Discrete functional stages of vaccinia virus early transcription during a single round of RNA synthesis in vitro.

We have developed a system for analysis of discrete steps in vaccinia virus early mRNA synthesis during a single round of transcription in vitro. A synthetic early promoter is used to direct transcription by vaccinia RNA polymerase of a G-less cassette in linear duplex DNA. Omission of GTP from transcription reactions leads to the formation of ternary elongation complexes paused stably at the end of the G-less cassette. These complexes can be induced to elongate by provision of GTP. While initiation of transcription is sensitive to low concentrations of salt and Sarkosyl, elongation is relatively resistant to these agents. Termination can be studied in a single synthetic cycle by forming transcription complexes paused just proximal to the termination signal TTTTTNT that can subsequently elongate and terminate. By selectively incorporating the termination-inhibiting analog BrUMP into proximal and distal portions of the nascent transcript, we localize the termination signal within or near the sequence UUUUUNU in the nascent RNA. We show that access of the vaccinia termination factor (VTF/capping enzyme) to the transcriptional apparatus can occur subsequent to initiation and synthesis of a 390-nucleotide nascent RNA. Termination is more sensitive to inhibition by salt and Sarkosyl than in elongation. This sensitivity is not reversed by preincubation of VTF with the transcription complex. Finally, we confirm the identity of VTF and vaccinia mRNA capping enzyme by demonstration of VTF activity associated with capping enzyme expressed in Escherichia coli.


Discrete Functional Stages of Vaccinia Virus Early Transcription during a Single Round of RNA Synthesis in Vitro*
(Received for publication, October 1, 1990) Yan Luo, Jeremiah Hagler, and Stewart ShumanS From the Program in Molecular Biology, Sloan-Kettering Institute, New York, New York 10021 We have developed a system for analysis of discrete steps in vaccinia virus early mRNA synthesis during a single round of transcription in vitro. A synthetic early promoter is used to direct transcription by vaccinia RNA polymerase of a G-less cassette in linear duplex DNA. Omission of GTP from transcription reactions leads to the formation of ternary elongation complexes paused stably at the end of the G-less cassette. These complexes can be induced to elongate by provision of GTP. While initiation of transcription is sensitive to low concentrations of salt and Sarkosyl, elongation is relatively resistant to these agents. Termination can be studied in a single synthetic cycle by forming transcription complexes paused just proximal to the termination signal TTTTTNT that can subsequently elongate and terminate. By selectively incorporating the termination-inhibiting analog BrUMP into proximal and distal portions of the nascent transcript, we localize the termination signal within or near the sequence UUUUUNU in the nascent RNA. We show that access of the vaccinia termination factor (VTFIcapping enzyme) to the transcriptional apparatus can occur subsequent to initiation and synthesis of a 390-nucleotide nascent RNA. Termination is more sensitive to inhibition by salt and Sarkosyl than is elongation. This sensitivity is not reversed by preincubation of VTF with the transcription complex. Finally, we confirm the identity of VTF and vaccinia mRNA capping enzyme by demonstration of VTF activity associated with capping enzyme expressed in Escherichia coli. Important insights into eukaryotic mRNA biogenesis have come from studies of RNA synthesis in vitro by infectious vaccinia virions (1)(2)(3). The recent development of soluble transcription systems derived from extracts of virus core particles (4,5) has led to the purification and characterization of viral proteins required for early transcription (6,7), and the identification of critical regulatory signals in the template DNA (8,9). Two protein components are necessary and sufficient for accurate initiation and effective elongation of RNA chains i n vitro. These are a DNA-dependent RNA polymerase (containing nine virus-encoded subunit polypeptides (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)) and an accessory factor, VETF' (composed of *This work was supported by Grants GM 42498-02 from the National Institutes of Health, and JFRA 274 from the American Cancer Society. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ Recipient of a Pew Scholarship in the Biomedical Sciences. I The abbreviations used are: VETF, vaccinia early transcription factor; VTF, vaccinia termination factor; nt, nucleotide(s); AdoMet, S-adenosylmethionine. two virus-encoded subunit polypeptides (20-22)). Duplex DNAs containing vaccinia early promoters can be transcribed i n vitro by highly purified vaccinia RNA polymerase supplemented with purified VETF (7), or by partially purified RNA polymerase preparations that include VETF (6). Purified RNA polymerase alone synthesizes RNA nonspecifically from single-stranded DNA templates (10)(11)(12) and is inactive on promoter-containing duplex DNAs (7). Purified VETF binds specifically to the vaccinia early promoter element and has an intrinsic DNA-dependent ATPase activity (20). The latter property may account for (or at least contribute to) the requirement for hydrolyzable ATP in early transcription (23-26).
