Meeting Report

Thousands of years ago humans discovered the medicinal properties of certain plant and animal products, most likely by trial and error. In modern times as the ability to isolate individual components progressed, drug discovery began to rely increasingly on an understanding of the molecular basis of disease. This trend culminated in the latter part of the 20th century with the concept of rational drug design, which often uses information about the atomic interactions of proteins and small molecules with their biological targets to design drug candidates.

The approach has proved difficult because many compounds that look promising in vitro fail to work in humans. While there have been several notable successes, there is always room for a parallel approach. Some scientists are even looking to the ancients for guidance in drug development.

On September 9, 2009, three researchers presented their work at a meeting of the Academy's Chemical Biology Discussion Group showcasing different ways to identify small molecules that could become promising drug candidates.

Restoring function to nuclear hormone receptors

John Koh of the University of Delaware uses rational design to target mutations in nuclear hormone receptors, including the thyroid, androgen, and vitamin D receptors. These receptors are modular, with distinct ligand-binding and activation domains. This allows researchers to manipulate one domain without affecting the activity of the other. Although the structures of their ligands are quite different, the overall architecture of the receptors' ligand-binding domains is the same. They also tend to operate by the same mechanism, said Koh.

Nuclear hormone receptors share a similar architecture in their ligand-binding domains despite marked differences in the ligands they bind.

Several diseases result from receptor mutations that impair ligand binding. One is resistance to thyroid hormone (RTH), which is caused by an autosomal dominant mutation. Koh and his team reasoned that if they could make a compensatory change in the ligand and "rescue" or restore binding, they would have an effective treatment for the disease.

Koh explained that this project faced a particular challenge because two different but related receptors bind thyroid hormone; his team needed to target the mutant specifically. Any modified ligand that bound to both the mutant TRβ and normal TRα would cause overstimulation of the latter, possibly resulting in tachycardia.

The team used a combinatorial approach to design a series of compounds that were predicted to bind only to the mutant receptor. They eventually found one that was both potent and selective, and could rescue the three most common mutations that are associated with RTH.

The mutations involved in ligand-dependent activation cause a 500-fold drop in receptor response, compared to a 3- to 10-fold drop caused by mutations in the ligand-binding domain. After understanding the mechanism of activation at the structural level, Koh's team designed compounds that could activate the mutant receptor. To rescue mutations that affect the coactivator domain's ability to function, they added an activation domain to the ligand itself, along with a scaffold to enable the connection. Although they were able to construct a good scaffold and add an activation domain to it, they found that their compounds have difficulty dislodging a corepressor from the DNA to which they need to bind.

Another big target for Koh's group is the androgen receptor. Anti-androgens, such as flutamide and bicalutamide, are effective in inhibiting prostate tumor growth. However the cancer can become refractory to this treatment, often because of mutations that develop in the androgen receptor. In some cases the mutations cause anti-androgens to act as agonists rather than antagonists.

Once Charles Bell and James Dalton of Ohio State University determined the crystal structure for such a mutant in 2005, Koh's team was able to begin to understand the physical mechanism causing the change in function. After examining this structure, they hypothesized that they could develop small molecules with bulky side chains that would act as antagonists of the mutant receptor. This added bulk would block part of the receptor from moving into the position it adopts in the agonist conformation. A number of these compounds acted as antagonists to both the wild-type receptor and several different bicalutamide-resistant mutants.

They tested their compound in a prostate cancer cell culture to see if resistant colonies of cells would develop over time. Comparing the structures of successful compounds to those to which resistance is developed, Koh showed that the successful compounds had bulkier structures. He suggested that no single mutation can change the receptor structure enough to overcome the blocking activity of the bulky structures. It remains to be seen if there are other mechanisms by which resistance to anti-androgens can develop in prostate cancer tumors. But this investigation suggests that rational design can be used to overcome at least one kind of mutation that renders the cancer treatment ineffective, or worse, detrimental to the patient.

The value of traditional medicine

Akira Kawamura's team at Hunter College is using ancient wisdom to guide new drug development. As he explained, natural products have been used for thousands of years to treat all kinds of conditions. Time has shown many of them to be safe and effective, but their mechanisms of action are often unknown. Kawamura's group is trying to identify the active ingredients of some widely used Oriental herbal medicines.

As an example of their approach, Kawamura described the analysis of Toki-shakuyaku-san (TSS), which is used to treat blood circulation disorders. The medicine is a mixture of six components—two roots, three rhizomes, and one mushroom. Kawamura's group treated human umbilical vein endothelial cells with the medicine for four hours, isolated mRNA, and performed gene chip expression analysis.

Among the genes that were upregulated in response to the treatment were SerpinB2 and COX2, both of which have been shown to affect the circulatory system. Fractionation of the medicine revealed that a simple polyacetylene compound was responsible for increasing SerpinB2 levels. Although the compound had previously been identified, its mechanism of action was not known. Further studies on its effects on coagulation disorders are underway, said Kawamura.

