Role for perinuclear chromosome tethering in maintenance of genome stability.

Repetitive DNA sequences, which constitute half the genome in some organisms, often undergo homologous recombination. This can instigate genomic instability resulting from a gain or loss of DNA. Assembly of DNA into silent chromatin is generally thought to serve as a mechanism ensuring repeat stability by limiting access to the recombination machinery. Consistent with this notion is the observation, in the budding yeast Saccharomyces cerevisiae, that stability of the highly repetitive ribosomal DNA (rDNA) sequences requires a Sir2-containing chromatin silencing complex that also inhibits transcription from foreign promoters and transposons inserted within the repeats by a process called rDNA silencing. Here we describe a protein network that stabilizes rDNA repeats of budding yeast by means of interactions between rDNA-associated silencing proteins and two proteins of the inner nuclear membrane (INM). Deletion of either the INM or silencing proteins reduces perinuclear rDNA positioning, disrupts the nucleolus-nucleoplasm boundary, induces the formation of recombination foci, and destabilizes the repeats. In addition, artificial targeting of rDNA repeats to the INM suppresses the instability observed in cells lacking an rDNA-associated silencing protein that is typically required for peripheral tethering of the repeats. Moreover, in contrast to Sir2 and its associated nucleolar factors, the INM proteins are not required for rDNA silencing, indicating that Sir2-dependent silencing is not sufficient to inhibit recombination within the rDNA locus. These findings demonstrate a role for INM proteins in the perinuclear localization of chromosomes and show that tethering to the nuclear periphery is required for the stability of rDNA repeats. The INM proteins studied here are conserved and have been implicated in chromosome organization in metazoans. Our results therefore reveal an ancient mechanism in which interactions between INM proteins and chromosomal proteins ensure genome stability.

Eukaryotic rDNA is tandemly repeated anywhere from ∼100 to over 10,000 times8. rDNA repeats provide the foundation for at least one ribosome-manufacturing compartment, the nucleolus. Budding yeast Saccharomyces cerevisiae has 100-200 rDNA units tandemly arranged on chromosome XII (Chr. XII) and forming one nucleolus (Fig. 1a, b)8. In addition to harboring rRNA-coding DNA sequences, each unit contains intergenic spacers (IGS1 and 2) that promote repeat integrity (Fig. 1a)9-11. Recruitment of nucleolar protein complexes RENT (regulator of nucleolar silencing and telophase exit; composed of Cdc14, Net1/Cfi1, and Sir2) and Cohibin (mitotic monopolin proteins Lrs4 and Csm1) to IGS1 suppresses unequal recombination at the repeats3,12-16. This suppression is seemingly linked to the ability of these complexes to induce rDNA silencing, which involves chromatin changes preventing RNA Polymerase II-driven transcription within IGSs of rDNA4,5,16-19.

Figure 1

Protein network extending from rDNA to the nuclear envelope

a, rDNA repeats on Chr. XII. Each unit yields a Pol I-transcribed 35S precursor rRNA (processed into 25S, 18S, and 5.8S) and a Pol III-transcribed 5S rRNA. CEN, centromere; TEL, telomere; IGS, intergenic spacer; RE, recombination enhancer;, replication fork block; TIR, Pol I transcription initiation region; O, DNA replication origin; Mbp, megabase pairs. Vertical arrowheads indicate insertion sites of mURA3 reporter genes used in this study. b, Nuclear organization at the G2/M cell cycle stage. No1., nucleolus; ONM, outer nuclear membrane. c-f, Purification of native complexes. c, e, Protein detection in silver-stained gels. TAP cleavage during protein purification leaves a calmodulin binding protein (CBP) fragment. d, f, LC-MS/MS analysis. Number of unique peptides followed by percent coverage of the protein sequence is shown. SI contains full protein lists (Supplementary Table 1, part A) and spectral counts (Supplementary Tables 2 and 3).

