The Recruitment of the Saccharomyces cerevisiae Paf1 Complex to Active Genes Requires a Domain of Rtf1 That Directly Interacts with the Spt4-Spt5 Complex

Abstract

Transcription elongation factors associate with RNA polymerase II and aid its translocation through chromatin. One such factor is the conserved Paf1 complex (Paf1C), which regulates gene expression through several mechanisms, including the stimulation of cotranscriptional histone modifications. Previous studies revealed a prominent role for the Rtf1 subunit in tethering Paf1C to the RNA polymerase II elongation machinery. Here, we investigated the mechanism by which Rtf1 couples Paf1C to active chromatin. We show that a highly conserved domain of Rtf1 is necessary and sufficient for mediating a physical interaction between Rtf1 and the essential transcription elongation factor Spt5. Mutations that alter this Rtf1 domain or delete the Spt5 C-terminal repeat domain (CTR) disrupt the interaction between Rtf1 and Spt5 and release Paf1C from chromatin. When expressed in cells as the only source of Rtf1, the Spt5-interacting domain of Rtf1 can associate independently with active genes in a pattern similar to that of full-length Rtf1 and in a manner dependent on the Spt5 CTR. In vitro experiments indicate that the interaction between the Rtf1 Spt5-interacting domain and the Spt5 CTR is direct. Collectively, our results provide molecular insight into a key attachment point between Paf1C and the RNA polymerase II elongation machinery.

INTRODUCTION

Packaging of the eukaryotic genome into chromatin hinders the movement of RNA polymerase II (Pol II) during transcription. Hence, eukaryotes have evolved numerous initiation and elongation factors that orchestrate the recruitment and movement of RNA Pol II across active genes. The Paf1 (polymerase-associated factor I) complex (Paf1C) is one such conserved elongation factor. In Saccharomyces cerevisiae, Paf1C consists of five subunits: Paf1, Ctr9, Cdc73, Leo1, and Rtf1 (1–6). Paf1C regulates transcription and chromatin structure through several mechanisms. The most well studied functions of Paf1C are its roles in regulating cotranscriptional histone modifications. Through a mechanism that requires the histone modification domain of Rtf1, Paf1C is required for the efficient monoubiquitylation of lysine 123 (K123) of histone H2B by the ubiquitin-conjugating enzyme Rad6 and the ubiquitin-protein ligase Bre1 (7–10). This mark is a prerequisite for the di- and trimethylation of histone 3 lysine 4 (H3 K4) and K79 by the Set1 and Dot1 histone methyltransferase, respectively (7, 11–14). Paf1C also facilitates H3 K36 trimethylation on the bodies of active genes (15). Additional functions of Paf1C include the Rtf1-mediated recruitment of the Chd1 chromatin remodeling protein (16), the stimulation of serine 2 phosphorylation on the C-terminal domain (CTD) of RNA Pol II (17, 18), the maintenance of nucleosome occupancy on highly expressed regions of the genome (19), and the recruitment of factors involved in transcription termination and RNA processing (17, 18, 20, 21). Either individually or collectively, the multiple functions of Paf1C can influence the rate of transcription elongation in vitro and in vivo (22–24). In addition to its roles in elongation, Paf1C has also been shown to affect the initiation and termination stages of the transcription cycle (17, 18, 20, 21, 25–27).

Members of Paf1C were first discovered in a search for proteins that associate with RNA Pol II (28). Consistent with its physical association with RNA Pol II, Paf1C is enriched on the bodies of actively transcribed genes at levels that correlate with gene expression (29). Previous studies have implicated several proteins in the recruitment of yeast Paf1C to active chromatin, including the transcription elongation factors Spt16-Pob3/FACT, the Ccr4-Not complex, and Spt4-Spt5/DSIF (30–32). The Bur1-Bur2 protein kinase stimulates the recruitment of Paf1C to RNA Pol II through the phosphorylation of the C-terminal repeats (CTRs) of Spt5 and through a pathway independent of this function (30, 33–37). In addition, the Kin28 protein kinase promotes the recruitment of Paf1C through phosphorylation of the CTD of RNA Pol II and by facilitating the chromatin association of Bur1-Bur2 (35, 38). With respect to members of Paf1C, loss of the Rtf1, Cdc73, or Leo1 subunits reduces the occupancy of Paf1C on chromatin (18, 39, 40). In vitro, recombinant Cdc73, Rtf1, and Ctr9 can bind to peptides corresponding to the RNA Pol II CTD phosphorylated on serines 2 and 5 and to peptides corresponding to the phosphorylated Spt5 CTR (35). For Cdc73, this peptide binding activity maps to a domain that adopts a Ras-like fold and is important for Paf1C recruitment in vivo (35, 41). The ability of Leo1 to facilitate chromatin association of Paf1C correlates with its ability to bind RNA (39). The mechanism of chromatin association of Paf1C through the Rtf1 subunit remains obscure.

Through genetic deletions, we previously identified a highly conserved region within Rtf1 that is important for the chromatin association of Paf1C and termed this region the open reading frame (ORF) association region (OAR) of Rtf1 (40). The Rtf1 OAR contains a Plus3 motif, highlighted by the presence of three conserved, positively charged amino acids (42). A nuclear magnetic resonance (NMR) study of the human Rtf1 Plus3 domain demonstrated that this domain is structurally similar to Tudor domains, which participate in protein-protein interactions, and the PAZ domains found in Dicer and Argonaute proteins (42–44). In this study, we sought to identify the mechanism of recruitment of Paf1C to active chromatin through this highly conserved region of Rtf1. Using a subtractive proteomics approach, we discovered that the OAR of Rtf1 mediates the interaction of Paf1C with Spt5. Using purified recombinant proteins, we obtained evidence for a direct interaction between the Rtf1 OAR and Spt5, independent of the other members of Paf1C. We also show that both the deletion of the Spt5 CTR and mutation of the Bur1-Bur2 complex impair the recruitment of full-length Rtf1 and the OAR alone, suggesting that the OAR is a crucial target for the recruitment of Paf1C by the Spt5 CTR and Bur1-Bur2.

MATERIALS AND METHODS

Yeast strains and growth.

The S. cerevisiae strains used in this study (Table 1) are isogenic to FY2, a GAL2⁺ derivative of strain S288C (45). Yeast transformations and matings were performed as previously described (46, 47). Rich (yeast extract-peptone-dextrose [YPD]), synthetic complete (SC), and minimal (synthetic dextrose [SD]) media were made as previously described (47). 6-Azauracil (6-AU) was added to SC-uracil (Ura) medium at a final concentration of 50 μg/ml. Except where otherwise indicated below, cells were grown at 30°C. For serial dilution growth assays, yeast strains were grown overnight to saturation in the appropriate medium and washed twice with sterile water. Using sterile water, cells were 10-fold serially diluted from a starting concentration of 1 × 10⁸ cells/ml. An amount of 2.5 μl of each dilution was then spotted onto the appropriate medium and incubated at 30°C or 37°C for the specified number of days.

Table 1.

