What’s an antibody?
The iconic (IgG) antibody is a Y-shaped molecule with two arms, called the Fab regions, and a base termed the Fc region. An IgG is composed of four amino acid chains—a pair of identical “heavy chains” that run from the tip of the arms to the base, and a pair of identical “light chains,” one in each arm.
The complementarity determining regions (CDRs), made up of portions of one light chain and one heavy chain, are at the tips of the arms; these are responsible for recognizing and binding antigen. Both CDRs see the same antigen, making the antibody bivalent but monospecific.
The Fc region will bind to Fc receptors found on natural killer (NK) cells, macrophages, neutrophils and other cells of the (principally innate) immune system. Such binding, in turn, is responsible for induction of phagocytosis, cytokine release, antibody-dependent cell-mediated cytotoxicity (ADCC) and other downstream effects of antibody-antigen engagement. Similarly, binding the Fc region to a particular Fc receptor (called FcRn) conveys a half-life of days on the IgG by protecting it from degradation.
But is it human?
A typical monoclonal antibody for research is made by injecting a mouse with the antigen against which the antibody is to be generated. Antibody-producing cells are harvested, fused to a myeloma cell line to create a hybridoma and screened for how well the antibody recognizes its antigen.
Yet such an antibody used in a therapeutic context likely would be recognized by the human immune system as a foreign protein. “We make our own antibody response to the drug antibody, and therefore the concentration of the drug is diminished, perhaps drastically,” explains David Bramhill, founder of Bramhill Biological Consulting, who leads workshops on antibody engineering. “There are only three on the market that are purely mouse antibody.”
Most therapeutic antibodies are “chimeric” or “humanized”—meaning the variable region of the mouse Fab, or its CDR, respectively, has been cloned into a human antibody scaffold—or the sequence has been altered so it isn’t rejected by the human immune system.
Among the disadvantages of generating antibodies by traditional immunization is that the vast majority of antibodies are subject to a process termed “central tolerance,” in which the animal recognizes and deletes antibodies that are self-reactive. “If you line up a human and mouse protein, they’re generally homologous in regions of importance … so you’re potentially losing a lot of therapeutically valid and useful epitopes,” notes Paul Kang, chief scientific officer at Innovative Targeting Solutions (ITS).
ITS uses a system in which all the genetic elements required for efficient production of antibodies that mimic progenitor B cells are engineered into cultured human embryonic kidney (HEK) cells. “We expand the cell, induce recombination so that each cell in that population undergoes a unique V(D)J recombination reaction [to generate a diverse repertoire of antibodies], and the surface displays the antibody, just like it would happen in vivo” states Kang.The resulting library of billions of cells, each expressing a unique antibody, is not subject to central tolerance.
There are more options for utilizing powerful and efficient molecular cloning methods to consistently produce fully humanized therapeutic antibodies. “In my experience, the dominant way to raise antibody candidates for therapeutic purposes is based on fully human synthetic antibody libraries coupled to in vitro panning” by phage display, yeast display or ribosome display, notes Daniel Ivansson, staff research engineer, strategic technologies, for GE Healthcare’s Life Sciences business.
Another option is to make use of engineered mice, such as those offered by pharmaceutical company Regeneron, in which the native immunoglobulin coding sequences have been replaced with human sequences. “You immunize that animal, and it’s going to make fully human antibodies from human germline sequences,” notes Bramhill.
Antibodies these days are engineered to be “developable”—meaning not only that they possess the right combination of biological properties, but that they are “manufacturable” . Residues that may subject the antibody to glycosylation, oxidation or other unwanted modifications during manufacturing or storage are eliminated, and “they are selected and tested to be among the more stable to temperature and other denaturants,” explains Bramhill. These antibodies are designed to be expressed better, to be more soluble and fold better, and to give higher yields—typically in the 2- to 5-g/L range using a fed batch culture.
Among the attributes that can be engineered recombinantly is the Fc region. The four human IgG isotypes each have a different interaction with the various Fc gamma receptors, with associated abilities to activate and inhibit those receptors’ functions (such as triggering phagocytosis or ADCC); researchers are experimenting with swapping or mutating sections of these regions. Similarly, antibody recycling can be fine-tuned by engineering the region that interacts with the FcRn.
Some tumors and pathogens can produce proteases that will cleave an antibody at the hinge structure (which connects the Fc to the Fabs), making it less likely to induce ADCC or the complement cascade. It is possible to modify that region in antitumor or antibacterial antibodies so that it is no longer susceptible to proteolysis, Bramhill says. Similarly, this region can be modified to afford the antibody greater flexibility or to span greater distances between the Fabs and the Fc.
Antibodies are being created in a host of configurations besides that of the standard Y-shaped IgG. One of the fastest growing areas is that of the antibody-drug conjugate (ADC). As the name implies, a toxin is attached to a molecule capable of recognizing a specific sequence, enabling it to be delivered directly to a tumor, for example.
A therapeutic affinity reagent need not be a standard IgG, either. At least three Fab fragments—in which the antigen-recognition portion of the antibody has been separated from Fc region—have been licensed to date. These may be capable of blocking a receptor, for example, without invoking effector function associated with the Fc. There are many variations on this theme. In one, Fab2 fragments (in which two Fab regions have been chemically linked in the absence of the Fc) can be used to crosslink targets, such as receptors, or to bind up soluble molecules.
The two Fabs—with or without an Fc or conjugated drug—need not recognize the same epitope, either. Bi-specific antibodies are being produced that can simultaneously recognize a tumor-associated antigen and a T cell receptor, for example. It’s also possible to add additional antigen-recognition domains to enable tri-valent antibodies, for example.
There are many things people are looking to target. “The trick is to find the magic ‘right combination,’” says Kang. That combination of drugs, mono- and bi-specific antibodies, and effector function are really likely to be “target-by-target, disease indication-by-disease indication, cell type-by-cell type specific questions.”
The field of therapeutic-antibody engineering is rapidly evolving. Numerous conferences have brought together experts in the field to discuss improvements that can be applied to antibody therapeutics. Recent publications, and a soon to be released 25-page Annual Review, share the recent advances in recombinant antibody generation . If there’s some property of an antibody that doesn’t quite work right, there’s probably somebody devising a solution as you read this.
 Irani, V, et al., “Molecular properties of human IgG subclasses and their implications for designing therapeutic monoclonal antibodies against infectious diseases,” Mol Immunol, 67(2 Pt A):171-82, 2015. [PMID: 25900877]
 Larrick, JW, et al., “Antibody engineering and therapeutics conference,” MAbs, 6(5):1115-23, 2014. [PMID: 25517297]
 Parren, WHI, Lugovskoy, AA, “Therapeutic antibody engineering,” mAbs, 5(2):175-177, 2013. [PMID: 26274600]
 Tiller, TE, Tessier, PM, “Advances in Antibody Design,” Annu Rev Biomed Eng, 17:8.1-8.26, 2015. [PMID: 26274600]