Formation and Crystallization of Yeast RNA Polymerase II Elongation Complexes*

Minimal templates were devised for the efficient generation of yeast RNA polymerase II transcription elongation complexes. A 33-base pair DNA with a 15-residue dC tail at one 3′-end supported the formation of a complex containing the polymerase paused at nucleotide 11 of the duplex region and an RNA of 14–16 residues. The same template could yield an arrested complex with the enzyme at nucleotide 13–15 and RNA of 15–17 residues. These complexes were stable for at least a week under various conditions and could be resolved by gel electrophoresis or purified by ion exchange chromatography. The purified paused complex formed crystals capable of x-ray diffraction to 3.5 Å resolution. The complex remained active in the crystal and, in the presence of nucleoside triphosphates, could efficiently extend the transcript in situ.

RNA chain elongation, once viewed as a monotonous process of ribonucleotide polymerization, now appears mechanistically varied, with important regulatory consequences. RNA polymerase may be unable to extend an RNA chain due to impediments in the DNA template, such as special sequences, damage sites, or protein binding sites (1,2). The resulting halted complexes may become arrested as a consequence of the polymerase translocating backwards on the template, extruding the 3Ј-(growing) end of the RNA chain (3,4). Release from arrest is effected by the elongation factors TFIIS in eukaryotes (5)(6)(7) and GreA/B in bacteria (8), which enable the cleavage of extruded RNA and resumption of chain elongation from the newly generated 3Ј-end (9,10).
Pausing and arrest afford opportunities for regulatory intervention. Pausing immediately downstream of promoters may be a general rate-limiting step following initiation (11). Genespecific activator proteins, and also general factors such as elongin, stimulate the elongation process (12)(13)(14)(15). Elongin function can play a role in tumorigenesis. Elongin is bound and inhibited by the VHL tumor suppressor protein, and mutations of the VHL protein found in tumors prevent binding (16,17).
The mechanism and regulation of elongation have been pursued by studies of isolated elongation complexes. In the case of the bacterial enzyme, the complexes are generally formed by promoter-dependent transcription, halted by withholding a single nucleoside triphosphate (18). The halted complexes have been extensively characterized by enzymatic and chemical footprinting, chemical cross-linking, electron microscopy, and other means (19 -22). The complexes contain an unwound region ("bubble") of about 15 base pairs of DNA around the nucleotide addition site (20,23,24) and nascent RNA hybridized to some extent with the template strand of the DNA (20,25). Outstanding questions include the paths followed by DNA and RNA, the length of the DNA-RNA hybrid, and the mechanisms of unwinding and rewinding of DNA/DNA and RNA/ DNA duplexes.
Answers to these questions will come from determination of the three-dimensional structure of halted complexes. To this end, we have formed halted and arrested complexes with yeast RNA polymerase II. Halted complexes are more difficult to prepare with eukaryotic than bacterial polymerases because initiation at eukaryotic promoters is far less efficient and requires multiple general initiation factors. Here we employ "tailed" templates, bearing 3Ј-protruding single stranded ends, which support transcription without general initiation factors, starting at a cluster of sites about three residues from the single strand-double strand junction (26). We report on characteristics of polymerase II elongation complexes pertinent to their crystallization and structure determination.

EXPERIMENTAL PROCEDURES
Transcription Templates-A 220-base pair fragment of pGR220 (kindly provided by Dr. Caroline Kane, University of California, Berkeley), containing a 135-residue T-less cassette, was prepared with a 3Ј-dC tail by cleavage with AvaI, dephosphorylation with shrimp alkaline phosphatase (U. S. Biochemical Corp.), heat inactivation of the phosphatase, ligation to a phosphorylated synthetic oligonucleotide with an AvaI end and 18 dC residues, cleavage with KpnI, and purification by electrophoresis in low melting agarose, removed by treatment with Gelase (Epicentre Technologies). Plasmids containing shortened T-less cassettes were constructed from the 220-base pair fragment of pGR220 by PCR with the use of oligonucleotide primers containing AvaI and KpnI sites, yielding V51 (from residues 113 to 158 of the pGR220 fragment), V102 (from residues 113 to 206), and V115 (from residues 97 to 206).
