Effectors of Lysine 4 Methylation of Histone H3 in Saccharomyces cerevisiae Are Negative Regulators of PHO5 and GAL1–10 *

Post-translational modifications of histone amino-ter-minal tails are a key determinant in gene expression. Histone methylation plays a dual role in gene regulation. Methylation of lysine 9 of histone H3 in higher eukaryotes is associated with transcriptionally inactive heterochromatin, whereas H3 lysine 4 methylation cor-relates with active chromatin. Methylation of lysine 4 of H3 via Set1, a component of the Saccharomyces cerevisiae COMPASS complex, is regulated by the transcriptional elongation Paf1-Rtf1 and histone ubiquitination Rad6-Bre1 complexes, which are required for the expression of a subset of genes. This suggests that lysine 4 methylation of histone H3 may play an activating role in transcription; however, the mechanism of Set1 function remains unclear. We show here that H3 lysine 4 methylation also negatively regulated gene expression, as strains without Set1 showed enhanced expression of PHO5 , wherein chromatin structure plays an important transcriptional regulatory role. Di- and trimethylation of H3 lysine 4 was detected at the PHO5 promoter, and a strain expressing a mutant version of histone H3 with lysine 4 changed to arginine, (which cannot be methylated) exhibited PHO5 derepression. Moreover, PHO5 was derepressed in strains that lacked components of either the Paf1-Rtf1 elongation or Rad6-Bre1 histone ubiquitination complexes. Lastly, PHO84 and

In eukaryotes, DNA is packaged with histone proteins to form nucleosomes that are further condensed into higher order chromatin structures. Generally, this compaction serves as a barrier for the binding of factors that elicit important cellular processes such as transcription and DNA replication. Thus, genes found in heavily condensed regions, such as heterochromatin, are transcriptionally silent. Expression of genes located in euchromatic regions, which are typically less compacted, is also regulated by chromatin structure.
Post-translational modifications of the amino-terminal tails of histone proteins are a key determinant in defining active and repressed chromatin. These modifications may alter chromatin structure directly by affecting histone-DNA and histone-histone interactions (1). They also allow for the recruitment of transcriptional activators or repressors. Acetylation of histone H3 at lysines 9 and 14 is strongly correlated with transcriptionally active and accessible chromatin. Histone methylation is associated with both active and repressed chromatin states. In eukaryotes other than budding yeast, heterochromatic silencing is marked by methylation of histone H3 at lysine 9. Conversely, euchromatic regions are associated with histone methylation of histone H3 at lysine 4 (H3 Lys-4) (2). Set1 is the catalytic subunit of a large complex named COMPASS (3), which is responsible for all H3 Lys-4 methylation observed in yeast (4). Set1 is required for full activation of a subset of euchromatic genes, including RAM2, HAS1, INO1, PPH3, and MET16 (5,6). Paradoxically, despite its presumed role in transcriptional activation, the absence of Set1 or other components of COMPASS leads to loss of ribosomal DNA (4, 7) and telomeric (8) silencing.
Set1-dependent methylation requires histone ubiquitination of lysine 123 of histone H2B via the ubiquitin-conjugating Rad6-Bre1 complex (9). Set1 is still recruited in the absence of Rad6; however, no resulting H3 Lys-4 methylation is observed (10). This is the first evidence that a modification of one histone regulates the trans-modification of another histone. Recent reports have also indicated that Set1 methylation is associated with transcriptional elongation (11). The Paf1-Rtf1 complex, which associates with RNA polymerase II, is required for H3 Lys-4 methylation as well as recruitment of the COMPASS complex (12). A paf1 null strain shows no detectable histone ubiquitination; however, Rad6 is still recruited (13). As seen with set1⌬ mutants, strains with deletions of PAF1, RTF1, or RAD6 show loss of telomeric silencing (9,12,14).
In this report, we explored the role of Set1 in the transcriptional regulation of select genes of the phosphate-repressible PHO cluster (15). We found that loss of Set1 led to increased levels of expression of PHO5, coding for the repressible acid phosphatase (rAPase), 1 under both repressed and active conditions. The expression of the high affinity phosphate transporter PHO84 is also higher in set1⌬ than in SET1 ϩ strains. Deletions of genes encoding critical components of the Rad6-Bre1 and Paf1-Rtf1 complexes, which are required for Set1-dependent methylation, also exhibited derepression of PHO5. Finally, we observed derepression of the GAL1-10 locus in strains lacking SET1. Our results suggest that methylation at Lys-4 of histone H3 may provide an activating signal at some genes although being a repressive one at others. Further, our data are consistent with a model in which H3 Lys-4 methylation establishes a negative feedback loop on basal transcription, perhaps enforcing a temporary transcriptional elongation checkpoint (16,17).

