Intrinsic Transcript Cleavage in Yeast RNA Polymerase II Elongation Complexes*

Transcript elongation can be interrupted by a variety of obstacles, including certain DNA sequences, DNA-binding proteins, chromatin, and DNA lesions. Bypass of many of these impediments is facilitated by elongation factor TFIIS through a mechanism that involves cleavage of the nascent transcript by the RNA polymerase II/TFIIS elongation complex. Highly purified yeast RNA polymerase II is able to perform transcript hydrolysis in the absence of TFIIS. The “intrinsic” cleavage activity is greatly stimulated at mildly basic pH and requires divalent cations. Both arrested and stalled complexes can carry out the intrinsic cleavage reaction, although not all stalled complexes are equally efficient at this reaction. Arrested complexes in which the nascent transcript was cleaved in the absence of TFIIS were reactivated to readthrough blocks to elongation. Thus, cleavage of the nascent transcript is sufficient for reactivating some arrested complexes. Small RNA products released following transcript cleavage in stalled ternary complexes differ depending upon whether the cleavage has been induced by TFIIS or has occurred in mildly alkaline conditions. In contrast, both intrinsic and TFIIS-induced small RNA cleavage products are very similar when produced from an arrested ternary complex. Although α-amanitin interferes with the transcript cleavage stimulated by TFIIS, it has little effect on the intrinsic cleavage reaction. A mutant RNA polymerase previously shown to be refractory to TFIIS-induced transcript cleavage is essentially identical to the wild type polymerase in all tested aspects of intrinsic cleavage.

The ternary elongation complex consists of the RNA polymerase, the template DNA, and the nascent RNA transcript. Surratt et al. (2) discovered that bacterial RNA polymerase ternary complexes have a mechanism for transcript shortening, endonucleolytic cleavage near the 3Ј-end of the transcript. After transcript cleavage, the 3Ј-fragment is released while the 5Ј-fragment is retained in an active ternary complex. A single cleavage event can liberate from 1 to 17 nt 1 of RNA bearing a 5Ј-monophosphate (3)(4)(5)(6). Remarkably, transcription accurately resumes from the site of transcript cleavage to re-synthesize the excised RNA.
Transcript cleavage activity is conserved in many DNA-dependent RNA polymerases, including bacterial RNA polymerases, vaccinia virus RNA polymerase (7), RNA polymerase I (8,9), RNA polymerase II (10 -14), and RNA polymerase III (15). Accessory factors that stimulate transcript cleavage in ternary complexes have been identified in both prokaryotes (GreA and GreB) and eukaryotes (reviewed in Refs. 3, 9, 16 -18). A subunit of vaccinia virus RNA polymerase, rpo30, is 25% similar to one such elongation factor, TFIIS (19), and this vaccinia subunit is thought to affect the transcript cleavage reaction of the vaccinia polymerase.
Several observations suggest that the catalytic site for this transcript hydrolysis resides within the polymerase. First, RNA polymerase III complexes catalyze transcript cleavage, but a separate cleavage-stimulating factor has not been reported for RNA polymerase III. Second, ternary complexes of Escherichia coli RNA polymerase purified from a greA Ϫ greB Ϫ strain retain a low level of transcript cleavage activity (20), an activity stimulated by basic pH. Third, mammalian RNA polymerase II pyrophosphorylizes its transcript to give products identical to those induced by treatment with TFIIS (21). However, how TFIIS or the prokaryotic Gre factors stimulate this intrinsic cleavage remains unknown.
For yeast RNA polymerase II (RPII), the Rpb9p subunit has been shown to mediate the signal between TFIIS and the polymerase catalytic center for stimulating transcript cleavage (22). Arrested ternary complexes formed with yeast RNA polymerase II lacking Rpb9 (RPII⌬9) are much less responsive to TFIIS, and yet they can carry out the intrinsic cleavage reaction. These results led us to characterize the transcript cleavage reaction of RPII in more detail, particularly to compare the intrinsic cleavage in stalled and arrested complexes. Furthermore, we compared the wild type and the RPII⌬9 polymerases in their intrinsic cleavage properties. The analysis of intrinsic transcript cleavage activity in several discrete elongation complexes suggests extensive structural and/or functional hetero-geneity during transcript elongation. The results provide insights for how TFIIS might influence the properties of an elongation complex, aside from accelerating transcript cleavage.

Plasmids and DNA Templates
Plasmid pRW102 was constructed by inserting the "AGC cassette" from pGR220 (26) isolated on a SacI-KpnI fragment into the same sites of pTK242A:⌬8 (27), respectively. This plasmid pRW102 has two SmaI sites. The pRW102 plasmid was digested completely with SmaI and re-recircularized, generating pRW104 with a single SmaI site. Plasmid pRW106 was constructed by digesting pRW104 with NsiI, treating with T4 DNA polymerase to trim the overhangs, then digesting with HindIII to isolate the T1a-containing insert. This fragment was inserted into the HincII and HindIII sites of pN2⌬17. pN2⌬17 is a derivative of pN2 (28) in which 17 bp were deleted by digesting with EcoRI and BamHI, treating these ends with Klenow, and finally re-ligation.
Transcription templates from pRW106 were generated by digestion with SmaI, then 3Ј-deoxycytidine extensions were added with terminal nucleotidyl transferase (29). Templates constructed with such 3Ј tails are referred to as pCpRW106 templates. For 3Ј-biotinylated templates, unincorporated dCTP was removed by using two consecutive Biospin-30 columns (blocked with 20 g/ml tRNA), followed by incorporation of biotin-14-dCTP with terminal nucleotidyl transferase. Unincorporated biotin-14-dCTP was again removed by centrifugal gel filtration. Finally, pCpRW106 templates were restricted with NdeI and EcoRI to generate a run-off transcript and to disable transcription from the undesired end of the template. The pCpGEMTERM template preparation was as described previously (14); this template harbors a series of arrest sites in the first intron of the human histone H3.3 gene (30).

