The Structures of Life
Chapter 4: Structure-Based Drug Design: From the Computer to the Clinic
In 1981, doctors recognized a strange new disease in the United States. The first handful of patients suffered from unusual cancers and pneumonias. As the disease spread, scientists discovered its cause—a virus that attacks human immune cells. Now a major killer worldwide, the disease is best known by its acronym, AIDS.
Although researchers have not found a cure for AIDS, structural biology has greatly enhanced their understanding of HIV and has played a key role in the development of drugs to treat this deadly disease.
HIV was quickly recognized as a retrovirus, a type of virus that carries its genetic material not as DNA, as do most other organisms on the planet, but as RNA. After entering a cell, retroviruses "reverse transcribe" their RNA into DNA.
Long before anyone had heard of HIV, researchers in labs all over the world studied retroviruses, some of which cause cancers in animals. These scientists traced out the life cycle of retroviruses and identified the key proteins the viruses use to infect cells.
When HIV was identified as a retrovirus, these studies gave AIDS researchers an immediate jump-start. The previously identified viral proteins became initial drug targets.
Our story begins in 1989, when scientists determined the X-ray crystallographic structure of HIV protease, a viral enzyme critical in HIV's life cycle. Pharmaceutical scientists hoped that by blocking this enzyme, they could prevent the virus from spreading in the body.
With the structure of HIV protease at their fingertips, researchers were no longer working blindly. They could finally see their target enzyme—in exhilarating, color-coded detail. By feeding the structural information into a computer modeling program, they could spin a model of the enzyme around, zoom in on specific atoms, analyze its chemical properties, and even strip away or alter parts of it.
Most importantly, they could use the computerized structure as a reference to determine the types of molecules that might block the enzyme. These molecules can be retrieved from chemical libraries or can be designed on a computer screen and then synthesized in a laboratory. Such structure-based drug design strategies have the potential to shave off years and millions of dollars from the traditional trial-and-error drug development process.
These strategies worked in the case of HIV protease inhibitors. “I think it's a remarkable success story,” says Dale Kempf, a chemist involved in the HIV protease inhibitor program at Abbott Laboratories. “From the identification of HIV protease as a drug target in 1988 to early 1996, it took less than 8 years to have three drugs on the market.” Typically, it takes 10 to 15 years and more than $800 million to develop a drug from scratch.
The structure of HIV protease revealed a crucial fact—like a butterfly, the enzyme is made up of two equal halves. For most such symmetrical molecules, both halves have a "business area," or active site, that carries out the enzyme's job. But HIV protease has only one such active site—in the center of the molecule where the two halves meet.
Pharmaceutical scientists knew they could take advantage of this feature. If they could plug this single active site with a small molecule, they could shut down the whole enzyme—and theoretically stop the virus' spread in the body.
Several pharmaceutical companies started out by using the enzyme's shape as a guide. “We designed drug candidate molecules that had the same two-fold symmetry as HIV protease,” says Kempf. “Conceptually, we took some of the enzyme's natural substrate [the molecules it acts upon], chopped these molecules in half, rotated them 180 degrees, and glued two identical halves together.”
To the researchers' delight, the first such molecule they synthesized fit perfectly into the active site of the enzyme. It was also an excellent inhibitor—it prevented HIV protease from functioning normally. But it wasn't water-soluble, meaning it couldn't be absorbed by the body and would never be effective as a drug.
Abbott scientists continued to tweak the structure of the molecule to improve its properties. They eventually ended up with a nonsymmetrical molecule they called Norvir® (ritonavir).
Between December 1995 and March 1996, the Food and Drug Administration approved the first three HIV protease inhibitors—Hoffman-La Roche's Invirase™ (saquinavir), Abbott's Norvir™ (ritonavir), and Merck and Co., Inc.'s Crixivan® (indinavir). Initially, these drugs were hailed as the first real hope in 15 years for people with AIDS. Newspaper headlines predicted that AIDS might even be cured.
Although HIV protease inhibitors did not become the miracle cure many had hoped for, they represent a triumph for antiviral therapy. Antibiotics that treat bacterial diseases abound (although they are becoming less effective as bacteria develop resistance), but doctors have very few drugs to treat viral infections.
Protease inhibitors are also noteworthy because they are a classic example of how structural biology can enhance traditional drug development. “They show that with some ideas about structure and rational drug design, combined with traditional medicinal chemistry, you can come up with potent drugs that function the way they're predicted to,” says Kempf.
“That doesn't mean we have all the problems solved yet,” he continues. “But clearly these compounds have made a profound impact on society.” The death rate from AIDS went down dramatically after these drugs became available. Now protease inhibitors are often prescribed with other anti-HIV drugs to create a "combination cocktail" that is more effective at squelching the virus than are any of the drugs individually.
Traditionally, scientists identify new drugs either by fiddling with existing drugs or by testing thousands of compounds in a laboratory. If you think of the target molecule—HIV protease in this case—as a lock, this approach is rather like trying to design a key perfectly shaped to the lock if you're given an armload of tiny metal scraps, glue, and wire cutters.
Using a structure-based strategy, researchers have an initial advantage. They start with a computerized model of the detailed, three-dimensional structure of the lock and of its key (the natural molecule, called a substrate, that fits into the lock, triggering viral replication). Then scientists try to design a molecule that will plug up the lock to keep out the substrate key.
