We have analyzed subunit interactions of Escherichia coli RNA polymerase by measuring the rate of cleavage of each subunit by trypsin. We have modified a standard sodium dodecyl sulfate gel technique to include sodium tetradecyl sulfate, which results in a large separation of the beta and beta' subunits. A comparison of enzyme lacking the sigma subunit (core) with holoenzyme reveals that the presence of sigma does not alter the cleavage rates of alpha or beta, but it dramatically slows the initial cleavage of beta. Simple addition of purified sigma to core results in reconstitution of holoenzyme as measured both by glycerol gradient sedimentation and protection of beta from trypsin cleavage. However, Mg2+ is required for the beta protection but not for the binding of sigma to core. Although the enzyme reconstituted in the absence of Mg2+ and lacking the sigma-beta association is similar to holoenzyme with respect to promoter binding, it catalyzes the transition from closed to open complexes much more slowly than does enzyme reconstituted in the presence of Mg2+. Thus, the specific association between beta and sigma may be involved in the DNA melting phase of the RNA polymerase-DNA interaction. We have also examined the effects of the polyanion heparin, an RNA polymerase initiation inhibitor, on the rates of trypsin cleavage of holoenzyme subunits. Again, subunits alpha and beta' were unaffected, but heparin increased the cleavage rate of beta such that the rate was indistinguishable from the beta cleavage rate of core enzyme. Since heparin did not cause the release of sigma as measured by glycerol gradient analysis, the inhibitor may simply disrupt the sigma-beta association.