The Fourth Blade within the β-Propeller Is Involved Specifically in C3bi Recognition by Integrin αMβ2*

Interactions between the complement degradation product C3bi and leukocyte integrin αMβ2 are critical to phagocytosis of opsonized particles in host defense against foreign pathogens and certain malignant cells. Previous studies have mapped critical residues for C3bi binding to the I-domains of the αM and the β2 subunits. However, the role of the αM β-propeller in ligand binding remains less well defined, and the functional residues are still unknown. In the present study, we studied the function of the αM β-propeller in specific ligand recognition by αMβ2 using a number of different approaches, and we report four major findings. 1) Substitution of five individual segments (Asp398–Ala402, Leu412–Leu419, Tyr426–Met434, Phe435–Glu443, and Ser444–Thr451) within the W4 blade of the β-propeller with their homologous counterparts in integrin α2 abrogated C3bi binding, whereas substitution of eight other segments outside this blade had no effect. 2) These five mutants defective in C3bi binding supported strong αMβ2-mediated and cation-dependent cell adhesion to fibrinogen, suggesting that the conformations of these five defective mutants were intact. 3) Polyclonal antibodies recognizing sequences within the W4 blade significantly blocked C3bi binding by wild-type αMβ2. 4) A synthetic peptide corresponding to Gln424–Gly440 within W4 interacted directly with C3bi. In conclusion, our data demonstrate that the W4 blade (residues Asp398 to Thr451) is involved specifically in C3bi but not fibrinogen binding to αMβ2. Altogether, our study supports a model in which three separate domains of αMβ2 (the αMI-domain, the αM β-propeller, and the β2I-domain) function together and contribute to the formation of the C3bi-binding site.

Like other integrins, ␣ M ␤ 2 is notorious in its ability to recognize multiple structurally unrelated ligands, including fibrinogen (Fg) 1 (9), ICAM-1 (10), C3bi (1), and neutrophil inhibitory factor (NIF), a specific ␣ M ␤ 2 antagonist isolated from canine hookworms (11). Studies from our laboratory and others suggested that the underlying molecular mechanism for such broad ligand specificity of the receptor is several overlapping but not identical binding pockets within ␣ M ␤ 2 . For example, both C3bi and Fg recognize an inserted region of ϳ200 amino acids in the ␣ M subunit termed the I-domain (␣ M I-domain) (12), but different residues are involved (13)(14)(15). Additionally, ligand binding by ␣ M ␤ 2 is mediated by other regions including the ␤-propeller region of ␣ M (16, 17) and the I-domain region (residues 125-385) of ␤ 2 (18,19).
In the past few years, we have used homolog-scanning mutagenesis (20) to map the ligand-binding sites in ␣ M ␤ 2 (13,14,21). The ligand-binding sites identified have been confirmed using a number of different complementary approaches, including gain-in-function mutations, synthetic peptides, and epitope mapping studies of function-blocking mAbs. Importantly, the recently published crystal structure of the collagen-␣ 2 I-domain complex (the ␣ 2 I-domain has 47% homology with the ␣ M Idomain) agrees well with our results (22). Therefore, we chose to employ homolog-scanning mutagenesis in this study to identify critical regions within the ␣ M ␤-propeller for C3bi binding. We report here that the W4 blade, containing residues Asp 398 to Thr 451 , is involved specifically in C3bi but not Fg binding.  as well as mAbs TS1/18,  TS1/22, OKM1, M1/70, M1/87, 44a, 904, and LM2/1 were from the  ATCC (ATCC, Manassas, VA). mAb 44 was from Sigma, and mAb 31H4  was from BIOSOURCE (Camarillo, CA).
Site-directed Mutagenesis and Establishment of Stable Cell Lines-The detailed procedures for homolog-scanning mutagenesis and establishment of stable cell lines expressing wild-type and mutant ␣ M ␤ 2 in human kidney 293 cells have been published (21). Similar procedures were used to generate 13 ␣ M ␤-propeller mutants that contain individual switches between the ␣ M and ␣ 2 ␤-propellers, which have 52% homology. To obtain cell lines that express equivalent receptor numbers as wild-type ␣ M ␤ 2 , each mutant cell line was sub-cloned by cell sorting using ␣ M -specific mAb 44a. Up to 20 colonies were picked and analyzed for integrin expression by FACS analysis. Cells expressing similar levels of receptor to those expressing wild-type ␣ M ␤ 2 were selected and subcloned. To exclude the possibility of subcloning artifacts, all of our studies have been repeated using the original pool for every mutant.
