• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of embojLink to Publisher's site
EMBO J. Mar 24, 2004; 23(6): 1301–1312.
Published online Mar 11, 2004. doi:  10.1038/sj.emboj.7600144
PMCID: PMC381416

Separation of silencing from perinuclear anchoring functions in yeast Ku80, Sir4 and Esc1 proteins

Abstract

In budding yeast, the nuclear periphery forms a subcompartment in which telomeres cluster and SIR proteins concentrate. To identify the proteins that mediate chromatin anchorage to the nuclear envelope, candidates were fused to LexA and targeted to an internal GFP-tagged chromosomal locus. Their ability to shift the locus from a random to a peripheral subnuclear position was monitored in living cells. Using fusions that cannot silence, we identify YKu80 and a 312-aa domain of Sir4 (Sir4PAD) as minimal anchoring elements, each able to relocalize an internal chromosomal locus to the nuclear periphery. Sir4PAD-mediated tethering requires either the Ku complex or Esc1, an acidic protein that is localized to the inner face of the nuclear envelope even in the absence of Ku, Sir4 or Nup133. Finally, we demonstrate that Ku- and Esc1-dependent pathways mediate natural telomere anchoring in vivo. These data provide the first unambiguous identification of protein interactions that are both necessary and sufficient to localize chromatin to the nuclear envelope.

Keywords: Esc1, Ku, nuclear organization, SIR proteins, telomeric heterochromatin

Introduction

Chromosomes assume a nonrandom distribution in interphase nuclei (Marshall, 2002). The function of their spatial arrangement is largely unknown, although recent evidence suggests that subnuclear compartments contribute to the establishment and maintenance of epigenetic controls over eukaryotic gene expression (Fisher and Merkenschlager, 2002).

In yeast, flies and man, the protective ends of chromosomes known as telomeres nucleate the formation of an altered chromatin structure that represses the transcription of adjacent RNA pol II genes in a heritable fashion (termed telomere position effect or TPE; reviewed in Huang, 2002; Perrod and Gasser, 2003). Repressed telomeric chromatin in yeast is found near the nuclear envelope (NE) in discrete foci that are far fewer than the number of chromosomal ends. These clusters of telomeric DNA colocalize with pools of silent information regulatory proteins (Sir2–4; Gotta et al, 1996), which are essential for TPE. Intriguingly, a similar spatial juxtaposition of telomeres can be observed in the parasite Plasmodia, where the clustering favors subtelomeric gene conversion events between virulence factor loci (Freitas-Junior et al, 2000).

In yeast, DNA elements known as silencers bind the sequence-specific factors Abf1, Rap1 and ORC, which nucleate the binding of the SIR complex at the cryptic mating-type loci, HML and HMR (Rusche et al, 2003). At telomeres, on the other hand, Rap1 binds the terminal [TG1–3]n repeat and cooperates with the end-binding complex YKu70/YKu80 to recruit SIR proteins (Moretti et al, 1994; Cockell et al, 1995). Once bound, the Sir complex spreads along adjacent nucleosomes preferentially binding underacetylated histone tails (Renauld et al, 1993; Hecht et al, 1996; Rusche et al, 2002).

Several observations support the proposal that subnuclear organization influences silencing. First, the delocalization of SIR factors from telomeres promotes repression at silencer-flanked loci that usually remain active due to their distance from chromosomal ends (Maillet et al, 1996, 2001; Marcand et al, 1996). This suggests that telomeric foci sequester silencing factors from nontelomeric sites of action. Second, the repression of HML and other silencer-flanked reporters is aided by their proximity to telomeres (Thompson et al, 1994; Maillet et al, 1996; Marcand et al, 1996). Third, silencing can be restored to a derepressed allele of HMR by targeting proteins of the endoplasmic reticulum to the locus via a Gal4 DNA-binding domain (e.g. Yif1, Yip1; Andrulis et al, 1998). These membrane-spanning hybrids accumulate in NE and are thought to promote repression by tethering the reporter gene near telomeric pools of SIR proteins. It has never been demonstrated, however, that the Yif-targeted reporter genes are actually recruited to the NE.

The proteins that mediate the perinuclear anchoring of yeast telomeres and silent loci are unknown, although mutations in YKU or SIR2–4 do affect telomere position. Indeed, the deletion of either subunit of the Ku heterodimer delocalizes roughly half of the yeast telomeres from the NE (Laroche et al, 1998), but leaves truncated telomeres tethered (Tham et al, 2001; Hediger et al, 2002a). The complete delocalization of certain telomeres, such as Tel 14L or a truncated Tel 6R (Tel 6Rt), was detected only after deletion of both sir4 and yku70, suggesting that redundant anchoring mechanisms exist (Hediger et al, 2002a). Because the restoration of silent chromatin allowed anchoring in the absence of Ku but in a Sir4-dependent manner, it was proposed that chromatin-bound SIR complexes and Ku define two partially redundant tethering pathways.

Among the silencing factors in yeast, Sir4 is most likely to act as a chromatin anchor. First, ectopic expression of its coiled-coil carboxy-terminal fragment leads to the release of full-length Sir3 and Sir4 from an insoluble chromatin fraction (Cockell et al, 1995) and loss of TPE. Indeed, based on the homology between this domain and human lamins A and C (Diffley and Stillman, 1989), it has been speculated that Sir4 might substitute for the nuclear lamina, which yeast lack. Finally, a penultimate subdomain of Sir4 (aa 950–1262 called Sir4PAD for partitioning and anchoring domain) was found to confer efficient mitotic partitioning on otherwise unstable plasmids and to impair the free rotation of DNA (Ansari and Gartenberg, 1997). One mechanism that could account for both phenotypes would be the anchoring of DNA to a symmetrically segregating nuclear component.

Recent work shows that plasmid partitioning by Sir4PAD requires Esc1 (establishes silent chromatin), a protein that associates with Sir4, imparts silencing when targeted to a crippled silencer-flanked reporter at HMR, and accumulates at the nuclear periphery when overexpressed (Andrulis et al, 2002). These data gave rise to the proposal that the Sir4–Esc1 interaction might link silent chromatin to a scaffolding at the nuclear periphery. Unfortunately, this hypothesis could not be tested in loss-of-function alleles, because Esc1, like Ku and Sir4, promotes and stabilizes silent chromatin. Indeed, to determine which elements mediate chromatin anchoring, it was necessary to separate the silencing functions of the major telomere-associated proteins from any tethering activity they might have.

