report
Report RePort
Cell Cycle 8:22, 3750-3769; November 15, 2009; © 2009 Landes Bioscience
High resolution imaging of changes in the structure and spatial organization of chromatin, γ-H2A.X and the MRN complex within etoposide-induced DNA repair foci Graham Dellaire,1,* Rosemarie Kepkay1 and David P. Bazett-Jones2,* Departments of Pathology and Biochemistry and Molecular Biology; Dalhousie University; Halifax, NS CA; 2The Hospital for Sick Children; Toronto, ON Canada
1
Key words: DNA repair foci, chromatin, γ-H2A.X, MRE11, electron microscopy Abbreviations: DDR, DNA damage response; DSB, double-strand break; ESI, electron spectroscopic imaging; LM, light microscopy; MRN, MRE11/Rad50/NBS1; NHDF, normal human diploid fibroblast; PCF, pair correlation function; PCCF, pair cross-correlation function; PML NB, promyelocytic leukemia nuclear body
The focal accumulation of DNA repair factors, including the MRE11/Rad50/NBS1 (MRN) complex and the phosphohistone variant γ-H2A.X, is a key cytological feature of the DNA damage response (DDR). Although these foci have been extensively studied by light microscopy, there is comparatively little known regarding their ultrastructure. Using correlative light microscopy and electron spectroscopic imaging (LM/ESI) we have characterized the ultrastructure of chromatin and DNA repair foci within the nuclei of normal human fibroblasts in response to DNA double-strand breaks (DSBs). The induction of DNA DSBs by etoposide leads to a global decrease in chromatin density, which is accompanied by the formation of invaginations of the nuclear envelope as revealed by live-cell microscopy. Using LM/ESI and the immunogold localization of γ-H2A.X and MRE11 within repair foci, we also observed decondensed 10 nm chromatin fibers within repair foci and the accumulation of large non-chromosomal protein complexes over three hours recovery from etoposide. At 18 h after etoposide treatment, we observed a close juxtapositioning of PML nuclear bodies and late repair foci of γ-H2A.X, which exhibited a highly organized chromatin arrangement distinct from earlier repair foci. Finally, the dual immunogold labeling of MRE11 with either γ-H2A.X or NBS1 revealed that γ-H2A.X and the MRN complex are sub-compartmentalized within repair foci at the sub-micron scale. Together these data provide the first ultrastructural comparison of γ-H2A.X and MRN DNA repair foci, which are structurally dynamic over time and strikingly similar in organization.
Introduction Genetic integrity relies on the efficient detection and repair of DNA lesions. Cancer and genomic instability arise from a compromised ability to repair DNA.1-3 Among the common lesions that occur in DNA, DNA double-strand breaks (DSBs) are particularly mutagenic, contributing to the formation of chromosomal rearrangements including translocations, inversions, insertions and deletions.4,5 DNA DSBs are repaired by both nonhomologous end-joining (NHEJ) and homologous recombination (HR) and many of the factors involved in these pathways are highly conserved from yeast to man.6-8 During DNA repair the cell must coordinate lesion detection with a transient arrest in cell cycle progression and the activation of DNA repair machinery.8 DNA within the eukaryotic nucleus is packaged with histones to form chromatin9 and lesion detection
