Transcription elongation factor SII (TFIIS) enables RNA polymerase II to elongate through a block to transcription in a human gene in vitro.

Elongation and termination by RNA polymerase II are important regulatory steps for eukaryotic gene expression. We have previously studied the transcription of linear DNA templates where specific initiation of transcription by highly purified RNA polymerase II can be achieved in the absence of promoters and promoter-specific factors. Using these templates we have shown that a human histone gene, H3.3, contains sequences (intrinsic terminators) within which purified RNA polymerase II will efficiently terminate transcription (Reines, D., Wells, D., Chamberlin, M.J., and Kane, C. M. (1987) J. Mol. Biol. 196, 299-312). Curiously, these signals were found within an intron, 3'-untranslated, and protein-encoding regions of the gene suggesting that they might act to attenuate transcription of H3.3 in vivo. Here we show that intrinsic terminator sequences from an H3.3 gene intron also block in vitro transcript elongation by RNA polymerase II when the enzyme has initiated transcription from a promoter using highly purified transcription initiation factors. However, under the conditions used for promoter-specific transcription there is little transcript release. Instead the polymerase can pause at these sites for periods exceeding 60 min. We have identified and partially purified an activity from HeLa cells that causes the transcription complex to read through this block to transcription elongation. This readthrough activity fractionates with a previously characterized elongation factor (SII) over three chromatographic columns. A homogeneous preparation of calf thymus SII can also provide this activity in trans. This factor may facilitate passage of the RNA polymerase II transcription complex through such intragenic sites in cellular genes in vivo.

Transcription Elongation Factor SI1 (TFIIS) Enables RNA Polymerase I1 to Elongate through a Block to Transcription in a Human Gene in Vitro* (Received for publication, January 10, 1989) Daniel Reines, Michael J. Chamberlin, and Caroline M. Kane

From the Department of Biochemistry, University of California, Berkeley, California 94720
Elongation and termination by RNA polymerase I1 are important regulatory steps for eukaryotic gene expression. We have previously studied the transcription of linear DNA templates where specific initiation of transcription by highly purified RNA polymerase I1 can be achieved in the absence of promoters and promoter-specific factors. Using these templates we have shown that a human histone gene, H3.3, contains sequences (intrinsic terminators) within which purified RNA polymerase I1 will efficiently terminate transcription (Reines, D., Wells, D., Chamberlin, M. J., and  J. Mol. Biol. 196,[299][300][301][302][303][304][305][306][307][308][309][310][311][312]. Curiously, these signals were found within an intron, 3'untranslated, and protein-encoding regions of the gene suggesting that they might act to attenuate transcription of H3.3 in vivo. Here we show that intrinsic terminator sequences from an H3.3 gene intron also block in vitro transcript elongation by RNA polymerase I1 when the enzyme has initiated transcription from a promoter using highly purified transcription initiation factors. However, under the conditions used for promoter-specific transcription there is little transcript release. Instead the polymerase can pause at these sites for periods exceeding 60 min. We have identified and partially purified an activity from HeLa cells that causes the transcription complex to read through this block to transcription elongation. This readthrough activity fractionates with a previously characterized elongation factor (SII) over three chromatographic columns. A homogeneous preparation of calf thymus SI1 can also provide this activity in trans. This factor may facilitate passage of the RNA polymerase I1 transcription complex through such intragenic sites in cellular genes in vivo.
The completion of a primary transcript by RNA polymerase I1 is essential for gene expression. A growing number of genes transcribed by RNA polymerase I1 are regulated, a t least in part, at the level of transcript elongation (Bentley and Groudine, 1986;Eick and Bornkamm, 1986;Nepveu and Marcu, 1986;McGeady et al., 1986;Mechti et al., 1986;Bender et al., 1987;Fort et al., 1987;Kao et al., 1987;McCachren et al., 1988;Watson, 1988;Bhat and Padmanaban, 1988). RNA polymerase I1 stops transcription within these genes before * This work was supported by National Institutes of Health Postdoctoral Fellowship 1F32-GM11296-01 (to D. R.) and National Institutes of Health Grant 5R01-GM34963 (to M. J. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement'' in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
completing the primary transcript; in many cases the ability of polymerase to read through the elongation block is conditional. The biochemical basis for this type of regulation is not understood.
One valuable approach to understanding the regulatory mechanisms that govern gene expression involves in uitro reconstitution of specific transcription using purified proteins. This approach has been used successfully for dissecting prokaryotic gene regulation (reviewed in Platt, 1986;Hoopes and McClure, 1987;and Yager and von Hippel, 1987). Advances in identifying protein factors required for transcription initiation at promoter sites have been an important step toward this goal in eukaryotes (Samuels and Sharp, 1986;Reinberg and Roeder, 1987a;Reinberg et al., 1987;Conaway et al., 1987;Zheng et al., 1987;Conaway and Conaway, 1989;Flores et al., 1988;Buratowski et al., 1988;Cavallini et al., 1988). However, it seems likely that, as is the case for prokaryotic transcription, multiple factors will also be needed for elongation and termination.
In our previous studies of transcription elongation and termination we have used an alternative template system that allows highly purified RNA polymerase I1 to initiate transcription in the absence of any other cellular proteins (Kadesch and Chamberlin, 1982;Dedrick and Chamberlin, 1985;Kane and Chamberlin, 1985;Dedrick et al., 1987;Reines et al., 1987;Kane, 1988;Kerppola and Kane, 1988). Such transcription elongation must involve only the polypeptides present in the RNA polymerase itself. With this system we have identified a number of sites recognized by the purified polymerase as potent intrinsic transcription terminators. At these terminators the purified enzyme stops RNA chain elongation and releases the newly synthesized RNA products. We SI1 is an elongation factor that stimulates transcription by RNA polymerase I1 and was first purified from mouse cells by Natori and co-workers (reviewed by Natori, 1982). Similar proteins have also been identified in other species including humans (TFIIS; Reinberg and Roeder, 1987b). For the purposes of this report we will refer to these proteins collectively as SII.
were initially surprised to find that these sites occurred within the body of a human histone gene (H3.3; Reines et al., 1987). T h e strength and location of these intrinsic termination sites suggested that RNA polymerase I1 required accessory elongation factors in transcribing full-length transcripts or that most transcribing RNA polymerase I1 molecules fail to complete a full-length transcript. Several observations favored the notion of a n accessory elongation factor. Transient transfection experiments suggested to us that these sequences can be read through in vivo.' Furthermore, transcription of this: as well as other genes, in isolated nuclei also implies that such factors exist and are important for regulating eukaryotic gene expression (Bentley and Groudine, 1986;Eick and Bornkamm, 1986;Nepveu and Marcu, 1986;McGeady et al., 1986;Mechti et al., 1986;Bender et al., 1987;Fort et al., 1987;Kao et al., 1987;McCachren et al., 1988;Watson 1988;Bhat and Padmanaban, 1988).
