Systematic profiling of subtelomeric silencing factors in budding yeast

Abstract Subtelomeric gene silencing is the negative transcriptional regulation of genes located close to telomeres. This phenomenon occurs in a variety of eukaryotes with salient physiological implications, such as cell adherence, virulence, immune-system escape, and ageing. The process has been widely studied in the budding yeast Saccharomyces cerevisiae, where genes involved in this process have been identified mostly on a gene-by-gene basis. Here, we introduce a quantitative approach to study gene silencing, that couples the classical URA3 reporter with GFP monitoring, amenable to high-throughput flow cytometry analysis. This dual silencing reporter was integrated into several subtelomeric loci in the genome, where it showed a gradual range of silencing effects. By crossing strains with this dual reporter at the COS12 and YFR057W subtelomeric query loci with gene-deletion mutants, we carried out a large-scale forward screen for potential silencing factors. The approach was replicable and allowed accurate detection of expression changes. Results of our comprehensive screen suggest that the main players influencing subtelomeric silencing were previously known, but additional potential factors underlying chromatin conformation are involved. We validate and report the novel silencing factor LGE1, a protein with unknown molecular function required for histone H2B ubiquitination. Our strategy can be readily combined with other reporters and gene perturbation collections, making it a versatile tool to study gene silencing at a genome-wide scale.


Introduction
The condensation level of chromatin varies along the genome and impinges on a variety of cellular processes.One of the most important consequences of chromatin compactness is the accessibility of the transcriptional machinery that orchestrates gene expression.In general, highly compacted chromatin regions (heterochromatin) are associated with low transcription rates whereas loosely packed regions (euchromatin) are accessible chromatin sites that are transcriptionally active.In Saccharomyces cerevisiae, heterochromatic-like regions are well localized to telomeres and the silent mating type loci, making the budding yeast an excellent model organism to study chromatin conformation.At telomeres, the chromatin condensed state extends to its adjacent regions (subtelomeres) producing transcriptional inactivation or "silencing" of the genes in these loci.This phenomenon has also been termed telomere position effect (TPE) and, overall, it has been associated in different eukaryotic organisms with a variety of traits such as ageing (Kaeberlein et al. 1999), cell adherence (Su et al. 1995;Castano et al. 2005), virulence (Tham and Zakian 2002;De Las Penas et al. 2003;Janzen et al. 2004;Duraisingh et al. 2005), along with other features of industrial relevance (Halme et al. 2004;Bauer et al. 2010).
The silenced state at telomeres in S. cerevisiae is produced mainly by the Silent Information Regulation (SIR) complex constituted by Sir4, Sir3, and Sir2 (Aparicio et al. 1991).This complex is recruited to telomere ends by Rap1 (Liu and Lustig 1996), which binds specific DNA sequences at the telomere repeats termed silencers.Occupancy of the SIR complex at the subtelomeric regions is propagated inward, continuously toward the centromere through the action of the histone deacetylase Sir2 (Hoppe et al. 2002); propagation may occur discontinuously at some telomeres (Fourel et al. 1999;Pryde and Louis 1999).Interestingly, the SIR complex is also one of the multifunctional complexes involved in telomere homeostasis (Kupiec 2014).The deletions of Sir3 and Sir4 each cause shortening of telomeric repeats and mitotic instability of chromosomes (Palladino et al. 1993).TPE is also influenced by gene-dosage balance of telomeric and subtelomeric complex components (Renauld et al. 1993).For instance, Sir3 https://doi.org/10.1093/g3journal/jkad153Advance Access Publication Date: 11 July 2023

Mutant Screen Report
overexpression causes the spreading of silencing over longer distances from the telomere (Hecht et al. 1996), but a more recent study suggests that Sir3 overexpression does not extend heterochromatin domains (Brothers and Rine 2022).Besides the protein complexes that exert silencing, the size and structure of the telomere tract also influence TPE.It has been observed that short telomeres are associated with diminished TPE (Kyrion et al. 1993) and that telomere folding is also relevant for the maintenance of TPE (de Bruin et al. 2000).In addition, it is known that chromosome context influences silencing levels; regulatory elements at the subtelomeric regions contribute to the intrinsic basal silencing level of each subtelomere (Mondoux and Zakian 2007).
Over 100 genes have been reported to affect Sir-mediated silencing levels at different telomeres in S. cerevisiae.These genes were identified mostly on a gene-by-gene basis, usually using the URA3 reporter gene.The classic assay is based on the experiments that unintendedly led to the original discovery of TPE in yeast (Gottschling et al. 1990); it involves growing a strain carrying the URA3 gene in a silenced subtelomeric region in the presence of 5-fluoro-orotic acid (5-FOA).In 5-FOA-containing media, Ura3 activity produces a toxic metabolic intermediate, causing cell death.Therefore, colony growth can be used as a readout for the intensity of subtelomeric silencing, whereby further genetic modifications with an impact on gene silencing result in URA3 expression and, thus, cell death (Boeke et al. 1984).This semiquantitative assay has inherent drawbacks since it has been reported that 5-FOA induces metabolic changes leading to apparent TPE effects in some gene mutants (Rossmann et al. 2011).In addition, the assay is labor-intensive and not amenable to testing hundreds or thousands of mutant strains.
In principle, any methodology to measure gene expression such as RT-qPCR or RNA-seq can be employed to assess subtelomeric gene silencing.However, due to labor and cost, most of these methods cannot be readily used in combination with genedeletion or other available strain collections allowing genomewide genetic analysis.In this work, we developed a screening approach based on a novel URA3-GFP dual reporter integrated into subtelomeric loci to evaluate the effect of nonessential gene knockouts (Giaever et al. 2002) on silencing using high-throughput quantitative flow cytometry.In contrast to other techniques to measure gene expression, flow cytometry is less expensive, suitable for large-scale screenings, and does not require nucleic acid isolation.In addition, gene expression data is obtained at singlecell resolution in live cells, allowing the analysis of not only changes in average expression levels but also changes in the distribution of gene expression levels across a population.By using this robust and sensitive approach, we reveal variation in gene silencing among different subtelomeric regions of the genome and score genes influencing this phenomenon.Our study provides a large-scale screening approach to pinpoint genes and functions with an impact on subtelomeric gene silencing.

