Limited role of charge matching in the interaction of human immunoglobulin A with the immunoglobulin A Fc receptor (FcαRI) CD89

Abstract

Human immunoglobulin A (IgA) mediates protective effector mechanisms through interaction with specific cellular Fc receptors (FcαRI). Two IgA Fc interdomain loops (Leu257–Leu258 in the CH2 domain and Pro440–Phe443 in the CH3 domain) have previously been identified as critical for binding to FcαRI. On the receptor, the interaction site for IgA has been localized to the EC1 domain. The essential FcαRI residues involved are Tyr35, Tyr81 and Arg82, with contributions also from Arg52 and to a lesser extent from His85 and Tyr86. The basic nature of the side chains of some of the receptor residues implicated in ligand binding suggested that charge matching might play some role in the interaction. To address this possibility, we have generated five IgA1 mutants with point substitutions in acidic residues lying close to the putative interaction site and assessed their abilities to bind FcαRI on human neutrophils. Mutants E254A, E254L and E437A displayed affinities for FcαRI comparable to that of wild-type IgA1, while mutants D255A and D255V had only slightly reduced affinities for the receptor. Therefore, electrostatic interactions appear unlikely to play a significant role in the IgA–FcαRI interaction. Moreover, the lack of effect of mutations in residues adjacent to those previously implicated in binding, reaffirms the importance of the interdomain loops in FcαRI binding.

Introduction

Immunoglobulin A (IgA), a major serum antibody and the predominant immunoglobulin class in the seromucous secretions that bathe mucosal surfaces, serves as a key first line of defence against many invading pathogens. It also appears to function as a second line of defence mediating elimination of pathogens that have breached the mucosal surface.¹ In humans there are two subclasses of IgA, termed IgA1 and IgA2, the latter existing as two or possibly three allotypic variants.

An important component of the protective function of IgA relies on interaction with specific myeloid Fc receptors (FcαRI, CD89), present on the surface of a range of immune cells including neutrophils, macrophages, monocytes, eosinophils, Kupffer cells and dendritic cells.² Both IgA subclasses bind FcαRI. Receptor-ligation by IgA-coated targets initiates a tyrosine kinase signalling cascade mediated via the FcR γ chain associated with the ligand-binding α chain of the receptor, culminating in potent responses such as phagocytosis, respiratory burst, and release of cytokines.²

The α chain of the FcαRI has two extracellular immunoglobulin-like domains (ectodomain (EC) 1 and EC2) and displays homology to other human FcR specific for IgG and IgE, namely FcγRI, FcγRII, FcγRIII and FcɛRI. However, FcαRI is clearly a more distantly related member of the family, and in fact shares greater homology to killer cell immunoglobulin-like receptors (KIR), LAIR-1 and -2, and leucocyte immunoglobulin-like receptors, which lie in the leucocyte receptor cluster alongside FcαRI on chromosome 19.^3,4

The interaction site for IgA has been localized to the EC1 domain of FcαRI.^5–7 This site localization is in marked contrast to the ligand sites of FcγRI, FcγRII, FcγRIII and FcɛRI, which are all located in their membrane-proximal EC2 domains.^8–14 The essential FcαRI residues involved in binding to IgA, as identified by scanning mutagenesis, are Tyr35, Tyr81 and Arg82, with contributions also from Arg52 and to a lesser extent from His85 and Tyr86.^5,7 These residues are all closely arranged at the apical tip of the EC1 domain.

On IgA, residues lying on loops at the interface between the two heavy chain domains (CH2 and CH3) of the Fc region have been shown to be essential for interaction with FcαRI.^15,16 In particular, residues Leu257 and Leu258 on a CH2 loop, and residues Pro440, Leu441, Ala442 and Phe443 on a CH3 loop, appear critical for binding to and triggering of FcαRI. Molecular models of human IgA1 predict the loops to lie close in three-dimensional space¹⁷ so an FcαRI domain could presumably interact readily with both loops (or close-lying residues). Recent work using analytical ultracentrifugation and equilibrium gel filtration indicates that two FcαRI molecules bind to a single IgA Fc.¹⁸ The respective interaction site localizations on FcαRI and IgA are clearly compatible with such a stoichiometry. Indeed, a homology model for cellular FcαRI binding of IgA based on these interaction sites had earlier suggested that a 2 : 1 stoichiometry was sterically feasible.⁷

