Signal Regulatory Protein (SIRPα), a Cellular Ligand for CD47, Regulates Neutrophil Transmigration*

Recent studies have demonstrated that CD47 plays an important role in regulating human neutrophil (PMN) chemotaxis. Two ligands for CD47, thrombospondin and SIRPα, have been described. However, it is not known if SIRP-CD47 interactions play a role in regulating PMN migration. In this study, we show that SIRPα1 directly binds to the immunoglobulin variable domain loop of purified human CD47 and that such SIRP-CD47 interactions regulate PMN transmigration. Specifically, PMN migration across both human epithelial monolayers and collagen-coated filters was partially inhibited by anti-SIRP monoclonal antibodies. Similar kinetics of inhibition were observed for PMN transmigration in the presence of soluble, recombinant CD47 consisting of the SIRP-binding loop. In contrast, anti-CD47 monoclonal antibodies inhibited PMN transmigration by markedly different kinetics. Results of signal transduction experiments suggested differential regulation of PMN migration by SIRPversus CD47 by phosphatidylinositol 3-kinase and tyrosine kinases, respectively. Immunoprecipitation followed by Western blotting after SDS-PAGE under nonreducing conditions suggested that several SIRP protein species may be present in PMN. Stimulation of PMN with fMLP resulted in increased surface expression of these SIRP proteins, consistent with the existence of intracellular pools. Taken together, these results demonstrate that PMN migration is regulated by CD47 through SIRPα-dependent and SIRPα-independent mechanisms.

CD47 is a transmembrane Ig superfamily member that is expressed in most tissues, and its function has been broadly implicated in multiple cellular processes including neutrophil phagocytosis, T cell activation (1,2), T and B cell apoptosis (3,4), platelet activation (5,6), and stroma-supported erythropoiesis (7). Other studies have demonstrated that antibodies to CD47 interfered with ␣ v ␤ 3 -mediated cell functions (8) and inhibited endothelial Ca 2ϩ influxes during cell adhesion to fibronectin-or vitronectin-coated surfaces (9). Previous studies from our group and others have shown that CD47 plays an important role in the regulation of PMN 1 migration (10 -12). It was shown that monoclonal antibodies against CD47 inhibited the rate of PMN migration across both epithelial monolayers and cell matrix-coated transwell filters without reducing the total amount of PMN migration (10), suggesting a positive regulatory role of CD47 in PMN migration. The precise mechanism of CD47-mediated regulation of PMN migration is not known. However, specific tyrosine phosphorylation events appear to convey the downstream signals from cell surface CD47 to regulate the rate of PMN migration (10).
SIRPs form a family of transmembrane glycoproteins expressed in a variety of tissues (24). However, within these tissues, SIRPs are only selectively expressed in certain cell types (25). In mice, SIRPs (termed SHPS-1) are richly expressed in hematopoietic cells including macrophages and myeloid cells, but not in T and B cells (24). In humans, SIRPs are expressed in monocytes, granulocytes, dendritic cells, and CD34 ϩ CD38 Ϫ CD133 ϩ bone marrow stem/progenitor cells but not in lymphocytes (20,26). Through cDNA library screening, multiple homologous sequences that account for at least 15 additional SIRP members have been reported (27).
Primary structural analysis indicate that all SIRPs share common structural motifs including a single transmembrane segment and an N-terminal extracellular domain that contains three Ig-like loops connected by three pairs of disulfide bonds. The C-terminal intracellular domain structurally separates two subfamilies of SIRPs, termed SIRP␣ and SIRP␤ (27). SIRP␣ has a long intracellular domain containing four tyrosine residues that form two immunoreceptor tyrosine-based inhibi-tory motifs, whereas SIRP␤ contains a basic lysine residue followed by a short intracellular tail that serves as a receptor for DAP12, a protein with an immunoreceptor tyrosine-based activation motif (27,28).
Similar to other immunoreceptor tyrosine-based inhibitory motif domain-containing proteins such as CD3, T cell receptor , FcR␥, and B cell receptor (29,30), SIRP␣ has been shown to play an important role in regulating cellular responses to a wide variety of different stimuli. For example, treatment of tissue-cultured cells with growth factors (including plateletderived growth factor and epidermal growth factor), growth hormone, insulin, colony-stimulating factor, lysophosphatidic acid, etc., has been shown to induce phosphorylation of tyrosines on the intracellular immunoreceptor tyrosine-based inhibitory motif domain of SIRP␣, resulting in binding to Src homology 2 domain-containing tyrosine phosphatase-1 or 2 (SHP-1 and 2) (24,27,31). SIRP␣ binding to SHP-1 or SHP-2 mediates positive or negative signals that regulate a variety of cellular functions, respectively (27,32,33).