Immobilized DNA Templates
Biotinylated DNA templates (prepared as described above) were coupled to streptavidin-coated magnetic Dynabeads prior to transcrip-tion. 5 pmol of biotinylated DNA template was incubated at 37°C for 5 min in buffer containing 10 mM Tris-Cl, pH 8, 200 mM NaCl, and 25 g/ml BSA-OAc in a volume of 25 l. Then, 25 g of streptavidin-coated magnetic Dynabeads was added to the solution, and incubation proceeded for 30 -60 min at room temperature with slow rotation. The Dynabead-conjugated DNA templates were collected by magnetic precipitation (45 s), washed twice with 50 l of a high salt solution (10 mM Tris-Cl, pH 8, 1 M NaCl, 25 g/ml BSA-OAc), washed twice with 50 l of a low salt solution (10 mM Tris-Cl, pH 8, 20 mM NaCl, 25 g/ml BSA-OAc), and washed twice with 50 l of 1ϫ transcription buffer.

Formation of Ternary Complexes on Immobilized Templates
G135 ternary complexes were formed by incubating 1 pmol of RNA polymerase II with 5 pmol of Dynabead-conjugated pCpRW106 in 1ϫ transcription buffer containing 1 M [␣-32 P]CTP (3000 Ci/mmol), 800 M GTP and ATP, and 1 unit of InhibitAce, in a final volume of 25 l for 1.5 min at 30°C. At this time, ternary complexes containing short transcripts (ϳ13 nucleotides) predominate (data not shown). After this 5Ј-end-labeling step, 175 l of 1ϫ yeast transcription buffer containing 100 M CTP and 0.1 mg/ml heparin was added, and incubation continued for 1.5 min at 30°C. Acetylated BSA at 25 g/ml was included in all transcription buffers to inhibit the nonspecific adsorption of streptavidin to the microcentrifuge tubes. Unincorporated nucleotides were removed by collecting the ternary complexes via magnetic precipitation and washing three times with 50 l of transcription buffer.
G135 ternary complexes were "walked" to form U138 ternary complexes by addition of transcription buffer containing 10 M UTP followed by incubation for 3 min at 30°C. U138 ternary complexes were purified from unincorporated UTP by magnetic precipitation, and two washes with 50 l of transcription buffer. U138 ternary complexes were walked to form G143 ternary complexes by addition of transcription buffer containing 10 M CTP and GTP followed by incubation for 3 min at 30°C. G143 ternary complexes were purified from unincorporated NTPs by magnetic precipitation and two washes with 50 l of transcription buffer. G143 ternary complexes were walked to form G152 ternary complexes by addition of transcription buffer containing 10 M ATP and GTP followed by incubation for 3 min at 30°C. G152 ternary complexes were purified from unincorporated NTPs by magnetic precipitation and two washes with 50 l of transcription buffer. These were walked to C158 upon addition of 10 M CTP and ATP for 10 min at 30°C.
Ternary complexes halted on pCpGEMTERM, containing the T1a block to elongation (31), were formed by incubating 1 pmol of RNA polymerase II with 5 pmol of Dynabead-conjugated pCpGEMTERM in 1ϫ transcription buffer containing 1 unit of InhibitAce, 1 M [␣-32 P]CTP (3000 Ci/mmol), 800 M UTP, GTP, and ATP, in a volume of 25 l for 1.5 min at 30°C. After the addition of CTP to 100 M, heparin to 0.1 mg/ml, and BSA-OAc to 25 g/ml, reactions were incubated for 1.5 min at 30°C to generate ternary complexes halted at T1a, T1b, and TII. Ternary complexes were purified from unincorporated NTPs by magnetic precipitation and three washes with 50 l of transcription buffer.

3Ј-End Labeling of U138 and T1a Ternary Complexes
3Ј-End-labeled U138 Complexes-Unlabeled G135 complexes were formed on immobilized pCpRW106 as described above with the exception that unlabeled CTP was substituted for [␣-32 P]CTP. The unlabeled G135 complexes were purified from unincorporated NTPs by magnetic precipitation and washed three times with 50 l of 1ϫ standard transcription buffer. 3Ј-End-labeled U138 complexes were generated by addition of 3 M [␣-32 P]UTP for 5 min at 30°C. Unincorporated nucleotides were removed as above.
3Ј-End-labeled T1a Complexes-Unlabeled T1a complexes were formed on immobilized pCpGEMTERM as described above with the exception that unlabeled CTP was substituted for [␣-32 P]CTP. The unlabeled T1a complexes were purified from unincorporated NTPs by magnetic precipitation and washed three times with 50 l of 1ϫ standard transcription buffer. 1 nM TFIIS and 25 M GTP and CTP in 1ϫ standard transcription buffer were added to the unlabeled elongation complex, and incubation was continued for 5 min at room temperature to stimulate limited transcript cleavage in T1a-arrested elongation complexes. TFIIS and unincorporated NTPs were removed by three washes with 50 l of transcription buffer. 3Ј-End-labeled T1a complexes were generated by addition of 3 M [␣-32 P]UTP in 1ϫ standard transcription buffer and incubating at 30°C for 10 min. Unincorporated nucleotides were removed as above.

