We have used trimethylpsoralen to measure localized levels of unconstrained DNA supercoiling in vivo. The data provide direct evidence that plasmid and chromosomal DNA supercoiling is altered in vivo in an hns mutant. This increase in supercoiling is independent of transcription or changes in the activity of topoisomerase I. These data have implications for the mechanisms by which the chromatin-associated protein H-NS may influence chromosome organization and gene expression.The Escherichia coli chromosome is organized as a highly condensed structure, the nucleoid. The two most abundant architectural proteins in the nucleoid are HU and H-NS (7). H-NS is a 16.5-kDa polypeptide which binds DNA relatively nonspecifically, although it exhibits a preference for curved DNA (15,27,44). Mutations in the hns gene are highly pleiotropic, affecting genome stability (16), recombination-related events (5, 11), and transcription from a variety of promoters (9,10,13,21). Many H-NS-dependent promoters are sensitive to factors which alter DNA supercoiling, and it has been suggested that H-NS may influence transcription through changes in DNA topology (10,11,14,26). Consistent with this hypothesis, H-NS has been shown to constrain DNA supercoils in vitro (42), and hns mutants show changes in the linking number of plasmid DNA isolated from cells (10,12,13).Although changes in plasmid linking number provide an indication of the level of DNA supercoiling in vivo, this method of assessment suffers from several limitations. First, linking number reports the level of supercoiling of naked DNA after purification from the cell. Because of the constraining influence of proteins bound to DNA in vivo, changes in linking number do not necessarily correspond to changes in unconstrained supercoiling in vivo (1, 17). Second, linking number can report only the mean level of supercoiling throughout an entire DNA molecule; localized domains of supercoiling have been shown to exist within a plasmid (25,33,45). Finally, of course, changes in plasmid linking number cannot be used to determine levels of chromosomal supercoiling (8,30).To circumvent these limitations, and to measure plasmid and chromosomal DNA supercoiling directly in vivo, we have used the DNA cross-linking reagent trimethylpsoralen. Psoralen derivatives are able to penetrate into living cells and intercalate into DNA in a supercoiling-dependent fashion (38). Upon irradiation with long-wavelength UV light, psoralen forms covalent cross-links. The rate of formation of these cross-links is proportional to the level of supercoiling of the DNA in the cell (4, 25) and can detect changes in superhelical density of as small as 12% (25). Using this approach, we have shown that hns mutants have increased negative supercoiling of plasmid DNA in vivo. Furthermore, we demonstrate that chromosomal supercoiling is similarly altered. These studies provide a direct demonstration that the net level of negative supercoiling of both plasmid and chromosomal DNA in vivo is increased in hns mutants. These ...