Dysregulation of the DNA damage response by phosphorothioate antisense oligonucleotides

TL;DR

This paper shows how certain antisense oligonucleotides disrupt DNA repair processes, leading to toxic DNA damage.

Contribution

The study reveals that phosphorothioate ASOs trigger DNA repair enzyme activation and condensate formation without DNA damage.

Findings

Phosphorothioate ASOs bind DNA repair enzymes and form nuclear condensates.

ASO-induced condensates activate DNA damage response and disrupt repair processes.

This dysregulation leads to toxic DNA lesions and cell cycle checkpoint activation.

Abstract

Phosphorothioate (PS)-modified antisense oligonucleotides (ASOs) are widely used to modulate gene expression in basic research and therapy. Within cells, these ASOs seed nuclear structures with unclear functions and consequences. At DNA breaks, endogenous nucleotide polymers drive the assembly of biomolecular condensates that recruit repair proteins, but the underlying mechanism(s) and effects on repair enzyme activation are poorly understood. Here, we show that ASOs bind to DNA-PKcs, ATM, and PARP1, triggering phase separation and formation of nuclear condensates containing ASOs and these essential repair enzymes. Condensates assembly is stimulated by ASO concentration and ATM activity, while limited by DNA-PKcs activity. Notably, these condensates become enzymatically active and erroneously elicit the DNA damage response in the absence of DNA damage, activating cell cycle checkpoints, disturbing endogenous repair and causing accumulation of toxic DNA lesions. These findings uncover mechanisms for ASO toxicity and the activation of DNA repair enzymes by nucleotide polymers.

Antisense oligonucleotides (ASOs) are used in research and therapy. Here, the authors show that PS-modified ASOs bind key DNA repair enzymes and form nuclear condensates that dysregulate the DNA damage response, disrupt DNA repair, and cause toxic DNA lesions, revealing mechanisms underlying ASO-induced toxicity.

Figures

D). The extent of PS body formation varied among ASOs, which could be explained by different affinities for proteins^[19](#CR19)^. Moreover, the PS backbone was required for the formation of PS bodies and filaments, as neither formed by ASOs with a phosphodiester (PO) backbone (Supplementary Fig. 1E, F), as previously reported^[24](#CR24),[26](#CR26),[27](#CR27)^. ASOs with a mixed PS/PO backbone, containing PS modification only at the termini, were also unable to form PS bodies and filaments (Supplementary Fig. 1F). Taken together, under standard transfection conditions, ASOs with a PS backbo

Activated DNA-PKcs, ATM, ATR, and PARP1 accumulate in ASO-seeded PS bodies and ASOs modulate DNA damage signaling in vivo.A Western blot of indicated DNA repair proteins extracted from U2OS control cells (No ASO) or from cells transfected with ASO 25 nM or 50 nM for 24 h. The graphs show the densitometric quantification of phosphorylated proteins, after background subtraction, normalized to their total counterparts, and to No ASO (means ± SD, *n* = 4–6 independent experiments). *(pDNA-PK^S2056^/DNA-PK) *p* = 0.0198, *(pATM^S1981^/ATM) left to right *p* = 0.0405, *p* = 0.0475, *(pATR^T1989^/ATR

F). This revealed that ASO co-precipitated DNA-PKcs, ATM, and PARP1, as well as the marker protein NONO, but not ATR or β-actin (Fig. [3F, G](#Fig3), Supplementary Fig. 3G, H). ASOs with PS backbone showed stronger affinity for these proteins than ASOs with a PO backbone, which underscores the PS modification as the mediator of this binding. TCP1β and nucleolin were only detected in ASO pull-down from lysates (Fig. [3F, G](#Fig3)), but not after transfection into cells (Supplementary Fig. 3G, H). We also verified that biotinylated ASOs with a PS backbone seeded PS bodies upon transfection into

A). In contrast, inhibition of ATR (ATRi) or PARP1 (PARPi) had no significant effect (Fig. [4A](#Fig4)). ATMi reduced PS body formation (Supplementary Fig. 4B), but not maintenance of pre-formed bodies (Supplementary Fig. 4C) while PARPi had no effect on either (Supplementary Fig. 4B, C). The number of PS bodies *per cell* remained unchanged in cells that still had them post-inhibitor treatment (Supplementary Fig. 4D). Assessment of ASO nuclear intensity as a proxy for its uptake showed no reduction by ATMi (Supplementary Fig. 4E), ruling out this indirect effect.Fig. 4PS bodies are regulated