Transcript Shortening Reactions
Ternary complexes were purified from unincorporated NTPs either by centrifugal gel filtration with Biospin-30 columns or through use of magnetic precipitation, as indicated in the figure legends. The method of ternary complex purification did not qualitatively influence results. However, all transcript-shortening reactions were more efficient when magnetic precipitation was used to isolate the ternary complexes (i.e. the overall level of transcript cleavage by T1a or U138 complexes was greater). Transcript shortening induced by TFIIS, pyrophosphate, or manganese, was performed in 70 mM Tris-Cl (pH 8), 100 mM KCl, 5 mM MgCl 2 (or MnCl 2 ), 5% glycerol, and 1 mM DTT. Alkaline-stimulated transcript shortening was performed in 70 mM CAPS, K ϩ (pH 9.5), 100 mM KCl, 5 mM MgCl 2 , 5% glycerol, and 1 mM DTT. Reactions were incubated for the times indicated in the figures or figure legends.

RESULTS
Transcript Cleavage in Arrested Complexes-Purified RNA polymerase II stops and arrests in vitro at several sites within the human histone H3.3 first intron (T1a, T1b, and TII (30)). The strongest block, T1a, has frequently been used to study the elongation properties of RNA polymerase II from yeast and mammalian cells (27,(31)(32)(33)(34). Approximately 50% of yeast or mammalian RNA polymerase II elongation complexes stop within the first tract of nontemplate thymidines at the T1a site (14,27,35) (Fig. 1A) and do not efficiently resume elongation in the absence of elongation factor TFIIS. Thus, they are "arrested" (16,36) at the T1a site. RNA polymerase II arrested at T1a requires a transcript cleavage event, stimulated by TFIIS, to promote readthrough of the elongation block (37). Thus the T1a site has been used to investigate the mechanism of transcription arrest as well as the relief of arrest in the presence of TFIIS.
Biochemical Properties of the Intrinsic Cleavage Reaction-The addition of TFIIS to yeast RNA polymerase II elongation complexes arrested at T1a in the absence of NTPs generates a characteristic pattern of shortened transcripts (Fig. 2, lane 7) (see also Ref. 25). Two major cleavage products, C1 and C2, remain associated with active ternary complexes; the C1 cleavage product is generated prior to C2 (25). A low but detectable level of these cleaved transcripts was observed in the absence of TFIIS (Fig. 2, lane 2). This transcript cleavage intrinsic to RNA polymerase II was accelerated in mildly alkaline solutions (Fig.  2, lanes 3 and 4), as has been seen with E. coli RNA polymerase (20). Like the TFIIS-stimulated transcript cleavage, this intrinsic transcript cleavage yielded shortened transcripts associated with active ternary complexes that resumed elongation upon Shortened transcripts were elongated upon the addition of NTPs and thus were associated with active ternary complexes (Fig. 3, lanes 7 and 9). Other divalent cations such as Zn 2ϩ and Ca 2ϩ did not support TFIIS-stimulated transcript cleavage or intrinsic transcript cleavage at pH 8 (data not shown).
The effects of monovalent anions (Cl Ϫ and OAc Ϫ ) and cations (K ϩ , Na ϩ , and NH 4 ϩ ) on the rate of intrinsic transcript cleavage also were examined. In contrast to the pronounced stimulation of transcript cleavage by mild alkaline or certain divalent cations, the overall effects of salts (100 mM) were generally subtle and did not resolve into cation-specific or anion-specific effects. However, 100 mM KCl was optimal for intrinsic transcript cleavage at pH 9.5 or with manganese at pH 8.0. Curiously, 100 mM NH 4 OAc was least effective in supporting intrinsic transcript cleavage (data not shown), despite the fact that ammonium is capable of stimulating the overall elongation reaction of RNA polymerase II (38,39). Apparently, the stimulation of elongation by ammonium is not mediated by an increase in transcript cleavage at sites blocking elongation by RNA polymerase.
The Effect of NTPs and Pyrophosphate on the Intrinsic Cleavage Reaction-The addition of NTPs stimulates the transcript cleavage activity of vaccinia virus RNA polymerase (7) and The template was pCpGEMTERM (14). A minimal sequence sufficient to block RNA polymerase II elongation complexes is boxed (27). C1 demarcates the position of yeast and mammalian ternary complexes after the initial transcript cleavage event from the T1a site (25,37). B, sequences used to prepare G135, U138, G143, G152, and C158 ternary complexes with the immobilized pCpRW106. The transcript length in nucleotides and the positions (boldface and underlined) of the ternary complexes are indicated. RNA polymerase III (15). Thus, we also tested whether the intrinsic cleavage reaction of yeast RNA polymerase II was stimulated by the addition of NTPs. In contrast to the results with the viral polymerase and RNA polymerase III, NTPs did not stimulate the intrinsic transcript cleavage activity (data not shown) of two different yeast RNA polymerase II elongation complexes halted at either T1a or U138 (Fig. 1B, and see below). The TFIIS-stimulated transcript cleavage by RNA polymerase II ternary complexes also is not stimulated by NTPs (data not shown and Ref. 13).
Pyrophosphate stimulates an endolytic event in an arrested complex containing mammalian RNA polymerase II (21). When yeast RNA polymerase II complexes arrested at the T1a site were treated with 2 mM pyrophosphate at pH 8, only a small number of shortened transcripts with mobility similar to C1 were observed (data not shown). Significantly higher levels of shortened transcript accumulated when the reaction was performed at pH 7 (Fig. 4, lane 1). The lack of stimulation under basic conditions is more consistent with a pyrophosphorolytic than a hydrolytic mechanism. Pyrophosphate-shortened transcripts were associated with active ternary complexes as evidenced by their ability to resume elongation upon addition of NTPs (Fig. 4, lane 2). However, these results suggest that pyrophosphorolysis does not account for the observed intrinsic cleavage activity.
