SIRPbeta1 is expressed as a disulfide-linked homodimer in leukocytes and positively regulates neutrophil transepithelial migration.

Signal regulatory proteins (SIRPs) comprise a family of cell surface signaling receptors differentially expressed in leukocytes and the central nervous system. Although the extracellular domains of SIRPs are highly similar, classical motifs in the cytoplasmic or transmembrane domains distinguish them as either activating (beta) or inhibitory (alpha) isoforms. We reported previously that human neutrophils (polymorphonuclear leukocytes (PMN)) express multiple SIRP isoforms and that SIRPalpha binding to its ligand CD47 regulates PMN transmigration. Here we further characterized the expression of PMN SIRPs, and we reported that the major SIRPalpha and SIRPbeta isoforms expressed in PMN include Bit/PTPNS-1 and SIRPbeta1, respectively. Furthermore, although SIRPalpha (Bit/PTPNS-1) is expressed as a monomer, we showed that SIRPbeta1 is expressed on the cell surface as a disulfide-linked homodimer with bond formation mediated by Cys-320 in the membrane-proximal Ig loop. Subcellular fractionation studies revealed a major pool of SIRPbeta1 within the plasma membrane fractions of PMN. In contrast, the majority of SIRPalpha (Bit/PTPNS-1) is present in fractions enriched in secondary granules and is translocated to the cell surface after chemoattractant (formylmethionylleucylphenylalanine) stimulation. Functional studies revealed that antibody-mediated ligation of SIRPbeta1 enhanced formylmethionylleucylphenylalanine-driven PMN transepithelial migration. Co-immunoprecipitation experiments to identify associated adaptor proteins revealed a 10-12-kDa protein associated with SIRPbeta1 that was tyrosine-phosphorylated after PMN stimulation and is not DAP10/12 or Fc receptor gamma chain. These results provide new insights into the structure and function of SIRPs in leukocytes and their potential role(s) in fine-tuning responses to inflammatory stimuli.

Signal regulatory proteins (SIRPs) 3 are a family of transmembrane receptor-like signaling proteins that are abundantly expressed in hema-topoietic cells, including granulocytes, monocytes, dendritic cells, and lymphocytes (1)(2)(3). In addition, SIRPs are expressed in neuronal cells (4 -6) and certain types of cancer cells (7)(8)(9)(10). SIRPs can be divided into two subfamilies, SIRP␣ and SIRP␤, based on the putative structures of their C-terminal intracellular domains (11). SIRPs share typical immunoglobulin superfamily structures with an N-terminal extracellular domain containing three cysteine-bound Ig-like loops, a single membrane-spanning transmembrane domain, and a C-terminal intracellular domain (11). The C-terminal intracellular domains of the SIRP␣ subfamily contain a relatively long amino acid sequence (110 amino acids for SIRP␣1) that includes four tyrosine residues to form two immunoreceptor tyrosine-based inhibition motifs (ITIM). Conversely, SIRP␤ subfamily members have a short intracellular domain containing only a few amino acids (4 amino acids for SIRP␤1). Despite a short cytoplasmic tail, SIRP␤1 contains a positively charged lysine in the transmembrane domain that can mediate interactions with an immunoreceptor tyrosine-based activation motif (ITAM),containing adaptor protein. This type of protein structure and adaptor protein interaction has been shown for other immunoreceptors such as NK cell receptors and Fc receptors (12)(13)(14) where the adaptor proteins DAP12 (also termed KARAP (15)), DAP10, and Fc receptor ␥ chain (FcR␥), interact with the parent receptor through ionic interactions within the transmembrane domain to mediate outside-in signaling (16). However, not all SIRP␤ family members share this type of protein structure. SIRP␤ 2 (also termed SIRP␥ (1)) lacks an apparent adaptor-binding element in its transmembrane domain. These different structures of SIRP family members and their associated distinctive signaling elements suggest potential divergent roles of SIRP isoforms in regulating cellular functions.
In a previous study (17), we reported evidence of several SIRP proteins that are expressed in human PMN. In addition, we demonstrated that SIRP␣ regulates PMN transmigration across epithelial monolayers through interaction with another Ig superfamily cell surface protein CD47. However, given the highly homologous extracellular domain structures of SIRP family members, it was difficult to rigorously define the identities and functional roles of other SIRP proteins in PMN. In the present study, we utilized SIRP␣ and -␤ isoform-specific antibodies and RT-PCR to identify the major isoforms of SIRP␣ and -␤ in human PMN. From these studies, we discovered important structural differences between SIRP␣ and -␤. In particular, we report for the first time that SIRP␤1 is expressed not as a monomer but as a disulfide-linked homodimer. Furthermore, our results suggest that, in contrast to SIRP␣, antibody-mediated ligation of SIRP␤1 enhances PMN transmigration.

