A Block of Transcription Elongation by RNA Polymerase I1 at Synthetic Sites in Vitro*

We have previously suggested that transcription elongation by RNA polymerase I1 can be blocked when the nascent RNA is folded into a stem-and-loop structure followed by polyuridines. As an approach to test this suggestion in vitro, several GC-rich deoxyoligonucleotides with dyad symmetries were chemically synthesized and inserted following the adenovirus 2 major late promoter. These constructs were tran- scribed in vitro using HeLa whole cell extract. The transcripts of the synthetic inserts can potentially form stem-and-loop structures with destabilization energy from 0 to -48 kcal followed by 3, 5, and 8 U residues. The results obtained show that transcription elongation is blocked by these synthetic inserts and that the extent of the elongation block is directly correlated to the stabilities of the potential stem-and-loop struc- ture and the proceeding number of U residues. Three levels of elongation blocks were observed: a brief pause of the polymerase occurs when the RNA could be folded into a secondary structure or when there were 5-6 T residues on the sense DNA strand. An extended pause occurred when the number of T residues on the sense DNA strand was increased to 8. Transcription termination, with a partial release of the attenuated tran- script occurred when a stable RNA secondary structure (AG = -48 kcal) was followed by 8 U residues. The relevancy of these in vitro results to the in vivo mech- anism of a transcription elongation block is discussed.


A Block of Transcription Elongation by RNA Polymerase I1 at Synthetic Sites in Vitro*
(Received for publication, October 11, 1988)

Eyal Bengal and Yosef AloniS
From the Department of Genetics, The Weizmann Institute of Science, Rehouot, Israel 76100 We have previously suggested that transcription elongation by RNA polymerase I1 can be blocked when the nascent RNA is folded into a stem-and-loop structure followed by polyuridines. As an approach to test this suggestion in vitro, several GC-rich deoxyoligonucleotides with dyad symmetries were chemically synthesized and inserted following the adenovirus 2 major late promoter. These constructs were transcribed in vitro using HeLa whole cell extract. The transcripts of the synthetic inserts can potentially form stem-and-loop structures with destabilization energy from 0 to -48 kcal followed by 3, 5, and 8 U residues.
The results obtained show that transcription elongation is blocked by these synthetic inserts and that the extent of the elongation block is directly correlated to the stabilities of the potential stem-and-loop structure and the proceeding number of U residues. Three levels of elongation blocks were observed: a brief pause of the polymerase occurs when the RNA could be folded into a secondary structure or when there were 5-6 T residues on the sense DNA strand. An extended pause occurred when the number of T residues on the sense DNA strand was increased to 8. Transcription termination, with a partial release of the attenuated transcript occurred when a stable RNA secondary structure (AG = -48 kcal) was followed by 8 U residues. The relevancy of these in vitro results to the in vivo mechanism of a transcription elongation block is discussed.
Transcription termination plays an important role in regulating gene expression in bacteria (1)(2)(3). The elements that are required to form a bacterial terminator are well defined. They fall into two major classes: factor-dependent and factorindependent terminators. Factor-dependent termination usually occurs at multiple sites and no common sequence has been identified. Factor-independent termination is characterized by a GC-rich stretch of DNA with dyad symmetry followed by T residues in the sense DNA strand (for a review see Ref. 3). A wide variety of experimental results support the hypothesis that the GC-rich sequence in the DNA that enables the RNA transcript to fold into a stem-and-loop structure impedes the progress of the prokaryotic RNA polymerase, and then the instability of the dA:rU base pairing between *This work was supported by Grant CA 14995 from the United States Public Health Service and by grants from the Minerva Foundation, Munich, West Germany, the United States-Israel Binational Science Foundation (BSF), Jerusalem, Israel, the Leo Forcheimer Center for Molecular Genetics, The Henry Gutwirth Fund, and The Rockefeller-WIF Collaboration Trust Fund. 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.