Intrinsic Cleavage Is Sufficient to Promote Readthrough-Transcript cleavage is a necessary prerequisite for reading through the T1a block to elongation (13,33,37). However, some TFIIS mutants (25,35,40) and at least one RNA polymerase II mutant (RPII⌬9 (25,41)) are able to stimulate the formation of cleavage products but not reactivate elongation complexes arrested at the T1a site. TFIIS is clearly necessary to stimulate the transcript cleavage reaction in elongation complexes containing RNA polymerase II at physiological pH, but the additional TFIIS-dependent step required for readthrough still is not clear. The ability to generate cleaved transcripts in the absence of TFIIS allowed us to assess the role of TFIIS in promoting transcript cleavage and in any additional step in the reactivation process. We also could determine if intrinsic cleavage itself could be sufficient to promote readthrough.
RPII complexes stopped at T1a were purified from unincorporated NTPs and resuspended in buffers optimized for stimulating either intrinsic cleavage or TFIIS-stimulated cleavage. After transcript cleavage (Fig. 5, lanes 1, 3, 5, and 7), the ternary complexes were washed and resuspended in standard transcription buffer containing 800 M CTP, UTP, and GTP. This subset of nucleotide substrates permits active elongation complexes to move beyond the T1a site, generating readthrough products halted downstream of T1a where the next ATP substrate would be required. In four separate experiments, the level of readthrough products (Fig. 5, lanes 2, 4, 6, and 8) generated in this assay was roughly the same, regardless of whether alkali, manganese, or TFIIS was used to stimulate transcript cleavage. These results indicate that intrinsic cleavage can be sufficient to promote readthrough for the T1a complex.
Intrinsic Cleavage in Complexes Stalled by Withholding Nucleotides-Elongation complexes stopped at T1a represent naturally arrested ternary complexes that inefficiently resume elongation in the presence of optimal NTPs; the TFIIS cofactor is essential to rescue the complexes from arrest at physiological pH. To examine whether intrinsic transcript cleavage occurs in other types of elongation complexes, five discrete complexes were examined that had been artificially halted by NTP omission. These complexes were prepared using a "walking" proto- col (42,43) along the pGR220 template (26) to form G135, U138, G143, G152, and C158 complexes (Fig. 1B). Hence, these five elongation complexes contain transcripts with 135 nucleotides of identical 5Ј proximal sequence and differ only in their 3Ј proximal sequence.
Elongation complexes vary in their susceptibility to cleavage stimulated by TFIIS (4,44,45). Likewise, the different elongation complexes tested herein varied significantly in their intrinsic cleavage activities (Fig. 6). For example, the U138 complex undergoes intrinsic cleavage at pH 9.5 within 60 min (Fig.  6A, lane 4) and in even shorter time periods (see below). In contrast, the G135 complex does not undergo detectable intrinsic transcript cleavage under these conditions (Fig. 6A, lane  11). Nevertheless, both complexes exhibited transcript cleavage in response to TFIIS within 5 min (Fig. 6A, top, lanes 6 and  13), underwent rapid pyrophosphorolysis (Fig. 6A, lanes 7 and  14), and efficiently resumed elongation upon addition of NTPs after each type of treatment (data not shown). Thus although the G135 and U138 complexes differ in their propensity to undergo intrinsic transcript cleavage, no differences were observed in other biochemical properties of these two complexes.
Intrinsic transcript cleavage was also observed with G152 and C158 complexes (Fig. 6B, lanes 4 and 6) but not in G143 elongation complexes in the timeframe of the experiment (Fig.  6B, lane 2, and data not shown). All of these complexes were active, and their transcripts could be elongated upon the addition of NTPs (data not shown). Hence, over the course of 23 bp of template sequence, only three of five elongation complexes exhibited detectable intrinsic transcript cleavage during the experiment, whereas all complexes remain competent for elongation upon addition of NTPs. The reason that the other two stable complexes were refractory to detectable intrinsic cleavage is not known. None of the five ternary complexes had arrested, and yet the catalytic center of the enzyme apparently had greater mobility to move away from the 3Ј-end of the transcript in some but not other active complexes. Both G135 and G143 are G:C-rich upstream of the stalled position. The complex most susceptible to intrinsic cleavage, U138, has three Us at the 3Ј-end of the transcript. It has been proposed that a more mobile catalytic center, and thus more cleavage, might be seen over A:T-rich sequences (46). Indeed, both G152 and C158 complexes undergo intrinsic cleavage to a more limited degree than U138, and those two complexes have terminal G or C residues.
Similarities and Differences in the Sites of Intrinsic and TFIIS-stimulated Transcript Cleavage-If the primary role of TFIIS were to stimulate the intrinsic RNA polymerase endonuclease activity, then one would expect that the pattern, but not necessarily the kinetics, of cleavage would be identical in the presence or absence of TFIIS. To test this hypothesis, we compared the cleavage patterns produced by the intrinsic reaction and by TFIIS in the T1a and U138 elongation complexes. In 3Ј-end-labeled U138 complexes, mild alkaline (pH 9.5) or manganese stimulated the release of a predominant RNA product with a mobility consistent with a trimer, although the exact size is not known. However, it is clear that intrinsic cleavage generates a set of products distinct from those released follow-  2, 4, 6, and 8, respectively). This NTP subset allows C1 (asterisks) and T1a ternary complexes to elongate past the T1a site to generate a specific readthrough product (RT).