Trityl purification of synthetic oligonucleotides (Fig. 1, Anagene Inc.) was sufficient for transcription experiments. Further purification in 6 M urea, 19% polyacrylamide gels was done for purposes of crystallization. Oligonucleotides were electroeluted, dried, dissolved in water, and desalted by filtration through Sephadex G-50. Complementary oligonucleotides, in equimolar amounts on the basis of optical density measurements, were hybridized in 30 mM ammonium sulfate by heating to 80°C for 5 min and cooling to 20°C. The presence of a small excess of one strand had no deleterious effect on transcription assays, presumably because their affinity for the polymerase was considerably lower than that of the tailed templates.
RNA Polymerase II-Enzyme lacking subunits 4 and 7 was purified from Saccharomyces cerevisiae by immunoaffinity chromatography as described (27) with minor modifications. Polymerase was eluted from the antibody column with 40% propylene glycol instead of glycerol (28). Protein for use in crystallization trials was futher purified on either Poros (HE1) heparin or DEAE. Purified polymerase was precipitated by the addition of an equal volume of saturated ammonium sulfate, frozen in liquid nitrogen, and stored at Ϫ80°C.
Transcription-RNA polymerase II (0.1-1.5 g) was incubated with * This work was supported in part by National Institutes of Health Grants GM49985 and AI21144 (to R. D. K.). 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. ). Halted complexes were formed by omission of UTP. After 30 min at room temperature or 20 min at 28°C, reactions were terminated by the addition of SDS to 0.4%, proteinase K to 0.8 g/ml, and incubation for 30 min at 40°C in a total of 100 l. RNA was extracted with a half volume of phenol, followed by extraction with a half volume of chloroform and precipitation in the presence of 1 g of glycogen, 0.3 M sodium chloride, and 2.4 volumes of ethanol for 16 h at Ϫ20°C. Pellets were collected by centrifugation for 30 min at 4°C and resuspended in 13 l of formamide loading buffer, and 7 l was loaded on a 6 M urea, 15-19% polyacrylamide gel, developed at 50°C until the bromphenol blue dye reached the bottom. A mixture of oligonucleotides ranging from 15 to 34 bases, end-labeled as described (29), served as size markers.
Reaction mixtures containing hexahistidine-tagged polymerase (30) and ␤-mercaptoethanol instead of DTT were incubated with an equal volume of Ni 2ϩ -nitriloacetic acid resin (Qiagen), previously equilibrated in the reaction buffer, for 10 min at room temperature with gentle agitation. The resin was washed three times (or more if needed to remove radioactivity from the supernatant) by suspension in 1 ml of buffer and centrifugation for 30 s. The resin was suspended in buffer (25 l) with nucleotides at 1 mM or as indicated, and reaction was allowed to continue for an additional 20 min at 28°C, followed by extraction and analysis of RNA as above.
For quantitation, halted complexes were separated from free nucleotides by filtration through Sephadex G-50, concentrated by ultrafiltration (Microcon 100; Amicon) and passed through a heparin-Sepharose 6B column. Radioactivity incorporated in halted complexes was compared with total protein determined by the Bradford method (38).
Crystallization-A transcription reaction mixture (1 ml) containing RNA polymerase (1 mg), a 1.5-3-fold molar excess of template E (see below), and 1 mM ribonucleotides without UTP was incubated for 20 min at 28°C and then passed through a 1-ml heparin HiTrap heparin column (Pharmacia). The yield of homogenous paused complex was about 50% of the starting polymerase. The buffer was changed to 5 mM Tris, pH 7.5, 40 mM ammonium sulfate, 10 mM DTT, 50 M zinc sulfate, 50 M EDTA by filtration with a Microcon 100. Crystallization was effected as described (31) with the addition of PEG 6000 to lower the ammonium sulfate concentration to 200 mM, preventing dissociation of the complex.

Formation and Stability of Paused Elongation Complexes-
Yeast RNA polymerase II lacking subunits Rpb4 and Rpb7 was employed throughout this work because of its homogeneity and its propensity to form both two-and three-dimensional crystals (31). While Rpb4 and Rpb7 are required for promoter-dependent initiation (32), polymerase lacking these subunits is able to transcribe tailed templates and is indistinguishable from wildtype enzyme in RNA chain elongation (not shown). For initial characterization, enzyme with a hexahistidine tag at the C terminus of the largest subunit, Rpb1, was used, but in subsequent work directed toward crystallization, the tag was omitted.
