Abstract

The conserved eukaryotic Paf1 complex regulates RNA synthesis by RNA polymerase II at multiple levels, including transcript elongation, transcript termination, and chromatin modifications. To better understand the contributions of the Paf1 complex to transcriptional regulation, we generated mutations that alter conserved residues within the Rtf1 subunit of the Saccharomyces cerevisiae Paf1 complex. Importantly, single amino acid substitutions within a region of Rtf1 that is conserved from yeast to humans, which we termed the histone modification domain, resulted in the loss of histone H2B ubiquitylation and impaired histone H3 methylation. Phenotypic analysis of these mutations revealed additional defects in telomeric silencing, transcription elongation, and prevention of cryptic initiation. We also demonstrated that amino acid substitutions within the Rtf1 histone modification domain disrupt 3′-end formation of snoRNA transcripts and identify a previously uncharacterized regulatory role for the histone H2B K123 ubiquitylation mark in this process. Cumulatively, our results reveal functionally important residues in Rtf1, better define the roles of Rtf1 in transcription and histone modification, and provide strong genetic support for the participation of histone modification marks in the termination of noncoding RNAs.

PROPER regulation of genomic transcription is essential for cell survival. Many proteins contribute to the regulation of RNA polymerase II (Pol II) transcription by influencing one or more steps in RNA synthesis. One protein complex that regulates RNA Pol II transcription in eukaryotes is the conserved Paf1 complex. The RNA polymerase-associated factor 1 (Paf1) protein was initially identified as a RNA Pol II-interacting protein in S. cerevisiae (Shiet al. 1996; Wadeet al. 1996). Together with Paf1, four additional proteins comprise the Paf1 complex in budding yeast: Cdc73, Ctr9, Rtf1, and Leo1 (Shiet al. 1997; Mueller and Jaehning 2002; Squazzoet al. 2002). The Paf1 complex colocalizes with RNA Pol II at open reading frames (ORFs) from the transcriptional start site to the poly(A) site and regulates multiple processes coupled to transcription elongation (Kimet al. 2004; Mayeret al. 2010). These processes include the post-translational modification of histone proteins, the phosphorylation of RNA Pol II on its carboxy-terminal domain (CTD), and the correct generation of RNA 3′ ends (Stolinskiet al. 1997; Muelleret al. 2004; Penheiteret al. 2005; Sheldonet al. 2005; Nordicket al. 2008; Jaehning 2010).

The homologous Paf1 complexes in organisms ranging from plants to humans share similar functions, such as regulating elongation and histone modifications, recruiting chromatin modifying proteins, and targeting proteins involved in RNA 3′-end formation (Zhuet al. 2005a; Adelmanet al. 2006; Tenneyet al. 2006; Ohet al. 2008; Kim et al. 2009, 2010; Rozenblatt-Rosenet al. 2009; Jaehning 2010). The Paf1 complex in higher eukaryotes is important for organismal development, cellular differentiation, and prevention of diseases, such as cancer (Jaehning 2010). For example, the murine Paf1 complex promotes stem cell pluripotency by binding to promoters to activate key regulatory genes, such as OCT4 (Dinget al. 2009). Reduction of Rtf1 function causes disruptions in Notch signaling and developmental defects in flies and zebrafish, respectively (Tenneyet al. 2006; Akanumaet al. 2007). Additionally, defects in the Paf1 complex are associated with human cancers, which may be in part due to misregulation of the Wnt signaling pathway and alterations in c-myc and cyclin D expression (Mosimannet al. 2006;Chaudharyet al. 2007; Linet al. 2008).

In budding yeast, the Paf1 complex associates with RNA Pol II at active ORFs, where it physically interacts with other elongation factors and affects the transcription of numerous genes (Kroganet al. 2002; Pokholoket al. 2002; Porteret al. 2002; Penheiteret al. 2005; Mayeret al. 2010). Therefore, it is not surprising that cells lacking subunits of the Paf1 complex have phenotypes that are associated with defects in transcription elongation, such as sensitivity to the base analog 6-azauracil (6-AU) (Costa and Arndt 2000; Squazzoet al. 2002). By depleting intracellular nucleotide pools, 6-AU is thought to make RNA Pol II more dependent on accessory factors involved in transcription (Exinger and Lacroute 1992). As a result, cells that already have impaired transcription elongation, such as those with mutations in RNA Pol II or cells lacking Paf1 complex subunits, show sensitivity to 6-AU. Importantly, the Paf1 complex has also been shown to stimulate transcription elongation in vitro (Rondonet al. 2004; Pavriet al. 2006; Chenet al. 2009; Kimet al. 2010).

A less well-understood function of the Paf1 complex is its role in transcript termination and 3′-end processing. Specifically, Paf1, Ctr9, and Rtf1 regulate proper 3′-end formation of small nucleolar RNAs (snoRNAs), including SNR13 and SNR47, by participating within a larger regulatory pathway that includes the RNA binding proteins, Nrd1 and Nab3, and the helicase Sen1 (Steinmetzet al. 2001, 2006; Sheldonet al. 2005; Houseley and Tollervey 2009). The >70 snoRNAs in yeast help catalyze distinct modifications and cleavage reactions during ribosomal RNA (rRNA) processing and ribosome biogenesis and comprise one group of noncoding RNAs (ncRNAs) (Materaet al. 2007). In a range of eukaryotes, snoRNAs can be further processed into other functional small RNAs (Enderet al. 2008; Taftet al. 2009). Many classes of eukaryotic ncRNAs have essential cellular functions and their perturbation has been implicated in human cancers (Costa 2005; Goodrich and Kugel 2006; Amaralet al. 2008; Mallardoet al. 2008). Interestingly, the human Paf1 complex also regulates RNA 3′-end formation through physical association with mRNA processing factors (Rozenblatt-Rosenet al. 2009).

Components of the Paf1 complex are also required for the post-translational modification of histone proteins, which control access of DNA to the transcriptional machinery. Paf1 and Rtf1 promote the monoubiquitylation of histone H2B on lysine 123 (K123) by Rad6 and Bre1, the ubiquitin-conjugating enzyme and ubiquitin protein ligase, respectively (Nget al. 2003a; Woodet al. 2003). H2B K123 ubiquitylation is a prerequisite to achieve both di- and trimethylation of histone H3 K4 by the methyltransferase Set1 and histone H3 K79 methylation by the Dot1 enzyme (Sun and Allis 2002; Kroganet al. 2003; Ng et al. 2003a,b; Woodet al. 2003; Shahbazianet al. 2005). Histones modified in this way are associated with sites of active transcription and help recruit additional proteins involved in transcription (Santos-Rosaet al. 2002; Nget al. 2003b; Martinet al. 2006; Kim and Buratowski 2009; Wanget al. 2009). Paf1 and Ctr9 are also needed for the trimethylation of histone H3 at lysine 36 by Set2, another modification that is enriched at actively transcribed genes (Xiaoet al. 2003; Bannisteret al. 2005; Morillonet al. 2005; Pokholoket al. 2005; Raoet al. 2005; Chuet al. 2007). As in budding yeast, the human Paf1 complex plays roles in recruiting hRad6 and hBre1 to monoubiquitylate histone H2B, which in turn stimulates hSet1 to methylate H3 K4 and hDot1 to methylate H3 K79 (McGintyet al. 2008; Kimet al. 2009).

We previously identified a region within the Rtf1 subunit of the yeast Paf1 complex that is responsible for promoting H2B ubiquitylation and downstream histone modifications (Warneret al. 2007). Specifically, we showed that deletion of amino acids 62–109 or amino acids 112–152 prevented ubiquitylation of histone H2B. Therefore, we have termed the region defined by these combined deletions the histone modification domain (HMD) of Rtf1. Interestingly, the HMD is one of the most highly conserved sections of Rtf1, indicating its functional importance (Warneret al. 2007). In this study, we set out to better understand how the Paf1 complex contributes to transcriptional regulation by defining the specific residues of the Rtf1 HMD necessary for its functions. Using two different genetic approaches, we identified amino acids within the Rtf1 HMD that are important for histone H2B ubiquitylation and downstream histone H3 K4 and K79 methylation. Additionally, we found that these mutations within the HMD coding sequence confer common phenotypes associated with defects in transcription elongation, telomeric silencing, and prevention of cryptic initiation, which are likely to be consequences of impaired H2B ubiquitylation. By demonstrating that amino acid substitutions within the Rtf1 HMD cause defects in snoRNA transcript 3′-end formation, we were also able to provide strong genetic evidence that histone H2B ubiquitylation is involved in termination of transcription of yeast noncoding RNAs. Cumulatively, our results define functional roles for specific amino acids within the Rtf1 HMD that are conserved throughout eukaryotes.

