|Home | About | Journals | Submit | Contact Us | Français|
Telomere-associated position effect variegation (TPEV) in budding yeast has been used as a model for understanding epigenetic inheritance and gene silencing. A widely used assay to identify mutants with improper TPEV employs the URA3 gene at the telomere of chromosome VII-L that can be counter-selected with 5-fluoroorotic acid (5-FOA). 5-FOA resistance has been inferred to represent lack of transcription of URA3 and therefore to represent heterochromatin-induced gene silencing. For two genes implicated in telomere silencing, POL30 and DOT1, we show that the URA3 telomere reporter assay does not reflect their role in heterochromatin formation. Rather, an imbalance in ribonucleotide reductase (RNR), which is induced by 5-FOA, and the specific promoter of URA3 fused to ADH4 at telomere VII-L are jointly responsible for the variegated phenotype. We conclude that metabolic changes caused by the drug employed and certain mutants being studied are incompatible with the use of certain prototrophic markers for TPEV.
Constitutive heterochromatin is characterized by low or variegated transcriptional activity (reviewed by Dimitri et al., 2009). The budding yeast Saccharomyces cerevisiae has been used as a genetic model to study inheritance of heterochromatin. Initially, experiments focused on the two silent mating type loci HMR and HML when it was discovered that for an α cell to behave like an a cell genetic information did not have to be deleted but could be silenced by the silent-information regulator (SIR) proteins (reviewed by Herskowitz, 1988). Epigenetically variegated inheritance of chromatin states at the endogenous HM loci can be observed in single cells by the distinct morphology of yeast cells preparing to mate (Osborne et al., 2009; Pillus and Rine, 1989).
Telomeres, the ends of chromosomes, can also cause genes to be expressed in a variegated manner. This phenomenon, called telomere position effect variegation (TPEV), was recognized in Drosophila melanogaster and subsequently in other species including human cells (reviewed by Tham and Zakian, 2002). In budding yeast, the use of various prototrophic markers spurred characterization of TPEV. One marker, the URA3 gene, placed at a truncated left arm of chromosome VII (Gottschling et al., 1990), was particularly useful in that it could be selected both positively and negatively. The uracil biosynthetic pathway analog 5-fluoroorotic acid (5-FOA) enabled selection against URA3 expression since the Ura3 enzyme participates in conversion of 5-FOA into a lethally toxic metabolite (Boeke et al., 1984). Using this assay, the artificial telomere was found to be transcriptionally silenced over a distance of 3.5 kb in a SIR-dependent manner.
In addition to SIR proteins, many proteins have been implicated in telomeric heterochromatin formation using TPEV reporters such as URA3-VIIL including those involved in DNA replication-coupled chromatin assembly as well as histone modifiers.
The Disruptor Of Telomeric silencing 1 (DOT1; Singer et al., 1998) encodes the only known histone H3 K79 methyltransferase in budding yeast (van Leeuwen et al., 2002). By methylating H3 K79 in euchromatin Dot1 is thought to prevent the limiting SIR proteins from localizing from telomeric and HM loci to euchromatin (van Welsem et al., 2008). Human and yeast PCNA, the processivity clamp for DNA polymerases, have been shown to physically interact with Chromatin Assembly Factor-1 (CAF-1), a three-subunit chaperone that deposits newly synthesized histones H3 and H4 onto replicated DNA (reviewed by Ransom et al., 2010). Both budding yeast PCNA (POL30) and CAF-1 were found to be required for heterochromatin formation using the TPEV assay (Enomoto et al., 1997; Kaufman et al., 1997; Zhang et al., 2000).
We initiated this study asking which additional factors are required for PCNA-dependent TPEV. Surprisingly, our work led to the conclusion that the widely used URA3-VIIL assay does not necessarily reflect a defect in TPEV of either the mutant PCNA or DOT1 cells, but rather is a read-out of metabolic changes in these mutants that are further induced by 5-FOA. In support, work by Takahashi et al. (2011) shows that H3 K79 methylation by Dot1 is not required for gene silencing at most native telomeres.