We report here on the properties of a promoter-initiated RNA polymerase I1 transcription complex that efficiently stops RNA chain elongation in response to an intragenic signal within a human histone gene. Furthermore, we can promote readthrough by this complex when the reactions are supplemented with fractions from human cells or with a homogeneous preparation of an elongation factor (SII) from calf thymus.

Plasmids and Templates
The plasmid pAdTerm-2 contains the histone H3.3-intron intrinsic terminators TI., Tb, and TI] (Reines et al., 1987) subcloned downstream of the adenovirus major late promoter. These histone sequences reside on a 285-bp' TaqI restriction fragment removed from the genomic subclone pHuH3-640 (Reines et al., 1987). This DNA by highly purified RNA polymerase I1 at sites Ia, Ib, and 11: The fragment contains sufficient sequence information for termination T q I restriction fragment was inserted into the unique AccI site of the plasmid pDNAdML. The plasmid pDNAdML (Conaway and Conaway, 1988) carries a synthetic oligonucleotide containing the adenovirus-2 major late promoter sequence from -50 to +10 inserted into the KpnI and XbaI sites of pUC18. A runoff template was generated from this plasmid by cleaving it with the restriction enzyme NdeI which cuts the plasmid once (see also Fig. 1A).
Transcription by purified calf thymus RNA polymerase I1 was carried out on a 3'-extended ("tailed") DNA template. This template was prepared as described by Reines et al. (1987) from the plasmid pUC18-EF2 provided to us by E. Falck-Pederson (Cornell University Medical College, New York). This plasmid contains mouse &major globin gene segments E and F (Hofer and Darnell, 1981) inserted into the BamHI site of pUC18. The plasmid was linearized at the unique SmaI site, and oligodeoxycytidylate "tails" were added with terminal transferase as described by Kadesch and Chamberlin (1982). The DNA was further cleaved with PstI to generate a runoff template of approximately 1200 bp and EcoRI to remove the "distal" tail. The tailed and cleaved DNA was purified by extraction with phenokchloroform and ethanol precipitation. The DNA was dissolved * T. Kerppola, unpublished results. ' The abbreviations used are: bp, base pair(s); NTPs, nucleoside triphosphates; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid.

Proteins and Enzymes
Bovine serum albumin and placental ribonuclease inhibitor were purchased from Sigma. Bovine serum albumin was further purified by chromatography on Bio-Rex 70 to remove contaminating ribonuclease activity (Hirs et al., 1952). Proteinase K was purchased from Boehringer-Mannheim and Sigma. Restriction enzymes were purchased from New England Biolabs, Bethesda Research Laboratories, and Boehringer-Mannheim and were used according to the supplier's recommendations.
Rat liver RNA polymerase I1 transcription factors were provided by, or purified in collaboration with, Drs. Joan Conaway and Ron Conaway (University of Texas, Austin). RNA polymerase 11, fraction D, and a were purified from rat liver as described by Conaway et al. (1987). Rat liver fraction B' was purified by a modification of the procedure described by Conaway et al. (1987). The purification included phosphocellulose and AcA 44 chromatography as described (Conaway et al., 1987). This protein was then chromatographed on a TSK DEAE-5-PW column as recently described (Conaway and Conaway, 1989) and concentrated by chromatography on hydroxylapatite as described by Conaway et al. (1987). Calf thymus RNA polymerase I1 was purified as described by Hod0 and Blatti ((1977) modified by Dedrick and Chamberlin (1985)). This preparation of calf thymus RNA polymerase I1 contains trace amounts of RNA polymerase I11 activity (DNA-dependent RNA polymerase resistant to 1 pg/ml aamanitin). This has been noted previously by us (Kerppola and Kane, 1988) and others (Watson et al., 1984). Calf thymus elongation factor SI1 was purified to apparent homogeneity by the procedure of Rappaport et al. (1987) and was generously provided by J. Rappaport and R. Weinmann, Wistar Institute, Philadelphia, PA. In Vitro Transcription Reactions scribed above (pUC18-EF2) was transcribed with purified calf thymus Transcription on 3"Extended Templates-The tailed template de-RNA polymerase I1 as previously described (Reines et al., 1987). Two 25-pl transcription reactions were assembled in transcription buffer (70 mM Tris-HC1 (pH 8.0), 150 mM NaCl, 20% (v/v) glycerol, 6 mM MgCl,, 5 mM spermidine, 0.15 mM dithiothreitol) containing 800 p M each of ATP, UTP, and GTP, 20 p~ [a-32P]CTP (66,000 cpm/pmol), 0.6 pg of tailed pUC18-EF2 prepared as described above, and 12.5 milliunits of calf thymus RNA polymerase I1 (1 unit is the amount of enzyme needed to incorporate 1 nmol of CMP into acid-insoluble form in a 10-minute reaction under the standard conditions described by Hod0 and Blatti (1977) using sheared calf thymus DNA as a template). The reaction was started by incubation at 37 'C. A 6-pl sample was withdrawn from each reaction after 1 min and made 220 pg/ml in yeast RNA and 100 mM in EDTA (final volume of 260 ~1 ) on ice. Forty-five seconds later the remainder of each reaction was diluted 10-fold with transcription buffer containing 800 pM each of ATP, UTP, and GTP, 100 p~ unlabeled CTP, and 100 pg/ml heparin. One reaction was made 0.23% in Sarkosyl (w/v). Incubation was times and placed into the EDTA/RNA solution on ice as described continued at 37 "C, and 60-pl aliquots were withdrawn at various above for the 1-min samples. Nucleic acids were extracted with phenol:chloroform, precipitated with ethanol and analyzed by electrophoresis and autoradiography as described below.