$ To whom correspondence should he addressed.
the DNA template and the RNA transcript facilitates the release of the transcript from the template (4,5). As in prokaryotes, RNA polymerase I1 in eukaryotes is capable of prematurely terminating transcription within a gene. This mechanism has been termed a block of transcription elongation or attenuation and was shown to occur within the leader region of the yeast LEU 2 gene (6), in the long terminal repeat of HIV-1 (7), in the c-myc gene , in the c-fos gene (11), and in the c-myb gene (12). In addition, premature transcription termination a t precise sites was shown to occur both i n vitro in SV40 (13-19) and i n vivo and i n vitro in the parvovirus minute virus of mice (MVM)' (20, 21) and in the adenovirus 2 (Ad2) (22)(23)(24). We have suggested that the termination signal, at least in the viral systems, includes in addition to a run of U residues at the 3' of the terminated RNA, an upstream stem-and-loop structure. This would suggest that the prokaryotic polymerase and the eukaryotic polymerase I1 can respond to a similar transcription termination signal (16,19).
To test this suggestion directly we have inserted several GC-rich synthetic deoxyoligonucleotides that have dyad symmetries and are proceeded by various numbers of T residues in the sense DNA strand, following the Ad2 major late promoter (MLP). These constructs were transcribed in the HeLa whole cell extract system (WCE) (25). The results obtained show that transcription elongation is blocked by these synthetic inserts and that the extent of the elongation block is directly correlated to the stability of the stem-and-loop structure and the number of U residues proceeding them.
DNA Template Preparation-Three pairs of complementary oligomers of 21, 23, and 26 nucleotides (a total of six oligomers) (see Fig. 1) were synthesized on DNA synthesizer model 380B using the phosphoramidite method (Applied Biosystems, Inc.). The full length products were purified from a 20% polyacrylamide nondenaturing gel. Complementary oligomers were annealed at 80 "C followed by slow cooling to room temperature. Unphosphorylated double-stranded deoxyoligomers were blunt-end-ligated into the StuI site of pSV2CAT plasmid DNA. To delete a part of the inverted repeat sequences, plasmid DNAs were digested with SmaI, which cut twice at the inserted oligomers. The digested DNAs were then gel-purified and religated. Positive clones which contained a 10-bp deletion of the expected fragment were selected by sequencing. To further stabilize the original potential secondary structure, a 12-bp EcoRI linker, 5'-GGCCTTAAGGCC-3', (New England, BioLabs) was inserted at the unique EcoRV site of the inserted deoxyoligomers. Each clone of the original constructs was cut with either NcoI-Hind111 or BglI-Hind111 to produce two fragments that were inserted into the polylinker of the pGEM-MLP downstream from the Ad2-MLP. The target vector pGEM-MLP is a pGEM-1, containing a 450bp SacI-ScaI (5634-6083) fragment of Ad2 inserted into the unique SacI-SmaI sites of pGEM-1. For this, the BglI was first blunt ended by T4 polymerase, and BglI-Hind111 was inserted at the HincII-HindIII-digested plasmid DNA. To insert the NcoI-Hind111 fragment, an XbaI-NcoI adaptor was first inserted into the XbaI site of the polylinker of pGEM-MLP, thus creating a unique NcoI site.
In order to map the attenuated transcripts, a 14-bp XbaI-ClaI adaptor, 5'-CTAGGCATCGATGC-3', that was synthesized and purified as described above, was inserted at the unique XbaI site of the polylinker of all pGEM-MLP-TER constructs.
Plasmid DNA was prepared by the alkali lysis method (26), and purified by CsCl density gradients.