MATERIALS AND METHODS
Antibodies-Rabbit anti-SIRP␣1.ex antibody that is reactive with multiple SIRP protein species in leukocytes was kindly provided by Dr. Axel Ullrich (11) and was used in immunoblotting as described previously (17). A murine polyclonal antibody specific to SIRP␣ was generated by immunizing mice with a fusion protein consisting of the extracellular domain of human SIRP␣1 (Ig loops 1 ϩ 2 ϩ 3) fused to rabbit Fc (SIRP␣1.ex-Fc) (18). Monoclonal antibodies B4B6 and B1D5 that specifically bind to SIRP␤ were generated as described previously (19). Anti-SIRP␣ mAbs SE5A5 and SE7C2 were generated and used as described previously (17,20). Anti-DAP12 mAb DX37 (21) was kindly provided by Dr. Lewis Lanier (University of California, San Francisco). In addition, we produced murine polyclonal anti-DAP12 and DAP10 antibodies by immunizing BALB/c mice with a glutathione S-transferase fusion protein containing the putative intracellular domain of DAP12 and DAP10. Rabbit polyclonal antibody against FcR␥ was obtained from Upstate Biotechnology, Inc. As a noninhibitory antibody binding control for PMN transmigration assays, we used anti-JAM-A mAb J10.4 (22). As an inhibitory control for PMN transmigration assays, we used anti-CD47 mAb C5D5 as described previously (23).
PMN, Peripheral Blood Mononuclear Cells (PBMC), and HL60 Cells-To isolate PMN and PBMC, fresh blood from healthy donors and anticoagulated with 0.38% sodium citrate was centrifuged for 10 min (1000 rpm) at room temperature, and the upper layer of platelet-rich plasma was removed. The lower layer of cells was subjected to dextran (Amersham Biosciences) sedimentation to separate leukocytes from red blood cells. The leukocyte-containing fraction was further separated by Ficoll-Paque (Amersham Biosciences) sedimentation followed by separate collection of PMN-and PBMC-enriched fractions. After washing with cold HBSS(Ϫ), leukocytes were resuspended in HBSS(Ϫ) and used within 4 h. PMN prepared in this way typically represented Ͼ90 -95% of cells in such isolates. HL60 cells were obtained from American Tissue Culture Collection (ATCC) and maintained in RPMI medium containing 20% fetal bovine serum.
Immunoprecipitation and Immunoblotting Experiments-Cells were lysed with buffer containing 100 mM Tris (pH 7.5), 150 mM NaCl, 2 mM CaCl 2 , 2 mM MgCl 2 , 1% Triton X-100 or octyl glucoside, 1:100 dilution of proteinase inhibitor mixture (Sigma), and 1 mM phenylmethylsulfonyl fluoride. Cell lysates (200 -400 g of total protein) were pre-cleared with 2-5 g of control IgG-conjugated Sepharose before incubation with 1-2 g of specific mAb and protein A-Sepharose (Sigma) for 4 h (4°C). Washed immunoprecipitates were boiled in SDS-PAGE sample buffer with (reducing condition) or without ␤-mercaptoethanol (nonreducing condition) and subjected to SDS-PAGE followed by transfer to nitrocellulose under standard conditions. Nonspecific binding was blocked by 5% nonfat dry milk in TBST (2 mM Tris, 50 mM NaCl, 0.5% Tween 20 (pH 7.4)) followed by immunoblotting with the specific antibody. To detect SIRP␤1 in leukocytes, immunoprecipitation was performed using mAb B4B6 or B1D5 followed by immunoblotting with the same mAb or with rabbit anti-SIRP␣1.ex. To confirm SIRP␤1 existing as a disulfide-bonded dimer, total cell lysates or SIRP␤1 immunoprecipitates were treated with 30 -300 mM iodoacetamide for 30 min at 25°C before SDS-PAGE under nonreducing conditions and Western blot. To biotinylate SIRP␤1 in PMN, freshly isolated PMN (10 7 ) were incubated with N-hydroxysuccinimide-biotin (Pierce) at a final concentration of 1 mg/ml in HBSS for 1 h on ice. Cells were then washed three times with HBSS followed by quenching with 20 mM Tris (pH 7.5) and 100 mM NH 4 Cl for 1 h before cell lysis. After immunoprecipitation of SIRP␤1, Western blots of biotin-labeled proteins were probed with streptavidin-peroxidase followed by enhanced chemiluminescence (Amersham Biosciences). SIRP␣ protein was immunoprecipitated with a murine polyclonal anti-SIRP␣ antibody. In a subset of experiments, SIRP␣ was co-precipitated using CD47-AP fusion protein as described below. To chemically cross-link proteins in PMN, unstimulated and fMLP-stimulated PMN (2 ϫ 10 7 ) were incubated with 1 mM dithiobis[succinimidylpropionate] or dithiobis[sulfosuccinimidylpropionate] (Pierce) in HBSS for 1 h. The reaction was terminated by washing and quenching with 0.2 M Tris (pH 7.5) and 0.1 M NH 4 Cl for 30 min. SIRP␤1, DAP12, and CD11b were detected by immunoblotting of the cell lysates using mAb B4B6, mouse anti-DAP12 antibody, and rabbit anti-CD11b antibody R7928A (23).
