Two Dissociable Subunits of Yeast RNA Polymerase I1 Stimulate the Initiation of Transcription at a Promoter in Vitro*

polymerase 11 the fourth and seventh largest subunits (pol I1 A4/7) was purified from Sac- charomyces cerevisiae strain rpb-4, in which the gene for the fourth largest subunit is deleted. pol I1 A417 was indistinguishable from wild-type pol I1 (holoen-zyme) in promoter-independent initiationlchain elongation activity (400-800 nmol of nucleotide incorpo-rated110 minlmg of protein at 22 “C), in rate of chain elongation (20-25 nucleotidesls), and in the recognition of pause sites in the DNA template. In contrast to pol 11 holoenzyme, pol I1 A417 inactive in pro- moter-directed initiation of transcription in vitro. The of an equimolar complex of the fourth and seventh largest subunits, purified from pol I1 holoenzyme by ion-exchange chromatography in the presence of urea, restored promoter-directed initiation activity to pol I1 A4/7. The transcriptional activator protein Gal4-VP16 promoter-directed initiation by pol I1 A417 from a promoter with a Gal4 binding was observed between extracts of rpb-4, the

The two largest subunits of RNA polymerase, which account for about two-thirds of the mass of the molecule, have been conserved in amino acid sequence from prokaryotes to eukaryotes (1). These subunits of the Escherichia coli enzyme contain binding sites for the DNA template, nucleotides, and RNA product, so the conservation of amino acid sequences probably reflects conservation of the basic mechanism of RNA synthesis (2). By contrast, both the number and sequences of the smaller subunits vary widely, presumably because these subunits are involved in promoter specificity and regulation. For example, the u subunit of E. coli RNA polymerase is responsible for promoter selection and is released from the core enzyme following the initiation phase of the transcription reaction (2). In the eukaryotic enzymes, the role of the u subunit appears to be divided among multiple subunits and J Fellow of the National Cancer Institute of Canada.
** To whom correspondence should be addressed. accessory factors. In the case of RNA polymerase I1 (pol II),' responsible for the synthesis of mRNA in eukaryotes, a factor termed TFIID, BTF1, or T recognizes the "TATA" sequence associated with most (pol 11) promoters (3). At least three more accessory factors appear to be required in addition to TFIID for initiation at pol I1 promoters (3,4). Here we report that a complex of two small subunits of pol I1 also plays an important role in the initiation of transcription.
Pol I1 is made up of 10 subunits, ranging in size from about 6 to 190 kDa (5). In addition to the two largest subunits (denoted 1 and 2), conserved across species, subunits 5,6, and 8 are common to the three types of eukaryotic nuclear RNA polymerase (I, 11, and 111) (6) and so are presumed to play fundamental roles in RNA synthesis. Both biochemical and genetic evidence have been obtained bearing on the functions of subunits 4 and 7. A form of pol I1 that lacks these two subunits (here denotedpol I1 A4/7) was first isolated by anionexchange chromatography of the 10-subunit enzyme (holoenzyme) in the presence of urea (7,8). Only slight differences were detected between pol I1 A417 and holoenzyme in the capacity to synthesize RNA on a denatured DNA template, so there is evidently no requirement for subunits 4 and 7 in promoter-independent (nonspecific) initiation and RNA chain elongation (8).
Two yeast mutant strains are thought to contain predominantly the pol I1 A417 enzyme. One strain, rpo B l , contains a mutation in the largest subunit (8), whereas the other strain, rpb-4, lacks the gene for subunit 4 (9). pol I1 purified from these strains lacks both subunits 4 and 7, suggesting that subunit 7 must interact with subunit 4 to associate with the polymerase. Although rpb-4 is viable, it grows slowly, is sensitive to heat and cold, and is auxotrophic for inositol. Because pol I1 A417 is apparently fully functional in RNA chain elongation, the phenotype of rpb-4 presumably reflects the involvement of subunits 4 and 7 in initiation or in some other aspect of polymerase assembly, maintenance, or activity. A system was recently described for the promoter-directed (specific) initiation of transcription by yeast pol I1 in vitro (10,ll). With the use of this system, we have now characterized pol I1 A417 isolated from strain rpb-4. Differences in specific initiation activity between pol I1 A417 and holoenzyme were found that might account for the phenotype of the mutant strain.