Transcription of early viral genes terminates downstream of the signal T T T T T N T in the nontranscribed DNA strand (8,27). Purified vaccinia RNA polymerase is itself unable to terminate appropriately but is rendered competent to terminate in the presence of a transcription termination factor, VTF (6). VTF contains two subunit polypeptides of M , 95,000 ad M, 31,000 and is, by the criteria of copurification and thermal coinactivation, seemingly identical to the vaccinia virus mRNA-capping enzyme (6). The two capping enzyme subunits are themselves encoded by early viral genes (28,29). Initial mechanistic studies of termination suggest (i) that the termination signal may be recognized at the RNA level as the sequence UUUUUNU in the nascent transcript (30), and (ii) that VTF-depe ndent termination is the predominant pathway of mRNA 3'-end formation by vaccinia virions (31).
The dynamics of protein-protein and protein-nucleic acid interactions during the synthesis of early mRNAs are unknown. In seeking to address this point we were confronted by two constraints of the i n vitro system as presently constituted, these being (i) difficulty in restricting synthesis to a single round of transcription, and (ii) lack of methodology to analyze discrete steps in the transcription cycle, i.e. initiation, elongation, and termination. These impediments have been surmounted through the use of specialized DNA templates and pulse-chase RNA labeling strategies, as described in this report. We focus on the formation and properties of a stable ternary transcription complex, the nature of the termination signal in the nascent transcript, and the access of VTF to the ternary complex. Furthermore, we show that vaccinia-capping enzyme expressed in (and purified from) Escherichia coli (32) promotes transcription termination i n vitro.

EXPERIMENTAL PROCEDURES
DNA Templates-Plasmid pSB24, containing a G-less cassette (33) downstream of a synthetic vaccinia early promoter element, was constructed by Dr. Steven Broyles (Purdue University). The synthetic promoter (shown in Fig. le) is considerably more active in directing vaccinia transcription in vitro than are various "natural" early promoters; the choice of this particular sequence was inspired by the comprehensive study of early promoter strength reported by Davison and Moss (9). Plasmid pYL1, containing a vaccinia early termination Initiation, Elongation, and Termination by Vaccinia RNA Polymerase signal downstream of the G-less cassette, was constructed by replacing the sequences from the BamHI site to the Hind111 site of the pSB24 polylinker with a duplex oligonucleotide (shown in Fig. IC) using standard molecular cloning techniques. The inserted sequence is identical to the well-studied terminator at the 3'-end of the vaccinia growth factor gene (34) and includes three tandem copies of the TTTTTNT termination signal. Plasmids pC2AT and pML are described by Sawadogo and Roeder (33), and were the gift of Dr. Danny Reinberg (Rutgers/University of Medicine and Dentistry of New Jersey). Enzyme Purification-Vaccinia DNA-dependent RNA polymerase was purified as described (11). The phosphocellulose enzyme fraction (150 units/ml; 1 unit catalyzed incorporation of 1 nmol of UMP into acid-insoluble material under standard conditions (11)) was used routinely unless specified otherwise. The glycerol gradient polymerase fraction (6) was used in certain experiments where indicated, All RNA polymerase preparations contained core polymerase and the vaccinia early transcription factor, VETF. VTF/capping enzyme was purified from vaccinia virions as described (6). The phosphocellulose preparation of the virion-capping enzyme was used in the present study. Capping enzyme was expressed in E. coli and purified from bacterial lysates as described (32). The molar concentration of active capping enzyme was determined by enzyme-GMP complex formation as described (6).
Transcription in Vitro-Reaction mixtures containing 20 mM Tris-HCl, pH 8.0, 2 mM dithiothreitol, 6 mM MgC12, NTPs, linear duplex DNA template, RNA polymerase, and other proteins as indicated were incubated at 30 "C. Samples were processed as described (6). Transcription reaction products were analyzed by electrophoresis through 4% polyacrylamide gels under denaturing conditions. Radiolabeled RNAs were visualized by autoradiographic exposure of the gels.
Materials-Radiolabeled nucleotides were purchased from Amersham Corp. BrUTP was synthesized chemically from UTP (30). Restriction endonucleases were obtained from Bethesda Research Laboratories and New England Biolabs.

A Synthetic Early Promoter Directs Transcription of a G-
less Cassette by Vaccinia RNA Polymerase-Partially purified vaccinia RNA polymerase (that contained VETF and core polymerase) was used to transcribe the linear duplex template pSB24. This DNA contained a 382 nt G-less cassette situated immediately downstream of a synthetic vaccinia early promoter ( Fig. 1). Plasmid DNA was linearized at the end of the G-less cassette by treatment with endonuclease SmaI and tested for template activity as a function of DNA concentration. In the absence of added DNA, no radiolabeled RNA products were detected (Fig. 2B, lane 0). Addition of template lead to the appearance of a prominent labeled species of 390 nt, consistent with transcription of the entire G-less cassette from the promoter-driven initiation site to the end of the linear molecule. A less abundant transcript of approximately 350-360 nt was evident at lower DNA concentrations; the level of this RNA diminished steadily as template concentration was increased. In other experiments in which low template concentrations were employed, longer autoradiographic exposures revealed additional smaller RNA products with regular electrophoretic spacing (data not shown). The smaller RNA could arise by synthesis of more than one transcript/ template with inefficient release of the leading polymerase from the linear end of the DNA. This would result in the backup of lagging polymerase molecules such that the entire G-less cassette cannot be transcribed by those enzymes. Increasing template concentration would be expected to limit multiple initiations and reduce the abundance of the smaller RNA without affecting overall synthesis, just as was observed in the present case (Fig. 2B). Parenthetically, the size interval between the leading and lagging transcripts (30-40 nt) should reflect the amount of template DNA occupied by a single polymerase molecule. The presence of regularly spaced  smaller RNAs has also been detected in studies of transcription of the G-less cassette by RNA polymerase I1 (33,35,36).