SIRP␣ Co-precipitation Experiments-A recombinant fusion protein consisting of the putative extracellular domain of CD47 and alkaline phosphatase (CD47-AP) was generated as described previously (17). A fusion protein containing the extracellular domain of human JAM-A (JAM-AP) was used as a control. Both CD47-AP and JAM-AP (2 g of each) were incubated with PMN cell lysates for 2 h at 4°C followed by further incubation with 20 l of anti-AP conjugated agarose (Sigma) for 2 h. The agarose bead-protein complexes were washed three times followed by SDS-PAGE under nonreducing conditions and blot with antibodies against SIRP␣ and SIRP␤.
Site-directed Mutagenesis of Cysteine Residues in SIRP␤1-Fulllength, wild-type SIRP␤1 was cloned into pCDNA3.1 and used as a template for mutagenesis. Cys-320 was mutated to serine and alanine using forward primers 5Ј-agctggctcctggtgaacacctctgcccacagg and 5Ј-agctggctcctggtgaacaccgctgcccacagg, respectively. The reverse primer for constructing both C320S and C320A was 5Ј-aggtgttcaccaggagccagctcatccagt. Cys-393 was mutated to valine using forward primer 5Ј-ggtgtctctgccatctacatcgtctggaaacag and reverse primer 5Ј-gatgtagatggcagagacaccaaccaccag. Mutations of Cys-320 and Cys-393 were performed using the GeneTailor site-directed mutagenesis system (Invitrogen). After confirmation by DNA sequencing, at least two clones of each mutant were selected and transiently transfected into COS-7 cells using the DEAE-dextran method. COS cells were then lysed 48 -72 h after transfection, and SIRP␤1 expression was analyzed by SDS-PAGE and immunoblotting. In addition, expression of SIRP␤1 was also assessed by immunofluorescence staining and flow cytometry analysis (fluorescence-activated cell sorter).
Subcellular Fractionation Experiments-Stimulation of PMN was performed in Petri dishes using 1 M fMLP in HBSS and incubation at 37°C (30 min). Unstimulated and stimulated PMN (2 ϫ 10 8 cells per condition) were then resuspended in 5 ml of cavitation buffer (0.34 M sucrose, 10 mM HEPES, 1 mM EDTA, 0.1 mM MgCl 2 , 1 mM Na 2 ATP (pH 7.4)) and disrupted by nitrogen cavitation (15 min, 400 p.s.i., 4°C). The lysates were centrifuged at low speed (1000 rpm), and the supernatants were subjected to isopycnic sucrose density gradient fractionation on linear 20 -55% sucrose gradients in a Beckman SW-28 swinging bucket rotor (100,000 ϫ g, 3 h, 4°C) as described previously (23). From each gradient, 1.5-ml fractions were collected and were analyzed for sucrose density and protein. The subcellular localization of plasma membrane and primary and secondary granules was determined by assays of alkaline phosphatase, myeloperoxidase, and lactoferrin, respectively, as described previously (23).
PMN Transepithelial Migration Assay-PMN transepithelial migration in the physiologically relevant basolateral to apical direction was performed using isolated PMN and intestinal epithelial monolayers exactly as described previously (23,24). Briefly, confluent inverted T84 monolayers were washed twice with HBSS (20°C), and PMN (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 adding 1 ml of 1 M fMLP (in HBSS) to the lower chamber followed by incubation at 37°C. PMN migration across epithelial monolayers into the fMLP-containing lower chambers was quantified by myeloperoxidase assay (23,24).

Identification of SIRP Proteins in Leukocytes-
In previous experiments, we used a polyclonal rabbit antibody against the SIRP␣1 extracellular domain, anti-SIRP␣1.ex (11), to study SIRP␣ expression in PMN (17). We observed that this antibody labels multiple protein bands in Western blots from detergent-solubilized PMN (17). As shown in Fig.  1A, immunoblots of PMN using anti-SIRP␣1.ex revealed broad bands with molecular mass values of 110 -120, 65-75, and/or 52-58 kDa (Fig.  1A, arrows) after nonreduced SDS-PAGE. Similar immunoblotting patterns of SIRP proteins were also obtained from detergent-solubilized PBMC and PMN-like Me 2 SO-induced HL60 cells by anti-SIRP␣1.ex under nonreducing conditions. Because of the high degree of homology between the extracellular domains of SIRP family isoforms, we hypothesized that anti-SIRP␣1.ex is likely cross-reactive with several SIRP isoforms.
