By David N. Leff

It takes two to tango, and at the molecular level, the partners had better be married.

As protein engineer Shane Atwell put it: "A lot of biology is regulated by protein-protein interactions." Atwell is a visiting scientist in the protein engineering department of Genentech Inc., of South San Francisco.

Genentech made its bones in biotechnology 12 years ago, by cloning and bringing to market injectable, recombinant human growth hormone, which has since become one of the company's mainstay drugs.

Atwell is first author of a paper in today's Science, dated Nov. 7, 1997, which explores the mating dance between growth hormone and its receptor at their molecular interface. His article bears the title: "Structural plasticity in a remodeled protein-protein interface."

Before growth hormone (GH) and its receptor (GH-R) can start dancing cheek to cheek — interfacially — these molecules have to find each other and exchange signals.

The way this signaling occurs," Atwell told BioWorld Today, "is that the GH, floating around the space outside a cell, binds to one GH-R on the cell surface. The receptor is anchored to that surface," he continued, "where it has its extracellular hormone-binding domain. And on the inner side of its outer membrane layer, the GH-R has a domain responsible for signaling."

In a mode more like squaredancing than tangoing, this hormone-receptor pair finds another receptor on the cell surface. Then, Atwell explained, "When the domains on the inside of the cell are brought together, that starts a cascade of phosphorylation, which ultimately leads to gene expression.

"Given the complexities of these interactions," he went on, "we wondered three years ago how a functionally disruptive situation on one side of the interface could be complemented by mutations in its binding partner."

To test this double cutting-in on the dance floor, he and his co-authors resorted to mutating many amino acids on the contact surfaces of both proteins. For starters, they transformed tryptophan, a large-size molecule critical for binding, into alanine, one of the smaller amino acids.

"So basically," Atwell recounted, "we just created a big, putative hole or cavity where the tryptophan used to sit — and found that the alanine GH no longer bound to its receptor."

Then they went on to systematically mutate all side-chain amino acids to alanine, all over that interface, basically cutting off all of their binding atoms. "And if any of those atoms were critical for interaction," Atwell continued, "when we did our binding assay we saw very much reduced binding."

Paired Peptides Complement Like Base Pairs

He drew an analogy between the match-up of these complementary mutant proteins, and that between base pairs in an RNA gene sequence: "In the RNA world," Atwell pointed out, "you have four bases — adenine, thymine, cytidine and guanine. If you have a structure there where the A interacts with a T, and you mutate that to A-C, you've ruined that structure. But then if you mutate its partner to a G, they can base-pair again, so you get complementary mutations and reestablish the structure."

In searching for mutants that would bind, Atwell had recourse to phage display. "This technology," he observed, "is incredibly potent. You display on the surface of a bacterial virus a heavily mutagenized protein. The DNA that encodes these variants is packaged in the phage's genome."

In this project, Atwell created a pool of 10⁷ (10 million) diverse mutants, and incubated them with an immobilized target receptor. "The virus-displaying mutants that will bind tightly," he explained, "will be captured; the rest washed away. We stripped off the captured ones and propagated them in Escherichia coli host cells, in order to iterate the process."

Since completing the experiment last year, the protein engineering lab has upped its diversification ante from 10⁷ to 10¹⁰ — 1 trillion mutants.

Swinging At Far Edge Of Dance Floor

One surprise surfaced with the X-ray crystallization structure of the GH-GH-R complex, Atwell recalled. "We saw changes in the interface that were very far away from the residues [amino acids] we had mutated. Those mutants," he added, "were in the center of the interface, and when we solved the crystal structure we found that the hormone rotates, slides across the receptor, such that there are hydrogen bonds and other interactions that change very far away from the site of mutation."

Atwell made the point that "we weren't trying to solve any growth-hormone medical problem, other than a long-term goal of understanding those interface side chains."

But he did mention another long-term goal of medical interest: drug design. This potential application derives, he explained, "from a functional hot spot — a set of residues, on both hormone and receptor, that are really responsible for binding. The general strategy," he continued, "would then be to take those residues in combination with a crystal structure, and attempt to design a small molecule that would place all those atoms, those side chains, in the right position so that they could bind, as a small molecule, to the hormone."

Whoever mentions small molecules these days is alluding to orally ingestable, rather than injectable, drugs.

In a separate experiment, the team mutagenized a second tryptophan. "I did the analogous experiment with it," Atwell recalled, "and didn't find any mutant hormone that complemented that receptor. Now," he added, "there's another project in the lab that is attempting to use these two critical tryptophans in order to design drugs." *