A WHOLE GENOME APPROACH TO INVIVO DNA-PROTEIN INTERACTIONS IN ESCHERICHIA-COLI

THE increasingly rapid pace at which genomic DNA sequences are being determined has created a need for more efficient techniques to determine which parts of these sequences are bound in vivo by the proteins controlling processes such as gene expression, DNA replication and chromosomal mechanics. Here we describe a whole-genome approach to identify and characterize such DNA sequences. The method uses endogenous or artificially introduced methylases to methylate all genomic targets except those protected in vivo by protein or non-protein factors interfering with methylase action. These protected targets remain unmethylated in purified genomic DNA and are identified using methylation-sensitive restriction endonucleases. When the method was applied to the Escherichia coli genome, 0.1% of the endogenous adenine methyltransferase (Dam methylase) targets were found to be unmethylated. Five foreign methylases were examined by transfection. Database-matched DNA sequences flanking the in vivo-protected Dam sites all fell in the non-coding regions of seven E. coli operons (mtl, cdd, flh, gut, car, psp and fep). In the first four operons these DNA sequences closely matched the consensus sequence that binds to the cyclic AMP-receptor protein. The in vivo protection at the Dam site upstream of the car operon was correlated with a downregulation of car expression, as expected of a feedback repressor-binding model.