IgA Binding Characteristics of Golden Hamster and Ferret Fcα Receptors

Abstract

Golden hamster (Mesocricetus auratus) and ferret (Mustela putorius furo) are important animal models in studies of human infectious disease. They are used widely to investigate pathogen spreading mechanisms, host immunology, and to evaluate the safety and efficacy of small molecule and biologic drugs and vaccines. To this end, immunoglobulin A (IgA) and its Fcα receptor (FcαR) play critical roles in humans but are not well characterized in these two species. Golden hamster and ferret IgA and FcαR were recombinantly expressed, purified, characterized for N-linked glycosylation site occupancy, and binding affinity. Based on sequence and structural alignments, hamster IgA showed greater similarity to human IgA than did ferret, and hinge domains in both small animal models suggested greater structural homology to human IgA2 than IgA1. Despite considerable sequence divergence in both immunoglobulins and receptors, and the lack of binding between ferret FcαR and ferret IgA, human IgA bound to both hamster and ferret FcαR with high affinity. Further, differences in dissociation rates were dependent on test format suggesting that the 2:1 stoichiometry of human FcαR:IgA is recapitulated in these animals. Overall, this work suggests the suitability of these animals to model protection or pathology driven by interactions between human IgA and host FcαR and will aid in critical and confident interpretation of infection and immunization studies in each species.

Introduction

In humans, immunoglobulin A (IgA) accounts for ~10-20 % of total serum antibody after the more abundant IgG (70-75 %); however, at 70-95% depending on location and inflammatory state, it is the most abundant antibody type at mucosal sites, where it typically presents as secretory IgA (sIgA) (1-3). Lacking a recycling receptor to prolong its plasma half-life, human IgA has a faster catabolic rate (4-6 day half-life) in serum than IgG (~21 day half-life) (4, 5). Serum IgA plays an important role in eliminating pathogens by initiating inflammatory responses through Fcα receptor- (FcαR, CD89) mediated cell responses, antibody-dependent cellular cytotoxicity (ADCC), phagocytosis (ADCP), and NETosis, but not the classical complement pathway (6). Post transport to the mucosa, and despite resistance to proteases and better stability in harsh environments like the intestinal lumen and respiratory tract, sIgA only has a half-life of hours to two days (1, 7). As a result of its rapid systemic and mucosal clearance, it is estimated that over 4 g of IgA is produced and secreted daily in adults (8). In combination with the many microbial factors designed to interfere with IgA (9), this impressive output identifies IgA a key player in the first line of defense at mucosal surfaces, where it interacts with both friendly bacteria and harmful pathogens (10). As such, it is an important contributor to homeostasis and host defense not only of humans, but in species ranging from reptiles and birds to mammals (11).

Human FcαRI, found on myeloid cells such as neutrophils, eosinophils, monocytes/macrophages, dendritic cells, and Kupffer cells (3), has two extracellular immunoglobulin-like domains, a transmembrane region containing a charged arginine residue that enables association with the FcR γ-chain, and a short cytoplasmic tail devoid of intrinsic signaling motifs (12). Human FcαRI binds both IgA1 and IgA2 subclasses with low affinity (10⁻⁷ M) (13), and can drive either pro- or anti-inflammatory responses depends on whether it is crosslinked by IgA-immune complexes or bound to monomeric IgA (12). Multimeric FcαRI crosslinking induces ADCC, ADCP, reactive oxygen species production, antigen presentation, and cytokine release, while monomeric interaction of IgA with FcαRI triggers inhibitory signals that can downregulate IgG- or IgE-Fc receptor mediated phagocytosis, oxidative burst activity, chemotaxis, and cytokine release (12). IgA-associated diseases in humans, including IgA nephropathy, rheumatoid arthritis, vasculitis, and skin blistering diseases, are all associated with aberrant stimulation of FcαR (14).

However, expression of FcαR is not uniform across species. While present in humans, nonhuman primates (NHP), cattle, and horses, it is notably absent in mice and dogs (3). As a result, investigation of IgA-FcαR mediated immune responses has mainly been conducted in humans and NHP. With the exception of the hinge region, NHP IgA is highly conserved (15, 16), and NHP FcαR exhibits comparable affinity to its own and human IgA subclasses (15). Other species present greater and sometimes surprising divergence: bovine FcαR is able to bind its bovine and human IgA, while equine FcαR can bind human but not equine IgA (17). Differing FcαR-IgA binding profiles across species are thought to have arisen in association with competition with microbes (18). The absence of FcαR in mice has led to creation of CD89-transgenic mice to permit modeling of IgA-mediated human diseases and immunotherapies (19-21). In contrast, important small animal models such as hamsters and ferret exhibit robust IgA responses, including at mucosal sites, but remain poorly characterized with respect to FcαR engagement, even though they are widely used in respiratory infection studies and active and passive vaccine development (22-24). In this study, we report the complete sequence analysis, recombinant expression and affinity characterization of golden hamster and ferret FcαR and IgA—filling a translational gap in our understanding of IgA biology in these two model animals and their relevance to human health.

