Crystal Structure of the Ectodomain of Human Fc (cid:1) RI*

Human Fc (cid:1) RI (CD89) is the receptor specific for IgA, an immunoglobulin that is abundant in mucosa and is also found in high concentrations in serum. Although Fc (cid:1) RI is an immunoglobulin Fc receptor (FcR), it differs in many ways from FcRs for other immunoglobulin classes. The genes of most FcRs are located on chromosome 1 at 1q21–23, whereas Fc (cid:1) RI is on chromosome 19, at 19q13.4, a region called the leukocyte receptor complex, because it is clustered with several leukocyte receptor families including killer cell inhibitory receptors (KIRs) and leukocyte Ig-like receptors (LIRs). The amino acid sequence of Fc (cid:1) RI shares only 20% homology with other FcRs but it has around 35% homology with its neighboring LIRs and KIRs. In this work, we analyzed the crystal structure of the ectodomain of Fc (cid:1) RI and into a Novagen pET-28a vector using the Nco I and Xho I restriction sites and an E. coli BL21(DE3) strain. Two additional amino acids (Met-Ala) were added to the 5 (cid:1) end of the gene, and a histidinetag(His 6 )wasaddedtothe3 (cid:1) endtofacilitatetheexpressionand purification. The protein was first expressed in an inclusion body form and then reconstituted in vitro . The isolation of the inclusion bodies was started with an intense combined lysozyme/sonification procedure to open virtually all cells. Subsequent washing steps with Triton X-100 and NaCl yielded a product with a purity of (cid:2) 80% as estimated by SDS-PAGE. The grown from a of in m M buffer, 2 SO, and m M 2 as and same BL41XU under K at wavelengths Å, 0.9800, and 0.9000 Å and processed using HKL2000 (19). crystals belong to the space group C222 1 with the unit cell dimensions of a (cid:4) 59.0, b (cid:4) 69.5, c (cid:4) Å and one each

In humans, IgA is the most abundant immunoglobulin in secretions, and it constitutes about 20% of the immunoglobulin pool in serum (1,2). Since its turnover rate is faster than other immunoglobulins, the daily production of IgA exceeds all other immunoglobulins combined (3). Undoubtedly, IgA should to play important roles in immune defense against invaded pathogens.
Five types of IgA receptors have been recognized so far. They are Fc␣RI (CD89), the polymeric Ig receptor, Fc␣/R, the transferrin receptor, and the asialoglycoprotein receptor (1). Among them, Fc␣RI is the only one that specifically binds IgA. On ligation of IgA complexed with antigens, Fc␣RI is able to mediate various cellular responses including phagocytosis, anti-body-dependent cell cytotoxicity, oxidative bursts, and release of inflammatory mediators (1).
Fc␣RI belongs to the immunoglobulin superfamily and contains an extracellular region of 206 amino acids, a transmembrane domain of 19 amino acids and a cytoplasmic region of 41 amino acids (4). The extracellular region of Fc␣RI consists of two Ig-like domains, EC1 and EC2, and six potential sites for N-glycosylation. The receptor binds IgA1 and IgA2 with an equal affinity (5). A number of residues including Tyr 35 , Arg 52 , Tyr 81 , Arg 82 , Ile 83 , Gly 84 , His 85 , and Tyr 86 on Fc␣RI are potentially involved in IgA binding (6,7).
Although Fc␣RI is an immunoglobulin Fc receptor (FcR), 1 it differs in many ways with FcRs for other immunoglobulin classes. IgG receptor Fc␥RIII and IgE receptor Fc⑀RI bind antibodies in the near hinge regions and form 1:1 complexes (8,9), whereas Fc␣RI binds the C H 2-C H 3 interface of Fc␣ (10,11) and preferably forms 2:1 complex with a single Fc␣ homodimer (12). It has been reported that Fc␥Rs and Fc⑀RI use their membrane proximal-domain and linker region binds immunoglobulin (8,9,13,14), whereas Fc␣RI uses its membrane-distal domain EC1 to bind IgA (15).
