Ctr9, Rtf1, and Leo1 Are Components of the Paf1/RNA Polymerase II Complex

Yeast strains, growth conditions, and genetic techniques.

The S. cerevisiae strains used in this study were derived from strain YJJ662 (MATa leu2Δ1 his3Δ200 ura3-52) (47) and are all isogenic except for the specific deletion or modification indicated: YJJ577 (paf1Δ::HIS3), YJJ1303 (rtf1Δ::kan^r), YJJ1326 (rtf1Δ::kanr paf1Δ::HIS3), YJJ1336 (leo1Δ::kan^r), YJJ1339 (trp1Δ::kan^r), YJJ1361 (leo1Δ::kan^{r paf1Δ}::HIS3), and YJJ1364 (ctr9Δ::kan^{r paf1 Δ}::HIS3). Details of TAP-tagged and hemagglutinin (HA)-tagged strain construction are described below. YJJ1339 was used to create HA-tagged Leo1, resulting in strain YJJ1330 [LEO1::LEO1^HA6(KlTRP1)]. The TAP-tagged strains were derived from YJJ662 and include YJJ1307 [CDC73::CDC73^{CBP-TEV-ProtA}(KlURA3)], YJJ1308 [SRB5::SRB5^{CBP-TEV-ProtA}(KlURA3)], YJJ1329 [CTR9::CTR9^{CBP-TEV-ProtA}(KlURA3)], and YJJ1334 [HPR1::HPR1^{CBP-TEV-ProtA}(KlURA3)]. Strains containing both Leo1-HA and TAP tags are YJJ1309 [CDC73::CDC73^{CBP-TEV-ProtA}(KlURA3) LEO1::LEO1^HA6(KlTRP1)], YJJ1331 [SRB5::SRB5^{CBP-TEV-ProtA}(KlURA3) LEO1::LEO1^HA6(KlTRP1)], and YJJ1332 [CTR9::CTR9^{CBP-TEV-ProtA}(KlURA3) LEO1::LEO1^HA6(KlTRP1)]. Strains were grown in YPD with 2% glucose using standard methods (15).

Construction of TAP-tagged strains.

Yeast chromosomal genes CDC73, SRB5, HPR1, and CTR9 were epitope tagged with a C-terminal TAP tag (43). The tagging cassette consists of two immunoglobulin G (IgG) binding domains of Staphylococcus aureus protein A and a calmodulin binding peptide separated by a tobacco etch virus (TEV) protease site. The Kluyveromyces lactis URA3 gene located at the 3′ end of the tag served as a selectable marker. PCR fragments were generated using a 5′ oligonucleotide corresponding to the last 51 bp of the gene of interest, excluding the stop codon, plus 17 bp matching the TAP tag, and a 3′ oligonucleotide corresponding to the 51 bp downstream of the stop codon plus 17 bases of the tagging cassette. Oligonucleotide sequences are available upon request. The PCR fragments were transformed into strain YJJ662, generating strains YJJ1307 (Cdc73-TAP), YJJ1308 (Srb5-TAP), YJJ1329 (Ctr9-TAP), and YJJ1334 (Hpr1-TAP). TAP-tagged constructs were confirmed by Western blotting and PCR.

Construction of Leo1-HA strains.

Leo1 was C-terminally epitope tagged using the pYM3 plasmid (19) containing a six-HA peptide and a K. lactis TRP1 marker. To permit selection on synthetic complete medium lacking Trp, TRP1 was deleted in YJJ662, using the KANMX4 cassette (54), resulting in YJJ1339. TAP-tagged constructs were generated as described above. The Leo1-HA PCR fragment was transformed into YJJ1339. The resulting strains containing Leo1-HA were YJJ1330 (Leo1-HA), YJJ1309 (Cdc73-TAP, Leo1-HA), YJJ1331 (Srb5-TAP, Leo1-HA), and YJJ1332 (Ctr9-TAP, Leo1-HA). HA-tagged constructs were confirmed by Western blotting and PCR.

Extract preparation and purification of TAP-tagged complexes.

