RNA Polymerase II Subunit Rpb9 Regulates Transcription Elongation in Vivo *

RNA polymerase II lacking the Rpb9 subunit uses alternate transcription initiation sites in vitro andin vivo and is unable to respond to the transcription elongation factor TFIIS in vitro. Here, we show thatRPB9 has a synthetic phenotype with the TFIIS gene. Disruption of RPB9 in yeast also resulted in sensitivity to 6-azauracil, which is a phenotype linked to defects in transcription elongation. Expression of the TFIIS gene on a high-copy plasmid partially suppressed the 6-azauracil sensitivity of Δrpb9cells. We set out to determine the relevant cellular role of yeast Rpb9 by assessing the ability of 20 different site-directed and deletion mutants of RPB9 to complement the initiation and elongation defects of Δrpb9 cells in vivo. Rpb9 is composed of two zinc ribbons. The N-terminal zinc ribbon restored the wild-type pattern of initiation start sites, but was unable to complement the growth defects associated with defects in elongation. Most of the site-directed mutants complemented the elongation-specific growth phenotypes and reconstituted the normal pattern of transcription initiation sites. The anti-correlation between the growth defects of cells disrupted for RPB9 and the selection of transcription start sites suggests that this is not the primary cellular role for Rpb9. Genome-wide transcription profiling of Δrpb9 cells revealed only a few changes, predominantly in genes related to metabolism.

Rpb9 has roles both in transcription initiation and in transcription elongation. In the initiation reaction, Rpb9 modulates the selection of the transcription start site. In cells lacking Rpb9 and in reconstituted transcription reactions lacking Rpb9, the population of start sites is shifted upstream at a variety of promoters (9 -11). In the elongation reaction, Rpb9 is required, along with TFIIS, to effect transcription through blocks to elongation encoded by the DNA template (12). A role in the modulation of initiation and elongation is consistent with the localization of Rpb9 in the three-dimensional structure of yeast RNA polymerase II. Rpb9 is located at the tip of the so-called "jaws" of the enzyme, which is thought to function by clamping the DNA downstream of the active site (3,13,14). The Rpb9 homologue in RNA polymerase III, C11, also has been implicated in regulating RNA chain elongation (15).
Rpb9 comprises two zinc ribbon domains joined by a 30amino acid linker. The C-terminal zinc ribbon is a sequence homologue of the zinc ribbon in the transcription elongation factor TFIIS (16,17). The roles of each domain of Rpb9 in transcription elongation were determined by assaying a series of alanine-scanning mutants of Rpb9 in in vitro reactions (18). Alanine substitutions in the C-terminal zinc ribbon domain of Rpb9, like amino acid substitutions in the homologous part of TFIIS, completely eliminated elongation activity. Mutating the first zinc ribbon had no effect on elongation activity, although deleting this domain entirely abrogated activity. The linker region mediated the interaction of Rpb9 with the rest of the RNA polymerase. In this study, we used this series of mutations to probe the cellular role of Rpb9 in both initiation and elongation.

Yeast Expression Plasmid
The yeast expression plasmid pRS314RPB9 containing the RPB9 open reading frame plus ϳ500 base pairs upstream and 2200 base pairs downstream was obtained from Dr. Rolf Furter (9). This plasmid was adapted by inserting a BamHI restriction site immediately upstream of the start codon and an EcoRI restriction site immediately downstream of the stop codon, creating the plasmid pRS314RPB9BE. These sites were inserted using the QuikChange protocol and PfuI DNA polymerase (Stratagene). Incorporation of these restriction sites allowed for the insertion of each of the previously constructed rpb9 mutants into pRS314RPB9BE. The resulting plasmids, containing each of the rpb9 mutant alleles under control of the endogenous RPB9 promoter, were transformed into yeast to determine their effects on growth and the use of initiation start sites.

