By the mid-70s, structural knowledge had started to accumulate on antibodies and their interaction with antigens (foreign substances). The structures of whole antibodies were determined, as well as of various isolated parts of the molecule. The most studied part was the Fab (the antigen-binding fragment), by itself or, in many instances, in complex with antigen. It soon became clear that different antibodies had very similar structures, except in the region used for binding antigens. (The antigen-binding site is formed by the Complementarity-Determining Regions [CDRs], as originally predicted by Elvin Kabat and T.T. Wu.)
The binding of antibody to antigen is very specific. In fact, antibodies are known to be able to distinguish between two molecules that differ by only a few atoms. A notable example is the case of the blood group substances A and B, both large molecules, but which differ only in that the group B substance has a hydroxyl group at one position in its terminal sugar, while the group A substance has an acetylated amino group at the same position. This small structural difference is enough to preclude blood transfusion between individuals of these blood types (individuals with blood type A have antibodies to group B substance and vice versa, and those antibodies can cause the death of the recipient if transfused with as little as 30 ml of the wrong type blood).
The exquisite specificity of antibody-antigen interactions makes antibodies very useful in medicine. For example, antibodies could be used to target tumor cells – and only tumor cells – so that cancer treatment with antibodies would have fewer side effects than, say, chemotherapy which kills cancer cells as well as many other (normal) cells. So, we only need to produce antibodies against the various cancer types and we have a cure for cancer. Alas, it is not that simple.
While large amounts of specific antibodies could be easily produced, they are often obtained from non-human sources, e.g. mice. There are disadvantages to using murine and other non-human antibodies in human therapy. First, our immune system will recognize those antibodies as foreign molecules and will try to get rid of them. Second, the cells of our immune system, which cooperate with our antibodies in mounting an effective response to an invasion by foreign cells or substances, don’t interact with non-human antibodies as well as they do with our own antibodies. There is then a need to make the non-human antibodies look like human molecules – we need to "humanize" them.
Various procedures have been developed to humanize a non-human antibody while preserving its specificity. One obvious method, called "CDR-grafting," developed by Greg Winter’s group at Cambridge, England, retain the CDRs while replacing everything else with human parts (some non-CDR residues usually need to be kept in order to maintain the structure of the CDR segments). But we have shown from our structural studies that not all of the residues in the CDRs are involved in antigen binding. Therefore, my group proposed to retain only those amino acids which are involved in the interaction with the antigen (the Specificity-Determining Residues, or SDRs, as I have named them). Another method that I proposed is to replace the residues on the surface of the non-human antibody with amino acids that are usually found in human antibodies, so that the non-human molecule is given a human "veneer" and is thus camouflaged. Several antibodies have been humanized by me and by others using the SDR-transfer and veneering techniques.
The first molecules I helped humanize were anti-tumor antibodies. By the early 90s, I had forged a collaboration with Jeffrey Schlom (NCI) whose group had developed several murine antibodies against human tumor markers (molecules on the surface of cancer cells). A particular human tumor marker, TAG 72 (Tumor Associated Glycoprotein No. 72), that is targeted by some of Jeff Schlom’s antibodies, is present in many different cancer types, and in a large percentage of the cancer cells but only rarely in normal tissues, so that the antibodies specific for that marker have great therapeutic (and diagnostic) potential. I was asked to design humanization protocols for some of Jeff Schlom’s murine anti-tumor antibodies.
The first anti-tumor antibody that we worked on was CC49. The target of CC49 is TAG 72. First, we humanized CC49 by CDR-grafting. Then, we guessed the possible SDRs of CC49 (we don’t have a structure for the complex of CC49 with TAG 72 so that we could not identify with certainty which residues of CC49 are actually involved in the binding to the tumor marker) and transferred only those to the human template molecule. Those procedures worked, so that now we have humanized versions of CC49 that have very much reduced immunogenicity and with their specificity for TAG 72 intact.
However, humanization of antibodies often results in reduced affinity for antigen. Such was the case for both the CDR-grafted and the SDR-transferred versions of humanized CC49. We tried to offset this by designing a molecule that had four ligand-binding sites for TAG 72. That tetravalent molecule has an effective affinity for TAG 72 that is many times better than the humanized versions of CC49 and even the original antibody.
The second anti-tumor antibody that we humanized was COL-1, which is specific for carcinoembryonic antigen (CEA). COL-1 also was initially humanized by CDR-grafting and later by retaining only those parts of the CDRs which we felt contain the residues responsible for its specificity for CEA. Since CEA is found in normal tissues also, although over-expressed in tumor cells, COL-1 is not as tumor-specific as CC49. Nonetheless, the work on COL-1 is worth pursuing and is being pursued.
The work on CC49 and COL-1 involved two Filipinas who were postdoctoral fellows in my laboratory: Dr. Ameurfina D. Santos of NIMBB, UP Diliman, and Dr. Noreen R. Gonzales, formerly of Ateneo.
I was involved in various other structural ventures, including work on a pathological hemoglobin, on other molecules of the immune system, on some aspects of allergy and autoimmune disease, and some enzymes, as well as in computer analyses of various biological phenomena. Some ventures resulted in publications; many were failures.
There are diseases that could be analyzed through sequence analysis, e.g. prion and other "conformational" disorders. A paper on the analysis of "mad-cow" disease has resulted from one such analysis done in collaboration with Dr. Gisela P. Padilla-Concepcion of the Marine Science Institute (MSI), UP Diliman; that paper attracted the attention of a science reporter who wrote about it in the National Post, a Canadian newspaper. Giselle and I are continuing our studies of protein conformational disorders. With Ma. Pamela C. David, a member of Giselle’s group, we recently found a possible structural explanation for the lack of direct transmission of scrapie, the prion disease of sheep and goats, to humans despite the fact that we have been eating sheep and goats since biblical times (the prevailing notion is that scrapie has been transmitted to other animals, e.g. "mad-cow" disease, and we then get the disease from eating infected parts of those other animals). A paper describing that analysis has been published. Giselle and I have also published a paper proposing an explanation for the apparent susceptibility to prion disease of individuals, who have a particular version of the prion gene.
We have also analyzed antibody sequences in an attempt to explain the high affinity of antibodies for their specific antigens. (Yes, not only are antibody-antigen interactions noted for their exquisite specificity, they are also characterized by high affinity.) This work was done principally by Pam David, with the help of several other research assistants in Giselle’s lab, and a paper describing the analysis is in press. Other sequence analyses are in the works. In fact, we feel that we are doing enough in silico experiments to justify establishing a "Virtual Laboratory of Biomolecular Structures" – something we formally did a few months ago.
And then there’s our major effort (which got its start when I visited Giselle’s lab at MSI in 1998) to generate "molecular guided missiles" against breast cancer. This is the DOST-funded AMOR (Antibody and Molecular Oncology Researchers) project which involves a number of scientists from different disciplines from several institutions. The objective of AMOR is to produce antibody constructs that will specifically target breast cancer cells and kill those cells, either by direct action of the antibodies or through the delivery of cytotoxic drugs.
In recent months, Giselle and I have started looking at the other side of the antibody-antigen coin – we are now into vaccines. I think I have developed a possible strategy for designing vaccines against constantly mutating pathogens – the influenza virus, the cold virus, the AIDS virus, and the malarial parasite are examples of such pathogens. At the moment, we are gearing up to see if the strategy can succeed against influenza. If the strategy works, then we have a means to combat not only naturally mutating pathogens, but also intentionally mutated ones.
There are lots of molecular studies that I can do in the Philippines. Very importantly, I am doing them with Filipino colleagues. I’m home!