Progerin impairs 3D genome organization and induces fragile telomeres by limiting the dNTP pools

Chromatin organization within the nuclear volume is essential to regulate many aspects of its function and to safeguard its integrity. A key player in this spatial scattering of chromosomes is the nuclear envelope (NE). The NE tethers large chromatin domains through interaction with the nuclear lamina and other associated proteins. This organization is perturbed in cells from Hutchinson–Gilford progeria syndrome (HGPS), a genetic disorder characterized by premature aging features. Here, we show that HGPS-related lamina defects trigger an altered 3D telomere organization with increased contact sites between telomeres and the nuclear lamina, and an altered telomeric chromatin state. The genome-wide replication timing signature of these cells is perturbed, with a shift to earlier replication for regions that normally replicate late. As a consequence, we detected a higher density of replication forks traveling simultaneously on DNA fibers, which relies on limiting cellular dNTP pools to support processive DNA synthesis. Remarkably, increasing dNTP levels in HGPS cells rescued fragile telomeres, and improved the replicative capacity of the cells. Our work highlights a functional connection between NE dysfunction and telomere homeostasis in the context of premature aging.

Telomeres are specialized nucleoprotein complexes capping the ends of linear chromosomes. They consist of repetitive nucleotide sequences (-TTA GGG -in mammals) that end with a 3′-overhang, and are shielded by a specific protein complex called shelterin 1,2 . The primary function of telomeres is to delineate chromosome ends and avoid illicit DNA repair that would result in chromosome fusions and subsequent genomic instability and aneuploidy 3,4 . Shelterin plays a major role in repressing the DNA damage response at chromosome ends. Consistently, the depletion or the loss of function of shelterin components leads to the activation of ATM-or ATRdependent DNA damage responses, cell cycle arrest, and chromosome instability 5 . Another essential function of telomeres is to protect chromosome ends from DNA resection occurring after each round of replication due to the end replication problem. Therefore, telomeres from human somatic tissue shorten during log-phase cell growth, leading to a progressive change in their structure that entails replicative senescence 4,6 . Restricting the growth of aged cells serves as a tumor suppressor mechanism, but also contributes to cellular and organismal aging 7 . Telomere shortening can be counteracted by the activity of telomerase, the enzyme responsible for telomere elongation. In telomerase-positive cells such as stem cells and germline cells, telomeres are maintained to a stable length, bypassing senescence and resulting in cellular immortalization 8 .
The connection between telomere length and human organismal aging is highlighted by a large panel of age-related human syndromes associated with short telomeres and signs of premature senescence 9 . Premature aging disorders are usually caused by mutations impairing the DNA damage response and repair pathways (Ataxia Telangiectasia), shelterin complex structure (Dyskeratosis Congenita), and telomere length regulation (Dyskeratosis Congenita, Bloom and Werner Syndromes). These telomere-related syndromes share overlapping traits with human syndromes linked to lamins, grouped under the term of laminopathies 10 . In metazoans, the lamina forms a thin meshwork lining the inner side of the nuclear envelope (NE) and is composed of a family of Scientific Reports | (2021) 11:13195 | https://doi.org/10.1038/s41598-021-92631-z www.nature.com/scientificreports/ proteins consisting of A-type and B-type lamins. In addition to a structural function of the lamina in maintaining the shape and the mechanical properties of the nucleus, there is growing evidence suggesting that nuclear lamins also play a major role in genome organization and stability 11 . One of the most severe laminopathies, Hutchinson-Gilford progeria syndrome (HGPS), is predominantly caused by a de novo point mutation G608G (nucleotide 1824 C > T) in exon 11 of the LMNA gene encoding both LaminA and C isoforms. This leads to the expression of an aberrantly spliced and processed protein termed progerin 10,12 , causing premature aging at the organismal level. Compared to normal fibroblasts, the growth potential of cultured HGPS fibroblasts is strongly impaired. They also display a low but persistent activation of DNA damage checkpoints, and carry short telomeres [13][14][15][16][17][18][19] . One key feature of HGPS cells is their abnormal nuclear architecture, with lobulation of the NE, thickening of the nuclear lamina, and defects in nuclear pore organization 20 . The NE represents a main organizer of chromatin within the 3D nuclear space. During the last decade, it has become clear that the position of chromatin in the nucleus impacts genome stability and gene regulation 21,22 . Large heterochromatin domains are anchored to the nuclear lamina to form the so-called LADs (lamina-associated domains), regions that are essentially transcriptionally repressed and known to replicate late [23][24][25] . The nuclear architecture defects observed in progerin-expressing cells have dramatic consequences on chromatin organization, with a loss of some heterochromatin-lamina domains, decreased interactions within heterochromatin domains, and epigenetic alterations in LADs [26][27][28] . How these changes in 3D chromatin organization account for the proliferative defects of HGPS cells remains unclear. Telomere maintenance is a key factor, as telomere elongation by telomerase expression prevents progerindependent growth defects 17,29 . In fact, several lines of evidence indicate that telomeres and lamins are directly connected, and that nuclear substructures impact telomere function and stability. In interphase cells, human telomeres are organized close to the center of the nucleus 30,31 , but a few late-replicating telomeres are anchored to the NE 32,33 . In contrast, a large subset of telomeres is tethered to the NE during post-mitotic nuclear assembly 34 , in close proximity to Lamin B1 35 . Connection between telomeres and lamins also occurs in the nucleoplasm and plays a role in telomere protection 36 .