FIG. 6. Intrinsic cleavage in elongation complexes formed by withholding nucleotides. A, lanes 1 and 8 contain purified G135 and U138 complexes formed from the pCpRW106 template (see Fig. 1B). Each complex was suspended as indicated in either standard transcription buffer (lanes 3 and 10), alkaline buffer (lanes 4 and 11), TFIIScontaining buffer (lanes 6 and 13; TFIIS at a 1:10 molar ratio to RNA polymerase II), or 1 mM pyrophosphate (PPi) containing buffer (lanes 7 and 14) for the indicated times at 30°C. B, ternary complexes were walked from U138 to template positions G143, G152, or C158 as described under "Experimental Procedures," and the complexes were purified. Aliquots from each ternary complex were resuspended in either standard transcription buffer at pH 8 (lanes 1, 3, and 5, respectively) or alkaline buffer at pH 9.5 (lanes 2, 4, and 6) and incubated for 20 min at 30°C.
ing stimulation by TFIIS (Fig. 7A, bottom, compare lanes 4 and  6 with lane 9). That is, TFIIS stimulated the release of three products, none of which migrated with the mobility of the product detected from the intrinsic cleavage reaction (Fig. 7A,  bottom, lane 9). When TFIIS was incubated with the U138 complexes under the intrinsic cleavage stimulating conditions, oligomers were the same size as those released by TFIISstimulated cleavage under standard conditions (Fig. 7A, bottom, lane 10, and data not shown). Thus, the differences in cleavage products could not be explained simply by differences in the reaction conditions. Furthermore, the efficiency of cleavage induced by TFIIS was greater than that induced by basic pH. These results strongly suggest that TFIIS influences the site of transcript hydrolysis, in addition to stimulating the rate of transcript hydrolysis.
Complexes containing 3Ј-end-labeled T1a transcripts were also examined for the sizes of liberated cleavage products (Fig.  7B, bottom). The analyses of the T1a-derived transcripts are complicated somewhat because of the presence of minor contaminating transcripts, but two general features are evident. First, many of the intrinsic and TFIIS-stimulated cleavage products from the 3Ј-end-labeled arrested T1a complexes are of the same mobility (compare lanes 3, 4, and 6 in "released products"), which is in contrast to the disparity in sizes of released cleavage products from the stalled U138 complex. Second, pyrophosphate-treated 3Ј-labeled T1a complexes also liberate two products with the same low mobility as seen with TFIIS or intrinsic cleavage (Fig. 7B, bottom, lane 7). The behavior of the U138 and T1a elongation complexes is clearly distinct for transcript cleavage. Thus, the intrinsic cleavage reaction is affected by both the site of the elongation block as well as the method used to create the block.
␣-Amanitin Inhibits TFIIS-stimulated Cleavage but Not Intrinsic Cleavage-The fungal toxin ␣-amanitin is a well known inhibitor of RNA polymerase II transcription. The toxin binds tightly to RNA polymerase II (47), blocks TFIIS-stimulated transcript cleavage for some complexes (14,48), and significantly slows elongation and pyrophosphorolysis (49). Given these effects, ␣-amanitin might also interfere with the intrinsic transcript cleavage reaction. Both the T1a and U138 ternary complexes were tested. Reactions with 150 g/ml ␣-amanitin effectively blocked TFIIS-stimulated transcript cleavage in both the T1a complex (Fig. 8, A and B, lanes 3-5) and U138 complex (Fig. 8, C and D, lane 6). Surprisingly, intrinsic transcript cleavage, stimulated by either mild alkaline or manganese at pH 8, was not inhibited by 150 g/ml ␣-amanitin in either the T1a (Fig. 8, A and B, lane 2) or U138 complexes (Fig.  8, C and D, lanes 4 and 5). For the T1a complex reconstituted in 5 mM manganese, 23 and 25% intrinsic cleavage was observed in the absence and presence, respectively, of 150 g/ml ␣-amanitin. For the alkaline-treated U138 complex, 35 and 25% intrinsic cleavage was observed in the absence and presence, respectively, of 150 g/ml ␣-amanitin. These results demonstrate that ␣-amanitin does not directly inhibit the intrinsic hydrolytic activity of RNA polymerase II. Because the toxin does inhibit TFIIS-induced cleavage, it must interfere with some step specific to TFIIS in the stimulation of transcript hydrolysis. Alternatively, these conditions of mild alkali or manganese, pH 8, might interfere with the binding of amanitin to the polymerase. However, this explanation does not hold for Mn 2ϩ , because amanitin has been shown to inhibit transcription in vitro in the presence of manganese (data not shown).

An RNA Polymerase II Mutant with Altered TFIIS Responsiveness Has Wild Type Intrinsic Cleavage Properties-Work
with mutants of TFIIS suggested that the reactivation of arrested complexes requires at least two steps that are stimulated by TFIIS: cleavage of the nascent transcript and a hypothetical conformational change within the complex (35,25). A mutant yeast RNA polymerase II lacking subunit Rpb9 (RPII⌬9) is significantly impaired in its ability to respond to TFIIS-stimulated readthrough of elongation blocks, although the mutant polymerase binds TFIIS as well as wild type polymerase (25). It seemed possible that the molecular defect of RPII⌬9 in response to TFIIS was in the intrinsic cleavage activity. However, complexes formed at U138 or T1a with RPII⌬9 had no detectable reduction in intrinsic transcript cleavage relative to the wild type enzyme (data not shown). Mild alkaline (pH 9.5) or manganese at pH 8 liberated the same intrinsic cleavage products at about the same rate and extent as with the wild type enzyme (Table I). Also, for both U138 and T1a complexes the intrinsic cleavage reaction was sufficient to promote readthrough with RPII⌬9 (Table I provides data with  T1a complexes). Thus, there appears to be no inherent defect in the intrinsic hydrolytic activity in RPII⌬9. The impaired ability of this mutant polymerase to respond to TFIIS cannot be easily explained by these results. Further work using TFIIS mutants in combination with RPII⌬9 should be informative.