To better characterize the isoforms of SIRPs expressed in PMN and to study their functional roles PMN, we performed experiments using SIRP␤ isoform-specific antibodies. Monoclonal antibodies B4B6 and B1D5 were generated against the SIRP␤1 extracellular domain (19), and binding was confirmed to be noncross-reactive with SIRP␣ by enzymelinked immunosorbent assay (data not shown) and cell binding assays (19). These specific mAbs were used to immunoprecipitate and/or immunoblot SIRP␤ from detergent-solubilized PMN and other leukocytes. As shown in Fig. 1B, under nonreducing conditions, mAbs B4B6 and B1D5 exclusively immunoprecipitated and immunoblotted a protein band of 110 -120 kDa, which correlates with the highest molecular weight SIRP protein species shown in Fig. 1A. These results were surprising given that SIRP␤ has a predicted core molecular mass of ϳ45 kDa (398 amino acids for SIRP␤1). Because these results were obtained after nonreducing SDS-PAGE, we hypothesized that SIRP␤ protein may exist as a disulfide-linked homodimer or hetero-oligomer. Indeed, as shown in Fig. 1C, after reducing SDS-PAGE, the anti-SIRP␤1 mAbreactive band decreased to 55 kDa. To exclude hetero-oligomerization of SIRP␤ with other protein(s) through disulfide bonding, PMN were biotinylated with cell-permeable N-hydroxysuccinimide-biotin followed by immunoprecipitation and nonreduced SDS-PAGE. In these experiments, we excised the 110 -120-kDa SIRP␤ protein band from the nonreduced acrylamide gels and performed a second electrophoresis under reducing conditions. Western blots of the second running were then performed using peroxidase-conjugated streptavidin. No protein band other than the characteristic 55-kDa SIRP␤ protein band was detected (results not shown). Finally, to rule out the possibility that the dimerization observed is a gel artifact due to oxidation of unpaired cysteine residues during electrophoresis, we confirmed that iodoacetamide treatment of PMN had no effect on the electrophoretic mobility of SIRP␤ (data not shown). These results thus support the notion that SIRP␤ likely exists as disulfide-bonded homodimer in leukocytes.
To confirm further that this 65-75-kDa protein is indeed SIRP␣, we performed co-precipitation assays with a SIRP␣-binding CD47 extracellular domain fusion protein (CD47-AP), which was previously confirmed to bind specifically to SIRP␣ but not to SIRP␤1 (17). As shown in Fig. 1E, compared with the control, co-precipitation using another abundantly expressed PMN immunoglobulin superfamily member JAM-A (JAM-AP) (17), CD47-AP specifically precipitated a protein of 65-75 kDa that was labeled by anti-SIRP␣. This result is consistent with the immunoblotting results (Fig. 1D) and suggest that the protein of 65-75 kDa is SIRP␣. In contrast to SIRP␤, the apparent molecular mass of SIRP␣ is not significantly decreased after reduction and remains at 65-75 kDa (Fig. 1D), indicating that SIRP␣ most likely exists as a monomer with a molecular mass of 65-75 kDa in human leukocytes.
The Predominant SIRP␣ and ␤ Subfamily Isoforms in PMN Are Bit/ PTPNS-1 and SIRP␤1-Because previous studies suggest that there are multiple isoforms of SIRP␣ (SIRP␣1 and -␣2, Bit (25)/PTPNS-1, and MFR (26)) and SIRP␤ (SIRP␤1, -␤2), we performed RT-PCR to define the specific SIRP␣ and -␤ sequences from human PMN. Total RNA samples were isolated from transmigrated PMN (see "Materials and Methods") to minimize contamination of other leukocytes. Multiple oligonucleotide primers that correspond to multiple conserved and variable regions of SIRP␣ isoforms were synthesized according to the sequences of SIRP␣1 (NCBI accession number Y10375), SIRP␣2, Bit (accession number AB023430)/PTPNS-1 (accession number AL117335), and were used in PCRs with different primer annealing temperatures. DNA sequencing revealed one predominantly amplified sequence that matches Bit/PTPNS-1. As shown in Fig. 2, two pairs of primers corresponding to the variable regions of SIRP␣1 and Bit/PTPNS-1 extracellular domains were synthesized and used in PCRs with gradient annealing temperatures and different amplification cycles. As can be seen, higher annealing temperatures (Ն60°C) eliminated potential primer cross-annealing and produced predominant amplification of Bit/PT-PNS-1. A similar strategy was designed to define the specific SIRP␤ isoform and revealed that all differentially amplified DNA fragments matched SIRP␤1.
