Opposite Consequences of Two Transcription Pauses Caused by an Intrinsic Terminator Oligo(U)

The RNA oligo(U) sequence, along with an immediately preceding RNA hairpin structure, is an essential cis-acting element for bacterial class I intrinsic termination. This sequence not only causes a pause in transcription during the beginning of the termination process but also facilitates transcript release at the end of the process. In this study, the oligo(U) sequence of the bacteriophage T7 intrinsic terminator Tϕ, rather than the hairpin structure, induced pauses of phage T7 RNA polymerase not only at the termination site, triggering a termination process, but also 3 bp upstream, exerting an antitermination effect. The upstream pause presumably allowed RNA to form a thermodynamically more stable secondary structure rather than a terminator hairpin and to persist because the 5′-half of the terminator hairpin-forming sequence could be sequestered by a farther upstream sequence via sequence-specific hybridization, prohibiting formation of the terminator hairpin and termination. The putative antiterminator RNA structure lacked several base pairs essential for termination when probed using RNases A, T1, and V1. When the antiterminator was destabilized by incorporation of IMP into nascent RNA at G residue positions, antitermination was abolished. Furthermore, antitermination strength increased with more stable antiterminator secondary structures and longer pauses. Thus, the oligo(U)-mediated pause prior to the termination site can exert a cis-acting antitermination activity on intrinsic terminator Tϕ, and the termination efficiency depends primarily on the termination-interfering pause that precedes the termination-facilitating pause at the termination site.

Multisubunit bacterial RNA polymerase transcription terminates upon various termination signals, causing disassembly of elongation complexes (ECs) 2 with or without the assistance of termination factors (1,2). The most common class of intrinsic (factor-independent) termination signal relies on an oligo(U) sequence in the RNA preceded by an RNA hairpin structure (class I intrinsic termination signal) (3). The termination mechanism of class I signals is shared by virtually all multisubunit bacterial and single-subunit phage RNA polymerases. However, it is distinct from the mechanism of class II intrinsic ter-mination, which is caused by specific DNA sequence recognition of single-subunit phage T7 RNA polymerase (4).
The oligo(U) sequence allows ECs to pause prior to the release of RNA transcripts (5,6). This transcription pause has been shown to be the first signal for the class I termination pathway, providing additional time for the formation of a terminator RNA hairpin that opens or unwinds the RNA-DNA hybrid at an upstream region, shortening and weakening the hybrid (5,7,8). Because the remaining downstream part of the hybridized region consists mostly of rU:dA base pairs, it becomes thermodynamically unstable, facilitating the release of nascent RNA transcripts (9).
In this study, an opposite effect of the oligo(U)-mediated pause was found with a typical class I signal of T7 terminator T for T7 RNA polymerase (see Fig. 1). The oligo(U) sequence of this terminator induces a pause not only at the termination site (referred to hereafter as position T) but also at a position 3 bp upstream (position T-3). Although the pause at position T facilitates termination, the pause at position T-3 exerts an antitermination activity on the complex.
In many of the antitermination mechanisms of class I intrinsic termination, regulatory factors interact with RNA polymerase to convert the transcription complex into a terminationresistant form. For example, phage -encoded N protein interacts with Escherichia coli RNA polymerase, exerting an antitermination effect on nutL and nutR terminators of the phage genome (1). In other examples, RNA polymerase interacts with a product transcript rather than an external factor; hence, the antitermination is intrinsic, requiring no external factors, such as the antitermination at the phage HK022 put operon leader (10).
In contrast, in other antitermination mechanisms, antitermination factors interact with transcript RNA rather than RNA polymerase to alter RNA secondary structure. For example, tiny abortive initiation transcript RNAs (11) or uncharged tRNAs (12,13) interact with transcript RNA, interfering with terminator hairpin formation and exerting antitermination effects on phage T7 terminator T and on the Bacillus subtilis terminator of the T box gene family, respectively.
