The Fun30 Chromatin Remodeler Fft3 Controls Nuclear Organization and Chromatin Structure of Insulators and Subtelomeres in Fission Yeast

In eukaryotic cells, local chromatin structure and chromatin organization in the nucleus both influence transcriptional regulation. At the local level, the Fun30 chromatin remodeler Fft3 is essential for maintaining proper chromatin structure at centromeres and subtelomeres in fission yeast. Using genome-wide mapping and live cell imaging, we show that this role is linked to controlling nuclear organization of its targets. In fft3∆ cells, subtelomeres lose their association with the LEM domain protein Man1 at the nuclear periphery and move to the interior of the nucleus. Furthermore, genes in these domains are upregulated and active chromatin marks increase. Fft3 is also enriched at retrotransposon-derived long terminal repeat (LTR) elements and at tRNA genes. In cells lacking Fft3, these sites lose their peripheral positioning and show reduced nucleosome occupancy. We propose that Fft3 has a global role in mediating association between specific chromatin domains and the nuclear envelope.


Introduction
Nuclear architecture, i.e. non-random positioning of chromosomal loci and nuclear components in three dimensions, is important in organizing genome processes such as transcription, DNA replication and DNA repair [1][2][3].One crucial aspect of this organization is the interaction between chromatin and the nuclear periphery.In many eukaryotic species, chromosomal domains near the nuclear envelope show low expression levels, repressive chromatin marks and low gene density [4][5][6][7].These domains interact with the nuclear lamina [4,5] or inner nuclear membrane (INM) proteins, such as the LEM-domain proteins [6,7].How chromosomal loci are targeted to the nuclear envelope is still poorly understood, although step-wise methylation of lysine 9 on histone H3 (H3K9me) [8,9] and interaction with different nuclear membrane proteins [10] are known to contribute.
In the fission yeast Schizosaccharomyces pombe, the INM protein Man1 interacts with about a third of the genome, mainly at lowly transcribed genes [7].Especially striking, we observed Man1 binding to large domains adjacent to the telomeres of chromosomes I and II, which are characterized by a unique type of chromatin [11].These subtelomeric domains show low levels of histone methylation, in both the repressive mark H3K9me2 and the active mark H3K4me2, as well as lower histone acetylation and H2A.Z levels compared to euchromatin [11,12].Many genes in these regions are lowly expressed in rapidly growing cells, but are induced during nutritional stress or meiosis [13].The borders of the subtelomeric domains are bound by the chromatin remodeling factor Fft3 [14].Maintenance of the special chromatin state inside these domains in rapidly growing cells depends on Fft3, since a deletion of the remodeler results in a strong upregulation of subtelomeric genes and spreading of euchromatin marks into the domains.
Here, we explore the interplay between genome organization and transcriptional regulation.We show that Fft3 maintains chromatin structure and peripheral positioning of subtelomeres by binding to and remodeling nucleosomes at their borders.In addition to the subtelomeres, Fft3 interacts with TFIIIC and Pol III at tRNA genes and affects nucleosome positioning and interaction with the nuclear periphery.We propose that Fft3 maintains nucleosome stability and peripheral positioning of these elements and thereby preserves proper genome organization.

Deletion of fft3 results in altered chromatin properties and decreased nuclear envelope interactions at subtelomeres
We previously showed that subtelomeric genes are upregulated in cells lacking the chromatin remodeler Fft3 [14].To see whether these expression changes coincide with an altered chromatin structure, we performed genome-wide ChIP-chip for three hallmarks of active chromatin: RNA polymerase II (Pol II), the histone variant H2A.Z and the histone modification H4K12Ac.All three marks show a striking increase over the subtelomeric chromatin domains on chromosomes I and II (Fig. 1A) in fft3Δ cells compared to wild type, allowing for a chromatin structure more permissive to transcription.These increases are significant when compared When we mapped interactions between the INM protein Man1 and chromatin, we observed that subtelomeres are strongly enriched for Man1 [7].We hypothesized that the expression and chromatin changes in fft3Δ cells might be accompanied by an altered nuclear organization.Indeed, DamID mapping revealed that Man1 interactions with the subtelomeres are strongly reduced in fft3Δ cells (Fig. 2A-C, S2 Fig) .Interestingly, Fft3 itself does not bind the subtelomeric domains, only their borders (Fig. 2A,D).
Taken together, these results suggest that deleting fft3 leads to drastic changes in chromatin composition and positioning of these large domains while interacting only with the domain borders.