Ligand Binding to ␣ M ␤ 2 -expressing Cells-C3bi binding was performed with slight modification of the method of Bilsland (23). Sheep erythrocytes coated with C3bi (EC3bi) were prepared using anti-sheep erythrocyte IgM antibody M1/87 and human C5-deficient serum (Sigma). Briefly, 7 ϫ 10 8 sheep erythrocytes (Colorado Serum Company, Denver, CO) were washed twice in HBSS, containing 5 mM HEPES and 1 mM Mg 2ϩ , and coated with IgM as described (24). The coated erythrocytes were surface-labeled with biotin using 1 mg of sulfosuccinimidyl-6-(biotinamido) hexanoate (Pierce) at 37°C for 20 min. The biotinylated cells were resuspended in 0.9 ml of HBSS with 5 mM HEPES, 1 mM Ca 2ϩ , and 1 mM Mg 2ϩ , mixed with 100 l of C5-deficient serum, and incubated at 37°C for 60 min. After washing twice, the resulting EC3bi was resuspended in 2 ml of the above solution.
To perform the EC3bi binding assays, a total of 2 ϫ 10 5 ␣ M ␤ 2expressing cells were seeded onto polylysine (50 g/ml)-coated 24-well non-tissue culture polystyrene plates (BD Biosciences) for 15 min at 37°C, followed by addition of 2 ϫ 10 7 EC3bi. After 60 min at 37°C, unbound EC3bi were removed by washing with PBS. Bound EC3bi were fixed with 2% paraformaldehyde overnight, and excess paraformaldehyde was neutralized with 1% BSA at 37°C for 2 h. Bound EC3bi were quantitated by addition of 300 l of avidin-alkaline phosphatase conjugate (1:2000 dilution) (Zymed Laboratories Inc., San Francisco, CA). After 90 min at 37°C, the plates were washed three times with PBS, and 250 l of 3 mg/ml p-nitrophenyl phosphate was added. After 15 min of incubation at 37°C, the absorbance at 405 nm was determined.
Cell adhesion to the Fg ␥-module was carried out as described previously (13,14,21). A total of 2 ϫ 10 6 ␣ M ␤ 2 -expressing cells were added to 24-well non-tissue culture polystyrene plates, which were pre-coated with recombinant ␥-module (10 g/ml) and subsequently blocked with 0.05% polyvinylpyrrolidone in DPBS. After incubation at 37°C for 20 min, the unbound cells were removed by three washes with DPBS, and the adherent cells were quantified by cell-associated acid phosphatase activity.
Solid Phase Binding Assay-To test interactions between identified sequences of the ␣ M ␤-propeller and the ␣ M ␤ 2 ligand C3bi, 100 l of the synthetic peptides at different concentrations (0 -5 mM in DPBS) were coated onto the center of a 24-well plate overnight at 4°C. The efficiency of peptide coating was determined by labeling the N-terminal free sulfhydryl (-SH) group within each peptide, using the EZ-link PEOmaleimide-activated biotin kit (Pierce), based on the product instruction. After washing with PBS and blocking with BSA, the amount of the immobilized biotin group within each well was determined using an avidin-alkaline phosphatase conjugate and p-nitrophenyl phosphate as the substrate and measuring the absorption at 405 nm. For C3bi binding, the peptide-coated plate was blocked with 1% BSA for 1 h at 22°C, and then biotinylated EC3bi in 300 l of HBSS containing 1 mM Ca 2ϩ and 1 mM Mg 2ϩ was added. After incubating for 1 h at 37°C, nonadherent EC3bi were removed by washing. The plate was then fixed with 2% paraformaldehyde and blocked with 2% BSA. C3bi binding was measured with a conjugate of avidin-alkaline phosphatase as described above and measuring the absorbance at 405 nm.
