Recombinant antibodies have reached blockbuster drug status. Herceptin, Avastin, Remicade, and others rack up billions of dollars per year in sales. Almost all the clinically approved antibodies in use today are naked, or unconjugated, antibodies. When participants recently met for the Strategic Research Institute’s “World Antibody Summit,” discussion focused on novel approaches to therapeutic antibody development, including immunoconjugates.
In the early days of mAbs, the magic bullet concept was vigorously pursued. Investigators followed the century old notion of designing antibodies that could carry killer molecules to the cell. The idea has great prima facie appeal in that an enzyme-antibody fusion protein is expected to be active against its target at very low concentrations. Theoretically, the conjugate could hone in on a cancer cell, perform a killing reaction, move to the next cell, and repeat the process over and over again.
Unfortunately, this approach was notoriously unsuccessful, since protein toxins stimulated a strong immune response in the host. Even using humanized or human antibodies, such toxins as ricin, diphtheria toxin, and pseudomonas exotoxin when linked to the antibody provoked violent immunological reaction. Moreover, many patients already have antibodies against bacterial toxins, developed in earlier bouts of infection.
Maytansinoids as Immunoconjugates
Rajeeva Singh, Ph.D., director of biochemistry at ImmunoGen (www.immunogen.com), adapted the magic bullet approach. His group focuses on the maytansinoids, highly potent cytotoxic agents that can be easily coupled to antibodies.
Since one would not anticipate that the conjugate molecule could work in a catalytic fashion, it is essential that compounds with extremely high potency be incorporated into the procedure. Many poisonous substances must be present in micromolar levels to exert significant cytotoxicity. Delivering such a high level of toxin to the cell is clearly impossible, precluding their use as immunotoxins. This requires that only the most potent candidates can be considered for this system.
Maytansinoids fit the bill, as these inhibitors of microtubular assembly boast IC50s in the subnanomolar range. The members of the maytansinoid family are bacterial compounds that are extensively modified to improve stability and solubility.
ImmunoGen chemists increased the stability of a disulfide linker by alkyl substitution, while a thioether conjugate allows a covalent attachment to the antibody.
In a model system of human colon cancer xenografts, the human antibody HuC242 coupled to a synthetic maytansinoid has effective killing power. The controls without antibody or applied to other cell lines that lack the target antigen, however, had little effect.
A surprising property of the system is a bystander effect by which the antibody-drug conjugate is seen to kill nontarget cells in close proximity to the tumor cell when challenged in vitro. When cells containing the antigen targeted by the HuC242 antibody are mixed with antigen-negative tumor cells, both tumor types are destroyed following treatment with the conjugate. This property could help to eliminate nearby tumor cells that have not taken up the drug.
Currently, ImmunoGen is moving the maytansinoid-antibody conjugates into the clinic through a collaboration with Genentec(www.gene.com). Using Herceptin conjugated with a noncleavable, thioether linker, a Phase I study of HER2 positive, metastatic breast cancer patients showed it was well-tolerated. Also, four out of 10 patients demonstrated an objective response. The study was reported at the recent “American Society of Clinical Oncology” meeting.
In another series of trials, patients were evaluated with ImmunoGen’s HuN901 antibody conjugate targeting multiple myeloma and CD56-expressing solid cancers as well as small-cell lung cancers. Once again, the drug was well tolerated, and several patients showed objective responses.
The antibody-maytansinoid conjugates with sterically hindered disulfide linkers appear to be an appealing and novel approach to the issue of malignancy. ImmunoGen is engaged in a number of clinical and preclinical programs in collaboration with Centocor(www.centocor.com), Biogen-Idec (www.biogen.com), Biotest(www.biotest.com), Genentech, and sanofi-aventis (www.sanofi-aventis.com).
Other Immunotoxin Options
Duocarmycins are another class of small molecules whose high toxicity makes them strong candidates for anticancer immunoconjugates, according to David J. King, Ph.D., director of molecular biology and biochemistry at Medarex (www.medarex.com). The company is known for its transgenic mice that are programmed to produce fully human antibodies. These purine ring structures are toxic in the picomole range, but cell lines resistant to duocarmycins frequently arise through the mechanism of multidrug resistance, presenting a challenge to the Medarex team.
Chemotherapy frequently fails because resistant populations of tumor cells arise within the body of the patient. The multidrug resistance mutation, as the name implies, is especially devastating because many different therapies no longer block progress of the malignancy.
To overcome this shortcoming of the original duocarmycin molecule, Dr. King and his associates modified the basic structure. The simplified form is referred to as CBI, which must be chirally synthesized. When further modified to form a DNA-binding alkylating agent, the compound is no longer subject to removal by the multidrug resistance pump mechanism.
Medarex is pursuing a program to develop immunoconjugates using the duocarmycins coupled to antibodies directed against different cancer-related markers including CD70 and the prostate specific membrane antigen (PSMA).
Dr. King and his colleagues have investigated the antigen CD70, a type II membrane protein of the TNF superfamily. Expressed on activated T and B cells, it functions in the regulation of lymphocyte activities. It is also expressed widely in cancer cells including lymphomas, myelomas, and especially clear-cell renal carcinoma.
