Natural Abs represent the indigenous immune repertoire and are thus present at birth and persist throughout life. Previously, human autoantibodies to the α domain of the high-affinity IgE receptor (FcεRIα) have been isolated from Ab libraries derived from normal donors and patients with chronic urticaria. To investigate whether these anti-FcεRIα Abs are present in the germline repertoire, we constructed a phage Fab display library from human cord blood, which represents the naive immune repertoire before exposure to exogenous Ags. All isolated clones specific to the FcεRIα had the same sequence. This single IgM Ab, named CBMα8, was strictly in germline configuration and had high affinity and functional in vitro anaphylactogenic activity. Inhibition experiments indicated an overlapping epitope on the FcεRIα recognized by both CBMα8 and the previously isolated anti-FcεRIα Abs from autoimmune and healthy donors. This common epitope on FcεRIα coincides with the binding site for IgE. Affinity measurements demonstrated the presence of Abs showing CBMα8-like specificity, but with a significantly lower affinity in i.v. Ig, a therapeutic multidonor IgG preparation. We propose a hypothesis of escape mutants, whereby the resulting lower affinity IgG anti-FcεRIα Abs are rendered less likely to compete with IgE for binding to FcεRIα.
The origin and function of natural Abs in the human repertoire are enigmatic (1). They appear spontaneously throughout life, without antigenic stimulation, have none or only a few somatic mutations in the V regions (2), and are usually polyreactive to phylogenetically conserved structures (3). They can be of the IgM, IgG, or IgA isotype (4); may be involved in innate immunity to viruses (5, 6) or elimination of cancer cells (7); and probably function as precursors for hypermutated Abs. Serum natural Abs are frequently identified as autoantibodies (8) and for a long time were considered indicative of ongoing pathological autoimmune disease. Many of their functions were extrapolated from murine model systems, and there is a paucity of human natural mAbs available for research. Nevertheless, extensive studies using affinity-purified autoreactive Ig fractions have illustrated the increasing importance of natural autoantibodies. Such autoreactive Abs are now thought to play an important role in regulation of the normal B cell repertoire and represent the major fraction of serum Abs (9). Recent literature suggests that natural autoantibodies react throughout life with a restricted set of self-Ags (10).
Using phage display technology, we have recently isolated autoantibodies to the high-affinity IgE receptor (FcεRIα) from tonsils of healthy children and from blood of chronic urticaria (CU)3 patients (11). These widespread Abs are particularly interesting because they cross-link the FcεRIα on basophils leading to cell degranulation and release of inflammatory mediators in vitro and may thus have a role in the pathogenesis of CU (12). This disease is diagnosed by regular appearance of short-lived wheals on the skin for at least 6 wk (13). We suggested that autoimmune CU may be caused by an imbalance of natural Abs, which bind to an epitope exposed after removal of IgE from FcεRIα. Moreover, we have previously detected IgM anti-FcεRIα Abs in human cord blood serum (14). These Abs most likely originated from the fetus itself, because it is believed that only maternal IgG Abs cross the placenta (15).
To verify the presence of anti-FcεRIα Abs in the natural Ab repertoire, we constructed a phage display library from IgM transcripts from pooled human umbilical cord blood samples. The library represents a naive immune repertoire before exposure to exogenous Ags and clonal expansion of hypermutated Abs. In contrast, the libraries from which we previously isolated anti-FcεRIα Abs were limited to donors exposed to environmental Ags. In this study we report the isolation and molecular characterization of a human cord blood natural Ab strictly in germline configuration, which reacted with an epitope common for the previously isolated Abs and overlapping with the IgE binding site on the FcεRIα. The interaction between IgE and FcεRIα is of high affinity; consequently, the autoepitope on the receptor is physiologically masked by IgE. Occasional dissociation of IgE from the receptor has been suggested (11) and could lead to an immune response and selection of specifically reacting B lymphocytes. According to current knowledge, Abs of predetermined specificities mutate and are selected for increased affinity, but against self-recognition. We found an autoantibody in germline configuration that has higher affinity than the overall affinity of the corresponding autoantibodies found in a therapeutic i.v. Ig (IVIg) preparation. This may suggest a previously unrecognized phenomenon, whereby autoantibodies undergo a maturation process resulting in reduced affinity, thus avoiding an autoimmune reaction. We have termed such potentially mutated autoantibodies escape mutants.