Growth Assays
The ⌬rpb9 yeast strain grows slowly at 30°C, is extremely sensitive to high-and low-temperature extremes, and is sensitive to the drug 6-azauracil. Expression of wild-type RPB9 corrects these defects. Haploid ⌬rpb9 cells were transformed with the RPB9 yeast expression plasmids to test each mutant for the ability to complement the ⌬rpb9 growth phenotypes. To test for complementation of cold and temperature sensitivity, the cells were grown on solid synthetic complete yeast medium lacking tryptophan. Suspensions containing ϳ10,000, 2000, 400, and 80 cells were spotted onto solid medium and grown at 12, 30, or 37°C for 2-6 days. Cells were grown on solid synthetic complete yeast medium lacking tryptophan and uracil and containing 50 g/ml 6-azauracil to measure ability to correct sensitivity to 6-azauracil. Suspensions containing ϳ10,000, 2000, 400, and 80 cells were spotted onto solid medium and grown at 30°C for 3-8 days. Each mutant construct was compared with wild-type RPB9 with respect to ability to restore growth characteristics.

Primer Extension
Primer extension assays were performed to identify the transcription start sites in the mutant yeast strains. Yeast strains YF2221, YF2222, YF2230, and YF2234 were grown in yeast extract peptone liquid medium with 2% glucose. YF2230 cells transformed with each of the pRS314RPB9BE constructs was grown in liquid complete synthetic medium lacking tryptophan. All cultures were grown at 30°C to A 600 nm ϭ 0.2 to 1.0. Cells (5 ϫ 10 7 ) were harvested, and total RNA was isolated using the RNeasy protocol (QIAGEN Inc.). The primer used for these experiments, 5Ј-AGAAGATAACACCTTTTTGAG-3Ј (Dalton Chemicals), is complementary to nucleotides ϩ37 to ϩ17 in the ADH1 gene. The primer was radiolabeled at the 5Ј-end by phosphorylating with polynucleotide kinase (New England Biolabs Inc.) and [␥-32 P]ATP. For each primer extension reaction, 15 g of total RNA from the appropriate yeast strain was annealed with 0.4 pmol of the 5Ј-radiolabeled primer for 45 min at 52°C. Reverse transcription from the annealed primer was done with Moloney murine leukemia virus reverse transcriptase (Life Technologies, Inc.) according to manufacturer's instructions. The reverse transcripts were collected by ethanol precipitation and resolved on a Tris borate/EDTA, 8.3 M urea, and 6% polyacrylamide gel and visualized by phosphorimaging.
RNA Isolation-Yeast cells were grown in yeast extract peptone 2% glucose medium at 30°C with constant agitation and aeration to A 600 nm ϭ 0.4 -0.6. Cells were washed once with diethyl pyrocarbonate-treated water and resuspended in buffer containing 10 mM Tris-HCl, pH 7.5, 10 mM EDTA, and 0.5% SDS. Total RNA was isolated using a hot phenol method (19) and further purified using a QIAGEN RNeasy midi kit essentially as described by the supplier. RNA concentrations were determined by measuring the absorbance at 260 nm. Two independent RNA preparations were made for each mutant strain. . 20 units of RNasin (Promega) was also added to the reaction. The mixture (minus the enzyme) was heated at 65°C for 5 min and then at 42°C for 5 min; the reverse transcription enzyme was added; and the reaction was incubated at 42°C for 2 h. The reverse transcription was stopped with EDTA (final concentration of 6.25 mM), and the RNA template was degraded by the addition of 10 N NaOH (final concentration of 0.5 N) and incubation at 65°C for 20 min. The mixture was neutralized by the addition of 5 M acetic acid (final concentration of 0.5 M), and the labeled cDNA was precipitated by the addition of 1 volume of isopropyl alcohol and incubation on ice for 30 min. After rinsing with 70% EtOH, the labeled cDNA was resuspended in 5 g of diethyl pyrocarbonate-treated water.