Saccharomyces cerevisiae strains used in this study

Strain	Genotypea
KY619	MATa rtf1Δ102::ARG4 arg4-12 his4-912δ leu2Δ1 lys2-173R2 trp1Δ63
KY995	MATα rtf1Δ::URA3 CTR9-6×MYC::LEU2 his3Δ200 leu2Δ(0 or 1) ura3(Δ0 or 52) trp1Δ63
KY1220	MATα HA3-PAF1 his3Δ200 leu2Δ1 ura3Δ0
KY1258	MATa rtf1Δ::URA3 RAD6-13×MYC::KanMX leu2Δ1 ura3-52 trp1Δ63
KY1758	MATα rtf1Δ101::LEU2 his4-912δ lys2-128δ leu2Δ1 trp1Δ63 arg4-12
KY1813	MATα paf1Δ::KanMX rtf1Δ::LEU2 his4-912δ leu2Δ1 trp1Δ63 ura3-52
KY2124	MATa rtf1Δ::KanMX4 hta1-htb1Δ::LEU2 hta2-htb2Δ::KanMX his3Δ200 lys2-128δ leu2Δ1 ura3-52 trp1Δ63 pJH23 WT [HTA1-HTB1/HIS3/CEN/ARS/Ampr]
KY2125	MATa rtf1Δ::KanMX4 hta1-htb1Δ::LEU2 hta2-htb2Δ::KanMX his3Δ200 lys2-128δ leu2Δ1 ura3-52 trp1Δ63 pJH23 FL [HTA1-FLAG-HTB1/HIS3/CEN/ARS/Ampr]
KY2195	MATα rtf1Δ101::LEU2 cdc73Δ::KanMX4 his4-912δ leu2Δ1 trp1Δ63
KY2410	MATa rtf1Δ101::LEU2 HA3-PAF1 his3Δ200 leu2Δ1 ura3(Δ0 or 52)
KY2413	MATa his3Δ200 leu2Δ1 ura3(Δ0 or 52) trp1Δ63
KY2414	MATa rtf1Δ101::LEU2 HA3-PAF1 leu2Δ1 ura3(Δ0 or 52) trp1Δ63
KA150	MATα rtf1Δ101::LEU2 spt5ΔCTR::NATR his3Δ200 lys2-128δ leu2Δ1 ura3Δ0 trp1Δ63 arg4-12
KA181	MATα spt5Δ::HIS3 rtf1Δ::KanMX his3Δ(1 or 200) leu2Δ(0 or 1) trp1Δ63 pHQ1494 [LEU2 SPT5-3×HA]
KA183	MATα spt5Δ::HIS3 rtf1Δ::KanMX his3Δ(1 or 200) leu2Δ(0 or 1) ura3Δ0 trp1Δ63 met15Δ0 pHQ1494 [LEU2 spt5-S1-15A-3×HA]
KA185	MATa spt5Δ::HIS3 rtf1Δ::KanMX his3Δ(1 or 200) leu2Δ(0 or 1) pHQ1494 [LEU2 spt5-S1-15D-3×HA]
AY777b	MATα bur2-1 rtf1Δ::KanMX4 his4-912δ lys2-128δ suc2Δuas(-1900 or -390) ura3-52 trp1Δ63

^aWT, wild type; FL, full length.

^bProvided by Greg Prelich.

Plasmid construction.

Some of the plasmids used in this study are described in Table 2. Mutations encoding amino acid substitutions in the OAR were constructed by site-directed mutagenesis using the QuikChange mutagenesis kit (Agilent) and pLS21-5 (48) as the template. pMM25 encoding glutathione S-transferase (GST)-OAR was created by subcloning the coding sequence for amino acids 235 to 373 of Rtf1, which was PCR amplified from pLS21-5 with primers that introduce EcoRI and BamHI sites at the 5′ and 3′ ends of the PCR product and ligated to EcoRI-BamHI-digested pGEX-3X (49). pAP21 expresses GST-Rtf1-His₆. pAP21 was generated by insertion of the coding sequence for the His₆ tag at the 3′ end of the RTF1 ORF in plasmid pJS4 to create a version of Rtf1 that has an N-terminal GST tag and a C-terminal His₆ tag (40). pMM26, which expresses GST-Rtf1ΔOAR-His₆, was constructed by deleting the sequence coding for Rtf1 amino acids 230 to 390 from pAP21 using a QuikChange mutagenesis kit (Agilent). pGH25, pGH100, and pGH258 were gifts from Grant Hartzog and express a maltose-binding protein (MBP)-Spt5 CTR fusion protein (residues 807 to 1063 of Spt5), MBP, and Flag-Spt5-His₆ plus Spt4, respectively. The plasmid expressing His₆-OAR (residues 235 to 373 of Rtf1) was a gift from Andrew VanDemark.

Table 2.

Plasmids used in this study

Plasmid	Construction	Origin	Promoter	Protein
pMM01	pLS20 derivative (4); created using site-directed mutagenesis	CEN/ARS	RTF1 promoter	HA3-Rtf1ΔOAR (lacks residues 230 to 390 of Rtf1)
pMM03	pPC59 derivative (4); created using site-directed mutagenesis	CEN/ARS	RTF1 promoter	Rtf1ΔOAR-TAP (lacks residues 230 to 390 of Rtf1)
pMM35	pAP37 derivative (10); PCR-amplified fragment encoding the OAR and containing NdeI and PstI restriction sites was subcloned into NdeI-PstI-digested pAP37	2μ	ADH1 promoter	NLS-Myc-OAR (residues 230 to 390 of Rtf1)
pMM36	pMM35 derivative; the Myc tag was removed, and an EcoRI site was introduced by site-directed mutagenesis to create a vector for introducing a PCR-amplified HA3 tag sequence from pLS21-5 having EcoRI and NcoI overhangs	2μ	ADH1 promoter	NLS-HA3-OAR (residues 230 to 390 of Rtf1)
pMM41	pLS20 derivative; a PCR-amplified NLS-HA3-OAR-encoding fragment with NdeI and AflII overhangs was subcloned into pLS21-5 digested with NdeI and AflII to replace the RTF1 gene in pLS21-5	CEN/ARS	RTF1 promoter	NLS-HA3-OAR (residues 230 to 390 of Rtf1)
pMM43	pMM40 derivative (10); NdeI fragment of pLS21-5 encoding the HA3 tag was subcloned into NdeI-digested pMM40	CEN/ARS	ADH1 promoter	HA3-Rtf1
pMM44	pMM41 derivative; the RTF1 promoter was replaced with the ADH1 promoter, which was PCR amplified from pGBKT7 (Clontech) using primers that introduce SalI and NdeI sites at the 5′ and 3′ ends of the PCR product	CEN/ARS	ADH1 promoter	NLS-HA3-OAR (residues 230 to 390 of Rtf1)
pMM47	pMM01 derivative; NdeI restriction fragment encoding the HA3 tag was deleted	CEN/ARS	RTF1 promoter	Rtf1ΔOAR (lacks residues 230 to 390 of Rtf1)
pMM61	pRS316 derivative; XhoI-SacI fragment from pLS20 including the RTF1 promoter and the RTF1 (untagged)-coding region was ligated to XhoI-SacI-digested pRS316	CEN/ARS	RTF1 promoter	Rtf1
pMM62	pRS316 derivative; XhoI-SacI fragment from pMM47 including the RTF1 promoter and the rtf1ΔOAR-coding region was ligated to XhoI-SacI-digested pRS316	CEN/ARS	RTF1 promoter	Rtf1ΔOAR (lacks residues 230 to 390 of Rtf1)

Immunoblot analyses.

Log-phase cultures of yeast cells grown to an optical density at 600 nm (OD₆₀₀) of approximately 0.8 were used to make trichloroacetic acid (TCA) extracts by bead beating as described previously (10). Proteins were resolved on SDS-polyacrylamide gels, transferred to nitrocellulose or polyvinylidene difluoride (PVDF) membranes, and probed with primary antibodies against Rtf1 (4), the hemagglutinin (HA) epitope (1:2,500 dilution; Roche), total histone H3 (1:30,000 dilution) (10), trimethylated H3 K4 (H3 K4 Me3) (39159, 1:2,000 dilution; Active Motif), H3 K4 Me2 (07-030, 1:2,000 dilution; Upstate), H3 K79 Me2/3 (ab2621, 1:2,000 dilution; Abcam), Paf1 (1:1,000 dilution; gift from Judith Jaehning), glucose-6-phosphate dehydrogenase (G6PDH) (A9521, 1:30,000 dilution; Sigma), tobacco etch virus (TEV) protease-cleaved tandem affinity purification (TAP) tag (CAB 1001, 1:2,500; Thermo Scientific), or an antibody that detects the uncleaved TAP tag (P1291, peroxidase-antiperoxidase, 1:2,000 dilution; Sigma). After incubation with the primary antibodies, membranes were probed with sheep anti-mouse or donkey anti-rabbit secondary antibodies (1:5,000 dilution; GE Healthcare) and visualized using enhanced chemiluminescence substrate (PerkinElmer).

Coimmunoprecipitation assays.

Yeast transformants grown in selective medium to a density of 3 × 10⁷ to 4 × 10⁷ cells/ml were lysed by bead beating in a lysis buffer containing 20 mM HEPES, pH 7.4, 100 mM sodium acetate, 2 mM magnesium acetate, 10 mM EDTA, 0.1% Tween 20, 10% glycerol, 1 mM dithiothreitol (DTT), and protease inhibitors (Halt protease inhibitor cocktail; Thermo Scientific). Five hundred to 1,000 μg of the clarified extract was incubated with anti-Spt5 antisera (1:1,000 dilution; gift from Grant Hartzog) for 2 h at 4°C, followed by incubation with protein A-conjugated agarose (GE Healthcare) for 1 h at 4°C. The agarose beads were then washed with lysis buffer containing 400 mM sodium acetate, and the immunoprecipitated proteins were subjected to immunoblot analysis using anti-Spt5 and anti-HA (1:2,500, Roche) antibodies to detect the immunoprecipitated Spt5 and HA-Rtf1 protein, respectively.

