Facilitated Recycling Pathway for RNA Polymerase III

Cell, Vol. 84, 245–252, January 26, 1996, Copyright 1996 by Cell Press

Facilitated Recycling Pathway for RNA Polymerase III Giorgio Dieci* and Andre´ Sentenac Service de Biochimie et Ge´ ne´tique Mole´culaire Commissariat a` l’Energie Atomique–Saclay F-91191 Gif-sur-Yvette Cedex France

Summary We show that the high in vitro transcription efficiency of yeast RNA pol III is mainly due to rapid recycling. Kinetic analysis shows that RNA polymerase recycling on preassembled tDNA•TFIIIC•TFIIIB complexes is much faster than the initial transcription cycle. High efficiency of RNA pol III recycling is favored at high UTP concentrations and requires termination at the natural termination signal. Runoff transcription does not allow efficient recycling. The reinitiation process shows increased resistance to heparin as compared with the primary initiation cycle, as if RNA polymerase was not released after termination. Indeed, template competition assays show that RNA pol III is committed to reinitiate on the same gene. A model is proposed where the polymerase molecule is directly transferred from the termination site to the promoter. Introduction The central event in the activation of class III genes is the assembly of stable preinitiation complexes containing transcription factors IIIC and IIIB (TFIIIC and TFIIIB). TFIIIC specifically binds to tRNA genes by recognizing their intragenic promoter elements. Bound TFIIIC then specifically recruits TFIIIB on the 59-flanking region of class III genes, and TFIIIB, in turn, recruits RNA polymerase III (pol III) (Geiduschek and Kassavetis, 1992). In yeast, TFIIIB is constituted of TATA box–binding protein (TBP) and at least two other components of 70 and 90 kDa (Kassavetis et al., 1992a). Yeast TFIIIB does not appear to be a stable molecular entity (Huet et al., 1994), and in vitro, its recruitment on class III genes follows a multistep pathway involving the sequential addition of protein components to the template (Kassavetis et al., 1992a). Once established, the TFIIIB–DNA interaction is highly stable in vitro under conditions that dissociate TFIIIC (Kassavetis et al., 1990). TFIIIB is alone responsible for the recruitment of pol III at the transcription start site. Even after stripping off TFIIIC, it can direct multiple rounds of transcription by pol III (Kassavetis et al., 1990). Therefore, after the slow step of preinitiation complex assembly, a class III gene remains stably committed to transcription during the following reinitiation events, in which pol III is the only recycling factor. The situation seems to be different in the case of RNA pol II. All the components of a pol II preinitiation complex dissociate during the initiation-to-elongation transition, with the *Permanent address: Istituto di Scienze Biochimiche, Universita` di Parma, Viale delle Scienze, I-43100 Parma, Italy.

exception of TFIID, and a new complex must be reassembled for each new cycle (Zawel et al., 1995). TFIID remains promoter-bound through the transcription cycle and therefore facilitates reinitiation on the same template (Hawley and Roeder, 1987; Jiang and Gralla, 1993). However, the release and reassociation of the other factors make the reinitiation process more complex than in the case of pol III and thus increase the number of critical checkpoints for transcriptional modulation. Some activators of pol II transcription have indeed been shown to be required at each new cycle (Arnosti et al., 1993; Roberts et al., 1995). Reinitiation on class III genes, which bypasses almost all the steps required for the initial transcription cycle, does not allow such a fine regulation, but potentially leads to more efficient RNA production. Pol III-transcribed genes are small and can efficiently direct multiple rounds of transcription in vitro (Wolffe et al., 1986; Kassavetis et al., 1989). This efficiency could be accounted for, at least in part, by the extreme stability of the TFIIIB•DNA complex, which does not dissociate after transcription initiation (Kassavetis et al., 1990). However, previous kinetic analyses of open complex formation (Kassavetis et al., 1992b) and overall initiation (Dieci et al., 1995a) by yeast RNA pol III on preformed preinitiation complexes showed that these processes are relatively slow. It was difficult to imagine how iteration of slow processes could account for a high pol III reinitiation frequency. In the present work, we have explored the process of reinitiation to solve this intriguing paradox. Our analysis revealed an unexpected property of the class III transcription machinery that is responsible for its high reinitiation efficiency. After the first initiation event on a given preinitiation complex, RNA pol III becomes committed to more rapidly transcribing the same gene in a way that is termination dependent. The general implications of such a mechanism for the control of gene expression are discussed. Results Kinetics of Initiation and Reinitiation by Yeast RNA Pol III Free RNA pol III needs several minutes to complete open complex formation on a preformed tDNA•TFIIIC•TFIIIB preinitiation complex (Kassavetis et al., 1992b), and we recently observed a similar time course for the initiation of RNA synthesis (Dieci et al., 1995a). The time course of initiation by yeast pol III on a preinitiation complex is shown in Figure 1A. Stable preinitiation complexes were formed by preincubation of SUP4 tDNA with TFIIIC and TFIIIB components; then RNA pol III was added together with ATP, CTP, and 32 P-labeled UTP. In the absence of GTP, a halted stable ternary complex is formed, containing a 17 nt nascent RNA (Kassavetis et al., 1989). Under the experimental conditions used, RNA chain initiation, as monitored by the formation of the 17-mer product, was completed in 5 min, with a t1⁄2 of about 2 min.

