|Home | About | Journals | Submit | Contact Us | Français|
Chromatin function requires specific three-dimensional architectures of chromosomes. We investigated whether Saccharomyces cerevisiae extra TFIIIC (ETC) sites, which bind the TFIIIC transcription factor but do not recruit RNA polymerase III, show specific intranuclear positioning. We show that six of the eight known S. cerevisiae ETC sites localize predominantly at the nuclear periphery, and that ETC sites retain their tethering function when moved to a new chromosomal location. Several lines of evidence indicate that TFIIIC is central to the ETC peripheral localization mechanism. Mutating or deleting the TFIIIC-binding consensus ablated ETC -site peripheral positioning, and inducing degradation of the TFIIIC subunit Tfc3 led to rapid release of an ETC site from the nuclear periphery. We find, moreover, that anchoring one TFIIIC subunit at an ectopic chromosomal site causes recruitment of others and drives peripheral tethering. Localization of ETC sites at the nuclear periphery also requires Mps3, a Sad1-UNC-84–domain protein that spans the inner nuclear membrane. Surprisingly, we find that the chromatin barrier and insulator functions of an ETC site do not depend on correct peripheral localization. In summary, TFIIIC and Mps3 together direct the intranuclear positioning of a new class of S. cerevisiae genomic loci positioned at the nuclear periphery.
The three-dimensional organization of the genetic material in nuclear space is integrally related to chromatin function (reviewed by Sexton et al., 2007 ). In some higher eukaryotic cells, for example, chromosomes occupy specific nuclear “territories” that reflect their gene density and heterochromatin content (Croft et al., 1999 ; Tanabe et al., 2002 ). Typically, chromosome regions containing a high proportion of nontranscribed sequence display localization to the nuclear periphery (reviewed in Towbin et al., 2009 ). Silenced chromatin in yeast cells is preferentially positioned to the nuclear periphery (Maillet et al., 2001 ), and artificial tethering to the nuclear rim partially restores transcriptional repression by a compromised silencer (Andrulis et al., 1998 ). Gene expression in metazoans can also be modified by manipulating its intranuclear positioning (Finlan et al., 2008 ). The spatial arrangement of metazoan chromosomes appears to be tissue specific (Parada et al., 2004 ) and becomes reorganized during differentiation (Kim et al., 2004 ).
Studies of chromosome spatial organization have revealed specific intranuclear positioning of particular chromosome domains. Localization of telomeres at the nuclear periphery has been described in the budding yeast Saccharomyces cerevisiae (Gotta et al., 1996 ), in fission yeast Schizosaccharomyces pombe (Funabiki et al., 1993 ), in human cells (de Lange, 1992 ; Croft et al., 1999 ), and in other organisms (Chung et al., 1990 ; Dawe et al., 1994 ). The 64 telomeres of diploid budding yeast cells cluster at the nuclear periphery in three to eight discrete foci (Klein et al., 1992 ; Gotta et al., 1996 ), with the subtelomeric sequences being subject to transcription silencing (Gottschling et al., 1990 ). Other genomic regions also exhibit specific spatial organization in the nucleus that is related to biological function. For example, the ribosomal DNA is localized to the nucleolus (Hartung et al., 1979 ; Dujon, 1998 ; Kalmarova et al., 2007 ), whereas during interphase yeast centromeres cluster near the spindle pole body, opposite the nucleolus (Guacci et al., 1997 ; Jin et al., 1998 ). In addition, it has been reported that active S. cerevisiae tRNA genes tend to be localized to the nucleolus (Bertrand et al., 1998 ; Thompson et al., 2003 ), which is important for tRNA gene-mediated silencing (Kendall et al., 2000 ). Additional examples of directed chromosome positioning include the relocalization to the nuclear periphery of specific genes upon transcriptional activation (e.g., GAL genes, INO1 locus; Brickner and Walter, 2004 ; Casolari et al., 2005 ; Schmid et al., 2006 ). The fact that peripheral localization has been implicated in transcription activation as well as repression led to models proposing that the nuclear envelope comprises a mosaic of zones favoring either transcriptional induction or silencing.
The S. pombe RNA polymerase III transcriptional apparatus is implicated in chromosome spatial organization. Eukaryotic RNA polymerase III (Pol III) is responsible for the transcription of small structural RNAs, including tRNAs, 5S rRNA, and other small nuclear and cytoplasmic RNAs (reviewed by Dieci et al., 2007 ). The Pol III transcription machinery is highly conserved and consists of the multisubunit Pol III polymerase and two transcription factor complexes (TFIIIB and TFIIIC; Kassavetis et al., 1990 ). An additional factor (TFIIIA) is required only for 5S rDNA transcription. Pol III–transcribed genes share specific sequence and structural properties, including conserved A and B box sequences typically found within the coding region (Galli et al., 1981 ). These internal control regions are recognized by the six-subunit complex TFIIIC (Baker et al., 1986 ; Bartholomew et al., 1990 ). The B box sequence is conserved in all eukaryotes (GGTTCGANTCC), and mutation of the internal cytosine inactivates both TFIIIC binding and Pol III transcription of a tRNA gene (Newman et al., 1983 ; Baker et al., 1986 ; Marzouki et al., 1986 ). Once assembled, TFIIIC recruits TFIIIB to an ~50–base pair, AT-rich region upstream of the transcription start site. After recruitment by TFIIIC, TFIIIB in turn recruits Pol III for transcription initiation (Kassavetis et al., 1990 , 1997 ).