GATCCAATTTTTTATAAATTTTTTTATGA G T T A A A A A A T A T T T A T A C T T C G A
Promoter Specificity-Two other plasmids were employed as control templates for in vitro transcription: (i) pCPAT, containing the G-less cassette alone with no viral promoter, and (ii) pML, containing a G-less cassette downstream of the adenovirus 2 major late promoter element. pML has been shown to be an effective template for transcription in vitro by RNA polymerase I1 (33). Each DNA was linearized with SmaI and tested at high and low DNA concentration for template activity in parallel with pSB24. No radiolabeled RNA was detected in the absence of DNA ( Fig. 2.4, lane 1) or in the presence of either pC2AT ( Fig. 2 A , lanes 4 and 5 ) or pML ( Fig. 2 A , lanes 6 and 7) at either DNA concentration. Thus, the vaccinia RNA polymerase was highly specific for its own early promoter and did not recognize a "strong" RNA polymerase I1 promoter even when the segment of DNA to be transcribed was invariant. The absence in transcription reactions containing the control templates of any RNA corresponding to the putative "lagging" nascent transcript seen at low concentration of pSB24 template ( Fig. 2 4 , lune 2) excluded the possibility that such transcripts arise via initiation within the G-less cassette in a promoter-independent reaction.
Analysis of a Single Round of Transcription-We have exploited the early promoter-driven G-less template to restrict our analysis to a single complete cycle of transcription in vitro. This has been accomplished as follows: (i) by developing a novel method to cotranscriptionally label RNA specifically in the 5' cap, and (ii) by defining a stable ternary complex of template, polymerase, and nascent RNA paused at the 3'-end of the G-less cassette that could be induced to resume elongation by the addition of GTP to the reaction.
In order to label the cap, advantage was taken of the ability of the vaccinia virus RNA guanylyltransferase (a component of the multifunctional capping enzyme) to utilize dGTP in lieu of GTP in the formation of a blocked 5' terminus (37). Because dGTP is not a substrate for RNA polymerase (ll), UTP, and 600 ng of SmaI-cut pSB24 DNA templat,e. Reactions were supplemented with NaCl prior to the addition of RNA polymerase (0.3 unit). Incubation was a t 30 "C for 60 min. An autoradiogram of the gel is shown. The concentration of added NaCl (mM) is indicated above each lane.
[a-'"PIdGTP added to the transcription reaction should be excluded from internal positions in the RNA chain. Also, each RNA molecule should be labeled only once with dGMP, obviating the need to normalize the extent of labeling to nucleotide composition and transcript length. Stable incorporation of dGMP into the cap should be enhanced by inclusion of AdoMet in the reaction. AdoMet serves as a methyl donor for N-7 methylation of the blocking guanylate (or deoxyguanylate) moiety (37,38). Methylation can be expected to render the dGMP-labeled cap structure resistant to pyrophosphorolytic reversal of the capping reaction that would otherwise be promoted by PPi generated during RNA synthesis (37). These predictions were borne out experimentally (Fig. 3, top panel). Inclusion of purified capping enzyme and [ a-'"PIdGTP in transcription reactions containing vaccinia RNA polymerase and SmaI-cut pSB24 template resulted in trace labeling of a 390-nt product, the size expected for a runoff transcript of the G-less cassette. dGMP labeling of this RNA was stimulated dramatically by the inclusion of AdoMet and was completely dependent on the presence of template DNA (Fig. 3 dinucleotide was not verified directly by chromatographic analysis of nuclease digests of the labeled transcript; however, it is extremely unlikely that the radionucleotide could have been assimilated at any other position in the RNA chain given that AdoMet has no stimulatory effect on the rate or extent of N T P incorporation into RNA by the fractionated in vitro system (6) or by permeabilized virions (31,39). Whether RNA polymerase could pause at the 3'-end of the G-less cassette was studied using a pSB24 template linearized with nuclease NdeI. This enzyme cleaves the plasmid 251 nt downstream of the SmaI site (where the first G residue is incorporated into the nascent RNA) (Fig. 1). Analysis of [a-"PIdGTP-labeled products after 5 min of synthesis revealed the presence of a single major transcript of 390 nt (Fig. 3, lower panel, denoted by P ) . This indicated that polymerase was largely unable to elongate past the G-less cassette in the absence of added GTP. This, in turn, could be attributable either to stable elongation block or to release of the RNA chain. These two possibilities were discriminated by analyzing the fate of the cap-labeled transcript after a "chase" with 1 mM unlabeled GTP. G T P addition should not only permit further elongation by polymerase (if paused on the template), but should also effectively preclude any subsequent cap labeling (e.g. on newly initiated transcripts) by virtue of overwhelming dilution of the specific activity of the radiolabeled cap donor. Also, G T P addition should not affect dGMP that had been incorporated into the cap during the "pulse" because those caps are stabilized against exchange with GTP by virtue of prior methylation. The observed effect of such a chase was to convert the 390-nt transcript to a species of approximately 630 nt, a size consistent with read-through transcription to the end of the template (Fig. 3, bottom panel, indicated by RT). The conversion to read-through product was nearly quantitative, indicating that virtually all polymerase molecules engaged during the pulse were stably paused at the end of the G-less cassette. The total amount of cap-labeled RNA remained relatively constant up to 15 min of chase, as predicted.
