By David N. Leff

Science Editor

It takes a myriad of genes to express the protein fragments and chains that assemble into a single antibody. This complex, Y-shaped molecule is designed to "remember" and grapple one specific epitope, among all the antigens on earth that are likely to challenge an individual immune system.

Ever since Cesar Milstein and Georges Kohler created the first monoclonal antibodies in 1975, this multi-purpose biotechnological tool has suffered one handicap to fulfilling its therapeutic potential: Generating the hybridomas that churn out monoclonals has required two starter cells -- one raised in mice immunized to embody the specific antibody of interest, the other capable of indefinitely replicating it.

Treating a human patient with a mouse-derived monoclonal might work the first time, but ever after, that recipient's immune system would reject the murine portion of the antibody.

For 20 years now, molecular geneticists, immunologists and biochemists have labored to humanize monoclonals by replacing their murine components with fully human ones. Their efforts range from genetic engineering to chemical synthesis to combinatorial chemistry. The latest fruit of these endeavors appears in an article in the February 1997 issue of Nature Genetics, titled: "Functional transplant of megabase human immunoglobulin [antibody] loci recapitulates human antibody response in mice."

Molecular geneticist/immunologist Aya Jakobovits, director of discovery research at Abgenix Inc., in Fremont, Calif., is the paper's senior author. (See BioWorld Today, July 1, 1996, p. 1.) She and her project group constructed two megabase-size recombinant YACs (yeast artificial chromosomes) containing a 970-kilobase stretch of the human heavy-chain genes and an 800-kb light-chain sequence.

"YACs are the only vehicle," Jakobovits told BioWorld Today, "that allows you to clone megabase-size DNA fragments, manipulate them and deliver them in intact form into the mouse embryonic stem cells. So we used these living yeast cells as a sort of processing plant," she continued, "where we recombined large, overlapping DNA fragments that contained 80 percent of the human antibody gene repertoire."

Her team inserted these loci, which encode the antibody's variable * antigen-grabbing * region, into the embryonic stem cells of mice, whose native murine B-cell (antibody-making) genes they had blocked.

In effect, they humanized the rodents' humoral (antibody-generating) immune systems, and named the transgenic rodents "xenomice."

As an ultimate test of this strategy, the group infected their transgenic animals with three antigenic human proteins of potential interest in treating inflammation or cancer * interleukin-8 (IL-8), tumor necrosis factor (TNF) and epidermal growth factor receptor (EGFR) * and raised antibody-yielding hybridomas to each.

The pseudo-human rodent B cells met all three challenges, generating large numbers of high-affinity monoclonal antibodies to each antigenic target.

Jakobovits reported: "The affinity values we obtained for xenomouse-derived antibodies are the highest reported for human antibodies against human antigens produced from either engineered mice or combinatorial libraries.

"Why this system can be so valuable for antibody therapy," she pointed out, "is because this large diversity of the antibody genes, which exist now in mice, provides a platform of technology that now can be utilized for chosen antigens, as easily and reliably as people have generated mouse monoclonal antibodies the last 20 years."

She enumerated the "obvious practical implications of being able to generate high-affinity fully human antibodies":

* minimizing the immunogenic and allergic responses intrinsic in mouse-derived monoclonals;

* thus, increasing therapeutic safety and efficacy;

* thus, enhancing the treatment of chronic and recurring human diseases, such as inflammation, autoimmunity and cancer, which require repeated antibody administration.

"We have a very large colony of xenomice that can support all our needs," Jakobovits observed. "They breed very well, and are like wild-type mice in all respects, except that they produce human antibodies instead of mouse antibodies."

"Abgenix," she concluded, "would like to use them toward clinically relevant antigens, as we disclosed in the paper. We are planning to initiate clinical trials for our anti-inflammatory human antibody against IL-8 in 1997, and for the anticancer EGF receptor in early 1998." *