Thrombospondin Reduce Inflammation Interactions Between

Interactions between CD47 and Thrombospondin Reduce Inflammation Laurence Lamy, Arnaud Foussat, Eric J. Brown, Paul Bornstein, Michel Ticchioni and Alain Bernard This information is current as of September 27, 2021. J Immunol 2007; 178:5930-5939; ; doi: 10.4049/jimmunol.178.9.5930 http://www.jimmunol.org/content/178/9/5930 Downloaded from

References This article cites 60 articles, 36 of which you can access for free at: http://www.jimmunol.org/content/178/9/5930.full#ref-list-1

Why The JI? Submit online. http://www.jimmunol.org/

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on September 27, 2021

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2007 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Interactions between CD47 and Thrombospondin Reduce Inﬂammation1

Laurence Lamy,* Arnaud Foussat,* Eric J. Brown,‡ Paul Bornstein,§ Michel Ticchioni,*† and Alain Bernard2*

CD47 on the surface of T cells was shown in vitro to mediate either T cell activation or, in the presence of high amounts of thrombospondin (TSP), T cell apoptosis. We report here that CD47-deficient mice, as well as TSP-1 or TSP-2-deficient mice, sustain oxazolone-induced inflammation for more than four days, whereas wild-type mice reduce the inflammation within 48 h. We observe that prolonged inflammation in CD47-, TSP-1-, or TSP-2-deficient mice is accompanied by a local deficiency of T cell apoptosis. Finally, we show that upon activation normal T cells increase the expression of the proapoptotic Bcl-2 family member BNIP3 (Bcl-2/adenovirus E1B 19-kDa interacting protein) and undergo CD47-mediated apoptosis. This finding is consistent with our previous demonstration of a physical interaction between BNIP3 and CD47 that inhibits BNIP3 degradation by the proteasome, Downloaded from sensitizing T cells to CD47-induced apoptosis. Overall, these results reveal an important role in vivo for this new CD47/BNIP3 pathway in limiting inflammation by controlling the number of activated T cells. The Journal of Immunology, 2007, 178: 5930–5939.

ymphocyte programmed cell death plays a crucial role in caspase-independent cell death triggered only 1 h after cell stim- controlling immune responses at several steps, notably in ulation in vitro (5, 7). These functional outcomes are actually re- L terminating peripheral T cell responses and achieving a lated either to the cell activation state (because T cells must have http://www.jimmunol.org/ return to homeostasis. Failure to eliminate inflammatory T cells received a prior TCR activation signal) or to the nature of the results in harmful reactions, leading to tissue injury and disease. ligand for CD47 (5, 7). Indeed, multiple nonoverlapping mechanisms act to inactivate or Indeed, two families of proteins have been described to be the to kill these activated cells. Although lymphocyte cell death can be natural ligands for CD47: two members of the signal regulatory mediated by cell surface signals through receptors from the TNFR protein (SIRP)3 receptor family, SIRP␣ and SIRP␥, and two superfamily (1), it can also be mediated by surface receptors that thrombospondin (TSP) family members. SIRP␣ is expressed on are not specialized in inducing cell death but are able to trigger the surfaces of macrophages and endothelial and dendritic cells other cell functions depending on the cell type, stimulus, and in- and inhibits tyrosine kinase-coupled signaling pathways via the teracting ligand(s) (2–6). phosphorylation of ITIM motifs in its cytoplasmic domain (15). by guest on September 27, 2021 CD47 is one of these equivocal initiators of lymphocyte death in SIRP␥, expressed mostly on peripheral blood leukocytes but also both T cells and B chronic lymphocytic leukemia cells (5, 7). In in many human tissues including the brain, lung, placenta, and addition to apoptosis, CD47 has been shown to trigger a wide liver, lacks the cytoplasmic ITIMs and is reported to mediate cell- variety of cellular functions in T cells, including activation, pro- cell adhesion and T cell costimulation (16, 17). However, neither liferation, arrest, and spreading on an inflamed vascular endothe- SIRP␣ nor SIRP␥ has been implicated in CD47-induced apoptosis lium (8–11). Yet, under certain conditions it can also induce an- in contrast to the TSP family members, which have been shown to ergy or generate regulatory T cells from CD4ϩCD25Ϫ T cells (12, bind CD47 specifically via their COOH-terminal cell-binding do- 13). Its role in T cell apoptosis is complex; on the one hand it has main. The latter interactions subsequently induce CD47-dependent been shown to interact with Fas and to sensitize the cells to Fas- apoptosis in several cell lineages (7, 18–22). induced apoptosis by enhancing Fas clustering (14). On the other TSPs are a family of five extracellular proteins composed of hand, CD47 can trigger direct death signals that lead to a rapid and several structural domains known to bind extracellular matrix components and cell surface receptors (23). As a consequence of this modular structure, TSPs display diverse biological functions, *Institut National de la Santeét de la Recherche Me´dicale Unit 576 and †Laboratoire d’Immunologie, Centre Hospitalier de l’Universite´ de Nice, University of Nice- including regulation of migration, adhesion, proliferation, differ- Sophia Antipolis, Nice, France; ‡Program in Host-Pathogen Interactions, University entiation, and apoptosis (23). Although TSP1 and TSP2 are widely of California, San Francisco, CA 94158; and §Departments of Biochemistry and Med- icine, University of Washington, Seattle, WA 98195 expressed, TSP3, TSP4, and TSP5 show a more limited distribu- tion and are unlikely to play a role in the events described here (24, Received for publication December 28, 2005. Accepted for publication February 1, 2007. 25). It should be noted that the TSP1 concentration is very low in Ͻ The costs of publication of this article were defrayed in part by the payment of page plasma ( 200 ng/ml) but increases locally within seconds when charges. This article must therefore be hereby marked advertisement in accordance released by activated platelets and can reach concentrations of with 18 U.S.C. Section 1734 solely to indicate this fact. 10–20 mg/ml at these sites of platelet activation (26). It is also 1 This work was supported by grants from the Institut National de la Santeétdela Recherche Me´dicale, the Association pour la Recherche sur le Cancer, the Fondation pour la Recherche Me´dicale and the Cance´ropoˆle Provence-Alpes-Coˆte d’Azur. The Ϫ/Ϫ Ϫ/Ϫ generation of TSP-1 and TSP-2 mice was supported by National Institutes of 3 Abbreviations used in this paper: SIRP, signal-regulatory protein; 7-AAD, 7-ami- Health Grant AR 45418 (to P.B.). noactinomycin D; BNIP3, Bcl-2/adenovirus E1B 19-kDa interacting protein 3; DTH, 2 Address correspondence and reprint requests to Dr. Alain Bernard, Institut National delayed type hypersensitivity; LN, lymph node; MFI, mean fluorescence intensity; de la Santeét de la Recherche Me´dicale unit 576, University of Nice-Sophia Anti- TSP, thrombospondin; WT, wild type. polis, Hoˆpital de l’Archet I, 151 rue Saint Antoine de Ginestie`re, Nice, France. E-mail address: [email protected] Copyright © 2007 by The American Association of Immunologists, Inc. 0022-1767/07/$2.00 www.jimmunol.org The Journal of Immunology 5931

