Cyclooxygenase-2 (COX-2) is constitutively expressed and highly regulated in the thick ascending limb (TAL). As COX-2 inhibitors (Coxibs) increase COX-2 expression, we tested the hypothesis that a negative feedback mechanism involving PGE(2) EP3 receptors regulates COX-2 expression in the TAL. Sprague-Dawley rats were treated with a Coxib [celecoxib (20 mg·kg(-1)·day(-1)) or rofecoxib (10 mg·kg(-1)·day(-1))], with or without sulprostone (20 μg·kg(-1)·day(-1)). Sulprostone was given using two protocols, namely, previous to Coxib treatment (prevention effect; Sulp7-Coxib5 group) and 5 days after initiation of Coxib treatment (regression effect; Coxib10-Sulp5 group). Immunohistochemical and morphometric analysis revealed that the stained area for COX-2-positive TAL cells (μm(2)/field) increased in Coxib-treated rats (Sham: 412 ± 56.3, Coxib: 794 ± 153.3). The Coxib effect was inhibited when sulprostone was used in either the prevention (285 ± 56.9) or regression (345 ± 51.1) protocols. Western blot analysis revealed a 2.1 ± 0.3-fold increase in COX-2 protein expression in the Coxib-treated group, an effect abolished by sulprostone using either the prevention (1.2 ± 0.3-fold) or regression (0.6 ± 0.4-fold vs. control, P < 0.05) protocols. Similarly, the 6.4 ± 0.6-fold increase in COX-2 mRNA abundance induced by Coxibs (P < 0.05) was inhibited by sulprostone; prevention: 0.9 ± 0.3-fold (P < 0.05) and regression: 0.6 ± 0.1 (P < 0.05). Administration of a selective EP3 receptor antagonist, L-798106, also increased the area for COX-2-stained cells, COX-2 mRNA accumulation, and protein expression in the TAL. Collectively, the data suggest that COX-2 levels are regulated by a novel negative feedback loop mediated by PGE(2) acting on its EP3 receptor in the TAL.