|Home | About | Journals | Submit | Contact Us | Français|
Swd2, an essential WD repeat protein in Saccharomyces cerevisiae, is a component of two very different complexes: the cleavage and polyadenylation factor CPF and the Set1 methylase, which modifies lysine 4 of histone H3 (H3-K4). It was not known if Swd2 is important for the function of either of these entities. We show here that, in extract from cells depleted of Swd2, cleavage and polyadenylation of the mRNA precursor in vitro are completely normal. However, temperature-sensitive mutations or depletion of Swd2 causes termination defects in some genes transcribed by RNA polymerase II. Overexpression of Ref2, a protein previously implicated in snoRNA 3′ end formation and Swd2 recruitment to CPF, can rescue the growth and termination defects, indicating a functional interaction between the two proteins. Some swd2 mutations also significantly decrease global H3-K4 methylation and cause other phenotypes associated with loss of this chromatin modification, such as loss of telomere silencing, hydroxyurea sensitivity, and alterations in repression of INO1 transcription. Even though the two Swd2-containing complexes are both localized to actively transcribed genes, the allele specificities of swd2 defects suggest that the functions of Swd2 in mediating RNA polymerase II termination and H3-K4 methylation are not tightly coupled.
The maturation of mRNA occurs cotranscriptionally in vivo and is facilitated by the recruitment of mRNA processing factors to the phosphorylated C-terminal domain (CTD) of the largest subunit of RNA polymerase II (RNAP II) (for a review, see reference 40). During transcription, the nascent pre-mRNA is capped at the 5′ end, introns are removed by splicing, and the 3′ end is cleaved and polyadenylated. The processing of the 3′ end promotes transcription termination and nuclear export, and acquisition of a poly(A) tail is essential for the production of functional mRNA.
The cleavage and polyadenylation steps take place within a large complex of proteins that for the most part are highly conserved between yeast and mammals (13). In the yeast Saccharomyces cerevisiae, CF IA, Hrp1/Nab4, and CF II are sufficient for the cleavage reaction in vitro, while specific poly(A) addition requires CF IA, Hrp1, poly(A) binding protein, and the CPF complex, which is composed of CF II, poly(A) polymerase, and additional proteins. A subset of these components are important in causing RNAP II to terminate downstream of poly(A) sites or of genes encoding some snRNAs and snoRNAs (2, 5, 11, 33).
Recent application of a tandem affinity purification strategy allowed several groups to isolate highly purified CPF from yeast whole-cell extract (10, 16, 21, 36). This complex contains the nine previously characterized CPF subunits (references 39 and 62 and references therein), several novel polypeptides (including Glc7, Pti1, Ssu72, and Swd2) with no previous links to the cleavage and polyadenylation machinery, and Ref2, an RNA-binding protein known to stimulate processing at weak poly(A) sites (46). Nedea et al. (36) proposed that these new proteins form a subcomplex of CPF called APT (for associated with Pta1) that is connected to core CPF by Pta1. Two of these APT subunits, Pti1 and Ref2, have minor roles in mRNA processing but are very important for maturation of snoRNA 3′ ends (9, 36, 46, 52) which are cleaved but not polyadenylated. Ssu72, a low-molecular-weight phosphatase (15, 30), is essential for 3′ end formation of both types of RNAP II transcripts (21, 36, 57) and also contributes to accurate initiation of RNAP II transcription (60). Glc7, the only type 1 serine/threonine protein phosphatase in S. cerevisiae, has been implicated in numerous nuclear and cytoplasmic processes, including transcription regulation and mRNA export (17, 56).
The fourth APT subunit, Swd2, is a 37-kDa protein without a proven cellular function. It has six WD repeats, which are domains of about 40 amino acids that usually end with tryptophan-aspartic acid (WD). WD repeat proteins are thought to have circularized β-propeller-like structures, which can interact sequentially or simultaneously with several different proteins (53). The presence of Swd2 in the CPF complex suggests that it may be directly involved in mRNA or snoRNA 3′ end formation or may help coordinate 3′ end processing with other events in mRNA synthesis and utilization.
Swd2 is also part of the Set1, or COMPASS, complex, which is composed of Set1, Bre2, Swd1, Swd2, Swd3, Spp1, Shg1, and Sdc1 (31, 34, 44). This complex, through its catalytic subunit Set1, adds one to three methyl groups specifically to lysine 4 of histone H3 (H3-K4) (6, 31, 34, 44). This function is required for position-dependent gene silencing at telomeres, the mating type locus, and ribosomal DNA (6, 8, 25, 38). Set1-mediated H3-K4 dimethylation in S. cerevisiae is found throughout the genome, while the trimethylated form is most prevalent in the coding regions of active or recently transcribed genes (3, 37, 48; reviewed in references 20 and 51) and may protect active coding regions from deacetylation (3).