The TAP-tagged strains were grown in YPD medium to a density of 2 × 10⁷ cells/ml, and transcriptionally active whole-cell extracts were prepared as described previously (56). Briefly, protein pellets were resuspended after NH₄SO₄ precipitation in buffer A (20 mM HEPES-KOH [pH 7.9], 10% glycerol, 10 mM EGTA, 10 mM MgSO₄, 1 mM dithiothreitol [DTT], 1 mM phenylmethylsulfonyl fluoride [PMSF], 2 mM benzamidine hydrochloride, leupeptin [0.5 μg/ml], bestatin [0.35 μg/ml], pepstatin [0.4 μg/ml]) to a final protein concentration of 20 to 30 mg/ml. The resuspended proteins were extensively dialyzed against buffer A containing 100 mM NH₄SO₄. Approximately 1 g of total protein was incubated with 500 μl of rabbit IgG agarose (Sigma) at 4°C for 2 h. Beads were washed three times with 50 ml of buffer A and then equilibrated in TEV protease cleavage buffer (20 mM HEPES-KOH [pH 8.0], 10% glycerol, 0.5 mM EDTA, 1 mM DTT, 1 mM PMSF, 2 mM benzamidine hydrochloride, leupeptin [0.5 μg/ml], bestatin [0.35 μg/ml], pepstatin [0.4 μg/ml]). Proteins were eluted from the beads by adding 200 U of TEV protease in 1 ml of TEV protease cleavage buffer and incubating the mixture at ambient temperature for 2 h. The eluted proteins were concentrated with a Centricon YM-30 filter (Millipore) and applied at 0.2 ml/min to a Superose 6 HR 10/30 (Pharmacia) gel filtration column equilibrated in buffer B [30 mM HEPES-KOH (pH 7.9), 8% glycerol, 2 mM EDTA, 2 mM EGTA, 0.1 M (NH₄)₂SO₄, 0.05% NP-40, 5 mM β-glycerophosphate, 1 mM DTT, 1 mM PMSF, 2 mM benzamidine hydrochloride, leupeptin (0.5 μg/ml), bestatin (0.35 μg/ml), pepstatin (0.4 μg/ml)]. Fractions (200 μl) were collected. Fractions were trichloroacetic acid (TCA) precipitated (38) and analyzed by Western blotting. Transcriptional activity was measured by a nonselective transcription assay as described elsewhere (58). To estimate the size of protein complexes, the following markers were used to calibrate the Superose 6 column: aggregated MCM protein (>2,000 kDa; gift from X. Chen, University of Colorado Health Sciences Center), thyroglobulin (669 kDa), apoferritin (443 kDa), and β-amylase (200 kDa).

Isolation of Leo1-HA complexes.

The C-terminally Leo1 HA-tagged strains were grown in 100 ml of YPD to a density of 1.4 × 10⁷ cells/ml. Cells were pelleted and washed in sterile water, freeze-thawed in liquid nitrogen, and then resuspended in 1.0 ml of whole-cell extract lysis buffer [0.2 M Tris-HCl (pH 7.9), 0.39 M (NH₄)₂SO₄, 10 mM MgSO₄, 20% (vol/vol) glycerol, 1 mM EDTA, 1 mM DTT, and 1 mM PMSF, plus protease inhibitors as above]. Cells were lysed with the addition of 1 ml of glass beads and vortexed four times for 1 min while cooling. Lysates were clarified at 13,000 rpm with a Baxter Biofuge 13 for 20 min at 4°C. Protein concentrations were determined by Bradford assay (GibcoBRL). Approximately 1.5 mg of total protein was used for TAP tag purification or immunoprecipitation. TAP-tagged extracts were incubated with 25 μl of rabbit IgG agarose at 4°C for 2 h. Beads, equilibrated in TEV protease cleavage buffer, were washed three times with 1 ml of buffer B. Proteins were eluted from the beads with the addition of 25 U of TEV protease in 50 μl of TEV protease cleavage buffer. Eluates were TCA precipitated and analyzed by Western blotting.

