By David N. Leff

When a monoclonal antibody comes in for a landing on a cell targeted for destruction, it makes contact by means of two antigen-binding sites.

Antibodies are Y-shaped molecules, and their business ends are at the extremes of the Y forks, like feet on the bottoms of legs.

Then, standing on the cell's surface, the monoclonal antibody (mAb) sends a signal to its enforcing agent, the complement system. Thus unleashed, the cascade of complement proteins takes lethal action to punch the cell full of holes.

Alternatively, the mAbs send one of two different signals into the cell. Either: "Stop dividing and quit growing," or: "Die!" -- that is, commit apoptosis, programmed cell death. Which of these two fates depends upon which cell-surface receptors the mAbs bind.

Though antibodies have long been hailed as magic bullets, they tend to misfire when the target is a malignant tumor.

One currently favored antitumor antibody strategy, among many, is to treat mAbs as the aiming sight and gunpowder, rather than the bullet itself. This means attaching the tumor-seeking antibody to a lethal toxin. This conjugation system sometimes works, but it has major drawbacks: Malignancies don't oblige by displaying totally specific antigens that lure the antibody. And mAbs can't gain access much below the surface of a solid tumor.

Scientists at the University of Texas Southwestern Medical Center, in Dallas, have found a way to turn these wimpy, double-whammy antibodies into muscular quadruple-whammy ones.

They tell how in the current Proceedings of the National Academy of Sciences (PNAS), dated July 8, 1997, in a paper titled: "Homodimerization of tumor-reactive monoclonal antibodies markedly increases their ability to induce growth arrest or apoptosis of tumor cells."

What the co-authors did was convert those two-legged monomeric mAbs into four-legged dimers, packing double the immunological oomph of the original molecules. With twice the number of antigen-binding sites, the article reports, "these proteins attach themselves more tightly, more avidly, to cancer cells than do monomers of monoclonal antibodies."

"It's necessary to bring a lot of receptors on the cell surface together in a very tight cluster," pointed out research oncologist Jonathan Uhr, who chairs microbiology at Southwestern, "in order to give the kind of 'negative' signal we want."

Uhr, a co-author of the PNAS paper, told BioWorld Today, "When we increase the capacity of the molecule to do this kind of clumping by simply making a single monoclonal antibody with four sites that could bind, instead of two, you have an enormous increase in the signaling capacity."

Spontaneous Dimerization Was The Clue

Semi-serendipity led first author Maria-Ana Ghetie and senior author Ellen Vitetta to see their monomeric mAbs, and raise them to homodimeric. "In preparing antibodies for clinical use," Vitetta recounted, "Gheti noticed that in one antibody, HD37, which signaled growth arrest, 15 to 20 percent of the molecules spontaneously formed 300-kiloDalton homodimers, so they were then double antibodies with four binding sites."

When Ghetie separated those "natural" Siamese-twin mAbs from the rest of the batch, Vitetta recalled, "all of the negative signaling capacity could be attributed to those molecules. That led us to explore the possibility that other, chemically generated, homodimeric monoclonal antibodies might work the same way."

The short answer: Yes, they did.

By chemical manipulation, the team converted a handful of ineffective, non-signaling monomeric mAbs to highly efficient dimeric antitumor agents, as demonstrated by in vitro and in vivo testing.

In disseminated human Burkitt's lymphomas and human breast cancer cell lines, the newly fledged tetravalent antibodies, as they reported, "had significant antigrowth activity . . . whereas the monomers showed no effect even at the highest concentrations tested."

SCID (severe combined immune deficiency) mice inoculated with Burkitt's lymphoma, then treated with either monomers or dimers, fared far better on the latter than on the former. What's more, the antitumor activity of the dimers went still higher after an additional dose of the standard antitumor chemotherapeutic drug, doxorubicin.

"The three or four receptors that we tried in the Burkitt's lymphoma all worked," Uhr observed. "The one we used in the breast cancer cells is already known to have the signaling receptor, and that worked. The receptors." he added, "are the eyes and ears and nose of the cell, anyhow, and it may be that a good proportion will be able to display this property."

"Importantly," the PNAS article reported, "the homodimers completely prevented tumor growth in the organs where the majority of the [lymphoma] cells grow [i.e., kidney, lung and ovaries] but had little effect on spinal lymphoma, which is the site of lethal tumor growth."

Uhr explained this adverse phenomenon: "In the in vivo model, the spinal lymphoma was less affected, presumably because the antibody dimer was too large to penetrate into that particular tumor site. One might want to design a smaller version by genetic engineering," he suggested, "making the smallest part of the molecule that would contain the binding sites, and leave off the rest of it."

In several controlled experiments, mice treated with one dose of dimer lived longer than control rodents, which received either the monomer or no antibody at all.

"We don't know all the answers," Vitetta commented, "but we do know that some of these molecules induce apoptosis, and some tell cells to stop dividing. Several pathways inside the cell convey these signals."

She and her co-authors will now fine-tune dose regimens of these homodimers, with a view to treating human lymphomas and breast cancers.

"These antibodies shouldn't be toxic," she observed; "they aren't carrying any warheads. They should work on accessible primary tumors or metastatic disease." She added: "This also raises the possibility of giving homodimers after conventional cancer therapy."

The Dallas team's results raise still another prospect, Vitetta observed: "We don't have to generate whole grocery stores full of new antibodies. The monomeric ones are already out there, stored in lab freezers around the world. They just need to be re-evaluated as dimers. And if they work, clinical-grade reagents can be generated by genetic engineering. This will greatly reduce their cost and increase their availability."

Uhr concluded: "I think we need more preclinical studies of human tumors in SCID mice, and then go into human clinical trials." *