Yuan-Ping Pang, Ph.D.

Star Trek's "holodeck" has nothing on the visualizations of Mayo Clinic chemist, Dr. Yuan-Ping Pang. While the 3-D holographic chamber on the starship Enterprise projected imaginary interactive images, those of Dr. Pang's lab are all very real. They are scientific visualizations of molecules that cause infectious diseases; he creates them to show researchers what their quarry looks like. In this dimension of inner space, he is going where no one has gone before.

The supercomputing capability needed to build computer-simulations was, in its early years, a resource developed and used within the physical sciences. Yet with the advent of the human genome project, supercomputing power needed to simulate complex biological systems. That necessity led Dr. Pang to establish CAMDL at Mayo's Rochester campus in 1997.

To have supercomputing capability on a medical campus is not unique, but amassing super computing power per user is.

"Other research institutions have the same or better supercomputing ability that is shared by many users. Funding from DARPA (Defense Advanced Research Projects Agency, of the Department of Defense) has enabled us to design and build dedicated, terascale supercomputing ability. This ability is extremely valuable for us in exploring novel microscopic solutions to challenging macroscopic problems," says Dr. Pang.

CAMDL specializes in developing computer simulated models aimed at the discovery of new treatments for infectious diseases and cancer. It is one of few labs conducting advanced research in computational, medicinal, synthetic and combinatorial chemistry under one roof.

var so = new SWFObject("upload/de07-1-trans-pang_001.swf", "photogallery", "320", "270", "8", "##000000"); so.write("flashcontent"); VIDEO

Dr. Pang describes the process of creating a molecule in a 3D environment.

The laboratory houses supercomputing hardware and software used to process highly complex biological data, and develop comprehensive databases of three-dimensional molecules. Dr. Pang has adapted his imaging concepts from the small computer screen to a large wall screen where visitors are drawn into a three-dimensional, sub-microscopic world. Researchers can examine and study, in simulation, microsecond-scale proportions of proteins and enzymes associated with malaria, avian flu and severe acute respiratory syndrome (SARS). This ability has led to significant discoveries in the lab that Dr. Pang says will impact the prevalence and spread of infectious diseases.

Avian Flu

Also known as bird flu, avian influenza was first identified in Italy more than a hundred years ago, but confirmed cases of human infection have been reported since 1997. According to the Centers for Disease Control, 200 confirmed cases of human infection with avian influenza A (H5N1) virus have been reported since 2004. In 2005, Mayo Clinic and University of Wisconsin researchers reported that many indicators suggest the H5N1 virus is closer to extending beyond Southeast Asia into the worldwide population.

Dr. Pang has developed a 3-D model of neuraminidase (NA) of the H5N1 virus. He found that the conformation (shape) of the inactive, monomeric NA of the H5N1 virus is very different from the conformation of the active, tetrameric NA of the H5N1 virus. He proposed that small molecules that stabilize the monomeric NA can disrupt the tetrameric NA and thereby inhibit NA of H5N1. This opens a possibility to use multiple antiviral agents that could simultaneously target multiple forms of the NA molecule to minimize the resistance problem of current anti-influenza drugs.

Malaria

In June 2005, Dr. Pang identified a target site within malaria-carrying mosquitoes that could be used to develop human-safe pesticides that are toxic only to the Anopheles gambiae mosquito and other mosquito species. If supported by further studies, the findings could offer a safer and more effective control of mosquito-borne diseases such as malaria.

Dr. Pang identified two unique amino acid residues cysteine (286) and arginine (339). These exist in three mosquito species and the German cockroach.

The findings are significant because the residues could potentially be used as a target site for a pesticide that would incapacitate only insects that carry these residues, which do not exist in mammals.

These findings suggest that new pesticides can be designed to target only the mosquito enzyme. Such pesticides could be used in small quantities to harm mosquitoes, but not mammals. Dr. Pang has developed a blueprint for a pesticide that could incapacitate malaria-carrying mosquitoes and he is currently developing a prototype of the new pesticide. "This discovery could potentially revolutionize the control of mosquito-borne diseases without using existing pesticides that are toxic to humans," Dr. Pang said.