Strains YJJ662 and YJJ1330 were used for immunoprecipitation studies. To reduce the binding of nonspecific proteins to the protein A-coupled beads, extracts were preincubated with 25 μl of Sepharose 4B (Sigma) for 1 h at 4°C. The beads were removed, 1 μl of 12CA5 antibody (Roche) was added, and extracts were incubated at 4°C for 1 h. To isolate Leo1-HA complexes, protein A immobilized on Sepharose 4B (Sigma) was added to the extract. After binding for 1 h at 4°C, the beads were pelleted and washed four times with buffer B. Beads were suspended in 20 μl of NuPAGE LDS Sample buffer (Invitrogen) and heated to 70°C for 10 min, and aliquots were resolved on a 4-to-12% Bis-Tris NuPAGE gel (Invitrogen).

Western blotting.

Proteins were resolved on a 4 to 12% bis-Tris NuPAGE gel and then transferred to a polyvinylidene difluoride (Millipore) membrane. Proteins were detected with the following primary antibodies: anti-Rpb1 (8WG16; D. Bentley); anti-Cdc73, anti-Paf1, and anti-TFIIB (X. Shi); anti-Srb5 (R. Young); anti-HA (Roche); anti-Spt5 (G. Hartzog); anti-Rtf1 (K. Arndt); and anti-Rpt5 and anti-Rpt6 (Sug1) (R. Deshaies). Horseradish peroxidase-conjugated antibody was the secondary antibody, and proteins were detected using a Chemiluminescent Reagent Plus kit from NEN.

Mass spectrometry analysis.

Fractions from the Superose 6 gel filtration column were pooled, TCA precipitated, and resolved on a 4-to-12% Bis-Tris NuPAGE gel. Proteins were visualized by staining with NOVEX colloidal blue. Protein bands were excised and processed as described previously (25). Briefly, samples were reduced and alkylated by incubating with 2.8 mM DTT in 100 mM NH₄CO₃ at 60°C for 15 min, followed by the addition of 10 μl of 100 mM iodoacetamide. After a 15-min incubation at ambient temperature in the dark, the gel slices were washed with 100 μl of 50% acetonitrile in 50 mM NH₄CO₃ with shaking for 20 min. To dehydrate the gel slices, 100 μl of 100% acetonitrile was added and incubated for 10 min. The solvent was removed, and the gel pieces were completely dried in a vacuum centrifuge. The gel pieces were swollen with the addition of 0.2 μg of modified trypsin (Promega) in 10 μl of 25 mM NH₄CO_3. The gel pieces were kept wet during enzymatic cleavage (37°C, overnight) with 25 mM NH₄CO₃. The supernatant containing the digested peptides was collected, and the remaining peptides were extracted from the gel twice with 20 μl of 1% trifluoroacetic acid. All supernatants were pooled and dried in a vacuum centrifuge. The sample was reconstituted in 3 to 5 μl of 1% trifluoroacetic acid and desalted using C18 Ziptips (Millipore). Samples were processed on a PE-Biosystems Voyager DE-PRO matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) spectrometer. The masses of the tryptic peptides were used to search for protein candidates using the programs Protein Prospector and ProFound. Up to two undigested trypsin sites and a mass tolerance of ±1.0 Da were allowed.

RNA isolation and analysis.

Cells were grown to a density of 1.3 × 10⁷ cells per ml, and RNA was isolated as described (4). Fifteen micrograms of total RNA per sample was fractionated by formaldehyde gel electrophoresis according to standard methods (33). RNA was transferred to ZetaProbe GT membranes (Bio-Rad). The CLN1 probe, a gel-purified 500-bp fragment from a plasmid which contains the 1.2-kb CLN1 coding region (R. Sclafani), was labeled using the random prime method (RadPrime kit; GibcoBRL). The 18S rRNA oligonucleotide probe (5′-GCTTATACTTAGACATGCAT-3′) was 5′ end labeled. Hybridizations and washes were performed at either 64°C for the CLN1 probe or at 41°C for the 18S rRNA probe. Blots were exposed to a phosphorimaging screen, and bands were quantitated using Quanity One imaging software. Signals were normalized to 18S rRNA.

Affinity purification of the Paf1 complex.