Chromatin immunoprecipitation (ChIP) assays.

Yeast strains were grown to a density of 1 × 10⁷ to 2 × 10⁷ cells/ml. As described previously, cells were cross-linked with formaldehyde, quenched with glycine, harvested, and lysed (10). Soluble chromatin was prepared by sonication (10) and then incubated with agarose-conjugated anti-HA antibody (sc-7392AC; Santa Cruz Biotechnology), anti-Spt5 antibody (gift from Grant Hartzog), anti-Spt16 antibody (gift from Tim Formosa), or anti-Rpb3 antibody (Neoclone) overnight at 4°C. This was followed by a 2-h incubation at 4°C with protein A-conjugated agarose (GE Healthcare) for anti-Spt5 and anti-Spt16 antibodies or protein G-conjugated agarose (GE Healthcare) for anti-Rpb3 antibody. Input and immunoprecipitated DNA were purified and analyzed by quantitative real-time PCR using Maxima SYBR green/ROX quantitative PCR (qPCR) master mix (Fermentas) and primers (+1 relative to ATG) for the 5′ coding region of PYK1 (+253 to +346), the 3′ coding region of PYK1 (+1127 to +1270), the region beyond the poly(A) site of PYK1 (+1803 to +1938), the 5′ coding region of PMA1 (+214 to +319), the 3′ coding region of PMA1 (+2107 to +2194), the region beyond the poly(A) site of PMA1 (+3373 to +3475), or a telomeric region of chromosome VI (coordinates, 269495 to 269598). The graphs in the figures represent the average values for three biological replicates, and the error bars indicate the standard errors of the means (SEM) for those values.

Purification of TAP-tagged proteins.

Transformants of rtf1Δ strains (KY1258 or KY619) expressing plasmid-encoded Rtf1 derivatives were grown to log phase (3 × 10⁷ to 4 × 10⁷ cells/ml). Whole-cell extracts were made by bead beating. TAP-tagged proteins were then subjected to one-step affinity purification using rabbit IgG (I5006; Sigma)-conjugated magnetic beads (M-270 epoxy; Invitrogen) as previously described (41). The bound proteins were eluted by cleavage with TEV protease (12575-015; Invitrogen) for 3 h at 15°C, concentrated by TCA precipitation, and then run on 7.5% SDS-polyacrylamide gels. For the identification of proteins that copurified with the Rtf1 derivatives, samples were run approximately 1 cm into the gel, excised, and analyzed by tandem mass spectrometry in an Orbitrap mass spectrometer (Fred Hutchinson Cancer Research Center Proteomics Facility). The peptides identified were validated as described previously (41). The average numbers of peptides identified from a minimum of three independent purifications are shown in Figure 2A.

Fig 2.

The OAR is important for the interaction of Rtf1 with Spt5. (A) Extracts of an rtf1Δ strain (KY1258) bearing plasmids expressing Rtf1 (pLS20), Rtf1-TAP (pPC59), and Rtf1ΔOAR-TAP (pMM03) were subjected to one-step affinity purification using IgG-conjugated magnetic beads, and mass spectrometric analysis was performed on the proteins isolated. The average number of peptides identified for each protein (minimum of three trials) is listed. A subset of the results is shown. (B) Immunoblotting analysis was performed to determine the levels of Rtf1 in transformants of an rtf1Δ strain (KY619) expressing Rtf1-TAP (pPC59) or Rtf1ΔOAR-TAP (pMM03) by loading 9 μl, 7 μl, and 5 μl of each of the indicated extracts. G6PDH served as the loading control. (C) Immunoblotting analysis of the affinity-purified Rtf1 proteins from extracts of transformants of an rtf1Δ strain (KY619) containing the plasmids described in the legend to panel A. (D) ChIP analysis was performed using anti-Spt16 antibody to determine the occupancy of Spt16 over 5′ and 3′ regions of PYK1 and PMA1 and a telomeric region of chromosome VI (Tel VI). Chromatin from transformants of an rtf1Δ strain (KY619) expressing HA₃-Rtf1 (pLS21-5) or HA₃-Rtf1ΔOAR (pMM01) was used. As a negative control, ChIP analysis was performed with chromatin from HA₃-Rtf1 (pLS21-5)-expressing transformants without the addition of antibody (No ab). (E) Spt5 was immunoprecipitated from extracts of transformants of an rtf1Δ strain (KY619) transformed with a plasmid expressing HA₃-Rtf1 (pLS21-5) or HA₃-Rtf1ΔOAR (pMM01) and a transformant of an rtf1Δ spt5ΔCTR strain (KA150) expressing HA₃-Rtf1 (pLS21-5). Extracts of transformants expressing HA₃-Rtf1 (pLS21-5) subjected to similar analysis without the addition of anti-Spt5 antibody (No ab) served as the negative control. Immunoblotting with anti-HA and anti-Spt5 antibodies was performed to detect Rtf1 and Spt5, respectively. The results shown are representative of those from three separate experiments. (F) Immunoblotting analyses were performed using indicated antibodies on transformants of an rtf1Δ (KY619) strain containing plasmids expressing HA₃-Rtf1 (pLS21-5) or HA₃-Rtf1ΔOAR (pMM01) and transformants of an rtf1Δ spt5ΔCTR (KA150) strain containing plasmid expressing Rtf1 (pLS21-5). An rtf1Δ strain (KY619) transformed with empty vector (pRS314) was used as the negative control. G6PDH and total H3 levels served as loading controls. (G) ChIP analysis of the occupancy of Spt5 over 5′ and 3′ regions of PYK1 and PMA1 and over a telomeric region of chromosome VI was performed with chromatin prepared from strains described in the legend to panel D.

Expression and purification of recombinant proteins.

The expression of Flag-Spt5-His₆ together with Spt4 in Escherichia coli codonplus-RIL cells transformed with plasmid pGH258 was induced by growth in LB medium containing 0.1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) and 0.3 mM zinc chloride overnight at 20°C. Cells were harvested and lysed using a homogenizer in a buffer containing 25 mM Tris, pH 8.0, 500 mM NaCl, 10% glycerol, and 5 mM imidazole. Spt5 was purified from the clarified lysate by nickel affinity chromatography (Qiagen). It was then subjected to a second round of purification by heparin affinity chromatography in a buffer containing 25 mM Tris, pH 6.5, 100 mM NaCl, and 8% glycerol. OAR-His₆ expression was induced in E. coli codonplus-RIPL cells in ZY autoinduction medium (50) for ∼24 h at 37°C and purified following the same procedure as that used for the purification of Spt5. The expression of GST in E. coli codonplus-RIL cells and GST-Rtf1-His₆ and GST-Rtf1ΔOAR-His₆ in E. coli codonplus-RIPL cells was induced in LB medium containing 0.1 mM IPTG at 37°C for 3 h. As described above for the purification of Spt5, GST-Rtf1-His₆ and GST-Rtf1ΔOAR-His₆ were purified by nickel affinity chromatography in a buffer containing 25 mM Tris, pH 8.0, 250 mM NaCl, 10% glycerol, and 5 mM imidazole and then subjected to a second round of purification with glutathione-Sepharose resin (GE Healthcare) in a buffer containing 25 mM Tris, pH 7.4, 150 mM NaCl, and 10% glycerol. The expression of GST and GST-OAR was induced using the procedure described for GST-Rtf1-His₆ and GST-Rtf1ΔOAR-His₆, and the proteins were purified from the clarified extract by affinity purification with glutathione-Sepharose resin (GE Healthcare) in a buffer containing 20 mM HEPES, pH 7.5, 150 mM NaCl, 10 mM MgCl₂, 10% glycerol, and 5 mM EDTA. The GST-bound proteins were used in the binding assay. MBP-tagged proteins were expressed in E. coli codonplus-RIPL cells by growth in LB medium containing 0.1 mM IPTG and 0.3 mM zinc chloride for 4 h at 37°C. The pellets were lysed in a homogenizer in a buffer containing 25 mM Tris-Cl, pH 7.4, 250 mM NaCl, and 10% glycerol, and the proteins were isolated by affinity purification of the clarified lysate with amylose resin (New England BioLabs).