The purpose of the work described in this report is to examine the role of Swd2 in the processes of histone methylation, transcription, and mRNA 3′ end formation. Here we demonstrate that Swd2, as a bona fide member of the Set1 complex, is required for global histone H3-K4 methylation and other Set1-regulated events. In spite of its tight association with the CPF complex, Swd2 is not directly involved in pre-mRNA 3′ end processing. However, we show that Swd2 plays a role in RNAP II transcription termination that is not dependent on its function in histone methylation. We propose that Swd2, as part of CPF, helps to coordinate the activities of the transcription and 3′ end processing machineries.
The FY23 (MATaura3-52 trp1Δ63 leu2Δ1) strain was described previously (28). The Degron-Swd2 strain was constructed by transformation of FY23 with the pDegron-Swd2 plasmid linearized at the MscI site. Integration of plasmid DNA at the SWD2 locus generated an open reading frame (ORF) encoding ubiquitin-Arg-temperature-sensitive dihydrofolate reductase (DHFRts)-hemagglutinin (HA)-Swd2. The heterozygous SWD2 deletion strain (BY4741/BY4742 background, swd2::KAN), obtained from the American Type Culture Collection (ATCC), was used to create the haploid strain YHC1 (BY4742 background), with a disrupted chromosomal copy of the essential SWD2 gene and pYCplac33-SWD2 [MATα swd2::KAN pYCplac33-SWD2 (CEN URA3 SWD2)]. The yeast set1Δ strain, MBY1217, and the isogenic wild-type syrain, MBY1198, as well as the strain with the K4R replacement in histone H3 and the isogenic wild type, were kindly provided by David Allis (6). The ref2Δ strain (BY4741 background) was obtained from ATCC. The strain used for the telomere-silencing assays, UCC1001 (MATaade2-101 his3Δ-200 leu2Δ1 trpΔ1 lys2-801 TELadh4::URA3), was kindly provided by Lorraine Pillus (38).
The pNOP-SWD2 plasmid was constructed by cloning the SWD2 ORF into the LEU2 plasmid pNOPPATAIL, which provides two protein A tags followed by a TEV site at the N terminus encoded by the ORF (constructed by Klaus Hellmuth). Plasmids carrying pNOP-SWD2 and truncated versions of swd2 (ΔN1, ΔN2, ΔN3, and ΔC1, as shown in Fig. Fig.1)1) were individually introduced into YHC1. To assess the viability of the mutant constructs, the transformants were then plated on 5-fluoroorotic acid (5-FOA) medium to force loss of the URA3-marked plasmid YCplac-33-SWD2. The pNOP-SWD2 plasmid was also used as template for error-prone PCR. The amplification was done with the oligonucleotides 5′-CGCGAATTCGGGGCTGCAGGAATTCGATATCCCAACGACC-3′ and 5′-CCTGAGAAAGCAACCTGACCTACAGGAAAGAGTTACTCAAG-3′, which anneal to flanking vector sequences in pNOP-SWD2. PCR was carried out with Taq DNA polymerase (Invitrogen) in the reaction buffer provided by the manufacturer. The reaction mixture contained 50 pmol of each primer, 200 μM deoxynucleoside triphosphates, 160 to 320 μM MnCl2, and 4 mM MgCl2. The PCR fragment was then cotransformed with the linearized vector plasmid pNOPPATAIL into YHC1. The resulting Leu+ transformants were forced to lose pYCplac33-SWD2 by growth on 5-FOA medium. Thermosensitive mutants were selected by growing short streaks of the same yeast colonies at 30 or 37°C. Retransformation of pNOP-swd2 plasmids from putative temperature-sensitive cells into the YHC1 strain confirmed temperature sensitivity upon removal of the covering plasmid pYCplac33-SWD2.
Preparation of yeast cell extracts, transcription of [α-32P]UTP-labeled full-length GAL7-1 RNA or precleaved GAL7-9 RNA, and processing assays have been described by Zhao et al. (65).
Total yeast RNA was prepared by the acid-phenol extraction method (54). Northern analyses were performed by separating 20 μg of total RNA per lane on a 1.2% formaldehyde-agarose gel, followed by transfer to a Hybond N+ membrane. Probes were made with the Prime-It II random primer labeling kit (Stratagene), PCR products corresponding to the coding regions of the various genes, and [α-32P]dATP. Northern hybridization was performed as described by Sparks and Dieckmann (54).
Transcription run-on (TRO) analysis was performed as described in Steinmetz and Brow (57). Degron-swd2 and its wild-type isogenic strain, FY23, harboring the G-less cassette TRO plasmids, were used in this study. These plasmids (pG-Leu-CYCds, pG-Leu-CYCpAmax, and pG-Leu-SNR13-CYCds) were kindly provided by David Brow (57).