Previous studies of the Paf1 complex used affinity purification techniques to establish the connection to Pol II and to identify some of the components of the complex. Paf1 and Cdc73 were originally identified as RNA Pol II-associated proteins using a monoclonal antibody directed against the C-terminal domain (CTD) of Pol II to isolate the enzyme from yeast transcription extracts (57). The RNA Pol II-associated proteins also included GTFs TFIIF and TFIIB. Subsequently, GST-tagged forms of Paf1, Cdc73, and TFIIF were used to demonstrate that the complex is distinct from the Srb-mediator form of Pol II and that Hpr1 and Ccr4 are also associated with the complex (47). Although the plasmid-borne GST-tagged constructs functionally complemented the genes they replaced (47), they were overexpressed relative to the natural chromosomal copies of the genes. In addition, dimerization of the GST tag and associated proteins (32, 41) causes difficulties in determining the size of the complex. Therefore, to further characterize the composition and function of the Paf1 complex, and to compare it to the Srb-mediator complex, we epitope-tagged the chromosomal copies of CDC73 and SRB5 with a TAP tag (43), as described in Materials and Methods. The TAP tag includes two IgG binding domains and a calmodulin binding peptide separated by a TEV protease site. This C-terminal tag should not alter the normal expression of the tagged genes from their native promoter. Consistent with this supposition, we found that the TAP-tagged strains did not display any of the pleiotropic phenotypes (including slow growth and temperature sensitivity) found for deletion mutants of these genes (data not shown).

Yeast whole-cell transcription extracts were prepared from the tagged strains, the TAP-tagged constructs were isolated by binding to IgG agarose, and bound proteins were eluted by TEV protease digestion. We found that both the Cdc73-tagged Paf1 complex and the Srb5-tagged Srb-mediator complex could be isolated by this procedure (see below) but that neither complex was stable to subsequent purification on calmodulin beads. Apparently the interactions with Pol II are not stable to the second stringent affinity step, although they are maintained under the high-salt (100 mM NH₄SO₄) conditions used for binding to the IgG agarose. Figure 1 shows the results from the IgG agarose isolation of proteins associated with TAP-tagged Cdc73 and Srb5. The input, flowthrough, and eluate fractions from both preparations were probed with antibodies to the largest subunit of Pol II (αRpb1), Paf1 (αPaf1), and Srb5 (αSrb5). Pol II is clearly associated with both tagged proteins, Paf1 is only present in the TAP-tagged Cdc73 eluate, and Srb5 (in its tagged form) is uniquely found in the TAP-tagged Srb5 eluate (Fig. 1). These results establish that the TAP tag protocol specifically and efficiently isolates distinct Pol II complexes from tagged strains.

To further purify the Paf1 complex it was fractionated by chromatography on a Superose 6 gel filtration column (Fig. 2). The elution profile of Paf1 complex components was monitored by immunoblotting using the antibodies indicated in Fig. 2A. A high-molecular-mass complex containing Pol II, Paf1, Cdc73, and TFIIB, but not Srb5, was observed eluting from the column at approximately 1.7 MDa. Cdc73 is present in the high-molecular-mass complex and also in an intermediate position overlapping with a small amount of Pol II. Since Cdc73 is known to bind Pol II directly (47), this may represent a subcomplex of the larger Paf1 complex. We also observed significant amounts of free core Pol II in the lower-molecular-mass fractions (∼600 kDa), a position in the fractionation that also contained proteolyzed forms of Paf1 (Fig. 2, see also below). When the TAP-tagged Srb5 complex was fractionated on Superose 6 under identical conditions, we found that all of the Pol II was in the lower-molecular-mass position of free core, indicating that the Paf1 complex may be somewhat more stable to purification than the Srb-mediator complex.

In addition to the antibody identification of Pol II, we also performed a nonspecific transcriptional assay for RNA polymerase activity on the Superose 6 fractions. Transcriptional activity was detected in fractions 39 to 47 corresponding to the high-molecular-mass Paf1 complex and also in fractions 69 to 73 corresponding to core Pol II (data not shown). Similar nonselective transcription assays performed on GST-tagged constructs (see reference 47) revealed that approximately 2% of the total Pol II was associated with the Paf1 complex (data not shown).

Identification of Paf1 complex components by mass spectrometry.