Strains and strain construction
All strains used in this study are listed in Supplementary Table 1.Knockout strains are from the yeast deletion collection xxxΔ:: KANMX4 in the BY4741 background (Giaever et al. 2002).The Y8205 parental mCherry (SCA52) and mTagBFP2 (Subach et al. 2011) (SCA89) fluorescent strains were generated by integrations of fluorescent-nourseothricin resistance (NAT) cassettes at the HO locus by homologous recombination.Fluorescent-NAT cassettes were constructed on a pFA6 (Addgene) based plasmid.
All the primers used for the construction of the strains are listed in (Supplementary Table 2).The URA3-GFP reporter was PCR amplified from pAJ69 and integrated at subtelomeric loci by homologous recombination using primers sharing 40 bp identity with subtelomeric regions.This reporter was integrated into parental strain mCherry (SCA52) and mTagBFP2 (SCA89).Using this methodology, we replaced several subtelomeric genes.The PCR primers in all cases were designed to replace entirely the subtelomeric ORFs of selected genes.At the chromosomal internal locus, CUP9 integration occurs at the 5´ intergenic region leaving intact ORFs.The construction of the library of mutants to study silencing at different loci was based on synthetic genetic array (SGA) methodology (Tong and Boone 2006).The sir3Δ strains for each locus were generated by homologous recombination over the parental strains.Not all crosses and further screening and data acquisition were successful and therefore the final data sets consisted of 3,716 knockouts for the COS12 locus and 4,193 for the YFR057 locus.

Plasmid construction
Plasmid pAJ69 was constructed to contain a translational fusion of URA3 and GFP genes under the control of a minimal URA3 promoter (216 bp) and the ADH1 terminator.URA3 gene and promoter were amplified from pRS416 (Stratagene) and the GFP gene and ADH1 terminator were amplified from pFA6a-GFP (S65T)-His3MX6 (Huh et al. 2003).These PCR fragments were fused by double-joint PCR and cloned in pUC19 EcoRI-HindIII sites.URA3 and GFP genes were cloned in the frame using a 27 bp linker (primers in Supplementary Table 2).

Growth conditions for flow cytometry GFP measurements
For large-scale screenings at COS12 and YFR057W loci, the reference blue fluorescent mTagBFP2 protein (BFP) strains SCA93 and SCA91 and plates of the respective subtelomeric reporter knockout collection were grown overnight in YPD medium on 96-well plates at 30°C without shaking.Each pair of reference-mutants was then pinned inoculated in 160 µl SC medium with 20 mg/l uracil in 96-well microtiter plates.Strains were grown at 30°C, 1000 rpm for 14-17 hours (7-9 cell generations), and OD 600nm in microtiter plate reader was 0.4-0.6.Cells were treated with 20 µl TE 2X and immediately measured with a flow cytometer.

Flow cytometry: instrumentation, acquisition, and data analysis
Large-scale flow cytometry was performed on a Stratedigm S1000EX cytometer.mTagBFP2 was excited with a violet laser (405 nm) and fluorescence was acquired through a 445/60 bandpass filter.GFP was excited with a blue laser (488 nm) and fluorescence was acquired through a 530/30 band-pass filter.mCherry was excited with a yellow laser (561 nm) and fluorescence was acquired through a 615/30 band-pass filter.For each co-culture in each well, the flow cytometer was set to measure 15,000 events or to stop acquiring after 50 seconds.More than 10,000 cells were counted for most samples.In smaller-scale validation experiments, BD FACSCalibur and BD LSR Fortessa X-20 flow cytometers were used.All data analyses and plots were performed with custom scripts on MATLAB.

Confocal microscopy
Cells were grown in YPD at 30°C to late exponential phase (OD 600nm = 0.6-0.9)and then collected and washed three times with 1 ml PBS 1X (NaCl 8.0 g/L, KCl 0.2 g/L, Na 2 HPO 4 1.44 g/L, KH 2 PO 4 0.24 g/L) and paraformaldehyde 4%; fixed, washed again, and resuspended in sorbitol 1 M. Cells were visualized in a LSM800 Zeiss confocal microscope, using 40X or 63X objectives.GFP was excited with a 488 nm laser, and mCherry with a 561 nm laser; fluorescence was captured using standard parameters and two different channels using SP620nm and LBF640 filters.

5-FOA growth assays
Strains were grown in YPD medium to stationary phase at 30°C and 200 rpm.The cultures were adjusted to an optical density of 1 at 600 nm with sterile water and 10-fold serial dilutions were made.A total of 5 µl of each dilution was spotted onto YPD, SC -ura, and 5-FOA agar plates and incubated at 30°C for 48 h for YPD and SC -ura agar plates, and 72 h for 5-FOA agar plates before imaging.

Nucleosome scanning assay
Nucleosome scanning experiments were performed adapted from Infante et al. (2012).The his3Δ S. cerevisiae was the wild type (WT) reference and isogenic mutants were grown to late exponential growth phase (45 mL of an O.D 600 = 0.8-1.0).Cells were treated with formaldehyde (1% final concentration) for 20 min at 37°C and then glycine (125 mM final concentration) for 5 min at 37°C.Formaldehyde-treated cells were harvested by centrifugation, washed with Tris-buffered saline, and then incubated in Buffer Z2 (1 M Sorbitol, 50 mM Tris-Cl at pH 7.4, 10 mM β-mercaptoethanol) containing 2.5 mg of zymolase 20 T for 20 min at 30°C on a rocker platform.Spheroplast was pelleted by centrifugation at 3000X g and resuspended in 1.5 mL of NP-40 spermidine (NPS) buffer (0.5 mM spermidine, 0.075% NP-40, 50 mM NaCl, 10 mM Tris pH 7.4, 5 mM MgCl 2 , 1 mM CaCl 2 , and 1 mM β-mercaptoethanol).Samples were divided into three 500 µL aliquots that were then digested with 22.5 U of MNase (Nuclease S7, Roche) for 50 min at 37° C. Digestions were stopped with 12 µl of 50 mM EDTA and 1% SDS and were treated overnight with 100 µg of proteinase K at 65°C.DNA was extracted twice by phenol/chloroform and precipitated with 20 µL of 5 M NaCl and an equal volume of isopropanol for 1 h at −20°C.Precipitates were resuspended in 40 µL of TE and incubated with 20 µg RNase A for 1 h at 37°C.DNA digestions were separated by gel electrophoresis from a 1.5% agarose gel.Monosomal bands were cut and purified by Wizard SV Gel Clean-Up System Kit (Promega, REF A9282).DNA samples were diluted 1:30 and used in qPCR using primers listed in Supplementary Table 2 to quantify the relative MNase protection of the query template.qPCR analysis was performed using a Corbett Life Science Rotor Gene 6000 machine using SYBR Green (2× KAPA SYBR FAST qBioline and Platinum SYBR Green, Invitrogen).Real-time PCR was carried out as follows: 94° for 2 min (1 cycle), 94° for 15 sec, 58° for 20 sec, and 72° for 20 sec (30 cycles).Relative protection was calculated as a ratio to the control VCX1 template which is within a well-positioned nucleosome in +250 bp of the ORFs.The PCR primers amplify from around −650 to +222 bp of YFR057W locus; coordinates are given relative to the autophagy (ATG) (+1).