Materials and Methods
Phage display library
In brief, members of the VH, κ, and λ gene families (16, 17, 18, 19, 20) from the IgM class were amplified independently from cDNA of cord blood lymphocytes. The downstream primers used for this PCR were tested in silico in a BLAST (21) search and were found to match exclusively the first constant domain of human IgM Fc fragment. The PCR products were cloned into an expression phagemid, yielding a total of 4 × 107 clones.
The cord bloods (20 ml each) were provided from deliveries of 17 male and eight female newborns (Lindenhofspital). Great care was taken not to contaminate the cord blood samples with maternal blood. The lymphocyte fraction was isolated by Ficoll (Amersham Biosciences) density gradient centrifugation. Total RNA was prepared as described previously (22) and pooled. The SMART cDNA library construction system (23) (BD Clontech) was used to ensure complete RT. A first long-distance PCR was performed according to BD Clontech manual PT3000-1. The conditions were as follows: 50 μl of a mixture containing Pfx Amplification Buffer, Pfx Enhancer Buffer, 1 mM MgSO4
The primers for H chain families VH1a, VH1f, VH3a, VH3f, VH4f, and VH6a were previously described (24). Additionally designed primers were VH2 (cag gtc acc ttg ctc gag tct ggt), VH4g (cag gtg cag cta ctc gag tgg gg), VH5 (gag gtg cag ctc gag cag tct gg), and VH7 (cag gtg cag ctc gag caa tct gg). The following downstream primer for the first constant domain of human IgM was used: gct cac act agt cta ggc aat cac tgg aag agg.
The primers for κ families: Vκ1a, Vκ2a, Vκ3a, and Vκ3b were previously described (24). Additionally designed primers were Vκ4 (identical with the Vκ1a primer), Vκ5 (gaa acg gag ctc acg cag tct cca), and Vκ6 (gaa att gag ctc act cag tct cca). The downstream primer for the 3′ end of human κ constant domain was tcc ttc tag aac act ctc ccc tgt tga agc tct ttg tga cgg gcg aac t.
The designed primers for λ families were Vλ1 (cag tct gag ctc acg cag cc(g/a) ccc tc), Vλ2 (cag tct gag ctc act cag cct gcc tc), Vλ3 (gcc tcc tat gag ctc act cag cca), Vλ4a (cag cct gag ctc act caa tca tcc tc), Vλ4b (cag cct gag ctc act cag ccc ccg tc), Vλ5 (cag cct gag ctc act cag ccg (g/t)ct tcc), Vλ6 (aat ttt gag ctc act cag ccc cac), Vλ7 (cag act gag ctc act cag gag ccc), Vλ8 (cag act gag ctc acc cag gag cca tcg ttc), Vλ9 (cag cct gag ctc act cag cca cct tc), and Vλ10 (cag gca gag ctc act cag cca ccc tcg). The downstream primer for the 3′ end of the human λ constant region was gca ttc tag atg aac att ctg tag ggg cca c.
The H and L chains were separately pooled. DNA was purified by extraction from bands in 2% agarose using a kit (Qiagen). The H chains were cut with XhoI and SpeI, and L chains were cut with XbaI and SacI (Roche). The resulting fragments were separately pooled and sequentially cloned into pMVS (12), a vector based on pComb3H that allows the expression of Fab on the surface of the filamentous phage M13 (25). Ligation, transformation of Escherichia coli XL-1 Blue strain, and production of phage particles were conducted as described previously (26, 27, 28) (the phagemid DNA was directly amplified in bacteria without performing affinity maturation in vitro). Total phagemid DNA was purified (Plasmid Maxi kit; Qiagen) from infected bacteria.