Transcription in HeLa Nuclear Extracts-HeLa cells were grown in spinner culture at 37 "C to a density of 4-6 X lo5 cells/ml in Joklik's modified Eagle's medium supplemented with 2 mM glutamine and 5% calf serum. Extracts from HeLa cell nuclei were prepared, dialyzed into buffer " D (20 mM Tris-HCI, pH 7.9, 20% glycerol (v/ v), 0.1 M KCl, 0.2 mM EDTA, 1 mM dithiothreitol), and used for in uitro transcription essentially as described by Dignam et al. (1983). Fifty microliter reactions contained NdeI-cleaved pAdTerm-2 and 50% (v/v) nuclear extract protein (250 pg). RNA was synthesized in vitro for 50 min at 30 "C and was labeled with [a-32P]GTP at 25 p M (5-10 Ci/mmol). ATP, UTP, and CTP were each present at 600 PM. Placental ribonuclease inhibitor was included at 5 units/reaction. Labeled RNA was isolated as described (Dignam et al., 1983) by pheno1:chloroform extraction and ethanol precipitation, and analyzed by electrophoresis and autoradiography as described below.
In Vitro Transcription Reconstituted from Purified Rat Liver Transcription Initiation Factors-Transcription with rat liver transcription factors utilized fractions D, B', and pure a-polypeptide, as well as partially purified RNA polymerase I1 from rat liver (Conaway et al., 1987). A standard reaction included a 20-4 preincubation step containing 0.4 milliunits of rat liver RNA polymerase 11, 100-200 ng of pAdTerm-2 DNA (cut with NdeI), fraction D (2-4 pg), and 5 units of placental ribonuclease inhibitor in 20 mM Hepes-NaOH, pH 7.9, 20 mM Tris-HCI, pH 7.9, 2% polyvinyl alcohol (w/v), 0.4 mg/ml bovine serum albumin, 0.15 M KCI, 2 mM dithiothreitol, and 7% (v/ v) glycerol. This mixture was incubated at 28 "C for 30 min. The reaction was then diluted with 33 p1 of a solution containing fraction B' (1 pg) and 1 ng of a-polypeptide in the same buffer lacking KC1 and incubated for another 20 min. Magnesium chloride, ATP, UTP, and [cY-~'P]CTP (>400 Ci/mmol) were added in a volume of 6 p1 to final concentrations of 7 mM, 20 p M , 20 pM, and approximately 0.6 p~, respectively. Incubation proceeded under these limiting nucleotide conditions for another 20 min. This procedure resulted in the formation of elongation complexes containing a 14-nucleotide transcript (the first GTP in the transcript is at position 15). Heparin or Sarkosyl was added as indicated to prevent reinitiation by RNA polymerase 11. These transcription complexes are referred to as "ternary complexes" (see the legend to Fig. 1B). The next phase of the reaction was referred to as the "elongation phase" in which higher concentrations of ATP, CTP, GTP, and UTP were added as indicated. Incubation was continued at 28 'C for an additional 30 min unless otherwise indicated. Also, where indicated, HeLa protein or purified calf thymus SI1 protein was added at the start of the elongation phase of the reaction. Reactions were stopped with an equal volume of proteinase K buffer (2% sodium dodecyl sulfate (w/v), 0.2 M Tris, pH 7.5, 25 mM EDTA, 300 mM NaCI), 5 pg of proteinase K, and 20 pg of yeast RNA and incubated at room temperature for 5 min. Nucleic acids were precipitated with ethanol and prepared for electrophoresis and autoradiography as described below.

Fractionation of HeLa Nuclear Extract Proteins
HeLa nuclear extract protein was prepared from 30 liters of cells as described by Dignam et al. (1983). The protein was chromatographed over P11, DE52, and single-stranded DNA-agarose as described by Reinberg and Roeder (1987b) for the purification of TFIIS.
Electrophoresis and Autoradiography gels (38:2, acry1amide:bisacrylamide) in TBE (89 mM Tris, 89 mM Generally, electrophoresis was carried out on 5% polyacrylamide boric acid, pH 8.0, 1 mM EDTA) containing 8.3 M urea. Ethanol precipitates of nucleic acids were dissolved in TBE, 80% formamide, 0.025% xylene cyanol, and 0.025% bromphenol blue, heated at 95 'C for 5 min, and chilled on ice. Twenty-centimeter gels were run in TBE until the xylene cyanol dye was approximately 15 cm from the origin. Gels were dried onto Whatman No. 3MM paper. In some cases, discontinuous polyacrylamide/urea gels (gels containing a 5% polyacrylamide gel poured on top of a 15% polyacrylamide gel, see legend to Fig. 2 A ) were used to resolve both small (14 nucleotides) and large (530 nucleotides) transcripts. These gels were not dried. All gels were exposed to X-Omat film (Kodak) at -80 "C with an intensifying screen (Du-Pont). Marker RNAs of known sizes were prepared by in vitro transcription of plasmids pKK5-1 (lanes M ) and pKK34-121 (lanes M ' ) (Brosius, 1984;Thayer and Brosius, 1985) with Escherichia coli RNA polymerase holoenzyme purified by the method of Gonzalez et al. (1977).

