What They Saw
They looked at purified enzymes from wild-type R. capsulatus and a nifHDK deletion mutant. The latter should only produce the iron-only nitrogenase, if anything. They had to treat the medium to remove as much Mo as possible so as to be able to control the concentration; this reduced the Mo present from around 1 ppb to less than 0.05 ppb (the detection limit).
They found that adding 10μM Mo to cultures growing with no Mo (and thus producing the Fe nitrogenase) greatly increased the amount of ethane produced from acetylene reduction (up to 40% of the ethylene produced). Without Mo, ethane remained constant at about 2% of ethylene. The amount of ethane increased over 72 hours too, up to 68% of ethylene; however, total activity decreased greatly over that time, down to only about 5% of what it had been before adding Mo. The ethane production rate increased for 24 hours and then decreased more slowly than the ethylene rate. This may be due to repression of the Fe nitrogenase, but the authors claim it is not, because the rate decreases more quickly in late-log cultures with chloramphenicol + Mo than with just chloramphenicol (or neither, which was about the same as with chloramphenicol alone). This shows that no new nitrogenase protein is being made even when chloramphenicol is absent, but adding Mo speeds the loss of it.
They also found that the more Mo they added, the higher the proportion of ethane produced (and also the lower the total acetylene reduction activity).
They tested the sensitivity of the system to oxygen, both with and without Mo: in both cases, more oxygen meant less acetylene reduction, though the system with Mo seemed a bit more sensitive (dropping to almost 0% with 1% oxygen while that without Mo only fell to about 20%), but also they noticed that increased oxygen increased the proportion of ethane produced after Mo was added. So somehow oxygen enhanced the Mo effect.
Rhenium, tungsten, and vanadium did not cause anything similar to the Mo effect. The Mo effect was also absent in mutants unable to produce FeMo cofactor (nifE knockouts), so it seems like the cofactor is part of the system. nifQ knockouts seemed to show the effect only at high concentrations of Mo (0.1-1mM); this gene's product is involved in cofactor synthesis at a different step. Mo uptake wasn't an issue; all strains had the same intracellular concentrations.
Using EPR spectroscopy on purified enzyme, they claim to show that the spectrum for Fe nitrogenase with added Mo is similar to that of the Mo nitrogenase from the wild-type, so it seems like the FeMo cofactor is incorporated into the Fe nitrogenase. I would've liked to see their result for the Fe nitrogenase without added Mo as a control, but I'll have to take their word for it. Though they did do metal analysis that found ratios of Fe to Mo similar to that of the Mo nitrogenase.
The fact that chloramphenicol didn't prevent the Mo effect seemed to show that the proteins required to make FeMoco were present before Mo was added. This was confirmed with lacZ fusions to related genes to observe expression more directly.
What This Means
Is FeMoco actually replacing FeFeco in completed enzymes? Seems more likely that FeMoco is inserting into incomplete apoprotein, but it's hard to distinguish between these possibilities. Considering that the process seems to continue over several days, maybe the FeMoco is actually displacing FeFeco from completed proteins over time. This is supported by the observation of oxygen enhancement of the Mo effect; oxygen seems to make the enzyme more labile.