In this study, we further explored the interconnection between telomere maintenance, lamin-dependent nuclear organization, and aging. We demonstrate that progerin expression perturbs telomere 3D organization and alters their chromatin state. Telomere replication is challenged, leading to an accelerated telomere shortening that limits cell growth. We also linked the replication defect to nuclear organization by studying the replication timing (RT) signature of progerin-expressing cells. A significant fraction of the genome indeed shifted to earlier replication, in agreement with the loss of peripheral heterochromatin that characterizes HGPS. Finally, we uncovered that dNTP pools seem limiting in progerin cells, as supplementation with nucleosides restored their replication competence and alleviated the telomere replication stress. We propose that an alteration of the 3D nuclear architecture of HGPS cells modifies their replication timing program, leading to a higher density of active replication forks that limits dNTPs availability and promotes fork stalling. We show that telomeres are particularly sensitive to this replication defect, and the resulting short telomeres trigger premature growth arrest. Western blots of whole cell extracts of HDF cells expressing the indicated constructs. Two independent transductions are shown (TR1 and TR2). The membrane was probed with the LaminA/C antibody to detect endogenous LaminA (A), endogenous LaminC (C) and the EGFP-fusion proteins EGFP-LA (E-A) and EGFP-PG (E-P). Actin is shown as a loading control. (B) Live HDF cells expressing EGFP-LA or EGFP-PG, 3 or 6 days post-transduction. The EGFP signal (grey) is detected using a benchtop fluorescent microscope. (C) Representative immunostaining of HDF cells expressing the indicated constructs. EGFP (cyan), TRF1 (yellow), and merge. The max projection is shown. Scale bar: 10 μm. (D) Scheme of MadID-dotblot principle. Genomic DNA is extracted from cultured cells, and digested with frequent cutter restriction enzymes that do not cut the telomeres. Telomeres are isolated using biotinylated oligonucleotides complementary to the TTA GGG telomeric sequence and streptavidin beads. The purity and the amount of isolated telomeres are verified by qPCR, before they are blotted on a membrane for hybridization with a m6Aspecific antibody. (E) Dotblot of increasing amount of lambda DNA isolated from Dam− (M−) or Dam+ (M+) bacterial strains. The membrane is hybridized with a m6A-specific antibody. The m6A signal intensity in the M+ DNA is indicated on the right, relative to the signal obtained with 75 ng of lambda DNA. (F) qPCR analyses of MadID using TelC/TelG primers to amplify telomeric DNA. Left panel: a fragment of 800 bp of telomeric repeats was used as a template, before or after in vitro methylation with M.EcoGII. Right panel: isolated telomeres from HDF expressing NLS-EGFP, EGFP-LA, or EGFP-PG as indicated. Enrichments are shown for three independent telomere purifications. Fold enrichments are calculated against INPUTs, i.e. genomic DNA. (G) Representative example of a MadID-dotblot experiment performed on HDF cells expressing NLS-EGFP, EGFP-LA and EGFP-PG. INPUT DNA and isolated telomeres were blotted on a membrane hybridized with a m6A-specific antibody. (H) Quantification of 3 independent MadID experiments similar to (E). The graph represents the normalized m6A intensity (normalized to the amount of isolated telomeres measured by qPCR) relative to the EGFP-NLS control. Mean with SD is shown, n = 3. Statistical significance was determined using one-way Anova (**indicates p < 0.01). (I) ChIP analyses on HDF cells expressing NLS-EGFP, EGFP-LA, and EGFP-PG. Immunoprecipitations were performed with the indicated antibodies. The graphs represent the enrichment relative to input material (left) or to H3 (right) (Mean ± SD, n = 4). Statistical significance was determined using two-way Anova (*indicates p < 0.05, **indicates p < 0.01, ***indicates p < 0.001, ****indicates p < 0.0001).

Results
Progerin impairs 3D-telomere organization in human primary skin fibroblasts. The impact of A-type lamins on telomere organization was previously investigated in Lmna −/− mouse cells, where a significant change in distribution of telomeres toward the nuclear periphery was observed 37 . We hypothesized that the expression of progerin in HGPS cells may also affect human telomere organization and be detrimental for their maintenance. To model the physiopathology of HGPS fibroblasts, we established normal Human Dermal Fibroblasts (HDF) cell lines stably expressing EGFP-tagged human LaminA∆50/progerin (EGFP-PG), and used wild type LaminA (EGFP-LA) or a nuclear localized EGFP (NLS-EGFP) as controls (Fig. 1A, Supplemental  Fig. S1A). Within 3 days post-transduction, progerin expression led to strongly misshapen nuclei (Fig. 1B). In contrast, wild-type LaminA was also properly integrated into the lamina but did not affect nuclear morphology. Some nuclei from EGFP-LA cells displayed additional foci in the nucleoplasm, which correspond to LaminA precursors before their processing 38 . EGFP-PG expression resulted in previously reported phenotypes such as mislocalization of the protein SUN1 39 , as well as a decreased expression of LAP2alpha, H3K9me3, H3K27me3, and LaminB1 40 (Fig. S1B-E). Expression of EGFP-LA also impacted LAP2alpha levels (Fig. S1C,D) as previously described 41 . To assess the impact of progerin on telomere distribution in the nuclear volume, we imaged telomeres by immunostaining of the TRF1 shelterin protein (Fig. 1C). However, the extent of NE blebbing with deep membrane invaginations reaching the nucleus center made it difficult to evaluate telomere localization with regards to the nuclear lamina by microscopy. To solve this issue, we applied MadID, a proximity labeling technique that we recently developed to probe telomere-NE contact sites 35,42 . MadID relies on the expression of a bacterial methyltransferase called M.EcoGII, which catalyzes the methylation of adenine residues (m6A). Fused to a protein of interest, adenines within and in the vicinity of the protein binding sites will be decorated with m6A, a modification that can be easily detected using a m6A-specific antibody. We previously fused M.EcoGII to LaminB1 to probe chromatin-NE contact sites, and were able to map LADs with high specificity, resolution and genome coverage 35 . This experiment also revealed telomeres and NE contact sites in normal and cancer cells. We therefore performed MadID-dotblot as previously described 42 (Fig. 1D). Briefly, intact telomeric repeats were released by restriction enzymes digestion and purified using biotin-(CCC TAA ) 3 oligos and streptavidin-coated magnetic beads. Input DNA and purified telomeric DNA samples were next blotted and probed with a m6A-specific antibody. We confirmed the dynamic range of our