We show here that a silencing-incompetent YKu80 monomer and a subdomain of Sir4 can each relocalize chromatin to the nuclear periphery without nucleating silent chromatin. Furthermore, we show that Esc1 functions as a perinuclear anchor for this silencing-deficient domain of Sir4, but not as its exclusive binding site. Sir4PAD tethers chromatin through two pathways: one requiring yKu and the second Esc1. Finally, we show that these two pathways cooperate to anchor native telomeres in a cell-cycle-regulated manner. We propose that, unlike the chromatin components Sir4 and yKu, Esc1 plays a structural role determining chromosome domain position in vivo.

Results

An assay for chromatin relocalization

We have developed a cytological screen for proteins that impart a specific nuclear localization to an otherwise randomly positioned chromosomal segment. Four LexA operators (lexAop) were linked to an array of 256 lacop-binding sites and integrated at an internal active locus >50 kb from either the centromere or the right telomere of chromosome 6, near ARS607 (called Chr6int; Figure 1A). When LacI–GFP is expressed in these cells, it binds the array creating a single focus of GFP fluorescence. The ability of various LexA fusion proteins to influence the subnuclear position of the array can then be assessed.

Figure 1
An assay for perinuclear chromatin anchoring. (A) Strain GA-1461 bears lacop repeats and LexA-binding sites at an internal locus on the right arm of Chr 6 (Chr6int) and a GFP–Nup49 fusion. A typical single-plane confocal image superimposed on ...

Three-dimensional (3D) focal stacks were collected to measure the distance between LacI–GFP foci and the nuclear membrane (tagged with GFP–Nup49). Positions measured in several hundred cells show that the Chr6int lacop array is randomly distributed among three concentric nuclear zones of equal surface, with or without LexA (Figure 1B). Spot position is monitored in one focal plane with zones normalized to the measured nuclear diameter in this plane. To show that relocalization can occur, we first targeted LexA fused to an integral membrane protein called Yif1 (Andrulis et al, 1998) to the Chr6int locus. We observe a significant enrichment of Chr6int in the outermost nuclear zone throughout the cell cycle (Figure 1C). Given that the fluorescent GFP–Nup49 ring occupies a large fraction of zone 1, the presence of a chromosomal signal in this zone indicates close NE–DNA contact (<180 nm). We conclude that the LexA–Yif1 fusion, which promotes repression when targeted to a crippled silencer-flanked reporter gene (Andrulis et al, 1998), can indeed relocalize a chromosomal locus to the nuclear periphery.

Because neither the silencing nor anchorage is 100% efficient, we monitored the residence time of the tagged locus at the periphery using time-lapse microscopy (Hediger et al, 2002a; Heun et al, 2001b). A total of 300–400 sequential images were collected at 1.5 s intervals from individual cells expressing either LexA or LexA–Yif1. Without LexA–Yif1, the locus moves freely in the nucleoplasm without any discernible perinuclear constraint (Figure 1D). In the presence of LexA–Yif1, the tagged locus oscillates back and forth along the NE, much like the movement of natural telomeres (Hediger et al, 2002a). Movements away from the nuclear periphery are occasional. We conclude that an internal nonsilenced chromosomal locus becomes significantly but reversibly associated with the NE by binding LexA–Yif1. Importantly, these results demonstrate that our chromatin ‘relocation' assay can be used to identify protein domains that tether chromosomes to the nuclear periphery.

Identification of silencing-defective YKU80 and SIR4 alleles

Sir4 and the Ku complex represented logical anchoring candidates based on loss-of-function analyses. However, YKu80 and full-length Sir4, as well as its N- or C-terminal domains, nucleate silent chromatin when targeted to DNA (Marcand et al, 1996; Martin et al, 1999). Thus, in order to test a chromatin relocation activity in the absence of silent chromatin, we sought fusions of Sir4 and Ku that would separate silencing and anchoring activities. For Sir4, we fused LexA to the Sir4PAD domain, which unlike other Sir4 domains has no dominant-negative effect on TPE (Ansari and Gartenberg, 1997). For the Ku heterodimer, on the other hand, we created mutant forms of yku80 by degenerate PCR. Two alleles were identified (yku80-4 and yku80-9; see Figure 2A) that are unable to restore TPE in the yku80 null background.

Figure 2
Silencing-incompetent yku80-4 and Sir4PAD anchor chromatin at the nuclear periphery. (A) Scheme of the different domains and mutants used in targeted silencing and relocation assays. Dashed boxes correspond to putative coiled-coil domains. (B) Different ...

To test whether these proteins nucleate silencing when targeted to a crippled HM locus, we inserted a TRP1 reporter gene at HMR and replaced the Rap1 and Abf1 sites of the E silencer with four lexAop sites (creating a crippled silencer; c.f. Andrulis et al, 1998). Growth of serially diluted cultures in the presence or absence of tryptophan allows a qualitative measure of TRP1 repression. Figure 2B shows that whereas targeted LexA–yku80-9 confers robust silencing relative to LexA–Yif1 (100 × difference), no silencing was seen with LexA–yku80-4, LexA–Sir4PAD or LexA alone. LexA fused to either full-length Sir2 or the C-terminal 540 residues of Esc1 (Esc1c) also silenced efficiently, in agreement with previous reports (Cockell et al, 2000; Andrulis et al, 2002).

It was important to show not only that the Sir4PAD domain and yku80-4 fail to promote repression, but that they also fail to bind other SIR components. By two-hybrid analysis we detect no significant interaction of Sir4PAD with any of the domains of Sir2, Sir3 and Sir4 that are known to be critical for repression (Figure 2C). On the other hand, we detect a strong interaction between Sir4PAD and Esc1C, as reported, when either protein is used as bait or prey (Andrulis et al, 2002; Figure 2C and D). We next tested the two yku80 mutant proteins for two-hybrid interaction with Sir4C, an interaction previously reported for YKu70 (Tsukamoto et al, 1997). The mutant that supports targeted silencing (yku80-9) binds Sir4C robustly, while the one deficient for targeted silencing (yku80-4) does not. Importantly, the yku80-9/Sir4C interaction does not require YKu70 (Figure 2E). Because LexA-yku80-4 and LexA–Sir4PAD neither promote targeted silencing nor bind the silencing relevant domains of SIR proteins, we are able to test their anchoring activity independently of Sir complex formation.