Cell 246

Figure 1. Transcription Initiation and Reinitiation on the SUP4 tRNA Gene (A) Overall time course of initiation by pol III. A stable preinitiation complex was formed on the SUP4 tRNA gene as described in Experimental Procedures. Pol III (50 ng) was then added together with ATP (500 mM), CTP (500 mM) and [a-32P]UTP (10 mM; 40 mCi/nmol). At the indicated times, samples were taken from the mixture and transferred to tubes chilled in dry ice to stop the reaction. The reaction products were separated on a 15% polyacrylamide, 7 M urea gel. The position of the 17 nt transcript is indicated. The upper part of the figure shows a scheme of the incubation protocol. (B) Time course of pol III recycling. A stable preinitiation complex was formed on the SUP4 tRNA gene as described in Experimental Procedures. Pol III (50 ng) was then added together with ATP (500 mM), CTP (500 mM), and [a-32P]UTP (100 mM; 4 mCi/nmol), and the incubation was continued for 10 min. Transcription was restarted by the addition of GTP (500 mM) in the presence (lanes 1 and 2) or absence (lanes 3–5) of heparin (0.3 mg/ml). At the indicated times, samples were taken from the mixtures and transferred to tubes chilled in dry ice to stop the reaction. The products were separated on a 6% polyacrylamide, 7 M urea gel. The position of the tRNA gene primary transcript (SUP4 tRNA) is indicated. S, single-round transcription; M, multiple-round transcription. (C) UTP dependence of recycling. Transcription reactions were as described for (A), with the exception that UTP concentration was varied as indicated. To keep the specific radioactivity constant, reactions carried out at 20 mM, 100 mM, and 500 mM UTP contained, respectively, 2, 10, and 50 mCi of [a- 32P]UTP. After the addition of GTP only (lanes 2, 4 and 6; M, multiple-round transcription) or GTP plus heparin (lanes 1, 3 and 5; S, single-round transcription), the incubation was continued for 3 min. Products in lanes 2, 4, and 6 corresponded to 2.5, 5.8, and 7.0 rounds of transcription, respectively.

The time course of the first initiation event contrasted with the average duration of one transcription cycle under reinitiation conditions, as determined by the experiment shown in Figure 1B. The average time for one transcription cycle by recycling RNA pol III was estimated by comparing the amount of transcript produced

in single and multiple round transcription reactions, as described by Kassavetis et al. (1989). Note that this one and all subsequent transcription assays were carried out at limiting pol III concentration, to be sure that reinitiation was performed by the same pol III molecules that first initiated (see Experimental Procedures). Stable preinitiation complexes (tDNA•TFIIIC•TFIIIB) were formed; then limiting pol III was added together with ATP, CTP, and UTP, to form ternary complexes containing a 17 nt nascent RNA. These complexes were completely resistant to 0.3 mg/ml heparin (which abolishes reinitiation), as shown by the fact that the 17 nt RNA could be quantitatively chased to full-length RNA in the presence of this heparin concentration (Kassavetis et al., 1989; Dieci et al., 1995a; data not shown). RNA synthesis was allowed to resume by the addition of GTP, in the presence of heparin (single-round transcription) or in its absence (multiple-round transcription). The multiple-round versus single-round transcript ratio corresponded to 4.5, 7, and 9 new rounds of transcription after, repectively, 2, 4, and 6 min. This corresponded to an average of 35 s per transcription cycle, a value in good agreement with previous estimations (Kassavetis et al., 1989). Such a high efficiency of reinitiation was in marked contrast with the relatively slow rate of the primary initiation process. On stable preinitiation complexes, recycling pol III completed each new cycle 5- to 10-fold more rapidly than the first round of transcription. The efficiency of reinitiation was generally found to be influenced by concentration of nucleoside triphosphates (NTPs). Varying concentration of UTP had the most dramatic effect. As shown in Figure 1C, lowering UTP concentration from 100 mM to 20 mM did not change, as expected, the output of a single-round transcription reaction (compare lanes 3 and 1), nor did it alter the kinetics of initiation by free RNA pol III (data not shown), but it did decrease the efficiency of reinitiation by a factor of 2 to 3 (compare lanes 4 and 2). Initiation and Reinitiation by Pol III Have Different Heparin Sensitivities To try to differentiate the processes of initiation and reinitiation, we explored their sensitivity to heparin treatment (Figure 2). When free RNA pol III was added to preinitiation complexes together with heparin, ATP, CTP, and UTP, heparin almost totally inhibited 17-mer synthesis at a concentration of 2 mg/ml, and half-inhibitory concentration was 0.25–0.5 mg/ml (Figure 2A). On the other hand, when RNA pol III is engaged in transcription elongation, it becomes resistant to high heparin concentrations. Thus, a concentration of 300 mg/ml is normally used to allow single-round transcription (Kassavetis et al., 1989). To investigate the heparin sensitivity of recycling RNA polymerase molecules, ternary complexes were preformed (containing a 17-mer RNA); then, elongation and recycling (5 min) were allowed to proceed in the presence of varying heparin concentrations (Figure 2B). Under these conditions, 50% inhibition was obtained at 5 mg/ml, and a concentration of 100 mg/ml was required to prevent reinitiation completely. The residual transcription activity at 100–500 mg/ml corresponded, as expected, to heparin-resistant, singleround transcription. Initiating pol III is thus at least 10-fold more sensitive to heparin than recycling RNA polymerase molecules.