FACS Analysis-A total of 1 ϫ 10 6 cells in HBSS containing 1 mM Mg 2ϩ were incubated with 5 g of mAb for 30 min at 4°C. A subtypematched mouse IgG served as a control. After 3 washes with PBS, cells were mixed with FITC-goat anti-mouse IgG(HϩL) F(abЈ)2 fragment ((Zymed Laboratories Inc.), kept at 4°C for another 30 min, washed with PBS, and then resuspended in 500 l of PBS. FACS analysis was then performed using FACScan (BD Biosciences), counting 10,000 events. Mean fluorescence intensities were quantitated using the FAC-Scan program, and the values were used to compare ␣ M ␤ 2 expression levels and reactivity of the cells with various mAbs.

Contribution of the ␣ M ␤-Propeller to C3bi
Binding-Previously, we showed that switching the ␤-propeller and the stalk region between ␣ M and ␣ L (␣ L ␤ 2 does not interact with C3bi) had no effect on C3bi binding (13), suggesting that the functional residues within the ␤-propellers of ␣ M and ␣ L are well conserved. Therefore, we chose another I-domain containing integrin ␣ subunit (␣ 2 ) in this study, which is less homologous to ␣ M . The ␣ M and ␣ 2 ␤-propellers have 52% homology, yet the ␣ 2 subunit does not recognize C3bi as its ligand. Because the ␣ M I-domain plays a major role in ligand binding, we tested whether its ligand binding property could be affected by other domains of the receptor, including the ␤-propeller. Thus, we placed the ␣ M I-domain in the context of ␣ 2 by substituting the ␣ 2 I-domain with the ␣ M I-domain, and we expressed the chimeric receptor ␣ 2 (I/␣ M )␤ 1 on Chinese hamster ovary cells. FACS analyses showed that the chimeric ␣ 2 (I/␣ M )␤ 1 receptor was well expressed (Fig. 1A); it reacted with a mAb (31H4) that recognizes the ␣ 2 ␤-propeller (Fig. 1A, e) but not with a mAb (P1E6) that recognizes the ␣ 2 I-domain (Fig. 1A, d). The presence of the ␣ M I-domain within ␣ 2 (I/␣ M )␤ 1 was confirmed by its reactivity toward the ␣ M -specific mAb 44a (Fig. 1A, f), whose epitope is located within the ␣ M I-domain (12,13). As a control, ␣ 2 ␤ 1 reacted with both ␣ 2 -specific mAbs (31H4 and P1E6) but did not bind the ␣ M -specific mAb 44a (Fig. 1A, a-c), whereas ␣ M ␤ 2 reacted with mAb 44a but not with mAb 31H4 or P1E6 (data not shown). To test if the expressed chimeric receptor is functional, we incubated the Chinese hamster ovary cells expressing either wild-type ␣ 2 ␤ 1 or the chimeric ␣ 2 (I/␣ M )␤ 1 receptor with biotinylated NIF, which recognizes specifically ␣ M ␤ 2 (11,24), and we then detected bound NIF with avidin-FITC by FACS analysis. As shown in Fig. 1B, wild-type ␣ 2 ␤ 1 did not interact with biotinylated NIF significantly. However, insertion of the ␣ M I-domain in ␣ 2 conferred ␣ 2 ␤ 1 the ability to bind NIF. Compared with wild-type ␣ 2 ␤ 1 , NIF binding to ␣ 2 (I/␣ M )␤ 1 was increased 7-fold (mean fluorescence intensity 4 versus 28), demonstrating that the ␣ M I-domain, when placed in the context of the ␣ 2 ␤-propeller, is still functional. Verifying specificity of the FACS analysis, NIF binding to either ␣ M ␤ 2 or ␣ 2 (I/ ␣ M )␤ 1 could be blocked by addition of 1 mM EDTA or 50-fold excess of non-labeled NIF (data not shown).
