Purified PGH synthase when acting on arachidonic acid in the presence of reduced nicotinamideadenine dinucleotide or reduced nicotinamide-adenine dinucleotide 3'-phosphate generated superoxide in burst-like fashion. In eight experiments using different batches of enzyme, the mean ± SE rate of superoxide generation from 100 U of enzyme measured as the superoxide dismutase-inhibitable reduction of cytochrome c was 5.06 ± 0.19 nmol/min in the first minute and 0.35 ± 0.03 nmol/min subsequently. Optimum rates of superoxide were seen at concentrations of reduced nicotinamideadenine dinucleotide in excess of 80 /xM and reduced nicotinamide-adenine dinucleotide 3-phosphate in excess of 100 juM. Using prostaglandin G 2 or linoleic acid as substrate rather than arachidonate also resulted in superoxide generation. When prostaglandin H 2 was used as substrate, no superoxide was generated. The rate of superoxide generation was markedly inhibited by cyclooxygenase inhibitors. Superoxide generation was also observed during the action of lipoxygenase on either linoleic or arachidonic acid in the presence of reduced nicotinamide-adenine dinucleotide or reduced nicotinamide-adenine dinucleotide 3-phosphate but not in their absence. Indomethacin had no effect on superoxide generation from lipoxygenase. We conclude that PGH synthase and lipoxygenase produce superoxide via a side-chain reaction dependent on the presence of suitable reducing cosubstrate. This mechanism is analogous to that described for peroxidases in general. and topical application of arachidonate 89 cause a spectrum of functional, morphological, and biochemical cerebral arteriolar abnormalities. These consist of sustained dilation, reduced responsiveness to vasoconstrictor and vasodilator influences, focal destructive lesions of the endothelium and vascular smooth muscle, and reduced oxygen consumption of the vessel wall. Superoxide anion and other reactive agents derived from it, such as hydrogen peroxide and the hydroxyl radical, are the likely mediators of the cerebral arteriolar abnormalities seen in these conditions, since the latter are inhibited by scavengers of superoxide anion, hydrogen peroxide, and of the hydroxyl radical. 6 Received January 21, 1986; accepted August 25, 1986. ioles to these agents for more than a few minutes causes vascular injury. The enzymatic source of superoxide anion in the experimental conditions noted above is not firmly established. The following evidence suggests strongly that superoxide is produced via the action of the prostaglandin hydroperoxidase. First, the cerebral arteriolar abnormalities noted above can be reproduced by topical application of prostaglandin G 2 (PGG 2 ) 8 or 15-hydroperoxy-eicosatetraenoic acid (15-HPETE)," which are good substrates of the prostaglandin hydroperoxidase. The cerebral arteriolar abnormalities are not produced by prostaglandin H 2 (PGH 2 ), the product of the reaction catalyzed by prostaglandin hydroperoxidase.8 Second, the vasodilation and cerebral arteriolar abnormalities due to hype...