assay by blotting increasing amounts of lambda DNA obtained from dam+ (with methylation) or dam− (no methylation) E. coli strains. The resulting m6A signal was specific and proportional to the amount of DNA (Fig. 1E). To implement MadID in our setting, M.EcoGII-LaminB1 was expressed using an inducible retroviral vector in our established HDF cell lines. We confirmed the proper expression of M.EcoGII-LaminB1 fusion protein 24 h after induction, with a level of induction comparable in HDF NLS-EGFP, EGFP-LA, and EGFP-PG (Fig. S1F). These cells were used to purify telomeric DNA. The efficiency of telomere capture and purity was verified by qPCR, based on an amplification method previously established to measure telomere length 43 . A known concentration of a fragment of 800 bp of TTA GGG repeats was used to verify that qPCR efficiency was close to 100%. This fragment was also in vitro methylated by recombinant M.EcoGII 35 in order to assess the impact of m6A residues on PCR amplification. We did not detect any differences in fold enrichment with or without m6A methylation, confirming that qPCR is a suitable approach to assess the amount of purified telomeres in HDF samples, regardless of their methylation status ( Fig. 1F-left panel). We successfully purified telomeres from our HDF samples ( Fig. 1F-right panel) and confirmed that they did not contain detectable genomic DNA using single copy gene amplification 43 (Fig. S1G). Analysis of the resulting blots revealed that M.EcoGII-LaminB1 was able to methylate genomic DNA in NLS-EGFP, EGFP-LA and EGFP-PG cells (Fig. 1G, see Input samples). m6A signal was also detectable on isolated telomeres for all three samples, which indicates that a subset of telomeres contact the nuclear lamina in asynchronous HDF cells 35 . Interestingly, progerin expression induced a 1.5 fold increase in m6A signal at telomeres ( Fig. 1G,H). This result suggests that telomeres contact the lamina more frequently in progerin-expressing cells. Changes in 3D-chromatin environment could impact telomere chromatin structure, such as the establishment of heterochromatin marks. This is particularly relevant in the context of HGPS patients, since HGPS fibroblasts show an alteration of constitutive heterochromatin histone marks. We carried out ChIP experiments using H3K9me2 and H3K9me3 specific antibodies, as well as TRF1 and H3 as internal controls. We observed a marked decrease of H3K9me2 at telomeres in progerin cells, when signals were normalized to Input or H3 levels (Fig. 1I). Similarly, although not statistically significant, we reproducibly found that levels of H3K9me3 were also lower. Overall, these results indicate that progerin expression affect telomere 3D organization as well as their chromatin state, in line with the global loss of heterochromatin previously reported in HGPS cells. Interestingly, expression of wild-type LaminA also decreased the levels of H3K9me2 and me3 levels at telomeres. This suggests that a modification in the LaminA/Lamin B1 balance is detrimental for chromatin maintenance at telomeres.
Progerin impairs telomere length maintenance and induces premature senescence in primary skin fibroblasts. Next, we aimed to determine whether these changes in 3D telomere organization and chromatin state correlate with defects in telomere length maintenance. HDF cell lines expressing EGFP-NLS, EGFP-LA and EGFP-PG were cultured until they reached their growth plateau. Physiological oxygen conditions (5%) were used to limit oxidative stress that strongly affects the fitness of progerin-expressing cells 44 . Remarkably, although ectopic expression of progerin instantly affected NE integrity (Fig. 1B), several population doublings were necessary to visualize a change in the growth rate (Fig. S2A,B). We consistently observed a growth defect in progerin-expressing cells, together with an early detection of senescent cells that accumulated over time (Fig. S2C). Using southern blots of terminal restriction fragments (TRFs), we first confirmed that ectopic expression of progerin accelerated the rate of telomere shortening 45 ( Fig. 2A). Telomeres shortened at a rate of 100-150 bp/PD in NLS-EGFP and EGFP-LA control HDFs, which is comparable to what has been described in    (Fig. 2B). In contrast, telomeres shortened 1.5-times faster in EGFP-PG cells. Importantly, we did not observe an acceleration of telomere shortening after wild-type LaminA expression, unlike what was previously described 45 . As critically short telomeres are the trigger of senescence entry in primary fibroblasts, these results suggest that the presence of short telomeres in HGPS cells is responsible for their premature senescence. Consistent with this idea, telomere elongation by telomerase expression was shown to rescue the growth defect of progerin-expressing cells 17,29,47,48 . Other studies suggested that telomerase fails to immortalize HGPS cells without the concomitant repression of p53. Indeed, hTERT expressing HGPS cells still reached senescence, while normal fibroblasts were immortalized 49,50 . To investigate this further, we assessed the growth potential and telomere length of HGPS patients skin fibroblasts obtained from Coriell Cell Repositories (see "Method" section for further details). Both HGPS M14 and HGPS F8 cell lines grew very poorly (Fig. S2D), with senescent cells present in the cell population at the onset of culturing (Fig. S2E). HGPS M14 cells presented the strongest growth defects, with up to 75% of senescent cells. Senescence entry was due to checkpoint activation, since suppression of pRB and p53 pathways by expression of the human papilloma virus serotype 16 (HPV16) E6 and E7 oncoproteins allowed HGPS cells to bypass senescence (Fig. S2D,E). We confirmed that initial telomere length was much shorter in both patient cells (6-7 kb) than in control HDFs from healthy donors (< 9 kb) 13,16,48 (Fig. 2C). In our hands, expression of the catalytic subunit of telomerase hTERT rescued the growth potential of these HGPS cell lines, even in the presence of functional checkpoints (Fig. S2D). hTERT also increased the percentage of BrdU positive cells (Fig. S2F,G). The growth advantage driven by telomerase was stronger for HGPS M14 compared to HGPS F8, with an increase from 9% to about 18% of S phase cells after hTERT expression, in agreement with the greater growth defect initially observed and higher percentage of senescence cells. Interestingly, hTERT expression in HGPS M14 cells also increased laminB1 levels, known to decrease upon senescence ( Fig. S2H). Shortly after hTERT expression, telomere length increased in both HGPS cells, with an efficiency of telomere elongation comparable to HDF (Fig. 2D,E). Telomere length was then stabilized similarly in patients and healthy cells, suggesting that progerin did not impair the regulation of telomere elongation by telomerase 51 .