I n Vitro Transcription Using HeLa Whale Cell Extract-WCE was prepared according to Manley (27). A 20-p1 reaction contained 10 pl of WCE, -1 pg of DNA, 4 mM creatine phosphate, 500 p~ ATP, GTP, UTP, 50 p~ CTP, and 20 pCi of [LI-~'P]CTP, and transcription was performed a t 30 "C. All reactions were preincubated 20 min before the addition of the nucleotide mixture. Sarkosyl or heparin was added 45 s after the addition of nucleotides. In the experiments where UTP was used as the labeled nucleotide, the reaction mixtures contained 500 p~ ATP, GTP, CTP, 50 p~ UTP, and 20 pCi of [CY-~~PIUTP.
Labeling under pulse-chase conditions was done as follows: the pulse labeling was for 1 min and the mixture contained 500 g~ ATP, GTP, UTP, and 20 pCi of [a-"PICTP. For the chase, SarkosylO.1-0.2% or heparin, 1 mg/ml, was added followed 45 s later, by 500 p M of unlabeled CTP. The reactions were allowed to proceed for the times indicated, stopped by the addition of proteinase K buffer (50 mM Tris-HC1, p H 7.5, 10 mM EDTA, 0.2% sodium dodecyl sulfate, 10 mM NaCl), 200 pg/ml proteinase K, and 30 pg of tRNA, and incubated for an additional 15 min. RNA was phenol/chloroformextracted and collected by ethanol precipitation and centrifugation.
RNase Mapping-[[3ZP]RNA probes were synthesized according to Melton et al. (28). The DNA template was removed by treating the mixture with 10 units of RNase free DNase I (Worthington) for 15 the RNA was purified by a quick spin through a Sephadex G-25 min. The mixture was extracted twice with phenol/chloroform, and column followed by ethanol-precipitation and centrifugation. The RNA probe (50,000 cpm) was hybridized with 30 pg of unlabeled RNA in 30 p1 containing 0.4 M NaCl, 40 mM PIPES, pH 6.7, 1 mM EDTA, and 80% formamide. The RNA was heat-denatured at 85 "C for 5 min and then transferred to 45 "C and incubated for an additional 3 h. The mixture was then diluted 10-fold with 0.3 M NaC1, 10 mM Tris-HC1, pH 7.5, 5 mM EDTA, 40 pg/ml RNase A, 20 units/ml T1 RNase and incubated for 1 h at 30 "C. The reaction was stopped by the addition of 0.4% sodium dodecyl sulfate and 50 pg/ml proteinase K and incubated for an additional 15 min. Nucleic acids were phenol/chloroform-extracted followed by a quick spin through a Sephadex G-25 column. Nucleic acids were collected by ethanol precipitation and centrifugation.
Polyacryhmide Gel Electrophoresis-RNA was resuspended in 10 pl of 90% formamide and analyzed on 6% polyacrylamide gels (bis/ acrylamide, 1:19) containing 7 M urea (Schwarz/Mann) and 1 X T B E (TBE: 89 mM Tris borate, pH 8.3, 89 mM boric acid, 2 mM EDTA), Electrophoresis was carried out a t a constant current of 20 mA. The constructs that have been tested for their ability to block transcription elongation can potentially form RNA secondary structures with the following destabilization energy and number of consecutive U residues immediately following the RNA secondary structures:

Construction
-48 kcal/8 U residues, -48 kcal/5 U residues, -48 kcal/3 U residues, -23 kcal/8 U residues, -23 kcal/5 U residues, 0 kcal/8 U residues. Note that because the inserted deoxyoligonucleotides are of variable lengths the distance between the transcription start site and the elongation block site, if it occurs immediately following the RNA secondary structures, can vary within a few nucleotides. Fig. 2 shows schematics of two constructs that were tested in the present study. The two constructs are almost identical and they differ only by the length of the SV40 DNA fragment into which the synthetic deoxyoligonucleotides were inserted: BglI-Hind111 in Fig. 2A as compared to NcoI-Hind111 in Fig. 2 B (see under "Materials and Methods" for details). The plasmids were linearized by NaeI or NheI. The lengths of the expected runoff and attenuated transcripts are also indicated in Fig. 2. Note that the distance between the transcription start site and the potential elongation block site is different between the two plasmids. Since preliminary results showed no distance effect on the elongation block (results not shown) the constructs were designated only by the stability of the potential RNA second-  Fig. 1 and "Materials and Methods." The functional promoter is the MLP of Ad2 (5634-6083, Ad2 numbering, black box). This is followed by the BglI-HindIII(5235-5171) in A or NcoI-Hind111 (37-5171) in B fragments of SV40 (SV40 numbering, hatched box).