In vitro protein interaction assays with purified proteins.

Four micrograms of purified recombinant Flag-Spt5-His₆/Spt4 was added to glutathione-Sepharose-bound, GST-tagged proteins in a binding buffer containing 25 mM Tris, pH 7.4, 150 mM NaCl, 10% glycerol, 5 mM EDTA, 2% NP-40, and 1 mM DTT, and the mixtures were incubated for 1 h at 4°C with gentle agitation. The glutathione-Sepharose beads were washed six times in binding buffer containing 250 mM NaCl, and the bound proteins were eluted from the beads by boiling for 5 min in 4× SDS sample buffer. Samples were then loaded onto 12% SDS–polyacrylamide gels. Spt5 and the GST-tagged proteins were detected by immunoblot analysis with antibodies against Spt5 and GST (A5800, 1:1,000; Molecular Probes). For the binding assay with MBP-tagged proteins, 4 μg of purified recombinant OAR-His₆ was incubated with 4 μg of MBP or MBP-CTR and magnetic nickel beads (Qiagen) for 1 h at 4°C in a binding buffer containing 25 mM Tris, pH 7.4, 150 mM NaCl, 10% glycerol, and 20 mM imidazole. After incubation, the beads were washed three times with buffer containing 25 mM Tris, pH 7.4, 250 mM NaCl, 10% glycerol, and 25 mM imidazole. Bound proteins were eluted from the beads with a buffer containing 500 mM imidazole and run on a 15% SDS–polyacrylamide gel. MBP-tagged proteins and OAR-His₆ were detected by immunoblotting with anti-MBP (E8030S, 1:10,000 dilution; New England BioLabs) and anti-His₆ (24-4710-01, 1:500 dilution; GE Healthcare) antibodies.

GST pulldown assays using yeast extracts.

Assays to test the interaction between recombinant GST-OAR protein and Spt5 derivatives, provided in the form of yeast extracts, were performed as described previously (40). Briefly, log-phase cultures (2 liters at an OD₆₀₀ of 0.6 to 1.0) of E. coli codonplus-RIL cells containing a plasmid encoding GST (pGEX-3X) and E. coli CodonPlus-RIPL cells containing a plasmid encoding GST-OAR (pMM25) were induced with 0.1 mM IPTG for 2 h 45 min and lysed in phosphate-buffered saline (PBS) buffer containing 1 mM EDTA and protease inhibitors using a homogenizer. Clarified lysates were incubated with 1 ml of bovine serum albumin (BSA)-blocked 50% glutathione-Sepharose resin (GE Healthcare) for 1 h at 4°C to purify GST and GST-OAR. The purified glutathione-bound proteins were incubated with 2 mg of clarified yeast extracts prepared by homogenization of strains KA181, KA183, and KA185 for 1.5 h in binding buffer (150 mM NaCl, 20 mM HEPES, pH 7.4, 10 mM MgCl₂, 5 mM EDTA, 10% glycerol, and protease inhibitors). SDS loading buffer was added to the washed beads. Samples were boiled for 5 min at 100°C and resolved on 8% and 15% SDS–polyacrylamide gels to analyze the presence of Spt5 and GST-tagged proteins, respectively, by immunoblotting.

RESULTS

The OAR of Rtf1 is important for the chromatin association and transcriptional functions of Paf1C.

Full association of Paf1C with active ORFs requires the Cdc73, Leo1, and Rtf1 subunits of the complex. In this study, we sought to explore the mechanism by which Rtf1 facilitates the recruitment of Paf1C to chromatin. In a previous analysis of internal deletion mutations in the S. cerevisiae RTF1 gene, we identified a region within Rtf1, termed the ORF association region (OAR), which is critical for the chromatin association of Paf1C (40). Together, the consecutive deletions that defined the OAR in this earlier study removed residues 201 to 395 of Rtf1. To better map the boundaries of the OAR, we examined the predicted secondary structure of this region, as well as the degree of amino acid conservation (Fig. 1A). Based on this analysis, we constructed a mutant with a single complete OAR deletion mutation that removes residues 230 to 390 of Rtf1. Confirming that the removal of these residues did not affect the stability of the protein, the levels of the new deletion derivative, Rtf1ΔOAR, and full-length Rtf1 were similar, as determined by immunoblotting for the triple-HA tag present on these proteins (Fig. 1B).

Fig 1.

The OAR is a highly conserved and a functionally important region of Rtf1. (A) Multiple sequence alignment of Rtf1 homologues developed using Clustal X 2.0 (65) and secondary structure prediction developed using jNET (66). The coloration indicates the percentage of identity among the residues of Rtf1 homologues. Darker shades of gray represent higher percentages of identity. Tubes and arrows denote α helices and β sheets, respectively. “§” symbols indicate the Plus3 residues (S. cerevisiae R251, R273, and K299). S. pombe, Schizosaccharomyces pombe; H. sapiens, Homo sapiens; M. musculus, Mus musculus; D. rerio, Danio rerio; D. melanogaster, Drosophila melanogaster; C. elegans, Caenorhabditis elegans. (B) Immunoblotting analyses were performed using indicated antibodies on transformants of an rtf1Δ strain (KY619) containing plasmids expressing HA₃-Rtf1 (pLS21-5), HA₃-Rtf1ΔOAR (pMM01), and Rtf1 (pLS20). An rtf1Δ strain (KY619) transformed with empty vector (pRS314) was used as the negative control. G6PDH and total histone H3 levels served as loading controls. (C and D) ChIP analysis with anti-HA beads was performed on strains used in the experiment whose results are shown in panel B to determine the localization of HA₃-tagged Rtf1 proteins at PYK1 (C) and PMA1 (D). (E) ChIP analysis was performed on the KY2413, KY1220, and KY2410 strains and the KY2414 strain transformed with pMM47 using anti-HA beads to determine the occupancy of HA₃-Paf1 over the indicated regions. For panels C to E, the average values of three biological replicates are shown, with the error bars representing the SEM. (F) Tenfold serial dilutions, ranging from 10⁸ cells/ml to 10⁴ cells/ml, of an rtf1Δ strain (KY619) transformed with an empty vector (pRS314) and the strains used for ChIP analysis in the experiment whose results are shown in panel B were spotted on SD-His-Trp or SC-Ura-Trp medium containing 50 μg/ml 6-AU and the appropriate control medium and incubated for 4 days at 30°C. (G) Immunoblotting analysis was performed using the indicated antibodies on transformants of an rtf1Δ strain expressing FLAG-tagged H2B (KY2125) and either Rtf1 (pLS21-5) or Rtf1ΔOAR (pMM01) from plasmids. An rtf1Δ strain expressing untagged H2B (KY2124) and Rtf1 from plasmid pLS21-5 served as a negative control for the FLAG antibody. An rtf1Δ strain expressing FLAG-tagged H2B (KY2125) and containing the empty vector pRS314 served as a negative control for H2B K123 ubiquitylation. G6PDH served as the loading control. Ub, ubiquitin.

Using this new deletion derivative, we assessed the impact of the OAR on chromatin association of Rtf1 at the 5′ and 3′ regions of two actively transcribed genes, PYK1 and PMA1, by performing chromatin immunoprecipitation (ChIP) assays. We observed that deletion of the OAR severely reduced the ability of Rtf1 to associate with these ORFs (Fig. 1C and D), indicating that residues 230 to 390 comprise the functional OAR of Rtf1. To test whether the absence of the OAR, like a complete deletion of RTF1 (18), diminished the association of Paf1C with chromatin, we performed ChIP analysis of the Paf1 subunit, using strains that express HA₃-tagged Paf1 and untagged full-length Rtf1, untagged Rtf1ΔOAR, or no form of Rtf1 (Fig. 1E). The absence of the OAR reduced the association of Paf1 with active ORFs to levels similar to those caused by complete deletion of RTF1. Therefore, consistent with our previous results (40), the OAR of Rtf1 is important for tethering not only the Rtf1 subunit but also other Paf1C subunits to actively transcribed chromatin.

