5 ␮g of a Gal4-Elk expression plasmid and an E1B-CAT construct by electroporation, as described in Materials and Methods. Transiently transfected cells were stimulated with either peptide 246 (20 ␮M) or 7N3 (20 ␮M) in the presence of precoated anti-human CD3 mAb (1 ␮g/well). Anti-CD47 Ab was also tested by coimmobilizing with the anti-CD3. CAT activity was measured 4 h after stimulation, as described in Materials and Methods. The relative CAT activity is presented as mean Ϯ SD of duplicate samples. B, Jurkat cells were transfected with 5 ␮g of an AP1-CAT construct. Transfected cells were then stimulated with peptides 246 (20 ␮M) or 7N3 (20 ␮M) in the presence of precoated anti-human CD3 mAb (1 ␮g/well). CAT activity was measured after 4 h. C, AP-1 activity induced by TSP1 peptide 246 is inhibited by heparin. Jurkat cells were transiently transfected with AP1-CAT construct. Cells were then stimulated with peptide 246 (20 ␮M) in the presence of precoated anti-human CD3 mAb (1 ␮g/well) either with or without 100 ␮g/ml of heparin. The relative CAT activity is presented as mean Ϯ SD for duplicates. D, Stimulation of AP-1-dependent CAT activity was measured in untreated Jurkat cells or cells pretreated with 10 ␮M of the MEK inhibitor PD98059. and the MEK-ERK pathway is required for induction of AP-1- TCR signaling may not be the only pathway that is modulated dependent transcription by this peptide. Binding of TSP1 peptides by TSP1 binding to T cells. Interaction of TSP1 peptides with to CD47 also directly activates the ERK pathway through a per- heparan sulfate proteoglycans or CD47 can also stimulate signal- tussis toxin-sensitive pathway, but this signal is less efficient than ing through integrin receptors in melanoma cells, endothelial cells, that induced by the heparin-binding peptide 246 for costimulating and platelets (3, 46, 51). Because TSP1 and the TSP1 peptides also ␤ AP-1 activity. Thus, 1 integrins and proteoglycans are also im- activated MAP kinase signaling in the absence of CD3 stimulation, portant signaling receptors for TSP1 in T cells. Further studies will signals arising from TSP1 binding in the absence of TCR signaling be required to understand how T cells integrate these three signal- may significantly modulate T cell responses to other stimuli. Sig- ing pathways activated by TSP1 and the physiological significance naling through the MAP kinase pathway in T cells can induce IL-2 of these signals to lymphocyte functions. expression, anergy, activation, proliferation, or apoptosis (52, 53). We have identified cell surface proteoglycans as a potential re- TSP1 is known to modulate proliferation of other cell types (4, 5, ceptor that modulates TCR/CD3-dependent AP-1 activity through 45, 54) and to induce apoptosis of endothelial cells (55). a signal transduction pathway that requires the MEK/ERK1/2 cas- Physiologically, T cells could encounter TSP1 or fragments of cade, which may be stimulated through Ras activation of Raf. TSP1 either in solution or immobilized in the extracellular matrix. TSP1 is a high affinity heparin-binding protein and contains sev- If TSP1 was bound to the surface of an APC, its engagement of ␤ eral sites implicated in heparin binding (reviewed in Ref. 6). Most 1 integrins, CD47, or proteoglycans on T cells could modulate TCR studies of signal transduction through cell surface proteoglycans have focused on the transmembrane syndecans (47), which are responses, as modeled in this study, using immobilized anti-CD3. At known to bind to TSP1 (48, 49). Clustering of syndecans following present, no data exist to demonstrate TSP1 on APCs, so a more plau- ligand binding activates protein kinase C pathways (50). Interac- sible hypothesis for TSP1 function is as a soluble modulator released tions between proteoglycans and Ras or MAP kinase signaling at sites of inflammation or bound to the extracellular matrix at these