Production of site-specific antibody conjugates using metabolic glycoengineering and novel Fc glycovariants

Abstract

Molecular conjugation to antibodies has emerged as a growing strategy to combine the mechanistic activities of the attached molecule with the specificity of antibodies. A variety of technologies have been applied for molecular conjugation; however, these approaches face several limitations, including disruption of antibody structure, destabilization of the antibody, and/or heterogeneous conjugation patterns. Collectively, these challenges lead to reduced yield, purity, and function of conjugated antibodies. While glycoengineering strategies have largely been applied to study protein glycosylation and manipulate cellular metabolism, these approaches also harbor great potential to enhance the production and performance of protein therapeutics. Here, we devise a novel glycoengineering workflow for the development of site-specific antibody conjugates. This approach combines metabolic glycoengineering using azido-sugar analogs with newly installed N-linked glycosylation sites in the antibody constant domain to achieve specific conjugation to the antibody via the introduced N-glycans. Our technique allows facile and efficient manufacturing of well-defined antibody conjugates without the need for complex or destructive chemistries. Moreover, the introduction of conjugation sites in the antibody fragment crystallizable (Fc) domain renders this approach widely applicable and target agnostic. Our platform can accommodate up to three conjugation sites in tandem, and the extent of conjugation can be tuned through the use of different sugar analogs or production in different cell lines. We demonstrated that our platform is compatible with various use-cases, including fluorescent labeling, antibody-drug conjugation, and targeted gene delivery. Overall, this study introduces a versatile and effective yet strikingly simple approach to producing antibody conjugates for research, industrial, and medical applications.

Keywords: Immunotherapy; antibody engineering; antibody-drug conjugate; glycoconjugate; glycoengineering; glycosylation; immunoglobulin G (IgG); nanoparticle.

Figure 1

Fc glycovariant design and manufacturing process.A, crystallographic structure of human immunoglobulin G1 (IgG1) fragment crystallizable (Fc) domain (PDB ID:5JII), with glycovariant sites and canonical N297 glycan site indicated. B, process of Fc glycovariant manufacturing through supplementation with 1,3,4-O-Bu3ManNAz, an azido-functionalized ManNAc analog. C and D, Nonreducing (C) and reducing (D) SDS-PAGE analysis of the F5111 antibody in wild type (WT), N297A, and engineered Fc glycovariant formats. HC, heavy chain; LC, light chain.

Figure 2

Fc glycovariants express robustly and incorporate azides to allow fluorescent labeling while retaining antigen and Fc receptor binding.A, yield of antibodies from human embryonic kidney (HEK) 293F cells following purification with protein G. B, average number of dye molecules per antibody molecule, determined by dye/protein ratio of azide-functionalized wild type (WT) F5111 antibody and glycovariants thereof labeled with dibenzocyclooctyne (DBCO)-linked fluorescent dye, as measured by UV/Vis spectroscopy. C, biolayer interferometry studies of the equilibrium binding between immobilized IL-2 and soluble F5111 glycovariants. D, biolayer interferometry studies of the equilibrium binding between immobilized FcRn and soluble F5111 glycovariants. E and F, biolayer interferometry studies of the equilibrium binding between immobilized FcγRI (E) or FcγRIIa (F) and soluble F5111 glycovariants with re-introduced N297 glycosylation site.

Figure 3

Fc glycosylation sites can be grafted onto antibodies of distinct specificity for fluorescent labeling.A, average number of dye molecules per antibody molecule, determined by dye/protein ratio of azide-functionalized wild type (WT) 1d3 antibody and glycovariants thereof labeled with DBCO-linked fluorescent dyes, as measured by UV/Vis spectroscopy. B, mouse CD19⁺ A20 B cell staining of fluorescently labeled 1d3 glycovariant antibodies, as detected via flow cytometry. CD19^- (HEK 293F) cells were used as a negative control.

Figure 4

Azide incorporation in Fc glycovariants can be optimized through use of different analogs, production in alternative cell lines, or incorporation of multiple glycosylation sites.A, average number of dye molecules per antibody molecule, determined by dye/protein ratio of azide-functionalized and DBCO-linked fluorescent dye-labeled S4 F5111 glycovariant antibody produced with varying amounts of 1,3,4-O-Bu₃ManNAz (ManNAz) or Bu₄GalNAz (GalNAz) in HEK 293F or Chinese hamster ovary (CHO)-S cells, as measured by UV/Vis spectroscopy. B, reducing SDS-PAGE analysis of wild-type (WT) F5111 antibody and glycovariants thereof, including double and triple glycomutant antibodies. C, the average number of dye molecules per antibody molecule upon expression of antibody glycovariants with supplemented azide-functionalized analogs. Either WT F5111 antibody and double or triple mutant glycovariants thereof transiently expressed in HEK 293F cells or the S146 trastuzumab glycovariant stably expressed in ExpiCHO cells in the presence of the sialyltransferase ST6GAL1 were labeled with DBCO-linked fluorescent dye, and dye/protein ratio was measured by UV/Vis spectroscopy. HC, heavy chain; LC, light chain.

Figure 5

Fc glycovariants demonstrate potent tumor cell killing upon formulation as antibody-drug conjugates.A, biolayer interferometry studies of the equilibrium binding between immobilized human epidermal growth factor receptor 2 (HER2) and soluble trastuzumab glycovariant antibodies. B, cytotoxicity of single mutant trastuzumab glycovariant antibody-drug conjugates (ADCs) against HER2-expressing SKBR3 human breast cancer cells. DBCO-linked monomethyl auristatin E (MMAE) was used as the drug payload only control. C–E, cytotoxicity of a single and triple mutant trastuzumab glycovariant ADCs against HER2⁺ SKBR3 (C), HER2⁺ HCC1954 (D), or HER2^- MDA-MB-231 (E) human breast cancer cells. DBCO-linked MMAE was used as the drug payload only control. Error bars represent standard deviation (n = 4).

Figure 6

Fc glycovariant conjugation to biomaterials enables targeted gene delivery.A, schematic of detection scheme for flow cytometry analysis of azido-modified trastuzumab glycovariants linked to DBCO-coated magnetic microparticles. Antibody conjugation is detected with a fluorescent anti-Fab antibody and target antigen binding is detected using biotinylated HER2 and secondary fluorescent streptavidin staining. Representative flow cytometry plots are shown below. B, cartoon depicting poly(beta-amino ester) (PBAE) nanoparticle encapsulating cyanine-5 (Cy5)-labeled enhanced green fluorescent protein (eGFP)-encoding mRNA conjugated to trastuzumab glycovariant antibodies. C and D, nanoparticle uptake in transfected HER2⁺ SKBR3 (C) and HER2^- MDA-MB-231 (D) cells, normalized to the unconjugated control for each concentration. E and F, eGFP expression in transfected SKBR3 (E) and MDA-MB-231 (F) cells, normalized to the unconjugated control for each concentration. Statistical significance was determined by two-way ANOVA with a Dunnett post hoc test. ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001 (n = 4).