Chromatin relocation activities of YKu80, Sir4PAD, Esc1 and Sir2

We expressed the LexA fusions discussed above in the tagged Chr6int strain to monitor their abilities to relocalize DNA to the NE (Figure 3). Quantitative analysis shows that both YKu80 fusions produce significant relocation of the GFP focus to the outermost zone of the nucleus. The tethering activity is observed in G1- and S-phase cells (zone 1 values ranging from 51%, p=10−10 to 61%, p=10−6), although not in G2-phase cells, where the locus remains random (24 to 29%, p=0.055). Statistical analyses are summarized in Table I. Importantly, the anchoring efficiencies of the two yku80 mutants are very similar, despite their opposite behavior with respect to silencing. This demonstrates the separation of function we sought to obtain (Figure 3A).

Figure 3
YKu80, Sir4PAD and Esc1C relocalize chromatin to the nuclear periphery. The position of Chr6int with respect to the three concentric nuclear zones was determined as in Figure 1 in GA-1461 expressing the following fusions: (A) LexA–yku80-4 and ...
Table 1
Localization of lacop -tagged loci (n) and significance (p value) for zone 1 enrichment

Sir4PAD and Esc1C fusions are both extremely efficient at chromatin relocalization (Figure 3B and C). The Sir4PAD tethers Chr6int to the NE efficiently throughout the cell cycle (60–65% in zone 1, p=10−14–10−2), and Esc1C does so even better (61–82% in zone 1, p=10−16–10−5). Intriguingly, and in contrast to these, a LexA–YKu70 fusion anchors Chr6int at the NE weakly and only during G1 phase (49% in zone 1; Table I).

Sir4PAD is able to promote efficient chromatin association with the NE without interacting with other SIR components. This makes Sir4 a good candidate for the previously reported Ku-independent perinuclear anchor that sequesters silent chromatin (Hediger et al, 2002a). If true, one would expect the tethering by other SIR factors to be SIR4 dependent. Indeed, LexA–Sir2 can efficiently tether the tagged locus to the nuclear periphery and this is entirely lost in a sir4 deletion (Figure 3D).

In conclusion, while YKu80, Sir4PAD, Esc1C and Sir2 all mediate significant chromatin relocation to the NE, we can separate chromatin anchoring from the establishment of silenced chromatin for Sir4PAD and yku80-4 fusions (Figures 2 and and3).3). Because they possess silencing-independent anchorage functions, we propose that Sir4 and YKu80 have direct rather than indirect roles in perinuclear tethering of chromatin.

Association of Esc1p with the NE in the absence of SIRs, Ku or silent chromatin

Unlike Sir4PAD and yku80-4, Esc1C promotes both silencing, presumably through its affinity for Sir4, and chromatin anchoring. To distinguish its role as a structural element as opposed to a silent chromatin factor, we determined Esc1 localization in silencing-deficient cells, since integral silent chromatin factors generally become dispersed from perinuclear foci when TPE is lost (Gotta et al, 1996; Martin et al, 1999).

Previous work has shown that overexpressed Esc1 is perinuclear (Andrulis et al, 2002). Fearing artifacts of overexpression, we created 3′ fusions of Esc1 with either GFP or Myc epitopes by integration at the endogenous ESC1 locus. These cells have no defects in TPE nor in NE morphology. Live epifluorescence of Esc1–GFP reveals a patchy perinuclear localization that overlaps only partially with CFP-tagged nuclear pores (Figure 4B), and the double Nop1–Esc1 labelling shows that Esc1–GFP is systematically excluded from the zone adjacent to the nucleolus (Figure 4A). Using an immunofluorescence nanogold labelling technique, we next localized Esc1 by both confocal and electron microscopy (Figure 4C and D). Electron micrographs show that Esc1 localizes mainly to the inner membrane of the NE between nuclear pores. Indeed, scoring gold particles in 50 cells we note that 85% (501/587) are within 100 nm of the NE (see table in Figure 4D). Of these, only 4% are located within 100 nm of a visible nuclear pore.

Figure 4
Esc1 is localized on the nucleoplasmic side of the inner nuclear membrane independently of YKu70 and Sir4. (A) Live visualization of GA-2121 bearing endogenous Esc1–GFP and the nucleolar protein Nop1 fused to CFP. Equatorial sections of nuclei ...

To determine whether the positioning of Esc1 is influenced by pore distribution, we examined its localization in a nup133Δ strain, a mutation that causes nuclear pores to cluster (Doye et al, 1994). Esc1–GFP remains evenly distributed around the nucleus despite the concentration of pores in one or two foci (Nup49–CFP; Figure 4B). This shows that Esc1 distribution, like that of telomeres (Hediger et al, 2002a), is unlinked to that of nuclear pores. Finally, we show that the localization of Esc1 is unchanged in strains mutant for yku70, sir4 or both, demonstrating that its perinuclear localization does not depend on either the presence of a functional Ku heterodimer or on silencing (Figure 4B, and data not shown). Thus Esc1 is a chromatin anchor whose localization is independent of both telomeric repression and telomere localization.

Esc1 and the Ku complex define two pathways for Sir4PAD anchorage

Because Sir4 interacts with both YKu70/YKu80 and Esc1, it was essential to investigate the interdependence of the Ku–Sir4PAD- and Esc1C-mediated anchoring functions. To do this, we scored for GFP–Chr6int relocation by all three LexA fusions in strains bearing deletions of yku70, esc1 or sir4 (Figure 5).

Figure 5
Redundant pathways anchor yku80-4 and Sir4PAD at the nuclear periphery. (A–D) LexA fusions were expressed in GA-1461 bearing complete disruptions of yku70, esc1, sir4 and yku70 esc1, as indicated. Chr6int position was scored as in Figure 1, and ...

LexA–Sir4PAD chromatin relocation activity was unaffected by loss of either Esc1 or YKu70 individually (Figure 5A), yet we see a complete loss of the Sir4PAD tethering activity in the absence of both Esc1 and Ku, throughout the cell cycle. This result strongly suggests that Esc1 and Ku provide two parallel anchorage pathways for Sir4PAD (Figure 5B). Indeed, the Sir4–Esc1 interaction can account for the yKu-independent anchoring described previously (Tham et al, 2001; Hediger et al, 2002a). Intriguingly, LexA–Sir4PAD anchors more efficiently in the absence of endogenous Sir4. This latter could reflect competition between the full-length Sir4 and Sir4PAD for a limiting number of perinuclear binding sites.

