Role of CD47 in the Induction of Human Naive T Cell Anergy¹

The maintenance of homeostasis and the induction of an immune response, whether tolerogenic or immunogenic, relies on an efficient immune system that distinguishes tissue-specific self-Ags from foreign Ags. T cell tolerance to peripheral self-Ags appears to be an important physiological mechanism to prevent autoimmune diseases while allowing protective immune response to pathogens (1). It is achieved by 1) clonal deletion, 2) T cell anergy, or 3) induction of regulatory cells (2, 3, 4).

Anergy is defined as a functional state in which viable lymphocytes fail to proliferate and secrete IL-2 upon optimal Ag receptor engagement and costimulation. The cellular and molecular mechanisms of anergy induction and maintenance have been extensively investigated in vitro and in vivo (2, 4). Most in vitro studies used primed T cells or T cell clones, often of the Th1/Th0 subsets. Note that in vitro induction of naive T cell anergy was found to be difficult (5). Several procedures were reported to induce in vitro T cell anergy, including 1) the delivery of TCR signal in the absence of costimulation, 2) IL-2R blockade during Ag receptor engagement and costimulation, 3) the delivery of nonmitogenic signal by altered peptide ligand, 4) T cell activation in the presence of IL-10 and/or TGF-β, and 5) treatment of Th1 clones by rapamycin in the absence or the presence of costimulation (6, 7, 8, 9, 10, 11). The in vitro induction of T cell anergy appears to trigger an incomplete T cell activation program, leading to impaired production and/or response to IL-2 (2, 11). Recent data suggest that anergy may result from a reduced or altered signaling through the common γ-chain of the IL-2R (γ_c)³, which is also used by IL-4, IL-7, IL-9, and IL-15 (12).

By contrast, naive T cells appear to be more sensitive than primed memory T cells to in vivo induction of anergy (5, 13). Tolerance induction may be achieved by 1) injection of superantigen, 2) administration of Ag without adjuvant, and 3) transfer of transgenic T cells to normal mice followed by i.v. administration of Ag in saline (14, 15, 16). In vivo experiments revealed that naive T cells may become anergic when the B7 costimulatory molecules expressed on APC engage T cell-associated CTLA-4 instead of CD28. The mechanisms leading to preferential interaction of B7 molecules with CTLA-4 rather than with CD28 remain to be determined; however, they are favored by a low level of B7 expression (17).

CD47, a multispan membrane protein expressed on all leukocytes, is physically and functionally associated with the vitronectin receptor (18, 19). CD47 has two ligands, i.e., the extracellular matrix protein thrombospondin (TSP) and signal regulatory protein, a transmembrane protein that is selectively expressed on myeloid cells (20, 21). It was recently reported that CD47 ligation by CD47 mAb or its natural ligand, TSP, down-regulated IL-12 production by monocytes and dendritic cells (DC) (22, 23). In the latter case CD47 ligation prevented functional DC maturation. However, T cells express much more CD47 than APC. Coengagement of TCR and CD47 by immobilized mAbs reportedly costimulated or killed lymphocytes (24, 25, 26, 27), whereas soluble CD47 mAbs inhibited allogeneic MLR (28). Here we present a novel procedure to induce the anergy of immunologically naive neonatal human T cells. Ligation of CD47 in the absence of exogenous cytokine during primary activation of umbilical cord blood mononuclear cells (CBMCs) inhibits the expression of IL-2 and IL-2R α-chain and induces a state of T cell unresponsiveness characterized by both reduced proliferation and cytokine expression. Although anergized T cells fail to respond to IL-2, induction of anergy may be prevented in primary cultures by including a large excess of IL-2 or IL-4 or by providing CD28 costimulation.