Results

Identification and phylogenetic analysis of golden hamster and ferret FcαR

FcαR and Fcα/μR (FCAMR or CD351) are the two main IgA receptors characterized in humans. The sequence of the Human FcαR (CD89) (Table S1) was used to search for homology within the published golden hamster and ferret (25) genomes, resulting in identification of hamster ortholog of FcαR (XP_021091006.1) and ferret FcαR (XP_012906223.2). Hamster and ferret FcαR exhibit similar putative domain architecture as human and rhesus FcαR (Figure 1A), which have two extracellular domains, and can signal through the common γ chain. Amino acid sequence identity with human FcαR is limited to 54.8 % for hamster and 39.4 % for ferret (Figure 1B). As compared to other common model organisms, this extent of sequence identity is lower than that observed for rhesus macaque (85.3 %), but greater than for mice, which lack an FcαR ortholog.

Figure 1. Domain structure and sequence homology of hamster and ferret FcαR.

A. Schematic illustrating the domain architecture of FcαR in humans and common mammalian model organisms. FcαR extracellular domains (ECD) are indicated as ECD1, and ECD2. Signaling motifs include immunoreceptor tyrosine-based activation motifs (ITAM, green). The γ-chain serves as a common signaling component. B. Percent identity observed among FcαR in human, rhesus, hamster and ferret. C. Phylogenetic tree depicting the evolutionary relationships among FcαR sequences in mammals. Bootstrap support values are indicated at the nodes. The tree is rooted using an outgroup, which is not displayed here.

As a second receptor for IgA, the ferret genome also presents a putative Fcα/μR (CD351) (XP_044942916.1). Though less well-studied than FcαR, human Fcα/μR binds with high affinity to both IgA and IgM(26). While golden hamster Fcα/μR could not be identified with sufficient confidence based on the published genomic data, that of the Chinese hamster was identified. This diversity in putative IgA receptors within the hamster family is mirrored by the relatively high divergence of FcαR of these two hamster species, which present only 83.4 % identity.

Broader phylogenetic analysis of both IgA receptor families makes clear the distinct grouping of FcαR from Fcα/μR (Figure 1C, Figure S1). Golden hamster FcαR grouped confidently with its close relatives Chinese hamster and rat, and human FcαR grouped well with the non-human primates; however, ferret FcαR placement in the dendrogram was less stable, here grouping with rat and hamsters is moderately supported (bootstrap = 78) (Figure 1C). Consistent with the lack of identification of an FcαR ortholog in mice, which has been attributed to loss during a gene translocation event (27), more species presented identified Fcα/μR than FcαR sequences (Figure S1 and Table S1). For example, dog and dingo Fcα/μR grouped with ferret Fcα/μR with 100 % bootstrap support (Figure S1), but neither species has yet had an FcαR ortholog identified or annotated.

Sequence alignment, structural modeling, and N-glycosylation analysis of ferret and golden hamster FcαRs

Sequence alignments across human, rhesus, hamster, and ferret FcαR extracellular domains demonstrate conservation (Figure 2A). Of the 13 predicted hamster FcαR contact residues based on the human FcαRI:IgA1 co-crystal structure (PDB: 1ow0) (28), five are identical to human FcαR (Figure 2A, 2B and 2D). In contrast, only two of 13 predicted ferret FcαR contact residues are identical to human FcαR (Figure 2A, 2C and 2D). For both species, many positions vary both within and outside of the putative IgA contact surface and across both domains (Figure 2B-C). Despite sequence diversity, models of hamster and ferret FcαR superimposed on the human IgA-FcαR complex structure showed some degree of consistency on contact residue side chain positioning. After pruning atom pairs that were poorly matched or outlier pairs (29, 30), root mean square deviation (RMSD) of 0.908 Å out of 154 atom pairs and 1.192 Å of 108 atom pairs were observed for hamster and ferret FcαRs, respectively (Figure 2D).

Figure 2. Sequence and structural analysis of hamster and ferret FcαR.

A. Sequence alignment of human, rhesus, hamster and ferret FcαR extracellular domains. Major domains (ECD1, ECD2) are denoted in blue, putative N-linked glycosylation sites in purple and numbered, and IgA contact residues in cyan. B-C. Structural models of human FcαR (blue) in complex with human IgA (gray) (PDB: 1ow0). The side chains of amino acids that exhibit variation between human and hamster (B) and human and ferret (C) FcαR within (cyan) and outside (red) the IgA contact interface are illustrated on the FcαR ribbon backbone. Putative N-linked glycosylation sites are denoted in purple. D. Contact region of human FcαR (blue) and IgA Fc (grey), superimposed with the predicted structure of hamster (top) or ferret (bottom) FcαR (green). E. N-glycan occupancy (%) of FcαR on N-linked glycosylation motifs. *Human FcαR data from (76).

Four predicted N-linked glycosylation motifs (NXS/T, where X is not P) were identified in hamster and two in ferret as compared to the six in the human FcαR extracellular domain (Figure 2A). Among these, one site (numbered motif 6) in domain 2 is common to all three species (N₁₁₉IS in human and N₁₁₇IS in hamster, N₁₁₆VT in ferret) and another site in domain 1 (motif 2) is shared between hamster (N₄₃ST) and human (N₄₄ST); others are unique (Figure 2E). While all motifs are outside of predicted contact regions, N₅₈ET (motif 3), in the human receptor is adjacent (Figure 2A). Analysis of site-specific N-glycan site occupancy by mass spectrometry demonstrated a high degree of occupancy for three of four hamster FcαR sites and both ferret FcαR sites (Figure 2E, Table S3). In contrast, N₁₁₇IS (motif 6) in hamster was found to exhibit only ~20 % glycan occupancy.