Rather than involving an alteration in RNA polymerase conformation, the antitermination mechanism described in this study appears to involve an alteration in transcript RNA secondary structure that prevents the formation of the terminator hairpin structure. This antitermination is made possible by a transcription pause prior to the termination site and does not require an external regulatory factor.
In this intrinsic antitermination, the causal pause associates with the oligo(U) sequence, which is also responsible for causing another pause and termination at the termination site. We suggest that the oligo(U)-induced pause could be an antitermination signal as well as a termination signal, depending on the RNA secondary structures that are formed during the pauses.
Stepwise Walking of T7 RNA Polymerase-EC with 15-mer RNA stalled at position T-77 (i.e. 77 bp upstream of the termination site) was obtained by incubating the linear biotinylated KM01 template (125 nM) bound to streptavidin-coated magnetic beads (Invitrogen) in a 160-l reaction containing 40 mM Tris-HCl (pH 7.9), 6 mM MgCl 2 , 100 mM KCl, 10 mM dithiothreitol, 500 M ATP, 500 M GTP, 50 M CTP, 40 units of RNasin (Promega), and 2000 units of T7 RNA polymerase at room temperature for 20 min (15). EC at position T-77 was advanced to position T-74 in a walking reaction using 80 Ci of [␣-32 P]UTP (800 Ci/mmol; PerkinElmer Life Sciences); after washing, the radiolabeled EC was farther advanced to the desired downstream positions by repetition of washing and incubation in 160 l of transcription buffer containing 0.5 M rNTPs required for the next walking step. For incorporation of IMP into RNA transcripts, ECs were incubated with 5 M ITP (Ambion).
Single-round Transcription-EC stalled at a particular position was incubated in a 20-l solution of 40 mM Tris-HCl (pH 7.9), 6 mM MgCl 2 , 100 mM KCl, 10 mM dithiothreitol, 200 M rNTPs, and 4 units of RNasin at room temperature for various amounts of time, and the incubation was stopped by the addition of 20 l of gel loading buffer (12 M urea, 10 mM EDTA, and 0.1% bromphenol blue) prewarmed at 65°C. After being heated at 95°C for 5 min, reaction products were separated by electrophoresis on 8 M urea-12% polyacrylamide gels. Gels were dried and scanned using a phosphorimaging analyzer (Fuji BAS 3000), and band intensities were quantified using TINA 2.0 software.
RNA Synthesis-For structure probing, the terminator and antiterminator RNAs were synthesized and labeled by incubating 10 pmol of a DNA template as described above in a 200-l reaction containing 40 mM Tris-HCl (pH 7.9), 6 mM MgCl 2 , 50 mM KCl, 10 mM dithiothreitol, 500 M ATP, 500 M GTP, 500 M CTP, 100 units of T7 RNA polymerase, and 50 Ci of [␥-32 P]GTP (6000 Ci/mmol; PerkinElmer Life Sciences) at 37°C for 4 h. DNA templates were obtained by PCRs of pKM01 using 5Ј-CGGCGTAGAGGATCGAGA-3Ј as the forward primer and 5Ј-CAAGACCCGTTTAGAGGCCC-3Ј (for an EC pausing at position T-3) or 5Ј-CCTCAAGACCCGTT-TAGAGG-3Ј (for an EC pausing at position T) as the reverse primer and were digested by DNase I (Takara), followed by phenol/chloroform extraction and ethanol precipitation.
Enzymatic Probing of RNA Secondary Structures-Synthesized RNAs were refolded by heating in RNA structure buffer (10 mM Tris-HCl (pH 7), 100 mM KCl, and 10 mM MgCl 2 ) at 95°C for 2 min, followed by cooling to room temperature for 1 h. For enzymatic cleavage, radioactive RNA (5 nCi) was incubated at room temperature with RNase A, T1, or V1 (Ambion) for 15 min. The reactions were stopped by the addition of inactivation buffer (Ambion), followed by ethanol precipitation. Samples were analyzed by electrophoresis on 15% polyacrylamide gels containing 8 M urea. RNA was denatured and digested by RNase T1 to produce G ladders and then boiled in an alkaline buffer (50 mM NaOH and 10 mM EDTA) for 5 min to make all ladders.