Anti-CD3 mAb (UCHT-1) was provided by Dr. P. Beverley (University College Middlesex School of Medicine, London, U.K.). Recombinant human (rh)IL-2 and rhIL-12 were provided by Drs. D. Bron (Bordet Institute, Brussels, Belgium) and M. Gately (Hoffmann-LaRoche, Nutley, NJ), respectively, and rhIL-4 was received from Immunex (Seattle, WA). The rhIL-7 was purchased from Genzyme (Ontario, Canada). The CD32 and B7.1 double-transfected mouse L fibroblasts have been described previously (29). The anti-CD47 mAb was used in soluble form (clone B6H12, mouse IgG1; BioSource International, Montreal, Canada). Neutralizing anti-IL-10 mAb (clone 19 F1.1) was obtained from American Type Culture Collection (Manassas, VA). Agonist anti-CD28 mAb (clone CD28.2) was purchased from BD PharMingen (Ontario, Canada). Mouse IgG1 control mAb anti-Rye was produced in our laboratory. PMA and ionomycin were purchased from Sigma (Ontario, Canada) and Calbiochem (La Jolla, CA), respectively.

CBMCs were isolated by density gradient centrifugation of heparinized umbilical cord blood from normal healthy volunteers using Lymphoprep (Nycomed, Oslo, Norway). CBMCs (1 × 10⁶ cells/ml) were cultured in triplicate in 24-well culture plates in 1 ml RPMI 1640 medium (BioWhittaker, Walkersville, MD) containing 5% FCS, 2 mM l-glutamine, 50 IU penicillin, and 100 μg streptomycin, and 2 μg PHA. After 3 days, cells were washed, counted with trypan blue, and cultured at a starting concentration of 0.5 × 10⁶ cells/ml in culture medium supplemented with 25 U/ml rhIL-2, in 24-well plates. After 9–12 days of IL-2 culture, cellular expansion in control cultures was ∼35-fold. Cells were then washed and restimulated for cytokine production at 1 × 10⁶ viable cells/ml with either plastic-bound anti-CD3 mAb (10 μg/ml) alone, plastic-bound anti-CD3 mAb (1 μg/ml) together with soluble anti-CD28 mAb (2 μg/ml), anti-CD3 mAb (200 ng/ml) immobilized on irradiated CD32/B7.1 L cells (2.5 × 10⁵ cells/ml), or PMA (10 ng/ml) plus ionomycin (1 μg/ml). Some cultures were directly restimulated after 6–7 days priming with no expansion in IL-2.

Cells were stained with FITC- or PE-conjugated mAbs to CD3, CD25, CD28, CD45RA, CD45RO, or isotype-matched negative controls (Ancell, London, Canada) and analyzed using a FACScan (BD Biosciences, Mountain View, CA). To detect intracytoplasmic IL-2, T cells were stimulated with anti-CD3 mAb immobilized on irradiated CD32/B7.1 L cells for 5 h in the presence of monensin (3 mM final concentration; CEDARLANE Laboratories, Hornby, Ontario, Canada). Cells were stained with an ICScreen Intracellular Staining Kit from Medicorp (Montreal, Canada). Fas (CD95) and Fas ligand (FasL; CD95L) surface expression was assessed using a two-step procedure. Cells were first incubated with biotinylated anti-Fas (M3), biotinylated anti-FasL (N0K-1; Immunex, Seattle, WA) or biotinylated class-matched negative control at 5–10 μg/ml in the presence of normal human IgG (150 μg/ml) for 1 h at 4°C. Cells were then incubated with PE-labeled streptavidin for 30 min at 4°C. The mean fluorescence intensity (MFI) was calculated as follows: MFI of sample − MFI of negative control.

Phosphatidylserine exposure was performed by flow cytometric analysis using a FACScan. Cells were double-stained with 2 μg/ml FITC-labeled annexin V (Biodesign, Kennebunk, ME) and propidium iodide (Sigma).