Kappa-based functional analysis
Gene Ontology (GO) and phenotype terms were downloaded from the Saccharomyces Genome Database (SGD, last updated October 2019) to build two m by n matrices, where m is the number of analyzed genes 266 and n is the number of GO and phenotypic terms (2,234).Each term evaluates the overall agreement between genepairs to calculate Cohen's kappa (kappa = Pra(a)−PrPr (e) 1−Pr(e) ).Where Pra(a) is the number of GO and phenotypic terms in which each genepair shares an agreement, divided by the total number of terms downloaded from SGD, and Pr(e) is the hypothetical probability for each member of the gene-pair to be associated by chance.Then a matrix of genes representing the agreement as a kappa value between each gene-pair was built.Gene-pairs with a kappa > 0.35 were considered as functionally associated, values above this threshold represent the top 5% kappa values, this threshold has also been used in previous reports for large datasets (Campos et al. 2018).In the first step, only gene-pairs are associated, these pairs were then used as cluster seeds to form larger groups of genes with subsequent iterations of the analysis, where clusters sharing over 50% of its members were merged.Later, the clusters were named by manually inspecting for enriched functions or by using GO term finder tool (version 0.86) at SGD.The algorithm for kappa analysis was written in Matlab.Cytoscape was used to create a network where associated genes displayed kappa agreement above the threshold (kappa > 0.35).Analyzed genes are listed in Supplementary Table 3.

Gene expression by Rt-qPCR
Expression of the two subtelomeric genes YFR057W and COS12 was evaluated by RT-qPCR analysis on the WT strain.In addition, the sir3Δ was included as a strain with strong defects in silencing for comparison purposes.For RNA extraction, RiboPure-yeast kit (Ambion by life technologies) was used.Cells were grown in YPD at 30°C until 0.6 OD 600nm before harvesting.Total RNA extraction was carried out for each strain and cDNA was obtained in triplicate for each RNA extraction, following manufacturer's instructions.RNA integrity was assayed by gel electrophoresis and quantified in NanoDrop ND-1000.cDNA was obtained of 2 µg of total RNA using SuperScript III Reverse transcriptase and quantified again in NanoDrop.RT-qPCR was performed on StepOne Real-Time PCR system (Applied Biosystems), for 40 cycles using power SYBR Green PCR Master Mix (Applied Biosystems) and primers listed in Supplementary Table 2 with a T a = 60°C.A ΔCt was normalized to ACT1 for each COS12 or YFR057W Ct on each sample and then a ΔΔCt was calculated for each replicate as described in (Livak and Schmittgen 2001;Schmittgen and Livak 2008) relative to WT strain BY4741; average fold-change expression and SD were calculated.

A dual URA3-GFP gene reporter system allowing a quantitative assessment of gene silencing
To screen for genes that influence subtelomeric gene silencing in budding yeast, we constructed a dual-reporter system consisting of a translational fusion of the URA3 and GFP genes under the transcriptional control of the silencing-sensitive URA3 promoter (Materials and Methods).The URA3 gene with its native promoter has been widely used to detect gene silencing (Gottschling et al. 1990), but the addition of the GFP gene to the construct allows for assessing gene silencing by fluorescence microscopy, and, more importantly, by flow cytometry which makes the system amenable to quantitative, high-throughput screening.
To test the dual-reporter system, we inserted the cassette at two loci that are known to be silenced, the COS12 and YFR057W genes (Vega-Palas et al. 2000;Mondoux and Zakian 2007) at the subtelomeric regions of chromosomes VII (left arm) and VI (right arm), respectively (Fig. 1a.These two genes display amongst the highest fold increase in expression in a sir3Δ mutant (Wyrick et al. 1999), suggesting that the silenced state is mediated by the SIR complex.Furthermore, the telomere where COS12 resides often localizes to the nuclear periphery, a naturally silencing-promoting nuclear location (Tham et al. 2001).COS12 belongs to the large family of COS subtelomeric genes, a poorly studied set of genes, most of which are the first protein-coding gene next to the conserved core X element of the chromosome.The reporter was integrated by replacing the entire open ORFs of COS12 or YFR057W, in a strain that lacks the native URA3 gene.In this way, expression of the reporter can be tested both by growing the strains in media lacking uracil or containing 5-FOA and by measuring GFP fluorescence.As a control for non-silenced gene expression, we inserted the dual cassette at the large intergenic region between the CUP9 and TRE1 loci in the left arm of chromosome XVI (hereinafter the CUP9-5′ nonsubtelomeric loci).For following flow-cytometry experiments, strains also expressed mCherry from a strong, constitutive promoter (Pdc1-mCherry fusion).
GFP fluorescence above the background (parental strain) was detected by confocal fluorescence microscopy in cells carrying the reporter in all tested loci (Fig. 1b; Supplementary Fig. 1).Importantly, we observed reduced GFP signal in yeast cells with the GFP reporter inserted at both subtelomeric loci, especially in COS12.It must be noted that such silencing occurred in a variegated manner, namely, that the GFP signal was very low in some cells and higher in others.Such variegated gene expression is usually observed at silenced loci in yeast (Gottschling et al. 1990).The GFP signal at the nonsubtelomeric CUP9-5′ locus was also variable, but clearly higher than observed at the subtelomeric loci (Supplementary Fig. 1) To test the reporting potential of the system based on uracil metabolism, we grew the strains in media lacking uracil or containing 5-FOA (Fig. 1c).Strains carrying the dual reporter at both subtelomeric COS12 and YFR057W loci were able to grow in medium lacking uracil.However, only the strain with the reported inserted at COS12 was able to grow in 5-FOA medium.This result confirmed that silencing is incomplete at either locus and is indeed stronger at COS12.In fact, growth of the strain with the reporter at YFR057W in the presence of 5-FOA was not observed, as if silencing was not occurring at this locus.It must be noted that insertions were designed to replace the native loci, hence promoters driving the expression of the URA3-GFP module are differentially oriented relative to the telomere, which could explain the observed differences in silencing levels of the dual reporter (Fig. 1a; see chromosomal features in Supplementary Table 4).
We also confirmed the dependency of silencing on the SIR complex by inserting the reporter in a sir3Δ strain.As expected, in this background even the strain with the reporter inserted at COS12 was not able to grow on a medium containing 5-FOA, showing that reporter silencing is fully dependent on the integrity of the SIR complex.This observation was quantitatively confirmed by flow cytometry of cells in the late-log growth phases.The mean GFP expression in cells with the reporter inserted at the COS12 and YFR057W loci was higher in the sir3Δ compared to the WT background.This indicates that repression of the URA3-GFP reporter at the selected subtelomeric loci depends at least partially on Sir3 activity (Fig. 1d).
Together, these results show that the dual URA3-GFP reporter system allows the measurement of gene silencing level by two independent readouts.First, silencing can be estimated in a semiquantitative manner using the classical 5-FOA assay based on URA3 expression and its effect on cell growth, allowing a more direct comparison with previous findings.In addition, GFP fluorescence measurements by flow cytometry provide a quantitative readout that is amenable to high-throughput screening with a larger dynamic range due to increased sensitivity to subtle silencing effects.