Purification of Cohibin suggested an association with INM proteins of unknown function16. To gain insight into the possible role of this association in nucleolar organization, we purified native Cohibin and INM proteins using tandem affinity purification (TAP). The TAP-tagged proteins are functional in vivo (Ref.16 and below). We detected purified complexes by silver staining and total protein mixtures were analyzed by liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). Negative controls were untagged cells. Purification of Lrs4 and Csm1 yielded peptides of INM proteins Heh1 (helix extension helix 1, also called Src1) and Nur1 (nuclear rim 1, Ydl089w) (Fig. 1c, d; Supplementary Table 1, part A)16. Heh1, the orthologue of human Man1, is a member of a family of INM proteins containing a highly conserved LAP-Emerin-Man1 domain (LEM, also called HEH; Supplementary Fig. 2)20-22. LEM-domain proteins are linked to multiple clinical conditions via emerging roles in fundamental cellular processes, including gene expression and chromatin organization6,7,20,21,23,24. Little is known about Heh1 and Nur120, which we define here as chromosome linkage INM proteins (CLIP). Purification of either INM protein yielded peptides for both Heh1 and Nur1 (Fig. 1e, f; Supplementary Text, section A). Purification of Heh2, an Heh1 homologue (Supplementary Fig. 2)20, did not yield peptides for CLIP or Cohibin proteins (Fig. 1e, f; Supplementary Fig. 3c; Supplementary Table 1). Moreover, TAP-tagged Heh1, Lrs4, and Csm1 coimmunoprecipitated with Myc13-tagged Lrs4, Heh1, and Nur1, respectively (Supplementary Fig. 3d). Migration of Heh1 to 115 kDa, instead of the predicted 95 kDa, led us to identify multiple post-translational modifications of the protein and fluctuation of Heh1 levels over the cell cycle with peaks at interphase and mitosis (Supplementary Figs 3a, e, f, and 4). These findings physically link rDNA-associated complexes to INM proteins.

Peripheral association of genes is linked to silent chromatin assembly, which seemingly stabilizes repeats by limiting access to recombination proteins2,7. Thus, CLIP may assemble at IGS1 to cooperate with RENT and Cohibin to silence transcription and inhibit unequal rDNA recombination. Therefore, we monitored unequal sister chromatid exchange (USCE) by measuring the rate of loss of an ADE2 marker gene from rDNA repeats. Deletion of Sir2, Lrs4, or Csm1 increased USCE, as expected (Fig. 2a, b; Supplementary Table 4)16,25. USCE also increased following deletion of Heh1 or Nur1, but not Heh2 (Fig. 2a, b; Supplementary Table 4). heh1Δ nur1Δ cells displayed additive USCE defects compared to single mutants, suggesting that INM proteins play partially overlapping roles at rDNA. Moreover, deletion of Heh1, Lrs4, or Csm1 exacerbated the effect of loosing Sir2 (Fig. 2a; Supplementary Table 4)16 suggesting that Sir2 stabilizes rDNA via CLIP/Cohibin-dependent and -independent processes. Since increases in USCE affect rDNA copy-number on Chr. XII, we analyzed its size using contour-clamped homogeneous electric field (CHEF). Chr. XII measured ∼2.83 Mbp in wild-type cells (∼190 rDNA units) and chromosome smearing in sir2Δ cells was indicative of severe changes in rDNA copy-number (Fig. 2c; Supplementary Fig. 5a), as expected4,17. Interestingly, deletion of Lrs4, Csm1, Heh1, or Nur1 resulted in marked changes in rDNA copy-number averages and chromosome smearing patterns (Fig. 2c; described in Supplementary Text, section B). Together, these data suggest that the perinuclear protein network studied here is required for rDNA repeat stability.

Figure 2

Role of perinuclear protein network at rDNA repeats

a, b, Rates of ADE2 marker loss (mean ± s.d.) relative to wild-type (WT) (a) and representative colonies (b) are shown. c, CHEF analysis of rDNA stability. Chromosomes resolved by CHEF were probed with IGS1 rDNA (Chr. XII, top). Ethidium bromide (EtBr) staining of Chr. IV and smaller chromosomes shows quality of the preparation (bottom). Values corresponding to rDNA copy-number averages and smear edges are indicated. Chr. XII size for sir2Δ was too heterogeneous to estimate copy-number. d, Unlike Sir2 and Cohibin, CLIP is dispensable for rDNA silencing. Ten-fold serial dilutions of cells with the mURA3 reporter gene inserted at IGS1/IGS2 (see Fig. 1a for locations) or outside rDNA at the LEU2 locus are shown.