Fft3 and Bqt4 cooperate in anchoring subtelomeres to the nuclear envelope
Our data suggest that Man1 and Fft3 have essential roles in tethering subtelomeres to the NE.To explore this further, we monitored the intranuclear position of the telomeres in live cells using the telomere associated protein Taz1.Based on their relative distance to the nuclear envelope, the Taz1 signals in each cell were assigned to one of three zones of equal volume (Fig. 3A).As expected, about 75% of telomere signals are localized in the outermost zone, close to the nuclear envelope, in wild type cells (Fig. 3B).We observed no change in telomere localization in fft3Δ cells (Fig. 3B), suggesting that the telomeres themselves are unaffected by the changes in the subtelomeres.The anchoring of telomeres to NE has been shown to depend on the Bqt4 protein [23].Indeed, we observed that deleting bqt4 reduced the number of telomeres in zone 1 to below 60%, with an increased number of cells showing telomere signals in the inner zones (Fig. 3C).The loss of peripheral localization is exacerbated in cells lacking both Bqt4 and Fft3 (Fig. 3D-E), with more than 50% of the cells now showing telomere signals in the two innermost zones.We obtained similar results using fluorescence in situ hybridization (FISH, S3 Fig) .Signals from a subtelomeric FISH probe showed a modest shift towards the interior in fft3Δ cells compared to wildtype.In fft3Δ bqt4Δ cells, the majority of FISH signals were found in the two innermost zones, while deletion of bqt4 by itself had no effect on localization.Taken together, these results demonstrate that Bqt4 and Fft3 work together in anchoring the subtelomeres to the nuclear envelope, with Bqt4 attaching the telomere end and Fft3 preserving the interaction with Man1 over the subtelomeres.

Fft3 interacts with solo LTRs at subtelomeric borders and genome-wide
We next asked what marks the subtelomeric borders aside from Fft3 binding.Interestingly, all four borders feature long terminal repeat elements (LTRs) either at the Fft3 binding sites or in close proximity (Fig. 4A).Assuming that this is a sequence feature that Fft3 recognizes, we looked at all LTRs in the S. pombe genome and found that Fft3 is strongly enriched over these elements genome-wide (Fig. 4B).To study how Fft3 affects nucleosome positioning and occupancy, we performed micrococcal nuclease digestion followed by sequencing (MNase-seq).Interestingly, we observed a significant decrease in nucleosome occupancy over LTR elements in fft3Δ cells compared to wild type (Fig. 4C).This suggests that Fft3 is required for either positioning or maintaining a nucleosome over LTRs.Since Fft3 alters interactions with the nuclear envelope at subtelomeres, we wondered whether LTRs are affected similarly.In wild type cells, LTR elements are enriched for binding of both Man1 and the nucleoporin Nup85 [24]

Catalytic activity of Fft3 is required for subtelomeric silencing
This raises the question how Fft3 affects chromatin structure at subtelomeric borders.MNaseseq data shows several changes in nucleosome occupancy close to the Fft3 binding sites at the borders (Fig. 5A, S5 Fig) .In most cases, nucleosome occupancy was reduced in fft3Δ cells compared to wild type, suggesting that loss of Fft3 leads to reduced nucleosome stability in these regions.Based on these observations, we hypothesized that the catalytic activity of Fft3 is essential for its functions in subtelomeric chromatin regulation.We therefore constructed a variant of Fft3 with a point mutation in the ATPase domain (fft3-K418R, Fig. 5B), which is known to lead to loss of enzymatic activity [25,26].The resulting catalytically inactive variant of Fft3 is expressed at similar levels (Fig. 5C) and recruited to the same targets as unmodified Fft3 (Fig. 5D Together, these results strongly suggest that ATP-dependent remodeling by Fft3 is directly required to maintain silencing of subtelomeric genes and chromatin structure at the boundaries to the subtelomeres.

