Telomere-to-mitochondria signalling by ZBP1 mediates replicative crisis

Cancers arise through the accumulation of genetic and epigenetic alterations that enable cells to evade telomere-based proliferative barriers and achieve immortality. One such barrier is replicative crisis—an autophagy-dependent program that eliminates checkpoint-deficient cells with unstable telomeres and other cancer-relevant chromosomal aberrations1,2. However, little is known about the molecular events that regulate the onset of this important tumour-suppressive barrier. Here we identified the innate immune sensor Z-DNA binding protein 1 (ZBP1) as a regulator of the crisis program. A crisis-associated isoform of ZBP1 is induced by the cGAS–STING DNA-sensing pathway, but reaches full activation only when associated with telomeric-repeat-containing RNA (TERRA) transcripts that are synthesized from dysfunctional telomeres. TERRA-bound ZBP1 oligomerizes into filaments on the outer mitochondrial membrane of a subset of mitochondria, where it activates the innate immune adapter protein mitochondrial antiviral-signalling protein (MAVS). We propose that these oligomerization properties of ZBP1 serve as a signal amplification mechanism, where few TERRA–ZBP1 interactions are sufficient to launch a detrimental MAVS-dependent interferon response. Our study reveals a mechanism for telomere-mediated tumour suppression, whereby dysfunctional telomeres activate innate immune responses through mitochondrial TERRA–ZBP1 complexes to eliminate cells destined for neoplastic transformation.

Replicative senescence and crisis constitute two anti-proliferative barriers that human cells must evade to gain immortality 3 . Senescence is p53 and RB dependent and occurs when shortened telomeres elicit a DNA-damage response 4,5 . Loss of cell cycle checkpoints renders cells incapable of senescence activation, resulting in continued proliferation and telomere shortening. Such cells eventually succumb to crisis, characterized by extensive cell death and genome instability 1,2,6,7 . Crisis provides a redundant tumour-suppressor mechanism for replicative senescence, whereby CGAS-STING signalling triggers a non-canonical form of autophagy capable of executing cell death rather than sustaining cell survival 1 . Although this discovery established an essential role for autophagy during cell death in crisis, it remained unclear how dysfunctional telomeres engage nucleic-acid-sensing machineries and activate innate immune signalling pathways that are required for cell death.
The inherent stringency of replicative crisis offered a powerful system in which to conduct a positive selection CRISPR-Cas9 knockout screen ( Fig. 1a (left)). Survivors are expected to have lost pathways that are required for crisis, such as the ones linking dysfunctional telomeres to innate immune activation. Enrichment analysis revealed first that Gene Ontology (GO) terms associated with inflammation and innate immunity were predominant (Extended Data Fig. 2c). Second, the innate immune sensor ZBP1 (also known as DAI and DLM-1) emerged as a top hit, along with the previously characterized CGAS-STING-IFN pathway 1 (Fig. 1a (right)). ZBP1, an ISG product 12 , was described as cytosolic nucleic acid sensor that induces type I IFNs and regulates innate immunity and cell death 13 . CRISPR-mediated deletion of ZBP1 demonstrated its essential role in crisis. Control cells (sgLUC, sgGFP) https://doi.org /10.1038/s41586-023-05710-8 Article entered crisis around PD90 for WI38 SV40LT and PD107 for IMR90 E6E7 , when cell death was frequent and replicative ability was reduced (Extended Data Fig. 3a-d). However, pooled ZBP1-knockout cells continued to proliferate for an additional 7-10 population doublings beyond the crisis plateau with a notable reduction in cell death and maintenance of growth potential (Extended Data Fig. 3a-d). Cells that have bypassed crisis showed reduced expression and secretion of IFNβ, indicative of impaired type I IFN activity (Extended Data Fig. 3e,f). Depletion of ZBP1 abrogated the ISG signature, confirming that ISG induction during crisis is attributable to ZBP1 (Fig. 1b). These results directly linked ZBP1 to innate immune activation and cell death during crisis.

Innate signalling requires telomere dysfunction
Replicative crisis is a telomere-dependent program. We therefore depleted the telomere protection factor TRF2 in growing IMR90 E6E7 cells, which resulted in telomere deprotection and fusion (Extended Data Fig. 6f-h), accompanied by ZBP1(S) upregulation and a type I IFN response through TBK1 phosphorylation (Fig. 2a). Depletion of ZBP1 dampened TBK1-IRF3 signalling, reduced ISG expression and attenuated autophagy, without affecting telomere fusions (Fig. 2a,b and Extended Data Fig. 6i,j). Expression of the catalytic subunit of telomerase (hTERT) in growing cells resulted in telomere maintenance and continued growth past crisis (Extended Data Fig. 7a,b). These immortalized cells did not express ZBP1 or activate IFN signalling ( Fig. 2c and Extended Data Fig. 7c), whereas control cells entered crisis, upregulated ZBP1(S) and activated the TBK1-IRF3-IFN signalling axis ( Fig. 2c and Extended Data Fig. 7c). hTERT expression did not affect the ability of cells to launch an IFN response or stimulate autophagy when treated with exogenous double-stranded RNA or DNA (Fig. 2c). Expression of either ZBP1(L) or ZBP1(S) in growing or telomerase-positive cells did not activate the IFN pathway ( Fig. 2d (left and right)). ZBP1(S) potentiated an IFN response only when expressed in pre-crisis cells with short telomeres (Fig. 2d (middle)), suggesting the requirement for an additional immunostimulatory molecule specific for telomere dysfunction. In conclusion, two stimuli are required for launching a ZBP1-dependent IFN response in crisis: (1) upregulation of ZBP1(S) by CGAS-STING, and (2) a signal provided by dysfunctional telomeres.