Previous studies demonstrated that a mutant RNA polymerase II lacking both the RPB4-and RPB7-encoded subunits is similar to wild type enzyme in the rate of transcript elongation, the recognition of intrinsic blocks to elongation, and the response to TFIIS-stimulated cleavage and readthrough (14,50). The intrinsic cleavage properties of this enzyme were tested with the U138 complex, and both the intrinsic and the TFIISstimulated cleavage products and rates were indistinguishable from that of the wild type enzyme (data not shown). Thus, RPB4, RPB7, and RPB9 are dispensable for intrinsic transcript cleavage by RNA polymerase II.

DISCUSSION
This work was carried out to explore the mechanisms of pausing, arrest, and readthrough by RNA polymerase II. The protein factor, TFIIS, promotes readthrough of RNA polymerase II ternary elongation complexes that have arrested in vitro. This factor stimulates transcript cleavage at the 3Ј-end of the nascent transcript as a prerequisite for readthrough. The presented work focused on the factor-independent, intrinsic ability of RNA polymerase II to carry out this transcript cleavage reaction in both arrested and stalled complexes. Comparing intrinsic and TFIIS-induced cleavage in these two types of complexes has identified both similarities and differences that highlight what parts of the readthrough reaction are ripe for regulation.
The transcript cleavage reaction has been conserved among all tested RNA polymerases and is important in the control of RNA elongation (16). For RNA polymerase II, a fraction of ternary complexes cease elongation within the transcription unit in a regulated fashion (1). In vivo, it is experimentally difficult to determine if these complexes have paused or have arrested. In vitro, arrested complexes appear to require transcript cleavage as a prerequisite for the resumption of elongation (13,37,51), and, at physiological pH, stimulation of cleavage requires TFIIS (3,52). However, by altering the biochemical conditions, intrinsic cleavage by the polymerase can be accelerated in arrested or stalled complexes, and this cleavage can be sufficient for promoting continued elongation.
In contrast to arrested complexes, interruption of transcript elongation by limiting subsets of NTPs often results in stable, paused elongation complexes that readily resume elongation upon addition of NTPs (Ref. 4, although see Ref. 53). These active ternary complexes often will arrest if incubated for an extended period in the absence of the next nucleotide needed for elongation (54 -56). However, only a subset, of the complexes stalled by depriving them of nucleotides, detectably carries out the intrinsic cleavage reaction. At the least, these observations indicate that stable, paused elongation complexes differ kinetically in the intrinsic cleavage reaction, and it remains possible that some complexes simply will not carry out this reaction. These results raise several questions. How do arrested complexes differ from stalled elongation complexes in their susceptibility to intrinsic cleavage? Why do different stalled elongation complexes differ in their susceptibility to intrinsic cleavage? Transcending these mechanistic questions is the regulatory dilemma: how can the cleavage reaction intrinsic to RNA polymerase II be stimulated by protein factors such as TFIIS?
There are several models proposed to explain such functional differences (4,46,(57)(58)(59). All of the models suggest that hydrolytic cleavage of the nascent transcript within an arrested or stalled ternary complex generates a new transcript 3Ј-end at the active site. As hypothesized, the cleavage reaction becomes necessary as the catalytic center of the polymerase is proposed to be mobile and can move away from the 3Ј-OH necessary for continued catalysis. This mobility can be in a readily reversible equilibrium, or it can result in a catalytic center arrested away from the transcript 3Ј-end. Transcript cleavage generates a new 3Ј-end, immediately proximal to the catalytic center, which permits continued incorporation of nucleotide substrates for catalytic elongation of the transcript.
A focus on the intrinsic cleavage reaction in this work removed one additional level of biochemical complexity, the stimulation of this reaction by protein factors. However, the comparison between the reaction intrinsic to the polymerase and that accelerated by TFIIS was essential to evaluate what was carried out solely by the polymerase versus what was affected by the regulatory protein, TFIIS.
As would be anticipated for a hydrolysis reaction, mild alkali stimulated intrinsic cleavage in several yeast RNA polymerase II ternary complexes. However, the present analysis does not distinguish between general base catalysis and specific base catalysis stimulated by elevated pH. Specific base catalysis has been proposed for the 3Ј35Ј-exonuclease (proofreading) active site of the Klenow fragment, in which the two divalent metal ions in the exonuclease active site of DNA polymerase I apparently promote rapid hydrolytic cleavage of the phosphodiester bond (60). In contrast to specific base-catalyzed reactions, if an active site functional group were participating in catalysis as a general base, the rate of the reaction should become independent of pH after the pK a of the functional group is reached (61). For ternary transcription complexes, intrinsic transcript cleavage was significantly stimulated over the range of pH 8.0 -9.5, but further elevation in pH (10.5) was only slightly more effective than pH 9.5 for intrinsic cleavage. However, at pH 10.5 a significant proportion (50%) of RNA polymerase II ternary complexes were inactivated for subsequent elongation (data not shown). A similar pH plateau has been observed for intrinsic cleavage with E. coli RNA polymerase ternary complexes (20), and these authors proposed that titration of a specific amino acid (i.e. histidine, lysine, and cysteine) is more likely than general (passive) base catalysis. It is not possible to distinguish between general and specific base-catalyzed hydrolysis reactions for the yeast RNA polymerase II, because the polymerase is inactivated in strong alkaline solutions, and there is differential stimulation of cleavage by some divalent cations.