Activation of repair enzymes by ASOs is indirect but linked to chromatin-association and the microenvironment of PS bodies.A In vitro assay of DNA-PK kinase activity employing recombinant human DNA-PK and double-stranded DNA, ASO or PO ASO, detected by Western blotting. This experiment was performed three times with similar results. The asterisks (*) next to the blot indicate degradation products and/or a smaller isoform of phosphorylated DNA-PKcs, as determined by MS analysis^[11](#CR11)^. B In vitro assay of PARP1 PARylation activity employing recombinant human PARP1 and double-stranded DNA,

A, B). pRPA32^S4/S8^ (target of DNA-PK) was enriched in PS bodies (Fig. [6B](#Fig6), Supplementary Fig. 6A), while total RPA32 was not (Supplementary Fig. 6C), indicating phosphorylation of chromatin-associated RPA, rather than recruitment of free/soluble RPA into these bodies. Downstream repair factors BRCA2, RAD51, XRCC4 and DNA ligase 4 were also detected in PS bodies (Fig. [6B](#Fig6), Supplementary Fig. 6A). XRCC1, a PAR-binding protein, was also enriched in PS bodies while other PAR-associated proteins (FUS, EWSR1, TAF15, TDP-43) were absent (Fig. [6B](#Fig6), Supplementary Fig. 6A, C).

A). Concurrently with ASO treatment leading to reduced RAD51 levels (Supplementary Fig. 7B), we also observed decreased assembly of the upstream protein BRCA2 in ASO-treated cells (Fig. [7D](#Fig7)), despite unchanged BRCA2 expression (Supplementary Fig. 7B). Since BRCA2 is responsible for loading RAD51 onto resected DNA^[2](#CR2)^, this suggest that impaired RAD51 recruitment may result from disrupted BRCA2 function rather than reduced expression. Supporting this, siRNA-mediated RAD51 depletion did not significantly affect BRCA2 assembly at FokI break sites (Supplementary Fig. 7C), despite re

Full text

Introduction

The cellular DNA damage response involves DNA damage recognition followed by signaling to initiate repair locally and to regulate cellular responses globally. Central to this response are the phosphatidylinositol 3-kinase-related protein kinases (PIKK): DNA-dependent protein kinase catalytic subunit (DNA-PKcs), Ataxia-telangiectasia mutated (ATM), and Ataxia telangiectasia and Rad3-related protein (ATR), which upon detection of DNA damage phosphorylate numerous proteins to activate repair and cell cycle checkpoints^1,2^. Another key player is Poly(ADP-ribose) polymerase 1 (PARP1), which rapidly binds DNA breaks and synthesizes negatively charged poly-ADP-ribose (PAR) chains that recruit downstream factors^3^.

Upon DNA damage, phosphorylation and PARylation by the above-mentioned enzymes promote concentration of repair factors into structures known as repair foci. Emerging evidence suggests that these foci form via liquid-liquid phase separation (LLPS)^4,5^, driven by multivalent interactions between nucleotide polymers (often RNA or PAR chains) and intrinsically disordered proteins (often RNA/PAR-binding proteins)^4,6,7^. The tendency of proteins to phase separate can be influenced by phosphorylation and its resulting change of charge^8^, emphasizing both phosphorylation and PARylation in the regulation of repair condensates. However, the molecular details underlying these processes and effects on repair enzyme activation are not clear.

While initiation of DNA damage signaling generally involves DNA damage, this response can also be initiated without lesions, for example, through persistent chromatin binding of upstream repair factors ATM or ATR/ATRIP^9,10^. Moreover, in the absence of DNA lesions, DNA-PK and PARP1 can be activated by RNA^11–13^, or by synthetic DNA mimicks disrupting DNA repair fidelity and sensitizing cells to genotoxic stress^14,15^.