The basic nature of the side chains of some of the receptor residues implicated in ligand binding has lead to speculation that these residues might interact with acidic residues on IgA⁵ with electrostatic interactions potentially making a significant contribution to the free energy of binding. A search in the vicinity of the Leu257–Leu258 and the Pro440–Phe443 interface loops reveals a number of acidic residues that might be involved in charge–matching interactions with FcαRI. These include Glu254 and Asp255 adjacent to the Leu257–Leu258 loop, and Glu437 lying close to the Pro440–Phe443 loop (Fig. 1). In order to ascertain the possible contribution of these acidic amino acids to the interaction with FcαRI we have generated mutant human IgA1s with substitutions at these residues. The mutants, expressed in Chinese hamster ovary (CHO) K1 cells, have been purified and assayed for their ability to bind FcαRI on human neutrophils. We report that mutation of these acidic residues appears to have, at most, only a minor impact on FcαRI binding. Thus charge-matching, mediated via Glu254, Asp255 or Glu437, does not appear to play a major role in the interaction of IgA with FcαRI.

Figure 1.

Amino acid alignment of human IgA1, human IgA2 of allotypes IgA2m(1) and IgA2m(2) and bovine IgA in the two interdomain loop regions. Sequences are taken from the translations of nucleic acid sequences using the following accession numbers: J00220, human IgA1; J00221, human IgA2m(1) and IgA2m(2); and AF109167, bovine IgA. The sequences are numbered according to the commonly adopted scheme, originally used for IgA1 Bur.²⁹ The larger boxes contain the conserved motifs implicated in FcαRI binding. The smaller boxes indicate the acidic residues lying close to these motifs, which were mutated in this study.

Materials and methods

Generation of mutant IgA1 expression vectors

Recombinant IgA1 vectors with mutations in the Fc region were prepared by polymerase chain reaction (PCR), either with or without overlap extension¹⁹ using the plasmid pMB2 containing the wild-type human α1 heavy chain sequence as template DNA, as described previously.²⁰ For the overlap extension the 5′ flanking primer (5′GCGCGCGCGGATCCGGTCCAACTGCAGGC3′) annealed around 140 bp 5′ of the start of the Cα1 domain sequence and incorporated a BamHI restriction site (italics) to facilitate cloning of the PCR product. The 3′ flanking primer (5′-CGCCAGCAACGCGGCCCGAGGTCGA-3′) annealed 3′ of a unique SalI site in the α1 vector. In each case, the mutated PCR products were ligated into unique BamHI and SalI restriction sites in the expression vector, replacing the wild-type sequence in that region.

In mutant E254A (using mismatch primer 5′-ACCGGCCCTCGCGGACCTGCTCT-3′ and its complement) a GAG to GCG substitution changed Glu254 to Ala, while in mutant E254L (using mismatch sense primer 5′-GACCGGCCCTCCTGGACCTGCTCT-3′ and its complement) this same codon was changed to CTG coding for Leu. For mutant E437A, mismatch primer 5′-GGTGGGCCACGCGGCCCTGCCGC-3′ and its complement replaced wild-type GAG with GCG, thereby encoding Ala instead of wild-type Glu at residue 437.

In an alternative approach, mutants D255A and D255V were generated using primers 5′-ACCGGCCCTCGAGGCCCTGCTCTT-3′ and 5′-ACCGGCCCTCGAGGTCCTGCTCTT-3′, respectively (each incorporating a XhoI site in italics) as 5′ primers in a single round of PCR with the same 3′ (flanking) primer as above. The XhoI–SalI fragments generated were used to replace the wild-type sequence in that region.

Each resultant plasmid was sequenced on an automated sequencer to confirm the presence of each mutation.