To identify proteins that associate with the 1.7-MDa TAP-tagged Cdc73 complex, we performed mass spectrometry on individual protein bands excised from a sodium dodecyl sulfate (SDS) gel (Fig. 3). Peptides from the trypsin-digested proteins were eluted and analyzed by MALDI-TOF mass spectrometry as described in Materials and Methods. The tryptic peptide masses were used to search for protein candidates in several Web-based data sets (ProFound, Protein Prospector, and MASCOT). Identification was considered positive if the top candidate had a probability score near 1.0 and the second-best candidate had a significantly lower probability. Proteins that were identified in this way are marked in Fig. 3. Newly identified proteins of the Paf1 complex include Ctr9, Rtf1, and Leo1. The ribosomal proteins identified in Fig. 3 are nonspecific contaminants in the high-molecular-mass fractions, as also observed by others isolating yeast complexes (28, 29, 45). A few peptides corresponding to Spt5 were also identified in the same gel band that contained Rpb2 and Ctr9, but the lower probability score precluded a confident match (indicated by the parentheses in Fig. 3). We also analyzed the proteins in the lower-molecular-mass fractions containing free Pol II (Fig. 3, 580 to 670 kDa). We were able to unambiguously identify most of the Pol II subunits, plus Leo1 and both intact and proteolyzed forms of Paf1.

Validation of newly identified Paf1 complex factors.

To confirm the identifications from mass spectrometry, we showed by Western blotting that Rtf1 and Spt5 were present in the same high-molecular-mass complex with Pol II and Paf1 (Fig. 2B). In addition, we observed Rtf1 and Spt5 in an intermediate fraction between the Pol II complex and free core Pol II, suggesting that they might exist in a separate subcomplex.

The Koch laboratory described an association between Ctr9 and Paf1 and Cdc73 but did not establish a connection to Pol II (22). To confirm that Ctr9 is part of the Paf1 complex, we constructed a strain bearing a TAP-tagged form of Ctr9 and confirmed that the tagged strain had a wild-type phenotype. We isolated TAP-tagged Ctr9-associated proteins as described above for TAP-tagged Cdc73. Fractions from the Superose 6 gel filtration column were immunoblotted with antibodies specific for Pol II and Paf1 (Fig. 4A). Both Paf1 and Pol II were found in these fractions, again in the high-molecular-mass complex and at the position of free core Pol II. Based on these observations, and the mass spectrometry identification, we conclude that Ctr9 is part of the Paf1 complex.

We have previously used a GST-tagged form of Hpr1 to establish that this protein is associated with Pol II and Paf1 (4). Because Hpr1 has also been found in another, non-Pol II-containing complex, the Hpr1/Tho complex (5), we confirmed the Pol II association by analyzing proteins associated with TAP-tagged Hpr1. The comigration of Pol II, Paf1, and Hpr1 in the high-molecular-mass fractions of the Superose 6 column (Fig. 4B) reaffirms that Hpr1 associates with the Paf1 complex.

There is little information about Leo1 other than the fact that it is a very hydrophilic protein encoded by a nonessential gene (30). We therefore constructed strains containing chromosomally HA-tagged Leo1 in wild-type and in various TAP-tagged backgrounds (Materials and Methods). Using an anti-HA antibody to precipitate HA-tagged Leo1 we found both Paf1 and Rtf1, but not Srb5, in the bound fraction (Fig. 5A). However, although Paf1 was not present in the precipitate from the untagged strain, a small amount of Rtf1 was detected. We have observed low levels of nonspecific association of Rtf1 in many of the different affinity assays performed in these studies (see also below); however, the specific association of Rtf1 is clearly apparent in Fig. 5A over this background. To confirm that Leo1 associates specifically with the Paf1 complex, we used double-tagged strains (TAP tags plus Leo1-HA), first isolating the complexes with the TAP tag and then probing for the presence or absence of HA-tagged Leo1 (Fig. 5B). We found that Leo1, like Paf1 and Rtf1, associates with TAP-tagged Cdc73 and Ctr9, but not with TAP-tagged Srb5. These results confirm that Leo1 is part of the Paf1 complex and is not part of the Srb-mediator.

Proteasome components are not part of the Paf1 complex.