Primer extension analysis was carried out as described by Nedea et al. (36) using primers specific for sequences downstream of the mature snoRNAs as described previously (58). The probe for U6 was AAAACGAAATAAATCTCTTTGTAAAAC. The amount of reverse-transcribed U6 loaded was adjusted to match the signal given by snoRNA cDNAs.
Yeast genomic DNA was prepared as described previously (14). Ten micrograms of each sample was digested with XhoI and resolved on a 0.75% agarose gel in 1× Tris-borate-EDTA buffer. The gel was then capillary blotted onto a Hybond N+ membrane and hybridized in sodium dodecyl sulfate (SDS)-phosphate buffer (0.5 M sodium phosphate [pH 7.2], 1 mM EDTA [pH 8.0], 7% SDS, 1% bovine serum albumin) at 65°C with an [α-32P]dATP-labeled telomere-specific PCR fragment as described by Roguev et al. (44).
Isogenic SWD2, swd2-1, swd2-2, swd2-3, swd2-5, swd2-6, and ref2Δ strains were transformed with plasmid pJS325 (INO1-lacZ), and the activity of β-galactosidase produced in each strain was assayed as described previously (64). For the promoter-targeting experiments, the SWD2 coding sequence was amplified by PCR and cloned into pBTM116 (provided by S. Fields) to give C-terminal LexA DNA-binding domain fusion plasmid pBTM116-SWD2. Plasmid pM1175 (LexA-SAP30) was a gift from M. Hampsey (64). β-Galactosidase assays were performed as previously described (64). Briefly, pBTM116, pBTM116-SWD2, or pM1175 was transformed into the wild-type strain W303 (MATaade2-1 can1-100 his3-11,15 leu2-3,112 trp1-1 ura3-1) along with the LacZ reporter construct (pCK30; provided by K. Johnson) that contained one LexA operator site upstream of the CYC1 promoter. The amount of repression was calculated by using the vector only (pBTM116) as the comparison.
The SWD2 ORF was cloned into pHMTc (a gift from J. Williamson) and used to express Swd2 in E. coli BL21(DE) pLysS as a fusion protein with the His6 and maltose binding protein (MBP) epitopes at the N terminus. The recombinant protein was denatured and renatured with a protein refolding kit (Novagen). In vitro-translated proteins were generated with the TNT rabbit reticulocyte lysate system (Promega) in the presence of [35S]methionine. For pull-downs, His6-MBP-Swd2 was prebound to amylose resin (New England Biolabs) and then incubated with in vitro-translated 35S-labeled proteins for 2 h at 4°C. The amylose resin was washed four times with IP-150 buffer (21). Proteins were eluted in 15 μl of SDS sample buffer and resolved by SDS-10% polyacrylamide gel.
For Western blot analysis of histone H3-K4 methylation, yeast extract was prepared as described by Briggs et al. (6). Approximately 50 μg of extract was resolved by SDS-15% polyacrylamide gel electrophoresis, transferred to a polyvinylidene difluoride membrane, and probed with the α-di-Me(Lys4)H3 rabbit antibody (a gift from David Allis) or the α-tri-Me(Lys4)H3 rabbit antibody (Abcam). Protein A-tagged proteins were detected by using the PAP (peroxidase-antiperoxidase-soluble complex) antibody (Sigma).
Plasmid pYES-REF2 was constructed by cloning the REF2 ORF downstream of the GAL1 promoter in the pYES plasmid (Invitrogen) and introduced into isogenic SWD2, swd2-1, swd2-3, and swd2-5 strains. Ura+ transformants were deposited onto complete medium lacking leucine and uracil and containing 2% galactose as the sole carbon source. Duplicate plates were incubated at 30 or 37°C.
To assess the role of Swd2 in S. cerevisiae, we sought to create a growth-defective swd2 mutant that could be tested for effects on RNA 3′ end formation and H3-K4 methylation. For this analysis, we fused a protein A tag in frame with the amino terminus of the wild-type Swd2 protein. Cells containing only this version of Swd2 have normal growth (data not shown). Swd2 belongs to a subgroup of the WD repeat superfamily in which nearly all of the protein is composed of WD motifs. We found that swd2 proteins with truncations from the N terminus or C terminus, as shown in Fig. Fig.1A,1A, could not support cell growth in the absence of wild-type Swd2. The mutants also did not cause a dominant-negative effect on growth of wild-type cells (data not shown). Immunoblot analysis using antibodies to detect the protein A tag confirmed the presence of all four truncated proteins when expressed in wild-type cells (data not shown), eliminating protein instability as a cause of the lethality. The results with the truncations encompassing the first or the sixth WD repeat of Swd2 are perhaps not unexpected, since removal of any part of the WD repeat portion might disrupt the propeller structure and consequently destroy the biological activity of the protein. However, the lethality of swd2ΔN1 suggests a critical role for the short N-terminal part that lies outside of the WD repeats.