MATERIALS AND METHODS

Yeast strains:

S. cerevisiae strains used in this study are isogenic to FY2, a GAL2+ derivative of S288C and are listed in supporting information, Table S1 (Winstonet al. 1995). Standard methods for yeast transformations, gene disruptions, genetic crosses, and tetrad dissections were performed as previously described (Ausubelet al. 1988; Roseet al. 1991). Constructs expressing HA-tagged wild-type and mutant Rtf1 derivatives, which are tagged at the amino terminus with three copies of the HA epitope, were integrated to replace the chromosomal RTF1 gene by a two-step gene replacement method (Rothstein 1991). As previously observed in plasmids encoding HA-Rtf1 and Rtf1 proteins (Warneret al. 2007), no phenotypic difference was detected between strains containing the integrated HA-RTF1 and untagged RTF1 alleles in any growth conditions tested. All integrations and gene disruptions were confirmed by PCR, Southern analysis, or sequencing. To construct a strain in which H2B-K123R is expressed from the chromosomal HTB1 locus, pEC1 (a URA3-marked plasmid; see Plasmid construction) was linearized with PflMI and transformed into a hta2-htb2Δ::KanMX4 ura3Δ0 strain (KY2027) and transformants were selected on SC −Ura medium. Transformants were subsequently plated on 5-FOA medium, and 5-FOA–resistant colonies that had retained htb1-K123R were identified by PCR amplification of the genomic htb1 locus followed by restriction digestion, as the mutation encoding the K123R substitution destroys a ScaI site and creates a MluI site. These strains were confirmed by their lack of histone H3 K4 trimethylation using Western analysis.

Media and growth assays:

Rich (YPD), minimal (SD), synthetic complete (SC), and 5-fluoroorotic acid (5-FOA) media were prepared as described (Roseet al. 1991). Where indicated, 6-azauracil was added to SC −Ura medium at a final concentration of 50 μg/ml. SC −his +gal medium contained 2% galactose. Cells were grown at 30° unless otherwise specified. For yeast dilution growth assays, strains were grown to saturation at 30° in appropriate media and washed twice. Cultures were 10-fold serially diluted in sterile water (starting at 1 × 108 cells/ml) and each dilution was spotted by pipetting onto the indicated media, followed by incubation at 30° for the specified number of days.

Plasmid construction:

Standard cloning techniques were used to construct plasmids (Ausubelet al. 1988). pLS20 and pLS21-5 are derivatives of pRS314 that express Rtf1 or triple HA-tagged Rtf1 (N-terminal tag), respectively (Sikorski and Hieter 1989; Stolinskiet al. 1997). pKR27 and pKR28 express HA-Rtf1Δ3 and HA-Rtf1Δ4, respectively, and were created as part of the Rtf1 internal deletion series (Warneret al. 2007). Rtf1∆3 lacks amino acids 62–109 and Rtf1∆4 lacks amino acids 112–152. Alanine substitution mutations were created by site-directed mutagenesis using the QuikChange system (Stratagene) and confirmed by DNA sequencing. By this method, pLS21-5 was used as the template plasmid to create pJB1, which encodes the HA-Rtf1-102-104A protein, and pJB2, which encodes the HA-Rtf1-108-110A protein. In both of these mutant proteins, three consecutive amino acids (e.g., 102–104) have been replaced by alanines. Plasmids pKB1 (HA-Rtf1-F80V, F123S), pDS2 (HA-Rtf1-E104K, E186D), and pDS3 (HA-Rtf1-E104G) are derived from pLS21-5 and were isolated in the RTF1 HMD mutational screen as described below. The double mutations in pKB1 and pDS2 were separated by subcloning to generate pRS314-derived plasmids pDS1 (HA-Rtf1-F123S) and pCD1 (HA-Rtf1-E104K). To construct pDS1, the EcoRI–BglII fragment of pKB1 was replaced with that of pLS21-5. For pCD1, the 2853bp XhoI–SpeI fragment of pDS5 (see below) was subcloned into pRS314. URA3-marked integrating plasmids were constructed to perform gene replacements at the endogenous RTF1 locus. Plasmid pMW9 (HA-Rtf1) contains the SalI–SpeI fragment of pLS21-5 inserted into pRS306. Plasmids pDS5 (HA-Rtf1-E104K) and pDS6 (HA-Rtf1-E104G) contain the 956bp EcoRI–BglII fragments of pDS2 and pDS3, respectively, inserted to replace wild-type RTF1 sequence in pMW9. Plasmid pDS4 was constructed by inserting the SalI–SpeI fragment of pDS1 into pRS306. To generate pEC1, a BamHI–SacII fragment containing HTA1-htb1 K123R was excised from pJH23KR and inserted into pRS306 cut with the same enzymes (Nget al. 2002). pADH1-HIS3-CYC1 and pADH1-SNR47(70)-HIS3-CYC1 were gifts from Dr. Jeffry Corden (Carrollet al. 2004).

Genetic screen for mutations in the RTF1 HMD coding region:

To create random substitution mutations in the Rtf1 HMD coding region, a plasmid-based gap repair strategy was used (Muhlradet al. 1992). A PCR product encompassing nucleotides −23 through +1008 of RTF1, relative to the start codon of Rtf1, was generated using Promega Taq DNA polymerase under low nucleotide conditions (50 μm each dNTP). The resulting PCR product was transformed into KY960 along with plasmid pKR27, which had been digested with BglII and SmaI to remove nucleotides 190–783 of the RTF1 coding region (regions 3–6 as defined by Warneret al. 2007). Transformants were selected on medium lacking tryptophan (SC −Trp). Strain KY960 is a rtf1∆ rkr1∆ double mutant strain, which is kept viable by the presence of a CEN/ARS URA3-marked RTF1 plasmid, pKA69 (Stolinskiet al. 1997). Trp+ transformants were plated on medium containing 5-FOA to counterselect plasmid pKA69. Transformants that were Trp+, but sensitive to 5-FOA, were further characterized by replica plating on media lacking histidine to assess the Spt phenotype. Plasmids were recovered from candidates that were 5-FOAS and weakly Spt, confirmed by retransformation, and analyzed by DNA sequencing.

Immunoblotting analysis of histone H3 methylation:

Cells were grown to approximately 3 × 107 cells/ml in YPD medium at 30°. Whole-cell extracts were prepared by glass bead lysis in RIPA buffer as described previously (Warneret al. 2007). To assess levels of histone modification and HA-Rtf1 expression, 50 μg of extract was resolved on a 15% SDS-polyacrylamide gel and transferred to a nitrocellulose membrane. Membranes were then probed with primary antibodies at the following concentrations: anti-H3 dimethylated (me2) and trimethylated (me3) at K79 (1:1000 dilution; Abcam ab2621), anti-H3 K4me3 (1:2000 dilution; Active Motif 39159), anti-H3 K4me2 (1:2000 dilution; Millipore 07-030), anti-H3 (1:50,000 dilution; described below), anti-HA (1:2500 dilution; Roche 11666606001), and anti-glucose-6-phosphate dehydrogenase (G6PDH; 1:50,000 dilution; Sigma A9521). Antibodies against total histone H3 were raised in rabbits (GenScript) against a peptide from the C terminus of S. cerevisiae histone H3 (peptide sequence: CKDIKLARRLRGERS). After primary antibody incubation, membranes were probed with sheep antimouse or donkey antirabbit horseradish peroxidase-conjugated secondary antibodies (1:5000 dilution; GE Healthcare), and immunoreactivity was detected through enhanced chemiluminescence (Perkin-Elmer) using a 440 CF digital imaging station (Kodak). Like Rtf1, Rtf1∆4 runs somewhat abnormally on SDS–PAGE, but this construct has been verified by sequencing (Stolinskiet al. 1997).