Biopanning on immobilized Ag
FcεRIα-HSA-FcεRIα (provided by Novartis) is a double-fusion construct of the extracellular part of FcεRIα and human serum albumin (HSA), produced in Chinese hamster ovary cells (14). One of the FcεRIα is fused by its N terminus, and the other by the C terminus. Binding of human IgE and recombinant human anti-FcεRIα mAbs was confirmed by ELISA (data not shown). For panning, four wells of microtiter plates (Corning; product 3690) were coated with a total of 100 μl (panning rounds one to four) or 400 μl (rounds five to eight) of FcεRIα-HSA-FcεRIα at 50 μg/ml (in 0.05 M NaHCO3 (pH 9.6) at 4°C overnight). HSA (ZLB Behring) was coated at the same concentration and used for two preabsorptions in each panning round. Blocking, washing, and elution were performed essentially as previously described (27), except for preabsorptions on HSA. Wells coated with HSA were each filled with 180 μl of phages (∼1 × 1013 phages/ml in blocking buffer) at room temperature; after 1 h, phages were transferred to the wells with FcεRIα-HSA-FcεRIα for 2 h. Eluted phages were amplified in bacteria (26, 27). After each round of panning, the enrichment was monitored by titrating the eluted phage CFU, and the presence of both H and L chain genes was confirmed by restriction analysis of DNA (data not shown).
Infected bacteria were cloned on selective media as previously described (29), and phagemid DNA was purified (Wizard Plus SV Minipreps; Promega). Clones were sequenced at Microsynth. The following primers were used: 5′-gga gga att taa aat gaa ata c (for variable H chain) and 5′-gtg gaa ttg tga gcg gat aac (for variable L chains). The most similar germ lines were found using the University of Cambridge School of Biological Sciences computing facilities and the V BASE index (30).
Phage ELISA and phage inhibition assay
Phages were titrated and diluted (∼1012 CFU/ml in PBS with 2% FCS and 0.05% Tween 20). To test the specificity of the phages, ELISA plates (same as for biopanning) were coated with FcεRIα-HSA-FcεRIα (5 μg/ml in 0.05 M NaHCO3 (pH 9.6), 50 μl/well, 4°C, overnight) or, for controls, with equimolar HSA or 10 μg/ml anti-human Fab (The Binding Site; product PC005). The plates were washed three times with 0.1% Tween 20 in PBS (pH 7.4), blocked with 5% BSA in PBS (2 h, 37°C), washed six times, incubated with the phages (100 μl/well, 2 h, ∼21°C), washed eight times, incubated with rabbit anti-phage Abs labeled with HRP (100 μl/well, 90 min, ∼21°C), washed eight times, and developed with 3,3′,5,5′-tetramethylbenzidine (Fluka)/hydrogen peroxide solution (100 μl/well, 5 min). The reaction was quenched with 1 M H2SO4 (100 μl/well). OD was measured in an ELISA plate reader (450 nm). The experiments shown in Fig. 2⇓ were performed in essentially the same way, except that for the experiment shown in Fig. 2⇓A FcεRIα-HSA-FcεRIα was coated at 1 μg/ml and CBMα8 phages were at a concentration of 5 × 1010 CFU/ml after addition. The Abs tested for the ability to inhibit CBMα8 were incubated in the coated wells for 2 h at room temperature (50 μl/well at the concentrations indicated in Fig. 2⇓), followed by CBMα8 phages (50 μl/well) for 2 h. Mean OD values from duplicate measurements are shown. Bars indicate the values obtained. The anti-phage Abs (raised in our laboratory) were conjugated with HRP (Sigma-Aldrich; P-8375) as previously described (31). LTMα15 and LTMα35 were previously produced in our laboratory as IgG Abs (11). Human monoclonal IgE-SUS11 was purified (29) from hybridomas generated in our laboratory (32). Immunoaffinity-purified murine monoclonal anti-human FcεRIα (5H5F8) was provided by F. Kricek (Novartis, Vienna, Austria). Sheep anti-mouse IgG-HRP was purchased from ICN (product 55558); anti-human serum albumin-HRP and anti-human IgE-HRP were obtained from The Binding Site (products PP032 and AP014).