have not been observed previously, nor has their ligation been sites. Intact TSP1 and fragments of TSP1 produced by local proteol- shown to alter Elk- or AP-1-dependent transcription. Based on the ysis may transduce distinct signals to T cells by interacting with one present data, signals arising from binding to T cell proteoglycans or more of the TSP1 receptors identified in this work. Based on our are a new pathway for modulating TCR signaling. It will be im- evidence that TSP1 induces signals that can alter gene expression in portant to identify the T cell proteoglycans that mediate these sig- Jurkat T cells, T lymphocytes must be considered as one of the targets nals and the pathways through which the TSP1 peptides induce of TSP1 when it is expressed or released during hemostasis, inflam- phosphorylation of ERK, JNK, and p38 kinase. matory disease, wound repair, and in cancer. 3628 T CELLS USE MULTIPLE THROMBOSPONDIN RECEPTORS

Acknowledgments 28. Sturmhofel, K., C. Brando, F. Martinon, E. M. Shevach, and J. E. Coligan. 1995. Antigen-independent, integrin-mediated T cell activation. J. Immunol. 154:2104. We thank Drs. Ken Yamada, Henry Krutzsch, Silvio Gutkind, Larry 29. Woodside, D. G., T. K. Teague, and B. W. McIntyre. 1996. Specific inhibition of ␤ ␤ T lymphocyte coactivation by triggering integrin 1 reveals convergence of 1, Samelson, and Stephen Taylor for providing reagents. ␤ ␤ 2, and 7 signaling pathways. J. Immunol. 157:700. ␣ ␤ 30. Udagawa, T., D. G. Woodside, and B. W. McIntyre. 1996. 4 1 (CD49d/CD29) References integrin costimulation of human T cells enhances transcription factor and cyto- kine induction in the absence of altered sensitivity to anti-CD3 stimulation. J. Im- 1. Roberts, D. D., J. A. Sherwood, and V. Ginsburg. 1987. Platelet thrombospondin munol. 157:1965. mediates attachment and spreading of human melanoma cells. J. Cell Biol. 104: 31. Rosales, C., and R. L. Juliano. 1995. Signal transduction by cell adhesion recep- 131. tors in leukocytes. J. Leukocyte Biol. 57:189. 2. Murphy-Ullrich, J. E., S. Gurusiddappa, W. A. Frazier, and M. Hoö¨k. 1993. 32. Roberts, D. D., J. Cashel, and N. Guo. 1994. Purification of thrombospondin from Heparin-binding peptides from thrombospondins 1 and 2 contain focal adhesion- human platelets. J. Tissue Culture Methods 16:217. labilizing activity. J. Biol. Chem. 268:26784. 33. Akiyama, S. K., S. S. Yamada, W. T. Chen, and K. M. Yamada. 1989. Analysis 3. Gao, A. G., F. P. Lindberg, J. M. Dimitry, E. J. Brown, and W. A. Frazier. 1996. of fibronectin receptor function with monoclonal antibodies: roles in cell adhe- ␣ ␤ Thrombospondin modulates v 3 function through integrin-associated protein. sion, migration, matrix assembly, and cytoskeletal organization. J. Cell Biol. J. Cell Biol. 135:533. 109:863. 4. Phan, S. H., R. G. Dillon, B. M. McGarry, and V. M. Dixit. 1989. Stimulation of 34. Hemler, M. E., F. Sanchez-Madrid, T. J. Flotte, A. M. Krensky, S. J. Burakoff, fibroblast proliferation by thrombospondin. Biochem. Biophys. Res. Commun. A. K. Bhan, T. A. Springer, and J. L. Strominger. 1984. Glycoproteins of 210,000 163:56. and 130,000 m.w. on activated T cells: cell distribution and antigenic relation to 5. Majack, R. A., L. V. Goodman, and V. M. Dixit. 1988. Cell surface throm- components on resting cells and T cell lines. J. Immunol. 132:3011. bospondin is functionally essential for vascular smooth muscle cell proliferation. 35. Zhang, W., J. Sloan-Lancaster, J. Kitchen, R. P. Trible, and L. E. Samelson. 1998. J. Cell Biol. 106:415. LAT: the ZAP-70 tyrosine kinase substrate that links T cell receptor to cellular 6. Roberts, D. D. 1996. Regulation of tumor growth and metastasis by throm- activation. Cell 92:83. bospondin-1. FASEB J. 10:1183. 36. Gillis, S., and J. Watson. 