August 1, 2005 (Vol. 25, No. 14)
Improving Selection and Affinity
At a recent CHI meeting in Cambridge, MA, “Recombinant Antibodies from Concept to Clinic”, a wide range of new antibody technologies were covered.
Dana Ault-Rich, Ph.D., CEO of Pointilliste (Mountain View, CA), discussed the company’s innovative technology and its application to the isolation of new therapeutic antibodies directed against non-Hodgkin’s lymphoma.
“Pointilliste takes its name from the painting technique used by Georges Seurat in the 19th century French artistic movement in which canvases were painted using many small spots of pigment,” explained Dr. Ault-Rich. By the same token, the Pointilliste technology displays large, complex molecular libraries on unique surfaces, or “Canvases,” of extremely small spots (addresses) of defined molecular composition.
Nearly 1,000 antibodies are displayed at each address, with each Canvas containing nearly 70,000 unique addresses. In this way, billions of different potential molecules for drug discovery are rapidly displayed and tested. For example, over two billion different antibodies can be screened using a set of 30 different Canvases.
The addressing system provides Pointilliste with an informational trace back to the clone identified in the screen. The single-chain antibody sequences are cloned into sublibraries containing different epitope tags, detectable with appropriate reagents, allowing sets of antibodies to be co-purified and then sorted into different addresses.
“Doing two billion different purifications is obviously impractical, but doing 100,000 purifications is not,” according to Dr. Ault-Rich. Once purified, the antibody library can be screened thousands of times, minimizing the cost per screen.
Dr. Ault-Rich discussed the use of this technology to develop anti-idiotypic antibody therapies for lymphoma, an approach successfully pioneered in the 1980s by Ron Levy, M.D., a member of the company’s scientific advisory board. Dr. Levy and his colleagues successfully treated patients with mouse anti-idotypic antibodies directed against immunoglobulins displayed on the surface of the patients’ cancer cells.
However, the cumbersome hybridoma technology of that period did not permit the isolation of large numbers of antibodies, as is possible with the Pointilliste approach.
Non-Hodgkin’s lymphoma is a devastating malignancy. The anti-idiotypic approach is an alternative to Rituxin, the antibody that targets the CD-20 receptor. Rituxin, although widely used in cancer treatment, does not result in a cure. The Pointilliste team screens their single-chain antibody libraries with cancer cells in order to isolate patient-specific reagents.
The single-chain antibody sequences are cloned into sub libraries containing different epitope tags, detectable with appropriate capture reagents, allowing sets of antibodies to be isolated. By rescreening of these antibody subsets it is possible to isolate individual antibodies that can be tested for their apoptotic-inducing ability. In addition to these screening platforms, Pointilliste can couple its single-chain antibodies to full-size immunoglobulins, thus preserving their stability in the circulation.
Pointilliste’s fascination with lymphoma as a target for its technology rides high on the wave of current interest in individualized drug therapies. Because the approach promises to be rapid and economical, the concept of designing a therapy tailored to the individual patient could be feasible. This is a tantalizing target for small, nimble biotechs, leaving the big pharma companies to focus on the blockbuster drugs.
Although recent setbacks in the industry have taken some of the gloss off this goal, the promise of billion-dollar returns from a single drug is still too enticing for the giant pharmaceutical companies to abandon.
Pointilliste is still some years away from the marketplace, but the notable successes achieved by Dr. Levy and his colleagues years ago, at a time when antibody therapies were in the doldrums, bodes well for the future of the company.
A Mammalian Display System
Screening systems, such as that described by Dr. Ault-Rich, represent a promising path to the isolation of new therapeutic proteins. However, antibodies can also be generated through selection based on display systems, of which phage display is the most widely exploited. Alternative display technologies include ribosomal, yeast, and mammalian display.
W. David Shen, Ph.D., of the Amgen (Thousand Oaks, CA) antibody engineering group, presented the rationale for his company’s mammalian antibody display system. In contrast to other display systems for antibody fragments, mammalian display offers the advantages of screening for affinity and fully operational immunoglobulins.
“If you focus entirely on high affinity of your antibodies, you may limit the number of epitopes you deal with and lose your most important functional antibodies,” Dr. Shen said.