Fft3 colocalizes with the Pol III machinery and TFIIIC
After observing the function of Fft3 at subtelomeres and LTR, we set out to explore whether Fft3 also plays a role in regulating chromatin structure elsewhere in the genome.Among other features, we found a preference of Fft3 for snRNA genes, snoRNA genes and replication origins (S8A Fig) .Most prominently, we observed a significant enrichment of Fft3 at tRNA genes (Fig. 6A) compared to the rest of the genome.tRNA genes are transcribed by the RNA polymerase III (Pol III) machinery and require the transcription factor TFIIIC for recruitment of Pol III (reviewed in [27]).We also observed Fft3 enrichment at 5S rRNA genes, which are transcribed by Pol III (S8B Fig) .When we compared the Fft3 binding profile to Pol III and TFIIIC maps [28], we found a strong overlap in binding sites (Fig. 6B).Several reports have identified loci called ETC (Extra-TFIIIC) or COC (chromosome-organizing clamps) sites across the genome that are occupied by TFIIIC but not by any other components of the Pol III machinery [28][29][30].While Fft3 binding coincides with Pol III binding, we did not observe enrichment of Fft3 at ETC/COC sites (Fig. 6C).This suggests that Fft3 binds to actively transcribed Pol III sites.
To search for interaction partners of Fft3, we carried out a yeast two-hybrid screen using the full-length Fft3 protein as the bait against an S. pombe cDNA library.This screen identified several cDNAs encoding Sfc4 (Fig. 6D), a subunit of TFIIIC and homologous to S. cerevisiae Tfc4p and human TFIIIC102 [31].The clones cover a set of tetratricopeptide repeats (TPR) which function as sites for protein-protein interactions [32], suggesting that Fft3 could interact with Sfc4 through this domain.To examine whether Fft3 and Sfc4 also interact in vivo, we constructed a double-tagged strain (Fft3-TAP / Sfc4-Myc) and performed a co-immunoprecipitation assay.We observed an enrichment of Sfc4-Myc in the anti-TAP purified material from the double-tagged strain compared to the single-tagged strain (Fig. 6E), confirming that Fft3 and Sfc4 physically interact in vivo.Taken together, our data indicate that the Fft3 chromatin remodeler directly interacts with the TFIIIC transcription complex at Pol III transcribed loci.Although tRNA genes are dispersed throughout the fission yeast genome, most of them cluster into a few foci in close proximity to centromeres, facilitated by condensin [33,34].We found condensin binding strongly to tRNA genes both in wild type cells and fft3Δ cells (S9 Fig) , suggesting that clustering is unaffected the absence of Fft3.Since the tRNA genes cluster close to the centromeres and therefore the nuclear periphery [34], we asked whether their nuclear positioning changes in fft3Δ cells.While tRNA genes are enriched for Nup85 and Man1 interactions in wild type (Fig. 6F-G), we observed a reduction of Man1 association in fft3Δ cells (Fig. 6H).This suggests that tRNAs move away from the nuclear envelope in the absence of Fft3.Interestingly, this effect seems to be specific to tRNA genes, since the 5S rRNA genes stay enriched for Man1 association when Fft3 is deleted (S8B-D Fig).
We then wondered if there are further effects at tRNA genes in fft3Δ cells.Therefore, we examined expression levels of two tRNA classes, proline and alanine, by northern blot and RT-qPCR (S10A-C Fig) .We observed an increase in transcription for proline, but not for alanine, suggesting there is no clear-cut effect of Fft3 on tRNA transcription.A clearer trend emerged when we looked at MNase sequencing data: like LTRs, tRNA genes show a pronounced decrease in nucleosome occupancy (Fig. 6I).Similarly, 5S rRNA genes also showed a decrease in nucleosome occupancy (S6E Fig) .Together, these data indicate that Fft3 is involved in maintaining chromatin structure and intra-nuclear positioning of Pol III transcribed genes.