Given the importance of the OAR for the chromatin association of Paf1C, we next tested whether a full deletion of the OAR caused mutant phenotypes indicative of transcriptional defects, such as the suppressor of Ty phenotype (Spt⁻ phenotype) and 6-AU sensitivity (51). To assess the Spt⁻ phenotype, we introduced plasmids expressing full-length Rtf1, Rtf1ΔOAR, or no form of Rtf1 into an rtf1Δ his4-912δ strain. Suppression of the transcriptional effects of the Ty δ-element insertion mutation within the HIS4 promoter of the his4-912δ allele is indicated by growth on medium lacking histidine (52). Whereas the pattern of transcription initiation at his4-912δ in RTF1 cells leads to an extended, nonfunctional HIS4 transcript and a His⁻ phenotype, the deletion of RTF1 restores transcription initiation at the normal start site of the HIS4 gene and growth on medium lacking histidine (Fig. 1F) (48). Similar to the effect of deleting RTF1 entirely, the absence of the OAR caused a strong Spt⁻ phenotype (Fig. 1F). Sensitivity to the uracil analogue 6-AU, which depletes certain nucleotide pools, is frequently used as an indicator of a defect in transcription elongation (53). At the 6-AU concentrations used in our assay, wild-type cells grew normally, but cells lacking Rtf1 or just the OAR of Rtf1 grew poorly (Fig. 1F). Thus, the removal of the OAR of Rtf1 caused strong transcription-related phenotypes that are similar to those caused by a complete loss of the protein (51).

An important function of Rtf1 is in facilitating the modification of histones during transcription elongation. To test the importance of the OAR in promoting histone H2B K123 monoubiquitylation and downstream histone methylation events, we performed immunoblotting analyses on extracts prepared from RTF1, rtf1ΔOAR, and rtf1Δ strains. The absence of the OAR caused a reduction in the global levels of H3 K4 Me3, H3 K4 Me2, and H3 K79 Me2/3 but not to the same degree as deletion of the entire RTF1 gene (Fig. 1B). (Note that the antibody used for the analysis of H3 K79 methylation detects both the dimethyl and trimethyl modification states). Consistent with the reduction in H3 K4 and K79 methylation, the absence of the OAR also led to reduced levels of H2B K123 ubiquitylation, as revealed by immunoblotting analysis of RTF1 and rtf1ΔOAR strains that express FLAG-tagged H2B as the only source of H2B (Fig. 1G). These results indicate that the OAR-mediated chromatin association of Paf1C is important for full levels of transcription-coupled histone modifications, in agreement with our previous observations (40).

The OAR is required for the interaction of Rtf1 with Spt5.

We hypothesized that interactions of Paf1C mediated by the OAR could be involved in tethering Paf1C to chromatin. To uncover these interactions, we performed one-step affinity purification of TAP-tagged full-length Rtf1 and a TAP-tagged Rtf1ΔOAR derivative, lacking amino acids 230 to 390, from yeast extracts (see Materials and Methods). This one-step affinity purification approach allowed us to isolate the Rtf1 proteins under native conditions, thereby preserving potentially weak or transient interactions. In parallel, untagged Rtf1 was subjected to the same purification scheme to identify nonspecific associations. Of the proteins identified by mass spectrometry analysis of our purified samples, the essential transcription elongation factor Spt5 was unique in its ability to interact with full-length Rtf1 but not with Rtf1 lacking the OAR (Fig. 2A). Immunoblotting analysis indicated that the loss of this interaction was not due to reduced expression or general instability of the Rtf1ΔOAR-TAP protein relative to that of the full-length Rtf1-TAP protein or to reduced Spt5 levels in cells lacking the OAR (Fig. 2B and E, lanes 1 and 2). Furthermore, the mass spectrometry results showed that the Rtf1ΔOAR protein retained its interactions with the four other Paf1C subunits, Paf1, Ctr9, Cdc73, and Leo1 (Fig. 2A), and an assessment of Paf1C composition by gel filtration chromatography revealed similar coelution profiles for all five Paf1C subunits in fractions obtained from RTF1 and rtf1ΔOAR extracts (data not shown). Together, these observations suggest that the OAR does not play a major role in governing overall Paf1C integrity.

To validate the mass spectrometry results, we performed immunoblotting analysis of the affinity-purified Rtf1 derivatives. This analysis showed that both full-length Rtf1 and Rtf1 lacking the OAR interacted with Paf1 (Fig. 2C), in accordance with the mass spectrometry and gel filtration data and our identification of a Paf1-interacting region at the C terminus of Rtf1, which is retained in the Rtf1ΔOAR protein (40). Interestingly, our mass spectrometry data showed that the Rtf1ΔOAR-TAP protein, like full-length Rtf1-TAP, interacted with Spt16, a result confirmed by immunoblotting with antibodies against Spt16 (Fig. 2C). Since deletion of the OAR greatly impaired the interaction between Rtf1 and chromatin, we asked whether the chromatin occupancy of Spt16 was reduced in the rtf1ΔOAR strain. Surprisingly, ChIP analysis revealed that the Spt16 levels over actively transcribed regions were not greatly altered in cells lacking the OAR (Fig. 2D). These results suggest that Paf1C interacts with some factors involved in transcription even when it is not tightly associated with chromatin. They also indicate that the chromatin occupancy of Spt16 is not strongly dependent on the chromatin occupancy of Paf1C and that the interaction between Spt16 and Rtf1ΔOAR is insufficient to recruit the latter to chromatin.

To uncover the mechanism of recruitment of Paf1C to chromatin through the Rtf1 OAR, we focused on the interaction between Rtf1 and Spt5. Both mass spectrometry analysis and immunoblotting analysis of the affinity-purified proteins demonstrated that deletion of the OAR disrupted the interaction between Rtf1 and Spt5 (Fig. 2A and C). Reciprocally, immunoprecipitation of Spt5 resulted in the coimmunoprecipitation of full-length Rtf1 but not Rtf1 lacking the OAR (Fig. 2E, lanes 5 and 6). Based on reports demonstrating that the CTR of Spt5 is important for the chromatin association of Paf1C (33, 34) and that recombinant Paf1C subunits can bind to CTR peptides in vitro (35), we hypothesized that the CTR of Spt5 could mediate the interaction of Spt5 with Rtf1. Using an antibody against an internal region of Spt5, we immunoprecipitated Spt5 or an Spt5 mutant protein lacking the CTR from yeast extracts and asked if HA-tagged full-length Rtf1 could coimmunoprecipitate with either Spt5 protein. Interestingly, full-length Rtf1 coimmunoprecipitated with full-length Spt5 but not with Spt5 lacking the CTR (Fig. 2E, lanes 5 and 7). To determine whether removal of the Spt5 CTR affected the cotranscriptional histone modification functions of Rtf1, we performed immunoblotting analysis of the spt5ΔCTR strain used in our studies. In agreement with previous results (33, 34), the absence of the Spt5 CTR caused a reduction in H3 K4 Me3 and H3 K79 Me2/3 levels (Fig. 2F). Together, these results demonstrate that the interaction of Rtf1 with Spt5 requires the OAR of Rtf1 and the CTR of Spt5 and that this interaction is important for the function of Rtf1.

Our protein interaction studies revealed a requirement for the Rtf1 OAR in mediating an interaction between Paf1C and Spt5. We were, therefore, interested in determining whether the OAR was important for the chromatin association of Spt5. Analysis of Spt5 occupancy at the constitutively active genes PYK1 and PMA1 by ChIP in strains expressing full-length Rtf1 or Rtf1ΔOAR showed that the levels of Spt5 associated with these genes were not greatly impacted by the lack of the OAR (Fig. 2G). This finding indicates that Spt5 acts upstream from Rtf1 in promoting Paf1C recruitment.

The OAR of Rtf1 interacts directly with the CTR of Spt5.

Our data are consistent with the possibility that a direct physical interaction between the Spt4-Spt5 complex and the Rtf1 OAR mediates the coupling of Paf1C to RNA Pol II. To determine whether the interaction between Rtf1 and Spt4-Spt5 is direct or indirect, we performed in vitro binding assays with bacterially expressed Spt4-Spt5 complex and Rtf1 derivatives. As in yeast cells (40), we have found bacterially expressed Rtf1 to be susceptible to proteolytic breakdown. To maximize the recovery of recombinant, intact wild-type Rtf1 and Rtf1ΔOAR, we designed expression constructs that encode doubly tagged forms of these proteins. The N-terminal GST tag and C-terminal His₆ tag were exploited in a two-step affinity purification strategy (see Materials and Methods) to enrich for the intact Rtf1 proteins. Subsequently, the GST-tagged-Rtf1 proteins were bound to glutathione-Sepharose and mixed with purified recombinant Spt4-Spt5 to test for binding. We found that full-length Rtf1 interacted with Spt4-Spt5 in vitro (Fig. 3A, lane 2), and removal of the OAR from Rtf1 diminished this interaction (Fig. 3A, lane 3). Together, the results of our in vitro and in vivo experiments indicate that the OAR is important for the direct interaction of Rtf1 with Spt4-Spt5. To determine whether the OAR is sufficient for the interaction with Spt4-Spt5, we performed an in vitro binding assay using a purified recombinant GST-OAR protein (amino acids 235 to 373 of Rtf1) and recombinant Spt4-Spt5. Interestingly, the OAR alone interacted strongly with Spt4-Spt5 (Fig. 3A, lane 4). These results show that the Rtf1 OAR is both necessary and sufficient for a direct interaction between Rtf1 and Spt4-Spt5.

