|Home | About | Journals | Submit | Contact Us | Français|
Set2 methylates Lys36 of histone H3. We show here that yeast Set2 copurifies with RNA polymerase II (RNAPII). Chromatin immunoprecipitation analyses demonstrated that Set2 and histone H3 Lys36 methylation are associated with the coding regions of several genes that were tested and correlate with active transcription. Both depend, as well, on the Paf1 elongation factor complex. The C terminus of Set2, which contains a WW domain, is also required for effective Lys36 methylation. Deletion of CTK1, encoding an RNAPII CTD kinase, prevents Lys36 methylation and Set2 recruitment, suggesting that methylation may be triggered by contact of the WW domain or C terminus of Set2 with Ser2-phosphorylated CTD. A set2 deletion results in slight sensitivity to 6-azauracil and much less β-galactosidase produced by a reporter plasmid, resulting from a defect in transcription. In synthetic genetic array (SGA) analysis, synthetic growth defects were obtained when a set2 deletion was combined with deletions of all five components of the Paf1 complex, the chromodomain elongation factor Chd1, the putative elongation factor Soh1, the Bre1 or Lge1 components of the histone H2B ubiquitination complex, or the histone H2A variant Htz1. SET2 also interacts genetically with components of the Set1 and Set3 complexes, suggesting that Set1, Set2, and Set3 similarly affect transcription by RNAPII.
In Saccharomyces cerevisiae RNA polymerase II (RNAPII) initiates transcription in concert with general transcription factors and a 20-subunit mediator complex (16, 38, 59). During initiation it becomes phosphorylated by the Kin28 subunit of the general transcription factor TFIIH on Ser5 of the heptapeptide repeats, YSPTSPS, in the carboxy-terminal domain (CTD) of its largest subunit, Rpb1, resulting in recruitment of the mRNA-capping enzyme to the transcription complex (23, 32, 49, 69). Ser5 phosphorylation declines during the early stages of elongation and is replaced by Ser2 phosphorylation, mediated mainly by the cyclin-dependent kinase, Ctk1, during the later stages of elongation by RNAPII (10, 23). General transcription factors and mediator dissociate from the transcription complex or remain behind at the promoter and are replaced by various elongation factors during chain elongation by RNAPII (25, 43). Among these elongation factors are Spt4/Spt5, Spt6/Iws1, and Spt16/Pob3, whose patterns of genetic interactions suggest that they participate in transcription on chromatin templates (66). Another elongation factor is the Paf1 complex, which consists of Paf1, Rtf1, Cdc73, Ctr9, and Leo1 and associates with RNAPII (25, 34, 52). Chromatin immunoprecipitation (ChIP) experiments have shown that all of these polypeptides are associated with transcribed regions (25, 43).
Histone methylation by SET domain-containing histone lysine methyltransferases has important roles in chromatin structure and function (44). Lysines 4, 9, 27, and 79 are well-studied sites of methylation on histone H3, while lysine 20 is the only known methylated lysine in histone H4 (55, 64). There are at least six proteins with recognizable SET domains in S. cerevisiae. Set1, a component of an eight-protein complex (COMPASS), methylates Lys4 of histone H3 (7, 24, 33, 36, 46). Recruitment of COMPASS to the early transcribed region, as well as histone H3 Lys 4 trimethylation in this region, requires both Ser5 phosphorylation of the RNAPII CTD by Kin28 and the Rtf1, Paf1, and Ctr9 subunits of the Paf1 complex (26, 37, 48). Methylation of H3 Lys4 by Set1, as well as methylation of H3 Lys79 by Dot1, also requires ubiquitination of histone H2B by Rad6 and Bre1 (13, 19, 58, 67).
Set2 methylates Lys36 of histone 3 (54). Members of the mammalian nuclear receptor-binding SET-domain-containing family (NSD1, NSD2, and NSD3) contain a SET domain that is highly related to that of yeast Set2. NSD genes have been implicated in acute myeloid leukemia (21, 47, 53), whereas the NSD2 gene maps to the region associated with Wolf-Hirschhorn syndrome (47), which is characterized by mental retardation and developmental defects.
