Software Reveals Structure of Vancomycin, “Last Resort” Antibiotic

BUFFALO, N.Y. -- Humans are one up in the war against drug-resistant bacteria, thanks to a software package used to solve molecular structures developed at the University at Buffalo and Buffalo’s Hauptman-Woodward Medical Research Institute.

Called SnB, the software program has allowed scientists to unravel the structure of vancomycin, providing information needed to synthesize new forms of the drug targeted to fight killer bacteria.

And just in the nick of time.

Vancomycin is medicine’s “drug of last resort,” the only antibiotic effective against certain life-threatening infections that resist all other known antibiotics. Just last week, the Centers for Disease Control reported the appearance for the first time in the U.S. of a strain of bacteria that is resistance to vancomycin, raising the specter of medicine having no weapon to fight often-fatal staphylococcus infections.

The user-friendly SnB allowed University of Pennsylvania pharmacologists to determine the structure of vancomycin.

“There just aren’t any other methods that could have solved this structure,” said Russ Miller, Ph.D., professor of computer science at UB, senior research scientist at Hauptman-Woodward and a member of the research team that developed SnB.

Patrick Loll, Ph.D., assistant professor of pharmacology in the University of Pennsylvania School of Medicine, who led the team that solved the structure, said “the evidence that some bacteria strains are now resistant to vancomycin poses a major public health threat.

“Our hope is that now that we have the structure, we will be able to design a new form of the drug that will circumvent the resistance problem and preserve the last-resort utility of this drug.”

But unraveling the structure wasn’t easy. According to Loll, the vancomycin molecule is extremely complicated.

“Despite 30 years of effort, no one has ever succeeded in synthesizing it,” he explained. The drug is isolated from soil microorganisms, which are cultured in huge fermentors.

“Because it’s not even possible to synthesize the drug itself, it’s not possible to blithely synthesize lots of variants of it, the traditional trial-and-error method used in drug discovery,” he said.

The only alternative was to rationally decide beforehand which modifications would have the best chance of eluding the resistant bacteria.

But doing that required the structural information. Before trying SnB, the team spent about eight months of CPU (central processing unit) time trying other methods. All of them failed.

“That’s where SnB came in -- it enabled us to determine the structure and bypass most of the trial and error part,” said Loll.

Loll noted that even SnB, which has solved other complex structures in as little as hours or days, took about a month to yield the correct result.

“SnB is very well-written and very user-friendly so once you set it up, it will run on your computer very happily for weeks without any intervention,” he said.

That let the software try out lots of possible solutions, many more than would have been possible or even feasible in the laboratory.

According to Miller, vancomycin’s structure fell into a middle range where few -- if any -- methods of structural determination are successful.

Some methods are successful mostly with molecules of about 100 atoms, and others work with much larger molecules, such as proteins, of several thousand atoms. Miller explained that neither of them are especially successful with molecules in the mid-range, such as vancomycin, which contains approximately 400 atoms.

Without the structural information, efforts to disarm the resistant bacteria would most likely be unsuccessful.

“Now that we have the structure, we can much more easily determine how the drug does what it does,” explained Loll. “We will use that information to design new variants of the drug that can recognize the altered target produced by the resistant bacteria and destroy it.”

Loll’s team includes Paul H. Axelsen, Ph.D., assistant professor of pharmacology at the University of Pennsylvania, and doctoral candidates Anthony Bevivino and Brian Korty.

According to its developers, the key advantage of the SnB software lies in its ability to solve complex structures with virtually no user assistance.

“SnB solves molecular structures like a black box, with no human intervention, given data with sufficient resolution,” said Herbert Hauptman, Ph.D., president of the Hauptman-Woodward Medical Research Institute and UB research professor of computer science.

Hauptman’s idea for the minimal principle, on which the program is based, originated in an elegant, 18th-century mathematical principle formulated by mathematician Carl Friedrich Gaus about the orbits of asteroids.

Initially successful when it ran on massively parallel computers, the SnB program has been continually refined and adapted by Miller and Charles M. Weeks, Ph.D., senior research scientist at Hauptman-Woodward.

Now, just two years after its introduction, its developers are about to release the second version of it, which is even faster than the original. The front end is written in Java with a graphical interface.

“SnB has been wildly successful,” said Miller. “People are finding they just input the basic crystallographic information, type in a few answers about their structure, hit return and off it goes. It’s solving things people cannot solve with any other method.”

Information on how to obtain SnB is available at http://www.hwi.buffalo.edu/SnB or by contacting Russ Miller or Charles M. Weeks at (716) 856-9600.