The S. cerevisiae PCNA mutation pol30-8 (RD61,63AA; Ayyagari et al., 1995) was reported to have a heterochromatin silencing defect at the HMR locus marked with ADE2 and at URA3-VIIL adjacent to a truncated ADH4 gene (Figure 1A; Zhang et al., 2000). This phenotype correlated with reduced binding of PCNA to Cac1, the large subunit of CAF-1, in vitro and less chromatin-bound Cac1 in vivo. To further elucidate the role of pol30-8 in TPEV, a high-copy suppressor screen was performed. By screening for 5-FOA resistant colonies indicative of a silenced URA3-VIIL reporter, POL30 was isolated as well as five other genes, one of which was CDC21. CDC21 is an essential gene encoding the metabolic enzyme thymidylate synthase (Hartwell et al., 1973). It catalyzes the addition of a methyl group to deoxyuridylate (dUMP) to form thymidylate (dTMP), required for synthesis of thymidine triphosphate (dTTP). The 5-FOA sensitivity of pol30-8 or cac1Δ but not that of sir3Δ URA3-VIIL was suppressed by overexpressed CDC21 (Figure 1B and data not shown). Furthermore, CDC21 overexpression did not suppress the 5-FOA sensitivity of a strain wild-type for URA3 at its endogenous chromosome V locus, unlike GFA1, an essential gene for chitin biosynthesis also identified in the screen (Figure S1A).
A PROSITE motif search predicts the Cdc21 catalytic-site cysteine residue at position 177 (see Supplemental Experimental Procedures). Cdc21 also contains the EUK1 motif that when deleted decreased catalytic activity to 1 % of the wild type (Munro et al., 1999). The cdc21-ΔEUK1 mutant decreased and the cdc21-C177A mutant eliminated suppression of pol30-8 URA3-VIIL (Figure 1B), although cdc21-C177A was expressed at levels comparable to the wild-type protein (Figure 1C). Moreover, CDC21 overexpression increased dTTP levels by about 1.8-fold (Figure S1B). In contrast, compromised Cdc21 function in URA3-VIIL strains carrying the cdc21-216 allele (Zhou and Elledge, 1992 and Figure S1C) resulted in pronounced 5-FOA sensitivity, supporting an apparent role of CDC21 in modulating TPEV (Figure 1D).
Interestingly, CDC21 also suppressed the apparent TPEV defect of dot1Δ, but not pol30-8 dot1Δ URA3-VIIL mutants (Figure 1E). The histone chaperone Anti-Silencing Function 1 (ASF1) aids CAF-1 in the assembly of H3 and H4 onto replicating DNA but also acts in a different genetic pathway from PCNA-CAF-1 (Tyler et al., 1999). Surprisingly, deletion of ASF1 partially rescued the dot1Δ URA3-VIIL 5-FOA sensitivity, which could not be further alleviated by overexpressing CDC21 (Figure 1E).
These data suggest that CDC21 is a high-copy suppressor of the TPEV defect of pol30-8 and dot1Δ URA3-VIIL strains, and likely depends on the catalytic activity of CDC21. Furthermore, deletion of ASF1 and CDC21 overexpression act in the same genetic pathway with regard to alleviating the apparent dot1Δ URA3-VIIL TPEV defect.
In the original description of TPEV, four genes were placed at telomere VII-L and three had substantially silenced expression (Gottschling et al., 1990). To address the general effect of CDC21 on TPEV, we proposed to test whether its ability to suppress the telomeric silencing defect of pol30-8 or dot1Δ cells was specific to URA3-VIIL or applied to other reporter genes, such as HIS3 at telomere VII-L (Figure 2A). HIS3 expression can be positively selected for on medium containing 3-amino-1,2,4-triazole (3-AT), a competitive inhibitor of the HIS3 gene product. sir3Δ mutants (a control for this assay) grew even in the presence of 50 mM 3-AT (Figure 2B). Unexpectedly, pol30-8 (or cac1Δ, data not shown) HIS3-VIIL mutants only had a slight growth advantage compared to wild type on medium lacking histidine (less than 10-fold). Moreover, addition of 3-AT did not result in a growth advantage of these mutants over wild type. Thus the pol30-8 mutation had only a minimal effect on HIS3-VIIL expression, suggesting that the pol30-8 mutant weakly increased expression of a poorly expressed gene. Surprisingly, in dot1Δ cells HIS3-VIIL was not expressed and therefore it is unlikely that Dot1 silences the HIS3 telomeric gene.