Discussion
It has been established in different eukaryotic systems that the periphery of the cell nucleus harbors chromosomal domains with transcriptionally repressed chromatin.Insulators contribute to this three-dimensional nuclear organization by forming clusters to separate different chromatin domains [35].
Here we show that cells lacking the Fft3 remodeling enzyme or carrying a catalytically inactive version display drastic changes in chromatin marks and gene expression in the subtelomeric regions (Fig. 7A).Fft3 is required to ensure the association of these domains with the nuclear envelope through the LEM domain protein Man1.The budding yeast homolog of Man1, Heh1p, is involved in regulating silencing of rDNA repeats [36].Also in budding yeast, subtelomeres are associated with the INM protein Src1p [37] and subtelomeric genes change in expression when the Fft3 homolog Fun30 is missing [38].Our data highlights that Fft3-even though itself not enriched at the subtelomeres per se-affects gene expression in these domains by acting on their borders.We show that local chromatin structure at the borders is altered in fft3Δ cells.This agrees with observations on the action of other chromatin remodelers at insulators, such as incorporation of H3.3 by PBAP at border elements in fruit flies [39] and binding of ISWI to ArsI insulators in sea urchins [40].
The borders of subtelomeres in S. pombe coincide with retrotransposon-derived LTR elements.In mouse cells, LTR elements can function as insulators [41].Furthermore, the wellcharacterized gypsy insulator in Drosophila is derived from the gypsy retrotransposon (reviewed in [42]).The Tf2 retrotransposons in fission yeast cluster within the nucleus to form Tf-bodies, and de-cluster in response to oxidative stress [43].Our results show that Fft3 binds It has been suggested that insulator elements function by changing higher-order chromatin structure (reviewed in [44]).Insulators can interact with each other and tether the chromatin fiber to structural elements within the nucleus, e.g.nuclear pores.In this way, chromatin loops can form which separate euchromatin and heterochromatin domains.In S. pombe, Tf2 transposons cluster into Tf bodies that play a role in genome organization [43].Furthermore, the Drosophila gypsy insulators form specialized insulator bodies at the nuclear periphery, and transposon-derived MAR/SAR sequences can create chromatin loops [42,45].In agreement with this, we found that S. pombe LTR elements localize close to the nuclear periphery and In addition to LTR elements, Fft3 binds to tRNA genes and physically interacts with the Pol III transcription factor TFIIIC.These features may be conserved in the Fun30 remodeling family, since the S. cerevisiae homolog, Fun30, is enriched over tRNA genes and the human homolog, SMARCAD1, purifies together with the TFIIIC complex [16,17,21].Based on our findings in S. pombe it is plausible that budding yeast and human Fun30 homologues also directly interact with TFIIIC.Like in S. pombe, tRNA genes function as insulators in other eukaryotes and play a role in genome organization by clustering at specific sites in the nucleus (reviewed in [46]).It is noteworthy that both the insulating function and clustering of tRNA genes in yeast depend on TFIIIC [46].It is therefore likely that Fft3 is recruited to tRNA genes through its interaction with TFIIIC and maintains the chromatin state at these insulators.Fft3 does not appear to be involved in condensin recruitment to tRNA genes, suggesting that condensindependent clustering of tRNA genes may not be affected.However, our results indicate that Fft3 is required for their anchoring at the nuclear envelope (Fig. 7B).
At the nucleosome level, we observed a reduction in nucleosome occupancy at Fft3 targets when the remodeler was missing.A similar decrease has also been observed at genes regulated by Fun30 in budding yeast [38].This suggests that Fft3 is required to incorporate these nucleosomes or to stabilize them after incorporation.The nucleosomes could then serve either as a platform for binding or as a barrier for other proteins.Alternatively, Fft3 remodeling could affect chromatin mobility.INO80, another Snf2 remodeling factor, facilitates chromatin movement inside the budding yeast nucleus by altering the stiffness of the chromatin fiber [47].It is conceivable that Fft3 acts in a similar manner to allow for correct positioning of its targets within the nucleus.
Beyond yeasts, subtelomeres are important in several other eukaryotic systems.In parasites such as Trypanosoma brucei and Plasmodium falciparum, the subtelomeres habor genes encoding surface markers that can be varied through recombination [48] and contribute to antigen variation.In T. brucei, these loci move from the nuclear periphery towards the nuclear interior when activated during differentiation [49].In human cells, the D4Z4 insulator is located in the subtelomere of chromosome 4q and maintains subtelomeric heterochromatin [50].Deletion of D4Z4 causes a type of muscular dystrophy, Facio-Scapulo-Humeral Dystrophy (FSHD).Interestingly, insulator function of D4Z4 depends on CTCF and lamin A, and is involved in peripheral positioning of the telomere [51,52].Taken together, these studies highlight the importance of links between subtelomeric chromatin states and nuclear positioning.
In this study, we provide an example how a chromatin remodeler affects both local chromatin structure and genome-wide nuclear organization.Further studies will be necessary to shed light on this interplay between nuclear architecture and transcriptional regulation that seems to have a major role genome function in eukaryotes.

