Nucleolar Reorganization Upon Site-Specific Double-Strand Break Induction

Abstract

DNA damage response (DDR) in ribosomal genes and mechanisms of DNA repair in embryonic stem cells (ESCs) are less explored nuclear events. DDR in ESCs should be unique due to their high proliferation rate, expression of pluripotency factors, and specific chromatin signature. Given short population doubling time and fast progress through G1 phase, ESCs require a sustained production of rRNA, which leads to the formation of large and prominent nucleoli. Although transcription of rRNA in the nucleolus is relatively well understood, little is known about DDR in this nuclear compartment. Here, we directed formation of double-strand breaks in rRNA genes with I-PpoI endonuclease, and we studied nucleolar morphology, DDR, and chromatin modifications. We observed a pronounced formation of I-PpoI-induced nucleolar caps, positive on BRCA1, NBS1, MDC1, γH2AX, and UBF1 proteins. We showed interaction of nucleolar protein TCOF1 with HDAC1 and TCOF1 with CARM1 after DNA injury. Moreover, H3R17me2a modification mediated by CARM1 was found in I-PpoI-induced nucleolar caps. Finally, we report that heterochromatin protein 1 is not involved in DNA repair of nucleolar caps.

Introduction

Embryonic stem cells as well as numerous other rapidly proliferating cells are most threatened by DNA damage that can come from various sources. Many different DNA repair mechanisms evolved to repair any possible damage inflicted on the DNA. Some of these remove bulky adducts from the DNA (such as nucleotide excision repair [NER]); others correct for mismatched nucleotides (mismatch repair) or fix double-stranded breaks (non-homologous end joining and homologous recombination repair, among others). Detailed reviews on different types of DNA repair can be found elsewhere.^1–3 Proteins that carry out different steps of DNA repair are nowadays more or less well described, but other factors important for this process, such as chromatin remodelers, began to emerge in the last decade. Chromatin microenvironment seems crucial for DNA repair, and it changes considerably during DNA repair. Nucleosomes at the very site of double-strand breaks (DSBs) and in close proximity (few kilobases) are evicted or disrupted by various ATP-dependent chromatin remodelers,^4,5 whereas surrounding chromatin is modified as a part of the DNA damage response (DDR). The best described chromatin mark for the repair of DSBs is the phosphorylation of the histone variant H2A.X on serine 139, which is known as γH2Ax.^6,7 This histone modification, altogether with histone ubiquitination mediated by RNF8 and RNF168,^8–10 is an essential part of the DDR signaling cascade. Other chromatin marks, such as histone acetylation and histone methylation, seem important from the point of view of accessibility of chromatin to proteins. It has been shown that transient formation of a more compact, hypoacetylated, and inaccessible chromatin precedes a more loose, accessible state of hyperacetylated chromatin during DDR. This is intuitively plausible, as open chromatin states grant “access” for proteins of the DDR.^5,11,12 Histone modifications involved in DNA repair are not limited to DSB repair. Trimethylation of lysine 36 on histone H3 has been shown to be important for mismatch repair, where it interacts with MutSα protein.¹³

Nucleolus is the tripartite nuclear center of ribosomal biogenesis.¹⁴ It is involved in transcription of rRNA genes and their modification by ribonucleoproteins, such as fibrillarin.¹⁵ Beyond being a Pol-I-driven factory for rRNA synthesis, nucleolus is also involved in processes such as cell-cycle regulation, stress sensing, and many others.^16–18 Genes coding for rRNA are organized as tandem repeats, and often hundreds of copies of pre-rRNA genes are found clustered in mammalian genomes (~400 copies in human genome, ~50% transcriptionally active).¹⁷ These clusters of rRNA genes (also called nucleolar organizer regions [NORs], because nucleolus forms around these loci)¹⁹ reside on short arms of different acrocentric chromosomes in humans. It also appears that epigenetic regulation of transcription in the nucleolus differs from nuclear epigenetic regulation in some aspects (for a review, see Bártová et al.²⁰ and Mcstay et al.²¹). In response to stress, nucleolus undergoes a radical restructuralization, where rRNA genes are relocated from the interface of fibrillar center and dense fibrillar component to the periphery of a remnant of the nucleolus termed central body.

From a methodical standpoint, DSBs in cells are usually induced by ultraviolet (UV) microirradiation,^22,23 using the I-SceI system^24–26 or chemical agents such as bleomycin.^27–29 We decided to use the I-PpoI endonuclease system^30–32 that generates double-stranded breaks in rRNA genes, which is very useful for studying nucleolar reorganization and DDR in the nucleolus. Using this system, we sought to examine the complex nucleolar reorganization in response to directed DSB introduction. First of all, we examined the morphology of nucleoli upon DNA damage and compared this morphology with typical actinomycin D (ActD)–induced nucleolar caps (NCs). Next, we analyzed the recruitment of different DDR factors to upstream binding factor (UBF)–positive NCs. As recent articles reported the interaction of TCOF1 and NBS1,^33,34 we wanted to examine the role of TCOF1 in nucleolar DDR more closely. We show that TCOF1 interacts with chromatin modifiers such as CARM1 and HDAC1 following rRNA gene cleavage. We confirm findings of van Sluis and McStay³⁵ by detecting numerous DDR factors in the vicinity of UBF-positive NCs and show that NBS1 is also involved in DSB recognition in the nucleolar periphery after rRNA gene cleavage. Last but not least, we analyze the role of histone acetylation and histone arginine methylation in nucleolar reorganization and DNA repair and report that arginine methylation is present in NCs.