Fig. 2 |
Telomere dysfunction is required for ZBP1-dependent IFN signalling. a, Immunoblot analysis of growing (PD35) IMR90 E6E7 fibroblasts expressing non-targeting control short hairpin RNA (shRNA) or shRNA against TRF2 (shTRF2). shTRF2 cells were transfected with either two individual siRNAs against ZBP1 (siZBP1) or non-targeting control siRNA at day 4 after shRNA transduction. Protein extracts were collected at days 6, 9 and 12 as shown in the timeline of the experiment in Extended Data Fig. 6i. Mock represents non-transduced cells. GAPDH was the loading control. Two independent experiments were performed. p, phosphorylated. b, RT-qPCR analysis of ISGs in growing (PD35) IMR90 E6E7 fibroblasts expressing non-targeting control shRNA or shRNA against TRF2. shTRF2 cells were transfected with either two individual siRNAs against ZBP1 or non-targeting control siRNA (siCtrl) at day 4 after shRNA transduction. RNA extracts were collected at day 12. Expression levels were normalized to control cells with non-targeting shRNA. Data are mean ± s.d. of technical replicates.
n values indicate the number of technical replicates. Statistical analysis was performed using one-way ANOVA; ***P < 0.001. Three independent experiments were performed. c, Immunoblot analysis of IMR90 E6E7 fibroblasts expressing either empty vector or hTERT at the indicated PDs. Growth curves and telomere restriction fragment (TRF) analysis are shown in Extended Data Fig. 7a,b. Cells stimulated with 2 μg ml −1 of poly(deoxyadenylic-deoxythymidylic) (poly(dA:dT)) or poly(inosinic:cytidylic) (poly(I:C)) for 24 h were used as positive controls. GAPDH was used as the loading control. Two independent experiments were performed. d, Immunoblot analysis of growing (PD42), pre-crisis (PD100) and telomerase-positive (PD105) IMR90 E6E7 fibroblasts expressing ZBP1(L) or ZBP1(S). Flag tag was added to either the N terminus or the C terminus. GAPDH was the loading control. Two independent experiments were performed. Lipo, lipofectamine.