Initial studies employed a plasmid containing a unique AvaI site, adjacent to a 135-bp T-less cassette, terminated by three T residues. Following AvaI digestion and ligation with a synthetic oligonucleotide to introduce a 3Ј-protruding tail of 18 C residues, transcription in the absence of UTP generated a 135nucleotide product (Fig. 2), as described by others (25). No transcript from the opposite end of the DNA, also bearing a C tail, was detected, presumably due to the presence of several T residues in the coding strand near this end, resulting in the formation of unstable elongation complexes. The 135-nucleotide RNA resided in stable elongation complexes, as shown by 1 The abbreviations used are: DTT, dithiothreitol; bp, base pair(s).
FIG. 1. Synthetic transcription templates used in this work. Synthetic templates A-F are aligned at their halt sites, consisting of at least three T residues in the nontemplate strand. The V51 template, derived from the T-less cassette in pGR220 sequence, was prepared by restriction enzyme digestion and ligation of a C tail. Templates A-D are based on the pGR220 sequence. Template F retains some similar sequences at the beginning of the double-stranded region but is different afterward. Template E is based on another oligonucleotide (not shown) whose sequence upstream of the halt site induced a high level of arrest. To that sequence was added an oligo(T) stretch to cause nearly total arrest. three further findings. First, the RNA remained associated with the hexahistidine-tagged polymerase following adsorption on Ni 2ϩ -agarose and removal of unbound material. Second, incubation of the Ni 2ϩ -bound elongation complex with UTP extended the transcript by three residues (Fig. 2, first lane), as expected if pausing occurred precisely at the end of the T-less cassette. Third, incubation of the Ni 2ϩ -bound elongation complex with all four nucleoside triphosphates resulted in conversion of most of the RNA to the length expected for runoff at the opposite end of the DNA or shorter due to arrest between the pause site and the end (Fig. 2).
The fraction of paused complexes capable of elongation varied from 50 to 95%, depending on the manner of preparation. For example, vigorous vortex mixing or exposure to elevated ionic strength diminished the number of active complexes. Manipulations such as chromatography and filtration, however, had no deleterious effect. Even following washes of Ni 2ϩ -bound halted ternary complex with 2 M potassium acetate, 32% of paused complexes could be elongated (Fig. 2). Complexes were also stable over time, showing no significant loss of activity after a week or more at room temperature (see below).
Minimal Template Size-With a view to eventual crystallization of paused complexes, which might be prevented by protruding DNA, we wished to determine the minimal size of template required for complex formation. DNase I footprint analyses (33), showing protection of 20 -24 bp upstream and 19 -23 bp downstream of the pause site, suggested the size could be reduced to about 50 bp, if not further. A nested set of templates derived from the T-less cassette by insertion of restriction sites showed no loss of complex formation, stability, or capacity to elongate upon reduction of the template to a 51-bp fragment with the pause site in the middle, at nucleotide 27 (Fig. 3). Smaller templates were obtained by oligonucleotide synthesis (Figs. 1 and 4). Upon removal of 5 bp from the upstream end and 2 bp from the downstream end of the 51-bp sequence, 47% of the paused complexes were still capable of elongation (Fig. 4). Further removal of 6 and 10 bp from the downstream end had no effect on complex formation but diminished elongation to 9 and 5%, respectively. A length of 22 bp downstream of the halt site was therefore taken as the minimum for the generation of fully active complexes.
Transcripts from shorter templates were better resolved in gels (Figs. 3 and 4), revealing a small degree of length heterogeneity, arising from initiation at a variable location in the single-stranded tail 3-5 residues upstream of the junction with double-stranded DNA. This heterogeneity is a general characteristic of transcription from tailed templates (34). It posed no impediment to crystallization (see below).
Early Arrest Site-All templates derived from the T-less cassette yielded not only the desired products, extending to the halt site, but also a set of shorter RNAs, 15-20 residues in length (Figs. 3 and 4). These shorter RNAs did not derive from abortive transcription and release from the polymerase since they were retained with the hexahistidine-tagged enzyme on Ni 2ϩ -agarose. They evidently resided in arrested complexes since they were not extended by incubation with all four nucleoside triphosphates and since they could be reactivated by the addition of TFIIS (Fig. 5).