Two Transcription Pauses at and near the Termination Site
Caused by Oligo(U)-An oligo(U) sequence is immediately adjacent to the 3Ј-terminal G residue in most of the RNA terminated at phage T7 terminator T by T7 RNA polymerase ( Fig. 1) (16,17). To study the pattern of transcription pausing in T, we monitored transcription reactions of a template with T in a time course manner. Elongation paused for several seconds at two positions, T-3 and T ( Fig. 2A).
In a modified template with substitutions that disrupt the terminator hairpin, elongation still paused at both positions ( Fig. 2A). In contrast, both pauses disappeared in a broken oligo(U) variant ( Fig. 2A). Thus, the transcription pauses at positions T-3 and T were caused by oligo(U) rather than by RNA hairpin.
Pause at Position T-3 Associates with Read-through at Position T-Although the T-3 pause was caused by oligo(U), a causal element for termination, quantitative analysis of the results of time course elongation on the wild-type template showed that it associated with read-through rather than termination ( Fig. 2, B and C). After reaching a peak at 4 or 5 s, the amount of the T-3 pause complex decreased by 22% during the next 2-s period (from 5 to 7 s). During the same period, the amount of read-through complexes increased by 28%, but that of terminated complexes increased by only 7%. Accordingly, read-through complexes rather than terminated complexes appeared to emerge in a reciprocal manner as the T-3 pause complexes diminished.
When time course transcriptions were repeated at a lower NTP concentration (10 M), it was also observed that readthrough complexes emerged in a reciprocal manner as the pause complexes diminished (Fig. 2, D and E). From 16 to 31 s, the pause decreased by 31%, and read-through increased by 41%, although termination increased by only 4%.
When the NTP concentration was increased to 250 M, the T-3 pause reached a lower maximum peak (5% at 2 s), and subsequently, a lower plateau of read-through (30%) was achieved than at 50 or 10 M NTP. Comparison of the data obtained at the three different NTP concentrations revealed a high correlation (r ϭ 0.89) between the T-3 pause strength and the final read-through efficiency. The read-through efficiency was 30, 40, and 68% when the maximum pause was 5, 28, and 39%, respectively (as measured at 250, 50, and 10 M NTP, respectively).
Thus, the more strongly a pause occurred at position T-3, the more frequently a read-through occurred at position T. These results suggest that the pause at position T-3 facilitates readthrough at position T rather than termination, whereas the pause at position T facilitates termination.
Stalling or Slowing Down at Position T-3 Facilitates Read-through-To confirm the finding that the T-3 pause facilitates read-through, the pause was prolonged by stalling ECs in transcription reactions that lacked a specific NTP before resuming elongation with supplementation of the NTP. Stalled complexes could be isolated on biotinylated templates using streptavidin-coated magnetic beads (15), and stalling could be induced virtually at any position of the template for any length of time in this way.
When ECs were stalled at position T-3 for varying periods of time on a modified template in which an A residue had been substituted for the T-2 residue, breaking the oligo(T) stretch (this substitution decreased termination efficiency to 57%), termination occurred less frequently as the duration of stalling increased (Fig. 3A). When ECs were stalled briefly (for 5 min) at varied positions of the wild-type template before being chased with 50 M NTP, termination efficiency was decreased upon stalling at position T-12, T-11, or T-7 (Fig. 3B). However, brief stalling at positions between T-33 and T-13 had little effect on termination efficiency.