Chromatin boundary elements function to separate chromatin domains, either by insulating promoters from transcriptional activation or by acting as barriers to the propagation of repressive heterochromatin (West et al., 2002 ). A study in the fission yeast S. pombe revealed a role for the RNA polymerase III apparatus, and TFIIIC in particular, in boundary function and genome organization. Chromatin boundary elements called “inverted repeats” (IRs) contain multiple B box sequences but are not transcribed. IR elements were shown to bind TFIIIC but not other Pol III factors or Pol III itself, suggesting that TFIIIC binding may mediate chromatin boundary function (Noma et al., 2006 ). These TFIIIC-bound IR insulators were found to be predominantly associated with the nuclear periphery. It was suggested that such loci act as so-called chromosome-organizing-clamp (COC) sites that tether chromosomal regions to the nuclear periphery, possibly mediating three-dimensional organization of the fission yeast genome (Noma et al., 2006 ). However, the mechanism of peripheral localization is unclear.
In a genome-wide survey of Pol III apparatus occupancy in S. cerevisiae, eight intergenic loci were identified that display TFIIIC occupancy but no significant recruitment of other Pol III factors (Moqtaderi and Struhl, 2004 ). These loci were called extra TFIIIC (ETC) sites. Of interest, these loci tend to lie in divergent intergenes. All eight loci share a sequence that resembles a B box consensus, but with an additionally conserved 10-base extension to the 3′ side (Moqtaderi and Struhl, 2004 ). The ETC extended B box consensus sequences are highly conserved among sensu stricto yeast species, suggesting an important biological function. Two additional S. cerevisiae genome-wide studies of Pol III–binding sites (Harismendy et al., 2003 ; Roberts et al., 2003 ) identified several of the same ETC loci, as well as other sites that recruit partial Pol III complexes. Recently ETC loci were shown to be able to function as chromatin insulators, blocking gene activation if artificially inserted between an upstream activation sequence (UAS) and its transcriptional start site (Simms et al., 2008 ; Valenzuela et al., 2009 ). In addition, ETC6 was shown to have an insulator-like function in its natural context. Two ETC sites (ETC2 and ETC4) can also function in reporter constructs as barriers to the spread of heterochromatin, suggesting a role for TFIIIC in regulating Pol II–transcribed genes (Simms et al., 2008 ; Valenzuela et al., 2009 ; Kleinschmidt et al., 2011 ). Although several studies suggested that TFIIIC binding in the absence of TFIIIB is sufficient for such insulator and barrier activities, a recent investigation found TFIIIC-mediated insulation was increased in the presence of bound TFIIIB and was compromised in various histone modifier and remodeler mutants (Valenzuela et al., 2009 ).
In a recent interesting development, Moqtaderi et al. (2010 ) identified >1865 ETC loci in the human genome that recruit TFIIIC but not other Pol III apparatus components. Human ETC (hETC) loci are preferentially located between closely spaced, divergently transcribed Pol II genes, reminiscent of S. cerevisiae ETCs. hETC loci are characterized by one of two sequence motifs: either a B box sequence or a novel motif loosely related to the binding motif for the ET transcription factor family (Moqtaderi et al., 2010 ). Thousands of ETC sites have also been identified in the mouse genome (Carriere et al., 2012 ).
We investigated whether S. cerevisiae ETC sites mediate positioning to the nuclear periphery. By fluorescently tagging all eight known ETC sites, we showed that the majority of ETC loci localize predominantly to the nuclear periphery. The ETC B box consensus is necessary for peripheral positioning, and an ETC locus is sufficient to cause peripheral localization if transferred to another chromosomal region. We find that the TFIIIC complex itself directs peripheral tethering through a pathway that involves the inner nuclear membrane protein Mps3. Surprisingly, however, it appears that peripheral localization of an ETC site is not required for its insulator or barrier activity.
We investigated whether ETC loci in S. cerevisiae are tethered to the nuclear periphery. ETC sites can be microscopically visualized in live cells using the chromosome dot system. An ETC site is tagged by inserting an array of lac operator repeats (Figure 1A) in cells expressing the Lac repressor protein fused to green fluorescent protein (GFP; LacI-GFP; Robinett et al., 1996 ). Recruitment of LacI-GFP to the tagged ETC locus allows its visualization as a bright dot (Figure 1B). To facilitate measurement of the position and movement of the ETC dot, the nuclear envelope is also marked using a GFP-fused allele of the nuclear pore component NUP49 (Belgareh and Doye, 1997 ). Quantification of the chromosomal ETC dot position is performed using the “three-zoning” method (Taddei and Gasser, 2004 ), in which the dot is scored to one of three concentric zones of equal surface area (Figure 1C). A randomly positioned locus shows equal distribution among the three zones (~33% in zones 1–3), whereas a locus positioned at the nuclear periphery is preferentially observed in zone 1 (Figure 1, C and D). Cell cycle position is assessed according to bud size (see Materials and Methods).
ETC2 lies on the left arm of chromosome XV (genome coordinate: XV, 58539–58758), more than 59 kb from telomere XV-L (Figure 1A; Moqtaderi and Struhl, 2004 ). The intergene neighboring ETC2 was GFP tagged as described. We found that in the majority of interphase cells, the ETC2 locus resides in zone 1 (localization to zone 1 in 71.5% of G1 cells, 78.9% of S cells, and 75.7% of G2 cells; Figure 1D). In most cases in which ETC2 was observed within zone 1, the fluorescent signal from the chromosome dot and the nuclear envelope appeared juxtaposed. Of importance, ETC2 remains predominantly localized to the nuclear periphery in G2 phase. The cell cycle regulation of ETC2 positioning therefore differs from that of telomeres, which become randomly localized in G2 phase. A χ2 analysis confirmed that ETC2 positioning during all three cell cycle phases differs significantly from random (Figure 1D).
