Superresolution microscopy reveals linkages between ribosomal DNA on heterologous chromosomes

Abstract

The spatial organization of the genome is enigmatic. Direct evidence of physical contacts between chromosomes and their visualization at nanoscale resolution has been limited. We used superresolution microscopy to demonstrate that ribosomal DNA (rDNA) can form linkages between chromosomes. We observed rDNA linkages in many different human cell types and demonstrated their resolution in anaphase. rDNA linkages are coated by the transcription factor UBF and their formation depends on UBF, indicating that they regularly occur between transcriptionally active loci. Overexpression of c-Myc increases rDNA transcription and the frequency of rDNA linkages, further suggesting that their formation depends on active transcription. Linkages persist in the absence of cohesion, but inhibition of topoisomerase II prevents their resolution in anaphase. We propose that linkages are topological intertwines occurring between transcriptionally active rDNA loci spatially colocated in the same nucleolar compartment. Our findings suggest that active DNA loci engage in physical interchromosomal connections that are an integral and pervasive feature of genome organization.

Figure 1.

SIM revealed rDNA linkages between acrocentric human chromosomes. (A) Normal human karyotype with highlighted acrocentric chromosomes 13, 14, 15, 21, and 22 bearing rDNA loci on their short arms, provided courtesy of Karen Miga (Genomics Institute, University of California Santa Cruz, Santa Cruz, CA) and Amalia Dutra (Cytogenetic and Microscopy Core, National Human Genome Research Institute, National Institutes of Health, Bethesda, MD). (B) Schematic representation of rDNA repeat units and coverage of rDNA probes used in this study. In the human karyotype, rDNA genes are arranged as repeats on the short arms of the acrocentric chromosomes between centromeres and telomeres, flanked by proximal and distal junctions (PJ and DJ). Each unit consists of a coding region (encoding pre-mRNA for 18S, 5.8S, and 28S ribosomal RNA subunits) and intergenic spacer. Boundaries of the coding region contain external transcribed spacers (5′ETS and 3′ETS), and coding parts of the 45S sequence are separated by internal transcribed spacers (ITS1 and ITS2). The human rDNA probe used in this study was derived from BAC clone RP11-450E20 and spans the intergenic spacer and the transcription initiation site of the next repeat. The mouse rDNA probe was derived from BAC clone RP23-225M6, spanning the end of the coding part and the intergenic spacer. (C) Wide-field illumination (WF) and SIM images of mitotic chromosome spread from 184FMY2 HMEC cell labeled by FISH with rDNA probe (green) and CenB probe (red). Arrows 1–3 point to acrocentric chromosomal rDNA associations. Panels on the right show individual acrocentric chromosomes: six large rDNA chromosomes and four small rDNA chromosomes. Bar, 10 µm. (D) Quantification of the number of interchromosomal rDNA linkages in chromosomal spreads from isogenic HMEC cell lines labeled by FISH with rDNA probe. Cell lines overexpressing c-Myc highlighted in blue. Images of ≥10 chromosomal spreads from each cell line were examined. The difference between 184DTERT and each of the other cell lines was evaluated using the Mann–Whitney U test. *, P < 0.05; **, P < 0.001; ns, not significant. (E) Wide-field illumination and SIM images of the interphase nucleus of 184FMY2 HMEC cell labeled by FISH with rDNA probe (green) and centromere CenB probe (red). While centromere loci form compact dots, most of the rDNA forms long thin filaments within the nucleolar compartment. Bar, 10 µm. (F) Panels 1–3 show corresponding enlarged wide-field illumination and SIM images of rDNA associations marked by arrows in C. While wide-field illumination images accurately depict rDNA associations, SIM images reveal the network of thin filamentous rDNA linkages between different acrocentric chromosomes. Bar, 1 µm.

Figure 2.

