Computer-Based Drug Design: Advancing the Discovery of New Cancer Medicines
Finding drugs to treat and prevent cancer is a difficult and expensive process. But, thanks to NCI-funded research, we have identified many biological targets for drugs based on a deeper understanding of how cancer develops and progresses. Decades of basic research combined with genome sequencing projects, such as The Cancer Genome Atlas, have revealed many genes that cause cancer and how they work in molecular pathways to either drive or suppress tumor growth. An increasing number of cancer drugs target these processes, thereby suppressing or killing cancer cells while minimizing the effects on normal cells. Research on the immune system and cancer is also leading to new therapies that engage the immune system in killing cancer cells.
Many innovative approaches are on the horizon, from drugs that specifically tag cancer cells for destruction to targeted signals that stop cancer cells from growing. While immense opportunities exist, drug discovery and development remain time consuming and costly. It can take 10 to 15 years for a new medicine to complete the journey from its initial discovery to its use in patients. These timelines do not necessarily include the many years of basic research discoveries that make clinical advances possible. Only 3%–5% of investigational cancer drugs that reach clinical trials receive Food and Drug Administration approval.
Typically, researchers find drugs by screening large numbers of small molecules for their ability to block the activity of a specific target that causes cancer. But even the largest collections of molecules—those with as many as one or two million compounds—represent a minuscule portion of all possible drug candidates. And testing these molecules is slow and expensive. Advances in computer-based drug design could greatly increase the odds of finding the best candidate drugs. Using computational methods, researchers can rapidly screen billions of molecules to find those that interact with targets in the body without having to physically make and test molecules in the lab. This sort of large-scale screening to find candidate drugs could dramatically speed drug discovery.
Computer-based drug design requires determining both the structure of a target molecule (such as an abnormal protein driving cancer cell growth) and the computational design of drug molecules that will bind to specific areas in that structure. Computational methods are already being used to fine-tune the shape of candidate drugs to help them attach more tightly and more selectively to their targets.
Improved computational techniques could identify the structure of key sites in targets where drugs can be most effective and predict how potential drugs may affect the behavior of the target molecules. In addition, improved computational methods could help predict the stability of candidate drugs in the body and, in the future, could be used to assess whether they may have unintended side effects.
Computer-based drug design will rely on the increasing availability of large amounts of information generated from laboratory research in biomedical sciences, without which computational techniques will not have enough data to return reliable results.
As computers get faster, and when quantum computing becomes available, the cancer research community is poised to make important leaps forward. Cancer is a particularly complex problem, but the field of cancer research has accumulated a great deal of knowledge about the mechanisms that drive cancer. Advances in computation could help greatly in identifying matches between targets in the body and prospective drugs.
Ending cancer as we know it includes a future where safe and effective medicines are available for every patient. Ultimately, having rapid computational methods, based on experimental data to predict interactions between potential drugs and targets in specific cancers, could open up a world of treatment and prevention options that work more effectively to save lives and cause fewer toxic side effects.
A Virtual Test Tube of Biological Structures and Medicines
One key challenge that computational techniques can help address is treatment side effects. Many molecules drop out of the drug development pipeline during clinical trials because of their toxic side effects. Avoiding side effects is especially important with drugs for cancer prevention that would be given to healthy people. With intensive computational modeling that builds on a rich trove of data, it will be possible to screen drug candidates for interactions with molecules in the body other than the intended target. For example, healthy cells and tissues may have slightly different versions of the same protein as drug targets. Computational modeling can help design drugs that match the target precisely enough to reduce the chance of side effects in such cases.
To enable this and other promising forms of computational modeling, more experimental data must be gathered in laboratory research and made publicly accessible. Proteins are the most important class of targets for cancer drugs, and researchers need clear pictures of many more proteins, including the same protein in different states. Support is needed to develop rapid, large-scale laboratory technologies to visualize the shapes of proteins in all their different configurations. The structures of some classes of proteins that are important cancer targets have been difficult to figure out using existing research techniques, and support is also needed to develop ways to solve their structures.
A developing technology called cryo-electron microscopy (cryo-EM) may help satisfy the need for more protein structures. Cryo-EM uses extremely low temperatures to capture the shapes of biological samples while they are suspended in water, very similar to the state in which they may exist in the human body. Higher-resolution techniques are leading to a boom in the number of protein structures that can be revealed by cryo-EM. Cryo-EM has the advantage of allowing researchers to solve protein structures quickly using samples that are not necessarily pure. Further research is needed to refine cryo-EM technology so that it can reveal structures with an even higher level of resolution that shows in detail the interactions between targets and potential drugs.
Over the last few years, cryo-EM has been used to reveal several protein structures that are important to understanding cancer, some of which are potential targets for drugs. For example, investigators in the NCI Intramural Research Program have used cryo-EM to get exceptionally high-resolution snapshots of a protein bound to an inhibitor and to obtain an image of a protein called p97 that is considered an attractive target for cancer drug development. NCI is making cryo-EM technology available to the research community through facilities at the Frederick National Laboratory for Cancer Research.
As computer-based drug design evolves, it can be used to screen not only small-molecule drug candidates, but also larger molecules, some of which mimic naturally occurring molecules in the body. Many such drug candidates take advantage of the body’s immune response to target cancer cells.
One promising new category of cancer drugs that was developed with NCI support is called a proteolysis targeting chimera (PROTAC). These are two-headed molecules where one head attaches to a target protein, and the other head recruits an enzyme to degrade the protein to which the PROTAC attaches. This treatment approach does not require designing a drug that can disable a protein by interfering with how it works, which can be complicated to achieve. It simply requires designing a drug that clings to the target protein long enough to label it for destruction, an interaction that may be easier for computers to model.
What the Government Can Do
While pharmaceutical companies ultimately bring most drugs to market, the federal government plays an important role in improving technology and developing the knowledge available to catalyze drug discovery and development. NCI can make critical data publicly available and get tools into the hands of researchers.
The federal government can make massive computational power available to analyze systems that are too complex for individual organizations to tackle on their own. For example, NCI has been working with the Department of Energy to make computational tools and data available to the research community. The government could potentially make virtual libraries containing billions of molecules available to the broad scientific community. NCI also establishes standards and quality control measures to ensure that its databases are uniformly of high quality. The government is well poised to ensure that aggregated data and computational resources are accessible to those who can use them to expand knowledge and develop drugs.