Facilitated RNA Polymerase Recycling 247

Figure 3. Gene Commitment by Recycling RNA Pol III

Figure 2. Heparin Inhibition of Initiation and Reinitiation (A) Heparin sensitivity of initiation. A stable preinitiation complex was formed on the SUP4 tRNA gene as described in Experimental Procedures. Pol III (50 ng) was then added together with ATP (500 mM), CTP (500 mM), [a-32P]UTP (10 mM; 40 mCi/nmol), and variable concentrations of heparin, as indicated. Synthesis of the 17-mer was allowed to proceed for 15 min. The products were separated on a 15% polyacrylamide, 7 M urea gel. The position of the 17 nt transcript is indicated. (B) Heparin sensitivity of reinitiation. A stable preinitiation complex was formed on the SUP4 tRNA gene as described in Experimental Procedures. Pol III (50 ng) was then added together with ATP (500 mM), CTP (500 mM) and [a-32P]UTP (100 mM; 4 mCi/nmol) and the incubation continued for 10 min. GTP was then added, together with variable concentrations of heparin, as indicated. The reactions were stopped after 5 min (equivalent to about eight rounds of transcription in the absence of heparin) and the products separated on a 6% polyacrylamide, 7 M urea gel. The position of the tRNA gene primary transcript is indicated.

Gene Commitment of Recycling RNA Pol III The above results suggested that, once RNA pol III has initiated, it may rapidly reinitiate on the same template, possibly without being released at each cycle. If this were the case, a recycling pol III would be committed to transcribe the same gene. To explore a possible gene commitment, recycling pol III was given the choice between two genes (Figure 3A). The tRNALeu3 and SUP4 tRNATyr genes were chosen, because their primary transcription products have different electrophoretic mobilities. In addition, stable ternary complexes can be formed on both genes in the presence of the same subset of nucleotides, ATP, CTP, and UTP (Kassavetis et al., 1989). Pol III efficiently recycled on both genes (Figure 3A; compare lanes 1 and 2 with lanes 3 and 4, respectively). The reinitiation frequency (i.e., the multiple-round versus single-round transcript ratio) on tDNALeu3 (6 cycles/5 min) was slightly lower than on tDNA SUP4 (8 cycles/5 min). When a limiting amount of pol III was added to a mixture of the two genes (each being preassembled into preinitiation complexes) and then multiple

(A) The tRNALeu3 and SUP4 tRNA genes were separately preincubated with TFIIIC and reconstituted TFIIIB, to allow for the formation of stable preinitiation complexes. Pol III (50 ng) was then added to one of the two complexes (template 1) together with ATP (500 mM), CTP (500 mM), UTP (100 mM), and [a-32P]UTP. For the reaction in lane 5, pol III (50 ng) was added to a mixture of the two preinitiation complexes. The incubation was continued for 10 min, and then GTP (500 mM) was added alone (lanes 3–5), together with heparin (lanes 1 and 2), or together with the competitor template preassembled in a stable preinitiation complex (template 2; lanes 6 and 7). Transcription was allowed to proceed for 5 min, and then reactions were blocked and the products separated on a 6% polyacrylamide, 7 M urea gel. The position of the two primary transcripts is indicated. Reactions in lanes 1–4 contained 20 mCi of [a-32P]UTP in a final volume of 50 ml. Reactions in lanes 5–7 contained 40 mCi of [a-32P]UTP in a final volume of 100 ml. S, single-round transcription; M, multiple-round transcription. The ratios between SUP4 and Leu3 transcription were 1.7, 7.0, and 0.3 in lanes 5, 6, and 7, respectively. (B) Lanes 1 and 3, single-round transcription reactions done as in lanes 1 and 2 of (A), respectively. Lanes 2 and 4, SUP4 tDNA or tDNALeu 3 were preincubated with TFIIIC and TFIIIB to form stable preinitiation complexes, then added to a mixture containing pol III (50 ng) and all four NTPs, at the same concentrations as in (A), and multiple rounds of transcription were allowed for 5 min. The ratios between the signals in lanes 2 and 4 and the signals of corresponding single-round transcription reactions were 5.6 and 5.0 for the SUP4 and Leu3 genes, respectively.

rounds of transcription were allowed, pol III equally transcribed the two templates (lane 5), indicating that transcription of neither gene was intrinsically favored. In lanes 6 and 7, pol III was first sequestered in ternary complexes on template 1 (either SUP4 or Leu3 gene), and then multiple rounds of transcription were allowed to proceed for 5 min in the presence of an equimolar amount of template 2 preassembled into preinitiation complex with TFIIIC and TFIIIB. In both cases, pol III did not redistribute itself equally on both genes, as would be expected if it had been released after the first cycle. Instead, pol III preferentially recycled on template 1, while transcription of the competitor template 2 was significantly excluded. Figure 3B additionally shows that when template 1 was omitted, template 2 was much more actively transcribed, during 5 min of incubation, than when pol III had initiated on template 1 prior to template 2 addition (as in lanes 6 and 7 of Figure 3A).

Cell 248

Figure 5. Time Course of Initiation on a Previously Transcribed Template

Figure 4. Time Course of 17-mer Synthesis by a Complete Initiation Complex The experiment was as in Figure 1A, but pol III alone was preincubated 10 min with the preinitiation complex before the addition of ATP, CTP, and UTP. A scheme of the incubation protocol is shown at the top.