Antisense oligonucleotides (ASOs) are synthetic, negatively charged, single-stranded nucleic acid polymers, commonly used to silence gene expression in research and therapy. Their ability to target virtually any disease-related gene product has made them promising drug candidates, with several ASOs already approved drugs and many in development^16,17^. Most ASOs are chemically modified with a sulfur replacing a non-bridging oxygen in the phosphodiester bond - known as a phosphorothioate (PS) modification^16,18^. This enhances ASO stability, cellular uptake, and protein interactions, especially with RNA- and DNA-binding proteins^19^. Consequently, ASOs can mislocalize proteins and disrupt their function, known to cause nucleolar stress and cell death^20–22^. ASOs also accumulate in different cellular compartments, such as nucleoli, stress granules, and paraspeckles, interfering with their normal cellular processes^23–25^.

Furthermore, ASOs can induce new nuclear structures, including filaments and PS bodies, where the ASO itself is present. The ability to form these structures depends on ASO concentration, PS modifications and the single-stranded attribute, but not on sequence or sugar backbone (i.e., RNA, DNA or 2’ modifications), although both sequence and sugar modifications can influence the propensity of ASOs to form these structures^24,26,27^. While nuclear filaments are linked to paraspeckle protein interactions^24^, less is known about PS bodies. Since their discovery in 1998^26^, only one protein - TCP1β, a subunit of the TCP1 (TRiC/CCT) complex involved in folding aggregation-prone proteins^28,29^ - has been identified within them^27^. However, the molecular composition, mechanisms of ASO sequestration, and the functional consequences of PS bodies remain largely uncharacterized.

Here, we show that ASOs bind to DNA repair proteins, including DNA-PKcs, ATM, and PARP1, triggering phase separation and formation of condensates resembling repair foci, identified as PS bodies, even in the absence of DNA breaks. This activates the DNA damage response, leading to cell cycle arrest, suppression of homologous recombination (HR), and increased radiosensitivity. Based on these results, we propose a model where high amounts of ASOs trigger a phase separation event that acts as a scaffold for the multimerization of DNA-PKcs, ATM, ATR, and PARP1 on chromatin. Through this mechanism, ASOs induce activation of repair enzymes and initiate a physiological DNA damage response, independent of DNA damage.

Results

ASOs can seed nuclear structures, of which PS bodies are the most prevalent type

To examine the intracellular distribution of ASOs in human cells, we transfected U2OS cells with fluorescently labeled ASOs (with fluorescein amidite (FAM) at the 3’ end), and visualized their localization using fluorescence microscopy (Fig. 1A). The ASO used was a gapmer (i.e., contained a central DNA fragment, flanked on each side by locked nucleic acids (LNAs)) with a scrambled 15 nucleotide sequence and a PS backbone - commonly used as a non-targeting control in knockdown experiments. To enhance visualization of nuclear structures, soluble proteins were removed using cytoskeleton (CSK) buffer prior to paraformaldehyde (PFA) fixation.Fig. 1ASOs with PS backbones can seed nuclear structures, with PS bodies being the most prevalent type.A Schematic illustration of the PS backbone modification in ASOs and the experimental workflow. B Immunostaining of nucleolin, TCP1β or NONO in U2OS control cells (No ASO) or in cells transfected with 3’FAM ASO (ASO-FAM) 50 nM for 24 h. Filaments are marked by white arrowheads. Scale bar (white line) = 20 μm. Intensity line scans across transfected nuclei show enrichment of ASO in nucleoli, PS bodies and filaments (gray zones). (a.u., arbitrary units). C Assessment of ASO localization by fluorescent microscopy in U2OS cells transfected with ASO-FAM 50 nM for 24 h. The graph shows the percentage of cells with nuclei containing ASO in PS bodies, nuclear filaments or nucleoli ( ≥ 100 cells per experiment, means ± SD, n = 3). D Quantification of PS body formation by TCP1β immunostaining of U2OS control cells (No ASO) or of cells transfected with ASO-FAM 50 nM for 6 or 24 h. The graph shows the percentage of cells with nuclei containing PS bodies ( ≥ 100 cells per experiment, means ± SD, n = 3). E Quantification of PS body formation by TCP1β immunostaining of U2OS cells transfected with ASO-FAM at indicated concentrations for 24 h. The graph shows the percentage of cells with nuclei containing PS bodies ( ≥ 100 cells per experiment, means ± SD, n = 3). F Quantification of PS body formation by TCP1β immunostaining of U2OS cells transfected with indicated ASOs 50 nM or ASO#3 (αMALAT1) 25 nM, for 24 h. The graph shows the percentage of cells with nuclei containing PS bodies ( ≥ 100 cells per experiment, means ± SD, n = 3). Source data are provided as a Source Data file.

