A minimal RNA polymerase III transcription system
1999; Springer Nature; Volume: 18; Issue: 18 Linguagem: Inglês
10.1093/emboj/18.18.5042
ISSN1460-2075
Autores Tópico(s)RNA and protein synthesis mechanisms
ResumoArticle15 September 1999free access A minimal RNA polymerase III transcription system George A. Kassavetis Corresponding Author George A. Kassavetis Department of Biology and Center for Molecular Genetics, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA, 92093-0634 USA Search for more papers by this author Garth A. Letts Garth A. Letts Department of Biology and Center for Molecular Genetics, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA, 92093-0634 USA Search for more papers by this author E. Peter Geiduschek Corresponding Author E. Peter Geiduschek Department of Biology and Center for Molecular Genetics, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA, 92093-0634 USA Search for more papers by this author George A. Kassavetis Corresponding Author George A. Kassavetis Department of Biology and Center for Molecular Genetics, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA, 92093-0634 USA Search for more papers by this author Garth A. Letts Garth A. Letts Department of Biology and Center for Molecular Genetics, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA, 92093-0634 USA Search for more papers by this author E. Peter Geiduschek Corresponding Author E. Peter Geiduschek Department of Biology and Center for Molecular Genetics, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA, 92093-0634 USA Search for more papers by this author Author Information George A. Kassavetis 1, Garth A. Letts1 and E. Peter Geiduschek 1 1Department of Biology and Center for Molecular Genetics, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA, 92093-0634 USA *Corresponding authors. E-mail: [email protected] or E-mail: [email protected] The EMBO Journal (1999)18:5042-5051https://doi.org/10.1093/emboj/18.18.5042 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Transcription factor (TF) IIIB recruits RNA polymerase (pol) III for specific initiation of transcription. All three subunits of TFIIIB, TBP, Brf (the TFIIB-related subunit) and B″, are required for transcription of supercoiled and linear duplex DNA, but we show here that B″ is non-essential on a promoter that has been partly pre-opened by unpairing a short segment of the transcription bubble. These findings expose a striking similarity between transcriptional initiation by pol II, pol III and bacterial RNA polymerases: a preformed single-stranded DNA bubble upstream of the transcriptional start removes the dependence of pol II on TFIIE, TFIIH and ATP hydrolysis, and the dependence of pol III on B″; the favored placement of the transcription bubble for B″-independent transcription by pol III overlaps a DNA segment that interacts sequence specifically as single-stranded DNA with the σ70 initiation subunit of Escherichia coli RNA polymerase holoenzyme. Introduction The core transcription apparatus of yeast RNA polymerase (pol) III comprises the 18-subunit polymerase (Chédin et al., 1999), together with its transcription initiation factors TFIIIA, B and C. TFIIIB, which is responsible for directing pol III to its promoters, is composed of the TATA-binding protein (TBP), Brf, the TFIIB-related and archaeal TFB-related subunit, and the pol III-specific B″ (Buratowski and Zhou, 1992; Colbert and Hahn, 1992; López-De-León et al., 1992). (Brf is also referred to as TFIIIB70 and B″ is also referred to as TFIIIB90.) TFIIIC is a large general assembly factor that places TFIIIB on the DNA template, and the multi-zinc finger TFIIIA is a 5S RNA gene-specific assembly factor for TFIIIC. TFIIIB also binds autonomously to promoters with strong TATA boxes through a direct interaction of its TBP subunit. When pol III is brought to the transcriptional start site by TFIIIB, the DNA duplex around the transcriptional start site is spontaneously and thermoreversibly opened in linear as well as in supercoiled DNA (Kassavetis et al., 1990, 1992a; White, 1998). Recent functional dissections of the Brf and B″ subunits of TFIIIB have shown that certain internal deletions in B″ and N-terminal deletions in Brf generate TFIIIB assemblies that retain competence for directing accurately-initiating transcription of supercoiled plasmid DNA by pol III, but are inactive in the context of linear DNA. The defective TFIIIBs recruit pol III and position it appropriately over the transcriptional start site, but the promoter does not open. In consideration of these findings, we have proposed that the normal function of TFIIIB combines pol III recruitment with an essential role in subsequent steps of transcriptional initiation (Kassavetis et al., 1998b). The subsequent analysis of these defective TFIIIB assemblies has focused on how to restore promoter opening in linear DNA. It is known