Single-round transcription conditions could also be achieved using a conventional pulse-chase protocol in which the nascent transcript was labeled by inclusion of [m-:"P]UTP in the reaction mixture (Fig. 4). Analysis of UMP-labeled products after a 5-min pulse phase of synthesis in the presence of HindIII-cut pSB24 template revealed a major transcript of 390 nt consistent with pausing at the end of the G-less cassette (Fig. 4, lane 1, indicated by P ) . The chase was performed by addition of excess unlabeled UTP and 1 mM GTP. This resulted in the near-quantitative conversion of the 390-nt transcript to a species of approximately 430 nt, indicative of read-through to the site of linearization by HindIII, 37 n t downstream of the 3'-end of the G-less cassette (Fig. 4, lane   3, indicated by H ) . The size of the read-through transcript was strictly a function of the site of linearization of the plasmid DNA; when similar pulse-chase experiments were performed using as template pSB24 cut with NdeI, elongation of the pulse-labeled transcript yielded a 630-nt product (Fig.  4, lane 2, indicated by N). When NdeI-cut and HindIII-cut pSB24 templates were present together at the same concentration during the pulse-phase of the reaction, the transcripts were chased into approximately equal amounts of 630 and 430 nt run-off RNAs (Fig. 4, lane 4 ) . However, when NdeIcut template was included in the pulse-phase and a second HindIII-cut template was added upon initiation of the chase, only the 630-nt transcript of the NdeI-cut DNA was radiolabeled (Fig. 4, lane 5 ) . Thus, no new transcripts were labeled during the chase.
These experiments (Figs. 3 and 4) established the feasibility of analyzing a single round of RNA synthesis via a pulsechase protocol and provided a method for temporal separation of the initiation/elongation and termination phases of the transcription reaction (see below).
Properties of the Ternary Transcription Complex: Stability to Salt and Sarhosyl-Transcription of pSB24 template by vaccinia RNA polymerase was extremely sensitive to increasing NaCl concentration. Addition of salt to the reaction mixture prior to the addition of RNA polymerase resulted in a concentration-dependent inhibition of synthesis (Fig. 2C). Reactions supplemented with 2100 mM NaCl (over the ap-  1 unit), and linear IINA template, as follows: lanes I and 3,300 ng of HindIII-cut pSH24; lnnrs 2 and 5 , 300 ng of NdrI-cut pSR24; lanP 4 , 300 ng of' HindIIIrut pSR24 plus 300 ng of NdrI-cut. pSR24. Reactions were incuhated for . 5 min at 30 "C and either processed directly for elect.rophoretic analysis (lanr I ) or "chased" by addition of 1 mM unlaheled U T P and 1 mM GTP ( / a r m 2-4). The reaction in lane 5 was supplemented with a second template (300 ng of HindIII-cut pSR24) immediately following the initiation of the chase. Incubation was continued for 5 min at 30 "C. An autoradiogram of the gel is shown. Transcription reaction products corresponding to paused nascent RNA ( P ) , and read-through RNAs synthesized from either the NdrI-cut templat,e (N) or the HindIII-cut template ( H ) are indicated on the right.  3 unit), and 100 fmol of purified vaccinia-capping enzyme. Samples were incuhated for 5 min at 30 "C and t.hen either heated at 65 "C for 5 min and processed for electrophoresis (lane A ) or supplemented with NaCI. The chase-phase was then initiated by addition of 1 mM G T P followed by incuhation a t 30 "C for another 5 min. An autoradiogram of the gel is shown. The concentration of NaCl (mM) added t.0 each reaction mixture after the pulse phase is indicated ahove the lanes. Transcript,ion react.ion products corresponding to paused nascent RNA ( P ) and read-through RNA (HT) species are indicated on the right. proximately 25 mM NaCl contributed to the standard reaction by the polymerase) were completely inactive. In contrast, the ternary transcription complex of RNA polymerase, template, and nascent RNA was remarkably insensitive to salt (Fig. 5). Ternary complex containing cap-labeled nascent RNA was generated by pulse-labeling for 5 min as described above. Analysis of the RNA at the end of the pulse revealed a product of the size expected for a paused transcript (Fig. 5 , , lane A).