We next studied the ability of cells to silence an RNA Pol II-transcribed mURA3 reporter gene positioned within IGS1 or IGS2 by assessing cellular growth on synthetic complete (SC) medium that is either lacking uracil (-Ura, silencing disrupts growth) or supplemented with 5-fluoro-orotic acid (+5FOA, silencing allows growth). Deletion of Sir2, Lrs4, or Csm1 disrupted IGS1 silencing (Fig. 2d), as expected16. Surprisingly, in contrast to rDNA repeat stability, Heh1 and Nur1 were dispensable for silencing (Fig. 2d; Supplementary Fig. 5b), suggesting that silencing is insufficient for proper repeat size regulation.

We next asked whether tethering rDNA repeats to INM proteins via Cohibin limited recombination independently of silencing. Using immunofluorescence, we visualized the functional green fluorescent protein (GFP)-tagged Net1 and Myc13-tagged Heh1 (Supplementary Table 4)3,16. Net1 associates with rDNA in the nucleolus throughout the cell cycle and recruits Sir2 to IGS13. However, enrichment of Sir2 at rDNA in chromatin immunoprecipitation (ChIP) experiments is unaffected by deletion of Cohibin (Supplementary Figs 1 and 8c). To measure the limit of Net1-GFP internalization, the nucleus, delineated by peripheral Heh1-Myc13 signal, was divided in three concentric zones of equal area, zone I being most peripheral26. Cells were categorized according to whether the centre of the least peripheral Net1-GFP focus localizes to zone I, II, or III. Most wild-type cells displayed peripheral Net1-GFP localization (zone I, 74%) while few contained central Net1-GFP staining (zone III, 2%) (Supplementary Fig. 6a). Lrs4 or Csm1 deletion drastically shifted Net1-GFP to zone II or III (Supplementary Fig. 6a) and expanded the volume occupied by Net1-GFP within nuclear space in 3D (Fig. 3a), suggesting that optimal perinuclear localization of the rDNA-associated Net1 requires Cohibin.

Figure 3

Protein network tethers rDNA to the nuclear envelope

a, Representative 3D reconstructions of cells from double immunofluorescence analysis reveal the relative organization of Net1-GFP (green) and Heh1-Myc13 (red). Signal boundaries are shown with clipping planes for Heh1-Myc13. Percentages of green relative to full red volumes are shown (mean ± s.d.; n = 5). b, Nocodazole-arrested (G2/M) cells were subjected to FISH to visualize rDNA and DAPI staining (pseudocoloured red) to visualize bulk nuclear DNA. Cells with different rDNA morphologies are indicated by double arrowheads (spooled lines), arrows (amorphous), or triangles (two bodies) with quantifications in Supplementary Fig. 6b. Scale bar, 3 μm. c, Live cells with TetI-RFP-marked rDNA and expressing Rad52-YFP and nucleolar Nop1-CFP were imaged. Images depicting most observed phenotypes are shown. More images and quantifications are in Supplementary Fig. 6d-f. Scale bars are 1 μm (white) and 3 μm (black). d, e, Relative fold enrichment of indicated TAP-tagged proteins are shown. rDNA organization schematics are shown on graphs. Gels for d and e are shown in Supplementary Figs 7a and 7d, respectively. Detailed IGS1 ChIP is shown in Supplementary Fig. 7b, c.