Following transfection of cells for 24 hours with 50 nM ASO, conditions commonly used for ASO-mediated gene silencing, the ASO was detected in the nucleus, enriched in the nucleolus (marked by nucleolin; Fig. 1B), but not in other endogenous nuclear organelles, e.g., Cajal bodies (marked by coilin), PML bodies (marked by PML) or splicing speckles (marked by SC35; Supplementary Fig. 1A). In addition, the ASO seeded distinct nuclear structures of spherical or filamentous shape that were absent in mock-transfected cells: PS bodies (marked by TCP1β) and nuclear filaments (marked by NONO; Fig. 1B). PS bodies were the most common site of ASO localization (91% of cells), followed by filaments (57%) and nucleolar accumulation (49%; Fig. 1C). The localization of ASOs was unaffected by cell permeabilization or fixation (Supplementary Fig. 1B, C).

Having established that PS bodies form in almost all cells following ASO transfection, the features of these structures were further evaluated. PS bodies formed rapidly (within hours post ASO transfection; Fig. 1D) and in a concentration-dependent manner (with more cells containing PS bodies at higher concentrations; Fig. 1E). They also formed by non-fluorescent ASOs of different sequences, length, and type (e.g., scramble, gene targeting, or mixmer with alternating LNA/DNA units; Fig. 1F and Supplementary Table 1), which excludes dependence on fluorophore, specific sequence, and RNase H-mediated target degradation in PS body formation. Additionally, PS bodies formed by ASOs composed solely of either RNA or DNA with a PS backbone (i.e. lacking LNA; Supplementary Fig. 1D). The extent of PS body formation varied among ASOs, which could be explained by different affinities for proteins^19^. Moreover, the PS backbone was required for the formation of PS bodies and filaments, as neither formed by ASOs with a phosphodiester (PO) backbone (Supplementary Fig. 1E, F), as previously reported^24,26,27^. ASOs with a mixed PS/PO backbone, containing PS modification only at the termini, were also unable to form PS bodies and filaments (Supplementary Fig. 1F). Taken together, under standard transfection conditions, ASOs with a PS backbone (here referred to as just ASOs) can localize to nucleoli and induce the formation of PS bodies and nuclear filaments.

ASO treatment cause activation of DNA repair enzymes and these proteins are enriched in ASO-seeded PS bodies

In lysates of ASO treated cells, phosphorylation of repair factors DNA-PKcs, ATM, and ATR was elevated, with more phosphorylation at higher ASO concentrations (Fig. 2A), suggesting that ASO treatment triggers enzyme activation.Fig. 2. Activated DNA-PKcs, ATM, ATR, and PARP1 accumulate in ASO-seeded PS bodies and ASOs modulate DNA damage signaling in vivo.A Western blot of indicated DNA repair proteins extracted from U2OS control cells (No ASO) or from cells transfected with ASO 25 nM or 50 nM for 24 h. The graphs show the densitometric quantification of phosphorylated proteins, after background subtraction, normalized to their total counterparts, and to No ASO (means ± SD, n = 4–6 independent experiments). *(pDNA-PK^S2056^/DNA-PK) p = 0.0198, *(pATM^S1981^/ATM) left to right p = 0.0405, p = 0.0475, *(pATR^T1989^/ATR) p = 0.0169, ns = not significant as determined by One-sample two-tailed t-test. The asterisks next to the blot indicate degradation products and/or a smaller isoform of phosphorylated DNA-PKcs, as determined by MS analysis^11^ and utilizing recombinant DNA-PK (Fig. 5A). B Immunostaining of pDNA-PKcs^S2056^ (ab124918) in U2OS control cells (No ASO) or in cells transfected with indicated ASOs 50 nM for 24 h. Scale bar (white line) = 10 μm. Intensity line scan across transfected nuclei shows enrichment of pDNA-PKcs in ASO-positive PS bodies (gray zones), n > 3. C Immunostaining of indicated DNA repair proteins in U2OS cells transfected with ASO-FAM 50 nM for 24 h. Scale bar (white line) = 5 μm. Intensity line scans show enrichment of all indicated repair factors except total DNA-PKcs (MA5-13238), in ASO-positive PS bodies (gray zones), n > 3. The dashed line separates factors detected in PS bodies from those that are not. D Immunostaining of pDNA-PKcs^S2056^ and pATR^T1989^ in mouse brain tissue isolated 48 h after ICV injection of Cy3-labeled ASO#7. Brain regions with low or high ASO uptake were selected based on cytoplasmic accumulations of ASO. Scale bar (white line) = 20 μm. The graphs show the nuclear intensity of pDNA-PKcs^S2056^ (ab18192) and pATR^T1989^ (GTX128145) as determined by CellProfiler, and normalized to the image mean ( > 300 nuclei per condition). Outliers were identified and removed using the ROUT method. The solid lines are medians and the dashed lines represent quartiles. ****p < 0.0001 as determined by Welch’s two-tailed t-test. E Schematic illustration of ASO localization in the nucleus and proteins enriched in ASO-seeded PS bodies. Source data are provided as a Source Data file.