To further explore the function of Swd2, we used error-prone PCR mutagenesis to generate conditionally lethal mutations in the SWD2 gene. Five different thermosensitive alleles were isolated. At 37°C, swd2-3 and swd2-5 strains were inviable and the swd2-1 strain barely grew, while swd2-2 and swd2-6 strains showed slower growth than the wild-type strain (Fig. (Fig.1B).1B). All mutants were inviable at 39°C (Fig. (Fig.1B),1B), but none of the mutants were cold sensitive (data not shown). DNA sequencing identified multiple point mutations in all five alleles that would result in amino acid substitutions (data not shown), so it was not possible to ascribe defects to particular domains of Swd2.
The mutants were also tested for their ability to grow on media containing various chemicals (Fig. (Fig.1C).1C). The swd2-1 mutant was the only strain that was sensitive to caffeine, an indication of weakened cell walls (7) and defects in mitogen-activated protein kinase pathways (19), and the swd2-1 mutant showed the greatest sensitivity to 6-azauracil, a phenotype generally associated with transcriptional elongation defects (19). The swd2-3 mutant also grew more slowly on this medium. The swd2-1 strain, as well as a strain containing an unmodifiable histone H3-K4R replacement, exhibited significant growth inhibition when grown in 0.9 M NaCl, suggesting a defect in responding to osmotic shock. The swd2-1, swd2-3, swd2-5, and H3-K4R strains also grew poorly in the presence of hydroxyurea, which prevents DNA synthesis by inhibiting ribonucleotide reductase. It has been reported that null mutants missing any one of the six nonessential subunits of the Set1 complex also show sensitivity to hydroxyurea (31), which is consistent with a proposed role for Set1 in Rad53-dependent DNA repair pathways (50).
We also generated a heat-labile form of Swd2 by integrating a heat-inducible degradation signal (Arg-DHFRts-HA) in frame with the Swd2 coding sequence using what is known as the Degron strategy (12). Western blot analysis of extracts from cells expressing the Degron-Swd2 fusion indicated that Swd2 is not detectable after 30 min of incubation at 37°C (Fig. (Fig.2B).2B). In accord with the Swd2 gene being an essential gene, these cells do not continue to grow at this temperature (data not shown).
Recruitment of Swd2 to CPF requires the Ref2 subunit, as Swd2 does not copurify with CPF if cells have a deletion of the REF2 gene (36). We used a recombinant MBP-Swd2 fusion protein in pull-down assays to determine if other CPF or CF IA subunits interacted directly with Swd2. Positive interactions of Swd2 were detected with 35S-labeled in vitro-translated Cft1, Pfs2, and Pcf11, but not with Brr5/Ysh1, Fip1, Mpe1, Pap1, Pta1, Pti1, Rna14, Rna15, Ssu72, and Swd2 itself (Fig. (Fig.2A).2A). This analysis suggests that, while Ref2 may be the primary contact of Swd2 with CPF, Swd2 has the potential to interact with Cft1/Yhh1 and Pfs2, CPF subunits that also contain β-propeller repeats (11, 39). Furthermore, Swd2 may provide another bridge to the CF IA factor through Pcf11, in addition to the one previously reported for Pfs2, which interacts with Rna14 of CF IA (39).
The presence of Swd2 in the CPF complex and its interactions with subunits essential for pre-mRNA 3′ end processing prompted us to test if Swd2 has a role in this processing event. Extracts prepared from Degron-Swd2 cells were examined for their ability to cleave and polyadenylate a synthetic, radiolabeled GAL7 precursor RNA or to polyadenylate a GAL7 substrate that ended at the cleavage site. Surprisingly, in vivo depletion of Swd2 had no effect on processing of mRNA 3′ ends in vitro, even after 4 h at 37°C, except for a slightly longer poly(A) tail at the later time points (Fig. (Fig.2C2C).
Wild-type cells overexpressing the Pti1 subunit of the APT subcomplex have been shown to give extract defective for poly(A) addition (9), leading to the proposal that the role of Pti1 is to suppress CPF polyadenylation activity on snoRNA transcripts. We found that overexpression of Swd2 had no effect on in vitro poly(A) addition activity (data not shown), ruling out this possible function for Swd2. These results indicate that Swd2 does not have a direct role in cleavage and polyadenylation of the mRNA precursor.
At least one subunit of CPF, Cft1/Yhh1, contacts the CTD of RNAP II and is important for efficient termination of transcription downstream of poly(A) sites (11). The APT subunits Ref2 and Pti1 appear to have a less significant role than Cft1 in this type of termination (10, 21, 36, 46, 52, 57), and mutations in CPF components specific for the poly(A) addition step, such as Fip1 and Pap1, have no impact on termination (5).