The genes of most FcRs are located in chromosome 1 at 1q21-23 (16), whereas Fc␣RI is in chromosome 19, at 19q13.4 (17,18), a region called the leukocyte receptor complex because it is clustered with several leukocyte receptor families including killer cell inhibitory receptors (KIRs) and leukocyte Ig-like receptors (LIR/LILR/ILTs) (17,18). The amino acid sequence of Fc␣RI shares only 20% homology with other FcRs, but it has around 35% homology with its neighboring LIRs and KIRs (1).
In this paper, we report our analysis of the crystal structure of the ectodomain of Fc␣RI expressed in Escherichia coli and its comparison with FcRs, LIR, and KIR.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-The construction and expression of the extracellular ligand binding domain of a human Fc␣RI will be described in detail elsewhere. 2 Briefly, residues 1-207 of the mature sequence were subcloned into a Novagen pET-28a vector using the NcoI and XhoI restriction sites and an E. coli BL21(DE3) strain. Two additional amino acids (Met-Ala) were added to the 5Ј end of the gene, and a histidine tag (His 6 ) was added to the 3Ј end to facilitate the expression and purification. The protein was first expressed in an inclusion body form and then reconstituted in vitro. The isolation of the inclusion bodies was started with an intense combined lysozyme/sonification procedure to open virtually all cells. Subsequent washing steps with Triton X-100 and NaCl yielded a product with a purity of Ͼ80% as estimated by SDS-PAGE. The The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The inclusion bodies were dissolved in a buffer containing 6 M guanidine hydrochloride and 5 mM dithiothreitol, and incubated for 2 h to unfold completely the misfolded protein of inclusion bodies. Refolding was achieved by dilution of the guanidine-dissolved inclusion bodies dropwise with stirring into the refolding buffer (0.1 M Tris/HCl, 1.5 M guanidine hydrochloride, 5 mM reduced glutathione, 0.5 mM oxidized glutathione, pH 8.5) at 4°C. The mixture was stirred for 2-3 days, and then the renatured Fc␣RI was applied to a Q-Sepharose high performance ion-exchange column and further purified on a Superdex-75 column.
Crystallization, Data Collection, and Structure Determination-The Fc␣RI crystals were obtained by the hanging-drop method. The crystals were grown from a buffer of 11.4% polyethylene glycol-8000 in 100 mM sodium Hepes buffer, pH 7.6, containing 8% (v/v) ethanol glycol, 3% (v/v) Me 2 SO, and 50 mM MgCl 2 as an additive, and protein concentration of ϳ12 mg/ml. The Se-Met derivative crystal was grown from the same conditions. The Se-Met derivative data were collected at the Spring8 beamline BL41XU under 100 K at wavelengths 0.9798 Å, 0.9800, and 0.9000 Å and processed using HKL2000 (19). The crystals belong to the space group C222 1 with the unit cell dimensions of a ϭ 59.0, b ϭ 69.5, c ϭ 106.4 Å and one molecule in each asymmetric unit. The SOLVE program (20) was used to locate Se sites and to calculate initial phases. Following density modification by RESOLVE (21), the resultant electron density map was of sufficient quality that the entire model except for one flexible loop and several residues at the termini could be built. The initial chain tracing and all subsequent model building were done using the program O (22), version 8.0. Refinement was performed using CNS1.0 (23) and merged synchrotron data with F obs Ͼ 0. The Bijvoet pairs of the data used in refinement are unmerged. The model was initially refined as a rigid body with data 8.0 -4.0 Å resolution. The resolution was extended gradually, and subsequent refinement used protocols including anisotropic temperature factor refinement, energy minimization, and slow cool simulated annealing. Several rounds of manual refitting using omit maps permitted the missing loop regions to be traced and side chains built. 68 water molecules were built into the electron density when a F o Ϫ F c map, contoured at 3.5, coincided with well defined electron density of a 2F o Ϫ F c map contoured at 1. The N-terminal 2 additional residues (MA), C-terminal 15 residues (DSIHQDYTTQNLILE), and residues 56 -59 (FWNE) were disordered in the crystal. The final model contained 191 residues of Fc␣RI and 68 solvent molecules. R cryst and R free were 0.210 and 0.239, respectively, for data in the resolution range 40.0 -2.1 Å. The structure contains two cis prolines at position 154 and 161. None of the main-chain torsion angles are located in disallowed regions of the Ramachandran plot. Statistics for data collection, phasing, and refinement are shown in Table I. For analyses of interdomain angles, contacts, and buried surface areas, D1 was defined as residues 1-100 and D2 was defined as residues 101-195, following the structure-based definition of KIR2DL1 domain boundaries (24). Interdomain contact residues were defined as being within 3.6 Å of the partner domain and identified using CON-TACT (35). Buried surface areas were calculated using SURFACE (35) with a 1.4-Å probe radius.