The conditions identified here that stimulate intrinsic tran- script cleavage are consistent with the prior work leading to the proposal that transcript cleavage is mediated at the active site of RNA polymerase. Pyrophosphorolysis can liberate large RNA oligomers from complexes stopped at natural blocks to elongation (this work) or by nucleotide exclusion (21). Furthermore, the mobilities of the RNA products that result from transcript cleavage within a single arrested ternary complex are similar whether cleavage is mediated intrinsically, by TFIIS, or by pyrophosphate. The same subset of divalent metal cofactors that support (Mg 2ϩ , Mn 2ϩ , and Co 2ϩ ) and inhibit (Ca 2ϩ and Zn 2ϩ ) the TFIIS-mediated transcript cleavage reaction also support and inhibit the intrinsic reaction. This observation supports the notion that TFIIS, which accelerates the rate of transcript cleavage, acts through the intrinsic hydrolytic site of RNA polymerase II. However, the TFIIS-mediated transcript cleavage reaction demonstrated a divalent cation preference (Mg 2ϩ Ͼ Mn 2ϩ Ͼ Co 2ϩ ) distinct from that of the intrinsic cleavage reaction (Co 2ϩ Ͼ Mn 2ϩ Ͼ Mg 2ϩ ) at physiological pH (data not shown). Several differences also were observed between the intrinsic transcript cleavage properties of the complexes arrested at the T1a site and stalled complexes formed by nucleotide exclusion. First, detectable levels of intrinsic cleavage occurred in only three of the five tested elongation competent complexes (U138, G152, and C158). Indeed for the G135 complex, intrinsic transcript cleavage was not detected over the course of 1 h at pH 9.5. Despite their distinct intrinsic cleavage properties, both the G135 and U138 complexes were responsive to TFIIS-stimulated transcript cleavage, efficiently underwent pyrophosphorolysis, and were equally capable of resuming elongation. What the different propensity for intrinsic cleavage reflects functionally or structurally in the elongation complex is not known. As mentioned above, it remains possible that intrinsic cleavage occurs at a significantly reduced rate in some complexes, but these results clearly emphasize that ternary complexes can differ substantially in the process of transcript cleavage. Distinct differences also have been seen between mammalian RNA polymerase II ternary complexes following TFIIS-stimulated transcript cleavage (4,44,45). If the catalytic site of RNA polymerase does mediate transcript hydrolysis, such dramatic differences in the intrinsic cleavage reaction likely reflect differences in the relative mobility of the catalytic site, or transcript, in different stalled elongation complexes (4,46).
Nuclease activities are associated with many nucleic acid polymerases, including DNA polymerases and reverse transcriptases, and provide extremely important functions during the polymerization of nucleic acids. However, the intimate association between the polymerization and hydrolytic sites in RNA polymerase II appears unique. Perhaps the non-distributive nature of RNA polymerization has enforced tight linkage between these catalytic sites evolutionarily.
How TFIIS influences the site of transcript cleavage is not known. A priori the expectation would be that TFIIS would simply stimulate the intrinsic cleavage reaction, and, other than reaction rate, no differences in cleavage pattern or released products would be seen with and without TFIIS. However, the distinct sizes of released cleavage products from stalled complexes generated with and without TFIIS suggest there is another layer of mechanistic complexity. Reines and colleagues (62) have reported that human TFIIS cross-links to the 3Ј-terminal nucleotide of the transcript in an arrested complex, and thus TFIIS is in close proximity with the RNA, and perhaps the polymerase active site. TFIIS has a low affinity for single-strand nucleic acids in solution. However, if there is a relevant nucleic acid binding domain in TFIIS, it might be "unmasked" upon interaction with the ternary complex (63,64). It remains to be determined whether the apparent differences in the site of transcript cleavage are the consequence of direct interaction of TFIIS with RNA in the ternary complex or a conformational alteration induced by TFIIS upon binding to RNA polymerase. TFIIS also can induce the cleavage reaction when purified RNA polymerase II is bound to purified RNA in the absence of DNA, and the polymerase can then add nucleotides to the newly formed 3Ј-end (6). Thus, the TFIIS stimulation of the cleavage reaction does not require a ternary complex but can operate directly on the polymerase when bound to RNA.
Alternatively, it may be that TFIIS-mediated cleavage is influenced by RNA sequence (4), and certain RNA sequence contexts may have greater, or lesser, lability to hydrolysis. In this regard, it is noteworthy that the pattern of DNase I protection conferred by the U138 complex with mammalian RNA polymerase II is strikingly different from that of G135 and G143 complexes, whose DNA footprints are quite similar (65). In relation to the catalytic site, the DNase I footprint of the leading edge of the U138 complex is shifted upstream by ϳ5 bp compared with the G135 and G143 complexes. Mammalian RNA polymerase II U138 complexes, like the yeast ternary complexes, are also susceptible to intrinsic transcript cleavage (data not shown). Perhaps the efficiency with which ternary complexes carry out intrinsic transcript cleavage correlates with particular types of structural alterations, such as the mobility of the catalytic center forward or back on the DNA template (46), or the positioning of the RNA transcript within the proposed RNA exit channel (66), or extruded from the ternary complex (58). Clearly all these models make testable predictions for future work.
In addition to the rate of hydrolysis, the particular site of transcript cleavage in the ternary complex can be influenced by protein factors. In U138 ternary complexes, TFIIS-induced cleavage generates RNA products distinct from those produced by the polymerase in the absence of TFIIS. This effect on the site of cleavage is also seen with two factors for E. coli RNA polymerase, GreA and GreB. The actions of these two proteins can be distinguished by the distinct sizes of cleavage products generated in their presence from the same ternary complex (67). A distinction in sizes of released products in stalled versus arrested complexes has been documented in ternary complexes containing mammalian RNA polymerase II as well (4). However, for both the bacterial and the eukaryotic enzymes, a size range of released cleavage products can be found from arrested and stalled complexes (Ref. 4 and this work). 2 Thus, there is unlikely to be a discrete differential between arrested and stalled complexes but rather a continuum of cleavage products determined by the many influences that define a ternary elongation complex. Additionally, the prokaryotic elongation factor NusA can change the site of transcript cleavage within a ternary complex (68). 2 Another difference between TFIIS-stimulated and intrinsic cleavage is the effect of ␣-amanitin. This fungal toxin blocks TFIIS-stimulated transcript cleavage, without significantly affecting the rate of intrinsic cleavage. The toxin, ␣-amanitin, binds tightly to the largest subunit of RNA polymerase II (47,69) and dramatically slows elongation and pyrophosphorolysis (48,49). However, it appears to interfere with translocation rather than phosphodiester bond formation (70). That neither intrinsic cleavage nor polymerization is inhibited by ␣-amanitin might be expected if the hydrolytic and polymerase activities are catalyzed by the same active site.