We constructed strains in which the other seven ETC loci were individually fluorescently tagged and examined their subnuclear localization. Figure 1E shows the percentage of GFP dots observed at the nuclear periphery (i.e., in zone 1) for each ETC-tagged strain, shown as a cumulative total throughout interphase (cell cycle–staged results in Supplemental Figure S1). ETC4, ETC5, ETC6, ETC7, and ETC8 reside in zone 1 in the majority of interphase cells (Figure 1E). All of these loci retained peripheral localization throughout interphase (Supplemental Figure S1), similar to the pattern observed for ETC2. A control locus (ChrVIint) displayed random positioning.
ETC1, in contrast, exhibited virtually random positioning throughout the cell cycle (33.5%; Figure 1E). ETC3 was also positioned largely randomly, displaying only a slight tendency toward peripheral localization (47.5%; Figure 1E).
To summarize, we found that six of the eight ETC loci (ETC2, ETC4, ETC5, ETC6, ETC7,and ETC8) exhibited clear peripheral subnuclear localization. The S. cerevisiae genome therefore contains at least six peripherally positioned ETC chromosome loci, which we propose are equivalent to S. pombe COC sites.
ETC loci share certain sequence and structural properties with tRNA genes—in particular, a B box consensus and TFIIIC binding. Because some tRNA genes are proposed to localize to the nucleolus (Bertrand et al., 1998 ; Thompson et al., 2003 ), we tested whether ETC loci also associate with the nucleolus. The S. cerevisiae nucleolus occupies a crescent adjacent to the nuclear envelope, so ETC positioning to the nuclear periphery does not preclude nucleolar localization. ETC7 positioning was examined in a strain bearing an mCherry-tagged version of the nucleolar marker protein Nop1 (Schimmang et al., 1989 ). Colocalization of ETC7 with the nucleolus was scored if the chromosome dot and Nop1 fluorescent signals coincided or were juxtaposed when observed at the equatorial region in a Z-stack of images (Figure 2A). We found no tendency for ETC7 to be located close to or within the nucleolus. The nucleolus occupies on average 30% of the nuclear volume, but ETC7 coincided with the nucleolus in only 32.4% of interphase cells (Figure 2Aii). A χ2 analysis confirmed this value as consistent with only random coincidence of ETC7 with the nucleolus (p = 0.086). Similar results were obtained from analysis of ETC5, which showed 30.1% colocalization with the nucleolus (unpublished data).
S. cerevisiae telomeres form clusters at the nuclear periphery (Klein et al., 1992 ; Gotta et al., 1996 ). To test whether ETC sites colocalize with telomeres, we used a mCherry-Rap1 fusion protein to visualize the telomere foci, which appear as bright fluorescent foci at the nuclear periphery (Hayashi et al., 1998 ; Hiraga et al., 2006 ). In G1- and S-phase cells, these foci were predominantly localized at the nuclear envelope, as expected. Colocalization of an ETC7 dot with a telomere focus was scored if the fluorescent signals coincided or were juxtaposed when observed in the equatorial region in a Z-stack of images (Figure 2B). We found no tendency for ETC7 to colocalize with telomere clusters at greater-than-random incidence (11.1%; Figure 2Bii). Similar results were obtained with ETC2 (12.9% colocalization; unpublished data).
To summarize, S. cerevisiae ETC sites do not appear to be localized to nucleoli or telomeric foci.
We next tested whether the extended B box consensus (and by extension TFIIIC binding) is required for ETC-site perinuclear localization. Starting with the chromosome dot–tagged ETC6 strain, we deleted the 23–base pair ETC consensus along with 10 base pairs of intergenic sequence on either side, resulting in a total deletion of 43 base pairs (illustrated in Figure 3A). No other B box–like sequence is present in the intergenes where ETC6 lies, and this etc6Δ mutant no longer binds TFIIIC (Figure 3B).
Deleting the ETC6 consensus caused the locus to become randomly positioned in all three cell cycle phases (Figure 3, C and D). The B box extended ETC consensus therefore appears essential for tethering ETC6 to the nuclear periphery. Analogous data were obtained on deleting the ETC7 consensus (Supplemental Figure S2, A and B). To summarize, the conserved B box extended consensus sequence is critical for perinuclear localization of ETC loci.
We examined whether an ETC locus inserted at a randomly positioned chromosomal region can direct its localization to the nuclear periphery. We constructed a strain in which an internal chromosomal intergene (YNL179C-RPS3; ChrXIV-302) was fluorescently tagged and confirmed that this ChrXIV-302 locus is randomly distributed in the nucleus throughout interphase (Figure 4, A and B). We next inserted at ChrXIV-302 a 91–base pair fragment of RAD2-TNA1 intergenic sequence from ChrVII, encompassing ETC4. Subnuclear localization revealed that the resulting “ectopic” ETC site was positioned in zone 1 in the majority of interphase cells (Figure 4C). In contrast, insertion of an ETC4 fragment containing a mutated consensus sequence incapable of binding TFIIIC (etc4mut; Simms et al., 2008 ) was unable to direct peripheral localization (Figure 4D). An ETC site can therefore direct peripheral tethering even if moved to a new chromosomal context, with positioning dependent on an intact TFIIIC-binding consensus. Larger genomic fragments containing ETC2 or ETC6 were also able to direct peripheral positioning when inserted at the ChrXIV-302 site (Supplemental Figure S2, C and D).
The eight ETC sites were discovered on the basis of their TFIIIC occupancy (Moqtaderi and Struhl, 2004 ). To test directly whether TFIIIC mediates ETC site peripheral tethering, we fused an auxin-inducible degron (Nishimura et al., 2009 ) to Tfc3 in the strain containing the fluorescently tagged ETC4 locus and tested the effects of inducing Tfc3 degradation on ETC4 localization. Microscopic examination of cells 1–2 h after addition of the auxin 3-indoleacetic acid (IAA) revealed that ETC4 peripheral localization was ablated (Figure 5A). In contrast, ETC4 localization remained largely intact in a control strain with untagged Tfc3 (Figure 5B). The rapid loss of ETC4 peripheral positioning on induction of Tfc3 degradation suggests that TFIIIC is directly responsible for tethering ETC sites at the nuclear periphery.