In this study, we found that Set2 interacts physically with RNAPII. Similar observations have been made recently by other groups (28, 29, 68). We used ChIP assays to show that Set2 is recruited and methylates histone H3 Lys36 in the coding regions of actively transcribed genes. Synthetic growth defects were observed when a set2 deletion was combined with deletions of known or suspected transcriptional elongation factors, including Chd1, Soh1, and members of the Paf1 elongation complex, as well as components of the Set1 and Set3 complexes. A deletion of set2 also results in slight sensitivity to the drug 6-azauracil (6-AU) and a decrease in β-galactosidase from a lacZ reporter plasmid, implicating Set2 in transcriptional regulation. Finally, recruitment of Set2 and Lys36 histone H3 methylation relies on components of the Paf1 complex as well as the RNAPII CTD kinase Ctk1. These results suggest that binding of Set2 to RNAPII and histone H3 methylation occur during transcriptional elongation and depend on both the Paf1 elongation factor and the phosphorylation state of the CTD. The involvement of SET domain proteins in transcriptional elongation would not be entirely unexpected, because a translocation that joins part of the human elongation factor ELL to part of the SET domain protein MLL leads to myelogenous leukemia (12).
The following yeast strains were used in this study: NJK581 MATa ura3-1 leu2-3,112 his3-11,15 trp1Δ ade2-1 can1-100 set2-TAP::TRP1 and NJK742 MATa ura3-1 leu2-3,112 his3-11,15 trp1Δ ade2-1 can1-100 set2(Δ476-733)-TAP:TRP1. Strains with the following genes replaced by a kanamycin resistance cassette were obtained from Research Genetics (http://sequence-www.stanford.edu/group/yeast_deletion_project/deletions3.html): CDC73, RTF1, CTK1, and SET2. Tandem affinity purification (TAP)-tagged versions of Set2 were then made in these backgrounds (4, 17), as follows: YSB971, MATa ura3Δ0 leu2Δ0 his3Δ1 met15Δ0 set2-TAP::HIS3 ctk1Δ::KanMX; YSB972, MATa ura3Δ0 leu2Δ0 his3Δ1 met15Δ0 set2-TAP::HIS3 rtf1Δ::KanMX; and YSB973, MATa ura3Δ0 leu2Δ0 his3Δ1 met15Δ0 set2-TAP::HIS3 cdc73Δ::KanMX. Sensitivity to 6-AU was tested by plating strains harboring pRS316 (51) onto plates lacking uracil and containing 25 μg of 6-AU per ml.
ChIP assays were performed essentially as described earlier (23). Yeast strains were grown at 30°C to an optical density at 595 nm (OD595) of 0.4 to 0.6. For the ChIP with Rpb3 on the lacZ gene, both wild-type and set2Δ strains harboring p416GAL1-lacZ were incubated until they reached an OD595 of approximately 1 in medium containing 2% glucose. Cells were harvested, washed three times with sterile water, and divided into two aliquots. The cells were reinoculated into media containing either 2% glucose or 2% galactose. After 4 h at 30°C, an aliquot was removed for the β-galactosidase assay and the remaining cells were cross-linked with formaldehyde for ChIP assays. The antibody against Rpb3 was obtained from Neoclone Biotechnology. Unpublished primers used for PCR in this study are designated by the name of the gene followed by the position of its 5′ end relative to the translation initiation codon, as follows: Gal1-lacZ−340 (CGAACAATAAAGATTCTACAATACTAGC); Gal1-lacZ−70 (CATTTGAATAAGAAGTAATACAAACCGA); Gal1-lacZ715 (TGTACTGGAGGCTGAAGTTCAGAT); Gal1-lacZ1025 (CAGACCATTTTCAATCCGCACCTC); Gal1-lacZ1567 (AATGGCTTTCGCTACCTGGAGAGA); Gal1-lacZ1827 (CAAAGACCAGACCGTTCATACAGA); Gal1-lacZ2247 (ATCGAGCTGGGTAATAAGCGTTGG); Gal1-lacZ2465 (GTATCTGCCGTGCACTGCAACAAC); Gal1-lacZ3094 (AATGGCGATTACCGTTGATGTTGAA); Gal1-lacZ3414 (GTTTCCATCAGTTGCTGTTGACTGT).