The recent reports of CD47 binding to SIRP␣ prompted us to examine the role of SIRP in CD47-mediated regulation of PMN transmigration. In this report, we confirm that human CD47 binds specifically to SIRP␣ and find that SIRP␣-CD47 interactions partially inhibit fMLP-driven PMN chemotaxis, but with kinetics that are markedly different from the inhibition observed with anti-CD47 mAbs. Furthermore, our results suggest that SIRP and CD47 regulate PMN migration through different intracellular signaling pathways. Finally, we report that several protein bands reactive with SIRP antibodies under nonreducing conditions are present in PMN and can be upregulated to the cell surface after chemotactic stimulation. The significance of these findings is discussed in the context of PMN migration during the acute inflammatory response.
Cells-Culture and maintenance of T84 human intestinal epithelial cells (passages 50 -90) were as described previously (36). Briefly, T84 cells were grown in a 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F-12 medium supplemented with 15 mM HEPES buffer, pH 7.5, 14 mM NaHCO 3 , 40 g/ml penicillin, 8 g/ml ampicillin, 90 g/ml streptomycin, and 5% newborn calf serum. For transmigration experiments, T84 cells were grown on collagen-coated, permeable polycarbonate filters (inserts) with a surface area of 0.33 cm 2 and pore size of 5 m (Costar, Cambridge, MA) as previously described (11). Neutrophils (PMN) were isolated from whole blood of normal human volunteers by Dextran sedimentation followed by Ficoll-Hypaque separation as previously described (37). Residual RBC were lysed with isotonic NH 4 Cl. Isolated PMN were resuspended in Hanks' balanced salt solution devoid of calcium or magnesium (HBSS(Ϫ)) (4°C) at a concentration of 5 ϫ 10 7 cells/ml and were used within 4 h.
Recombinant SIRP␣1, SIRP␤1, and GST Fusion Proteins-Fusion proteins consisting of the extracellular domains of SIRP␣1 and SIRP␤1 fused to GST were produced after transfection of constructs into 293E cells and purified as previously described (20).
Recombinant CD47-Alkaline Phosphatase (AP) Fusion Protein-A construct consisting of the extracellular domain of human CD47 fused to the catalytic domain of human placental AP was prepared as previously described (10) using the AP-tag2 (19, 38) expression vector. Culture supernatant containing the recombinant protein was prepared by either transient transfection of COS-7 cells or by stable transfection of CHO-K1 cells. For the CHO-K1 transfections, the CD47-AP construct was subcloned from AP-tag2 into the tetracycline response element containing pTRE2 vector (CLONTECH). This plasmid was co-transfected into CHO-K1 cells with a plasmid (KnlstTA) consisting of a Tet-Off transactivator with a nuclear localization signal cloned into the pIRES1neo vector (CLONTECH). Clones were picked after G418 selection and assayed for AP activity. Clones with the highest expression were used for production of recombinant protein. Expressed CD47-AP fusion protein in medium was affinity-purified using PF3.1-Sepharose and eluted with 50 mM triethylamine, pH 10.5, 150 mM NaCl followed by neutralization, concentration, and dialysis. A second AP fusion protein containing the extracellular domain of JAM was produced and purified by similar methods (10).
Enzyme-linked Immunosorbent Assay and in Vitro SIRP␣1-CD47 Binding Assay-96-well microtiter plates were coated with SIRP-GST fusion protein for 2 h at room temperature (5 g/ml). After blocking nonspecific binding with 1% BSA (in HBSS), monoclonal anti-SIRP antibodies were added at 10 g/ml, followed by incubation at room temperature for 30 min. After washing, the wells were incubated with peroxidase-conjugated goat anti-mouse secondary and developed with the substrate ABTS. To assay CD47 binding to SIRP-GST, purified human CD47 (10 -100 g/ml in buffer containing 1% octyl glucoside) was diluted 50-fold in HBSS in microtiter wells (50 l) and allowed to bind for 2 h (room temperature). After blocking with BSA, the wells were incubated with SIRP-GST for 1 h. After washing, bound SIRP was assayed by incubation with goat anti-GST (Amersham Biosciences, Inc.) followed by peroxidase-conjugated rabbit anti-goat secondary. CD47 extracellular domain binding to SIRP was assayed by incubation SIRP-GST-coated wells with CD47-AP fusion protein (2-10 g/ml) for 30 min (room temperature), and binding was detected by alkaline phosphatase activity. Controls included microtiter wells coated with BSA only or with similar concentrations of immunopurified human CD11b/CD18 or JAM in the same buffers.