1980. Biochemical and biological characterization of 7. Bornstein, P. 1992. Thrombospondins: structure and regulation of expression. lymphocyte regulatory molecules. V. Identification of an interleukin 2-producing FASEB J. 6:3290. human leukemia T cell line. J. Exp. Med. 152:1709. 8. Mosher, D. F. 1990. Physiology of thrombospondin. Annu. Rev. Med. 41:85. 37. Landgren, U. 1984. Measurement of cell numbers by means of the endogenous 9. Schuepp, B. J., and T. W. Jungi. 1991. Thrombospondin-exposed human mono- enzyme hexosaminidase: applications to detection of lymphokines and cell sur- cytes display augmented luminol-enhanced chemoluminescence upon receptor face antigens. J. Immunol. Methods 67:379. triggering. Biochem. Biophys. Res. Commun. 177:1087. 38. Guo, N. H., H. C. Krutzsch, T. Vogel, and D. D. Roberts. 1992. Interactions of 10. Mansfield, P. J., and S. J. Suchard. 1994. Thrombospondin promotes chemotaxis a laminin-binding peptide from a 33-kDa protein related to the 67-kDa laminin and haptotaxis of human peripheral blood monocytes. J. Immunol. 153:4219. receptor with laminin and melanoma cells are heparin-dependent. J. Biol. Chem. 11. Savill, J., N. Hogg, Y. Ren, and C. Haslett. 1992. Thrombospondin cooperates 267:17743. with CD36 and the vitronectin receptor in macrophage recognition of neutrophils 39. Taylor, S. J., and D. Shalloway. 1996. Cell cycle-dependent activation of Ras. undergoing apoptosis. J. Clin. Invest. 90:1513. Curr. Biol. 6:1621. 12. Pierson, B. A., K. Gupta, W.-S. Hu, and J. S. Miller. 1996. Human natural killer 40. Butscher, W. G., C. Powers, M. Olive, C. Vinson, and K. Gardner. 1998. Coor- cell expansion is regulated by thrombospondin-mediated activation of transform- dinate transactivation of the interleukin-2 CD28 response element by c-Rel and ing growth factor ␤-1 and independent accessory cell-derived contact and soluble ATF-1/CREB2. J. Biol. Chem. 273:552. factors. Blood 87:180. 41. Duncan, R., I. Collins, T. Tomonaga, T. Zhang, and D. Levens. 1996. A unique 13. Schultz-Cherry, S., and J. E. Murphy-Ullrich. 1993. Thrombospondin causes ac- transactivation sequence motif is found in the carboxyl-terminal domain of the tivation of latent transforming growth factor-␤ secreted by endothelial cells by a single-strand-binding protein FBP. Mol. Cell. Biol. 16:2274. novel mechanism. J. Cell Biol. 122:923. 42. Gorman, C. 1985. DNA Cloning. D. M. Glover, ed. IRL Press, Oxford, p. 143. 14. Crawford, S. E., V. Stellmach, J. E. Murphy-Ullrich, S. M. F. Ribeiro, J. Lawler, 43. Dawson, D. W., S. F. Pearce, R. Zhong, R. L. Silverstein, W. A. Frazier, and R. O. Hynes, G. P. Boivin, and N. Bouck. 1998. Thrombospondin-1 is a major N. P. Bouck. 1997. CD36 mediates the in vitro inhibitory effects of throm- activator of TGF-␤1 in vivo. Cell 93:1159. bospondin-1 on endothelial cells. J. Cell Biol. 138:707. 15. Lawler, J., M. Sunday, V. Thibert, M. Duquette, E. L. George, H. Rayburn, and 44. Frazier, W. A., A.-G. Gao, J. Dimitry, J. Chung, E. J. Brown, F. P. Lindberg, and R. O. Hynes. 1998. Thrombospondin-1 is required for normal murine pulmonary M. E. Linder. 1999. The thrombospondin receptor integrin-associated protein homeostasis and its absence causes pneumonia. J. Clin. Invest. 101:982. (CD47) functionally couples to heterotrimeric Gi. J. Biol. Chem. 274:8554. 16. Bornstein, P. 1995. Diversity of function is inherent in matricellular proteins: an 45. Guo, N., V. S. Zabrenetzky, L. Chandrasekaran, J. M. Sipes, J. Lawler, appraisal of thrombospondin 1. J. Cell Biol. 130:503. H. C. Krutzsch, and D. D. Roberts. 1998. Differential roles of protein kinase C 17. Lahav, J. 1993. Thrombospondin. CRC Press, Boca Raton. and pertussis toxin-sensitive G-binding proteins in modulation of melanoma cell 18. Guo, N. H., H. C. Krutzsch, E. Ne`gre, T. Vogel, D. A. Blake, and D. D. Roberts. proliferation and motility by thrombospondin-1. Cancer Res. 58:3154. 1992. Heparin- and sulfatide-binding peptides from the type I repeats of human 46. Sipes, J. M., H. C. Krutzsch, J. Lawler, and D. D. Roberts. 