TAP-tagged Set2 was purified on immunoglobulin G (IgG) and calmodulin columns from extracts of yeast cells (3 liters) grown in yeast extract-peptone-dextrose medium to an OD600 of 1.0 to 1.5. The cell pellets (7 to 10 g) were frozen in liquid nitrogen and lysed by grinding with dry ice in a Krups coffee grinder (model no. 203-70). An equal volume of buffer (250 mM KCl, 100 mM HEPES-KOH [pH 7.9], 1 mM EDTA, 2.5 mM dithiothreitol [DTT]) was added, and following centrifugation in a Beckman Ti70 rotor at 4°C for 2 h at 34,000 rpm, the supernatant was dialyzed against IPP buffer (10 mM Tris-Cl [pH 7.9], 0.1% Triton X-100, 0.5 mM DTT, 0.2 mM EDTA, 20% glycerol, 100 mM NaCl). After dialysis, the extract was again centrifuged in a Ti70 rotor at 4°C for 30 min at 34,000 rpm and the supernatant was mixed for 3 h with 200 μl of IgG-Sepharose (Pharmacia) equilibrated with IPP buffer. Following binding, the IgG-Sepharose was washed with 1 ml of IPP buffer followed by 400 μl of TEV protease cleavage buffer (50 mM Tris-Cl [pH 7.9], 1 mM DTT, 0.1% Triton X-100, 100 mM NaCl). The beads were then incubated overnight at 4°C with 100 U of TEV protease (Life Technologies) in 200 μl of TEV cleavage buffer. The eluate was combined with a 200-μl wash with TEV cleavage buffer. To this were added 200 μl of calmodulin binding buffer (10 mM Tris-Cl [pH 7.9], 10 mM β-mercaptoethanol, 2 mM CaCl2, 0.1% Triton X-100, 100 mM NaCl) and 200 μl of calmodulin beads (Pharmacia) equilibrated with the same buffer. After binding for 1 to 2 h at 4°C, the calmodulin beads were washed with 200 μl of calmodulin binding buffer and 200 μl of calmodulin wash buffer (10 mM Tris-Cl [pH 7.9], 10 mM β-mercaptoethanol, 0.1 mM CaCl2, 0.1% Triton X-100, 100 mM NaCl). The purified protein complexes were eluted from the calmodulin beads with 5 × 100 μl of calmodulin elution buffer (10 mM Tris-Cl [pH 7.9], 10 mM β-mercaptoethanol, 3 mM EGTA [pH 8.0], 0.1% Triton X-100, 100 mM NaCl). The purified proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on gels containing 10% polyacrylamide, and the proteins were visualized by silver staining. Peptide samples were spotted onto a target plate with a matrix of α-cyano-4-hydroxycinnamic acid (Fluka). Matrix-assisted laser desorption ionization-time of flight (mass spectrometry) (MALDI-TOF) analysis was conducted by utilizing a Reflex IV (Bruker Daltonics, Billerica, Mass.) instrument in positive-ion reflectron mode (31, 50). Western blotting was performed by using standard techniques with the monoclonal antibodies 8WG16, H5, and H14 (6, 40, 60).
Synthetic genetic array (SGA) analysis was carried out as previously described (61).
In order to further characterize S. cerevisiae Set2, we used single-step transformation to place a TAP tag (45) containing a calmodulin-binding peptide and Staphylococcus aureus protein A, separated by a TEV protease cleavage site, at the C terminus of Set2. To confirm the success of the tagging procedure, Western blotting was performed on extract derived from the tagged strain by making use of an irrelevant immunoglobulin to recognize the protein A component of the TAP tag. The tagged protein was then purified on IgG and calmodulin columns and analyzed by SDS-PAGE followed by staining with silver. Protein bands absent from a control preparation and corresponding to the tagged protein and any associated proteins were then identified by MALDI-TOF (50). Set2 copurified with substoichiometric amounts of two larger polypeptides identified as Rpb1 and Rpb2, the two largest subunits of RNAPII (Fig. (Fig.1A).1A). This physical association was confirmed when Set2 and several elongation factors copurified with RNAPII from a strain harboring a TAP tag on Rpb1 (data not shown). We were not able to identify by mass spectrometry any of the other minor polypeptides that copurified with the Set2-TAP shown in Fig. Fig.1A,1A, other than Rrp5 and fragments of Set2. Tagged Rrp5 did not copurify with Set2 (data not shown). Moreover, the minor polypeptides other than Rpb1 and Rpb2 that copurified with Set2-TAP have not been reproducibly detected in other preparations.