The stark difference between the telomeric effects using URA3-VIIL and HIS3-VIIL reporters prompted us to assess URA3-VIIL mRNA expression levels. RT-qPCR analysis of RNA extracted from strains grown to logarithmic phase in rich medium showed that URA3-VIIL expression was elevated by only 6.2- to 7.2-fold in pol30-8 compared to wild-type cells (Figure 2C). In contrast, in MATα sir3Δ cells URA3-VIIL expression levels were 187 times above wild-type levels. Unexpectedly, POL30 overexpression did not significantly lower the elevated URA3-VIIL expression in pol30-8 mutants (Figure 2D), while it suppressed their 5-FOA sensitivity very well (Figure 1B). This key observation suggests that URA3 expression is not the sole determinant of 5-FOA sensitivity. In pol30-8 HIS3-VIIL strains however, POL30 suppressed expression below wild-type levels (Figure 2E). Thus HIS3-VIIL reported TPEV more accurately than URA3-VIIL.
Since the dot1Δ HIS3-VIIL phenotype opposes that of dot1Δ URA3-VIIL cells, we sought to investigate further the heterochromatin defect in dot1Δ cells. If euchromatic H3 K79 trimethylation acts as a barrier to SIR protein spreading away from heterochromatin (van Welsem et al., 2008), we would expect the limiting SIR proteins to be less confined to telomeres in dot1Δ cells. However, using chromatin immunoprecipitation (ChIP) we could not observe a reduced Sir2 and Sir4 occupancy at two telomeres in dot1Δ compared to wild-type cells (Figures 2F and S2).
Together these results suggest that neither pol30-8 nor dot1Δ led to a general telomeric silencing defect. Rather, the phenotypes were specific for the URA3-VIIL reporter when counter-selected with 5-FOA.
To address in an unbiased fashion whether pol30-8 and dot1Δ had a general effect on telomere-associated gene expression, mRNA levels were measured using an Affymetrix-platform based microarray, comparing three biological replicates of wild-type, pol30-8, dot1Δ and pol30-8 dot1Δ strains carrying the heterochromatin reporter constructs hmrADE2 and URA3-VIIL (deleted for the endogenous alleles ADE2 and URA3). We observed down-regulation (or deletion) of gene expression in a 40-kb region on chromosome V common to all three mutant strains tested (Figure S3A). Since the pol30-8 mutation otherwise altered gene expression differently from the dot1Δ mutation, we did not follow up this effect further.
A heatmap comparing dot1Δ mutant to wild type, considering 20-kb segments from each of the 32 telomeres in yeast, confirmed the genetic results with HIS3-VIIL: no elevation, but rather a mild down-regulation of telomeric gene expression, except for that of ADH4 (5.8-fold up-regulated) and URA3 (2.9-fold up-regulated) at telomere VII-L (Figure 3A and Table S1). To analyze regional gene expression in the different mutants, we applied two different methods, “generally applicable gene-set enrichment” (GAGE; Luo et al., 2009) as well as “fold change” to plot the gene expression change compared to wild type in 10-kb windows from all 32 pooled telomeres towards the pooled centers of all 16 chromosomes. This analysis demonstrated very modest global up-regulation of gene expression in the telomere-proximal 50 kb in pol30-8, but not in dot1Δ cells (Figure 3B). Globally, the pol30-8 mutation seemed to be dominant over the dot1Δ deletion (Figures 3B, S3B and S3C). A previous study of cac2Δ and asf1Δ cells synchronized in G2/M did not find any bias of gene expression changes towards telomeres (Zabaronick and Tyler, 2005). Reanalyzing replicates of this dataset revealed a similar phenotype for these two mutants to that observed for pol30-8, but not the dot1Δ mutant (Figures 3C and S3D). Genome-wide, the up-regulated genes in pol30-8 cells were those that were normally poorly expressed in wild-type cells (Figure 3D), while there was no such bias when comparing the dot1Δ mutant or single wild-type replicates over the average of wild-type signal (Figures 3E, S3E and S3F). The same correlation was seen for cac2Δ and asf1Δ mutants (Figures S3G, S3H and S3I). In agreement, the top up-regulated Gene Ontology processes in pol30-8 cells concern sporulation which is normally suppressed in vegetative cells by the transcriptional repressor Sum1 (Table S2; Pierce et al., 2003).
