Martin wanted to test the models by mixing and matching functional and nonfunctional ClpX subunits. But the six subunits are identical, so coexpression of wild-type and mutant subunits would yield only messy mixtures. “If you were going to make headway you needed to connect the subunits” into a single unit, says Sauer. Martin set to work, but “progress was not encouraging,” says Sauer. “After six months there were only insoluble proteins. I was encouraging him to think of other projects, but he refused to give up.” As a last attempt Martin deleted a nonessential part of ClpX. The deletion—which was done “because we could,” says Sauer—worked.
The resulting construct of six linked subunits could be reassorted at will. Amazingly, activity per functional subunit barely decreased as more mutants crept in. Hexamers with only two active subunits had almost a third of the activity and ATP efficiency of a fully active hexamer. With only one active subunit the hexamers fell down further in activity assays, but still showed impressive digestive powers.
If random mixtures of mutant and wild-type subunits can function so efficiently, Sauer figures that the fully wild-type hexamers probably operate by a random firing of their subunits. This would allow the hexamer to keep processing a protein no matter which ClpX subunit the substrate happened to bump up against.