For the experiments described below, the templates were digested with either NaeI or NheI as indicated. The expected lengths (in nucleotides ( n t ) ) of the runoff and attenuated RNA (Att. RNA) are indicated.
ary structure of the transcripts and the number of proceeding U residues. For example, a Construct that upon transcription would yield RNA with a potential secondary structure of AG = -48 kcal followed by 8 uridine residues is termed -48 kcal/ 8 U residues.
The Potential R N A Secondary Structure and the Run of U Residues Contribute to the Efficiency of the Elongation Block-In the following experiments moderate concentrations of Sarkosyl or heparin were included in the WCE system. The inclusion of these chemicals in the WCE system inhibits transcription initiation but not elongation by eukaryotic RNA polymerase I1 (29, 30). Moreover, in the WCE transcription system Sarkosyl and heparin cause pausing of polymerase I1 at the attenuation sites of Ad2 (24,30), MVM (21), and SV40.' Termination at two of these sites has also been shown to occur in vivo (21,23,24), which indicates that the Sarkosyl/ heparin WCE in vitro system accurately reflects the in vivo situation a t least concerning termination within a gene.
In order to test the effect of the potential RNA secondary structure and the stretch of U residues on the production of the attenuated RNA, we compared the different constructs in the in vitro transcription assay. Fig. 3A shows the results of the kinetics of a transcription reaction using the -48 kcal/3 U residues construct. A short RNA band of about 135 nucleotides is recognized at the first 1-2 min of incubation, and the band disappears after prolonged incubation. The 3' end of the 135 nucleotides RNA maps within the stem-and-loop structure. We suggest that the stem-and-loop structure of about -48 kcal could lead the polymerase to pause briefly and reversibly at this site on the DNA template.
We next compared transcription from two templates that differ only in the stability of their potential RNA secondary structure: -23 kcal and -48 kcal (Fig. 3, B and C). As the incubation time with the -23 kcal/5 U residues construct increases the attenuated RNA (of about 175 nucleotides), gradually disappears and it is completely absent after 30 min of incubation with a high CTP concentration (Fig. 3B). In contrast, as the incubation time with the -48 kcal/5 U residues construct increases the intensity of the band corresponding to the attenuated RNA, of about 187 nucleotides is not changed and it remains with almost the same intensity even following a 30-min incubation with a high CTP concentration (Fig. 3C). It appears that when the signal includes 5 consecutive uridine residues, the potential secondary structure has a prominent effect on the elongation block.
The synergistic nature of the secondary structure and the stretch of U residues can further be seen in Fig. 3, D and E , that show a comparison of transcription kinetics of two constructs: -23 kcal/8 U residues and -48 kcal/8 U residues. With both constructs an attenuated RNA is recognizable at all incubation times, including following a chase for 30 min with a high CTP concentration. However, the efficiencies of the elongation blocks, as reflected by the intensity of the attenuated RNAs (filled arrowheads), as compared to the runoffs (open arrowheads) clearly shows that it is higher with the -48 kcal/8 U residues construct as compared with the -23 kcal/8 U residues construct. Scanning the autoradiogram for determining the ratio of the attenuated to the runoff RNA ( Fig. 3, D and E, lunes 1+30) revealed that the -48 kcal/8 U residues signal blocked elongation three times more effectively than the -23 kcal/8 U residues signal. It is also evident that the increase in the number of U residues from 5 to 8 (compare Fig. 3, B and D) without changing the potential stem-and-loop structure (-23 kcal) had a pronounced effect on the efficiency of the elongation block. Thus, whereas, with the -23 kcal/5 U residues construct the attenuated RNA is chased into runoff transcript (Fig. 3B) with the -23 kcal/8 U residues construct the attenuated RNA is recognized also following the chase reaction (Fig. 3 0 ) . We therefore conclude that both the potential stemand-loop structure and the run of U residues contribute to the efficiency of the elongation block.