How rDNA is separated from the bulk of nuclear DNA is unknown (Fig. 1b). We tested if the perinuclear network studied here affects this subnuclear separation. Nocodazole-arrested cells were analyzed by fluorescence in-situ hybridization (FISH) to visualize rDNA and 4',6-diamidino-2-phenylindole dihydrochloride (DAPI) staining to visualize bulk nuclear DNA. Wild-type cells exhibited line-shaped rDNA spooling away from the DNA bulk towards the nuclear periphery (Fig. 3b and quantification in Supplementary Fig. 6b)27. Deletion of Lrs4, Csm1, or Heh1, but not Heh2, caused rDNA to adopt amorphous distributions often overlapping DAPI signal and a small percentage of cells exhibited two separable rDNA bodies (Fig. 3b; Supplementary Fig. 6b), which may reflect severe loss of interactions between rDNA repeats on Chr. XII sister chromatids. Nur1 deletion caused smaller changes in rDNA morphology, which appeared less condensed (Fig. 3b; 65 ± 6% of cells, mean ± s.d.). Disorganization of rDNA was also observed in asynchronous cells (Supplementary Fig. 6c).

We next studied the localization of a specific site within rDNA repeats in live cells harboring a tetO array at rDNA repeats and expressing tetO-binding TetI-RFP (TetI fused to red fluorescent protein) and the nucleolar Nop1-CFP (Nop1 protein fused to cyan fluorescent protein)28. TetI-RFP localized inside or at the periphery of the nucleolus in most wild-type cells (93%; Fig. 3c; Supplementary Fig. 6d, e). Deletion of Lrs4 or Heh1 shifted TetI-RFP outside of the nucleolus or to its periphery (Fig. 3c; Supplementary Fig. 6d, e). sir2Δ cells exhibited less severe mislocalization of TetI-RFP (Fig. 3c; Supplementary Fig. 6d, e). Deletion of Lrs4, Heh1, or Sir2 also induced the formation of extranucleolar DNA repair centers, as marked by clustering of the yellow fluorescent protein-tagged Rad52 recombination protein, Rad52-YFP (Fig. 3c; Supplementary Fig. 6d, f). Fewer sir2Δ cells exhibited Rad52 foci compared to heh1Δ or lrs4Δ cells (Supplementary Fig. 6f). This is in contrast to USCE in sir2Δ cells, which is higher than that of heh1Δ or lrs4Δ cells (Fig. 2a), suggesting that more recombinations in sir2Δ cells are unequal crossovers. Alternatively, a higher incidence of Rad52-YFP foci in cells lacking Heh1 or Lrs4 might suggest that these proteins stabilize several genetic loci. Most Rad52-YFP foci (61%-65%) did not overlap TetI-RFP signal in lrs4Δ, heh1Δ, or sir2Δ cells (Fig. 3c; Supplementary Fig. 6d, f), likely reflecting the occurrence of one or few repair events per rDNA array and their distance from tetO sequences. While we cannot exclude the possibility that the tetO array contributes to rDNA mislocalization in lrs4Δ, heh1Δ, or sir2Δ cells, disruption of rDNA organization in lrs4Δ and heh1Δ cells lacking tetO sites, as revealed by FISH and immunofluorescence (Fig. 3a, b; Supplementary Fig. 6a-c), argues against this possibility and suggests that the perinuclear complexes studied here help stabilize wild-type rDNA repeats. Together, these results suggest that Heh1 and Lrs4, and to a lesser extent Sir2, are required for sequestration of rDNA in the peripherally located nucleolus, and show that loss of sequestration correlates with increased repeat instability (Fig. 2) and Rad52 recombination foci.

To further analyze CLIP-Cohibin links, we performed ChIP using a combination of dimethyl adipimidate and formaldehyde crosslinkers. We observed 2.85±0.37 and 2.35±0.12 fold enrichments for IGS1 sequences in Lrs4-TAP and Heh1-TAP immunoprecipitations, respectively (Fig. 3d, e; Supplementary Fig. 7). We did not detect an enrichment using Nur1-TAP, likely due to its low abundance or weaker association with Cohibin (Fig. 3d; Supplementary Figs 3a and 7a-c). More importantly, deletion of Lrs4 abolished the IGS1 enrichment of Heh1-TAP without affecting its levels (Fig. 3e; Supplementary Fig. 7d-f). In contrast, no enrichment was detected for Heh2-TAP (Fig. 3e; Supplementary Fig. 7d), an INM protein that neither interacts with Cohibin (Fig. 1) nor affects rDNA stability (Fig. 2), although expressed to similar levels as Heh1 (data not shown)29. Together, these data indicate that CLIP/Cohibin-mediated tethering of rDNA repeats to the INM is required for repeat stability.