Generation of full-length CBMα8
Phages were not optimal for IAsys (Affinity Sensors) measurements due to possible steric effects caused by the phage capsids. To generate CBMα8 as an Ig, the H and L chain genes were fused to the IgG Fc by cloning to modified (33) vectors from the VH and VKExpress system (34), and the constructs were verified by sequencing. Human embryonic kidney cells (HEK 293; American Type Culture Collection; CRL-1573; adjusted to suspension culture) were cotransfected with the H and L chain constructs (Swiss Federal Institute of Technology) and cultured in EX-CELL 293 serum-free medium (JRH Biosciences) for 5–6 days. The Ab was purified on a protein G-Sepharose 4 Fast Flow (Amersham Biosciences) column, and the concentration was measured by UV absorption and with the DC Protein Assay (Bio-Rad). The Ab reacted with recombinant FcεRIα, and both chains were detected with anti-human Fd or anti-human λ Abs in an ELISA and visualized by SDS-PAGE (data not shown). The Abs used in the ELISA were purified in our laboratory from hybridoma culture supernatants (American Type Culture Collection; HP6045 and HP6054, respectively).
Purification of anti-FcεRIα from IVIg
IVIg (ZLB Behring) was reconstituted with ultrapure water (Millipore) and dialyzed against PBS. Abs reacting with FcεRIα were purified using an affinity column of Sepharose coupled to FcεRIα-HSA-FcεRIα (4.5 mg of protein was taken for coupling). IVIg (1.4 g) was applied to the column, and the eluate was absorbed on a Sepharose-HSA column (10 mg of HSA was taken for coupling) before concentration on a protein G-Sepharose column.
Affinities were measured with the IAsys surface plasmon resonance instrument (Affinity Sensors) using previously described methods (14). For the linear analysis of association kinetics of CBMα8 to FcεRIα, a monophasic association model was assumed. The previously determined affinity of LTMα35 was reconfirmed. Inhibition assays were performed on a carboxylate IAsys cuvette coated with FcεRIα-HSA-FcεRIα (10 μg was used for the immobilization) in the first channel and with equimolar HSA in the second channel as a control. Both channels were treated in the same way, and the HSA background was subtracted. Acid elution of bound Abs was performed with 10 mM HCl (three washes, followed by three washes with PBS). The concentrations of Abs in the experiment shown in Fig. 3⇓ were as follows: 7.5 μg/ml (50 nM IgG) for 5H5F8, LTMα35, LTMα15, and CBMα8; and 19 μg/ml (100 nM IgE) for the mAb SUS11. For testing the inhibition of enriched anti-FcεRIα from IVIg, CBMα8 was incubated in the cuvette at 30 μg/ml for 2 h and replaced with fresh CBMα8 solution for an additional 20 min (to ensure saturation). The enriched anti-FcεRIα Abs were added together with CBMα8 at 7.5 μg/ml (to maintain the saturation).
Hexosaminidase release from RBL-2H3 cells expressing human FcεRIα
The mast cell line RBL-2H3E5.D12.8 (RBL-2H3; American Type Culture Collection; CRL-2256; transfected with human FcεRIα) was a gift (S. Mécheri, Institut Pasteur, Paris, France) and was confirmed to express human FcεRIα by FACS analysis. Release of β-hexosaminidase from the cells was performed similarly to previously described procedures (35, 36). Briefly, the cells were distributed to 24-well tissue culture plates (5 × 104 cells in 0.5 ml of antibiotic-free complete DMEM/well). The medium was supplemented with 0.1 mM hydroxyurea to enhance the expression of FcεRIα (27). After 2 days, the medium was replaced with prewarmed medium either with or without 50 μg/ml IgE SUS-11 for 90 min. Cells were washed twice with Tyrode’s salts (Sigma-Aldrich) supplemented with 10 mM HEPES buffer and treated with Abs (concentrations shown in Fig. 5⇓) in 100 μl of Tyrode’s buffer for 45 min at 37°C. Secreted and intracellular β-hexosaminidase relative levels were measured by means of a chromogenic reaction as previously described (37). The release was expressed as a percentage of the activity of total available hexosaminidase activity of unstimulated cells. Spontaneous release was measured using a buffer control without added Abs. Omalizumab (human anti-IgE Ab used for control) was a gift from Novartis. Triggering with the CBMα8 Ab was analyzed statistically using the Student’s t test.