Homolog-scanning Mutagenesis of the ␣ M ␤-Propeller-To localize functional residues within the ␤-propeller, we carried out homolog-scanning mutagenesis using our established methods (21). We focused our attention primarily on four individual blades (W4 to W7) of the ␤-propeller, particularly the W4 blade, based on our earlier study showing that the C3bibinding site within the ␤ 2 I-domain resides proximal to the W4 and W5 blades of the ␤-propeller (19). Guided by the crystal structure of ␣ V ␤ 3 (25) and the protein model of the ␣ M ␤-propeller (16), the connecting loops on both upper and lower faces of the ␣ M ␤-propeller were individually substituted with their homologous counterparts of ␣ 2 ( Fig. 2A), using the mutagenic primers listed in Table I. The presence of the expected mutations and the correctness of the rest of the ␣ M sequence were confirmed by DNA sequencing. The mutated ␣ M was then expressed, together with wild-type ␤ 2 , on human 293 cells using our previously published methods. Altogether, a total of 13 mutants was constructed, and their corresponding stable cell lines were established. We found that all 13 ␣ M ␤ 2 mutants could be well expressed on the cell surface and could react with both an ␣ M -specific mAb (44a) and a ␤ 2 -specific mAb (IB4). In addition, we have used a panel of conformation-dependent mAbs to probe the folding of these ␣ M ␤-propeller mutants, and we found that reactivity toward these mAbs was preserved for all the 13 mutants (data not shown), suggesting that substitu-tion of these individual sequences within ␣ M did not significantly affect the gross structure of the ␣ M ␤ 2 receptor. Furthermore, surface labeling and immunoprecipitation experiments using mAb 44a showed that all 13 mutants formed correct heterodimers with ␤ 2 on the cell surface and exhibited expected molecular weights as judged by SDS-PAGE (Fig. 2B).
A Critical Role of the W4 Blade in C3bi Binding to ␣ M ␤ 2 -To evaluate the impact of the homolog-scanning mutations of the ␣ M ␤-propeller on ligand binding function of ␣ M ␤ 2 , we con- After washing, bound NIF (filled lines) was detected with an avidin-FITC conjugate. Verifying specificity of NIF binding, addition of 1 mM EDTA completely blocked NIF binding by ␣ 2 (I/␣ M )␤ 1 (open lines). C, C3bi binding. Biotinylated EC3bi (2 ϫ 10 7 ) was added to 2 ϫ 10 5 cells expressing wild-type ␣ M ␤ 2 , the chimeric receptor ␣ 2 (I/␣ M )␤ 1 , or ␣ 2 ␤ 1 , which had been pre-seeded onto polylysine-coated 24-well plates. After 60 min at 37°C, the amount of bound EC3bi was determined using avidin-alkaline phosphatase and pnitrophenyl phosphate, measuring the absorbance at 405 nm. Specificity was demonstrated by addition of 1 mM EDTA (gray bars). Data are the means Ϯ S.D. of two independent experiments. ducted ligand binding assays using two representative ligands of ␣ M ␤ 2 , Fg and C3bi. As shown in Fig. 3A, all 13 homologscanning mutants interacted with the ␥-module (a major recognition domain within Fg for ␣ M ␤ 2 (13,14,21)) in a cationdependent manner very similar to the wild-type receptor, indicating that 1) all 13 mutants exhibited correct conformations capable of ligand binding, and 2) switching these sequences did not alter the cation-dependent nature of the ␣ M ␤ 2 / ligand interaction. Verifying the specificity of the cell adhesion assay, mock-transfected cells did not adhere to the ␥-module, and for all 13 mutants, cell adhesion could be completely blocked by addition of 10 nM NIF (data not shown). We next evaluated C3bi binding activity of the 13 mutant ␣ M ␤ 2 receptors using our established procedures (15). As shown in Fig. 3B These five mutated segments are all located within the W4 blade of the ␤-propeller (Fig. 2A). These data suggested that the W4 blade contributed significantly to the formation of the C3bi binding pocket. Similar expression levels for the wild-type receptor and the 13 mutants were verified by FACS analysis using mAb 44a, and the mean fluorescence intensity did not vary by more than 1.5-fold compared with wild-type ␣ M ␤ 2 -bearing cells. Most importantly, these five mutants exhibited similar cell adhesion to Fg (Fig. 3A). Thus, differences in receptor expression level or global conformation of the receptors were not responsible for the loss of C3bi binding by these five mutants.
