A bacterial toxin tipped with a tiny fragment of antibody may heraldthe cancer therapy of the future. The new "search and destroy"molecules, which scientists hope will be able to track down cancercells throughout the body and kill them, could enter trials on humanswithin months if tests on animals go well.

The latest results, reported by Ira Pastan, chief of the Laboratory ofMolecular Biology at the National Cancer Institute in Bethesda, Md.,and Darell Bigner, of Duke University Medical Center in Durham,N.C., and their colleagues, in this month's Proceedings of theNational Academy of Sciences raise hopes that a new treatment forone of the most common brain tumors, as well as many tumors of thebreast and ovary, could be just around the corner. Their paper isentitled: "Recombinant immunotoxins specific for a mutantepidermal growth factor receptor: Targeting with a single chainantibody variable domain isolated by phage display."

Pastan, who has been working on developing targeted therapy forcancer for more than 15 years, describes the latest incrementaladvance as "very exciting." We are cautious but we think this is asubstantial step forward in trying to find specific agents to killcommon tumors for which there is really very little therapy to offerpeople," he said.

Arthur Frankel, associate professor of medicine at the MedicalUniversity of South Carolina, in Charleston, said the new antibody-toxin conjugate has "real potential for clinical usefulness." Theoverriding problem in the field has been in identifying molecularstructures on target cells that are specific to tumors. "This particularconjugate made by Dr. Pastan is a state of the art development in thatthe target is truly tumor specific, and present in a large number ofpatients with certain tumors."

The strategy Pastan and his colleagues are following involvesidentifying antigens specific to tumors, making antibodies whichrecognize these antigens, and then attaching powerful toxins to theantibodies. The hope is that the resulting molecules _ calledrecombinant immunotoxins _ will home in on cells bearing thecancer-specific antigens, rather like a heat-seeking missile locks on tothe heat of a fighter-jet's engines.

Until recently, several stumbling blocks have remained in the way ofthis ideal scenario. First, antigens specific to cancer cells have beenhard to find. Second, antibodies that recognize them have not beenspecific enough: they fail to lock on to all cancer cells bearing theantigen, or they stick to other antigens found on normal cells. Butnow, Pastan said, he and his collaborators have manufactured arecombinant immunotoxin which is highly specific for an antigencommon on brain tumors. Furthermore, it is very stable, and smallenough to allow it to penetrate well into tumors.

The quest for an antigen specific to cancer cells took a big leapforward when, in 1990, Bigner and his colleagues and otherresearchers discovered that the cells in a type of brain tumor called aglioblastoma frequently had a specific mutation in their receptor forepidermal growth factor (EGF). The mutant EGF receptor, which hasbeen called EGFRvIII, has a protein sequence specific to the tumor.

Bigner's group identified and described several antibodies whichrecognize the mutant receptor but not the normal receptor. Workingwith Bigner, Pastan conjugated these antibodies to the toxin andshowed that the conjugates could kill cancer cells. But they did notbind to the cancer cells as strongly as the researchers would haveliked. Their next step was to clone the variable regions of theseantibodies and fuse them to the toxin. "For reasons which we do notunderstand, these experiments were not successful," Pastan said.

Antibody Phage Display Played Important Role

At this point, the teams turned to a new technology, called antibodyphage display, developed by Greg Winter of Cambridge University.The starting point for this technique is the M13 bacteriophage, asmall single-stranded DNA bacteriophage which has only a fewgenes. On its surface is a protein called the gene 3 protein, which thebacteriophage uses to attach to the surface of Escherichia coli toinfect it. The next step is to take the spleen of an immunized mouse,which has many cells making a huge number of antibodies, clone thevariable domains of all these antibodies and attach these fragments _separately _ to the chromosome of the M13 bacteriophage. Thisgives a collection of bacteriophages, each with a different antibodyfragment on its surface, attached to the gene 3 protein.

Pastan's group therefore took a mouse which Bigner's group hadimmunized with EGFRvIII and made a library of bacteriophagesbearing antibody fragments. They then fixed a portion of the mutantEGF receptor protein to an immobilized surface and Ian Lorimer,Pastan's postdoctoral researcher, searched through the library for thebacteriophage which had variable region fragments on its surfacewhich bound to the mutant protein. Once they had found the rightbacteriophage, they could look in its chromosome for the DNAencoding the variable fragment of the antibody they were seeking.

"The whole selection process ensures that the antibody fragmentmade by the bacteriophage has to be very stable for you to find it,"Pastan said. "So you end up with very stable scFv fragments, and byvarying the selection process and making it more stringent, you canalso get those of very high affinity."

This made it possible to identify an antibody fragment which binds tothe mutant EGF receptor with high affinity and which is very stable."When this is fused to the toxin, we have a compound which is veryspecifically cytotoxic to cancer cells with the mutation, because theantibody binds only to cancer cells with this mutation, not to othercancer cells, and not to other cells in the body," Pastan said. "If youput it in tissue culture with cancer cells expressing the mutant protein,it kills them."

The potential for clinical application is considerable. EGFRvIII ispresent in 50 to 60 percent of glioblastomas, the most common typeof brain tumor. In addition, various groups have reported finding it inup to 70 to 80 percent of breast and ovarian cancers, and 16 percentof non-small cell lung cancers. And, encouragingly, the newimmunotoxin retained all its cytotoxic activity after incubation at 37degrees centigrade for 24 hours in human serum.

Pastan, Bigner and their colleagues are now going ahead with tests onan animal model. They are implanting human glioblastoma cells intothe brains of immunodeficient rats. The rats' immunological defectwill allow the human cells to grow. The teams will then administertheir new recombinant immunotoxin, in the hope of killing the tumorsand curing the rats.

If this works, Pastan said, the next step will be to find a way to makeenough of the immunotoxin to test it in patients with brain tumors. n

-- Sharon Kingman Special To BioWorld Today

(c) 1997 American Health Consultants. All rights reserved.