Levels of gene silencing at different subtelomeric regions
The subtelomeric loci YFR057W and COS12 have been thoroughly used to study gene silencing in budding yeast (Vega-Palas et al. 2000;Mondoux and Zakian 2007).Yet, there are 30 other subtelomeric regions in S. cerevisiae, many of which remain poorly characterized.To determine the level of gene silencing throughout the genome and to understand whether silencing at YFR057W and COS12 are representative of overall subtelomeric silencing, we integrated the dual URA3-GFP reporter at seven other members of the COS gene family, each located in the vicinity of different telomeres (see Supplementary Table 4 for insertion sites and chromosome features).These genes are not essential and represent, in all but one case, the first gene adjacent to the subtelomeric core X element at the centromere-proximal side.As for YFR057W and COS12, the reporter was integrated by full replacement of each ORF.
Different reporter expression levels were observed in the subtelomeric-insertions, as inferred from the strain's capacity to grow on 5-FOA, ranging from full growth of the COS8 insertion (strongest reporter silencing) to almost no growth in the COS5 insertion (no silencing) (Fig. 2a).These COS8 and COS5 extreme cases behaved similarly to the parental no-expression and nonsubtelomeric unsilenced controls, respectively.In terms of silencing reported by 5-FOA growth, the COS12 and YFR057W insertions were also two extreme cases of strong and undetectable silencing, respectively.Several insertions in the telomere vicinity resulted in little or no apparent silencing in the semiquantitative 5-FOA assay; such was the case of the COS4, COS2, COS10, YFR057W, and COS5 insertions.However, all strains with subtelomeric integrations showed decreased average GFP expression compared to integrations at the nonsubtelomeric CUP9-5′ locus (Fig. 2b; Supplementary Fig. 1).Silencing strength determined by growth on 5-FOA ranked similar to that determined by GFP expression across the subtelomeric loci tested, except for the COS5 locus which showed partial silencing in the 5-FOA assay but high GFP expression.In addition, we note that the GFP assay provides a wider dynamic range.For instance, there was no growth on 5-FOA of the strain with the reporter at the YFR057W locus, which was mostly indistinguishable from the nontelomeric CUP9-5′ control insertion; in contrast, the quantitative flow-cytometry assay showed that mean GFP expression was 1.64-fold lower in the YFR057W compared to the CUP9-5′ insertion.Sir3-dependency of such decreased expression level confirms silencing at YFR057W, which was successfully scored by GFP signal but not by growth on 5-FOA (Fig. 1c, d) Therefore, these assays show that GFP expression at single-cell level, measured by flow cytometry, resolves slight differences in silencing levels compared to the more qualitative, conventional assay based on URA3 expression.
Our results indicate that there is a varying level of gene silencing in the subtelomeric regions of the S. cerevisiae genome.A simple explanation for such variation could be the differences in the distance of the COS genes to the telomere.However, we did not observe such a relation of ORF's ATG distance to the telomere (r = −0.18,P > 0.05; Pearson) or to the core X element (r = −0.17,P > 0.05; Pearson) (Supplementary Table 4).Hence, it is likely that other factors of the subtelomeric context contribute to the observed differences in silencing of the same reporter.In this study, we focused on the COS12 and YFR057W loci, which not only have been previously studied at a smaller scale but also cover the range of silencing strengths of the subtelomeric regions in S. cerevisiae as revealed from our results.

Genome-wide screening for potential subtelomeric-silencing factors
Over 100 genes are known to influence gene silencing at subtelomeric regions in S. cerevisiae (reviewed in Supplementary Note 1).To screen for novel genes or pathways involved in gene silencing in a systematic and comprehensive manner, we sought to generate two collections of nonessential gene knockouts bearing the dual URA3-GFP reporter at the COS12 or YFR057W loci.These subtelomeric loci are at the extremes of the silencing intensity spectrum (Fig. 2a, b) and have previously been used to study mechanisms involved in TPE.To confirm that insertion of the URA3-GFP reporter did not have major effects on the local chromatin state, we performed nucleosome-scanning assays (NuSA) of the YFR057W promoter in the wild-type and the yfr057wΔ:: URA3-GFP strain in which the insertion is 1.5 kb larger than the native YFR057W ORF.Nucleosome positioning was very similar in the two strains, suggesting that nucleosome distribution was not modified by the sequence replacement, or the presence of the new promoter caused by the insertion (Supplementary Fig. 2).
Deletion collections bearing the silencing reporter were generated using a SGA approach (Tong and Boone 2006), by crossing strains with the integration of the URA3-GFP at either loci with a collection of ∼4,500 knockout strains, each one with a nonessential gene replaced by the KanMX cassette (Fig. 3a).For competitive flow-cytometry analysis, the URA3-GFP integrations were done in a strain background constitutively expressing the fluorescent mCherry protein (RFP) in the neutral HO locus and an isogenic wild-type strain was labeled with the BFP.The resulting haploid strain collections bear the URA3-GFP reporter at COS12 or YFR057W, a KanMX gene replacement, and express RFP constitutively.
To measure the effect of the deletion of each nonessential gene on subtelomeric silencing in the two query strains, we used highthroughput flow cytometry to measure GFP expression.For increased comparative resolution, each RFP-labeled knockout strain bearing the dual reporter at COS12 or YFR057W was grown in coculture with the isogenic BFP-labeled wild type bearing the URA3-GFP reporter at the corresponding locus, allowing to tell apart the GFP signal of the knockout and wild-type populations in each sample (Fig. 3b).Typically, between 5,000 and 15,000 cells were measured from each competitive population.A silencing score (Si score) was defined as the ratio of average GFP signals of the mutant and the wild-type reference strains.Based on this metric, we observed that, while most deletions caused no effects on gene expression, many gene deletions resulted in diminished gene silencing (higher GFP signal, Si > 1), while others resulted in increased silencing (lower GFP signal, Si < 1) (Fig. 3c, d).Results of both genomewide screens are provided in Supplementary Dataset 1.
Using a 10% false-discovery rate, 69 and 55 deletions resulted in decreased gene silencing at the COS12 and YFR057W loci, respectively, while 8 and 25 resulted in increased silencing.There was a trend of more gene deletions having a negative effect on silencing at the COS12 locus; this trend was less evident at YFR057W, which could be associated with the higher basal expression at the later compared to the former locus.

