Science Writer

Blending physics, mathematics and computer science, Protein Mechanics Inc. is working to design and develop drugs in silico, though knowledge of such methods grew out of another industry.

Its founders spent 20 years working in physics and as computational geometry software developers, creating virtual models for the mechanical engineering industry. The group - Michael Sherman, Dan Rosenthal, Michael Hollars and Daniel Filip - viewed drug design and development as a logical next step in applying their virtual technology.

And with a newly appointed president and CEO on board, Protein Mechanics is looking to push its virtual drug discovery technology even further.

Ken Haas, named to the top post late last month after Sherman moved into the role of executive vice president, is aiming to spearhead such an effort.

"The company was looking for a CEO to hone the business model, start the partnering effort and go out and raise a second round of financing," Haas told BioWorld Today, noting that his background includes software and biotechnology companies. "Initially, this company is technologically a software company, but business-wise, it's a biotech company."

Founded in late 2000, the Mountain View, Calif.-based firm soon raised $3.5 million from a pair of venture capital firms. Palo Alto, Calif.-based Alloy Ventures and London-based Abingworth Management Ltd. laid down their initial investments in April 2001.

Following a second round of investment, which Haas said he would begin working toward in a month, the 15-person company plans to hire more employees in its chemistry department as it looks to broaden its development efforts. He said Protein Mechanics would seek to close about $7 million in additional funding by this fall.

In the meantime, the company continues to sharpen its focus on its Imagiro simulation technology.

"In mechanical and electrical areas, virtual engineering had to deal with severe complexity and inaccessibility," Haas said. "Drug design has those same problems and the same opportunities, but size does not matter. Virtual engineering works as well at the nano level of computer chips as it does at the space station level."

Similar to such prior modeling and simulation methods, Protein Mechanics believes its technology would lead to greater accuracy and therefore better efficiency compared to current drug discovery techniques of synthesis and experimentation. Initially focusing on lead optimization, the company is working to develop in silico three components of the process - creating co-structures, calculating binding affinities and demonstrating dynamic live structures, in particular confirmational change.

"Our goal is chemical accuracy," Haas said. "The idea is to reach a level of accuracy where medicinal chemists use our virtual molecules in a way that a mechanical engineer viewed a virtual motorcycle when there was an inflection point in mechanical engineering."

Already, Protein Mechanics has validated that its simulation produces the same binding image as that shown in the known X-ray structure of the protein lysosyme with the peptide N-acetylglucosamine. Other validations include in silico-produced images of binding that predict the co-structure between the SH3 domain in kinases and a peptide known to bind with the domain, as well as an area of binding weakness in the co-structure. Protein Mechanics' technology also has shown that the peptide beta amyloid, when binding with a beta-sheet breaker peptide, undergoes an allosteric change that reshapes it and makes for harder aggregation. Increased accumulation of beta amyloid is thought to play a large role in Alzheimer's disease.

"It creates novel insights that dramatically help medicinal chemists find small molecules that create these kinds of changes," Haas said. "It's not possible in any other vehicle - certainly not X-ray crystallography or nuclear magnetic resonance, the current leading vehicles for lead optimization."

In time, Haas said, such virtual modeling could become more universally used because of added acceptance of its level of accuracy as well as inherent cost efficiency built into a potentially quicker drug design timeline. In pushing its technology, Protein Mechanics does not plan to sell its platform but rather attract partners to which the company would deliver its virtual targets and compounds.

"We're not a full-on chemistry company with all their expenses, and on the other hand we're not just providing the platform," Haas said. "We're delivering the same kinds of things traditionally delivered in vials, but we're delivering it in silico."

Such a business model, he added, is designed to produce cost and revenue leverage for Protein Mechanics. Assuming such research collaborations prove fruitful, the company eventually plans to enter development collaborations and begin internal work.

With a future of virtual drug design envisioned for Protein Mechanics, the company is eyeing a bright horizon.

"Ultimately we expect at first to complement and then start to compete with real-world techniques like X-ray crystallography or nuclear magnetic resonance," Haas said. "This is one of those things that's a potentially revolutionary technology that unlike others will not be that hard to sell. You read about chemistry veterans who say that in silico is the wave of the future, and drugs will be designed in silico when such techniques get to the level of accuracy and performance where they are competitive. We think we are on the verge of that."