Mapping of the Attenuated R N A Transcripts and Confirmation of the Existence of R N A Secondary Structure-In order to map the attenuated RNA a fragment of 14 base pairs was inserted in all templates between the transcription start site and the site of the elongation block (see under "Materials and Methods"). If the attenuated RNA originates from the Ad2-MLP, in the extended construct both the runoff and the attenuated transcripts bands should shift upward by the corresponding 14 nucleotides. Fig. 4C shows that the expected shift of the attenuated and runoff bands is clearly seen. The attenuated RNA therefore originates from the Ad2 MLP and terminates immediately following the RNA secondary structure. The expected 14-nucleotide shift was observed with the other constructs used throughout this study.
We also attempted to map the in vitro synthesized RNA directly using RNase mapping (28). For the probe synthesized by SP6 polymerase (see Fig. 1 We suspect that this is due to the fact that both the probe and the RNA to be mapped are in a strong secondary structure conformation and thus only partially hybridized. The 103-107-and 90-nucleotide bands could result if the labeled RNA probe hybridizes to the RNA to be mapped as is shown in Fig.  4B and the RNase nicks this structure within the loop and at the base of the stem. In a control experiment, runoff transcripts with no potential secondary structure fully protected the labeled riboprobe and thus allowed mapping of the runoff RNA (not shown). These results support the presence of a secondary structure in the RNA.
The Efficiency of the Elongation Block Can Be Modulated by the Number of T Residues ut the Block Site-An examination of the sequence on the sense strand in the vicinity of the inserted synthetic deoxyoligonucleotides revealed additional T-rich sequences (see Fig. 5C where the T sequences are represented as U residues in the RNA transcript). In addition to the 3 T, 5 T, and 8 T residues that are part of the inserted sequences, there are stretches of 6 T residues upstream and 4 T residues downstream of the inserted synthetic sequences (in the SV40 sequences). Grass et ul. (31) using a similar system found several termination sites in the SV40 sequences including the sites of the 6 and 4 U residues (see Fig. 5C). In order to examine directly the effect of stretch of U residues on termination, we transcribed a construct which had no secondary structure inserted, only an insert containing 8 U residues. In Fig. 5A, the 148-and 183-nucleotide bands represent the location of an RNA transcript which reached the 6 or 4 consecutive uridines (from the SV40 sequence) in the transcript (filled arrowheads), while the 168-nucleotide band represents a transcript which reached the 8 U residues of the insert. It is apparent that in the present in uitro system, a stretch of 6 consecutive uridines led to a brief pause of the polymerase which was chased after 5 min, whereas a stretch of 4 consecutive uridine residues did not cause a noticeable pausing of the polymerase. However, 8 consecutive uridine residues had a more pronounced effect. Notice that although the intensity of the 168-nucleotide band decreases gradually with the incubation time, it is still recognizable even after 50 min of incubation.