Expression and purification of mutated IgA1 antibodies

CHO K1 cells stably transfected with an appropriate mouse λ light chain²¹ were transfected with the plasmid vectors for each of the mutated IgA1 antibodies and positive transfectants selected as described previously.²¹ Clones secreting high levels of mutated recombinant antibody were identified by an enzyme-linked immunosorbent assay measuring binding to the antigen NIP (3-nitro-4-hydroxy-5-iodophenylacetate) as described previously²¹ before they were expanded into large cultures. The recombinant mutated antibodies were purified from transfectant supernatants by affinity chromatography on NIP-Sepharose as described previously.²¹ The purified antibodies were supplemented with 0·1% sodium azide and stored in small aliquots at −20°.

Rosette formation

Human erythrocytes were derivatized by incubation in isotonic borate buffer pH 8·5 containing 100 µg/ml NIP-caproate-O-succinimide (Genosys, Cambridge, UK) for 1 hr at room temperature. The cells were then washed five times with phosphate-buffered saline (PBS) before and after fixing for 30 min with 3% glutaraldehyde. The NIP-derivatized erythrocytes were incubated with varying amounts of antibody in PBS in a total volume of 200 µl overnight at 4° and then washed three times in PBS. Coating levels for each antibody were found to be equivalent as assessed by reactivity with rabbit F(ab′)₂ anti-human IgA–fluorescein isothiocyanate conjugate (Dako, Ely, UK) analysed by flow cytometry. Rosetting of sensitized erythrocytes to neutrophils, isolated from the heparinized blood of healthy individuals, was performed as described previously.¹⁶ The percentage of neutrophils forming rosettes, where a rosette was defined as a fluorescent neutrophil with three or more attached erythrocytes, was determined. The assays were performed in duplicate with neutrophils from two different donors.

Results

Expression of IgA1 mutants

In our mutagenesis we chose to substitute each of the three acidic residues investigated with Ala. Alanine possesses a small side chain that lacks the negative charge of the original Glu and Asp side chains at physiological pH. It therefore represents a significant change to the character of the residue at that position in the heavy chain. We also mutated Glu254 to Leu because the latter amino acid has a side chain of similar size to that of Glu that is uncharged at physiological pH. Similarly we substituted Val for Asp255, because valine's side chain, while uncharged at physiological pH, has a reasonably similar bulk to that of aspartate. Using this range of substitutions we aimed to test whether the size and particularly the charge of these acidic amino acids might provide a contribution to the interaction between FcαRI and IgA. We reasoned that if the charge of the acidic residues was critical for interaction with the receptor, the substitutions should render the mutant IgAs significantly less able to mediate rosette formation with neutrophils.

DNA sequence analysis confirmed the incorporation of the desired mutations into the heavy chain expression vectors. After expression in CHO K1 cells, the antibodies were purified by hapten affinity chromatography before quantitation and use in the rosette assays.

Functional analysis of mutant antibodies

We used a well-established rosette assay to assess the interaction of neutrophil FcαRI with IgA coated onto the surface of erythrocytes. These sensitized red cells represent physiologically relevant mimics of antibody-coated target cells. Rosette formation with wild-type IgA1 is presumably mediated via constitutively expressed FcαRI because very few if any rosettes formed in the absence of coating IgA. We find the rosette assay sufficiently sensitive to detect apparent differences in binding affinity of the order of twofold.¹⁶ In our earlier studies some point mutations in the Fc domain interface (e.g. L257R, P440A, A442R, F443R, and LA441-442MN), have resulted in complete ablation of rosette formation, and this has been interpreted as an indication that these residues play a critical role in interaction with FcαRI.¹⁶

We found that mutation of Glu254 to either Ala or Leu had no effect on the ability of IgA to form rosettes (Fig. 2). Similarly, mutation of Glu437 to Ala appeared to have no significant effect on rosette formation (Fig. 2). Mutation of residue Glu255 to either Ala or Val resulted in a slightly reduced ability to form rosettes. Half maximal rosette formation was achieved at coating levels of around 100 µg/ml (Fig. 2), consistent with an affinity approximately two- to threefold less than that of wild-type human IgA1. These results contrasted sharply with those obtained previously, where the majority of mutations in the Leu257–Leu258 and Pro440–Phe443 loops resulted in the loss of ability to bind the receptor on neutrophils.¹⁶ No major impact on rosette formation was observed upon point mutation of the group of acidic residues tested here, suggesting that these residues probably do not play major roles in the IgA–FcαRI interaction.

Figure 2.