FIGURE 1. Prolonged contact hypersensitivity elicited in CD47-deficient mice. A, Several groups of six mice, either CD47Ϫ/Ϫ or WT mice, were treated with a sensitizing dose of oxazolone (oxa) (2%) on the shaved abdomen followed by a challenging dose of oxazolone (0.4%) on the right ear. The left ear was treated with vehicle alone. Ear thickness was measured (displayed as the means Ϯ SEM from six animals) in both types of mice at different time intervals after challenge (representative experiment, n ϭ 3). B, Ear tissue sections were performed in CD47Ϫ/Ϫ mice and WT Downloaded from mice at 24 and 48 h after challenge. Ear sections were stained with H&E. Scale bar represents 150 ␮m. C, Thirty hours after oxazolone challenge, cells were collected from the ears of CD47Ϫ/Ϫ or WT mice and the http://www.jimmunol.org/ proportion of dead T cells was determined using an annexin V assay kit. Mouse T cells were identified by FITC-labeled CD3 mAb. Cell debris was electronically gated out based on forward light scatter (representative experiment, n ϭ 3). D, Twenty-four hours after oxazolone challenge or treatment with the vehicle alone, cells were collected from the draining LNs by guest on September 27, 2021 of CD47Ϫ/Ϫ or WT mice and the proportion of dead T cells was determined using an annexin V assay kit. Naive and memory T cells were identified by PE-labeled CD3 mAb and FITC-labeled CD62L. Cell debris was electronically gated out based on forward light scatter (representative experiment, n ϭ 3).

overexpressed by fibroblasts under mechanical stress and by en- required for its mitochondrial localization and its homodimeriza- dothelial cells following a disturbance in blood flow (18–20). tion (22, 30). BNIP3 is overexpressed during hypoxia and in cy- TSP2 has also been shown to be up-regulated in the skin of mice totoxic effector T cells and kills cells via the opening of the mi- subjected to cutaneous inflammation (27). All of these phenomena tochondrial transition pore, which leads to profound mitochondrial occur in damaged tissues where inflammation has developed. dysfunction without caspase activation (31–34). We have shown in a previous article that the CD47 apoptotic In this study, by using CD47Ϫ/Ϫ mice and an in vivo model of pathway in T cells is peculiar (22). In a resting state, the Bcl-2 T cell-determined skin inflammation (35, 36) we show that CD47 homology (BH3) domain-only protein BNIP3 (Bcl-2/adenovirus plays a crucial role in limiting the course of an inflammatory im- E1B 19-kDa interacting protein 3) is bound to CD47. Upon a pro- mune response by eliminating activated lymphocytes. Our study apoptotic signal, BNIP3 translocates to mitochondria and triggers also shows that TSP-1Ϫ/Ϫ and TSP-2Ϫ/Ϫ mice display a similar mitochondrial death. BNIP3 regulates programmed cell death from deficient phenotype to that displayed by CD47Ϫ/Ϫ mice in con- mitochondrial sites by interacting selectively with Bcl-2 and trolling cutaneous inflammation, implying a role for these two li- Bcl-XL (28, 29). BNIP3 has a COOH-terminal transmembrane gands of CD47 in T cell clearance. We demonstrate that the ex- domain that is required for its association with the multiple mem- pression of BNIP3, which mediates CD47-induced cell death, is brane spanning domain of CD47 at the plasma membrane and also increased following CD3 stimulation, providing sensitization of T 5932 CD47 DOWN-REGULATES INFLAMMATION

FIGURE 2. Thrombus generation and up-regulation of TSP-1 expression in cutaneous inflammation. Thrombosis and inflammation were induced in the ear tissues of CD47-deficient (CD47Ϫ/Ϫ) and WT mice using oxazolone (oxa). A, Ear tissue sections from CD47Ϫ/Ϫ and WT mice were made 24 h after oxazolone challenge. Ear sections were stained with H&E. Arrows indicate clotted vessels. B, Fro- zen ear sections of WT mice were made 30 h after challenge and stained with anti-TSP1 or CD31 mAb. Immuno- staining for TSP-1 and CD31 was performed on serial cryosections. Arrows Downloaded from and arrowheads show the same vessel or interstitial area, respectively. Coun- terstaining with hematoxylin was performed (representative experiment, n ϭ 3). Scale bars represent 150 ␮m (A) and 75 ␮m(B). http://www.jimmunol.org/

cells to CD47-induced apoptosis. Finally, we discuss a biphasic Cell preparation model for controlling T cell-dependent inﬂammation via CD47 Murine blood was collected from the retro-orbital venous plexus in citrate, based on a critical threshold of TSP ligands.

and PBMCs were isolated by Ficoll-Isopaque density gradient centrifuga- by guest on September 27, 2021 tion (Nycomed/ALTANA Pharma). Splenocytes were harvested by gently Materials and Methods forcing them through a 70-␮m cell strainer (Falcon, catalog no. 2350; BD Biosciences) into 3.0 ml of HBSS containing 2% FBS and then depleted of Reagents and antibodies RBC using ammonium chloride (ACK) lysis buffer. The draining lymph The FITC-conjugated CD47 mAb “B6H12” and an Ab against BNIP3 were nodes (LN) of mouse ears (3 to 5 mice per experiment) were collected 24 h from BD Pharmingen; the mAb against ubiquitin was from Sigma-Aldrich; after oxazolone challenge or treatment with vehicle alone (n ϭ 3 per time point the Ab against ␤-actin was from Cell Signaling Technology; the mAb and genotype). Cells were isolated from the nodes by gently forcing them against human CD3 (X3) was produced in our laboratory and described through a 70-␮m cell strainer into 3.0 ml of HBSS containing 2% FBS. elsewhere (8); the mAb “Ab-4” against TSP-1 was from NeoMarkers; and Blood from healthy adult donors was obtained by venipuncture in citrate the conjugated Abs anti-human CD25 (PE), anti-human CD4, and CD8 and PBMCs were isolated by Ficoll-Isopaque density gradient centrifuga- (PerCP), the anti-mouse CD3 (PE and FITC), and the anti-mouse CD31 tion (Nycomed/ALTANA Pharma). Donors did not take any drugs during were from BD Biosciences. The 4N1K (KRFYVVMWKK) and 4NGG the previous 10 days. Approval for this study was obtained from the in- (KRFYGGMWKK) peptides were purchased from Genosys Biotechnolo- stitutional review board of the French National Institute of Health and gies. The proteasome inhibitor MG132 was purchased from Calbiochem. Medical Research; informed consent was provided according to the Dec- laration of Helsinki. Mice Flow count fluorospheres (Beckman Coulter) were used to quantify the number of living cells by flow cytometry. Under each condition, 20,000 Wild-type (WT) C57BL/6 mice were purchased from Charles River Lab- microbeads were added just before the analysis, 2000 microbeads were oratories-Iffa Credo. C56BL/6 mice deficient in CD47 were generated as collected, and the number of lymphocytes in region R1 was determined previously described (37). WT 129/SvJ mice and TSP-1- and TSP-2-defi- using a FACScan flow cytometer (BD Biosciences). cient mice on a mixed 129SvJ/129SvEmsϩTer background were previously described (38, 39). Female mice between 7 and 12 wk of age were Immunofluorescence analysis used. Mice were maintained in our animal facility. All mice used were cared for in accordance with the Institut National de la Santeétdela Three-color flow cytometry analysis was performed on a FACScan device. Recherche Me´dicale (INSERM) guidelines. Studies were conducted after PBMCs were incubated at 4°C for 30 min in the dark in 100 ␮l of PBS and full approval of the ethical committee of INSERM. 0.1% BSA with saturating concentrations of directly conjugated mAbs. Cells were washed twice with PBS and 0.1% BSA. Forward and right- Induction of cutaneous inflammation angle scatter gatings were set to include lymphocytes. Mean fluorescence intensity (MFI) values were computed from 10,000 gated cells and are Cutaneous inflammation was induced in the skin of 10-wk-old female WT presented as the difference between the MFI of tested cells and the MFI of ϭ C57BL6 mice (n 6 for each group) as previously described (40). The background staining. Experiments were repeated three times. mice were sensitized by topical application of a 2% oxazolone (4-ethoxym- ethylene-2-phenyl-2-oxazoline-5-one; Sigma-Aldrich) solution in acetone/ Assay for apoptosis olive oil (4:1; v/v) to the shaved abdomen (25 ␮l). Five days after sensitization the right ears were challenged by topical application of 16 ␮lofa Mouse ears were excised 30 h after oxazolone challenge, diced, and di- 0.4% oxazolone solution, whereas the left ears were treated with vehicle gested in 1 mg/ml collagenase D (Sigma-Aldrich) in HBSS medium for 1 h alone. The extent of inflammation was measured daily using the mouse ear at 37°C. Cells were washed in HBSS medium with 5% FCS and resus- swelling test (40). pended in PBS. Cells were then labeled with a FITC-labeled CD3 Ab and The Journal of Immunology 5933 Downloaded from http://www.jimmunol.org/