Article
pathway, but also on mitochondrial antiviral signalling protein (MAVS) (Extended Data Fig. 4e)-an innate immune adapter that is located at the mitochondrial outer membrane (MOM). MAVS can be activated by the viral RNA sensors retinoic-acid-inducible gene I (RIG-I) and melanoma differentiation-associated protein 5 (MDA5) 32-35 , which did not have a role in ZBP1 upregulation during crisis (Extended Data Fig. 4e). Dysfunctional telomeres are actively transcribed into long non-coding RNA species termed TERRA, resulting in subtelomeric RNA sequences followed by a variable number of telomeric UUAGGG repeats 36,37 . RNA-dot blot and quantitative PCR with reverse transcription (RT-qPCR) analysis showed enhanced TERRA transcription from critically short telomeres in crisis or telomeres depleted of TRF2 38 (Fig. 3a,b and Extended Data Fig. 8a,b). Given that long non-coding RNAs interact with cytosolic sensors to regulate immune responses 39,40 and that TERRA has been linked to inflammation previously 41,42 , we reasoned that TERRA could be the immunostimulatory nucleic acid species that is recognized by ZBP1(S). We first depleted TRF2 in growing IMR90 E6E7 fibroblasts expressing WT ZBP1(S) and then performed formaldehyde cross-linking combined with ZBP1-associated RNA immunoprecipitation (Fig. 3c and Extended Data Fig. 8c,d). Formaldehyde cross-linking combined with RNA immunoprecipitation-sequencing (fRIP-seq) reads were mapped, and enrichment profiles were generated by calculating the fold change of the read counts between the fRIP and input samples. The analysis revealed reads for subtelomeres, with sharp peaks appearing at the telomere-proximal region located 5-10 kb upstream of TTAGGG repeats, implying that ZBP1 senses TERRA-containing transcripts  TFEC  TIMP1  TLR3  TMEM140  TMEM51  TNFAIP3  TNFAIP6  TNFRSF10A  TNFSF10  TXNIP  TYMP  ( Fig. 3c and Extended Data Fig. 8c,d). To confirm ZBP1-TERRA interactions, we suppressed TRF2, immunoprecipitated WT or mutant ZBP1 lacking Zα2 (ZBP1(ΔZα2)) and assessed the presence of TERRA using an RNA-dot blot analysis ( Fig. 3d and Extended Data Fig. 8e). A strong signal for TERRA was detected in immunoprecipitates from WT ZBP1 cells but not in those expressing mutant ZBP1 (Fig. 3d). Treatment with RNase A confirmed RNA-specific signals (Fig. 3d). These results indicate that telomere dysfunction contributes to the accumulation of TERRA molecules that physically interact with ZBP1 through its Zα2 domain.
To evaluate the role of TERRA-ZBP1 interactions in innate immune activation, we expressed a mutant form of TRF1 lacking the N terminus (TRF1(ΔN)) alone or fused to the VP16 transcriptional activation domain (VP16-TRF1(ΔN)) 42 in growing IMR90 E6E7 fibroblasts. Stable expression of these proteins was confirmed, and upregulation of TERRA was validated ( Fig. 3e and Extended Data Fig. 9a-c). WT ZBP1-expressing cells responded to TERRA induction by activating the IFN signalling pathway and ISG expression (Fig. 3e,g). Accordingly, cells entered a crisis-like state associated with cell death and elevated ISG levels ( Fig. 3f and Extended Data Fig. 10a-d). No loss of telomere protection, no activation of ATM-and ATR-dependent DNA-damage response and no formation of fused chromosomes were observed (Extended Data Fig. 11a,b). Partial depletion of TERRA in cells expressing WT     replicates. Statistical analysis was performed using one-way ANOVA. ***P < 0.001. Three independent experiments were performed. f, Immunoblot analysis of growing IMR90 E6E7 fibroblasts as described in e. GAPDH was used as the loading control. Two independent experiments were performed. g, RT-qPCR analysis of ISGs in growing (PD66) IMR90 E6E7 fibroblasts. Cells were transfected with siMAVS or non-targeting control siRNA at day 10 after transduction. RNA extracts were collected at day 14. Expression levels were normalized to empty vector siCtrl cells. Data are mean ± s.d. from technical replicates. n values indicate the number of technical replicates. Statistical analysis was performed using one-way ANOVA. NS, not significant; ***P < 0.001. Three independent experiments were performed.

Article
ZBP1 delayed STAT1 activation (Extended Data Fig. 10e-g). Cells expressing mutant ZBP1 with disrupted nucleic-acid-binding activity were insensitive to TERRA and did not activate the innate immune response ( Fig. 3e-g and Extended Data Fig. 10a-d). These findings demonstrate a mechanism of innate immune signalling triggered by dysfunctional telomeres, in which TERRA functions as the messenger molecule that binds to ZBP1(S) in the cytoplasm to promote an enhanced type I IFN response.

ZBP1 forms filaments at mitochondrial membranes
To decipher the mechanism of ZBP1 signalling, we performed immunostaining of ZBP1 in crisis cells, which revealed filamentous structures in the cytoplasm (Fig. 4a). ZBP1 lacking functional Zα2 and RHIM1 domains did not form these filaments, whereas no significant effect was observed after disruption of RHIM2 (Fig. 4a). A lack of filaments correlated with impaired type I IFN signalling (Fig. 4b,c), indicating that the assembly of ZBP1 filaments is essential for innate immune signalling and requires both TERRA binding through Zα2 and self-oligomerization through homotypic RHIM1 interactions. Multimerization is used by several sensors for signal propagation, including RIG-I 43 and MDA5 44 , which translocate to mitochondria to interact and activate MAVS 32-35 ; however, they often form aggregated structures rather than filaments 45 , suggesting an alternative oligomerization mechanism for ZBP1. We found co-localization of ZBP1 filaments with a subset of mitochondria based on MOM (TOM20) and matrix (MitoTracker and TFAM) markers (Extended Data Fig. 11c). Inhibition of mtDNA replication with 2′,3′-dideoxycytidine 46 had no effect on the number of ZBP1 filaments even though mtDNA and mtRNA levels were significantly reduced (Extended Data Fig. 11d). Depletion of either SUV3 or PNPase enzymes, components of the mitochondrial degradosome 47 , led to significant accumulation of mtRNA without altering ZBP1 filament formation (Extended Data Fig. 11e). We therefore concluded that ZBP1-mediated innate immune response requires formation of filaments on mitochondria, independently of sensing mtDNA or mtRNA.

