Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Jun 9;292(23):9745-9759.
doi: 10.1074/jbc.M117.782433. Epub 2017 Apr 27.

Immunoglobulin domain interface exchange as a platform technology for the generation of Fc heterodimers and bispecific antibodies

Affiliations

Immunoglobulin domain interface exchange as a platform technology for the generation of Fc heterodimers and bispecific antibodies

Darko Skegro et al. J Biol Chem. .

Abstract

Bispecific antibodies (bsAbs) are of significant importance to the development of novel antibody-based therapies, and heavy chain (Hc) heterodimers represent a major class of bispecific drug candidates. Current technologies for the generation of Hc heterodimers are suboptimal and often suffer from contamination by homodimers posing purification challenges. Here, we introduce a new technology based on biomimicry wherein the protein-protein interfaces of two different immunoglobulin (Ig) constant domain pairs are exchanged in part or fully to design new heterodimeric domains. The method can be applied across Igs to design Fc heterodimers and bsAbs. We investigated interfaces from human IgA CH3, IgD CH3, IgG1 CH3, IgM CH4, T-cell receptor (TCR) α/β, and TCR γ/δ constant domain pairs, and we found that they successfully drive human IgG1 CH3 or IgM CH4 heterodimerization to levels similar to or above those of reference methods. A comprehensive interface exchange between the TCR α/β constant domain pair and the IgG1 CH3 homodimer was evidenced by X-ray crystallography and used to engineer examples of bsAbs for cancer therapy. Parental antibody pairs were rapidly reformatted into scalable bsAbs that were free of homodimer traces by combining interface exchange, asymmetric Protein A binding, and the scFv × Fab format. In summary, we successfully built several new CH3- or CH4-based heterodimers that may prove useful for designing new bsAb-based therapeutics, and we anticipate that our approach could be broadly implemented across the Ig constant domain family. To our knowledge, CH4-based heterodimers have not been previously reported.

Keywords: CH3; T-cell receptor (TCR); antibody; antibody engineering; bispecific; exchange; heterodimer; interface; monoclonal antibody; protein engineering.

PubMed Disclaimer

Conflict of interest statement

The authors declared that they were all employees of Glenmark Pharmaceuticals at the time the work presented here was conducted