Sequence alignment and structural modeling of golden hamster and ferret IgA

The fundamental immunoglobulin structure and Fc region are kept relatively consistent across mammalian species, with more variation in the hinge and CH2 domains (11, 31). Between the two human IgA subclasses, the hinge region of IgA1 is longer and has several O-glycosylation sites, while IgA2 hinge region is shorter, more compact and more resistant to bacterial proteases (32). The ratio of these subclass changes depending on site in humans: IgA1 is more common in serum and upper respiratory secretions (up to 96 %), while IgA2 is found in increased concentrations in the lower gut (up to 60 %) (33)—a distribution thought to reflect adaptations to the increasing microbial density and protease activity of the colon (1, 33).

The constant region of hamster (WEL32140.1) and ferret (34) IgA (Table S4) were aligned with those of rhesus and mouse and the two distinct human IgA allotypes (IgA1, IgA2m(2)) (Figure 3A). Like those of rhesus and mouse, the shorter hinge of ferret and hamster IgA suggests greater structural homology to human IgA2 than human IgA1, though overall percent identity values are essentially equal (Table S5).

Figure 3. Sequence and structural analysis of hamster and ferret IgA.

A. Sequence alignment of human, rhesus, hamster, ferret, and mouse IgA Fc domains. Human IgA CH1, Hinge, CH2 and CH3 domains are denoted in blue, putative N-linked glycosylation sites in purple, and IgA contact residues in cyan. B-C. Structural model of human FcαR (gray) in complex with human IgA Fc (blue) (PDB: 1ow0). The side chains of amino acids that exhibit variation between human and hamster (B) or human and ferret (C) within (cyan) and outside (red) the contact interface are illustrated on the human IgA Fc ribbon backbone. Putative N-linked glycosylation sites are denoted in purple. D. N-glycan occupancy (%) of IgA heavy chain on N-linked glycosylation motifs. *Human IgA1 and IgA2 data from (75). E. Top-down view of hamster (top) and ferret (bottom) IgA Fc in space fill with the FcαR contact surface bounded by yellow lines and residues colored as in B-C.

As defined by the human FcαRI:IgA1 co-crystal structure (28), hamster differs from human IgA at four putative FcαR contact residues and ferret IgA differs at seven. (Figure 3A). Substitutions outside of putative contact residues are numerous (Figure 3A-C). N-linked glycosylation motifs vary in number from two (human IgA1), to six (human IgA2m(2)) for hamster (Figure 3A, 3D and Table S6). Among glycosylation motifs in hamster IgA, only N₄₅₂VS (site 9) at the end of the heavy chain in shared with human IgA, exhibiting 41% glycan occupancy. N₁₅₈VS (motif 1) has 82 % glycan occupancy, while N₄₂₂YS (motif 7) was only about 1 % occupied. Ferret N₃₃₇IS (motif 6) and N₄₆₁VS (motif 9) that are shared with human IgA2, are 15 % and 63 % glycan occupied, respectively, while the two sites different to human IgA2, N₂₆₁GS (motif 4) and N₄₄₄FT (motif 8) are highly glycosylated (Figure 3A and 3D). Intriguingly, both mouse and ferret IgA bear an N-linked motif in a putative FcαR contact residue (N₄₄₄FT, motif 8) (Figure 3A and 3D). Whereas nine O-linked glycosylation sites are present in the human IgA1 hinge, six of which have been reported to be occupied (35), these sites are not present in human IgA2 or the IgA of other species.

A top-down view of the IgA Fc contact region with FcαR shows a number of substitutions along the periphery of the FcαR binding interface (Figure 3E). At about 59% overall sequence identity, ferret IgA is slightly more similar than hamster (about 55 %) to human IgA, despite apparently more divergent binding region to FcαR. Also, interestingly, N glycan site N422 (motif 7) that next to the deduced binding pocket of hamster IgA to FcαR only has 1% glycan occupancy, but N261 (motif 4) (close to binding pocket) and N444 (motif 8) in the binding pocket of ferret IgA are both highly glycosylated. Overall, the degree of variation in IgA glycosylation between these species (Figure S2), stands in stark contrast with their single, perfectly conserved IgG Fc domain glycosylation site.

Affinity characterization of golden hamster FcαR and IgA

To profile the recognition of hamster IgA and FcαR, a series of biolayer interferometry (BLI) and surface plasmon resonance (SPR) experiments were conducted (Figure 4 and S3). For BLI, the hamster FcαR was immobilized to the BLI tips through a C terminal biotin, mimicking the natural presentation of FcαR extracellular domains on the surface of a cell (Figure 4A). When expressed recombinantly, hamster IgA tended to form dimers when expressed in CHO cells, a form that also exhibited high affinity to hamster FcαR; while in 293F cells, monomers were dominant (Figure S4). Binding to monomeric hamster IgA in solution was measured and compared with human IgA1 (Figure 4B) and human IgA2 (Figure 4C). Hamster FcαR showed the expected high affinity binding profile for each of these analytes (K_D 3.0 nM, 2.6 nM, and 2.7 nM, respectively), and both human IgA subclasses in particular showed relatively slow dissociation rates, suggesting that this test format permitted avid bivalent capture of individual IgA molecules (Figure 4D, top), as would be expected from the 2:1 (FcαR:IgA) stoichiometry of this interaction in solution and crystal structures (28).