We next examined whether anchoring via Esc1C requires Sir4 or Ku. Consistent with Esc1 localization results (Figure 4B), the Esc1C-mediated relocation of Chr6int shows little dependence on either SIR4 or YKU70 (Figure 5C). It is sensitive, on the other hand, to the deletion of the genomic ESC1 gene. The simplest explanation for this is that Esc1 homodimerizes through its C-terminal domain, which is present in the LexA fusion, but requires its N-terminus for perinuclear localization. It is not clear why the requirement for full-length Esc1 is more pronounced in G1 phase. To rule out nonspecific effects of these mutations, we show that the anchoring activity of LexA–Yif1 promotes significant anchoring in all deletion strains tested (Figure 1E, Table I).

The YKu80 anchoring activity requires YKu70 only in G1-phase cells

We next examined whether the chromatin relocation activity associated with yku80-4 requires the Ku heterodimer. Surprisingly, we find that yku80-4-mediated anchoring is lost in the yku70 mutant in G1 phase, yet is maintained in this strain during S phase (Figure 5D). It appears, therefore, that there is one anchorage site in G1 for the Ku heterodimer, and another in S phase that recognizes YKu80 directly. Coupled with the fact that LexA–YKu70 can relocate chromatin to the periphery only in G1 phase (Table I), we propose that cell-cycle-regulated changes in the Ku complex influence anchoring. We further observe that the yku80-4 anchorage function is sensitive to deletion of ESC1 uniquely in S phase (Figure 5D). This suggests an S-phase-specific interaction between YKu80 and Esc1, or between YKu80 and an Esc1-dependent factor, distinct from the G1-phase Ku anchor (Figure 5D).

In summary, YKu80-mediated anchoring in G1 phase requires YKu70 but is independent of Esc1, while in S-phase cells it requires Esc1 (Figure 5B). Importantly, these cell-cycle variations mirror differences observed for the positioning of GFP-tagged telomeres in yku70 mutants (Hediger et al, 2002a).

Esc1 tethers natural telomeres and promotes repression

Recent analysis of telomere localization in living cells has shown telomere-specific effect of YKU mutations: Tel 6R was randomly localized in a yku70 deletion strain, while Tel 14L and a truncated Tel 6Rt remained significantly perinuclear. Importantly, however, the residual Ku-independent anchoring was lost upon SIR4 deletion (Hediger et al, 2002a).

To test whether the Sir4-dependent anchorage of telomeres involves the Sir4PAD–Esc1 interaction, we deleted ESC1 and YKU70 in strains bearing GFP-tagged telomeres and scored telomere position relative to the NE. Whereas the deletion of ESC1 alone has little effect, the esc1 and yku70 effects are additive for Tel 14L delocalization, leading to a random distribution in S phase (Figure 6A). For Tel 6Rt, the Esc1 dependence even is more pronounced: esc1 deletion alone delocalizes Tel 6Rt significantly, and random distributions are detected in both G1- and S-phase esc1 ku70 cells (Figure 6B). Thus, the esc1 yku70 double mutant mimics the sir4 yku70 mutant with respect to both the Tel 6Rt and Tel 14L localization (Hediger et al, 2002a), arguing that the two redundant pathways of telomere anchoring act directly through Ku and Esc1–Sir4, respectively. We observe an inherent cell-cycle variation in their use: the Ku anchor is dominant in G1 phase, while yku70 deletion has little effect on anchorage in S phase (Figure 6E).

Figure 6
The Ku heterodimer and Esc1 anchor telomeres in vivo and regulate TPE. (A) Tel 14L position was determined in G1- and S-phase cells of GA-1985, yku70 (GA-1983), esc1 (GA-2074) and yku70 esc1 (GA-2082) strains as in Figure 1. Graphs and symbols are as ...

Consistent with this redundancy, we note that the esc1 deletion induces only a minor reduction in silencing efficiency (10- to 100-fold) for a URA3 reporter at Tel 7L (Andrulis et al, 2002; Figure 6C). To correlate the effects on telomere position and silencing at a single telomere, we monitored repression of an ADE2 reporter inserted in the lacop-tagged Tel 6Rt. Consistent with the partial Tel 6Rt delocalization observed in the esc1 null strain, ADE2 becomes partially derepressed, producing pink rather than the red/white sectored colonies that characterize wild-type cells (Figure 6D). Loss of yku70 leads to near total derepression because Sir4 recruitment is impaired. Neither a yku70 nor esc1 deletion has any effect on URA3 or ADE2 expression at their normal genomic loci based on genome-wide microarray analysis (data not shown). The fact that repression and telomere positioning are altered at the same telomere (Tel 6Rt::ADE2) in the same strain supports the notion that Esc1-mediated telomere anchoring directly influences TPE.

As further proof that Esc1 is the physiological tether of at least a subset of natural telomeres, we assessed the presence of Esc1 at telomeres by chromatin immunoprecipitation (ChIP). Using an endogenous Myc-tagged Esc1 and quantitative real-time PCR, we find that a DNA fragment located 600 bp away from the TG repeat of the native Tel 6R was enriched 2.8-fold over a nontelomeric fragment (located within SMC2 gene; Figure 6F). Enrichment is calculated from real-time PCR accumulation curves that are normalized to signal ratios in the input DNA. This enrichment is not detected when crosslinking is omitted or a control antibody is used. The presence of Ku and SIR complexes at this telomere was previously shown (Martin et al, 1999).

Esc1/Sir4 colocalization increases as TPE improves

If Esc1–Sir4 interaction does anchor a subpopulation of silent yeast telomeres, one might expect that Sir4 foci would colocalize with Esc1 at the NE. To address this point, we performed double immunolocalization of Esc1–13Myc and Sir4 in a GFP–Nup49-tagged strain. We find that 26% of the Sir4 foci colocalize with Esc1 (Figure 7C). This is consistent with the partial colocalization monitored for Sir4 and overexpressed Esc1–GFP (Andrulis et al, 2002). In contrast, we find that only 12% of the Sir4 foci colocalize with the pore staining in the same cells (Figure 7B). We then asked if we could improve the amount of Esc1/Sir4 colocalization by increasing the amount of Sir4 associated with telomeres. This can be achieved by eliminating the Rap1 interacting factor Rif1, which competes with Sir4 for binding to the telomere-bound Rap1. Loss of Rif1 is known to improve Sir4 recruitment by Rap1, which in turn improves silencing efficiency (Buck and Shore, 1995). We show in Figure 7D that the colocalization of Esc1 and Sir4 increases to 55% of the Sir4 foci in rif1 deletion cells. Thus, the number of Sir4 foci that colocalize with Esc1 increases when silencing improves, consistent with the idea that Esc1 anchors telomeric silent chromatin through Sir4.