PMN Transmigration Assay-PMN transepithelial migration experiments were performed in the basolateral to apical direction using T84 intestinal epithelial cell monolayers as previously described (10,11). Briefly, confluent inverted T84 monolayers were washed twice with HBSS (20°C). PMN (1 ϫ 10 6 ) in 150 l of HBSS with or without antibody were added to the upper chamber of the monolayer setup. Transmigration was initiated by transferring of PMN-containing monolayers to 24-well tissue culture plates containing 1 ml of 1 M fMLP in HBSS followed by incubation at 37°C. PMN migration across monolayers into the fMLP-containing lower chambers was quantified by myeloperoxidase assay as previously described (11). PMN migration across acellular, collagen-coated filters was performed as described previously (10).

RESULTS
The Extracellular Domain of SIRP␣1 Directly Binds to Purified Human CD47 in Vitro-In a previous report, human leukocytes were shown to adhere to SIRP␣-GST-coated surfaces via CD47-mediated events (20). We confirmed that CD47 binds directly to SIRP␣1 extracellular domain by in vitro assays using purified CD47. Native CD47 was affinity-purified from human PMN, RBC, spleen, and intestinal epithelial cells (HT-29). SDS-PAGE analysis of the CD47 purified from PMN is shown in the inset of Fig. 1A, demonstrating characteristic broad protein staining bands between 45 and 65 kDa. CD47 purified from RBC, spleen, and intestinal epithelium had the same characteristics (data not shown). Additional higher molecular mass bands represent oligomers of CD47 as previously described (39) that was confirmed by reactivity with CD47 mAbs (Fig. 1A).
Microtiter wells were coated with purified CD47 followed by the addition of SIRP␣1-GST. Binding of SIRP␣1-GST was detected by a goat anti-GST antibody. To test the binding specificity of SIRP␣1, parallel experiments were performed using SIRP␤1-GST. As shown in Fig. 1A, SIRP␣1-GST but not SIRP␤1-GST bound to CD47 purified from all different tissues, FIG. 1. SIRP␣1 binds directly to human CD47 purified from different cell types. A, CD47 binds to SIRP␣1 but not SIRP␤1. 96-well microtiter plates were coated with CD47 purified from human RBC, PMN, spleen, and intestinal epithelial cells (HT29) before blocking with 1% BSA and incubation for 1 h with 1 g/ml SIRP␣1-GST or SIRP␤1-GST fusion protein. Binding of SIRP-GST was then detected by incubation with a goat anti-GST antibody followed by peroxidase conjugated rabbit anti-goat secondary and ABTS assay. Control binding experiments were done by coating wells with 1% BSA alone. The inset shows analysis of CD47 immunopurified from human PMN by nonreducing SDS-PAGE where SS denotes the silver-stained protein, and WB denotes the Western blot probed with CD47 mAb PF3.1. Similar profiles were obtained for CD47 purified from human spleen, RBC, and intestinal epithelial cells (HT-29) (data not shown). B, absence of SIRP␣1-GST binding to purified transmembrane Ig superfamily members JAM and CD11b/CD18, isolated as detailed under "Experimental Procedures." C, specific inhibition of SIRP␣1-CD47 binding by antibodies. In these experiments, binding assays of SIRP␣1-GST to PMN CD47 were performed in the presence of HBSS containing 1% BSA and 10 g/ml anti-SIRP mAbs (P3C4, SE5A5, and SE7C2) or anti-CD47 mAbs (C5D5 and B6H12.2) and compared with binding in the presence of isotype-matched IgG. Data in this figure represent one of four experiments performed in triplicate Ϯ S.D. *, p Ͻ 0.01. even though these two SIRP members are highly homologous in their extracellular domains (27). These results confirm those of others (20) and demonstrate that the extracellular domain of SIRP␣1 binds directly to CD47. To confirm the binding specificity, additional experiments were performed. As shown in Fig.  1B, no binding was observed between SIRP␣1-GST and two other transmembrane Ig superfamily members JAM and CD11b/CD18. The specificity of SIRP␣ binding to purified CD47 is further demonstrated in Fig. 1C. In this panel, the effects of some of our anti-SIRP and anti-CD47 mAbs on SIRP␣ binding to CD47 purified from PMN are shown. As can be seen, SIRP mAbs SE5A5 and SE7C2 and CD47 mAb C5D5 markedly inhibited recombinant SIRP␣ binding to CD47. In contrast, SIRP mAb P3C4 had little, if any, inhibitory effect.