RNAPII phosphorylated on Ser2 on the Rpb1 CTD heptapeptide repeats has been shown to localize to transcribed coding regions and 3′ ends of genes, whereas Ser5 phosphorylation on the CTD is found primarily in 5′ early elongation regions (23). Western blotting analyses performed with monoclonal antibodies H5 and H14 (6, 40), which recognize the Rpb1 CTD repeats phosphorylated on Ser2 and Ser5, respectively, showed that the RNAPII copurifying with Set2 was phosphorylated on both residues (Fig. (Fig.1B),1B), suggesting that Set2 might be associated with coding regions during transcription. As a control, Western blotting with H5 and H14 did not detect Rpb1 when a parallel purification was done by using an extract from an untagged strain. The Rpb1 that copurified with Set2 from a Set2-TAP strain was also recognized weakly by the monoclonal antibody 8WG16 (60), which recognizes both unphosphorylated and partially phosphorylated Rpb1 (Fig. (Fig.1B).1B). These results suggested that Set2 might be associated with elongating RNAPII.
ChIP assays in which proteins were cross-linked to DNA in vivo by using formaldehyde were employed to analyze the in vivo distribution of Set2 and histone H3 Lys36 methylation along various transcribed genes (23). Following isolation and shearing of chromatin, Set2-TAP or Lys36-methylated histone H3 was immunoprecipitated with either rabbit IgG agarose (directed against protein A on the TAP tag) or antibody directed against methylated histone H3 Lys36 (Upstate Biotechnology). After reversal of the cross-links, PCR analyses were performed on the coprecipitated DNA. Primer pairs directed against promoter regions, coding regions, and 3′ untranslated regions were used to analyze the ADH1, PYK1, and PMA1 genes, encoding alcohol dehydrogenase, pyruvate kinase, and plasma membrane ATPase, respectively (Fig. (Fig.2A).2A). Set2-TAP, of all three genes that were tested, was found to cross-link most strongly to the coding regions and not to the promoter or 3′ untranslated regions. Similarly, when ChIP experiments were carried out with antibody directed against Lys36-methylated histone H3, the same pattern was observed: an enrichment of cross-linked DNA corresponding primarily to the coding regions of the genes that were studied (Fig. (Fig.2B2B).
To directly test whether the presence of Set2 is correlated with active transcription, ChIP was performed on the galactose-inducible gene, GAL1, which encodes galactokinase. In the absence of galactose, virtually no Set2 cross-linked to any part of the GAL1 gene. Upon induction, however, both Set2-TAP and methylated histone H3 Lys36 were detected primarily in the coding region of GAL1 (Fig. (Fig.2C),2C), providing further evidence that the presence of Set2 and its histone-methylating activity require elongation by RNAPII.
The presence of Set2 in transcribed regions and its interaction with RNAPII suggested that Set2 might influence transcription by RNAPII. To determine whether yeast Set2 influences transcription elongation by RNAPII in vivo, we initially examined a set2Δ strain for sensitivity to the pyrimidine analog 6-AU. Treatment with 6-AU leads to reduction of the UTP and GTP concentrations in yeast cells and impairs elongation by RNAPII (14). Studies using 6-AU have helped to show that genes like DST1, which encodes TFIIS, encode factors having a role in transcriptional elongation (3). Deletion of SET2 resulted in slight sensitivity to 6-AU, suggesting that Set2 might stimulate elongation by RNAPII (Fig. (Fig.3A).3A). Recently, similar observations have been made by two independent groups (28, 29).
To substantiate this conclusion, we examined whether a set2 deletion would reduce lacZ expression from the plasmid, p416GAL1-lacZ, which contains the Escherichia coli lacZ gene fused to a GAL1 promoter (9). Based on results obtained by deleting genes encoding TFIIS or components of the TREX complex, it has been argued that a decrease in β-galactosidase production with this plasmid correlates with a defect in transcriptional elongation (9, 42, 56; our unpublished data). After 4 h of galactose induction, β-galactosidase synthesis was reduced about threefold in a set2Δ strain compared to that of a strain with wild-type SET2 (Fig. (Fig.3B).3B). The addition of 20 μg of 6-AU/ml to a set2 deletion strain harboring the lacZ reporter plasmid resulted in an approximately 20-fold reduction of β-galactosidase compared to that of a wild-type strain (Fig. (Fig.3B).3B). These results, combined with the slight sensitivity to 6-AU of a set2Δ strain, suggested that Set2 is a positively acting transcription factor, perhaps an elongation factor. To further study the effect that a set2 deletion had on the amount of β-galactosidase produced from the reporter plasmid, ChIP was performed to monitor the approximate concentration of Rpb3, the third largest subunit of RNAPII, at various positions across the lacZ gene (Fig. (Fig.3C).3C). As has been previously observed, there was an apparent decline in the density of RNAPII from the 5′ end to the 3′ end of the gene even in a wild-type strain (23), and there was a similar 5′-to-3′ decline in a set2Δ strain. However, the apparent concentration of RNAPII in the promoter-proximal region was lower in a set2Δ strain than in a wild-type strain, suggesting that transcriptional initiation might be defective when SET2 is deleted. The uniform decline of RNAPII across the lacZ gene in a set2Δ strain suggests either that Set2 does not affect elongation or else that Set2 uniformly accelerates elongation without preventing the release of RNAPII at particular sites on the lacZ gene.
