Evidence that protein thiols are not primary targets of intracellular reactive oxygen species in growing Escherichia coli

Abstract

The oxidizability of cysteine residues is exploited in redox chemistry and as a source of stabilizing disulfide bonds, but it also raises the possibility that these side chains will be oxidized when they should not be. It has often been suggested that intracellular oxidative stress from hydrogen peroxide or superoxide may result in the oxidation of the cysteine residues of cytoplasmic proteins. That view seemed to be supported by the discovery that one cellular response to hydrogen peroxide is the induction of glutaredoxin 1 and thioredoxin 2. In this study we used model compounds as well as alkaline phosphatase to test this idea. Our results indicate that molecular oxygen, superoxide, and hydrogen peroxide are very poor oxidants of N-acetylcysteine and of the protein thiols of alkaline phosphatase in vitro. Copper could accelerate thiol oxidation, but iron did not. When alkaline phosphatase was engineered to remain in the cytoplasm of live cells, unnaturally high concentrations of hydrogen peroxide were required to oxidize it to its active, disulfide-dependent form, and toxic levels of superoxide had no effect. At the same time, far lower concentrations of these oxidants were sufficient to poison key metalloenzymes. The elimination of glutaredoxin 1 and thioredoxin 2 did not change these results, raising the question of why E. coli induces them during peroxide stress. In fact, when catalase/peroxidase mutants were chronically stressed with hydrogen peroxide, the absence of glutaredoxin 1 and thioredoxin 2 did not impair growth at all, even in a minimal medium over many generations. We conclude that physiological levels of reduced oxygen species are not potent oxidants of typical protein thiols. Glutaredoxin and thioredoxin must either have an alternative purpose or else play a role under culture conditions that differ from the ones we tested.