These experiments were repeated with stalling at various positions for varied durations (Fig. 3C). For reactions in which stalling occurred at positions between T-12 and T-3, termination efficiency exponentially decreased with increasing duration of stalling. In contrast, stalling at position T enhanced the termination efficiency, consistent with previous observations (5). Stalling at positions between T-17 and T-13 reduced termination efficiency only a bit, regardless of the duration of stalling. Thus, upstream stalling facilitated read-through at position T more effectively when it was longer or closer to position T, but it was effective only with stalling at positions between T-12 and T-3.
A much shorter or weaker pause than stalling can be induced by lowering the concentration of an NTP(s) instead of completely removing it. According to previous studies, elongation slowed down and termination increased with lower concentrations of NTPs (18,19). In contrast, we found here that lowering specific NTP concentrations decreased termination efficiency. After stalling at position T-17, which, as described above, did not affect termination efficiency, ECs were chased with NTPs in which the GTP concentration was varied from 5 pM to 500 M (Fig. 3D).
In the wild-type template, GTP was required at positions T-12, T-10, T-9, T-8, T-7, and T, among other positions (Fig. 1); for this experiment, the G residue at position T was replaced with a T residue (this substitution did not affect termination efficiency) to remove the effect of slowing down the extension from position T-1 to T. When the GTP concentration was lowered from 5 nM (Fig. 3D, lane 7) to 5 pM (lane 10), termination was dramatically reduced from 76 to 28%.
In similar experiments using other template variants, termination was suppressed by lowering the concentration of the NTP required at positions T-11, T-4, and T-3 (data not shown). Accordingly, slowing down the elongation at positions from T-13 to T-3 resulted in enhanced read-through at position T rather than termination.
Ternary Complexes at Position T-3 or Upstream Are Stable-Two explanations can be conceived for the finding that upstream pausing facilitates read-through rather than termination. The first is that termination-prone complexes are selectively destabilized during the upstream pause, resulting in increased relative amounts of read-through complexes. The second is that the conformation of the nascent RNA in the complex is changed to a read-through-favorable form during the upstream pause.
To test the first possibility, the stability of stalled ECs was examined as described previously (4). ECs stalled at a position from T-7 to T-1 were obtained using a modified template with a broken oligo(T) sequence, and aliquots of the reaction were incubated at room temperature for 0 -120 min. The ECs, which contained biotinylated DNA, were purified using streptavidincoated magnetic beads, and their radioactive RNA transcripts were measured after gel electrophoresis (Fig. 4A). When RNA transcripts dissociate from ECs or when binary complexes of polymerase-RNA are released from biotinylated DNA during the incubation, a reduction in radioactivity will be seen.
In this assay, the stability of diverse ECs was measured in a time course manner. A dramatic reduction in radioactive RNA was observed with ECs isolated after stalling at position T-2 or T-1; after 20 min of incubation, only ϳ20% remained (Fig. 4B). In contrast, ECs collected after stalling at position T-3 or upstream were stable. For example, 77% of the T-3 ECs remained even after 2 h of incubation (Fig. 4B). Accordingly, the enhanced read-through after pausing at position T-3 (or any position between T-12 and T-3) was probably caused by structural change rather than by dissociation of ECs.
The Upstream Pause Allows for Formation of an Antiterminator RNA Structure-To uncover the structural difference in the EC related to the two pauses at positions T and T-3 that exert opposite effects on T, the RNA secondary structures were estimated by the mfold program for each pause complex (20). We assumed that 9 residues at the 3Ј-end (from positions T-8 to T) cannot participate in the formation of a secondary structure because 7 or 8 residues at the 3Ј-end are engaged in an RNA-DNA hybrid (21,22), and the 9th residue from the 3Ј-end is in contact with T7 RNA polymerase (23,24). A stable terminator hairpin structure with an 11-bp stem (⌬G ϭ Ϫ14.8 kcal/ mol) was estimated for the T pause complex (Fig. 5A).