Immunoblot analysis of histone H2B K123 monoubiquitylation:

Levels of histone H2B K123 monoubiquitylation were assessed using a protocol provided by Brian Strahl (derived from Adamset al. 1997). hta1-htb1∆ hta2-htb2∆ strains with a covering HTA1/HTB1/URA3 plasmid were transformed with CEN/ARS/HIS3 plasmids encoding HTA1/HTB1 or HTA1/FLAG-HTB1. The HTA1/HTB1/URA3 plasmid in each strain was removed by counterselection on SC −Ura +5-FOA medium. Transformants were grown to an OD600 of ∼0.8–1.2. Cell pellets from 5 OD600 units of each culture were subjected to glass bead lysis in SUTEB buffer [1% sodium sodecyl sulfate (SDS), 8 m urea, 10 mm Tris pH 8.0, 10 mm EDTA pH 8.0, 0.01% bromophenol blue]. Extracts (15 μl) were resolved on 15% SDS-polyacrylamide gels and transferred to polyvinylidene fluoride (PVDF) membranes. Membranes were probed with antibodies against the FLAG epitope to detect histone H2B (1:1000 dilution; Sigma F3165) and ubiquitylated H2B (specificity of H2B-Ub upper band was confirmed with htb1-K123R extracts; data not shown). Anti-G6PDH was used to demonstrate equal loading of extracts. To assess expression of HA-tagged Rtf1 proteins, 5 μl of extract was run on a 15% SDS-polyacrylamide gel, transferred to PVDF membrane, and probed with a primary antibody specific for the HA epitope (1:2500 dilution; Roche). Secondary antibodies were used as described above.

Northern analyses:

Total RNA was isolated from cells grown in YPD at 30° to a density of 1–2 × 107 cells/ml and subjected to Northern analysis with random-prime–labeled, PCR-amplified DNA probes as described previously (Swansonet al. 1991). The probes used in these studies were generated against the following regions, with regions described being relative to the translation start site: HIS3 (+5 to +654), SCR1 (−242 to +283), SNR47-YDR042C (−325 of YDR042C to −33 of YDR042C), and SNR13-TRS31 (−231 of TRS31 to +449 of TRS31). SNR47-YDR042C and SNR13-TRS31 probes were made using [α32P]-dATP and [α32P]-dTTP; HIS3 and SCR1 probes were made using [α32P]-dATP. Signals were quantified through the use of a phosphorimager and ImageQuant software or the ImageJ software. In SNR47 and SNR13 Northern blots, relative signal from the wild-type control strain was set equal to one, and gene-specific signals relative to loading control (SCR1) signals were averaged for at least three independent sample preparations. Error bars represent one standard deviation from the mean.

RESULTS

Generation of mutations that alter the identity of conserved amino acids within the Rtf1 HMD:

To elucidate the functions of the Rtf1 HMD as well as the interdependencies of these functions, we used two different strategies to generate specific mutations in the HMD coding region. First, to target highly conserved and potentially surface-exposed amino acids in the HMD, we used site-directed mutagenesis to create three alanine-scanning mutations, termed rtf1-102-104A, rtf1-108-110A, and rtf1-120-121A. Eight invariant residues reside in a 33-amino-acid span between residues 88 and 120, which localizes within the genetically defined HMD, composed of amino acids 62–152 (Figure 1A; Warneret al. 2007). Each of the alanine-scanning mutations changes at least one invariant and one neighboring residue to alanines, resulting in the neutralization of a charged patch. These particular residues were chosen because they flank the border between the two regions of Rtf1, previously termed regions 3 and 4, which were shown by deletion analysis to be essential for Rtf1-dependent histone modifications (Figure 1A). Because subsequent phenotypic analysis of the rtf1-120-121A mutation showed it was not impaired in HMD functions (data not shown), we focused on studying the rtf1-102-104A and rtf1-108-110A mutations in greater detail.

Figure 1.—

Isolation of mutations that alter the Rtf1 HMD.

(A) Alignment of Rtf1 protein sequences from a range of eukaryotic species generated using the Clustal X algorithm (Larkinet al. 2007). Notations above the amino acids indicate the degree of conservation with * denoting invariant residues. S. cerevisiae Rtf1 mutants were generated through targeted substitutions (underlined) and unbiased genetic screening (arrows). The F80V substitution (open arrowhead) was obtained in combination with F123S. Subcloning of these mutations revealed that F123S is primarily responsible for the phenotypes of this double mutant. (B) Dilution growth assays of HA-RTF1 RKR1 (KY2076), HA-RTF1 rkr1∆ (KY2075), HA-rtf1∆3 (KY2032), rkr1∆ HA-rtf1∆3 (KY2029), HA-rtf1∆4 (KY2033), rkr1∆ HA-rtf1∆4 (KY2030), HA-rtf1-F123S (KY1531), rkr1∆ HA-rtf1-F123S (KY1957), HA-rtf1-F80V,F123S (KY2034), rkr1∆ HA-rtf1F80V,F123S (KY2031), HA-rtf1-E104K (KY1532), rkr1∆ HA-rtf1-E104K (KY1961), HA-rtf1-E104G (KY1533), and rkr1∆ HA-rtf1-E104G (KY1964) cells. Ten-fold serial dilutions were spotted on SD complete (left) and SC complete (right) media and incubated at 30° for 3 days.

In addition to the targeted substitutions, we also conducted an unbiased genetic screen to identify functionally important residues in the Rtf1 HMD. Specifically, we exploited our previous observation that an rtf1∆ mutation is synthetically lethal with a deletion of RKR1, a gene encoding a ubiquitin ligase (Braunet al. 2007). Although the basis for this genetic interaction remains to be elucidated, the loss of Rtf1-dependent histone modifications, especially H2B monoubiquitylation, is at least partially responsible for the synthetic interaction (Braunet al. 2007). Thus deletions of regions 3 and 4 in Rtf1 cause growth defects on SD medium when combined with a rkr1∆ mutation (Figure 1B). We therefore decided to utilize the poor growth of rkr1∆ rtf1 double mutants as a basis for a genetic screen to identify amino acids important for the histone modification functions of Rtf1. Briefly, we used error-prone PCR and a gap repair strategy to introduce mutations in the HMD coding region of a plasmid-encoded RTF1 gene and screened among the transformants for those that had impaired growth in a rkr1∆ genetic background (see materials and methods for details). rtf1 null mutations were ruled out through a secondary screen, which was based on the ability of the mutations to suppress the effects of the Ty long terminal repeat (δ element) integrated at the his4-912δ allele (i.e., the Spt or suppression of Ty phenotype; Winston and Sudarsanam 1998). A complete RTF1 deletion causes a strong Spt phenotype in a his4-912δ strain, whereas the effects of partial deletions within the HMD on this phenotype are relatively mild (Warneret al. 2007). Candidate plasmids that caused a mild Spt phenotype were isolated, confirmed by retransformation, and sequenced. Through this approach, and subcloning analysis to separate some multiple mutations, we pursued four mutations that altered amino acids in the Rtf1 HMD and caused growth defects in rkr1∆ cells: rtf1-E104K, rtf1-E104G, rtf1-F80V,F123S, or rtf1-F123S (Figure 1B). It is interesting to note that the F123S substitution was isolated in two independent double mutant clones, and after separation of the mutations, was shown to be primarily responsible for the observed phenotypes. Nonetheless, in several of our experiments we have included the F80V, F123S double mutant, because the F80V substitution appears to have a modulating effect on phenotypic strength (see Figures 1B and 2B for examples). Therefore, targeted mutation and unbiased genetic screening provided a set of Rtf1 HMD mutations in highly conserved residues that we have subsequently found to be functionally important.

Mutations in RTF1 cause phenotypes associated with transcriptional defects:

Having generated a new set of mutations that alter residues within the Rtf1 HMD, we tested how these mutations impact the myriad of functions of Rtf1 in transcription. One way to assess the ability of rtf1 mutations to promote transcription elongation is testing for sensitivity to the base analog 6-azauracil. Wild-type cells can grow in media containing 6-AU, but cells with impaired transcription elongation, such as those lacking RTF1, show sensitivity to 6-AU (Figure 2A). No 6-AU sensitivity was observed in control cells expressing HA-RTF1 or RTF1 (Figure 2; data not shown). Yet cells containing the rtf1-102-104A, rtf1-108-110A, rtf1-E104K, rtf1-E104G, rtf1-F80V,F123S, or rtf1-F123S mutations show a sensitivity to 6-AU similar to that of rtf1∆3 and rtf1∆4 cells (Figure 2, A and B). These results suggest that these conserved residues within the Rtf1 HMD are required to properly promote transcription elongation.