Paf1, Ctr9, Rtf1, and Leo1 were recently reported to associate with the 19S cap of the proteasome in the absence of ATP (53). However, we did not identify any proteasome components in the mass spectrometry analysis of the Paf1 complex. In addition, when we used antibodies specific to 19S proteasome components Sug1 and Rpt1 to probe TAP-tagged Cdc73 fractions, no proteasome components were found associated with the Paf1 complex (Fig. 6). The reported association of the Paf1 complex with the proteasome may therefore be an artifact of the affinity isolation used in these earlier studies. In particular, the nonspecific association of Rtf1 with many affinity matrices noted above (Fig. 5) may have led to these observations. Using antibodies directed against components of several other high-molecular-mass yeast complexes, we have found no evidence for the presence of Spt6, Bur2, Snf12, or Caf1 in the purified Paf1 complex (data not shown).

Genetic interactions between PAF1, RTF1, and LEO1.

We have previously reported that although PAF1 is not an essential gene, its loss results in many phenotypic changes, including slow growth; increased sensitivity to low and high temperature, cell wall-damaging agents, and hydroxyurea; and abnormal morphology relative to wild-type cells (4, 48; Betz et al., submitted). Deletion of CTR9 results in an identical spectrum of phenotypes (22; Betz et al., submitted). Loss of CDC73 results in much more modest phenotypic changes that are in general a subset of those seen for paf1Δ (4, 47; Betz et al., submitted). Combining a PAF1 deletion with deletions of CDC73 or CTR9 does not result in any phenotypic enhancement (22, 47), consistent with the fact that they function in the same cellular processes. If Rtf1 and Leo1 are also involved in the same cellular process, then combining these mutations with mutations in other Paf1 complex components should not result in an enhanced phenotype (14). We therefore constructed leo1Δ and rtf1Δ mutants and used these to create double mutants by mating to an appropriate isogenic paf1Δ strain. The diploids were sporulated, and tetrads were dissected and tested for genotype and phenotype.

A point mutation in RTF1 was originally characterized as a suppressor of a mutation in TATA-binding protein (TBP) (48). Deletion of RTF1 also suppresses the same allele of TBP, and both the point mutation and the deletion result in alterations in transcriptional start sites. We found that rtf1Δ strains grew a little more slowly than wild type at 30°C in rich media (Fig. 7A) (doubling times, 97 versus 80 min) and were slightly temperature sensitive at 36°C (Fig. 7B). However, in contrast with a recent report (8), we observed that loss of Rtf1 in our yeast genetic background did not result in sensitivity to 6-azauracil (6AU) (data not shown). When we analyzed several rtf1Δ paf1Δ double mutants, we found that loss of Rtf1 dramatically suppressed many paf1Δ phenotypes. An rtf1Δ paf1Δ double mutant grew significantly faster than paf1Δ in rich media at 30°C (Fig. 7A) (doubling times of 145 versus 175 min), although more slowly than rtf1Δ. In addition, even though rtf1Δ alone is slightly temperature sensitive, the mutation suppressed the temperature-sensitive phenotype of paf1Δ (Fig. 7B). Loss of Rtf1 also restored growth of paf1Δ on caffeine and hydroxyurea (Fig. 7B). We also found that rtf1Δ suppressed many of these phenotypes when combined with ctr9Δ (data not shown). As previously reported (30), we found that leo1Δ strains did not demonstrate any obvious phenotype and grew essentially as well as wild type (Fig. 7C). However, we found that a leo1Δ paf1Δ double mutant was less temperature sensitive and slightly more resistant to hydroxyurea than a paf1Δ strain (Fig. 7C). Therefore, removal of either Rtf1 or Leo1 can suppress some of the pleiotropic defects caused by loss of Paf1.

Loss of Rtf1 restores CLN1 expression in paf1Δ.