FIGURE 3. Prolonged contact hypersensitivity elicited in TSP-1- and TSP-2-deficient mice. A, Several groups of six mice, either TSP-1Ϫ/Ϫ or WT, were treated with a sensitizing dose of oxazolone (oxa) (2%) on the shaved abdomen followed by a challenging dose of oxazolone (0.4%) on the right ear. The left ear was treated with vehicle alone. Ear thickness was measured (displayed as the means Ϯ SEM) at different time intervals after challenge (representative experiment, n ϭ 3). B, The same experiment was performed on several groups of six mice, either TSP-2Ϫ/Ϫ or WT. Ear thickness was measured (displayed as the means Ϯ SEM) at different time intervals after challenge (representative experiment, n ϭ 3). C, Thirty hours after oxazolone challenge, cells were collected from the ears of TSP-1Ϫ/Ϫ, TSP-2Ϫ/Ϫ, or WT mice, and the proportion of dead T cells was determined using an annexin V assay kit. Mouse T cells were identified by a FITC-labeled CD3 mAb. Cell debris was electronically gated out based on forward light scatter. The numbers at the top of the dot plots indicate the percentage of cells in the lower left (LL) and upper right (UR) regions, respectively (representative experiment, n ϭ 3). PI, propidium iodide. by guest on September 27, 2021 the degree of cell death was quantified using annexin V-PE/7-aminoacti- TACTGT-3Ј) and 3Ј primer (5Ј-CTCTTATTCGTATGGCTG-3Ј) for mu- nomycin D (7-AAD) staining as described above. The percentage of apo- rine CD47; and the 5Ј primer (5Ј-GAACTGCACTTCAGCAATG-3Ј) ptotic cells was calculated by scoring 7-AAD and annexin V-PE-binding and 3Ј primer (5Ј-AACCTTCAGAAAGTCTGC-3Ј) for murine BNIP3. cells (as described by the manufacturer, Boehringer Mannheim) after back- ␤-Actin was used as loading control. gating on CD3ϩ cells in the third color. The cell suspensions obtained from the draining LNs were stained with PE-labeled CD3 and cell death was Real-time quantitative RT-PCR Assay quantified using annexin V-allophycocyanin/7-AAD staining. BNIP3 mRNA expression was measured by real-time quantitative PCR, Western blot analysis using the SYBR Green PCR core reagents kit in an ABI PRISM 5700 sequence detection system (Applied Biosystems) according to the manu- Freshly isolated PBMCs and splenocytes were activated for 24, 48, or 72 h facturer’s instructions. Primers (MWG Biotech) were designed using the with anti-CD3 or incubated with medium alone. Cells were then washed computer program Primer Express (Applied Biosystems) to span exon- twice in cold PBS and lysed in 2% Triton X-100 isotonic buffer with intron junctions to prevent the amplification of genomic DNA and to am- freshly added protease inhibitors (50 mM HEPES (pH 7.6), 150 mM NaCl, plify DNA fragments in the size range of 100–250 bp that is optimal for 20 mM EDTA, 10 mM sodium orthovanadate, 100 mM NaF, 2% Triton real-time RT-PCR. Reaction data were expressed as cycle thresholds, X-100, 10 ␮g/ml aprotinin, 10 ␮g/ml leupeptin, and 0.5 mM PMSF). Cell which are the PCR cycle numbers at which the fluorescent signal in each debris was removed by centrifugation and 50 ␮g of lysate per lane, sepa- reaction reaches a threshold above background. Because the precise rated by electrophoresis on 10% SDS-polyacrylamide gels, transferred to amount of total RNA added to each reaction mix and its quality were nitrocellulose membranes, and immunoblotted with rabbit polyclonal anti- difficult to assess, we also quantified the level of four different housekeep- BNIP3 (28), anti-CD47 “B6H12”, anti-ubiquitin, or anti-␤-actin Abs. The ing genes (␤-actin, ubiquitin, GAPDH, and hypoxanthine-guanine-phos- immune complexes were detected by HRP-conjugated secondary Abs phoribosyltransferase) as endogenous RNA controls. Th expression of tar- (DakoCytomation) and developed using ECL (Amersham Biosciences). get genes was measured after normalization of RNA concentrations with Densitometric assessment of Western blots has been performed using NIH the four different housekeeping genes, and values were expressed as fold Image 1.61 software (http://rsb.info.nih.gov/nih-image). increased expression above a theoretical negative sample, as previously described (41). RNA isolation, cDNA synthesis, and PCR amplification Histology and immunohistochemistry RNA was prepared from human PBMC and murine splenocytes using an RNeasy mini kit (Qiagen, Germany). cDNA was prepared from RNA using Tissues were fixed in 10% neutral buffered formalin, processed routinely, the SuperScript first strand synthesis system (Invitrogen Life Technolo- and embedded in paraffin. They were then sectioned at 6-␮m thickness and gies) for RT-PCR. RT-PCR was typically performed for 35 cycles (dena- stained with H&E. Frozen sections of ears measuring 5 ␮m, prepared on a turation at 95°C for 20 s, annealing at 68°C for 1 min, and extension at cryostat, were fixed in acetone for 10 min at 4°C and then air dried. After 72°C for 1 min). The primers used in PCR were as follows: 5Ј primer blocking with 3% BSA for 20 min, tissue sections were incubated with Abs (5Ј-GATCAGCTCAGCTACTAT-3Ј) and 3Ј primer (5Ј-ACAATGACAG against CD31 or biotinylated Abs against TSP-1 for 30 min. After washing, TGATCACT-3Ј) for human CD47; the 5Ј primer (5Ј-GTTCAGCTCAAC bound Abs were detected by incubation with HRP-conjugated secondary 5934 CD47 DOWN-REGULATES INFLAMMATION

Abs and HRP-conjugated streptavidin and revealed with diaminobenzidine (DakoCytomation). Control stainings were performed by omitting the primary Abs.

Results CD47-deficient mice suffer prolonged cutaneous inflammation To investigate the role of CD47 in inflammation in vivo, we elicited delayed-type hypersensitivity (DTH) reactions in WT and in CD47Ϫ/Ϫ mice by using oxazolone (36). The resulting inflammation is known to develop and resolve quickly and is associated with marked interstitial edema, vascular enlargement, and mononuclear cell infiltration (36). CD47Ϫ/Ϫ and WT mice (six for each group) were treated as indicated and the extent of inflammation was measured daily using the mouse ear swelling test (40) (Fig. 1A). Re- markably, ear thickness increased to the same extent, maximizing at 24 h in both types of mice. Nevertheless, WT mice displayed a sharp decrease in ear thickness during the second day, with the ear tissue returning to normal by approximately day 3. By contrast, Downloaded from CD47Ϫ/Ϫ mice retained their thick and inflamed ears for Ͼ4 days followed by a slow recovery (Fig. 1A). Forty-eight hours after challenge, histological examinations revealed that edema and mononuclear infiltration were markedly resorbed in WT mice but still elevated in CD47Ϫ/Ϫ mice (Fig. 1B). Because contact hypersensitivity is a T cell-mediated cutaneous inflammatory reaction (36), we recovered the cells infiltrating the http://www.jimmunol.org/ ears after 30 h and measured CD3ϩ T cell viability. The percentage of CD3-positive cells measured 30 h after challenge represents 25–35% of total cells, with the remaining cells being monocytes Ϫ/Ϫ and neutrophils (results not shown). As expected, CD47 mice FIGURE 4. A TSP peptide induces cell death of CD3-activated but not had a higher proportion of living, inflammatory T cells compared resting T cells via CD47. A, PBMCs from WT or CD47Ϫ/Ϫ mice (1 ϫ 106 with WT mice, suggesting that a delay in T cell death occurs in cells/ml) were stimulated for 24 h with medium alone or with a soluble CD47Ϫ/Ϫ mouse-inflamed tissues (Fig. 1C). It should be noted that anti-CD3␧ mAb (1 ␮g/ml). Cells were then treated for 1 h with the 4N1K Ϫ/Ϫ peptide (400 ␮M), the 4NGG peptide (400 ␮M), or vehicle only (co).