Materials and Methods

Cell Culture, Transfection, γ-Irradiation, and Actinomycin D Treatment

Mouse embryonic stem cells (cell line D3; ATCC CRL-1934) were cultivated on culture dishes (Nunc; Sigma-Aldrich, St. Louis, MO) coated with sterile 0.1% gelatin. Cells were maintained in high-glucose DMEM (#P04-03600; PAN-Biotech, Aidenbach, Germany) with 15% FBS (#FB1001; Biosera, Boussens, France), supplemented with non-essential amino acids (#1140-035; Thermo Fisher Scientific, Waltham, MA), mouse leukemia inhibitory factor (mLIF; #ESG1106; Merck Millipore, Darmstadt, Germany), and monothioglycerol. Media were changed daily, and cells were passaged every 2 to 3 days depending on their confluency. Cells were passaged by trypsinization and seeded as single cells. For immunofluorescence experiments, mouse embryonic stem cells were seeded onto µ-dishes (#81166; ibidi, Martinsried, Germany).

Cells were transfected with plasmid DNA using either Metafectene PRO (#T040-1.0; Biontex, Munich, Germany) or Effectene (#301425; Qiagen, Hilden, Germany) transfection reagent. Majority of experiments in this work have been conducted either 8 or 24 hr posttransfection. Time interval is included for each experiment in the “Results” section. pICE-HA-PpoI-NLS was a gift from Steve Jackson (Addgene plasmid #46963; Cambridge, MA). UBF-GFP (green fluorescent protein) was a gift from Tom Misteli (Addgene plasmid #17656). NBS1-GFP plasmid used in this study was purchased from OriGene (RG214682; OriGene, Rockville, MD). Irradiation with 5 Gy was delivered by cobalt-60, and cells were analyzed 1 hr after irradiation. Cells were treated with either 50 or 500 ng/ml of ActD (#A9415; Sigma-Aldrich) for 2 hr.

Double-Strand Break Induction

DSBs were induced in mouse embryonic stem cells by transient transfection with I-PpoI endonuclease. Proper plasmid delivery and presence of the I-PpoI endonuclease in cells was verified by an antibody against the HA-tag that this nuclease carries. Both immunofluorescence analysis and Western blotting confirmed the presence of I-PpoI in stem cells.

Immunofluorescence Analysis

For immunofluorescence staining, cells were washed thoroughly with PBS and then fixed for 20 min in 4% paraformaldehyde. Then, cells were rinsed in PBS and permeabilized in 0.1% Triton X-100 (#02300221; MP Biomedicals, Santa Ana, CA) for 8 min and in 0.1% Triton X-100/saponin for 12 min. Subsequently, cells were washed 2 times in PBS for 15 min each. Afterward, samples were blocked for 1 hr in 1% BSA in PBS. Incubation with primary antibodies followed, which took place either overnight at 4C or for 1 hr at room temperature. Primary antibodies used for immunofluorescence are listed in Table 1. After incubation with primary antibodies, cells were washed 3 times in PBS and then incubated with a secondary antibody conjugated to a fluorescent dye for 1 hr. We used Alexa Fluor 405, 488, 546, and 594-conjugated secondary antibodies (Thermo Fisher Scientific). Then, cells were washed in PBS 2 times and either stained with 4′,6-diamidino-2-phenylindole (DAPI) dihydrochloride or simply mounted in an antifade agent, Vectashield (#H-1000; Vector Laboratories, Burlingame, CA).

Table 1.

Antibodies Used in This Study.

Antibody	Catalog Number	Dilution
α-53BP1	ab21083 (Abcam; Cambridge, UK)	1:100 (IF)
α-BRCA1	sab2702136 (Sigma-Aldrich)	1:150 (IF)
α-CARM1	ab128851 (Abcam)	1:100 (IF); 1:1000 (WB)
α-cyclinB1	MS868 (Thermo Fisher Scientific)	1:1000 (WB)
α-Fibrillarin	ab5821 (Abcam)	1:200 (IF); 1:1000 (WB)
α-HA	H3663-100 (Sigma-Aldrich)	1:100 (IF); 1:1000 (WB)
α-HDAC1	06-720 (EMD Millipore); sc-7872 (Santa Cruz Biotechnology; Dallas, TX)	1:100 (IF); 1:1000 (WB)
α-MDC1	ab41951 (Abcam)	1:100 (IF)
α-NBS1	N3162 (Sigma-Aldrich)	1:100 (IF); 1:3000 (WB)
α-NBS1	ab23996 (Abcam)	1:300 (IF)
α-PML	mab3738 (EMD Millipore)	1:150 (IF)
α-PRMT1	ab73246 (Abcam)	1:100 (IF); 1:1000 (WB)
α-TCOF1	ab65212 (Abcam)	1:1000 (WB)
α-UBF1	sc9131 (Santa Cruz Biotechnology)	1:200 (IF)
α-H4Ac	382160 (EMD Millipore)	1:200 (IF)
α-H3K9Ac	06942 (EMD Millipore)	1:200 (IF)
α-γH2Ax	ab2251 (Abcam)	1:100 (IF); 5 µl/sample (ChIP)
α-H3R17me2a	ab8284 (Abcam)	1:100 (IF); 5 µl/sample (ChIP)
α-H3	ab1971 (Abcam)	2 µl/sample (ChIP)
α-Suv39h1	ab38637 (Abcam)	1:100 (IF)
α-Jmjd2b	ab103129 (Abcam)	1:100 (IF)

This table summarizes the types of antibodies used, their dilution, catalog number, and manufacturer. Abbreviations: ChIP, chromatin immunoprecipitation; IF, immunofluorescence; WB, western blotting; UBF, upstream binding factor; PML, promyelocytic leukemia.