Figure 7
Sir4 foci–Esc1 colocalization increases with increased TPE. (A) Equatorial confocal sections with Sir4 (red) and Esc1–13Myc (blue) revealed by immunofluorescence and Nup49–GFP (green) by direct fluorescence in GA-2110. (B, C) To ...

Discussion

Many results suggest a correlation between perinuclear position and silencing in yeast. Nonetheless, establishing a causal relationship between subnuclear organization and repression has been difficult, because the mutants known to alter the position of silent domains are also directly implicated in the repression machinery. Here we have exploited a chromatin relocalization assay that allows us to monitor protein domains that determine subnuclear position independently of their interaction with silencing factors. The minimal protein domains that are sufficient to relocate a tagged locus to the NE include YKu80, Sir4PAD and the C-terminal domain of Esc1. By coupling this chromatin relocation assay with complete gene disruptions of interacting factors, we have identified interactions that can account for the redundant pathways that anchor telomeres in living cells. Indeed, monitoring the position of tagged telomeres in double mutants shows that these same proteins tether yeast telomeres in vivo.

Esc1: a nonchromatin element involved in telomere positioning

Esc1 was originally identified as a ligand for the Sir4PAD fragment, and was recovered in a screen for fusion proteins that enhance subtelomeric repression in rap1ΔC cells, in which SIR proteins are dispersed from telomeres (Andrulis et al, 2002). Using immunoEM studies, we show that Esc1 is almost exclusively localized along the inner face of the NE, yet it does not colocalize with nuclear pores, nor is its position affected by mutants that lead to pore clustering (Figure 4). Esc1 has no membrane-spanning domain but appears to bear significant post-translational modification. Its solubilization properties are consistent with membrane association through lipid modification (data not shown).

When fused to a DNA-binding domain, Esc1C very efficiently relocalizes nontelomeric chromatin to the nuclear periphery. The chromatin relocation activity of Esc1C requires full-length Esc1 protein, suggesting that its C-terminus mediates homodimerization. Endogenous Esc1 is also necessary for the perinuclear tethering of chromatin by Sir4PAD in the absence of YKu70 and for the S-phase-specific anchorage mediated by YKu80. Finally, and most importantly, we have shown that Esc1 cooperates with Ku to anchor natural telomeres, being a physiological anchor for both Sir4 and silent chromatin. Consistently, Sir4 is homogenously distributed throughout the nucleoplasm in a double yku70 esc1 mutant, whereas residual Sir4 foci can still be detected in the single yku70 mutant (data not shown). In support of these mutant studies, Esc1 can be recovered crosslinked to telomeres and we observe a positive correlation between Sir4/Esc1 colocalization and TPE (Figure 7).

The fact that silent telomeres only partially colocalize with Esc1 suggests that there is more than one anchorage site for silent chromatin. Indeed, Sir4 itself can tether chromatin through Ku as well as through Esc1, and we cannot exclude that other weak anchorage sites exist for other telomeres. The dual pathways for tethering Sir4 are mirrored in the dual pathways that tether telomeres (Hediger et al, 2002a), and we predict that, as for telomeres, both pathways must be disrupted to release silent chromatin (Figure 7E). In recent studies that allow the separation of a silent chromatin domain from its chromosomal context, we find that the combined deletion of yku70 and esc1, but neither mutation alone, releases an excised silent chromatin ring from perinuclear constraints (Gartenberg et al, submitted).

Because endogenous Esc1 localization, as well as chromatin tethering through the targeted LexA–Esc1C fusion, requires neither Sir4, Ku nor chromatin repression, we propose that Esc1 is a structural component of the yeast nuclear periphery, possibly replacing lamin-associated proteins as a chromatin anchorage site in yeast. Esc1 is highly acidic, with three short coiled-coil domains that may be critical for homodimerization and interaction with other proteins. There are no obvious Esc1 homologs in animal genomes, although recent data suggest that emerin and the lamin B receptor are involved in chromatin anchoring at the mammalian nuclear periphery (Holaska et al, 2002). The functional similarities between emerin and Esc1 include evidence that neither is an integral chromatin factor, yet both interact with chromatin components to recruit DNA to the NE. By immunoEM we monitor on average 12 Esc1-specific grains per 90 nm nuclear section. This, together with its weak fluorescence, suggests that Esc1 is a low-abundance protein.

Cell-cycle variations in anchoring pathways

The relative efficiency of the Ku and Sir4–Esc1 anchorage pathways varies with the cell cycle. Strikingly, the cell-cycle variations that were reported for natural telomere anchorage in yeast (Tham et al, 2001; Hediger et al, 2002a) were accurately reproduced by the efficiency of LexA–yku80-4 tethering. The Ku pathway is prevalent in G1-phase cells, while the Esc1 pathway dominates S-phase anchoring; both are compromised in metaphase cells (Tham et al, 2001; Hediger et al, 2002a).

Cell-cycle modification of Ku, Sir4 or their NE partners could be responsible for these cell-cycle variations in chromatin anchoring pathways. Indeed, Ku interaction with the RNA component of telomerase may be under cell-cycle control (Stellwagen et al, 2003), and telomeric chromatin has been shown to change its composition through the cell cycle (Smith et al, 2003). We feel it is unlikely that the S-phase changes in anchoring reflect a destabilization of chromatin structure due to passage of the replication fork, because telomeres are replicated late in S phase, while the change in telomere/chromatin anchoring is detected as soon as buds emerge. On the other hand, these cell-cycle fluctuations in anchorage mechanisms correlate well with the fact that yeast chromatin must traverse S phase, but not necessarily replicate, to establish SIR-mediated repression (Kirchmaier and Rine, 2001; Li et al, 2001).