Silencing effects are reproducible and consistent between the two readouts of the reporter system
To assess experimental replicability within and between screens, we first assayed a fraction of the deletion strains with the COS12 insertion in two independent experiments, which showed a good rank correlation (Supplementary Fig. 3; r = 0.63, P < 10 −168 , Spearman).Importantly, the Si score at both loci was also significantly correlated for the 3,677 gene-deletion mutants that were successfully screened in both COS12 and YFR057W assays (Supplementary Fig. 3; r = 0.56, P < 10 −301 , Spearman).
To quantitatively verify some hits from our screens, we selected a subset of 41 hits above the 10% False Discovery Rate (FDR)cutoff.These hits included subunits of protein complexes identified as PAF1 (cdc73Δ), pyruvate dehydrogenase complex (PDC) (pdb1Δ, pda1Δ, pdx1Δ), KU (yku70Δ, yku80Δ), INO80 (ies2Δ), RNA pol II interactors (rtr1Δ, iwr1Δ), 18 mitochondrial genes, and individual genes that were not part of a clear complex or functional category.We used the same flow cytometry strategy to measure changes in the GFP signal in the COS12 insertion by performing five technical replicates in competition assays with the BFP-labeled WT reference strain.We used the RFP-labeled sir3 deletion mutant and the parental WT strain bearing the URA3-GFP insertion as positive and negative controls, respectively.Of the retested hits, 92.6% showed a significant increase in GFP signal compared to that of the parental reference (Supplementary Fig. 4; P < 0.05, t-test).It must be noted that most validated hits showed a modest Si score, yet several mutants showed average values above 0.5, including the sir3Δ control.Together, these results suggest that screening of changes in the expression of the GFP reporter   (Tong and Boone 2006).For this, a parental strain carrying the reporter at either locus was crossed with a gene-deletion collection generated using the KanMX marker (Giaever et al. 2002).b) Each of the resulting mutants was grown in co-culture with a BFP-expressing reference strain harboring the subtelomeric URA3-GFP reporter at the same locus, with no gene deleted.GFP expression of each pair of mutant RFP and reference BFP strains was measured simultaneously by flow cytometry; separation of the populations was done using their constitutive RFP or BFP signals.The ratio of the average GFP signals of the mutant strain and reference strains was defined as the silencing score (Si score).c) Cumulative distribution of Si score obtained from screening the gene-deletion collection with the reporter at COS12 (n = 3,716) and YFR057W (n = 4,193).
Strains that overexpress GFP are marked in red (FDR < 10%).d) Distribution of GFP expression of representative strains with distinct Si scores; vertical lines are the average GFP signal of mutant and reference populations.
inserted at subtelomeric loci provides a robust, straightforward way to screen for genetic factors involved in gene silencing.
To compare the results of the screens using the conventional method based on repression of Ura3 activity, we assayed a subset of the top-ranked hits for growth on 5-FOA medium.We used the ura3Δ knockout strain and the parental cos12Δ::URA3-GFP insertion as controls.Sixteen out of 21 strains tested (76.2%) showed mild to strong growth defect in 5-FOA, suggesting impaired gene silencing at the reporter (Fig. 4).As expected, deletion of the FUR4-encoded uracil permease results in a mild growth defect, likely due to a direct regulatory effect on the URA3 promoter and not a telomere-position effect.Among the strains with the strongest silencing defect were mutants of genes known to be involved in subtelomeric gene silencing, such as yku70Δ, yku80Δ, and spt21Δ, which was consistent with their high GFP-signal increase in flow-cytometry validation experiments.However, we note that several hits with a modest increase in GFP signal did not result in reduced growth in 5-FOA; this was the case of the pdb1Δ and pda1Δ strains deleted for genes of the PDC complex, among others.We confirmed that this was not the result of experimental variation in the 5-FOA assay, as replicate assays showed very similar semiquantitative trends (Supplementary Fig. 5).Together, these results indicate that both readouts of the reporter system are consistent with each other, but with a larger dynamic range observed for the GFP signal readout for gene deletions resulting in modest impairment of gene silencing.

Scored silencing effects are consistent with the literature
We tested whether previously described silencing genes were overrepresented at the tails of the Si score distribution, as would be expected if the screens recapitulated current knowledge of silencing factors in yeast.To this end, we assembled a catalog of 132 genes from the SGD (GO Term: chromatin silencing at telomere), an extensive revision of the subject (Mondoux and Zakian;2006), and our own curation of the literature (Supplementary Note 1).Of the 132 genes, 72 were evaluated in the COS12 screen and 54% belong to the two higher or lower deciles of the Si score distribution, while 54% of the 85 that were measured in the YFR057W insertion were in the extreme deciles (Fig. 5a, b).The observed enrichments strongly suggest that our large-scale screens revealed genetic factors involved in subtelomeric gene silencing, especially if we consider that the reference catalog includes genes that had been identified in many independent studies, using different methodologies.
Examples of silencing factors confirmed by our screens include genes known to influence telomere length, such as RIF1 (Hardy et al. 1992), RIF2 (Wotton and Shore 1997), YKU70, and YKU80 (Williams et al. 2014).The SPT21 deletion was also part of the top hits in both subtelomeric silencing screens and its knockout mutant has been previously reported to show loss of silencing at subtelomeric positions and altered telomere length (Gatbonton et al. 2006).Spt21 physically interacts with Spt10 (Kurat et al. 2014) and both are required for proper silencing in a native YFR057W telomere context (Chang and Winston 2011).In addition, our screens scored other genes related to telomere length (Askree et al. 2004) (CDC73, RAD50, UPF3) and telomere capping (Addinall et al. 2008) (MTC7).Likewise, different gene knockouts of the elongator complex have been previously reported to diminish the silencing of subtelomeric reporters at the VII-L subtelomeric locus, where COS12 is located (Li et al. 2009).In our work, in both the COS12 and YFR057W screens, deletions of genes of this complex, ELP2, ELP3, and ELP4, were among the top hits.
Another group of genes related to transcriptional regulation obtained at the top positions of the screens were members of the SET3 chromatin remodeling complex (HOS2, SIF2, SET3, and SNT1).Interestingly, subunits of the SAS complex showed opposite effects depending on the query locus.For the YFR057W locus screen, sas4Δ and sas5Δ showed a negative effect on silencing, while the subunits Sas2, Sas4, and Sas5 had a positive effect on silencing at the COS12 locus.These opposite effects were expected since previous studies have shown that components of the SAS complex display locus-dependent opposite silencing effects.In particular, Sas2 activity weakens silencing at a defective HMR-E silencer in the HMR locus but promotes it at the HML locus and telomeres (Reifsnyder et al. 1996;Ehrenhofer-Murray et al. 1997).