Calculation of solvent-accessible surface of FcεRIα
We used the MSMS program (M. F. Sanner, The Scripps Research Institute, La Jolla, CA) and crystallography model 1F6A from PDB. Residues not belonging to FcεRIα (the cocrystallized Fcε) were removed from the model. The probe radius was 1.4 Å.
Analysis of human cord blood Fab library
To examine the complexity of the library, 14 random clones were sequenced. The H and L chains of clones consistently revealed different germline sequences with only occasional scattered mutations and with differing hypervariable region 3 (CDR3). Thus, the library can be considered to represent the natural Ab repertoire. The most frequently occurring H chain germline families were VH4 and VH1 (four clones each), then VH3 and VH6 (two clones each), and VH5 and VH7 (one clone each); no VH2 clones were found. These frequencies are similar to the reported VH use: VH3 > VH4 > VH1 > VH5 > VH2/VH6/VH7 (38, 39). The following L chain (VL) gene families were found: Vλ3 (three clones), Vλ4 (one clone), Vλ7 (two clones), Vκ2 (three clones), Vκ3 (two clones), and Vκ5 (three clones).
Cord blood-derived natural Ab recognizes FcεRIα
The library phages were expressed in bacteria and selected by biopanning on recombinant FcεRIα fused to HSA. Phages were preabsorbed on HSA. To take account of possible low affinity clones present in this cord blood library, washing was less stringent for the first round of panning.
Analysis by ELISA confirmed the enrichment scores, in that the strongest reaction to FcεRIα was observed after the eighth panning round. Interestingly, we also found a strong reactivity of the cord blood library phages to CRM, a Diphtheria toxin mutant (data not shown), which points at a possible role of natural Abs in innate immunity to bacteria. This initial reactivity was diluted out by the rounds of panning on FcεRIα. Nine clones specific to FcεRIα were randomly selected from the enriched library and sequenced. All these clones had identical VH and VL, indicating that only one Ab, named CBMα8, was isolated. As expected, the VH and VL sequences of CBMα8 were entirely in germline configuration. The sequences were aligned to the previously isolated anti-FcεRIα Abs for comparison (Fig. 1⇓). UGα8, and an Ab identical with LTMα15, previously described as UMα16, were isolated from libraries generated from blood of CU patients (11). The most similar Ab to CBMα8 is LTMα35. The LTMα15 and LTMα35 Abs were previously isolated from a library constructed from tonsils of healthy children (11). Among the human anti-FcεRIα Abs isolated to date, CBMα8 had a unique CDR3 of the H chain. The CDR3 of the L chain of CBMα8 was similar to that of LTMα35 (one amino acid difference). The VL sequences of CBMα8 and LTMα35 differed by several amino acid residues. The VH and VL sequences of both CBMα8 and LTMα35 belonged to the VH5 and Vλ1 families (17, 40), respectively (Table I⇓). These Abs had the same canonical classes of CDR loops 1 and 2 of VH and VL. The VH of CBMα8 belonged to the DP-73 germ line, and that of LTMα35 belonged to the VHVCW germ line. These germ lines differ only at the DNA level by one nucleotide (silent mutation) (16).
Human anti-FcεRIα Abs and IgE inhibit binding of the natural Ab CBMα8 to FcεRIα
Using an inhibition ELISA, we evaluated whether the described sequence differences between CBMα8 and the recombinant human anti-FcεRIα Abs affect the epitope specificity. Binding of phages expressing the Fab of CBMα8 to human FcεRIα was inhibited by increasing concentrations of the human anti-FcεRIα Ab LTMα35 or human IgE, but not by a control murine mAb 5H5F8 (37, 44) (Fig. 2⇓A), which has a membrane-proximal epitope on FcεRIα. The data indicate that the epitope of the germline-encoded Ab CBMα8 can be masked with IgE, as previously shown for the human anti-FcεRIα Abs LTMα15 and LTMα35, which have somatic mutations. Moreover, binding of the phage-displayed CBMα8 Fab to FcεRIα was equally well inhibited by LTMα15 and LTMα35 individually and when added together (Fig. 2⇓B). The result suggests that the three epitopes are overlapping, and binding of one of the Abs renders the epitope inaccessible for the remaining Abs.