In the T-3 pause complex, an alternative secondary structure (⌬G ϭ Ϫ17.4 kcal/mol) (Fig. 5B) was predicted, and the 5Ј-half of the terminator hairpin-forming sequence hybridized with a farther upstream sequence. This structure is incapable of inducing termination because it is located 16 nucleotides distant from oligo(U). If this antiterminator RNA structure persists at position T, the terminator hairpin close to oligo(U) cannot be formed efficiently, and antitermination would occur at T.
The prediction of RNA secondary structures was checked by enzymatic probing assays using three structure-specific endonucleases. Two RNA molecules radioactively labeled at the 5Ј-end were synthesized in multiround transcription reactions using [␥-32 P]GTP to represent the terminator and antiterminator RNA structures formed in nascent transcripts. The terminator RNA was produced to match the 83-nucleotide sequence from positions T-91 to T-9, and the antiterminator RNA to match the 80-nucleotide sequence from positions T-91 to T-12.
The radioactive transcripts were subjected to partial digestion with RNases A, T1 and V1 separately. Cleavage sites revealed by gel electrophoresis (Fig. 5, C and D) were consistent with the expectations made from the predicted secondary structures, except for cleavage between the 2 bp at the extreme bottom of the antiterminator stem by RNase A, which cleaves the 3Ј-side of unpaired U or C residues, likely due to breathing of RNA (Fig. 5, A and B).
The best distinction between the terminator and antiterminator structures was observed with the 5Ј-GGGU-3Ј sequence at positions T-19 to T-16 (boldface in Fig. 5, A and B), the base pairing of which has been demonstrated to be essential for termination (11). It was cleaved well by RNase T1, which cleaves the 3Ј-side of unpaired G residues, and by RNase A but not by RNase V1 in the antiterminator RNA (Fig. 5B). The same sequence was not cleaved by RNase A or T1 but was cleaved by RNase V1 in the terminator RNA (Fig. 5A), although RNase V1 cleaves stacked unpaired residues as well as paired residues at the 5Ј-side (25). to T-1 of the biotinylated DNA template were divided into eight aliquots and incubated for varied amounts of time (0 -120 min) at room temperature, and stable complexes were purified using streptavidin-coated magnetic beads for analysis on denaturing 12% polyacrylamide gels. Experiments for T-7 and T-1 stalling used the wild-type T template. Experiments for stalling at positions of the T residue stretch from T-6 to T-2 required modified templates in which the T residue next to the stalling position was replaced with an A residue. B, quantitative analysis of the data in A. The relative amounts of RNA retained in the complexes (y axis) after stalling at various positions are plotted against the incubation time in minutes (x axis).

RNA Structure Responsible for cis-Acting Antitermination-
To examine the hypothesis that stable antiterminator RNA structure at position T-3 causes antitermination, termination efficiency was measured under conditions in which formation of the antiterminator structure was blocked or made inefficient. Incorporation of IMP into RNA transcripts at G residue positions destabilizes RNA secondary structures by reducing the number of hydrogen bonds (26). IMP incorporation at 12 G residue positions from T-66 to T-29 (circled in Fig. 5B) would disrupt the antiterminator RNA structure of Fig. 5B, allowing the terminator hairpin structure of Fig. 5A to be formed.
