Report

Divergent Immunoglobulin G Subclass Activity Through Selective Fc Receptor Binding

See allHide authors and affiliations

Science  02 Dec 2005:
Vol. 310, Issue 5753, pp. 1510-1512
DOI: 10.1126/science.1118948

Abstract

Subclasses of immunoglobulin G (IgG) display substantial differences in their ability to mediate effector responses, contributing to variable activity of antibodies against microbes and tumors. We demonstrate that the mechanism underlying this long-standing observation of subclass dominance in function is provided by the differential affinities of IgG subclasses for specific activating IgG Fc receptors compared with their affinities for the inhibitory IgG Fc receptor. The significant differences in the ratios of activating-to-inhibitory receptor binding predicted the in vivo activity. We suggest that these highly predictable functions assigned by Fc binding will be an important consideration in the design of therapeutic antibodies and vaccines.

Antibodies have evolved into classes with specific assigned functions. Within these classes, further subclassification extends immunoglobulin diversity, most strikingly in the four subclasses of IgG of mammals (1). In rodents and primates, these subclasses have evolved specialized effector responses, such as cytotoxicity, phagocytosis, and release of inflammatory mediators (2, 3). IgG subclass expression is influenced by multiple factors, including the prevailing cytokine environment. For example, the T helper cell TH2 cytokine interleukin 4 (IL-4) preferentially induces switching to IgG1 and IgE, whereas transforming growth factor–β (TGF-β) induces switching to IgG2b and IgA (46). Alternatively, TH1 cytokines such as interferon-γ (IFN-γ) result in IgG2a, 2b, and 3 switching (7). Switching is also strongly influenced by the nature of the stimulating antigen. For example, protein antigens elicit a thymus-dependent response generally dominated by IgG1, 2a, and 2b, whereas carbohydrate antigens can induce so-called thymus-independent responses that result in IgG3 antibody expression (8). Among IgG subclasses, IgG2a and 2b are generally considered to be the most potent for activating effector responses and dominate antiviral immunity and autoimmune diseases (911). Such functional distinctions among these IgG subclasses have been attributed to differences in their capacity to fix complement (12, 13). However, studies in complement-deficient mice have challenged this assumption and have focused attention on the cellular receptors for IgG, the FcγRs, as the primary mediators of IgG effector responses (14, 15). We hypothesized that unified mechanisms accounting for the different potencies of IgG subclasses might be based on the differential binding to the known FcRs.

Activation FcγRs are expressed on all myeloid cells, and their cross-linking results in sustained cellular responses (3). Balancing these activation receptors is the inhibitory Fc receptor, FcγRIIB, which, when coligated to activation receptors, dampens the cellular response (3, 16). The coexpression of activation and inhibitory receptors establishes a threshold for cellular triggering by IgG antibodies. Although all FcγRs can bind IgG immune complexes, we have observed that individual Fc receptors display significantly different affinities for IgG subclasses (17). We described this differential affinity for functionally distinct FcRs by specific IgG subclasses as a ratio, referred to as the activating-to-inhibitory (or A/I) ratio (17). These A/I ratios were found to differ by several orders of magnitude between IgG subclasses and thus raised the possibility that the variation in in vivo IgG subclass activity could be directly linked to the specific A/I ratio. To address this hypothesis, we established an in vivo system for testing antibodies that differed in their A/I ratios. The variable portions of the immunoglobulin heavy chain (VH regions) of the cloned hybridomas that recognize either the melanosome gp75 antigen (TA99) or a platelet integrin antigen (6A6) were grafted onto IgG1, 2a, 2b, or 3 Fc regions and coexpressed with the appropriate light chains (17, 18). These recombinant antibodies were purified, and subsequent testing revealed that switching the constant regions of IgG did not affect antigen binding affinity (18) (table S1). As expected, however, specific differences in binding affinity of each subclass to specific FcγRs were observed, resulting in different A/I ratios for each subclass (17) (fig. S1, table S2).