IL-2, IL-4, IL-5, IL-10, IL-13, lymphotoxin-α, TNF-α, and IFN-γ were measured by two-site sandwich ELISA or RIA, exactly as previously described (29, 30). TGF-β1 was measured by ELISA (R&D Systems, Minneapolis, MN) after acidification of the culture supernatants. Unless otherwise indicated, cytokines were measured after 24 h (IL-2 and IL-4) or 48 h (IFN-γ, TNF-α, lymphotoxin-α, IL-5, IL-10, and IL-13) of restimulation. For TGF-β measurement, cultures were incubated for 72 h in serum-free HB101 medium (Irvine Scientific, Santa Ana, CA). The percentage of inhibition of IL-2 secretion at restimulation and the percent inhibition of expansion (end of expansion phase in IL-2) were calculated as follows, respectively: 1 − (IL-2 (U/ml) in anti-CD47-treated cells)/(IL-2 (U/ml) in control mAb-treated cells) × 100%; and 1 − (number of viable cells in anti-CD47-treated cells)/(number of viable cells in control mAb-treated cells) × 100%.

CBMCs were collected after 6 h of stimulation in nonpolarizing conditions (medium) in the presence of anti-CD47 mAb or isotype-matched control mAb. Total RNA was prepared with the RNeasy Total RNA kit (Qiagen, Chatsworth, CA). Exactly as previously described (30), 1 μg RNA from each sample was reversed transcribed by the GeneAmp RNA PCR kit from PerkinElmer/Cetus (Emeryville, CA) with oligo(dT)₁₆ as the first-strand cDNA primer. One-twentieth volume of RT product was mixed with known quantities of serially diluted competitive internal standards (PCR MIMICs; CLONTECH Laboratories, Palo Alto, CA) and subjected to quantitative RT-PCR according to the manufacturer’s protocol. Target-specific primer pairs of IL-2 and G3PDH were also purchased from CLONTECH Laboratories. After 36 cycles of amplification, the PCR products were resolved on a 1.8% agarose gel containing ethidium bromide. The intensities of competitor-generated bands and cDNA sample-generated bands were compared with determine the quantity of target gene product. The amount of target cDNA was ascertained by determining the amount of competitor required to produce equal molar quantities of target and competitor products. The photographs of agarose gels were further analyzed by computer imaging (NIH image, version 1.61, National Institutes of Health, Bethesda, MD), and the ratio of the mean histogram of target-generated to competitor-generated band was calculated. The specificity of the amplified bands was validated by their predicted size.

Thymidine incorporation was assessed by adding 1 μCi/well [methyl-³H]thymidine (10 Ci/mmol; Amersham, Arlington Heights, IL) during the last 6 h of the culture. Triplicate cultures were then harvested onto glass-fiber filters, and the radioactivity was counted using liquid scintillation.

Student’s paired t test was used to determine the statistical significance of the data. Values of p < 0.05 were chosen for rejection of the null hypothesis.

We recently reported that CD47 ligation on CBMCs by CD47 mAb or its natural ligand, TSP, selectively impaired IL-12-induced differentiation of naive T cells into Th1 effectors with no immune deviation toward Th2 (31). In the present study CBMCs were activated with PHA in the presence of CD47 mAb, but in the absence of exogenous cytokine. After 3 days cells were washed, and 0.5 × 10⁶ viable cells/ml were cultured for 9–12 days in IL-2 (25 U/ml) to allow cellular expansion of effector T cells. Cells were washed again, adjusted to 1 × 10⁶ viable cells/ml, and restimulated with either plastic-coated anti-CD3 mAb alone or anti-CD3 together with anti-CD28 mAb. As shown in Fig. 1,A, cells that had been pretreated with CD47 mAb, referred to hereafter as CD47-primed cells, displayed an impaired proliferative response to optimal TCR stimulation. Of interest, a similar state of hyporesponsiveness was found when CBMCs were activated for 6 instead of 3 days in the presence of CD47 mAb and immediately restimulated without expansion in IL-2 (Fig. 1,B). The phenotype of CD47-primed cells was similar to that of control mAb-pretreated cells (at the end the IL-2 expansion and before restimulation). T cells expressed the phenotype of newly generated effector cells: CD3⁺, CD45RA⁺, CD45RO⁺, and CD25⁺ (Fig. 1,C). CD47-primed cells not only failed to proliferate, but also showed a drastic decrease in IL-2 production in response to TCR and CD28 coengagement (Fig. 1,D). The impaired IL-2 secretion was confirmed at the single-cell level (Fig. 1,E). In contrast, CD47-primed cells did proliferate and produce IL-2 as well as control mAb-treated cells in response to PMA plus ionomycin (Fig. 1, A and D), suggesting that the hyporesponsive T cells remained alive. From these data, we proposed that CD47 ligation induced T cell anergy, a cellular state in which viable T lymphocytes fail to proliferate and secrete IL-2 when optimally restimulated through Ag-specific receptor (signal 1) and CD28 (signal 2).