On the basis of all these data, we conclude that recycling RNA pol III is predominantly committed to the first transcribed gene. Mechanisms of Facilitated Reinitiation by RNA Pol III Reinitiation could proceed faster if one or more ratelimiting steps of initiation are performed at a higher rate in the following rounds. The kinetics of initiation shown in Figure 1A are determined by the rate of three successive steps: recruitment of pol III, DNA melting, and initiation of RNA synthesis. The contribution of the latter step, i.e., the time required for the formation of the 17-mer RNA, was determined by the experiment shown in Figure 4. RNA pol III was allowed to associate with tDNA• TFIIIC•TFIIIB complexes for 10 min at 228C before the addition of ATP, CTP, and UTP. Under these conditions, 17-mer synthesis was completed within 10 s (Figure 4). Initiation by free RNA pol III thus involves a slow association or DNA-opening step, followed by rapid RNA chain initiation. The uncoupling of initiation and reinitiation rates is most likely due to the ability of recycling pol III to carry out this slow step at a higher rate. The use of limiting amounts of RNA polymerase made unlikely the possibility that fast reinitiation was simply due to a rapid recruitment of excess, DNA-bound enzyme molecules. This was confirmed by the observation that facilitated reinitiation also occurred on very short DNA templates such as a 270 bp DNA fragment harboring the SUP4 tRNA gene (data not shown). Facilitated reinitiation could proceed through a modification of preinitiation complexes, occurring during the first transcription cycle, that would increase the rate of subsequent reinitiation cycles. This point was addressed by measuring the time course of de novo RNA chain initiation by free pol III on a template that had been through one transcription cycle. Preinitiation complexes were formed, subjected to single-round transcription in the presence of heparin, isolated by gel filtration to remove RNA polymerase, and reused for RNA chain initiation. As a control, preinitiation complexes were

A stable preinitiation complex was formed on the SUP4 tRNA gene as described in Experimental Procedures. The incubation mixture was split into two parts. One part received pol III (50 ng) together with ATP (500 mM), CTP (500 mM), and UTP (100 mM); after a 10 min incubation, GTP (500 mM) was added together with heparin (0.25 mg/ml), to generate preinitiation complexes transcribed once (Kassavetis et al., 1990). The other half of the mixture, containing nevertranscribed complexes, received only heparin. Each of the two mixtures (110 ml) was loaded onto a 1 ml Sepharose CL-2B gel filtration column. Fractions 9–11 from each column, containing stable preinitiation complexes but not free RNA polymerase (data not shown), were collected and pooled. Pol III (50 ng) was then added to each pool, together with ATP (500 mM), CTP (500 mM), and [a-32P]UTP (10 mM; 40 mCi/nmol), and samples of the reactions were blocked at the indicated times. The reaction products were separated on a 15% polyacrylamide, 7 M urea gel. The position of the 17 nt transcript is indicated.

treated with heparin and gel-filtrated in the absence of RNA polymerase. As shown in Figure 5, initiation on previously transcribed preinitiation complexes proceeded with a t1⁄2 comprising between 1 and 2 min. After 2 min, about 70% of the complexes had initiated in this case (Figure 5, bottom), and 50% in the case of preinitiation complexes that had not been previously transcribed (Figure 5, top). This slight increase in initiation rate on previously transcribed complexes, which was reproducibly observed in several experiments, could not account by itself for the fast reinitiation rate observed in Figures 2B and 2C. We have not explored further the basis for this small increase in initiation efficiency. It should be noted, however, that transient modifications of the preinitiation complex would likely be lost during heparin treatment followed by gel filtration. Thus, we can only conclude that facilitated recycling does not involve long-lived and stable modifications of the preinitiation complex. The Pol III Terminator Sequence Is Required for Efficient Reinitiation Since termination and RNA release immediately precede reinitiation, the modality of termination by pol III could influence reinitiation efficiency. In class III genes, the termination signal is a run of T residues in the nontranscribed strand of DNA (Geiduschek and Tocchini-Valentini, 1988). To examine the role of this signal in facilitated reinitiation, we compared the recycling efficiency on a linear template with or without a functional terminator. We chose for this purpose the truncated form of the yeast U6 gene lacking the extragenic B box (Brow and Guthrie, 1988). Transcription of this gene in a purified system does not require TFIIIC and is only dependent on

Facilitated RNA Polymerase Recycling 249

Figure 6. Requirement of the Terminator Element for Efficient Reinitiation on the U6 Gene (A) Decreased reinitiation frequency after runoff termination. TFIIIB was assembled on the truncated version of the yeast U6 gene (contained in pTaq6 plasmid) as described in Experimental Procedures. Reactions in lanes 1 and 2 contained the uncut, circular plasmid. For reactions in lanes 3 and 4, the plasmid was linearized at the polylinker SmaI site, downstream of the U6 terminator sequence. For reactions in lanes 5 and 6, the plasmid was linearized at the EcoNI site in the coding region, 30 bp upstream of the terminator (Moenne et al., 1990). Pol III (50 ng) was added together with CTP (500 mM), GTP (500 mM), and [a-32P]UTP (200 mM; 2 mCi/nmol), and the incubation was continued for 15 min to allow for the formation of a ternary transcription complex containing a 7 nt nascent transcript (Brow and Guthrie, 1988). ATP was then added in the presence (lanes 1, 3, and 5) or the absence (lanes 2, 4, and 6) of heparin to inhibit reinitiation, and the incubation was continued for 6 min. Reaction products were separated on a 6% polyacrylamide, 7 M urea gel. The positions of the complete and runoff U6 RNA gene products are indicated. S, single-round transcription; M, multipleround transcription. (B) The terminator sequence does not influence the affinity of the preinitiation complex for RNA pol III. TFIIIB was separately assembled on SmaI- and EcoNI-linearized pTaq6, and then pol III was added to a mixture of the two templates, together with CTP (500 mM), GTP (500 mM), and [a-32P]UTP (200 mM; 2 mCi/nmol). After 15 min, ATP (500 mM) was added with (lane 1) or without (lane 2) heparin, and transcription was allowed to proceed for 6 min in a final volume of 100 ml. Reaction products were separated on a 6% polyacrylamide, 7 M urea gel. The positions of the complete and runoff U6 RNA gene products are indicated. S, single-round transcription; M, multiple-round transcription.