Inhibition of C3bi Binding by Antibodies Specific for the W4 Blade-To confirm the importance of the W4 blade in C3bi binding, we prepared polyclonal antibodies against the functional sequences of the W4 blade. One of the polyclonal antibodies, recognizing the sequence NMTRVDSDMNDAYL at the beginning of W4, gave a high titer in enzyme-linked immu- The targeted regions are shown in brackets. B, surface labeling and immunoprecipitation. The ␣ M ␤ 2 -expressing cells (1 ϫ 10 6 ) were surface-labeled with biotin and then immunoprecipitated with a mAb (44a) to ␣ M overnight at 4°C. After washing, the immunoprecipitates were subjected to 7% SDS-PAGE, transferred to a polyvinylidene difluoride membrane, and immunoblotted with an avidin-horseradish peroxidase conjugate. The membrane was then incubated with ECL substrate and exposed to Kodak X-Omat film for 1 min.  nosorbent assays. In addition, an antibody specific for a sequence on the upper face of the W2 blade (QEIVAAN-QRGSLYQ) was prepared as a control. This antibody gave a similar titer as the antibody specific for the W4 sequence. The ability of these two antibodies to react with ␣ M ␤ 2 was confirmed by FACS analysis (data not shown). To see if the two antibodies were capable of blocking C3bi binding, we pre-incubated these two antibodies with the wild-type ␣ M ␤ 2 -expressing cells, and we then performed the C3bi binding assays. Fig. 4 shows that addition of the W4-specific antibody reduced C3bi binding significantly (p Ͻ 0.01), whereas neither a rabbit IgG control nor the W2-specific antibody displayed significant inhibition. As the ␣ M I-domain plays an important role in C3bi binding by ␣ M ␤ 2 , we expected that the W4-specific antibody would exert stronger inhibition when combined with NIF, a high affinity antagonist of ␣ M ␤ 2 that is specific for the ␣ M Idomain (13,24). Indeed, addition of NIF (10 nM) alone reduced C3bi binding to ␣ M ␤ 2 by 60%, and combining NIF with the W4-specific antibody resulted in more than 90% inhibition of C3bi binding. In contrast, addition of the W2-specific antibody or a rabbit IgG control did not produce any further inhibition over NIF alone (Fig. 4). These data demonstrated that the W4 blade of the ␤-propeller, along with the ␣ M I-domain, contributed significantly to C3bi binding by ␣ M ␤ 2 .
C3bi Binding to the W4 Peptides-As an ultimate proof for the direct involvement of the W4 blade in C3bi binding, we tested whether C3bi would bind directly to peptides derived from the W4 blade. Based on the three-dimensional model of the ␣ M ␤-propeller and our above mutagenesis data, we synthesized four peptides pN 391 -L 404 (C 391 NMTRVDSDMNDA-YL 404 ), pI 411 -V 420 (C 411 ILRNRVQSLV 420 ), pP 424 -G 440 (C 424 PRYQHIGLVAMFRQNTG 440 ), and pM 441 -G 454 (C 441 MWESNANVKGTQIG 454 ), and their corresponding scrambled controls. In our preliminary experiments, we tested the ability of these synthetic peptides to support direct C3bi binding, when immobilized on the surface of a 24-well plate. All four peptides were immobilized readily on the plastic surface with similar coating efficiencies (ϳ0.3 nmol of each peptide were immobilized per well when 100 M of the peptide were used for coating). Of the four synthetic peptides, one peptide pP 424 -G 440 bound C3bi well. The other three peptides did not support detectable C3bi binding (data not shown), possibly due to their less optimal conformations