Ku and Sir4 function as perinuclear anchors independently of silencing function

The separation of tethering and silencing activities of the Ku and SIR complexes is manifest in the yku80-4 allele and the Sir4PAD domain, which efficiently relocalize a targeted locus to the NE, but which are unable to promote targeted silencing or bind other SIR proteins. Recent reports have described four mutant forms of yku80 bearing substitutions in a conserved region of the gene. Of these, two coincide fortuitously with our yku80-4 mutation (residue 437; Bertuch and Lundblad, 2003; Roy et al, 2004). Bertuch and Lundblad show that the yku80-4 mutant loses TPE and telomere protection activities, but not the nonhomologous end-joining function of Ku. In a related yku80 allele that bears P437Aand F438A substitutions, interaction with the N-terminus of Sir4 is impaired (Roy et al, 2004). These results support our argument that this domain contributes to silencing by mediating YKu80–Sir4 interactions.

Anchoring and silencing: the self-perpetuation of subnuclear compartmentation

An important conclusion of this study is that perinuclear chromatin anchoring can occur prior to or independently of transcriptional repression (see Figure 7E). Nonetheless, our results suggest a functional relationship between anchoring and silencing, because the two are promoted by the same proteins (Ku and Sir4). Moreover, silencing and anchoring, while separable, reinforce each other.

We propose that pools of SIR factors result from the initial anchoring of telomeres through Ku-NE interactions. Indeed, by grouping nonsilent telomeres at the NE, the TG repeat-bound Rap1 can directly recruit Sir4 to this zone. The resulting high density of Sir4-binding chromatin would in turn attract Sir2 and Sir3 through protein–protein interactions. The concentrated Sir factors should then reinforce repression of loci located within this compartment. As silent chromatin spreads, the number of Sir4 molecules bound will increase, thus reinforcing interaction with the NE. Consistent with this model, ChIP studies suggest that Sir4 is the first component bound, and it may be limiting for nucleating the assembly of the Sir2/3/4 complex (Bourns et al, 1998; Hoppe et al, 2002; Luo et al, 2002; Rusche et al, 2002).

The dual mechanism for localization described here (i.e. the ability of Ku to anchor in the absence of silencing and the binding of silent chromatin through Sir4) suggests a self-perpetuating mechanism for the formation telomere-associated SIR foci. We suggest that the clustering of repetitive sequence elements and spatial concentration of their ligands may be a general mechanism for the formation of nuclear subcompartments with biochemically distinct characters. In higher eucaryotes, similar mechanisms may facilitate the clustering of centromeric satellite and the propagation of heterochromatin, although in some cases the association of repetitive sequences in trans might eliminate the need for NE anchoring domains.

Materials and methods

Plasmid, strains and yeast methods

All strains used are indicated in Table II. PCR-based gene deletions (Longtine et al, 1998) created complete null alleles of yku70, esc1 and sir4 that were checked by PCR and phenotypic assays. The endogenous ESC1 gene was tagged C-terminally with either GFP or 13Myc epitope as described (Goldstein and McCusker, 1999).

Table 2
Yeast strains used in this study

Standard culture conditions at 30°C were used unless otherwise indicated (Rose et al, 1990). Silencing assays were performed as described (Gotta et al, 1998). All lacop-tagged strains are based on GA-1320 (Heun et al, 2001a). Integration of <10 kb of multimerized lac operators and four LexA operators 272 bp upstream ARS607 (Chr6int) into GA-1320 (Heun et al, 2001a) gave rise to GA-1461.

All LexA fusions were cloned into pAT4, a 2 μm LEU2 vector expressing LexA from the ADH promoter, which was obtained by fragment fusion from pBMT116 and pRS425. Fusion proteins were created by cloning appropriate gene fragments of Sir4PAD (aa 960–1262), Esc1C (aa 1124–1658), Yif1 and full-length Sir2 from existing plasmids: pLS4, pEDA114 and pGBC12–YIF1 (Ansari and Gartenberg, 1997; Andrulis et al, 1998, 2002; Cockell et al, 2000). The mutants yku80-4 and yku80-9 were produced by degenerate PCR and chosen for study because they do not complement TPE. Bidirectional sequencing identified three substitutions (L149V, N349D and E518A) in yku80-9, and one in yku80-4 (P437L), which by chance is identical to and named for one recently published (Bertuch and Lundblad, 2003).

Liquid culture two-hybrid analysis was performed as described (Aushubel et al, 1994). The lacZ-reporter pSH18-34, pEG-202-derived bait, and pJG4-6-derived prey plasmids were transformed into GA-180. Quantitative β-galactosidase assays were performed on permeabilized cells, using four independent transformants per sample (Adams et al, 1997) after 6 h on 2% galactose.

Immunological techniques

For IF assay, cells were fixed prior to spheroplasting, and stained with affinity-purified rabbit anti-Sir4 whose specificity has been previously characterized (Gotta et al, 1996), the anti-pore MAb414 monoclonal (CRP Inc.) or the anti-Myc MAb 9E10.

For ImmunoEM, cells were permeabilized for 1 min with 0.1% Triton X-100 in phosphate buffer after fixation and spheroplasting (Gotta et al, 1996), and incubated sequentially with MAb 9E10 and IgG FAb fluoronanogold antibody coupled to Alexa 488 (Nanonprobes, Inc., Yaphank, NY). After washing, part of the cells was analyzed by confocal microscopy (Leica SP2 AOBS; Leica Microsystems, Mannheim, Germany), and part fixed with 2% glutaraldehyde for 20 min. Gold enhancement was performed with Goldenhance-EM (Nanonprobes, Inc., Yaphank, NY) followed by fixation with OsO4 and uranyl acetate. Epon embedding, ultrathin sectioning and EM images capture (Philips TEM 410, Eindhoven, Netherlands) were as described (Bauer et al, 2001).

Live fluorescence microscopy and quantitation

Cultures were grown in SD-His, to <1 × 107 cells/ml. Live microscopy was performed between 22 and 25°C as described (Heun et al, 2001a). A Tillvision driven Olympus IX70 microscope was used to capture 19-image stacks of 170 nm step size. Cell-cycle stage classification was as follows: unbudded cells=G1 phase; budded cells with round nucleus not at the bud neck=S phase; large bud with bud-neck nucleus=G2 phase. In the focal plane in which the GFP spot is brightest, its position is mapped to one of three concentric zones of equal surface by dividing the spot-to-pore distance by the nuclear diameter (the width of the three equal surface zones is also normalized to the diameter; Hediger et al, 2002a). We eliminate cells in which the GFP spot falls into the top or bottom three focal planes because of the error arising from the small zone width and background signal from the pore. Statistical analysis used a proportion method to compare zone I percentages among different samples or with the predicted random distribution. Significance was determined with a 95% confidence interval. Time-lapse imaging was performed using a Zeiss LSM510 confocal microscope as described (Heun et al, 2001b). Deconvolution was performed using Metamorph Software with the following parameters: fast algorithm with five iterations, sigma 0.7 and a frequency of 4.