Telomere length changed upon E6-E7-dependent checkpoint inactivation in M14 cells, with a wider range of size reaching very short (2 kb) and very long (more than 10 kb) telomeres ( Fig. 2C-F). These results indicate that critically short telomeres are a trigger for growth arrest in progerin-expressing cells, and their elongation is sufficient to prevent checkpoint-induced senescence.

Progerin expression induces fragile telomeres in human cells.
While the impact of progerin expression on telomere length was previously suggested, the mechanism of action is unknown. To address this key question, we monitored telomere integrity in EGFP-PG, EGFP-LA and NLS-EGFP cells. First, we observed that the expression of the six subunits of the shelterin complex in whole cell extracts was unaffected by ectopic progerin expression (Fig. 3A, Supplemental Fig. S3A). In addition, ChIP experiments confirmed that both TRF1 and TRF2, the core shelterin proteins that directly bind double stranded TTA GGG repeats, were correctly addressed at telomeres in EGFP-PG expressing cells (Fig. 3B, Supplemental Fig. S3B). Telomere integrity was therefore analyzed on metaphase spreads prepared from HGPS patient cells or HDF cells expressing NLS-EGFP, EGFP-LA and EGFP-PG. As HGPS patient cells have a low mitotic index, cells were pre-synchronized by a double thymidine block, released for 7 h and treated with colcemide to arrest them in metaphase by microtubule depolymerization. To visualize telomeres, metaphase chromosomes were stained with a telomere specific PNA FISH probe fused to FITC. Telomere aberrations, such as telomere loss, sister chromatid fusions, telomere fragments, and end-to-end telomere fusions were screened using this method but no significant difference was observed compared to control cells (Fig. S3C). However, fragile telomeres, telomeric regions where replication is impaired and that are visualized as multi-telomere signals (MTS) 52,53 , were found at both progerin-expressing HGPS cell lines (Fig. 3C,D). Interestingly, this phenotype was more pronounced in HGPS M14 compared to HGPS F8, indicating patient-to-patient variability and a possible link to the age of the donor. We also found a significant increase of MTS in HDF expressing EGFP-PG compared to controls (Fig. 3D). These results indicate that telomere replication is affected in progerin-expressing cells, which explain their length maintenance defect and the rescue by telomerase elongation. This prompted us to assess the presence of ultrafine anaphase bridges (UFBs), residual secondary structures observed in anaphase that must be processed to ensure faithful chromosome segregation. One class of UFBs arises from under-replicated intermediate structures, often associated with fragile sites at genome regions difficult to replicate. UFBs can be identified indirectly by immunostaining of the resolvase PICH that binds UFBs to promote their dissolution at the onset of mitosis 54 . NLS-EGFP, EGFP-LA and EGFP-PG HDF were synchronized with a single thymidine block to enrich the cell population in anaphase cells. At this stage, the diffuse EGFP staining reflected NE breakdown that occurs during open mitosis and releases lamins (Fig. 3E). PICH staining was also diffuse in cells with no UFBs, but formed thread-like structures when UFBs were present (Fig. 3E). We found that the number of UFBs per anaphase was significantly higher in progerin cells compared to controls (Fig. 3F). Altogether, our results point to a DNA replication defect triggered by progerin expression that affects telomere replication.

Progerin expression sensitizes cells to replication stress. Replication defects impact DNA integrity
and induce endogenous checkpoint activation and DNA damage responses. Several studies reported an increase of the DNA damage markers γH2AX and 53BP1 in mouse models defective for LaminA processing, in HGPS cells, and in normal cells following ectopic expression of progerin 15,17,19,44 . This was observed in the absence of exogenous DNA damage, suggesting that A-type lamin integrity can affect chromatin maintenance. Indeed, we confirmed that endogenous levels of γH2AX were higher in both HGPS cell lines compared to normal HDFs (Fig. 4A,B). To assess the impact of replicative stress on the level of endogenous DNA damage, we performed quantitative image based cytometry (QIBC) after ectopic progerin expression, using EdU incorporation com-  55,56 . QIBC allows to simultaneously assess cell-cycle distribution, replication competence, and DNA damage signaling. γH2AX was found enriched in HDF expressing EGFP-PG compared to controls (Fig. 4C), with a higher intensity in S and G2/M phases of the cell cycle (Fig. 4D). When DNA replication was challenged with 0.2 μM of aphidicolin for 24 h, we observed an accumulation of cells in  www.nature.com/scientificreports/ S phase in all three cell types, but more pronounced for LaminA-and progerin-expressing cells (Fig. S4A). This treatment did not induce a significant increase of γH2AX in EGFP-NLS control cells (Fig. 4C,D). However, γH2AX staining was strongly increased in progerin cells, pointing to their sensitivity to replication stress. The accumulation of γH2AX was evident in S phase, but foci also accumulated in G1 and G2/M, arguing for a transmission of the damage signaling outside of S phase (Fig. 4D). We noticed that expression of wild-type LaminA also perturbed chromatin integrity, although to a lower extent compared to the progerin mutant. Interestingly, scoring 53BP1 spots in these cells yielded in a completely different outcome. Although endogenous levels of 53BP1 foci were detected in HDF NLS-EGFP and EGFP-LA, a drastic drop of the number of foci was consistently observed in EGFP-PG cells (Fig. 4E, Supplemental Fig. S4B-D). These results confirm the defective recruitment of 53BP1 to sites of DNA damage that was previously described in progerin cells 14 . Interestingly, the induction of 53BP1 DNA damage foci was observed after a short term induction of Progerin expression using an inducible expression system 17 . This suggests that Progerin effect on the DNA damage response might require longer expression periods, and is not an immediate effect. In undamaged normal fibroblasts, 53BP1 binds to A-type lamins and this interaction was proposed to serve as a storage to facilitate 53BP1 recruitment to sites of DNA damage 57 . Indeed, in absence of functional LaminA, 53BP1 is degraded by a cysteine protease 58 . Although the pool of functional LaminA is impaired in HGPS patient, it remains unclear how progerin expression impacts 53BP1 and its recruitment to DNA damage sites. To test this, we performed co-immunoprecipitation experiments using EGFP pull-down in HDF NLS-EGFP, EGFP-LA and EGFP-PG, before and after bleomycine treatment to induce DNA damage. In order to improve 53BP1 protein recovery and the sensitivity of our assay, cell extracts were fractionated to separate the cytoplasmic fraction, here enriched in actin, and the nuclear fraction enriched in LaminA/C (Fig. 4F, see INPUTs). After bleomycin treatment, γH2AX was detected in the nuclear fraction, confirming the checkpoint activation following induced DNA damage. Pull-downs were performed on nuclear extracts only, and IP fractions clearly recovered EGFP-LA and EGFP-PG (Fig. 4F, see E-A and E-P bands). Consistent with the fact that LaminA interacts with itself and with progerin, endogenous LaminA/C bands were also detected in the IP lanes (Fig. 4F, see A and C bands). 