Antibodies specific for a segment of human membrane IgE deplete IgE-producing B cells in humanized mice

Abstract

IgE-mediated hypersensitivity is central to the pathogenesis of asthma and other allergic diseases. Although neutralization of serum IgE with IgE-specific antibodies is in general an efficacious treatment for allergic asthma, one limitation of this approach is its lack of effect on IgE production. Here, we have developed a strategy to disrupt IgE production by generating monoclonal antibodies that target a segment of membrane IgE on human IgE-switched B cells that is not present in serum IgE. This segment is known as the M1' domain, and using genetically modified mice that contain the human M1' domain inserted into the mouse IgE locus, we demonstrated that M1'-specific antibodies reduced serum IgE and IgE-producing plasma cells in vivo, without affecting other immunoglobulin isotypes. M1'-specific antibodies were effective when delivered prophylactically and therapeutically in mouse models of immunization, allergic asthma, and Nippostrongylus brasiliensis infection, likely by inducing apoptosis of IgE-producing B cells. In addition, we generated a humanized M1'-specific antibody that was active on primary human cells in vivo, as determined by its reduction of serum IgE levels and IgE plasma cell numbers in a human PBMC-SCID mouse model. Thus, targeting of human IgE-producing B cells with apoptosis-inducing M1'-specific antibodies may be a novel treatment for asthma and allergy.

Figure 1. M1′-specific antibody specifically binds human M1′ in the context of membrane IgE with high affinity.

(A) M1′-specific antibody 47H4 bound human membrane IgE-transfected Daudi B cells but not Daudi transfectants expressing membrane IgE that lacks the M1′ sequence. M1′-specific antibody 47H4 also bound the U266 myeloma cell line, which naturally expresses low levels of membrane IgE. Isotype control antibody staining is shown as the gray areas, and 47H4 antibody staining is shown as black lines. (B) Scatchard analysis of M1′-specific antibody 47H4 binding to membrane IgE-transfected Daudi cells. The left panel shows a competition binding curve, where each data point represents the ratio of iodinated 47H4 antibody bound to cells/total iodinated and unlabeled 47H4 antibody on the y axis vs. the total concentration of iodinated and unlabeled 47H4 antibody on the x axis. The right panel shows a Scatchard plot, where each data point represents the ratio of iodinated 47H4 antibody bound to cells/unbound iodinated 47H4 antibody on the y axis vs. the concentration of bound iodinated 47H4 antibody on the x axis The mean binding affinity ± SD from 2 separate experiments is 0.54 ± 0.02 nM.

Figure 2. Human M1′/GFP knockin mouse has normal antibody responses and generates M1′⁺ GFP⁺ IgE B cells.

(A) Targeting scheme for insertion of human M1′ and a bicistronic GFP reporter gene in the mouse IgE locus. Human M1′ (red) is inserted into the mouse M1 exon splice acceptor site in-frame, with the M1 exon coding sequence. An IRES-GFP (green) bicistronic reporter gene is inserted 26-bases downstream of the end of the M2 exon. Wild-type and M1′ knockin (KI) mice immunized with TNP-OVA/alum have identical baseline and immunization-induced serum (B) IgE and (C) IgG1, as measured by ELISA. (D) Ex vivo culture of M1′ knockin mouse splenocytes with LPS and IL-4, but not LPS alone, induced M1′⁺ GFP⁺ B cells. Numbers indicate the percentage of CD19⁺ GFP⁺ cells and are representative of at least 3 experiments. Flow cytometry data from LPS⁺IL-4–stimulated cultures in the center panel was further gated on GFP vs. CD19 expression as defined by the box and analyzed for M1′ expression as indicated by the arrow to the right-hand panel. Isotype control antibody staining is shown as the gray areas, and 47H4 antibody staining for M1′ expression is shown as black lines. (E) A small population of GFP⁺ B cells was detectable in the mesenteric lymph nodes of N. brasiliensis–infected M1′ knockin mice. Numbers indicate the percentage of CD19⁺ GFP⁺ cells and are representative of at least 3 experiments.

Figure 3. M1′-specific antibody prevents primary and memory IgE responses in human M1′ knockin mice.