We were able to enhance the accumulation of the RNA a t the 8 U residues by adding RNA secondary structure of -48 kcal (Fig. 5B). The 188-nucleotide transcript, which represents an RNA with a 3' terminus within the 8 U residues, is present a t all the time points examined. Again, a brief pause at the 6 U residues led to an RNA band of 148 nucleotides. A comparison of Fig. 5, A and B, shows that, whereas with the 0 kcal/8 U residues, at the 50-min time point <lo% of the radioactivity is present at the attenuated RNA, with the -48 kcal/8 U residues construct >25% of the radioactivity is present in the attenuated RNA, indicating that there is a synergistic effect between the stem-and-loop structure and U residues on the elongation block. The Elongation Block Is Temperature-dependent-Based on the above studies we suggest that premature termination of transcription at the inserted sequence is a two-step event involving first pausing of the polymerase at the block site, followed by the release of the transcript due to the instability of the dAxU interaction (5). If instability of the dA:rU interaction is a dominant part of termination at this site it is expected that as the temperature of the reaction is increased, more attenuated RNA will accumulate. We compared the amount of attenuated RNA transcripts obtained when transcription elongation was carried out a t 30, 37, 40, and 45 "C for 50 min (Fig. 6). The temperature of the reactions was only increased after adding the "chase" solution in order to insure that the extent of preinitiation complex formation and the degree of pulse labeling was the same for each reaction. It is apparent that the transcription temperature has a strong effect on the level of the elongation block. With increased temperature there is a gradual increase in the amount of attenuated RNA. A comparison of the amount of attenuated and runoff transcripts a t 30 and 45 "C shows that the attenuated transcripts account for 25% of the total transcription a t 30 "C as compared to 60% at 45 "C. Interestingly, a similar effect of temperature on the elongation block was observed in C shows the mapping of the attenuated RNA and runoff transcripts by inserting a 14-bp adaptor in the XbaI site of the pGEM-1 polylinker. The polylinker is located between the Ad2 MLP and the inserted SV40 sequences (see Fig. 1). Lane R shows the lengths, in nucleotides, of the transcripts synthesized on the regular -48 kcal/8 U residues construct cut with NaeI (See Fig. M ) , and lane E shows the lengths of the transcripts produced on the extended construct that contains the 14-bp adaptor.
the Ad2 and SV40 systems (32).2 The possibility that the augmentation of the elongation block with increased transcription temperature is the result of a corresponding decrease in activity of the RNA polymerase is excluded by the observation that the rate of transcription elongation at 45 "C is even higher than that at 30 "C (32). 2 A Substantial Fraction of the Nascent RNA Is Released from the -48 kcall8 U Residues Template at the Elongation Block Site-Two approaches were used in order to verify whether the attenuated RNA transcript synthesized on the -48 kcall 8 U residues template results from pausing of the polymerase orland from transcription termination. We first examined whether the elongation block is reversible when the Sarkosyl concentration is diluted 10-fold. This approach was used by Hawley and Roeder (29) and by Kessler et al. (32) in their studies on the elongation block in the Ad2-MLP system. In the Ad2-MLP system, dilution of the Sarkosyl caused an almost complete chase of the attenuated RNA into the runoff transcript (29, 32). Fig. 7A (middle lane) shows that dilution of the Sarkosyl concentration from 0.2 to 0.02%, resulted in only a partial reversal of the elongation block at the stretch of 8 U residues. This can be seen by comparing lanes 0.2-0.02 and 0.2-0.2. This experiment does not exclude the possibility that the 0.2% Sarkosyl caused an irreversible pausing of the polymerase rather than transcription termination which is characterized by the release of the nascent RNA from the template.