Reaction mixtures were supplemented with increasing concentrations of NaCl (indicated above the lanes in Fig. 5 ) and then chased by the addition of 1 mM GTP. NaCl concentrations up to 300 mM had little effect on the ability of polymerase to elongate to the end of the template. NaCl concentrations 2400 mM reduced incrementally the amount of fulllength read-through RNA and resulted in the appearance of new transcripts of discrete size between that of the paused RNA and the read-through RNA. At 500 mM NaCI, a significant fraction of the pulse labeled RNAs were not extended during the chase (Fig. 5). Was this due to inhibition of nucleotide incorporation by polymerase or to salt-induced destabilization of the ternary complex leading to transcript release (termination)? Additional experiments suggested that the nonelongated RNAs observed a t high salt had been released from the transcription complex, insofar as paused ternary complexes that were adjusted to 450 mM in NaCl and then diluted to 150 mM NaCl in transcription buffer did not catalyze further elongation to full-length transcripts when chased with GTP (data not shown).
Inclusion of the detergent Sarkosyl in the reaction mixture prior to the addition of RNA polymerase resulted in a concentration-dependent inhibition of transcription of a SmaI-cut pSB24 template (Fig. 6A). UMP incorporation into the runoff transcript (assayed by cutting out the appropriate region of the dried gel and counting the gel slice in a liquid scintillation fluid) was inhibited by 80% a t 0.0125% Sarkosyl and by 96% at 0.02% Sarkosyl; no labeled RNA product was discernible a t detergent concentrations in excess of 0.02% (Fig. 6A). Similar experiments were performed using NdeI-cut pSR24 template in reactions lacking GTP. Under these reaction conditions, only a single cycle of initiation events was permitted. It was observed that the concentration-dependent inhibition of synthesis of paused nascent RNA by Sarkosyl was essentially identical to that depicted in Fig. 6A (data not shown). In contrast, elongation of preformed ternary complex was refractory to Sarkosyl up to 0.03% (Fig. 6R). Partial inhibition of elongation was observed at 0.05% and elongation was completely abrogated a t 0.1% detergent. Studies of Transcription Termination Nature of the Termination Signal-In order to study transcription termination in a single round of RNA synthesis, we constructed the template pYLl by inserting downstream of the G-less cassette of pSB24 a synthetic oligonucleotide containing three copies (in tandem or overlapping) of the TTTTTNT early termination signal (Fig. 1). NdeI-cut pYLl was used to program transcription in uitro in the absence of GTP, leading to the accumulation of pulse-labeled ternary complexes that were paused at the 3'end of the G-less cassette (Fig. 3, bottom panel). Addition of G T P chased the nascent RNA into two classes of product (i) a read-through transcript, as was observed when pSB24 was used as template, and (ii) an RNA of 435 nt, arising via termination approximately 35 nt downstream of the first encountered T T T T T N T signal. The synthesis of this later class of transcript depended on the presence in the template of the cis-acting termination motif (Fig. 3, bottom panel, compare SB24 and YLl templates).
Prior studies had indicated that the cis-acting signal for factor-dependent termination was contained within the nascent RNA; this was based on the selective abrogation of factordependent termination when BrUTP or IUTP was substituted for UTP in the in uitro transcription reaction (30). We have now applied this strategy to the pulse-chase transcription reaction in order to probe the influence of specific regions of the nascent transcript on the termination signal transduction. As shown in Fig. 7, inhibition of termination by BrUTP was observed only when analog substitution occurred during the chase period, i.e. specifically when the TTTTTNT signal was being transcribed into UUUUUNU. BrUMP substitution for UMP throughout the 390-nt portion of the nascent RNA immediately preceding the termination signal had no appreciable inhibitory effect, provided that UTP was present in 20-

FIG. 7. BrUTP prevents transcription termination by incorporation of BrUMP into the distal segment of nascent RNA
containing the UUUUUNU termination signal. Transcription reactions were performed in two stages. In the pulse-phase, reactions contained 1 mM ATP, 1 mM CTI', 50 p~ UTP, or RrUTP (as indicat.ed), 0.33 p~ [o-:"P]dCTP (3000 Ci/mmol), 600 ng of NdeI-cut pYL-1 template, 50 p~ AdoMet, RNA polymerase (0.3 unit), and 100 fmol of purified vaccinia-capping enzyme. Samples were incubated for 5 min at 30 "C and then either processed immediately for electrophoresis (indicated in the figure by asymbol for the chase phase) or chased by adding 1 mM U T P or BrUTP (as indicated) followed by addition of 1 mM GTP. Incubation was continued for 5 min a t 30 "C. An autoradiogram of the gel is shown. Transcription reaction products corresponding to paused nascent RNA ( P ) , read-through RNA ( R T ) , and terminated RNA (7') species are indicated on the left. fold excess over BrUTP during the chase. Only about 45 nt separate the pause site at the end of the G-less cassette and the approximate site of termination; this region of RNA includes 19 U residues of which 14 are within the tandemly arranged termination signals. The data suggested that BrUMP specifically prevented interaction of VTF and/or RNA polymerase with the segment of the RNA containing the termination signal and that no other region of the transcript contributed to the analog effect.