Transcription Elongation in Vitro Studied with a Purified
Promoter-initiating System-In prior studies we have demonstrated that purified RNA polymerase I1 can cease transcript elongation and release RNA chains at discrete sites within a human histone gene (Reines et at., 1987). Because the polymerase stops transcription and releases its transcripts, we have termed such sites intrinsic termination sites. Such studies of transcription using 3"extended templates assume that the properties of elongating RNA polymerase I1 are similar regardless of whether the enzyme has initiated transcription from a promoter with initiation factors or from a 3'-extended template in the absence of initiation factors. To test this directly, we have analyzed transcription through three sites in an intron of the human histone H3.3 gene (Ia, Ib, and 11; Reines et al., 1987) by RNA polymerase I1 which had initiated transcription from the adenovirus major late promoter in the presence of additional factors (Fig. lA). This promoter is well characterized and relatively strong i n uitro. Accurate transcription initiation by RNA polymerase I1 can be reconstituted from purified chromatographic fractions derived from animal cells (Matsui et al., 1980;Tsai et al., 1981;Dynan and Tjian, 1983;Samuels et al., 1982;Davison et al., 1983;Parker and Topol, 1984;Conaway et al., 1987). However, in order to distinguish the activity of initiation factors from that of potential elongation factors it was important to obtain the initiation factors in highly purified form. As well, relatively large amounts of initiation factors were needed for biochemical studies of termination. Recently, a highly purified specific transcription system for RNA polymerase I1 has been described (Conaway et al., 1987;Conaway, 1988,1989). These workers exploited rat liver as an inexpensive and abundant source of material from which large amounts of transcription factors could be extensively purified. With these proteins, RNA polymerase I1 can initiate transcription from the adenovirus major late promoter and transcribe through the histone intron termination sites (Fig. lA). Elongation by RNA polymerase I1 was studied using an experimental protocol limiting transcription to a single round of initiation events. This was achieved by forming defined ternary complexes (DNA, RNA, and protein) arrested at a unique position on the template (+15) and preventing subsequent reinitiation by RNA polymerase I1 (see Fig. 1B and "Experimental Procedures").
An analysis of transcript elongation from the adenovirus major late promoter demonstrated that RNA polymerase I1 stops elongation at the same sites previously characterized as intrinsic termination sites (Fig. 2.4). The precise location of the 3' termini of these transcripts was confirmed using S1nuclease protection experiment^.^ Highly purified calf thymus RNA polymerase I1 also stopped transcription at these sites when it was substituted for the partially purified rat liver RNA polymerase I1 used in the reconstituted transcription r e a~t i o n .~ These initial studies were done using relatively low concentrations of NTPs, the conditions employed in earlier studies of transcription initiation in the rat liver system (Conaway et al., 1987). However, it is known that pausing in uitro by RNA polymerase I1 is sensitive to N T P concentrations.' This led us to test the effect of increasing concentrations of NTPs on elongation in this system. Under these conditions the pattern of transcription elongation and termination was significantly changed (Fig. 2B).
RNA polymerase I1 elongated efficiently through sites Ib and I1 although site Ia still halted transcription elongation. This raised the question of whether the cessation of elongation in the earlier experiments using low levels of NTPs was due t o true termination or to long pausing at these sites. In prior studies we have followed transcript release by nitrocellulose filter binding. However, this method is not applicable to the current experiments since the high levels of protein in the reactions interfere with this assay?
To attempt to distinguish between long pausing and termination/release, transcription was allowed to proceed at a low concentration of nucleotides until RNA polymerase I1 had reached sites 11, Ib, and Ia. The nucleotide concentration was then increased (Fig. 2C). The results show clearly that transcripts associated with sites I1 and Ib were chased to larger sizes. Transcripts in band Ia were not chased under these conditions; however, we show below that these transcripts can also be further elongated in the presence of a protein factor. Hence all of the transcripts with 3' ends at sites 11, Ib, and Ia are components of paused ternary com-  FIG. 1. A, map of plasmid pAdTerm-2. pAdTerm-2 is a plasmid containing the histone H3.3 intrinsic terminators TI., Tm, and TII (Reines et al., 1987) inserted downstream of the adenovirus major late promoter (see "Experimental Procedures" for its construction).
A 285-bp Tap1 restriction fragment (H3.3) containing these termination sites was inserted into the unique AccI site of a pUC18-based plasmid (pDNAdML) which contained the adenovirus-2 major late promoter (MLP) sequence from -50 to +lo. A runoff template was generated from this plasmid by cleaving it with the restriction enzyme NdeI. The runoff transcript should be 530 nucleotides. Transcripts with 3'-ends at sites Ia, Ib, and I1 should be 205, 185, and 145 nucleotides, respectively. I?, schematic representation of in vitro transcription experiments using purified rat liver transcription factors. Preincubation: rat liver transcription factors a, p', and D were incubated at 28 "C with RNA polymerase I1 and NdeI-cleaved pAdTerm-2 DNA as described under "Experimental Procedures." Ternary complex formation: the reaction was made 20 PM in ATP, 20 PM in UTP, and 0.6 PM in [w3*P]CTP, and incubation was continued at 28 "C as described under "Experimental Procedures." Note that there is no G residue in the nontranscribed DNA strand until position +15 (RNA start site is +l). This population of RNA polymerase I1 molecules that have synthesized a 14-nucleotide transcript will henceforth be referred to as ternary complexes. Heparin or Sarkosyl was added to prevent reinitiation by additional RNA polymerase I1 molecules. Elongation phase: ATP, UTP, CTP, and GTP were added to 800 PM unless otherwise indicated, and incubation plexes that have very long lifetimes.
An RNA transcript of 325 nucleotides was also seen in this experiment (Fig. 2) and other experiments (Figs. 4-6 and 8). Its appearance was reproducible and apparently resulted from transcription initiation at the adenovirus major late promoter and the subsequent stopping of RNA polymerase I1 within vector sequences.
An Activity in HeLa Cell Extracts That Allows Efficient Transcription through Site la-Site Ia appeared to be a strong block to elongation in vitro under all the reaction conditions we tested (Fig. 2). This result suggested to us that eukaryotic cells might contain a factor(s) that would enable polymerase to read through this site since H3.3 is expressed in viuo. Transient expression experiments in which this site is interposed between a strong promoter and a reporter gene confirmed this expectation for several cell types since there was no diminution of reporter gene expression.2 In order to investigate this possibility we transcribed the plasmid pAdTerm-2 using an unfractionated HeLa nuclear extract (Dignam et al., 1983). In contrast to the results found with the purified transcription system from rat liver, RNA polymerase I1 did not stop transcription at the histone intrinsic termination sites, and the only specific RNA synthesized in vitro was the runoff transcript (Fig. 3). Smaller RNAs resulting from RNA polymerase I1 stopping at sites Ia, Ib, or I1 were not detected. It was important to rule out the possibility that transcript Ia had been synthesized in the presence of HeLa protein but was exceptionally sensitive to degradation. To test for RNA degradation, transcript Ia was synthesized by RNA polymerase I1 in a reconstituted rat liver transcription reaction in the absence of the HeLa nuclear proteins; a-amanitin and the nuclear extract were added and incubation was continued. Although some transcript breakdown occurs during this prolonged incubation, no preferential degradation of transcript Ia was observed (Fig. 4, lune 5). This makes it unlikely that polymerase had stopped transcription at site Ia but that the transcripts with 3'-ends at site Ia were completely degraded. We concluded, therefore, that the HeLa extract contained a "readthrough" factor for RNA polymerase 11.
