ERROR 1
ERROR 1
ERROR 2
ERROR 2
ERROR 2
ERROR 2
ERROR 2
Password and Confirm password must match.
If you have an ACS member number, please enter it here so we can link this account to your membership. (optional)
ERROR 2
ACS values your privacy. By submitting your information, you are gaining access to C&EN and subscribing to our weekly newsletter. We use the information you provide to make your reading experience better, and we will never sell your data to third party members.
Computational chemists are advancing in their ability to account for water in molecular simulations for drug discovery. At a recent meeting, theorists described new developments in computational chemistry software that allow them to include water in simulations of protein-drug binding that are more explicit and more widely usable than ever before.
Theoreticians have long known that the simple “key in lock” picture of protein-drug binding—that protein-ligand binding is based on shape complementarity—is an extreme simplification. For one thing, ligands inserting themselves into the pockets of proteins are generally not rigid, like keys. In reality, proteins wiggle and ligands flex in a soup of watery biological fluids, creating chemical interactions that may help or hinder binding. As computational power, both in computers and theory, has grown, modelers have been able to add ligand flexibility and in some cases protein flexibility into their programs.
But what about water molecules in proteins? Protein pockets are filled with water, the molecules packing reluctantly against hydrophobic protein areas in some cases while snuggling against hydrophilic sites in others. The energy it takes to displace, or desolvate, this water can have a dramatic effect on the binding characteristics of proteins and ligands.
Until just a few years ago, no methods existed that explicitly accounted for the effects of water molecules in protein-ligand binding on a timescale suitable for interactive drug design. Some scientists believed the inability to account for those effects hampered drug discovery efforts.
Now, three programs that directly address solvation—two of them brand new—have burst onto the scene. On Feb. 23, several dozen researchers in one of the country’s hotbeds of drug discovery, the San Francisco Bay Area, got a rare opportunity to hear about all of them at a symposium hosted by Amgen. Speakers at the session discussed Schrödinger’s WaterMap, Chemical Computing Group’s (CCG) 3D-RISM, and OpenEye Scientific Software’s SZMAP.
“We wanted to get experts on each of the three programs together in the same session and have each talk about the science,” said Vickie Tsui, a computational chemist at Genentech and one of the symposium’s organizers.
As computational drug architects, the attendees have a major concern: Will these methods speed up the search for successful drug candidates? Theoretically, they should, because “from a theoretical point of view, receptor desolvation is a big component of binding affinity,” said Alan C. Cheng, a computational chemist at Amgen who attended the symposium. “Being able to capture explicit water character, which all these methods try to do, is a big theoretical advantage for rational drug design.”
Using different theoretical strategies, the programs all identify areas in a protein pocket where water is stable and therefore energetically costly to displace, or unstable and thus more easily displaced.
WaterMap, which debuted in 2007, was, for a while, the only game in town. It combines molecular mechanics and molecular dynamics to treat each water molecule in a pocket explicitly, mapping their locations and thermodynamic properties.
At the meeting, Schrödinger computational chemist Robert Abel demonstrated WaterMap’s relative maturity, citing its successes in drug discovery efforts. For example, it has made possible development of an inhibitor of the enzyme interleukin-1 receptor-associated kinase 4 (IRAK4), Abel told attendees.
IRAK4 is involved in inflammatory diseases such as lupus and rheumatoid arthritis, and inhibitors have been difficult to develop. Nimbus Discovery, in partnership with Schrödinger, used WaterMap to examine likely candidates, eventually evaluating 60,000 possible compounds based on a hit from their virtual screen. From that list, Nimbus made just over 300 compounds, resulting in a very potent inhibitor that is active in rat disease models. Going from virtual screening to in vivo proof-of-concept took only nine months.
Despite this success, WaterMap has a reputation among some users for being time-consuming and expensive, which some say has helped drive the search for faster, less expensive methods.
Last year, CCG released a novel application of a long-standing solvation theory known as 3D-RISM and incorporated it into the company’s flagship product, Molecular Operating Environment. As computational chemist Jean-François Truchon told the audience, CCG’s 3D-RISM predicts densities of water molecules inside a protein pocket—a strategy analogous to that of density functional theory in electronic structure theory. The result is that 3D-RISM calculations are faster than WaterMap calculations, while capturing the thermodynamics, he said.
Even newer is SZMAP (pronounced “shzmap”), which Santa Fe, N.M.-based OpenEye released last fall. OpenEye computational chemist Matt Geballe described SZMAP’s strategy of moving an explicit water molecule over a grid, in different orientations, to build up a picture of water energies over the area.
Though 3D-RISM and SZMAP have just come out of the gate, researchers at the symposium said they are already beginning to test the programs’ capabilities. Cheng, for example, said the Amgen group has been able to use 3D-RISM, and “it does run relatively quickly.” Tsui’s group has used both WaterMap and SZMAP, and “we’ve had successes with both,” she said.
Addressing water solvation represents a new level of sophistication in modeling protein-ligand binding, but computational programs still rely on some undesirable or at least nonoptimal assumptions. For instance, the three solvation programs discussed at the meeting still treat proteins as rigid molecules, which is not realistic. On the other hand, adding protein flexibility into the mix would create complexity that’s currently out of the reach of most computers and most modeling programs.
“There certainly is still a lot we’re willfully ignoring,” Geballe noted. “Water and solvent effects used to be in that category too, but now we’re getting to the point where we’re trying to tackle them.”
Join the conversation
Contact the reporter
Submit a Letter to the Editor for publication
Engage with us on Twitter