Recombinant Capping Enzyme Promotes Termination in Vitro-Vaccinia capping enzyme is a multifunctional protein that catalyzes three mechanistically distinct reactions leading to the synthesis of the cap zero structure (38,40). Studies to date indicate that capping enzyme and VTF are intimately associated (6). Active capping enzyme was recently produced in E. coli by coexpressing the viral genes encoding the M , 95,000 and 31,000 subunit polypeptides (32, 41). In the heterologous system, the two subunits associated with the proper stoichiometry to form a heterodimeric protein with RNA triphosphatase, RNA guanylyltransferase, and RNA (guanine-7) methyltransferase activities (32).
Does the capping enzyme expressed in E. coli also have VTF activity? This was addressed by assaying the ability of "native" and "recombinant" capping enzyme preparations to restore termination competence to termination-defective RNA polymerase (6). As shown in Fig. 8, the RNA polymerase synthesized a 1000-nt read-through transcript when programed with SspI-cut pYLl template (lane -). Addition of capping enzyme purified from E. coli resulted in the appear-Initiation, Elongation, and Termination by Vaccinia RNA Polymerase ance of a 435-nt terminated transcript (Fig. 8, lane E on left) identical in size to that of the terminated RNA made in the presence of capping enzyme purified from vaccinia virions (Fig. 8, lane V on left). The amount of recombinant capping enzyme used in this experiment was less than the amount of native enzyme. This was because the addition of higher amounts of the E. coli protein preparation resulted in complete inhibition of overall transcription. We attributed this inhibitory effect to bacterial proteins present in the enzyme fraction (which was shown to be 32% pure with respect to the vaccinia capping enzyme (32)). Nonetheless, the termination-promoting effect of the bacterial protein was dependent on the amount added within a narrow range of concentration (data not shown). Significantly, the appearance of the 435-nt transcript depended upon the presence in the template of the cisacting termination signal, i.e. no RNA species the size of the terminated transcript was induced by addition of E. coli capping enzyme to reactions programed by SspI-cut pSB24 (Fig. 8, lane E on right). We infer from this experiment that capping enzyme and VTF are indeed identical.

V T F Promotes Termination by Preformed Ternary Complex-Capping of vaccinia mRNA (in virions) is believed to occur on short nascent chains,
i.e. soon after initiation of transcription. The present study shows clearly that capping in the in vitro system also occurs on nascent transcripts (Fig.  3). Does the same enzyme molecule that caps the 5'-ends of an RNA mediate termination hundreds or even thousands of bases downstream and, if so, how? Or are capping and termination wholly unrelated, mediated by different enzyme molecules (e.g. from a pool of exchangeable molecules) that act at opposite ends of the transcription unit? The fact that the number of enzyme molecules/virus particle is estimated at 80-100 for both capping enzyme and RNA polymerase (12,42,43), and is roughly the same as the number of early transcription units, argues against a physiologic pool of excess capping enzyme.
One plausible model is that capping enzyme is engaged in the transcription process from the outset and remains associated with the transcription complex (template, polymerase, nascent RNA, etc.) during elongation. In this way, efficient capping of nascent chains might be ensured, and VTF would remain poised to "read" the UUUUUNU termination signal in the nascent RNA and induce the polymerase to terminate. We examined one facet of this proposal by asking if VTF/ capping enzyme had to be present at the time of initiation in order to promote 3'-end formation. An experiment was performed in which nascent RNAs were labeled with [a-:"PP]CTP during a 5-min pulse in the absence of GTP, then chased in the presence of excess cold CTP and GTP. The major product synthesized by the glycerol gradient polymerase fraction a t the end of the pulse was a 390-nt transcript, consistent with pausing at the end of the G-less cassette (Fig. 9, lane 1 ). A minor transcript of approximately 410 nt was also evident. Identical transcripts were synthesized when capping enzyme was included during the pulse (Fig. 9, lane 3 ) . During the chase, transcripts made by polymerase alone from the pYLl template were elongated to produce a major run-off product (Fig. 9, lane 2 ) . In contrast, RNAs synthesized in the presence of capping enzyme were converted predominantly into terminated products during the chase (Fig. 9, lane 4 ) . Paused ternary complexes formed by polymerase alone were supplemented with capping enzyme and incubated for an additional 1 min (Fig. 9, lune 5 ) or 5 min (lune 6 ) a t 30 "C. Although most of the paused transcripts were stable during this period, we noted an increase in the amount of the minor 410-nt transcript between 1-and 5-min post-pulse. We suspect that trace levels of G T P contamination permitted a low level of elongation past the G-less cassette until a distal G residue was encountered (at a point 30 nt downstream of the 3'-end of the cassette) resulting in a second elongation block. This notwithstanding, the addition of GTP to paused complexes that had been incubated with capping enzyme after the pulse resulted in the chase of nascent paused RNAs into predominantly terminated products (Fig. 9, lune 7). Thus, the presence of capping enzyme at the time of transcription initiation or during synthesis of the proximal portion of the complete RNA was not required in order for termination to occur.