A Complementation Assay for Readthrough Activity-Using these observations we developed a complementation assay for "readthrough" activity, since RNA polymerase 11, which initiated transcription using purified rat liver initiation factors, efficiently stopped transcription at histone site Ia (Fig. 2) and HeLa cell extracts contained a trans-acting factor which enabled RNA polymerase I1 to read through this block to elongation (Fig. 3).
Increasing amounts of HeLa cell nuclear extract were added to transcription reactions containing ternary complexes formed with RNA polymerase I1 and rat liver initiation factors (see Fig. 1B). Elongation was continued in the presence of the HeLa protein and all four nucleotides. Essentially all polymerase molecules were able to elongate past site Ia when 5 pg or more of HeLa protein were added to the elongation phase of the reaction (Fig. 4, lunes 1-4). These results show that HeLa nuclear extracts contain an activity that can allow RNA polymerase 11 to read through a block to elongation in a human gene and that this activity could be titrated. Further, actively elongating RNA polymerase I1 was required to obtain readthrough activity as amanitin-treated ternary complexes was continued for 30 min at 28 "C. In some experiments HeLa cell extract protein (Figs. 4 and 8), chromatographically fractionated HeLa protein (Fig. 5), or pure SI1 protein (Fig. 6) was added during the elongation phase. FIG. 2. Transcription of intrinsic termination sites by RNA polymerase I1 using purified rat liver transcription initiation factors. A, elongation a t "low" nucleotide concentrations. Transcription of the plasmid pAdTerm-2 (Fig. 1A) was carried out with purified transcription initiation factors from rat liver (Conaway et al., 1987). A 300-pl ( 5 standard reaction volumes) reaction was assembled as described ( Fig. 1R and "Experimental Procedures"). After ATP, UTP, and [a-"PlCTP were added, incubation proceeded until ternary complexes had been formed (see legend to Fig. 1B). A 60-pl aliquot was removed and diluted into proteinase K buffer as described under "Experimental Procedures" (0'). Sarkosyl (0.24%) was added to prevent the reinitiation of transcription by additional RNA polymerase I1 enzymes. Unlabeled GTP and CTP were added to final concentrations of 200 and 620 p M , respectively, while the final concentrations of ATP and UTP remained at 20 p~ during elongation. The reaction was incubated a t 28 "C. At the indicated times, 60-pl samples were withdrawn, and each was diluted into proteinase K buffer as described under "Experimental Procedures." Nucleic acids were isolated from the sample taken a t each time point and electrophoresed on a discontinuous gel system consisting of a 5% polyacrylamide gel stacked on a 15% polyacrylamide gel. The interface between the 5 and 15% gels is indicated to the right of the figure. The migration positions of radiolabeled RNAs of known size that were also run on this gel are indicated. The RNA species with 3'-ends a t intrinsic termination sites TI., Tb, and T I 1 are so indicated. B, elongation a t "high" nucleotide concentration. This experiment was carried out identically to that described in part A above, except that heparin (10 pg/ml) was added to prevent reinitiation. The final concentration of each N T P during the elongation phase was 790 p~ each for ATP and UTP and 770 p~ each for GTP and CTP. The 14-nucleotide transcript synthesized by ternary complexes before the elongation phase of the reaction was started is indicated (14rner). The discontinuous polyacrylamide gel (5%/15%) system was also used here. C, ternary complexes a t sites Ib and I1 can continue elongation after increasing the nucleotide concentration. Ternary complexes were formed as described in part A. The reaction was then brought to 200 pM in GTP and 620 p~ in unlabeled CTP; the concentration of ATP and UTP remained unchanged. Elongation under these conditions took place for 30 min. Half of the reaction was then quenched (lane marked LO). The remainder of the reaction was brought to 760 p~ each in ATP and UTP, 1 mM in GTP, and 1.4 mM in CTP, and incubation proceeded for another 20 min ( l a n e marked HI). RNA was isolated from both samples and analyzed on a 5% polyacrylamide gel as described under "Experimental Procedures." Lane M', RNA markers of 260,380,420, and 540 nucleotides. An extract from HeLa cell nuclei was prepared essentially as described by Dignam et al. (1983). Transcription reactions (Dignam et ul., 1983) contained the concentrations of DNA (NdeI-cleaved pAdTerm-2) indicated at the top of the figure. In this experiment RNA was synthesized in a single incubation reaction with all four NTPs. The resulting RNA was uniformly labeled with 25 pM [a-"PIGTP. ATP, UTP, and CTP were present a t 600 pM each. a-Amanitin (1 Fg/ml) was included from the start of the reaction in the lane labeled +a. Labeled RNA was isolated after 50 min of transcription and analyzed by electrophoresis on a 5% polyacrylamide gel and autoradiography. The runoff transcript and sizes of reference RNAs that were run on this gel are indicated. stopped a t site Ia did not produce longer transcripts (Fig. 4,  lane 5 ) .
It is interesting to note that HeLa readthrough activity was also effective in allowing RNA polymerase to efficiently elongate past the site defined by the 325-nucleotide transcript with a 3'-end in vector sequences (see above and Fig. 4).
Identification of Readthrough Activity-Natori and coworkers first reported the purification and characterization of a protein, SII, that stimulates transcription by RNA polymerase I1 (reviewed in Natori, 1982). This factor also enables 540-pl reaction (9 standard-reaction volumes) was prepared as described under "Experimental Procedures," and ternary complexes were formed as described in the legend to Fig. 1B under "Experimental Procedures." The reaction was separated into 60-pl aliquots on ice and made 400 p~ in all 4 NTPs and 10 pg/ml in heparin. Sixty microliters of buffer D (see "Experimental Procedures") containing the indicated amounts of HeLa nuclear extract protein were added to each reaction aliquot (lunes 1-4). Incubation was continued for 30 min. The RNA was extracted with pheno1:chloroform and precipitated with ethanol. Labeled RNA was analyzed by gel electrophoresis and autoradiography as described under "Experimental Procedures." In lane 5 one aliquot of "ternary" complexes was incubated for 30 min with 400 p~ NTPs without HeLa protein. a-Amanitin (1 pg/ml) was added followed by the HeLa protein, and the incubation was continued for an additional 30 min before processing for electrophoresis. The runoff transcript and transcript ending a t site Ia are indicated to the right of the figure. RNA transcripts of 260,380,420, and 540 nucleotides were run in lane M '.