the upstream assembly of TFIIIB (Moenne et al., 1990). It is thus possible to obtain runoff transcription from the U6 promoter by linearizing the template at the EcoNI site located in the coding region, 30 bp upstream of the terminator (Moenne et al., 1990). The pTaq6 plasmid was linearized at the unique EcoNI (runoff transcription) or SmaI (normally terminated transcription) sites, and the recycling ability of pol III on these templates was tested by comparing single- and multiple-round transcription reactions (Figure 6). Linearization downstream of the terminator region did not affect the reinitiation efficiency compared with that of a circular template: during 6 min of recycling, about six cycles of transcription were performed by pol III on both templates (Figure 6A; compare lanes 3 and 4 with lanes 1 and 2). In contrast, linearization upstream of the terminator strongly reduced transcription reinitiation (compare lanes 5 and 6): fewer than two cycles were performed on this template during the same period of time. The signal corresponding to a single round of runoff transcription (lane

Figure 7. Model for the Pathways of Transcription Reinitiation by RNA Pol III For simplicity, TFIIIC has been omitted from the scheme. Steps 1 and 2 summarize a simple transcription cycle, from initiation by free pol III to termination. Step 1 can be inhibited by 2 mg/ml heparin. Concomitant with the termination step, two alternative pathways can be followed by pol III: it can either dissociate (step 3b) and then slowly reinitiate a new cycle from step 1, or undergo a terminationcoupled, fast reinitiation event without being released from the template (steps 3a and 4). The latter pathway can take place in the presence of 2 mg/ml heparin and can be inhibited by 50–100 mg/ml heparin. A single arrow in the figure does not imply that there is only one biochemical step involved. For details, see Discussion.

5) was about half the signal produced by a single round of complete transcription (lane 3), but this was expected, because the runoff transcript contains half as many Us as the complete U6 transcript. When pol III was incubated with a mixture of the two linear templates (each preassembled with TFIIIB), the same number of singleround transcripts was produced from the two templates, indicating that limiting pol III had been equally recruited by the two genes (Figure 6B, lane 1). The terminator element has thus no influence on the affinity of the preinitiation complex for pol III. However, RNA polymerase molecules reinitiated much more efficiently on the terminator-containing template than on the terminator-deficient gene (lane 2). In experiments not shown, a circular pTaq6 template preassembled with TFIIIB was added to a reaction in which pol III had performed one cycle of runoff transcription. These pol III molecules could initiate and reinitiate transcription on pTaq6 template with the same efficiency as freshly added pol III. This result indicated that the reduced recycling ability on the U6 runoff template was not due to the artifactual inactivation of pol III by runoff transcription. Discussion The analysis reported here reveals a new mechanism allowing a high level of RNA polymerase recycling after a single gene activation event. We show that pol III preferentially recycles on the same template, in a way that allows it to complete new cycles more rapidly than the initial one. This conclusion is summarized in the model presented in Figure 7. In the first step of initiation (step 1 in Figure 7), free

Cell 250

RNA pol III is recruited by a stable tDNA•TFIIIC•TFIIIB preinitiation complex to form a closed complex that, at room temperature, easily isomerizes to an open complex (Kassavetis et al., 1992b). This overall slow process of RNA polymerase recruitment and open complex formation is very sensitive to heparin. When pol III has been allowed to associate with the preinitiation complex in the absence of NTPs, the resulting complex is able to initiate RNA synthesis very rapidly, in less than 10 s (step 2 in Figure 7), and becomes resistant to high heparin concentrations (Kassavetis et al., 1989). After the elongation phase, which is very rapid on the short tRNA genes (Matsuzaki et al., 1994), a pol III molecule must undergo the termination process, which includes, as potentially separate steps, the release of the terminated RNA chain and the release of RNA polymerase itself from the template (Chamberlin, 1974). These steps have the potential to influence strongly the ability of polymerase to reinitiate. The common view of reinitiation is that RNA polymerase needs to be released in order to be recruited by preinitiation complexes to start a new transcription cycle (step 3b in Figure 7). We propose that, in the case of pol III, a preferential termination pathway allows RNA release and transcription reinitiation to be performed without release of RNA polymerase (steps 3a and 4 in Figure 7). This pathway leads to an increased initiation rate and to a distinctive heparin resistance. The marked dependence of the reinitiation rate on NTP and, in particular, UTP concentration (Figure 1C) can be easily explained in the context of this pathway: the initiation phase being no longer rate-limiting, the overall transcription process can be significantly stimulated by the UTP-induced increase in the rates of elongation and termination. The terminator element appears to be required for pol III to enter the facilitated initiation pathway, since runoff termination leads to slow reinitiation (Figure 6), probably owing to spontaneous pol III dissociation (as in step 3b of Figure 7). The existence of a pathway avoiding dispersion of recycling pol III into the free RNA polymerase pool is supported by the finding that, during multiple rounds of initiation, pol III is committed to the gene on which it has first initiated (Figure 3). By template exclusion assays and glycerol gradient sedimentation of transcription complexes, Jahn et al. (1987) previously provided some evidence that human pol III is retained in the original transcription complex during the normal reinitiation cycle. Our data are in agreement with those of Jahn et al., even if, in this case, the extent of reinitiation was not evaluated, but they are at variance with the conclusions of Bieker et al. (1985), who suggested that human pol III equilibrates during multiple rounds of transcription on different 5S RNA templates. In the template exclusion assays of Bieker et al., however, transcription was allowed to proceed for long periods of time. At each cycle, a small percentage of pol III molecules can be released from the first gene (as in step 3b of Figure 7) and can therefore initiate transcription on the second template. Increasing the number of cycles will eventually lead to a complete redistribution of pol III. It should be noted that, under our short incubation conditions, pol III commitment was not absolute (see Figure 3).

