P rolonged hypokalemia causes vasopressin-resistant polyuria. We have recently shown that another cause of severe polyuria, chronic lithium therapy, is associated with decreased aquaporin-2 (AQP2) water channel expression (Marples, D., S. Christensen, E.I. Christensen, P.D. Ottosen, and S. Nielsen, 1995. J. Clin. Invest., 95: 1838-1845). Consequently, we studied the effect in rats of 11 days' potassium deprivation on urine production and AQP2 expression and distribution. Membrane fractions were prepared from one kidney, while the contralateral kidney was perfusion-fixed for immunocytochemistry. Immunoblotting and densitometry revealed a decrease in AQP2 levels to 27+/-3.4% of control levels (n=11, P<0.001) in inner medulla, and 34+/-15% of controls (n=5, P<0.05) in cortex. Urine production increased in parallel, from 11+/-1.4 to 30+/-4.4 ml/day (n=11, P<0.01). After return to a potassium-containing diet both urine output and AQP2 labels normalized within 7 d. Immunocytochemistry confirmed decreased AQP2 labeling in principal cells of both inner medullary and cortical collecting ducts. AQP2 labeling was predominantly associated with the apical plasma membrane and intracellular vesicles. Lithium treatment for 24 d caused a more extensive reduction of AQP2 levels, to 4+/-1% of control levels in the inner medulla and 4+/-2% in cortex, in association with severe polyuria. The similar degree of downregulation in medulla and cortex suggests that interstitial tonicity is not the major factor in the regulation of AQP2 expression. Consistent with this furosemide treatment did not alter AQP2 levels. In summary,hypokalemia, like lithium treatment, results in a decrease in AQP2 expression in rat collecting ducts, in parallel with the development of polyuria, and the degree of downregulation is consistent with the level of polyuria induced, supporting the view that there is a causative link.