Frequency of rDNA linkages and rRNA synthesis in HFF-1 and derivative iPS cells. (A) Confocal images of mitotic chromosome spread from HFF1 (left) and iPSC (right) cell labeled by FISH with rDNA probe (green) and CenB probe (red). Bar, 10 µm. Magnified insets depict acrocentric chromosomal associations between rDNA (bar, 1 µm). (B) Quantification of the number of interchromosomal rDNA linkages in chromosomal spreads from HFF-1 and iPSC labeled by FISH with rDNA probe. High-resolution confocal images of 20 chromosomal spreads from each cell line were examined. The Mann–Whitney U test was used to compare iPSC with parental HFF-1. *, P < 0.05. (C) Real-time qPCR analysis of iPSC compared with the parental HFF-1 cell line. The expression of pre-rRNA was normalized to the expression of GAPDH mRNA. Bar heights represent an average fold change of three primer sets to the 5′ETS; error bars represent SD. Statistical significance was evaluated using t test; **, P < 0.001.

Figure 3.

Overexpression of c-Myc leads to an elevated number of rDNA linkages, increased nucleolar size and merging, and a higher level of rRNA synthesis. (A) Western blot analysis of c-Myc protein levels of parental RPE1 cell line and c-Myc–overexpressing single-cell clone derivatives cMyc-1, cMyc-2, cMyc-3, and cMyc-4. (B) Quantification of the number of interchromosomal rDNA linkages in chromosomal spreads from parental RPE1 cells and c-Myc–overexpressing derivatives labeled by FISH with rDNA probe. High-resolution confocal images of ≥10 chromosomal spreads from each cell line were examined. The Mann–Whitney U test was used to compare c-Myc–overexpressing samples with parental RPE1. *, P < 0.05; **, P < 0.001. (C) Representative spinning disk confocal images of nucleolin immunofluorescence (green) of parental RPE1 cells and c-Myc–overexpressing derivatives. Nuclei were counterstained with DAPI (blue). Note enlarged and merged nucleoli. Bar, 10 µm. (D and E) Quantification of nucleolar area (D) and number (E) based on the nucleolin immunofluorescence labeling as in C show enlargement of the nucleoli and their decreased number in c-Myc–overexpressing cells. Bars show averages of three large fields of view (montage images) containing tens to hundreds of cells. *, P < 0.05; **, P < 0.001. Statistical significance was evaluated using one-way ANOVA with Dunnett multiple comparisons, comparing c-Myc samples to parental control. Error bars denote SD. (F) 5-EU incorporation in parental RPE1 cells and c-Myc–overexpressing single-cell clone derivatives. 5-EU–labeled RNA was detected with fluorescent azide and quantified by high-throughput imaging. Statistical significance was evaluated using one-way ANOVA with Dunnett multiple comparisons. *, P < 0.05; **, P < 0.001. (G) Real-time qPCR analysis of pre-rRNA expression of c-Myc–overexpressing derivatives compared with the parental RPE1 cells. The expression of pre-rRNA was normalized to the expression of GAPDH mRNA. Bar heights represent an average fold change of three primer sets to the 5′ETS; error bars represent SD. Statistical significance was evaluated using t tests; **, P < 0.001.

Figure 4.

siRNA minilibrary screen identified UBTF as an essential gene for the formation of rDNA linkages. (A) Schematic representation of the siRNA minilibrary used to screen for genes that eliminate rDNA linkages in c-Myc–overexpressing cells. c-Myc–overexpressing RPE1 derivative cMyc-3 cells were transfected with siRNAs for 72 h, followed by mitotic spread preparation and FISH labeling for the rDNA. The number of interchromosomal linkages was scored manually. Genes whose knockdown prevented mitotic entry, precluding chromosomal spread preparation, are highlighted in gray. The sole hit, UBTF, is highlighted in red. (B) Representative chromosomal spreads from c-Myc–overexpressing cells and untransfected (untr.) control or transfected with UBTF siRNAs. Arrows point to interchromosomal linkages in the control spread. In the UBTF knockdown spread, rDNA linkages are absent. Bar, 10 µm. (C) Quantification of the number of interchromosomal rDNA linkages from parental RPE1 cells and c-Myc–overexpressing derivative cell lines cMyc 1–4 transfected with MYC or UBTF siRNAs. Knockdown of UBTF causes a stronger decrease in the frequency of rDNA linkages than the knockdown of MYC in all derivatives. Images of ≥10 chromosomal spreads from each sample were examined. The Mann–Whitney U test was used to compare siRNA-treated samples with control. *, P < 0.05; **, P < 0.001. (D) Western blot analysis of c-Myc and UBF protein levels in cells treated with indicated siRNAs for 72 h. Both MYC and UBTF siRNAs induced strong protein knockdowns.