RNA polymerase I1 to efficiently read through a well characterized pause site within the adenovirus major late transcription unit (Maderious and Chen-Kiang, 1984;Mok et al., 1984;Hawley and Roeder, 1985;Reinberg and Roeder, 1987b;Rappaport et al., 1987). Therefore, we wanted to determine whether the readthrough activity identified here for histone site Ia was similar to elongation factor SII.' To this end we fractionated the HeLa nuclear extract as described by Reinberg and Roeder (1987b) for the purification of SII. On phosphocellulose, DEAE-cellulose, and singlestranded DNA-agarose, readthrough activity partitioned with fractions reported by Reinberg and Roeder (1978b) to contain SI1 (Fig. 5)

.s
These results suggested that at least part of the readthrough activity detected in HeLa cell nuclear extracts was due to elongation factor SII. In fact, the addition of NTPs and a homogeneous preparation of calf thymus SI1 (Rappaport et PhosbhOCelblOSe DEAE runoff al., 1987) to preformed rat liver ternary complexes resulted in virtually quantitative readthrough of site Ia (Fig. 6A, compare  lanes 2 and 3). This result demonstrated that SII, added after 13 phosphodiester bonds were synthesized, can promote readthrough by RNA polymerase I1 at a site almost 200 bp downstream (Fig.

B sn
heparin Sarkosyl 5. Fractionation of HeLa readthrough activity. HeLa nuclear extract was chromatographed on phosphocellulose and DEAE-cellulose as described by Reinberg and Roeder (1987b). No protein (-) or 1 pg of protein from the most concentrated fraction of each step-eluted protein peak was assayed for readthrough activity in the elongation phase of a standard rat liver transcription reaction (as described for the unfractionated nuclear extract in Fig. 4) , or 50 ng of SI1 protein ( S I I ) were added to aliquots of ternary complexes. Each reaction was also made 800 *M in ATP, CTP, UTP, and GTP, and incubation was continued for 30 min. RNA was isolated and analyzed by electrophoresis as described under "Experimental Procedures." In lanes 1 3 , heparin (10 pg/ml) was added to the ternary complexes to prevent reinitiation. In lanes 4-6 Sarkosyl (0.25%) was added to the ternary complexes to prevent reinitiation. Runoff transcripts and transcript Ia are designated on the left of the figure.
Marker RNAs were run in l a m s M and M ' as described above. B, histone site Ia is a pause site for RNA polymerase 11. Ternary complexes treated with heparin (10 pg/ml) were formed as described under "Experimental Procedures" and the legend to Fig. 1R. These RNAs were extended in the presence of 800 p~ ATP, CTP, UTP, and GTP for 30 min in the absence of added SI1 protein. Half of the reaction was removed and incubated with SI1 buffer (10 mM Tris-HCI, pH 7.9, 0.1% Triton X-100 (v/v), 72 mM NaCI) for an additional 30 min (-). Fifty nanograms of SI1 protein was added to the remaining half of the reaction, and it was also incubated for an additional 30 min (+). The RNA was isolated, electrophoresed, and autoradiographed as described under "Experimental Procedures." Runoff transcripts and transcripts with 3' termini a t sites Ia are indicated. RNAs of known sizes (505, 625, 665, 785, and 1620 nucleotides from fastest to slowest migrating) appear in lane M.

RNA Polymerase I1
6 ) . Thus, as described by others (Reinberg and Roeder, 1987b;Rappaport et al., 1987), SI1 is not needed during the initiation reaction to exert its effect during elongation.
Our earlier experiments (Fig. 2C) revealed that the apparent termination at sites Ib and I1 was instead due to formation of stable paused complexes. This raised the question of whether site Ia was acting as a terminator or a pause site. T o determine whether an active elongation complex was arrested a t site Ia, SI1 was added to the reaction after transcript Ia had been synthesized (Fig. 6 B ) . Since SI1 enabled RNA polymerase I1 to continue chain elongation beyond site Ia, we can conclude that transcript Ia was still associated with an active enzyme complex. Therefore, this factor can exert its effect on the stalled complex.
The Effect of Sarkosyl on the Readthrough Actiuity of SII-Sarkosyl (N-lauroylsarcosine) is an anionic detergent that has been extensively used to study transcription initiation and elongation (Gariglio et al., 1974(Gariglio et al., , 1981McKnight and Palmiter, 1979;Ackerman et al., 1983;Coppola and Luse, 1984;Tolunay et al., 1984;Roeder, 1985, 1987;Conaway and Conaway, 1988;Zhang-Keck and Stallcup, 1988). Initiated RNA polymerase I1 molecules are resistant to relatively high levels of this detergent. Uninitiated or "free" RNA polymerase molecules are sensitive to inactivation by Sarkosyl (Gariglio et al., 1974). In addition, Sarkosyl has been reported to cause enhanced pausing or premature termination a t a specific site within the adenovirus major late transcription unit (Hawley and Roeder, 1985). This could result from an effect of Sarkosyl on either RNA polymerase or an associated elongation factor. Our use of purified transcription systems, one that uses only RNA polymerase I1 protein (3'extended templates; Reines et al., 1987) and one that utilizes the polymerase and accessory initiation factors (this report), provided an opportunity to identify the target of Sarkosyl action.