53BP1 was found in both EGFP-LA and EGFP-PG pull-downs, but not in the control fraction NLS-EGFP, regardless of the bleomycin treatment. This confirms the specificity of the 53BP1-LaminA interaction, and indicates that progerin expression did not prevent this interaction. LaminA-53BP1 interaction is supposed to decrease upon DNA damage, in order to release 53BP1 that can be addressed to sites of damage 57 . We did not detect a clear drop in 53BP1 signal after bleomycin treatment in our LaminA IP fraction. However, progerin IP did retrieve 53BP1 efficiently, suggesting that progerin could sequester the DNA damage marker and prevent its recruitment to sites of DNA damage. Our results suggest that progerin cells are sensitive to drugs challenging DNA replication. A low but persistent DNA damage checkpoint is activated, but the pathway is impaired due to a decreased recruitment of 53BP1. Previous work reported that the DNA damage response in HGPS originates from telomeres 47 . In our hands, DNA damage markers did not colocalize with telomeres using TRF1 co-staining in immunofluorescence, and we could not detect γH2AX at telomeric repeats using ChIP. This is in accordance with the work by Wheaton et al. 49 .
Progerin expression perturbs replication fork progression at telomeres. To further examine the impact of progerin on DNA replication, we decided to assess the progression of replication forks on DNA fibers. This technique was used in recent reports to demonstrate that progerin expression globally impaired fork progression 49 , generating a replication stress that induced a cell-intrinsic innate immune response 59 . However, how this relates to short telomeres observed in progerin-expressing cells and why telomerase expression rescued the cell growth is still unclear. The expression of fragile telomeres in both HGPS cells and progerin-expressing  www.nature.com/scientificreports/ HDFs (Fig. 3D) suggested that the progression of forks might be affected at telomeric repeats. To test this idea, we performed single-molecule analysis of replicated DNA (SMARD) to measure replication progression specifically at telomeres 53,60 . Cells were pulsed successively with the thymidine analogues EdU and IdU, which are incorporated into the newly synthesized DNA, followed by a chase period. This pulse/chase cycle was repeated four times before collecting the cells for DNA stretching on glass coverslips. EdU and IdU incorporation was visualized on the stretched DNA fibers via immunofluorescence, and telomeres were labeled using telomere FISH (Fig. 5A). This experiment allows evaluation of several parameters related to the progression of the DNA replication forks, including track length, the inter-origin distance, and the proportion of stalled forks (Fig. S5A,B).
Telomere fibers with no EdU/IdU incorporation were considered non-replicating (Fig. 5B). Telomere FISH on stretched DNA fibers was first used to measure telomere length 61 in all three cell lines. Consistent with our previous data using Southern blots (see Fig. 2), telomeres were on average 1 kb shorter in EGFP-PG cells compared to NLS-EGFP and EGFP-LA control HDFs (Fig. 5C). Next, we examined replication fork progression and stability at the whole-genome level. Progerin expression in HDFs provoked a decrease in track length and replication fork rate, and an increase in fork stalling (Fig. 5D-F). As a consequence, the measured inter-origin distance (IOD) decreased in progerin cells, suggesting an activation of dormant origins to compensate for fork failures (Fig. 5E). Then, we specifically assessed replication progression at telomeres, only scoring fibers with a terminal telomere FISH signal next to EdU and/or IdU labelling (Fig. S5C). Although these experiments were technically challenging due to relatively short telomeres in untransformed HDF cells and a limited number of cells in S phase, we nevertheless managed to score enough telomere-fibers for accurate statistical analysis. We observed that replication forks were moving at a slower pace at telomeres than in other genomic regions, with rates close to 1 kb/ min and 2 kb/min, respectively (Fig. 5F). This result confirmed that telomeric DNA is a difficult substrate for replication fork progression in normal cells. Progerin expression further decreased replication fork progression at telomeres with a drop to 0.7 kb/min. Consistent with a slower replication rate, we also observed shorter track length (Fig. 5G). Origins of replication were observed mostly at subtelomeric regions, and their number was not affected by progerin expression (Fig. S5D). Importantly, changes in the lamina organization by increasing the level of wild-type A-type lamins also moderately affected DNA replication. This was more obvious at telomeres, with an intermediate phenotype between EGFP-NLS control and EGFP-Pg.
Progerin affects the replication timing signature of normal HDFs, which impairs their replication capacity. The DNA replication timing (RT) program is a highly organized and defined temporal sequence of S phase replication events. Replication timing and 3D chromatin organization are tightly connected, with early replication taking place in the nuclear interior, while chromatin at the nuclear periphery replicates late in S phase 62 . Recent work reported RT programs of a large number of fibroblasts derived from HGPS progeroid patients and healthy donors 63 . They discovered a specific RT signature of cells from HGPS donors, with regions that replicate early only in progeria but replicate later in cells from all healthy donors. This study highlighted a particular region on chromosome 3 encompassing the TP63 gene that replicates early compared to healthy donors. In order to determine whether exogenous progerin expression could also impact HDF cells RT, we examined at the genome-wide level their spatiotemporal replication program before and after retroviral transduction of EGFP-LA and EGFP-PG. Cells were pulse-labeled with BrdU and sorted into early and late S-phase fractions by flow cytometry. Nascent BrdU-substituted DNA of each fraction was then isolated and labeled with fluorescence dyes before hybridization on microarrays covering the human genome, with the exception of centromeric and telomeric regions, with one probe every 13 kb. The RT was then determined using the Agilent CGH algorithm. A positive log ratio value corresponds to early-replicating regions, while a negative log ratio value corresponds to late-replicating regions. We compared the RT profiles of EGFP-LA and EGFP-PG cells with the profile from non-transduced cells and found that progerin expression induced a significant change in RT for about 3% of the genome (Fig. S6A). As an example, the RT profiles of chromosome 1 are displayed Fig. S6B. Progerin expression induced a shift to earlier replication for 1.6% of chromosome 1, and a shift to later replication for only 0.2% of that chromosome. To compare our data with the one from Rivera-Mulia et al., we plotted the chromosome 3 RT profiles from LA and PG cells compared to non-transduced HDFs (Fig. S6C). Expression of progerin did shift TP63 towards earlier replication, confirming TP63 as a marker of HGPS as previously proposed. In conclusion, exogenous progerin expression can alter the RT signature of normal HDFs, with a tendency to hasten their replication program.