Figure 4. Affinity of hamster FcαR for hamster and human IgA.

A-C. (Top) BLI sensorgrams (solid lines) and fitting curves (dashed lines) show the interactions of recombinant hamster FcαR with hamster IgA (black, A), human IgA1 (magenta, B) and human IgA2 (purple, C). (Bottom) Equilibrium fits. Binding affinities (K_D, M) and corresponding errors for are listed in inset. D. Assay schematic. For BLI, FcαR was immobilized as the ligand on the biosensor probe and IgA flowed as analyte (top). For SPR, IgA was conjugated to the prism surface and FcαR was flowed as analyte. E-G. (Top) Exemplary SPR association and dissociation profiles for hamster FcαR binding to hamster IgA (E), human IgA1 (F) and human IgA2 (G). (Bottom) Equilibrium fits. Binding affinities (K_D, M) and for each are listed in inset. H-I. Kinetic (H) and equilibrium (I) binding affinities (K_D, M) of hamster FcαR binding hamster IgA (black), human IgA1 (pink), and IgA2 (purple) across replicates. Bars indicate mean and standard deviations.

In multiplexed SPR, the assay format was inverted: IgA was covalently immobilized onto a SPR chip surface through conjugation with multiple technical replicates (Figure 4D, bottom). The SPR chip printed with IgA in discrete, replicated spots was loaded with hamster FcαR within the same SPR flow cell, providing for simultaneous evaluation of FcαR affinity for each IgA. In this format, the high affinity of golden hamster FcαR for golden hamster IgA (Figure 4E), human IgA1 (Figure 4F), and human IgA2 (Figure 4G) was also clear. Excellent agreement was observed across replicates for both kinetic (average K_D 435.7 nM, 49.7 nM and 79.3 nM, respectively, Figure 4H and S3) and equilibrium (average K_D 433.3 nM, 72.7 nM and 132.3 nM, respectively, Figure 4I and S3) fits for hamster IgA and human IgA1, and IgA2. Consistent with the inability to achieve an avid interaction, the off-rates observed in this format were considerably faster, with similar trends of faster dissociation for hamster than human IgA as observed by BLI.

Affinity characterization of ferret FcαR and IgA

To profile the recognition of ferret IgA and FcαR with greater resolution, similar BLI and SPR experiments were conducted (Figure 5). When ferret FcαR was immobilized to the BLI tips, binding affinities of monoclonal ferret IgA with ferret FcαRI were measured and compared with human IgA1 and IgA2 (Figures 5 and S5). Whereas no binding was detected between ferret FcαR and recombinant ferret IgA, either expressed with a lambda or kappa light chain (Figure 5A, S5), ferret FcαR showed high affinity binding to human IgA1 (Figure 5B) and IgA2 (Figure 5C). Human IgA2 exhibited a slightly higher affinity than human IgA1 (K_D 1.4 nM and 2.2 nM, respectively); both showed slow dissociation rates, again consistent with the possibility of avid binding of the symmetric sites of one IgA molecule to two immobilized FcαR molecules (Figure 4D, top).

Figure 5. Affinity of ferret FcαR for ferret and human IgA.

A-C. Binding profiles with FcαR immobilized on surface. Binding (top) BLI sensorgrams (solid lines) and fitting curves (dashed lines) show the interactions of recombinant ferret FcαR with recombinant ferret IgA (kappa light chain, black, A, top) native ferret IgA (A, bottom), human IgA1 (magenta, B) and human IgA2 (purple, C). (Bottom right) Equilibrium fits with resulting binding affinities (K_D, M, B-C) and corresponding errors listed in inset. D-H. Binding profiles with IgA immobilized on surface (dotted green box). D-F. Exemplary SPR association and dissociation profiles for ferret FcαR binding recombinant ferret IgA (top) and native ferret IgA (bottom), human IgA1 (D) and IgA2 (E) in fits for kinetic (top) and equilibrium (bottom) binding affinity. G-H. Kinetic (G) and equilibrium (H) binding affinities (K_D, M) of ferret FcαR binding human IgA1 (pink), and human IgA2 (purple). Bars indicate mean and standard deviations among replicates.

The lack of binding of ferret FcαR to recombinant ferret IgA was unexpected given its ability to bind to human IgA and suggested a potential defect in recombinant ferret IgA. This possibility was further investigated by comparison to serum-derived, native ferret IgA, which also failed to demonstrate robust signal (Figure 5A). Because N-linked glycosylation of ferret IgA, especially at N261 (motif 4) and N444 (motif 8), near/in the deduced binding regions might impair its binding to ferret FcαR, PNGase-F digested IgA was tested. However, the resulting deglycosylated IgA showed minimal evidence of binding (Figure S6). Lastly, these glycosylation sites were genetically eliminated; ferret IgA amino acid point mutants N261E, N444A, and double mutant N261E/N444A were expressed, and the binding affinity of these mutants to ferret FcαR were evaluated by BLI. Still, no binding between ferret FcαR and three ferret IgA mutants was detected (Figure S7).