We also examined the effects of divalent cations, pH, and different detergents on SIRP-CD47 binding. We found that CD47-SIRP␣1 binding occurred over a wide range of pH (from pH 5 to 10), was independent of Ca 2ϩ /Mg 2ϩ , and was not affected by the nonionic detergent Triton X-100 or 1% Nonidet P-40 (data not shown). However, CD47-SIRP␣1 binding was markedly reduced in the presence of 1% octyl glucoside, 0.5% deoxycholate, or 0.25% SDS (data not shown).
Because human leukocyte adherence to SIRP␣-coated surfaces is mediated through cell surface CD47 (20), we performed experiments to test the direct binding of the IgV loop of CD47 to SIRP␣1 (Fig. 2). A recombinant CD47 IgV loop fused to alkaline phosphatase (CD47-AP) was constructed as detailed under "Experimental Procedures" and expressed in mammalian (CHO) cells to allow for proper folding and glycosylation, which was confirmed by binding assays with multiple functionally blocking anti-CD47 extracellular domain-reactive mAbs (10) (data not shown). As shown in Fig. 2, CD47-AP bound to SIRP␣1-GST fusion protein resulting in a high value of AP activity (A 405 ). Specificity of this interaction was confirmed by the absence of binding of another AP fusion protein containing the extracellular domain of junction adhesion molecule (JAM-AP) (35) and the lack of binding to either SIRP␤1-GST or BSA (Fig. 2B). The effects of our SIRP and CD47 mAbs on binding of CD47-AP to SIRP␣1 were identical to those observed in binding assays with purified native CD47 (data not shown). These results indicate that CD47 binds directly to SIRP␣1 via the extracellular IgV domain of CD47.
As shown in Fig. 3A, the tested anti-SIRP mAbs inhibited PMN transmigration to various degrees. After 1 h of migration, the fraction of applied PMN that had migrated into the lower chamber in the presence of antibodies was 40.1 Ϯ 4.4% for P3C4, 24.9 Ϯ 3.5% for SE5A5, 36.2 Ϯ 4.6% for SE8A3, 34.4 Ϯ 2.4% for SE7C2, 21.6 Ϯ 0.6% for SE12B6, and 14.9 Ϯ 2.3% for SE12C3, compared with 54.3 Ϯ 2.7% of applied PMN migrated in the presence of control mouse IgG. In contrast to the partial inhibition by anti-SIRP antibodies, anti-CD47 mAb C5D5 and anti-CD11b/CD18 mAb CBRM1/29 inhibited PMN transmigration almost completely, which is in agreement with our previous observations (10). Similar inhibitory profiles were obtained with PMN migration across cell-free collagen-coated transwell filters in the presence of anti-SIRP mAbs (data not shown). In summary, the inhibitory effects of this panel of anti-SIRP mAbs ranged from 20 to 60%. Interestingly, mAb P3C4, which failed to inhibit in vitro CD47-SIRP␣1 binding (Table I) and Jurkat cell adhesion to SIRP␣-GST (20), weakly affected PMN transmigration (ϳ10 -20% inhibition).