Deletion of LGE1 results in robust impairment of subtelomeric gene silencing
The lge1Δ deletion strain was among the top potential hits of impaired subtelomeric silencing in both our genome-wide screens.This strain resulted in high GFP-signal increase of the lge1Δ strain in our highly replicated flow-cytometry experiments (Supplementary Fig. 4), which was consistent with the fact that the lge1Δ strain had the most severe 5-FOA growth defect in our validation experiments (Fig. 4).Lge1 is involved in H2B ubiquitination mediated by the Rad6/Bre1 complex (Kim et al. 2018), but its precise molecular activity remains unknown.Functionally, Lge1 has been shown to play a role in histone modification and DNA repair, although a direct connection to subtelomeric silencing has not been reported.To validate the effects of lge1Δ on both query loci, we first measured changes in the GFP reporter in an independent, replicated experiment (Fig. 6a).Importantly, this experiment monitored expression at different growth phases, showing that LGE1 impaired strains show consistent and robust increased gene expression at the silenced loci, relative to the WT.
Given that our reporter system could result in increased expression due to activation of the URA3 promoter and not a general effect on TPE, we used RT-qPCR to test whether the observed effect of LGE1 impairment was still observed on the native COS12 and YFR057W genes, with no URA3-GFP insertion (Fig. 6b).We observed that both query genes showed a significant 2-fold expression increase in the lge1Δ compared to the parental strain, indicating that Lge1 activity influences gene silencing independently of effects on the reporter system used.Together, these data confirm that Lge1 is a novel positive subtelomeric silencing factor in budding yeast.

Mitochondrial impairment is not associated with changes in subtelomeric gene expression or nucleosome positioning
We investigated the enrichment of a large set of mitochondrial genes among the mutants with the highest Si score in our screens.To this end, we focused on the subunits of the PDC involved in the conversion of pyruvate to acetyl-CoA (PDB1, PDA1, PDX1, and LAT1).To confirm the effect of PDC subunits on the silencing, these knockouts were measured again by flow cytometry at both subtelomeric query loci, which we compared in parallel to subunits of chromatin remodeling complexes well known to affect chromatin structure (Sir3) and were hits in our screens (Set2 and Ies2).All PDC knockouts showed significant Si score differences when compared to the WT strain (Fig. 7a; P < 10 −2 and P < 10 −3 for YFR057W and COS12 insertions, respectively).Even though most mutants of the PDC subunits showed stronger effects on the silencing of the dual reporter than the mutants of chromatin remodeling complexes that were used as a reference, their effects were less evident or nondetectable in the 5-FOA assay (Fig. 4).Given that gene expression effects could be due to an indirect metabolic effect on the URA3 reporter used (Rossmann et al. 2011), we evaluated the effects of mutants of PDC subunits pdb1Δ and pda1Δ on native YFR057W expression by RT-qPCR (Fig. 6b).This analysis showed no significant effects on gene expression for all but the COS12 locus on the pdb1Δ background, suggesting that the observed reporter effects are associated with the promoter system used, rather than a telomere-position effect.
We also tested whether the effects on silencing observed in the mutants of the PDC subunits were due to chromatin changes at the nucleosome level, which are expected in bona fide TPE.We carried out NuSA at the YFR057W promoter and the URA3-GFP insertion sequences in the set2Δ strain and, as a reference, mutants of subunits of chromatin remodeling complexes.In each NuSA assay, nucleosome positioning was compared to the parental strain.Deletion of PDB1 did not influence nucleosome occupancy at the promoter; nucleosome distribution was very similar to the Fig. 4. Some changes in gene expression are not resolved by the conventional 5-FOA silencing assay.Gene silencing assessment by 5-FOA growth assays at the COS12 locus of selected hits of the quantitative screen of overrepresented functional categories or protein complexes.Each strain was grown to saturation and plated in serial dilutions onto YPD, SC lacking uracil, and SC plus 5-FOA media.parental strain.In contrast, the absence of IES2, SIR3, and SET2 results in a reduction in nucleosome positioning over the entire promoter region, as expected, even at the well-positioned nucleosome at the −96 bp position (Fig. 7b).In agreement with its relatively lower Si score, the deletion of SET2 showed the weakest effect on nucleosome distribution at the YFR057W query region.Together, these results suggest that, at least for the top PDC hits, the effects of mitochondrial function on gene expression depend on the reporter system used, either due to a higher basal expression or a direct activation effect of the URA3 promoter.