Loss of Paf1 results in changes in expression patterns of many yeast genes (4, 48; Betz et al., submitted). Since loss of Rtf1 does not completely suppress the paf1Δ phenotype, we would not expect that all transcription would be restored to normal levels. However, since the double mutant grows significantly faster than the paf1Δ strain, we analyzed the level of expression of the G₁ cyclin CLN1 in these mutants. Koch and coworkers found that Paf1 and Ctr9 are required for full expression of CLN2, another G₁ cyclin (29), and we have found that Paf1 is required for normal expression of many other cell cycle-regulated genes, including CLN1 (Betz et al., submitted). We found that in paf1Δ, CLN1 mRNA was reduced about threefold, whereas in an rtf1Δ paf1Δ strain CLN1 expression was restored to wild-type levels (Fig. 8). We have observed a similar restoration of mRNA levels for RNR1 and CYC1 in rtf1Δ paf1Δ relative to paf1Δ (data not shown; E. Amiott, personal communication). The restoration of RNR1 transcripts provides a potential molecular explanation for the suppression of hydroxyurea sensitivity seen in Fig. 7B. The fact that rtf1Δ, and to a lesser extent leo1Δ, can suppress the transcriptional defects caused by loss of Paf1 is consistent with the idea that a partially defective Paf1 complex may block transcription and that removal of additional complex components may relieve the block.

Ctr9, required for G₁ cyclin expression, is part of the Paf1 complex.

CTR9, also known as CDP1, has been isolated in two genetic screens, one looking for genetic interactions with a centromere binding factor (10) and the second searching for genes required for G₁ cyclin transcription (22). Loss of Ctr9 results in chromosome instability (10); reduction in the expression of G₁ cyclins (21); slow growth; and sensitivity to extremes of temperature, hydroxyurea, and many cell wall-damaging agents, including caffeine and SDS (21; Betz et al., submitted). Except for the chromosome instability phenotype, which has not been tested, all other known phenotypes of ctr9Δ are shared by paf1Δ. Loss of Paf1 does, however, lead to an increase in recombination (4), which may be related to the loss of chromosome stability seen for ctr9Δ. Like ctr9Δ, paf1Δ strains have a severe defect in the expression of the G₁ cyclins CLN1 and CLN2. In addition Paf1 is required for full expression of many other cell cycle-regulated genes, including HO and RNR1 (Porter et al., submitted). Paf1 is also critical for signaling from the protein kinase C-mitogen-activated protein kinase cascade to target genes required for maintenance of cell integrity (4). The requirement for Paf1 and Ctr9 in the expression of a significant subset of yeast genes is true both for the target genes in their chromosomal locations and for a variety of promoter-reporter constructs (4, 10, 22). In fact, CTR9 was originally identified as a factor required for G₁ cyclin expression in a screen using cell cycle-regulated promoter elements fused to a reporter gene (22). Therefore, Paf1 and Ctr9 act through promoter elements to effect normal transcript levels and to respond to signals during the cell cycle and in response to cell damage.

Rtf1, important for both start site selection and elongation, is also part of the Paf1 complex.

RTF1 was originally identified in a genetic screen for mutations that would suppress a defective form of TBP (50). The spt15-122 allele of TBP has an altered specificity for TATA box recognition, which results in changes in transcriptional start sites (2). The rtf1-1 allele suppresses the phenotypes caused by the spt15-122 mutation and further alters start site selection but does not restore initiation sites to their wild-type positions (50). However, complete deletion of RTF1 also suppresses the spt15-122 defects and restores wild-type start site selection (50). It is interesting that mutations in Ccr4, also found in the Paf1 complex (4), also have extensive genetic interactions with TBP and the spt15-122 allele (3). Based on these studies it is clear that Rtf1, presumably as part of the Paf1 complex, plays an important role at an early stage of transcription.

In addition to its role in initiation, Rtf1 was recently reported to also have extensive genetic interactions with transcriptional elongation factors (8). In particular, rtf1Δ is synthetically lethal in combination with mutations in CTK1, a Pol II CTD kinase (8); FCP1, a CTD phosphatase (8); and POB3, a chromatin remodeling factor that plays roles in both initiation and elongation (11). In addition, RTF1 genetically interacts with other elongation factors, including TFIIS (PPR2) and SPT5 (8). The interaction with SPT5 is of particular interest since we detected peptides from Spt5 in our mass spectrometry analysis of TAP-tagged Cdc73, and an anti-Spt5 antibody detects a protein of the correct size for Spt5 in the high-molecular-mass Paf1 complex (Fig. 2B). In addition, Paf1, Cdc73, and Rtf1 have been identified as Spt5-interacting factors (G. Hartzog and K. Arndt, personal communication).