Interestingly, as previously observed with cac1Δ mutants (Tamburini et al., 2006), ChIP analysis for pol30-8 cells showed less histone H3 bound in all chromosomal regions tested, including an intergenic region on chromosome VIII-R (Figure 3F). Therefore, the global up-regulation of poorly expressed genes could be possibly due to lower histone density on chromatin in pol30-8 mutants. In dot1Δ cells, however, the URA3 inserted into the ADH4 locus was specifically up-regulated in the absence of a general telomeric silencing defect (Figure 3A). Overexpression or deletion of DOT1 also leads to increased expression of ADE2 inserted at telomere V-R or hmr (Singer et al., 1998 and white colony color in Figures 1E and and3G).3G). Together these phenotypes raised the possibility that purine (Ade) and pyrimidine (Ura) synthesis pathways were connected. Such cross-regulation was suggested as deletion of either of two transcription factors BAS1 or PHO2 required for de novo purine synthesis almost abolished URA3 transcription for pyrimidine synthesis in conditions limiting for purines (Figure S3J; Denis et al., 1998). Indeed, decreasing the activity of the purine synthesis pathway lowered URA3-VIIL expression, as dot1Δ bas1Δ pho2Δ URA3-VIIL mutants grew better on 5-FOA than dot1Δ alone (Figure 3G). This effect was specific to dot1Δ since 5-FOA sensitivity of pol30-8 cells was unchanged by deletion of BAS1 and PHO2. These results confirm a co-regulation of ADE2 and URA3 and thus call into question the independence of the two prototrophy markers at different heterochromatic loci. They also further underscore that different mechanisms cause 5-FOA sensitivity in the dot1Δ and the pol30-8URA3-VIIL mutants. We can conclude, however, that in both cases the URA3-VIIL reporter assay did not reflect an essential role for either of these genes in silencing of telomere-associated genes.
If the pol30-8 mutation only causes mildly elevated URA3-VIIL mRNA expression, then why are pol30-8 URA3-VIIL cells so 5-FOA sensitive? We hypothesized that the latter phenotype might be explained by the up-regulation of other genes in pol30-8 cells. Two microarray analyses for pol30-8 mutant strains revealed elevated expression of RNR2, RNR4 (both 1.7-fold, data not shown) and RNR3 (3.3-fold, Table S1). These genes encode subunits of RNR which generates the four deoxyribonucleoside triphosphates (dNTPs) required for DNA synthesis. In S. cerevisiae, this enzyme contains a large catalytic R1 subunit comprised of either Rnr1 or Rnr3, or their combination, and a small R2 subunit containing Rnr2 and Rnr4 (Nordlund and Reichard, 2006). The microarray results were confirmed by RT-qPCR analysis (Figure 4A). In the case of RNR2 and RNR4, the mild transcriptional up-regulation was reflected at the protein level (Figure 4B). However, dNTP levels were not significantly altered from wild type in pol30-8 cells (Figure S4A). One of the well described consequences of DNA damage is the derepression of the RNR genes (Zegerman and Diffley, 2009), transcription of which is normally repressed by Crt1 (Huang et al., 1998). Interestingly, our screen also identified CRT1 as a high-copy suppressor of pol30-8URA3-VIIL 5-FOA sensitivity (data not shown). This result is consistent with the observed RNR up-regulation in pol30-8 cells.