Figure 5.

UBF is associated with rDNA loci throughout mitotic progression and with rDNA linkages. (A) Localization of UBF and rDNA during a progressive sequence of mitotic stages. Immuno-FISH of fixed RPE1 cells labeled with rDNA FISH probe (green) and UBF antibody (red). UBF is associated with rDNA in interphase and throughout mitotic progression. Bar, 10 µm. (B) A chromosomal spread from an RPE1 cell labeled by immuno-FISH with rDNA probe (green) and UBF antibody (red). The white box on the left shows the rDNA linkage shown separately on the right (top). Both rDNA and UBF form a bridge between two chromosomes. The panel on the lower right shows individual acrocentric chromosomes labeled with rDNA probe and UBF antibody, respectively. All rDNA loci in this cell line contain UBF. Bar, 10 µm. (C) SIM images of rDNA-linked mitotic chromosomes from c-Myc–overexpressing RPE1 cell line cMyc-3 labeled by immuno-FISH with rDNA probe (green) and UBF antibody (red). Both rDNA and UBF form filamentous connections between chromosomes. Bar, 1 µm.

Figure 6.

Human–mouse hybrid cell line GM15292 displays rDNA linkages only between active (UBF⁺) mouse rDNA loci. (A) A representative mitotic chromosome spread from mouse–human hybrid cell line GM15292 labeled by immuno-FISH with mouse rDNA probe (green), human rDNA probe (red), and UBF antibody (magenta). Boxes 1 and 2 highlight rDNA linkages (magnified on the right; bar, 1 µm). Red arrows point to human rDNA chromosomes. The lower panel shows individual mouse and human rDNA chromosomes arranged according to their species and size. The top row shows rDNA probe labeling, and the bottom row shows UBF antibody labeling of the same chromosomes. This particular spread had 112 total chromosomes, of which 33 were mouse rDNA chromosomes (19 UBF⁺) and five were human rDNA chromosomes. Note that rDNA linkages were formed only between mouse rDNA chromosomes positive for UBF (active). In all human rDNA chromosomes, loci with rDNA were UBF negative (silenced) and did not form linkages. At least 10 chromosomal spreads were examined. Bar, 10 µm. (B) An example of the interphase nucleus from mouse–human hybrid cell line GM15292 labeled by immuno-FISH with mouse rDNA probe (green), human rDNA probe (red), and nucleolin antibody (magenta). While most mouse rDNA is decompacted and associated with nucleolin, human rDNA loci remain compacted and are not incorporated in nucleoli. At least 10 nuclei were examined. Bar, 10 µm. (C) The fraction of rDNA area overlapping with nucleolin in mouse–human hybrid cell line GM15292. The area of overlap was determined between binary masks of mouse rDNA (green) and nucleolin (magenta), and between human rDNA (red) and nucleolin (magenta). Images of masks correspond to the nucleus shown in B. The graph on the right shows fractions of areas overlapping with nucleolin from 10 nuclei. A t test was used to compare nucleolin-overlapping area fractions of mouse and human rDNA. **, P < 0.0001.

Figure 7.