(A) Experimental design for TNP-OVA–induced primary and memory immune responses. Human M1′ knockin mice were immunized with TNP-OVA/alum on day 0. In the primary IgE response model (n = 26 per group), mice were treated with 10 mg/kg M1′-specific antibody 47H4 or mIgG1 control antibody 3 times a week from day 0–28. Treatment with M1′-specific antibody 47H4 antibody (B) prevented the primary TNP-OVA–specific IgE response, but (C) did not affect the primary TNP-OVA–specific IgG1 response. In the memory IgE response model (n = 10 per group), mice were challenged with TNP-OVA alone on day 28 after the primary immunization, and mice were treated with 10 mg/kg M1′-specific antibody 47H4 or mIgG1 control antibody 3 times a week from day 28–49. Treatment with M1′-specific antibody 47H4 antibody (D) reduced the memory TNP-OVA–specific IgE response, but (E) did not affect the memory TNP-OVA–specific IgG1 response. Results are mean ± SD. *P < 0.05 (Bonferroni correction for pairwise comparisons).

Figure 4. Therapeutic treatment with M1′-specific antibody specifically reduces IgE in N. brasiliensis infection.

(A) Experimental design for therapeutic M1′-specific antibody treatment of N. brasiliensis infection. Human M1′ knockin mice (n = 10 per group) were infected with 500 N. brasiliensis L3 larvae on day 0. Mice were treated with 10 mg/kg M1′-specific antibody 47H4 or mIgG1 control antibody 3 times per week from day 11 after infection to the end of the study at day 21 after infection. Treatment with M1′-specific antibody 47H4 antibody reduced (B) total serum IgE levels and (C) the number of IgE-producing cells in the mesenteric lymph nodes, but did not affect (D) the percentage of total syndecan⁺ plasma cells in the mesenteric lymph nodes. Results are mean ± SD. *P < 0.05 (Bonferroni correction for pairwise comparisons); **P < 0.0001 (Dunnett’s test).

Figure 5. Therapeutic treatment with M1′-specific antibody specifically reduces IgE in a mouse model of allergic asthma.

(A) Experimental design for therapeutic M1′-specific antibody treatment of mouse allergic asthma model. Human M1′ knockin mice (n = 8 per group) were immunized with TNP-OVA/alum on day 0 and challenged with 7 daily aerosol administrations of TNP-OVA, starting on day 35. Mice were treated with 100 μg of M1′-specific antibody 47H4 or control mIgG1 antibody daily from day 39 through day 45. Treatment with M1′-specific antibody 47H4 antibody reduced (B) TNP-OVA–specific IgE levels, but did not affect (C) TNP-OVA–specific IgG1 levels. Treatment with M1′-specific antibody 47H4 antibody also reduced (D) the number of IgE-producing cells, but did not affect (E) the percentage of total syndecan⁺ plasma cells in the spleen, as measured on day 63. Results are mean ± SD. *P < 0.05 (Bonferroni correction for pairwise comparisons); **P < 0.05 (Dunnett’s test).

Figure 6. Efficacy of M1′-specific antibody is mediated by apoptosis.

(A) M1′-specific antibody 47H4 antibody induced apoptosis of human membrane IgE-transfected Daudi cells. (B) Caspase inhibitor z-VAD inhibited M1′-specific antibody 47H4 antibody–induced apoptosis. Apoptosis is measured by flow cytometry using anti-Annexin V antibody. The control antibody is mIgG1. (C) M1′-specific antibody 47H4 antibody reduced the percentage of GFP⁺ IgE-switched B cells on day 4 in human M1′ knockin mouse splenocyte cultures stimulated with anti-CD40 antibody and recombinant IL-4. (D) M1′-specific antibody 47H4 antibody reduced the generation of soluble IgE on day 4 in human M1′ knockin mouse splenocyte cultures stimulated with anti-CD40 antibody and recombinant IL-4. (E) Experimental design for M1′-specific antibody treatment of N. brasiliensis infection. Human M1′ knockin mice (n = 9–10 per group) were infected with 500 N. brasiliensis L3 larvae on day 0. Mice were treated with 10 mg/kg M1′-specific antibody 47H4 wild-type, 47H4-DANA, or mIgG1 control antibody 3 times a week from day 0 to 21. (F) Treatment with 47H4 wild-type and 47H4-DANA antibody resulted in equivalent inhibition of N. brasiliensis–induced serum IgE. Results are mean ± SD. *P < 0.05 (Bonferroni correction for pairwise comparisons). (G) Representative flow cytometry plots of IgE-switched GFP⁺ B cells in the spleens of N. brasiliensis–infected mice treated with M1′-specific antibody 47H4 wild-type or mIgG1 control antibody on day 21. Numbers indicate the percentage of CD19⁺ GFP⁺ cells and are representative of at least 3 experiments. (H) M1′-specific antibody 47H4 antibody reduced the percentage of IgE-switched GFP⁺ B cells in the spleens of N. brasiliensis–infected mice on day 21. (A–D and H) Results are mean ± SD. (C, D, and H) *P < 0.05 (Dunnett’s test).