The increased heparin resistance of recycling pol III is probably not restricted to the case of the yeast pol III system. In their study of the effects of sarkosyl on human pol III transcription, Kovelman and Roeder (1990) found a sarkosyl concentration (0.015%) that inhibited a pol III–dependent step in the first round of transcription of the VA1 RNA gene, but that did not inhibit reinitiation. It is thus tempting to speculate that reinitiation by human pol III also proceeds through the facilitated pathway. Facilitated reinitiation by pol III could involve a specific contact between terminating pol III and a component of the preinitiation complex. As represented in the model in Figure 7, this contact could involve DNA looping, facilitated by the short size of the transcribed region, and may require a conformational change of the pol III molecule. It has been recently shown that the oligo(dT) tract of Escherichia coli transcription terminators can induce a conformational change in E. coli RNA polymerase by acting as an inchworming signal (Nudler et al., 1995). The pol III terminator element could be required to give pol III the proper conformation for the interactions leading to facilitated reinitiation. Campbell and Setzer (1992) have shown that termination signal recognition by pol III can be experimentally uncoupled from polymerase and transcript release. Pausing of pol III at the termination signal before release would give time for the facilitated reinitiation event to take place; on the other hand, runoff termination would directly lead to the release of elongating polymerase molecules. Some reports have indicated that the RNA-binding protein La stimulates human pol III transcription by facilitating multiple rounds of transcription initiation (Gottlieb and Steitz, 1989a, 1989b; Maraia et al., 1994). Even if efficient class III gene transcription has been reconstituted from homogeneous or highly purified components (Kassavetis et al., 1992a; Joazeiro et al., 1994), the possibility cannot be excluded that a yeast La homolog (Yoo and Wolin, 1994; Lin-Marq and Clarkson, 1995), or another factor, might play a role in the facilitated reinitiation pathway. There are noticeable differences in the sensitivity of pol II and pol III transcription systems to inhibitors such as heparin or sarkosyl, which most likely reflect differences in reinitiation mechanisms. In the case of pol II transcription, reinitiation is inhibited by the same sarkosyl concentration (0.025%) that inhibits initiation complex assembly (Hawley and Roeder, 1987). This indicates that, in contrast with what was observed with pol III, the sarkosyl-sensitive step in pol II initiation has to be repeated at each reinitiation cycle. Zawel et al. (1995) recently showed that, following the transition from initiation to elongation, only TFIID remains promoter-bound, whereas TFIIB, TFIIE, TFIIF, and TFIIH are released and must reassociate at each cycle. The need for a partial iteration of preinitiation complex assembly could explain the sarkosyl sensitivity of pol II reinitiation. On the other hand, the fact that TFIID is left behind on the template after the first round of initiation can explain the uncoupling of the rates of initiation and reinitiation that has been observed for GAL–AH-dependent pol II in vitro transcription (Jiang and Gralla, 1993). In the case of pol III, the reinitiation process seems

Facilitated RNA Polymerase Recycling 251

to have gained its maximal efficiency: not only is a functional preinitiation complex constantly left at the promoter, but, in addition, the polymerase can initiate new cycles more rapidly than the first one without being completely released. This optimization of transcript production after a single gene activation event is probably essential to satisfy the high cellular needs for class III gene products. However, a facilitated reinitiation pathway might also exist, in variant forms, in the other polymerase systems. Since rDNA transcription by RNA pol I must also be highly efficient, it would not be surprising if the pol I transcription system had features allowing facilitated reinitiation to take place. Indeed, the tandem arrangement of eukaryotic rRNA genes has led to the hypothesis that a specialized mechanism may exist for the direct transfer of pol I from the terminator to the adjacent promoter without dissociation (Baker and Platt, 1986; Johnson and Warner, 1989). Mechanisms allowing RNA pol II to terminate and reinitiate without being dispersed into the free enzyme pool are also possibly used in situations where a maximal level of transcription is required. Experimental Procedures Purification of RNA Pol III and Class III Transcription Factors Purification of transcription proteins was carried out following previously described procedures. Pure RNA pol III (a gift of J. Huet) was prepared as described by Huet et al. (1985). Affinity-purified TFIIIC was prepared as described by Gabrielsen et al. (1989). TFIIIB activity was reconstituted from the three subcomponents TBP, TFIIIB70, and fraction B99. Yeast recombinant TBP, expressed in E. coli from plasmid pET3b(rTFIIDY) (a gift from J.-M. Egly), was purified by the procedure of Burton et al. (1991). Recombinant histidinetagged TFIIIB70 was expressed in E. coli from plasmid pSH360 (a gift from S. Hahn; Colbert and Hahn, 1992) and purified by chromatography on Ni2 1–NTA–agarose (Qiagen) under native conditions, as described (Dieci et al., 1995b). Fraction B99 was extracted from chromatin pellets and partially purified on BioRex 70 (Bio-Rad) according to the protocols of Kassavetis et al. (1992a). Specific In Vitro Transcription Assays Plasmid DNAs used for in vitro transcription reactions were the following: pRS316-SUP4, containing the yeast SUP4 tRNA gene (S. Shaaban, personal communication); pGE2-WT, containing the yeast tRNALeu3 gene (Baker and Hall, 1984); and pTaq6, containing a truncated form of the yeast U6 RNA gene lacking the extragenic B box (Brow and Guthrie, 1988). Transcription reactions were carried out in a final volume of 50 ml and contained 20 mM Tris–HCl (pH 7.9), 90 mM KCl, 5 mM MgCl2, 10% glycerol (v/v), 0.1 mM EDTA, 0.5 mM each of ATP, CTP, and GTP, variable UTP and [a-32P]UTP (Amersham) concentrations, 250 ng of plasmid DNA, 50 ng of affinitypurified TFIIIC, 40 ng of rTBP, 50 ng of rTFIIIB70, 0.4 mg of B99 fraction, and 50 ng of RNA pol III. Reactions programmed with the U6 gene did not contain TFIIIC. In all transcription reactions, RNA pol III was limiting. The limiting pol III concentration was chosen on the basis of titration experiments in which pol III concentration was varied under single-round transcription conditions. In these assays, a plateau was reached in the presence of 100 ng of pol III, and 40 ng of pol III was half saturating. In all transcription assays, the template was first preincubated (15 min at 228C) with TFIIIC and the TFIIIB subcomponents (or with TFIIIB only, in the case of the U6 RNA gene) in a final volume of 42 ml, to allow for the formation of stable preinitiation complexes. Pol III and subsets of NTPs were then added, to form heparin-resistant ternary complexes containing a nascent RNA chain. Transcription was eventually allowed to resume by the addition of the missing NTP under single- or multipleround transcription conditions (Kassavetis et al., 1989). Heparin for