Strain construction and cell culture
S. pombe cells were grown at 30°C and in YES medium unless stated otherwise.The S. pombe strains used in this study are listed in S1 Table .The point mutation in the ATPase domain of Fft3 was created using PCR with a primer containing the mutation.Genomic DNA was isolated from the Fft3-myc::hph strain (Hu1911) and part of the gene, the myc-tag, the hygromycin resistance gene and part of the 3'UTR were amplified by PCR using primers ATPase-F and ATPase-R (see S2 Table ).The ATPase-F primer contained a mismatch introducing the K418R substitution (AAA to AGA).To improve the recombination efficiency, the fragment was elongated by a second PCR using ATPaseL-F and ATPase-R primers.The PCR product was then purified and electroporated into a wild type strain (Hu0029).Strains in which the construct had replaced the wild type fft3+ gene through homologous recombination were selected by hygromycin resistance and confirmed by DNA sequencing.
For microarray hybridization, immunoprecipitated DNA was amplified to 5 μg DNA as described in [53], with the exception that in the second PCR, 5 mM dUTP was added to the reaction.Fragmentation, labeling and hybridization to the Affymetrix GeneChip S. pombe Tiling 1.0FR was performed by the Affymetrix core facility at Novum (http://apt.bea.ki.se) according to Affymetrix standard protocols.
All experiments were done as biological duplicates.
For real-time quantitative PCR, immunoprecipitated DNA was amplified in the presence of SYBR Green using Applied Biosystems 7500 real-time PCR machine.The primers used are listed in S2 Table.

DamID
Genomic DNA was extracted as described previously [7], with an additional clean-up of genomic DNA using the DNeasy Blood and Tissue Kit (Qiagen).DamID experiments were done as described by [54].The amplified material was cleaned using the Qiagen PCR purification kit and fragmented with DNAse.Fragmented DNA was labeled and hybridized to the Affymetrix GeneChip S. pombe Tiling 1.0FR using standard protocols by the Affymetrix core facility at Novum (http://apt.bea.ki.se).
All experiments were done as biological duplicates.

RNA extraction
RNA was extracted using the hot phenol method as described in [55] and reverse transcribed using random hexamers and the SuperScript II Reverse Transcriptase kit (Invitrogen).Realtime quantitative PCR was performed in the presence of SYBR Green using an Applied Biosystems 7500 real-time PCR machine.Primers used are listed in S2 Table .All experiments were done as biological duplicates.