Cell Visualization and Microscopy

Both living cells and fixed samples have been visualized on a Leica TCS SP-5X confocal microscope (Leica Microsystems, Wetzlar, Germany). Images have been captured by a 63× HCX PL APO immersion objective, whose numerical aperture is 1.4. A diode laser (405 nm) was used to visualize DAPI and Alexa Fluor 405–stained proteins, whereas white light laser (470–670 nm) was used to visualize GFP- or RFP (red fluorescent protein)-tagged proteins as well as proteins stained with Alexa Fluor 488/594 secondary antibodies.

Image Deconvolution

Image deconvolution has been performed in Leica LAS X Core software (Leica Microsystems). Point spread function has been generated in the computer software for each image, based on objective parameters using Lorentz fit. The resolution full width at half maximum for the generation of the point spread function has been set to 139 nm (measured resolution in x–y plane at 488 nm). Wiener filter method has been used, with regularization parameter set to 0.3.

Image Analysis

Images taken on the Leica TCS SP-5X confocal microscope were analyzed on Leica LAS X Core software. Fluorescence intensity evaluation (scan lines in different channels) and two-dimensional colocalization scatterplots were created in Leica LAS X Core software. Colocalization analysis was performed in Coloc2 plugin on ImageJ software (Rasband, W.S., ImageJ, US National Institutes of Health, Bethesda, MD; http://imagej.nih.gov/ij/; 1997–2016).

Statistical Analysis of Colocalization

We conducted statistical analysis of colocalization in ImageJ Coloc2 plugin. For each pair of factors analyzed, we measured colocalization in 15 different NCs (n = 15). We present a number of coefficients that describe colocalization, and we describe them briefly in this section. We used the Pearson’s correlation coefficient (R) and Manders’ colocalization coefficients (MCCs)³⁶ to show the extent to which two proteins colocalize. R is calculated as follows:

$$R=\frac{{\displaystyle \sum _i\left(R_i-\overline{R}\right)\times \left(G_i-\overline{G}\right)}}{\sqrt{{\displaystyle \sum _i{\left(R_i-\overline{R}\right)}^2}\times {\displaystyle \sum _i{\left(G_i-\overline{G}\right)}^2}}},$$

where R_i = fluorescence intensity of red channel in pixel I; G_i = fluorescence intensity of green channel in pixel I; $\overline{G}$, $\overline{R}$ = mean fluorescence intensities.

Pearson’s correlation coefficient, <−1,1>, is sensitive to the proportionality of signals, is independent of signal levels, and measures the linear relationship of the pair of probes.

MCCs can be calculated simply as follows:

$$t{M}_1=\frac{{\displaystyle \sum _i{R}_{i,col.}}}{{\displaystyle \sum _i{R}_i}}\mathrm{and}t{M}_2=\frac{{\displaystyle \sum _i{G}_{i,col.}}}{{\displaystyle \sum _i{G}_i}},$$

where R_i,_col. = R_i, if G_i > 0; G_i,_col. = G_i, if R_i > 0.

Manders’ coefficients (tM₁, tM₂), <0,1>, are indicators of overlap between two probes and are independent of signal proportionality. We have calculated thresholded Manders’ coefficients and used the Costes thresholding method to determine the threshold values.³⁷

Although R and MCCs give an estimate of the degree of colocalization, they do not offer an estimate of significance of these values. Calculating the p value of significance for R is not useful in this study, because the high number of pixels analyzed in every region of interest (ROI) makes even very low values of R significant at p = 0.05 or p = 0.01 (for n = 100, R = 0.254 is a significant correlation at p = 0.01; Statistical Tables).³⁸ For this reason, we used two different randomization methods to evaluate the significance of our results. One of these is the Costes randomization method (50 iterations), the other is Fay randomization (x,y,z translation of the image; 25 iterations). We present both p and confidence values, <0,1>, in the colocalization analysis. For more in-depth discussion on colocalization analysis, see Dunn et al.³⁹

NOR Staining

Cells were rinsed in PBS and then fixed in 4% paraformaldehyde for 20 min. Cell membrane was permeabilized by incubation in 0.1% Triton X-100 and saponin, as was described for immunofluorescence. Following permeabilization, samples were dehydrated by subsequent application of 70%, 80%, and 96% prechilled ethanol, each for 1 min. Staining solution was prepared by mixing solution A (which consists of 2% gelatin and 1% formic acid in distilled water) and solution B (which consists of 0.5-g AgNO₃ dissolved in 1 ml of distilled water) in a 1:2 ratio. Staining solution was applied to samples for 30 min in dark. Afterward, samples were exposed to 96%, 80%, and 70% ethanol (in the presented order), each for 1 min, and mounted in Vectashield (Vector Laboratories).

Cell Lysate Preparation

Cells grown to 70% to 90% confluency were rinsed in PBS twice and lysed in 100 µl of 1% SDS. Lysates were then sonicated at maximum power for 5 sec. Protein concentrations were determined on a µQuant microplate spectrophotometer (BioTek, Winooski, VT). Then, 5 µl of a 1:1 β-mercaptoethanol/bromophenol blue mixture was added to lysates which were then boiled for 5 min.