The company’s anti-CD70 human antibody is rapidly internalized in renal cancer cell lines, according Medarex. In Severe Combined Immunodeficiency Disease (SCID) mice, the antibody-drug conjugate was much more potent against the tumor than the drug alone, reported Medarex.
In another set of studies, the team built immunoconjugates between CBI and a human anti-PSMA antibody with high affinity for the antigen. The PSMA antigen is a glycosylated 100 kD homodimer molecule, highly expressed in prostate epithelial cells and in some prostate cancers. In the LNCaP prostate cancer cell line, the antibody is rapidly internalized, according to the company.
The Medarex team designed the conjugate using either a hydrazone or peptide linkage, both of which proved to be stable. After subjection to the linkage reaction, the antibody retained its binding capability. The antibody has high affinity for and rapidly internalizes into PSMA expressing cells, according to Medarex. Moreover, in a SCID mouse model system, the antibody conjugate was much more effective at suppressing tumor growth than the free drug.
A hallmark of the cancer cell is its ability to avoid apoptosis. Targeting the TNF-related apoptosis inducing ligand (TRAIL) receptor with appropriate antibodies can trigger the self-destruction of cancer cells, providing a new approach to tumor therapy, noted Gilles Gallant, Ph.D., vp, clinical oncology at Human Genome Sciences www.hgsi.com).
Dr. Gallant discussed how activation of the TRAIL receptor family prompts tumor regression and inhibits new tumor growth. The TRAIL receptor is, in fact, a family of receptors that when stimulated embark upon a cascade of destruction inducing caspases that rampage through the cell with resulting blebbing, fragmentation, and finally engulfment by macrophages.
A group of scientists at HGS developed a mAb, Lexatumumab (HGS-ETR2), capable of inducing apoptosis via activation of the TRAIL-R2 receptor. The antibody binds to TRAIL-R2 with high specificity and is effective in inducing apoptosis in vitro in the colo205 colorectal cell line, reported HGS. This is not surprising, given the high level of expression of TRAIL-R2 on colorectal cancer xenografts.
Lexatumumab causes regression of large colorectal tumors in the mouse model as well as non-small-cell lung carcinoma in a comparable system. Moreover, the antibody works effectively in vitro in combination with traditional therapies. The HGS team observed that the antibody in combination with cisplatin is much more effective at suppressing tumor growth than cisplatin alone.
Further investigations by the team showed that Lexatumumab is effective in a wide variety of tumor types in preclinical settings. This encouraged the team to embark upon a series of Phase I trials to evaluate tolerance and consider adverse effects of the therapy. The results of the these investigations showed that the antibody is well tolerated. Lexatumumab is currently being studied in a Phase Ib study in combination with different chemotherapy agents.
Alternative Receptor Targets for Cancer
Another appealing target for antibody-based cancer therapy is the protein EphA2, a transmembrane receptor tyrosine kinase, vital for the regulation of cell growth, survival, angiogenesis, and migration. “Because EphA2 is overexpressed on a wide variety of solid tumors, selective targeting of this protein with an antibody conjugate may represent a novel therapeutic approach,” stated Steve Coats, Ph.D., director of cancer research at MedImmune (www.medimmune.com).
Preclinical studies demonstrate that increased tumorigenicity is correlated with overexpression of EphA2 in cancers including breast, ovarian, cervical, renal, and prostate. Furthermore, this overexpression is frequently associated with poor survival.
MedImmune researchers humanized mouse mAbs using framework shuffling of human germline genes. Light and heavy chains of each parental mAb were simultaneously or sequentially humanized through one- and two-step strategies. The resulting mAb, 3F2-3M, was raised against EphA2 and retains all of its murine functional activities.
It has been shown to activate natural killer (NK) cells and requires them for antibody-dependent cell cytotoxicity. It is also effective in suppressing tumor growth in a SCID mouse model in the presence of NK cells, according to the company.
Further research by MedImmune involved in vivo and in vitro study of a fully human phage-derived antibody candidate against EphA2 that was conjugated to the potent inhibitor auristatin. The antibody was shown to inhibit tumor cell growth in vitro and to induce cell apoptosis, as measured by caspase activation. Currently, the company has several antibodies conjugated to auristatin in preclinical development.
Engineering of Therapeutic Bispecific Antibodies
A wealth of options are now open to antibody engineers as years of investigation bears fruit, remarked Alexey Lugovsky, Ph.D., head of molecular modeling at Biogen Idec. The availability of user-friendly approaches such as computer modeling and in vitro evolution techniques allow sophisticated design and exquisite fine tuning of antibody molecules.
Dr. Lugovskoy emphasized the benefits of structure-guided protein design, by which a vast number of candidate sequences can be enumerated and assayed in silico, and promising representatives can be tested experimentally to identify fine changes leading to properties of interest. Some recent successes of structure-guided protein design are the identification of novel protein folds, protein functions, as well as protein stabilization and affinity optimization.
