Using recombinant sulfotransferases (SULTs) expressed in E. coli, beta-estradiol (E2) sulfonation was examined to determine which SULT enzyme is responsible for producing E2-17-sulfate (E2-17-S). SULTs 1A1*1, 1A1*2, 1A3, 1E1 and 2A1 all sulfated E2 to varying extents. No activity was observed with SULT1B1. Among the SULTs studied, SULT2A1 produced primarily E2-3-sulfate (E2-3-S), but also some E2-17-S and trace amounts of E2 disulfate. SULT2A1 had a K(m) value of 1.52 microM for formation of E2-3-S and 2.95 microM for formation of E2-17-S. SULT2A1 had the highest V(max) of 493 pmol/min/mg protein for formation of E2-3-S, which was 8.8- and 47-fold higher than the maximal rates of formation of E2-17-S and E2 disulfate, respectively. SULT2A1 formed E2-3-S more efficiently. However, when celecoxib (0-160 microM) was included in the incubation with either SULT2A1 or human liver cytosol, sulfonation switched from E2-3-S to E2-17-S in a concentration-dependent manner. The ratio of E2-17-S/E2-3-S went up to 15 with SULT2A1, and was saturated at 1 with human liver cytosol. In both cases, more E2-17-S was formed, with the unreacted E2 remained unchanged, suggesting celecoxib probably bound to a separate effector site to cause a conformational change in SULT2A1, which favored production of E2-17-S. The ability of celecoxib to alter the position of sulfonation of E2 may in part explain its success in the experimental prevention and treatment of breast cancer.