In the second approach transcription elongation in the presence of 0.1% Sarkosyl was performed on a circular -48 kcall8 U residues template. After 40 min of incubation transcription was stopped by the addition of 10 mM EDTA to the reaction mixture which was then layered onto a 5-30% (w/v) sucrose gradient. After centrifugation, 32 fractions were collected and each fraction was analyzed by gel electrophoresis under denaturing conditions. The attenuated RNA of about 185 nucleotides was found in both the top fractions of the gradient as a free RNA and in the middle of the gradient together with long RNA transcripts where the circular DNA sedimented (Fig. 7B). The transcripts which sedimented with the circular DNA were probably nascent RNA attached to the Comparison of transcription kinetics of constructs containing 8 U residues with and without potential RNA secondary structure preceding them. Preincubation of the constructs 0 kcal/8 U residues (0 kcal/8 UJ in A and -48 kcal/8 U residues (-48 kcal/8 U ) in B linearized with NaeI (see Fig. 2 B ) with WCE, transcription initiation, addition of heparin (1 mg/ml), and transcription elongation were as described under "Materials and Methods." At the indicated times RNA was extracted and analyzed by polyacrylamide gel electrophoresis. C is a schematic showing in the hatched box the inserted synthetic oligonucleotides and in the open boxes the number of consecutive uridine residues in the vicinity of the inserted synthetic oligonucleotides. The numbers above the boxes show the calculated lengths, in nucleotides (nt), of potential transcripts that stop at the first and last U residues. Where two numbers appear, those in parentheses pertain to the transcripts in A and the others to the transcripts in B. The locations, in the gels, of the potential transcripts that stop at the first U residue of the boxes (shown in C ) are indicated by filled arrowheads and the location of the runoff by open arrowheads. transcription complex (TC). In control experiments with constructs whose attenuated RNA is chased into runoff transcripts no free RNA was found at the top of the gradient. We conclude that about 70% of the attenuated RNA from the -48 kcal/8 U residues plasmid is released from the template and represent the level of termination in this system. The remaining 30% is attached to the template and represents pausing of the polymerase.
Dilution of the Reaction during the Elongation Step Enhances the Elongation Block to a Similar Extent as the Presence of Sarkosyl-One explanation for the enhancement of the elongation block by Sarkosyl and heparin is that both chemicals inactivate a readthrough or a stimulating factor. If this putative stimulating factor binds to the RNA polymerase during transcription elongation then dilution of the transcription reaction during the elongation step may reduce the binding of the factor to the polymerase and thus enhance the elongation block at the attenuation site. To investigate this possibility the 0 kcal/8 U residues construct was incubated in vitro for 1 min in the presence of [cx-~'P]CTP followed by the addition of unlabeled CTP (500 p~) to the transcription reaction either in the same volume (22 pl) or in a final volume of 320 pl. For a comparison, transcription reaction was carried out under the same labeling conditions but instead of diluting the reaction following the pulse labeling with [cx-~'P]CTP, Sarkosyl to 0.2% was added during the chase with 500 p~ unlabeled CTP. Indeed, the dilution of the reaction mixture during the elongation step enhanced the synthesis of the attenuated RNA by approximately the same level as did the addition of 0.2% Sarkosyl (compare lanes b in Fig. 8, A and  B). In both cases, the attenuated RNA represents about 50% of the total RNA synthesized. A similar enhancement was observed when the same protocol was used with the other constructs. The attenuated RNA in the dilution protocol remained the same length, while that of the Sarkosyl protocol is about 5 nucleotides longer. At present we do not have an explanation for this phenomenon. Methods." B, preincubation of the uncut -48 kcal/8 U residues construct (see Fig.   3B), transcription initiation, addition of Sarkosyl to 0.1%, and transcription elongation was as described under "Materials and Methods." Transcription was stopped by the addition of 10 mM EDTA and the reaction mixture was layered onto 5-30% (w/v) sucrose gradient containing 50 mM Tris-HC1, pH 7.9,l mM MgC12,l mM dithiothreitol, and 50 mM NaCl. Centrifugation in Beckman SW41 rotor was for 5 h at 32,000 rpm at 4 "C. Fractions were collected and the RNA a t the top two fractions of the gradient (Free RNA) and two fractions sedimenting with the DNA (TC) in the middle of the gradient were analyzed by polyacrylamide gel electrophoresis. The attenuated RNA is indicated by filled arrowhead. DISCUSSION In the present study we have demonstrated that the two elements of the p-independent prokaryotic terminator, i.e. a stem-and-loop structure in the RNA followed by polyuridine, are sufficient to block transcription elongation by the eukaryotic RNA polymerase I1 in the WCE transcription system. Moreover, the extent of the elongation block is directly correlated with the stability of the stem-and-loop structure and 180- the number of U residues proceeding them.

