Contraction of canine pulmonary artery to hypoxia in vitro is both endothelium dependent and independent. The mechanisms which underlie this phenomenon were studied. Rings of canine pulmonary artery were suspended for isometric force recording in tissue baths containing modified Krebs-Ringer bicarbonate solution. Tissues were first contracted with norepinephrine [effective dose at 35% (ED35) concentration]. Subsequent exposure to hypoxia induced a triphasic response: an initial phasic transient contraction (phase 1), a transient reduction in force (phase 2), followed by a sustained tonic contraction (phase 3). In the absence of endothelium, all phases of the hypoxic response were reduced, and phase 2 was reversed from a contraction to a relaxation (with endothelium: 0.68 +/- 0.2 g; without endothelium: -0.34 +/- 0.1 g). Similar data were obtained in the presence of nitro-L-arginine (3 x 10(-5) M). In the absence of endothelium, indomethacin (10(-5) M) abolished the phase 2 relaxation and converted phase 3 from a contraction to a relaxation (control: 0.99 +/- 0.2 g; indomethacin: -0.44 +/- 0.1 g); and ONO-3708 (thromboxane A2/prostaglandin H2 receptor antagonist) diminished phase 3 (control: 0.99 +/- 0.2 g; ONO-3708: 0.3 +/- 0.04 g). In the absence of endothelium, but in the presence of indomethacin (10(-5) M), K(+)-free solution diminished phase 1 (contraction) and converted phase 2 (relaxation) to a contraction (control: -0.74 +/- 0.1 g; K(+)-free solution: 0.1 +/- 0.06 g). Similar results were obtained with ouabain (4 x 10(-7) M), and cooling of the bathing medium (20 degrees C).(ABSTRACT TRUNCATED AT 250 WORDS)