Intriguingly, mutants in the DNA damage checkpoint pathway have been shown to increase “silencing” of the URA3-VIIL reporter (Craven and Petes, 2000; Longhese et al., 2000) and to rescue CAF-1 dependent TPEV defects (Sharp et al., 2005). Indeed, rad53K227A pol30-8 URA3-VIIL cells grew much better on 5-FOA than pol30-8 cells alone (Figure S4B). Since in rad53 mutants RNR cannot be induced and pol30-8 mutants had increased RNR levels, we wondered whether the increased RNR activity in pol30-8 URA3-VIIL cells could directly contribute to 5-FOA sensitivity.
The toxicity caused by 5-FOA stems from a product generated by its conversion to fluoroorotidine monophosphate (5-FOMP) and further decarboxylation to 5-fluorouridine monophosphate (5-FUMP) by orotidine monophosphate decarboxylase, encoded by URA3. After phosphorylation, the diphosphate can be either incorporated into RNA from 5-fluorouridine triphosphate (5-FUTP) or it can be reduced by RNR to the deoxydiphosphate (5-FdUDP), which is either phosphorylated (to 5-FdUTP) and used for DNA synthesis or dephosphorylated to 5-fluorodeoxyuridine monophosphate (5-FdUMP). 5-FdUMP forms a covalent complex with Cdc21 and the methyl donor methylenetetrahydrofolate, blocking the enzymatic methylation reaction of dUMP to dTMP (Figure 4C; Hardman et al., 2001; Jones and Fink, 1982; Longley et al., 2003). We asked whether directly inhibiting RNR function could rescue the 5-FOA sensitivity phenotype. Indeed, the RNR inhibitor hydroxyurea (HU) at sub-lethal concentrations rescued the 5-FOA sensitivity of pol30-8 or cac1Δ strains, but not strains mutant for both the PCNA-CAF-1 and ASF1 chromatin assembly pathways (Figure 4D). The striking effect of HU in the presence of rather mild up-regulation of RNR2, RNR3 and RNR4 expression in the pol30-8 mutant prompted the question whether 5-FOA itself would stimulate RNR transcription. Logarithmically growing wild-type and pol30-8 URA3-VIIL cells were treated with 5-FOA (or DMSO as a control) for up to four hours and displayed a marked increase in RNR transcript levels. For RNR4, transcript levels increased 2.8- and 4.8-fold, respectively, with final levels being 3.1-fold higher in pol30-8 URA3-VIIL cells (Figure 4E). Similar results were observed for RNR2 and RNR3 (data not shown). Importantly, while POL30 overexpression only marginally lowered URA3-VIIL expression in pol30-8 cells (Figure 2D), it lowered Rnr4 levels very well (Figure 4F). These results indicate that 5-FOA causes a transcriptional DNA damage response and thereby increases the transcriptional up-regulation of RNR genes in pol30-8 cells. This together with mild up-regulation of URA3-VIIL expression in pol30-8 cells is likely the reason for their markedly increased 5-FOA sensitivity. The latter can be either overcome by overexpressing CDC21, the target of the drug, by reducing RNR activity or by overexpressing POL30 which reduces RNR levels.
Since deletion of ASF1 partially rescued the dot1Δ URA3-VIIL 5-FOA sensitivity (Figure 1E), we asked whether this could be due to a failure to increase RNR levels in this strain. dot1Δ URA3-VIIL cells had wild-type levels of RNR that could be induced by DNA damage with 4-nitroquinoline 1-oxide (4-NQO; Figure 4G). In contrast, dot1Δ asf1Δ URA3-VIIL was defective in elevating RNR levels in response to DNA damage. CDC21 overexpression did not cause an additional growth advantage of asf1Δ dot1Δ URA3-VIIL mutants on 5-FOA (Figure 1E). These observations further support the hypothesis that up-regulation of RNR expression upon 5-FOA treatment is a main component of 5-FOA sensitivity in these mutant strains. We conclude that the 5-FOA assay in the context of low URA3-VIIL expression can function as an indicator of RNR levels. Failure to up-regulate RNR can counteract 5-FOA metabolism so that overexpression of the target of 5-FdUMP, CDC21, is without effect.