Progerin expression limits dNTPs accessibility. Physiological nucleotide levels are known to be limiting for DNA synthesis. The temporal regulation of origin firing controls the number of replication forks that are active at any given time and moderates the demand for dNTPs. When the RT is not strictly respected, too many origins of replication are simultaneously activated, which in turn promotes a shortage of available dNTPs 62 . We hypothesized that a similar scenario may occur in progerin cells. Nucleotide metabolism was previously shown to be impaired in HGPS-derived cells 64 , which could further sensitize cells to a dNTP shortage. The shift to ear-  Mean with SD is shown, n = 3. At least 30 metaphases were scored per condition in total. Statistical significance was determined using Mann Whitney test (***indicates p < 0.0005, and ****p < 0.0001). www.nature.com/scientificreports/ lier RT as well as the smaller IOD measured on DNA fibers (Fig. 5E) indeed suggest that more replication forks are simultaneously active in these cells, which could limit dNTPs availability. To test this idea, HDF cells expressing NLS-EGFP, EGFP-LA and EGFP-PG were supplemented with dNTPs for 12 or 24 h, and pulse labeled with EdU for 30 min (Fig. 6A). The percentage of EdU positive cells was scored to reflect the replication capacity of the cells. As expected, progerin expression induced a twofold decrease of EdU positive cells compared to controls. Complementation of the cell medium with dNTPs was sufficient to rescue this phenotype and to restore the percentage of EdU positive cells to a level similar to controls (Fig. 6A). In patient cells, we observed that HGPS M14 cells also positively responded to dNTPs supplementation, with an increase in EdU positive cells after 24 h of treatment (Fig. 6B). The growth capacity of HGPS F8 cells was not significantly improved, but the initial percentage of EdU positive cells was much higher in this cell line. In line with this, dNTPs supplementation of HGPS cells expressing hTERT did not further increase the percentage of EdU positive cells (Fig. 6B). These cells already reached a steady replicative pace, comparable after dNTPs supplementation or hTERT expression. LAP2α expression was increased in HGPS cells after hTERT expression, as previously described 17 , and surprisingly dNTPs supplementation also impacted LAP2α levels, with an increased observed in most samples including normal HDFs (Fig. 6C). Finally, we tested whether dNTPs addition could prevent the formation of MTS during telomere replication. HDF cells expressing NLS-EGFP, EGFP-LA and EGFP-PG were grown with or without addition of dNTPs in the cell medium for 24 h, and the presence of MTS was scored on metaphase spreads. Remarkably, dNTPs supplementation completely reversed the occurrence of this phenotype (Fig. 6D).
Overall, these results confirm that progerin expression alters the RT program, most likely due to changes in 3D-chromatin organization and structure, and in turn impacts replication capacity by limiting access to dNTP pools. This dNTPs shortage directly affects telomere replication, triggering the expression of fragile telomeres.

Discussion
Overall, our work proposes a mechanistic model to describe the effects of NE dysfunction on telomere homeostasis in the context of premature aging. We propose that progerin expression alters the 3D telomere organization and their heterochromatin status. The genome-wide replication signature is perturbed, with a shortage of dNTPs that weight on telomere maintenance by inducing a telomeric replicative stress. As an outcome, telomeres shorten faster, inducing premature senescence and premature organismal aging. The strong connection between lamins, aging and telomeres we unraveled is also highly relevant for normal aging, as sporadic endogenous cryptic site leading to progerin expression has been discovered in wt aged human cells 40 .
Progerin and telomere organization. The NE is the barrier that confines the nucleoplasm and confers, among other functions, the essential scaffold required to organize the nuclear content. Besides being tissue-and cell-type specific, the spatial organization of chromatin influences major biological processes such as regulation of gene expression, replication timing, genome stability, and senescence 21,65,66 . Using our proximity labeling technique MadID, we discovered that expression of progerin increases the interaction between telomeres and the nuclear lamina. It is unlikely that the higher level of methylation we observed is due to increased methylation accessibility of telomeric chromatin. Indeed, both wild-type LaminA and Progerin affected heterochromatin marks at telomeres (discussed below), while only progerin impacted its methylation. Whether this interaction occurs at the nuclear periphery, in the nuclear interior, or at sites of membrane invagination is still unclear.
Blebbing of the NE could increase the chances of contact between lamins and chromatin. This is reminiscent of changes in nuclear architecture during senescence of human mesenchymal stem cells, where centromeres and telomeres colocalize with lamina intranuclear structures as shown by chromatin immunoprecipitation experiments 67 . Nevertheless, this enriched contact can alter telomere maintenance and perturb their dynamic properties. Lmna −/− mouse cells redistribute the telomeres to the nuclear periphery 37 , however, expression of progerin most likely results in a very different outcome when it comes to chromatin dynamics. The anomalous diffusion of interphase telomeres depends on the LaminA network that creates a crosslink to restrict chromatin movements 68,69 . Therefore, progerin expression could further stabilize telomeres and limit their diffusion, potentially creating a topological stress. With these elements in mind, it is also not surprising that expression of wild-type LaminA had some consequences on chromatin marks at telomeres. LaminA can occupy both heterochromatin and euchromatin domains 70 . An increase in LaminA will create an unbalanced A-and B-Type organization within the nuclear lamina, perturb the pool of nucleoplasmic LaminA 71 , and alter chromatin state.