To gain further insight into the potential roles of CD47 and SIRP␣ interactions in PMN migration, we examined the effects of our CD47 extracellular domain fusion protein CD47-AP on PMN transmigration. Recombinant CD47-AP was used instead of purified native protein because of the obligatory detergent requirements for native CD47. As shown in Fig. 3A, CD47-AP fusion protein inhibited PMN migration by ϳ40 -60%, which is essentially indistinguishable from the inhibition observed with anti-SIRP mAbs. In contrast, no inhibition was observed in the presence of another AP fusion protein JAM-AP. These results suggest that direct interaction between CD47 and SIRP␣ plays an important role in regulating PMN transmigration. We next examined potential signaling pathways that might be involved in SIRP-mediated regulation of PMN migration. Previously, we observed that inhibition of PI3-kinase by wortmannin or LY23009 enhanced PMN transmigration, but only partially reversed the inhibitory effects of CD47 mAbs (10). On the other hand, the tyrosine kinase inhibitor genistein both enhanced PMN transmigration and completely reversed anti-CD47 inhibition (10). Thus, we tested whether wortmannin or genistein could reverse the inhibitory effects of anti-SIRP mAbs (Fig. 3B). In these experiments, PMN were pre-treated with 100 nM wortmannin or 200 M genistein for 10 min (25°C) prior to initiation of transmigration across epithelial-free collagen-coated filters (to avoid potential effects on epithelial cells). To allow for direct comparison of the ability of wortmannin or genistein to reverse the inhibitory effects of anti-SIRP and anti-CD47 mAbs, transmigration in the presence of antibody and drug was normalized with respect to that observed in the presence of drug alone. In this way, we compared the effects of wortmannin and genistein on antibody-mediated inhibition of transmigration to that observed in the absence of any drug pretreatment. As shown in Fig. 3B, in the absence of drug treatment, transmigration at 1 h was inhibited Ͼ90% by anti-CD47 mAb C5D5 and ϳ55% by a mixture of anti-SIRP mAbs (P3C4, SE5A5, and SE7C2, 20 g/ml each). Inhibition of PI3kinase by wortmannin did not decrease the inhibition observed with anti-CD47 in this condition, whereas genistein reversed the antibody-mediated inhibition as we have previously reported (10). In contrast, PI3-kinase inhibition resulted in near complete reversal of the inhibitory effects of anti-SIRP mAbs, whereas genistein had little effect. These results indicate that PI3-kinase, but not genistein-sensitive tyrosine kinase(s), is an important regulatory element in SIRP signaling during PMN chemotaxis.

The Kinetics of PMN Transmigration in the Presence of Anti-SIRP mAbs Are Distinct from Those Observed in the Presence of
Anti-CD47-Although CD47 interacts directly with SIRP␣, our functional studies demonstrated that anti-CD47 mAbs are more effective than anti-SIRP mAbs or CD47-AP in blocking PMN transmigration. This is clearly demonstrated in Fig. 3, where ϳ15-40% of the total applied PMN migrated across epithelial monolayers in the presence of anti-SIRP mAbs, compared with only 5.1 Ϯ 3.4% of PMN in the presence of C5D5. We have recently shown that ligation of CD47 by antibody results in a delay of PMN transmigration but no inhibition of the total amount of migrated PMN (10). Thus, we examined the kinetics of PMN migration in the presence of anti-SIRP mAbs (Fig. 4A). As can be seen, the majority of control PMN migrated across the epithelial monolayers within 1 h. In the presence of anti-CD47 mAbs, PMN transmigration was minimal for 2 h, which is in agreement with our previous observations (10). Surprisingly, the transmigration time course observed in the presence of anti-SIRP mAbs had significantly different kinetics from that observed for anti-CD47, which is best appreciated in Fig.  4B. In the presence of anti-SIRP mAbs or CD47-AP, PMN migration occurred as in controls, but at reduced amounts. Thus, although anti-SIRP mAbs and CD47-AP inhibited PMN transmigration in all time periods, these mAbs did not delay PMN migration across the monolayers as was observed with anti-CD47 mAbs (Fig. 4B). Inhibition of PMN transmigration by SIRP mAbs and soluble CD47 resulted in an overall 30 -60% inhibition compared with the noninhibitory control. Thus, although CD47 binding to SIRP␣ clearly modulates PMN transmigration, the pathways that regulate migration downstream of CD47 and SIRP␣ are distinct.
Expression and Distribution of SIRP in PMN-Using hematopoietic cells, many studies on SIRP have focused on macro- To assay the binding of CD47-AP to SIRP␣1, 96-well microtiter plates were coated with 2 g/ml SIRP␣1-GST followed by blocking with 1% BSA and incubation with CD47-AP (1-2 g/ml) in BSA blocking buffer as detailed under "Experimental Procedures." After washing, CD47 binding was quantified by measurement of alkaline phosphatase activity. Control experiments were performed by using another AP fusion protein, JAM-AP, generated in a similar fashion as CD47-AP (35). Anti-CD47 or anti-SIRP mAbs were included in incubations as listed at concentrations of 10 g/ml. B, CD47-AP fusion protein binds specifically to SIRP␣1. Similar assays were conducted as in A, except that microtiter wells were coated with SIRP␣1-GST, SIRP␤1-GST, or BSA (control) and probed with CD47-AP followed by detection of AP activity. Data in this figure represent one of four experiments with triplicate determinations for each condition Ϯ S.D.