Many studies on TPEV in budding yeast have utilized truncated telomeres with the majority employing URA3 as a reporter gene. This work reports the surprising finding that for two different mutants, pol30-8 and dot1Δ, their previously suspected essential roles in heterochromatin induced gene silencing do not withstand more detailed analysis. Rather, both mutants are differently sensitized to 5-FOA (Figure 4C). pol30-8 cells globally up-regulate expression of normally poorly expressed genes, leading to modest up-regulation of URA3-VIIL as well 5-FOA induced RNR gene transcription. Alternatively, dot1Δ cells exhibit more pronounced up-regulation of URA3-VIIL transcription due to a specific effect on the ADH4 locus into which URA3 is embedded. These modest metabolic imbalances only reveal themselves in the context of low expression of telomeric URA3 and can then be counteracted by increased levels of Cdc21, the target of the 5-FdUMP toxic product. Thus 5-FOA sensitivity depends on subtle changes in enzymes that metabolize the drug and for RNR, the drug itself induces changes in enzyme levels. Interestingly, resistance of human colorectal tumor xenografts in mice to 5-FU, which like 5-FOA is converted to 5-FUMP, is accompanied by an almost 5-fold reduced RNR activity (Fukushima et al., 2001). In the case of sir2, sir3 or sir4URA3-VIIL mutants, possible subtle effects on 5-FOA metabolism are overridden by their strong derepression of telomere-associated reporter genes. However, the 5-FOA sensitivity assay does not allow distinction between etiologies since 5-FOA counter-selection of pol30-8 or dot1ΔURA3-VIIL cells reveals a phenotype that is nearly as strong as that seen in sir mutants. In the future, the direct measurement of expression of genes adjacent to telomeres would be the ultimate test for TPEV.
pol30-8 as well as cac1Δ mutants showed increased sensitivity to various DNA damaging agents but no growth defect and for pol30-8 no reduced interaction with DNA replication proteins (Ayyagari et al., 1995; Linger and Tyler, 2005). Moreover, the DNA damage checkpoint response, as assessed by Rad53 hyper-phosphorylation and Sml1 degradation, was not overtly activated in the pol30-8 mutant (our unpublished data). Thus, while we cannot exclude a subtle DNA repair defect in pol30-8 cells, we propose that increased transcription of RNR and URA3-VIIL occurs because of a reduction in global histone density due to a defect in DNA replication-coupled nucleosome assembly. In support, Wyrick et al. (1999) observed elevated gene expression upon H4 depletion, extending distally from SIR-silenced telomeres, similar to that seen for pol30-, cac2Δ and asf1Δ mutants. PCNA and CAF-1-dependent nucleosome assembly, while not essential, is thus necessary for correct global chromatin structure and in their absence, leaky expression of genes not normally expressed or genes that are expressed at low levels can occur. A similar, global defect in histone deposition leading to altered gene expression may explain why CAF-1 mutations in plants cause variegation among individuals (stem fasciation) and stem cell proliferation defects (Kaya et al., 2001) and the life span defects in yeast (Feser et al., 2010).
It was suggested that Dot1 is involved in anti-silencing mechanisms by methylating H3 K79 in euchromatin (van Leeuwen et al., 2002; van Welsem et al., 2008). The specific context at the truncated telomeric ADH4 locus is likely important for the transcriptional up-regulation of URA3, since a dot1Δ HMRURA3 strain shows wild-type levels of 5-FOA resistance (Osborne et al., 2009). One might speculate that mutants of other genes with assigned anti-silencing properties are 5-FOA sensitive for different reasons than a TPEV defect at URA3-VIIL.