Hybridization-For each DNA microarray, 5 l of purified Cy3-labeled cDNA and 5 l of purified Cy5-labeled cDNA were added to 75 l of DIG Easy hybridization buffer (Roche Molecular Biochemicals). 2 l of yeast tRNA (10 mg/ml; Sigma) and 2 l of single-stranded salmon sperm DNA (10 mg/ml; Sigma) were also added to the hybridization buffer, and the solution was heated at 65°C for 2 min. The solution was then applied under a coverslip to a custom-made yeast whole genome microarray (Microarray Center, Ontario Cancer Institute). The microarrays were incubated at 37°C in a humid hybridization chamber for 8 -12 h. Before scanning, the slides were washed with 0.1ϫ SSC and 0.1% SDS (3 ϫ 15 min at 50°C), rinsed with 0.1ϫ SSC (3 ϫ 5 min at room temperature), and dried by centrifugation. A total of eight slides were hybridized for each mutant strain. The arrays were read on a laser confocal scanner (ScanArray 4000, GSI Lumonics), and the images obtained were quantified using QuantArray 2.0 software (GSI Lumonics).

The Phenotype of ⌬rpb9 Cells Is Consistent with a Role in
Elongation-Our previous studies implicated Rpb9 in transcription elongation in vitro (12,18). In cells lacking Rpb9, a proportion of the Rpb9-deficient RNA polymerase II molecules (pol II⌬9) 1 initiated transcription at many promoters at upstream DNA sequences. This defect could be rescued by the addition of wild-type RPB9, but not a mutant altered in the N-terminal zinc ribbon domain. Subsequently, biochemical studies revealed a role for Rpb9 in transcription elongation in vitro (12,18). In our original study (12), the mutant enzyme (pol II⌬9) was shown to have the same maximal elongation rate as did the wild-type RNA polymerase II, but stopped less frequently at DNA sequences (e.g. a sequence from the histone H3.3 intron) that promote pausing of the transcription complex. The addition of Rpb9 to pol II⌬9 restored its in vitro FIG. 1. Phenotypes of ⌬rpb9 cells. A, sensitivity of ⌬rpb9 cells to 6-azauracil. Approximately 10,000 cells of the YF2221 (wild-type) or YF2230 (⌬rpb9) strain containing the indicated plasmids were spotted onto agar containing synthetic complete medium lacking tryptophan and uracil and including, where indicated, 100 g/ml 6-azauracil (6AU) and incubated at 30°C. B, synthetic phenotype of ⌬rpb9 and ⌬tfiis cells. Approximately 10,000 cells of the YF2221 (wild-type) or YF2234 (⌬rpb9 ⌬tfiis) strain containing the indicated plasmids were spotted onto agar containing synthetic complete medium lacking tryptophan and uracil and including, where indicated, 100 g/ml 6-azauracil. The plates were incubated at 30°C for 2 days (without 6-azauracil), 30°C for 4 days (with 6-azauracil), 35°C for 2 days, or 37°C for 2 days. elongation properties. Occasionally, the pol II⌬9 enzyme did form arrested elongation complexes at the histone H3.3 arrest site. Unlike wild-type arrested complexes, these arrested pol II⌬9 complexes were unable to be rescued by the addition of the elongation factor TFIIS. In general, these studies revealed a role for Rpb9 in transcription elongation. The parts of Rpb9 that contributed to the elongation activity were determined using a set of Rpb9 deletion and alanine-scanning mutants (18). These studies showed that the C-terminal zinc ribbon domain was important for elongation, as was the linker region connecting the two zinc ribbons composing Rpb9. The linker region was shown specifically to be important for the binding of Rpb9 to pol II⌬9. We were unable to show that the N-terminal zinc ribbon, which is important for start site selection (10), played a role in elongation.
TFIIS and Rpb9 are linked biochemically and are related in structure. We were interested whether there is also a genetic interaction between RPB9 and TFIIS gene (the TFIIS gene is also known as PPR2). Yeast cells lacking the gene for Rpb9 are sensitive to both low-and high-temperature extremes and grow more slowly than do wild-type strains even at the optimal growth temperature (10,21). These phenotypes were also observed in our ⌬rpb9 strain, and normal growth was restored by expressing RPB9 from a low-copy plasmid under the control of its own promoter. Yeast strains lacking the TFIIS gene are sensitive to the drug 6-azauracil. This phenotype is thought to reflect a defect in transcription elongation (20). Since Rpb9 is required for the functional interaction between RNA polymerase II and TFIIS, we tested the ⌬rpb9 strain for sensitivity to 6-azauracil. The ⌬rpb9 strain grew more slowly on medium containing 6-azauracil than did the parent strain (Fig. 1). Transforming the ⌬rpb9 strain with the RPB9 gene on a lowcopy plasmid fully complemented the 6-azauracil sensitivity (Fig. 1).