Why ␣-amanitin interferes with TFIIS function but not with intrinsic cleavage is of interest. The elongation factors ELL, TFIIF, and elongin also interfere with TFIIS-stimulated cleavage (71); whether these factors also interfere with intrinsic cleavage is not known. The toxin ␣-amanitin, a bicyclic octapeptide, could physically interfere with the approach of TFIIS to the active site of the polymerase or the 3Ј terminus of the RNA. Alternatively, TFIIS-stimulated cleavage might require a conformational change of the polymerase, which ␣-amanitin restricts, directly or indirectly, through binding to a region of the polymerase important in transcript elongation and termination (72). Indeed ␣-amanitin-resistant mutants cluster in a conserved region of the largest subunit of the polymerase; some mutations in this region of the largest subunit of RNA polymerase III enhance transcript cleavage by 10-fold (73). RNA polymerase II lacking the RPB9-encoded subunit is significantly impaired in its ability to respond to TFIIS for readthrough stimulation (25) and is less responsive to TFIISmediated transcript cleavage. However the Rpb9p-deficient polymerase has no distinguishable difference in the rate of intrinsic transcript cleavage within T1a or U138 complexes. Furthermore, the cleavage products released from U138 or T1a ternary complexes containing this mutant polymerase are the same as those released from complexes containing wild type polymerase. Thus, this polymerase subunit is not necessary for catalyzing intrinsic transcript cleavage, nor in the TFIIS-induced repositioning of the site of transcript hydrolysis. However, it appears to transmit the signal from TFIIS to the rest of the polymerase (25,41).
Curiously, the homologous subunit in RNA polymerase III, C11, also is essential for the intrinsic cleavage reaction carried out by that polymerase (74). However, there is no TFIIS-like cofactor required to stimulate cleavage for RNA polymerase III. Indeed, RNA polymerase III carries out this reaction in vitro under physiological conditions in the absence of an additional protein factor.
The mutant polymerase, RPII⌬9, possesses "wild type" intrinsic cleavage activity (this work) and binds TFIIS efficiently, yet RPII⌬9 responds poorly to TFIIS as assayed by either cleavage or readthrough (22). These results suggested that Rpb9p transmits the signal from TFIIS to the active site of the polymerase rather than impacting the intrinsic ability of the polymerase to cleave the nascent transcript. For both the wild type polymerase and RPII⌬9, intrinsic cleavage was sufficient to promote readthrough. However, some mutants in TFIIS that stimulate cleavage in arrested ternary complexes nonetheless do not promote readthrough (25,35,40). The effect of these mutants led to the hypothesis that TFIIS not only stimulates cleavage but also promotes an additional conformational change needed for efficient readthrough. Clearly this work indicates that the intrinsic cleavage event itself is sufficient to allow readthrough, although we cannot rule out that elevated pH may affect some other feature of the polymerase in addition to stimulating its cleavage activity. However, there were distinct differences observed between intrinsic and TFIIS-stimulated cleavage. First, within stalled but active elongation complexes, the pattern of intrinsic cleavage differed from that generated by TFIIS. In contrast, arrested complexes responded similarly in the intrinsic and the TFIIS-stimulated reactions. The specific molecular distinctions between the stalled and arrested complexes await discovery, and these distinctions will likely explain the differences seen here. Second, the drug ␣-amanitin, which blocks the translocation step of elongation but not the polymerization step, had little to no effect on the intrinsic cleavage reaction in stalled or arrested complexes yet strongly inhibited TFIIS-dependent reactions.
How have the properties of the intrinsic cleavage reaction reported here and elsewhere (20,21) informed the elongation mechanism of RNA polymerase? First, the results reinforce the kaleidoscope of potential differences among ternary elongation complexes. It is difficult to define "rules" that pertain always to arrested complexes or always to stalled complexes re: kinetics or released products from intrinsic or factor-induced cleavages. The potential flexibility of regulatory targets is significant for the cell but difficult for the experimentalist. Second, for yeast RNA polymerase II, the subunit encoded by RPB9 is not a major contributor to the transcript cleavage reaction. This input was a possibility because of the phenotype of cells without this subunit and because of the biochemistry of transcription reactions with polymerase lacking this subunit. This subunit clearly remains a mediator between the cleavage stimulatory factor TFIIS and the remaining subunits of the polymerase. Although Rpb9 and TFIIS bind in distinct locations on RNA polymerase II, 3 they share very similar regions of amino acid sequence that contribute to elongation stimulation (41). However, the intrinsic transcript cleavage reaction itself is clearly unaffected by the absence of Rpb9.
The plethora of recently reported crystal structures allows predictions about regions of the polymerase that are involved in the intrinsic cleavage reaction. These regions are likely targets of the activity of TFIIS. The stability of arrested complexes makes them good candidates for structural work as well. Such structures not only will pinpoint amino acid residues important in transcript hydrolysis by the polymerase but also may provide an explanation for how TFIIS stimulates this reaction.