RESULTS AND DISCUSSION
The crystal structure of the extracellular region of Fc␣RI consists of two Ig-like domains, EC1 (residues Gln 1 to Gly 100 ) and EC2 (residues Pro 105 to His 199 ) (Fig. 1a). EC1 and EC2 obey the typical heart-shaped arrangement, and a short linker (Leu 101 to Lys 104 ) connects them together. Both domains are primarily composed of ␤-structure arranged into two antiparallel ␤ sheets with a KIR-like folding topology. The sheets are closely packed against each other with the conserved disulfide bridge connecting the strands B and F on the opposing sheets. Three 3 10 helices are found in N termini of EC1 (Glu 2 to Asp 4 ), EF loops of EC1 (Ala 71 , Asn 72 , and Lys 73 ) and EC2 (Leu 164 , Asn 165 , and Val 166 ), respectively.
The EC2 domain is built up from eight ␤-strands arranged such that three stands (A, B, E)  The interdomain angle of EC1 and EC2 is calculated to be 85°. The bent shape of the Fc␣RI produces a large interface between the D1D2 domains that buries 1134 Å 2 of the accessible surface area (Fig. 1c). Most of the residues involved in the EC1 and EC2 interdomain interaction are hydrophobic, including Val 17  As Fc␣RI shows a relatively high degree of homology to the TABLE I Data collection, phasing, and refinement statistics R merge ϭ ⌺͉I i Ϫ ͗I͉͘/⌺͉I͉, where I i is the intensity of an individual reflection and ͗I͘ is the average intensity of that reflection. R-factor ϭ ⌺͉F p ͉ Ϫ ͉F c ͉/⌺͉F p ͉, where F c is the calculated and F p is the observed structure factor amplitude. Phasing power ϭ F hcalc /E, where F hcalc ϭ the heavy atom structure factor amplitude and E ϭ the residual lack of closure error. R cullis ϭ ⌺ʈF ph Ϯ F p ͉ Ϫ ͉F hcalc ͉/⌺͉F ph Ϯ P p ͉, where F ph is the derivative structure factor amplitude. Data 1. Crystal structure of Fc␣RI ectodomain. a, stereo ribbon drawing of the structure of Fc␣RI. EC1 is the N-terminal domain, and EC2 is the C-terminal domain. Disulfide bonds are shown in green. The residues 56 -59 and 196 -199 were disordered in the electron density map. b, topological diagram of the ectodomain of Fc␣RI. The arrows show the directions of ␤-strands, whereas the 3 10 helix structures are represented by two circles. The amino acid residues at each end of ␤-strands and helices are numbered. c, close-up stereo view of the hydrophobic core in the interdomain interface of Fc␣RI. The 12 residues responsible for stabilizing the hydrophobic core are shown in ball-and-stick representation. Tyr 173 (Y173) is colored yellow, Tyr 181 (Y181) blue, and Trp 183 (W183) green. Other residues are colored using the CPK (Corey-Pauling-Kendrew) convention (blue, nitrogen; red, oxygen; gray, carbon; yellow, sulfur) color scheme. and KIR2DL2. Hence, the C-CЈ loop in Fc␣RI EC1 forms earlier (Fig. 2b), allowing the C-CЈ loop and F-G loop to adopt a clamp-like arrangement. A similar feature can also be found in many other FcRs (Fig. 2c) even though they share low degree of identity with Fc␣RI. The significance of this is still unknown.
Although LIR-1 does not contain of CЈ and D strands in the D1 domain, it more closely resembles Fc␣RI in the other part of the molecule compared with KIR2DL1 and KIR2DL2. The root mean square deviation (r.m.s.d) values for the C␣ atoms are 1.44 Å for Fc␣RI EC1 and LIR-1 D1, 1.54 Å for Fc␣RI EC1 and KIR2DL2 D1 and 1.77 Å for Fc␣RI EC1 and KIR2DL1 D1. It seems the lack of CЈ and D strands in LIR-1 D1 have little effect on its overall structure, although the sequence of CЈ and D regions are variable within KIR2DL1, KIR2DL2, and LIR-1 (Fig. 3).