Cysteine 320 Mediates Homodimer Formation in SIRP␤1-Given that SIRP␣ (Bit/PTPNS-1) appeared to exist as a monomer in PMN, our data suggesting that SIRP␤1 exists as a disulfide-bonded homodimer were intriguing because the primary structure of the extracellular and transmembrane domains of SIRP␤1 are so similar to those of Bit/PT-PNS-1 or SIRP␣1. As shown in Fig. 3A, alignment of SIRP␤1 and Bit/ PTPNS-1 revealed three pairs of cysteine residues that most likely bridge the putative Ig loops given their conserved nature. Two additional cysteine residues, Cys-320 and Cys-393, are unique to SIRP␤1 and are not present in Bit/PTPNS-1 (Fig. 3A). Cys-320 is within the membrane-proximal extracellular Ig loop, whereas Cys-393 is within FIGURE 1. Expression of SIRP␣ and -␤ isoforms in human leukocytes. A, expression of SIRP proteins in human PMN, PBMC, and HL60 cells. The human leukocytes (5 ϫ 10 7 /each) were lysed and subjected to SDS-PAGE under nonreducing conditions followed by immunoblotting with a rabbit polyclonal anti-SIRP antibody (anti-SIRP␣1.ex (11)). Note that although the antigen for this antibody was the extracellular domain of SIRP␣1, it labels multiple SIRP isoforms in PMN and other leukocytes as highlighted by arrows. B, identification of SIRP␤1 in PMN. Left panel, immunoprecipitations (IP) were performed using either normal mouse IgG (control, Ctl) or SIRP␤1-specific mAbs B4B6 and B1D5 from PMN cell lysates. The immunoprecipitates were then subjected to nonreducing SDS-PAGE and Western blot (WB) using either anti-SIRP␣1.ex as in A or a murine polyclonal SIRP␣-specific antibody (anti-SIRP␣). Right panel, the whole cell lysates of PMN, PBMC, and HL60 cells were directly Western blotted by anti-SIRP␤1 mAb B4B6 under nonreducing conditions. C, SIRP␤1 is a dimer. PMN cell lysates were subjected to SDS-PAGE under nonreducing and reducing conditions followed by immunoblotting with anti-SIRP␤1 mAbs B4B6 and B1D5. In the far right panel, PMN lysates were deglycosylated (degly.) with N-glycosidase (New England Biolabs) followed by reducing SDS-PAGE and immunoblot with anti-SIRP␤1. D, identification of SIRP␣ in leukocytes. A SIRP␣-specific antibody, anti-SIRP␣, was generated by immunizing mice with the extracellular domain of SIRP␣1 ("Materials and Methods"). As shown, the antibody labels SIRP␣ protein of 65-75 kDa from PMN, PBMC, and HL60 cells. E, precipitation of SIRP␣ using recombinant CD47. PMN lysates were incubated with the CD47 extracellular domain fusion protein (CD47-AP) that binds to SIRP␣ but not SIRP␤1 (17). As a control, PMN lysates were incubated in parallel with another AP fusion protein JAM-AP. After precipitation by antialkaline phosphatase-conjugated agarose (Sigma), the protein complexes were subjected to SDS-PAGE under nonreducing conditions, followed by Western blot using anti-SIRP␣ antibody. Total PMN lysate was also loaded on the right as a control. Western blots of the precipitates were also reprobed with anti-SIRP␤1 mAb B4B6, and anti-SIRP␣1.ex and failed to detect any SIRP␤1 (results not shown).
the transmembrane segment. To test if these cysteine residues are involved in intermolecular disulfide bond formation, we performed sitedirected mutagenesis and swapped these cysteine residues with the corresponding residues in Bit/PTPNS-1. Specifically, we changed Cys-320 to serine and Cys-393 to valine followed by transient transfection of these mutants into COS cells.
As shown in Fig. 3B, no SIRP␤1 expression was detected in mocktransfected COS cells. Conversely, a prominent SIRP␤1 band of 110 -120 kDa under nonreducing conditions was detected in Western blots of cells transfected with a full-length DNA sequence encoding the wildtype SIRP␤1. Similar to that observed in leukocytes, SDS-PAGE under reducing conditions caused a decrease in the apparent molecular mass of the SIRP␤1 band to 55 kDa (Fig. 3B), consistent with SIRP␤1 forming a disulfide-linked homodimer in COS cells. As shown in Fig. 3B, mutation of Cys-393 to valine (C393V) had no effect on SIRP␤1 dimerization. In contrast, mutation of Cys-320 to serine (C320S) resulted in loss of SIRP␤1 dimerization. In particular, the immunoblot of the C320S transfectant did not reveal an apparent 110 -120-kDa SIRP␤1 dimer band under nonreducing conditions (Fig. 3B); however, the 55-kDa SIRP␤1 monomer was detected as with wild-type and C393V transfectants in Western blots under reducing conditions (Fig. 3B). These results implicate cysteine 320 as the key residue that mediates SIRP␤1 inter-molecular disulfide bond formation and thus protein dimerization. To confirm the role of Cys-320 and rule out individual amino acid effects, we also mutated Cys-320 to alanine. As shown in Fig. 3, no SIRP␤1 dimer was detected in the C320A transfectant, confirming that Cys-320 is the  essential residue for SIRP␤1 dimerization. Interestingly, we have consistently observed that the expression levels of C320S/C320A mutants were lower than those of wild-type and C393V SIRP␤1 (Fig. 3, B and C), suggesting that dimerization of the protein may contribute to stabilize the SIRP␤1 structure.