Knowing the exact three-dimensional shape of the lock, scientists can discard any of the metal scraps (small molecules) that are not the right size or shape to fit the lock. They might even be able to design a small molecule to fit the lock precisely. Such a molecule may be a starting point for pharmaceutical researchers who are designing a drug to treat HIV infection.
Of course, biological molecules are much more complex than locks and keys, and human bodies can react in unpredictable ways to drug molecules, so the road from the computer screen to pharmacy shelves remains long and bumpy.
HIV is a moving target.When it reproduces inside the body, instead of generating exact replicas of itself, it churns out a variety of slightly altered daughter virus particles. Some of these mutants are able to evade, or "resist," the effects of a drug—and can pass that resistance on to their own daughter particles.While most virus particles initially succumb to the drug, these resistant mutants survive and multiply. Eventually, the drug loses its anti-HIV activity, because most of the virus particles in the infected person are resistant to it.
Some researchers now are working on new generations of HIV protease inhibitors that are designed to combat specific drug-resistant viral strains.
Detailed, computer-modeled pictures of HIV protease from these strains reveal how even amino acid substitutions far away from the enzyme's active site can produce drug resistance. Some research groups are trying to beat the enzyme at its own game by designing drugs that bind to these mutant forms of HIV protease. Others are designing molecules that latch onto the enzyme's Achilles' heels—the aspartic acids in the active site and other amino acids that, if altered, would render the enzyme useless. Still others are trying to discover inhibitors that are more potent, more convenient to take, have fewer side effects, or are better able to combat mutant strains of the virus.
While the HIV protease inhibitors are classic examples of structure-based drug design, they are also somewhat unusual—at least for now. Although many pharmaceutical companies have entire divisions devoted to structural biology, most use it as a complementary approach, in partnership with other, more traditional, means of drug discovery. In many cases, the structure of a target molecule is determined after traditional screening, or even after a drug is on the market.
This was the case for Celebrex®. Initially designed to treat osteoarthritis and adult rheumatoid arthritis, Celebrex® became the first drug approved to treat a rare condition called FAP, or familial adenomatous polyposis, that leads to colon cancer.
Normally, the pain and swelling of arthritis are treated with drugs like aspirin or Advil® (ibuprofen), the so-called NSAIDs, or non-steroidal anti-inflammatory drugs. But these medications can cause damage to gastrointestinal organs, including bleeding ulcers. In fact, a recent study found that such side effects result in more than 100,000 hospitalizations and 16,500 deaths every year. According to another study, if these side effects were included in tables listing mortality data, they would rank as the 15th most common cause of death in the United States.
A fortunate discovery enabled scientists to design drugs that retain the anti-inflammatory properties of NSAIDs without the ulcer-causing side effects.
Although the enzymes share some of the same functions, they also differ in important ways. COX-2 is produced in response to injury or infection and activates molecules that trigger inflammation and an immune response. By blocking COX-2, NSAIDs reduce inflammation and pain caused by arthritis, headaches, and sprains.
In contrast, COX-1 produces molecules, called prostaglandins, that protect the lining of the stomach from digestive acids.When NSAIDs block this function, they foster ulcers.
To create an effective painkiller that doesn't cause ulcers, scientists realized they needed to develop new medicines that shut down COX-2 but not COX-1. Such a compound was discovered using standard medicinal chemistry and marketed under the name Celebrex®. It quickly became the fastest selling drug in U.S. history, generating more prescriptions in its first year than the next two leading drugs combined.
At the same time, scientists were working out the molecular structure of the COX enzymes. Through structural biology, they could see exactly why Celebrex® plugs up COX-2 but not COX-1.
The three-dimensional structures of COX-2 and COX-1 are almost identical. But there is one amino acid change in the active site of COX-2 that creates an extra binding pocket. It is this extra pocket into which Celebrex® binds.
In addition to showing researchers in atom-by-atom detail how the drug binds to its target, the structures of the COX enzymes will continue to provide basic researchers with insight into how these molecules work in the body.
“I really like to study retroviruses,” says Kristi Pullen, who majored in biochemistry at the University of Maryland, Baltimore County (UMBC). “I also like highly infectious agents, like Ebola. The more virulent something is, the less it's worked on, so it opens up all sorts of fascinating questions. I couldn't help but be interested.”
In addition to her UMBC classwork, Pullen helped determine the structure of retroviruses in the NMR spectroscopy laboratory of Michael Summers. This research focuses on how retroviruses package "RNA warheads" that enable them to spread in the body. Eventually, the work may reveal a new drug target for retroviral diseases, including AIDS.
University of California, Berkeley
Until her senior year in high school, Pullen wanted to be an orthopedic surgeon. But after her first experience working in a lab, she recognized "there's more to science than medicine." Then, after taking some science courses, she realized she had an inner yearning to learn science and to work in a lab.
Pullen is now a graduate student at the University of California, Berkeley in the Department of Molecular and Cell Biology. She plans to continue studying structural biology, to earn a Ph.D., and possibly also to earn an M.D. She also has some longer-term goals.
“Ultimately what I want to do way, way, way down the line is head the NIH [National Institutes of Health] or CDC [Centers for Disease Control and Prevention] and in that way affect the health of a large number of people—the whole country.”
What is structure-based drug design?
How was structure-based drug design used to develop an HIV protease inhibitor?
How is the structural difference between COX-1 and COX-2 responsible for the effectiveness of Celebrex®?
How do viruses become resistant to drugs?