In order to test whether Set2 functions during transcriptional elongation, the recruitment of Set2, as well as the presence of Lys36 H3 methylation, was monitored on the PMA1 gene when several known elongation factors were deleted. The Paf1 complex, which contains five subunits and associates with RNAPII, is thought to be an RNAPII elongation factor on the basis of both biochemical and genetic criteria (25, 34, 52). In particular, it colocalizes with RNAPII in the coding regions of various genes (25, 43). Interestingly, deletion of genes encoding either of two components of the Paf1 complex, Rtf1 or Cdc73, resulted in a marked decrease in the recruitment of Set2 across PMA1 (Fig. (Fig.4A)4A) and abolished Lys36 H3 methylation (Fig. (Fig.4B),4B), demonstrating that the Paf1 complex is important for the recruitment of Set2 and essential for its methylation activity.
The RNAPII CTD kinase Ctk1 (27) phosphorylates Ser2 on the CTD heptapeptide repeats during transcriptional elongation (10). Deletion of the gene encoding Ctk1 also nearly eliminated the recruitment of Set2 and its histone H3 Lys36 methylation activity on the PMA1 gene (Fig. 4A and B). The RNAPII CTD consists of proline-rich heptapeptide repeats (2). Set2 contains a WW domain in its C-terminal region (Fig. (Fig.2A),2A), and WW domains are known to recognize proline-rich peptides (30). Consistent with this, we found that deleting the C-terminal portion of Set2, including its WW domain, significantly reduced the recruitment of Set2 to the PMA1, ADH1, and PYK1 genes, especially toward their 3′ ends and virtually eliminated histone H3 Lys36 methylation even though the catalytic SET domain of Set2 was still intact (Fig. 2A and B). These data suggest that an interaction between the WW domain or C-terminal region of Set2 and the Ser2-phosphorylated CTD might stabilize the association of Set2 with elongating RNAPII and trigger Lys36 methylation of histone H3.
To further evaluate the function of Set2, we used a method for systematic construction of double mutants, termed SGA analysis (61), in which the set2Δ mutation was crossed to an array of ~4,700 mutant strains, each carrying a unique gene deletion. Inviable or slow-growing double-mutant meiotic progeny identify functional relationships between genes. The set2 deletion was first introduced into a haploid starting strain of mating type MATα and then crossed to the array of gene deletion mutants of the opposite mating type, MATa. Sporulation of the resulting diploid cells led to the formation of double-mutant meiotic progeny. This resulted in an ordered array of double-mutant haploid strains whose growth rate was monitored by visual inspection and image analysis of colony size. Putative genetic interactions were then confirmed by either tetrad dissection or random sporulation.
There were approximately 60 double-deletion combinations that resulted in synthetic growth defects. Of these, seven are currently thought to function in transcriptional elongation, namely RTF1, CDC73, LEO1, CTR9, PAF1, SOH1, and CHD1 (Fig. (Fig.5).5). Rtf1, Cdc73, Leo1, Paf1, and Ctr9 are components of the Paf1 complex, which is associated with RNAPII and the RNAPII elongation factor Spt16/Pob3 (25, 34, 52). SOH1 was originally identified as a suppressor of HPR1 (15), a component of the transcription elongation complex, TREX (56). Soh1 is also thought to associate with the Mediator initiation complex in higher eukaryotic cells (18). Chd1 is a member of the chromodomain-helicase-DNA-binding family. It has been shown to be involved in ATP-dependent nucleosome remodeling (62) and has also recently been implicated in transcriptional elongation (11, 25). These genetic interactions with known elongation factors implied that Set2 functions during transcriptional elongation.