Figure 7. Characterization of a humanized M1′-specific antibody.

(A) Humanized M1′-specific 47H4 antibody (h47H4) bound human membrane IgE-transfected Daudi B cells but not Daudi transfectants expressing membrane IgE that lacks the M1′ sequence. Humanized M1′-specific 47H4 antibody also bound the U266 myeloma cell line, which naturally expresses low levels of membrane IgE. (B) Scatchard analysis of humanized M1′-specific 47H4 antibody binding to membrane IgE-transfected Daudi cells. The left panel shows a competition binding curve, where each data point represents the ratio of iodinated h47H4 antibody bound to cells/total iodinated and unlabeled h47H4 antibody on the y axis vs. the total concentration of iodinated and unlabeled h47H4 antibody on the x axis. The right panel shows a Scatchard plot, where each data point represents the ratio of iodinated h47H4 antibody bound to cells/unbound iodinated h47H4 antibody on the y axis vs. the concentration of bound iodinated h47H4 antibody on the x axis. The mean binding affinity ± SD from 2 separate experiments is 1.50 ± 0.14 nM.

Figure 8. Humanized M1′-specific antibody induces apoptosis of membrane IgE-transfected Daudi cells and specifically reduces IgE in an atopic human PBMC-SCID model.

(A) Humanized M1′-specific 47H4 antibody (h47H4) induced apoptosis of human membrane IgE-transfected Daudi cells. (B) Experimental design for atopic human PBMC-SCID model. Sublethally irradiated SCID-beige mice (n = 11–12 per group) were injected with 10⁸ PBMCs from an atopic human donor. Mice were treated on days 2, 3, and 4 with 100 ng recombinant human IL-4 and on days 0 and 3 with 100 μg each of anti–human IFN-γ and anti–human IL-12 neutralizing antibodies. Then, 300 μg humanized M1′-specific 47H4 antibody or control hIgG1 antibody was delivered 3 times a week, starting on day 0. Treatment with humanized M1′-specific 47H4 antibody reduced (C) total serum human IgE levels and (D) the number of splenic human IgE-producing plasma cells, but did not affect (E) total human IgM levels or (F) the frequency of total CD38⁺ PC⁺ splenic plasma cells. Results are mean ± SD. *P < 0.0001 (Dunnett’s test).

Figure 9. Crystal structure of humanized M1′-specific 47H4 antibody Fab in complex with M1′-derived peptide.

(A) The crystal structure of the humanized M1′-specific 47H4 antibody Fab (space filling mode; heavy chain is shown as blue, light chain is shown as gray) in complex with an M1′-derived peptide (stick model, Ac-S⁶AQSQRAPDRVLCHSGQQQGLPRAAGGSVPHPRCH⁴⁰-OOH) shows hydrogen bonds (dotted lines) between the h47H4 Fab and the M1′ peptide and burial of an arginine side chain from the M1′ peptide (Arg 11) in a central deep pocket in the h47H4 Fab. Numbers correspond to amino acid residues in the entire M1′ sequence. (B) The arginine binding pocket is formed by CDRs L1 (magenta), L3 (red), and H3 (purple), but CDR H2 (yellow) makes more contacts with the peptide than any other CDR. Amino acid side chains in the Fab that make important contacts with the peptide are shown as stick representations. (C) Four antibody aspartic acid residues in the Fab peptide-binding region combine to create a net negative electrostatic potential (red), which is complementary to the positive charges from M1′ peptide arginine side chains. H, heavy chain; L, light chain.