What They Saw
The normal Azotobacter medium (Burk's) has enough contaminating nickel that adding it is unnecessary. But when they added a chelator (nitrilotriacetate) to bind it up, the hydrogenase activity decreased by 80% without affecting growth. This inhibition was lessened by adding nickel.
Nickel availability seemed to affect which form of the alpha subunit was present: the larger, unprocessed form, or the smaller, mature form. With nickel available, only the smaller form was seen; when it was bound up, only the larger. But when excess nickel was added, following the proteins over time showed that gradually the population shifted from larger to smaller as the nickel was used. These two forms are found in different places: the smaller is bound to the membrane (as it should be), and the larger is soluble.
Inhibiting protein synthesis, such as with chloramphenicol, and then adding nickel led to a similar increase in activity as a control without an inhibitor, up to 70 minutes; so for this period, increasing activity wasn't due to protein synthesis. But after this point, the inhibited cultures stopped increasing while the uninhibited continued. The processing of the large form into the small continued regardless of inhibition. So it seems that nickel is important partially for processing and partially for stimulating protein synthesis.
In vitro, ATP or GTP was important for processing. Membranes and oxygen (or lack thereof) were not important. No divalent cation could substitute for nickel: zinc inhibited processing completely, and cobalt or calcium some too. The only protease inhibitor that prevented processing was 1,10-phenanthroline, which inhibits metal-activated proteases.
What This Means
It seems that nickel and processing are both essential for hydrogenase activity, and apparently they are interrelated. It's possible that the processing is regulated by the presence of nickel; without the metal, there isn't much point. Or maybe processing without nickel available will lead to nonfunctional product that can't be fixed. Alternatively, the protease that does the processing could require nickel. It's hard to distinguish these possibilities though. Anyway, it seems like when nickel is absent, the hydrogenase subunits are present but in a premature form, waiting for nickel. How poetic.