Polyacrylamide Gel Electrophoresis and Western Blotting

Proteins were separated according to their size by SDS-PAGE. Depending on the target, proteins were separated on either 8%, 10%, or 12% polyacrylamide gels. The electrophoresis ran at 130 V, 400 mA, for 90 to 150 min. After the electrophoresis, proteins were transferred to a polyvinylidene fluoride membrane (#10600021; GE Healthcare Life Sciences, Little Chalfont, UK) by “wet/tank” Western blotting. Afterward, membranes were incubated for 1 hr in a blocking solution (either 2% milk in high- or low-salt TBS or 2% gelatin in high- or low-salt TBS). Incubation with primary antibody took place overnight at 4C. Membranes were then washed 3 times with corresponding TBS and incubated with a secondary antibody conjugated to horseradish peroxidase for 90 to 120 min. Proteins were then visualized by the ECL Western Blotting Detection Reagent (#RPN2106; GE Healthcare Life Sciences) on an LAS-3000 system (Fujifilm, Tokyo, Japan).

Protein Immunoprecipitation

Proteins were immunoprecipitated by the Catch and Release v2.0 reversible immunoprecipitation kit (#17-500; Merck Millipore), according to the manufacturer’s recommendation. Briefly, cells were lysed on ice in 500 µl of chilled Pierce IP lysis buffer (#87787; Thermo Fisher Scientific), to which protease inhibitors (aprotinin + PMSF) have been added to a final concentration of 1 µg/ml. After 10 min of incubation with IP lysis buffer, lysates were centrifuged at 13,000 × g for 10 min at 4C. Supernatant was retained and protein concentration was calculated as described in the “Polyacrylamide Gel Electrophoresis and Western Blotting” section. In the meantime, immunoprecipitation capture tubes have been washed 5 times in a wash buffer by centrifugation (5000 rpm, 30 sec). Then, 10 µl of antibody capture affinity ligand, 4 µg of primary antibody, and 500 µg of cell lysate were added to the spin columns. Wash buffer was added to the spin column to a final volume of 500 µl, and samples were incubated overnight at 4C. The next day, spin columns have been washed 5 times in the wash buffer. After that, immunoprecipitate has been eluted by adding 70 µl of denaturing elution buffer directly onto the columns. Inputs have been prepared in the same way as whole-cell lysates described above.

Chromatin Immunoprecipitation

Cells were fixed in fresh 1% formaldehyde for 15 min at 37C. Then, cells were washed twice in prechilled 1× PBS containing protease inhibitors (aprotinin and PMSF, both at 1 µg/µl concentration). SDS lysis buffer (#20-163; Merck Millipore) was then added to cells (200 µl of lysis buffer per 1 × 10⁶), and cells were incubated on ice for 10 min. Cell lysates were sonicated on Hielscher sonicator (6 times for 10 sec for each sample). Afterward, sonicated lysates were centrifuged at 13,000 rpm at 4C for 10 min. Supernatant was transferred to a new Eppendorf tube (#72.690.007; Sarstedt, Nümbrecht, Germany) and pellet was discarded. Samples were then diluted 10 times in chromatin immunoprecipitation (ChIP) dilution buffer (#20-153; Merck Millipore). Diluted supernatant was then precleared with Sperm DNA/Protein A Agarose Slurry (#16-157; Merck Millipore; 75 µl/sample, rotation at 4C, 30 min). Samples were then briefly centrifuged (1000 × g, 1 min, 4C), and supernatant was transferred to a new tube. Incubation with primary antibody took place overnight with rotation at 4C. Amounts of antibodies used for ChIP are listed in Table 1. Next day, Salmon Sperm DNA/Protein A Agarose Slurry (#16-157; Merck Millipore) was added to samples (60 µl/sample), and the samples were left rotating for 1 hr at 4C. Then, agarose beads were pelleted and supernatant was discarded. A series of wash steps followed, where each step included supernatant aspiration and incubation in a specific buffer for 5 min with rotation. First, agarose beads were washed in Low Salt Immune Complex Wash Buffer (#20-154; Merck Millipore), then in High Salt Immune Complex Wash Buffer (#20-155; Merck Millipore), then in LiCl Immune Complex Wash Buffer (#20-156; Merck Millipore), and finally twice in TE Buffer (#20-157; Merck Millipore). After aspiration of TE Buffer, agarose beads were resuspended in 250 µl of fresh elution buffer (0.1-M NaHCO₃, 1% SDS, pH = 8). Samples rotated for 15 min at room temperature. Then, supernatant was transferred to a new tube, and the procedure was repeated with another 250 µl of elution buffer. Supernatants were then joined (total volume of 500 µl). Cross links were reversed by the addition of 20-µl of 5-M NaCl (#20-159; Merck Millipore) and overnight incubation in 65C water bath. After the overnight incubation, 10 µl of 0.5-M EDTA, 20 µl of 1-M Tris–HCl, and 2 µl of Proteinase K (#19133; Qiagen) were added to each eluate and this mixture was incubated for 1 hr at 45C. After this incubation, DNA isolation followed as described below.

DNA and RNA Isolation

DNA was recovered using the QIAamp DNA isolation kit (#51304; Qiagen). RNA was isolated by the RNeasy Mini Kit (#74104; Qiagen). Total of 2 × 10⁶ cells were used for RNA isolation for each treatment.