The interdomain angle of Fc␣RI is closer to that of LIR-1 (84 to 90°) (25), but larger than that of KIR2DL2 (60 to 80°) (26). The hydrophobic core interface observed in Fc␣RI also exists in LIR-1 and KIR2DL2 (25,26). Amino acid sequence alignment shows the 12 hydrophobic residues, especially Tyr 181 and Trp 183 , which play an important role in stabilizing the interdomain angle in Fc␣RI are also conserved in LIR-1 and KIRs, having only one residue (Leu 101 3 Ala) different for LIR-1, three residues (Val 17 3 Leu, Val 98 3 Ile, and Tyr 173 3 Phe) different for KIR2DL2 and four residues (Val 17 3 Leu, Val 98 3 Ile, Thr 99 3 Ile and Tyr 173 3 Phe) different for KIR2DL1. These 12 residues are also conserved in a KIR from cow, with only one residue (Tyr 102 3 Ser) different from Fc␣RI in this region (Fig. 3). Moreover, a bovine IgG2 FcR, Fc␥2R, also possesses most of these hydrophobic residues and only four  residues (Val 97 3 Leu, Thr 99 3 Ala and Tyr 102 3 Arg and Trp 183 3 Leu) are different from Fc␣RI. This Fc␥2R has been previously found to share more similarities to Fc␣RI (41%) than to other types of human Fc␥Rs (less than 28%) (27), and it is located on the same chromosome as bovine KIR (28,29), indicating that it belongs to the bovine leukocyte receptor complex. In contrast, such a hydrophobic core does not exist in human FcRs for IgG and IgE (11, 30 -33). This suggests that the hydrophobic core is a common feature of receptors from the leukocyte receptor complex and Fc␣RI is evolutionally closer to LIR and KIR than to other human FcRs.
As an immunoglobulin Fc receptor, Fc␣RI differs from other FcRs not only in structure but also in its ligand binding characteristics. IgG receptor Fc␥RIII and IgE receptor Fc⑀RI bind antibodies in the near hinge regions and form 1:1 complexes (8,9). In contrast, Fc␣RI binds the C H 2-C H 3 interface of Fc␣ (10,11) and preferably forms a 2:1 complex with a single Fc␣ homodimer (12). It has been reported that Fc␥Rs and Fc⑀RI use their membrane-proximal domain and linker region to bind immunoglobulin (8,9,13,14), whereas Fc␣RI uses its membrane-distal domain EC1 to bind IgA (33). A number of residues have been implicated in IgA binding, including Tyr 35 , Arg 52 , Tyr 81 , Arg 82 , Ile 83 , Gly 84 , His 85 , and Tyr 86 (6). The crystal structure of Fc␣RI shows that Tyr 35 is located in the B-C loop, Arg 52 is located in CЈ strand, Tyr 81 and Arg 82 are located in F strand, and the remainder are in the F-G loop. All these residues lie on the receptor surface except Tyr 81 , which is buried inside the receptor and is unlikely to be involved directly in IgA binding (Fig. 4).
Fc␣RI has six potential N-linked glycosylation sites (Asn 44 , Asn 58 , Asn 120 , Asn 156 , Asn 165 , and Asn 177 ). Unglycosylated Fc␣RI has a molecular mass of 30 kDa. When expressed in vivo, its molecular mass is increased to 50 -100 kDa due to different degrees of glycosylation (1). Although the effect of glycosylation still needs to be elucidated, carbohydrates seem to play an important role in IgA binding since desialylated Fc␣RI binds five times more strongly to IgA (1). Fig. 4 shows the position of potential N-linked glycosylation site at Asn 44 , which is close to the docking sites of IgA.
In conclusion, the crystal structure and sequence alignment show that Fc␣RI is a member of the leukocyte receptor complex and evolutionally closer to LIR than KIR. All members of this complex found so far share a common hydrophobic core structure. The crystal structure also locates the residues that are involved in Fc␣RI binding to IgA.