An interest of Dr. Lugovskoy’s group is the building of bispecific therapeutic antibodies with the appearance of two Y- shaped molecules joined end to end, where a single-chain Fv fragment (scFv) of a particular specificity is linked to full IgG of a different specificity.
The proof of concept “Hercules” bispecific antibody is an engineered construct consisting of an anti-TRAIL antibody fused to a single-chain anti-LTbR antibody (lymphotoxin beta-receptor, another member of the TNF receptor family). The Hercules antibody can crosslink the two receptors, thus enhancing an antitumor response. Through intensive collaborative engineering, the molecule was stabilized, taking advantage of sequence databases and proprietary crystal structures as well as screening of antibody libraries.
These in silico engineering techniques such as homology modeling, electrostatic optimization, sidechain repacking, statistical sequence analysis, and knowledge-based design yield candidate sequences that then are constructed and screened in a thermal challenge stability assay. These include an assay that checks the real life stability of the molecules as compared with computer predictions. By using these structure-guided methods the sequence space was reduced to a manageable size by which 100 candidate mutants in just over 40 positions were screened from four genetic antibody libraries.
A stable, robust, and nonaggregating bispecific antibody is all well and good, but does the new entity have therapeutic potential? An improved Hercules II antibody possesses these properties and, moreover, exhibited potent killing in a cell death assay comprising multiple tumor cell lines, according to Biogen Idec. The merit of the bispecific antibody concept was born out of the fact that the Hercules II Ag is more potent than combinations of the single antibodies in some of the tumor cell lines.
Targeting Tumors Through Their T-cell Receptors
“Are the 90 percent of proteins localized within the cell possible targets for antibody therapeutics?” asked Jon Weidanz, Ph.D., chief scientist and founder of Receptor Logic (www.rectptorlogic.com). Dr. Weidanz’ company uses the major histocompatibility complex, or HLA, as a bridge for delivering therapeutic antibodies to the cancer cell. The company’s rationale is based on exploitation of T-cell recognition systems.
“We took advantage of a novel protein, an RNA helicase known as p68,” Dr. Weidanz continued. “This protein is overexpressed in cancer cells, and peptides from it are displayed in the cell’s MHC I receptors.”
MHC class I molecules display peptides derived from self proteins in a special peptide binding cleft. Malignant and diseased cells display intracellular peptides on their surfaces, which allow them to be recognized and eliminated by killer T cells.
Receptor Logic researchers produced an antibody specific for the peptide (YLLPAIVHI)-HLA-A2 complex that they had found on the surface of a breast cancer cell line. This peptide, a component of the helicase protein, was shown to be overexpressed in a variety of cancer cell lines, both at the RNA and the protein level.
The antibody behaves as a T-cell receptor mimic and can bind to and be internalized by the appropriate cancer cells. So the question is, can the antibody, which will be internalized by the cancer cell, be used as a therapeutic device? Potential killing mechanisms include complement activation, antibody-dependent cellular cytotoxicity, and cross linking at the cell surface.
Dr. Weidanz found that a peptide from human chorionic gonadotropin beta could be used to trigger an antitumor response. This protein also is an overexpressed, cancer-related marker observed in the human breast cancer cell line MDA-MB-231. Using an antibody against the peptide-MHC I receptor complex, Dr. Weidanz showed that the tumor cells were killed by complement activation as well as through antibody-dependent cell cytotoxicity.
Moreover, in an in vivo nude mouse model, the antibody prevents tumor growth in the human breast cancer cell line. In these experiments, the mechanisms by which the tumors are eliminated have not yet been defined. It may be that the antibodies trigger the host immune response, as appears to be the case for naked antibodies such as Rituximab used to treat B cell non-Hodgkin lymphoma and Alemtuzumab for patients with B cell chronic lymphocytic leukemia.
Shooting at a Moving Target
With the announcement of each new advance in cancer treatment there is great euphoria, followed by optimistic predictions, and then depression when these starry-eyed prophesies are not realized. It is interesting to compare the last 25 years of cancer research with the vast improvements that have occurred during this period in the field of HIV investigation. Although there is no cure or vaccine for AIDS, the condition for many sufferers is a manageable, long-term, chronic illness, and afflicted individuals can, with luck, anticipate a long life span.
In many ways the cancer dilemma is comparable; a population of highly unstable, mutatable cells that is constantly evolving resistance to the cancer treatment de jour. There is now a host of candidates waiting in the wings for their opportunity to prove their mettle.
No drug moves into human cancer trials without extensive preclinical investigation, and many of the agents covered in this review are quite effective in suppressing xenografted tumors in the mouse SCID model system. However, human tumors are highly heterogeneous and there is every reason to believe that no matter how effective innovative antibody therapies are in preclinical models, resistant populations will always arise in the cancers of real patients, making the victory a short-term one.
This means that more than one of the options discussed here may be required in adjuvant protocols, and there will be room for a number of antibodies aimed at totally unrelated targets.
This strategy, which has met with such notable success in the struggle against AIDS in the last 25 years, may represent the guiding principle in the next round of therapeutic cancer intervention.