Antibody Fucosylation Lowers the FcγRIIIa/CD16a Affinity by Limiting the Conformations Sampled by the N162-Glycan

Abstract

Therapeutic monoclonal antibodies (mAbs) are largely based on the immunoglobulin G1 (IgG1) scaffold, and many elicit a cytotoxic cell-mediated response by binding Fc γ receptors. Core fucosylation, a prevalent modification to the asparagine (N)-linked carbohydrate on the IgG1 crystallizable fragment (Fc), decreases the Fc γ receptor IIIa (CD16a) binding affinity and mAb efficacy. We determined IgG1 Fc fucosylation reduced the CD16a affinity by 1.7 ± 0.1 kcal/mol when compared to that of afucosylated IgG1 Fc; however, CD16a N-glycan truncation decreased this penalty by 1.2 ± 0.1 kcal/mol or 70%. Fc fucosylation restricted the manifold of conformations sampled by displacing the CD16a Asn162-glycan that impinges upon the linkage between the α-mannose(1-6)β-mannose residues and promoted contacts with the IgG Tyr296 residue. Fucosylation also impacted the IgG1 Fc structure as indicated by changes in resonance frequencies and nuclear spin relaxation observed by solution nuclear magnetic resonance spectroscopy. The effects of fucosylation on IgG1 Fc may account for the remaining 0.5 ± 0.1 kcal/mol penalty of fucosylated IgG1 Fc binding CD16a when compared to that of afucosylated IgG1 Fc. Our results indicated the CD16a Asn162-glycan modulates the antibody affinity indirectly by reducing the volume sampled, as opposed to a direct mechanism with intermolecular glycan-glycan contacts previously proposed to stabilize this system. Thus, antibody engineering to enhance intermolecular glycan-glycan contacts will likely provide limited improvement, and future designs should maximize the affinity by maintaining the CD16a Asn162-glycan conformational heterogeneity.

Figure 1

N-glycans are required for IgG1 Fc to bind CD16a. (a) Both IgG and CD16a are N-glycosylated and multiple N-glycans contribute to affinity. (b) Cartoon schematics for the IgG1 Fc and CD16a N-glycans; individual carbohydrate residues are defined in the inset.

Figure 2

Predominant conformations sampled by CD16a are unaltered by N-glycan truncation. A ¹H-¹⁵N heteronuclear single quntum coherence spectrum of ¹⁵N[Tyr,Phe]-CD16a. Seventeen peaks are visible and correspond to eight Tyr and eight Phe residues in the CD16a construct.

Figure 3

Representative SPR sensograms for CD16a N-glycoforms binding IgG1 Fc. Response units from each sensogram were extracted once binding reached a steady state level to generate the binding curves shown on the bottom sections of each panel. Binding curves result from fitting the observed data shown as individual points in these plots.

Figure 4

Structure of CD16a N38Q/N74Q/N169Q (Man5) in complex with afucosylated IgG1 Fc. (a) The binding interface shows an interaction between the Fc Asn297-glycan and clear density around the (1)GlcNAc residue from the CD16a Asn162-glycan scaled to σ=1.5. (b) An overall view of the complex. (c) The Asn45 glycan of CD16a shows clear density for only the (1)GlcNAc residue scaled to σ=1.0.

Figure 5

Molecular dynamics simulations of CD16a reveal the space sampled by CD16a N-glycans. The top row shows a grey surface representation for the area samples in simulations for the CD16a Asn162 N-glycan. The bottom row shows the same features for the CD16a Asn45 N-glycan. The volumes sampled by each glycan are indicated.

Figure 6

A linkage in the CD16a Asn162-glycan explains the conformation difference between binding afucosylated Fc and fucosylated Fc. The area for each color in the discs represents the proportion of a distinct conformational state sampled by the Asn162-glycan in all-atom simulations. The dashed box highlights differences in conformations sampled by the linkage connecting the 3 and 4' mannose residues. Conformation states for each linkage are defined in Supplemental Figure 7.