PCR Amplification

We used following primers for PCR amplification (primer positions are annotated relative to transcription start site [TSS] of the mouse 45S pre-rRNA [X.82564.1]).

Mouse rDNA promoter fwd: −165/−145: 5′-GACC-AGTTGTTCCTTTGAGG-3′, mouse rDNA promoter rev: −20/−1: 5′-ACCTATCTCCAGGTCCAATAG-3′ (fragment length = 165 bp) and mouse 28S rDNA fwd: +8124/8145: 5′-GCGACCTCAGATCAGACGTGG-3′, mouse 28S rDNA rev: +8529/8549: 5′-CTTAACGGTTTC-ACGCCCTC-3′ (fragment length = 426 bp). Combi PPP Master Mix (#C-208; Top-Bio, Prague, Czech Republic) was used for amplification. We performed 25 cycles of denaturation at 95C for 30 sec, while rDNA promoter primers were annealed at 55C, and rDNA-encoding 28S rRNA was annealed at 58C, both for 40 sec. The extension temperature was 72C for 40 sec for both primer pairs. Four percent DMSO was added to the 28S rRNA PCR reaction mix. The final extension step was set to 5 min. The PCR reaction was performed using DNA Engine, Peltier Thermal Cycler (Bio-Rad, Hercules, CA).

Agarose Gel Electrophoresis

PCR products as well as isolated RNA were separated on 2% agarose gels (#50001, SeaKem LE Agarose; Lonza, Basel, Switzerland). Electrophoresis ran for 60 to 90 min at 100 V. Nucleic acids were visualized using GelRed (#41003; Biotium, Hayward, CA).

Results

Analysis of Nucleolar Morphology Upon Transcriptional Silencing and Directed DNA Damage

We performed staining for known markers of nucleolus, such as Fibrillarin, TCOF1, and UBF to analyze the possible changes in nucleolar structure and morphology. We have compared control cells that display large heterogeneous nucleoli (Fig. 1A and Fig. 2A), with cells treated by different concentrations of ActD (Fig. 1B and C) and cells transfected by PpoI endonuclease (Fig. 1D). It is clear that nucleolus is reorganized after both transcriptional silencing (triggered by ActD) and rRNA gene cleavage mediated by PpoI endonuclease. NCs are easily identified by Fibrillarin and UBF staining, and these markers occupy distinct positions in NCs (Fig. 1B, C, and D). Whereas transcriptional silencing induces the formation of a few large, crescent-like caps, DSBs in the rRNA genes induced by PpoI endonuclease lead to the formation of smaller UBF foci and less pronounced externalization of nucleolar factors such as Fibrillarin and TCOF1 (Fig. 1D, and Fig. 2C and D). “Ring-like” pattern of both Fibrillarin and TCOF1 can be observed in cells already 8 hr after cell transfection with PpoI endonuclease. We note that there are no significant differences between cells transfected for 8 and 24 hr (Fig. 2C and D). Silver nitrate staining is in accord with immunofluorescence analysis (Fig. 2; top half). To further examine the distribution of TCOF1 and its ring-like morphology, we performed a three-dimensional reconstruction of cells stained for TCOF1 and HA-PpoI-NLS, which clearly shows the occurrence of this ring-like pattern (Supplementary Video 1).

Figure 1.

Changes in nucleolar morphology after transcriptional silencing and double-strand break induction. A, Controls cells stained for Fibrillarin and UBF showing typical nucleolar distribution. B, C, Cells treated with low (50 ng/ml) and high (500 ng/ml) concentration of ActD displaying nucleolar caps with a hierarchical nucleolar marker organization. D, Fibrillarin and UBF localization upon cleavage of rRNA genes (8 hr posttransfection with HA-PpoI-NLS). Scale bars: A–D = 10 µm. Abbreviations: DAPI, 4′,6′-diamidine-2′-phenylindole; ActD, actinomycin D, UBF, upstream binding factor.

Figure 2.

Analysis of nucleolar morphology by AgNOR and TCOF1 staining. A, Untreated cells stained by silver nitrate and visualized by TCOF1 staining. B, Cells treated with high (500 ng/ml) concentration of ActD showing “crescent-like” nucleolar caps. C, Reorganized nucleoli showing a “ring-like” pattern upon directed double-strand breaks induction (8 hr posttransfection). Cells with nucleolar caps are highlighted by red arrowcaps. D, Cells shown 24 hr posttransfection. We did not observe any notable difference between the 8 and 24 hr periods. Scale bars: A–D = 25 µm. Abbreviation: AgNOR, silver nitrate staining of nucleolar organizer regions; ActD, actinomycin D.

We have analyzed the levels of 28S rRNA and 18S rRNA in cells after treatment with ActD and cell transfection with PpoI for 8 and 24 hr. We observe no changes in the levels or the ratio of 28S rRNA to 18S rRNA, which suggests that in the time points used in this study (2-hr treatment with ActD, 8-24 hr transfection with PpoI), the amount of rRNA in cells is stable (Fig. S1).