Chromatin immunoprecipitation

ChIP was performed with or without a 15 min fixation with formaldehyde on GA-2141 as described by Cobb et al (2003) using Mab against Myc (9E10) or HA epitopes (12C5A), and real-time PCR (Perkin-Elmer ABI Prism 7700 Sequence Detector System and software). Real-time PCR monitors the ‘threshold cycle (CT)' at which the exponential curve of the accumulated product passes a threshold. Bar heights represent the ratio of the specific signal divided by that of an internal locus (SMC2), after normalization to the ratio obtained in the input fraction, which monitors relative probe efficiency. Fold increase = 2(CTIPSMC2-CTIPtelo)-(CTinputSMC2-CTinputtelo).

Acknowledgments

We thank Thierry Laroche for assistance with microscopy, Dr Marek Blaszczyk for statistical and image analysis advice, Drs R Sternglanz, D Shore and M Gartenberg for reagents, and the Gasser laboratory and M Gartenberg for editing and discussions. Our research is supported by the Swiss National Science Foundation and the NCCR program ‘Frontiers in Genetics'. AT is supported by an EMBO fellowship.

References

  • Adams A, Gottschling DE, Kaiser CA, Stearns T (1997) Methods in Yeast Genetics. Cold Spring Harbor, NY: CSHL Press
  • Andrulis ED, Neiman AM, Zappulla DC, Sternglanz R (1998) Perinuclear localization of chromatin facilitates transcriptional silencing. Nature 394: 592–595 [PubMed]
  • Andrulis ED, Zappulla DC, Ansari A, Perrod S, Laiosa CV, Gartenberg MR, Sternglanz R (2002) Esc1, a nuclear periphery protein required for Sir4-based plasmid anchoring and partitioning. Mol Cell Biol 22: 8292–8301 [PMC free article] [PubMed]
  • Ansari A, Gartenberg MR (1997) The yeast silent information regulator Sir4p anchors and partitions plasmids. Mol Cell Biol 17: 7061–7068 [PMC free article] [PubMed]
  • Aushubel FM, Brent R, Kinston R, Moore D, Seidman JJ, Smith J, Struhl K (1994) Current Protocols in Molecular Biology. New York: John Wiley and Sons
  • Bauer C, Herzog V, Bauer MF (2001) Improved Technique for electron microscope visualization of yeast membrane structure. Microsc Microanal 7: 530–534 [PubMed]
  • Bertuch AA, Lundblad V (2003) The ku heterodimer performs separable activities at double-strand breaks and chromosome termini. Mol Cell Biol 22: 8202–8215 [PMC free article] [PubMed]
  • Bourns BD, Alexander MK, Smith AM, Zakian VA (1998) Sir proteins, Rif proteins, and Cdc13p bind S. telomeres in vivo. Mol Cell Biol 18: 5600–5608 [PMC free article] [PubMed]
  • Buck SW, Shore D (1995) Action of a RAP1 carboxy-terminal silencing domain reveals an underlying competition between HMR and telomeres in yeast. Genes Dev 9: 370–384 [PubMed]
  • Cobb JA, Bjergbaek L, Shimada K, Frei C, Gasser SM (2003) DNA polymerase stabilization at stalled replication forks requires Mec1 and the RecQ helicase Sgs1. EMBO J 22: 4325–4336 [PMC free article] [PubMed]
  • Cockell M, Palladino F, Laroche T, Kyrion G, Liu C, Lustig AJ, Gasser SM (1995) The carboxy termini of Sir4 and Rap1 affect Sir3 localization: evidence for a multicomponent complex required for yeast telomeric silencing. J Cell Biol 129: 909–924 [PMC free article] [PubMed]
  • Cockell M, Perrod S, Gasser SM (2000) Analysis of Sir2p domains required for rDNA and telomeric silencing. Genetics 154: 1069–1083 [PMC free article] [PubMed]
  • Diffley JF, Stillman B (1989) Transcriptional silencing and lamins. Nature 342: 24. [PubMed]
  • Doye V, Wepf R, Hurt EC (1994) A novel nuclear pore protein Nup133p with distinct roles in poly(A)+ RNA transport and nuclear pore distribution. EMBO J 13: 6062–6075 [PMC free article] [PubMed]
  • Fisher AG, Merkenschlager M (2002) Gene silencing, cell fate and nuclear organisation. Curr Opin Genet Dev 12: 193–197 [PubMed]
  • Freitas-Junior LH, Bottius E, Pirrit LA, Deitsch KW, Scheidig C, Guinet F, Nehrbass U, Wellems TE, Scherf A (2000) Frequent ectopic recombination of virulence factor genes in telomeric chromosome clusters of P. falciparum. Nature 407: 1018–1022 [PubMed]
  • Goldstein AL, McCusker JH (1999) Three new dominant drug resistance cassettes for gene disruption in S. cerevisiae. Yeast 15: 1541–1553 [PubMed]
  • Gotta M, Laroche T, Formenton A, Maillet L, Scherthan H, Gasser SM (1996) The clustering of telomeres and colocalization with Rap1, Sir3, and Sir4 proteins in wild-type S. cerevisiae. J Cell Biol 134: 1349–1363 [PMC free article] [PubMed]
  • Gotta M, Palladino F, Gasser SM (1998) Functional characterization of the N terminus of Sir3p. Mol Cell Biol 18: 6110–6120 [PMC free article] [PubMed]
  • Hecht A, Strahl-Bolsinger S, Grunstein M (1996) Spreading of transcriptional repressor SIR3 from telomeric heterochromatin. Nature 383: 92–96 [PubMed]
  • Hediger F, Dubrana K, Gasser SM (2002b) Myosin-like proteins 1 and 2 are not required for silencing or telomere anchoring but act in the Tel1 pathway of telomere length control. J Struct Biol 140: 79–91 [PubMed]
  • Hediger F, Neumann FR, Van Houwe G, Dubrana K, Gasser SM (2002a) Live imaging of telomeres. yKu and Sir proteins define redundant telomere-anchoring pathways in yeast. Curr Biol 12: 2076–2089 [PubMed]