Fig 3.

The OAR of Rtf1 interacts directly with the CTR of Spt5. (A) Recombinant GST (pGEX-3X), GST-Rtf1-His₆ (pAP21), GST-Rtf1ΔOAR-His₆ (pMM26), and GST-OAR (pMM25) proteins, bound to glutathione-Sepharose beads, were incubated with the same amount of purified recombinant Flag-Spt5-His₆/Spt4 (pGH258). Beads were washed, and bound samples were analyzed by immunoblotting for Spt5 or GST. The latter served as a control for the amount of GST-tagged derivatives used in the assay. The results shown are representative of three experiments. (B) In vitro binding assay of the OAR of Rtf1 with the CTR of Spt5 was performed by incubating equal amounts of recombinant purified His₆-OAR with MBP alone or MBP-CTR. The OAR was pulled down using magnetic nickel beads. The amount of MBP or MBP-CTR and OAR was determined by immunoblotting with anti-MBP and anti-His₆ antibodies. The results shown are representative of three experiments. (C) Extracts of rtf1Δ strains expressing wild-type Spt5 (KA181), Spt5 with a nonphosphorylatable CTR (KA183), or Spt5 with a phosphomimetic version of the CTR (KA185) were used for GST pulldown assays with bacterially purified GST-OAR (pMM25) bound to glutathione beads. A reaction mixture with GST bound to glutathione beads and extracts of an rtf1Δ strain expressing wild-type Spt5 (KA181) served as a negative control. The results shown are representative of two experiments.

Our coimmunoprecipitation experiments revealed a requirement for the CTR of Spt5 in the in vivo association of Spt5 with Rtf1 (Fig. 2E, lane 7). We therefore hypothesized that the CTR of Spt5 could be facilitating the recruitment of Paf1C by interacting directly with the OAR of Rtf1. To test this idea, we performed an in vitro binding assay to examine the binding of the isolated OAR to the Spt5 CTR. We purified His₆-tagged OAR and MBP-tagged Spt5 CTR proteins from bacterial expression strains, mixed the purified proteins, and used affinity chromatography to pull down the His₆-OAR protein. We then performed immunoblotting analysis to determine whether the MBP-CTR protein or MBP alone was retained on the beads after multiple washes. This analysis revealed that, relative to MBP alone, significantly greater amounts of MBP-CTR were bound to His₆-OAR (Fig. 3B, lanes 1 and 3), providing support for a direct interaction between the OAR of Rtf1 and the CTR of Spt5.

To address the importance of CTR phosphorylation in regulating the physical interaction between Spt5 and the OAR, we performed a GST pulldown assay using GST-OAR and extracts prepared from yeast cells that express either wild-type Spt5 or mutant derivatives of Spt5 in which the phosphorylated serines are replaced with alanines, as a nonphosphorylatable version (Spt5-S1-15A), or with aspartic acids, as a phosphomimetic version (Spt5-S1-15D) (35). In accordance with the results we obtained with recombinant proteins, the GST-OAR protein but not GST alone interacted with wild-type Spt5 provided in the yeast extract (Fig. 3C, lanes 1 and 2). A similar level of association was observed between the GST-OAR protein and the Spt5-S1-15D phosphomimetic derivative (Fig. 3C, lane 4). In contrast, the interaction between the GST-OAR protein and Spt5 was diminished by the replacement of the Bur1-phosphorylated serines (33) with alanines (Fig. 3C, lane 3). These results are consistent with the conclusion that phosphorylation of the Spt5 CTR stimulates its interaction with the Rtf1 OAR.

The Plus3 residues of the OAR are functionally important.

We next sought to identify functionally important residues within the OAR and focused on the three highly conserved, positively charged residues (R251, R273, and K299) that characterize the Plus3 domain (42) (Fig. 1A). To test their biological significance, we altered these Plus3 residues, either singly or in combination, to glutamic acids and then tested the ability of the resulting Rtf1 mutant proteins to confer an Spt⁻ phenotype. Two of the three single-residue charge swap substitutions, R251E and K299E, did not suppress the his4-912δ allele at 30°C (Fig. 4A). R273E was the only single-residue substitution that caused an Spt⁻ phenotype at 30°C, albeit weakly (Fig. 4A). However, the replacement of multiple Plus3 residues enhanced the Spt⁻ phenotype (Fig. 4A). Immunoblotting analysis indicated that the mutant Rtf1 protein levels were comparable to the wild-type Rtf1 levels at 30°C (Fig. 4B), suggesting that the enhanced mutant phenotypes in the double and triple mutants were not due to lower protein levels.

Fig 4.

The conserved Plus3 residues of the OAR are functionally important. (A and C) An rtf1Δ strain (KY619) was transformed with derivatives of HA₃-Rtf1 (pLS21-5) having the indicated amino acid substitutions in the OAR. Tenfold serial dilutions of these mutants, ranging from 10⁸ cells/ml to 10⁴ cells/ml, were spotted on SD-His-Trp plates to assess their Spt⁻ phenotype and on SC-Trp as a growth control and incubated for 4 days at 30°C (A) or 37°C (C). (B and D) Immunoblotting analysis with anti-HA antibody was performed with strains used in the experiments whose results are shown in panels A and C to determine Rtf1 protein levels in strains at 30°C (B) or 37°C (D). G6PDH levels were measured as a loading control. Quantification of the immunoblotting analysis revealed approximately 40% reductions in the levels of the Rtf1-R273E, Rtf1-R251E, R273E, and Rtf1-R251E, R273E, K299E derivatives at 37°C.

In addition to examining the Spt⁻ phenotype at 30°C, we also tested this phenotype at 37°C to determine whether the higher temperature exacerbated the Spt⁻ phenotype of the mutants. Of the three single-amino-acid substitutions, the R273E substitution caused the strongest Spt⁻ phenotype at 37°C (Fig. 4C). The double mutant (R251E, R273E) and the Plus3 triple mutant (R251E, R273E, K299E) also showed stronger Spt⁻ phenotypes at 37°C, although this may be partly due to reduced stability of the mutant proteins at 37°C (Fig. 4D). To test whether the Plus3 residues are important for the recruitment of Rtf1 to chromatin, we assessed the chromatin association levels of two of the Rtf1 mutant proteins, Rtf1-R273E and the Plus3 triple mutant, using ChIP analysis. Even when the cells were grown at 30°C, these Rtf1 mutant proteins exhibited significant chromatin association defects. Under these conditions, the levels of the mutant proteins were similar to the level of wild-type Rtf1 (Fig. 5A and B). Impairment of the ability of Rtf1 to associate with chromatin by deletion of the OAR significantly reduced the levels of Rtf1-dependent histone modifications (Fig. 1B). Likewise, replacement of the three Plus3 residues lowered the global levels of H3 K79 Me2/3 and H3 K4 Me3 (Fig. 5C, lane 5). Quantification of the immunoblots revealed a very modest reduction in the levels of H3 K79 Me2/3 in the Rtf1-R273E mutant strain but no effect on H3 K4 Me2 or H3 K4 Me3 levels (Fig. 5C, lane 4). We considered the possibility that the mutant phenotypes of the rtf1 mutant strains, including the reduction in chromatin-associated Rtf1 and Rtf1-dependent histone modifications, were due to a loss of Rtf1 from the nucleus. However, indirect immunofluorescence assays combined with confocal microscopy revealed that the Rtf1ΔOAR, Rtf1-R273E, and Rtf1-R251E, R273E, K299E proteins, like wild-type Rtf1, were localized to the nucleus (data not shown). Taken together, our results show that the Plus3 residues of the OAR are important for the chromatin association of Rtf1 and full levels of Rtf1-mediated histone modifications.