VTF-dependent Termination Is Sensitive to Salt and Sarkosyl-We began to address the functional interaction of VTF with the transcription elongation complex by examining the sensitivity of the termination reaction per se to salt and Sarkosyl. In these experiments, VTF was added to preformed UMP-labeled ternary complex on an NdeI-cut pYLl template and further elongation was permitted by provision of cold UTP and GTP (Fig. 10, A and 23). The appearance of an appropriately terminated transcript during the chase (denoted by T in Fig. 10) depended entirely on VTF added after the pulse (compare lanes -VTF and 0 in Fig. 10, A and R ) . Our analysis showed that Sarkosyl abolished termination a t 0.02% concentration, i.e. a t a level of detergent well below that required to inhibit elongation (Fig. 10B). A parallel series of experiments was performed in which VTF was included in the reactions prior to addition of Sarkosyl and subsequent chase (eg. VTF added at the same time as polymerase). It was observed that the Sarkosyl sensitivity of termination was unaltered (compared with the results of Fig. 10R) by varying the order of addition (data not shown). Further experiments were performed in which Sarkosyl was added to paused ternary complexes synthesized on NdeI-cut pYLl template and elongation was then resumed by chasing in the absence of added VTF. Under these circumstances, only read-through RNA was produced during the chase (eg. see Fig. 10R, lune -VTF). The effects of increasing Sarkosyl concentration on resumption of elongation was essentially identical to the results of Fig. 6B (data not shown). Thus, the addition of Sarkosyl during elongation and synthesis of RNA containing the UUUUUNU signal did not confer upon the elongating polymerase the ability to terminate in a factor-independent immediately for electrophoresis (lane -VTF (pulse)). Next, reactions were supplemented with the indicated concentrations of NaCl prior to the addition of capping enzyme (100 fmol). Reactions were chased immediately thereafter by addition of 1 mM unlabeled UTP and 1 mM GTP. Incubation was for another 5 min a t 30 "C. A control reaction was performed in which capping enzyme was omitted during the chase (lane -VTF (chase)). An autoradiogram of the gel is shown. Transcription reaction products corresponding to paused nascent RNA ( P ) , read-through RNA ( R T ) , and terminated RNA ( T ) species are indicated on the right. Panel R, pulse-chase transcription reactions were performed as described for panel A except that reactions were adjusted to the indicated concentrations of Sarkosyl (rather than salt) after the pulse and prior to the addition of capping enzyme and commencement of the chase. Transcription reaction products corresponding to paused nascent RNA ( P ) , read-through RNA ( R T ) , and terminated RNA (7') species are indicated on the right. manner. Nor were any new transcripts shorter than the readthrough RNA generated by addition of Sarkosyl prior to the chase (data not shown).
The salt sensitivity of termination is shown in Fig. 1OA. In this experiment, we estimated (by densitometric scanning) that only 70% of the pulse-labeled transcripts made in the control reactions (e.g. Fig. lOA, lanes -VTF and 0) were extended during the chase; this efficiency of elongation was less than that seen in other experiments (Figs. 3-6 and 10B). I t was found that supplementation of the reactions with increasing concentrations of NaCl prior to addition of VTF and commencement of the chase resulted in progressive inhibition of termination (Fig. lOA, lanes 10-100). At 100 mM NaCl, when termination was abrogated, the elongation efficiency was reduced slightly (such that only 50% of the pulselabeled RNAs were extended). Although this effect of 100 mM NaCl on elongation was not observed in other experiments (e.g. Fig. 5, in which the chase was more efficient overall), it was clear nonetheless that salt was more inhibitory to termination than to elongation. The salt sensitivity of termination was unaltered by addition of VTF at the time of transcription initiation (data not shown).
