Historical Perspective to Humanized Antibodies

A century ago, Paul Ehrlich proposed that antibodies could be used as "magic bullets" to target and destroy human diseases. This vision is still being pursued today since antibodies combine specificity (the ability to exquisitely discriminate diverse harmful molecules) and affinity (the ability to tightly lock onto those targets) with the ability to recruit effector functions of the immune system such as antibody- and complement-mediated cytolysis and antibody-dependent cell-mediated cytotoxicity (ADCC). Alternatively, a "toxic payload" (such as a radioactive element or a plant toxin) attached to the antibody can be accurately delivered to the target. This makes them suitable for homing in on and killing cancer cells, infectious diseases (bacteria, viruses and their toxins) as well as modulating the immune system by binding and inhibiting or enhancing its regulatory molecules and thus curing autoimmune and inflammatory diseases.

Antibodies are large, glycoprotein molecules produced by white blood cells (B-lymphocytes) of the immune system in higher organisms. Their function is to recognise and attach matter harmful to the organism, thereby marking it out for other components of the immune system to destroy. The organism makes millions of different types of antibodies, each designed to bind a surface feature (the epitope or antigenic determinant) on the foreign body (the antigen). The most common human antibody, IgG, is shaped like the capital letter "Y". It consists of four protein chains: two identical long "heavy" chains and two identical short "light" chains. The two heavy chains are parallel in the trunk of the "Y" but fan out further up to form the arms which are flexible due to the hinge region in the heavy chains at the intersection between the arms and the trunk. One light chains runs parallel to each arm. Each heavy chain is segmented into four regions: a variable domain plus three constant domains. The light chain consists of only a variable domain plus one constant domain (Figure 1). Other antibody classes (IgE, IgD, IgA, IgM) are variations on this theme, differing in the number of chains or the number of constant regions and sometimes an additional J chain. Only the tips of the variable regions on the arms of the "Y" bear the structures that recognise and attach to the antigenic epitope. These tips are called the antigen-binding sites of which there are two identical ones per IgG molecule. The main points of attachment to the epitope on the antigen-binding sites are called complementarity-determining regions (CDRs) of which there are six on each arm of the "Y"; three on the heavy chain, three on the light (Figure 2). The analogies of lock and key, and hand-in-glove have been used to describe the antibody-antigen interaction, where the antibody resembles the lock or glove and the antigen resembles the key or hand. More on the structure of antibodies can be found in Introduction to Antibody Structure.

Historically, antibodies have been produced from the serum of animals (that part of the blood left once the cells have been removed). Serum contains a cocktail of antibodies (polyclonals), some of which will attach to the antigen. Since 1890, when Emil Behring published a paper demonstrating that diphtheria antitoxin serum could protect against a lethal dose of diphtheria toxin; antisera has been used to neutralise pathogens in acute disease as well as prophylactically. Antisera is also widely used in vitro as a diagnostic tool to establish and monitor disease. The problem with using antisera for treatment is that it leads to "serum sickness" - basically the patient's immune system reacts against the harmful proteins causing fevers, rashes, joint pains and sometimes life-threatening anaphylactic shock. Also, the serum is a crude extract containing not only the antibodies against the disease-causing pathogen (often at low concentration), but also unrelated antibodies (plus other non-antibody proteins).

In 1975, César Milstein and Georges Köhler at the Medical Research Council's (MRC) Laboratory of Molecular Biology (LMB) in Cambridge (UK) worked out a way to produce "custom-built" antibodies in vitro with relative ease. They fused rodent antibody producing cells with immortal tumour cells (myelomas) from the bone marrow of mice to produce a hybridoma. A hybridoma has the cancer's ability to reproduce almost indefinitely, plus the immune cell's ability to make antibodies. Once screened, to isolate the hybridomas producing antibodies of a determined antigen specificity and required affinity - and given the right nutrients - a hybridoma will grow and divide almost indefinitely, mass-producing antibodies of a single type (monoclonals). It seemed like a production-line of batch consistency for Ehrlich's magic bullets and for this breakthrough the scientists won the Nobel Prize in Medicine in 1984.

Although monoclonal antibodies (mAbs) from hybridoma technology have proved to be immensely useful scientific research and diagnostic tools, they have not fulfilled the possibilities inherent in Erhlich's vision. The problems included identifying better antigenic targets of therapeutic value with which to raise mAbs against; making useful fragments of mAbs (whole antibodies are rather too large to penetrate solid tumours for instance); and attaching toxic payloads to the mAbs, since rodent antibodies are not as effective as human in recruiting the other cells of the immune system to complete their destructive function. However, the major hurdle has proven to be similar to that of serum therapy. Namely, that when the rodent mAbs are administered in multiple doses, the patient invariably raises an immune response to the mAbs with similar symptoms to serum sickness and violent enough to endanger life. This HAMA (Human Anti-Mouse Antibody) response can occur within two weeks of the initiation of treatment and precludes long-term therapy. The obvious answer would be to raise human mAbs to the therapeutic targets, but this is difficult both practically and ethically using the route of immortalisation of human antibody-producing cells. Human hybridomas beside being difficult to prepare are unstable and secrete low levels of mAbs of the IgM class with low affinity.

In an effort to realise Erhlich's dream of a magic bullet with high binding affinity, reduced immunogenicity (HAMA response), increased half-life in the body and adequate recruitment of effector functions (ie the ability to summon help from the body's own natural defences), scientists have used techniques from molecular biology to design, engineer and express mAbs from hybridoma technology to produce humanised mAbs. The first step was to produce a chimaeric antibody where the xenogeneic variable (V) and human constant (C) domains were constructed by linking together the genes encoding them and expressing the engineered, recombinant antibodies in myeloma cells ( Figure 3). However, when these antibodies were used therapeutically in humans, some still generated HAMA response directed against the V regions, although the level of immunogenicity varied depending on the chimaeric antibody. Going one stage further, Greg Winter, also at the MRC Cambridge, realised that only the antigen binding site from the human antibody needed to be replaced by the antigen binding site from the rodent. Since this consisted of the six CDR regions (see above), only these were grafted into the human frameworks (Figure 3). Antibodies made this way are called humanised, reshaped or CDR-grafted. In some cases, pure CDR-grafting could produce a humanised antibody with roughly the same antigen specificity and affinity as the original rodent antibody. This was not always true and it was soon evident that a more detailed design of the engineered antibody was needed before it was constructed.

Home Page José Saldanha © 2000. Birkbeck College, London WC1E 7HX.