Synthetic growth defects were also detected between set2Δ and all seven components of the Set3 complex, namely CPR1, HOS2, HST1, SIF2, SNT1, YIL112w, and SET3 (Fig. (Fig.5)5) (41). The Set3 complex contains two known histone deacetylases, Hos2 and Hst1, but no specific methylation activity has yet been attributed to the complex. Similarly, deletions of six of the eight subunits of the Set1-containing complex, COMPASS (CPS60, CPS40, CPS35, CPS30, CPS25, and CPS15) (7, 24, 33, 36, 46), were found to be synthetically sick with set2Δ (Fig. (Fig.5).5). The remaining two components of COMPASS were not tested, since the CPS35 gene is essential and no set1Δ strain was present in the original deletion array. These results suggested that the function of Set2 is similar in some way to the functioning of COMPASS and the Set3 complex.
A set2 deletion also generated synthetic growth defects when combined with a deletion of the gene encoding either of two components of the histone H2B ubiquitination complex, Bre1 or Lge1 (Fig. (Fig.5)5) (19, 67; our unpublished data), consistent with observations that histone H2B ubiquitination is essential for histone H3 methylation by Set1 (13, 58). Interestingly, a set2Δ htz1Δ double mutant also has a synthetic growth defect. Htz1 is a histone H2A variant (20), and this result suggested that Htz1 might have a role in histone methylation or transcriptional elongation or their consequences.
The specificity of the synthetic genetic interactions uncovered by the SGA screen with SET2 was quite striking. In 64 other full-scale SGA screens, synthetic genetic interactions were only rarely detected with genes encoding subunits of COMPASS, the Set3 complex, or the Paf1 complex (e.g., for the Set3 complex: 1 of 64 were detected for SET3, 2 of 64 for YIL112W, 2 of 64 for HOS2, 0 of 64 for HST1, 5 of 64 for SIF2, 3 of 64 for SNT1, 2 of 64 for CPR1) and then only when screens were done with genes that are likely to be involved in transcriptional elongation (A. Tong and C. Boone, unpublished data). In total, there were 45 interactions with subunits of COMPASS, the Set3 complex, or the Paf1 complex in 64 screens, averaging less than one hit per screen. Of the 37 other synthetic genetic interactions with set2Δ that were detected in our screen (data not shown), 17 were open reading frames of unknown function and none are known to be involved in transcription, although further study may indeed reveal that some are involved in transcriptional elongation and/or histone methylation.
Histone H3 methylation on Lys36 was recently attributed to the SET domain-containing protein Set2 in S. cerevisiae (54). In an effort to further characterize this protein, we used single-step transformation to place a TAP tag on the C terminus of Set2. Following purification and analyses by SDS-PAGE and mass spectrometry, we discovered that a substoichiometric amount of RNA polymerase II (RNAPII) copurified with Set2.
This physical interaction suggested that methylation of histone H3 Lys36 might be intimately linked to transcription. ChIP analyses demonstrated that Set2 localizes primarily to the coding regions of three genes that were tested, and this correlated well with the locations of histone H3 Lys36 methylation. Consistent with the notion that the recruitment of Set2 is linked to elongation by RNAPII, the presence of Set2 was found to correlate with active transcription when ChIP was performed on the GAL1 gene following induction with galactose. Strahl et al. tethered Set2 to a heterologous promoter and demonstrated that, under these conditions, Set2 represses transcription (54). However, we found little or no Set2 or histone H3 Lys36 methylation in various promoter regions, suggesting that Set2 activity is not normally directed towards this region. Deletion of the SET2 gene caused slight sensitivity to 6-AU and a large reduction in β-galactosidase produced by a reporter gene. Both the sensitivity to 6-AU and the reduced β-galactosidase synthesis from the reporter gene that we used in our experiments have been used as indicators that a gene is involved in enhancing elongation by RNAPII (3, 9, 25, 42, 52). A similar suggestion has been made based on the enhanced sensitivity to 6-AU generated when a SET2 deletion is combined with a deletion of DST1, which encodes TFIIS (28). Follow-up ChIP experiments showed that the concentration of RNAPII is lower across the lacZ gene in a set2Δ background than in a wild-type strain. Therefore, we propose that wild-type Set2 normally has a positive role in transcription. This experiment did not, however, prove that Set2 specifically stimulates elongation by RNAPII.