Th1 clones rendered anergic in vitro following TCR stimulation in the absence of signal 2 or by blocking IL-2R signaling reportedly respond to exogenous IL-2, but not IL-12 (7, 32, 33). Here we show that the proliferative response of CD47-primed cells to IL-2, IL-4, or IL-12 and anti-CD3 mAb remained significantly lower than that of control mAb-treated cells (Fig. 2,A). These data extended the above observations (Fig. 1) that expansion of CD47-primed cells for 9–12 days in IL-2 did not prevent the induction of anergy.

Human T cells stimulated for 10 days in IL-10 were reported to become anergic and lose IL-2 responsiveness (9). Also, IL-10 suppressed IL-2 production by activated T cells (34). In the present case the induction of anergy by CD47 mAb was not likely to be mediated by IL-10, inasmuch as CD47 mAb inhibited IL-10 production in primary cultures, and inclusion of neutralizing anti-IL-10 mAb at priming did not prevent induction of anergy (Fig. 2, B and C).

We concluded that CD47 ligation on naive cells induced a profound state of anergy that could not be reversed by restimulating the T cells in IL-2 with or without TCR engagement.

In most studies anergized Th1 clone cells failed to produce IL-2, but largely retained their capacity to secrete other cytokines (6, 7, 32). T cells that were rendered unresponsive after priming of PHA-activated CBMCs in the presence of CD47 mAb were severely impaired in their capacity to produce both Th1 (IFN-γ, TNF-α, and lymphotoxin-α) and Th2 (IL-4, IL-5, and IL-13) cytokines (Table I). Note that CD47-primed cells produced a decreased amount of IL-10 and TGF-β, which were reported to be up-regulated in anergic cells displaying immunoregulatory functions (35). Importantly, the hyporesponsiveness could not be ascribed to increased activation-induced cell death as determined by annexin/propidium iodide staining on day 1 or 2 postrestimulation (Fig. 3 A and data not shown).

Phenotypic analysis of anergized T cells at restimulation revealed no significant difference compared with control mAb-treated cells in terms of CD3, CD28, Fas, and FasL expression (Fig. 3,B). This indicated that hyporesponsiveness did not result from TCR/CD3 or CD28 down-modulation. In contrast, CD47-primed cells showed a reduction in CD25 (IL-2Rα) expression after CD3 restimulation compared with control cells (MFI reduction from 435 to 245, respectively; Fig. 3,C). This decrease in CD25 might be correlated with the reduced IL-2 responsiveness of anergic T cells (Fig. 2,A). However, the expression of CD25 on CD47-primed cells was significantly up-regulated after CD3 stimulation (MFI increase from 12 to 245, respectively; Fig. 3 C), suggesting that signaling through TCR was not totally impaired, but was insufficient to promote proliferation and IL-2 secretion.