EXPERIMENTAL PROCEDURES
Yeast Strains and Genetic Manipulations-The genotypes of the Saccharomyces cerevisiae strains used are listed in Table I. The SET1 open reading frame was completely replaced in the diploid strain CCY694 (18) with the kanMX4-selectable marker by a PCR-based method using the plasmid pRS400 as described by Brachmann et al. (19). Gene replacement was confirmed by PCR, and the resulting diploid was sporulated and tetrads were dissected to obtain two independent SET1 ϩ (CCY1467 and CCY1468) and set1⌬ haploid segregants (CCY1471 and CCY1472). The wild-type strain YPH500⌬L (20) as well as the wild-type and histone H3 mutant strains WZY43 and JDY2 (generous gifts from Dr. Sharon Dent) were described previously (21). MBY1198 and MBY1217 are gifts from Dr. Mary Bryk and are described elsewhere (7). Strain CCY2895 used in Fig. 4 was constructed by replacing SET1, as described above, in strains BY4741 and BY4742 and crossing to make the homozygous diploid deletion strain. All other yeast deletion strains were obtained from the homozygous deletion panel (Research Genetics).
Growth Conditions-For PHO5 expression experiments, strains were pregrown in minimal medium containing 0.7 g of yeast nitrogen base (without ammonium sulfate, inorganic phosphate (P i ), and amino acids) (Bio 101), 2 g of glutamine (Sigma), 20 g of dextrose (Fisher), and 3.9 g of 2-N-morpholino ethanesulfonic acid (JT Baker), pH 5.5, per liter, which was supplemented with 13.4 mM KH 2 PO 4 . The cells were then washed and resuspended in minimal medium containing either 13.4 mM KH 2 PO 4 or 13.4 mM KCl and incubated at 30°C with shaking for 6 h. PHO5 activity was measured by either rAPase activity assays or Northern hybridization as described by Neef and Kladde (18). In other experiments, cells were grown in rich YPD medium supplemented with 13.4 mM KH 2 PO 4 (YPPD) overnight at 23°C without shaking. For GAL1-10 experiments, strains were pregrown overnight in YPD and then resuspended in YPD or YP galactose (YPG) ϩ 0.5% glucose and incubated at 30°C with shaking for 4 h. Northern hybridization probes were generated by PCR amplification using the oligonucleotides listed in Table II. Transcript levels were quantified by Storm 860 phosphorimaging analysis.
Chromatin Immunoprecipitation (ChIP) Assay-To analyze histone H3 Lys-4 methylation levels, antibodies specific for di-and trimethyl-ated forms of histone H3 Lys-4 were used to immunoprecipitate formaldehyde cross-linked chromatin from strains MBY1198 and MBY1217 grown in YPD as described previously (7). Quantitative PCR amplification of input and immunoselected DNA was performed using primers ADO236 and LFO740 as described elsewhere (22).

Deletion of SET1 Leads to Increased Levels of PHO5 Expression-
To determine the role of histone H3 Lys-4 methylation at PHO5, we analyzed PHO5 expression levels in SET1 ϩ and set1⌬ strains under both repressive and activating conditions. To distinguish rAPase (Pho5) from constitutive acid phosphatase (Pho3) activity as well as to avoid potential cross-hybridization in Northern analyses, we used strains in which the entire coding sequence of PHO3 was deleted (18). Under the repressive conditions of minimal medium supplemented with P i , set1⌬ strains reproducibly showed significantly higher levels of rAPase activity than SET1 ϩ cells (Fig. 1A). This increased rAPase activity correlated with the higher PHO5 mRNA levels in set1⌬ cells, suggesting that the derepression was because of increased transcription (Fig. 1B). We also observed enhanced PHO5 expression in fully activating conditions of minimal medium lacking P i (Fig. 1C). Although the fold increase was lower, the increase by nearly 800 Miller units was substantial. Similarly, under no-P i conditions, a modest but reproducible increase in PHO5 transcript was observed (Fig. 1D).
