Escherchia coli RNA polymerases containing mutated a subunits were tested for their ability to respond to three different positive regulators (activators) in vitro. The two a (rpoA) mutants, a-256 and a-235, have deletions of the C-terminal 73 and 94 amino acids, respectively. In runoff transcription assays catalyzed by reconstituted holoenzyme, the effects of the mutations on each of three promoters tested were different: activation of the XpRm promoter by cI protein (repressor) was nearly normal, activation of the XpRE promoter by cI protein was reduced approximately fivefold, and direct activation of the trpPB promoter ofPseudomonas aeruginosa was completely inhibited. We also found that the reconstituted mutant enzyme was defective in recognition of trpPI in the absence of activator. The differential responses of the three promoters to their activators in the presence of the mutant enzymes indicate that the location of an activator-binding site does not by itself determine the region of RNA polymerase with which the activator interacts.Recently, several groups have reported evidence implicating the a subunit of Escherichia coli RNA polymerase (RNAP) in positive control of transcription (13,20,21,23,26). Since at least some activators appear to interact directly with RNAP (18,27), it is possible that activation is mediated by direct contact between particular activators and a. The simplest explanation of the available data would be that the contact point is in the C-terminal third of a, since most rpoA mutations that cause defects in activation are located in this region. However, since RNAP is a multisubunit enzyme (a%p'o%), indirect effects of the mutations on RNAP function cannot be excluded.The effects of two particular mutant subunits, a-235 and a-256, which have deletions of the C-terminal 94 and 73 amino acids, respectively, have been tested extensively in vitro, by using RNAP holoenzyme reconstituted with purified mutant a protein (20,21). Transcription catalyzed by the mutant enzymes was activated normally when cyclic AMPcatabolite gene activator protein (CAP) was bound near the -35 region of the galPi promoter at a site whose midpoint is at -41.5 (41.5 nucleotides preceding the transcription start site) but not when the activator was bound upstream from the lacPl promoter at a site with a midpoint at -61.5. Extension of these studies to several other activators led to the hypothesis that activators bound close to the -35 regions of their target promoters would, in general, contact RNAP in a region not affected by the a deletions, while activators bound farther upstream would contact the region removed by the deletions (20, 21).To test this hypothesis, we examined activation of three promoters: (i) X pRm, which is activated by binding of cI protein (repressor) to OR2 (24); (ii) XpRE, which is activated by binding of cII protein to two TTGC sequences that flank the -35 region (29); and (iii) the Pseudomonas aeruginosa promoter, trpPB, which is activated by TrpI protein bound to recognition site II, th...