Ira Pastan

The goal of our research group is to develop a targeted therapy for human cancer. We do this by modifying a powerful bacterial toxin, Pseudomonas exotoxin A (PE), so that instead of binding to its own receptor, which is widely distributed on normal human cells, it will preferentially bind to antigens present on tumor cells. PE is a three-domain toxin in which domain I binds to the PE receptor, domain II catalyzes toxin translocation, and domain III catalyzes the ADP-ribosylation of elongation factor 2 that leads to protein-synthesis arrest and cell death. Removal of domain I generates a molecule (PE38) that has very low toxicity unless chemically attached to an antibody or fused to the Fv portion of an antibody that will target the toxin to tumor cells.

Initially, we made a recombinant toxin in which we deleted domain I; it is termed LysPE38 because it contains an extra lysine residue at the amino terminus of domain II. We used this lysine residue to couple LysPE38 to the B3 antibody. The antibody recognizes the LewisY (LeY) antigen, which is present on the surface of many epithelial tumor cells but very few normal cells. B3-LysPE38 (named LMB-1) kills cells bearing the LeY antigen, causes complete regressions of LeY-containing tumors growing in mice, and has been used in a Phase I clinical trial at NCI. In this trial, seven responses (tumor regressions) were obtained.

We next used recombinant-DNA techniques to make a smaller recombinant immunotoxin termed LMB-7.

This immunotoxin contains the Fv fragment of monoclonal antibody B3 in a single-chain form fused to PE38. B3(Fv)-PE38 (LMB-7) is produced in Escherichia coli as inclusion bodies. The protein is solubilized, renatured, and readily purified to homogeneity. LMB-7 is very cytotoxic to cancer cells bearing the LeY antigen. A Phase I clinical trial with LMB-7 is now being conducted at NCI.

Many single-chain immunotoxins, including LMB-7, are unstable due to dissociation of the Fv heterodimer, which is not prevented by the peptide linker. To overcome this difficulty, we have developed a general method of making immunotoxins in which the Fv heterodimer cannot dissociate because it is held together by a disulfide bond. LMB-9 is an improved form of LMB-7 in which the Fv fragment is disulfide-linked.

LMB-9 is stable at 37 degrees C for more than two weeks and shows excellent antitumor activity in mice. LMB-9 is currently being prepared for a Phase I clinical trial due to begin late in 1997.

The research described here is the result of a very productive collaborative effort to which many outstanding scientists have contributed. Among them are David FitzGerald, Mark Willingham, Lee Pai, Robert Kreitman, Ulrich Brinkmann, Vijay Chaudhary, and B.K. Lee.

QUESTIONS

Q: What was your starting point in this research, and how have your questions evolved?

A: Our goal was to use what we knew about genetic engineering, the process of endocytosis of cell surface proteins, and the biochemical properties of Pseudomonas exotoxin to develop a targeted therapy for cancer by targeting the toxin to cancer cells. Initially, we did simple experiments in which we conjugated the whole toxin to antibodies or growth factors. Once the three-dimensional structure of Pseudomonas exotoxin was solved, we could use this information to determine the function of different parts of the molecule and combine this with genetic engineering to make sophisticated chimeric toxic proteins.

Q: Which findings have been most surprising to you or to other scientists?

A: Our most surprising finding was that we could design molecules that would produce complete regressions of solid tumors growing in mice without damaging normal cells.

Q: What were the greatest stumbling blocks, and what new observations, techniques, reagents, or insights helped you get past them?

A: There were many scientific problems, including finding specific antibodies to target the toxin to tumor cells and figuring out how to make active proteins in E. coli. Currently, the greatest stumbling block is producing sufficient amounts of clinical–grade immunotoxin to carry out clinical trials. Unfortunately, trials in small animals, or even monkeys, do not accurately predict how the drug will behave in humans. So each clinical trial in patients is an experiment that teaches us how to improve our immunotoxins. The difficulty is that each trial takes one to two years.

Q: How can basic and clinical scientists capitalize on this research?

A: Our research, which uses basic science techniques and approaches to develop drugs to treat cancer, is best characterized as applied or translational rather than clinical. We have within our own group basic scientists who design molecules, protein engineers who make these molecules, and clinicians who carry out our clinical trials. This combination of skills in one department enables us to carry our research into patient trials. We hope what we do will encourage others trained in nondirected basic science to try and translate what they do into clinical applications.

Q: How are you following up on this work, and what questions would you ultimately like to answer?

A: Our current goal is to produce regressions of solid tumors in patients on a regular basis. We believe we can do this by using protein modeling and genetic engineering to make molecules that are more active and have fewer side effects.

LMB-7: B3(Fv)-PE38, Single-chain Immunotoxin

Return to Table of Contents