Accurate replication timing is crucial for genome maintenance. Alteration of heterochromatin at
the genome level has been linked to changes in the replication timing signature in HGPS 63 . In our setting, a significant proportion of the genome replicated earlier in progerin cells compared to controls. The changes were mild, but we analyzed the profiles of cells 2 weeks post selection after EGFP-PG transduction, to collect exponentially growing cultures. Changes would probably increase after longer periods of progerin expression as it happens in HGPS patient cells. We did not specifically determine the replication timing of telomeres, which are known to replicate throughout S phase in normal human cells. While heterochromatin marks were decreased at telomeres, which could favor a shift to earlier replication, their stronger interaction with the nuclear lamina could in contrary favor a delay. Nevertheless, the shift toward earlier replication observed genome-wide will limit the levels of "building blocks", i.e. dNTPs, that can in turn affect any active replication fork. This is what we observed at telomeres, with a supplementation of dNTPs sufficient to alleviate the replication stress induced by progerin expression. This result is consistent with the finding that HGPS cells have an altered de novo purine synthesis machinery with a downregulation of ribose-phosphate pyrophosphokinase 1 (PRPS1), essential for nucleotide synthesis 64 . www.nature.com/scientificreports/ Replication stress, lamins and telomeres. Compared to the global genome, our SMARD analysis revealed that forks progress slower at telomeres, a region known to be difficult to replicate 53 . Progerin expression further impaired the fork progression, consistent with the fragile telomere we observed on metaphase spreads. Recent work suggest that heterochromatin loss is a determinant of progerin-induced DNA damage that occurs in late S phase 72 . Our work reveal a similar behaviour at telomeres, with a diminution in H3K9 methylation and replication defects that promotes faster telomere shortening. Other mechanisms could also promote replication stress in progerin cells. Several lines of evidence link nuclear A-and B-Type lamins to DNA replication aptitude, via interaction with sites of DNA replication, and/or PCNA 49,[73][74][75][76] . These findings also explain why expression of wild-type LaminA mildly influenced fork progression. The protein AKT-interacting protein (AKTIP) could also be at play. AKTIP has been reported to preferentially localize at the nuclear periphery, but also to interact with shelterin proteins and be required for normal telomere replication 77,78 . Since AKTIP is depleted in HGPS cells 78 , its implication should be further assessed. Lastly, although we could not detect a change in shelterin expression in total extracts, and no obvious change in TRF1 and TRF2 binding to telomeric repeats using ChIP-dot blot (Fig. 3B), ChIP-Q-PCR experiments revealed a small decrease in TRF1 binding to telomeres after progerin expression (Fig. 1I). Since TRF1 promotes efficient telomere replication 53 , it would be interesting to further assess the role of TRF1 in progerin-induced fragile telomeres.
Relevance of telomere defects for HGPS organismal premature aging. HGPS patients appear normal at birth, but develop severe abnormalities within the first two-years of life. As telomere dysfunction stands out to be a key element in the fitness of progerin cells, long telomeres at birth might delay the clinical signs of the disease in patients. Even if only two patient cell lines were studied here, all phenotypes appeared more pronounced in HGPS M14 compared to F8, which correlates with patient age and initial telomere length. Similarly, up to 5 PDs were necessary to observe cell growth delays after ectopic expression of progerin in normal cells. These observations suggest that telomere length is determinant for the onset of HGPS-related phenotypes, such as premature senescence. It could explain the discrepancy in the literature related to the severity of the phenotypes induced by progerin expression, notably on the extent of DNA damage or onset of growth delay 45,79 . Immunofluorescence. Human fibroblasts were grown to 50% confluence on glass coverslips (1.5 thickness) and fixed with 4% paraformaldehyde for 10 min at RT. After three washes with PBS, cells were permeabilized with 0.5% Triton X-100 in PBS and incubated with a blocking solution (0.2% (w/v) cold water fish gelatin, 0.5% (w/v) Bovine Serum Albumin (BSA) in 1× PBS) for 30 min at RT. Incubation of primary antibodies was performed for 2 h at RT or overnight at 4 °C. Alexa488/546/647-conjugated secondary antibodies (Invitrogen) were incubated for 45 min at RT. For the detection of Ultrafine Anaphase Bridges (UFBs), the procedure described in Bizard et al., was followed 54   Replication timing profiles. Exponentially growing HDF cells non-transduced or expressing NLS-EGFP, EGFP-LA and EGFP-PG were pulse-labeled with 50 μM BrdU for 90 min, washed three times in PBS, and then a minimum of 10 million cells per sample were fixed in 70% ethanol and stored at − 20 °C. Fixed samples were resuspended in PBS-RNAseA (0.5 mg/ml) and 50 μg/ml propidium iodide, incubated for 30 min at RT prior to cell sorting. 100,000 cells were sorted in two fractions, S1 and S2 that correspond to early and late fractions respectively, using the INFLUX 500 or the Astrios cell sorter. Both S1 and S2 fractions were treated with 0.2 mg/ ml proteinase K in lysis buffer (50 mM Tris pH 8, 10 mM EDTA, 300 mM NaCl, 0.5% SDS) for 2 h at 65 °C in the dark or overnight. DNA was extracted and sonicated to obtain fragments of about 500-1000 bp. DNA was then denatured at 95 °C for 5 min and snap cooled for 10 min. Nascent DNA was immunoprecipitated using the IP-STAR apparatus with the indirect method option (Diagenode) using 10 μg of anti-BrdU antibodies (BD Biosciences, #347,580), and purified by phenol-chloroform. The quality of the enrichment of S1 and S2 fractions was verified by qPCR using primers specific to regions replicating early or late. Next, whole genome amplification was performed using the WGA amplification kit (Sigma) to obtain at least 500 ng of DNA. After amplification, S1 and S2 were labeled with Cy3 and Cy5 ULS molecules (Genomic DNA ULS labeling Kit, Agilent) according to the manufacturer's protocol. The samples were hybridized on 4 × 180 K human microarrays (Agilent, genome reference hg18) that covers the whole genome with one probe every 13 kb (11 kb in RefSeq sequences) following the manufacturer's protocol. Microarrays were scanned with an Agilent's Hi-Resolution C Scanner with a resolution of 3 μm and the autofocus option. Replication timing data were analyzed using the limma R package as in Ref. 82 . The loess method was used to normalize within arrays, and the quantile method was used to normalize in between arrays. The normalized profiles of each chromosome were then smoothed using loess regression fitting with a bandwidth of 3 Mb. We combined the smoothed profiles of two replicates to obtain the average RT profile of each chromosome in each condition. The 95% confidence interval (CI) obtained from the loess regression fitting of the average profiles using the loess.ci function of the spatialEco R package, and the non-transduced condi-

Southern blots of terminal restriction fragments (TRFs).