A global picture of subtelomeric silencing in yeast
Our genetic screens provide an opportunity to revise the general cellular and molecular functions contributing to subtelomeric silencing, using results from a comprehensive genetic dataset.To this end, we used a functional analysis based on Cohen's kappa (Huang da et al. 2009), as previously described (Campos et al. 2018).This analysis evaluates the relationship between gene pairs by establishing the overall agreement between a set of associated evaluators.Here, evaluators included GO and phenotypic terms that have been previously ascribed to the genes of interest, as reported in SGD.We tested a set of 266 genes (Supplementary Table 3) including 141 potential factors from our two screens (top and bottom Si score rank, FDR <10%) and 125 silencing factors reported in the TPE literature (Supplementary Note 1).
Using the kappa-based functional analysis, we identified 11 clusters of genes, each composed of three to a dozen genes (Fig. 8).The main cellular processes associated with the clusters were histone and chromatin modification and telomere maintenance.Two high-scoring hits from the screen, YKU70 and YKU80, clustered together with RRM3 forming a cluster related to maintenance, with roles subtelomeric silencing.Most of the observed clusters included genes to different categories that impact chromatin structure or function, which had previously been reported as a factor mediating subtelomeric silencing in yeast.These included genes with roles in nucleosome positioning or remodeling (ISW2 and CHD1) that were connected to the FUN30 and INO80 genes.The latter two genes have been previously reported to participate in chromatin silencing.Another cluster included CDC73, LEO1, and RTF1 which all are part of the multifunctional Paf1 complex involved in RNA polymerase II transcriptional elongation, RNA processing, and histone modification during elongation.Interestingly, the novel silencing factor LGE1 was part of two chromatinrelated clusters, linked to histone deacetylases, histone methyltransferases involved in chromatin silencing at telomeres, and other chromatin remodeling complexes.This finding is consistent with the function of Lge1 as a histone H2B-ubiquitination cofactor, and it suggests a possible mechanism through which Lge1 impacts subtelomeric silencing.Our screens also revealed a cluster of several genes involved in ribosomal function and another of uncharacterized ORFs.Further validation is needed to confirm the potential role of these genes in subtelomeric silencing.The log 2 Si score of the WT was normalized to zero for each locus and each deletion replicate was compared to the corresponding parental average (t-test; *P < 0.05, **P < 0.01; ***P < 0.005).b) Nucleosome positioning at theYFR057W promoter in the sir3Δ, pdb1Δ, ies2Δ, and set2Δ mutant strains.NuSAs were performed on strains that do not have the double reporter by growing them in SC medium containing uracil (20 mg/L) at 30°C and harvested at late-log phase (see Materials and Methods).Relative protection was calculated as a ratio using the VCX1 gene as a reference since a well-positioned nucleosome is found at the +250 bp position of this ORF.For each primer pair, the midpoint of the PCR fragment is shown as a solid dot and overall, they amplify from around −650 to +222 bp of the YFR057W locus.The coordinates are given relative to the ATG (+1).
URA3-GFP reporter gene system coupled with high-throughput flow cytometry.Our method proved to be more sensitive than classical 5-FOA assays, allowing the detection of subtle differences in gene silencing across a variety of subtelomeric loci in the yeast genome.Telomere length is a known determinant of gene silencing, and we did observe changes in gene expression in mutants altered in telomere length and maintenance.However, the distance from the insertion sites of the reporter to the telomere did not correlate with silencing levels, at least within the range of distances that we assessed in nine different insertion sites.Six reporters oriented away from the telomere were those with stronger silencing, compared with two cases that were oriented toward the telomere.Further research will be needed to determine which other factors, such as telomere structure and orientation of the gene relative to the telomere, contribute to the silencing variation that we observed in the different subtelomeric regions of the yeast genome.
We show that the measurement of GFP expression by flow cytometry provides a readout with a wider dynamic range than the growth on medium containing 5-FOA, even though the results of the two assays were in general agreement; this was also the case of a subset of genes that were measured repeatedly by flow cytometry.Furthermore, the top hits identified by our screens are enriched in genes previously known to affect gene silencing.Clear examples are those known to affect telomere length and members of the Su(var)3-9 Enhancer-of-zeste and Trithorax Fig. 8.A functional view of subtelomeric gene silencing in yeast.Functional annotation of top-ranked Si score genes and genes previously known to influence gene silencing by kappa-based functional analysis.Functional clusters are represented as a network where genes are oval nodes that are connected by edges when the kappa value indicates functional agreement between genes (k > 0.35).A large cluster of genes with mitochondrial function is not shown, given that effects of mitochondrial function on gene expression do not seem to be due to subtelomeric silencing, but rather to the reporter system used.Colors indicate the different clusters within the network.Asterisks indicate association with different clusters besides the one indicated by node color; the chromatin-organization cluster in light orange includes three subclusters.Ovals with a solid outline indicate potential silencing factors from the top FDR 10% of the screens, while ovals with no outline are genes associated with TPE silencing in previous reports.
(SET), Something About Silencing (SAS), Ku70/80, and PAF1 complexes.These results indicate that our approach is robust and amenable to large-scale analysis of gene silencing.
Most of the functional categories associated with silencing are related to chromatin conformation and modification, in one way or another, and it is worth noting that most potential factors are known players of gene silencing in yeast.Nonetheless, LGE1 was among the potential silencing factors that had not been directly associated with subtelomeric gene silencing.Its knockout was one of the mutants that showed the strongest effect in our screens and its role was further validated by growth in 5-FOA and, importantly, by qPCR-analysis of gene expression on the native gene.One possible explanation for the role of Lge1 in subtelomeric silencing may be its connection with Set1 and Dot1 (Wood et al. 2003).It is known that monoubiquitination of H2B mediated by Rad6-Bre1-Lge1 is a prerequisite to the H3K4 and H3K79 methylation produced by Set1/Complex Proteins Associated with Set1 (COMPASS) and Dot1, respectively (Wood et al. 2003).Lysine methylation of H3K79 by Dot1 has been shown to be important during transcriptional elongation by the Paf1 complex and to regulate telomeric silencing (Ng et al. 2002).Thus, it is possible that the loss of Set1 dependent methylation in a lge1 knockout could affect silencing by disrupting the ability of Sir2 and Sir3 to form heterochromatin.An alternative route, although less clear, could be through the association of Lge1 with the DNA repair protein Ku70.The mutants of these genes show synthetic lethality at high temperature (L.2008), and Ku70 also shows a synthetic lethal interaction with Set1, the histone methyltransferase that is central for subtelomeric silencing.
Unexpectedly, a large set of genes with mitochondrial function were enriched in our screens.The strongest effect was observed in the mutants of the pyruvate decarboxylase complex, which showed clear and robust increases in subtelomeric GFP expression and impaired growth in 5-FOA.However, further direct expression measurements of the native promoters by RT-qPCR and nucleosome-position analysis of some of the mitochondrial mutants suggested that the effect on silencing is specific to the reporter system used (Figs. 6a and 7a,b).Even though NuSA was carried out only at the YFR057W locus showing weaker silencing, the most plausible explanation is that deletion of the mitochondrial genes is specifically interacting with pathways that affect expression from the URA3 promoter, as previously suggested (Rossmann et al. 2011).This interpretation is indeed the case for genes detected by our screens and that are involved in pathways related to the availability of uracil, e.g. the plasma membrane uracil permease Fur4 and the uracil biosynthetic genes URA5 and URA1.Although further work is required to understand the exact connection between mitochondrial genes and the silencing effects observed in our reporter system, our results raise a word of caution for the use of the URA3 gene for assessing gene silencing, which is routinely done with the use of the 5-FOA growth assay.Indeed, a recent study has tested different fluorescent reporters to overcome the potential problems of using URA3-expressing cells and 5-FOA to measure gene silencing (Shaban et al. 2023).In future studies, in addition to using fluorescence reporters, it would be very informative to replace the URA3 promoter with other, less metabolic-sensitive promoters in a double reporter assay.
Our screens did not include mutants of genes that are essential or those resulting in sterile strains since they are not amenable to SGA.This is relevant given that some essential genes such as RAP1 (Kyrion et al. 1993) and ABF1 (Pryde and Louis 1999) are known to have strong roles in silencing.Similarly, SIR2 and SIR3, whose deletion causes sterility, are the main players of subtelomeric silencing.In this work, we generated some of the reference strains by direct PCR-based transformation, but essentiality and sterility limitations could be overcome by using strain collections of conditional mutants.
Most subtelomeric genes identified in previous studies were those with strong telomere-position effects.Our work shows that there are many other genes that have subtler effects and that are more readily detected by sensitive, quantitative methodologies.The flow-cytometry-based approach presented here also allows for obtaining single-cell expression data to identify variegation trends in large populations.Besides being quantitative and amenable to high-throughput screening, our strategy is versatile since different promoters and fluorophores can be combined with the many gene-deletion collections that are available for budding yeast.We anticipate that using our systematic approach with other promoters and reporters will allow overcoming the caveats detected in our screens, revisit previous studies, and globally understand the molecular mechanisms of subtelomeric gene silencing in budding yeast and other organisms.