In a TRO assay using permeabilized yeast cells developed by Birse et al. (5), we found that Degron-Swd2 cells analyzed 45 min after shutoff of Swd2 synthesis showed a 1.5- to 1.9-fold increase in transcription beyond the CYC1 poly(A) site, compared to the same cells without Swd2 depletion (data not shown). To further investigate this possible defect in transcription termination, we used a simplified version of the TRO assay developed by Steinmetz and Brow, in which the level of transcription extending through two G-less cassettes of different sizes is measured directly after T1 RNase treatment and denaturing gel electrophoresis (57) (Fig. (Fig.3A).3A). The effect of inserting elements that direct 3′ end formation between the two G-less cassettes can be determined by comparing the ratios of radioactivity incorporated into the 262-nucleotide (nt) upstream product and the 132-nt downstream product. To insure thorough depletion of Swd2, we extended the time after shutoff to 90 min. In the absence of Swd2, transcriptional read-through downstream of the CYC1 poly(A) site is increased fivefold when Swd2 is depleted, in comparison to results obtained with the isogenic wild-type strain exposed to the same culture conditions (Figs. 3B and C). A change of similar magnitude was observed when a strain with mutant SSU72 (ssu72-G33A), which encodes another subunit of the APT complex, was analyzed by this TRO assay (57).
Many recent studies have shown that the mRNA 3′ end processing machinery is also required for proper formation of the 3′ ends of some snRNAs and snoRNAs, which are transcribed by RNAP II but are not polyadenylated (9, 33, 36, 57). The TRO assay showed that removal of Swd2 caused a 2.6-fold increase in transcription beyond the SNR13 3′ end signal, suggesting that termination downstream of this element is also defective (Fig. 3B and C). Another indication of a defect in synthesis of SNR13 snoRNAs is the appearance of transcripts from this gene that extend into the downstream protein-encoding TRS31 gene (58), as diagrammed in Fig. Fig.3D.3D. When Northern blots were analyzed with a probe against the TRS31 gene, we observed a complete shift to the longer SNR13-TRS31 transcripts after Degron-Swd2 cells were grown for 45 or 90 min at 37°C (Fig. (Fig.3E),3E), confirming a defect in termination and subsequent processing of the SNR13 transcripts. The accumulation of the SNR13-TRS31 transcripts suggests that they are properly polyadenylated at the TRS31 poly(A) site and not degraded, supporting the conclusion from our in vitro results that Swd2 is not required for mRNA 3′ end processing.
We next tested SNR13 termination in the three alleles of swd2 (swd2-1, swd2-3, and swd2-5) that produced the strongest thermosensitive growth phenotypes. When the swd2-3 and swd2-5 mutants were grown at 37°C for 3 h, normal TRS31 mRNAs disappeared and extended transcripts accumulated (Fig. (Fig.4A),4A), as they did in the Swd2-depleted cells. These mutants also showed some read-through transcripts at 30°C. There was no difference in the wild-type strain at the two temperatures and only a slight increase in swd2-1 cells. From reverse transcription analysis of extended transcripts using downstream primers, we found that the swd2-5 allele also showed significant read-through compared with the wild type at the SNR33, SNR50, and SNR71 loci (Fig. (Fig.4B).4B). However, in other snoRNA genes (SNR3, SNR45, SNR39b, and SNR128) which are affected by mutation of NRD1, PTI1, SEN1, or SSU72 (9, 36, 58), no extended transcripts were detected (data not shown).
We also examined how the swd2-5 mutant allele affected transcription termination in two other mRNA genes, CUP1 and GLK1. Previous studies have demonstrated the accumulation of extended transcripts from CUP1 as a consequence of defects in components of the core cleavage and polyadenylation machinery, such as Pcf11, Rna14, and Rna15 of CF I (18) and Ssu72 of CPF (10, 15), or when CUP1 was transcribed by RNAP II lacking the CTD (29). Ganem et al. (15) also observed significant read-through of GLK1 transcripts into the downstream YCL039w gene in the ssu72-ts69 mutant. By Northern analysis, we observed only the normal-size CUP1 and GLK1 mRNAs in swd2-5 cells after a shift to 37°C for 3 h (Fig. (Fig.4C).4C). The results described in this section indicate that Swd2 helps direct RNAP II termination in response to some, but not all, poly(A) and snoRNA signal elements.
