Beyond the limits of protein building.

Some children may use their Tinker Toys to model the structures printed on the box, perhaps embellishing them with an extra hub or two. But other children may go beyond the toymakers' imaginations, creating structures all their own and adding pencils, forks, twigs and mousetraps to their edifices.

Protein engineering, an activity still in its early stages, attempts to improve upon nature by remodeling existing enzymes to make them more stable or more active or to change the type of reaction they catalyze. But Bruce Erickson of Rockefeller University in New York and his colleagues are taking a more radical approach. They are constructing from scratch a molecule that does not look like any protein in nature, and they plan to add binding and catalytic sites to its basic framework. Erickson reported at the AAAS meeting the successful chemical synthesis of the framework for a new class of molecules, which he calls betabellins.

"We are opening up a new vista,"

Erickson says. "With this approach we are not limited to the 20 L-amino acids in natural proteins. We can put in D-amino acids [the mirror images of the L-amino acids found in nature], and we can use any of the 2,000 non-genetically-coded amino acids."

The basic plan for betabellin's structure comes from a natural structure, idealized for chemistry. Several different proteins, such as immunoglobulins and the enzyme superoxide dismutase, have bell-shaped, slatted regions made up of an arrangement of amino acids called beta sheets. Erickson, in collaboration with Jane and David Richardson of Duke University in Durham, N.C., devised a sequence of amino acids that comprises a structure far more regular than any "beta barrels" seen in nature. Betabellin has two identical flat regions. Each of these sheets is made up of three rows of eight amino acids, with a sharp turn between them, and one row of seven amino acids attached to a special synthetic cross-linker.

The structure of this small, globular protein includes a variety of special features that appeal to organic chemists. Because its two halves are symmetrical, they can be synthesized simultaneously. Betabellin contains internal sites where it can be snipped into pieces appropriate for analyzing the amino acid sequence. And it contains end sites to which binding sites and active sites can later be added. Betabellin also can be crystallized easily.

Erickson found many constraints on the choice of amino acids in the framework. To form the beta sheets, he had to alternate small hydrophobic (water-shunning) and hyrophilic (water-loving) amino acids. At the tight turns, he used proline and asparagine--originally the L forms, but later he replaced them with the D forms, which fit better in the available space. He included one cysteine in each half of the molecule, and in his most advanced structure he has those amino acids attach to each other in a disulfide bond, thus holding together the two sheetlike sections.

Even the most carefully planned chemical scheme can generate surprises. When Erickson purified the betabellin structure containing the disulfide bond, he obtained two distinct materials with identical amino acid sequences. Erickson and his colleagues are now trying to determine the specific structural difference.

The next step will be to further modify the framework and to add binding and catalytic sites modeled after those found in nature. "We think the present technology is adequate to start evolving proteins with new functional groups at active sites," Erickson concludes. This research, like much of the protein engineering work, is sponsored by the Office of Naval Research.