In several in vitro models, induction of anergy was associated with inhibition of IL-2 production and/or responsiveness during initial TCR-mediated cell activation (2, 4). We recently reported that CD47 mAb decreased IL-12-induced, but not IL-4-induced, IL-2 production by PHA-activated CBMCs (31). Therefore, we examined how CD47 mAb regulates IL-2 production in primary cultures of CBMCs activated in the absence of exogenous cytokine. The data in Fig. 4 show that CD47 ligation not only drastically reduced the production of IL-2 (Fig. 4,A), but also down-regulated IL-2Rα expression (Fig. 4,B) after 3 days of priming. IL-2 expression was inhibited at the protein (>10-fold suppression; Fig. 4,A) and at the transcription level (8- to 15-fold suppression; Fig. 4,C). The down-regulation of CD25 expression by CD47 mAb was associated with a decrease in IL-2 responsiveness (Fig. 4,D). Indeed, after 3 days of priming, 0.5 × 10⁶ viable cells were cultured in IL-2-supplemented medium, and after 9 days we found that the recovery of viable cells was drastically reduced in the CD47-primed cell populations; this impaired cellular proliferation was significantly correlated with the reduced capacity of CD47-primed cells to secrete IL-2 at restimulation (Fig. 4 E). Note that inhibition of expansion in IL-2 was associated with complete abrogation of IL-2 production in four of 10 experiments. These data suggested that induction of anergy in naive T cells by CD47 mAb may result from a blockade of IL-2 production and/or IL-2R signaling.

TCR stimulation in the absence of costimulation results in low level of IL-2 production and induction of anergy (2, 4, 7). CD28-mediated signaling costimulates T cells and thus prevents the induction of anergy in T cell clones, whereas B7/CTLA-4 signaling is a initiator of anergy (17, 36, 37, 38). We recently reported that CD47 ligation on human DC down-regulated B7 molecule (CD80 and CD86) expression, preventing DC maturation; CD47 mAb-pretreated DC induced functional adult T cell hyporesponsiveness (23). Here we first showed that CD47 mAb slightly, but significantly, decreased CD28 expression on naive T cells after 3 days priming, whereas it failed to up-regulate CTLA-4 expression (Fig. 5,A and data not shown). Because we used unfractionated cells (CBMCs) expressing a low level of B7, we hypothesized that CD47 mAb-mediated inhibition of IL-2 production might result from a further decrease in B7 expression, lowering B7/CD28 interactions and/or favoring B7/CTLA-4 signaling. This led us to investigate the effect of CD28 costimulation on IL-2 production, CD25 expression, IL-2 responsiveness, and induction of anergy in CD47 mAb-treated cultures. As depicted in Fig. 5, B and C, agonistic CD28 mAb strongly increased IL-2 production and CD25 expression at priming, but did not completely overcome the CD47 mAb-mediated inhibition. However, and most importantly, this mAb largely, but incompletely, reversed the anergic state, because the CD47-primed cells proliferated as well as their nonanergized counterparts upon optimal restimulation, but still secreted a lower amount of IL-2 (Fig. 5 D). We concluded that induction of anergy in naive T cells by CD47 ligation could be largely reversed by providing full T cell activation (i.e., CD28 costimulation). However, these two signals failed to completely restore IL-2 production in primary and secondary CD47 mAb-treated cultures.