must occur in this context.10 A number of histone modifications have been characterized, including phosphorylation, methylation, acetylation, sumoylation and ubiquitinylation, which provide landmarks for the recruitment of chromatin remodelling factors and DNA repair proteins.11-13 These modifications include the transient acetylation of histone H4,14 which may serve to facilitate the relaxation of chromatin structure in order to allow greater access to the DNA lesion, and phosphorylation of the histone variant H2A.X (γ-H2A.X), which occurs rapidly within chromatin surrounding a DNA break and spreading as far as a megabase away from the lesion.15 The principal mediators of γ-H2A.X phosphorylation in response to DNA breaks are the phosphatidylinositol-3 kinase-like family of kinases (PI3), which include: ataxia telangiectasia mutated (ATM), ATM- and Rad3related (ATR), ATM related kinase (ATX) and DNA dependent protein kinase (DNA-PK).16
*Correspondence to: Graham Dellaire and David P. Bazett-Jones; Email:
[email protected] and
[email protected] Submitted: 07/31/09; Accepted: 09/14/09 Previously published online: www.landesbioscience.com/journals/cc/article/10065 3750
Cell Cycle
Volume 8 Issue 22
Report
Report
The cytological manifestation of DNA DSB formation is the focal accumulation of γ-H2A.X in so-called “repair foci,” whose number correlates linearly with the number of DNA breaks.17 Although the function of these γ-H2A.X repair foci in DNA repair is still under intense investigation, these structures are believed to play a role in the association and/or maintenance of DNA repair factors in the vicinity of damaged chromatin including Brca1, NBS1, MDC1 and 53BP1.18,19 NBS1 is a member of the MRE11/Rad50/NBS1 (MRN) complex that is thought be key in the detection and signaling of DNA damage,20 particularly through the activation of ATM.21 MDC1 is recruited to DNA breaks through direct binding to phosphorylated S139 of γ-H2A.X and the recruitment of MDC1 to DNA breaks is important for the co-accumulation of other cellular factors in the DNA damage response (DDR) including NBS1, 53BP1 and the phosphorylated form of ATM kinase.19 Both MDC1 and proteins of the MRN complex form ionizing radiation-induced foci or “IRIF.”19,22 Interestingly, MDC1 can also recruit and help maintain MRN complex proteins at IRIF through phospho-dependent interaction between the N-terminus of MDC1 and NBS1.23-25 The importance of the focal accumulation of these DNA repair factors has recently been highlighted by the fact that sequestration of NBS1, MRE11 or MDC1 to chromatin in the absence of DNA damage is sufficient to activate the DDR resulting in phosphorylation of H2A.X in an ATM and DNA-PK-dependent manner.26 Although γ-H2A.X forms numerous small repair foci immediately after DNA DSB-induction they become larger and less numerous as DNA repair proceeds.15,27 MRN foci become easily discernible from background nuclear staining after 2–3 hours following irradiation, peaking in size 4–8 hours after the induction of DNA damage.22 This is in contrast to the rapid recruitment of the MRN complex to the initial DNA break, which occurs even in the absence of γ-H2A.X.28 Therefore, rather than playing a direct role in repair of DNA DSBs, it has been suggested that large MRN repair foci may function in either DNA damage signalling and checkpoint control, or serve to mark un-repaired DNA lesions.22,28 It is also at these later time points following DNA damage that MRN and γ-H2A.X repair foci extensively overlap. Although recent studies by light microscopy have begun to address the mechanism responsible for repair foci formation and dynamics,27,29-33 the ultrastructure of the macromolecular complexes involved in repair, the degree of chromatin condensation or compaction and the spatial relationships of the repair complexes with chromatin, cannot be easily addressed by light microscopy and remain largely uncharacterized. In addition to the formation of DNA repair foci, a number of studies have described large-scale changes in chromatin structure following diverse forms of DNA damage including UV,34 ionizing radiation and chemotherapy agents.35,36 In these studies, either light microscopy observation of DNA density or the indirect biochemical analysis of chromatin organization by nuclease sensitivity was used to support the conclusion that chromatin within the nucleus undergoes relaxation or decondensation in response to DNA damage. However, light microscopy observations of DNA density can prove misleading with regard to actual changes in chromatin fiber organization due to the lack of spatial resolution