Phosphorylation and dephosphorylation of the CTD of the largest subunit of RNAPII, Rpb1, are required for gene regulation (22, 63). Phosphorylation on Ser5 of the CTD heptapeptide repeats, YSPTSPS, by the Kin28 subunit of the general transcription factor TFIIH is associated with early transcriptional elongation, whereas Ser2 phosphorylation by Ctk1 is found at later elongation steps (10, 23). Our Western blot analyses demonstrated that both types of RNAPII copurify with Set2, again implying that the recruitment of Set2 to a transcription unit and histone H3 methylation on Lys36 might be cotranscriptional. It is striking that the distribution of Set2 along various genes, strong in the coding region and weak in the promoter-proximal and 3′ untranslated regions, is virtually identical to the distribution of Ser2-phosphorylated CTD (10, 23) and the Ser2 kinase, Ctk1 (M. Kim, S. H. Ahn, and S. Buratowski, unpublished data). Moreover, deletion of CTK1 results in substantially less recruitment of Set2 to the PMA1 gene and virtually eliminates histone H3 Lys36 methylation, indicating that Lys36 methylation depends on Ser2 phosphorylation of the CTD during transcriptional elongation.
The WW domains of Ess1/Pin1 and Rsp5 are known to interact with the proline-rich CTD repeats of RNAPII (8, 35, 57). Since our ChIP experiments showed that deletion of the C-terminal region of Set2, including its WW domain, reduced the recruitment of Set2 and essentially eliminated its methylation activity, we propose that there is a physical interaction between the WW domain of Set2 and the Ser2-phosphorylated CTD of RNAPII that stabilizes the association of Set2 with RNAPII and triggers the histone H3 methylation activity of Set2. Consistent with this conclusion, Li et al. recently also found that Set2 copurifies with RNAPII and showed that Set2 can bind to high concentrations of a Ser2-phosphorylated CTD peptide (29). Although this study found that deletion of the WW domain of Set2 eliminates the association between RNAPII and Set2, we have observed that Set2 (Δ476-733)-TAP, which lacks the WW domain, still coimmunoprecipitates with some RNAPII (N. J. Krogan and J. Greenblatt, unpublished data) and is still recruited, albeit less efficiently, to actively transcribed genes (Fig. 2A and B). Xiao et al. (68) have found that a C-terminal portion of Set2 that is distal to its WW domain is sufficient to bind RNAPII. Therefore, a site on RNAPII other than the Ser2-phosphorylated CTD is also likely to mediate an interaction with Set2.
To further pursue Set2's role in transcriptional elongation, the presence of histone H3 Lys36 methylation and the recruitment of Set2 were examined on the PMA1 gene when genes encoding subunits of the Paf1 complex were deleted. The Paf1 complex contains five subunits (Paf1, Cdc73, Rtf1, Leo1, and Ctr9) and has recently been implicated, both biochemically and genetically, in the process of transcriptional elongation (11, 25, 34, 52). Specifically, deletions of genes encoding components of the Paf1 complex cause sensitivity to 6-AU and genetically interact with mutations in genes encoding other known elongation factors. Furthermore, the Paf1 complex interacts physically with both RNAPII and Spt16/Pob3, the yeast homologue of the human elongation factor FACT (39). ChIP experiments have also shown that the Paf1 complex colocalizes with RNAPII in the coding regions of various genes (25, 43) and, like Set2, declines in the region beyond the poly(A) signal (Kim et al., unpublished). We have shown here that when genes encoding two components of the Paf1 elongation complex, Cdc73 and Rtf1, are deleted, Set2 recruitment is significantly reduced and histone H3 methylation is virtually eliminated.
Consistent with a functional connection between Set2 and the Paf1 complex, synthetic growth defects were observed when a set2 deletion was individually combined with deletions of genes encoding all five subunits of the Paf1 complex. The SET2 gene also interacts with the genes encoding two other putative elongation factors, Soh1 and Chd1. Mutations in SOH1 were originally identified as suppressors of HPR1 (15), a component of the transcription elongation complex TREX (56). We have recently found that a soh1 deletion also interacts genetically with genes encoding a number of other known elongation factors, including TFIIS, and causes significant sensitivity to 6-AU and a marked decrease in expression of the lacZ gene on the plasmid p416GAL1-lacZ (Krogan and Greenblatt, unpublished). The CHD1 gene interacts genetically with a number of genes encoding elongation factors (11; Krogan and Greenblatt, unpublished) and was recently implicated in termination by RNAPII (1). Moreover, we recently found that Chd1 is associated with casein kinase II and Spt16/Pob3 (25) and can be effectively cross-linked to the coding regions of a number of genes (Kim et al., unpublished). The dependence of Set2 function on the Paf1 complex and the CTD kinase Ctk1, as well as the many genetic interactions between the SET2 gene and genes encoding known positive elongation factors, strengthens our conclusion that Set2 function is cotranscriptional and that Set2 may act as an activator of elongation.