The foregoing results implicate TFIIIC in the ETC tethering mechanism. We therefore tested whether TFIIIC alone can drive tethering of a chromosomal domain to the nuclear periphery. We used a system developed as a cytological assay for proteins that cause peripheral tethering (Taddei et al., 2004 ; Ebrahimi et al., 2010 ). Briefly, LexA-binding sites (lexAOp) are inserted at a randomly positioned chromosome locus (ChrVIint, adjacent to ARS607 on chromosome VI). Candidate anchoring proteins are expressed fused to the LexA DNA-binding domain and their effect on ChrVIint subnuclear position assessed. An array of lacOp repeats at the same site allows subnuclear positioning of ChrVIint to be monitored microscopically (Figure 6A; Taddei et al., 2004 ).
We tested the ability of LexA-fused TFIIIC components to cause peripheral localization of ChrVIint. Expression of LexA alone does not affect ChrVIint localization (Figure 6D), but expression of either LexA-Tfc1 or LexA-Tfc6 induces anchoring of ChrVIint to the nuclear periphery (Figure 6, B and C). In both cases, peripheral anchoring levels were highest in G1 and dropped slightly in S and G2 phases. Similar results were obtained on expression of LexA fused to other TFIIIC subunits (LexA-Tfc3, LexA-Tfc4, LexA-Tfc7, and LexA-Tfc8; Supplemental Figure S3).
The fact that all the TFIIIC subunits were able to induce some level of peripheral tethering suggested that binding of one Lex-Tfc protein to DNA might cause recruitment of other TFIIIC subunits. We tested this possibility, and found, using chromatin immunoprecipitation (ChIP) analysis, that binding of LexA-Tfc3 or LexA-Tfc6 causes corecruitment of Tfc1 (Figure 6E). Together with the positioning data, this result suggests that tethering any TFIIIC subunit can cause nucleation of the other complex subunits to direct peripheral localization.
We aimed to identify the nuclear envelope component responsible for anchoring ETC sites at the nuclear periphery. One candidate was Mps3, a Sad1-UNC-84 (SUN)–domain inner nuclear envelope protein. Mps3 functions as an integral membrane anchor for telomeres (Bupp et al., 2007 ) and is also involved in sequestering double-strand break sites at the nuclear periphery (Oza et al., 2009 ). Mps3 is an essential protein, so we examined the impact of a mutant version that lacks the N-terminal nucleoplasmic domain required for localizing telomeres (the previously described mps3Δ75–150 allele; Bupp et al., 2007 ).
Deleting this Mps3 N-terminal domain resulted in random positioning of the ETC6 locus in all three cell cycle phases (Figure 7A), demonstrating that Mps3 is important for anchoring ETC6 to the nuclear periphery. Similar data were obtained on subnuclear localization analysis of ETC2 in the mps3Δ75–150 mutant (Figure 7B). This loss of peripheral anchoring suggests that the SUN-domain protein Mps3, and specifically its N-terminal nucleoplasmic domain, plays an important role in the perinuclear tethering of ETC loci.
To exclude the possibility that loss of tethering is an indirect consequence of the mps3Δ75–150 mutation, we examined the effect of ectopically overexpressing a dominant-negative version of MPS3 containing only its nucleoplasmic N-terminal domain fused to tetR-mCherry to permit visualization. A similar fusion construct was previously shown to interfere with telomere anchoring at the nuclear periphery (Schober et al., 2009 ). Microscopic observation revealed that this Mps3-N-tetR-mCherry (Mps3-N′) protein localizes throughout the nucleoplasm (Supplemental Figure S4A), in contrast to full-length Mps3 (Bupp et al., 2007 ) and as expected, since this Mps3-N′ construct lacks the Mps3 membrane-spanning domain. We found that the overexpression of Mps3-N′ (from a multicopy vector in a wild-type MPS3 background) ablates peripheral positioning of the ETC4 locus (Figure 7C). Expression of Mps3-N′ also prevented peripheral positioning of ETC6 (Supplemental Figure S4B). These results support the conclusion that Mps3, and specifically its N-terminal domain, is involved in ETC locus anchoring to the nuclear periphery. We propose that the soluble Mps3 N′ domain competes with full-length, membrane-attached Mps3, preventing proper recruitment of the ETC4 site to the nuclear periphery and resulting in its random localization within the nuclear space. The finding that overexpressing the Mps3-N′ domain interferes with ETC-site peripheral positioning supports the idea that ETC nuclear membrane anchoring involves an interaction with the N-terminal domain of Mps3.
Because anchored TFIIIC subunits are central to ETC site positioning, we tested whether Mps3-N′ also antagonizes ectopic peripheral tethering driven by TFIIIC components. Using the LexA-based tethering system described previously (Figure 6A), we found that Mps3-N′ overexpression severely compromised the ability of LexA-fused TFIIIC subunits to drive localization to the nuclear rim. Specifically, neither LexA-Tfc7 nor LexA-Tfc1 is effective in anchoring ChrVIint when Mps3-N′ is overexpressed (Figure 8 and Supplemental Figure S4, C and D). In contrast, Mps3-N′ did not affect anchoring-mediated Yif1, a nuclear transmembrane protein previously found to cause peripheral tethering (Andrulis et al., 1998 ). Our observations favor a model in which TFIIIC mediates peripheral tethering of ETC sites based on either direct or indirect interactions between TFIIIC and the Mps3 N-terminal domain.
