Termination and slippage by bacteriophage T7 RNA polymerase.

We have examined the termination efficiency of T7 and T3 RNA polymerases (RNAPs) at a variety of termination signals. In agreement with previous investigators we find that termination occurs after the synthesis of an RNA product with a stable secondary structure followed by a run of U residues. Stem-loop structures that lack a 3' U-tract fail to terminate the phage enzyme. The distance (or the sequence) between the start site for transcription and the termination signal may also be important, as placing the terminator at different locations downstream from the promoter, or changing the promoter sequence, results in alterations in termination efficiency. We have explored termination at extended runs of homopolymers in the absence of an apparent stem-loop structure, and have observed that the enzyme terminates (inefficiently) when synthesizing U-rich transcripts, but not A- or C-rich transcripts. This is especially true at low concentrations of UTP. Strikingly, when an elongation complex (EC) encounters a dA-tract in the template strand it is able to slide on the template, resulting in the synthesis of products that have more or fewer U residues than predicted by the sequence of the DNA. This observation suggests that the formation of an RNA: DNA hybrid may be important to the lateral stability of the EC (its ability to maintain proper register with the DNA template). We have also explored the termination properties of a proteolytically nicked form of T7 RNAP. The nicked enzyme forms a less stable EC than the intact RNAP and dissociates more readily from the template in regions that encode inherently destabilizing RNAs (e.g. stem-loop structures, poly(U)-tracts). However, the nicked enzyme terminates less efficiently at the late T7 terminator (T7-T phi) or at a termination signal in the human preproparathyroid hormone gene. These results suggest that termination is a highly specific event, and not merely a consequence of decreased stability of the EC. Our observations are not consistent with previous models of termination by the phage RNAP and indicate that revisions to these models may be required.