MAVS is required for a ZBP1-mediated IFN response
Given that MAVS, a crucial adapter for RNA sensors anchored to the MOM, was enriched in our CRISPR-Cas9 knockout screen (Supplementary Information), and its depletion prevented IFN-dependent ZBP1 induction during crisis (Extended Data Fig. 4e), we reasoned that ZBP1-driven cell death could be mediated through MAVS. Indeed, depletion of MAVS prevented ISG expression in response to ZBP1 and TERRA co-expression independently of RIG-I and MDA5 (Extended Data Fig. 12a-c). Deletion of MAVS also reverted ISG expression during crisis, prevented autophagy upregulation and reduced the frequency of cell death (Extended Data Fig. 12d-f), suggesting that ZBP1 filaments at the mitochondria could be the critical event downstream of telomere stress signalling that leads to MAVS activation. To test this, ZBP1(S) was fused to the mitochondrial-targeting sequence of FIS1 to direct it specifically to the MOM in growing cells with functional telomeres (Extended Data Fig. 13a). ZBP1(S)-FIS1 colocalized with MAVS, triggered an IFN response, stimulated autophagy and induced cell death in fibroblasts and epithelial cells ( Fig. 4d-f and Extended Data Fig. 13c,d). ISG induction by mitochondria-targeted ZBP1(S)-FIS1 was dependent on MAVS ( Fig. 4g and Extended Data Fig. 13b). Finally, as shown by suppression of IFNAR2, cell death by autophagy required a secondary signalling pathway downstream of ZBP1-MAVS, by which secreted IFNs bind to and activate their cognate IFN receptor complexes in an autocrine and paracrine manner (Extended Data Fig. 13c,d). Together, these results suggest that ZBP1 signalling in crisis cells involves binding to immunostimulatory TERRA, followed by conformational changes allowing cytosol-to-MOM translocation and oligomerization to activate MAVS. We propose that these properties of ZBP1 serve as a signal-amplification mechanism enabling TERRA molecules to trigger a second wave of IFN signalling that, together with autophagy activation, causes cell death during crisis. Although ZBP1 functions as an RNA sensor with the ability to activate a MAVS and IFN signalling response, it differs from RIG-I and MDA5 in several aspects. First, ZBP1 filament formation occurs at the mitochondrial surface, whereas RIG-I 43,48 and MDA5 44,49,50 undergo oligomerization along double-stranded RNA structures. Second, ZBP1 lacks the caspase recruitment domain (CARD), whereas RIG-I and MDA5 activate MAVS through CARD-CARD interactions 35 . Finally, we propose that TERRA transcripts constitute a structure-specific ligand for ZBP1, with low or no binding affinity to RIG-I and MDA5. These differences may prime ZBP1 to propagate the inflammatory signalling cascade specifically in response to telomere dysfunction, revealing a unique mechanism of MAVS-dependent IFN activation and subsequent cell death by autophagy.
Our study reveals a mechanism for telomere-mediated tumour suppression, whereby dysfunctional telomeres in crisis stimulate two intertwined cytosolic nucleic-acid-sensing pathways and trigger a lethal IFN response (Extended Data Fig. 13e). We propose that the breakage of fused telomeres and the subsequent release of nuclear DNA into the cytoplasm drives initial activation of the CGAS-STING pathway and expression of ISGs, including ZBP1. However, crisis-associated cell death requires additional downstream activation of ZBP1(S) by TERRA and formation of ZBP1 filaments on mitochondria. This promotes an inflammatory loop leading to the expression of an ISG profile that, in concert with activation of autophagy through an as yet undetermined pathway, drives cell death specifically in replicative crisis. The simultaneous activation of CGAS-DNA-sensing and ZBP1-RNA-sensing pathways enables the innate immune system to orchestrate an efficient, type-I-IFN-dependent cell death response to eliminate precancerous cells with unstable telomeres. These findings highlight a synergy between critically short telomeres, mitochondria and innate immunity that has evolved to prevent age-associated cancer initiation in humans.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-023-05710-8.
14. Zhang, T. et al. Influenza virus Z-RNAs induce ZBP1-mediated necroptosis. Cell 180, 1115-1129 (2020 Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
T4 ligase (M0202S), T4 PNK (M0201S) and all restriction enzymes used, except for Esp3I (Thermo Scientific ER0452), were obtained from NEB. All primers were ordered from Eton Bioscience. PCR reactions for cloning were performed using KOD Hot Start DNA Polymerase (Novagen, 71086) according to the manufacturer's protocol.