in the absence of the structural constraints of the ␤-propeller. Subsequently, we conducted further experiments on this peptide. As shown in Fig. 5A, peptide pP 424 -G 440 bound C3bi in a dose-dependent and saturable manner, whereas its scrambled control (CRLGPIRHMVY-FQQGATN) did not show significant binding. The titration data could be fit to a single binding site model using non-linear regression analysis. The binding constant (K d ) was estimated to be around 20 M, which is in the same range as one of the most active ␣ M I-domain-derived peptides (A7, residues FIG. 3. Ligand binding by the wild-type and mutant ␣ M ␤ 2 transfectants. A, Fg adhesion. A total of 2 ϫ 10 6 ␣ M ␤ 2 -expressing cells were added to 24-well non-tissue culture polystyrene plates, which were pre-coated with recombinant ␥-module (10 g/ml) and subsequently blocked with 0.05% polyvinylpyrrolidone in DPBS. After incubation at 37°C for 20 min, the unbound cells were removed by three washes with DPBS, and the adherent cells were quantified. The value for wild-type ␣ M ␤ 2 was taken as 100%. Verifying specificity and cation dependence, cell adhesion could be blocked by addition of 1 mM EDTA (gray bars) (or 10 nM NIF, data not shown), and mock-transfected cells did not adhere to the ␥-module. Data are the means Ϯ S.D. of three independent experiments. B, C3bi binding. Binding of biotinylated EC3bi to ␣ M ␤ 2 was conducted as described in Fig. 1B 232-245) (26). These data strongly support our above mutagenesis results and establish the W4 blade of the ␣ M ␤-propeller as a direct binding interface for C3bi. To see if these synthetic peptides could function as soluble inhibitors of C3bi binding to the immobilized pP 424 -G 440 , we repeated the C3bi binding assay in the presence of different soluble peptides, including pN 391 -L 404 , pP 424 -G 440 , and pM 441 -G 454 , as well as two C3bi-binding peptides derived from the ␤ 2 I-domain, pQ 199 -K 209 (CQPPFAFRHVLK) and pS 303 -E 413 (CSVGQLAHKLAE) (19). Fig. 5B shows that only peptide pP 424 -G 440 blocked C3bi binding efficiently. The other four peptides did not exhibit significant inhibition. Verifying the specificity, scrambled pP 424 -G 440 had no inhibitory activity, and no specific C3bi binding was observed on BSA-coated wells.

DISCUSSION
C3bi binding by leukocyte integrin ␣ M ␤ 2 plays critical roles in host defense functions. The functional domains within ␣ M ␤ 2 involved in C3bi recognition have been mapped to the I-domains of the ␣ M and the ␤ 2 subunits (13,19,26). Recently, Yalamanchili et al. (17) have presented evidence for a critical role of the ␣ M ␤-propeller in C3bi binding by ␣ M ␤ 2 . However, the exact location of the C3bi-binding site within the ␤-propeller is still unknown. In this study, using a combination of different approaches, we have identified the W4 blade within the ␤-propeller, containing segments Asp 398 -Ala 402 , Leu 412 -Leu 419 , Tyr 426 -Met 434 , Phe 435 -Glu 443 , and Ser 444 -Thr 451 , as a major contact site for C3bi. Our results indicate that the W4 blade is specifically involved in C3bi but not Fg recognition, as the five mutants that were defective in C3bi binding exhibited normal cell adhesion to Fg. A direct involvement of the W4 blade in C3bi recognition was demonstrated by the ability of a synthetic W4 peptide Pro 424 -Gly 440 to bind C3bi, when immobilized on the surface of a microtiter plate.