To allow affinity measurements, a full-length CBMα8 Ab was constructed by DNA engineering and produced in eukaryotic cells. Although the Fab of this Ab was isolated from an IgM library, an IgG Fc was added. The IgG isotype was chosen because our interest was Ag binding characteristics, not Fc-mediated interactions, and because the previously isolated Abs LTMα15 and LTMα35 were also produced as IgG. The specificity of the purified Ab was verified by a sandwich ELISA in which the full-length CBMα8 Ab was shown to contain the L and H chains with the Fc and to react with recombinant FcεRIα. The affinity of CBMα8 was assessed by on-line monitoring of binding kinetics using the IAsys cuvette system. The Kd was 3 × 10−9 M and was similar to the values obtained for the previously isolated anti-FcεRIα Abs LTMα15 (7 × 10−9 M) and LTMα35 (1 × 10−8 M) (11). In comparison, our previous unpublished observations showed that the UGα8 Ab, which was isolated from an IgG library, did not bind FcεRIα as strongly as did the LTMα15 Ab (UGα8 used at 100 μg/ml gave an OD of 0.5, whereas LTMα15 at 5 μg/ml gave an OD of >1 in the same ELISA experiment).
The ELISA results were also validated using full-length CBMα8 in the IAsys. The interaction profiles of anti-FcεRIα Abs, human monoclonal IgE, and FcεRIα are shown in Fig. 3⇓. In these experiments FcεRIα was covalently bound to the solid phase. The first curve (Fig. 3⇓A) reflects binding of CBMα8 Ab to FcεRIα. Upon further addition of the Ab, there was no additional binding, which indicates saturation with CBMα8. Addition of LTMα35 or LTMα15 at the same concentrations as CBMα8 also showed no significant binding. However, a control murine mAb 5H5F8 bound to FcεRIα. The epitope of the 5H5F8 Ab is known to be separate from the IgE binding site (37, 44). The results show that CBMα8 interferes with binding of LTMα35 and LTMα15 and suggest that these Abs have an overlapping epitope different from that of the murine control Ab 5H5F8. After removal of the bound Abs with an acid wash, the LTMα15 and LTMα35 Abs could bind the receptor (∼60 arc s response; data not shown). In another experiment, FcεRIα was saturated with human monoclonal IgE (Fig. 3⇓B). Under these conditions, CBMα8 did not bind, indicating that its epitope overlaps with the IgE binding site on the FcεRIα receptor. As expected, binding of the control 5H5F8 Ab in the presence of saturating amounts of IgE was not affected.
Evidence for CBMα8-like specificity in normal human serum
To investigate whether the epitope specificity of CBMα8 can be found in human serum, we used IVIg (45), which represents IgG Abs from a large pool of healthy donors. Abs specific to recombinant FcεRIα were enriched by affinity chromatography and were shown to react with FcεRIα by ELISA. Using the IAsys system, we have demonstrated binding of the serum-derived anti-FcεRIα Abs to the FcεRIα in the absence of CBMα8 (Fig. 4⇓). The serum-derived anti-FcεRIα Abs were partially inhibited (∼45%) by the natural mAb CBMα8, suggesting that IVIg contains CBMα8-like Abs. These anti-FcεRIα Abs purified from IVIg may be oligo- or polyclonal. The average affinity of these Abs demonstrated a Kd value of 1 × 10−6 M, as determined by IAsys.
Natural Ab CBMα8 induces degranulation in vitro
The potential of CBMα8 to trigger mast cells was evaluated by monitoring the release of a marker of secretory granules, β-hexosaminidase (Fig. 5⇓), from RBL-2H3 cells expressing human FcεRIα (RBL-2H3E5.D12.8 cells). To up-regulate FcεRIα and increase the sensitivity of the assay, we used hydroxyurea (33), not IgE, because IgE interferes with binding of the natural Ab CBMα8. For a positive degranulation control, the cells were triggered with a mouse anti-human FcεRIα Ab 22E7 or with human IgE followed by a cross-linking anti-IgE Ab. Triggering using IgE is a suitable control, because previous experiments have shown that it strongly stimulates degranulation of purified human basophils (46). In the cell culture system, the release levels of β-hexosaminidase were maximally 17%, which is higher than published results obtained by cytotoxin-induced degranulation of RBL-2H3 cells (47). We stimulated the transfected cells with different concentrations of CBMα8. Hexosaminidase release was relatively strong and was dose dependent up to 16% of the total intracellular content with 4% spontaneous release. A negative control isotype-matched nonanaphylactogenic Ab, omalizumab (48, 49), showed 6% secretion.
