Biological function is inextricably embedded in molecular interaction. When Bernd Wollscheid and his colleagues at the ETH Zurich developed a bifunctional reagent to label glycosylated cell-surface proteins some years ago, they wondered if they might extend the approach to study extracellular interactions between receptors and ligands. A collaboration between the groups of Wollscheid and Erick Carreira, also at the ETH Zurich, has now led to TRICEPS, a trifunctional reagent designed for this purpose.

TRICEPS works as follows. First, its N-hydroxysuccinimide ester is used to conjugate the reagent to free amino groups on a protein or peptide ligand. Then, its protected hydrazine is reacted with cell-surface aldehydes under mild conditions compatible with cell survival. Although there are typically no natural aldehydes on the cell surface, they can be generated by gently oxidizing sugar groups that decorate cell-surface proteins. Finally, TRICEPS contains a biotin to capture the interacting receptor-ligand pair.

“We went through a number of iterations,” says Wollscheid. The key was to find a molecule that would be compatible with reactions on the surface of living cells, he explains. The researchers show that oxidation with low concentrations of sodium metaperiodate to generate cell-surface aldehydes, as well as subsequent cross-linking of the TRICEPS-conjugated ligand, can be carried out on living samples.

Using quantitative mass spectrometry, Wollscheid and colleagues compared peptides identified by streptavidin pulldown of a TRICEPS-conjugated ligand with those identified in a parallel random control. They observed that glycopeptides enriched with a specific ligand identified the known cellular receptors for several peptide and protein ligands, including insulin, transferrin and an antibody of therapeutic interest.

Importantly, TRICEPS does not link two proteins; rather, it links a ligand with the glycostructure of its receptor, thus directing the cross-linking away from the receptor's protein backbone. “We don't interfere with the protein domains of the receptor,” says Andreas Frei, first author on the paper, “but, in addition, this lets us address the reagent in space and time. We first want to couple it to the ligand, and only then do we want to capture the ligand-directed interaction with the receptor. Since we have one protein-reactive moiety and one sugar-reactive moiety, this is perfect.”

The TRICEPS reagent can also be fluorescently labeled, making it possible to follow ligand binding to the cell surface by flow cytometry. Some fraction of the ligand will be rendered nonfunctional by TRICEPS conjugation, the researchers say, but this merely increases the nonspecific signal. “The remaining ligands are more than sufficient to enrich for the receptor,” Frei says.

Not only were the researchers able to capture receptor-ligand interactions on cells, they could do so in living tissue samples as well. Using primary breast cancer tissue, they correctly identified ErbB2 as the receptor for the antibody Herceptin, even though the specific tissue sample expressed the receptor at low levels. “We have, for the first time, the possibility to look at complex surface interactions of ligands in the context of the particular microenvironment for the receptor,” Wollscheid says.

Finally, ligands of increased complexity could be accessible with this reagent. TRICEPS-labeled vaccinia viruses captured seven cell surface proteins, five of which the researchers could functionally validate with siRNA experiments, and which are therefore good candidates for cellular interactors with this virus. There are still a large number of viruses whose cellular receptors remain unknown.

Wollscheid points out that most interactome mapping efforts are focused on intracellular interactions. “I think we have the possibility to unravel the extracellular interactome,” he predicts. “We could use it to decode cellular communication as well as ... to study what are drugs doing at the cell surface. It's early on, but that is where we would like to go.”