Besides the ASO itself, and the TCP1β chaperone protein, the composition of PS bodies is unknown^27^. Immunostaining revealed that phosphorylated (p)DNA-PKcs co-localizes with ASO-seeded PS bodies in U2OS cells (Fig. 2B) and across multiple other cell types (HeLa, RPE-1, MelJuSo; Supplementary Fig. 2A). The colocalization was independent of ASO sequence (Fig. 2B). However, while the phosphorylated form of DNA-PKcs was enriched in PS bodies, and observed with different antibodies towards different sites of DNA-PKcs phosphorylation (i.e., at S2056 or T2609; Supplementary Fig. 2B), the total counterpart of DNA-PKcs did not accumulate in these structures (Fig. 2C, Supplementary Fig. 2B), indicating that phosphorylation occurs locally during PS body formation.

In addition to pDNA-PKcs, the related PIKK enzymes, ATM and ATR, also accumulated in PS bodies (Fig. 2C), in both phosphorylated and total forms. This location was not affected by cell fixation procedure (Supplementary Fig. 2C). ATRIP, the obligate binding partner of ATR, was also enriched in PS bodies, and so were PARP1 and PAR chains (Fig. 2C). Although these repair proteins were present in PS bodies in most cells (Supplementary Fig. 2D), none were detected in filaments (Supplementary Fig. 2E). Ku70/Ku80 (co-factors of DNA-PKcs), RAD50/NBS1 (co-factors of ATM), and TOPBP1 (co-factor of ATR/ATRIP) were not detected in either structure (Supplementary Fig. 2F, G).

To assess ASO-mediated DNA damage response activation in vivo, wild-type mice were treated with naked Cy3-labeled ASOs via intracerebroventricular (ICV) injection. At 48 hours post-injection, ASOs were detected in brain cells, predominantly within cytoplasmic vesicles (Supplementary Fig. 2H), consistent with previous reports^30–32^. Nuclear ASO accumulations were observed in some cells (Supplementary Fig. 2H), but lacked the characteristic PS body pattern seen in cultured cells, likely due to lower concentrations achieved through free uptake of ASOs. These nuclear enrichments did not clearly co-localize with PS body components pDNA-PKcs or pATR (Fig. 2D). Attempts to assess co-localization with pATM, γH2AX or TCP1β were inconclusive due to suboptimal antibody staining.

ASOs are known to act in the nucleus despite low detection in this compartment^31,33^ and their nuclear activity has been shown to correlate with their uptake efficiency, as assessed by cytoplasmic accumulations^32^. Therefore, we compared pDNA-PKcs and pATR nuclear intensity in brain regions with low versus high ASO uptake. Regions with high ASO uptake exhibited elevated levels of pDNA-PKcs and pATR compared to regions with low uptake (Fig. 2D). This positive correlation indicates that ASOs can influence the DNA damage response also in vivo and at the lower nuclear concentrations attained using clinically relevant delivery methods (i.e., naked ASOs). Furthermore, the presence of DNA damage signaling in the absence of visible PS bodies indicates that ASOs can affect the DNA damage response through multiple mechanisms, some independent of PS bodies.