Protein Purification-Wild-type yeast pol I1 (holoenzyme) was purified from strain BJ926 as described (11). pol I1 A4/7 was purified by the same procedure from strain rpb-4, grown at 24 "C, and harvested at an Am value of 6-8. For the isolation of subunits 4 and 7, holoenzyme (200 pg) was incubated for 3 h in 4.5 ml of 50 mM Tris-HCI (pH 8), 1 mM EDTA, 50 mM potassium acetate, 0.1 mM dithiothreitol, 5% glycerol, and 2 M urea (buffer A). The solution was then applied to a Bio-Gel SEC DEAE-5 PW column (7.5 X 75 mm, Bio-Rad) equilibrated in the same buffer. The column was developed with a 20-ml linear gradient of 0.05-1 M potassium acetate in buffer A. Fractions containing subunits 4 and 7 were dialyzed against 50 mM Tris-HC1 (pH 7.5), 1 mM EDTA, 30 mM ammonium sulfate, 50% glycerol, and were frozen in liquid nitrogen and stored at -70 "C. Ga14-VP16 fusion protein was enriched to 50% purity as described (13). Transcription Assays-Nonspecific initiation/chain elongation with poly(rC) as template was measured as described (14). Promoterspecific initiation assays were as described (15) in a reaction volume of 30 pl containing 4-6 pl of nuclear extract (12-20 mg of protein/ml of nuclear extract) and 20 pCi of [~u-~*P]UTP (about 1 p M UTP). The template (125 ng except where noted) was pGal4CG-(15) with a Gal4 protein binding site upstream of the yeast CYCl promoter fused to a 377-base pair sequence lacking guanosine residues in the coding strand. Promoter-specific transcripts were quantified with the use of an AMBIS radioanalytic imaging system (Ambis Systems, San Diego, CA).
Templates with single-stranded extensions on the 3' ends were prepared as described (16). In particular, template pCpTK243B, containing human histone H3.3 intron terminators (denoted TIa, TIb, and TII) (17), was produced by cleavage of plasmid pTK243B' at a unique SmaI site, followed by the addition of oligo(dC) extensions with terminal transferase and treatment with SstI and MluI. On this template, RNA polymerase I1 synthesizes transcripts of 326 nucleotides (run-off), 197 nucleotides (TIa), 182 nucleotides (TIb), and 138 nucleotides (TII). For preparation of template pCpTK202, used for measurement of elongation rates, plasmid pTK202' was linearized with SmaI, oligo(dC) extensions were added, and the DNA was treated with SacI. This template contained no strong termination or pause sites for calf thymus RNA polymerase 113 and supported production of a full-length transcript of 4770 nucleotides.
Transcription on 3'-extended templates was initiated in 50 mM Tris acetate (pH KO), 100 mM ammonium acetate, 5 mM magnesium acetate, 5% glycerol, 6 mM spermidine, 1 mM dithiothreitol, 0.8 mM ATP, GTP, and UTP, and 0.3 p M [a-"P]CTP (>400 Ci/mmol) with the indicated amounts of enzyme and template at 30 "C. After incubation for 1 min, 10 volumes of a solution were added to give the same reaction conditions except with 0.1 mM CTP. Heparin was also added to 100 pg/ml to prevent reinitiation. This procedure resulted in the labeling of the 5' ends of transcripts a t a high specific activity and such that the radioactivity in transcripts of a particular length was proportional to the number of transcripts of that length. Quantification was performed with an AMBIS radioanalytic imaging system.

RESULTS
Purification of Yeast RNA Polymerase 11 Holoenzyme, RNA Polymerase 11 A4/7, and Subunit 417 Complex-Pol I1 holoenzyme was purified from the diploid strain BJ926 by heparin-Sepharose and immunoaffinity chromatography as previously described (11), with the use of a monoclonal antibody directed against the carboxyl-terminal repeat domain of the largest subunit (18). pol I1 A417 was purified from the haploid strain rpb-4 by the same procedure. The chromatographic characteristics of the two enzymes were indistinguishable, but the yield of pol I1 A417 was about 25% of that for the holoenzyme, primarily because of a lesser amount of enzyme activity in the starting cell extract. This may be attributable to the lower yields of nuclear material commonly obtained from haploid rather than from diploid strains. The purified pol I1 A417 was 'Kerppola, T. K., and Kane, C. M., EMBO. J., submitted for publication.

