Nonequilibrium mechanism of transcription termination from observations of single RNA polymerase molecules

Cessation of transcription at specific terminator DNA sequences is used by viruses, bacteria, and eukaryotes to regulate the expression of downstream genes, but the mechanisms of transcription termination are poorly characterized. To elucidate the kinetic mechanism of termination at the intrinsic terminators of enteric bacteria, we observed, by using single-molecule light microscopy techniques, the behavior of surface-immobilized Escherichia coli RNA polymerase (RNAP) molecules in vitro. An RNAP molecule remains at a canonical intrinsic terminator for ≈64 s before releasing DNA, implying the formation of an elongation-incompetent (paused) intermediate by transcription complexes that terminate but not by those that read through the terminator. Analysis of pause lifetimes establishes a complete minimal mechanism of termination in which paused intermediate formation is both necessary and sufficient to induce release of RNAP at the terminator. The data suggest that intrinsic terminators function by a nonequilibrium process in which terminator effectiveness is determined by the relative rates of nucleotide addition and paused state entry by the transcription complex.

Balanced branching in transcription termination

The theory of stochastic transcription termination based on free-energy competition [von Hippel, P. H. & Yager, T. D. (1992) Science 255, 809–812 and von Hippel, P. H. & Yager, T. D. (1991) Proc. Natl. Acad. Sci. USA 88, 2307–2311] requires two or more reaction rates to be delicately balanced over a wide range of physical conditions. A large body of work on glasses and large molecules suggests that this balancing should be impossible in such a large system in the absence of a new organizing principle of matter. We review the experimental literature of termination and find no evidence for such a principle, but do find many troubling inconsistencies, most notably, anomalous memory effects. These effects suggest that termination has a deterministic component and may conceivably not be stochastic at all. We find that a key experiment by Wilson and von Hippel [Wilson, K. S. & von Hippel, P. H. (1994) J. Mol. Biol. 244, 36–51] thought to demonstrate stochastic termination was an incorrectly analyzed regulatory effect of Mg2+ binding.

Transcription termination factor La is also an initiation factor for RNA polymerase III.

La RNA-binding protein is a transcription termination factor that facilitates recycling of template and RNA polymerase (pol) 111. Transcription complexes preassembled on immobilized templates were depleted of pol III after a single round of RNA synthesis in the presence of heparin and sarkosyl. The isolated complexes could then be complemented with highly purified pol III and/or recombinant La to test if La is required for transcription reinitiation. VA1, 7SL, and B1 transcription complexes cannot be transcribed by supplemental pol III in single or multiple-round transcription assays unless La is also provided. La mediates concentration-dependent activation of pol III initiation and thereby controls the use of preassembled stable transcription complexes. The initiation factor activity of La augments its termination factor activity to produce a novel mechanism of activated reinitiation. A model in which La serves pol III upon transcription initiation and again at termination is discussed.

Transcription termination of RNA polymerase I due to a T-rich element interacting with Reb1p.

All transcription terminators for RNA polymerase I (pol I) that have been studied so far, ranging from yeast to humans, require a specific DNA binding protein to cause termination. In yeast, this terminator protein has been identified as Reb1p. We now show that, in addition to the binding site for Reb1p, the yeast pol I terminator also requires the presence of a T-rich region coding for the last 12 nucleotides of the transcript. Reb1p cooperates with this T-rich element, both to pause the polymerase and to effect release of the transcript. These findings have implications for the termination mechanism used by all three nuclear RNA polymerases, since all three are known to pause at this terminator.

Transcription termination at intrinsic terminators: the role of the RNA hairpin.