Analysis of telomere length was performed as in Ref. 83 . Quantification of the signal was done using the ImageJ software.
Reactions were adjusted to 1× SCC and 0.1% Triton X-100, and the digested DNA was then annealed with a biotinylated oligonucleotide (Bio-5′-ACTCC(CCC TAA ) 3 -3′) (3.5 pmol) by controlled stepwise cooling from 80 to 25 °C (1 °C/min) using a thermocycler. Then 5% of samples was collected as an input and streptavidin-coated magnetic beads (18 µl, Invitrogen, M-280) prewashed with 1× PBST and blocked for 1 h with 5× Denhardt solution (0.1% Ficoll (type 400), 0.1% polyvinylpyrrolidone, and 0.1% bovine serum albumin), were incubated with the annealed samples overnight in a rotator end-over-end at 6 r.p.m. and 4 °C. Beads were collected against the side of the tubes by applying a magnet (Invitrogen), and the unbound fraction was collected. The beads were washed four times with 1× sodium chloride-sodium citrate (SSC), 0.1% Triton X-100, and once with 0.2× SSC. Beads were resuspended in 50 µl elution buffer and telomeres were slowly eluted by heating the tubes at 50 °C for 20 min. The elution was repeated with 50 µl of elution buffer. The quality of telomere purification was assessed by qPCR. Telomeric DNA was detected using primers as described in Ref. 43 . The presence of genomic DNA was assessed using primers directed against a single genomic locus as in Ref. 43  www.nature.com/scientificreports/ for 1 h in 5% nonfat dry milk and 0.1% TBST (0.1% Tween-20 in 1xTBS, pH7.4). Subsequently, m6A antibody (Synaptic Systems) was diluted to 1:2000 in 5% nonfat dry milk and 0.1% TBST, and incubated overnight at 4 °C. Following 3 washes with 0.1% TBST, a HRP-conjugated secondary antibody was applied for 45 min at room temperature. After 3 washes with 0.1% TBST, the chemiluminescence signal was visualized using ChemiDoc™ Imaging System (BioRad). The signal was quantified using the ImageJ software. The intensity of the m6A signal was normalized to the amount of telomeric DNA purified in each sample measured by qPCR.
EdU incorporation, QIBC and DNA damage labeling. For 53BP1 and yH2AX immunostaining, HDFs cells expressing NLS-EGFP, EGFP-LA and EGFP-PG were seeded on coverslips (1.5 thickness; 13 mm of diameter). When indicated, cells were treated with 0.2 μM aphidicolin (Sigma) for 24 h and 10 μM of EdU (Invitrogen C10340) was added for 30 min before fixation in 4% PFA/1X PBS. Cells were then permealized with 0.5% Triton X-100 in 1× PBS and incubated with a blocking solution (0.2% (w/v) cold water fish gelatin, 0.5% (w/v) bovine serum albumin (BSA) in 1X PBS) for 30 min at RT. Detection of EdU was performed prior to incubation with the primary antibodies using the Click-iT™ Plus EdU Alexa Fluor™ 647 Imaging Kit according to the manufacturer's instructions (Thermo-Fisher Scientific). 53BP1 primary antibody (Santa Cruz sc-22760-1/200) or yH2AX (Biolegend 613402-1/1000) were used for immunofluorescence following the above protocol. Image acquisition of multiple random fields was carried out on a wide field DM6000 Leica microscope equipped with a 40 × dry objective using a Qimaging Retiga R6 camera driven by Metamorphe software. 10 to 20 images per condition were acquired randomly and processed for automated analysis with the Cell Profiler 2.1.1. image analysis software. DAPI signal was used for segmentation of the nuclei according to intensity threshold, generating a mask that identified each individual nucleus as an individual object. This mask was applied to quantify pixel intensities in the different channels for each individual cell/object. The values quantified for EdU and DAPI staining per cell were graph plotted by dual-parameter (EdU vs DNA) generating diagrams in a flow-cytometry-like fashion (QIBC Quantitative-based image Cytometry) for each cell condition. This approach allows the assignment of cells to G1, S or G2/M phases. Images were assembled with ImageJ software.
Nucleoside supplementation. Embryomax ® Nucleosides 100× (Merck, ES-008-D) were added to cell culture media at a final concentration of 1× for 12 or 24 h as indicated in the figures. EdU incorporation and detection were performed as described in the previous section.

SMARD analysis.
Single-molecule analysis of replicated DNA was performed as described previously 53,60 . In Coverslips were denatured for 15 min in alkali-denaturing buffer (0.1 M NaOH, 0.1% β-mercaptoethanol in 70% ethanol) and fixed by addition of 0.5% glutaraldehyde for 5 min. Replication tracks were visualized by detection of halogenated nucleotides using anti-IdU and anti-EdU antibodies. For the replication fork progression analysis at telomeric regions, telomeric DNA was visualized by hybridization with an TAMRA-OO-(CCC TAA ) 3 PNA probe and replication tracks co-localized with PNA staining were scored. Track length was determined using the length of the IdU track (second pulse). Stalled forks are seen as labelled during the first pulse of EdU, but not IdU signal was detected.
Statistical analysis. GraphPad Prism version 7.0c for Mac was used for statistical analysis.