Normally aloof, two proteins (red and green) huddle when they carry the same GPI anchor.

One way to switch the function of a cell surface protein is to force it into a different membrane domain. On page 647, Nicholson and Stanners describe a second method: diluting it within its domain with an abundance of defective partners. The approach might provide a way to neutralize a protein that is overproduced in most cancers.

Some surface proteins lie completely outside the cell, attached by molecules called glycophosphatidyl inositols (GPIs). These anchors dictate protein function by steering their proteins into particular membrane domains called rafts, clusters crucial for transmitting signals. Nicholson and Stanners tested whether dilution of raft clustering could thwart CEA, an adhesion protein whose overexpression in many cancers blocks differentiation and allows tumor cells to continue dividing.

The researchers worked with two hybrid proteins. One, called ΔNCEA, is a pruned version of CEA that carries the same GPI tether but doesn't block cell specialization. The second hybrid, called NCB, sports the GPI anchor of CEA linked to a different adhesion protein, called NCAM. CEA and NCAM don't congregate on the membrane, but ΔNCEA and NCB do, the authors found. The CEA-specific GPI anchors thus bring them into the same membrane rafts.

With its own anchor, NCAM does not halt differentiation. But NCB, which has CEA's GPI anchor, does, suggesting that the external adhesive domain is only important because it clusters the CEA rafts, resulting in downstream signaling that blocks differentiation.

ΔNCEA lifted this block and boosted the size of the rafts. The researchers speculate that these oversized patches dilute NCB so much that the proteins can't link up, a necessary step for signaling. Differentiation stalled—indicating that signaling had resumed—when the researchers dosed cells with antibodies that force NCB proteins to cluster.

Overall, the work reveals that one way GPI anchors regulate a protein's activity is by controlling which other molecules it consorts with by placing it in the appropriate rafts. A key remaining question, Nicholson and Stanners say, is whether a dilution procedure similar to the one they used can impede CEA in cancer cells.