Fluorescence in-situ hybridization
Cells were grown to early log phase at 30°C in YES medium, fixed with adding paraformaldehyde to 1,75% for 40 min and processed for IF with mouse monoclonal anti-myc antibodies (9E10, Sigma, 4μg/ml).For FISH, cells were refixed with 3% paraformaldehyde for 30min at room temperature.Chromosomal DNA was denatured by successive 15 min incubation time in 2SSC, 2SSC 10% formamide, 2SSC 20% formamide and 2SCC 40% formamide at room temperature.Subtelomeric FISH probes were generated using the sequencing cosmid c186 (provided by the Wellcome Trust Sanger Institute, Hinxton, UK) as template.c186 spans 30176bp on chr1R (Range 5524764 to 5554939), from SPNCRNA.283 to SPAC186.09.Images were taken using a Nikon A1+ laser scanning microscope with a 60x Lambda S oilimmersion objective (NA 1.4).Imaging was set up fulfilling the Nyquist criterion in xy and z, with a minimum zoom of 2. For each cell, z-stacks were acquired with 0.2μm spacing and subjected to a blind 3D deconvolution algorithm using the NIS Elements Advanced Research software version 4.12.After decapping (removing the top three and bottom three z-stacks) of each nucleus, distances from the center of each FISH spot to the nuclear periphery were measured.Measurements were then assigned to three concentric equal volume zones [56].Significance tests were carried out using the Chi-square test.

Live cell microscopy
Cells were grown over night at 30°C in PMG liquid medium.For mounting, 35 mm glass bottomed culture dishes (MatTek Corp) were coated with soy lectin (Sigma).A log phase cell suspension was added on top of the coated surface and cells were left to sediment for five minutes before pre-warmed PMG medium was added on top.During imaging, the culture dish was kept at 30°C inside the microscope incubator and cells were imaged for no longer than three hours.
Images were taken using an inverted Nikon A1R laser scanning microscope with a 60x Lambda S oil-immersion objective (NA 1.4).Imaging was set up fulfilling the Nyquist criterion in xy and z, with a minimum zoom of 2. For each nucleus, z-stacks were acquired with 0.125μm spacing in a 5 μm range around the center of the nucleus.Deconvolution of each stack was carried out using a 3D non-blind algorithm in NIS Elements (Nikon, High Content Analysis software version 4.20).After decapping (removing the top three and bottom three zstacks of the nuclear volume) of each nucleus, distances from the center of each mCherry spot to the nuclear periphery were measured in the z-plane with the brightest signal.The radius of each nucleus was measured and used to normalize the distance measurement.Normalized distances were then assigned to one of three concentric equal volume zone [56].Significance tests were carried out using a two-sided Chi-square test.

Micrococcal nuclease digestion and sequencing
Mononucleosomal DNA fragments were digested and purified as described in [57].Fragments were then amplified and labeled with the NEBNext ChIP-Seq Library Prep Master Mix Set for Illumina (NEB #E6240) followed by sequencing on an Illumina Miseq v.3.Paired-end reads were mapped to the S. pombe genome using Bowtie2 [58] with standard parameters.DANPOS [59] was then used to remove clonal reads, normalize by quantile normalization and calculate nucleosome occupancy.All calculations and visualization were carried out using R.All experiments were done as biological duplicates.MNAse-seq data can be accessed at NCBI GEO under the accession number GSE58013.

Protein extraction and western blot
Proteins were extracted using a FastPrep-24 machine (MP Biomedicals) and separated by SDS PAGE.Immunoblot analysis was carried out using anti-myc (9E10, Sigma) and anti-actin (ab8224, abcam) antibodies.

Yeast-2-hybrid assay
The yeast-2-hybrid screen and data analysis were performed by Hybrigenics, Paris, France (www.hybrigenics.com).The full-length open-reading frame of the fft3 gene (SPAC25A8.01c)was cloned into the pB27 vector as a C-terminal fusion to LexA.The construct was use as bait to screen a S. pombe cDNA library.62.4 million interactions were tested.

Fig 1 .
Fig 1. Chromatin changes in fft3Δ cells are pronounced at the subtelomeres of chromosomes I and II.(A) Subtelomeric regions show strong changes in Pol II and H2A.Z occupancy, as well as H4K12Ac levels.H2A.Z and H4K12Ac ChIP-chip data from [14].Subtelomeric regions are marked in gray.(B-D) Pol II, H2A.Z and H4K12Ac are significantly enriched over subtelomeric regions of chromosomes I and II in fft3Δ cells compared to wild type cells.doi:10.1371/journal.pgen.1005101.g001 (Fig. 4D, S4A Fig).Binding for both is slightly reduced when Fft3 is deleted (Fig. 4E, S4B Fig), but still higher than at other loci in the genome

