Repair of oxidized iron-sulfur clusters in Escherichia coli

Abstract

The [4Fe-4S]2+ clusters of dehydratases are rapidly damaged by univalent oxidants, including hydrogen peroxide, superoxide, and peroxynitrite. The loss of an electron destabilizes the cluster, causing it to release its catalytic iron atom and converting the cluster initially to an inactive [3Fe-4S]1+ form. Continued exposure to oxidants in vitro leads to further iron release. Experiments have shown that these clusters are repaired in vivo. We sought to determine whether repair is mediated by either the Isc or Suf cluster-assembly systems that have been identified in Escherichia coli. We found that all the proteins encoded by the isc operon were critical for de novo assembly, but most of these were unnecessary for cluster repair. IscS, a cysteine desulfurase, appeared to be an exception: although iscS mutants repaired damaged clusters, they did so substantially more slowly than did wild-type cells. Because sulfur mobilization should be required only if clusters degrade beyond the [3Fe-4S]1+ state, we used whole cell EPR to visualize the fate of oxidized enzymes in vivo. Fumarase A was overproduced. Brief exposure of cells to hydrogen peroxide resulted in the appearance of the characteristic [3Fe-4S]1+ signal of the oxidized enzyme. When hydrogen peroxide was then scavenged, the enzyme activity reappeared within minutes, in concert with the disappearance of the EPR signal. Thus it is unclear why IscS is required for efficient repair. The iscS mutants grew poorly, allowing the possibility that metabolic defects indirectly slow the repair process. Our data did indicate that damaged clusters decompose beyond the [3Fe-4S]1+ state in vivo when stress is prolonged. Under the conditions of our experiments, mutants that lacked other repair candidates--Suf proteins, glutathione, and NADPH: ferredoxin reductase--all repaired clusters at normal rates. We conclude that the mechanism of cluster repair is distinct from that of de novo assembly and that this is true because mild oxidative stress does not degrade clusters in vivo to the point of presenting an apoenzyme to the de novo cluster-assembly systems.