Cell death assays
For real-time assessment of cell death, 2,000 cells per well per condition in triplicate were seeded in 96-well plates (CytoOne, CC7682-7596) the day before the experiment. The next day, the medium was changed to medium containing 250 nM IncuCyte Cytotox green dye (Sartorius, 4633) and IncuCyte Nuclight Rapid red dye (1:100 dilution) (Sartorius, 4717). Cells were seeded in 96-well plates the day before the experiment and the medium changed to medium containing 250 nM IncuCyte Cytotox green dye on the day of the experiment. Cells were imaged with at least two fields per well every 2 h using the ×10 objective on the IncuCyte S3 or IncuCyte Zoom live-cell analysis system (Sartorius). The total number of nuclei and the Cytotox-green-dye-positive nuclei were quantified using the IncuCyte analysis software. The percentage of dead cells was calculated by dividing the Cytotox green dye positive nuclei by the total number of nuclei multiplied by 100. In Extended Data Figs. 6e and 13d, cells were transfected with siRNA 48 h before cell death measurements. Then, 24 h after transfection, the medium was changed to medium containing 250 nM IncuCyte Cytotox green dye (Sartorius,4633) or IncuCyte Cytotox red dye (Sartorius, 4632). Then, 48 h after transfection, nuclei were stained with Hoechst (1 μg ml −1 ), images were taken using the Revolve fluorescence microscope and analysed with CellProfiler v.4.2.1 using a customized pipeline. Total nuclei and Cytotox-dye-positive nuclei were segmented using an integrated intensity-based object detection module. The percentage of dead cells was calculated by dividing the number of Cytotox-dye-positive nuclei by the total number of nuclei multiplied by 100.

Retroviral and lentiviral transduction
Lentiviral and retroviral particles were produced by the laboratory. Production of lentivirus was performed as described previously 56 . In brief, HEK293T (ATCC, CRL-11268) cells were transfected with 7 μg of DNA using Lenti-X Packaging Single-Shot system (Clontech, 631276). Then, 48 h after transfection, viral supernatant was collected, supplemented with serum and used for transduction in the presence of Lenti-Blast (Oz Biosciences, LB00500). To produce retrovirus, Phoenix cells were transfected with 20 μg of DNA using 100 μM of chloroquine. Then, 5 h after transfection, fresh medium was added. The viral supernatant was collected 24 h later and used for transduction in the presence of polybrene 4 μg ml −1 . Then, 48 h after infection, cells were washed and selected with 1 μg ml −1 puromycin, 600 μg ml −1 G418, or 90 μg ml −1 hygromycin. IMR90 E6E7 and WI38 SV40LT fibroblasts were subjected to long-term culturing under antibiotic selection.

Transfections
DNA transfections were performed using the Lipofectamine 3000 kit (Thermo Fisher Scientific, 1857482) according to the manufacturer's instructions. siRNA transfections were performed using the Lipofectamine RNAiMAX kit (Thermo Fisher Scientific, 13778030) according to the manufacturer's instructions.

Immunofluorescence-FISH analysis of interphase nuclei and metaphase spreads
For interphase nuclei, cells were seeded onto glass coverslips 24 h before the experiment as described previously 60 . Cells were washed in PBS, fixed in 4% formaldehyde in PBS for 10 min and permeabilized in 0.1% Triton X-100 in PBS for 10 min. For metaphase spreads, cells were treated with 20 ng ml −1 of Colcemid (Gibco, 15212-012) for 1 h, collected and incubated in hypotonic solution (27 mM KCl, 6.5 mM tri-sodium citrate) for 5 min. The cell suspension was cytocentrifuged, fixed in 4% formaldehyde in PBS for 10 min and permeabilized in KCM buffer (120 mM KCl, 20 mM NaCl, 10 mM Tris pH 7.5, 0.1% Triton X-100) for 10 min. In both settings, the samples were incubated in blocking buffer (20 mM Tris pH 7.5, 2% BSA, 0.2% fish gelatin, 150 mM NaCl, 0.1% Triton X-100, 0.1% sodium azide and 100 μg ml −1 RNase A) for 1 h at 37 °C. Cells were incubated with the primary antibody (γH2AX) for 2 h, washed in PBS and incubated with secondary antibody for 1 h at room temperature. The secondary antibodies used were AlexaFluor 568 anti-IgG Mouse (Thermo Fisher Scientific, A-11004) or AlexaFluor 647 anti-IgG Mouse (Thermo Fisher Scientific, A-21235). The samples were finally fixed in 4% formaldehyde in PBS before fluorescence in situ hybridization (FISH) analysis.

Immunofluorescence
Cells were seeded onto glass coverslips 24 h before the experiment. Cells were fixed in 4% paraformadehyde in PBS for 10 min, washed in PBS and incubated in blocking solution of 5% BSA in PBS at room temperature. Cells were then incubated with primary antibodies for 2 h, washed in PBS and incubated with secondary antibodies for 1 h at room temperature. The secondary antibodies used were AlexaFluor 488 goat anti-rabbit IgG (H+L) (Thermo Fisher Scientific, A-11034) and AlexaFluor 568 goat anti-Mouse IgG (H+L) (Thermo Fisher Scientific, A-11004). The samples were finally washed in PBS and mounted in ProLong Diamond with DAPI (Invitrogen, P36971). For MitoTracker staining, MitoTracker Red CMXRos (Thermo Fisher Scientific, M7512) was added to the culture medium at a final concentration of 200 nM and incubated for 20 min under tissue culture conditions. After staining, cells were washed twice in prewarmed PBS, then fixed in paraformaldehyde 4% PBS for 10 min and washed in PBS. Imaging was performed using the Zeiss LSM 880 with the Airyscan microscope. ZEN (Zeiss) and ImageJ were used for image analysis.