Several studies reported that CD28 costimulation does not directly prevent anergy. Rather, costimulation prevents anergy by up-regulating the production of IL-2 (2, 7, 36). For instance, incubating T cell clones with live APC along with Ag in the presence of anti-IL-2 plus anti-IL-2R Abs resulted in the induction of anergy (39). Therefore, we examined whether inclusion of excess amount of IL-2 at priming might overcome the down-regulation of CD25 expression and IL-2 responsiveness of CD47 mAb-treated cells. Exogenous IL-2 was added in primary cultures for 3 days, cells were washed, examined for CD25 expression, expanded in IL-2 for 9–12 days, and restimulated by coengagement of TCR and CD28. As shown in Fig. 6, IL-2 strongly up-regulated CD25 expression and totally overcame CD47-mAb mediated suppression of 1) IL-2Rα expression (Fig. 6,A), 2) cellular expansion (data not shown), and 3) functional hyporesponsiveness at restimulation (Fig. 6,B). We previously reported that CD47 mAb did not impair the IL-4-induced development of naive T cells into Th2 effectors (31). We confirmed and extended these observations by showing that IL-4 prevented the down-regulation of IL-2 production and the induction of anergy (Fig. 6, C and D). Note that IL-7, using the same γ_c of IL-2R as IL-2 and IL-4, up-regulated IL-2 secretion in primary cultures of control mAb-treated cells. However, it failed to overcome CD47-mediated inhibition of IL-2 production at priming and functional hyporesponsiveness of CD47-primed cells at restimulation (Fig. 6, C and D). Taken together, the data indicate that CD47 ligation down-regulated IL-2 production and the IL-2 response of naive T cells and that providing IL-2 during primary cultures completely prevents the induction of anergy.

In the present study we show that ligation of CD47 by CD47 mAb during primary activation of CBMCs in the absence of exogenous cytokine led to the development of hyporesponsive T cells. CD47-primed cells failed to proliferate and produce IL-2 along with a diffusely suppressed cytokine production capacity. Functional T cell inactivation was associated with the acquisition of an effector/memory phenotype (CD3⁺, CD45RO⁺, and CD25^low). TCR by-passing by PMA and ionomycin restimulation induced a normal response, indicating that CD47-primed cells remained viable. This functional impairment did not result from increased activation-induced cell death after stimulation. Rather, it reflected a profound state of anergy with a defect of IL-2 production, documented at the single-cell level and a failure to proliferate that could not be reversed by exogenous IL-2, IL-4, or IL-12 at restimulation. Thus, CD47 appears as a novel target to induce in vitro anergy of human naive T cells.

We recently reported that CD47 ligation during primary activation of CBMCs in the presence of IL-12 and anti-IL-4 selectively impaired the development of naive T cells into Th1 effectors without inducing Th2 (31). In the course of this study we also noticed that CD47-primed Th1 effectors not only secreted reduced amounts of IFN-γ at the single-cell level, but also displayed an impaired proliferative response (M.-N. Avice, unpublished observations). Note that inclusion of IL-2 at priming restored IL-12-induced Th1 development and was shown to fully prevent induction of anergy (Fig. 6). Thus, the ability of CD47 ligation to both inhibit IL-12-induced Th1 differentiation and anergize naive T cells with no immune deviation toward Th2 cells appeared to be unique. In vitro or in vivo procedures reported to inhibit Th1 cell development are not known to induce anergy, and conversely, the induction of anergy usually does not prevent the polarization of the lymphokine-producing phenotype of anergic T cells (reviewed in Ref. 1). Treatment of mice with soluble Ag in the absence of adjuvant generated Ag-specific anergic T cells that failed to clonally expand in response to subsequent optimal immunization in the presence of adjuvant (13). In that case, induction of tolerance was associated with selective inhibition of Th1 development. However, the administration of IL-12 during tolerance induction (13) or immunization of tolerant mice with Ag in CFA (40) did not correct anergy, but allowed the acquisition of a Th1 phenotype.

In other similar types of experiments, anergy was associated with immune deviation, i.e., suppression of Th1, but enhancement of Th2 cytokine production (15, 41). Most interestingly, the anergic T cells generated in these in vivo experiments remained capable of serving Th1 or Th2 effector functions (1, 40). We suggest that CD47-primed cells should lack such functional activity because of 1) the diffuse suppression of Th1 and Th2 cytokine production and 2) their impaired expression of CD154 (M.-N. Avice, unpublished observations). Th1 clones rendered anergic in vitro by TCR engagement in the absence of costimulation still produced substantial amounts of IFN-γ and, unlike CD47-primed cells, remained responsive to IL-2 (11). Similarly, anergized Th0 clones produced normal levels of IL-4, contrasting with a marked suppression of IL-2 production and IL-4 unresponsiveness (42, 43).