Several ETC sites have been shown to function as “insulators” (blocking transcriptional activation by an enhancer) or as “barriers” (interrupting the spread of heterochromatin; Sun and Elgin, 1999 ; Simms et al., 2008 ; Valenzuela et al., 2009 ). To examine whether positioning at the nuclear periphery is required for these ETC functions, we tested the effect on ETC4 insulator and barrier activity of overexpressing the Mps3-N′ domain, which, as shown previously, is a dominant inhibitor of peripheral localization. We used an established assay for enhancer blocking transcriptional insulator activity (Figure 9A; Simms et al., 2008 ), in which ETC4 inserted between the GAL10 ORF and its UASG activator sequences prevents GAL10 transcription and, as a consequence, growth on galactose medium (Figure 9B, lower left quadrant). Growth on galactose was not improved by overexpressing Mps3-N′ (Figure 9B, lower right quadrant), showing that GAL10 still fails to be transcribed. This finding suggests that ETC4 retains insulator activity even when it is not localized at the nuclear periphery. Consistent with this interpretation, ETC4 also retained insulator function in a strain background in which the Mps3 N-terminal perinuclear domain was deleted (Supplemental Figure S5).
The function of ETC4 as a barrier to heterochromatin was assessed using the assay construct illustrated in Figure 9C, which tests whether silenced chromatin spreading from the silenced HMRa mating locus represses transcription of ADE2 (Jambunathan et al., 2005 ; Simms et al., 2008 ). Reduced ADE2 transcription results in pink colony pigmentation. The tRNA gene (tDNA) lying next to HMRa provides barrier activity that prevents spread of silent chromatin, allowing efficient ADE2 transcription and white colony color (Figure 9D, tDNA and tdnaΔ; left; Donze et al., 1999 ). Barrier function can be provided by a copy of ETC4 replacing the tDNA but not by a mutated version etc4mut (Figure 9D, ETC4 and etc4mut; left). We found that ETC4 (and the tDNA) can still block the spread of heterochromatin when Mps3-N′ is overexpressed, as indicated by the formation of white colonies, implying successful transcription of ADE2 (Figure 9D, ETC4 and tDNA; right). We conclude that ETC4 can continue to function as a heterochromatin barrier element even when its peripheral localization is disrupted. We also found that ETC1, which is not peripherally localized (Figure 1), is not effective as a transcription-blocking insulator or as a heterochromatin barrier element (Supplemental Figure S6, A and B).
Here we described S. cerevisiae ETC sites as a new class of sequence loci positioned at the nuclear periphery. We found that six of eight identified S. cerevisiae ETC loci exhibit peripheral localization. ETC loci therefore represent distinct chromosome sites conserved in eukaryote genomes, involved in directing correct spatial positioning within the eukaryotic nucleus (Noma et al., 2006 ). ETC sites are not colocalized with telomere foci, nor are they positioned within the nucleolus. Our experiments suggest that TFIIIC bound at ETC sites directly mediates their peripheral localization, since mutating its binding site or degrading a TFIIIC subunit abolishes positioning. We find indeed that anchoring TFIIIC subunits at an ectopic chromosomal site can drive localization to the nuclear periphery. Disrupting function of the Mps3 N-terminal domain prevents ETC-site localization, but ETC site chromatin boundary function remains intact.
Our results clearly implicate TFIIIC as central for the ETC-site positioning mechanism, a finding that raises interesting questions about the involvement of the RNA Pol III apparatus in spatial organization of the genome. Active Pol III–transcribed tRNA genes appear preferentially localized to the nucleolus (Thompson et al., 2003 ), but we found no significant colocalization of either ETC5 or ETC7 with the nucleolus. It has been suggested that another category of tRNA genes may tend to colocalize with centromeres (Duan et al., 2010 ), but we saw no tendency for ETC sites to associate with centromeres or telomere clusters. Perinuclear anchoring of ETC sites therefore appears to represent a new mode of TFIIIC-mediated positioning, acting aside from and independent of nucleolar and telomere localization. The fact that ETC-site peripheral localization is retained throughout interphase also differs from previously described peripheral positioning mechanisms. In particular, ETC sites do not appear to undergo the replication-triggered release from the nuclear periphery that leads to delocalization of telomeres during G2 (Ebrahimi and Donaldson, 2008 ).
ETC sites all contain an extended B box sequence that is conserved among sensu stricto Saccharomyces species (Moqtaderi and Struhl, 2004 ). Deletion of the B box extended consensus and immediately surrounding sequence ablated tethering of the ETC7 and ETC6 sites to the nuclear envelope, showing that the TFIIIC binding sequence is required for tethering, in agreement with studies in S. pombe (Noma et al., 2006 ). Moreover, a version of ETC4 mutated in its TFIIIC-binding consensus was unable to cause localization of an ectopic site. One possibility is that the variant, extended B box consensus present at ETC sites may allow TFIIIC to direct peripheral localization rather than TFIIIB recruitment, perhaps by altering its mode of binding.
We addressed the importance of the B box–based consensus by moving ETC loci to a new chromosomal context. All three loci tested (ETC4, ETC2, and ETC6) can direct peripheral tethering even in a new chromosomal context, with retention of peripheral positioning throughout interphase (as at endogenous ETC loci). Our results support the suggestion that TFIIIC binding alone is enough to drive peripheral localization, overriding other limitations presented by chromatin context. It is notable that ETC1 and ETC3, the two sites that display little or no peripheral localization, displayed the lowest TFIIIC binding in the study that originally identified the S. cerevisiae ETC loci (Moqtaderi and Struhl, 2004 ), and ETC3 has the weakest homology to the B box consensus. A role for TFIIIC as a major component in the positioning mechanism is further suggested by our finding that artificial recruitment of TFIIIC subunits mediates peripheral anchoring of a randomly positioned locus (Figure 6).
