Combinatorial Libraries: A New Fast Track To Monoclonals

Back in 1975, Georges Köhler and Cesar Milstein of the Medical Research Council, Cambridge, England, fused a mouse myeloma cell with a B cell to manufacture monoclonal antibodies (MAbs). Their breakthrough was viewed by the scientific community as nothing short of brilliant. In nearly every situation requiring a single component to be isolated - in distinguishing a tumor from normal tissue, for example - monoclonals proved a valuable tool. The natural diversity of these immune system proteins had long beckoned investigators to somehow harness the antibodies' specificity for use in diagnostics and therapeutics, as well as in the research laboratory. But this was difficult, because the immune system generates a polyclonal antibody response, spewing out a mixed bag of antibodies that respond to different portions of an antigen. Köhler and Milstein's hybridomas - the B cell-cancer cell duo - offered a way to isolate the single-antibody potential of an individual B cell and, courtesy of the myeloma cell's immortality, mass-produce it. So useful were MAbs that they even found their way into the corner drugstore, in pregnancy and ovulation predictor tests. Ingenious in concept, in reality MAbs proved problematic. Once a hybridoma was supplied, the process was clear-cut and profitable, but obtaining that hybridoma was often a frustrating trek down a flow chart. Would the B cell and cancer cell fuse? If the cells fuse, will they secrete an antibody? If they secrete an antibody, will it be a useful one? The procedure was, as one biotech company described it, labor-intensive, time-consuming, and expensive. What's more, even if useful MAbs do result from a hybridoma, the rodent origin often poses problems when encountered by a human immune system, limiting MAbs' utility in the clinic. In addition, mouse antibodies are not specific enough to recognize some key human antigens, such as those of the HLA system. Manufacturing human MAbs is even more technically cumbersome than the mouse approach, because human cells are more finicky about joining together. Another method, using genetic engineering to fashion fused antibody proteins that are mouse-human chimeras, still elicits unwanted immune responses. And even if these hurdles could be overcome, hybridomas simply cannot generate the diversity required for some projects, such as looking for catalytic antibodies that have structures not normally encountered in nature. Genetic engineering indeed provides a way to rapidly mass-produce human antibody proteins. Could the appropriate antibody genes or gene fragments be stitched into vectors harbored in standard E. coli, and then be coaxed to mix and match pieces to construct a diverse collection, much as happens in vivo? This would mimic the spectacular diversity of the natural antibody response, and possibly even exceed it. A team of investigators at the Research Institute of Scripps Clinic and at the biotech company Stratagene Inc., both in La Jolla, Calif., did just that, becoming the first to clone the entire antigen binding site (see story on page 22). The team consists of Richard Lerner, S.A. Iverson, Lakshmi Sastry, and William Huse from Scripps; Joe Sorge and Michelle Alting-Mees from Stratagene; and Dennis R. Burton and Angray Kang, working at Scripps on sabbatical from the Krebs Institute at the University in Sheffield, England. Similar work is being done by Gregory Winter at the Medical Research Council Laboratory of Molecular Biology, and by chemist Peter Schultz at the University of California, Berkeley. With the development of the revolutionary technique, dozens of biotech companies - such as the Houston-based Monoclonetics International Inc., and Epitope Inc., of Beaverton, Oreg. - have expressed interest in adapting the "combinatorial library" method to their own antibody searches. The simplest type of antibody is a Y-shaped structure built of four polypeptide chains, and therefore specified by four genes. The two longer chains are termed heavy (H); the two shorter chains, light (L). The two arms of the Y are each formed by one H and one L portion, and their specific conformation is the antigen binding site. Here, the amino acid sequences vary from antibody to antibody, so the tips of the Y are termed "variable." The body's finite number of antibody genes can combine with one another in a million or more different ways, enabling the organism to mount a defense against nearly any invader. The antibodies alert other parts of the immune system to finish off the interloper. To create the combinatorial libraries, the researchers immunized a mouse against the well-studied antigen NPN (p-nitrophenyl phosphonamidate antigen 1), isolated lymphocytes from its spleen, and amplified mRNA for the light and heavy genes plus sequences overseeing expression and secretion using the polymerase chain reaction (PCR). The PCR primers included oligonucleotides containing restriction sites inserted in particular orientations. Separate libraries harboring H or L sequences were then mixed and ligated with the appropriate restriction enzymes, such that the only surviving phage retain both H and L sequences in different combinations. Once expressed, the heavy and light proteins attach by sulfhydryl bonds between key amino acids to form the portion of the antibody responsible for binding the antigen. Lambda was chosen as a vehicle because screening plaques - each of which churns out a unique antibody and is therefore a clone - is simple. The plaque pattern is lifted onto a nitrocellulose filter and screened with labeled antigen. The site of bound antigen indicates the position of the phage secreting the antibody with the desired binding specificity. Advantages of combinatorial libraries over now-standard hy-bridomas are many. Use of whole animals will be minimal. One nitrocellulose filter can screen 50,000 clones, thus examining a million or more in a day. Antibody searches can be conducted systematically by the combination of one heavy chain gene, for example, with the entire repertoire of light chain genes, or vice versa. Combinations of H and L chains not seen in nature, but nonetheless possible, could be generated: a little like considering all of the possible combinations of pants and shirts in a wardrobe, not just the ones that match. With combinatorial libraries, even the staggering in vivo potential for antibody diversity can be exceeded and directed. Site-directed mutagenesis can be used to stitch particular alterations into genes. Also, antigens not likely to be encountered, or survive, in nature can be developed. For example, a highly toxic antigen could be directed to kill all but those bacteria harboring a phage whose antibody binds to it. Most exciting to Lerner and Benkovic, who came up with the original idea, is the potential for generating catalytic antibodies. Rather than seeking antibodies that bind transition state analogs - an indirect approach entailing sometimes tricky organic syntheses - catalytic activity can be assayed directly with combinatorial libraries. And perhaps the biggest plus of all is that the system will produce fully human MAbs. The Scripps/Stratagene research team is confident that combinatorial libraries will replace hybridomas. Although other companies are working on the technology, at present the constructs - called ImmunoZap antibody expression vectors - are available only from Stratagene. An original purchase for noncommercial use is $1,000; for commercial use, $10,000. For more information, contact the company at 11099 N. Torrey Pines Rd., La Jolla, Calif. 92037; (619) 535-5146. Ricki Lewis teaches biology at the State University of New York, Albany.

Combinatorial Libraries: A New Fast Track To Monoclonals

Interested in reading more?

Become a Member of