Figure 2.—

Amino acid substitutions in the Rtf1 HMD cause phenotypes associated with defects in transcription.

(A) Ten-fold serial dilutions of RTF1 (KY1010), rtf1Δ (KY619), HA-rtf1Δ3 (KY1506), HA-rtf1Δ4 (KY1508), HA-rtf1-102-104A (KY1510), and HA-rtf1-108-110A (KY1511) cells were spotted on SC −Ura medium containing 50 μg/ml 6-AU to assess 6-AU sensitivity (left), SD −His medium to examine the Spt phenotype (middle), or SC complete medium (right) as a control for growth. (B) Ten-fold serial dilutions of RTF1 (KY1021), rtf1Δ (KY656), HA-rtf1Δ3 (KY1507), HA-rtf1-E104K (KY1532), HA-rtf1-E104G (KY1533), HA-rtf1Δ4 (KY1508), HA-rtf1-F80V-F123S (KY1512), and HA-rtf1-F123S (KY1531) cells were spotted on SC −Ura medium containing 50 μg/ml 6-AU to assess 6-AU sensitivity (left) or SC −Ura medium as a control for growth (right). (C) Ten-fold serial dilutions of RTF1 (KY292), rtf1Δ (KY522), HA-rtf1Δ3 (KY1507), HA-rtf1-E104K (KY1514), HA-rtf1-E104G (KY1515), HA-rtf1Δ4 (KY1508), HA-rtf1-F80V,F123S (KY1512), and HA-rtf1-F123S (KY1513) cells were spotted on SD −His medium to examine the Spt phenotype (left) or SD complete medium (right) as a control for growth. Plates were incubated at 30° for 4 days.

After demonstrating that these Rtf1 residues are necessary for 6-AU resistance, a phenotype that is often associated with transcription elongation, we then asked whether they were involved in other aspects of transcriptional regulation. We tested whether the rtf1 mutations cause an Spt phenotype in cells containing the his4-912δ allele. Cells with the his4-912δ allele can express HIS4 if they contain mutations in trans-acting genes involved in transcription or chromatin regulation, such as those encoding histones, TATA-binding protein (TBP), or subunits of the Paf1 complex that restore utilization of the proper HIS4 transcriptional start site (Winston and Sudarsanam 1998). Therefore, growth on media lacking histidine indicates an Spt phenotype. Whereas his4-912δ cells containing HA-Rtf1 or Rtf1 do not have an Spt phenotype, cells containing the his4-912δ allele and lacking RTF1 or regions 3 or 4 of the HMD grow in the absence of histidine, as has been previously described (Figure 2A; data not shown; Warneret al. 2007). However, rtf1-102-104A and rtf1-108-110A cells do not show an Spt phenotype, indicating that alanine substitutions of these residues in Rtf1 do not impact transcription start-site utilization at his4-912δ. Interestingly, whereas rtf1-F80V,F123S or rtf1-F123S cells do not grow well in the absence of histidine, cells containing rtf1-E104K or rtf1-E104G mutations do cause an Spt phenotype (Figure 2C). These phenotypic differences indicate that, among the conserved residues tested in the Rtf1 HMD, the E104 residue is most important for regulating transcriptional start-site selection as measured by the Spt phenotype. Since rtf1-102-104A cells did not share this phenotype, it is likely that an alanine substitution at E104 is a less disruptive substitution of this residue than a lysine or a glycine, with respect to the Spt phenotype.

Transcriptional silencing of genes near telomeres is disrupted by deletions within the Rtf1 HMD (Nget al. 2003a; Warneret al. 2007). To determine whether substitutions of specific amino acids within the Rtf1 HMD affect telomeric silencing, we integrated an ectopic copy of URA3 next to the telomere on the right arm of chromosome V (TEL-VR::URA3) in strains containing a ura3-52 mutation at the endogenous URA3 locus (Warneret al. 2007). In these cells, impaired telomeric silencing will also derepress expression of the ectopic URA3, and such cells will not be able to grow on media containing 5-FOA. Unlike cells expressing HA-RTF1 or RTF1, which grow on media containing 5-FOA, the rtf1∆ strain and all the rtf1 HMD mutants fail to grow on 5-FOA (Figure 3, A and C; data not shown). These data indicate that telomeric silencing is especially sensitive to the functions of the Rtf1 HMD.

Figure 3.—

Amino acid substitutions in the Rtf1 HMD cause phenotypes related to defects in chromatin structure.

(A) To analyze telomeric silencing in strains expressing Rtf1 substitution mutants, 10-fold serial dilutions of RTF1 (KY1226), rtf1Δ (KY1227), HA-rtf1Δ3 (KY2035), HA-rtf1Δ4 (KY2036), HA-rtf1-102-104A (KY2037), and HA-rtf1-108-110A (KY2038) cells were spotted on SC −Ura medium containing 5-FOA to assess telomeric silencing defects (left) or SC complete medium (right) as a control for growth. Plates were incubated at 30° for 3 days. (B) To measure aberrant initiation from a cryptic promoter within the FLO8 open reading frame, 10-fold serial dilutions of RTF1 (KY1116), rtf1Δ (KY1207), HA-rtf1Δ3 (KY1208), HA-rtf1Δ4 (KY1209), HA-rtf1-102-104A (KY1210), and HA-rtf1-108-110A (KY1211) cells were spotted on SC −His +galactose medium to induce and monitor expression of the GAL1pr-FLO8-HIS3 reporter (left) and SC complete medium (right) as a control for growth. Plates were incubated at 30° for 7 days. (C) Ten-fold serial dilutions of RTF1 (KY1491), rtf1Δ (KY1370), HA-rtf1Δ3 (KY1230), HA-rtf1-E104K (KY1526), HA-rtf1-E104G (KY1527), HA-rtf1Δ4 (KY1231), HA-rtf1 F80V-F123S (KY1234), and HA-rtf1-F123S (KY1525) were spotted on SC −Ura medium containing 5-FOA to assess telomeric silencing defects (left), SC −His +galactose medium to induce and monitor expression of the GAL1pr-FLO8-HIS3 reporter (middle), and SC complete medium as a control for growth (right). Plates were incubated at 30° for 6 days.

Members of the Paf1 complex are also involved in suppressing transcription initiation from cryptic promoters within ORFs (Chuet al. 2007; Cheunget al. 2008). We therefore sought to determine whether Rtf1 and conserved residues in the Rtf1 HMD play a role in this process. FLO8 has an internal cryptic promoter, and expression from that promoter can be monitored using a reporter strain in which the HIS3 gene is integrated downstream of the cryptic FLO8 promoter and the native FLO8 promoter has been replaced with the inducible GAL1 promoter (Cheunget al. 2008). Growth of these strains on galactose-containing medium that lacks histidine is a sensitive indication of initiation at the FLO8 cryptic internal promoter. Cells containing HA-Rtf1 or Rtf1 do not demonstrate cryptic initiation (Figure 3; data not shown). However, using this reporter, we found that the rtf1-102-104A, rtf1-108-110A, rtf1-E104K, rtf1-E104G, rtf1-F80V,F123S, and rtf1-F123S mutations result in internal initiation at FLO8 at a level equivalent to that observed when RTF1 is completely deleted (Figure 3, B and C). Surprisingly, this effect is more severe than that caused by the absence of Rtf1 regions 3 and 4, suggesting that changing the identity of conserved residues in the HMD is not functionally equivalent to completely removing them from the protein. Cumulatively, we have demonstrated that conserved residues within the Rtf1 HMD can contribute to various functions of Rtf1 in transcription such as start-site selection, prevention of cryptic initiation, telomeric silencing, and transcription elongation.