In multiplexed SPR, the assay format was inverted: IgA molecules of interest were covalently immobilized onto a SPR chip surface through conjugation via primary amines with multiple technical replicates (Figure 4D, bottom). The SPR chip containing IgA printed in discrete, replicated spots was simultaneously loaded with ferret FcαR within the same SPR flow cell, providing for coincident evaluation of FcαR affinity for each IgA. In this format, which does not permit avid binding and thus assesses the intrinsic affinity of a single binding site, considerably faster off rates of ferret FcαR for both human IgA1 (Figure 5E) and IgA2 (Figure 5F) were clear. Again, no binding to ferret IgA was observed (Figure 5D). Excellent agreement was observed across replicates calculated based on kinetic (average K_D 235.3 nM and 150.8 nM, respectively, Figure 5G) and equilibrium (average K_D 285.0 nM and 155.0 nM, respectively, Figure 5H) fits, and this testing confirmed the somewhat higher affinity of human IgA2 than human IgA1 observed in BLI.

Discussion

In this study, we have characterized recombinant golden hamster and ferret FcαR and IgA through sequence alignments, affinity assays, and site-specific glycosylation analysis. These results provide genetic, structural, and biophysical analysis of golden hamster and ferret FcαR and IgA features that underpin species-specific and cross-species Fc-FcαR interactions relevant to the use of these important small animal models in studies of human disease. Overall, despite extensive sequence divergence in both receptor and antibody, including in and around putative contact residues, cross-reactivity among IgA and FcαR between species was often observed – including in one case when recognition within a species was apparently lost.

We found that human IgA1 and IgA2 bind to both hamster and ferret FcαR, suggesting the potential utility of these models in passive antibody transfer studies in which human IgA is administered. Of the two human IgA subclasses, IgA2 presented similar affinities within model organisms. Though more commonly performed in NHP (36) or mice (37), our data provides insight into studies of protection or pathology associated with human IgA in the hamster model (38), and suggests that hamsters, whose FcαR binds human IgA might be superior to mice, which lack expression of an FcαR ortholog, while being more resource efficient than NHP. While we did not identify an example of a study in which human IgA was passively transferred to ferrets, our data likewise supports that FcαR-driven activities may also be recapitulated in this model. In contrast, because ferret IgA was not observed to bind to ferret FcαR, of the two species, hamsters may provide a better suited model of active immunization. Indeed, many studies report immunogenicity and efficacy of systemic and mucosal vaccines in hamsters, including in association with robust induction of IgA (39-47).

In terms of IgA biology, human IgA exists as monomers, dimers covalently linked by J chain, and higher order polymers. Secretory IgA, comprising dimeric IgA complexed with secretory component, the cleaved extracellular domains of the polymeric Ig Receptor (pIgR) following translocation from systemic circulation across the epithelium, plays an important role in limiting the invasion of pathogens in mucosal secretions. In contrast, serum IgA can function through triggering effector functions via binding to FcαR (48). While it has been presumed from structural studies that pIgR blocks access of FcαR (49), we have observed binding of FcαR to IgA in mucosal secretions (50). Human IgA2m(1), IgA2m(2) are well-characterized allotypic variants that differ in how well they form covalent bonds with the J chain and secretory component/pIgR, but present similar affinity for FcαR (51). One limitation of our study is that the rich structural and allotypic variance among human IgA molecules was not profiled here. However, in the absence of J chain, we observed secreted dimers of ferret and golden hamster IgA that showed similar affinity to FcαR as monomers. Multiple IgA subclasses were not identified in these species, consistent with the differing diversification of IgA molecules across species. Based on similarity in the hinge region that accounts for differing human IgA structures, a single, human IgA2-like form of IgA was found in ferret and hamster genomes, as in old world monkeys (such as rhesus macaques and crab-eating macaque), while some great apes have two IgA subclasses and orangutan has IgA1-like isotype (31).

Human IgA glycosylation contributes to structural stability, mucosal persistence, microbial interactions, and modest effects on FcαR binding affinity (3, 52). Glycosylation of recombinant golden hamster and ferret IgA are not conserved to human IgA in terms of site or occupancy, differences that, together with differences in glycan composition, which was not evaluated here, may influence IgA aggregation, tissue localization, or pathogen binding. As for the receptors, human FcαR glycosylation, in contrast, can significantly affect ligand binding (53). The composition of glycans incorporated on recombinant ferret and hamster FcαR, or consistency with native receptor, was not evaluated here, but may influence affinity and activity. FcαR is known to be secreted in human serum, in two forms—one slightly glycosylated (30 kDa) (54), and the other presenting with more extensive glycosylation modification (50-60 kDa), which together comprise FcαR-IgA complexes and mediate IgA nephropathy (Berger's Disease) (55). Whether secretion or varying glycosylation features are conserved in golden hamster and ferret FcαR remains to be determined.

