The marRAB multiple antibiotic resistance operon of Escherichia coli controls the expression of a large number of genes resulting in low level antibiotic and superoxide resistance through a complex network of reactions. MarR auto-represses the mar operon but is inactivated upon interaction with salicylate, losing its DNA binding capacity. This, in turn, results in derepression of the operon and expression of MarA, which activates the transcription of some 40 to 60 promoters (the mar/sox/rob regulon) including the marRAB promoter itself (auto-activation). SoxS, a very closely related transcriptional activator of the same promoters, is separately regulated. Both are members of the AraC family which in E. coli contains 22 members. Stimulation of soxS expression by exposure to superoxides also leads to low level antibiotic resistance but to a far higher resistance to superoxides than does the stimulation of MarA expression. The critical question thus becomes how MarA and SoxS differ at the different promoters. We have shown previously that the capacity to bind to all of the many promoter binding sites alone is not the answer. For promoters where the DNA binding site of MarA does not overlap the DNA binding site of RNA polymerase (Class I promoters) activation by MarA and SoxS has been shown to require interaction with the alpha subunit of RNA polymerase at positions where MarA and SoxS are nearly identical. We have now found that the difference between the activation by MarA and SoxS of these promoters is largely influenced by a single glutamic acid on the surface of MarA (but absent in SoxS) that may interfere with DNA binding of MarA at certain binding sites and thus may account for the different abilities of MarA and SoxS to stimulate. Thus, binding affinity alone may be the answer for Class I promoters. For a far larger subclass of the promoters, namely those where the DNA binding region overlaps the DNA binding region of RNA polymerase (Class II promoters), an additional polar interaction between the sigma subunit of RNA polymerase and the activator is thought to be essential. Curiously, MarA shares virtually no homology with the region of SoxS that has been reported to be responsible for the polar interaction with the sigma subunit. How then is it possible for MarA to activate these promoters? Through a series of analyses we now have preliminary evidence that both SoxS and MarA proffer different surface acidic amino acids for their interaction with RNA polymerase depending on the precise location of the binding site from the -10 signal for the binding of RNA polymerase. Thus, when the binding site is separated by 17 bp from the -10 signal, SoxS principally uses the aspartic acid at position 79, but when the separation increases to 18 or 19 bp this aspartic acid becomes less important and that at position 75 becomes crucial. SoxS is virtually non-functional when the separation becomes 20 bp. In general, SoxS is a far better activator than MarA at separations of 17 bp, whereas MarA is the better activator when the separation is 20 bp. The critical surface acidic amino acid for MarA appears to be the glutamic acid located at position 84 when the separation is 19 or 20 bp. We have been unable to identify a critical acidic acid for MarA at shorter separations. This work was carried out in collaboration principally with Drs. J.L. Rosner and Michael Wall (Computer and Computational Sciences &Bioscience, Los Alamos National Laboratory, Mail Stop B256,Los Alamos NM 87545 USA).