Histone H3 methylation marks associated with active transcription are altered in cells with rtf1 mutations:

Central to regulation of transcriptional processes is the control of post-translational modifications of histone proteins. Rtf1 is required for proper histone H3 K4 and K79 methylation (Nget al. 2002, 2003a,b; Kroganet al. 2003). Given the transcriptional phenotypes associated with our RTF1 mutations and the locations of the affected residues within the region of Rtf1 responsible for histone modifications, we examined the effects of these mutations on global Rtf1-dependent histone modifications. Using antibodies specific to H3 K4 trimethylation, H3 K4 dimethylation, and H3 K79 di/trimethylation, immunoblot analysis was performed using whole-cell extracts prepared from wild-type cells and the various rtf1 mutant strains (Figure 4). All of the rtf1 mutant strains are defective in histone H3 K4 trimethylation, whereas the rtf1-102-104A, rtf1-108-110A, and rtf1-F123S mutants still retain significant levels of H3 K4 dimethylation (Figure 4). Therefore, while all tested substitutions contribute to achieving full levels of H3 K4 trimethylation, the rtf1-E104K and rtf1-E104G mutants are unique in their inability to promote H3 K4 dimethylation. Cells expressing rtf1-E104K or rtf1-E104G also show the strongest defects in H3 K79 di/trimethylation (Figure 4). To a lesser extent, cells with rtf1-102-104A, rtf1-108-110A, and rtf1-F123S are also impaired in H3 K79 di/trimethylation (Figure 4). The alterations in histone modification levels are not due to changes in expression levels of the rtf1 mutant proteins, as they are expressed at wild-type levels (Figure 4). These results demonstrate that key residues in Rtf1 are critical for achieving proper levels of histone H3 K4 and K79 methylation, conserved post-translational modifications that mark active genes throughout eukaryotes.

Figure 4.—

Immunoblot analysis reveals amino acids in Rtf1 important for histone H3 methylation.

Immunoblot analysis of Rtf1-dependent histone H3 modifications was performed on strains containing Rtf1 internal deletions and amino acid substitutions. Whole cell extracts from RTF1 (KY1491), HA-RTF1 (KY2088), rtf1Δ (KY1370), HA-rtf1Δ3 (KY1230), HA-rtf1Δ4 (KY1231), HA-rtf1-102-104A (KY1232), HA-rtf1-108-110A (KY1233), HA-rtf1-F123S (KY1525), HA-rtf1-E104K (KY1526), HA-rtf1-E104G (KY1527), set1∆ (KY907), and dot1∆ (KY903) strains were probed with antibodies specific for histone H3 K4me3, H3 K4me2, H3 K79me2/3 (both methylation states are detected by this antibody), and total histone H3 protein. A nonspecific band detected with the H3 K79me2/3 antibody is noted by an asterisk. Immunoblots were probed with antibodies specific for the HA epitope tag to confirm Rtf1 expression and for G6PDH to serve as a loading control. Note that Rtf1 is sensitive to proteolytic attack under certain extraction methods, such as that used in this experiment, and occasionally runs as two bands.

Amino acid substitutions in the Rtf1 HMD impair histone H2B K123 ubiquitylation:

The establishment of histone H3 K4 and K79 methylation depends on the proper monoubiquitylation of histone H2B at K123 (Briggset al. 2002; Sun and Allis 2002). H2B K123 ubiquitylation is absent in cells lacking RTF1, and we previously demonstrated that the Rtf1 HMD is essential for this modification (Warneret al. 2007). Given that substitutions of conserved residues in the HMD caused alterations in H3 K4 and K79 methylation, it is possible that they also impact H2B K123 ubiquitylation. Alternatively, they could separate the histone methylation and ubiquitylation functions of Rtf1. To test these ideas, we generated hta1-htb1∆ hta2-htb2∆ cells that express a FLAG-tagged version of H2B from a plasmid and used immunoblot analysis to detect a mobility shift when FLAG-H2B is ubiquitylated (see materials and methods). Using this approach, we measured H2B ubiquitylation in wild-type and rtf1 mutant backgrounds by performing immunoblot analysis on extracts prepared under denaturing conditions that preserve the modification. As controls for the FLAG signal, hta1-htb1∆ hta2-htb2∆ cells containing a plasmid with untagged HTA1-HTB1 were used. As expected, immunoblot analysis of extracts prepared from wild-type cells expressing either HA-Rtf1 or untagged Rtf1 show H2B monoubiquitylation, whereas rtf1∆, rtf1∆3, and rtf1∆4 extracts lack detectable levels of H2B ubiquitylation (Figure 5, A and B). Interestingly, rtf1-102-104A, rtf1-108-110A, rtf1-E104K, rtf1-E104G, rtf1-F80V,F123S, and rtf1-F123S cells all lack H2B ubiquitylation (Figure 5, A and B). Therefore, the substitution of any of these conserved residues of Rtf1 prevents the ability of Rtf1 to properly promote H2B ubiquitylation and suggests that the primary function of Rtf1 in histone modification is most likely at the level of this mark. Further, our results provide a useful set of alleles to better study the central functions of both Rtf1 and H2B monoubiquitylation in transcription.

Figure 5.—

Histone H2B K123 ubiquitylation requires E104 and F123 residues within Rtf1.

Immunoblot analysis of histone H2B K123 ubiquitylation in strains expressing alanine scanning mutations (A) or screen-derived point mutations (B). For A, the following strains were used: HA-RTF1 (KY1459), rtf1Δ (KY982), HA-rtf1-102-104A (KY2039), and HA-rtf1-108-110A (KY2040). For B, the following strains were used: RTF1 (KY943), HA-RTF1 (KY1459), rtf1Δ (KY982), HA-rtf1Δ3 (KY1216), HA-rtf1Δ4 (KY1217), HA-rtf1-F80V-F123S (KY1462), HA-rtf1-F123S (KY1534), HA-rtf1-E104K (KY1535), and HA-rtf1-E104G (KY1536). Cells were transformed with plasmids expressing either untagged or FLAG-tagged histone H2B, and extracts from these strains were probed with an antibody specific for the FLAG epitope. The bands corresponding to unmodified H2B (FLAG-H2B) and ubiquitylated H2B (FLAG-H2Bub) are indicated. Extracts were probed with antibodies specific for the HA epitope to confirm Rtf1 expression and for G6PDH, which serves as a loading control.

Conserved residues in Rtf1 contribute to proper snoRNA synthesis and implicate a role for H2B ubiquitylation in the regulation of 3′-end formation of these ncRNAs:

In addition to the many important functions of Rtf1 during general transcription processes, Rtf1 has also been shown to be involved in the termination of noncoding RNA transcripts in budding yeast. Specifically, Rtf1 participates in regulating the 3′-end formation of small nucleolar RNAs, including SNR13 and SNR47 (Sheldonet al. 2005). Although Rtf1 has been implicated in snoRNA termination by the presence of extended read-through snoRNA transcripts in rtf1∆ cells, it is not known whether this involves the H2B K123 ubiquitylation function of Rtf1. Therefore, we used cells with Rtf1 HMD mutations that have impaired H2B ubiquitylation to better understand the snoRNA termination pathway.

To explore snoRNA 3′-end formation in these cells, an established plasmid reporter system for snoRNA termination was used. This plasmid reporter contains the HIS3 gene positioned downstream of a 70-bp SNR47 3′-end formation element that is sufficient to direct termination of the primary snoRNA transcript (Carrollet al. 2004). Wild-type cells can properly terminate transcription within the SNR47 3′ sequence and prevent HIS3 expression. However, cells with improper termination at the SNR47 locus will express a read-through transcript containing HIS3 that allows for strong growth in media lacking histidine (Carrollet al. 2004; Sheldonet al. 2005). To test whether substitutions in the Rtf1 HMD cause a snoRNA termination defect, rtf1∆ his3∆200 cells were first transformed with either a SNR47 reporter plasmid pADH1-SNR47(70)-HIS3-CYC1 or a control plasmid pADH1-HIS3-CYC1, which lacks the SNR47 3′ sequence and were then subsequently transformed with pRS314 or pRS314-derived plasmids expressing HA-tagged Rtf1 or HA-tagged Rtf1 mutant proteins. As expected, all cells containing the pADH1-HIS3-CYC1 control plasmid grew in media lacking histidine (Figure 6A, right). Compared to wild-type cells, rtf1∆ cells containing the pADH1-SNR47(70)-HIS3-CYC1 termination reporter plasmid grew more robustly on plates lacking histidine (Figure 6A, left). This result suggests that rtf1∆ cells express more HIS3-containing read-through transcripts, which arise from impaired RNA 3′-end formation within the SNR47 3′ sequence. Interestingly, examination of growth of the rtf1-102-104A, rtf1-E104K, rtf1-108-110A, or rtf1-F80V,F123S mutants containing the reporter plasmid revealed that these cells also have defects in snoRNA 3′-end formation (Figure 6A, left).