To determine if perinuclear tethering suppresses recombination in the absence of Cohibin proteins, which are required for rDNA silencing and suppression of recombination, we created a strain in which rDNA was linked to Heh1 via Sir2. We fused HEH1 and SIR2 genes in lrs4Δ cells creating a hybrid HEH1-SIR2 gene (Fig. 4a). This yielded a fusion protein of expected size (∼175 kDa) detectable in anti-Sir2 immunoblotting (Supplementary Fig. 8a). Fusion of Heh1 and Sir2 restored the separation of rDNA from bulk nuclear DNA in lrs4Δ cells (Fig. 4b), reduced unequal recombination (Fig. 4c; Supplementary Table 4), and increased homogeneity in the size of Chr. XII in cell populations (Fig. 4d; Supplementary Fig. 8b). Furthermore, ChIP revealed that Heh1-Sir2 associated with rDNA to similar levels as Sir2 (Supplementary Fig. 8c). Moreover, Heh1-Sir2 did not rescue IGS1-specific increases in histone H3 acetylation, a marker for loss of silencing, caused by Lrs4 deletion (Supplementary Fig. 8c). The inability of Heh1-Sir2 to fully restore rDNA stability may be due to other Lrs4 functions, such as silencing or perhaps chromosome condensation, which suppress recombination at repeats. Attempts to fuse Heh1 with other perinuclear proteins, such as Tof2 or Ku70, did not yield viable cells (data not shown). Thus, tethering rDNA to the INM can promote repeat stability at least partially independently of silencing.

Figure 4

Targeted perinuclear tethering promotes rDNA repeat stability

a, Fusion of Heh1 to Sir2 in lrs4Δ cells. Heh1 is shown embedded in the INM. b, FISH reveals that fusion of Heh1 and Sir2 restores the separation of rDNA signal from DAPI staining in nocodazole-arrested cells lacking Lrs4. Scale bar, 3 μm. c, Relative rates of ADE2 marker loss (mean ± s.d.). Asterisk, P < 0.0001 for Student's t test. d, CHEF analysis of rDNA stability. Chromosomes resolved by CHEF were probed with IGS1 rDNA (Chr. XII, top). EtBr staining of Chr. IV and smaller chromosomes shows quality of the preparation (bottom). e, Functional organization of the perinuclear molecular network tethering rDNA to the nuclear periphery. Repeat instability results from the loss of either the INM CLIP proteins or rDNA silencing complexes.

Our results suggest that Sir2-dependent silencing alone cannot inhibit recombination within the repetitive rDNA locus and that INM-mediated perinuclear chromosome tethering ensures repeat stability (Fig. 4e; Supplementary Fig. 1). Extranucleolar Rad52 focus formation in lrs4Δ, heh1Δ, or sir2Δ cells concurs with suggestions that while early rDNA recombination steps occur inside the nucleolus, Rad52 sumoylation and a high local concentration of the Smc5-Smc6 complex preclude Rad52 focus formation within nucleolar space28. Thus, our findings suggest that rDNA repeats unleashed from the INM accumulate lesions that can better access the nucleoplasm where high concentrations of functional Rad52 promote DNA repair by homologous recombination. Therefore, perinuclear tethering likely sequesters repeats away from recombination factors and may be required for Cohibin and RENT to stably align rDNA sister chromatids during replication to prevent unequal crossovers (Fig. 4e). Recombination between homologous repeats dispersed in the genome often instigates catastrophic chromosomal rearrangements. We anticipate that proteins studied here are members of perinuclear networks that control recombination at multiple loci to maintain genome stability.

Methods Summary

Standard co-immunoprecipitations3,14, ChIP14, TAP purification16, immunofluorescence16, rDNA silencing16, USCE16, FISH27, and live cell28 assays were performed as previously described. Modified ChIP (employing both formaldehyde and dimethyl adipimidate crosslinkers) and modified TAP purifications (employing different detergents) are described in Methods. Information about strain construction, materials, LC-MS/MS, CHEF, microscopy/imaging, whole-cell protein analysis, and cell-cycle arrest is available in Methods.