T o see if Sarkosyl affects the polymerase itself, we first transcribed 3'-extended templates with highly purified calf thymus RNA polymerase I1 in the presence and absence of Sarkosyl. When added before nucleotides, Sarkosyl (0.25%) serves to inhibit the initiation of transcription by RNA polymerase I1 on 3"extended templates5 as has been shown for initiation a t promoter sites (Hawley and Roeder, 1985). Conversely, the addition of Sarkosyl to the elongation phase of the transcription reaction (after the addition of nucleotides) did not result in a significant change in the average elongation rate of RNA chains (Fig. 7). Sarkosyl-induced pausing was not observed either. Therefore Sarkosyl did not serve to generally disable the elongation capacity of purified RNA polymerase 11. This suggested that the exaggerated transcriptional pausing brought about by Sarkosyl (Hawley and Roeder, 1985) resulted from an interaction of Sarkosyl with a component of the transcription machinery other than RNA polymerase II.
We wanted to test directly whether Sarkosyl inhibited the readthrough function of SII. This seemed likely given the following observations. First, in the absence of SII, RNA polymerase I1 pauses a t specific sites within the adenovirus major late transcription unit (Reinberg and Roeder, 1987b). The addition of SI1 (Rappaport et al., 1987;Reinberg and Roeder, 1987b) has been shown to provide RNA polymerase I1 with the ability to read through these pause sites. Second, Sarkosyl dramatically enhanced the pausing at these sites when transcription was reconstituted from chromatographic fractions containing partially purified HeLa initiation factors and SI1 (Hawley and Roeder, 1985).
When Sarkosyl was added to the elongation phase of a FIG. 7. Sarkosyl does not affect transcription elongation by purified RNA polymerase I1 on a "tailed" template. Transcription using purified RNA polymerase I1 on a tailed template (plasmid pUC18EF-2, a subclone of the mouse @-major globin gene) was carried out as described under "Experimental Procedures." RNA was synthesized in uitro in two separate (+ and -) two-part reactions such that ternary complexes could be formed and chased. During the "chase" phase of transcription (after dilution, see "Experimental Procedures") one reaction was made 0.23% (w/v) in Sarkosyl (+), and the other was not (-). At the indicated times after the start of the reactions, samples were withdrawn and RNA was isolated and analyzed by electrophoresis on a 5% polyacrylamide gel as described under "Experimental Procedures." RNAs of known sizes were run in lanes M and M '. reconstituted rat liver transcription reaction (Fig. 6A, lanes 4  and 5 ) , polymerase molecules accumulated at site Ia to the same extent as the control (Fig. 6A, lanes 1 and 2). Therefore, Sarkosyl does not influence stopping a t site Ia in this purified system. However, when the reconstituted rat liver transcription reaction was supplemented with SII, Sarkosyl inhibited the ability of this protein to provide readthrough activity (Fig.  6A, compare lanes 3 and 6). This same level (0.25%) of Sarkosyl also inactivated the readthrough activity present in a HeLa nuclear extract (Fig. 8). Thus Sarkosyl, at levels which have been shown to cause the accumulation of prematurely terminated (or paused) transcripts in the adenovirus major late transcription unit (Hawley and Roeder, 1985), prevented SI1 from promoting readthrough a t site Ia. It is reasonable to conclude that the inhibitory effect of Sarkosyl on SI1 function is responsible for the apparent enhancement HeLa nuclear extract. The complementation assay shown here was identical to the one described in Fig. 4. In lanes 3 and 4, 300 pg of HeLa nuclear extract protein in 60 pl of buffer D (see "Experimental Procedures") were added during the elongation phase of the reaction. Lanes I and 2 were buffer controls lacking protein. In lanes 2 and 4, heparin (10 pg/ml) was included in the elongation phase of the reaction. In lanes I and 3, Sarkosyl (0.25%) was included in the elongation phase of the reaction. Nucleic acids were isolated and analyzed by electrophoresis on a 5% polyacrylamide gel as described under "Experimental Procedures." of pausing observed in the adenovirus transcription unit (Hawley and Roeder, 1985).
studies have been useful in understanding the activities carried out by purified RNA polymerase 11. For example, we have obtained provocative evidence that sequences within eukaryotic genes block transcript elongation and appear to have physiological significance (Reines et al., 1987;Kerppola and Kane, 1988). ' This approach is predicated on the assumption that the elongating enzyme faithfully responds to template signals regardless of the means by which it has initiated transcription (i.e. from a promoter or 3'-extended template). In this report we have explicitly tested this assumption. We have found that intragenic signals originally defined on 3'-extended templates which block transcription through a human histone gene also serve as a block to elongation for RNA polymerase I1 which had initiated from a promoter. These findings support the validity of using 3"extended templates to study transcription by purified RNA polymerase.
Surprisingly, although sites Ia, Ib, and I1 stop RNA polymerase 11, there is an unexpected difference in the transcript release reaction at these sites with the promoter-dependent system as compared to our previous studies with 3'-extended templates. In particular, transcription using the promoterinitiated system leads to extended pausing at the same sites a t which we demonstrated termination/transcript release with 3'-extended templates. Further analysis has confirmed that release does occur on several different 3"extended tem-p1ates.'as There are at least three differences between the two transcription systems that could influence transcript release. First, the initiation reaction may influence the recognition by RNA polymerase of a site as a terminator or a pause. The accessory factors required for promoter-specific initiation may alter the conformation or subunit composition of the elongating polymerase in a way that influences the release reaction a t sites which stop the enzyme. Whereas SI1 can allow readthrough of such sites, there may be proteins involved in the initiation reaction which associate with the elongating polymerase to stabilize the ternary complex when it stops at such sites.
Second, the reaction conditions differ rather dramatically between transcription on the 3'-extended and promoter-containing templates. For example, the ionic strength of the rat liver reconstituted transcription reaction is lower (60 mM KCl) than that previously used with 3"extended templates (150 mM NaC1). The termination reaction of purified E. coli RNA polymerase can show extremely large changes when the ionic conditions are altered (Neff and Chamberlin, 1980). In addition, changes in ionic strength can specifically affect the transcript release reaction of the bacterial enzyme.' The effect of ionic strength on the elongation reaction of purified RNA polymerase I1 is currently being examined in detail.