With the exception of ferret IgA, which lacked binding to ferret FcαR or those from other species, common themes emerged from our affinity analysis. Human IgA1 and IgA2 presented similar affinities, and testing in the BLI format resulted in higher apparent affinity in association with slower off-rates. We hypothesize that this discrepancy results from the ability of the IgA in solution to bind avidly to the FcαR surface in the BLI format. Human FcαR binds IgA symmetrically through IgA CH2-CH3 region and FcαR extracellular domain 1 with a stoichiometry of 2 FcαR: 1 IgA (28). In contrast, the stoichiometry IgG to FcγR is 1:1, and divergent off-rates between these BLI and SPR test formats was not observed (56-58). At present, we lack direct structural data, but expect that hamster and ferret FcαR: IgA stoichiometry follow the 2:1 human FcαR and IgA binding ratio. It has been reported that immobilizing human FcαR or IgA can result in discrepant K_D values (13), thought to be driven by bivalent binding of IgA Fc to immobilized FcαR, results in much lower k_off (59, 60).

Four other IgA receptors are known in humans: the polymeric Ig receptor (pIgR), is involved in transport of IgM and polymeric IgA across epithelial barriers (61), Fcα/μR that can bind both IgM and IgA, the asialoglycoprotein receptor, and the transferrin receptor ⁽³⁾. Among these, FcαR expression is limited to myeloid cells and is capable of strong inflammatory responses (62). Fcα/μR, binds IgM with high affinity and polymeric IgA with lower affinity, and is better conserved through evolution, found in a range of vertebrates, including mammals and amphibians, which hints at ancient origins in adaptive immunity (63). It is found on mature B lymphocytes, follicular dendritic cells and certain macrophage subsets, but not on immature B cells or circulating B cells in the blood. Fcα/μR can mediate the endocytosis of immune complexes for clearance or antigen presentation but doesn’t initiate the direct cytotoxic responses associated with human FcαR (63-65), suggesting that it may play an important role in immune regulation and the bridging of innate and adaptive defenses. This could be the main IgA receptor for mucosal defense in animals that lack FcαR.

Similarities and differences among the species in these IgA receptors and their interactions with IgA were not evaluated. Other limitations include the recombinant expression of ferret and hamster FcαR. Neither receptor expression patterns nor signaling properties were evaluated. Lastly, a molecular explanation for the loss of binding between ferret FcαR and ferret IgA was investigated but not identified. While post-translational modification of ferret IgA in and near putative contact residues at the CH2-CH3 interface, known to be responsible for the compromised FcαR recognition of mouse IgA (66) was evaluated, it was not found to be responsible for the loss of the binding to ferret FcαR, Numerous other amino acid sequence differences exist, including the insertion of a glycine residue immediately preceding putative contact residues, but the impacts of these changes were not evaluated. Intriguingly, our data are consistent with the model proposed by Abi-Rached et al. (18) and supported by Wines et al. (66), in which counter-evasion of pathogen-derived IgA evasion factors, such as the decoy IgA receptor SSL7 from Staphylococcus aureus, drives host IgA diversification that can result in loss of FcαR binding. FcαR, then, must mutate in response to restore binding. When host IgA and FcαR coevolution is outpaced by pathogen adaptation, their interaction may be lost, in turn resulting in receptor loss from the genome, as seen in mice and rabbits. It is tempting to speculate that this process has yet to fully play out between receptor adaptation and loss in ferrets, and that the more extensive variation in IgA glycosite and IgA subclass variation may relate to pressure mediated by Ig proteases and receptor decoys (14, 67).

Overall, the divergence in IgA structure, sequence, and potentially function, coupled to the presence or absence of FcαR provides evidence as to the impact of evolutionary pressure across species. Given the many microbes that express IgA proteases and binding proteins (32), these differences may reflect unique host-pathogen adaptions (18). In this context, reliable interpretation of results from small animal models of human disease relies on knowledge of functional homology and distinctions among the species. The similar affinity of hamster and ferret FcαR for human IgA suggests that passive immunization with human IgA may result in similar activity as endogenous IgA. However, the apparent absence of a functional FcαR in ferrets suggests that active immunization, particularly in the context of mucosal vaccines, which elicit a greater IgA response, may rely on other IgA receptors, or fail to recapitulate aspects present in humans. We hope that this work also provides insight into the potential utility of these model systems in the study of IgA mediated diseases. In sum, while questions remain about various aspects of even human IgA and mucosal immunology, greater insight into the animals used to model human disease and therapies promises to result in more confident interpretation and translation.

Materials and methods

Receptor and IgA identification, homology assessment, and structural modeling

Ferret (Mustela furo) and golden hamster (Mesocricetus auratus) FcαR and Fcα/μR orthologs were identified by homology to human FcαR. Ferret (XP_012906223.2), and hamster (XP_021091006.1) FcαR candidates and ferret Fcα/μR (XP_044942916.1) were selected for further analysis.

Ferret IgA Fc region sequences were identified by B cell sequencing(34), while a Fab region was selected from the IMGT database (Table S3). The hamster IgA heavy chain CH1 to CH3 (WEL32140.1) and light chain CL (XP_040607057.1) domain sequences were identified through NCBI. VH and VL regions for a murine TCR-specific antibody were based on a prior report (Table S3) (68). As defined by the human FcαRI:IgA1 co-crystal structure (PDB: 1ow0)(28),

Multiple sequence alignments of selected FcαR, Fcα/μR, and IgA sequences were performed by Clustal Omega (69). IQ-tree were used to build the phylogenetic tree by maximum likelihood with 2000 times of bootstrapping (70). Geneious Prime was used to make the alignment figures and calculate the sequence identity.