This approach substantially advances the final goal of obtaining intact therapeutic antibodies. The antibodies are displayed on the cell surface where they can be detected by fluorescent-activated cell sorting (FACS). The system employs site-specific recombination in order to generate homogeneous cell populations.
The single integration site per cell ensures that there will be similar levels of expression per cell, and the levels are high (ug/mL). The molecules are anchored in large numbers to the cell surface through their transmembrane domain.
Using one round of FACS sorting, positive cells can be enriched more than 10,000-fold, much more efficient than other display systems. By adjusting the gating on the FACS and embarking upon rounds of successive screening, high-affinity antibodies (between 125 and 23 pM) can be distinguished and isolated. This system allows selection of antibodies with neutralization functions directly on the cell surface as well as direct determination of their affinity at this point.
Another advantage of the system is the fact that conversion of Fabs fragments into complete IgG molecules is much more straightforward with mammalian display as opposed to individual cloning of fragments, followed by subsequent conversion into entire IgGs.
Dr. Shen and his coworkers were able to move the affinity of their antibodies from 660 to 100 pM by generating a small library (~106) by mutagenesis and moving it through several rounds of sorting.
Despite the advantages of the mammalian display system, Dr. Shen cautioned that there are certain disadvantages associated with the technology. The library size is quite small (106) relative to display technologies such as ribosomal display, in which library sizes of 1014 are common. Background may be a problem, as some antigens tend to stick to cell membranes, although using different cell lines may represent a solution to this issue.
“The technology is ideal for screening libraries made from immunized human or mouse lymphocytes,” Dr. Shen stated. “In this case you aren’t dealing with a nave lymphocyte population and so you don’t need a huge library size. This means that the investigator can concentrate on the functional performance of the antibodies.”
The coming months and years will see a track record accumulate that should allow a side-by-side comparison of mammalian display with the more commonly exploited alternative display systems.
Antibody optimization is the goal pursued by EvoGenix (Mountain View, CA and Melbourne, Australia). The company has focused on the goal of superhumanization, as explained by David Wilson, Ph.D., vp of antibody products.
Over the years, antibody engineers have grappled with the problem of the antigenicity of therapeutic antibodies, with varying degrees of success. In the early days of clinical trials with mouse monoclonals, patient reactions were so severe that the antibodies were ineffective after a first round of treatment.
The situation improved with the introduction of chimeric antibodies, in which murine constant regions were replaced with human sequences. Later humanized antibodies provided even fewer immunogenic sequences, in which the mouse framework regions of the variable portions of the antibody molecule were replaced by human amino acid sequences.
This technology enabled even more successful responses in patients. Finally, mice engineered with human immune systems offer the possibility of fully human recombinant antibodies containing no murine sequences whatsoever. Fully human monoclonal antibodies are produced by transgenic mice, developed by Medarex and Abgenix.
While fully human antibodies have provided reagents of low antigenicity, they carry with them certain disadvantages, including the tightly locked intellectual property. On the other hand, humanized antibodies retain their original epitope specificity, and according to Dr. Wilson, rarely provoke a significant immunogenic response in patient trials. In fact, 7 of 16 therapeutic antibodies currently on the market are humanized.
Humanization improved the response of patients to the injection of a foreign protein, but it frequently resulted in a depression of the antibody’s affinity, as the molecule was stretched into an inappropriate configuration.
A more sophisticated approach is to introduce human framework regions that occur naturally in association with complementarity determining regions (CDRs) that are structurally similar to those of the mouse antibody. This sort of grafting technique results in a modest drop (approximately 14-fold) in affinity.
As Dr. Wilson explained, this loss of affinity could be compensated for by the EvoGene affinity improvement technology. This approach combines RNA-based mutagenesis and ribosome display.
In searching for an ideal mutagenesis platform, Wilson’s colleagues were attracted to the RNA viruses, well known for their genetic instability. This feature of the RNA viruses has confounded the search for successful vaccines for diseases caused by them. An RNA virus that infects bacteria, called Qb, copies its genome with an error-prone replicase that has ideal properties to be exploited for directed molecular evolution.