Figure 6.—

Amino acid substitutions in Rtf1 impair transcription termination in a SNR47 reporter plasmid.

(A) rtf1∆ cells (KY2041) transformed with either pADH1-SNR47(70)-HIS3-CYC1 (left) or pADH1-HIS3-CYC1 (right) were subsequently transformed with pRS314 or pRS314-derived plasmids containing HA-tagged RTF1 (pLS21-5), rtf1-102-104A (pJB1), rtf1-E104K (pCD1), rtf1-108-110A (pJB2), or rtf1-F80V,F123S (pKB1) as indicated. Ten-fold serial dilutions of these cells were spotted on SC −Leu −Trp −His and SC −Leu −Trp media and incubated for 3 days at 30°. (B) Northern blot analysis was used to examine HIS3 transcripts. Analysis was done using 20 μg of total RNA from the following his3∆200 strains: RTF1 (KY2090), RTF1 (KY2042) transformed with pADH1-HIS3-CYC1, rtf1∆ (KY2041) containing pADH1-SNR47(70)-HIS3-CYC1 subsequently transformed with pRS314 or pRS314-derived plasmids containing HA-tagged RTF1 (pLS21-5), rtf1-102-104A (pJB1), or rtf1-E104K (pCD1) as indicated by listed genotype.

While our results indicate that specific residues within the Rtf1 HMD are required to promote proper 3′-end formation at SNR47, it remains a possibility that the results obtained through the use of this reporter plasmid could be due to events of cryptic initiation arising from within the reporter. No evidence for cryptic initiation has been found at HIS3 in previous genomic studies (Lickwaret al. 2009). However, to be certain that cryptic initiation was not occurring in the reporter, we examined HIS3 transcripts arising from within the reporter plasmids by Northern blot analysis. As expected, no HIS3 transcript was observed in a wild-type strain carrying the his3∆200 allele (Figure 6B, lane 1). When this strain was transformed with the pADH1-HIS3-CYC1 control plasmid, which lacks the 70 bp of SNR47 3′ sequence, a strong single HIS3 transcript was now observed (Figure 6B, lane 2). In RTF1, rtf1∆, rtf1-102-104A, and rtf1-E104K strains transformed with the SNR47 reporter plasmid, a single HIS3 transcript was also detected (Figure 6B, lanes 3–6). This transcript is slightly longer than that produced from the control plasmid, indicating HIS3 transcripts are initiating from within the promoter sequence in the reporter plasmid and carry the additional 70 nucleotides of SNR47 absent in the control plasmid. If cryptic transcription was occurring in the rtf1 mutant strains, we would have expected to detect smaller HIS3 transcripts compared to those observed in the wild-type control strain, which we did not. This evidence argues against the idea that cryptic initiation is occurring in the SNR47 reporter plasmid and instead supports the idea that the growth of our mutants in medium lacking histidine is due to read through of the SNR47 3′ sequence.

To further verify the results obtained with the SNR47 reporter plasmid and confirm that defective transcription termination of snoRNAs occurs in these rtf1 mutant strains, we examined transcripts produced from the endogenous SNR47 locus by Northern analysis. To detect SNR47 read-through transcripts, we designed a probe to hybridize to a region between the SNR47 termination element and the translation start site of YDR042C, the downstream gene (see materials and methods for details). Analogous to the results obtained using the reporter plasmid, Northern analysis revealed an extended read-through transcript (SNR47-YDR042C) in rtf1∆, rtf1-102-104A, rtf1-E104K, rtf1-108-110A, and rtf1-F80V,F123S cells, suggesting that residues in the Rtf1 HMD contribute to its role in snoRNA termination (Figure 7, A and B). In addition to SNR47, the SNR13 locus was also shown to be regulated by Rtf1 (Sheldonet al. 2005). Therefore, we examined this locus for read-through transcripts in the Rtf1 mutants by Northern analysis, using a probe that hybridizes to a region downstream of the SNR13 termination element and extending partially into the coding sequence of TRS31, the downstream gene (see materials and methods). This probe will hybridize to both TRS31 transcripts (lower band, Figure 7C) and the read-through transcript (SNR13-TRS31; upper band, Figure 7C). As with the SNR47 locus, rtf1∆, rtf1-102-104A, rtf1-E104K, rtf1-108-110A, and rtf1-F80V,F123S cells all show read-through transcription at SNR13 (Figure 7, C and D). We found that the rtf1-F80V-F123S mutant consistently showed the highest levels of read-through transcription, indicating these hydrophobic amino acids make the greatest contribution of the HMD residues tested to snoRNA termination. Taken together, our results suggest that conserved residues in the Rtf1 HMD contribute to the regulation of proper snoRNA 3′-end formation.

Figure 7.—

Amino acid substitutions in Rtf1 prevent proper snoRNA 3′-end formation.

(A) Representative Northern analysis of extended SNR47 transcripts (SNR47-YDR042C) in wild-type (KY1698), rtf1∆ (KY1703), HA-rtf1-102-104A (KY1981), HA-rtf1-E104K (KY1982), HA-rtf1-108-110A (KY1983), and HA-rtf1-F80V,F123S (KY1984) cells. SCR1 serves as a loading control for all Northern analyses. Twenty micrograms of total RNA was analyzed in these experiments. (B) Quantification of SNR47-YDR042C transcript levels from three independent Northern analyses, performed as described in A, with relative signal of wild type set to one as described in materials and methods. (C) Representative Northern analysis of extended SNR13 transcripts (SNR13-TRS31) and TRS31 transcripts in wild-type (KY1698), rtf1∆ (KY1703), HA-rtf1-102-104A (KY1981), HA-rtf1-E104K (KY1982), HA-rtf1-108-110A (KY1983), and HA-rtf1-F80V,F123S (KY1984) cells. Twenty-five micrograms of total RNA was analyzed in these experiments. (D) Quantification of SNR13-TRS31 transcript levels from three independent Northern analyses, performed as described in C, with relative signal of wild type set to one.

Since the mutated Rtf1 residues are required for proper H2B K123 ubiquitylation, we tested directly whether H2B K123 has a role in the regulation of snoRNA 3′-end formation. To do this, we generated a strain in which an integrated htb1-K123R mutation provides the only source of H2B (see materials and methods for details) and used the SNR47 termination reporter plasmid to determine whether snoRNA 3′-end formation is impaired. Compared to hta2∆ htb2∆ control cells, hta2∆ htb2∆ htb1-K123R cells with the reporter plasmid exhibited stronger growth in the absence of histidine, indicative of a termination defect (Figure 8A, left). To test this result at endogenous loci, Northern blot analysis of SNR47 and SNR13 transcripts was performed in H2B-K123R cells. Consistent with results obtained with the reporter plasmids, snoRNA read-through transcripts were detected in hta2∆ htb2∆ htb1-K123R mutant cells but were not detected in hta2∆ htb2∆ control cells (Figure 8, B, C, F, and G). Importantly, this result reveals a previously uncharacterized role for H2B ubiquitylation in regulating yeast ncRNA transcript 3′-end formation. To further support our finding that H2B K123 ubiquitylation is involved in snoRNA termination pathways, we examined cells lacking RAD6 and BRE1, the ubiquitin conjugase and ligase responsible for H2B ubiquitylation in S. cerevisiae. Cells deleted for RAD6, BRE1, or both genes showed impaired snoRNA 3′-end formation at SNR47 and SNR13 (Figure 8, A and D–G). These experiments demonstrate that conserved Rtf1 residues participate in snoRNA 3′-end formation, likely by promoting proper H2B ubiquitylation.

Figure 8.—

Identification of a role for H2B K123 ubiquitylation in promoting 3′-end formation of snoRNA transcripts.