DDR Factors and Their Recruitment to NCs

We have tested the recruitment of a number of DDR factors to sites of DNA damage in the nucleolar context. Besides factors identified previously to accumulate near NCs such as BRCA1 and 53BP1, we have found both NBS1 and MDC1 to be recruited to NCs (Figs. 3 and 4). To analyze the relative position of DDR factors and nucleolar markers, we compared the fluorescence intensities of DNA repair factors and UBF. It is clear that most DDR factors are found adjacent to UBF-positive NCs (Fig. 3A), but they colocalize with one another in most cases (Fig. 3B). Results from colocalization analysis between NBS1 and UBF confirm NBS1 involvement in DSB recognition at NCs (Fig. 4E). We have also tested the mutual arrangement of Fibrillarin, UBF, and BRCA1 in caps, and we observe that DDR factors can be found in contact with UBF and that DDR factors are found at the outer rim of NCs (Fig. S2).

Figure 3.

DNA damage response factors and their relative position in nucleolar caps. A, Positioning of DNA repair factors with respect to UBF. Fluorescence intensity graphs show the relative position of different factors in the region of interest (ROI = scan line, each 3 µm long and 0.5 µm wide). Both MDC1 and γH2Ax can be found surrounding UBF-positive nucleolar caps. B, Colocalization of DNA repair proteins at sites of DNA damage, as seen for BRCA1/53BP1 and γH2Ax/53BP1, respectively. Scale bars: A, B = 5 µm. Abbreviation: UBF, upstream binding factor.

Figure 4.

NBS1 localization in nucleolar caps. A, B, Control cells (transfected with UBF-GFP) and cells transfected with both UBF-GFP and HA-PpoI-NLS showing the distribution of NBS1. C, Detailed view of endogenous NBS1 after double-strand break (DSB) induction indicating its recruitment to UBF-positive nucleolar caps (HA-PpoI-NLS—blue, NBS1—green, UBF—red). D, Exogenous UBF is recruited to the nucleolar periphery upon DSB induction (HA-PpoI-NLS—blue, NBS1-GFP—green, TCOF1—red). E, Two-dimensional scatterplot for NBS1/UBF colocalization and related statistics (described in the “Materials and Methods” section). Scale bars: A–E = 10 µm. Abbreviations: UBF, upstream binding factor; GFP, green fluorescent protein.

Role of Chromatin Remodelers in DDR of Nucleolar Periphery

We analyzed possible interactions between nucleolar proteins and chromatin modifiers. To this end, we immunoprecipitated TCOF1, a nucleolar protein that has been previously reported to interact and cooperate with NBS1 during DDR. The immunoprecipitation has been carried out in control cells, in cells transfected with I-PpoI, and in cells irradiated by 5 Gy of gamma radiation. We identified HDAC1 as a TCOF1 interactor upon DNA damage in the rRNA gene locus 8 hr posttransfection (Fig. 5B). This interaction has been verified by reverse immunoprecipitation (precipitation of HDAC1 and subsequent detection of TCOF1 by Western blotting, Fig. 5A). We also report an interaction between CARM1 and TCOF1 during DNA repair in rRNA genes, both 8 and 24 hr after cell transfection (Fig. 5B). To validate our results, we checked the interaction of TCOF1 with some of its known interaction partners, such as UBF (Fig. 5A). We also tested proteins such as cyclin B, Oct3/4, and PRMT1, which all showed no interaction (Fig. 5C).

Figure 5.

Changes in protein interactions of TCOF1 upon rRNA gene cleavage. A, Detection of TCOF1 after HDAC1 protein IP. B, IP of TCOF1 and identification of novel interaction partners such as HDAC1 and CARM1 after nucleolar reorganization. C, Detection of a number of proteins that do not display interaction with TCOF1, negative IP controls. Abbreviations: IP, immunoprecipitation; NC, negative control (IP without the antibody).

As many chromatin remodelers have been recently found to cooperate with DDR factors, we investigated their potential role in DSB repair in the nucleolus. Immunofluorescence staining after 24 hr did not reveal any significant enrichment or focal distribution of factors such as Jmjd2b, Suv39h1, or PRMT1 in UBF-positive NCs (data not shown). HDAC1 and CARM1 did not show any specific pattern, neither 8 nor 24 hr posttransfection (Fig. 6A and C). However, when we analyzed the distribution of histone modifications after NC formation, we discovered that H3R17me2a, an activating histone modification mediated by CARM1, relocates with UBF to NCs. We also checked the status of histone acetylation in NCs, given that histone acetylation marks are removed by HDAC1. Histone acetylation seems to be absent from UBF-positive caps. For all three histone modifications analyzed, we performed colocalization analysis, which is in agreement with visual cues suggesting H3R17me2a presence in UBF caps and absence of histone acetylation (Fig. 6B, D, and E). Results from colocalization analysis are presented in aforementioned figures and also in graph/table in Supplementary Fig. 3.

Figure 6.

Analysis of selected chromatin modifiers and histone modifications in nucleolar caps. A, CARM1, a histone arginine methyltransferase, is not recruited to nucleolar caps. B, Colocalization of H3R17me2a with UBF in nucleolar caps. C, D, E, Neither HDAC1 nor histone acetylation marks (H3K9Ac, H4Ac) are found in nucleolar caps (DAPI—blue, histone modifier/histone modification—green; UBF—red). Two-dimensional scatterplots and relevant statistics are presented for each factor analyzed. Scale bars: A–E = 10 µm. Abbreviation: UBF, upstream binding factor.