The double deletion strain (⌬rpb9/⌬tfiis) was constructed, and its phenotype was tested to explore the genetic interaction between the TFIIS gene and RPB9. In agreement with the biochemical studies, the double mutant possessed a much more severe phenotype than did either of the individual gene disruptions (Fig. 1). These observations are consistent with a functional interaction between TFIIS and Rpb9 in vivo.
The ⌬tfiis and ⌬rpb9 strains were each transformed with FIG. 2. Suppression of ⌬rpb9 sensitivity to 6-azauracil by the expression of the TFIIS gene on a high-copy plasmid. Approximately 10,000, 2000, 400, or 80 cells of the YF2221 (wild-type), YF2230 (⌬rpb9), or YF2222 (⌬tfiis) strain containing the indicated plasmids were spotted onto agar containing synthetic complete medium lacking tryptophan and uracil and including the indicated amounts of 6-azauracil. The plates were incubated at 30°C for 2 days (without 6-azauracil) or at 30°C for 6 days (with 6-azauracil).  Rpb9 ϩϩ ϩϩ Ϫ a ϩϩ, allele completely restores normal growth at 30°C on synthetic complete solid medium lacking tryptophan; ϩ, allele partially restores normal growth at 30°C on synthetic complete solid medium lacking tryptophan; Ϫ, allele does not restore normal growth at 30°C on synthetic complete solid medium lacking tryptophan.
high-copy plasmids bearing the wild-type TFIIS gene and RPB9, respectively (Fig. 2). These and various control strains were tested for growth on plates containing 0, 25, 50, and 100 g/ml 6-azauracil. The TFIIS gene on a high-copy plasmid partially suppressed the 6-azauracil sensitivity of ⌬rpb9 cells. In contrast, RPB9 on a high-copy plasmid did not suppress the 6-azauracil sensitivity of ⌬tfiis cells. These data suggest that the elongation defect caused by the absence of Rpb9 can be restored partially by increasing the cellular concentration of TFIIS. We conclude that the effects on cell growth caused by disrupting RPB9 arise in part from defects in transcription elongation.
Complementing the Growth of ⌬rpb9 Cells with rpb9 Mutants-Rpb9 was shown originally to have a role in regulating the choice of the transcription start sites (9 -11); subsequently, Rpb9 was implicated in transcription elongation (12,18). Here, we analyzed the properties of the set of rpb9 mutants in vivo to gain insight into the physiological role of Rpb9. The 20 mutants were assayed for their ability to restore normal growth to haploid ⌬rpb9 and ⌬rpb9/⌬tfiis yeast cells. To accomplish this, each strain was transformed with the low-copy plasmid pRS314 carrying each of the rpb9 mutant alleles, and the phenotypes were monitored.
All of the 20 alanine-scanning Rpb9 mutants restored normal growth to ⌬rpb9 cells. These mutants could, however, be divided into two classes: those that had no effect on any of the other Rpb9 properties (start site selection and elongation; data not shown) and those that affected one or the other (Table I, first and second columns). Other mutants had no effect on growth rate, initiation in vivo, or elongation in vitro and are not included in Table I Several rpb9 alleles with internal or C-or N-terminal deletions in either of the two zinc domains had cell growth phenotypes (Table I, first and second columns). A C-terminal truncation mutant, Rpb9-(1-47), which contains the N-terminal zinc-binding domain and part of the linker region but lacks the second zinc domain, was unable to restore normal cell growth to ⌬rpb9 cells. Two deletion mutants in the first zinc (Zn1) region (Rpb9-(⌬12-27) and Rpb9-(⌬16 -23)) also were unable to complement ⌬rpb9 cell growth. We conclude that Rpb9 requires both zinc-binding regions for normal growth.