To better quantify the level of derepression due to the deletion of SET1, we grew cells under conditions of higher basal expression in rich medium supplemented with P i (YPPD) at 23°C. Under these conditions, rAPase activity was ϳ10-fold higher than when grown in minimal medium containing P i at 30°C (compare the levels observed for SET1 ϩ cells in Fig. 1A with those observed in Fig. 2A). Nevertheless, PHO5 expression was further derepressed in cells lacking Set1 (Fig. 2, A and  B). Additionally, a strain containing a mutated version of histone H3 with Lys-4 replaced by arginine (K4R mutant), which can no longer be methylated, showed increased expression of PHO5 (Fig. 2C). It is important to note that the H3 Lys-4 strain, in which endogenous genes encoding histones H3 and H4 were deleted and complemented by a single copy episome containing histones H3 and H4, exhibited a significantly higher basal expression of PHO5 when compared with a congenic SET1 ϩ strain. It is possible that this is because of the lower overall expression levels of these core histones, which have been shown previously to derepress PHO5 (23). Our results suggested that histone H3 Lys-4 methylation via Set1 plays a role in the repression of the euchromatic gene PHO5.
Methylation of Lys-4 of Histone H3 Is Present at the PHO5 Promoter-To see whether H3 Lys-4 methylation directly affects PHO5, we investigated the methylation state of histone H3 at the PHO5 promoter. ChIP analysis was performed using TAAGAAATAGTCCATTTGAGATTT CCO1139 a Primer contains a 5Ј tail with core T7 promoter sequence.

Set1 Negatively Regulates PHO5 and GAL1-10 Transcription
antibodies specific for di-and trimethylated lysine 4 of histone H3 (6). When cells are grown in YPD, which is limiting for P i and hence partially activates PHO5 (18), considerable amounts of both di-and trimethylated forms of H3 Lys-4 are present at the PHO5 promoter (Fig. 3). This enrichment is abolished in the absence of Set1, a result that confirms the presence of H3 Lys-4 methylation at PHO5 (24) and suggests that Set1 is recruited to the PHO5 promoter. Upstream Effectors of Set1-dependent H3 Lys-4 Methylation Also Regulate PHO5-Histone of histone H2B at lysine 123 by the Rad6-Bre1 complex is required for Set1 to methylate histone H3 Lys-4. Additionally, recent reports have linked Set1dependent Lys-4 methylation of histone H3 to transcriptional elongation via the Paf1-Rtf1 complex. Because both the Rad6-Bre1 and Paf1-Rtf1 complexes are necessary for H3 Lys-4 methylation, we surmised that strains with defects in these complexes should have a similar phenotype as that observed in a set1⌬ strain. Consistent with this idea, deletions of either RTF1 or PAF1 lead to significant increases in PHO5 expression, with paf1⌬ having a larger effect than rtf1⌬ (Fig. 4A). Previous results have shown that deletion of PAF1 leads to a larger transcriptional effect than an RTF1 deletion (25). Further, the absence of Paf1 reduces the association of Set1 with coding regions more than does the loss of Rtf1 (10). Similarly, deletions of SET1 or BRE2 of the COMPASS complex (3) also show enhanced rAPase expression (Fig. 4A). In addition, deletions in RAD6, BRE1, or LGE1, encoding subunits of the Rad6-Bre1 complex (26), lead to increased PHO5 expression (Fig.  4B). It is interesting to note that the Paf1-Rtf1 complex, which is genetically upstream of the Rad6-Bre1 and COMPASS complexes, has the largest effect on derepression. This suggests that Paf1-Rtf1 may recruit other components that regulate PHO5 expression.
Set1 Is a Negative Regulator of PHO84 -To test whether other genes of the PHO cluster are regulated by Set1, we examined the expression of PHO84, which codes for the highaffinity phosphate transporter. Like PHO5, PHO84 is only minimally expressed in high P i conditions and is highly expressed in media where P i is limiting. PHO84 mRNA levels were analyzed in SET1 ϩ and set1⌬ cells grown in minimal high-P i medium. As for PHO5 (Figs. 1 and 2), PHO84 expression is also derepressed in a set1 null strain, showing a marked 13-fold increase in transcription over the SET1 ϩ strain (Fig.  5A). Conversely, a slight reduction in the mRNA levels transcribed from the gene for the constitutive protein phosphatase PPH3 is observed in set1⌬ cells (Fig. 5B), as has been shown previously (6). These results demonstrate that Set1 may be a general repressor of PHO genes although it is required for full expression of some genes that do not respond to phosphate levels.
