No abstract

Unlike IgA and IgM, IgG has not yet been shown to form covalent polymers. However in the presence of specific Ag, murine IgG3 has been shown to polymerize through noncovalent interactions. In contrast to the noncovalent oligomers found with murine IgG3, we have detected covalent dimers in three different recombinant human IgG2 Abs produced in myeloma cells. Both IgG2,κ and IgG2,λ can form dimers. In addition, analysis of pooled human γ globulin and several normal sera revealed the presence of IgG2 dimers. The IgG2 dimers are in contrast to the noncovalent IgG dimers found in pooled sera of multiple donors resulting from idiotype/anti-idiotype (Id/anti-Id) interactions. Cyanogen bromide cleavage analysis suggests that one or more Cys residues in the γ2 hinge are involved in dimer assembly. The potential role of IgG2 dimers in immunity against carbohydrate Ags is discussed.

No abstract

Several approaches have been developed to reduce the human immune response to nonhuman antibodies. However, chimeric antibodies and humanized antibodies often have decreased binding affinity. We described a new approach for reducing the immunogenicity of chimeric antibodies while maintaining the affinity. This approach seeks to prevent the recognition of murine immunogenic peptides from the antibody variable region by human lymphocytes. Putative immunogenic epitopes in the variable region are identified and subjected to site directed mutagenesis to make them human and/or to break the amphipathic motifs. The R3 antibody, which blocks the epidermal growth factor (EGF) receptor, was used as a model system to test this approach. Four segments containing possible amphipathic epitopes were found in the heavy variable domain using the program AMPHI. Six amino acids within two of these segments were substituted by the corresponding residues from a homologous human sequence. No mutations were made in the murine light variable domain. Experiments in monkeys suggested that the "detope" R3 antibody was less immunogenic than its chimeric analogue. A search for possible amphipathic epitopes in the Kabat database revealed the presence of conserved patterns in the different families of variable region sequences, suggesting that the proposed method may be of general applicability.

The development and production of recombinant monoclonal antibodies is well established. Although most of these are IgGs, there is also great interest in producing recombinant IgAs since this isotype plays a critical role in providing immunologic protection at mucosal surfaces. The choice of expression system for production of recombinant antibodies is crucial because they are glycoproteins containing at least one N-linked carbohydrate. These glycans have been shown to contribute to the stability, pharmacokinetics and biologic function of antibodies. We have produced recombinant human IgA1 and all three allotypes of IgA2 in murine myeloma and CHO cell lines to systematically characterize and compare the N-linked glycans. Recombinant IgAs produced in murine myelomas differ significantly from IgA found in humans in that they contain the highly immunogenic Galalpha(1,3)Gal epitope and N-glycolylneuraminic acid residues, indicating that murine myeloma is not the optimal expression system for the production of human IgA. In contrast, IgAs produced in CHO cells contained glycans that were more similar to those found on human IgA. Expression of IgA1 and IgA2 in Lec2 and Lec8 cell lines that are defective in glycan processing resulted in a less complex pool of N-glycans. In addition, the level of sialylation of rIgAs produced in murine and CHO cells was significantly lower than that previously reported for serum IgA1. These data underscore the importance of choosing the appropriate cell line for the production of glycoproteins with therapeutic potential.