single-round transcription was from Sigma (H-2149 type). Reaction products were purified by phenol extraction and double ethanol precipitation and separated on 0.8 mm 6% or 15% polyacrylamide gels (20:1 [w/w], acrylamide:bisacrylamide) containing 7 M urea. Transcripts were quantitated by a PhosphorImager with Image Quant software (Molecular Dynamics). The number of transcripts produced in a single-round transcription reaction was determined by comparison, on the same gel, of the radioactive signal produced by single-round transcripts with the signal of a known amount of a 32 P-labeled 200 bp DNA fragment of known specific radioactivity. A single round of transcription produced 8 fmol of SUP4-specific transcript. Therefore, only 10% of the template molecules could be assembled into active transcription complexes. This was expected, because in our assay, transcription components are limiting with respect to template DNA. Indeed, increasing the amounts of transcription components without changing the DNA concentration resulted in proportional increase of active transcription complexes (data not shown). Ternary complex purification was carried out as previously described (Dieci et al., 1995a). Acknowledgments Correspondence should be addressed to A. S. We thank Janine Huet for the gift of pure RNA pol III and TBP; Michel Werner, Christophe Carles, Olivier Lefebvre, and Marie-Claude Marsolier for helpful discussions and critical comments on the manuscript; and Carl Mann for improving the text. G. D. was supported by a long-term European Molecular Biology Organization Postdoctoral Fellowship. This work was supported by Biotechnology Programme Grant B102CT92-0090 from the European Community. Received September 13, 1995; revised October 30, 1995. References Arnosti, D.N., Merino, A., Reinberg, D., and Schaffner, W. (1993). Oct-2 facilitates functional preinitiation complex assembly and is continuously required at the promoter for multiple rounds of transcription. EMBO J. 12, 157–166. Baker, R.E., and Hall, B.D. (1984). Structural features of yeast tRNA genes which affect transcription factor binding. EMBO J. 3, 2793– 2800. Baker, S.M., and Platt, T. (1986). Pol I transcription: which comes first, the end or the beginning? Cell 47, 839–840. Bieker, J.J., Martin, P.L., and Roeder, R.G. (1985). Formation of a rate-limiting intermediate in 5S RNA gene transcription. Cell 40, 119–127. Brow, D.A., and Guthrie, C. (1988). Spliceosomal RNA U6 is remarkably conserved from yeast to mammals. Nature 334, 213–218. Burton, N., Cavallini, B., Kanno, M., Moncollin, V., and Egly, J.-M. (1991). Expression in Escherichia coli purification and properties of the yeast general transcription factor TFIID. Protein Exp. Purif. 2, 432–441. Campbell, F.E., and Setzer, D.R. (1992). Transcription termination by RNA polymerase III: uncoupling of polymerase release from termination signal recognition. Mol. Cell. Biol. 12, 2260–2272. Chamberlin, M.J. (1974). Bacterial DNA-dependent RNA polymerase. Enzymes 10, 333–374. Colbert, T., and Hahn, S. (1992). A yeast TFIIB-related factor involved in RNA polymerase III transcription. Genes Dev. 6, 1940–1949. Dieci, G., Hermann-Le Denmat, S., Lukhtanov, E., Thuriaux, P., Werner, M., and Sentenac, A. (1995a). A universally conserved region of the largest subunit participates in the active site of RNA polymerase III. EMBO J. 14, 3766–3776. Dieci, G., Duimio, L., Peracchia, G., and Ottonello, S. (1995b). Selective inactivation of two components of the multiprotein transcription factor TFIIIB in cycloheximide growth-arrested yeast cells. J. Biol. Chem. 270, 13476–13482.

Cell 252

Gabrielsen, O.S., Marzouki, N., Ruet, A., Sentenac, A., and Fromageot, P. (1989). Two polypeptide chains in yeast transcription factor t interact with DNA. J. Biol. Chem. 264, 7505–7511.