SARS

Close-up view of the computer-derived SARS viral enzyme (surface model in green) in complex with its peptide target (stick model in orange, blue and red) that will be cleaved by the enzyme upon binding.

In 2004, Dr. Pang was the first to develop a series of three-dimensional models of an enzyme responsible for replicating the SARS virus. These models, instantaneous "structures-in-time," are central to designing an anti-SARS drug - and are therefore a welcome advance as the virus continues to threaten public health.

"We successfully developed a computer 3-D model of a SARS viral protease using the SARS viral genome information only, computationally identified small molecules against the 3-D model, and experimentally confirmed that these computer-identified small molecules are able to penetrate cells, inhibit the human SARS-CoV Toronto-2 strain, and rescue the cells from viral infection," Dr. Pang said.

Dr. Pang analyzed the SARS viral genome and built, atom by atom, the instantaneous 3-D structures of the viral enzyme (called chymotrypsin-like cysteine proteinase) - each of composed of 8,113 atoms - just 20 days after the SARS viral genome was made public by the Centers for Disease Control and Prevention (CDC).

By performing exceptionally large-scale computer simulations with his powerful computer system, Dr. Pang quickly and correctly converted a genomic sequence into the 3-D model of a protein that encoded the blueprint for an anti-SARS drug. This ability is crucial in digesting the information available from the emerging fields of genomics and proteomics and in combating emerging infectious diseases.

Shape and polarity properties of the SARS viral enzyme (red: negatively charged; blue: positively charged) use to match the properties of small molecules as potential anti-SARS drugs.

Teraflops on Demand

The CAMDL computing system is a terascale system, one the most powerful available. Terascale measures computational power in teraflops, or floating point operations per second, and is capable of processing one trillion calculations per second. CAMDL consists of two terascale systems of 1,060 processors that have been strung together for a combined power of 3.8 teraflops, which reflects teraflop power per user or research group. Dr. Pang is pushing toward faster computer systems and within two years, hopes to have achieved petaflops status, equivalent to 1,015 teraflops or 1,000 trillion floating-point operations per second.

"Our supercomputer is just a Strad (a Stradivarius violin): whether the sweet music can get out of this instrument is largely dependent upon the player," says Dr. Pang.

It would take Dr. Pang 20 days to complete a simulation of a protein structure using the terascale systems he built as compared to the same simulation using a single conventional desktop computer which would take about 28 years.

Analyzing biological data of the smallest scale can shed new light on disease processes, but it was once thought that computer simulations would provide less detail, not more. "Many biological events, such as protein folding, take place in time beyond the nanosecond scale and scientists questioned whether simulations would provide useful data," Pang says. "What we've done is break one 28-year-long, microsecond-scale calculation into 590 nanosecond-scale calculations performed simultaneously on 590 processors for 20 days, a process called multiple molecular dynamics simulations (MMDSs). Without this capability, this kind of work wouldn't be possible in my lifetime."

Prior to the onset of supercomputer-aided 3-D modeling prediction for proteins, researchers use X-ray crystallographic or Nuclear Magnetic Resonance (NMR) Spectroscopy method to determine protein structures. This is still common today, however, there are challenges in crystallographically determining flexible regions of proteins and in determining large proteins with the NMR method. Dr. Pang's use of the MMDS method on powerful computers to predict 3-D protein models complements the X-ray and NMR methods.

Computer simulations hold endless potential for medicine from significantly reducing the time in which it takes discoveries to reach the bedside to reducing the cost of expensive clinical trials that could be done in simulation.

"Our computational approach potentially offers a solution to reduce manpower requirement which represents one of the greatest limitations to drug discovery and in addition, it ultimately improves the success rate of moving discoveries from the laboratory to the patient," Dr. Pang says.

Other Ongoing Studies

Dr. Pang and colleagues are also working on developing chemical antidotes for botulism, caused by the bacteria Clostridium botulinum. The toxin disrupts nerve function, causing paralysis. An ongoing anti-cancer research project seeks to use genetic code, high-performance computing, modern chemical synthesis, and cancer biology to develop chemicals that can selectively block the function of a cancer-related protein called JNK2. Funded by a Minnesota Partnership grant, this project is in collaboration with Zigang Dong, M.D., Dr.Ph., of the Hormel Institute, University of Minnesota .