the amount of cell death in WT and CD47 mice could have by guest on September 27, 2021 been overestimated as a consequence of the digestion procedure Tubes containing a known number of flow count fluorospheres were used to isolate infiltrated T cells. Nevertheless, this isolation pro- used to determine by flow cytometry the relative numbers of lympho- Ϫ Ϫ cytes following treatment. The region representing living lymphocytes cess was the same for CD47 / mice and WT mice, and the dif- was defined as R1, and the number of cells in R1 was determined ferences in cell viability are still significant. To investigate whether following a sampling of 2000 reference beads (population R2) (repre- T cell apoptosis could begin earlier during the immune response at sentative experiment, n ϭ 3). The values represent means Ϯ SEM. B,A another location, we collected the draining LNs 24 h after chal- representative dot plot of forward light scatter (FSC) vs side light scat- lenge and looked for cell death (Fig. 1D). In contrast to the in- ter (SSC) of human PBMCs treated as described in A. The number of flammatory infiltrate, we could not see any difference in the via- living lymphocytes is indicated above each plot (representative exper- bility of the cells isolated from the WT or the CD47Ϫ/Ϫ LNs. iment, n ϭ 3). Taken together, these results suggest that the prolonged cutaneous Ϫ/Ϫ inflammation in CD47 mice results from a defect in T cell Ϫ Ϫ Ϫ Ϫ TSP-1 / and TSP-2 / mice were subjected to DTH reactions apoptosis in peripheral tissues. as described in Fig. 1A. Twenty-four hours after oxazolone challenge the increase in ear thickness in WT mice was similar to that TSP-1- and TSP-2-deficient mice also suffer prolonged observed in TSP-1Ϫ/Ϫ and TSP-2Ϫ/Ϫ mice (Fig. 3, A and B). In cutaneous inflammation contrast, the two types of mutant mice were slower to resolve com- Members of the TSP family have been described as natural ligands pared with WT animals after 24 h, as indicated by the prolonged ear for CD47, and TSP-1 has been shown to induce T cell apoptosis swelling (Fig. 3, A and B). These findings suggest that TSP-1 and (22, 42). TSP-2 has previously been shown to be highly up-regu- TSP-2 are both implicated in the control of inflammation. Moreover, Ϫ Ϫ Ϫ Ϫ lated in inflamed skin 24 h after oxazolone challenge and is found TSP-1 / mice suffer more prolonged inflammation than TSP-2 / Ϫ Ϫ prominently in blood vessels and dermal cells (27). We have found mice, with inflammation lasting 10 days for TSP-2 / mice and Ϫ Ϫ that in this inflammation model the expression of TSP-1 is also Ͼ13 days for TSP-1 / mice. Finally, we recovered the cells in- Ϫ Ϫ enhanced. Twenty-four hours after oxazolone challenge, an exten- filtrating the ears after 30 h and observed, as for CD47 / mice Ϫ Ϫ sive vascular thrombosis was observed in ear tissue sections (Fig. (Fig. 1C), a decrease in inflammatory T cell apoptosis in TSP-1 / Ϫ Ϫ 2A). This was identical in CD47Ϫ/Ϫ mice and their normal coun- and TSP-2 / mice (Fig. 3C). terparts. A strong staining for TSP-1 was detected not only in blood vessels, as indicated by the arrows in Fig. 2B, but also in the Death of CD3-activated, but not resting T cells, is induced by interstitium, in contact with inflammatory cells (Fig. 2B, arrow- an active peptide derived from a cell-binding domain of TSP heads). These data and the previous observations show that TSP-1 The COOH-terminal cell-binding domain of TSP-family members and TSP-2 are present in large amounts during inflammation in has been identified as the site of their interaction with CD47 (43, areas where leukocyte infiltration has developed (27). 44),and the peptide 4N1K, which is conserved in all TSP isoforms The Journal of Immunology 5935 Downloaded from

FIGURE 5. CD47 expression is constant during T cell activation while BNIP3 expression is increased. A, RT-PCR was performed on RNA prepared http://www.jimmunol.org/ from human PBMCs stimulated for 0, 6, 12, 18, and 24 h with an anti-CD3␧ (␣CD3) mAb (1 ␮g/ml) using specific primer sets for human CD47. ␤-Actin was used as a loading control (representative experiment, n ϭ 3). B, Using three-color flow cytometry, human PBMCs stimulated for 0, 24, and 48 h with anti-CD3␧ mAb (1 ␮g/ml) were stained with FITC-conjugated CD47 mAb, PE-conjugated CD25 mAb, and a mix of PerCP-conjugated CD4 and CD8 mAb. Forward/side scatter dot plots were used to segregate the live lymphocyte population (R1 gate as described in Fig. 4), and T lymphocytes were defined as the CD4ϩ and CD8ϩ fractions of the R1 gated cells. The activation level of peripheral T lymphocytes was visualized with CD25 staining (n ϭ 3). The MFI of CD47 surface expression was measured. C, RT-PCR was performed on RNA prepared from murine splenocytes stimulated for 0, 6, 12, 18, and 24 h with anti-CD3␧ mAb (1 ␮g/ml) using specific primer sets for murine BNIP3 (n ϭ 3). D, Real-time RT-PCR was used to determine the relative levels of expression of BNIP3 and Bax in human PBMCs stimulated for 0, 6, 12, 18, and 24 h with anti-CD3␧ mAb (1 ␮g/ml) normalized against the mean of ␤ ϭ four endogenous controls ( -actin, GAPDH, hypoxanthine-guanine-phosphoribosyltransferase, and ubiquitin) (n 3). E, Murine splenocytes were treated by guest on September 27, 2021 with anti-CD3␧ mAb (1 ␮g/ml) for the indicated period of time and then subjected to Western blot analysis with an anti-BNIP3 Ab. The same membrane was blotted with a mouse anti-actin Ab as a control for equal loading and transfer (n ϭ 3). and mimics the activity of this domain, has been used extensively mice. Notably, we obtained similar results to those obtained in the in place of TSP1, although the crystal structures of recombinant mouse studies when we used PBMC obtained from healthy vol- COOH-terminal regions of TSP have shown that this region is not unteers (Fig. 4B). Thus, these results show that TSP induces the solvent exposed on the protein surface (45–47). We have shown death of activated, but not resting T cells via a CD47-dependent that this domain induces apoptosis in Jurkat cells (22). Thus, we mechanism. investigated whether it affects normal peripheral T cell survival. Previous studies have shown that CD3 ligation was required to induce the sensitization of normal human peripheral T cells to Expression of CD47 and BNIP3 during T cell activation CD47-induced cell death (48). Therefore, in this study we tested in Recently it has been observed that the expression of CD47 in- a murine model whether TSP also requires normal peripheral T creases in several cell types (e.g., fibroblasts and endothelial cells) cells to be activated to induce T cell death. For this, PBMCs ob- following appropriate stimulations. Moreover, this overexpression tained from CD47Ϫ/Ϫ and WT mice were incubated with either a triggers apoptosis when CD47 is allowed to interact with its ligand soluble CD3 mAb or in medium alone for 24 h at 37°C. An in- (18–20). We thus assessed whether such up-regulation of CD47 cubation of this duration did not affect the viability of T cells would occur on T cells upon their stimulation via the TCR. As obtained from either type of mice as assayed by annexin V binding shown in Fig. 5A, RT-PCR analysis detected no increase in CD47 and propidium iodide uptake (data not shown). After 24 h, cells mRNA expression following CD3 stimulation, indicating that were incubated with either the CD47-binding agonist 4N1K pep- CD47 is not induced during T cell activation. To confirm this tide or with the 4NGG control peptide (Fig. 4A) (49). We found result, we monitored the surface expression of CD47 on T cells comparable numbers of living lymphocytes among unstimulated after CD3 stimulation. Fig. 5B shows that CD47 was constitutively PBMCs (gated in R1) in CD47Ϫ/Ϫ and control mice, regardless of expressed on resting T cells (MFI, 63.2 Ϯ 4.22) and CD3 did not whether they were incubated with 4N1K or 4NGG peptides. In affect significantly CD47 surface expression at 24 h (MFI, 72.1 Ϯ contrast, when T cells were prestimulated for 24 h with a CD3 4.82) and at 48 h (MFI, 71.7 Ϯ 5.03). The induction of CD25 mAb and then further incubated for 1 h with 4N1K peptide, we expression was used as a control for CD3-induced T cell activa- observed a marked decrease in the number of living lymphocytes tion. These observations demonstrate that the sensitization of ac- in normal mice as compared with cells left unstimulated or stim- tivated T cells to apoptosis is not due to an increase of CD47 at the ulated with the 4NGG control peptide. In contrast, we observed no cell surface, in contrast to the reported effects in fibroblasts and proapoptotic effect of the 4N1K peptide on T cells from CD47Ϫ/Ϫ endothelial cells. 5936 CD47 DOWN-REGULATES INFLAMMATION

