The image of a molecule dances on the surface of a drug crystal. Computer-generated lines appear, showing where hydrogen bonds may form between the molecule and the surface. A stream of energy data scrolls through another window on the screen. A formulation chemist sits in front of the workstation comparing the numbers with those generated for other additives and solvents. Welcome to the virtual laboratory.
This scene is increasingly commonplace in the world's commercial and academic research labs. Molecular modeling and simulation are maturing. The technology, combining computer graphics with scientific computation, was born in the early 1980s and first used by pharmaceutical companies seeking to understand the interactions between drug candidates and protein-active sites. Simulation steadily diversified--first to address problems in polymer science and then to deal with more general organic and inorganic materials. It is now routinely used by chemists, biologists, chemical engineers, physicists, and materials scientists to direct and explain an enormous variety of experiments.
The link with experiment is vital, for the virtual laboratory is not some esoteric concept that is isolated from reality. It is at the heart of today's most successful research environments, a vital adjunct to experimental work. The number of reported commercial applications for modeling and simulation grows almost daily. Simulation now forms part of the solution to almost any problem that can be related to a molecular-level structure.
DESIGNING DRUGS
The longest standing application of simulation is research into drugs and the
biological compounds critical to health, and simulation tools are regularly
used at many major pharmaceutical companies. "Modeling as a partner with
synthetic work is now an accepted part of the drug discovery process," says
Michael J. Savage, president and CEO of Molecular Simulations, Inc. (MSI,
San Diego, CA).
Recent successes reported by MSI collaborators include the stabilization of a hormone used to lower blood pressure, the patenting by a Harvard research group of a method used to design antiviral drugs for fighting polio and the common cold, and the development of a novel strategy in the search for anti-AIDS drugs.
This last example provides an excellent illustration of how modeling is used to coordinate and integrate research. The HIV-1 protease plays a critical role in the replication of the HIV virus and is a key focus in AIDS research. Many potential inhibitors of this enzyme have been reported, but most of these suffer from poor bioavailability or are structurally complex, making them either ineffective or difficult to synthesize. Researchers seeking better inhibitors use empirical approaches involving the synthesis and testing of thousands of candidate chemicals. This approach is exceedingly inefficient, even with automated synthesis and testing facilities. Modeling rationalizes this process.
A model of the enzyme's active site is used to generate a hypothesis of a potential drug: a 3-D map of the required chemical characteristics. Databases of millions of compounds are then searched for molecules that match the hypothesis (e.g., they have groups that are liable to hydrogen bonding, that are hydrophobic or hydrophilic, or that are in approximately the right places). Models of these candidate compounds are "docked" into the enzyme's active site model to see whether they interact with it correctly. In this way, a few potential candidates are generated for synthesis and testing. Modeling in no way replaces experimentation but rather focuses and guides it.
The starting point for this work was a model of the relevant enzyme.
Determining the structure of the protein is of great benefit to researchers
as they begin to consider drug development. Computational methods are once
again pivotal; using data from X-ray studies to refine models of proteins is
now a common practice in the pharmaceutical
PHYSICAL SCIENCES
Although the use of simulation in the life sciences continues to expand
steadily, the most marked recent trend is an exponential growth in its
application to the physical sciences. The range of problems addressed is
quite astonishing. Following are three brief examples.
MSI scientists have studied a novel exhaust catalyst--dispersed copper cations in the zeolite ZSM-5--that could be used to improve the performance of automobile catalytic converters. Zeolites are aluminosilicate frameworks with complex pore structures providing enormous internal surfaces on which the reduction of pollutant gases can occur. The effectiveness of the catalyst depends on how the gas molecules diffuse through its complex 3-D structure, as well as the way in which these molecules are absorbed and react at the catalyst's active sites. Modeling offers unique insights into all the relevant problems. It is used to determine structure and build models of the zeolite. Likely adsorption sites are identified and characterized, and their accessibility is assessed by simulating gas diffusion through the zeolite pores. Novel quantum mechanics techniques now permit detailed study of the reaction chemistry at these sites.
Companies such as Ford can gain an understanding of the key catalytic processes to assist in their design of more effective catalysts. Similar zeolite applications, including the purification of nuclear waste materials, catalytic processes crucial to petrochemical manufacturing, and the separation of industrially pure gases from air, are worth billions of dollars annually.
Researchers at Schlumberger Research in Cambridge, UK, recently reported an unusual application of simulation work that improves the setting of cement. Their particular interest was the use of cements in oil exploration, where it is vital that the rate of setting be controlled as the material is pumped deep into a well. The researchers modeled the surfaces of ettringite, a key crystalline phase in the setting process, to establish the interaction of these surfaces with the phosphonate molecules typically used as cement retarders. Although these chemicals control the setting rate of the cement, their action is poorly understood, and new retarders have typically been discovered empirically. The modeling work helped the Schlumberger researchers to understand how phosphonates prevent growth of the ettringite phase by binding to particular crystal faces. More importantly, by identifying the salient geometric features, they were able to propose more effective inhibitors, which are now the subject of synthesis and testing.
The final example brings us full circle, back to the pharmaceutical industry. Drug discovery is just the start of the process of product development. This is taken up by formulation chemists who take a molecule and find the best method of delivering it, often producing a finished crystalline form. At this stage, polymorphism--the ability of the molecule to pack in different arrangements in the crystalline solid--is a major issue. Different polymorphs can have varying processibilities, densities, stabilities and, most significantly, therapeutic effects. Polymorph structures are conventionally determined by using single crystal X-ray diffraction, but often samples of the required purity cannot be crystallized. The experimental method also fails to guarantee that all possible polymorphs have been identified. MSI has developed a new computational method that predicts possible polymorphs starting from a model of the molecular structure. This new technology is already in use at companies such as Bayer, Organon, Zeneca Pharmaceuticals, and BASF.
ON YOUR DESKTOP SOON?
Modeling is increasingly becoming a mainstream research tool. Powerful
simulation packages are becoming easier to use, and the standards of desktop
computing are used to increase their productivity. The user base is
expanding from expert computational chemists to include more and more
experimentalists.
Savage believes that this trend is bound to continue: "Experiments are becoming more expensive, whereas computer power is increasingly affordable, with practical solutions beginning to appear on laboratory desktops. Simulation is demonstrating that it saves time and money by focusing experimental work on effective pathways. At MSI, we foresee a future where every research scientist will employ simulation at some level." The virtual and real laboratories are truly beginning to merge.
ACS Home Page