We have confirmed these results using isolated human basophils stripped of IgE with lactic acid as described previously (11). The degranulation response to stimulation with CBMα8 was positive: 0.2 μg/ml CBMα8 induced 10% release, 2 μg/ml induced 17% release, and 20 μg/ml induced 18% release (data not shown). The buffer control was only 3%, indicating that degranulation resulting from any possibly remaining IgE-allergen complexes was negligible.
Human autoantibodies specific to either FcεRIα or IgE have been ascribed a central role in up to 30% of CU cases (50). Our former data suggested that cross-linking anti-FcεRIα Abs are also present in healthy individuals (11). We have now cloned and characterized an IgM Ab from a representative human natural Ab repertoire. This anti-FcεRIα Ab, CBMα8, is remarkable, because it originates from human umbilical cord blood and is entirely in germline configuration. The human anti-FcεRIα Abs previously isolated by phage display (from tonsils of children and peripheral blood of CU patients) differ from CBMα8 by several mutations in the L chains and have totally different CDR3s of the H chains. Nevertheless, the germline Ab CBMα8, like the previous Abs, interfered with the binding of IgE to FcεRIα. By means of an IAsys inhibition assay, we have also shown that IVIg, the IgG fraction from plasma of multiple healthy donors (51), contains the epitope specificity of the natural mAb CBMα8. Remarkably, this single Ab inhibited by ∼45% the anti-FcεRIα Abs affinity purified from IVIg. This suggests that the epitope specificity of CBMα8 persists throughout life. The observed inhibition of Abs from IVIg by the mAb CBMα8 may be partially accounted for by steric hindrance. We calculated the solvent-accessible surface of an FcεRIα crystallographic model to be ∼1.1 × 104Å2, of which even less is available to Abs due to the bent shape of the receptor. The solvent-accessible surface of a typical Ab binding site is known to be ∼1.7 × 103Å2 (52). Therefore, a large part of the available surface on FcεRIα is already buried under a single binding site of an anti-FcεRIα Ab. However, CBMα8 did not block the membrane-proximal epitope on FcεRIα, as seen by binding of the control mAb 5H5F8. In CU cases, the presence of anti-FcεRIα Abs in serum has been frequently reported (53, 54, 55), but the epitopes of these Abs have not been elucidated.
We have confirmed that CBMα8 induces degranulation in vitro by measuring the release of the preformed granular enzyme β-hexosaminidase from RBL-2H3 cells expressing functional human FcεRIα and by measuring histamine release from purified human basophils. Thus, an Ab in strict germline configuration can be isolated from human cord blood by phage display and exhibits biological activity in vitro. This is in line with previous results describing biologically active germline Abs cloned from human cord blood B cells (56).
The affinity of CBMα8 to FcεRIα is high (3 × 10−9 M). Other reported natural Abs have affinities from 10−5 to 10−8 M (57), and mature Abs typically have affinities from 10−7 M upward. We hypothesized that mutations in autoantibodies to FcεRIα may facilitate escaping from tolerance mechanisms. Rather than undergoing a separate process, the escape mutants may occur just like other mutated Abs. Random somatic mutations in the V genes are likely to lower the affinity of Abs that have a relatively high affinity already in germline configuration. In addition, because FcεRIα appears saturated with strongly bound IgE (58), human autoantibodies to FcεRIα presumably do not physiologically exhibit a strong autoreactivity because their epitope is masked by the IgE. Thus, random mutations can occur to lower the affinities of the autoantibodies below a threshold level that would be needed for deletion of the autoreactive clone. To test the hypothesis of escape mutations, we measured the overall affinity of CBMα8-like natural autoantibodies found in normal human serum. Indeed it was relatively low (1 × 10−6 M), which is in agreement with previous observations (14). Because IVIg contains a pool of IgG Abs from human serum, these anti-FcεRIα may have undergone a class switch and may result from an immune response that lowers their affinity. This is also consistent with our previous observations that an anti-FcεRIα IgG Ab isolated by phage display from CU patients (UGα8) appeared to have a lower affinity than the IgM Ab LTMα15, as judged by functional assays (25). The recognition of FcεRIα was not as strong as that by the LTMα Abs (UGα8 used at 100 μg/ml gave an OD of 0.5, whereas LTMα15 at 5 μg/ml gave an OD of >1 in the same ELISA experiment; our unpublished observations).