www.landesbioscience.com
required to observe individual 10 nm or higher order fibers, for example using DAPI or Hoechst-staining.37 Biochemical analysis of nuclease sensitivity on the other hand, does not provide spatial information on the packaging of chromatin beyond the nucleosomal level and would not be appropriate for the analysis of DNA repair foci in situ. Therefore, we chose to use correlative light microscopy and electron spectroscopic imaging (LM/ ESI)38 to analyse in situ the ultrastructure of repair foci within the nuclei of normal human diploid fibroblasts in relation to DNA damage-induced changes in chromatin organization. Our results demonstrate at the ultrastructural level that the density of chromatin within the nucleus is reduced in response to etoposide-induced DNA DSBs, consistent with the global relaxation of chromatin. We also document for the first time a direct correlation between the induction of DNA damage and invagination of the nuclear envelope. Finally, through the analysis of DNA repair foci by LM/ESI combined with the immunogold localization of γ-H2A.X and the MRN proteins, we revealed the decondensation of chromatin into extended 10 nm fibers and the accumulation of non-chromosomal protein complexes within repair foci over time. Together these data provide the first ultrastructural comparison of γ-H2A.X and MRN DNA repair foci, which share surprising heterogeneity and structural changes over time. Results DNA DSB-induction by etoposide causes a global relaxation of chromatin structure in concert with alterations in nuclear shape. We have demonstrated that LM/ESI could be used to monitor changes in chromatin organization within tracks of UV laser-induced DNA damage.32 Chromatin decondensation seen by light microscopy within UV laser tracks correlated well to changes in chromatin density measured morphologically by direct observation of chromatin fiber density with line-scan analysis across the laser track area in mass-sensitive energy-filtered electron micrographs. We applied this technique to the analysis of global changes in chromatin structure within nuclei of normal human diploid skin fibroblasts (NHDFs) following DNA damage induced by the etoposide VP16 (Fig. 1), which creates DNA DSBs via inhibition of topoisomerase II.39 Although cells are generally more sensitive to cell killing by etoposide in mid-S and early G2 phases of the cell cycle, DNA breaks are created in cells throughout the cell cycle presumably due to the role of topoisomerase II in gene transcription as well as DNA replication.40,41 In these experiments we employ a pulsed dose of etoposide (20 μM, 0.5 h), which we have previously shown is equivalent to a single dose of ~2 Gy of ionizing radiation (i.e., ~70 DNA DSBs) using both neutral comet and clonagenic survival assays.42 Prior to DNA damage, a NHDF nucleus exhibits a relatively smooth nuclear lamina and is oval or kidney shaped (Fig. 1A). In contrast, 3 h after recovery from etoposide treatment the nuclear lamina appears jagged and there is evidence of one or more deep invaginations of the nuclear envelope that extend into the nucleus either in parallel (horizontal nuclear invaginations; HNI) (Fig. 1B, yellow arrow) or perpendicular (vertical nuclear invaginations; VNI) to the growth substrate (Fig. 1B, white asterisks).
Cell Cycle
3751
Figure 1. For figure legend, see page 3753.