Crystal violet assay for determining cell viability
Cells before crisis were seeded at low density (476 cells per cm 2 ) and kept at 37 °C for 10 days before fixing with 4% paraformaldehyde in PBS. Cells were then stained with 0.05% crystal violet in distilled water for 20 min. fRIP fRIP-seq was performed as previously described with minor modifications 61 . In brief, IMR90 E6E7 cells were cross-linked with 0.1% formaldehyde in PBS for 10 min at room temperature and unreacted formaldehyde was then neutralized with 125 mM glycine for 5 min at room temperature. Cells were washed twice with ice-cold PBS, collected by trypsinization and the cell pellets were resuspended in RIPA buffer (50 mM Tris pH 8, 150 mM KCl, 0.1% SDS, 1% Triton X-100, 5 mM EDTA, 0.5% sodium deoxycholate, 0.5 mM DTT (added fresh), protease inhibitor cocktail (Roche, 4693159001), 100 U ml −1 RNasin Ribonuclease Inhibitor (Promega, N251B)), and incubated for 20 min at 4 °C under slow rotation. The cell lysates were centrifuged at maximum speed at 4 °C for 10 min, and the supernatants were collected and diluted in equal volumes of freshly made fRIP binding/wash buffer (150 mM KCl, 25 mM Tris pH 7.5, 5 mM EDTA, 0.5% NP-40, 0.5 mM DTT (added fresh), protease inhibitor cocktail (Roche, 4693159001), 100 U ml −1 RNasin Ribonuclease Inhibitor (Promega, N251B)). Diluted lysates were first precleared with Dynabeads Protein G (Invitrogen, 10004D) at a concentration of 25 μl of beads per 5 million cells for 30 min at 4 °C under slow rotation, and then incubated with 10 μg anti-Flag M2 antibodies (Sigma-Aldrich, F3165) previously coupled to 40 μl protein G Dynabeads (per each 5 million cells) for 2 h at 4 °C under slow rotation. The beads were washed twice with fRIP binding/wash buffer and protein-RNA cross-links were reversed by resuspending the beads in 56 μl of RNase-free water + 33 μl of 3× reverse-cross-linking buffer (3× PBS (without Ca 2+ and Mg 2+ ), 6% N-lauroyl sarcosine, 30 mM EDTA, 15 mM DTT (added fresh), 10 μl of 20 mg ml −1 proteinase K (Millipore, 70663) and 1 μl of RNasin Ribonuclease Inhibitor. Protein degradation and reverse-cross-linking was performed for 1 h at 42 °C, followed by another hour at 55 °C. RNA was recovered by resuspending the beads and reaction buffer in 1 ml TRIzol Reagent (Invitrogen, 15596018) and purified using the Direct-zol RNA Microprep kit as recommended by manufacturer (Zymo Research, R2061).

Library generation and sequencing for fRIP-seq
RNA-seq libraries were prepared with immunoprecipitated RNA using the TruSeq stranded total RNA sample preparation kit according to the manufacturer's protocol (Illumina). RNA-seq libraries were multiplexed, normalized and pooled for sequencing. The libraries were sequenced on the MiniSeq system (Illumina) with paired-end 37 bp reads. Image analysis and base calling were performed using Illumina CASAVA v.1.8.2. on the MiniSeq system and sequenced reads were quality-tested using FASTQC. Read pairs were mapped individually to the most complete assembly available of human subtelomeres (http://www.wistar.org/ lab/harold-c-riethman-phd/page/subtelomere-assemblies) using STAR v.2.5.3a allowing up to 101 mapping locations 62 . Primary alignments were assigned randomly, and each read is represented by one mapping location. Library sizes were normalized to 1 × 10 7 for comparison and the log 2 ratio of enrichment of immunoprecipitates versus input was calculated at each base in the subtelomere using HOMER v.4.10 and the assembly available of human subtelomeres 63

RT-qPCR
Total RNA was isolated using TRIzol (Invitrogen, 15596018) and purified using the RNAeasy Mini Kit (Qiagen, 74106) according to the manufacturer's instructions. Genomic DNA was eliminated by double digestion with DNase I (RNase-free DNase Set, Qiagen, 79256). For RT-qPCR, 3.5 μg of RNA was reverse-transcribed either with random hexamers for measuring ISGs or with 2 μM of telC-and GAPDH-specific RT primers 64 using the SuperScript III First-Strand Synthesis System (Thermo Fisher Scientific, 18080-051). qPCR was performed on the CFX384 Touch Real-Time PCR Detection System (BioRad). Reactions were run in triplicates with Power SYBR Green Master Mix (Applied Biosystems, Thermo Fisher Scientific, 4367659) in a total volume of 10 μl with standard cycling conditions. Relative gene expression was normalized using GAPDH as a housekeeping gene and calculated using the Comparative CT Method (ΔΔC T method). The primers are listed below.