Examining other candidate molecular components that could mediate ETC-site peripheral tethering led us to identify the SUN-domain inner nuclear membrane protein Mps3 as a possible nuclear envelope anchor. Mps3 is a multifunctional protein previously found to act in spindle pole duplication, telomere peripheral tethering, and localization of DNA-break sites (Bupp et al., 2007 ; Schober et al., 2009 ). Deletion of the Mps3 N-terminal domain (mps3Δ75-150) severely compromised tethering to the nuclear envelope of two different ETC loci (ETC6 and ETC2). Mps3 may function as the ETC perinuclear anchor through its N-terminal acidic domain, which is located within the nucleoplasm and could interact with TFIIIC. Overexpressing a soluble N-terminal fragment of Mps3 in an MPS3 wild-type background ablated the perinuclear tethering of ETC loci and prevented LexA-Tfc–driven peripheral tethering of the ChrVIint locus (Figure 8 and Supplemental Figure S4, A and B), suggesting that Mps3-N′ competes with endogenous Mps3 for ETC-site interaction. These results support the idea that Mps3 acts as the ETC perinuclear anchor protein through its N-terminal nucleoplasmic domain. However, coimmunoprecipitation and two-hybrid tests did not reveal a direct interaction between Mps3 and any TFIIIC subunit (unpublished data), so we cannot exclude the possibility that the effect of Mps3 on ETC-site positioning is indirect and involves additional, unidentified components. We tested the effect of the variant histone Htz1, since Mps3 has been shown to interact with Htz1 (Gardner et al., 2011 ) and Htz1 is incorporated close to some ETC sites (Albert et al., 2007 ). Deleting Htz1, however, had only a marginal effect on ETC site positioning (unpublished data).
We previously found that ETC-site peripheral tethering required the activity of chromatin-remodeling proteins and in particular H3-K56 acetylation (Hiraga et al., 2008 ). Proteins like Yku70/Yku80 and Sir4, which are involved in telomere peripheral anchoring pathways, in contrast have only a marginal effect on ETC6 peripheral positioning (Hiraga et al., 2008 ). Further work will be required for a complete understanding of the ETC-anchoring pathway and identification of any additional protein components involved.
The conservation of ETC-site consensus sequences throughout sensu stricto Saccharomyces species suggests an important biological function for these loci. Six of the eight S. cerevisiae ETC loci lie between divergently transcribed genes, similar to the arrangement of most COC sites in S. pombe (Noma et al., 2006 ). ETC sites can behave as chromatin boundary elements, but copy number expression data (Ghaemmaghami et al., 2003 ) reveal no particular tendency for genes flanking ETC sites to be expressed at very different levels. There is a slight enrichment for genes within in the lowest 5% of expression levels in the vicinity of ETC sites (within the five flanking genes to the left and right). ETC sites might therefore tend to be associated with transcriptional suppression, but the significance of this observation is marginal, with the low number of identified sites limiting statistically significant conclusions. At least one S. pombe COC site behaves as a boundary element to limit the spread of silenced chromatin, and it was suggested that peripheral tethering of COC sites might facilitate boundary activity by creating a barrier to processive assembly of heterochromatin (Noma et al., 2006 ). However, we found that ETC4 retained both heterochromatin barrier and transcription-blocking insulator functions even under conditions in which ETC-site peripheral localization is ablated (Figure 9 and Supplemental Figure S5), implying that perinuclear localization is not required for these activities. Consistent with our observations, a recent study found although nuclear pore proteins associate with a tRNA gene barrier element at a modified HMRa locus, pore protein association is not essential for barrier activity (Ruben et al., 2011 ). The biological significance of ETC-site peripheral positioning is unclear, although one interesting possibility is of a relationship to condensin function, since the Pol III apparatus has been implicated in recruiting condensin to S. cerevisiae chromosomes (D'Ambrosio et al., 2008 ; Haeusler et al., 2008 ) and condensin is localized to a subset of the ETC sites. It will be interesting to explore further the relation between condensin, ETC-site function, and localization at the nuclear periphery.
The recent discovery of large numbers of ETC loci in the human and mouse genomes represents a particularly interesting addition to our knowledge of ETC/COC loci and reinforces the suggestion of additional roles for eukaryotic TFIIIC beyond its function in Pol III transcription (Simms et al., 2008 ; Moqtaderi et al., 2010 ; Carriere et al., 2012 ). The large number of mammalian ETC loci raises the question of whether additional ETC sites exist in the S. cerevisiae genome. Two recent studies hinted there may be uncharacterized TFIIIC-binding sites in S. cerevisiae (D'Ambrosio et al., 2008 ; Venters et al., 2011 ), which have yet to be validated or further investigated. Some of the human ETC sites contained a novel motif (instead of the known TFIIIC-binding motif), so it is even possible that additional yeast ETC sites might not contain a TFIIIC-binding consensus. Like yeast ETC sites, human ETC loci also tend to lie in closely spaced, divergently transcribed Pol II intergenic regions, hinting that human ETC loci could also act as chromatin boundary elements. Human ETC loci tend to occur near binding sites for CTCF, a protein implicated in higher-order organization of metazoan chromosomes through cohesin interaction, insulator function, and chromosome looping (Wallace and Felsenfeld, 2007 ; Parelho et al., 2008 ). Overall, the emerging evidence points toward an important role for ETC loci in chromosome spatial organization that is conserved throughout eukaryotes.
All yeast strains were constructed in the W303-1A background (ade2-1 trp1-1 leu2-3112 ura3-1 his3-11,15 can1-100). Strains are listed in Supplemental Table S1. Plasmids are listed in Supplemental Table S2. Standard techniques were used for DNA and yeast manipulations.