Subcellular Localization of SIRP␣ and -␤1 in PMN-In our previous study, we observed that SIRP␣ expression on the PMN cell surface is significantly up-regulated after chemoattractant stimulation (17), suggesting that intracellular pools might be exist for SIRP␣ in PMN. With our current results indicating that under nonreducing conditions the high molecular weight protein species is SIRP␤1, our previous observations also suggested that the cell surface expression of SIRP␤1 on PMN may not be increased by activation. To further characterize the subcellular localization of SIRP proteins in PMN, we performed subcellular fractionation experiments using isopycnic sucrose density gradients ("Materials and Methods").

Anti-SIRP␤1 mAbs Enhance PMN Transepithelial Migration-Be-
cause we demonstrated previously a role for SIRP␣ in regulating PMN transmigration (17), we performed experiments to test SIRP␤1-specific mAbs for inhibitory or stimulatory effects on fMLP-driven PMN transepithelial migration. In these experiments, T84 colonic epithelial cells were grown as inverted monolayers on permeable transwell filters (23). PMN were added to the upper chambers of transwells in the presence or absence of mAbs and were induced to migrate in a basolateral to apical direction toward the chemoattractant fMLP (23). As shown in Fig. 5, treatment with anti-SIRP␤1 mAb B4B6 resulted in enhanced PMN transepithelial migration after 1 h compared with migration in the absence of antibody or in the presence of the binding, noninhibitory antibody J10.4 (22) (52.2 Ϯ 4.2% migration for anti-SIRP␤1 versus FIGURE 5. Antibody-mediated ligation of cell surface SIRP␤1 enhances PMN transepithelial migration. Time course PMN transepithelial migration assays were performed using inverted T84 epithelial monolayers (23). In these experiments, migration in the presence of 20 mg/l of anti-SIRP␤1 mAb B4B6 was compared with migration in the presence of the same concentrations of control murine IgG (control IgG), a noninhibitory binding mAb J10.4, and an inhibitory anti-CD47 mAb C5D5 (23). Results obtained at different time points represent the sum of PMN that had migrated up to that time point.

FIGURE 4. Subcellular localization of SIRPs in PMN.
Unstimulated and fMLP-stimulated PMN were subjected to isopycnic sucrose density gradient centrifugation to separate subcellular organelles. Localization of plasma membrane and primary and secondary granules was determined by assays for alkaline phosphatase, myeloperoxidase, and lactoferrin, respectively. As shown in the figure, fractions 1-10 represent cytosol. Plasma membrane was distributed in fractions 11-17, secondary granules in fractions 17-20, and primary granules in fractions 20 -22. To localize SIRP isoforms, gradient fractions were immunoblotted using anti-SIRP␣ and anti-SIRP␤1 (mAb B4B6) antibodies. For reference, fractions were also immunoblotted for actin using anti-F-actin antibody (BD Transduction Laboratories). 31.8 Ϯ 4.5% and 38.2 Ϯ 5.6% for no antibody and mAb J10.4, respectively). Similar results were obtained using another specific anti-SIRP␤1 mAb B1D5 (data not shown). The enhanced PMN transmigration observed after incubation with anti-SIRP␤1 antibodies is in contrast to the inhibitory effects observed with anti-SIRP␣ mAbs or after incubation with a soluble form of the SIRP␣ ligand CD47-AP (17). Thus, these results suggest that regulation of PMN function by SIRP␤1 opposes that of SIRP␣, at least with regard to PMN migration.
Activation of PMN Results in Tyrosine Phosphorylation of a SIRP␤1associated Protein That Is Not DAP12-Previously, SIRP␤1 was reported to physically associate with the ITAM-containing adaptor protein DAP12 in other cell types (27,28). Thus, we investigated the possibility of SIRP␤1 binding to DAP12 in PMN as a mechanism for signal transduction and regulation of transmigration. As shown in the Western blots of PMN lysates in Fig. 6A, DAP12 is easily detected as a 24-kDa protein under nonreducing conditions, and the apparent molecular mass decreased to 12 kDa under reducing conditions, which is consistent with previous reports of DAP12 existing as a disulfide-linked homodimer (16). To determine whether SIRP␤1 associates with DAP12 in PMN, we performed co-immunoprecipitation assays. To our surprise, no DAP12 was detected in immunoprecipitates using anti-SIRP␤1 antibody from either unstimulated or fMLP-stimulated PMN. As shown in Fig. 6B, under nonreducing conditions, mAb B4B6 immunoprecipitated SIRP␤1 from both unstimulated and fMLP-stimulated PMN. Probing Western blots of the same immunoprecipitates using anti-DAP12 antibodies DX37 and a murine polyclonal anti-DAP12 antibody failed to detect any DAP12 band (Fig. 6B). However, the postbind cell lysates from such immunoprecipitations contained abundant DAP12 (not shown). To investigate further whether SIRP␤1 associates with DAP12 in PMN, we performed protein cross-linking experiments. We surmised that if DAP12 associates with SIRP␤1, its apparent molecular mass would increase after protein cross-linking. However, neither treatment of PMN with cell membrane-permeable (dithiobis[succinimidylpropionate]) nor cell impermeable (dithiobis[sulfosuccinimidylpropionate]) cross-linkers detected SIRP␤1 and DAP12 association under conditions where controls were successfully cross-linked (results not shown). Although our results did not support SIRP␤1 binding to DAP12 in PMN, probing Western blots of immunoprecipitated SIRP␤1 with the phosphotyrosine-specific antibody PY20 revealed a small protein of ϳ10 -12kDa that demonstrated increased phosphorylation after fMLP stimulation (Fig. 6B). Further Western blots using anti-DAP10 and FcR␥ antibodies indicated that the SIRP␤1-associated phosphorylated protein is neither DAP10 nor FcR␥. This result suggests that another, as yet undefined, adaptor protein associates with SIRP␤1 in PMN that is tyrosine-phosphorylated after chemoattractant stimulation and may mediate downstream signaling from SIRP␤1.