FIGURE 6. BNIP3 association with CD47 prevents its degradation by the proteasome. A, Splenocytes from WT and CD47Ϫ/Ϫ mice were ﬁrst stimulated with or without anti- CD3␧ mAb (1 ␮g/ml) for 48 h and then treated with or without 30 ␮M MG132 for 3 h. The samples were Downloaded from subjected to Western blot analysis using an anti-BNIP3 Ab. The same membrane was stripped and blotted with a mouse anti-ubiquitin Ab as a control for MG132 treatment and with a mouse anti-actin Ab as a con- http://www.jimmunol.org/ trol for equal loading and transfer. The ratio of the level of the BNIP3 protein to the level of the actin protein, normalized to 1 for untreated resting cells, was determined densito- metrically (n ϭ 3). B, RT-PCR was performed on RNA prepared from splenocytes from WT and CD47Ϫ/Ϫ mice stimulated with or without anti-

CD3␧ mAb (1 ␮g/ml) for 24 h and by guest on September 27, 2021 treated with or without 30 ␮M MG132 for 3 h, using speciﬁc primer sets for murine BNIP3, CD47, and actin (n ϭ 3). C, Western blot analysis was performed on WT and TSP1Ϫ/Ϫ splenocytes treated as described in A. D, RT-PCR was performed on RNA prepared from splenocytes from WT and TSP1Ϫ/Ϫ mice treated as described in B.

We have shown in a previous article, using coimmunoprecipi- mAb for different periods of time, and the cells were examined for tation, that CD47 is physically associated with the proapoptotic BNIP3 mRNA levels by RT-PCR. As shown in Fig. 5C, CD3 molecule BNIP3 and that CD47 ligation induces the apoptosis of ligation increased BNIP3 mRNA, which was apparent after 12 h. Jurkat T cells via BNIP3 (22). Because it has been shown that To quantify these data, cDNA from unstimulated and stimulated BNIP3 expression increases in effector CTLs (33), we next exam- cells was subjected to a real-time quantitative PCR assay (Fig. ined the effect of T cell stimulation on the expression of BNIP3. 5D). As a control, we compared gene expressions of BNIP3 and Human and murine PBMCs were incubated with soluble CD3 Bax, another member of the Bcl2 family, which is known to be The Journal of Immunology 5937 unaffected during T cell activation (33). As expected, no signifi- with TSP-1 and with the delayed up-regulation of TSP-2, at least cant difference was observed in the gene expression of Bax fol- in the case of wound healing (50, 51). lowing CD3 ligation. In contrast, BNIP3 gene expression in- However, given the diversity of TSP receptors, the delayed increased Ͼ8-fold after 24 h of induced activation. Western blot flammation observed in TSP1 and TSP2Ϫ/Ϫ mice probably reflects analysis also revealed higher levels of BNIP3 protein expression a more complex model. Indeed, other receptors on the T cell sur- after 48 and 72 h in CD3-stimulated cells compared with unstimu- ␣ ␤ face, such as the 4 1 integrin or heparan sulfate proteoglycan lated cells (Fig. 5E). Notably, cell incubation with the 4N1K pep- receptors, could also influence the extent and duration of inflam- tide did not alter BNIP3 expression regardless of whether or not mation (52, 53). In addition, the anti-inflammatory activity of they were activated by CD3 (data not shown). TSP1 can be driven by mechanisms other than apoptosis, such as inhibition of the production of proinflammatory cytokines by den- CD47 allows BNIP3 to accumulate in the cell by preventing its dritic cells through interactions with CD47 and CD36 (54) or fa- degradation by the proteasome cilitation of the clearance of damaged tissues and cells (55). Fi- BNIP3 mRNA expression has been shown to increase in response nally, through its interaction with CD47, TSP1 induces the to different stimuli, but the protein accumulates only after a pro- conversion of CD4ϩCD25Ϫ T cells into CD4ϩCD25ϩ regulatory longed exposure because of its degradation by the proteasome T cells in LNs and reduces effector T cell immune responses (13). (33). Cell death has been shown to occur in 3 or 4 days after a This model is further complicated by the fact that SIRP␣, another sufficient amount of BNIP3 accumulates in the cytoplasm (33). To CD47 ligand, is an inhibitory receptor on macrophages and the determine whether the inability to regulate degradation of BNIP3 lack of CD47 might affect phagocytosis (56). These considerations Ϫ/Ϫ Downloaded from could explain the failure of apoptosis in CD47 mice, we stud- highlight the fact that, even if this inflammatory model is princi- ied the effects of the proteasome inhibitor MG132 on BNIP3 pro- pally mediated by CD8ϩ T cells, the manifestation of the reaction Ϫ/Ϫ tein expression in CD47 and WT splenocytes after CD3 stim- depends on a much more complex cooperation between the cells Ϫ/Ϫ ulation. Fig. 6A shows that in CD47 cells the amount of BNIP3 participating in the inflammatory process, including endothelial protein is much lower than in WT cells after 48 h of CD3 stimu- cells, platelets, dendritic cells, and macrophages, and emphasize lation but increases after3hofMG132 treatment. Of note, this