FcαR and IgA structures were generated using ColabFold v1.5.5: AlphaFold2 ("use_templates": false, "relax_max_iterations": 200, "num_recycles": 3) (71). Multiple Sequence Alignments were generated using MMseqs2. Top ranked structures for both proteins yielded predicted Local Distance Difference Test (pLDDT) values greater than 80. The predicted ferret and hamster FcαR structures overlaid with human FcαR were generated and RMSD values calculated with UCSF Chimera (30). Contact residues between FcαR and IgA Fc were identified using buried solvent accessible surface area ≥15 Ų.

Recombinant protein expression and purification

The coding sequences of both golden hamster and ferret FcαR extracellular domains were modified to exchange their native for the human CD5 signal peptide (MPMGSLQPLATLYLLGMLVASCLG) on the N terminus, and to truncate at the transmembrane domain and add His6 and AVI tags, each preceded by a short GGG linker to the C terminus. Genes were codon optimized against human codons, synthesized (Genewiz), and cloned into the mammalian cell expression plasmid pCMV by HiFi assembly (NEBuilder^® HiFi DNA Assembly). Golden hamster IgA heavy chain CH1 to CH3 domains (WEL32140.1) and light chain CL (XP_040607057.1) sequences were identified through NCBI and VH and VL regions are hamster anti-mouse mAb H57 (68, 72). Ferret IgA expression plasmids were generously shared by the Kent and Wheatley labs (34). A His6 tag was added to the C-terminus of the IgA heavy chains. The genes were also optimized against human codons and synthesized (Genewiz), the inserts were cloned to the mammalian cell expression plasmid pCMV using HiFi assembly (NEBuilder^® HiFi DNA Assembly).

Hamster and ferret FcαR extracellular domains (Table S2), and IgA (Table S4) were produced by transient transfection of Expi293F human embryonic kidney cells (Thermo Fisher Scientific) following the Expi293 expression system user guide (Thermo Fisher Scientific). Protein samples were first purified from culture supernatants by gravity columns using Ni NTA resin (Cytiva). FcαR extracellular domains were further purified using a phosphate-buffered saline (PBS)-exchanged Superdex 75 Increase 10/300 GL size exclusion chromatography column (Cytiva), while IgA was further purified using Superdex 200 Increase 10/300 GL size exclusion chromatography column (Cytiva), both on the ÄKTA pure FPLC system (Cytiva). Protein monomers were used for further analyses. The purified FcαR extracellular domains were site-specifically biotinylated for BLI analysis on the C terminal AVI tag using BirA biotin-protein ligase BirA500 kit (Avidity LLC.). Protein concentrations were quantified by Braford Assay (Pierce). Essentially monoclonal Human IgA1 from human myeloma plasma (16-16-090701-1M, Athens Research & Technology) and IgA2 from human myeloma plasma (PI50013, Assaypro) were sourced commercially. Hamster IgA was also tried being expressed in CHO cells following ExpiCHO expression system user guide (Thermo Fisher Scientific).

Mass Spectrometry Sample Preparation:

Glycan occupancy was determined as previously described (73, 74) and compared to previous reports of IgA and FcαR glycosylation site occupancy (75, 76). Briefly, 50 μg of purified protein was denatured with 50 μL of 100 % 2,2,2-trifluoroethanol (TFE) and 5 μL of 100 mM dithiothreitol (DTT) at 55 °C for 45 minutes. Samples were then alkylated by incubation with 3 μL of 550 mM iodoacetamide (Sigma) for 30 min at RT in the dark. Alkylation was quenched with 892 μL of 40 mM Tris-HCl. Ferret IgA, hamster IgA and hamster FcαR were digested with trypsin at a ratio of 1:30 (w/w) trypsin/protein for 16 hours at 37 °C, while Ferret FcαR was digested with Chymotrypsin at a ratio of 1:74 (w/w) chymotrypsin/protein for 12 hours at 25°C. Formic acid was added to 1% (v/v) to quench the digestion, and the sample volume was reduced to ∼150 μL under vacuum. Peptides were then bound to C-18 (Pierce, 89870), washed three times with 0.1 % formic acid, and eluted with a 60 % acetonitrile and 0.1 % formic acid solution. C18 eluate was concentrated under vacuum centrifugation and resuspended in 10μL of 100mM ammonium bicarbonate in H₂¹⁸O. 2,500 units of Glycerol free PNGase (NEB, P0705) resuspended in H₂¹⁸O was then added to the samples and incubated at 37°C for 1 hour. The reaction was then denatured at 100°C for 5 minutes and concentrated under vacuum centrifugation. Prior to injection, final resuspensions were made in 0.1 % formic acid and 5 % acetonitrile in H₂¹⁸O.

LC-MS/MS analysis

Samples were analyzed by liquid chromatography-tandem mass spectrometry on an Easy-nLC 1200 (Thermo Fisher Scientific) coupled to an Orbitrap Fusion Tribrid (Thermo Scientific). Peptides were first loaded onto an Acclaim PepMap 100 C18 trap column (Thermo Scientific) prior to separation on a 75 μm × 50 cm EASY-Spray C18 analytical column (Thermo Scientific) using a 1.6 %–76 % (v/v) acetonitrile gradient with 0.1% formic acid over 90 mins at 300 nL/min. Eluting peptides were injected directly into the mass spectrometer using an EASY-Spray source (Thermo Scientific). The instrument was operated in data-dependent mode with parent ion scans (MS1) collected at 120,000 resolution. Monoisotopic precursor selection and charge state screening were enabled. Ions with charge ≥ +2 were selected for collision-induced dissociation fragmentation spectrum acquisition (MS2) in the ion trap, with a maximum of 20 MS2 scans per MS1. Dynamic exclusion was active with a 15-s exclusion time after an ion is selected once. Each sample was run three times to generate technical replicate datasets.