RNA-dot blot
Total RNA (10 μg) was blotted onto a positively charged nitrocellulose membrane (GE, RPNBL/02/10). For fRIP, 200 ng of RNA from IP or input samples were used. For RNase controls, RNA was incubated with RNase A (Invitrogen, 12091-039) at 37 °C for 1 h. RNA was fixed by ultraviolet cross-linking (Stratalinker, 2400) and TERRA was detected by hybridizing overnight at 55 °C with a Church mix containing telomeric repeat probes generated by [CCCTAA] 3 -primed Klenow labelling of an 800 bp [TTAGGG] n fragment in the presence of [α 32 P]dCTP. After hybridization, the membrane was washed twice in 2× SSC and 0.1% SDS for 10 min at room temperature and then once for 10 min at 50 °C. The radioactive signal was detected using the Typhoon FLA 9000 imager (GE Healthcare). After the signal acquisition, membranes were stripped and rehybridized at 50 °C overnight with 32 P-dCTP-labelled probes targeting GAPDH transcripts. Signals were measured using ImageJ.

TRF analysis
TRF analysis was performed as previously described 65 . In brief, genomic DNA was isolated by phenol-chloroform extraction and digested with AluI and MboI overnight. A total of 4 μg of digested gDNA was separated on 0.7% agarose gel at 40 V and transferred to a positively charged Nylon membrane (Amersham, RPN203B). After cross-linking the DNA and prehybridization (5× SSC. 0.1% N-lauroylsarcosine sodium salt solution, 0.04% SDS) for 2 h at 65 °C, the membrane was incubated with digoxigenin-labelled TelG probe diluted in hybridization buffer (1.3 nM final concentration) overnight at 65 °C. Digoxigenin-labelled TelG probe was generated as previously described 66 . Then, the membrane was washed three times with wash buffer 1 (2× SSC, 0.1% SDS), one time with wash buffer 2 (2× SSC) for 15 min each and blocked in freshly prepared blocking solution (100 mM maleic acid, 150 mM NaCl, pH 7.5, 1% (w/v) blocking reagent (Roche, 11096176001)) for 30 min. Next, the membrane was incubated for 30 min in anti-digoxigenin-AP antibodies (Roche, 11093274910) diluted in blocking solution, washed twice in wash buffer 3 (100 mM maleic acid, 150 mM NaCl, pH 7.5, 0.3% (v/v) Tween-20) for 15 min each and equilibrated in AP buffer (100 mM Tris, 100 mM NaCl, pH 9.5) for 2 min. Digoxigenin-labelled telomeric DNA was detected using CDP-star ready to use (Roche, 12041677001) solution.

Genome-wide CRISPR screen
The human Brunello CRISPR knockout pooled library was obtained from D. Root and J. Doench through Addgene (73179-LV) 67 . This ready-to-use lentiviral library has 77,441 gRNAs, targeting 19,114 protein-coding genes, with approximately 4 sgRNAs per gene. For adequate representation of each sgRNA, a total of 100 million pre-crisis (PD100) IMR90 E6E7 fibroblasts were transduced with the lentiviral library at a multiplicity of infection of 0.5. Transductions were performed in six-well plates (3 million cells per well) in medium containing 4 μg ml −1 polybrene while centrifuging at 1,000 rcf for 1 h at 33 °C (spinfection). The next day, cells were transferred to Cell Factory System (Thermo Fisher Scientific, 140360), and puromycin-containing medium (1 μg ml −1 ) was added for 7 days to eliminate uninfected cells and achieve genome-edited cell pools. After selection, cells were pooled together and divided into two technical replicates of 30 million cells each corresponding to the library baseline control at day 0. Another 30 million cells were replated into one Cell Factory System and positive selection was performed by growing cells for additional 15 days, at which point crisis-associated cell death was frequent. Two technical replicates of 30 million cells each were prepared by collecting cells at day 15, and their genomic DNA was extracted using a modified version of QIAGEN's DNeasy Blood and Tissue Kit provided by the FLI-Seq Library Prep for CRISPR kit (Eclipse Bioinnovations). Genomic DNA was also extracted from the baseline count control sample at day 0. DNA fragments containing the sgRNA sequences were first captured from sheared gDNA, amplified by PCR using the FLI-Seq Library Prep for CRISPR kit (Eclipse Bioinnovations) and processed for next-generation sequencing. CRISPR libraries were multiplexed, normalized and pooled for sequencing. To compensate for low base diversity in CRISPR libraries, high-diversity libraries or PhiX Control v3 Library were spiked in for sequencing and the libraries were sequenced on the HiSeq 2500 system (Illumina) as single reads. Image analysis and base calling were performed using Illumina CASAVA v.1.8.2 on the HiSeq 2500 system and sequenced reads were quality-tested using FASTQC. Fold changes of sgRNA read counts before (day 0) and after (day 15) enrichment were calculated with the PinAPL-Py software and hits were defined as genes with a read count ratio in the survival pool to the baseline control log 2 [fold change] ≥ 2 in both replicates. GO enrichment analysis was performed using the WEB-based Gene Set Analysis Toolkit (WebGestalt). The top 20 GO terms with an FDR value of <0.05 were considered to be statistically significant and visualized using the ClueGO and ggplot2 R packages.