Physiologically these Abs may not be able to bind FcεRIα expressed on mast cells unless the cryptic epitope on FcεRIα is unmasked by dissociation of IgE. This may be reflected by the occasional induction of mild urticaria, for example, by administration of the therapeutic anti-IgE Ab omalizumab, which complexes IgE and prevents its binding to FcεRIα (49). Additional mechanisms protecting from autoreactivity to FcεRIα may include anti-idiotypic Abs to the anti-FcεRIα Abs. Indeed, the symptoms of CU could be alleviated in clinical trials (50) with IVIg, which is hypothesized to contain natural Abs (45, 51) and is believed to restore an unbalanced anti-idiotypic network (59). In addition, the exposed receptor is likely to be protected from CBMα8-like Abs by its turnover and thus by its limited availability for the autoantibodies. Binding of IgE to FcεRIα reduces the receptor turnover (60, 61, 62), so conversely, dissociation of IgE is likely to increase the turnover. Despite having no apparent physiological function, anti-FcεRIα Abs may give rise to beneficial mature Abs to viruses or bacterial Ags, not necessarily cross-reactive with the FcεRIα.
To date, human Abs to other epitopes on FcεRIα have not been isolated, although this would be of potential therapeutic interest. Indeed, the anti-FcεRIα Ab (5H5F8) raised in immunized mice has been shown to down-regulate signaling by the human receptor in vitro (37). Additionally, pathological autoantibodies in CU may target other epitopes on the FcεRIα. Although the existence of such Abs cannot be excluded, only Abs specific to one site on FcεRIα were isolated both from the natural Ab repertoire and previously from donors exposed to environmental Ags, including CU patients.
In summary, we have shown that the cord blood-derived IgM Ab CBMα8 is entirely in germline configuration, but, nevertheless, has a high affinity, competes with IgE in vitro, and is biologically active in functional assays. All the human anti-FcεRIα Abs isolated to date are mutually inhibited. A similar specificity is found in IVIg. However, the overall affinity of the IVIg-derived IgG Abs was low, suggesting a mechanism leading to a decrease in the affinity of autoantibodies. We hypothesized that the anti-FcεRIα autoantibodies in IVIg may have escape mutations, reducing their affinity. It remains an open question whether these Abs have a role in immune regulation of FcεRIα-expressing cells or evasion of pathogens.
We thank Verena Ramseyer and Elsbeth Gautschi for technical assistance, Nicole Spiegl for help with the histamine release assay, Lucia Baldi (Swiss Federal Institute of Technology, Lausanne, Switzerland) for transfection and culture of cells in a bioreactor, and Franz Kricek (Research Institute of Novartis, Vienna, Austria) for the recombinant FcεRIα protein and the 5H5F8 Ab.
The authors have no financial conflict of interest.
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.
↵1 This work was supported by Swiss National Science Foundation Grant 3200B0-100651/1.
↵2 Address correspondence and reprint requests to Dr. Sylvia Miescher, Institute of Immunology, Inselspital, Sahli-Haus 2, UG11, Bern CH-3010, Switzerland. E-mail addresses: or
↵3 Abbreviations used in this paper: CU, chronic urticaria; IVIg, i.v. Ig; HSA, human serum albumin; IAsys, interaction analysis system.
- Received January 18, 2005.
- Accepted September 8, 2005.
- Copyright © 2005 by The American Association of Immunologists