Data analysis

Protein sequence databases were constructed by appending IgA or FcαR amino acid sequences to a database comprising a consensus human protein database (Ensembl 73, longest sequence/gene) and a list of common protein contaminants (MaxQuant). Spectra were searched against the database using SEQUEST (Proteome Discoverer 2.4; Thermo Scientific). A precursor mass tolerance of 10 ppm and fragment mass tolerance of 0.6 Da were used with an allowance of 2 and 4 missed cleavages for tryptic and chymotryptic peptides respectively. Modifications of deaminated asparagine (H₂¹⁸O, and standard) and oxidized methionine (dynamic) were selected. Peptide-spectrum matches (PSMs) were filtered using Fixed Value PSM Validator (Maximum Delta Cn = 0.05, Proteome Discoverer 2.4; Thermo Scientific). PSMs that were only detected once, yielded a ppm less than −5 or greater than 5, or identified asparagine deamidations outside of putative glycosylation sites, were removed from the dataset. Peptide abundance was calculated from the extracted-ion chromatogram (XIC) peak area, comprised of the sum of all unique peptide XIC areas of associated precursor ions. Final glycan occupancy was determined by calculating the XIC ratio between peptides possessing deamidated and non-deamidated asparagine residues within the putative N-linked glycosylation site.

Affinity assessments

Biolayer interferometry (BLI) was used to assess the interaction of IgA in solution with FcαR on a tip surface. Site-specifically biotinylated FcαR extracellular regions were loaded onto a streptavidin-coated biosensor (SAX2, Sartorius) at 40 nM in BLI running buffer (1×PBS, pH 7.4, with 0.02 % Tween-20, 0.1 % BSA) for 60 s. Antibodies were formulated at twice the intended concentration and diluted in BLI running buffer over a series of 1:3 dilutions starting from 222 nM for golden hamster IgA, 333 nM for human IgA1, and human IgA2, and 1 μM for ferret IgA (recombinant or native). Association of the complex was measured at 30 °C over a 300 s period followed by 500 s of dissociation by Octet^Red 96 (ForteBio). The boundary between association and dissociation phases is indicated by a vertical line in graphs. The data were collected and analyzed by Octet Data Acquisition and Data Analysis HT. Equilibrium binding affinities were calculated using the mean signal observed over the last 5-10 s of the association phase.

Surface plasmon resonance (SPR) was used to assess the binding of FcαR in solution to IgA on a prism surface. Antibodies of interest were prepared by covalent linkage to a carboxymethyldextran-functionalized biosensor (CMD200M, Xantec Bioanalytics) using a continuous flow microspotter (Carterra) capable of producing 96 discrete regions of interest. The microfluidic pathways were primed with 10 mM sodium acetate (pH 5.0) and activated for 5 minutes with 100 μL of 10.4 mM EDC and 1.4 mM sulfo-NHS (ThermoFisher) prepared in 10 mM MES (pH 5.0). Each antibody was prepared in 10 mM sodium acetate (pH 5.0) at concentrations ranging from 400 to 50 nM and along a 2-fold dilution series. Ligand was flowed over the activated regions for 7 minutes, followed by 5 minutes of washing with sodium acetate. The sensor chip was loaded on a PBS + 0.05 % Tween 20-primed imaging-based SPR (MX96, IBIS Technologies) and quenched with a 150 μL injection of 1 M ethanolamine.

Fc receptors were formulated at 5 μM (golden hamster FcαR) and 1.5 μM (ferret FcαR) in PBS + 0.05 % Tween 20 and serially diluted over an 7 or 8-point 3-fold dilution series. The association time between IgA and FcαR was set at 200 s, and disassociation time at 300 s. Between each dilution, the chip was regenerated using 10 mM glycine, pH 3 to fully remove bound Fc receptors and then re-equilibrated with PBS + 0.05 % Tween 20 buffer for the next analyte concentration. The instrument was maintained at 25 °C throughout the experiment. Initial processing of the SPR data was performed using SprintX (IBIS Technologies). The signal from each region of interest on the sensor was referenced using the nearest unconjugated inter spot to account for bulk shifts and nonspecific binding. The blank injection immediately preceding each series of receptors was subtracted from the signal of each of the injections within a receptor series. Affinity values were further processed in Scrubber 2 (BioLogic Software) using k_on and k_off values for kinetic values and R_max values of each analyte concentration for equilibrium fitting.

Abbreviations

FcαR

Fcα receptor

FcRs

Fc receptors

Fab

antigen-binding fragment

Fc

crystallizable fragment

Fv

variable fragment

mAb

monoclonal antibody

IgA

Immunoglobulin A

ITAM

Immunoreceptor tyrosine-based activation motif

N

asparagine

BLI

biolayer interferometry

SPR

surface plasmon resonance

Data availability

Raw data is available from the corresponding author on reasonable request.