It was conceivable that exposure to salt or Sarkosyl might inactivate VTF/capping enzyme, e.g. via global protein denaturation or through specific effects on individual functional domains. In the case of salt, we found that RNA guanylyltransferase activity of the capping enzyme, as assayed by the formation of covalent enzyme-GMP complex (42) was unaffected at 0.1 M added NaCl (a level that abolished termina-tion), was reduced by 10-20% at 0.2 M NaCl, and was inhibited by 50% a t 0.4 M NaCl (data not shown). Sarkosyl, in contrast to salt, was actually inhibitory to the guanylyltransferase activity of capping enzyme in the same range of detergent concentration that blocked termination. Addition of 0.01, 0.02, and 0.03% Sarkosyl to reaction mixtures resulted in 25, 75, and 94% inhibition of enzyme-guanylate formation (data not shown). Inhibition of guanylyltransferase by 0.03% Sarkosyl was reversible by 20-fold dilution of the detergent to 0.0015% (data not shown). VTF activity was also restored to 0.03% Sarkosyl-treated enzyme upon dilution of this enzyme into reaction mixtures containing preformed ternary complexes and no added Sarkosyl (data not shown). Thus, the effects of salt and Sarkosyl on termination were attributable to inhibition of VTF activity rather than irreversible enzyme inactivation.

DISCUSSION
Single Round of RNA Synthesis in Vitro-Discrete steps in transcription by DNA-dependent RNA polymerase have been studied in depth in bacteria (reviewed in Ref. 44) and, more recently, in eukaryotic systems (e.g. [45][46][47][48]. A common experimental strategy is the use of reagents that disrupt proteinnucleic acid and/or protein-protein interactions critical to a specific stage of the initiation or elongation reaction; such agents or treatments include Sarkosyl, heparin, high salt, thermal shift, and template challenge. It is a general property of RNA polymerases that the initiation and elongation phases of transcription display different sensitivities to such agents; typically the formation of a ternary complex is accompanied by enhanced stability to salt, Sarkosyl, etc. We have shown this to be the case for vaccinia virus RNA polymerase and have defined the sensitivities of the initiation, elongation and termination stages of vaccinia early transcription to both salt and Sarkosyl. In addition, we have employed pulse-chase labeling strategies (involving either cap labeling at the 5'-end with dGTP or internal labeling with rNTPs) to examine the fate of transcripts made within a single round of synthesis without having to specifically block the capacity for subsequent initiation events.
Factor-dependent Transcription Termination-We have examined transcription termination using a linear template in which the termination signal T T T T T N T is placed downstream of the G-less cassette. Polymerase paused at the end of the G-less cassette can, when allowed to resume elongation by addition of GTP, terminate transcription in response to the inserted element. The ability to temporally separate initiation and termination in a single transcriptional cycle has facilitated study of the termination signal. Specifically, by selectively incorporating the analog BrUMP into proximal and distal portions of the nascent transcript, we localize the termination signal within or near the sequence UUUUUNU in the nascent RNA. Selective incorporation of other labeled or derivatized precursors into specific segments of the transcript may prove useful as well in cross-linking experiments to address dynamics of protein-RNA interactions.
Our finding that VTF/capping enzyme can promote termination by preformed ternary complex containing a 390-nt nascent transcript rules out an obligate requirement for assembly of a "termination competent" apparatus either prior to initiation or during early phases of elongation. The data do not, however, exclude the possibility that VTF/capping enzyme does actually interact with polymerase and/or nascent RNA a t those phases when mRNA synthesis occurs in a physiologic setting (i.e. within the virion core particle). The issue of how the capping enzyme gains access to the elongating transcription complex in the reconstituted system, e.g. via RNA-protein interactions, protein-protein in$eraction, or both, remains open. Clearly, this putative interaction step for VTF is more sensitive to salt and Sarkosyl than is the maintenance of an intact elongation complex. The inability to override the salt inhibition of termination by prior addition of VTF can be interpreted in several ways. It is possible that (i) VTF does not interact at all with the ternary complex prior to the appearance of the UUUUUNU sequence in the nascent RNA (indeed the signal has not yet been transcribed by the paused complexes used in our experiments), (ii) VTF can bind to ternary complex but this interaction is either unstable or in rapid equilibrium with free VTF and therefore readily dissociated by salt, (iii) stable interaction of VTF with elongation complex occurs only after synthesis of the UUUUUNU signal and would therefore not be revealed by order of addition experiments using pYL1 as a template. Further studies addressing the timing and mode of ingress of VTF into the transcriptional apparatus will require the physical isolation of the ternary complex paused upstream of the termination signal (devoid of NTPs and free proteins) and the construction of new templates that allow pausing of the elongation complex downstream of the UUUUUNU signal.
Finally, the observation that heterologously produced capping enzyme promotes termination by vaccinia polymerase circumvents the caveats inherent in earlier studies demonstrating association of VTF and capping enzyme by copurification (6). It also opens up a molecular genetic approach to VTF function via the heterologous expression of mutated alleles of the genes encoding the enzyme subunits.