that pol III opens its SUP4 tRNA gene promoter non-coordinately, in two segments, one region surrounding the transcriptional start and extending downstream, and another region that opens at a lower temperature centered ∼7 bp further upstream (Kassavetis et al., 1992a). Accordingly, the effect on transcription of unpairing these segments of the transcription bubble by forming loops has been examined. In the course of that analysis, we have found conditions under which Brf and TBP alone secure accurately-initiating transcription by pol III: the requirement for participation of B″ is circumvented if the pol III promoter is partially pre-opened by unpairing DNA upstream of the transcriptional start site. The exploration of this B″-independent pol III transcription, and a discussion of its implications for the existence of a common reaction pathway to initiation of transcription by eukaryotic and bacterial RNA polymerases, are presented. Results B″-independent transcription The pol III transcription bubble opens non-coordinately: its upstream segment (bp −9 to −5) opens at lower temperature, and its downstream segment (bp −3 to +11) responds selectively to the presence of the initiating nucleotide in a temperature downshift (Kassavetis et al., 1992a). In order to explore the defects of certain Brf and B″ deletion proteins that assemble pol III over the start site of transcription on linear DNA templates but are unable to form a transcription bubble, we constructed a series of DNA templates, based on the TFIIIC-independent SNR6 (U6) gene, containing short single-stranded bubbles spanning the region unwound by pol III in the SUP4 gene open complex (Figure 1). One of these DNA templates, construct −9/−5, yielded the surprising finding that full-length Brf and TBP alone assemble pol III for transcription at a significant level (Figure 2): pol III alone only weakly utilizes the −9/−5 bubble construct, initiating transcription near bp +1, and the level of this background transcription is unchanged by TBP, B″ or Brf alone, or by B″ with TBP (Figure 2, lane a and data not shown). Addition of full-length Brf (wt) and TBP together (referred to as B′; Kassavetis et al., 1991) increases transcription 5- to 6-fold (Figure 2, lane b), but is ineffectual for promoting transcription of the fully duplex DNA (Figure 2, lane h), as expected. B′-directed transcription is at ∼5% of the level generated by the complete TFIIIB–DNA complex (compare Figure 2, lanes b and c). Removing Brf segments that are not present in the Candida albicans and Kluyveromyces lactis homologues (Khoo et al., 1994; based on two alternative alignments) generates two variant proteins that are somewhat more active for transcription of bubble-containing DNA and elevate B″-independent transcription to >10-fold over background (Figure 2, compare lanes d and f with a). We have therefore used these variant proteins for experiments that are shown below (but principal observations have been confirmed with wild-type Brf). Figure 1.Transcription templates with single-stranded DNA bubbles. The DNA, whose sequence is based on the SNR6 gene in pU6RboxB (Whitehall et al., 1995), extends from bp −60 to +138 (relative to the transcriptional start as +1). The template (bottom) strand is identical in all constructs. Download figure Download PowerPoint Figure 2.A TBP–Brf–DNA complex that assembles pol III for specific transcription. TFIIIB–DNA and B′ (i.e. TBP–Brf)–DNA complexes were formed on fully duplex and −9/−5 bubble DNA, as indicated at the top of the figure, for 60 min (TBP was present in all reaction mixtures). Pol III was added for a further 10 min and multiple rounds of transcription were then allowed for 30 min. The yield of U6 RNA is specified below each lane relative to transcription elicited by the TFIIIB(BrfΔ383–424)–duplex DNA complex, set at 100. The U6 transcript, 32P-labeled −9/−5 bubble construct ('DNA'; used as a tracer during template preparation) and a labeled DNA recovery marker (r.m.) are identified on the left. Download figure Download PowerPoint The ability of the B′–DNA complex to direct transcription is highly sensitive to bubble placement (relative to the SNR6 TATA box) (Figure 3A; BrfΔ383–424 was used for this experiment). Bubble-containing templates function less well for TFIIIB-directed transcription than does fully duplex DNA, and bubble templates −4/+1, +2/+6 and +7/+11 are consistently less active than −14/−10 and −9/−5 (Figure 3A, left panel, top, and data not shown). All templates specify TFIIIB-dependent initiation at bp +1 (Figure 3A, left panel, bottom), with bubble template +2/+6 and +7/+11 also specifying initiation at bp +5 and +7, respectively (Figure 3A, compare the left- and right-hand panels); initiation at bp +10 with construct +7/+11 is not factor-dependent. B′-directed transcription proves to be more selective (Figure 3A, middle panel): constructs −9/−5 and −4/+1 generate initiation at bp +1 that is 7 and 10%, respectively, of transcription provided by complete TFIIIB (left panel), and 30- and 5-fold, respectively, above the no-Brf background (Figure 3A, right panel; note that the background initiation at bp +1 is elevated with bubble template −4/+1). Fully duplex DNA, and constructs −14/−10, +2/+6 and +7/+11 fail to support B′ complex-dependent transcription at significant levels above the no-Brf background (at each potential site of initiation; compare the middle and right-hand panels). Thus, among the 5 bp bubble templates that were tested, construct −9/−5 is optimal for B′-dependent transcription. Reducing the size of the bubble to 3 bp (constructs −9/−7 and −7/−5) allows retention of some capacity for B′-dependent transcription, but at less than one-eighth the efficiency of the 5 bp bubble template (Figure 3B). Figure 3.Requirements for the location and size of the DNA bubble generating B″-independent transcription. (A) Effect of bubble placement on transcription and initiation site selection. TFIIIB–DNA (left panel), B′–DNA (middle panel) and TBP–DNA complexes (right panel) were used for multiple rounds of transcription by pol III. The RNA products are analyzed directly by denaturing gel electrophoresis (upper segment of each panel) and primer extension (lower segment). Upper segment of each panel: 32P-labeled RNA; the size of the more slowly migrating transcripts is consistent with read-through (rt) of the SNR6 transcriptional terminator. A labeled DNA recovery marker is not shown. Lower segment of each panel: otherwise identical samples with unlabeled transcripts were processed for mapping of RNA 5′ ends by primer extension with reverse transcriptase. The accompanying sequence ladder is not shown. Transcripts (excluding the read-through and +10 initiated products) and reverse transcripts were quantified and are specified below each lane: relative to transcription with TFIIIB on duplex DNA, set to 100, for the upper segments, and relative to the reverse transcript for +1-initiating RNA formed with TFIIIB on duplex DNA, set to 100, for the lower segments. BrfΔ383–424 was used for this experiment. (B) Three-base-pair bubbles also elicit B″-independent transcription. B′–pol III–DNA and TBP–pol III–DNA complexes (identified above each lane by the presence or absence of full-length Brf) were used for multiple rounds of transcription and analyzed as specified in (A) by direct examination of transcripts (left panel) and primer extension (right panel). U6 RNA synthesis and +1 initiation are quantified below each lane relative to B′-dependent transcription of the −9/−5 bubble template, set to 100. The lane at the far right (which is blank) shows a control reaction mixture lacking pol III analyzed by primer extension. Download figure Download PowerPoint Pol III is able to initiate transcription in the absence of TFIIIB on a DNA template containing a 4 nucleotide (nt) 3′ overhanging end, with initiation most efficiently primed by the dinucleotide corresponding to the first 2 nt of the recessed 5′ end. Dinucleotides further into the duplex region fail to prime initiation (Bardeleben et al., 1994). It is conceivable that the increased initiation at bp +1 that is mediated by B′ on the −4/+1 bubble construct (Figure 3A) results from its stabilization of pol III binding to the −4/+1 bubble, enhancing what would otherwise be a transcription factor-independent initiation process. On the other hand, the B′-dependent initiation at bp +1 with constructs −9/−5, −9/−7 and −7/−5 (Figure 3A and B) requires a promoter unwinding downstream of the preformed bubble that pol III appears to be incapable of executing on its own. Constraints on Brf and limitations on initiation The structure of the TFIIIB–DNA complex is redundantly supported by protein–protein (and protein–DNA) interactions: B″ and TBP interact with both the N- and C-proximal halves of Brf, and TBP also interacts with B″ (Khoo et al., 1994; Kassavetis et al., 1997, 1998a; Colbert et al., 1998; Shen et al., 1998); removal of some of these individual interactions does not destroy the transcriptional activity of TFIIIB for supercoiled DNA (Kumar et al., 1997; Kassavetis et al., 1998a). Omitting B″ removes a large part of this structural redundancy, as shown in Figure 4. In the context of a supercoiled SNR6 gene template with full-length B″ and TBP, the N-terminal, TFIIB-related half of Brf (amino acids 1–282) assembles into an unstable TFIIIB–DNA complex that is highly active for transcription (Figure 4A, lane f), while the C-terminal half of Brf (amino acids 284–596) assembles into a stable TFIIIB–DNA complex that retains only very weak transcriptional activity (lane g; barely visible on the original; see Kassavetis et al., 1998a). On linear duplex DNA, TFIIIB complexes containing the individual Brf halves are transcriptionally