We used ChIP to check the presence of histone modifications such as γH2Ax and H3R17me2a at rRNA gene loci. We used primers that cover both the 28S rRNA gene body and the rRNA promoter. As expected, we found an enrichment of γH2Ax at the rRNA promoter region, indicative of DNA lesions after PpoI transfection (Fig. 7A). Level of histone arginine methylation does not change after DNA cleavage by PpoI endonuclease, which suggests that the colocalization of H3R17me2a with UBF in NCs is a result of rRNA gene redistribution during nucleolar reorganization (Fig. 7B). We also performed histone H3 ChIP, which works both as a positive control and a tool to investigate possible changes in nucleosome occupancy. We do not observe any major changes in the rRNA promoter region, whereas we see an increase in H3 levels at the 28S rRNA region (Fig. 7C).

Figure 7.

Chromatin immunoprecipitation analysis of histone modifications at the rRNA gene promoter and 28S rRNA gene body (24 hr posttransfection). A, Analysis of the level of γH2Ax and H3R17me2a at the rRNA gene promoter. Significant increase of γH2Ax in the promoter region is visible, whereas the level of H3R17me2a does not change upon induction of DNA lesions. Histone H3 chromatin immunoprecipitation was included as a positive control. B, In the coding sequence of 28S rRNA, we have also found a very slight increase in γH2Ax. C, Schematic representation of the position of primers with respect to the I-PpoI cleavage site (indicated by the lightning). Annotation of primer annealing sites is relative to the transcription start site. Abbreviation: NC, negative control.

HP1 and PML Response to rRNA Gene Cleavage

Heterochromatin protein 1 (HP1) and its isoforms have been previously reported to accumulate in DSBs. Given their perinucleolar localization, we tested whether HP1 accumulates into NCs. Embryonic stem cells were transfected with wild-type as well as mutant HP1β lacking either chromo-, hinge-, or chromoshadow domain. Neither wild-type nor mutant HP1β was recruited to DNA damage sites in NCs 24 hr posttransfection (data not shown). Next, we tested whether a different isoform of HP1 protein might respond to DSB induction. We performed immunofluorescence analysis of stem cells 24 hr posttransfection and found that none of the isoforms of HP1 is recruited to NCs 24 hr posttransfection. At the same time, we reinforce the findings of other groups, which show that promyelocytic leukemia (PML) is recruited to UBF-positive NCs and that this protein is wrapped around UBF foci just like most of the factors examined in this study (Fig. S4).

Discussion

It is important to address nucleolar reorganization in response to DNA damage and cellular stress in general. This response of nucleolus to cellular stress has been known for decades.^17,40,41 Nucleolar reorganization is often studied by ActD exposure for a short period of time. This produces NCs, most likely as a consequence of transcriptional silencing. However, there is a certain ambiguity, because ActD works not only as a silencer for Pol-I-driven transcription but also as a DNA damage–inducing agent.⁴² In our work, we observed NC formation upon ActD treatment as well as after PpoI transfection.

A mechanistic explanation for the changes in nucleolar organization is still to come. A recent genome-wide analysis of genes involved with either nucleolar growth or reduction has shown that factors such as HIR (a histone chaperone complex) or Tip60 (a histone acetyltransferase) play a role in this process.⁴³ Our data indicate that other chromatin modifiers such as HDAC1 and CARM1 might be involved in this process. Relocalization of rRNA genes to the nucleolar periphery upon cellular stress might serve to expose this locus to nucleoplasm to enable a more efficient DNA repair. Nucleolar size itself is correlated with cellular metabolism and can be also used as a powerful prognostic tool in oncology, where detection of large, prominent nucleoli in tumors is correlated with poor prognosis in patients.⁴⁴ In recent years, a lot of reports have shown that nucleoli behave like liquid-phase droplets^45,46 and a lot of new insights came from biophysical studies of nucleoli, indicating a big role for phase separation processes as a means for functional organization.⁴⁷

Other groups have previously reported the occurrence of DDR factors in or near NCs with the use of I-PpoI system.^35,48 We observed similar behavior of DDR factors in NCs in our mouse embryonic stem cell model, but we note that there is a hierarchical organization of these factors with respect to Fibrillarin and UBF, where DNA repair factors can be usually found on the outer rim of NCs.

Another interesting finding is that we observe NBS1 localized in NCs, which, to our knowledge, has not been reported previously. NBS1-GFP that we used in our study surprisingly localizes to the nucleolus, even without DNA damage. After having a close look at the plasmid amino acid sequence, we observed that the linker sequence between NBS1 and GFP contains two additional basic amino acids (a lysine and an arginine residue). As the amino acid sequence of NBS1 itself contains a KRRR at the very C terminus, it is possible that the addition of basic residues in the C terminus generates a novel nucleolar localization signal (NoLS). A recent study indicates that as little as four basic arginine residues in tandem are sufficient to create an NoLS in a protein that already harbors a nuclear localization signal.⁴⁹ Even though the localization of NBS1-GFP in the nucleolus might be artefactuous, it remains nevertheless true that it responds to rRNA gene cleavage by accumulation to the NCs. We also show that endogenous NBS1 is recruited to NCs 8 hr after transfection, which underscores the consistency of our results.

