Design Principles for Bispecific IgGs, Opportunities and Pitfalls of Artificial Disulfide Bonds

Bispecific antibodies (bsAbs) are antibodies with two binding sites directed at different antigens, enabling therapeutic strategies not achievable with conventional monoclonal antibodies (mAbs). Since bispecific antibodies are regarded as promising therapeutic agents, many different bispecific design modalities have been evaluated, but as many of them are small recombinant fragments, their utility could be limited. For some therapeutic applications, full-size IgGs may be the optimal format. Two challenges should be met to make bispecific IgGs; one is that each heavy chain will only pair with the heavy chain of the second specificity and that homodimerization be prevented. The second is that each heavy chain will only pair with the light chain of its own specificity and not with the light chain of the second specificity. The first solution to the first criterion (knobs into holes, KIH) was presented in 1996 by Paul Carter’s group from Genentech. Additional solutions were presented later on. However, until recently, out of >120 published bsAb formats, only a handful of solutions for the second criterion that make it possible to produce a bispecific IgG by a single expressing cell were suggested. We present a solution for the second challenge—correct pairing of heavy and light chains of bispecific IgGs; an engineered (artificial) disulfide bond between the antibodies’ variable domains that asymmetrically replaces the natural disulfide bond between CH1 and CL. We name antibodies produced according to this design “BIClonals”. Bispecific IgGs where the artificial disulfide bond is placed in the CH1-CL interface are also presented. Briefly, we found that an artificial disulfide bond between V(H) position 44 to V(L) position 100 provides for effective and correct H–L chain pairing while also preventing the formation of wrong H–L chain pairs. When the artificial disulfide bond links the CH1 with the CL domain, effective H–L chain pairing also occurs, but in some cases, wrong H–L pairing is not totally prevented. We conclude that H–L chain pairing seems to be driven by V(H)–V(L) interfacial interactions that differ between different antibodies, hence, there is no single optimal solution for effective and precise assembly of bispecific IgGs, making it necessary to carefully evaluate the optimal solution for each new antibody.