CD47-primed cells shared a number of features with the recently described anergic T cells obtained after chronic in vitro TCR-mediated stimulation of human or mouse primary T cells in the presence of IL-10 (9, 35). Both types of cells were anergic, as defined by reduced proliferation and IL-2 production following anti-CD3 or anti-CD3 plus anti-CD28 stimulation. Moreover, similar to CD47-primed cells, IL-10-primed T cells failed to respond to IL-2 and to up-regulate CD25 upon stimulation. Several observations argue against the involvement of IL-10 in the CD47-mediated functional inactivation of naive T cells. First, CD47 ligation did not up-regulate the production of IL-10 in primary cultures (Fig. 2,B). Second, IL-10 production by CD47-primed cells was suppressed (Table I). Third, inclusion of neutralizing anti-IL-10 mAb at priming had no effect on the development or maturation of anti-CD47-treated cells (Fig. 2,C). More recently, IL-10-primed cells were shown to display in vivo and in vitro immunoregulatory activity that could be largely ascribed to their production of high levels of IL-10 and normal amounts of TGF-β (35). As indicated in Table I, the production of these two immunosuppressive cytokines was reduced in CD47-primed cells.

T cell anergy has been extensively investigated in vitro on T cell clones and more recently in vivo by using adoptively transferred TCR transgenic T cells (reviewed in Ref. 2). Activation of T cell clones in conditions leading to impaired IL-2 production (e.g., TCR engagement without costimulation; stimulation with altered peptide ligand in the presence of costimulation) or to impaired response to IL-2 (e.g., rapamycin) induces anergy. In the first case, as in the present study, inclusion of IL-2 prevents anergy induction, and in both cases anergy appears to result from reduced or impaired signaling via the IL-2R resulting in the blockade of activated T cells at the G₁ stage of the cell cycle (11). Induction of anergy in alloreactive human T cell clones (by TCR ligation in the absence of costimulation) was reportedly prevented by engaging the IL-2R γ-chain with either an agonist mAb or a cytokine using the same γ-chain receptor as IL-2 (IL-4 or IL-7) (7). Our results strongly suggested that CD47 ligation in primary cultures decreased IL-2R signals by inhibiting 1) IL-2 production, as shown at the protein and mRNA levels; 2) the response of naive T cells to IL-2; and 3) the expression of CD25, the IL-2R α-chain. IL-2 and IL-4 restored IL-2 production and CD25 expression and prevented anergy. However, our findings that IL-7 failed to prevent anergy induction did not support this without excluding a critical role of the signaling pathway in the CD47-mediated hyporesponsiveness. The availability of agonistic γ_c mAb should help to test this possibility. Note that IL-4 not only overcame CD47-mediated anergy, but also led to the normal development of Th2 effectors (not detailed).

The possible mechanisms underlying the induction of anergy by soluble CD47 mAb might be 1) an indirect effect on DC and subsequent down-regulation of B7/CD28 interactions, which then favors B7/CTLA-4; 2) a blockade of a positive signal delivered by CD47 ligands to T cells; or 3) a negative signal, mimicking CD47 ligands, directly transmitted to T cells. To date, attempts to investigate whether CD47 mAb may directly induce anergy of purified naive CD4 ⁺ or CD8⁺ T cells have failed, because naive T cells absolutely required two signals, TCR and CD28 costimulation, for full in vitro activation and differentiation into effector cells (29). However, CD47 ligation on purified activated adult human T cells or T cell clones reportedly enhanced or inhibited T cell activation (24, 28). Recent findings provided evidence that CD47 is localized in membrane lipid rafts (44). The way the CD47 molecule is engaged probably leads to opposite biologic effects. Immobilized anti-CD47 mAb or signal regulatory protein Ig costimulated T cell activation in response to suboptimal anti-CD3 mAb via CD28-independent signaling pathway (24). CD47 ligation converted antagonist peptides into agonists in a murine hybridoma T cell line. If either or both mAbs were used in solution, there was no costimulation. In contrast, soluble mAbs were reported to exert inhibitory activities (28), as shown in the present study. Note that similar opposite biological effects of the two molecular forms of CD47 mAb were reported in granulocyte function (45).