GAL1-10 Is Also Negatively Regulated by Set1-To determine whether Set1 is involved in the repression of other genes not under phosphate control, we examined the divergently transcribed GAL1-10 locus. SET1 ϩ and set1⌬ strains were grown in repressed conditions in the presence of glucose (YPD) and semi-activating conditions (YPG ϩ 0.5% glucose) prior to assaying for GAL1 and GAL10 mRNA accumulation. GAL1 and GAL10 transcripts are not detectable in YPD; however, more GAL1 and GAL10 mRNA accumulate in a set1⌬ strain in YPG ϩ 0.5% glucose (Fig. 6). These data are consistent with two previous microarray analyses that showed increased GAL1 transcript levels in a set1⌬ deletion strain (27,28). A recent study has also shown that, at early times of induction, GAL10 is expressed at higher levels in rad6⌬ as well as set1⌬ strains (29). Thus, Set1 may negatively regulate a myriad of genes with different functions and regulatory mechanisms. DISCUSSION Histone methylation at lysine 4 by Set1 via COMPASS is a prominent histone modification in yeast, with ϳ34% of the total histone H3 pool being methylated (30). Recent evidence has demonstrated that Set1-dependent methylation requires the monoubiquitination of histone H2B and the Paf1-Rtf1 complex, which has been implicated in transcriptional elongation through its interaction with the carboxyl-terminal domain of RNA polymerase II. The observation that H3 Lys-4 methylation is primarily associated with euchromatic genes although H3 Lys-9 methylation is correlated with heterochromatin in metazoan species suggests that each may play an integral role in the establishment of active versus inactive regions, respectively. Recent evidence has shown that Lys-9 methylation via the histone methyltransferase Suv39h leads to the recruitment of heterochromatin protein 1 (31,32). Artificial targeting of histone methylation or heterochromatin protein 1 via chimeric fusion proteins to euchromatic regions also leads to local gene silencing (33,34).
Although Set1-mediated Lys-4 methylation of histone H3 is associated with active chromatin, it is not known whether this modification leads to the recruitment of additional factors or serves another function. Furthermore, although this histone modification is associated with transcriptional elongation and appears to be prominent at core promoters and at the 5Ј end of transcribed regions of euchromatic genes, it has been shown to be required for the full expression of only a few genes (5,6). We presented evidence that Set1 is also involved in the repression of a subset of genes in active chromatin regions, showing that the loss of Set1 leads to higher levels of expression of PHO5 (Figs. 1, 2, and 4A), PHO84 (Fig. 5), and GAL1-10 (Fig. 6). This evidence indicated that, in addition to its previously characterized role in activation, H3 Lys-4 methylation has a repressive role in gene expression. General derepression of genes in our set1⌬ mutant strain was not observed, because we confirmed A, total rAPase activities of strains CCY1467 (SET1 ϩ ) and CCY1471/ CCY1472 (set1⌬) grown in minimal high-P i medium for 6 h. Activities were reported as Miller units as described previously (18). Results are representative of five independent experiments. B, RNA internally isolated in A for PHO5 and ACT1 mRNA levels. The PHO5 panel is pixilated as the transcript levels are extremely low in minimal medium containing P i . For quantification (fold relative to SET1 ϩ ), PHO5 transcript levels in each lane were normalized to the amount of ACT1 mRNA. C, total rAPase activities of strains grown in minimal no-P i medium for 6 h. D, Northern analysis of RNA internally isolated from cells in C. Set1 Negatively Regulates PHO5 and GAL1-10 Transcription that PPH3 expression is down-regulated, as has been shown previously (6).