160-
The types of elongation blocks that we observed can be divided into three levels: a brief pause of the transcription complex, an extended pause of the transcription complex, and transcription termination. The brief pause of the transcription complex occurs a t sites where the RNA can potentially fold into a stem-and-loop structure. The brief pause is quickly relieved and full length transcripts are synthesized. It may be caused by a physical constraint imposed on the transcription complex transversing the stem-and-loop structure of the RNA (3). However, the involvement of DNA secondary structure is also possible. The extended pause occurs at a T-rich region in the sense DNA strand. The duration of the extended pause is directly dependent on the number of T residues. In the present in vitro system 6 consecutive T residues caused the transcription complex to pause for about 5 min while 8 consecutive T residues dramatically enhanced the pause of the transcription complex to more than 50 min.
A similar observation was made when the prokaryotic E. coli RNA polymerase was used to transcribe synthetic termination signals in vitro. Although termination occurred at the same site on two templates containing 10 or 20 consecutive T residues, the efficiency of termination was significantly enhanced with the extended T-stretch (33). It appears that the polymerase can respond to sequences downstream of the actual termination site. Similarly, termination within the trp operon is significantly reduced when the number of T residues at the attenuation site are reduced (3435).
That consecutive T residues in the sense DNA strand can lead to transcription termination has been observed in several genes that are transcribed by RNA polymerase 11. In the gastrin gene and in yeast genes a T-rich region in the sense DNA strand is at least a part of the termination signal (36)(37)(38). However, when the T-rich sequence is interrupted by dC or dG following 5 T residues as in the sequence TTTTTGTTTGTTTT, neither pausing nor termination of transcription within this site in vitro was observed (32).
The third elongation block observed in the present study can be defined as true transcription termination because a major fraction of the nascent RNA is released from the template. It was found to occur at a U-rich stretch in the RNA when it follows a stem-and-loop structure of -48 kcal.
Although at first sight the effect of the stem-and-loop structure on the elongation block appears to be minimal (see for example Fig. 3A) it has a pronounced effect when it is followed by a stretch of U residues in the RNA transcript. Thus, for example, 5 U residues following a stem-and-loop structure of -23 kcal lead to a reversible block of elongation while the same number of U residues following a stem-andloop structure of -48 kcal lead to an irreversible block of elongation. It seems, therefore, that the stem-and-loop structure and the stretch of U residues are two elements of a termination signal recognized by the eukaryotic RNA polymerase 11. Thus, there is at least one mechanism for transcription termination shared by prokaryotes and eukaryotes. Indeed the extensive homology among the elongating subunits of the prokaryotic and eukaryotic RNA polymerases (39,40) and the similar features of the transcription bubble propagated during elongation (41), support this conclusion.
We have noted that Sarkosyl and heparin cause a five to tenfold enhancement of the elongation blocks. A similar enhancement was observed when the transcription reaction was diluted 10-15-fold during the elongation step. Our favored explanation for these observations is that the two chemicals and the dilution of the elongation reaction prevent the association of a putative readthrough factor with RNA polymerase. In support of this explanation is the observation that depletion of the stimulatory factor TFIIS from a fractionated HeLa nuclear extract results in a continued pausing of the polymerase 185 nucleotides downstream from the Ad2-MLP. The replacement of the stimulatory factor TFIIS to the extract relieved the pause (42, 43). Binding of the readthrough factor to the polymerase thus enables the polymerase to transverse through possible block sites. In the present in vitro system, the stimulatory factor TFIIS has no effect on the elongation block. However, we have isolated and partially purified a stimulating activity from HeLa and monkey cells that when added following transcription initiation prevented the elongation block and enhanced transcription el~ngation.~ That Sarkosyl or heparin enhance the elongation blocks at true termination sites is supported by the observations that with Ad2 and MVM systems the attenuated RNAs were identified and their 3' ends were mapped in vivo to the same sites where Sarkosyl or heparin enhance the elongation blocks in vitro (21,24).