Fig 5.

The Plus3 residues of the OAR are important for the chromatin association of Rtf1. (A) ChIP analyses of transformants of an rtf1Δ strain (KY619) bearing plasmids expressing HA₃-Rtf1 (pLS21-5) or the indicated mutant proteins, which were expressed from pLS21-5 derivatives, were performed using anti-HA beads to determine the occupancy of the Rtf1 proteins over 5′ regions of PYK1 and PMA1 and a telomeric region of chromosome VI. The average of three biological replicates is shown, with the error bars depicting the SEM. (B) Immunoblotting analysis of the strains used in the experiment whose results are shown in panel A using anti-HA antibody to determine the Rtf1 protein levels, with anti-G6PDH antibody as the loading control. (C) Immunoblotting analyses were performed using indicated antibodies on transformants of an rtf1Δ (KY619) strain containing plasmids expressing HA₃-Rtf1 (pLS21-5), HA₃-Rtf1ΔOAR (pMM01), or the indicated mutant proteins, which were expressed from derivatives of pLS21-5. An rtf1Δ strain (KY619) transformed with empty vector (pRS314) was used as the negative control. G6PDH and total histone H3 levels served as loading controls.

The OAR is sufficient for chromatin association and mimics the chromatin association pattern of Rtf1.

Since mutations that delete the OAR or alter Plus3 residues within the OAR reduced the occupancy of Rtf1 on chromatin, we hypothesized that the OAR alone might be able to associate with active chromatin. To test this hypothesis, we constructed a plasmid that expresses the OAR (amino acids 230 to 390) as a fusion protein with a nuclear localization sequence (NLS) and a triple-HA tag and transformed this plasmid into an rtf1Δ strain. ChIP was performed to analyze the chromatin association levels of NLS-HA₃-OAR and full-length HA₃-Rtf1 at PMA1 and PYK1. Interestingly, even in the absence of the remainder of the Rtf1 protein, the OAR associated with the PYK1 and PMA1 genes and not with an untranscribed telomeric region (Fig. 6A). The levels of association were, however, lower than those for full-length Rtf1, which is likely due to the decreased cellular levels of the NLS-HA₃-OAR protein compared to the levels of full-length HA₃-Rtf1 (Fig. 6B). Since the OAR lacks the C-terminal region of Rtf1 that is required for the interaction with other members of Paf1C, its association with chromatin would not be expected to be facilitated by other members of the complex (40). To test this idea, we assessed the chromatin occupancy of Rtf1 and the OAR in the absence of Cdc73 or Paf1. Consistent with previous observations that Cdc73 is important for Paf1C recruitment (18, 35, 41) and that Paf1 is important for Rtf1 stability (18), Rtf1 showed lower chromatin occupancy in the absence of Cdc73 or Paf1 subunits (Fig. 6C and D). This reduced chromatin occupancy correlated with reduced levels of Rtf1 in both the cdc73Δ and paf1Δ strains (Fig. 6B, lanes 2, 4, and 6). Surprisingly, the chromatin occupancy of the OAR alone was increased in the absence of CDC73 (Fig. 6C and D). Immunoblotting analysis showed that the levels of the OAR protein, unlike the levels of full-length Rtf1, were unaltered in the cdc73Δ and paf1Δ strains compared to its level in the PAF1 CDC73 control strain (Fig. 6B, lanes 3, 5, and 7). The higher chromatin levels of the OAR in a cdc73Δ strain may reflect a competition between the OAR and Cdc73 for a common binding partner on chromatin, consistent with observations that both can interact with the Spt5 CTR (Fig. 3) (35).

Fig 6.

The OAR is sufficient for chromatin association. (A) ChIP analyses of transformants of an rtf1Δ strain (KY619) expressing NLS-HA₃-OAR (pMM44), HA₃-Rtf1 (pMM43), and untagged Rtf1 (pMM40) were performed using anti-HA beads to determine the occupancy of HA-tagged Rtf1 derivatives over 5′ regions of PYK1 and PMA1 and a telomeric region of chromosome VI. (B) Immunoblotting analysis of transformants of rtf1Δ (KY619), rtf1Δ cdc73Δ (KY2195), and rtf1Δ paf1Δ (KY1813) strains expressing NLS-HA₃-OAR (pMM44) or HA₃-Rtf1 (pMM43) was performed using antibody against the HA tag. The anti-G6PDH immunoblot served as a loading control. (C and D) ChIP analyses were performed on strains used in the experiment whose results are shown in panel B to determine the chromatin occupancy of NLS-HA₃-OAR (pMM44) and HA₃-Rtf1 (pMM43) over 5′ regions of PYK1 (C) and PMA1 (D). For panels A, C, and D, graphs represent the average of three biological replicates, with error bars indicating the SEM.

Although the OAR is sufficient to bind chromatin, we were interested in knowing whether its occupancy pattern mimicked that of full-length Rtf1. Therefore, we used ChIP analysis to ask if the OAR alone, like Rtf1 and other subunits of Paf1C (29, 54), dissociated from chromatin near the poly(A) site of PYK1 and PMA1. To better visualize the decrease in Rtf1 association, we normalized the occupancy levels to those at the 5′ ends of the genes. Interestingly, we observed that the OAR occupancy levels dropped significantly beyond the poly(A) site, similar to the pattern of Rtf1 occupancy (Fig. 7A to C). The reduction in the levels of the OAR beyond the poly(A) site could be explained by reduced chromatin levels of Pol II or Spt5 in cells expressing the OAR alone compared to the levels in cells expressing full-length Rtf1. However, ChIP analysis revealed that the levels of Spt5 and Pol II across PYK1 and PMA1 were similar in both strains (Fig. 7D to G). These results indicate that the lower levels of the OAR beyond the poly(A) site are not due to lower levels of Spt5 or Pol II at these sites specifically in OAR-expressing cells but are due to the dissociation of the OAR, like full-length Rtf1, from chromatin near the poly(A) site. Together, our data show that the isolated OAR is capable of associating with chromatin independently of the other members of Paf1C in a pattern similar to that of full-length Rtf1.

Fig 7.

The OAR mimics the chromatin association pattern of Rtf1. Occupancy of Rtf1 derivatives (B and C), Spt5 (D and E), and RNA Pol II (F and G) at PYK1 and PMA1 was measured by ChIP. Locations of PCR products are shown in panel A. Transformants of an rtf1Δ strain (KY619) expressing either HA₃-Rtf1 (pMM43) or NLS-HA₃-OAR (pMM44) were used for ChIP assays with anti-HA beads (B and C), Spt5 antisera (D and E) or Rpb3 antibody (F and G). Occupancy levels of HA₃-Rtf1 and NLS-HA₃-OAR were normalized to their occupancy levels at the 5′ regions of the genes, which were set to 1 (B and C). Reaction mixtures with no antibody (No Ab) served as negative controls for the nonspecific association of Spt5 and RNA Pol II with the beads (D to G). All graphs depict the average of three biological replicates with SEM.

OAR occupancy on chromatin is dependent on the CTR of Spt5 and Bur2.

Our findings so far show that the OAR is necessary for chromatin association of Paf1C, that the OAR can interact directly with the CTR of Spt5 in vitro, and that the isolated OAR behaves similarly to Paf1C in its ability to interact with chromatin both in terms of the localization pattern and the preference for active genes. Also, previous studies have shown that the recruitment of Paf1C is dependent on the CTR of Spt5 and on the Bur1-Bur2 protein kinase, which targets the CTR of Spt5 and the CTD of RNA Pol II (30, 33, 34, 36–38). Based on all these findings, we hypothesized that the association of the isolated OAR with chromatin would be dependent on the CTR of Spt5 and Bur2. To test this hypothesis, we analyzed the chromatin occupancy of the NLS-HA₃-OAR protein and full-length HA₃-Rtf1 in spt5ΔCTR and bur2-1 strains. Strikingly, just as for full-length Rtf1, the levels of chromatin association of the OAR at PYK1 and PMA1 were significantly reduced in the spt5ΔCTR and bur2-1 strains (Fig. 8A and B). Immunoblotting analysis showed that the OAR protein levels were slightly reduced in the spt5ΔCTR and bur2-1 strains but not to the same degree as the reduction in the ChIP signals (Fig. 8C, lanes 3, 5, and 7). Together, these results suggest that the OAR is responsive to the same regulatory factors as Paf1C and underscore the central role of this domain in the chromatin association of Paf1C.