inactive (lanes b and c), but these Brf fragments complement each other efficiently (lane d; Kassavetis et al., 1998b). In addition, Brf(165–596), lacking its putative N-terminal zinc ribbon and first TFIIB-related repeat, is inactive for transcription of supercoiled DNA only when combined with certain B″ internal deletions (Kassavetis et al., 1997). In contrast, complementation of Brf(1–282) and Brf(284–596) is extremely inefficient for B′-only transcription (Figure 4B). Removal of the N-terminal 68 amino acids containing the putative zinc ribbon of BrfΔ366–407 does not interfere with TFIIIB-directed transcription of the supercoiled SNR6 gene (data not shown), but destroys B′-directed transcription of the −9/ −5 bubble template (Figure 4C). Figure 4.B″-independent transcription with N- and C-terminal truncations of Brf. (A) The N- and C-terminal halves of Brf complement for TFIIIB-dependent transcription of a linear duplex SNR6 gene template. Supercoiled pU6LboxB and a 366 bp linear DNA fragment derived from pU6LboxB were used for transcription directed by TFIIIB complexes containing the Brf variants indicated at the top. The U6LboxB construct generates two divergent transcripts (identified on the left) due to TBP binding to the SNR6 TATA box in either orientation. (B) The N- and C-terminal halves of Brf do not complement for B″-independent transcription of −9/−5 DNA. Primer extension analysis as described for Figure 3 with the Brf components specified above each lane. Initiation is quantified relative to B″-independent transcription with BrfΔ383–424 set to 100 below each lane. (C) B″-independent transcription of −9/−5 DNA requires the N-terminal zinc-ribbon domain of Brf. B′–pol III–DNA complexes were formed with the Brf variant protein specified above each lane. U6 RNA synthesis is analyzed directly and quantified relative to transcription with full-length Brf and B″ (set to 100 and not shown) below each lane. Download figure Download PowerPoint The relatively weak activity of B′-directed transcription has been further dissected in experiments that compare single and multiple rounds of transcription (Table I and Figure 5). B′ recruits pol III to transcription complexes that execute a single round of transcription less effectively and more slowly than does TFIIIB (Figure 5A). The rate difference is due to slower polymerase binding and/or formation of the open complex, since addition of the first 7 nt to an open complex (preformed during 60 min) proceeds at almost the same rapid rate in B′- and TFIIIB-directed transcription (data not shown). The number of rounds of transcription executed under the direction of B′ is also lower than with TFIIIB (Table I), and the relative disadvantage of B′ becomes more pronounced as the time of RNA synthesis increases (Figure 5B). Since the rate of RNA chain elongation has been shown to be unaffected by the presence or absence of promoter-bound TFIIIB (Bardeleben et al., 1994; Matsuzaki et al., 1994) and since termination, as measured by transcript release, is completed within 30 s in the absence of any transcription factor (Bardeleben et al., 1994), the deficit of B′ in directing multiple rounds of transcription must reside in a poorer ability to re-recruit pol III and reopen the promoter for reinitiation of transcription. Figure 5.Comparison of TFIIIB-directed and B′-directed assembly of pol III for multiple and single rounds of transcription. (A) Rate of formation of initiated transcription complexes. TFIIIB and B′ complexes with −9/−5 construct DNA (and using BrfΔ383–424) were formed for 60 min. A mixture of pol III, CTP, GTP and [α-32P]UTP (i.e. lacking ATP) was added for the indicated interval before further addition of ATP and heparin to complete a single round of transcription, and to strip any pol III that had failed to initiate transcription. U6 RNA synthesis is quantified relative to a labeled DNA recovery standard. (B) The B′–DNA complex is relatively inefficient in promoting multiple rounds of transcription. TFIIIB complexes (□) and B′ complexes (●) with −9/−5 DNA (and BrfΔ383–424) were formed for 60 min. Pol III and a labeled nucleotide mixture lacking ATP were added for an additional 15 min to initiate transcription, but block elongation past bp +7. ATP was added to resume elongation and reinitiation for the time interval indicated. ATP and heparin were added to separate samples for 15 min to generate the normalizing single round of transcription. Download figure Download PowerPoint Table 1. Single-round and multiple-round transcription with B′ and TFIIIB Single rounda Multiple rounda Roundsb Brf wt 0.16 ± 0.04 (5) 0.29 ± 0.09 (5) 1.8 (5) Brf(Δ366–407) 0.31 ± 0.07 (4) 0.79 ± 0.21 (4) 2.6 (4) Brf(Δ383–424) 0.40 ± 0.02 (4) 1.10 ± 0.07 (3) 2.8 (3) B″ + Brf wt 1 (5) 5.25 ± 1.34 (5) 5.3 (5) B″ + Brf(Δ366–407) 1.29 ± 0.11 (4) 6.88 ± 0.73 (3) 5.3 (3) B″ + Brf(Δ383–424) 1.29 ± 0.17 (4) 7.06 ± 1.94 (4) 5.5 (4) a Average of three to five experiments (as indicated in parentheses) normalized to single-round transcription with wild-type TFIIIB; the standard error of the mean is also specified. b After 15 min of transcription. The B″-less pol III promoter complex The physical basis of the limited ability of B′ to direct transcription of the −9/−5 bubble construct is explored in footprinting experiments that are summarized in Figures 6 and 7. Open complex formation was probed with KMnO4, which oxidizes thymine in single-stranded DNA, but not in duplex DNA (Hayatsu and Ukita, 1967; Sasse-Dwight and Gralla, 1989). Although the non-transcribed strand of the −9/−5 construct contains two T residues within its 5 bp bubble (at bp −8 and −7; Figure 1), 50%) is not fully open. The placement of pol III on B′–DNA and TFIIIB–DNA complexes was compared by two-dimensional DNase I footprinting (Figure 7), in which individual DNase I-cleaved protein–DNA complexes are separated by native gel electrophoresis prior to analysis. Pol III–B′ complexes of the −9/−5 bubble construct were not sufficiently stable to be isolated from native gels. Consistent with the notion that this instability might be contributed by incomplete promoter opening (Figure 6), we found that GTP + UTP improved the retention of B′–pol III–DNA complexes during gel electrophoresis ∼2.5-fold relative to pol III–DNA complexes. Because of its slightly lower mobility, conservative excision of bands from the gel minimized potential cross-contamination of the B′–pol III–DNA complex with pol III–DNA complexes (data not shown). The presence of GTP and UTP permits the formation of a trimer transcript that is synthesized reiteratively, along with dimers, as abortive initiation products (Bhargava and Kassavetis, 1999). The accuracy of pol III recruitment by Brf and TBP to the −9/−5 bubble template was examined by two-dimensional DNase footprinting. TFIIIB protects the −9/ −5 construct DNA between ∼bp −38 and −8, and enhances DNase I cleavage around the start site of transcription (Figure 7, top panel). Addition of pol III extends protection to bp +22, with enhanced DNase I cleavage at bp +28 and +30. The high degree of protection in this pol III footprint is unexpected, since TFIIIB does not bind to the SNR6 TATA box in a unique orientation, and places pol III either upstream or downstream of the TATA box according to its orientation (Whitehall et al., 1995). It appears, therefore, that the −9/−5 bubble further specifies a unique (or predominant) orientation for the TFIIIB–pol III–DNA complex. DNase I footprinting detects B′ protection of −9/−5 construct DNA between bp −32 and −18 (Figure 7, lower panel). Pol III binding to the B′–DNA and TFIIIB–DNA complexes generates nearly identical footprint additions: for the B′–pol III–DNA complex, protection extends to bp +22 with enhanced cleavage at bp +28. Pol III also generates protection of this complex between bp −6 and −12. Since this DNA segment is protected by the TFIIIB–DNA complex, this footprint difference cannot be interpreted as a difference in pol III placement. Indeed, on the basis of permanganate and DNase I footprinting, we conclude that the placements of pol III over the −9/−5 bubble promoter by B′–DNA and TFIIIB–DNA complexes are indistinguishable. The low level of complex formation of pol III with the TBP–DNA complex hampers an examination of pol III alone bound to the −9/−5 bubble. (Even the footprints of gel-isolated TBP–pol III–DNA complexes are less distinct and complete, possibly due to contamination with polymerase bound at DNA ends.) We have generally observed weak protection at bp −8 and between bp +1 and +17 (Figure 7, lower panel, inset), indicating that the −9/−5 bubble does not by itself 'correctly' place pol III over the SNR6 promoter. Discussion These experiments firmly establish a role for TFIIIB in the initiation of transcription by pol III that extends beyond polymerase recruitment. We show that the absolute requirement for the B″ subunit of TFIIIB in pol III transcription can be circumvented by opening a small part of the transcription bubble (Figure 2). This B″-less transcription requires Brf, specifically directs pol III to the normal start site of the SNR6 gene (Figure 7) and demands specific placement of a DNA loop (Figure 3A). Thus, the TATA box-bound B′ (i.e. TBP + Brf) complex imposes a rigid geometry on transcriptional initiation. Some residual B′-dependent transcription is even retained when the DNA loop is reduced in size from 5 to 3 bp (Figure 3B). Five-base-pair bubbles elicit a weak background of Brf-independent transcription, primarily initiating within the bubble, that varies considerably between different bubble constructs (Figure 3A, right-hand panel). This is probably due to sequence-dependent differences in internal structure within different 5 bp l
Referência(s)