phages. Because of our functional results, we examined SIRP expression in human PMN. As shown in Fig. 5A, immunoblots of detergent-solubilized PMN revealed two broad bands with molecular mass values of 50 -75 and 130 kDa after nonreduced SDS-PAGE using a polyclonal antibody against SIRP␣1 (anti-SIRP␣1 Ex). As shown in lane 3, the broadly stained lower molecular weight band appears to consist of two glycosylated protein species because N-linked deglycosy-5.2% and 72.5 Ϯ 6.1% versus 39.4 Ϯ 2.0% (% of total applied PMN) migrated PMN for wortmannin and genistein versus no drug, respectively). Data represent one of three individual experiments with three or four monolayers in each condition Ϯ S.D. *, p Ͻ 0.05 compared with control migration.

FIG. 3. Anti-SIRP mAbs inhibit PMN transmigration.
A, inverted T84 epithelial cell monolayers were cultured to confluence (high transepithelial resistance) on transwells as described under "Experimental Procedures." Transmigration assays were performed by placing PMN (1 ϫ 10 6 ) into the upper chamber of transwells followed by initiation of transmigration in a basolateral-to-apical direction by adding fMLP (1 M) to the lower chamber. After incubation for 1 h, PMN that had migrated into the lower chamber were quantified by myeloperoxidase assay. In these experiments, migration in the presence of 20 g/ml anti-SIRP mAbs P3C4, SE5A5, SE7C2, SE8A3, SE12B6, and SE12C3 was compared with migration in the presence of isotypematched normal mouse IgG (control IgG), as well as inhibitory mAbs against CD47 (C5D5) and CD11b/CD18 (CBRM1/29). Transmigration assays were also performed in the presence of both CD47-AP (50 g/ml) and the control, JAM-AP. Data represent one of four individual experiments with three monolayers in each condition Ϯ S.D. The bracket designates p Ͻ 0.05 (*) for inhibited migration compared with control migration. As we have previously shown, migration in the presence of binding isotype matched mAbs J10.4 and W6/32 was not different from migration in the presence of normal mouse IgG (data not shown) (11,35). B, effects of PI3-kinase and tyrosine phosphorylation inhibition on PMN migration. PMN (1 ϫ 10 6 ) were pre-treated with 100 nM PI3kinase inhibitor wortmannin, 200 M tyrosine phosphorylation inhibitor genistein, or the same dilution of vehicle (Me 2 SO) for 15 min at room temperature before use in migration assays across acellular collagencoated filters as detailed under "Experimental Procedures." PMN transmigration was performed in the presence of anti-SIRP mAbs (used as a mixture of P3C4, SE5A5, and SE7C2, 20 g/ml each), anti-CD47 mAb C5D5 (20 g/ml), and isotype-matched control IgG (20 or 60 g/ml). For each drug treatment condition (no drug, wortmannin, genistein), migration in the presence of either anti-SIRP or anti-CD47 is shown as a percentage of migration in the presence of isotype matched control IgG (percentage of control). As we have previously reported (35), control migration after wortmannin and genistein treatment was increased compared with migration in the absence of any drug treatment (69.1 Ϯ   FIG. 4. Kinetic analysis of the differential inhibitory effects of anti-SIRP, anti-CD47, and soluble CD47 on PMN transmigration. Time-course PMN transepithelial transmigration assays were initiated as detailed above in Fig. 3A. At 30-min intervals after initiation of migration, transwells were moved into new reservoirs containing the same concentration of fMLP. PMN that had migrated into the lower chamber were then quantified by myeloperoxidase assay as described previously (10). A, time course of PMN migration performed in the presence of control IgG (20 g/ml), anti-CD47 mAb C5D5 (20 g/ml), anti-SIRP mAb (mixture of P3C4 and SE5A5, 10 g/ml each), and 50 g/ml CD47-AP. lation of the immunoprecipitate (lane 3) resulted in three protein bands of 75, 54, and 38 kDa, respectively. Although these results suggest the presence of more than one form of SIRP in human PMN, we were not able to detect SIRP with our antibodies after SDS-PAGE under reducing conditions. Thus, the presence of disulfide-linked homo-or heterooligomers cannot be excluded.