Together, these findings demonstrate that DNA repair proteins are activated following ASO treatment, and DNA-PKcs, ATM, ATR and PARP1 localize with ASOs in PS bodies (Fig. 2E).

PS bodies are not sites of DNA damage and DNA-PKcs, ATM, and PARP1 bind directly to ASOs

To understand how repair enzymes localize to ASO-seeded PS bodies in cells, we first investigated whether these structures form at DNA damage sites, known to trigger local assembly of repair factors. However, established DNA damage markers MDC1, 53BP1, and BRCA1 were absent from PS bodies (Fig. 3A) and bromodeoxyuridine (BrdU) incorporation showed no signs of DNA resection at these sites, while often occurring at DNA lesions (Fig. 3B, Supplementary Fig. 3A). The terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay also failed to detect DNA breaks within PS bodies (Supplementary Fig. 3B, C).Fig. 3PS bodies do not contain DNA double-strand lesions, and repair proteins bind to ASOs.A Immunostaining of indicated DNA repair proteins in U2OS cells transfected with ASO-FAM 50 nM for 24 h. Scale bar (white line) = 5 μm. Intensity line scan across transfected nuclei shows no enrichment of these factors in ASO-positive PS bodies (gray zones), n > 3. B Immunostaining of BrdU and TCP1β in U2OS cells treated with BrdU for 24 h followed by transfection with ASO 50 nM for another 24 h. Scale bar (white line) = 5 μm. Intensity line scan across ASO transfected nuclei shows no enrichment of BrdU in PS bodies (gray zones). C Quantification of PS body formation by TCP1β immunostaining of U2OS cells transfected with ASO 50 nM for 24 h, irradiated with 2 Gy and recovery for 1 h or 24 h. The graph shows the percentage of cells with nuclei containing PS bodies, normalized to the No IR condition ( ≥ 100 cells per experiment, means ± SD, n = 6). ns = not significant as determined by One-sample two-tailed t-test. D Assessment of ASO localization by fluorescence microscopy in U2OS FokI cells transfected with ASO-FAM 50 nM for 24 h and FokI-mediated DNA break induction (i.e., treatment with Shield-1 and 4-hydroxytamoxifen) for an additional 4 h. No permeabilization prior to fixation. Scale bar (white line) = 20 μm. Intensity line scan across transfected nuclei shows exclusion of ASO from FokI site (gray zone). n > 3. E Experimental workflow of pull-down of biotinylated ASOs after incubation with nuclear lysates. F Pull-down of proteins bound to indicated biotinylated ASOs added to U2OS nuclear cell lysates at high (300 μM KCl) or low (137 μM KCl) salt conditions and detected by Western blotting. This experiment was performed three times and is quantified in (G). G Densitometric quantification of protein levels in (F) (normalized to background, and where a value of 1.0 indicates no pull-down of the protein by the ASO; means ± SD, n = 3 independent experiments). Source data are provided as a Source Data file.

Furthermore, PS body formation was unaffected by DNA damage induced via irradiation (IR) or the AsiSI endonuclease (Fig. 3C, Supplementary Fig. 3D), and neither PS bodies nor ASO co-localized with DNA double-strand breaks generated by the FokI or AsiSI endonucleases (Fig. 3D, Supplementary Fig. 3E). Instead, ASOs were excluded from all FokI- and some AsiSI-damage sites (Fig. 3D, Supplementary Fig. 3E), for unknown reasons. Although we cannot entirely rule out the presence of DNA lesions in PS bodies, our findings indicate that these structures are not sites of DNA damage and form independently of DNA breaks, which suggests a different mechanism for assembly of repair enzymes in PS bodies.