Fig 8.

OAR recruitment is reduced in strains lacking the Spt5 CTR or with a mutation in BUR2. (A and B) ChIP analyses of the isolated OAR and full-length Rtf1 in transformants of rtf1Δ (KY619), rtf1Δ spt5ΔCTR (KA150), and rtf1Δ bur2-1 (AY777) strains expressing NLS-HA₃-OAR (pMM44) or HA₃-Rtf1 (pMM43) over 5′ regions of PYK1 (A) and PMA1 (B) using anti-HA beads. The average results from three biological replicates with SEM are shown. (C) Immunoblot analysis of HA₃-Rtf1 and NLS-HA₃-OAR levels of strains used for ChIP analysis in the experiments whose results are shown in panels A and B was performed using antibody against the HA tag. G6PDH levels served as the loading control.

DISCUSSION

Both the initial purification of Paf1 as an RNA Pol II-associated factor (3) and the subsequent discovery that the subunits of Paf1C colocalize with RNA Pol II on open reading frames (55, 56) argue that the physical coupling of Paf1C to RNA Pol II is likely to be important for directing the functions of Paf1C to active genes. Our previous deletion analysis identified a region of Rtf1, the OAR, as being important for the chromatin association of Paf1C (40); however, the nature of the interaction between the OAR and the RNA Pol II elongation machinery was unknown. In this study, we investigated the mechanism of recruitment of Paf1C through the Rtf1 OAR. Affinity purification of Rtf1 proteins containing or lacking the OAR revealed that the OAR was critical for the physical association of Spt5 with Paf1C, and coimmunoprecipitation analysis demonstrated that the interaction between Rtf1 and Spt5 required the CTR of Spt5. Our in vitro binding assays performed with purified recombinant versions of Rtf1 and Spt4-Spt5 provided evidence for a direct interaction between these elongation factors. In addition, these experiments suggested that the OAR of Rtf1 and the CTR of Spt5 mediate the direct physical interaction between Rtf1 and Spt5. Using ChIP studies, we found that the OAR can occupy chromatin independently of other Paf1C components and exhibit a localization pattern similar to that of full-length Rtf1. Moreover, the chromatin occupancy of the isolated OAR was significantly reduced in strains lacking the Spt5 CTR or the Bur2 cyclin component of the Bur1-Bur2 protein kinase, which phosphorylates the Spt5 CTR, as well as the CTD of RNA Pol II (33, 34, 38). Taken together, our results suggest that the OAR of Rtf1 plays a prominent role in mediating the recruitment of Paf1C to elongating RNA Pol II through an interaction with the CTR of Spt5.

In organisms ranging from bacteria to humans, Spt5 and its homologues have fundamental roles in regulating transcription elongation. In eukaryotes, Spt5 has been shown to promote RNA Pol II pausing and processivity (57). Spt5 consists of several domains, including an acidic N-terminal domain, a NusG N-terminal domain (NGN), multiple KOW (Kyprides, Ouzounis, Woese) domains, and the CTR (58). There have been several structural studies on the interaction between Spt5 and Spt4 (59, 60). The crystal structure of the archaeal homologues of a fusion protein of Spt4 and the NGN domain of Spt5 revealed that Spt4 interacts with the NGN domain of Spt5 (60). Both the NGN and the KOW motifs of Spt5 are involved in the interaction of Spt5 with RNA Pol II (61, 62). Curiously, sequence homology between the KOW motifs of Spt5 and the Plus3 domain of human Rtf1 has been noted, although the significance of this similarity is unclear (42). In this work, we provide evidence that the Rtf1 OAR is an interacting partner for the CTR of Spt5; however, our results do not rule out the possibility that other regions within Spt5 or Rtf1 also enhance their interaction.

Sequence alignments indicate that the OAR is the most highly conserved region of Rtf1 across eukaryotes (42). This conservation suggests a key contribution of the OAR to the function of Paf1C, which in higher eukaryotes has prominent roles in preventing cancer, promoting stem cell pluripotency, and ensuring proper cellular differentiation and organismal development (6). In this study, we found that the three most highly conserved basic residues of the OAR, which confer upon the OAR its alternative name, the Plus3 domain (42), are functionally important. While replacement of the Arg251, Arg273, and Lys299 residues individually had little phenotypic consequence, combined mutations conferred a stronger transcription-related mutant phenotype (Spt⁻). Moreover, the strengths of the growth phenotypes correlated with the observed chromatin association defects. The Rtf1-R273E mutant protein, which caused a weak but reproducible Spt⁻ phenotype, exhibited a moderate defect in chromatin association, whereas the Plus3 triple mutant protein (Rtf1-R251E, R273E, K299E) caused a stronger Spt⁻ phenotype and was severely impaired in its ability to associate with chromatin. A possible explanation for these results is that residues within the OAR mediate multivalent interactions with the Spt5 CTR and, potentially, other OAR-binding partners.

In a screen for conditional rtf1 mutations, we previously identified two point mutations that alter residues within the OAR and confer an Spt⁻ phenotype at 37°C (21). The three-dimensional structure of the human Rtf1 Plus3 domain, solved by NMR analysis, indicated that the corresponding human residues are buried within the protein, and replacement of these residues is likely to destabilize the core of the structure (42). In contrast, based on predictions made from the human structure, the Plus3 residues that we mutated here are surface exposed. Interestingly, in vitro studies performed with the human Plus3 domain suggested that this domain has the ability to bind single-stranded DNA (42), suggesting models in which this domain interacts with the RNA Pol II transcription bubble. Our combined data strongly suggest that the yeast Rtf1 OAR interacts with the Spt4-Spt5 complex, at least in part, through the CTR of Spt5. However, we have not ruled out the possibility that the OAR mediates the recruitment of Paf1C through additional mechanisms, such as through interactions with single-stranded DNA.

The mechanisms that direct the recruitment of Paf1C to RNA polymerase II are likely to be important for targeting Paf1C functions to active genes, and previous studies have implicated several different transcription elongation factors, as well as the RNA Pol II CTD, in this process (30–32, 35). Particularly relevant to our results, mutations that alter the Spt5 CTR or its kinase, Bur1-Bur2, were previously shown to impair the recruitment of Paf1C to chromatin (33–37). Interestingly, the chromatin association levels of Paf1 in cells expressing a mutant form of Spt5 that could not be phosphorylated on the CTR were higher than those in cells expressing Spt5 lacking the CTR (33). This observation suggests that, while the phosphorylation of the Spt5 CTR by Bur1-Bur2 enhances the affinity of the OAR for the CTR, other aspects of the CTR sequence and structure are also important for this interaction. Consistent with this idea, our in vitro binding assays revealed an interaction between recombinant OAR and CTR proteins, even though the latter is unlikely to be properly phosphorylated in E. coli.

Our observation that the chromatin association pattern of the OAR was similar to that of full-length Rtf1 indicates that features of Paf1C that control its pattern of chromatin association and dissociation are contained within the OAR. Interestingly, other components of Paf1C, including the Cdc73 C domain, which interacts with the RNA Pol II CTD in a phosphospecific manner, and Leo1, which binds to RNA, are also important for ensuring full levels of Paf1C recruitment (35, 39, 41). Together, these results argue for the existence of more than one attachment point between Paf1C and transcribing RNA Pol II. Unlike its Spt5 and RNA Pol II interaction partners, which continue on to the transcription termination site, Paf1C dissociates from coding regions near the poly(A) site, where 3′-end processing factors are recruited to the RNA Pol II machinery (29, 54). The dissociation of Paf1C at this site may be governed by changes in the phosphorylation patterns of the RNA Pol II CTD, the Spt5 CTR, or both proteins. In addition, the association of Leo1 with RNA may contribute to the release of Paf1C near the RNA cleavage site. Interestingly, a recent study showed that the recruitment of the 3′-end RNA processing factor, RNA cleavage factor I, requires the CTR of Spt5 (63). Together with our discovery of a direct physical interaction between the Rtf1 OAR and the Spt5 CTR, this observation, as well as the previous identification of numerous RNA processing factors in complex with Spt5 (64), raises the possibility that the Spt5 CTR, like the RNA Pol II CTD, acts as a platform for the recruitment and exchange of proteins that coordinate the synthesis and processing of RNA Pol II transcripts.