There were only small changes in the pattern of arrested transcripts arising from the differences in length and sequence of the several templates used. The arrest site was reached when the RNA transcript was 16 -18 bases, and taking into FIG . 3. Transcription of plasmid-derived templates. Elongation complexes were formed from templates denoted by numbers above the lanes at the left and diagrammed at the right. Complexes were halted (H) at the first residue of the T-less cassette, adsorbed to Ni 2ϩ -agarose, and washed in transcription buffer. In some cases, this was followed by elongation (E) with all four nucleoside triphosphates (over 90% efficiency of elongation) on the agarose beads. Markers (M) were endlabeled oligonucleotides of 15, 20, 26, 29, and 34 residues. Templates were AvaI-KpnI fragments containing segments of T-less cassette (open bar; numbers at left and right are first and last residues of pGR220 insert), denoted V51 and V102 (numbered according to first residue of T-less sequence), and a plasmid derived from pGR220 as indicated, linearized with AvaI, denoted V115. Two kinds of arrest sites may be noted: a variable early arrest site with an RNA species of 15-16 bases in V51 and V102 and a late arrest site upon elongation with the same templates. account the variation in transcription start site mentioned above, the polymerase was about 13-15 residues from the beginning of the double-stranded region of the templates. This is in good agreement with the location of early arrest sites reported by others (35). Such early arrest is an intrinsic feature of transcription from tailed templates, which is not found for promoter-dependent transcription by the polymerase and general transcription factors. 2 Since heterogeneity in the population of paused complexes formed on tailed templates due to arrest was likely to interfere with crystallization, we sought to avoid the problem by placing the pause site before the early arrest site. The pause site, however, could not be located too far upstream or the transcript would be too short to form a stable complex. These considerations dictated the design of a template (denoted F in Fig. 6) with the pause site at nucleotide 11 of the double-stranded region. Transcription of this template in the absence of UTP yielded an RNA of about 14 residues (Fig.  1, template F, and Fig. 5). Elongation in the presence of all four nucleoside triphosphates was highly efficient, with about 43% arrest at the early site and most of the remainder continuing to run-off. A variant of this template, known to enhance arrest, was prepared with an oligo(T) tract at the arrest site (7,36) and with additional arrest-inducing sequences upstream of the halt site (template E in Fig. 6) to obtain more uniform behavior during elongation of the paused complex. Transcription again yielded an RNA of about 14 -15 residues (Fig. 6), and elongation resulted in almost complete conversion to the arrested complex.
Purification of Paused Elongation Complex-Template E was used in further work directed toward crystallization, with a view to structure determination of both paused and arrested complexes. The remaining source of heterogeneity to be addressed before crystallization trials was free polymerase in the paused complex preparation, due to the unavoidable presence of inactive polymerase molecules. To gauge the magnitude of the problem, the paused complex preparation was freed of unincorporated radiolabeled nucleotides by gel filtration, and 2 D. Price, personal communication. the ratio of radioactivity to protein was determined. The results for several preparations ranged from 50 -95% of the polymerase in paused complexes. This contrasted with previous estimates of 15-55% (18). We therefore sought another approach to the detection and quantitation of paused complexes. We found agarose gel electrophoresis effective for resolving paused complexes from polymerase-template complexes and free polymerase. Paused complexes formed a discrete band, while polymerase-template complexes and free polymerase trailed through the gel, due perhaps to polymerase-agarose interaction (Fig. 7). The gel electrophoretic analysis confirmed the occurrence of a majority of the polymerase in paused complexes.
We turned next to the removal of residual-free (inactive) polymerase. In light of reports that heparin inhibits transcription initiation but not elongation, we thought that heparin-Sepharose might bind free polymerase while allowing elongation complexes to flow through. Indeed, about 10 -25% of the polymerase in a paused complex preparation bound to heparin-Sepharose while the remainder, including all the radiolabeled RNA, flowed through (Fig. 8). The RNA to polymerase ratio in the flow-through was 1.13, compared with 0.8 for the material loaded on the column. (As a control, uncomplexed polymerase was applied and virtually all bound to the column.) The efficacy of this procedure was especially apparent when partially proteolyzed polymerase was used. SDS-PAGE revealed the removal of almost all degraded protein from the heparin-Sepharose flow-through (Fig. 9), indicating that only intact polymerase formed stable elongation complexes.