To determine whether the differences in A/I ratios for individual subclasses correlate to in vivo biological activity, we investigated the ability of these class-switched antibodies to mediate their specific biological functions: tumor clearance and platelet depletion (14, 19). Both TA99 (Fig. 1, A and B) and 6A6 (Fig. 1C) carrying IgG2a constant regions (A/I = 70) displayed enhanced activity compared with these antibodies bearing IgG1 constant regions (A/I = 0.1). IgG2a and 2b were equivalent in their ability to mediate platelet clearance, whereas IgG2a resulted in enhanced antibody-dependent cellular cytotoxicity (ADCC) in the metastatic melanoma model compared with IgG2b (A/I = 7) (Fig. 1, A and B). The hierarchy of activity for the IgG subclasses in these immune functions was thus IgG2a ≥ IgG2b > IgG1 » IgG3, mirroring the hierarchy based on the A/I ratios (Fig. 1D).

Fig. 1.

Hierarchy of antibody isotype–mediated effector functions in vivo. (A and B) B16-F10 lung metastasis in mice treated with TA99 switch variants (mean ± SEM). Mice were injected as described (18); the number of surface lung metastasis was evaluated on day 15. *P < 0.0001; **P < 0.01. (C) Platelet depletion with 6A6 switch variants (mean ± SEM). Mice were injected as described (18), platelet counts were determined at the indicated time points. (D). A/I ratio for the indicated switch variants.

We next tested the mechanism of this observed differential activity by repeating the experiments using strains of mice carrying specific deficiencies in, or blocked activation of, different activating FcγRs or complement components (Fig. 2; fig. S2). No differences in in vivo activity were observed in complement-deficient (C4, C3, or CR1/2) strains (fig. S2). In contrast, IgG1, 2a, and 2b all depended on expression of activating FcγR, because activity was abrogated in the common γ chain–deficient background (Fig. 2, A, B, and E). Because IgG2a has been shown to bind to all of the γ chain–containing activating FcγRs in vitro [with high affinity (108 to 109 M–1) to FcγRI, intermediate affinity (107 M–1) to FcγRIV, and low affinity (106 M–1) to FcγRIII], its in vivo capacity to deplete platelets or to mediate ADCC could, in principle, result from engagement of one or more of these FcγRs. We tested the contributions of each of these receptors to the in vivo activity of this subclass using mice deficient in or blocked for each receptor. TA99-IgG2a–mediated tumor clearance and 6A6-IgG2a–mediated platelet depletion were undiminished in mice deficient in either FcγRI or III (Fig. 2, C to E); in contrast, blocking FcγRIV binding with a specific monoclonal antibody significantly reduced these IgG2a-mediated activities (Fig. 2, C to E). Similarly, 6A6-IgG2b activity was only dependent on FcγRIV engagement, despite its in vitro binding to both FcγRIII and IV (17) (Fig. 2E). In contrast, the 6A6-IgG1 switch variant was only dependent on FcγRIII engagement (17) (Fig. 2E).

Fig. 2.

FcγR dependence of antibody isotype–mediated effector functions. (A and B) FcR-deficient mice, as indicated, were injected with B16-F10 melanoma cells and treated with TA99-IgG2a antibody or phosphate-buffered saline (mock) as described (18) (mean ± SEM). *P < 0.0001; N = 5. (C and D) FcγRI–/– mice were injected with B16-F10 melanoma cells, treated with TA99-IgG2a alone or together with an FcγRIV blocking or an isotype-matched control antibody, as described (18). *P < 0.001. (E) FcR dependence of 6A6 switch variants. Shown is the platelet count after injection of the indicated isotype variants (18).