Ref2 is needed for 3′ end formation of many snoRNA transcripts (9, 36) and recruits Swd2 to the APT complex (36). As shown in Fig. Fig.5A,5A, we found that overexpression of REF2 strongly suppressed the thermosensitive growth defects of the mutant swd2 alleles that exhibited the SNR13 termination defect (swd2-3 and swd2-5) but did not rescue swd2-1. A thermosensitive mutant ref2 has been shown to be defective in SNR13 3′ end formation (9), but in our strain background, a deletion of REF2 does not cause read-through into the TRS31 gene (Fig. (Fig.4A,4A, lane 9). Nevertheless, REF2 overexpression could partially rescue the termination defect of swd2-3 and almost completely rescue that of swd2-5 (Fig. (Fig.5B),5B), consistent with Swd2 and Ref2 having related functions.
Our observation that swd2 mutants exhibit chemical sensitivities similar to those produced by other mutations affecting H3-K4 methylation (Fig. (Fig.1C)1C) (31) lends credence to the idea that Swd2 is a bona fide subunit of the Set1 methylase. We tested this hypothesis more directly by examining whole-cell extracts from swd2 mutant cells, H3-K4R cells, and the corresponding wild-type isogenic strains for the presence of histone H3 di- and trimethylated at K4. As expected, the H3-K4R cells have no detectable H3-K4 dimethylation (Fig. (Fig.6A).6A). The swd2-1 and swd2-3 cells have a significantly decreased amount of this modification, while the swd2-2, swd2-5, and swd-6 mutants have wild-type levels. H3-K4 trimethylation is almost gone in swd2-1 and swd2-3 and is slightly reduced in swd2-5 and swd2-6 (Fig. (Fig.6B6B).
Deletion of SET1 and loss of H3-K4 methylation are associated with loss of silencing of telomere-proximal genes, which is detected by a very sensitive assay using a URA3 reporter gene placed adjacent to the telomere on the left arm of chromosome VII (38). Because of fluctuations in the boundaries of silent chromatin, some cells in a wild-type population express the URA3 gene and can grow on medium lacking uracil, while in others, transcription is repressed, allowing the cells to grow in the presence of the suicide substrate 5-FOA (Fig. (Fig.6C).6C). When the wild-type SWD2 gene was replaced with the swd2 mutant alleles in a strain containing the URA3 telomeric reporter gene, the swd2-1 and swd2-5 cells grew on 5-FOA. The extent of growth relative to that for wild-type cells is similar to that seen on media lacking 5-FOA. In contrast, the swd2-3 strain, which is the most severely impaired in H3-K4 methylation, shows no growth on 5-FOA, indicating complete derepression of URA3.
Another consequence of a silencing defect due to deletion of SET1 is a shortening of telomeres by about 50 bp (38). We asked whether telomeres were similarly affected in the swd2 mutants. This shortening can be detected by Southern blot analysis of an XhoI digest of yeast chromosomal DNA, which releases the telomeres as a unique species migrating between 1,000 and 1,500 bp. As shown in Fig. Fig.6D,6D, a reduction in telomere size, similar to the decrease reported for set1Δ cells (38), is observed for swd2-3 cells, but not for wild-type cells or the other mutant strains. These results suggest that H3-K4 methylation in swd2-1 and swd2-5 cells is sufficient to support silencing but is not sufficient in the swd2-3 mutant.
In contrast, H3-K4 methylation is not important for termination of SNR13 transcription. This conclusion is most evident from the normal TRS31 transcripts in the set1Δ strain (Fig. (Fig.4A,4A, lanes 10 and 11) and is supported by the differential effects of the swd2-1 and swd2-5 mutations on termination and H3-K4 methylation (Fig. (Fig.4A4A and and6).6). Furthermore, deletion of REF2 and the consequent release of Swd2 from CPF do not perturb H3-K4 methylation (Fig. 6A and B).
Several proteins containing 40-amino-acid WD repeats, such as Tup1 (reference 24 and references therein), Hir1 (55), Met30 (61), and Wtm1 and its homologs (42), have transcription repressor activity. To explore whether Swd2 might have similar activity, we investigated whether the swd2-1, swd2-3, and swd2-5 alleles could cause transcription derepression using the INO1-lacZ reporter plasmid. Transcription from this promoter is repressed when cells are grown in the presence of inositol (64). In strains carrying an INO1-lacZ reporter, β-galactosidase levels were enhanced 2.8- and 3.5-fold in swd2-1 and swd2-3 cells, respectively, but were not changed in the swd2-5 strain, in comparison to the isogenic wild-type strain (Fig. (Fig.7A).7A). A similar relief of repression was observed in the set1Δ strain, suggesting an involvement of the Set1 histone methylase complex.