Tetraploid derivatives of human hTERT CHON-002 cell line inactivate some rDNA loci. These UBF-negative rDNA loci are not incorporated in nucleoli and do not form rDNA linkages. (A) Quantification of total number of rDNA loci and active (UBF⁺) number of rDNA loci in RPE1 and CHON cell lines. Lines between data points connect total and UBF⁺ rDNA loci in the same chromosomal spread. In RPE1, all rDNA loci were UBF positive in all cells. In most parental CHON-002 cells, all rDNA loci were UBF positive, with a few cells silencing one rDNA locus. In tetraploid CHON derivatives, all cells had some silenced rDNA loci. 10 spreads per cell line were analyzed. (B) Ploidy analysis of asynchronously growing tetraploid single-cell clones, designated CHON tetraploid-1 and CHON tetraploid-2, generated from hTERT CHON-002 cell line. Cells were fixed and stained with propidium iodide. FACS profiles of CHON tetraploid-1 and CHON tetraploid-2 indicate the doubling of the DNA content. (C) Chromosome spreads from diploid CHON-002 and tetraploid CHON tetraploid-1 cells were labeled by immuno-FISH with rDNA probe (green) and UBF antibody (red). Boxes A–C highlight rDNA linkages that were formed between UBF⁺ rDNA loci (magnified on the right; bar, 1 µm). Arrows 1–3 point to rDNA chromosomes lacking UBF. Panels below show individual rDNA chromosomes labeled with rDNA probe (top row) and UBF antibody (bottom row). In the spread from the parental CHON-002 cell line, there were 10 rDNA chromosomes, all UBF positive. In the spread from the CHON tetraploid-1 derivative, there were 19 rDNA chromosomes, 16 UBF⁺ and 3 UBF⁻ (boxes 1–3, magnified on the right). All rDNA linkages were formed between UBF-positive rDNA chromosomes. Bar, 10 µm. (D) Interphase nuclei from diploid CHON-002 (top) and tetraploid CHON tetraploid-1 (bottom) cells were labeled by immuno-FISH with rDNA probe (green), nucleolin antibody (red), and UBF antibody (magenta). Bar, 10 µm. In the parental CHON-002 cell nucleus, all rDNA was decompacted and associated with UBF and nucleolin. In the tetraploid derivative nucleus, compact rDNA loci were present (arrows) that were UBF negative and were not incorporated in nucleoli (magnified insets; bar, 1 µm).

Figure 8.

rDNA linkages resolve at the metaphase-to-anaphase transition. (A) Stills from time-lapse live imaging of the normal mitotic progression in an RPE1 cell expressing GFP-UBF. UBF is localized to rDNA loci that segregate in anaphase as individual dots without connections to oppositely segregating chromatids. The top panels show maximum-intensity projections of spinning disk confocal images of eGFP-UBF, and the bottom panels show maximum-intensity projections of UBF overlaid with the best focal plane phase-contrast images of the cell. The top panels are magnified 1.5× with respect to the bottom panels. Bar, 10 µm. The complete video sequence is shown in Video 1. (B) A normal anaphase RPE1 cell was labeled by FISH with rDNA probe (green). rDNA loci segregate as individual dots. No rDNA-containing chromatin bridges were observed. Bar, 10 µm. (C) A chromatid spread was prepared from an anaphase RPE1 cell and labeled by FISH with rDNA probe (green) and CenB (red). rDNA loci form distinct individual spots that do not appear to be associated with other loci. Bar, 10 µm.

Figure 9.

