Mapping the site(s) of MgATP and MgADP interaction with the nitrogenase of Azotobacter vinelandii: lysine 15 of the iron protein plays a major role in MgATP interaction

Nitrogenase binds and hydrolyzes 2MgATP yielding 2MgADP and 2P, for each electron that is transferred from the iron protein to the MoFe protein. The iron protein alone binds but does not hydrolyze 2MgATP or 2MgADP and the binding of these nucleotides is competitive. Iron protein amino acid sequences all contain a putatitive mononucleotide-binding region similar to a region found in other mononucleotide-binding proteins. To examine the role of this region in MgATP interaction, we have substituted glutamine and proline for conserved lysine 15. The amino acid substitutions, K15Q and K15P, both yielded a non-N2-fixing phenotype when the genes coding for them were substituted into the Azotobacter vinelandii chromosome in place of the wild-type gene. The iron protein from the K15Q mutant was purified to homogeneity, whereas the protein from the K15P mutant could not be purified in its native form. Unlike wild-type iron protein, the purified K15Q iron protein showed no acetylene reduction, H2 evolution, or ATP hydrolysis activities when complemented with wild-type MoFe protein. The K15Q iron protein and the normal iron protein had a similar total iron content and both proteins showed the characteristic rhombic EPR signal resulting from the reduced state of the single 4Fe-4S cluster bridging the two subunits. Unlike the wild-type iron protein, addition of MgATP to the K15Q iron protein did not result in the perturbation necessary to change the EPR signal of its 4Fe-4S center from a rhombic to an axial line shape. Also unlike the wild-type iron protein, addition of MgATP to K15Q iron protein in the presence of the iron chelator, alpha,alpha'-dipyridyl, did not result in a time-dependent transfer of iron to the chelator. Thus, even though the K15Q iron protein contains a normal 4Fe-4S center, it does not respond to MgATP like the wild-type protein. Examination of the ability of the K15Q iron protein to bind MgADP showed no change from the wild-type iron protein, but its ability to bind MgATP decreased to 35% of the wild-type protein. Thus, in A. vinelandii iron protein, lysine 15 is not needed for interaction with MgADP but is involved in the binding of ATP, presumably through charge-charge interaction with the gamma-phosphate. Based on the above data, this lysine appears to be essential for the MgATP induced conformational change of wild-type iron protein that is required for activity. This is supported additionally by our finding that the K15Q iron protein did not compete with wild-type iron protein for interaction with the MoFe protein in activity assays. In spite of this, like the wild-type iron protein, the K15Q iron protein still formed the reported 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) cross-linked complex with the beta subunit of the MoFe protein. These results suggest that only a specific interaction of iron protein with MgATP forms a catalytically relevant complex with the MoFe protein and that the sites reported to be involved in EDC cross-linking are not relevant to this catalytically competent interaction.