Alfred Nobel (1833-1896), the world-famous chemist, engineer, industrialist and pacifist, made two lasting contrarian contributions to human civilization - dynamite and the Nobel Prize.

He invented TNT (trinitrotoluene) in 1867 and patented it that year in England, followed by U.S. patents in 1868. To this day the all-purpose explosive performs in war and peace - munitions and demolitions. On the military front, TNT fuels land mines and seabed weapons; on the civilian side it blasts mineral deposits, quarries and levels out-dated structures.

Fast forward to Duke University Medical Center, Durham, N.C., where biochemist Homme Hellinga fabricates TNT on his computer. He is professor of biochemistry, pharmacology and molecular cancer biology at Duke. Moreover, Hellinga is senior author of a paper in today's Nature dated May 8, 2003. It is titled: "Computational design of receptor and sensor proteins with novel functions."

"In this paper," Hellinga said, "we demonstrate that it's possible to design ligand binding sites at will, using computational design methods. We can dissect drastic changes in specificity of a protein for a given ligand," he told BioWorld Today. (Ligands are molecules that bind to a macromolecule, e.g. a receptor.)

Hellinga made two points: "First, we're able to create enormous changes in binding specificity; secondly, it's possible to do it by computational methods alone. There are several implications," he continued. "One is that we can now bind receptors as desired. These receptors," he added, "could function ex vivo as purification elements, or in vivo to drive signal transduction processes, for instance. Ultimately, we hope to be able to extend the technology not only to small molecules' interactions but also to protein-protein and nucleic interactions to help build enzyme active sites.

"One way to describe this concept is that we're doing directed evolution in silico. We create huge sequence libraries in the computer, with algorithms that allow us to sift through them and identify solutions that are likely to work well. The ideas go beyond protein engineering."

E. coli Hosts TNT, Lactate, Serotonin Proteins

"We are working with Escherichia coli bacterial proteins, which are members of a superfamily called periplasmic binding proteins," Hellinga recounted. (The periplasm is the space between the cell membrane and its wall.) "These protein clusters are not limited to E. coli; they occur everywhere in the biosphere. For instance, they are important domains in neuroreceptors and nerve terminals. We took the structures from some of the E. coli proteins, which normally bind sugars or amino acids. We then predicted mutations that change the binding specificity of an explosive like TNT instead of a sugar. Or a lactate or the neurotransmitter serotonin.

"Lactate," he explained, "is a product of anerobic metabolism. As for serotonin," he went on, "we did it just because it being a neuronal transmitter would make a receptor that would recognize a nerve."

In their Nature paper, Hellinga and his co-authors inventoried adapting E. coli proteins to detect four very different molecules of environmental and clinical importance:

"TNT is a nonbiological compound and carcinogen that the U.S. Navy is seeking to detect, as part of developing a TNT-sensing robot to aid its environmental cleanup effort. The TNT binding protein, because of being a deployable biosensor, is our detector for this explosive. We're working on that right now. The Navy is interested in having these to sniff' a plume of TNT in seawater emanating from unexploded ordinance.

"Lactate, an indicator of metabolic stress in the body, is also associated with certain cancers. We chose it because it's an interesting small molecule, an indicator of metabolic distress, either as a consequence of exercise or of tumors. So you can use it to monitor what's going on in the body. Also lactate has a left-handed' and a right-handed' molecular form. Demonstrating that an engineered protein could distinguish such chiral variations would be of great interest to pharmaceutical firms," Hellinga observed, "because they must purify drugs to eliminate unwanted chiral forms that could be highly toxic.

"Serotonin is a neurotransmitter that nerve cells use to trigger nerve impulses in neighboring cells. Fluctuating serotonin levels are linked with certain psychiatric disorders. Elevated levels are indicative of certain bowel tumors. We're now building receptors for drugs like ibuprofen and thalidomide, to show that we can work with those kinds of molecules as well.

"Diabetes mellitus: We actually have a glucose sensor built, and are trying to put this onto optical fibers, so we can make a catheter to insert the sensor to monitor blood glucose continuously in the patient. We're making headway on that project too, aiming ultimately at an artificial pancreas."

Here's what Hellinga and his co-authors do, and how they do it: "When you look at a protein binding site," he recounted, "you may notice that there are 15 to 20 amino acid residues, typically, that are responsible for interacting with the small molecules. In principle, each of those can change to any one of 20 amino acids. So you have a huge number of sequences that are generated if you explore this library. What the computer algorithm can do is calculate the atomic interactions between all the members of these sequences and the ensemble of ligands that you are trying to place in such a protein."

Scaling Down Astronomical Calculations

"And the tricks that the computer algorithm plays is by not enumerating all the possible sequences. That would be impossible, so we are able to sift out combinations that clearly cannot be any good. There are mathematical ways of finding what we mean by good and bad. Essentially, you're throwing away combinations that cannot possibly lead to the right answer. Keep doing that and the number of combinations goes down. It iterates, till ultimately you're left with a small number that you can enumerate."

As for patent protection, Hellinga revealed that "Wednesday, May 7, 2003, the University filed a U.S. Patent application. I'm the principal inventor, together with some of my students. The patent title is simply and generically, Protein design.' There's early talk at the University," Hellinga added, "about setting up a spin-off company out of Duke itself. There's no name for it yet; just the company.'

"Our research hints," he concluded, "that antibody-like molecules might soon be designed routinely on computers rather than discovered in the lab."