Inhibition of topoisomerase II in mitosis prevents the resolution of rDNA linkages at the metaphase-to-anaphase transition. (A) Time-lapse live imaging of the mitotic progression in RPE1 cells expressing GFP-UBF treated with 5 µM topoisomerase inhibitor ICRF-193. The cell treated with topoisomerase inhibitor shows impaired sister chromatid segregation and fails to segregate rDNA marked by GFP-UBF. The drug was added shortly before the initiation of imaging. The top panel shows maximum-intensity projections of spinning disk confocal images of GFP-UBF, and the bottom panel shows maximum-intensity projections of UBF overlaid with the best focal plane phase-contrast images of the cell. The top panel is magnified 1.5× with respect to the bottom panel. Bar, 10 µm. The complete video sequence is shown in Video 2. (B) Localization of rDNA and centromeres in cells that divided in the presence of topoisomerase inhibitor ICRF-193. Asynchronously growing c-Myc–overexpressing RPE1 derivative cell line cMyc 3 was untreated (first left panel) or treated with 5 µM ICRF-193 for 30 min (right panels). Cells were fixed and labeled by FISH with rDNA probe (green) and CenB probe (red). DNA was counterstained with DAPI. Maximum-intensity projections of confocal images depict cells that failed to segregate the rDNA. Four examples of individual cells are shown (1–4). Arrows point to rDNA trapped in the cleavage furrow. Bar, 10 µm. (C) A schematic of the experimental design of topoisomerase II washout experiment is shown. RPE1 cells stably expressing p53 shRNA were arrested in mitosis by colcemid for 10 h and collected by mitotic shake-off. Colcemid was removed, and 10 µM ICRF-193, 500 µM dexrazoxane, or control vehicle (DMSO) were added to the mitotic cells. Cells were allowed to rebuild the mitotic spindle, exit mitosis, and attach to the plates for 4 h. After this, topoisomerase inhibitors were washed out, and cells were allowed to progress through the cell cycle for another 14–15 h. Therefore, cells were treated with topoisomerase II inhibitors during mitotic exit only. Then, colcemid was added again for 10 h to collect cells in the next mitosis for chromosomal spreads. (D) Cells that exited mitosis in the presence of topoisomerase II inhibitors display complex rDNA linkages. Mitotic spreads prepared from cells in C were labeled by FISH with rDNA probe (green) and CenB probe (red) and imaged by SIM. Fragments of spreads containing complex multichromosomal linkages are shown. Two examples are shown per drug treatment. Bar, 1 µm.

Figure 10.

Topoisomerase IIα plays a major role in creating and resolving rDNA linkages. (A) Fluorescent images of interphase cells transiently transfected for 24 h with GFP-topoisomerase IIα (TopoIIα; top) or GFP only (bottom) are shown. GFP-TopoIIα shows propensity to localize in nucleoli. Bar, 10 µm. (B) The number of interchromosomal rDNA linkages in chromosomal spreads from cells transiently transfected with GFP or GFP-TopoIIα was quantified. For this experiment, RPE1 cells were transfected with indicated plasmids for 24 h and arrested in mitosis by addition of colcemid for 12–14 h. Mitotic cells were collected by shake-off and FACS sorted to isolate GFP-positive mitotic cells. Mitotic spreads from cells expressing GFP-TopoIIα show significantly more rDNA linkages compared with GFP only. The Mann–Whitney U test was used to compare GFP-TopoIIα–expressing samples with GFP-expressing control. **, P = 0.0002. (C) Representative confocal images of chromosome spreads from cells transiently transfected for with GFP-TopoIIα, GFP only (top), or TopoIIα (bottom). Spreads were labeled by FISH with rDNA probe (green) and CenB probe (red). Bar, 10 µm. Arrows point to rDNA linkages shown in magnified insets on the right; bar, 1 µm. (D) Chromatin spreads from interphase parental RPE1 cells (top) and cMyc-3 derivative (bottom) cells. Interphase cultures were treated with 100 nM calyculin A for 90 min to induce premature chromatin condensation and labeled by FISH with rDNA probe (green) and CenB probe (red). Asterisks denote individual separated rDNA spots. Bar, 10 µm. (E) Quantification of the number of separated rDNA spots in 100 nM calyculin A–treated interphase cells (D) labeled by FISH with rDNA probe. High-resolution confocal images of ≥10 condensed chromatin spreads were examined. The Mann–Whitney U test was used to compare c-Myc–overexpressing samples with parental RPE1. **, P < 0.001. (F) Working model for how interchromosomal rDNA linkages are generated and resolved. Inter- and intrachromosomal rDNA catenations may form due to the intertwining of transcriptionally active DNA from different chromosomes in the crowded nucleolar environment. Formation of rDNA linkages depends on UBF, and boosting rDNA transcription by overexpressing c-Myc increases the frequency of rDNA linkages. Silent (UBF⁻) loci are not incorporated in nucleoli and do not form linkages. Topoisomerase II generates more catenations by trying to correct the local topology in the interphase nucleus but then resolves these catenations during the metaphase–anaphase transition. Therefore, transcription-dependent formation of rDNA linkages does not lead to chromosomal missegregation and genomic instability under normal conditions.