Yeast BNA Polymerase 11
analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis ( Fig. 1) and, as has been noted previously (19), the enzyme lacked not only subunit 4 but also subunit 7.
Treatment of purified pol I1 holoenzyme with 2 M urea results in the dissociation of 90% or more of subunits 4 and 7 (Refs. 7 and 8, and data not shown). To resolve the dissociated subunits, pol I1 holoenzyme was fractionated on a DEAE-5PW HPLC column in the presence of urea (Fig. 1). Subunits 4 and 7 were found in a single peak in a molar ratio of 1:l (subunit 4subunit 7) as judged from staining with Coomassie Brilliant Blue (the ratio of the two subunits determined from dye binding was the same as that determined from radioactivity in polymerase purified from 35S-labeled cells). Subunits 4 and 7 also co-migrate in a nondenaturing polyacrylamide gel in the presence of 1 or 2 M urea (not shown), despite the differences in their sizes and ionic characteristics (9). These data strongly suggest that the two subunits in the DEAE HPLC peak were associated in a complex.
Purified pol I1 holoenzyme and pol I1 A417 were compared in assays of promoter-independent initiation and chain elongation with poly(rC) as template. It was previously reported that removal of subunits 4 and 7 had little effect on pol I1 activity in this assay (8). Our results lead to the same conclusion ( Fig. 2 FIG. 2. Activity of pol I1 holoenzyme and pol I1 A4/7 in promoter-independent initiation/chain elongation. pol I1 holoenzyme (0) and pol I1 A4/7 (0) were assayed with a poly(rC) template as described. One unit of activity is defined as 1 nmol of nucleotide incorporated into RNA/10 min at 22 "C. rates and termination properties.
The elongation rates of the two enzymes were indistinguishable, 20-25 nucleotides/s (data not shown). Both enzymes stopped at the same triplet of intrinsic termination sites in the human histone H3.3 gene (Fig. 3), and the efficiencies of stopping were nearly identical (53% for pol I1 holoenzyme, 57% for pol I1 A4/7 at TIa). Lack of Specific Initiution by pol 1 1 A4/7 in Vitro-Nuclear extracts from strain rpb-4 were assayed for specific initiation a t a minimal yeast promoter (TATA element and transcription site) as previously described (15). Five extracts, separately prepared, all failed to give any promoter-directed transcription (e.g. Fig. 4, lane I ) . The addition of purified pol I1 holoenzyme (Fig. 5 ) or subunit 4/7 complex restored transcriptional activity (Fig. 4, lane 3) in proportion to the amount of protein added (Fig. 5). The subunit 4/7 complex was 2-3fold more effective on a molar basis than purified pol I1 holoenzyme. Purified pol I1 A4/7 restored no activity at all. We conclude that the subunit 4/7 complex is required for initiation at a minimal promoter in uitro.
The capacity of an rpb-4 extract to support promoterdirected transcription was also restored by Ga14-VP16, a potent transcriptional activator protein (13) containing Gal4 binding sites (Fig. 4, lane 2). The Ga14-VP16 was present at amounts that saturated the Gal4 binding sites on the template and maximally stimulated transcription. Transcription was stimulated a further 7-fold when the subunit 4/7 complex was added as well (compare Fig. 4, lanes 2 and 4). Thus, although the subunit 4/7 complex is required for initiation at a minimal promoter, it can be partially supplanted by an activator at a promoter with an activator binding site.
Possible Exchange of Subunits 4 and 7 between Polymerase Molecules-Nuclear extracts from strain Y260-I were previously used to assay promoter-directed initiation of transcription by purified pol I1 holoenzyme (11). This strain is temperature-sensitive because of a mutation in the largest subunit of the enzyme (12), and warming an extract briefly prevents initiation. Activity is restored by holoenzyme in proportion to the amount added (Fig. 6, panel A ) . We were surprised to find that activity can also be restored to such an extract by pol I1 A4/7 (Fig. 6, panel B). Holoenzyme and pol I1 A4/7 were almost equally effective in this assay, raising the possibility that subunits 4 and 7 might shuttle from the inactive holoenzyme in the Y260-I extract to the added pol I1 A4/7. The converse possibility that the largest subunit can exchange from the pol I1 A4/7 to the heat-inactivated polymerase is unlikely; the dissociation of this subunit from the polymerase under transcription conditions has never been observed nor has the dissociation of the homologous subunit from the E. coli enzyme. was added to heat-treated extracts in the amounts indicated, and promoter-directed transcription was measured as described. The level of transcription is expressed relative to that obtained with unheated Y260-I extract (loo%, 40-80 amol of transcript). The data shown are from multiple assays and, thus, there is some scatter about the leastsquares fit lines; the amount of transcription in individual assays is linear with respect to enzyme concentration (11). Such exchange of subunits should permit in vitro complementation between rpb-4 and heated Y260-1 extracts. Indeed, although neither extract supported much initiation on its own, a mixture of the two extracts exhibited wild-type levels of activity (Fig. 7). Although these observations are consistent with an exchange of subunits between enzymes, they could also be explained by the occurrence of an excess of subunits 4 and 7 in Y260-I or even by complemention without dissociation of the subunits from the enzyme.