DISCUSSION
The SIRP family proteins are separated into two major groups or isoforms referred to as SIRP␣ and SIRP␤ that are characterized by the presence or absence of a long C-terminal intracellular domain, respectively (11). Despite the striking difference in their C-terminal domains, the extracellular domains of SIRP␣ and -␤ members share highly homologous primary structures that are predicted to form three Ig-like loops (11). However, despite highly similar extracellular domains, ligand-based interactions of the SIRP family proteins appear to be remarkably different. SIRP␣ has been clearly shown to be an extracellular ligand for CD47 (4,17,20), another cell surface Ig superfamily member that is expressed on nearly all cells and tissues (29). We and others have defined previously that CD47 exclusively binds to the membrane distal immunoglobulin variable domain loop of SIRP␣ (19,30) and that this binding interaction can regulate a multitude of cellular processes, including PMN and other cell migration (17,31), macrophage multinucleation (26,32), self-recognition of red blood cells (33), T cell and dendritic cell functions (34,35), and many others. Within the SIRP␤ subfamily, only SIRP␤2 (or SIRP␥) has been reported to bind to CD47 but with a much lower affinity than SIRP␣ (1,36). Whereas the physiological significance of SIRP␤2-mediated binding to CD47 is not known, recent reports suggest that CD47-SIRP␤2 interactions may play a role in regulating T cell proliferation (36) and apoptosis (1). SIRP␤1, on the other hand, does not bind to CD47 as demonstrated by in vitro CD47 binding assays using SIRP␤1 extracellular domain fusion proteins or SIRP␤1-transfected cells (17,19). In this work, we also failed to detect CD47 binding to SIRP␤1 by co-precipitation assays (Fig. 1E). Thus, it remains unknown what the extracellular ligand for SIRP␤1 is and how SIRP␤1 might regulate cellular function.
The highly similar extracellular domains of SIRP family proteins have created significant obstacles in determining the specific role(s) of indi- FIGURE 6. A, analysis of DAP12 expression in human leukocytes by Western blots (WB) under nonreducing and reducing conditions. B, SIRP␤1 does not associate with DAP12 in PMN. Unstimulated or fMLP-stimulated PMN were lysed in buffer containing 1% digitonin, protease, and phosphatase inhibitors. SIRP␤1 was immunoprecipitated (IP) by mAb B4B6 followed by SDS-PAGE under nonreducing conditions on 10 and 15% gels in order to detect SIRP␤1 and DAP12, respectively. Western blots were then probed with anti-SIRP␤1, anti-DAP12, and anti-phosphotyrosine antibody PY20 (Zymed Laboratories Inc.)). A control immunoprecipitation was performed using normal mouse IgG (IgG ctl.). vidual isoforms in regulating cellular function(s). We have observed that mAbs generated against the SIRP␣ extracellular domain cross-react with SIRP␤ isoforms under different conditions, which has complicated interpretation of results. 4 Given that SIRP␣ and -␤ are most often coexpressed in leukocytes, studies using cross-reactive anti-SIRP␣ antibodies have confounded the analysis of SIRP␣ function because the contribution of SIRP␤ isoforms could not be excluded. We observed previously that a rabbit polyclonal antibody against the extracellular domain of SIRP␣1 (termed anti-SIRP␣1.ex (11)) reacted with two or three SIRP isoforms in PMN (17) (also shown in Fig. 1). Thus, it was uncertain which protein band truly represented SIRP␣. These observations, coupled with the increasing appreciation of the importance of this family of proteins in immunology, prompted us to further define the specific SIRP isoforms in PMN and their functions.
In this study, we report for the first time, a clear distinction of SIRP␣ from SIRP␤ in PMN and other leukocytes. We report that SIRP␤1 exists as 110 -120-kDa disulfide-linked homodimer in leukocytes including PMN, PBMC, and HL60 cells. Our data suggest that SIRP␣, in its functional form, represents a smaller protein of 65-75 kDa due to differences in tertiary structure. In addition, from our results in Fig. 2, we conclude that the major SIRP␣ subfamily member in PMN includes Bit/PTPNS-1. We also report that the predominant SIRP␤ in PMN is SIRP␤1. We show by site-directed mutagenesis that Cys-320, which is in the membrane-proximal Ig loop of SIRP␤1, mediates intermolecular disulfide bond formation in SIRP␤1 dimers. Our finding of the dimeric structure of SIRP␤1 has implications for interpretation of previous studies because some of these reports included experiments with monovalent fusion protein constructs to assay potential SIRP␤1 binding interactions (19). These results provide new rationale for future experimental design aimed at characterizing the role of SIRP␤1 in cellular functions.