When a gene for an antibody fragment or any other protein is cloned between the recognition sequences of the Qb replicase, it can be rapidly copied, yielding 12 mutations per gene. The striking feature of this replicase is that there are virtually no biases in the positions or types of mutations that arise, which is far from the case using DNA-based or chemical mutagenesis methods.
This allows the best range of point mutations to be obtained, with minimal alteration of the overall sequence. By isolating improved antibodies and repeating the whole process, better and better antibodies can be isolated.
As Dr. Wilson puts it, “the important point is to follow a minimum mutation pathway’ to identify optimized variants of the starting molecule while minimizing the overall number of mutations.
“The best way to ensure this is to be thorough in the search of sequence variants in the immediate neighborhood of the starting human or humanized protein, which is made possible by exploiting the unique properties of the Qb RNA replicase.”
EvoGenix has employed the Q optimization technology for a number of projects, including improving the enzymatic performance of -lactamase, selective reduction of binding of a growth factor to one of two interacting proteins, and increasing the affinity of antibody fragments.
This approach allowed a 23-fold improvement in affinity of a single domain antibody directed against the malaria antigen AMA-1.
The Evogenix approach offers an innovative means to achieve potent molecular improvement through a minimalist approach that preserves as much as possible the native sequence of the human or humanized antibodies and other proteins, thus minimizing the likelihood of immunogenicity in patients. Although none of these evolved proteins has yet entered the clinic, the biochemical validation of the approach reveals great promise.
Therapeutic antibody technology has now matured to a point at which earlier successes are being advanced into the next generation of improved products.
Herren Wu, Ph.D., senior director of R&D at Medimmune (Gathersburg, MD), discussed his company’s experiences with Synagis (palivizumab), the first monoclonal antibody developed to combat an infectious disease. It is used for the prophylactic treatment of serious lower respiratory tract disease caused by respiratory syncytial virus (RSV).
Vaccine development for RSV has encountered several setbacks due to safety issues and poor efficacy. In contrast, a passive immunization protocol has been proven effective and safe. RespiGam, an early anti-RSV product developed by MedImmune, is a human hyperimmune globulin derived from exposed donors that has been shown to lower respiratory illness and rates of hospitalization. However, the product is far from ideal, being a blood-derived substance, which is cumbersome and expensive to produce.
The introduction of Synagis revolutionized RSV intervention, according to Dr. Wu. It consists of murine complementarity determining regions fused to a human IgG framework. In Phase III clinical trials it was shown to provide large reductions in RSV-induced hospitalization in high-risk infants, according to Dr. Wu. .
Medimmune has built on these early successes through development of a second-generation version of the antibody, with the goal of improved efficacy, penetration and pharmacoeconomics. Dr. Wu stated that the first step was to change several amino acids in the framework and complementarity determining regions of the molecule for further humanization and restoration of the parental light chain CDR1 sequence.
This intermediate molecule was secreted as a Fab fragment in E. coli, and single mutations were introduced in all 59 amino acid CDR positions. A number of mutations resulting in increased affinity due to decrease in koff were identified and these koff mutations were pooled in a combinatorial library.
The library was screened for the best “superantibody” which was expressed, purified, and evaluated for neutralizing activity.
Dr. Wu and his collaborators were able to combine the strongest mutations into a single IgG construct with an increased avidity of 2 logs over the original Synagis, but its increase in neutralizing activity was only two fold.
For this reason, the group concentrated on the isolation of kon mutants. They mutated all CDR residues of one of the best koff-mutants by single amino acid change.
These mutants were screened by a novel ELISA approach developed to pick up only high kon mutants. Identified kon mutations were combined in a library and screened for the best combination of mutations. The final product containing the optimal combination of both kon and koff mutations was given the trade name of Numax.
Numax has been evaluated in Phase I and II trials and two large-scale Phase III trials, comparing Numax with Synagis, are under way.
Numax represents another level in the continuing evolution of therapeutic antibodies in the war against infectious disease. “While we can’t predict how long the approval process will take, we’re looking at 2007 to 2008 as the likely date for arrival in the marketplace,” Dr. Wu stated.