Although a strong SIRP signal was detected by immunoprecipitation and Western blot experiments (Fig. 5A), cell surface fluorescence labeling revealed only modest labeling compared with that observed for CD47 and CD11b (Table I). In particular, cell surface labeling of freshly isolated, nonstimulated PMN with SIRP mAbs resulted in mean fluorescence intensity values ranging from 23 to 35, depending on which mAb was used. In contrast, staining with anti-CD47 mAb C5D5 and anti-CD11b/CD18 mAb CBRM1/29 resulted in mean fluorescence intensity values of 48 (average) and 124, respectively. Because both CD47 and CD11b/CD18 are up-regulated to the cell surface after PMN activation (10), we examined cell surface SIRP expression after stimulation of PMN with fMLP. In these experiments, PMN were stimulated with fMLP (0.1 M), followed by cell surface biotinylation and SIRP immunoprecipitation. As shown in Fig. 5B, cell surface SIRP was significantly increased after PMN were stimulated with fMLP as detected in Western blots by probing with streptavidin. Total SIRP in cell lysates, as detected by the polyclonal anti-SIRP␣1 antibody, was the same before and after fMLP stimulation (Fig. 5B). Analysis of the banding pattern in the biotinylated SIRP blots revealed both an increase in the amount of cell surface protein and a shift in the molecular mass of the lower SIRP␣ band (arrow) from 50 -65 kDa to a slightly higher molecular mass (52-75 kDa) after fMLP stimulation (Fig. 5B, top panel). DISCUSSION In this study, we examined the expression pattern and binding characteristics of PMN SIRPs along with the role of SIRP in modulating PMN transmigration. Although the majority of previous reports have focused on studying SIRP in rodents (21,25,31,40,41), recent studies using cell adhesion assays and recombinant SIRPs have demonstrated CD47-SIRP interactions in human cells (20). In this report, we confirm these observations by demonstrating that human CD47 purified from multiple tissues binds directly to recombinant SIRP␣1 (SIRP␣1-GST) in a specific manner. Our studies clearly demonstrate that the binding to SIRP␣1 is mediated through the IgV loop, defined by a disulfide bond between Cys-41 and Cys-114, in the first portion of the CD47 extracellular domain. Recently, a second long range disulfide bond was reported in CD47 extracellular domain between Cys-263 and Cys-33 (42). Using CD47 mutagenesis, recombinant SIRP␣, and binding assays, Rebres and co-workers (42) concluded that this long range disulfide bond between the first extracellular loop (IgV-containing) and a smaller third putative extracellular loop was important for strengthening CD47-SIRP␣ interactions. However, because our construct CD47-AP lacks this long range disulfide bond and is capable of inhibiting PMN transmigration with kinetics that are indistinguishable from that observed with multiple SIRP mAbs, the functional significance of this second disulfide bond in SIRP-CD47-mediated regulation of PMN migration remains to be determined.
The specificity of the observed CD47-SIRP␣1 interactions was confirmed by the absence of binding to SIRP␤1 and inhibition of binding by anti-SIRP and anti-CD47 mAbs. We demonstrate that anti-SIRP mAbs, previously shown to inhibit Jurkat cell adherence to SIRP␣1-GST (20), inhibited the direct binding of CD47 to SIRP␣1. These findings confirm that SIRP␣1 binds to Jurkat cells through direct interactions with cell surface CD47. We also demonstrate that anti-CD47 mAbs  2) was subjected to N-linked deglycosylation with N-glycanase prior to Western blot by the same antibody. B, increased PMN cell surface SIRP after fMLP stimulation. 2 ϫ 10 7 PMN were allowed to settle in 24-well tissue culture wells containing 1 ml of HBSS followed by fMLP stimulation (0.1 M, 37°C) for 1 h. PMN were then washed with cold HBSS followed by cell surface biotinylation and lysis in 400 l of buffer containing 1% Triton X-100 and protease inhibitors as detailed under "Experimental Procedures." SIRP proteins were then immunoprecipitated from cell lysates containing 200 g of total protein with anti-SIRP mAbs as indicated. Immunoprecipitates were separated by nonreducing SDS-PAGE followed by blotting with streptavidin-peroxidase to detect cell surface SIRP, and polyclonal anti-SIRP␣1 (SIRP␣1 Ex) to detect total cellular SIRP. that inhibit PMN transmigration (10) also inhibit CD47-SIRP␣1 binding, indicating that CD47-SIRP␣ interaction may play an important role in regulating PMN transmigration. In contrast to ligand binding characteristics of ␤ 2 integrins, we found that SIRP-CD47 interactions are independent of Ca 2ϩ and Mg 2ϩ and are stable over a broad pH range.