Crystallization of Paused Elongation Complexes-Paused elongation complexes prepared from template E and purified by passage through heparin-Sepharose formed plate-like crystals of approximate dimensions of 150 ϫ 60 ϫ 500 m, some reaching 800 m in length. About 10% of the plates were single crystals, and the remainder were multilayered (Fig. 10). Crystals were gradually adapted to mother liquor containing cryoprotectant. Cryoprotected crystals were flash frozen in liquid nitrogen using the flash cooling method (37). Diffraction data obtained with the synchrotron source at CHESS (beamline F1) were complete to about 6 Å and extended in some directions to 3.5 Å resolution. Processing with DENZO revealed a unit cell with parameters a ϭ 197 Å, b ϭ 220 Å, c ϭ 203 Å, and ␤ ϭ 103°a nd space group C2. The presence of paused elongation complexes in the crystals was confirmed by incubating a crystal in the presence of ATP, CTP, and [ 32 P]UTP, allowing elongation of existing RNA chains but not the initiation of new chains. The RNA product was detected by gel electrophoresis and autoradiography (Fig.  11). Extensive elongation of RNA in the crystal but not in the mother liquor was observed. Evidently, halted complexes were capable of resuming and completing RNA synthesis in the crystal. DISCUSSION Novel aspects of this work include the development of tailed templates for forming homogeneous paused and arrested RNA polymerase II transcription complexes, a gel electrophoretic FIG. 7. Resolution of RNA polymerase (P), polymerase-DNA complex (denoted "B" for binary complex), and halted elongation complex (denoted "T" for ternary complex) by agarose gel electrophoresis. Template E was employed for both binary and ternary complex. Gel was stained for protein (shown at left) and was autoradiographed (right) to reveal 32 P incorporated in RNA or DNA as indicated.
FIG. 8. Purification of halted elongation complexes by heparin-Sepharose chromatography. Complexes were formed with template E and 32 P-labeled nucleoside triphosphate, were freed of unincorporated nucleotides by filtration through Sephadex G-50, and were applied to the heparin column in 60 mM ammonium sulfate. After washing with transcription buffer containing 60 mM ammonium sulfate, the column was eluted with buffer containing 500 mM ammonium sulfate. Protein was determined by the Bradford method (38), and amounts of RNA are given in arbitrary units.
FIG. 9. Enrichment of intact RNA polymerase II in purified halted elongation complexes. Partially proteolyzed polymerase was employed for halted complex formation and heparin-Sepharose chromatography as in Fig. 8. The starting polymerase (Load) and purified halted complexes (Eluate) were analyzed in an SDS-10% polyacrylamide gel and stained with Coomassie Blue. Bands due to polymerase subunits Rpb1-8 (RBP1-8) are indicated. method for resolving the complexes, the use of heparin-Sepharose for separation from inactive polymerase molecules, and crystallization of paused complexes in a transcriptionally active state. The high stability of the complexes was anticipated from previous studies of the Escherichia coli enzyme (18). The fraction of active polymerase molecules, however, about 80%, was surprising in light of previous estimates for both yeast and E. coli enzymes of 15-55% (18). The reason for the discrepancy may lie in the methods of estimation. Our figure was obtained from the ratio of RNA to protein in the paused complex preparation, supported by direct visualization in agarose gels and by the results of heparin-Sepharose chromatography. The previous numbers relied on quantitation of RNA in gels, which may underestimate the amount, or on measurements of elongation rates, which may vary with the precise conditions used.
Early arrest at a transcript length of about 16 residues is evidently a general property of tailed templates (35). The similarity to the size of the unwound "bubble" in transcription complexes suggests a possible structural basis for arrest. Coding and noncoding strands ordinarily reassociate upstream of the bubble, for example in promoter-driven transcription. Such reassociation can only occur for tailed templates after transcription to about residue 16 and may fail for lack of suitable approximation of the strands, leading to arrest. Subtle differences in the pattern of arrest sites for different templates noted here may be due to the sequence dependence of DNA reassociation and may also reflect template dependence of polymerase conformational changes between elongating and arrested states.
The remarkable outcome of this work is the formation of paused complex crystals capable of diffraction to 3.5Å resolution. In view of the possibilities for modification of DNA and RNA with heavy atom compounds, as well as structural information forthcoming from studies of the native polymerase and polymerase-DNA complexes, 3 the prospects for structure determination of the paused complex appear favorable. The activity of the paused complex in the crystal, continuing RNA synthesis in situ with no adverse affect on crystal morphology, holds the further promise of structural analysis of the transcription mechanism. FIG. 11. Transcriptional activity of crystalline elongation complexes in situ. Intact crystals (X) of halted complexes formed on template E, as well as supernatant (S) from the crystallization well, and crystals dissolved in transcription buffer with 30 mM ammonium sulfate, along with the corresponding supernatant, were elongated with a mixture of ATP, CTP, and [␣-32 P]UTP, which supports synthesis of residues 15-29 but not initiation. Positions of markers as in Fig. 3 are indicated.