We also examined whether recruitment of Swd2 to a promoter could cause repression. For this experiment, we used a LexA-Swd2 fusion and a reporter containing the LexA operator, as diagrammed in Fig. Fig.7B.7B. The CYC1-lacZ or lexAop-CYC1-lacZ reporter plasmid was introduced into a wild-type strain harboring a plasmid that expresses LexA, LexA-Swd2, or LexA-Sap30. The LexA-Swd2 fusion resulted in a 3-fold repression that was dependent on LexAop (Fig. (Fig.7B),7B), which can be compared to the 5.3-fold repression seen with LexA-Sap30, a value similar to that reported previously for this subunit of the Rpd3 histone deacetylase complex (64). These findings indicate that Swd2 can affect transcription at the level of initiation.
Swd2 is an essential yeast protein present in the CPF complex necessary for mRNA 3′ end processing and in the Set1 complex that methylates histone H3-K4. In this paper, we establish that Swd2, as a subunit of CPF, is not involved in the cleavage and polyadenylation of mRNA transcripts. Conditional removal of Swd2 from cells had no effect on either of these steps. However, temperature-sensitive mutations or depletion of Swd2 causes termination defects in the CYC1 gene and some snoRNA genes transcribed by RNAP II. The swd2-3 and swd2-1 mutants, but not the swd2-5 mutant, were significantly defective in global histone H3-K4 methylation, consistent with Swd2 being an important component of the Set1 methylase complex.
Our results show that Swd2 is necessary for many activities that have been associated with Set1, such as H3-K4 methylation, silencing at telomeres, and negative regulation of transcription. The precise role of Swd2 in the Set1 complex remains unclear but could include activities such as complex assembly, substrate recognition, and directing the methylation activity to specific regions of chromatin. For example, H3-K4 trimethylation is focused to the 5′ end of actively transcribed genes (37). The pleiotropic growth defects of the swd2-1 mutant, such as 6-azauracil, caffeine, high-salt, and hydroxyurea sensitivities, are reminiscent of phenotypes associated with mutations in the Paf1 complex (4), which is important for transcriptional elongation and Set1 recruitment to genes transcribed by RNAP II (26). Swd2 may work with the Paf1 complex in recruiting Set1 to chromatin at the 5′ end of the gene or modulating its activity in this region. This role would be similar to that of other WD repeat proteins, such as Hat2, which increases the affinity of the Hat1 histone acetyltransferase for histone tails (41), or RbAp48, a subunit of the chromatin assembly factor CAF-1, which holds out histones for DNA to wrap around (22).
Deletion of the SET1 gene has been shown to alleviate the repression of DNA repair genes (50). The exact role of Set1 in this regulation is not clear, but the mechanism seems to involve inactivation of the Rfa2 repressor by phosphorylation. In our study, we have found that the swd2-1and swd2-3 mutations, as well as loss of SET1, caused derepression of INO1 transcription. Furthermore, direct recruitment of Swd2 to a promoter led to repression. We do not know the reason for these effects on transcription and whether it involves alterations in chromatin structure at the promoter. In this regard, it is interesting that Set2, which methylates lysine 36 of histone H3, is required for repression of GAL4 transcription (27) and that, similar to Swd2, Set2 represses transcription when tethered to a reporter gene (59). The mechanism associated with Swd2 may be related to a recent finding that human homologs of the Set1, Bre2/Ash2, and Swd3 histone methyltransferase subunits are tethered to the Sin3 histone deacetylase by HCF-1, a regulator of cell proliferation (63). Nakamura et al. (35) found that a different human H3-K4 methyltransferase, ALL-1, was part of a supercomplex that included histone deacetylase and acetylase activities, ATP-dependent nucleosome remodeling factors, homologs of the yeast Swd1 and Swd2 proteins, and some subunits of CPF, most notably Pta1. These complexes are thought to act at the promoter to activate transcription, and it is possible that the WD repeat proteins are important for their assembly. If a similar complex existed in yeast, mutation of Swd2 might upset the balance of activities that promote or repress transcription.
Transcription termination in mRNA-encoding genes is thought to be dependent on the association of CPF and CF IA with elongating RNAP II and the recognition of the poly(A) signal by these factors (2, 5, 11). In contrast, termination in some snoRNA genes requires the proteins Nrd1, Nab3, and Sen1, as well as CF IA and the APT subcomplex of CPF, but not necessarily the activity of core CPF (2, 5, 33). Like CPF and CF IA, the snoRNA-specific factors are thought to be recruited to the transcriptional complex by direct interactions with the CTD of RNAP II and to cooperate to provide an entry site for exonucleolytic trimming of snoRNA 3′ ends. In accord with its role as a CPF component important for transcription termination, Swd2 can be found at the 3′ ends of mRNA genes by chromatin immunoprecipitation assays (36).