JTT is a complex mixture with immune regulatory activities.

Kawamura's group is also working on Juzen-taiho-to, a combination of ten components used as an immune booster. Preclinical studies by others have shown that the beneficial effects are mediated by monocytes and macrophages. Kawamura's team performed gene expression profiling on cells treated with the medicine and found that many upregulated genes are regulated by NF-κB, a transcription factor important in the immune response to infection. They focused on the NF-κB target ICAM1 to guide the purification. After some difficulties, they found that diacylglyceryl glycosides were responsible for the immune stimulation and that the activity of these compounds was modulated by phytosteryl glucosides that eluted in a different fraction.

Kawamura noted that studying mixtures of compounds is a difficult process because activity can disappear when conditions change and interactions are difficult to tease out. But the group is starting with a treatment that many people already find helpful, giving them an enormous advantage.

Identification of immunobiomarkers

Tom Kodadek and his team at Scripps Jupiter are trying to develop diagnostics that can identify antibodies and T cells associated with a disease. The notion is not new, Kodadek said, but immunologists have traditionally believed that they first have to identify the specific antigen to which the antibody binds to isolate the antibodies they want. Because identifying the antigens has been difficult, very little progress has been made toward isolating antibodies for use as biomarkers.

Kodadek's team has instead developed peptoids—synthetic mimics of protein antigens that can be used in differential screening of serum of healthy people and patients with a specific disease. Kodadek emphasized that the screening technique, though microarray-based, was not designed to read "fingerprints" or develop a profile of a disease, but rather to identify synthetic molecules to which the difference in levels of antibody-binding in the healthy person and the patient is like "night and day."

A new way to identify antibodies indicative of a disease state.

As a proof-of-concept, they used the experimental allergic encephalomyelitis (EAE) model of multiple sclerosis in mice. Here, mice are injected with a peptide antigen and adjuvant that induces its immune system to attack the myelin sheath, leading to an MS-like condition. Kodadek and his colleagues incubated serum from EAE mice and control mice with peptoids on microarrays, looking for instances where the signal that indicates antibody binding to a peptoid was far more intense in diseased mice than in control mice. To ensure that the results were accurate, they repeated the experiment with the peptoids that gave a strong result in the "training set." A number of controls showed that they were capturing the antibodies to the antigen that induced the disease in EAE mice, though there was no obvious sequence similarity between the antigen and the peptoid.

Kodadek presented some preliminary results using the approach on patients with Alzheimer's disease. In the training set, his team compared serum from six patients known to have Alzheimer's disease based on postmortem autopsies with serum from control patients and some patients with Parkinson's disease. They identified three peptoids for further analysis in blinded studies. A similar study is being conducted with lung cancer patients. Kodadek noted that it was important to determine how early these signals appear during the development of the disease.

Ultimately, Kodadek would like to develop diagnostics and tools targeting the T cells and antibodies that cause diseases such as multiple sclerosis. He hypothesized that the main difference between the T cell populations of diseased individuals and healthy ones is that the former will have a high prevalence of cells bearing the T cell receptor (TCR) that recognizes the disease-causing antigen.

To find these cells in EAE mice, the group began by isolating CD4+ T cells from control and EAE mice, labeling the former with green quantum dots and the latter with red. The cells were mixed and screened against 500,000 peptoids. Only two peptoids were capable of binding the red-labeled T cells but not the green-labeled T cells, indicating that they were specific for the cells from the disease group. Kodadek's team went on show that the peptoids specifically targeted cells bearing the autoimmune T cell receptor, and have some evidence that it is binding the TCR itself. Titrating in the specific peptoid prevented antigen-presenting cells bearing the antigen from stimulating the proliferation of the autoimmune T cells.

Once the researchers were able to specifically target the autoimmune T cells, the next step was to destroy them. To do this they attached a "warhead" consisting of a rubidium tris-bypyramidyl compound to the end of the compound. After exposure to visible light, the compound mediates the conversion of triplet to singlet oxygen—a short-lived molecule that is very destructive to proteins. Ultimately, they would like to attach the peptoids to antibodies that can stimulate the destruction of the undesirable cells.

Although the evidence that Kodadek and the other speakers showed presented is still preliminary, their talks made clear how the powers of modern computational approaches, chemical synthetic techniques, and traditional knowledge can yield promising results.

Open Questions

Can scientists design ligands that are resistant to compensatory mutations in the receptors they bind?

Can ligands bearing activation domains be designed to displace corepressors from DNA?

How can the activities of traditional herbal medicines be assayed in culture?

When a cellular response is due to the activity of more than one compound, how can the active ingredients be isolated?

Does the immune response to a developing disease begin early enough to detect with a peptoid assay?

How does variability in the immune response of individuals affect the peptoid assay?