The group focused on the Phe and Gly (FG)-rich repeats present in many nuclear pore proteins. The effective concentration of the repeat proteins in the pore is close to 100 mg/ml. “At such a concentration, a gel must form,” says Görlich.
But proving that in vitro was not easy. Agarose makes a gel, but only if the powder is first boiled with water to make a homogeneous solution. A similar approach does not work with FG repeats, whose hydrophobic interactions are largely unresponsive to heat. Finally, the German group found that a regime of pH changes could give a similar result: a hydrogel whose properties are comparable to those of a 0.4% agarose gel. Gel formation runs counter to some other models for nuclear pore action, which only work if repeat proteins do not significantly interact with each other.
Repeat proteins with Phe residues mutated to Ser could no longer form the in vitro gels, bind to NTRs, or rescue nucleoporin mutants. Phe-to-Tyr mutants also failed to bind NTRs but could still rescue nucleoporin mutants. Apparently, the Tyr version interacted with the mesh enough to fulfill the essential function—maintenance of a barrier—and other nucleoporins helped out by interacting with NTRs.
The next step, says Görlich, will be to demonstrate that the in vitro gel can act as a selective permeability barrier. How might such a solid-looking structure perform this function? One might think the multivalent interaction between gel and NTR would freeze any movement, but “it's analogous to water,” says Görlich. “A protein diffuses through water even though it has bonds with water. It's just a matter of how fast the rearrangements occur.” The NTRs, he says, must use their hydrophobicity to catalyze the necessary exchange.