Binding of IgA1s to FcαRI on human neutrophils assessed by rosette formation. The results are normalized by expressing specific rosette formation as a percentage of the maximum seen with wild-type IgA1. The results from two separate experiments are shown. The lower curve represents the binding curve for D255A and D255V, which is shifted the right of that representing the binding of the other antibodies.

Discussion

Despite the prevalence of IgA in the body and mounting evidence to support a critical role in immune defence, understanding of the molecular basis of IgA effector function has lagged behind that of other immunoglobulin classes. Previously two interdomain loops in the Fc region of IgA, Leu257–Leu258 and Pro440–Phe443, have been shown to be critical for FcαRI binding^15,16 and subsequent triggering of a respiratory burst in neutrophils.¹⁶ The loops are predicted to lie close to each other most likely forming a single surface with which the receptor EC1 domain might dock (Fig. 3). Others have mapped the interaction site on the receptor domain, highlighting contributions from, amongst others, a number of basic residues including Arg 82, Arg 52 and to lesser extent His85.^5,7 This finding has prompted the suggestion that the association of these basic residues on FcαRI with acidic residues on IgA might contribute a significant electrostatic element to the free energy of binding.⁵

Figure 3.

Molecular model of the Fc region of human IgA1 (using coordinates from Boehm et al. (1999);³⁰ in Brookhaven Protein Data Bank as PDB code 1iga) showing residues previously implicated in interaction with FcαRI and those mutated in this study. For clarity a partial view of IgA1 is shown in which partial hinge regions are shown leaving the upper sides of the image and the two Fab arms (not shown) are situated one to each side of the image. One heavy chain is shown in orange and the other in blue. Residues previously shown to be critical for interaction with FcαRI are shown in pink, and acidic residues mutated in this study are shown in white.

Here we sought to investigate the possible role of charge matching in the interaction through targeted mutagenesis of acidic residues lying close to the IgA Fc domain interface. We therefore mutated residues Glu254, Asp255 and Glu437, which are predicted to lie in close proximity to the interdomain region (Fig. 3). We found that mutations designed to remove the charged nature of these residues appear to have little effect on the ability of the resultant antibodies to interact with FcαRI. It is necessary to consider that there is another close-lying Glu residue (Glu261) that has not been tested and we cannot rule out that it may make some contribution to FcαRI binding. However, bovine IgA, which binds human FcαRI with only slightly lower affinity (around 2·5 fold) than that of human IgA1,¹⁶ has Asn, which may be N-glycosylated, at this position (Fig. 1) suggesting that any charge matching contribution by this residue to the free energy of binding is likely to be minimal. Overall therefore it appears that electrostatic interactions are likely to play at most only a limited role in the binding event.

The above conclusion is in keeping with a recent report based on biosensor binding experiments using recombinant IgA1Fc and soluble FcαRI comprising EC1 and EC2 domains.¹⁸ The binding constants for the interaction were found to have very little dependence on the concentration of NaCl in the buffer, indicating that electrostatic interactions do not contribute significantly to the free energy of the binding of FcαRI to IgA1 Fc.¹⁸

In addition to the above conclusions the findings presented here add weight to the proposal that residues Leu257–Leu258 and Pro440–Phe443 play significant roles in the IgA–receptor interaction. The fact that mutation of these loop residues resulted in complete loss of FcαRI binding while mutation of adjacent residues Gly259,¹⁶ Glu254 and Glu437 (this paper) had no effect, would seem to argue against the idea that mutations in the loop residues merely trigger alterations in the conformation of close-lying residues that provide the direct receptor binding contacts. Instead it tends to argue for the loop residues mediating direct interactions with the receptor. In any case, the Leu257–Leu258 and Pro440–Phe443 loops remain important markers of the FcαRI binding site.

Information on the location of effector sites in antibodies is clearly important for the design of recombinant antibodies optimized for particular therapeutic applications. For example, IgG molecules featuring specific mutations to prevent FcγR binding are proving particularly valuable in the treatment of autoimmune diseases and transplant rejection.^22–26 With increasing interest in the therapeutic potential of IgA^27,28 it is important to have a thorough understanding of how this antibody mediates its function. The mutagenesis study detailed here provides further information on the structural features of IgA critical for interaction with FcαRI.