The A/I ratio of 0.1 for IgG1 suggested that IgG1 might exhibit a greater dependence in its in vivo activity on FcγRIIB expression. Consistent with this prediction, we observed that IgG1 displayed the most significant enhancement in activity in mice lacking the inhibitory receptor, both in tumor clearance (Fig. 3, A and B) and platelet depletion (Fig. 3C) relative to IgG2a or 2b switch variants (Fig. 3, A to C). In contrast, IgG2a (A/I = 70) showed the least enhancement in biological activity in FcγRIIB-deficient mice (Fig. 3, B and C). By comparison, IgG2b (A/I = 7) differed in the magnitude of enhancement displayed in the FcγRIIB-deficient strains between each of the two models, with a significant increase in tumor clearance (Fig. 3B), but only minimal enhancement in platelet depletion (Fig. 3C). The intermediate A/I ratio of this subclass may render it more sensitive to the absolute level of inhibitory receptor surface expression and the specific effector cell engaged. This makes the dependence of IgG2b on FcγRIIB consistent with the observation that expression of this FcγR is minimal on splenic macrophages (15, 20), the cell type responsible for platelet clearance, but higher on alveolar macrophages, which are involved in the metastatic melanoma model (21).

Fig. 3.

A/I ratio determines in vivo efficacy of native and modified antibodies. (A to C) Differential effects of inhibitory receptor expression on IgG subclass activity. (A and B) C57BL/6 wild-type or FcγRIIB–/– mice were injected with B16-F10 melanoma cells and treated with TA99 switch variants, as described (18). *P < 0.0001; **P < 0.05; N = 5. (C) C57BL/6 or FcγRIIB–/– mice were injected with 6A6 antibody switch variants, as described (18). (D to F) Modified antibodies with increased A/I ratio display enhanced cytotoxic activity. (D) Fold increase in association constants (KA) for C1q and FcγRs I-IV in binding to fucosylated or nonfucosylated TA99 switch variants. (E and F) Clearance of B16 melanoma lung metastasis with TA99-IgG2b with or without fucose (18). *P < 0.0001.

To further explore the relation between the A/I ratio of IgG subclasses and their activity, modified IgG constant regions were generated. FcR binding to IgG depends on the presence of N-linked glycosylation at position 297, and deglycosylation abrogates all FcR binding (22). However, selective removal of specific carbohydrates, such as fucose, has been suggested to enhance human IgG1 binding to human FcγRIII and, thus, to enhance NK cell–mediated ADCC in vitro (23, 24). We therefore prepared fucose-sufficient and fucose-deficient TA99-IgG1, 2a, and 2b subclasses and compared their binding to antigen, complement, and FcγRI, II, III, and IV (18). Fucose-deficient antibodies ranged in their binding affinities to their respective FcγRs, but not for antigen or C1q, the first component of complement (23) (table S1, S2; Fig. 3D). TA99-IgG1, with or without fucose, displayed minimal differences in binding to FcγRIIB and III, whereas fucose-deficient IgG2a and 2b antibodies bound with higher affinity (by an order of magnitude) to FcγRIIB and IV compared with fucose-sufficient versions. These differences in binding affinities resulted in alterations to the respective A/I ratios that were most pronounced for IgG2b, with defucosylation increasing its A/I ratio from 7 to 20 (Fig. 3E; table S2). Furthermore, this translated into significantly enhanced in vivo activity for fucose-deficient TA99-IgG2b (Fig. 3, E and F). This selective effect of IgG defucosylation on FcR binding further illustrates the specificity of IgG subclasses in their interactions with individual FcRs and the contribution of these affinities for IgG to in vivo activity.

The studies described here provide a mechanistic basis for the observed variation in IgG subclass activity in both active and passive vaccination and in the variable pathogenicity of the IgG subclasses in autoimmune conditions. The selective FcR binding affinities for the IgG subclasses and fucose-deficient antibodies appeared to be predictive of the in vivo activity for cytotoxic antibodies in models of tumor clearance and platelet depletion. Although significant differences between the mouse and human IgG subclasses and the FcγRs have been described (25), the principles that have emerged from these mouse studies are likely to apply to human antibodies as well as their respective FcRs. Such considerations may prove important in the design of antibody-based therapeutics and active vaccination protocols.

Supporting Online Material

www.sciencemag.org/cgi/content/full/310/5753/1510/DC1

Materials and Methods

Figs. S1 and S2

Tables S1 and S2

References and Notes

References and Notes

View Abstract

Navigate This Article