Whole-transcriptome analysis
Total RNA was isolated using TRIzol (Invitrogen, 15596018) and purified using the RNAeasy Mini Kit (Qiagen, 74106) according to the manufacturer's instructions. Genomic DNA was eliminated by double digestion with DNase I (RNase-free DNase Set, Qiagen, 79256). The quality of the isolated total RNA was assessed using the Agilent TapeStation 4200 and RNA-seq libraries were prepared with 500 ng total RNA using the TruSeq stranded mRNA sample preparation kit according to the manufacturer's protocol (Illumina). RNA-seq libraries were multiplexed, normalized and pooled for sequencing. The libraries were sequenced on the HiSeq 4000 system (Illumina) as 50 bp single reads or the NextSeq 500 system (Illumina) as 75 bp single reads. Image analysis and base calling were performed using Illumina CASAVA v.1.8.2 on the HiSeq 4000 system and sequenced reads were quality-tested using FASTQC.
Sequenced reads were quality-tested using FASTQC (v.0.11.8) 68 and mapped to the hg19 human genome using the STAR aligner (v.2.5.3a) 62 with the default parameters. Raw or transcripts per kilobase million (TPM) gene expression was quantified across all the exons of RefSeq genes with analyzeRepeats.pl in HOMER (v.4.11.1) 63 , which used the top-expressed isoform as proxy for gene expression. Differential gene expression was performed on the raw gene counts with the R package, DESeq2 (v.1.24.0) 69 , using replicates to compute within-group dispersion. Differentially expressed genes were defined as having a FDR < 0.05 and a log 2 [fold change] > 0.585 (~1.5 fold) when comparing two experimental conditions. GO enrichment analysis was performed using the WEB-based Gene Set Analysis Toolkit" (WebGestalt) and the R package clusterProfiler. The top 20 GO terms with an FDR value of <0.05 were considered to be statistically significant and visualized using the ClueGO and ggplot2 R packages. Heat maps and Venn diagrams were generated using the R packages ComplexHeatmap and VennDiagram, respectively.

ELISA assay
One million cells were seeded in a 10 cm plate for 24 h. Conditioned cell culture medium was collected and centrifuged at 14,000 rpm for 10 min at 4 °C to remove cellular debris. Three technical replicates of 50 μl each were collected for the assay. For preparation of standards, GlutaMax-DMEM (Gibco, 10569-010) supplemented with 0.1 mM non-essential amino acids (Corning, 25-025-Cl) and 15% fetal bovine serum (Thermo Fisher Scientific, SH3007103) was used. IFNβ secretion in cell culture supernatant was quantified using the IFNβ enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems Human IFNβ Quantikine ELISA Kit, DIFNB0) according to the manufacturer's instruction. Supernatants from the treated cells were collected and incubated in IFNβ ELISA kit for 2 h at room temperature and washed three times with wash buffer. The optical density at 450 nm for each sample was measured with a microplate reader and data obtained were plotted against a four-parameter logistic standard curve to determine the concentration of IFNβ.

Chemical reagents
The reagents were as follows: recombinant human IFN-beta protein:

Statistical analysis
Statistical analysis was performed using Prism 9. Comparisons between two groups were performed using unpaired two-tailed Student's t-tests. Multiple comparisons were performed using one-way ANOVA followed by Tukey's or Dunnett's multiple-comparisons test. For all representative findings, two or three independent experiments were performed, and similar results were obtained. Significance in all figures is denoted as follows: *P < 0.05, **P < 0.01, ***P < 0.001.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Corresponding author(s): Gerald S. Shadel, Jan Karlseder Last updated by author(s): Nov 16, 2022 Reporting Summary Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated Our web collection on statistics for biologists contains articles on many of the points above.

Software and code
Policy information about availability of computer code Data collection n/a Data analysis n/a For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A description of any restrictions on data availability -For clinical datasets or third party data, please ensure that the statement adheres to our policy RNA-sequencing data are available through the Gene Expression Omnibus (