Nudler, E., Kashlev, M., Nikiforov, V., and Goldfarb, A. (1995). Coupling between transcription termination and RNA polymerase inchworming. Cell 81, 351–357.

Geiduschek, E.P., and Kassavetis, G.A. (1992). RNA polymerase III transcription complexes. In Transcriptional Regulation, Volume 1, S.L. McKnight and K.R. Yamamoto, eds. (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press), pp. 247–280.

Roberts, S.G.E., Choy, B., Walker, S.S., Lin, Y.S., and Green, M.R. (1995). A role for activator-mediated TFIIB recruitment in diverse aspects of transcriptional regulation. Curr. Biol. 5, 508–516.

Geiduschek, E.P., and Tocchini-Valentini, G.P. (1988). Transcription by RNA polymerase III. Annu. Rev. Biochem. 57, 873–914.

Wolffe, A.P., Jordan, E., and Brown, D.D. (1986). A bacteriophage RNA polymerase transcribes through a Xenopus 5S RNA gene transcription complex without disrupting it. Cell 44, 381–389.

Gottlieb, E., and Steitz, J.A. (1989a). Function of the mammalian La protein: evidence for its action in transcription termination by RNA polymerase III. EMBO J. 8, 851–861.

Yoo, C.J., and Wolin, S.L. (1994). La proteins from Drosophila melanogaster and Saccharomyces cerevisiae: a yeast homolog of the La autoantigen is dispensable for growth. Mol. Cell. Biol. 14, 5412–5424.

Gottlieb, E., and Steitz, J.A. (1989b). The RNA binding protein La influences both the accuracy and the efficiency of RNA polymerase III transcription in vitro. EMBO J. 8, 841–850.

Zawel, L., Kumar, K.P., and Reinberg, D. (1995). Recycling of the general transcription factors during RNA polymerase II transcription. Genes Dev. 9, 1479–1490.

Hawley, D.K., and Roeder, R.G. (1987). Functional steps in transcription initiation and reinitiation from the major late promoter in a HeLa nuclear extract. J. Biol. Chem. 262, 3452–3461. Huet, J., Riva, M., Sentenac, A., and Fromageot, P. (1985). Yeast RNA polymerase C and its subunits. Specific antibodies as structural and functional probes. J. Biol. Chem. 260, 15304–15310. Huet, J., Conesa, C., Manaud, N., Chaussivert, N., and Sentenac, A. (1994). Interactions between yeast TFIIIB components. Nucl. Acids Res. 22, 3433–3439. Jahn, D., Wingender, E., and Seifart, K.H. (1987). Transcription complexes for various class III genes differ in parameters of formation and stability towards salt. J. Mol. Biol. 193, 303–313. Jiang, Y., and Gralla, J.D. (1993). Uncoupling of initiation and reinitiation rates during HeLa RNA polymerase II transcription in vitro. Mol. Cell. Biol. 13, 4572–4577. Joazeiro, C.A.P., Kassavetis, G.A., and Geiduschek, E.P. (1994). Identical components of yeast transcription factor IIIB are required and sufficient for transcription of TATA box–containing and TATAless genes. Mol. Cell. Biol. 14, 2798–2808. Johnson, S.P., and Warner, J.R. (1989). Unusual enhancer function in yeast rRNA transcription. Mol. Cell. Biol. 9, 4986–4993. Kassavetis, G.A., Riggs, D.L., Negri, R., Nguyen, L.H., and Geiduschek, E.P. (1989). Transcription factor IIIB generates extended DNA interactions in RNA polymerase III transcription complexes on tRNA genes. Mol. Cell. Biol. 9, 2551–2566. Kassavetis, G.A., Braun, B.R., Nguyen, L.H., and Geiduschek, E.P. (1990). S. cerevisiae TFIIIB is the transcription initiation factor proper of RNA polymerase III, while TFIIIA and TFIIIC are assembly factors. Cell 60, 235–245. Kassavetis, G.A., Joazeiro, C.A.P., Pisano, M., Geiduschek, E.P., Colbert, T., Hahn, S., and Blanco, J.A. (1992a). The role of the TATAbinding protein in the assembly and function of the multisubunit yeast RNA polymerase III transcription factor, TFIIIB. Cell 71, 1055– 1064. Kassavetis, G.A., Blanco, J.A., Johnson, T.E., and Geiduschek, E.P. (1992b). Formation of open and elongating transcription complexes by RNA polymerase III. J. Mol. Biol. 226, 47–58. Kovelman, R., and Roeder, R.G. (1990). Sarkosyl defines three intermediate steps in transcription initiation by RNA polymerase III: application to stimulation of transcription by E1A. Genes Dev. 4, 646–658. Lin-Marq, N., and Clarkson, S.G. (1995). A yeast RNA binding protein that resembles the human autoantigen La. J. Mol. Biol. 245, 81–85. Maraia, R.J., Kenan, D.J., and Keene, J.D. (1994). Eukaryotic transcription termination factor La mediates transcript release and facilitates reinitiation by RNA polymerase III. Mol. Cell. Biol. 14, 2147– 2158. Matsuzaki, H., Kassavetis, G.A., and Geiduschek, E.P. (1994). Analysis of RNA chain elongation and termination by Saccharomyces cerevisiae RNA polymerase III. J. Mol. Biol. 235, 1173–1192. Moenne, A., Camier, S., Anderson, G., Margottin, F., Beggs, J., and Sentenac, A. (1990). The U6 gene of Saccharomyces cerevisiae is transcribed by RNA polymerase C (III) in vivo and in vitro. EMBO J. 9, 271–277.