Fig 2 .
Fig 2. Chromatin changes at subtelomeres in fft3Δ cells coincide with changes in nuclear organization.(A) Association with the nuclear envelope in wild type is lost in fft3Δ cells.Shown over the left subtelomere of chromosome II are log2-ratios of Pol II enrichment in fft3Δ vs. wt, log2-enrichment of Fft3-myc ChIP over anti-myc control ChIP, log2-enrichment Man1-Dam over the Dam only control in wild type, log2-enrichment Man1-Dam over Dam only control in fft3Δ cells.(B) Subtelomeric chromatin is associated with the nuclear periphery in wild type cells.Man1-Dam enrichment over Dam-only control is shown as boxplot.(C) Association of the subtelomeres with the nuclear periphery is lost in fft3Δ cells.Man1-Dam enrichment over Dam-only control is shown as boxplot.(D) Fft3 is not enriched at subtelomeres.Fft3-enrichment over anti-myc control is shown as boxplot.doi:10.1371/journal.pgen.1005101.g002

Fig 3 .
Fig 3. Fft3 and Bqt4 anchor subtelomeres at the nuclear envelope.(A) Live cell microscopy examples for wild type and bqt4Δ fft3Δ cells.The nuclear envelope is stained by Ccr1N-GFP (green) and the telomeres by Taz1-mCherry (red).Cells were scored as belonging to either of three equal volume zones depending on the distance between the nuclear envelope and the telomere signal.(B-E) Telomeres are released from the nuclear envelope in bqt4Δ cells, but move even further into the nuclear interior in bqt4Δ fft3Δ cells, where subtelomeres lose association with the nuclear envelope.Percentages of cells in each of the zones in different strains are shown, with number of cells measured in the legend.Significance testing was done using a two-sided Chisquare test.doi:10.1371/journal.pgen.1005101.g003 , S6A-C Fig).However, fft3-K418R cells mimic the phenotype observed in fft3Δcells: they grow slightly slower than wild type cells at 30°C and show severe growth defects at 25°C and 37°C (S6D Fig).Importantly, subtelomeric genes are upregulated in cells carrying the point mutation in Fft3 (Fig. 5E), as observed in cells lacking Fft3.Furthermore, we observed similar increases in PolII, H3K9Ac and H2A.Z occupancy in fft3-K418R and fft3Δcells (S7A-C Fig).As in fft3Δ cells, Man1-interaction is reduced in fft3-K418R cells compared to wild-type (S7D Fig).

Fig 4 .Fig 5 .
Fig 4. Fft3 binds LTR elements at subtelomeric borders and genome-wide, affecting nucleosome occupancy and peripheral positioning.(A) Fft3 binding at subtelomeric borders occurs over or in close proximity to LTR elements.Graphs show Pol II enrichment in fft3Δ compared to wild type (green) and Fft3 enrichment compared to anti-myc control (blue) over chromosomes I and II.Subtelomeric regions are marked in gray and small panels show Fft3 binding at subtelomeric borders, with LTR elements marked in red.(B) Fft3 interacts with LTRs genome-wide.Fft3-enrichment over anti-myc control is shown as boxplot.(C) Nucleosome occupancy at LTRs is reduced in fft3Δ cells.The average number of reads mapping to each LTR is shown for wild type and fft3Δ.Pvalue was obtained using paired, two-sided Mann-Whitney U test.(D) LTRs interact with Man1 at the nuclear periphery.Man1-Dam enrichment over Damonly control is shown as boxplot.(E) Interaction with Man1 is reduced in fft3Δ cells.Man1-Dam enrichment over Dam-only control is shown as boxplot.doi:10.1371/journal.pgen.1005101.g004