the limitations of in vitro studies. Thus, our data reveal that a http://www.jimmunol.org/ short exposure to MG132 did not cause a detectable increase in the deficiency, either in CD47 or in TSP, reduces the rate of T cell BNIP3 level already present in WT cells stimulated by CD3. An apoptosis in inflamed tissues, and it seems reasonable to assume anti-ubiquitin antiserum showed the accumulation in the cells of that the prolonged inflammation observed in these animals is ubiquitinated proteins after3hofMG132 treatment. As expected, driven in part by a disrupted CD47/TSP proapoptotic pathway. MG132 had no effect on the mRNA levels of CD47 and BNIP3 It has been shown recently that endogenous TSP1 is also ex- (Fig. 6B). The BNIP3 mRNA induction and the accumulation of the protein were also detected in TSP1Ϫ/Ϫ cells after CD3 stimu- pressed on the membrane of blood T cells within minutes follow- lation and were similar to the ones observed in WT cells (Fig. 6, ing CD3 stimulation, provided the cells are in contact with an C and D). The quantification of the Western blots in Fig. 6, A and extracellular matrix substrate (57). It would be interesting to in- by guest on September 27, 2021 C, were confirmed by the measurement of intracellular BNIP3 and vestigate whether this endogenous TSP1 contributes to the induc- Bax using flow cytometry with specific Abs as an indication of tion of death signals via cis interactions with CD47. Clearly, trig- their intracellular accumulation (our unpublished data). Together, gering apoptosis via CD47 is critically dependent on the surface these data indicate that BNIP3 induction is independent of the densities of CD47 and TSPs in many cell types as shown by stim- presence of CD47 or TSP but that its accumulation is highly de- ulating fibroblasts with mechanical tension (an event occurring pendent on its association with CD47. during wound healing) or by stimulating endothelial cells with a disturbance in blood flow (18–20). In both cases, the surface density of CD47 increases, resulting in cell apoptosis. However, on T Discussion cells CD47 expression remains constant regardless of whether or The exact significance of the various and sometime opposite func- not they have been activated through the TCR. We have thus hy- tional effects that have been reported to be triggered via CD47 pothesized that, during T cell activation, another mechanism might requires clarification, and in vivo studies should shed some light on take place, such as the lowering of the apoptotic threshold, that is the role(s) played by CD47 in physiological and pathophysiolog- required for T cell apoptosis via CD47. Indeed, BNIP-3 expression ical conditions. CD47Ϫ/Ϫ mice are viable and healthy when raised increases greatly during this period of time in which T cells, under under normal conditions; however, when subjected to bacterial in- the influence of an activation stimulus, become sensitive to the fections they suffer from a defect in neutrophil migration and proapoptotic effect of CD47. This is quite consistent with the de- phagocytic activation and are more prone to succumb than their ϩ cisive role played by this protein in the proapoptotic intracellular WT relatives (37). In this study, by using a CD8 T cell-mediated skin inflammation model (35) we demonstrate in vivo that CD47 pathway triggered via CD47 (22). Accordingly, BNIP-3 has been on peripheral T cells is a powerful regulator of tissue inflamma- shown recently to be up-regulated at the transcriptional level in tion. Indeed, experiments conducted on mice deficient for CD47 or effector CTLs, leading to an increased susceptibility to activation- its ligands, TSP1 and TSP2, show a similar defect in controlling induced cell death (33). In the present study we show that this the length of inflammation and in apoptosis of infiltrating T cells. accumulation of BNIP3 in activated T cells is facilitated by its We assume that the similarity observed in these null mouse phe- physical association with CD47, which prevents its degradation by Ϫ/Ϫ notypes after the induction of DTH reactions can be attributed, at the proteasome. Indeed, in T cells obtained from CD47 mice least in part, to a defect in the TSP-CD47 interactions. Notably, BNIP3 mRNA increases following CD3 stimulation, but the over- TSP-1Ϫ/Ϫ mice suffer from a somewhat longer period of inflam- expression of the protein is reduced. In these CD47-deficient cells mation than TSP-2Ϫ/Ϫ mice, suggesting that TSP1 is a more pow- the overexpression of BNIP3 is restored by the use of a proteo- erful anti-inflammatory molecule than TSP2 and is potentially the some inhibitor. Recently it has been shown that CD47 enhances major in vivo proapoptotic ligand for CD47. This supposition is Fas-mediated cytotoxicity (14). Our data fit well with these results consistent with the more restricted expression of TSP-2 compared because it is likely that the absence of CD47 might, on the one 5938 CD47 DOWN-REGULATES INFLAMMATION hand, suppress BNIP-3 up-regulation and thus susceptibility to ac- 8. Ticchioni, M., M. Deckert, F. Mary, G. Bernard, E. J. Brown, and A. Bernard. tivation-induced cell death and, on the other, decrease Fas-depen- 1997. Integrin-associated protein (CD47) is a comitogenic molecule on CD3- activated human T cells. J. Immunol. 158: 677–684. dent mechanisms involved in cutaneous hypersensitivity (58, 59). 9. Reinhold, M. I., J. M. Green, F. P. Lindberg, M. Ticchioni, and E. J. Brown. Notably, these two mechanisms are not mutually exclusive but 1999. Cell spreading distinguishes the mechanism of augmentation of T cell activation by integrin-associated protein/CD47 and CD28. Int. Immunol. 11: might take place simultaneously in the model we describe here. 707–718. When inflammation develops in a tissue subjected to mononu- 10. Reinhold, M. I., F. P. Lindberg, G. J. Kersh, P. M. Allen, and E. J. Brown. clear/T cell infiltration, vascular thrombosis occurs in small ves- 1997. Costimulation of T cell activation by integrin-associated protein (CD47) is an adhesion-dependent, CD28-independent signaling pathway. sels, leading to the release of increasing amounts of TSPs, partic- J. Exp. Med. 185: 1–11. ularly TSP-1. Concomitantly, activation leads to an increase in 11. Waclavicek, M., O. Majdic, T. Stulnig, M. Berger, T. Baumruker, W. Knapp, and BNIP3 levels in T cells, allowing them to become more sensitive W. F. Pickl. 1997. T cell stimulation via CD47: agonistic and antagonistic effects to apoptotic signals. In the second phase, when TSP-1 reaches a of CD47 monoclonal antibody 1/1A4. J. Immunol. 159: 5345–5354. 12. Avice, M. N., M. Rubio, M. Sergerie, G. Delespesse, and M. Sarfati. 2000. CD47 critical level an apoptotic signal is delivered via CD47, leading to ligation selectively inhibits the development of human naive T cells into Th1 massive and rapid T cell apoptosis. effectors. J. Immunol. 165: 4624–4631. There are several consequences of a biphasic mode of CD47 13. Grimbert, P., S. Bouguermouh, N. Baba, T. Nakajima, Z. Allakhverdi, D. Braun, H. Saito, M. Rubio, G. Delespesse, and M. Sarfati. 2006. Thrombospondin/CD47 stimulation in terms of the functional outcomes. First, it provides interaction: a pathway to generate regulatory T cells from human CD4ϩ CD25Ϫ a powerful negative control of inflammation, because after being T cells in response to inflammation. J. Immunol. 177: 3534–3541. triggered inflammation will be dampened when both TSPs and 14. Manna, P. P., J. Dimitry, P. A. Oldenborg, and W. A. Frazier. 