(A) rtf1∆ (KY2041), wild-type (KY2042), hta2∆ htb2∆ (KY2043), htb1-K123R hta2∆ htb2∆ (KY2044), rad6∆ (KY2045), bre1∆ (KY2046), and rad6∆ bre1∆ (KY2047) cells were transformed with either pADH1-SNR47(70)-HIS3-CYC1 (left) or pADH1-HIS3-CYC1 (right). Ten-fold serial dilutions of these cells were spotted on SC −Leu −His and SC −Leu medium and incubated for 4 days at 30°. (B) Representative Northern analysis of extended SNR47 transcripts (SNR47-YDR042C) in hta2∆ htb2∆ (KY2043) and htb1-K123R hta2∆ htb2∆ (KY2044) cells. Twenty micrograms of total RNA was analyzed. SCR1 served as a loading control for all Northern analyses. (C) Quantification of SNR47-YDR042C transcript levels from three independent Northern analyses, performed as in B, with relative signal of hta2∆ htb2∆ set to one as described in materials and methods. (D) Representative Northern analysis of extended SNR47 transcripts (SNR47-YDR042C) in wild-type (KY2048), rad6∆ (KY2045), and bre1∆ (KY2046) cells. Twenty-five micrograms of total RNA was analyzed. (E) Quantification of SNR47-YDR042C transcript levels from three independent Northern analyses, performed as in D, with relative signal of wild type set to one. (F) Representative Northern analysis of extended SNR13 transcripts (SNR13-TRS31) in wild-type (KY2042), hta2∆ htb2∆ (KY2043), htb1-K123R hta2∆ htb2∆ (KY2044), rad6∆ (KY2045), and bre1∆ (KY2046) cells. Twenty-five micrograms of total RNA was analyzed. (G) Quantification of SNR13-TRS31 transcript levels from three independent Northern analyses performed as described in F. The relative signals of hta2∆ htb2∆ (left graph) or wild type (right graph) were set to one. (H) Wild-type (KY2042), dot1∆ (KY2091), set1∆ (KY2092), set1∆ dot1∆ (KY2093), rtf1∆ (KY2041), bre1∆ (KY2046), rad6∆ (KY2045), and rad6∆ bre1∆ (KY2047) cells were transformed with either pADH1-SNR47(70)-HIS3-CYC1 (left) or pADH1-HIS3-CYC1 (right). Ten-fold serial dilutions of these cells were spotted on SC −Leu −His and SC −Leu medium and incubated for 4 days at 30°.

The requirement for H2B K123 monoubiquitylation in proper snoRNA 3′-end formation at SNR47 and SNR13 suggests that downstream methylation events on histone H3 may be needed for snoRNA termination or alternatively K123 may be functioning independently of H3 K4 and K79 methylation. An examination of SNR47 and SNR13 transcripts in set1∆ and dot1∆ cells by Northern blot analysis, however, did not reveal a strong role for either of the H3 K4 and K79 methylation marks in snoRNA termination as compared to the defects observed in rtf1∆ cells (data not shown; Sheldonet al. 2005). The use of the SNR47 termination reporter plasmid in cells lacking either SET1 or DOT1 did show that these cells have a slight increase in growth in the absence of histidine as compared to wild-type cells. However this level of growth was considerably lower than that observed for rtf1∆, bre1∆, rad6∆, or rad6∆ bre1∆ cells, which all lack H2B K123 ubiquitylation (Figure 8H). These results suggest that SET1 and DOT1 have a modest role in snoRNA termination that can be observed when using a reporter construct that is driven by the highly expressed ADH1 promoter, but is only weakly detected at the endogenous SNR47 and SNR13 loci using Northern blot analysis. While a more sensitive technique may clarify whether H3 K4 or K79 methylation marks have any significant roles in termination at SNR47 and SNR13, our results suggest that the requirement for H2B K123 monoubiquitylation in this process is greater than that of downstream histone H3 modifications on K4 and K79. Further supporting this idea, all of our tested rtf1 mutants that lack detectable K123 ubiquitylation also demonstrate snoRNA read-through defects despite the fact that they exhibit varied states of K4 and K79 methylation. Taken together, our data reveal a new regulatory role for H2B K123 monoubiquitylation in snoRNA 3′-end formation. The many important functions of H2B ubiquitylation, which are dependent on specific residues in the Rtf1 HMD, likely explain why these residues have remained so conserved throughout eukaryotes.

DISCUSSION

The conserved Paf1 complex controls many important gene regulatory events in eukaryotes including transcription elongation, histone modifications, and RNA 3′-end formation. Understanding the mechanisms of how the Paf1 complex functions in a genetically tractable organism, like yeast, will shed light on its conserved functions in higher eukaryotes. Here we performed a genetic analysis of the histone modification domain of the Paf1 complex subunit Rtf1 in S. cerevisiae. Through targeted substitutions and a genetic screen, we identified mutations that alter highly conserved residues in the HMD and characterized their effects on the multiple functions of Rtf1. Specifically, we defined residues in the HMD that are required to achieve histone H2B K123 monoubiquitylation and full levels of downstream histone H3 K4 and K79 methylation. We also demonstrated that when these critical residues are altered, cells exhibit phenotypes associated with defects in transcription elongation, telomeric silencing, and prevention of cryptic initiation. This wide range of impaired functions may explain why so many of these residues are highly conserved. Interestingly, the rtf1 mutant strains also produce read-through snoRNA transcripts, implicating Rtf1 HMD-dependent functions in ncRNA 3′-end formation. In support of this idea, we found that the H2B monoubiquitylation site and proteins required for H2B K123 ubiquitylation are necessary for proper 3′-end formation of two different snoRNAs, providing new evidence for inclusion of histone modifications in this process.

While the alanine scanning approach highlighted the importance of the charged and conserved residues 102–104 and 108–110 in the HMD, our unbiased mutagenesis screen identified mutations that alter E104 and F123, with F80 playing a modulatory role. Notably, the F123 residue has been maintained as hydrophobic during evolution. Furthermore, among the Rtf1 proteins of 19 species examined, including the subset highlighted in Figure 1A, E104 and R110 are invariant. Therefore our discovery of many phenotypic consequences of alterations of the E104 and F123 residues highlight their functional importance to Rtf1 and may explain their evolutionary conservation. It was interesting to discover that rtf1-E104K and rtf1-E104G alleles showed similar defects, which were not always seen in the rtf1-102-104A cells. For example, only the rtf1-E104K and rtf1-E104G mutations caused an Spt phenotype, suggesting that an alanine substitution of this residue is likely a less disruptive change than substitution by a lysine or a glycine.

All of the amino acid substitutions within the Rtf1 HMD resulted in a dramatic and global loss of H2B ubiquitylation. Monoubiquitylation of histone H2B at lysine 123 is a prerequisite for full levels of di- and trimethylation of histone H3 K4 and K79. H2B ubiquitylation specifically participates in histone crosstalk in part by controlling the association of the Cps35/Swd2 protein with the Set1-containing COMPASS complex (Leeet al. 2007a). Indeed, when we examined rtf1-E104K and rtf1-E104G cells, we saw a near complete loss of di- and trimethylation of H3 K4. However, while rtf1-102-104A, rtf1-108-110A, and rtf1-F123S mutant cells had strong reductions of H3 K4 trimethylation, they retained substantial levels of H3 K4 dimethylation. Although it was not detected by our immunoblot assays, differences in the methylation profiles of the mutants might be due to differences in the levels of ubiquitylated H2B. These rtf1 mutations may be useful for defining the downstream functional consequences of these methylation marks. For example, the Spt phenotypes of the rtf1∆3, rtf1∆4, rtf1-E104K, and rtf1-E104G mutants correlate with a loss of H3 K4 dimethylation and H3 K79 di- and trimethylation. Cells containing rtf1-102-104A, rtf1-108-110A, and rtf1-F123S mutant alleles all show somewhat impaired H3 K79 di- or trimethylation, but not the complete loss of those modifications as seen in rtf1∆ cells. This result indicates that additional residues in Rtf1 are likely to be required for H3 K79 methylation. Furthermore, as substitutions of E104 showed the strongest K79 methylation defects, this again suggests that different residues in the HMD contribute to histone methylation to different extents and that the E104 residue appears to be the most important of the residues we tested in a broad range of Rtf1 functions. Therefore, these rtf1 HMD mutant alleles serve as genetic tools for further experiments to explain the mechanisms of how Rtf1, the Paf1 complex, and a range of histone post-translational modifications allow cells to properly transcribe their genetic material.