For all rpb9 mutants tested, the three phenotypes, temperature, cold, and 6-azauracil sensitivity, were strongly correlated. This correlation suggests that the lack of Rpb9 is the primary defect that underlies all three phenotypes and that they are not secondary effects of the gene disruption.
Effect of rpb9 Mutants on Transcription Start Site Selection in Vivo-rpb9 cells exhibit altered preference for transcription start sites on a variety of promoters (9 -11). In most cases, an upstream shift of the 5Ј-end of the transcript is observed. In this study, the ADH1 gene, which shows a distinctive difference in the transcription start site between the wild-type parent and the ⌬rpb9 strains (10), was used to analyze the effect of the various rpb9 alleles on the use of initiation sites. A tfiis knockout strain and a rpb9/tfiis double knockout strain were also analyzed to determine the effect of TFIIS on initiation start site selection. Primer extension analysis was performed on RNA isolated from the different yeast strains using a primer directed against the 5Ј-end of the ADH1 gene.
The pattern of initiation sites in the wild-type strain was compared with those in the ⌬rpb9, ⌬tfiis, and ⌬rpb9/⌬tfiis strains. The RNA for this primer extension analysis was prepared from these strains after they were grown in a rich me-dium. The patterns of initiation sites in the wild-type and ⌬rpb9 strains are easily distinguished; the reverse transcripts prepared from the ⌬rpb9 strain are longer and reflect an increase in the number of ADH1 transcripts that start upstream of position Ϫ37 (Fig. 3). The deletion of the TFIIS gene appeared to have no influence on start site selection, even in conjunction with the deletion of RPB9. The ⌬tfiis strain had a transcription initiation profile identical to that of the wild-type strain, and the profile for the rpb9/tfiis double knockout strain was identical to that of the ⌬rpb9 strain. The synthetic phenotype that occurs in the rpb9/tfiis double knockout strain does not appear to stem from a more severe defect in transcription initiation.
The majority of rpb9 alleles restored the normal pattern of initiation sites in rpb9-deleted yeast cells (see above and Table  I). One exception was an rpb9 mutant containing only the Zn1 domain. This mutant, which expressed at wild-type levels, 2 was able to complement the start site defect (Fig. 4), but not the other activities. The properties of this mutant serve to uncouple the role of Rpb9 in transcription initiation from the roles in transcription elongation and cell growth. The phenotypes of the rpb9 mutants are inconsistent with an important role for Rpb9 in transcription initiation.
Correlation between in Vivo and in Vitro Properties-Our previous studies linking Rpb9 structure to function identified several parts of the protein important for transcription elongation activity (18). Specifically, the C-terminal zinc ribbon was 2 S. M. Orlicky, personal communication.

FIG. 4. Primer extension analysis of the ADH1 transcript from a yeast strain carrying the truncated Zn1 allele of rpb9.
Primer extension assays of the ADH1 transcript were performed using total RNA prepared from the ⌬rpb9 yeast strain carrying either pRS314 or a derivative plasmid containing one of the various rpb9 mutant alleles. Shown here is an autoradiogram displaying the reverse transcripts from the ⌬rpb9 strains transformed with control plasmid (⌬9), the plasmid encoding wild-type RPB9 (Rpb9), or the truncated Zn1 rpb9 allele (Rpb9 1-47 ). Primer extension results for each of the ⌬rpb9 mutants are presented in Table I. required for RNA cleavage in arrested transcription complexes and for reactivating arrested complexes in conjunction with TFIIS, and the linker region mediated the interaction of Rpb9 with RNA polymerase II. When tested for activity in yeast cells, every rpb9 mutant that displayed a growth phenotype was inactive for transcription elongation. However, all of the alanine-scanning mutants that displayed reduced elongation activity had perfectly normal growth rates and were able to restore the normal pattern of initiation start sites (Table I). We suggest that the in vivo complementation assays are less sensitive indicators of Rpb9 function than the in vitro assays.