2005. CD47 augments Fas/CD95-mediated apoptosis. J. Biol. Chem. 280: 29637–29644. BNIP3 reach a critical threshold. Second, this model suggests that 15. Vernon-Wilson, E. F., W. J. Kee, A. C. Willis, A. N. Barclay, D. L. Simmons, there are physiologic mechanisms that can prevent undue apopto- and M. H. Brown. 2000. CD47 is a ligand for rat macrophage membrane signal Downloaded from ␣ sis because, under basal conditions, two major components are regulatory protein SIRP (OX41) and human SIRP 1. Eur. J. Immunol. 30: 2130–2137. lacking, namely TSPs in tissues and BNIP3 within T cells. Should 16. Brooke, G., J. D. Holbrook, M. H. Brown, and A. N. Barclay. 2004. Human an interaction of TSPs with CD47 occur for some reason other than lymphocytes interact directly with CD47 through a novel member of the signal inflammation, T cells are in a resting state and conditions for ap- regulatory protein (SIRP) family. J. Immunol. 173: 256–25702. 17. Piccio, L., W. Vermi, K. S. Boles, A. Fuchs, C. A. Strader, F. Facchetti, M. Cella, optosis are not achieved. Third, the apoptotic process should spare and M. Colonna. 2005. Adhesion of human T cells to antigen-presenting cells ␤ innocent unactivated bystander T cells, avoiding uncontrolled ap- through SIRP 2-CD47 interaction costimulates T-cell proliferation. Blood 105: http://www.jimmunol.org/ optosis. Fourth, in CD47Ϫ/Ϫ mice delayed apoptosis will finally 2421–2427. 18. Freyberg, M. A., D. Kaiser, R. Graf, J. Buttenbender, and P. Friedl. 2001. occur due to the slow accumulation of BNIP-3 in mitochondrial Proatherogenic flow conditions initiate endothelial apoptosis via throm- membranes, as is seen under hypoxic conditions (31, 32, 60). Ad- bospondin-1 and the integrin-associated protein. Biochem. Biophys. Res. Com- ditionally, it was observed that TSP-1Ϫ/Ϫ mice suffer from acute mun. 286: 141–149. 19. Freyberg, M. A., D. Kaiser, R. Graf, P. Vischer, and P. Friedl. 2000. Integrin- and chronic inflammations in various tissues, which is unexpected associated protein and thrombospondin-1 as endothelial mechanosensitive death in view of the supporting role of TSP-1 in neutrophil adhesion and mediators. Biochem. Biophys. Res. Commun. 271: 584–588. chemotaxis (38, 39) but is consistent with the present model. 20. Graf, R., M. Freyberg, D. Kaiser, and P. Friedl. 2002. Mechanosensitive induction of apoptosis in fibroblasts is regulated by thrombospondin-1 and integrin In summary, our results emphasize that the CD47 molecular associated protein (CD47). Apoptosis 7: 493–498. pathway can, in vivo, potently control the inflammation linked to 21. Johansson, U., K. Higginbottom, and M. Londei. 2004. CD47 ligation induces a by guest on September 27, 2021 T cell activation. This regulation appears to rely on a delicate rapid caspase-independent apoptosis-like cell death in human monocytes and dendritic cells. Scand. J. Immunol. 59: 40–49. balance between activation and apoptosis. 22. Lamy, L., M. Ticchioni, A. K. Rouquette-Jazdanian, M. Samson, M. Deckert, A. H. Greenberg, and A. Bernard. 2003. CD47 and the 19 kDa interacting protein-3 (BNIP3) in T cell apoptosis. J. Biol. Chem. 278: 23915–23921. Acknowledgments 23. Bornstein, P. 1995. Diversity of function is inherent in matricellular proteins: an We thank Drs. T. Lockwich, J. Hooks, P. Lagadec, D. D. Roberts, J. S. appraisal of thrombospondin 1. J. Cell Biol. 130: 503–506. Isenberg, M. Johansen, and A. K. Rouquette-Jazdanian for a critical reading of 24. Chen, F. H., A. O. Thomas, J. T. Hecht, M. B. Goldring, and J. Lawler. 2005. Cartilage oligomeric matrix protein/thrombospondin 5 supports chondrocyte at- the manuscript and helpful discussions, Professor P. Hoffman for histological tachment through interaction with integrins. J. Biol. Chem. 280: 32655–32661. studies, and D. Gautier and F. Hautem for animal care. 25. Hankenson, K. D., S. G. Hormuzdi, J. A. Meganck, and P. Bornstein. 2005. Mice with a disruption of the thrombospondin 3 gene differ in geometric and biome- chanical properties of bone and have accelerated development of the femoral Disclosures head. Mol. Cell. Biol. 25: 5599–5606. The authors have no financial conflict of interest. 26. Legrand, C., V. Dubernard, S. Rabhi-Sabile, and V. Morandi da Silva. 1997. Functional and clinical significance of thrombospondin. Platelets 8: 211–224. 27. Lange-Asschenfeldt, B., W. Weninger, P. Velasco, T. R. Kyriakides, U. H. von References Andrian, P. Bornstein, and M. Detmar. 2002. Increased and prolonged inflam- 1. Smith, C. A., T. Farrah, and R. G. Goodwin. 1994. The TNF receptor superfamily mation and angiogenesis in delayed-type hypersensitivity reactions elicited in the of cellular and viral proteins: activation, costimulation, and death. Cell 76: skin of thrombospondin-2-deficient mice. Blood 99: 538–545. 959–962. 28. Ray, R., G. Chen, C. Vande Velde, J. Cizeau, J. H. Park, J. C. Reed, R. D. Gietz, 2. Bernard, G., J. P. Breittmayer, M. de Matteis, P. Trampont, P. Hofman, A. Senik, and A. H. Greenberg. 2000. BNIP3 heterodimerizes with Bcl-2/Bcl-X(L) and and A. Bernard. 1997. Apoptosis of immature thymocytes mediated by E2/CD99. induces cell death independent of a Bcl-2 homology 3 (BH3) domain at both J. Immunol. 158: 2543–2550. mitochondrial and nonmitochondrial sites. J. Biol. Chem. 275: 143–14489. 3. Berndt, C., B. Mopps, S. Angermuller, P. Gierschik, and P. H. Krammer. 1998. 29. Yasuda, M., P. Theodorakis, T. Subramanian, and G. Chinnadurai. 1998. Ade- CXCR4 and CD4 mediate a rapid CD95-independent cell death in CD4ϩ T cells. novirus E1B-19K/BCL-2 interacting protein BNIP3 contains a BH3 domain and Proc. Natl. Acad. Sci. USA 95: 12556–12561. a mitochondrial targeting sequence. J. Biol. Chem. 273: 12415–12421. 4. Deas, O., C. Dumont, M. MacFarlane, M. Rouleau, C. Hebib, F. Harper, 30. Chen, G., R. Ray, D. Dubik, L. Shi, J. Cizeau, R. C. Bleackley, S. Saxena, F. Hirsch, B. Charpentier, G. M. Cohen, and A. Senik. 1998. Caspase-indepen- R. D. Gietz, and A. H. Greenberg. 1997. The E1B 19K/Bcl-2-binding protein dent cell death induced by anti-CD2 or staurosporine in activated human periph- Nip3 is a dimeric mitochondrial protein that activates apoptosis. J. Exp. Med. eral T lymphocytes. J. Immunol. 161: 3375–3383. 186: 1975–1983. 5. Pettersen, R. D., K. Hestdal, M. K. Olafsen, S. O. Lie, and F. P. Lindberg. 1999. 31. Bruick, R. K. 2000. Expression of the gene encoding the proapoptotic Nip3 CD47 signals T cell death. J. Immunol. 162: 7031–7040. protein is induced by hypoxia. Proc. Natl. Acad. Sci. USA 97: 9082–9087. 6. Skov, S., P. Klausen, and M. H. Claesson. 1997. Ligation of major histocompat- 32. Guo, K., G. Searfoss, D. Krolikowski, M. Pagnoni, C. Franks, K. Clark, K. T. Yu, ability complex (MHC) class I molecules on human T cells induces cell death M. Jaye, and Y. Ivashchenko. 2001. Hypoxia induces the expression of the pro- through PI-3 kinase-induced c-Jun NH2-terminal kinase activity: a novel apopto- apoptotic gene BNIP3. Cell Death Differ. 8: 367–376. tic pathway distinct from Fas-induced apoptosis. J. Cell Biol. 139: 1523–1531. 33. Wan, J., D. Martinvalet, X. Ji, C. Lois, S. M. Kaech, U. H. Von Andrian, 7. Mateo, V., L. Lagneaux, D. Bron, G. Biron, M. Armant, G. Delespesse, and J. Lieberman, R. Ahmed, and N. Manjunath. 2003. The Bcl-2 family pro-apop- M. Sarfati. 1999. CD47 ligation induces caspase-independent cell death in totic molecule, BNIP3 regulates activation-induced cell death of effector cyto- chronic lymphocytic leukemia. Nat. Med. 5: 1277–1284. toxic T lymphocytes. Immunology 110: 10–17. The Journal of Immunology 5939