Fig. 1 .
Fig. 1.A dual-reporter system to assess gene silencing by flow cytometry.a) Schematic representation of the URA3-GFP reporter cassette and its integration at the subtelomeric loci COS12 and YFR057W by replacing the open reading frames, and at the nonsubtelomeric CUP9-5′ intergenic region.b) Confocal fluorescence microscopy images of S. cerevisiae cells, bearing the URA3-GFP reporter integrated at subtelomeric loci COS12, YFR057W and at the nonsubtelomeric locus CUP9.Strains also express mCherry constitutively; GFP and mCherry channels are shown.c) 5-FOA growth assays of the parental strain (ura3Δ) and the URA3-GFP integrations at COS12 and YFR057W loci in a parental WT or sir3Δ background.d) Distribution of GFP signal measured by flow cytometry in the strains where the double reporter is inserted at COS12 or YFR057W in the parental or sir3Δ strain backgrounds.Dashed lines show the mean GFP signal for each cell population.

Fig. 2 .
Fig. 2. Gene silencing varies across different subtelomeric regions of the yeast genome.a) Strains bearing the URA3-GFP dual reporter integrated at the indicated subtelomeric locations by replacing the native ORF and maintaining the native orientation relative to the telomere, were subjected to 5-FOA growth assays.The parental strain (no reporter) is ura3Δ.YPD and SC minus uracil (SC -ura) plates were incubated for 48 h, while 5-FOA plates were incubated for 72 h, all at 30°C.b) Distribution of bulk GFP signal measured by flow cytometry.Strains were grown on liquid SC +20 mg/L uracil and assayed by flow cytometry in the late-log phase.The gray dashed line is the mean GFP background signal of the parental strain, while the red vertical lines indicate the mean GFP signal of each insertion.Integration at the CUP9-5′ intergenic region (asterisk) was used as a nonsubtelomeric reference.Parental and reporter-insertion strains are ordered in both panels by their ranked average GFP signal (red lines), from the lower (top) to the higher average signal (bottom).

Fig. 3 .
Fig.3.Genome-wide identification of genes affecting subtelomeric gene silencing by high-throughput flow cytometry.a, b) Schematic representation of the screen for subtelomeric gene silencing.a) A large collection of mCherry-expressing knockout mutants harboring the subtelomeric URA3-GFP reporter at either the COS12 or the YFR057w locus was generated by SGA(Tong and Boone 2006).For this, a parental strain carrying the reporter at either locus was crossed with a gene-deletion collection generated using the KanMX marker(Giaever et al. 2002).b) Each of the resulting mutants was grown in co-culture with a BFP-expressing reference strain harboring the subtelomeric URA3-GFP reporter at the same locus, with no gene deleted.GFP expression of each pair of mutant RFP and reference BFP strains was measured simultaneously by flow cytometry; separation of the populations was done using their constitutive RFP or BFP signals.The ratio of the average GFP signals of the mutant strain and reference strains was defined as the silencing score (Si score).c) Cumulative distribution of Si score obtained from screening the gene-deletion collection with the reporter at COS12 (n = 3,716) and YFR057W (n = 4,193).Strains that overexpress GFP are marked in red (FDR < 10%).d) Distribution of GFP expression of representative strains with distinct Si scores; vertical lines are the average GFP signal of mutant and reference populations.

Fig. 5 .
Fig.5.Silencing effects pinpoint previously known genes involved in subtelomeric silencing.Enrichment of genes previously known to be involved in gene silencing at the extreme deciles of the Si score distribution from the screens at a) COS12 and b) YFR057W.A list of 132 previously known silencing genes from SGD and an extensive literature curation of silencing factors (Supplementary Note 1) was scored at the cumulative distribution of Si score, divided into deciles.Most of the known silencing genes are at both ends of the distribution.The radius of each circle indicates the number of known genes at each decile, from n = 3 (the smallest) to n = 27 (the largest), and circles filled in gray were literature hits significantly enriched at the decile (P < 0.05, hypergeometric test).

Fig. 7 .
Fig. 7.No evidence for a direct role for mitochondrial function in subtelomeric gene silencing.a) Comparison of Si scores of knockouts of genes that code for the PDC complex subunits (pdx1Δ, pda1Δ, pdb1Δ, and lat1Δ) and chromatin remodeling factors (set2Δ, sir3Δ, and ies2Δ) measured by flow cytometry.The log 2 Si score of the WT was normalized to zero for each locus and each deletion replicate was compared to the corresponding parental average (t-test; *P < 0.05, **P < 0.01; ***P < 0.005).b) Nucleosome positioning at theYFR057W promoter in the sir3Δ, pdb1Δ, ies2Δ, and set2Δ mutant strains.NuSAs were performed on strains that do not have the double reporter by growing them in SC medium containing uracil (20 mg/L) at 30°C and harvested at late-log phase (see Materials and Methods).Relative protection was calculated as a ratio using the VCX1 gene as a reference since a well-positioned nucleosome is found at the +250 bp position of this ORF.For each primer pair, the midpoint of the PCR fragment is shown as a solid dot and overall, they amplify from around −650 to +222 bp of the YFR057W locus.The coordinates are given relative to the ATG (+1).