The critical roles of the ␤-propeller in ligand binding have been well documented in the literature, especially for those integrins which do not contain an I-domain in their ␣ subunits (non-I-domain integrins). Using different approaches, the ligand-binding sites have been mapped predominantly to the W2 and W3 blades for a number of integrins, including ␣ V , ␣ IIb , ␣ 4 , ␣ 5 , etc. (25,(27)(28)(29). Indeed, the recently published crystal structure of the ␣ V ␤ 3 -RGD complex reveals that Asp 150 within W3 and Asp 218 within W4 of the ␣ V ␤-propeller provide direct coordination to the bound RGD ligand (25). Compared with these non-I-domain integrins, less is known regarding the function of the ␤-propeller in integrins that contain an I-domain in their ␣ subunits (I-domain integrins), such as ␣ M , ␣ L , ␣ 2 , etc. It was reported that mAb OKM1, which recognizes an epitope outside the ␣ M I-domain, partially inhibits C3bi binding by ␣ M ␤ 2 (12). In addition, a mutant ␣ M ␤ 2 which does not contain an I-domain in its ␣ subunit is still capable of C3bi binding (17), implicating a role of the ␤-propeller in C3bi binding by ␣ M ␤ 2 , although such binding could also be mediated in part by the ␤ 2 subunit. Therefore, a definitive role of the ␤-propeller in ligand binding needs to be tested. Complicating this matter further, the ligand binding surface identified in the non-I-domain integrins is disrupted by insertion of the I-domain (between W2 and W3). Thus, the exact location of the ligand-binding site within the ␤-propeller of the I-domain integrins is unclear. Moreover, as the I-domain plays a dominant role in ligand recognition (12, 15, 21, 24, 26, 30 -32), the function of the ␤-propeller within the intact heterodimeric receptor remains to be defined. To address these issues, we used several complementary approaches in this study. Our results demonstrate that the ␤-propeller is involved directly in C3bi recognition. In addition, we found that the ligand binding property of the I-domain could be affected by the ␤-propeller. For example, the ␣ M I-domain within the ␣ M ␤-propeller recognized both NIF and C3bi, but it interacted only with NIF when placed in the ␣ 2 ␤-propeller (Fig. 1, B and C). These data suggested that the ␤-propeller could contribute to ligand binding in two different ways as follows: by participating directly in C3bi recognition, and by modulating the properties of the ␣ M I-domain.
The novel C3bi-binding site we identified in this study resides within the W4 blade of the ␤-propeller (Fig. 6). Surprisingly, this C3bi-binding site contained the entire W4 blade, including residues located in both the upper and lower faces of the ␤-propeller. This is in sharp contrast to the ligand-binding sites identified within several non-I-domain integrins, where all the functional residues were shown to reside on the upper face of the ␤-propeller (25,28,33). That the W4 blade of the ␣ M ␤-propeller is involved directly in C3bi recognition is supported by several observations. 1) Substitution of the five individual segments (Asp 398 -Ala 402 , Leu 412 -Leu 419 , Tyr 426 -Met 434 , Phe 435 -Glu 443 , and Ser 444 -Thr 451 ) within W4 abrogated C3bi binding, whereas segment substitutions within the other three blades (W5, W6, and W7) had no effect (Fig. 3B). Loss of C3bi binding by these W4 mutants did not result from alterations of gross conformations of the mutant ␣ M ␤ 2 receptors, as all 13 ␤-propeller mutants were expressed well on the cell surface, formed correct heterodimers with the ␤ 2 subunit (Fig. 2), and reacted with a panel of conformation-dependent ␣ M ␤ 2 mAbs (data not shown). Most importantly, all 13 ␤-propeller mutants, including the five defective receptors, supported cation-dependent cell adhesion to Fg in a manner similar to wild-type ␣ M ␤ 2 (Fig. 3A). 2) Antibodies specific for the W4 blade, but not the W2 blade, blocked C3bi binding by the wild-type ␣ M ␤ 2 receptor (Fig. 4). 3) a synthetic peptide (Pro 424 -Gly 440 ) corresponding to sequences within W4, but not its scrambled control, bound C3bi directly with a K d of 20 M (Fig. 5A). Although the other three peptides did not show significant binding of C3bi, they may simply adopt a less optimal conformation for ligand recognition, when existing outside the structural constraints of the ␤-propeller. Additionally, we found that the two C3bi-binding peptides identified within the ␤ 2 I-domain (19) did not compete with peptide Pro 424 -Gly 440 for C3bi binding (Fig. 5B), suggesting that either the ␤-propeller peptide Pro 424 -Gly 440 and the two ␤ 2 I-domain peptides recognized different regions of C3bi or that the ␤ 2 I-domain peptides bound C3bi much weaker and therefore were unable to compete with the ␤-propeller peptide. The importance of W4 in ligand binding has also been demonstrated in other integrins. For example, a region within the ␣ V ␤-propeller (Asp 218 in W4), which is homologous to the critical segment Asp 398 -Ala 402 found in this study, was shown to contact its ligand (RGD) directly in the crystal structure (25). In addition, Dickeson et al. (34) have reported that efficient colla-gen binding to ␣ 2 ␤ 1 required both the I-domain and the W3 to W5 blades of the ␣ 2 ␤-propeller, suggesting that the regions surrounding the W4 blade may play important roles in direct ligand recognition for other integrin subfamilies as well.