The temperature-sensitive growth defects of the swd2-3 and swd2-5 mutants, but not that of the swd2-1 mutant, could be rescued by overexpression of the Ref2 subunit of CPF, in agreement with the known functions of Ref2 in recruiting Swd2 to CPF and facilitating RNAP II termination in snoRNA genes (9, 36). The analysis of the architecture of the CPF holocomplex presented by Nedea et al. (36) showed that deletion of REF2 had no effect on CPF composition except for release of Swd2 and Glc7. Their finding suggests that Swd2 rests on the surface of CPF, consistent with a role in bridging CPF to other entities in the nucleus such as the transcriptional complex or one of the proteins dedicated to snoRNA 3′ end formation. The in vitro interaction between Swd2 and Pcf11 that we have observed indicates that Swd2 might also link CPF to CF IA. The interaction with the CPF subunits Cft1 and Pfs2 that we found suggests a role in CPF assembly. Given that Swd2 is not required for 3′ end processing, these contacts are probably not important for the integrity of core CPF but perhaps help to hold the APT subcomplex in the holoenzyme.
As a WD repeat protein with the potential to interact with multiple proteins, Swd2 would be an ideal adaptor for coordinating the assembly of a functional processing complex with RNAP II transcription termination. Such an adaptor would not necessarily be needed for 3′ end processing, just as the CTD-interacting domain of Pcf11 is needed for termination but not cleavage and polyadenylation (47). Our data with the swd2-5 mutant showed transcriptional read-through in the SNR13, SNR33, SNR50, and SNR71 genes, but not in mRNA-encoding genes such as CUP1 and GLK1 or the snoRNA genes SNR3, SNR45, SNR128, and SNR39b, which are affected to various degrees by mutations in CPF, CF IA, or snoRNA-specific factors (9, 10, 15, 18, 36, 58). Thus, either Swd2 is not needed for all termination events or the swd2-5 mutant protein retains sufficient function for termination in these loci. Furthermore, upon Swd2 depletion, transcription beyond the processing sites in the TRO assay does not reach the level of that seen in the absence of these processing elements, implying that Swd2 is only one channel by which 3′ end processing complexes communicate with the elongating RNAP II.
Swd2 may be an example of the cell using the same protein for two different purposes, and the differential behavior of the swd2-1 and swd2-5 mutants suggests that the function of Swd2 in transcription termination is not strongly linked to its role in H3-K4 methylation. This interpretation is supported by the normal SNR13 termination seen when SET1 is deleted and by the wild-type levels of H3-K4 methylation when REF2 is absent. However, a very recent study has demonstrated that the H3-K4 modification is necessary for the recruitment of the chromatin-remodeling ATPase Isw1 to some yeast genes (49). This finding is very intriguing in light of other studies showing that Isw1 is part of a checkpoint mechanism coordinating RNAP II elongation and termination (1, 32, 49). Furthermore, in humans a large complex called ALL-1 has been found to have both H3-K4 methylase activity and homologs of CPF subunits (35). Thus, we cannot rule out the possibility that an interaction of CPF with elongating RNAP II, paired with a simultaneous association of Swd2 with CPF and Set1, might help direct H3-K4 methylation to the chromatin of particular genes, and in this way affect downstream events in mRNA synthesis.
In the fission yeast Schizosaccharomyces pombe, there are two Swd2 homologs (23, 43, 45). Swd2.1 is found only in the Sp-Set1 complex and is necessary for H3-K4 methyltransferase activity. Swd2.2 is part of the Sp-CPF complex, but its role in CPF-related activities has not been investigated. Neither protein is essential in S. pombe, but the consequence of disrupting both genes is not known. Why is Swd2 required for cell viability in S. cerevisiae? The unique Set1 methylase is not essential in either species, though S. cerevisiae cells with SET1 deleted grow more slowly than wild-type cells (6, 31, 38). Thus, a combination of poor methylation and poor termination might be more than the cells can tolerate. Alternatively, the transcription termination defects alone may be responsible for the lethality when Swd2 is lost. The viability of the swd2-3 and swd2-5 mutants at 37°C when Ref2 is overexpressed, together with the recovery of SNR13 termination, is consistent with this interpretation. However, swd2-1, which is defective for H3-K4 methylation but only minimally impaired for SNR13 termination, is not rescued by an increase in Ref2. Moreover, swd2-1 is the only thermosensitive swd2 mutant allele that we isolated with sensitivity to caffeine and high salt, implying that some genes are particularly sensitive to this allele. These results suggest that a function of Swd2 in addition to histone methylation and transcription termination may yet be uncovered.
We thank M. Hampsey, D. Allis, L. Pillus, A. Johnson, J. Williamson, and D. Brow for sharing reagents with us; K. Havens for his help with the REF2 overexpression experiments; P. Nagy and M. Cleary for sharing data before publication; and M. Hampsey and A. Zhelkovsky for helpful discussions and for their critical reading of drafts of the manuscript.
This work was supported by NIH grants GM068887 and GM041752 to C.M.