Building Bridges between Academia and Industry

Stephen K. Burley contends that high-throughput structural biology is becoming feasible through a combination of industrial and academic efforts, particularly the Protein Structure Initiative (PSI) funded by the National Institutes of Health’s National Institute of General Medical Sciences (NIGMS). Burley should know because he bridges both worlds. He is chief scientific officer at the biotech firm Structural GenomiX and also a principal investigator and head of the New York Structural Genomics Research Consortium. The center, which is part of the PSI, is projected to solve 100 structures over the next year, Burley says.

John Norvell, who heads the protein structure initiative at NIGMS, says structural biology has become a leading tool for both understanding protein function and approaching targeted drug design. Industrial participants can become part of PSI. “The crucial rule is for openness and data release,” Norvell says. Just like their academic counterparts, industry participants must commit to complete public disclosure of data six weeks after the work is completed. Adherence to those rules is monitored.

A number of scientists are keen on filling some of the technical and methodological voids in structural biology. Lawrence J. DeLucas directs the Center for Biophysical Sciences & Engineering at the University of Alabama, Birmingham. The center, which offers high-throughput structural proteomics for both academic and industrial partners, is part of the Southeast Collaboratory for Structural Genomics, another PSI center. The Alabama center currently works with 10 different industrial partners—small biotechs as well as pharmaceutical companies. The center has also spawned companies, such as BioCryst Pharmaceuticals, a structure-based drug design firm.

An important bottleneck in structural proteomics is the fact that obtaining homogeneous proteins of just one species—a monomer, a dimer, or a trimer, but not a mixture—is not trivial but is necessary to obtain good crystals and accurate structures. “I don’t think we are exploring all the possibilities that can improve how the protein interacts with itself to form a good crystal,” DeLucas says.

One software tool under development, and, according to DeLucas, with commercial potential, uses a predictive neural net algorithm to cycle through all the variables that influence the crystallization of a protein to find the best experimental approach.