In addition to promoting histone H3 K4 and K79 methylation, histone H2B ubiquitylation in yeast has many important functions. H2B ubiquitylation cooperates with the yeast FACT (Spt16/Pob3) complex in promoting transcription elongation (Pavriet al. 2006;Fleminget al. 2008). The recruitment of Ctk1, the kinase responsible for phosphorylation of serine 2 on the RNA Pol II CTD, can be altered by H2B ubiquitylation (Wyceet al. 2007). H2B ubiquitylation also has an impact on gene repression by preventing histone eviction and stabilizing nucleosomes positioned over promoters of repressed genes, thereby preventing gene regulators from binding and activating transcription (Chandrasekharanet al. 2009). A role for H2B ubiquitylation in gene repression is also supported by microarrays on H2B-K123R cells, which showed that 75% of genes affected by the substitution had increased levels of expression (Mutiuet al. 2007). On the basis of the loss of histone H2B ubiquitylation in our Rtf1 HMD mutant strains, we expect that many of these functions downstream of H2B K123 ubiquitylation would also be affected in the Rtf1 HMD mutants. Recently, H2B ubiquitylation has been linked to preventing apoptosis (Walteret al. 2010). Therefore, as new roles for H2B ubiquitylation in transcription continue to be uncovered, it will be exciting to learn more about the complex interplay underlying how histone H2B ubiquitylation impacts so many different processes.

We previously showed that the Rtf1, Paf1, and Ctr9 subunits of the Paf1 complex make important contributions to 3′-end formation at SNR47 (Sheldonet al. 2005). snoRNAs such as SNR47 are critical in acting as guides for modifications of newly synthesized pre-rRNAs that are important for the generation of functional ribosomes and are one class of noncoding RNAs in eukaryotes. Furthermore, in higher eukaryotes, snoRNAs have been shown to generate other small ncRNAs, some of which function like microRNAs (Enderet al. 2008; Taftet al. 2009). Here we provide strong evidence for a new function of histone H2B ubiquitylation in snoRNA 3′-end formation. Interestingly, as the requirement of H2B K123 seems to be partially independent of its role in downstream histone H3 methylation marks on K4 and K79, this indicates that the K123 monoubiquitylation modification may play a more direct role in this termination pathway. Because the nucleosome density of yeast terminator regions differs from that of promoter or coding regions, regulation of nucleosome positioning may contribute to proper RNA transcript termination, and H2B ubiquitylation could influence 3′-end formation through its effects on nucleosome stability and chromatin compaction (Yuanet al. 2005; Leeet al. 2007b; Mavrichet al. 2008; Chandrasekharanet al. 2009; Fanet al. 2010; Fierzet al. 2011). Given that the Paf1 complex was shown to affect the recruitment to SNR47 of the RNA-binding proteins Nrd1 and Nab3, which are required for snoRNA termination, the ubiquitylation of H2B K123 may be needed to target these factors to snoRNAs (Sheldonet al. 2005). One way in which H2B ubiquitylation and the Paf1 complex could regulate the downstream recruitment of RNA processing factors, such as Nrd1 and Pcf11, is through effects on Pol II CTD phosphorylation (Steinmetzet al. 2001; Steinmetz and Brow 2003; Muelleret al. 2004; Wyceet al. 2007; Nordicket al. 2008). Alternatively, the Paf1 complex may control RNA 3′-end formation via direct physical interactions with RNA processing factors as previously reported for the yeast and human Paf1 complexes (Nordicket al. 2008; Rozenblatt-Rosenet al. 2009). SNR47 and SNR13 are part of the box C/D class of snoRNAs and have defined Nab3 and Nrd1 binding sites. Identification of new snoRNA targets of H2B K123 ubiquitylation and the Paf1 complex may broaden targets to other snoRNA classes, such as box H/ACA snoRNAs. Determining whether Nrd1 and Nab3 binding sites are always found in such targets may shed light on the mechanisms of this 3′-end formation pathway.

Beyond the importance of understanding why snoRNAs require alternate pathways of 3′-end formation, the Nrd1, Nab3, Sen1 regulatory pathway also impacts other interesting and unique transcriptional control mechanisms. For example, Nrd1 was recently shown to influence nuclear mRNA quality control through effects on the exosome component Rrp6 (Honorineet al. 2011). This function may be related to a nuclear RNA surveillance pathway recently found in humans that also involves Rrp6 (de Almeidaet al. 2010). In addition, Nrd1, Nab3, and Sen1 participate in the transcription of a recently defined set of cryptic unstable transcripts (CUTs) (Wyerset al. 2005; Arigoet al. 2006; Thiebautet al. 2006; Neilet al. 2009). Despite their transient nature, CUTs are important regulators of gene transcription in yeast and may resemble the short and unstable ncRNA transcripts recently found in human cells (Prekeret al. 2008; Carninci 2009). Furthermore, this regulatory pathway is also involved in regulation of upstream noncoding transcripts that result in transcription attenuation of downstream genes such as IMD2 and URA2 (Kuehner and Brow 2008; Thiebautet al. 2008). Interestingly, the Paf1 complex was very recently shown to be involved in a similar regulatory mechanism at FKS2 by influencing the recruitment of the Sen1-Nrd1-Nab3 termination complex (Kim and Levin 2011). Therefore, the Paf1 complex and the H2B K123 modification may participate in a range of regulatory pathways involving Nrd1, Nab3, and Sen1. Understanding how these proteins work together in promoting proper snoRNA 3′-end formation might therefore elucidate mechanisms underlying a number of interesting and novel transcriptional regulatory phenomena.

Importantly, the pathway establishing histone H2B ubiquitylation in humans is highly similar to that of yeast. The hRAD6 and hBRE1 homologs are responsible for ubiquitylation of the analogous lysine, 120, on human histone H2B, and hRtf1 and the rest of the human Paf1 complex play an important role in this modification by recruiting hRad6/Bre1(Zhuet al. 2005b; Kimet al. 2009). Similar to the crosstalk observed in yeast, human H2B K120 ubiquitylation stimulates histone H3 K4 methylation by hSet1 and H3 K79 methylation by hDot1 (McGintyet al. 2008; Kimet al. 2009). Underscoring the importance of these modifications in human cells, hBre1/Rnf20 functions as a tumor suppressor and the methyltransferase hSet1/MLL is often translocated in leukemias (Krivtsov and Armstrong 2007; Shemaet al. 2008). Additionally, components of the human Paf1 complex are involved in cancer regulation (Mosimannet al. 2006; Chaudharyet al. 2007; Linet al. 2008). This is especially true of the hCdc73/parafibromin tumor suppressor protein encoded by the HRPT2 gene, but misexpression of other hPaf1 complex subunits has also been found in tumor cells (Shattucket al. 2003; Rozenblatt-Rosenet al. 2005; Moniauxet al. 2006; Chaudharyet al. 2007). Given the dependence of human H2B K120 ubiquitylation on the hPaf1 complex, we expect that analysis of the mechanisms and contributions of H2B ubiquitylation in yeast will elucidate related processes in higher eukaryotes. It seems likely, for example, that the conserved residues in the HMD that we have shown to be critical for the H2B ubiquitylation and downstream modifications in yeast will also promote similar functions in hRtf1. Therefore, the rtf1 mutations identified in this study have not only allowed us to define specific residues important to H2B ubiquitylation, but as we have shown by uncovering a role for this modification in snoRNA 3′-end formation, they also represent useful genetic tools for identifying new functions of H2B ubiquitylation.

Acknowledgments

We are grateful to Elia Crisucci, Maggie Braun, Jill Dembowski, Daniel Sheidy, and Kristen Butela for generation of strains or plasmids used in this study. We thank Fred Winston and Jeffry Corden for yeast strains and plasmids and Brian Strahl for technical advice. We thank Joe Martens, Margaret Shirra, Elia Crisucci, Kristin Klucevsek, and Allen Ho for the critical reading of the manuscript. This research was supported by a National Institutes of Health grant GM52593 to K.M.A., by award F32GM093383 to B.N.T. from the National Institute of General Medical Sciences, and by undergraduate research fellowships from the Howard Hughes Medical Institute and the Beckman Scholars Program to C.P.D.

Footnotes

  • Received December 22, 2010.
  • Accepted March 14, 2011.

LITERATURE CITED

View Abstract