Deletions of genes coding components of complexes that regulate Set1 methylation also showed similar phenotypes as those observed in a set1 null strain (Fig. 4). Loss of Paf1, the most upstream regulator of H3 Lys-4 methylation, led to the largest derepression of PHO5. This suggests that other factors in addition to Set1 may be recruited by the Paf1-Rtf1 elongation complex. It was reported previously that a paf1 mutant causes both gene-specific increases or decreases in transcription, which demonstrates a dual role in gene regulation for transcriptional elongation (35). Interestingly, in this study, GAL10 and GAL7 were two of the genes that required Paf1 for full expression, suggesting further that Paf1-Rtf1 recruits additional proteins that affect gene expression positively or negatively. Similarly, loss of Ctk1, which phosphorylates the carboxyl-terminal domain repeat of RNA polymerase II, has both positive and negative effects on the transcription of various genes (36). More recently, the histone methyltransferase Set2, which methylates histone H3 at lysine 36, has also been associated with active chromatin and transcription elongation via the Paf1-Rtf1 complex and is required for full expression of a GAL1-lacZ reporter (37). However, when the Set2 protein is tethered to a heterologous promoter via LexA, it serves as a repressor lowering transcription by more than 20-fold (38). Consistent with this observation, Set2 is responsible for the repression of the basal expression of GAL4 (39).
repressed ϳ10-fold in the absence of Rad6 (40). Similarly, GAL10 is expressed at much higher levels in a rad6 null strain; however, the corresponding histone H2B lysine 123 to the arginine (K123R) mutant did not derepress GAL10 transcription (29). This same histone mutant was also used in another study to show that loss of histone H2B ubiquitination at GAL1 and PHO5 delays transactivation (41). However, in this report, Rad6 was dispensable for GAL1 activation. The reason for different phenotypes between rad6⌬ and histone H2B K123R mutant strains was unclear and may have resulted from strain differences or altered levels of expression of the episomal copy of the histone gene. It is also conceivable that histone H2B ubiquitination is required for basal repression as well as transcriptional activation as has been reported previously for the histone deacetylase Rpd3 (42).
A recent study demonstrates that the proteosomal ATPases Rpt4 and Rpt6 are required for H3 Lys-4 methylation (43), potentially linking ubiquitination of histone H2B and methylation of histone H3. Rpt4/6 recruitment is dependent on Rad6, suggesting that histone ubiquitination is required. Further, in agreement with our data (Fig. 6B), these authors show that a mutation in RPT6 leads to increased levels of GAL10 expression and state that a strain lacking Set1 shows a similar effect. Taken together, these results demonstrate that, despite their association with transcription elongation, specific effectors of H3 Lys-4 methylation may up-or down-regulate distinct subsets of genes.
The mechanism for the opposing regulatory role at different loci remains unclear. We propose that Set1 regulates gene expression positively and negatively. Although Set1 methylation activity depends on the transcriptional elongation complex Paf1-Rtf1, our data also show that PHO5 expression is derepressed in paf1⌬ and rtf1⌬ mutants that lack H3 Lys-4 methylation. This suggests that H2B Lys-123 ubiquitination and H3 Lys-4 methylation may establish a negative feedback loop on basal transcription. This view is consistent with the presence of a transcriptional elongation checkpoint in which Ctk1 phosphorylates serine 5 of the RNA polymerase carboxyl-terminal domain and recruits COMPASS that contains Set1 (16). At initial times of transcriptional initiation, transcriptional activators, such as Pho4 or Gal4, bind to their cognate DNA sites and recruit transcriptional coactivators, e.g. the histone acetyltransferase complex SAGA and the RNA polymerase machinery, including Paf1-Rtf1. During the transition to a stably elongating polymerase, Paf1-Rtf1 leads to increasing amounts of histone H2B ubiquitination and subsequent H3 Lys-4 methylation, possibly ensuring assembly of the necessary RNA capping and processing factors (17). However, later in the activation process, recruitment of SAGA is increased, and the histone deubiquitinase Ubp8 of SAGA may remove H2B ubiquitination. This model is supported by data demonstrating a transient peak of histone H2B ubiquitination following transactivation of GAL1 and PHO5 (41,44). Further studies are needed to elucidate the biochemical functions of histone methylation in transcription.  6. Set1 represses GAL1-10 expression. A, Northern analysis of GAL1 expression of CCY1467 (SET1 ϩ ) and CCY1471 (set1⌬) strains grown in YPD or YPG ϩ 0.5% glucose. For quantification (fold relative to SET1 ϩ ), GAL1 transcript levels in each lane are normalized to ACT1 mRNA levels. N.D., not detectable. B, Northern analysis of GAL10 transcript levels normalized to ACT1 mRNA levels.