DISCUSSION
An analogy may be drawn between pol I1 A4/7 and the core RNA polymerase from bacteria, which lacks the u subunit. Both enzymes possess RNA chain elongation and termination activities, but neither can recognize a promoter or initiate transcription specifically. The subunit 4/7 complex has some characteristics normally associated with a u factor. First, as already mentioned, the subunit 417 complex is required for initiation at a minimal promoter in vitro, but not for chain elongation or termination. Second, there are similarities in the amino acid sequences of subunit 4 and u factors (9). There are, of course, limits to the analogy between the subunit 4/7 complex and u. The subunit 417 complex is not sufficient for promoter recognition and initiation. Rather, a set of factors is required for these functions in eukaryotes, and the subunit 4/7 complex is a member of this set. Like other initiation factors, the subunit 4/7 complex may associate reversibly with the enzyme in the course of the initiation-elongation cycle. Our observation of in vitro complementation between defective holoenzyme and pol I1 A417 would be consistent with such reversible association.
rpb-4 represents the second of two yeast pol I1 mutants we have examined in which the measurement of promoter-directed initiation gives very different results from the traditional assay of nonspecific initiation and chain elongation. Extracts of strain Y260-I contain virtually no detectable activity in the nonspecific assay (12); but are as effective as wild-type extracts in promoter-directed initiation (11). Conversely, extracts of rpb-4 possess full activity in the nonspecific assay but fail to support promoter-directed initiation. The use of both types of assay is, therefore, warranted in future characterizations of pol I1 mutants and preparations.
Preparations of yeast pol I1 holoenzyme contain a mixture of holoenzyme and pol I1 A4/7? The less than stoichiometric amounts of subunits 4 and 7 relative to the other subunits may not be artifactual because of dissociation and loss during purification, but rather may be functionally significant if, as mentioned above, the subunit 417 complex serves as an initiation factor, interacting reversibly with the enzyme. It will be of interest to learn whether mammalian pol I1 exhibits similar subunit dissociation and whether there are functional consequences.
Although the subunit 4/7 complex is required for specific initiation at a minimal promoter in vitro, strain rpb-4, which lacks subunit 4, is viable. This apparent contradiction may be explained by our results with the transcriptional activator Ga14-VP16. An rpb-4 extract supports initiation at a promoter with a Gal4 binding site in the presence of Ga14-VP16. We suggest that transcription occurs at a low level in the absence of subunit 4 and is raised above the threshold for detection by Ga14-VP16. The low level of transcription in the absence of subunit 4 might be sufficient for viability. Alternatively, it may be argued that all yeast promoters contain activator binding sites, and the stimulatory effect of activators in the absence of subunit 4 results in sufficient transcription for cell growth. In either case, the level of transcription is lower than in the presence of both an activator and subunit 4, accounting for the slow growth of strain rpb-4.
Our results also bear on the question of what component of the pol I1 transcription apparatus is the target of activator proteins. The finding that Ga14-VP16 stimulates transcription in the absence of subunits 4 and 7 indicates that these two subunits cannot be the sole targets of the VP16 activation domain.
associates with yeast RNA polymerase I1 (pol 11) and migrates ahead of subunit 9 on denaturing gel electrophoresis (see Fig. 1 and Ref. 11) has been cloned and sequenced? This gene, rpb-11, is not related to any of the known yeast pol I1 genes and thus may constitute an additional subunit of yeast pol 11. This raises the total number of subunits in yeast pol I1 to 11.