Lastly, this study should also caution against the use of the ADE2 reporter to assess more subtle phenotypes in heterochromatin formation. While ADE2 expression allows for immediate phenotyping, it seems to be linked to URA3 expression via purine-pyrimidine cross-regulation. Instead of using these reporter genes, assaying for endogenous silencing properties such as mating efficiency reflective of HML and HMR silencing might be preferable (Osborne et al., 2009; Pillus and Rine, 1989).
SC medium was prepared as described with the exception that filter-sterilized adenine and uracil were added after autoclaving to a final concentration of 20 mg/l and 100 mg/l (or 20 mg/l for SC media containing 5-fluoroorotic acid [5-FOA]), respectively. Unless otherwise indicated SC 5-FOA media contained 1 g/l of 5-FOA (US Biologicals; for liquid cultures see Supplemental Experimental Procedures). Yeast strains were grown at 30 °C unless noted otherwise.
S. cerevisiae strains used in this study were in the W303, W1588-4C and YPH backgrounds. For a complete list see Table S3C.
Overnight (ON) cultures were adjusted to OD600 = 1.0 and 10-fold serially diluted. 5 μl of each dilution was spotted onto indicated media. Plates were photographed after at least 3 days of incubation at 4 °C.
Total RNA was prepared using the hot acidic phenol method. After DNase-I treatment and RT-PCR, samples were analyzed by quantitative PCR (qPCR). Primers used are listed in Table S3B.
Protein was extracted from 5 ODs of logarithmically growing cells at a density ~ 2 × 107 cells/ml (as determined by OD600) with either alkali lysis alone or with added TCA precipitation. For antibody sources and dilutions see Supplemental Experimental Procedures.
Chromatin immunoprecipitation was done with some modifications to published procedures. 50 ml of cell culture at a density of ~ 1.7 × 107 cells/ml was treated for 30 min with freshly prepared 1 % paraformaldehyde. After bead-beating lysis and sonication to yield ~ 0.5 kb DNA fragments, immunoprecipitations were carried out from 120 μg chromatin lysate ON, extensively washed and the cross-links reversed ON. Precipitated DNA was quantified by qPCR as above.
ON cultures were diluted to OD600 ~ 0.1 and grown until OD600 ~ 0.4. Cultures were then treated for 2 h with 4-nitroquinoline 1-oxide (Sigma-Aldrich) to a final concentration of 0.2 mg/l. Untreated or treated cells were harvested at a cell density of 1.8 – 2.2 × 107 cells/ml and processed for either RNA or protein extraction. For liquid 5-FOA culture experiments see Supplemental Experimental Procedures.
RNA was prepared as described above from 10 ml cells of 3 independent colonies per genotype grown to a cell density of ~ 1.7 × 107 cells/ml (OD600 = 0.78 – 0.84). Samples were hybridized onto Yeast Genome 2.0 GeneChips (Affymetrix) by the Cold Spring Harbor Laboratory Microarray Shared Resource. Affymetrix QC metrics were used to pass the image data. All raw data were processed using the FARMS method, followed by GAGE test statistics for gene set or pathway analysis.
We thank members of the Stillman laboratory for helpful suggestions. We are grateful for reagents from D. Gottschling, S. Elledge, P. Kaufman, M. Longhese, A. Neiman, M. Wigler and V. Zakian, to J. Tyler for providing the cac2Δ and asf1Δ raw expression data files, to C. Johns for help with the microarray and to N. Hollingsworth, B. Futcher, R. Sternglanz, R. Rothstein, W. Tansey, A. Stenlund and B. Böttner for discussion and reagents. This work was supported by NIH grant GM45436 and CA13106 (B.S.) and the Swedish Research Council (A.C.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Supplemental Data Supplemental data for this article include extended procedures with references as well as 4 figures and 3 tables.
Accession numbers Data files can be accessed online with Gene Expression Omnibus (GEO) repository number GSE27222.