Not only are the protein structures of SIRP␤ and SIRP␣ (Bit/PT-PNS-1) different, but we observed distinct cellular distributions of each protein in PMN. As highlighted in Fig. 4, SIRP␣ appears to localize in intracellular storage pools and is redistributed to the cell surface after chemoattractant stimulation. Such cell surface up-regulation after stimulation is similar to that reported for other proteins important in the regulation of PMN transmigration such as CD47 and CD11b/CD18 (23). We speculate that, like CD47 and CD11b/CD18, SIRP␣ may also regulate transmigration through adhesion-based signaling events. In contrast, we did not observe major redistribution of SIRP␤1 between subcellular organelles and the plasma membrane after stimulation. On the contrary, we observed that SIRP␤1 is constitutively present in the plasma membrane before and after fMLP stimulation. Most interestingly, as seen in Fig. 4, we also observed subtle changes in the profile of SIRP␤1 co-sedimention with plasma membrane after fMLP stimulation. Although the significance of this observation is unclear, these changes could be seen with activation-dependent redistribution of SIRP␤1 in membrane microdomains that might play a role in regulating SIRP function. Together, our findings of different distributions of SIRP␣ and -␤1 in PMN support distinct functional properties. This is supported by our results demonstrating that ligation of SIRP␣ inhibited PMN transepithelial migration (17), whereas ligation of SIRP␤1 in this study with specific mAbs resulted in enhanced PMN transmigration (Fig. 5).
Such opposite effects of SIRP␣ and -␤1 on PMN transmigration, however, might be expected given that the C-terminal structures would be predicted to have negative and positive regulatory roles. SIRP␣ has two inhibitory signaling motifs (ITIM) in the intracellular domain, whereas SIRP␤1 contains a positive charged Lys in the transmembrane domain that mediates association with an activation motif (ITAM)containing adaptor protein. These types of divergent regulatory pathways have been observed in natural killer cell receptors (KIRs and KARs) (14), Fc receptors (FcR) (14), novel immune-type receptors (37), and Ig-like transcripts (38,39). To date, these ITIM and ITAM domainmediated signaling pathways have been recognized to be characteristic of crucial cell surface receptors involved in regulating innate immune functions such as NK cell cytotoxicity and macrophage phagocytosis (14,33). Among such cell surface receptors, inhibitory elements contain ITIM structures in the C-terminal intracellular domain, and the activation isoforms associate with ITAM-containing adaptor proteins such as DAP12, CD3, or FcR␥.
Because previous studies suggested that SIRP␤1 associates with DAP12, we investigated whether SIRP␤1 regulates PMN function through DAP12-mediated signaling events. DAP12 is a 12-kDa homodimeric transmembrane protein that signals through tyrosine phosphorylation events in the intracellular ITAM domain. This adaptor protein was first described on the plasma membrane of natural killer cells and was associated with killer cell activatory receptors (15). DAP12 was later reported to be expressed in other leukocytes, including peripheral mononuclear cells and dendritic cells (40). In this study, we demonstrated expression of DAP12 homodimers in PMN (Fig. 6A). To assess potential SIRP␤1 and DAP12 binding interactions, we tried a variety of different detergents, buffers, and immunoprecipitation conditions (e.g. 1% digitonin (41), Brij/Nonidet P-40 lysis buffer (42)). However, none of our co-immunoprecipitation or cross-linking experiments revealed SIRP␤1 association with DAP12 in PMN. Furthermore, we also examined FcR␥ as a potential adaptor protein for SIRP␤1. Curiously, we observed that SIRP␤1 from PBMC but not PMN co-immunoprecipitated with FcR␥. 3 Although SIRP␤1 was not shown to associate with DAP12 in PMN, our data are consistent with the association of SIRP␤1 with an undefined 10-kDa adaptor protein. As shown in Fig. 6, immunoprecipitates of SIRP␤1 contained an ϳ10-kDa protein that was tyrosine-phosphorylated after chemoattractant stimulation. In parallel Western blotting experiments, we excluded the possibility that this protein is FcR␥ or DAP10 (results not shown). Thus, the identity of this associated protein and whether it mediates signaling events downstream of SIRP␤1 in regulating PMN transmigration remain unclear. Because PMN transmigration plays a central role in innate immunity and is vital for defense against invading pathogens, our findings implicating SIRP␣ and -␤1 as inhibitory and stimulatory signaling elements in the regulation of PMN migration provide additional insight into mechanisms that serve to finetune the innate immune response. Additional studies in this area will both advance our understanding of PMN transmigration and may provide ideas for new anti-inflammatory therapeutics.