Abstract
NQO1 [NAD(P)H:quinone oxidoreductase 1] has an integral role in cellular responses to oxidative stress. The expression of NQO1 is up-regulated in the mouse following challenge with electrophilic chemicals, in an Nrf2 (NF-E2 p45-related factor 2)-dependent fashion, but the molecular basis for this observation remains unexplained. Through characterization of the murine nqo1 5'-upstream region, we now show that Nrf2 regulates this gene directly via an ARE (antioxidant response element) that lies within a 24 bp region spanning nt -444 to -421. A comprehensive mutation study of this ARE revealed that it does not conform to the currently accepted ARE consensus sequence [(5'-TMAnnRTGAYnnnGCRwwww-3', with essential nucleotides shown in capitals); two cytosine residues (shown in bold in the following sequence) that have been designated 'n' previously because they were thought to be redundant (5'-gagTcA C aGTgAGt C ggCAaaatt-3') have now been found to be essential for enhancer activity; two guanines (also shown in bold) previously regarded as essential for ARE function (5'-gagTcACaGT g AGtCg g CAaaatt-3') have proven to be dispensable]. Examination of wild-type and nrf2 (-/-) mouse embryonic fibroblasts demonstrated that Nrf2 is essential for both constitutive expression of NQO1 and its induction by sulphoraphane. Electrophoretic mobility-shift and chromatin immunoprecipitation assays revealed that Nrf2 associates, in low amounts, with the nqo1 ARE under constitutive conditions, and following sulphoraphane challenge of cells, Nrf2 is recruited to the ARE in substantially greater quantities, as a heterodimer with the small Maf (musculoaponeurotic fibrosarcoma virus) protein, MafK. Also, MafK was found to bind the nqo1 ARE in an Nrf2-independent fashion, and may contribute to transcriptional repression of the oxidoreductase gene. These findings allow a model for transcriptional control of nqo1 through the ARE to be proposed. Furthermore, our results indicate that distinct AREs have differential sequence requirements, and a universally applicable consensus sequence cannot be derived.