Surprisingly, synthetic growth defects were obtained when a set2 deletion was combined with deletions of genes encoding all seven components of the Set3 complex (41). The Set3 complex has a negative effect on the expression of certain meiosis-specific genes, perhaps because the Set3 complex contains two histone deacetylases, Hos2 and Hst1, as well as Set3. The synthetic phenotypes generated when a set2 deletion is combined with deletions of genes encoding components of the Set3 complex suggest, however, that the Set3 complex may also have a positive role in transcription by RNAPII. This idea is supported by the recent observation that at least two components of the Set3 complex, Set3 and, surprisingly, Hos2, are recruited to actively transcribed genes (65).
A set2 deletion also generated synthetic growth defects when combined with deletions of genes encoding components of the Set1 complex, COMPASS. COMPASS methylates Lys4 of histone H3, a modification that has been shown to be needed for effective silencing at telomeres and ribosomal DNA loci (7, 24, 33, 36, 46). Interestingly, components of COMPASS as well as histone H3 Lys4 methylation, like Set2 and components of the Paf1 complex, localize to the transcribed regions of various genes and interact genetically with a number of known or suspected elongation factors (5, 48; our unpublished data). Moreover, methylation of Lys4 by COMPASS, like methylation of Lys36 by Set2, also requires components of the Paf1 complex, which associates with COMPASS (26, 37). Genetic interactions among components of three different Set protein-containing complexes imply that these complexes are functionally redundant and may all associate directly or indirectly with RNAPII and function during transcriptional elongation.
Our data and other published studies on COMPASS and Set2 have uncovered a remarkable coordination of RNAPII phosphorylation with histone H3 methylation during transcriptional elongation, as illustrated in Fig. Fig.6.6. COMPASS is recruited specifically to the early transcribed region (26, 37), and this depends on both the Paf1 complex and phosphorylation of the RNAPII CTD on Ser5 by Kin28. The consequence of COMPASS recruitment is histone H3 trimethylation on Lys4 in the same region (37). Upon further elongation by RNAPII, Ser5 phosphorylation declines and is replaced by Ser2 phosphorylation mediated by Ctk1 (10, 23). We found that Set2 is recruited to this region, where it methylates Lys36 of histone H3, and the recruitment of Set2 depends on both the Paf1 complex and phosphorylation of RNAPII by Ctk1 on Ser2, as has also been observed by Li et al. (28) and Xiao et al. (68). Finally, in the region beyond the poly(A) signal, Set2 disappears and histone H3 Lys36 declines in concert with dephosphorylation of RNAPII on Ser2 (10, 23) and the disappearance of the Paf1 complex (Kim et al., unpublished).
Finally, set2Δ bre1Δ and set2Δ lge1Δ double mutants also have synthetic growth defects. Bre1, an E3 ubiquitin ligase, is associated with Rad6 and Lge1, and all three are required for ubiquitination of histone H2B, a modification which is necessary for methylation of histone H3 on Lys4 by COMPASS (13, 19, 58, 67; our unpublished data). Therefore, disruption of Lys4 methylation by deleting genes encoding components of COMPASS or the Bre1/Lge1 complex results in a growth defect when SET2 is also deleted. This further confirms the functional redundancy between the Lys36 and Lys4 methylations on histone H3. Interestingly, SET2 also genetically interacts with HTZ1, which encodes a variant histone, H2A (20). This genetic interaction may predict a role for Htz1 in histone ubiquitination, histone methylation, and/or transcriptional elongation.
We are grateful to A. Aguilera for the gift of the plasmid p416GAL1-lacZ and to B. Strahl and K. Struhl for sharing unpublished data.
N.J.K. was supported by a PGS-B Scholarship Award from the Natural Sciences and Engineering Research Council of Canada (NSERC) and a Doctoral Fellowship from the Canadian Institute of Health Research (CIHR). This research was supported by grants to J.F.G. from the Canadian Institutes of Health Research, the Ontario Genomics Institute, and the National Cancer Institute of Canada with funds from the Canadian Cancer Society.
N.J.K. and M.K. contributed equally to this report.