We have shown that SIRP is involved in the regulation of PMN transmigration by demonstrating inhibition of PMN migration across epithelial monolayers and/or cell-free, collagencoated filters by anti-SIRP mAbs. In particular, anti-SIRP mAbs, which inhibited CD47-expressing Jurkat cells' adherence to recombinant SIRP␣, also inhibited PMN transmigration. Interestingly, anti-SIRP mAb P3C4, which did not block CD47-SIRP␣ binding, also partially inhibited PMN transmigration. We suspect that this may be because of the presence of other SIRP family members in PMN because this mAb, although generated against SIRP␣, recognized both SIRP␣1 and SIRP␤1 in vitro ( Table I).
As shown in Figs. 3 and 4, recombinant, soluble CD47 extracellular domain (CD47-AP) inhibited PMN transmigration by the same amount and kinetics as the anti-SIRP mAbs. These functional effects were observed irrespective of PMN migration across epithelial monolayers or cell-free filters. It is thus likely that CD47-SIRP␣ interactions within PMN mediate the observed effects. Interestingly, CD47, but not SIRP, is expressed on intestinal epithelial cells (11), 2 which raises the possibility of other functional interactions between epithelial CD47 and PMN SIRP␣. In support of this, we have observed that PMN migration is enhanced across CD47-AP-coated filters and CD47-transfected epithelial cells (10). These observations suggest that the functional consequences of CD47-SIRP␣ interactions may be influenced by how the binding partners are displayed.
Our transmigration results suggest that the signaling pathways downstream of SIRP␣ and CD47 are distinct. Recently, we reported that antibody-mediated ligation of PMN cell surface CD47 caused a delay in PMN transmigration without diminishing the total amount of migrated PMN (10). This is in contrast to the kinetics of inhibition by anti-SIRP mAbs and CD47-AP, which did not delay PMN transmigration but inhibited the amount of PMN transmigration by 30 -60%. In further support of these differences, in this report we show that the inhibitory effects by anti-SIRP antibodies were reversed by pre-treating PMN with the PI3-kinase inhibitor wortmannin, but not by the tyrosine phosphorylation inhibitor genistein (Fig. 3B). For CD47, on the other hand, antibody inhibitory effects are reversed by genistein, but not wortmannin (10). Therefore, although our findings suggest that CD47-SIRP␣ interactions occur during PMN migration, there appear to be different signaling pathways involved in CD47 and SIRP functions. Studies are under way to better define the relationship between the above signaling pathways with respect to regulation of PMN transmigration.
The effects of SIRP mAbs on PMN migration prompted us to examine the distribution of SIRP before and after chemoattractant stimulation in PMN. Interestingly, we found that PMN contain at least three proteins of different molecular weights that react with SIRP␣ mAbs after SDS-PAGE under nonreducing conditions. Although it is unclear whether some of the proteins visualized are the result of disulfide linkages, there is evidence for several different SIRP␣ proteins in mouse macrophages and brain (24). Furthermore, our results in Fig. 5 support the possibility of several different SIRP protein species in PMN. As observed, at least one of these proteins (ϳ60 -75 kDa) was strongly redistributed to the PMN cell surface after fMLP stimulation. This pattern of fMLP-induced cell surface up-regulation of PMN proteins is similar to that observed for CD11b/CD18 and CD47 (10). It is thus possible that translocation of this SIRP species to the plasma membrane after fMLP stimulation may play a role in functional interactions with CD47 and cell migration. A higher molecular mass protein of ϳ130 kDa that was also reactive with our SIRP␣ antibodies was also up-regulated to the cell surface (Fig. 5B) but to a lesser extent than that observed with the 60 -75-kDa species. Whether this higher molecular mass protein represents a different SIRP or a disulfide-linked multimer of lower molecular mass species remains to be determined. Further studies are under way to define the precise role of SIRP␣-CD47 interaction(s) during PMN migration as well as the molecular/biochemical characterization of SIRPs in PMN.