Like those in RTF1, mutations in SPT5 result in changes in transcription start sites and also have effects on elongation (16, 51), and Spt5 interacts with both the initiating and elongating forms of Pol II (26). Spt5 and Spt4, found together in a complex, are the yeast homologues of the human DSIF transcriptional elongation factor identified biochemically by Hartzog et al. (16). The DSIF elongation factor confers sensitivity to the drug DRB (5,6-dichloro-1-β-d-ribofuranosyl benzidazole) in vitro and is thought to play a negative role in transcription elongation in vivo (55). DRB inhibits the pTEFb kinase that phosphorylates the Pol II CTD and is required for an early stage of elongation in vitro (31). Some of the evidence that Spt5 and Rtf1 affect elongation comes from phenotypic analysis of mutants, specifically that mutation of either gene results in sensitivity to 6AU (8, 16). However, the connection between 6AU sensitivity and transcriptional elongation defects is complex. Mutation of the elongation factor TFIIS (PPR2) or the Rpb1 (1), Rpb2 (42) or Rpb9 (17) subunits of Pol II results in sensitivity to 6AU. In vitro, the 6AU-sensitive mutant of Rpb2 clearly has defects in elongation. However, in vivo the major effect of the various 6AU-sensitive mutants seems to be reduced expression of the PUR5 gene encoding IMP dehydrogenase (17, 46). Since the reduction in PUR5 expression can be transferred using a reporter construct fused to the promoter and 5′-untranslated region of the PUR5 gene (46), it is not yet clear what stage of transcription is affected by the 6AU-sensitive mutations in vivo.

Is the Paf1 complex involved in transcriptional elongation? Although paf1Δ strains are slow growing and sensitive to a variety of compounds, they are not sensitive to 6AU (Betz et al., submitted) although an isogenic ppr2 strain is strongly inhibited. In addition, rtf1Δ strains are not 6AU sensitive in our genetic background (C. Mueller, unpublished). Neither cdc73Δ nor hpr1Δ strains are 6AU sensitive, although ccr4Δ is very sensitive to the drug (9; Betz et al., submitted) and leo1Δ is slightly sensitive (C. Mueller, unpublished). As described above, sensitivity to 6AU neither proves nor disproves a protein's role in elongation. However, the many genetic connections between RTF1 and elongation factors (8) and the fact that the homologue of Spt5, DSIF, clearly is important for elongation in vitro are compelling evidence for a link.

The Paf1 complex is not the only factor that may straddle the line between initiation and elongation; many transcription factors seem to function at multiple stages of transcription. The GTF TFIIF is clearly required for initiation, promoter escape, and elongation (44, 59), and TFIIH's roles in both initiation and promoter escape can be traced to its action at different regions of downstream promoter DNA (49). Mutations in the Rpb9 subunit of Pol II reveal effects on both start site selection (12) and elongation (17). In addition, the Pob3-Spt16 chromatin remodeling complex, which has genetic links to RTF1 (8), also appears to be important for both initiation and elongation of transcription (11). Since initiation, promoter clearance, and entry into elongation are clearly a multistep process (39), perhaps it is not surprising that so many factors seem be involved in this critical transition. However, a central role in elongation for the Paf1 complex is not consistent with the fact that loss of Paf1 or other complex components affects the expression of only a small subset of genes. In addition, our observation that loss of both Paf1 and Rtf1 returns cells to a nearly wild-type phenotype does not support the concept that the Paf1 complex might be required for elongation of all yeast genes. Instead, our results fit a model wherein the Paf1 complex functions at a small subset of promoters to enhance expression in response to signaling pathways (i.e., protein kinase C, cell cycle). Partial defects in the complex, such as loss of Paf1 or Ctr9, could result in a defective complex that blocks transcription. Removal of a second component, such as Rtf1, may release the blocked complex and allow the return to nearly normal transcription patterns, perhaps through the action of the Srb-mediator form of Pol II. Testing this model will require the identification of the primary gene targets of the Paf1 complex.