When ECs were stalled at position T-3 after IMP incorporation at the 12 positions, termination efficiency was consistently high regardless of how long the complexes had been stalled (Fig. 6A), unlike in the case of normal GMP incorporation (Fig. 3A). When ECs were stalled briefly at other upstream positions, termination efficiency was high (Fig.   6B), unlike in the GMP incorporation case (Fig. 3B). Thus, antitermination was not observed when the secondary structure shown in Fig. 5B could not be formed. Accordingly, the FIGURE 5. Terminator and antiterminator RNA structures. A, a putative terminator RNA secondary structure representing the most stable one for the T pause complex predicted by the mfold program. Nine nucleotides at the 3Ј-end were excluded from formation of secondary structures because 7 or 8 nucleotides (dashed box) are secluded in the RNA-DNA hybrid (21,22), and the 9th residue (solid box) from the 3Ј-end makes contact with T7 RNA polymerase (23,24). The cleavage sites of RNases A (dashed arrows), T1 (solid arrows), and V1 (arrowheads) are shown with arrows or arrowheads, the length of which is roughly proportional to the cleavage efficiency. B, a putative antiterminator RNA structure for the T-3 pause complex as in A. C, enzymatic probing of the terminator RNA structure. Radioactively end-labeled RNA transcripts with the sequence from positions T-91 to T-9 were renatured and partially digested with RNases A, T1, and V1. G ladders were produced by treating denatured RNA with RNase T1 to provide RNA size markers, along with RNA ladders produced by incubation with 50 mM NaOH (OH lanes). Reactions without any RNase were performed as a negative control (Ϫ lanes). D, enzymatic probing of the antiterminator RNA structure. RNA transcripts with the sequence from positions T-91 to T-12 were used as in C. antiterminator RNA structure formed with the sequence from positions T-68 to T-50 was responsible for the cisacting antitermination activity of T.
Stability of Antiterminator Structure and Pause Duration Determine Antitermination Strength-To verify the hypothesis that the termination-interfering RNA secondary structure is responsible for the cis-acting antitermination of T, the correlation between stability of the antiterminator structure and antitermination strength was examined. Several antiterminator structures of varied ⌬G values were designed by substitution mutations (Fig. 7A). First, the A and T residues at positions T-62 and T-61 were exchanged with G residues, generating a more stable structure (S-AT RNA) with ⌬G ϭ Ϫ26.5 kcal/mol. Second, two G residues at positions T-64 and T-63 were replaced with A residues, producing a less stable structure (W-AT RNA) with ⌬G ϭ Ϫ10.3 kcal/mol.
When termination efficiencies were measured at 0.1 mM NTP using these modified templates, there was a positive linear correlation (r ϭ 0.99) between termination efficiency and the estimated ⌬G value of the RNA secondary structure (Fig. 7B,  solid line). Thus, antitermination strength depended on the thermodynamic stability of the antiterminator RNA secondary structure.
A linear correlation (r ϭ 0.99) was also observed when transcription reactions were carried out at 50-fold higher NTP concentration, 5 mM (Fig. 7B, dotted line). The antitermination strength was lower in 5 mM NTP than in 0.1 mM NTP; this can be explained by the fact that because the T-3 pause is shortened at higher NTP concentrations, there is insufficient time for formation of antiterminator structure. Accordingly, for cis-acting antitermination to occur, the pause must be long enough, and the antiterminator RNA structure must be sufficiently stable.

DISCUSSION
In this study, we found for the first time that in addition to facilitating termination by causing a pause at the termination site, RNA oligo(U) can induce antitermination on an RNA hairpin-dependent intrinsic terminator (class I) by provoking a transcription pause at a position upstream of the termination site. Because the antitermination pause precedes the termination pause, termination efficiency would be expected to depend primarily on the antitermination strength. This phenomenon could explain, at least in part, why termination efficiency is not strictly correlated with stability of the terminator hairpin and is affected by some changes in the upstream nucleotide sequence outside of a class I termination signal (27,28).
Transcription pausing can be induced or modulated by various signals, including RNA secondary structure, regulatory proteins, backtracking of ECs, certain downstream DNA sequences, and others (6,29). The RNA oligo(U) portion of class I termination signals has been well documented to make a transcription pause at the termination site (position T), triggering a termination process. The termination pause allows formation of an RNA hairpin that shortens or destabilizes the RNA-DNA hybrid and/or alters RNA polymerase conformation to facilitate release of RNA (5,8). It was confirmed in this study that the termination pause is caused by oligo(U) rather than by hairpin formation in terminator T RNA ( Fig. 2A) and that strengthening the pause at position T stimulates termination (Fig. 3C).