Line-scan analysis of the mass-sensitive energy-filtered electron micrographs of control versus etoposide treated nuclei demonstrated a reduced peak to background ratio in mass-density in DNA damaged nuclei, as well as a loss of the higher frequency peaks of mass-density. Changes in these parameters contribute to a significantly reduced mean coefficient of variation of the pixel
3752
intensity and thus mass-density of chromatin from 32 ± 1% for control nuclei to 19 ± 2% for etoposide treated nuclei (Fig. 1C; n = 20, p < 0.001). We also analysed the area of the nucleus occupied by either chromatin (yellow) or non-chromosomal protein complexes (cyan) within a single optical sections from control cells versus
Cell Cycle
Volume 8 Issue 22
Figure 1. Large scale ultrastructural changes in chromatin organization in response to DNA double-strand breaks induced by etoposide. (A and B) Normal human fibroblasts (GM05757) were treated with (B) or without (A) etoposide (VP16, 20 μm) for 0.5 h and left to recover for 3 h prior to fixation and processing for LM/ESI. Phosphorus-enhanced (155 eV) ESI micrographs are shown for both a control (A) and etoposide-treated normal human fibroblast (B) and a line-scan (between white arrow heads left to right) of the pixel intensity of phosphorus signals is shown below each image. White asterices indicate invaginations of the nuclear envelope that are perpendicular to the growth substrate (vertical nuclear invaginations; VNI) and yellow arrows indicate deep invaginations that are parallel to the growth substrate (horizontal nuclear invaginations; HNI). Scale bars = 5 μm. (C) Analysis of the coefficient of variation of mean phosphorus pixel intensities for control and etoposide-treated normal human fibroblast cells (n = 20; ±SE) *p < 0.001. (D) Untreated (Control) normal human fibroblasts (GM05757) or those treated with the etoposide VP16 (as in A and B) were fixed and processed for LM/ESI. Nitrogen (N, grayscale) and phosphorus (P, yellow) elemental ESI micrographs are shown at the left for control and treated fibroblasts. In the centre panel non-chromosomal protein complexes are distinguished by subtracting the nitrogen from the phosphorus elemental map (N-P) and this image is combined with the phosphorus elemental map to allow the visualisation of chromatin (yellow) from nonchromosomal protein complexes (cyan, red arrows) in the adjacent panel at the right (Merged, N-P/P). A PML nuclear body (PML NB) is indicated in the Control panels as a prototypical example of a non-chromosomal protein complex. The panel at the far right depicts the segmentation of pixels within the field of view that correspond only to chromatin (Segmented Chromatin, white). Scale bars = 500 nm. (E) Analysis of the average area of the nucleus occupied by either chromatin or non-chromosomal protein in control untreated normal human fibroblasts (GM05757) or those treated with the etoposide VP16 (as in A and B). Areas are depicted as percentages of the total area (n = 8) ± S.D. *p < 0.01.
cells treated with etoposide for 0.5 h or 3 h post-treatment (Fig. 1D). We observed a progressive drop in the fraction of the nuclear volume occupied by chromatin following the induction of DNA DSBs, as inferred from the area of the nucleus occupied by chromatin in control cells (21 ± 3%) versus chromatin in cells after 0.5 h of etoposide treatment (14 ± 2%; p < 0.01), and 3 h posttreatment (11 ± 2%; p < 0.01) (Fig. 1E; n = 8). Although we did not see a difference between the area of the nucleus occupied by non-chromosomal protein in control cells versus cells treated with etoposide for 0.5 h (both 7 ± 2%), a significantly greater accumulation of non-chromosomal protein complexes was observed after 3 h recovery from etoposide treatment (12 ± 1%; p < 0.01). Since chromatin density changes are accompanied by increased areas devoid of chromatin but enriched in non-chromosomal protein (Fig. 1D; 3 h recovery), it appears that this protein also contributes to the homogenization of mass-density seen in the low magnification mass-sensitive electron micrographs (Fig. 1A). Therefore, it is clear our mass-density measurements are detecting mass-changes in both chromatin and non-chromosomal protein within DNA damaged cells, and only by high resolution analysis of chromatin ultrastructure by ESI (Fig. 1D) can we interpret these mass-density changes as relaxation or decondensation of chromatin. Similar changes in the mean coefficient of variation of chromatin density and in the area of the nucleus occupied by chromatin were observed with ionizing radiation (data not shown) and corroborate our previous analysis of the decondensation of chromatin within tracks of UV-laser-induced DNA damage.32 Invaginations of the nuclear envelope correlate with DNA DSB-induction by etoposide. To further investigate the association between DNA damage and invagination of the nuclear envelope, we carried out immunofluorescence microscopy on fixed cells (Suppl. Fig. S1A) as well as live cell analysis of NHDFs following treatment with etoposide (Fig. 2). Although invaginations of the nuclear envelope are observed in a subset of untreated cells, we readily observed an overall increase in both vertical and horizontal nuclear invaginations (VNI and HNI, respectively) following etoposide treatment by light microscopy using lamin A/C immunostaining (Suppl. Fig. S1A). Direct observation of VNI and HNI by ESI, which is more sensitive for the detection of invaginations than fluorescence microscopy, demonstrated
www.landesbioscience.com
that 78% of control cells exhibited VNI (1.0 ± 0.3 VNI/cell) and 44% exhibited HNI (0.4 ± 0.2 HNI/cell) (Fig. S1B) (Table 1; n = 10). HNI were quite variable in morphology at the ultrastructural level, extending various distances into the nucleus (Suppl. Fig. S1B, white arrows). Therefore, for our analyses we only scored HNI that extended at least 50 nm into the nucleus. After 0.5 h of etoposide treatment a significant increase in VNI/cell was seen compared to control cells (100%; 3.8 ± 0.5 VNI/cell (p < 0.002)), whereas HNI remained less frequent (44%; 0.7 ± 0.05 HNI/cell) (Table 1). In contrast, after 3 h recovery from etoposide treatment HNI became both more prevalent with 78% of cells exhibiting HNI by ESI and significantly increased in overall number/cell (3.3 ± 0.8 HNI/cell (p < 0.02)); and although VNI were present in 100% of cells, their number per cell slightly decreased (2.7 ± 0.5 (p < 0.03 versus control)). More strikingly, when we visualized the nuclear lamina and histones within NHDFs transfected with GFP-laminB1 and mCherryhistone H1.4 (respectively) by live cell spinning-disk confocal microscopy, we observed the rapid onset of invaginations within 6–8 min after the addition of etoposide to the growth medium (Fig. 2 and Suppl. Movie S1). Continued evaluation of the etoposide treated cells by live-cell microscopy for 4 h indicated that the process of nuclear invagination is highly dynamic and that the number and size of invaginations changes over time. In particular, the most dramatic changes in the nuclear envelope appear to occur in the first 30 min, followed by a reduction in invaginations and almost full recovery of normal nuclear lamina morphology by 60 min. Between 90 min to 4 h after etoposide treatment, a progressive furrowing of the nuclear lamina and large HNI began to form, consistent with the increase in HNI seen by ESI at 3 h. The observed invaginations of the nuclear envelope were not due to imaging or culturing conditions as imaging of NHDFs in the absence of etoposide for 4–6 hours did not induce changes in the nuclear envelope (data not shown). Together these analyses indicate that the formation of nuclear invaginations correlates with the induction of DNA DSBs by etoposide. Chromatin decondensation and the accumulation of nonchromosomal complexes occur within repair foci in concert with the redistribution of γ-H2A.X over time. Although DNA repair foci containing γ-H2A.X have been extensively characterized by light microscopy, little is known about temporal changes in
Cell Cycle
3753
Figure 2. Live cell analysis of nuclear lamina dynamics in normal human fibroblasts treated with etoposide. (A–C) Normal human fibroblasts (GM05757) were transfected with GFP-LaminB1 and mCherry-histone H1.4 and left to recovery for 24 h prior to treatment with etoposide (VP16, 40 μm) and visualisation by live-cell microscopy using a spinning-disk confocal microscope. The morphology of the nuclear lamina, as indicated by the GFP-LaminB1 fluorescence, is shown prior to addition of etoposide (A). After the addition of etoposide, cells were initially imaged every two min for 52 min (B), and then every 30 min from 90 min until 240 min (i.e., 4 h). Folds and/or invaginations of the nuclear envelope are indicated by white arrow heads in the first frame they appear. Scale bars = 5 μm.
Table 1. Analysis of the incidence of nuclear invaginations in control versus etoposide treated cells over time Control
0.5 h VP16†
3 h Recovery VP16†
VNI
HNI
VNI
HNI
VNI
HNI
Number/Cell
1.0 ± 0.3
0.4 ± 0.2
3.8 ± 0.5
0.7 ± 0.3
2.7 ± 0.5
3.3 ± 0.8
% of Cells
78%
44%
100%
44%
100%
78%