1999. Cooperation ␣ ␤ thrombospondin promote melanoma cell adhesion. Proc. Natl. Acad. Sci. USA between thrombospondin-1 type 1 repeat peptides and integrin v 3 ligands to 89:3040. promote melanoma cell spreading and focal adhesion formation. J. Biol. Chem. 19. Asch, A. S., S. Silbiger, E. Heimer, and R. L. Nachman. 1992. Thrombospondin In press. sequence motif (CSVTCG) is responsible for CD36 binding. Biochem. Biophys. 47. Couchman, J. R., and A. Woods. 1996. Syndecans, signaling, and cell adhesion. Res. Commun. 182:1208. J. Cell. Biochem. 61:578. 20. Li, W. X., R. J. Howard, and L. L. Leung. 1993. Identification of SVTCG in 48. Rapraeger, A. 1989. Transforming growth factor (type ␤) promotes the addition thrombospondin as the conformation-dependent, high affinity binding site for its of chondroitin sulfate chains to the cell surface proteoglycan (syndecan) of mouse receptor, CD36. J. Biol. Chem. 268:16179. mammary epithelia. J. Cell Biol. 109:2509. 21. Panetti, T. S., B. J. Kudryk, and D. F. Mosher. 1999. Interaction of recombinant 49. Lebakken, C. S., and A. C. Rapraeger. 1996. Syndecan-1 mediates cell spreading procollagen and properdin modules of thrombospondin-1 with heparin and fi- in transfected human lymphoblastoid (Raji) cells. J. Cell Biol. 132:1209. brinogen/fibrin. J. Biol. Chem. 274:430. 50. Oh, E. S., A. Woods, and J. R. Couchman. 1997. Syndecan-4 proteoglycan reg- 22. Lawler, J., R. Weinstein, and R. O. Hynes. 1988. Cell attachment to throm- ulates the distribution and activity of protein kinase C. J. Biol. Chem. 272:8133. bospondin: the role of ARG-GLY-ASP, calcium, and integrin receptors. J. Cell 51. Chung, J., A. G. Gao, and W. A. Frazier. 1997. Thrombospondin acts via inte- ␣ ␤ Biol. 107:2351. grin-associated protein to activate the platelet integrin IIb 3. J. Biol. Chem. 23. Gao, A.-G., F. P. Lindberg, M. B. Finn, S. D. Blystone, E. J. Brown, and 272:14740. W. A. Frazier. 1996. Integrin-associated protein is a receptor for the C-terminal 52. Cantrell, D. 1996. T cell antigen receptor signal transduction pathways. Annu. domain of thrombospondin. J. Biol. Chem. 271:21. Rev. Immunol. 14:259. 24. Reinhold, M. I., F. P. Lindberg, G. J. Kersh, P. M. Allen, and E. J. Brown. 1997. 53. Van den Brink, M. R., R. Kapeller, J. C. Pratt, J. H. Chang, and S. J. Burakoff. Costimulation of T cell activation by integrin-associated protein (CD47) is an 1999. The extracellular signal-regulated kinase pathway is required for activa- adhesion-dependent, CD28-independent signaling pathway. J. Exp. Med. 185:1. tion-induced cell death of T cells. J. Biol. Chem. 274:11178. 25. Ticchioni, M., M. Deckert, F. Mary, G. Bernard, E. J. Brown, and A. Bernard. 54. Taraboletti, G., D. Roberts, L. A. Liotta, and R. Giavazzi. 1990. Platelet throm- 1997. Integrin-associated protein (CD47) is a comitogenic molecule on CD3- bospondin modulates endothelial cell adhesion, motility, and growth: a potential activated human T cells. J. Immunol. 158:677. angiogenesis regulatory factor. J. Cell Biol. 111:765. 26. Yabkowitz, R., V. M. Dixit, N. Guo, D. D. Roberts, and Y. Shimizu. 1993. 55. Guo, N., H. C. Krutzsch, J. K. Inman, and D. D. Roberts. 1997. Thrombospondin ␣ ␤ Activated T-cell adhesion to thrombospondin is mediated by the 4 1 (VLA-4) 1 and type I repeat peptides of thrombospondin 1 specifically induce apoptosis of ␣ ␤ and 5 1 (VLA-5) integrins. J. Immunol. 151:149. endothelial cells. Cancer Res. 57:1735. 27. Ybarrondo, B., A. M. O’Rourke, J. B. McCarthy, and M. F. Mescher. 1997. 56. Tolsma, S. S., O. V. Volpert, D. J. Good, W. A. Frazier, P. J. Polverini, and Cytotoxic T-lymphocyte interaction with fibronectin and vitronectin: activated N. Bouck. 1993. Peptides derived from two separate domains of the matrix pro- adhesion and cosignaling. Immunology 91:186. tein thrombospondin-1 have anti-angiogenic activity. J. Cell Biol. 122:497.