Intrinsic termination of transcription in Escherichia coli involves the formation of an RNA hairpin in the nascent RNA. This hairpin plays a central role in the release of the transcript and polymerase at intrinsic termination sites on the DNA template. We have created variants of the lambda tR2 terminator hairpin and examined the relationship between the structure and stability of this hairpin and the template positions and efficiencies of termination. The results were used to test the simple nucleic acid destabilization model of Yager and von Hippel and showed that this model must be modified to provide a distinct role for the rU-rich sequence in the nascent RNA, since a perfect palindromic sequence that is sufficiently long to form an RNA hairpin that could destabilize the entire putative 12-bp RNA-DNA hybrid does not trigger termination at the expected positions. Rather, our results show that both a stable terminator hairpin and the run of 6-8 rU residues that immediately follows are required for effective intrinsic termination and that termination occurs at specific and invariant template positions relative to these two components. Possible structural or kinetic modifications of the simple model are proposed in the light of these findings and of recent results implicating "inchworming" and possible conformational heterogeneity of transcription complexes in intrinsic termination. Thus, these findings argue that the structure and dimensions of the hairpin are important determinants of the termination-elongation decision and suggest that a complete mechanism is likely to involve specific interactions of the polymerase, the RNA terminator hairpin, and, perhaps, the dT-rich template sequence that codes for the run of rU residues at the 3' end of the nascent transcript.

Molecular coevolution of mammalian ribosomal gene terminator sequences and the transcription termination factor TTF-I.

Both the DNA elements and the nuclear factors that direct termination of ribosomal gene transcription exhibit species-specific differences. Even between mammals--e.g., human and mouse--the termination signals are not identical and the respective transcription termination factors (TTFs) which bind to the terminator sequence are not fully interchangeable. To elucidate the molecular basis for this species-specificity, we have cloned TTF-I from human and mouse cells and compared their structural and functional properties. Recombinant TTF-I exhibits species-specific DNA binding and terminates transcription both in cell-free transcription assays and in transfection experiments. Chimeric constructs of mouse TTF-I and human TTF-I reveal that the major determinant for species-specific DNA binding resides within the C terminus of TTF-I. Replacing 31 C-terminal amino acids of mouse TTF-I with the homologous human sequences relaxes the DNA-binding specificity and, as a consequence, allows the chimeric factor to bind the human terminator sequence and to specifically stop rDNA transcription.

The transcript release factor PTRF augments ribosomal gene transcription by facilitating reinitiation of RNA polymerase I

Termination of murine rDNA transcription by RNA polymerase I (Pol I) requires pausing of Pol I by terminator-bound TTF-I (transcription termination factor for Pol I), followed by dissociation of the ternary complex by PTRF (Pol I and transcript release factor). To examine the functional correlation between transcription termination and initiation, we have compared transcription on terminator-containing and terminator-less rDNA templates. We demonstrate that terminated RNA molecules are more efficiently synthesized than run-off transcripts, indicating that termination facilitates reinitiation. Transcriptional enhancement is observed in multiple- but not single-round transcription assays measuring either promoter-dependent or promoter-independent Pol I transcription. Increased synthesis of terminated transcripts is observed in crude extracts but not in a PTRF-free reconstituted transcription system, indicating that PTRF-mediated release of pre-rRNA is responsible for transcriptional enhancement. Consistent with PTRF serving an important role in modulating the efficiency of rRNA synthesis, PTRF exhibits pronounced charge heterogeneity, is phosphorylated at multiple sites and fractionates into transcriptionally active and inactive forms. The results suggest that regulation of PTRF activity may be an as yet unrecognized means to control the efficiency of ribosomal RNA synthesis.

Interaction between the phage HK022 Nun protein and the nut RNA of phage lambda.

The nun gene product of prophage HK022 excludes phage lambda infection by blocking the expression of genes downstream from the lambda nut sequence. The Nun protein functions both by competing with lambda N transcription-antitermination protein and by actively inducing transcription termination on the lambda chromosome. We demonstrate that Nun binds directly to a stem-loop structure within nut RNA, boxB, which is also the target for the N antiterminator. The two proteins show comparable affinities for boxB and they compete with each other. Their interactions with boxB are similar, as shown by RNase protection experiments, NMR spectroscopy, and analysis of boxB mutants. Each protein binds the 5' strand of the boxB stem and the adjacent loop. The stem does not melt upon the binding of Nun or N, as the 3' strand remains sensitive to a double-strand-specific RNase. The binding of RNA partially protects Nun from proteolysis and changes its NMR spectra. Evidently, although Nun and N bind to the same surface of boxB RNA, their respective complexes interact differently with RNA polymerase, inducing transcription termination or antitermination, respectively.