In this study, we also found that the same oligo(U) can induce another pause 3 bp upstream, i.e. at position T-3 ( Fig. 2A), and that the consequence of this upstream pause is opposite, stimulating read-through rather than termination at position T (Fig.  3A). This intrinsic antitermination, which requires no external factor, is presumably rendered by formation of an antiterminator RNA structure in which a farther upstream sequence sequesters the 5Ј-half of the terminator hairpin-forming sequence (Fig. 5B) to prevent formation of the terminator hairpin structure.
In this mode of intrinsic antitermination, the antitermination strength and the resulting termination efficiency will be determined by the formation and persistence of antiterminator secondary structure(s) over the terminator hairpin across position T. The persistence apparently depends on two parameters. First, the antitermination efficiency was positively linearly correlated with the stability (⌬G values) of antiterminator RNA secondary structure (Fig. 6B). Second, increasing the pause duration at position T-3 enhanced the antitermination efficiency (Fig. 3C), whereas decreasing the pause strength at a higher NTP concentration reduced the antitermination efficiency (Fig. 7B). Accordingly, both a sufficiently long pause and stable antiterminator RNA structure are required for the pausemediated cis-acting antitermination.
In this study, we did not examine whether this pause-mediated antitermination has any biological relevance or even whether it is effective in vivo, other than the fact that ϳ195 nucleotides upstream of this terminator are not translated, and no interference of ribosome is expected. However, this question must be considered because a transcription pause has been observed a little ahead of the termination site with many intrinsic terminators in vitro, and such pauses have been presumed to associate with termination events (30,31).
For example, an E. coli RNA polymerase transcription pause occurring 2 bp upstream of the termination site (position T-2) has been observed with the phage tR2 intrinsic terminator at low ATP concentrations (32); the biochemical properties of this pause complex have been regarded as part of the intrinsic termination mechanism (5). However, like the T-3 pause complex of T described in this study, the T-2 pause complex of tR2 would be more likely to result in read-through rather than termination, although the former was with T7 RNA polymerase and the latter with E. coli RNA polymerase. This is supported by a previous finding that tR2 termination efficiency was reduced at lower ATP concentrations (32). The ⌬G value for the tR2 terminator hairpin is estimated to be Ϫ10.1 kcal/mol (Fig. 8A) by the mfold program. In the T-2 pause complex with an 8-bp RNA-DNA hybrid (33), however, the hairpin would be shorter, with ⌬G ϭ Ϫ3.7 kcal/mol (Fig.  8B), which is much less stable than an alternatively formed antiterminator structure with ⌬G ϭ Ϫ10.1 kcal/mol (Fig. 8C). Thus, a pause upstream of the termination site should not be presumed to associate with termination without further characterization.
This intrinsic antitermination activity mediated by transcription pausing is distinct from the recently reported antitermination activity that is exerted by tiny abortive initiation transcripts (11), although both antitermination mechanisms were found for the T7 T terminator and T7 RNA polymerase. Antiterminator abortive transcripts are produced from some T7 promoters and interfere with formation of the terminator hairpin in a trans-acting manner, and therefore, the trans-antitermination strength increases over multiple rounds of transcription. In contrast, the pause mediates antitermination in a cis-acting manner and is effective even in single-round transcription that is experimentally performed after transcription ternary complexes are separated from all released products, including abortive initiation transcripts. In this study, termination efficiency was measured only in single-round transcription reactions and was therefore not compounded by trans-antitermination.
In summary, an oligo(U) that can induce a pause at the termination site to trigger a termination process on RNA hairpin-dependent intrinsic terminators can induce another pause upstream, which can exert antitermination on the same terminator. Thus, transcription pausing can induce either termination or antitermination depending on the subsequent RNA folding.