ACADEMIA
Research may lead to design of new drugs for breast cancer treatment
A team of biomedical researchers based at the San Diego Supercomputer Center (SDSC), the Moores UCSD Cancer Center, and the University of California, San Diego (UCSD) has made an important discovery in modeling the functions of enzymes in the human body based on their molecular structures -- a discovery with implications for drug design and cancer treatment.
In the journal Current Medicinal Chemistry, Igor Tsigelny, Vladimir Kotlovyi, and Linda Wasserman describe how they used high-performance computing techniques developed at SDSC and at the nearby Scripps Research Institute to model the effects of variations in a particular enzyme that have been shown by studies to increase the risk of breast cancer. The study provides insight into the chain of events that links the changes in the structure of the enzyme to the increased tendency for people with the mutation to develop breast cancer. "Medicinal chemists are increasingly using the 3-D structures of proteins, DNA, and RNA as a basis for developing new drugs," said Tsigelny, a Project Scientist at SDSC and in the Department of Chemistry and Biochemistry and the Department of Pharmacology at UCSD. "For example, structure-based drug design led to the development of HIV protease inhibitor. But scientists do not have very many completely solved 3-D structures of biomolecules to work with. In the absence of experimental data on protein structure, supercomputer-driven modeling programs can make useful predictions of molecular structures that can give us enough information to understand the interactions of biomolecules, and to make predictions about specific, functionally important regions. Our work demonstrates that structural predictions combined with genetic analysis can be a useful tool for drug design and possibly for individualized treatment of breast cancer." "One of the principal goals of the Integrative Biosciences program at SDSC is to use information technology to speed the integration of basic research discoveries into clinical appraisals, and to help in mitigating risks to individuals," said SDSC director Francine Berman. "The work reported here is a perfect of example of how a simulation can speed the first step. The results of analyzing the computer models suggest the nature of the problem, and that alone leapfrogs the seven or more years of investigation typically required to determine the structure of the enzyme by experimental methods." "This research helps us understand how a mutation can alter the functioning of a vital enzyme in an unexpected way, and thereby increase or decrease people's risks of developing certain kinds of cancers," said Mark Miller, Associate Program Director in Integrative Biosciences at SDSC. "Once we identify the basis for the risk, we can take steps to mitigate that risk." The cytochrome P450 family of enzyme proteins has more than 1,000 members, and individual members have a very high degree of structural similarity -- all of the family members can be sorted into five basic structural plans, with only minor variations. These enzymes are ideal candidates for simulations that predict variations in structure caused by minor changes in composition. Although all humans have basically the same set of P450 enzymes, subtle variations can be found within a human population. The variations are usually the result of single nucleotide polymorphisms (SNPs), variations that occur when one of the coding units (adenine, thymine, cytosine, or guanine) in the instructions for protein synthesis along the DNA molecule is changed. The SNPs cause subtle changes in the enzyme structure and function. Many of the P450 enzymes break down toxic chemicals, while others are extremely important in processing many kinds of pharmaceuticals, so minor changes can have major effects. Population studies of cancer risks indicate a significant correlation between SNP-induced changes in cytochrome P450 proteins and the incidence of breast cancer. The gene for cytochrome P450 type 19 (called CYP19 or aromatase for short) is involved in the production of the growth-promoting hormone estradiol, the most active known naturally occurring mammalian hormone of its class. "Our study comparing women with breast cancer to women with no history of breast cancer has shown that women who have two copies of the '264Cys' variant of the gene that codes for CYP19 have a greatly increased risk for developing breast cancer," said Wasserman, an Associate Professor of Clinical Medicine in the Department of Medicine at UCSD who, like Tsigelny, is affiliated with the Rebecca and John Moores UCSD Cancer Center. The researchers used molecular modeling to shed light on this correlation. "We wanted to push the envelope to see if we could understand why this particular SNP should be associated with increased risk of breast cancer," said Wasserman. "We created a structural prediction of CYP19 based on three other cytochrome P450 family members whose structures were known in detail from experimental data. Looking at a static image of the enzyme model, it was clear that the mutation did not occur in an area we would expect to alter the enzyme's ability to create estradiol. It left us puzzled. So we tried to go beyond the static picture of the enzyme, and model the dynamic interaction of the hormone molecules with CYP19 through the entire passageway, from the place where estradiol is created to its exit from the enzyme." Modeling the dynamic interaction of hormone molecules with CYP19 explained the mystery. A molecule of testosterone moves into its core of the enzyme, is transformed into estradiol, and then moves out of the core. Using a supercomputer-based simulation program called DOT developed at SDSC and the Scripps Research Institute, the team investigated the effect of mutations on the creation and migration of estradiol through the CYP19 molecule. They found that the 264Cys SNP modifies a pathway for movement of estradiol out of the enzyme core. This results in more efficient extraction of the hormone molecule from the enzyme and increases the general level of estradiol in the body. "We believe that mutations that change the electrostatic profile in this region can increase or decrease the release of estradiol and affect the probability of breast cancer development in women," Tsigelny said. "This would explain why women with the CYP19-264Cys mutation have higher rates of breast cancer. It should now be possible to design inhibitors of these P450 proteins that would be very promising drugs for treatment of post-menopausal breast cancer." The simulations were performed on "Ultra," a 64-processor Sun HPC 10000 parallel computer at SDSC. "The access to high-end computation was really key in this experiment," Miller noted. "Looking at the protein as a still picture just wasn't good enough. This is an area where computational power and good modeling techniques can help advance medicinal chemistry." "Protein structure prediction programs are improving very rapidly," Tsigelny said. "The number of publications on this topic has more than doubled in the last five years. I believe that protein structure prediction applied to high-throughput drug design is becoming an essential weapon in the fight against cancer. It is clear from our study that a clear hypothesis for cancer risk mitigation can be developed and tested without the need for years of experimental studies on enzymes' structures. I believe that success stories from this type of approach will become much more common in the next decade." Support for this work was provided by a grant to Linda Wasserman from the Susan G. Komen Foundation and a gift to the Moores Cancer Center from the Avon Foundation. Publication reference: Igor F. Tsigelny, Vladimir Kotlovyi, and Linda Wassermann, "SNP Analysis Combined with Protein Structure Prediction Defines Structure-Functional Relationships in Cancer Related Cytochrome P450 Estrogen Metabolism," Current Medicinal Chemistry, Vol. 11, No. 5, March 2004.