Figures

Figure 1.
Figure 1.
New heterodimeric interfaces can be built in two different ways. A, using heterodimers as donor interfaces, BEAT: a heterodimeric interface, in this case that of TCR Ca-Cb, is grafted onto a homodimeric interface such as that of IgG1 CH3. B, mixing homodimer interfaces, half of a homodimeric interface, in this case that of IgD CH3, is grafted onto another homodimeric interface such as that of IgG1 CH3.
Figure 2.
Figure 2.
Schematic diagrams depicting the interfaces of IgG1 CH3, TCR constant domain pairs, and BEAT CH3 interfaces. The interdomain interactions were calculated from solved structures or models. The IMGT numbering is used. Charged residues are colored in red (negative) or blue (positive). Hydrophobic interactions are in gray lines, and electrostatic interactions in dashed red lines. Grafted residue numbers are in yellow. The four key sets of residues selected for grafting the TCR Ca-Cb interface onto the IgG1 CH3 homodimer are circled in dashed black lines.
Figure 3.
Figure 3.
SDS-PAGE analysis of the new CH3 and CH4 heterodimers based on the TCR constant domain interfaces. Engineered Fc-like proteins were transiently expressed, purified by Protein A chromatography and analyzed by SDS-PAGE. A, BEAT Fc and variants thereof, an Fc-like protein with a CH3 heterodimer based on the TCR Ca-Cb interface; a control and Fc-like proteins based on previously described technologies are shown. B, BEAT Fc versus BEAT G/D Fc, an Fc-like protein with a CH3 heterodimer based on the TCR Cg-Cd interface. C, BEAT CH4 Fc, an Fc-like protein with a CH4 heterodimer based on the TCR Ca-Cb interface wherein Protein A binding was engineered by mutagenesis. The Fc-like protein IgM CH4 pA corresponds to a control Fc-like protein based on the IgM CH4 homodimer wherein Protein A binding was also engineered. D, summary of heterodimer content.
Figure 4.
Figure 4.
Crystal structure of the BEAT Fc. A, ribbon diagram. B, structural alignment of grafted residues from the BEAT interface with corresponding residues in TCR Ca-Cb (IMGT numbering). The structure of the BEAT CH3 heterodimer was superimposed on that of the TCR Ca-Cb heterodimer. For BEAT, only grafted residues are displayed at top of the TCR Ca-Cb structure (PDB code 1KGC). C, close-up showing the conservation of the side-chain orientation for positions 22 and 88 in monomer (A) and 22, 85.1, and 86 in monomer (B) between the BEAT and TCR interfaces. BEAT CH3 (A) residues are colored blue and BEAT CH3 (B) residues are colored red.
Figure 5.
Figure 5.
Schematic diagrams depicting the interfaces of the new CH3 and CH4 heterodimers based on a mix of homodimeric interfaces. The IMGT numbering is used. Naming convention: the domain type is stated in parentheses (CH3 or CH4) followed by the abbreviation of the engineered domain wherein position 88 was exchanged and then the abbreviation of the engineered domain wherein positions 85.1 and 86 were exchanged. Engineered domains are denoted by a two-letter abbreviation as follows: the 1st letter corresponds to the Ig class of position 88, and the 2nd letter corresponds to the Ig class of positions 85.1 and 86. Charged residues are colored in red (negative) or blue (positive). Hydrophobic interactions are shown as gray lines, and electrostatic interactions are shown as dashed red lines. Grafted residue numbers are in yellow.
Figure 6.
Figure 6.
SDS-PAGE analysis of the new CH3 and CH4 heterodimers based on a mix of homodimeric interfaces. Engineered Fc-like proteins were transiently expressed, purified by Protein A chromatography, and analyzed by SDS-PAGE. A, BEAT Fc and BEAT G/D Fc versus Fc-like proteins with a CH3 heterodimer based on half-donor interfaces from IgA CH3, IgD CH3, or IgM CH4 grafted onto the IgG1 CH3 domain pair: MI (CH3) AG/GA, MI (CH3) DG/GD, and MI (CH3) MG/GM, respectively; the Fc-like protein based on the SEED technology is shown. B, MI (CH4) GM/MG pA corresponds to an Fc-like protein with a CH4 heterodimer based on grafting half of the IgG1 CH3 interface onto the IgM CH4 domain pair; the engineered CH4 heterodimer was also engineered to bind Protein A. The Fc-like protein IgM CH4 pA is used as a control. C, summary of heterodimer content.
Figure 7.
Figure 7.
Differential Protein A chromatography of BEAT antibodies. A, schematics of asymmetric Protein A binding strategy for platform removal of trace amounts of homodimers. Upon transient transfection, the scFv × Fab BEAT antibody is assembled from three different chains (middle species); traces of two homodimeric contaminants can also be formed (top and bottom species) but are easily purified by Protein A chromatography; the Ig-like contaminant (top species) does not bind Protein A, whereas the scFv-Fc dimer contaminant (bottom species) binds more strongly than the bispecific antibody thereby allowing separation. A(0) means no Protein A-binding site; A(+) means one Protein A-binding site; and A(2+) means two Protein A-binding sites. B and C, BEAT 2/3. B, Protein A purification chromatogram. C, SDS-PAGE analysis of main peak fractions. D and E, BEAT 2/3 variant with the D84.4Q substitution. D, Protein A purification chromatogram. E, SDS-PAGE analysis of main peak fractions. Gels include a control lane (Load) corresponding to culture supernatant after Protein G purification that shows the hetero-to-homodimer ratio before loading on the Protein A column.
Figure 8.
Figure 8.
Biochemical characterization of BEAT 2/3. A, thermal stability by DSC. An overlay of the BEAT 2/3 without (solid line) and with the D84.4Q substitution (dashed line) is shown. The first transition corresponds to the melting of the scFv moiety and the BEAT Fc region (∼69.5 °C), and the second transition accounts for the melting of the Fab portion (∼81.5 °C). B, size-exclusion chromatography profiles of BEAT 2/3 and its D84.4Q variant: top, BEAT 2/3 (94% monomer); bottom, BEAT 2/3 variant with the D84.4Q substitution (92% monomer). C and D, co-binding of BEAT 2/3 to recombinant HER2 and HER3 extracellular domains monitored by SPR; the length of each antigen injection is indicated by a double-headed arrow.
Figure 9.
Figure 9.
In vitro and in vivo efficacy of BEAT antibodies. A–C, BEAT 1/3 and 2/3 inhibited tumor growth in vitro more potently compared with the monotherapies and their combinations. BxPC3 (A) and Calu-3 (B and C) cells were treated with antibodies or a combination of antibodies in a dose-dependent manner in the presence of heregulin. Cell proliferation was assessed after 72 h of treatment using Alamar Blue. Data are shown as mean ± S.E. plotted from one representative experiment. D, BEAT 2/3 inhibited tumor growth in vivo more potently compared with the monotherapies. Mice were treated every 3–4 days from day 12; tumor growth was monitored until day 40. Monotherapies and bispecific antibodies were dosed at 15 mg/kg; for combination therapies, each antibody was injected at 7.5 mg/kg. Data are presented as the tumor size (cm3) of each mouse at day 40 after xenograft; for each group, the mean ± S.E. is shown. Control (CTRL) indicates treatment with PBS. Asterisks denote statistically significant p values.

References

    1. Garber K. (2014) Bispecific antibodies rise again. Nat. Rev. Drug Discov. 13, 799–801 - PubMed
    1. Sheridan C. (2015) Amgen's bispecific antibody puffs across finish line. Nat. Biotechnol. 33, 219–221 - PubMed
    1. Sheridan C. (2016) Despite slow progress, bispecifics generate buzz. Nat. Biotechnol. 34, 1215–1217 - PubMed
    1. Weiner G. J. (2015) Building better monoclonal antibody-based therapeutics. Nat. Rev. Cancer 15, 361–370 - PMC - PubMed
    1. Klein C., Sustmann C., Thomas M., Stubenrauch K., Croasdale R., Schanzer J., Brinkmann U., Kettenberger H., Regula J. T., and Schaefer W. (2012) Progress in overcoming the chain association issue in bispecific heterodimeric IgG antibodies. MAbs 4, 653–663 - PMC - PubMed

MeSH terms