NusG is required to overcome a kinetic limitation to Rho function at an intragenic terminator.

Rho-dependent transcription termination at certain terminators in Escherichia coli also depends on the presence of NusG [Sullivan, S. L. & Gottesman, M. E. (1992) Cell 68, 989-994]. We have found that termination at the first intragenic terminator in lacZ (tiZ1) is strongly dependent on NusG when transcription is done in vitro with the concentrations of NTPs found in vivo. With a lower level of NTPs, and consequently a slower rate of RNA-chain growth, Rho causes some termination by itself that is enhanced with NusG. These results suggest that NusG serves to overcome a kinetic limitation of Rho to function at certain terminators. At a second intragenic terminator within the lacZ reading frame (tiZ2) the efficiency of Rho-mediated termination was unaffected by either NusG or by RNA polymerase elongation kinetics. Thus, using purified components and intracellular levels of NTPs, we have confirmed the in vivo finding that certain Rho-dependent terminators also depend on NusG, whereas others do not.

Escherichia coli RNA polymerase terminates transcription efficiently at rho-independent terminators on single-stranded DNA templates

Several models have been proposed for the mechanism of transcript termination by Escherichia coli RNA polymerase at rho-independent terminators. Yager and von Hippel (Yager, T. D. & von Hippel, P. H. (1991) Biochemistry 30, 1097–118) postulated that the transcription complex is stabilized by enzyme–nucleic acid interactions and the favorable free energy of a 12-bp RNA–DNA hybrid but is destabilized by the free energy required to maintain an extended transcription bubble. Termination, by their model, is viewed simply as displacement of the RNA transcript from the hybrid helix by reformation of the DNA helix. We have proposed an alternative model where the RNA transcript is stably bound to RNA polymerase primarily through interactions with two single-strand specific RNA-binding sites; termination is triggered by formation of an RNA hairpin that reduces binding of the RNA to one RNA-binding site and, ultimately, leads to its ejection from the complex. To distinguish between these models, we have tested whether E. coli RNA polymerase can terminate transcription at rho-independent terminators on single-stranded DNA. RNA polymerase cannot form a transcription bubble on these templates; thus, the Yager–von Hippel model predicts that intrinsic termination will not occur. We find that transcript elongation on single-stranded DNA templates is hindered somewhat by DNA secondary structure. However, E. coli RNA polymerase efficiently terminates and releases transcripts at several rho-independent terminators on such templates at the same positions as termination occurs on duplex DNAs. Therefore, neither the nontranscribed DNA strand nor the transcription bubble is essential for rho-independent termination by E. coli RNA polymerase.

Intron insertion facilitates amplification of cloned virus cDNA in Escherichia coli while biological activity is reestablished after transcription in vivo.

Insertion of introns into cloned cDNA of two isolates of the plant potyvirus pea seedborne mosaic virus facilitated plasmid amplification in Escherichia coli. Multiple stop codons in the inserted introns interrupted the open reading frame of the virus cDNA, thereby terminating undesired translation of virus proteins in E. coli. Plasmids containing the full-length virus sequences, placed under control of the cauliflower mosaic virus 35S promoter and the nopaline synthase termination signal, were stable and easy to amplify in E. coli if one or more introns were inserted into the virus sequence. These plasmids were infectious when inoculated mechanically onto Pisum sativum leaves. Examination of the cDNA-derived viruses confirmed that intron splicing of in vivo transcribed pre-mRNA had occurred as predicted, reestablishing the virus genome sequences. Symptom development and virus accumulation of the cDNA derived viruses and parental viruses were identical. It is proposed that intron insertion can be used to facilitate manipulation and amplification of cloned DNA fragments that are unstable in, or toxic to, E. coli. When transcribed in vivo in eukaryotic cells, the introns will be eliminated from the sequence and will not interfere with further analysis of protein expression or virus infection.