To tag each ETC locus with GFP, a suitable restriction site was identified in the genomic DNA near the ETC locus to be tagged. Primers were designed to amplify a ~400–base pair fragment containing this restriction site, and the fragment was cloned into lacOp repeat plasmid pAFS52 (Robinett et al., 1996 ). The resulting plasmid was cut at the unique restriction enzyme site and transformed into yeast strain GA-1320 (Heun et al., 2001 ), creating strains SBY1-SBY14 and SBY17-SBY25. In the cases of ETC1, ETC4, ETC5, and ETC8 the size of the intergene allowed the insertion of lacOp tagging sequences within the intergene occupied by the ETC site (Table 1); for ETC3, ETC6, and ETC7 the tag was inserted in a neighboring intergene. Table 1 shows ETC-site coordinates, ETC-site sequences, and relative distances of the lacOp insert from the ETC locus.
To test for ETC colocalization with the nucleolus, we tagged the endogenous NOP1 gene with mCherry (Shu et al., 2006 ). Briefly, SBY3 and SBY13 were transformed with a DNA fragment containing the natMX4 marker and mCherry targeted for in-frame insertion at the NOP1 3′ end, creating strains SBY31-33 and SBY49-51. To test for colocalization of ETC sites with telomeres, strains SBY84-86 and SBY87-89 were made by transforming SBY3 and SBY10, respectively, with plasmid pSB33 (YCp-mCherry-RAP1). The pSB33 plasmid resulted from exchanging GFP with mCherry (pKT355; Iwase et al., 2006 ) in plasmid YCp-GFP-RAP1 (Hiraga et al., 2006 ).
Strains SBY26, SBY27 (etc7Δ), and SBY37, SBY38 (etc6Δ) were derived from SBY3 and SBY1, respectively, by deleting the 23–base pair extended B box consensus sequence (23 base pairs) and 10 flanking base pairs on either side (43 base pairs total) using a fragment lacking this 43–base pair sequence but having 40–base pair homology on each side to the sequences flanking the deletion. This was performed in two steps. First the URA3 marker gene (YDp-U; Berben et al., 1991 ) was inserted at the appropriate ETC site, followed by disruption with either the etc6Δ or etc7Δ deletion fragment and selection of correct isolates by plating cells to 5-fluoroorotic acid. To create strains for Tfc1-FLAG ChIP (Figure 3B), we crossed SBY1 and SBY37 with DDY4058 and sporulated them to produce DDY4729 and DDY4732.
To insert an ETC site on another chromosome, we selected a suitable chromosomal locus (ChrXIV: RPS3-YNL179C intergene) and GFP tagged it (as described previously), creating SBY76, SBY77, and SBY78. A DNA fragment containing a kanMX marker flanked by loxP sites and a ~450–base pair sequence containing either ETC2 or ETC6 was targeted for integration next to the GFP tag. Removal of the kanMX marker by expressing Cre recombinase from a pSH47 plasmid (Guldener et al., 1996 ) then created SBY135, SBY136, and SBY137 and SBY139, SBY142, and SBY143. For transfer of ETC4, double-stranded synthetic oligonucleotides containing wild-type ETC4 or a mutated version of ETC4 with a single base substitution in B box consensus (etc4mut; Simms et al., 2008 ) were cloned in between BamHI and SalI sites of plasmid pU6H3FLAG (Katou et al., 2003 ). The resulting plasmids pSH136 and pSH138 contain wild type or etc4mut adjacent to loxP-kanMX-loxP, respectively. PCR fragments containing ETC4 and loxP-kanMX-loxP were PCR amplified with primers (with genomic sequence of chrXIV-302 at their 5′ ends) and used to transform strain SBY76. After insertion of ETC4 sequences, the kanMX marker was removed using galactose-inducible Cre recombinase of the plasmid pSH47. The resulting strains SHY465 and SHY468 have a 225–base pair sequence inserted at the chrXIV-302 locus containing ETC4 or etc4mut, respectively.
All LexA fusions were created in pAT4 (Taddei et al., 2004 ). Fusion proteins were created by inserting the full-length sequences of TFC1, TFC3, TFC4, TFC6, TFC7 (YOR110W), or TFC8 (made by PCR amplification) into pAT4. Error-free constructs were confirmed by sequencing, and the resulting plasmids were then used to transform strain GA-1461 (Hediger et al., 2002 ) to create SBY144-149, SBY155-166, and SBY211-212.
To construct strains suitable for ChIP analysis of Tfc1-FLAG recruitment by LexA-Tfc fusions, first we cloned double-stranded synthetic DNA containing four LexA operator sequences between the BsiWI and SalI site of plasmid pUG27 to obtain plasmid pSH142. Using pSH142 as a PCR template, we inserted the LexA operator array near the ARS607 locus of DDY4071 by one-step PCR replacement to obtain SHY451. The HIS3MX maker was then removed by Cre recombinase to obtain strain SHY457. Strain SHY457 was transformed with a plasmid pAT4, pSB48, or pSB50 to obtain strain SHY459, SHY461, or SHY463, respectively.
Strains SBY191, SBY195, and SBY196 (ETC6; mps3Δ75-150) and SBY194, SBY197, and SBY198 (ETC2; mps3Δ75-150) were derived from SBY1 and SBY10, respectively, by directing integration of BstEII-digested pSJ519 plasmid (mps3Δ75-150; Bupp et al., 2007 ) at the chromosomal LEU2 locus, followed by deletion of the chromosomal copy of MPS3 using a natMX4 cassette amplified from strain SLJ2059 (Bupp et al., 2007 ). Single-copy integration was verified by Southern blotting. Plasmid pSH129 was constructed by recloning the BamHI-SalI fragment of pSJ148 (bearing the MPS3 gene) into BamHI-SalI–cut pRS316.