  • Heun P, Laroche T, Raghuraman MK, Gasser SM (2001a) The positioning and dynamics of origins of replication in the budding yeast nucleus. J Cell Biol 152: 385–400 [PMC free article] [PubMed]
  • Heun P, Laroche T, Shimada K, Furrer P, Gasser SM (2001b) Chromosome dynamics in the yeast interphase nucleus. Science 294: 2181–2186 [PubMed]
  • Holaska JM, Wilson KL, Mansharamani M (2002) The nuclear envelope, lamins and nuclear assembly. Curr Opin Cell Biol 14: 357–364 [PubMed]
  • Hoppe GJ, Tanny JC, Rudner AD, Gerber SA, Danaie S, Gygi SP, Moazed D (2002) Steps in assembly of silent chromatin in yeast: Sir3-independent binding of a Sir2/Sir4 complex to silencers and role for Sir2-dependent deacetylation. Mol Cell Biol 12: 4167–4180 [PMC free article] [PubMed]
  • Huang Y (2002) Transcriptional silencing in S. cerevisiae and S. pombe. Nucleic Acids Res 30: 1465–1482 [PMC free article] [PubMed]
  • Kirchmaier AL, Rine J (2001) DNA replication-independent silencing in S. cerevisiae. Science 291: 646–650 [PubMed]
  • Laroche T, Martin SG, Gotta M, Gorham HC, Pryde FE, Louis EJ, Gasser SM (1998) Mutation of yeast Ku genes disrupts the subnuclear organization of telomeres. Curr Biol 8: 653–656 [PubMed]
  • Li YC, Cheng TH, Gartenberg MR (2001) Establishment of transcriptional silencing in the absence of DNA replication. Science 291: 650–653 [PubMed]
  • Longtine MS, McKenzie A III, Demarini DJ, Shah NG, Wach A, Brachat A, Philippsen P, Pringle JR (1998) Additional modules for versatile and economical PCR-based gene deletion and modification in S. cerevisiae. Yeast 14: 953–961 [PubMed]
  • Luo K, Vega-Palas MA, Grunstein M (2002) Rap1–Sir4 binding independent of other Sir, yKu, or histone interactions initiates the assembly of telomeric heterochromatin in yeast. Genes Dev 12: 1528–1539 [PMC free article] [PubMed]
  • Maillet L, Boscheron C, Gotta M, Marcand S, Gilson E, Gasser SM (1996) Evidence for silencing compartments within the yeast nucleus: a role for telomere proximity and Sir protein concentration in silencer-mediated repression. Genes Dev 10: 1796–1811 [PubMed]
  • Maillet L, Gaden F, Brevet V, Fourel G, Martin SG, Dubrana K, Gasser SM, Gilson E (2001) Ku deficient strains exhibit alternative states of silencing competence. EMBO Rep 2: 203–210 [PMC free article] [PubMed]
  • Marcand S, Buck SW, Moretti P, Gilson E, Shore D (1996) Silencing of genes at nontelomeric sites in yeast is controlled by sequestration of silencing factors at telomeres by Rap 1 protein. Genes Dev 10: 1297–1309 [PubMed]
  • Marshall WF (2002) Order and disorder in the nucleus. Curr Biol 12: R185–R192 [PubMed]
  • Martin SG, Laroche T, Suka N, Grunstein M, Gasser SM (1999) Relocalization of telomeric Ku and SIR proteins in response to DNA strand breaks in yeast. Cell 97: 621–633 [PubMed]
  • Moretti P, Freeman K, Coodly L, Shore D (1994) Evidence that a complex of SIR proteins interacts with the silencer and telomere-binding protein RAP1. Genes Dev 8: 2257–2269 [PubMed]
  • Perrod S, Gasser SM (2003) Long-range silencing and position effects at telomeres and centromeres: parallels and differences. Cell Mol Life Sci 60: 1–16 [PubMed]
  • Renauld H, Aparicio OM, Zierath PD, Billington BL, Chhablani SK, Gottschling DE (1993) Silent domains are assembled continuously from the telomere and are defined by promoter distance and strength, and by SIR3 dosage. Genes Dev 7: 1133–1145 [PubMed]
  • Rose MD, Winston F, Hieter P (1990) Methods in Yeast Genetics. Cold Spring Harbor, NY: CSHL Press
  • Roy R, Meier B, McAinsh AD, Feldmann HM, Jackson SP (2004) Separation-of-function mutants of yeast Ku80 reveal a Yku80p–Sir4p interaction involved in telomeric silencing. J Biol Chem 279: 86–94 [PubMed]
  • Rusche LN, Kirchmaier AL, Rine J (2002) Ordered nucleation and spreading of silenced chromatin in S. cerevisiae. Mol Biol Cell 7: 2207–2222 [PMC free article] [PubMed]
  • Rusche LN, Kirchmaier AL, Rine J (2003) The establishment, inheritance, and function of silenced chromatin in S. cerevisiae. Annu Rev Biochem 72: 481–516 [PubMed]
  • Smith CD, Smith DL, DeRisi JL, Blackburn EH (2003) Telomeric protein distributions and remodeling through the cell cycle in S. cerevisiae. Mol Biol Cell 14: 556–570 [PMC free article] [PubMed]
  • Stellwagen AE, Haimberger ZW, Veatch JR, Gottschling DE (2003) Ku interacts with telomerase RNA to promote telomere addition at native and broken chromosome ends. Genes Dev 17: 2384–2395 [PMC free article] [PubMed]
  • Tham WH, Wyithe JS, Ferrigno PK, Silver PA, Zakian VA (2001) Localization of yeast telomeres to the nuclear periphery is separable from transcriptional repression and telomere stability functions. Mol Cell 8: 189–199 [PubMed]
  • Thompson JS, Johnson LM, Grunstein M (1994) Specific repression of the yeast silent mating locus HMR by an adjacent telomere. Mol Cell Biol 14: 446–455 [PMC free article] [PubMed]
  • Tsukamoto Y, Kato J, Ikeda H (1997) Silencing factors participate in DNA repair and recombination in S. cerevisiae. Nature 388: 900–903 [PubMed]

Articles from The EMBO Journal are provided here courtesy of The European Molecular Biology Organization
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...