Results from this study and others in the literature suggest that ␣ M ␤ 2 possesses at least three independent ligand binding domains: the ␣ M I-domain, the ␤ 2 I-domain, and the ␣ M ␤-propeller (13,17,19,26). How these individual domains work together to form a functional ligand-binding site within the heterodimeric ␣ M ␤ 2 receptor is currently unknown. Given that the W4 blade of the ␤-propeller is located in close proximity to another C3bi recognition site we identified previously in the ␤ 2 I-domain (19), it is highly possible that these individual binding sites may function together to form a composite C3bi binding pocket. Accordingly, we propose that three distinct regions within ␣ M ␤ 2 , including the ␣ M I-domain, the ␤ 2 I-domain, and the ␣ M ␤-propeller, reside proximally within the receptor and contribute directly to the formation of the ligand binding pocket. The fact that the W4 blade is non-essential to Fg recognition suggests that the Fg/␣ M ␤ 2 interaction is likely mediated mainly through the two I-domains within the ␣ M and ␤ 2 subunits, whereas C3bi binding depends on three domains (the ␣ M and ␤ 2 I-domains and the ␣ M ␤-propeller). Based on this model, we predict that the MIDAS motif of the ␣ M I-domain should reside closely to both the W4 blade of the ␤-propeller and the ␤ 2 I-domain. In support of this model, we found that C3bi binding to intact ␣ M ␤ 2 was inhibited more effectively if the binding sites within the ␣ M I-domain and the ␣ M ␤-propeller were both blocked (Fig. 4). In addition, Yalamanchili et al. (17) reported that C3bi binding to the I domain-less ␣ M ␤ 2 could be inhibited completely by a ␤-propeller antibody (CBRM1/32), suggesting that the two binding surfaces (the ␣ M ␤-propeller and the ␤ 2 I-domain) within the I domain-less ␣ M ␤ 2 are located relatively close to each other in space (Ͻ70 Å of a typical antigen-binding site within a FabЈ fragment), such that the blocking mAb could occupy the two binding sites simultaneously, leading to complete inhibition. Alternatively, as mAb CBRM1/32 recognizes a region (corresponding to segment Ala 543 -Arg 550 ) in the upper face of the W6 blade (35), which itself is not involved directly in C3bi binding (Fig. 3), it is also possible that CBRM1/32 could function as a wedge to alter the relative orientation between the ␣ M ␤-propeller and the ␤ 2 I-domain, resulting in disruption of the optimal conformation for C3bi binding. Altogether, our model would expect that 1) the ␣ M I-domain plays a major role in C3bi binding to the intact receptor; 2) among the three individual domains that form the composite C3bi-binding site, the ␤-propeller and the ␤ 2 I-domain reside relatively close to each other in space, whereas the ␣ M I-domain is located farther apart; and 3) high affinity ligand binding to ␣ M ␤ 2 requires optimal orientations among all three ligand binding domains, and therefore spatial changes relative to each other could affect ligand binding and thereby provide a potential means to control the affinity state of the integrin receptor ("inside-out" signaling). Two other models have been proposed in the literature, in which the MIDAS motif of the ␣ M I-domain was shown to project away from the W4 blade of the ␤-propeller and the ␤ 2 I-domain (36,37). Further studies will be needed to test the validities of these different models.
In conclusion, using several different approaches, we have identified a novel C3bi-binding site within the W4 blade of the ␣ M ␤-propeller. Our data demonstrate that the W4 blade is involved differentially in ␣ M ␤ 2 binding to its two physiological ligands C3bi and Fg. When compared with the ligand-binding site within ␣ V ␤ 3 , which is located on the upper face of the ␤-propeller (25), the C3bi-binding site identified in this study encompasses residues on both the upper and lower faces of the ␣ M ␤-propeller. Altogether, our studies support a model in which three individual domains of ␣ M ␤ 2 , the ␣ M I-domain, the ␣ M ␤-propeller, and the ␤ 2 I-domain, reside together in space and contribute to the formation of a common C3bi-binding site.