The purposes of the present study were to examine the natural course of the impairment of endothelium-dependent relaxations during a regeneration and tissue repair process after balloon endothelium removal and to elucidate the cellular mechanism(s) underlying it. Twenty-three male Yorkshire pigs underwent balloon endothelium removal along the proximal portion of either the left anterior descending or circumflex coronary artery and were then maintained on a regular chow for 4, 8, 16, or 24 weeks. Endothelium-dependent responses were examined in vitro in rings taken from the control and previously denuded arteries studied in parallel. Morphometric analysis revealed that intimal thickening developed only at the previously denuded area. In the previously denuded arteries with regenerated endothelium, the endothelium-dependent relaxations to UK 14304 (a selective alpha 2-adrenergic agonist), serotonin, and aggregating platelets were impaired 4 weeks after endothelium removal and remained so throughout the study. The endothelium-dependent relaxations to thrombin and adenosine diphosphate became depressed 8 weeks after endothelium removal and those to bradykinin became depressed 16 weeks after endothelium removal, while those to the calcium ionophore A23187 were maintained throughout the study. Endothelium-dependent relaxations to all vasoactive agents were unaltered in the control arteries. In the control arteries, pertussis toxin, an inhibitor of certain G proteins, markedly inhibited the endothelium-dependent relaxations to UK 14304 and serotonin and partially inhibited those to thrombin and aggregating platelets. The responses inhibited by the toxin in control arteries were significantly reduced in the reduced in the previously denuded arteries with regenerated endothelium. The inhibitory effect of pertussis toxin was markedly reduced in those arteries with regenerated endothelium. In quiescent rings, the presence of normal endothelium inhibited the contractions caused by serotonin and aggregating platelets; this endothelium-dependent depression was markedly impaired in the previously denuded arteries throughout the study. Direct relaxation of the coronary smooth muscle to nitric oxide or sodium nitroprusside or direct contraction to KCl or serotonin were comparable between the control and previously denuded arteries. These experiments indicate that endothelium-dependent relaxations progressively worsen after regeneration of the endothelium and that the dysfunction of a pertussis toxin-sensitive G protein partly account for the endothelial dysfunction in the chronic regenerated state.