The interplay of chromatin remodelers and DDR factors in DNA repair is well documented.^11,50,51 Nucleolus itself is a good example of how transcription, chromatin remodeling, and DNA repair might go together. Cockayne syndrome B (CSB) protein is a DDR factor involved in the early steps of transcription-coupled NER that is also important in the regulation of rRNA gene expression.^52,53 As for HDAC1, it was first identified as a part of the nucleolar remodeling complex.⁵⁴ Subsequent research showed its importance for transcriptional silencing at the rRNA gene locus.⁵⁵ HDAC1 was later observed directly at sites of DNA damage induced by UV microirradiation, where it is important for non-homologous end joining repair pathway.⁵⁶ It has been also reported that histone deacetylase inhibitors influence the retention of DDR factors, such as 53BP1 and BRCA1, at sites of DNA damage.⁵⁷ HDAC1 might therefore be a critical factor in mediating both transcriptional silencing and DNA repair in the nucleolus, and this is why we decided to analyze its possible interaction with TCOF1. We were also keen to analyze the possible role of CARM1, and this was prompted by the research of Wu and Xu,⁵⁸ which have shown that H3R17me2a interacts with RNA polymerase-associated factor 1. Our previous experiments on CARM1, which have shown the decrease of nucleolar size upon CARM1 inhibition by ellagic acid,^59,60 were also an impulse to analyze the role of CARM1 more closely. Numerous nucleolar factors (such as Fibrillarin, TCOF1, and UBF) follow rRNA genes and relocalize to nucleolar periphery upon rRNA cleavage, which opens the possibility for novel protein–protein interactions, which turns out to be the case for both histone modifiers that we analyzed.

One last thing to address is the fact that we did not observe numerous chromatin modifiers previously reported to associate with DSBs. This can be caused by many factors. Although there are potentially hundreds of PpoI cleavage sites in rRNA loci, chromatin remodelers that can participate in chromatin modification at the cleavage site might be below the detection limit for immunofluorescence visualization. Next, it is plausible that nucleolus has important differences in its DDR when compared with non-nucleolar DNA repair. This is exemplified by the NBS1 TCOF1 interaction described recently,^33,34 but there are other examples. Nucleolin, an abundant nucleolar protein that works as a chromatin remodeler, has been shown to interact with Rad50, an integral protein of the Mre11-Rad50-NBS1 (MRN) complex.⁴ Nucleolin also participates in DSB repair in a γH2Ax-dependent pathway.⁶¹ Last but not least, it is known that brief condensation of damaged chromatin and its modification by repressive marks happen in a very narrow time window (in order of minutes),⁵ which makes it hard to detect in our transfection-based system. When looking at histone modifications themselves, these can be found altered in the NCs, which is the case for γH2Ax and H3R17me2a. Interestingly, each of these modifications shows a different distribution relative to UBF, where γH2Ax forms foci that are closely adjacent to UBF and H3R17me2 foci align with those of UBF. One explanation for this is that H3R17me2a is normally found in the nucleolus where it works as an activating mark and simply reorganizes to UBF-positive NCs with the underlying rRNA repeats when they are damaged. This explanation is also supported by the ChIP data in this study, showing no increase in H3R17me2a level after DNA damage.

HP1 and all its isoforms (α,β,γ) have been extensively characterized (for a review, see Kwon and Workman⁶²). HP1 has been also shown to be recruited to sites of DNA damage.^63,64 This protein is known to bind to H3K9me3 histone mark, which is a repressive histone modification. Binding of HP1 is thought to reinforce heterochromatin states, but there are some exceptions. Interestingly, HP1γ has been shown to be involved in the recognition of H3K9me2/3 in the context of ribosomal genes, and HP1γ altogether with CSB (mentioned previously as a DDR factor that is also a transcriptional transactivator) mark active genes.⁵² We did not observe recruitment of any of the HP1 isoforms to UBF-positive caps after DNA damage. However, it has been shown that HP1 proteins interact with UBF and that both of these factors are recruited to cyclobutane pyrimidine dimer (CPD).⁶⁵ It is therefore possible that UV-microirradiation techniques, which also induce CPDs and other DNA lesions, show HP1 recruitment, whereas in our system, HP1 is not found in NCs.

PML protein has been also reported to relocalize to nucleolus after cellular stress. PML forms a nucleolar necklace near NCs in response to UV-C irradiation or ActD-induced nucleolar silencing.^66,67 PML protein itself has been shown to interact with hundreds of factors and to participate in DDR.⁶⁸ In the context of this article, it is important to mention its interaction with the MRN complex as well as with HDAC1.^69,70 In this study, we show that DNA damage targeted to rRNA genes also triggers this PML response in mouse embryonic stem cells, which is in agreement with aforementioned articles.

In this article, we show that there are similarities as well as differences between ActD-induced and I-PpoI-induced NCs. To the list of DDR factors previously reported to accumulate in or near NCs (such as BRCA1 or 53BP1), we add MDC1 and, importantly, NBS1. We report that both exogenous and endogenous NBS1 can be found in UBF-positive NCs. We identify HDAC1 and CARM1 as TCOF1 interactors after rRNA gene cleavage. Finally, we show that a number of chromatin remodelers as well as HP1 are not found in DNA damage–induced NCs when analyzed by immunofluorescence. Our results indicate that DDR factors that are dependent on the γH2Ax amplification pathway can be readily detected in NCs, whereas chromatin remodelers that might be present close to the site of DNA damage cannot be identified in this system, which is probably due to its low resolution. Our data support the notion that nucleolar reorganization has functional significance, which is at least 2-fold. First, relocalization of rRNA genes to nucleolar periphery might serve to expose this locus to a microenvironment that is more effective in DNA repair, and second, relocalization of nucleolar proteins such as Fibrillarin, TCOF1, and UBF opens novel interaction possibilities for these factors.