Fig 6 .
Fig 6.Fft3 interacts with TFIIIC at Pol III transcribed loci and affects peripheral localization of tRNA genes.(A) Fft3 is significantly enriched at tRNA genes.Fft3-enrichment over anti-myc control is shown as boxplot.Fft3-myc ChIP-chip data from [14].(B) Fft3-myc binds to loci also bound by the TFIIIC-

subunitFig 7 .
Fig 7. Proposed mechanism of how Fft3 regulates subtelomeric chromatin and tRNA positioning.(A) Fft3 maintains subtelomeric chromatin through binding and remodeling activity at boundary elements.In the absence of Fft3, active chromatin marks and transcription in the subtelomeric domain increase and the domain is released from the nuclear periphery.When Bqt4 is deleted in addition to Fft3, the entire telomere can move away from the nuclear envelope towards the interior.(B) tRNA clusters at the nuclear envelope dissociate from the periphery in fft3Δ cells and lose their transcriptional regulation.doi:10.1371/journal.pgen.1005101.g007 standard deviation of duplicate experiments.(PDF) S2 Fig. Loss of peripheral association was confirmed by qPCR.Data shown as fold difference in enrichment over Dam-only control.Error bars represent the standard deviation of duplicate experiments.(PDF) S3 Fig. (A) FISH-Immunofluorescence examples for wt and fft3Δ cells.The nuclear envelope was stained with Nup61-myc-Cy3 (red) and the subtelomeric FISH probe was labeled with FITC-anti-DIG (green).Cells were scored as belonging to either of three equal volume zones depending on the distance between the nuclear envelope and the FISH signal.(B) Subtelomeres move towards the interior in fft3Δ bqt4Δ cells.Percentages of cells in each of the zones in different strains are shown, with number of cells measured in the legend.Significance testing was done using a two-sided Chi-square test.(PDF) S4 Fig. Fft3 is enriched at LTR elements.(A) Nup85 is significantly enriched at LTR elements.Nup85-Dam enrichment over Dam-only control is shown as boxplot.(B) Nup85 levels decrease at LTRs in fft3Δ cells.Data shown as fold difference in enrichment at LTR elements over Dam-only control.Error bars represent the standard deviation of duplicate experiments.(C) The subtelomeric border element (SBE) LTRs on the left arm of chromosome 2 loose their association with the nuclear envelope in fft3Δ cells.Data shown as fold difference in enrichment at SBE LTRs over Dam-only control.Error bars represent the standard deviation of duplicate experiments.(D) Insertion of an LTR into the ura4 gene is not sufficient to increase Fft3 interaction.Data from anti-myc ChIP qPCR of a strain with wild-type ura4 and a strain with an LTR inserted into the ura4 coding sequence (see schematic).Data is shown as enrichment over control locus (LTR close to SPBPB10D8.04c)with error bars representing the standard deviation of duplicate experiments.(PDF) S5 Fig. Nucleosome occupancy changes over subtelomeric borders in fft3Δ cells.Nucleosome occupancy at subtelomeric borders is affected by Fft3.Map shows nucleosome occupancy profiles (top panel) and Fft3 enrichment (blue) over the border of (A) the right subtelomere on chromosome I, (B) the left subtelomere on chromosome II and (C) the right subtelomere on chromosome II.Genes are shown in light grey, LTRs in dark grey.Asterisks mark positions were nucleosome occupancy is reduced.The left subtelomere border on chromosome II contains four identical copies of the gene SPBPB10D8.04cand an LTR, so the profiles in (B) are shown as an average over all four copies.(PDF) S6 Fig. Fft3-K418R-myc is recruited to the same locations as Fft3-myc, but mimics the fft3Δ phenotype.(A) The Fft3 ATPase mutant is recruited to the central core domain of centromere 1 (cnt1).Data from ChIP-qPCR of non-tagged wildtype, Fft3-myc and Fft3-K418Rmyc is shown as fold difference in enrichment at Cnt1 over dg, normalized to input.Error bars represent the standard deviation of duplicate experiments.(B) The Fft3 ATPase mutant is recruited to valine tRNA genes.Data/error bars as in (A).(C) The Fft3 ATPase mutant is recruited to proline tRNA genes.Data/error bars as in (A).(D) Fft3-K418R cells show the same temperature sensitivity as fft3Δ cells.Cell suspensions were diluted into five different concentrations, spotted on YES plates and grown at the indicated temperatures for 3 days.(PDF)