Comparison of Genome-wide Expression in TFIIS-and Rpb9disrupted Cells-The similar phenotypes of the tfiis-and rpb9deleted strains and their genetic interaction suggest that the proteins have a similar effect on elongation. To determine whether a common set of genes is regulated by the two transcription factors, we compared the patterns of gene expression in the ⌬rpb9, ⌬tfiis, and ⌬rpb9/⌬tfiis strains using yeast DNA microarrays.
Disruption of either the TFIIS gene or RPB9 in log-phase cells grown in rich medium had little effect on global gene expression (data not shown). 3 In ⌬rpb9 cells, where the effect was more dramatic, transcription of only 1-2% of yeast genes was altered more than 2-fold compared with the isogenic parent. Many of these genes belonged to metabolic clusters (Table  II)  suggests some form of metabolic response, the profile of gene expression did not resemble the global responses to glucose starvation (22). Further analysis will be required to determine whether the response is a direct effect on gene regulation or a more general indirect response. In cells lacking both TFIIS and Rpb9, the transcriptional changes were more pronounced, but the response resembled an amplified version of the RPB9 disruption. In general, the transcription profile correlated with the growth of the cells. There were more differences in the rpb9/tfiis double disruption than in the singly deleted rpb9 and tfiis cells. We were unable to relate any differences to specific elongation defects.

DISCUSSION
Rpb9 plays a role in selecting the sites of transcription initiation. Cells lacking Rpb9 initiate transcription at upstream sites on many promoters. We discovered that a derivative of Rpb9 containing only the Zn1 and linker domains of Rpb9 was sufficient to correct this defect. In addition, inactivating the Zn1 domain by deleting the majority of the region between the two pairs of cysteines or mutating the first cysteine (10) destroyed its ability to select the wild-type initiation sites. Together, these two results suggest that the Zn1 domain regulates the selection of transcription initiation sites.
In the crystal structure of yeast RNA polymerase II (3), Rpb9 is positioned near the largest subunit and is predicted to contact the DNA downstream of the active site. The Zn1 and Zn2 domains are positioned on opposite sides of a protein wall that separates the DNA cleft from the back of the enzyme with respect to the DNA. The Zn1 domain would be predicted to be closer to the DNA template than would the Zn2 domain. The activity of the Zn1 domain in selecting transcription start sites is therefore consistent with its position in the RNA polymerase.
In addition to altered patterns of transcription initiation sites, rpb9 strains also exhibit slow growth at optimal temperature, an increased sensitivity to high and low temperatures (20), and a lower tolerance for the drug 6-azauracil. One of the most significant observations of this work is that the temperature and drug phenotypes can be distinguished from the defect in start site selection. Two mutations in the Zn1 domain were unable to correct the defect in start site selection, yet corrected the growth defects. Therefore, in ⌬rpb9 cells, we conclude that the defect in start site selection does not appear to be the underlying basis for the defect in growth.
The Zn1 domain has a charged loop whose homologue in the Zn2 domain is essential for elongation activity. Point mutants or deletion mutants in the charged loop in the Zn1 domain do not appear to play an important role in selecting the transcription start sites. Each of the single amino acid substitution mutants within the loop as well as the ⌬16 -23 mutant, which has the entire predicted flexible loop removed, restored the wild-type pattern of initiation sites. The mechanism of action of the Zn1 domain likely differs than of the Zn2 domain.
Rpb9 must assemble with RNA polymerase to restore elongation activity to RNA polymerase II. In vitro, we showed that the mutant comprising the Zn1 domain and part of the linker interacted poorly with pol II⌬9 (18). However, this mutant was able to restore correct start site selection in vivo. The discrepancy between in vitro and in vivo observations is common; many of the mutants that bound poorly to pol II⌬9 in vitro restored both start site preferences and growth characteristics in vivo. Recently, we learned that the polymerase-binding surface of Rpb9 consists of more than just the conserved D-DPTLPR sequence; in the RNA polymerase crystal structure (3), there appear to be a range of contacts between Rpb9 and Rpb1. Many of these contacts involve the Zn1 domain, suggesting that the strength of these interactions is sufficient for the assembly of the Zn1 domain into RNA polymerase II in the cell.