34. Vande Velde, C., J. Cizeau, D. Dubik, J. Alimonti, T. Brown, S. Israels, 47. Kvansakul, M., J. C. Adams, and E. Hohenester. 2004. Structure of a throm- R. Hakem, and A. H. Greenberg. 2000. BNIP3 and genetic control of necrosis- bospondin C-terminal fragment reveals a novel calcium core in the type 3 repeats. like cell death through the mitochondrial permeability transition pore. Mol. Cell. EMBO J. 23: 1223–1233. Biol. 20: 5454–5468. 48. Pettersen, R. D. 2000. CD47 and death signaling in the immune system. Apo- 35. Kehren, J., C. Desvignes, M. Krasteva, M. T. Ducluzeau, O. Assossou, ptosis 5: 299–306. F. Horand, M. Hahne, D. Kagi, D. Kaiserlian, and J. F. Nicolas. 1999. Cytotox- ϩ 49. Bornstein, P., and E. H. Sage. 1994. Thrombospondins. Methods Enzymol. 245: icity is mandatory for CD8 T cell-mediated contact hypersensitivity. J. Exp. 62–85. Med. 189: 779–786. 50. Bornstein, P. 2001. Thrombospondins as matricellular modulators of cell func- 36. Desvignes, C., N. Etchart, J. Kehren, I. Akiba, J. F. Nicolas, and D. Kaiserlian. tion. J. Clin. Invest. 107: 929–934. 2000. Oral administration of hapten inhibits in vivo induction of specific cyto- toxic CD8ϩ T cells mediating tissue inflammation: a role for regulatory CD4ϩ T 51. Bornstein, P., A. Agah, and T. R. Kyriakides. 2004. The role of thrombospondins cells. J. Immunol. 164: 2515–2522. 1 and 2 in the regulation of cell-matrix interactions, collagen fibril formation, and 37. Lindberg, F. P., D. C. Bullard, T. E. Caver, H. D. Gresham, A. L. Beaudet, and the response to injury. Int. J. Biochem. Cell Biol. 36: 1115–1125. E. J. Brown. 1996. Decreased resistance to bacterial infection and granulocyte 52. Li, Z., M. J. Calzada, J. M. Sipes, J. A. Cashel, H. C. Krutzsch, D. S. Annis, defects in IAP-deficient mice. Science 274: 795–798. D. F. Mosher, and D. D. Roberts. 2002. Interactions of thrombospondins with 38. Kyriakides, T. R., Y. H. Zhu, L. T. Smith, S. D. Bain, Z. Yang, M. T. Lin, ␣4␤1 integrin and CD47 differentially modulate T cell behavior. J. Cell Biol. 157: K. G. Danielson, R. V. Iozzo, M. LaMarca, C. E. McKinney, et al. 1998. Mice 509–519. that lack thrombospondin 2 display connective tissue abnormalities that are as- 53. Li, Z., L. He, K. Wilson, and D. Roberts. 2001. Thrombospondin-1 inhibits TCR- sociated with disordered collagen fibrillogenesis, an increased vascular density, mediated T lymphocyte early activation. J. Immunol. 166: 2427–2436. and a bleeding diathesis. J. Cell Biol. 140: 419–430. 54. Doyen, V., M. Rubio, D. Braun, T. Nakajima, J. Abe, H. Saito, G. Delespesse, 39. Agah, A., T. R. Kyriakides, J. Lawler, and P. Bornstein. 2002. The lack of throm- and M. Sarfati. 2003. Thrombospondin 1 is an autocrine negative regulator of bospondin-1 (TSP1) dictates the course of wound healing in double-TSP1/TSP2- human dendritic cell activation. J. Exp. Med. 198: 1277–1283. null mice. Am. J. Pathol. 161: 831–839. 55. Savill, J., I. Dransfield, C. Gregory, and C. Haslett. 2002. A blast from the past: 40. Gad, S. C., B. J. Dunn, D. W. Dobbs, C. Reilly, and R. D. Walsh. 1986. Devel- clearance of apoptotic cells regulates immune responses. Nat. Rev. Immunol. 2: opment and validation of an alternative dermal sensitization test: the mouse ear 965–975. Downloaded from swelling test (MEST). Toxicol. Appl. Pharmacol. 84: 93–114. 56. Yamao, T., T. Noguchi, O. Takeuchi, U. Nishiyama, H. Morita, T. Hagiwara, 41. Wakkach, A., N. Fournier, V. Brun, J. P. Breittmayer, F. Cottrez, and H. Groux. H. Akahori, T. Kato, K. Inagaki, H. Okazawa, et al. 2002. Negative regulation of 2003. Characterization of dendritic cells that induce tolerance and T regulatory 1 platelet clearance and of the macrophage phagocytic response by the transmem- cell differentiation in vivo. Immunity 18: 605–617. brane glycoprotein SHPS-1. J. Biol. Chem. 277: 39833–39839. 42. Manna, P. P., and W. A. Frazier. 2003. The mechanism of CD47-dependent 57. Li, S. S., Z. Liu, M. Uzunel, and K. G. Sundqvist. 2006. Endogenous throm- killing of T cells: heterotrimeric Gi-dependent inhibition of protein kinase A. bospondin-1 is a cell surface ligand for regulation of integrin dependent T lym- J. Immunol. 170: 3544–3553. phocyte adhesion. Blood 9: 3112–3120. 43. Frazier, W. A., A. G. Gao, J. Dimitry, J. Chung, E. J. Brown, F. P. Lindberg, and 58. Gorbachev, A. V., and R. L. Fairchild. 2004. CD4ϩ T cells regulate CD8ϩ T M. E. Linder. 1999. The thrombospondin receptor integrin-associated protein (CD47) http://www.jimmunol.org/ functionally couples to heterotrimeric Gi. J. Biol. Chem. 274: 8554–1560. cell-mediated cutaneous immune responses by restricting effector T cell devel- 44. Gao, A. G., F. P. Lindberg, M. B. Finn, S. D. Blystone, E. J. Brown, and opment through a Fas ligand-dependent mechanism. J. Immunol. 172: W. A. Frazier. 1996. Integrin-associated protein is a receptor for the C-terminal 2286–2295. domain of thrombospondin. J. Biol. Chem. 271: 21–24. 59. Martin, S. F., J. C. Dudda, V. Delattre, E. Bachtanian, C. Leicht, B. Burger, ϩ 45. Carlson, C. B., D. A. Bernstein, D. S. Annis, T. M. Misenheimer, B. L. Hannah, H. U. Weltzien, and J. C. Simon. 2004. Fas-mediated inhibition of CD4 T cell ϩ D. F. Mosher, and J. L. Keck. 2005. Structure of the calcium-rich signature priming results in dominance of type 1 CD8 T cells in the immune response to domain of human thrombospondin-2. Nat. Struct. Mol. Biol. 12: 910–914. the contact sensitizer trinitrophenyl. J. Immunol. 173: 3178–3185. 46. Isenberg, J. S., L. A. Ridnour, J. Dimitry, W. A. Frazier, D. A. Wink, and 60. Kubasiak, L. A., O. M. Hernandez, N. H. Bishopric, and K. A. Webster. 2002. D. D. Roberts. 2006. CD47 is necessary for inhibition of nitric oxide-stimulated Hypoxia and acidosis activate cardiac myocyte death through the Bcl-2 family vascular cell responses by thrombospondin-1. J. Biol. Chem. 281: 26069–26080. protein BNIP3. Proc. Natl. Acad. Sci. USA 99: 12825–12830. by guest on September 27, 2021