The Mps3-N-tetR-mCherry construct, pSB79, was created in a pRS426 backbone through three steps of ligation. The promoter region and N-terminal domain of MPS3 N-terminal domain containing residues 1–151 of MPS3 were amplified from pSJ148 plasmid (Bupp et al., 2007 ) using primers that incorporated the XhoI and AatII, EcoRI restriction sites at the 5′ and 3′ ends of the fragment, respectively. The tetR coding region, flanked by SV40 NLS at its N-terminal end, was amplified from p3524 plasmid (Michaelis et al., 1997 ) using primers that incorporated the AatII and NheI, SpeI restriction sites, whereas the coding region and termination sequence for mCherry were amplified from pKT355 plasmid (Iwase et al., 2006 ) using primers that incorporated the NheI and NotI restriction sites. Initial ligation of Mps3-N′ under its own promoter using the XhoI and EcoRI restriction sites was followed by in-frame ligation of tetR using the AatII and SpeI restriction sites and concluded with in-frame ligation of mCherry and ADHter to the existing Mps3-N-tetR fusion using the NheI and NotI restriction sites. Strains SBY215, SBY216 (ETC6; Mps3-N′); SBY217, SBY218 (ETC4; Mps3-N′); SBY219, SBY220 (LexA-Tfc7; Mps3-N′); SBY221, SBY222 (LexA-Tfc1; Mps3-N′); and SBY223, SBY224 (LexA-Yif1; Mps3-N′) were derived from SBY1, SBY22, SBY147, SBY155, and SBY212, respectively, by transforming the aforementioned strains with the multicopy plasmid pSB79 (pRS426-Mps3-N′-tetR-mCherry).
To test for correct homologous insertion and replacement events, suitable PCR amplification reactions were designed to analyze the junction sites. ETC-site deletions, LexA fusions, and pSB79 (Mps3-N′) construct were confirmed by sequencing.
Insulator assays were as described (Simms et al., 2008 ) and barrier assays as in Jambunathan et al. (2005 ). Strains to test ETC1 barrier and insulator activity (Supplemental Figure S6) were constructed as described (Simms et al., 2008 ).
To make strains for the auxin-inducible degradation experiments, the OsTIR1 gene was inserted into the genomic URA3 locus of strain SBY22 as described (Nishimura et al., 2009 ) to obtain strain SHY472. To obtain strain SHY476, an auxin-inducible degron was added to the C-terminus of the genomic copy of the TFC3 gene in SHY472 as described (Nishimura et al., 2009 ). Strains SHY472 and SHY476 were cultivated in synthetic raffinose medium buffered at pH5.5 with appropriate auxotrophic selection. At OD600 = 0.2–0.3, galactose was added to a final concentration of 2%. One hour after addition of galactose, IAA (Sigma-Aldrich, St. Louis, MO I2886) was added to a final concentration of 0.5 mM. Cells were examined for ETC4 localization between 1 and 2 h after the addition of IAA.
Chromatin immunoprecipitation assays were performed essentially as described (Rusche et al., 2002 ).
Primers used to assess TFIIIC binding at ETC6 and etc6Δ delete loci (Figure 3B) were as follows:
Primers used to assess binding to lexAOp sequences (Figure 6E) were as follows:
The primers for the control tDNA R (CCG) on chromosome XII were as follows:
Other primer sequences are available upon request.
Microscopic techniques were performed as described in Hiraga et al. (2006 ). Briefly, a DeltaVision RT (Applied Precision, Issaquah, WA) microscope system with an UPlanApo 100× objective (1.35 numerical aperture; Olympus, Center Valley, PA), CoolSnap HQ monochrome cooled charge-coupled device camera (Photometrics, Tucson, AZ), and SoftWoRx (Applied Precision) acquisition software were used to acquire images. For observation of GFP and mCherry fluorescence, 30 Z-stack images were taken at 250-nm intervals with fluorescein isothiocyanate and tetramethylrhodamine isothiocyanate or DsRed filter sets. Differential interference contrast (DIC) images acquired at the same Z-intervals were used for determination of cell cycle stages by bud size: G1 phase, unbudded; S phase, cells with bud ≤2 μm; G2 phase, cells with a bud >2 μm and a spherical (i.e., nonmitotic) nucleus not at the bud neck.
Quantitative evaluation of GFP-tagged chromosomal dot localization was performed as described (Taddei et al., 2004 ). SoftWoRx Explorer (Applied Precision) was used to measure dot-to–nuclear envelope distance in yeast cells where the GFP dot was located within one of the three equatorial sections of its nucleus. Briefly, localization of the GFP dot was scored in two dimensions against three imaginary concentric zones of equal area, as shown in Figure 1B. At least 300 cells were scored for each isolate measurement (unless otherwise noted). p values were calculated by χ2 test against either random distribution or wild-type values.
Quantitative evaluation of GFP-tagged chromosomal dot colocalization either with the nucleolus or telomere foci was performed as follows. SoftWoRx Explorer was used to measure dot-to-nucleolus or dot-to-telomere foci distance in yeast cells. Briefly, colocalization of the GFP dot to either the nucleolus or telomere foci was scored in two dimensions if the two structures coincided or were juxtaposed (distance <0.26 μm) when observed within the equatorial region of a Z-stack of images (Figure 5, A and B). At least 200 cells were scored for each isolate measurement (unless otherwise noted). p values were calculated by χ2 test against random distribution.
We thank all the members of the Donaldson group for comments and technical advice. We thank Sue Jaspersen and Susan Gasser for providing strains and constructs. S.B. was funded by a University of Aberdeen Scholarship. This research was supported by Wellcome Trust Grant 082377/Z/07/Z to A.D. and National Science Foundation Grant MCB-0817823 to D.D.
This article was published online ahead of print in MBoC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E11-04-0365) on April 11, 2012.