Third, the template constructions studied here differ from those used previously. Thus the transcripts synthesized in the two different in vitro experiments contain different nucleotide sequences. For purified E. coli RNA polymerase, the first 20-25 nucleotides in the transcription unit can alter the termination properties of the enzyme a t sites hundreds of nucleotides farther downstream (Telesnitsky and Chamberlin, 1989;Goliger et al., 1989). The mechanism of this effect is not known. Further, it is possible that transcript release may be a function of primary sequence or specific secondary structures in the nascent transcript as has been shown for E. coli RNA polymerase.' While we know that specific RNA secondary structures are not required to stop RNA polymerase I1 (Dedrick et al., 1987;Reines et al., 1987; Kerppola and Kane, ' K. Arndt and M. Chamberlin, submitted for publication. 1988), we do not yet know if particular features of the template or nascent transcript may affect the transcript release reaction. All three of these possible explanations are being investigated.
Despite the fact that we have not yet defined exactly which parameters control transcript release, site Ia clearly serves as a strong block to elongation through this histone gene in both purified transcription systems. While RNA polymerase I1 pauses briefly at many sites during transcription, the very long pausing at sites Ia, Ib, and I1 is unusual. Indeed site Ia serves to block transcription in vitro for hours while RNA polymerase remains in a potentially active ternary complex that can respond to elongation factors. Experiments analyzing transcription of H3.3 in isolated nuclei from HeLa cells also suggest that transcript elongation is blocked within this H3.3 i n t r~n .~ In vivo, a paused complex may serve as an intermediate to a termination event, i.e. as a target for the influence of regulatory factors. A polymerase elongation complex poised within a transcription unit could likely respond to changing cellular conditions more rapidly than one which must reassociate with other proteins at the promoter prior to initiation. Thus, a cell might carefully control whether to prematurely terminate within a gene or to pause during the course of transcription across a gene.
We have shown here that extracts from mammalian cells of numerous species serve as a source of factors that permit RNA polymerase I1 to read through histone site Ia. Rat liver appears to contain an SII-like readthrough activity5 as do human and bovine cells. SII-like proteins have been identified in yeast, insect, and mammalian cells by immunological or chromatographic criteria Sawadogo et al., 1981;Egyhazi et al., 1984;Rappaport et al., 1987;Reinberg and Roeder, 1987b). Since readthrough activity (SII) from various species can interact with RNA polymerase I1 from heterologous species (this report, Rappaport et al., 1987), it appears that this feature of the transcription machinery is conserved at least between mammals. These results further define transcriptional readthrough as an important process in controlling eukaryotic gene expression.
Transcriptional pause sites for promoter-initiated RNA polymerase I1 have previously been detected in viral sequences through the use of purified HeLa transcription initiation factors (Hawley and Roeder, 1985;Reinberg and Roeder, 198713;Rappaport et al., 1987). In particular, a site in the adenovirus major late transcription unit stops the enzyme, and SI1 is effective in allowing RNA polymerase I1 to read through this site (Hawley and Roeder, 1985;Roeder, 1987b, Rappaport et al., 1987). It is plausible that sites such as that described in adenovirus and H3.3 occur more generally in other genes. However, they may have gone undetected since many investigators use unfractionated cell extracts containing readthrough activity, truncated (run-off) templates, and indirect assays (primer extension, S1 nuclease analysis) to study transcription in vitro; all of these may preclude the detection of polymerase stop sites.
Transcription in isolated nuclei also suggests that the readthrough process is important in regulating the synthesis of transcripts from adenovirus (Maderious and Chen-Kiang, 1984;Mok et al., 1984) and histone H3.33 as well as other genes (Bentley and Groudine, 1986;Eick and Bornkamm, 1986;Nepveu and Marcu, 1986;McGeady et al., 1986;Mechti et al., 1986;Bender et al., 1987;Fort et al., 1987;McCachren et al., 1988;Watson, 1988;Bhat and Padmanaban, 1988). It will be important to test whether SII, or SII-like proteins, are involved in the regulation of transcription in vivo and if this type of mechanism may be more generally utilized in cells. It is possible that the ability of a cell to properly express genes is determined by the abundance, or post-translational modification, of SII. Although SI1 can be phosphorylated in vivo (Egyhazi et al., 1984;Hirashima et al., 1985), the effect of this modification on readthrough activity remains to be determined.
How does SI1 operate to promote transcriptional readthrough? Previous work has shown that SI1 can bind to purified RNA polymerase I1 (Horikoshi et al., 1984;Reinberg and Roeder, 1987b;Rappaport et al., 1987) and RNA polymerase I1 in transcription complexes (Horikoshi et al., 1984). Perhaps the interaction of SI1 with RNA polymerase I1 converts the enzyme into a readthrough conformation which would be immune to the signal for transcriptional arrest. A similar function has been shown for the X phage-encoded N and Q proteins which interact with an elongating bacterial RNA polymerase transcription complex to render it resistant to termination downstream. Whereas N protein requires at least four other bacterial proteins for its function, Q can act on E. coli RNA polymerase alone although full activity requires the E. coli nusA protein (reviewed by Roberts, 1988). The Q protein can also suppress pausing by purified E. coli RNA polymerase? In fact, purified RNA polymerase I1 transcribing 3'-extended templates can respond to SI1 in the absence of any other protein (Sluder et al., 1989). 586 We have also shown that Sarkosyl, a reagent commonly used to study transcription in vitro, can inhibit the activity of SI1 but has little, if any, effect on elongation by highly purified RNA polymerase 11. Previous studies (Hawley and Roeder, 1985) have described the apparent induction of pausing by Sarkosyl. This observation can now be readily explained since Sarkosyl inhibits the pause-suppressing activity of SI1 and can therefore unmask latent (SII-suppressed) pause sites for RNA polymerase 11. The interpretation of experiments using Sarkosyl should take this into consideration.
A second transcription inhibitor, heparin, has been used routinely for studying transcription in vitro. Heparin inhibits initiation by RNA polymerase I1 (Reinberg and Roeder, 1987b) while having no observable effect on the elongation properties of purified RNA polymerase 11, as we show here. The SI1 protein also functions to promote readthrough in the presence of heparin. In fact, SI1 is active on purified RNA polymerase I1 in the presence of heparin concentrations 10fold greater than those used in the studies described here.' Thus, we think it likely that SI1 can function efficiently in the presence of this drug. However, SI1 does bind to heparin-Sepharose columns (Rappaport et al., 1987), and we have not excluded the possibility that heparin may interact with SI1 in solution to alter its reaction mechanism in some quantitative manner. Further studies will resolve this question.