Bipartite function of a small RNA hairpin in transcription antitermination in bacteriophage lambda.

Transcription of downstream genes in the early operons of phage lambda requires a promoter-proximal element known as nut. This site acts in cis in the form of RNA to assemble a transcription antitermination complex which is composed of lambda N protein and at least four host factors. The nut-site RNA contains a small stem-loop structure called boxB. Here, we show that boxB RNA binds to N protein with high affinity and specificity. While N binding is confined to the 5' subdomain of the stem-loop, specific N recognition relies on both an intact stem-loop structure and two critical nucleotides in the pentamer loop. Substitutions of these nucleotides affect both N binding and antitermination. Remarkably, substitutions of other loop nucleotides also diminish antitermination in vivo, yet they have no detectable effect on N binding in vitro. These 3' loop mutants fail to support antitermination in a minimal system with RNA polymerase (RNAP), N, and the host factor NusA. Furthermore, the ability of NusA to stimulate the formation of the RNAP-boxB-N complex is diminished with these mutants. Hence, we suggest that boxB RNA performs two critical functions in antitermination. First, boxB binds to N and secures it near RNAP to enhance their interaction, presumably by increasing the local concentration of N. Second, boxB cooperates with NusA, most likely to bring N and RNAP in close contact and transform RNAP to the termination-resistant state.

Requirement for the c-Maf transcription factor in crystallin gene regulation and lens development

The vertebrate lens is a tissue composed of terminally differentiated fiber cells and anterior lens epithelial cells. The abundant, preferential expression of the soluble proteins called crystallins creates a transparent, refractive index gradient in the lens. Several transcription factors such as Pax6, Sox1, and L-Maf have been shown to regulate lens development. Here we show that mice lacking the transcription factor c-Maf are microphthalmic secondary to defective lens formation, specifically from the failure of posterior lens fiber elongation. The marked impairment of crystallin gene expression observed is likely explained by the ability of c-Maf to transactivate the crystallin gene promoter. Thus, c-Maf is required for the differentiation of the vertebrate lens.

The Ets transcription factor ERM is Th1-specific and induced by IL-12 through a Stat4-dependent pathway

Interleukin 12 (IL-12)-induced T helper 1 (Th1) development requires Stat4 activation. However, antigen-activated Th1 cells can produce interferon γ (IFN-γ) independently of IL-12 and Stat4 activation. Thus, in differentiated Th1 cells, factors regulated by IL-12 and Stat4 may be involved in IFN-γ production. Using subtractive cloning, we identified ERM, an Ets transcription factor, to be a Th1-specific, IL-12-induced gene. IL-12-induction of ERM occurred in wild-type and Stat1-deficient, but not Stat4-deficient, T cells, suggesting ERM is Stat4-inducible. Retroviral expression of ERM did not restore IFN-γ production in Stat4-deficient T cells, but augmented IFN-γ expression in Stat4-heterozygous T cells. Ets factors frequently regulate transcription via cooperative interactions with other transcription factors, and ERM has been reported to cooperate with c-Jun. However, in the absence of other transcription factors, ERM augmented expression of an IFN-γ reporter by only 2-fold. Thus, determining the requirement for ERM in Th1 development likely will require gene targeting.

Induction of ARF tumor suppressor gene expression and cell cycle arrest by transcription factor DMP1

Expression of the DMP1 transcription factor, a cyclin D-binding Myb-like protein, induces growth arrest in mouse embryo fibroblast strains but is devoid of antiproliferative activity in primary diploid fibroblasts that lack the ARF tumor suppressor gene. DMP1 binds to a single canonical recognition site in the ARF promoter to activate gene expression, and in turn, p19ARF synthesis causes p53-dependent cell cycle arrest. Unlike genes such as Myc, adenovirus E1A, and E2F-1, which, when overexpressed, activate the ARF-p53 pathway and trigger apoptosis, DMP1, like ARF itself, does not induce programmed cell death. Therefore, apart from its recently recognized role in protecting cells from potentially oncogenic signals, ARF can be induced in response to antiproliferative stimuli that do not obligatorily lead to apoptosis.