Our data further showed that the induction of naive T cell anergy by soluble CD47 mAb can be prevented by CD28 costimulation, suggesting an indirect effect on APC. CD47 mAb down-regulated CD80/86 costimulatory molecule expression (23) The mechanisms responsible for CD28-mediated enhancement of IL-2 production and prevention of anergy induction are beginning to be elucidated (2, 11). Increased levels of IL-2 transcription and mRNA stability led to enhanced IL-2 production. Furthermore, CD28 signaling up-regulated G₁ cell cycle kinases and the degradation of the cell cycle inhibitor Kip-1 (17). It was also reported to block the production of inhibitory factors such as Nil2a (2). These CD28-mediated events led to entry of the T cells into the cell cycle. In that regard, rapamycin probably induced anergy by blocking the degradation of Kip-1 following IL-2R engagement (11).

CD28^−/− mice lead to donor-specific tolerance at least by favoring B7/CTLA-4 interactions (17). Several reports indicated that the CD28 homologue CTLA-4 is a negative regulator of T cell activation. CTLA-4^−/− mice developed a fatal lymphoproliferative disease (37, 38, 46). CTLA-4 blockade in vivo has been shown to augment the immune response, resulting in exacerbated autoimmune disease and enhanced allograft rejection (46). A recent report showed that the inhibitory effect of CTLA-4 was mediated by either its cytoplasmic domain or through the ability of its extracellular domain to compete for ligands (47). However, induction of anergy may occur in CTLA-4-deficient T cells (48). Our unpublished observations indicate that CD47 mAb did not increase, but, rather decreased, CTLA-4 expression on naive CD4⁺ T cells and that CTLA-4 mAb added during primary cultures was inefficient in preventing the induction of anergy. B7 family has been expanded to six members (49). Among those, PD-L1/PD-1 and PD-L2/PD-1 have been proposed as novel negative regulators of immune responses that may be involved in the maintenance of peripheral self-tolerance. Therefore, the signaling pathway mediated by TCR, CD28, CTLA-4, and new B7 ligands is rather complex, and the integration of these signals will dictate the outcome of T cell stimulation (50, 51, 52).

Whether CD47 and its two ligands, TSP and signal regulatory protein-α, may interfere with this complex signaling pathway to induce naive T cell anergy is still unknown. They may modulate B7/B7 ligand expression and/or interaction or directly deliver a negative signal to both T cells and/or APC. TSP reportedly inhibited TCR-mediated T cell activation and IL-2 secretion by T cells (53). These results, although in apparent contradiction with its costimulatory activity on inflammatory T cells (54), are in agreement with our observations that a CD47 binding peptide (4N1K) also induces naive T cell anergy (M.-N Avice and M. Sarfati, unpublished observations). 4N1K peptide prevented the development of naive T cells into Th1 effectors (31).

Inflammatory cytokines, such as IL-12, by themselves or local inflammation are insufficient to prevent anergy but play a central role in the development of type 1 autoimmune diseases (13, 55). Indeed, treatment of mice with IL-12 antagonists resulted in impaired Th1 differentiation, but not in anergy. In the present report CD47 ligation prevented both Th1 development and induced anergy even in the presence of IL-12. Combined with its inhibitory effect on IL-12 production by APC, it is proposed that CD47 is a potential target to anergize disease-promoting Ag-specific T cells and as such may be considered an attractive therapeutic approach in autoimmunity, graft-vs-host disease, and organ transplantation.

We fully appreciate the excellent secretarial assistance of Norma Del Bosco.