We next examined whether binding of repair enzymes to ASOs could provide an explanation. Utilizing biotinylated ASOs with PS or PO backbone, we performed pull-down of ASOs, added either directly into cell lysates or via transfection into cells (Fig. 3E, Supplementary Fig. 3F). This revealed that ASO co-precipitated DNA-PKcs, ATM, and PARP1, as well as the marker protein NONO, but not ATR or β-actin (Fig. 3F, G, Supplementary Fig. 3G, H). ASOs with PS backbone showed stronger affinity for these proteins than ASOs with a PO backbone, which underscores the PS modification as the mediator of this binding. TCP1β and nucleolin were only detected in ASO pull-down from lysates (Fig. 3F, G), but not after transfection into cells (Supplementary Fig. 3G, H). We also verified that biotinylated ASOs with a PS backbone seeded PS bodies upon transfection into cells (Supplementary Fig. 3I).

Increased salt concentration revealed that binding of PARP1 and NONO to ASOs was salt-sensitive, indicating interactions derived from electrostatic forces. DNA-PKcs and ATM maintained strong binding even at higher salt concentrations (Fig. 3G), indicative of high affinity interactions, and/or involvement of additional forces such as hydrophobic interactions. Together, these findings indicate that PS bodies are not formed at DNA damage sites and that DNA-PKcs, ATM, and PARP1 directly bind ASOs, supporting a model in which assembly of repair enzymes in PS bodies occurs via ASO binding rather than DNA damage.

PS body formation is regulated by ATM and DNA-PK activity, and these bodies exhibit liquid-like properties

To explore the role of repair enzymes in PS body assembly, we inhibited DNA-PK, ATM, ATR and PARP activity. Inhibition of ATM (ATMi) abolished PS bodies in more than half of the cells, while inhibition of DNA-PK (DNA-PKi) enhanced their formation (Fig. 4A). This was verified by another set of ATM and DNA-PK inhibitors (Supplementary Fig. 4A). In contrast, inhibition of ATR (ATRi) or PARP1 (PARPi) had no significant effect (Fig. 4A). ATMi reduced PS body formation (Supplementary Fig. 4B), but not maintenance of pre-formed bodies (Supplementary Fig. 4C) while PARPi had no effect on either (Supplementary Fig. 4B, C). The number of PS bodies per cell remained unchanged in cells that still had them post-inhibitor treatment (Supplementary Fig. 4D). Assessment of ASO nuclear intensity as a proxy for its uptake showed no reduction by ATMi (Supplementary Fig. 4E), ruling out this indirect effect.Fig. 4PS bodies are regulated by ATM and DNA-PK activity and form via LLPS.A Quantification of PS body formation by TCP1β immunostaining of U2OS cells transfected with ASO 50 nM and co-treated with indicated inhibitors for 24 h. The graph shows the number of cells with nuclei containing PS bodies, normalized to the No inhibitor condition (i.e., only ASO; ≥100 cells per experiment, means ± SD, n = 5-6). **left to right p = 0.0018, p = 0.0028, ns = not significant as determined by One-sample two-tailed t-test. B Fluorescence recovery after photobleaching (FRAP) analysis of PS bodies in U2OS cells transfected with 5’Cy3 ASO (Cy3-ASO). Scale bar (white line) = 5 μm. The graph shows the fractional intensity recovery (mean ± SD) of 10 PS bodies from 5 different cells, after background subtraction and normalization to pre-bleach intensity and bleaching depth. C Immunostaining of TCP1β and pATM in U2OS cells transfected with ASO 50 nM for 24 h followed by treatment with 1,6-hexanediol or ammonium acetate, prior to fixation, for the time points indicated. Scale bar (white line) = 20 μm. D Quantification of the TCP1β immunostaining in (C). The graph shows the number of cells whose nuclei contain TCP1β-positive PS bodies normalized to the Control condition (i.e., only ASO; ≥100 cells per repeat, means ± SD, n = 5-9). **left to right p = 0.0023, p = 0.0020, p = 0.0075, as determined by One-sample two-tailed t-test. E Quantification of the fraction of cells with pATM detected in the remaining TCP1β-positive PS bodies in (C, D). Percentages refer to the amount of TCP1β-positive PS bodies with detectable pATM and significance tests are based on deviations from 100%. ( ≥ 100 cells per repeat, means ± SD, n = 5–9). ****p < 0.0001, **p = 0.0027, ns = not significant as determined by One-sample two-tailed t-test. F Immunostaining of TCP1β in U2OS cells transfected with ASO 50 nM for 24 h followed by treatment with RNase A or DNase I. Scale bar (white line) = 20 μm. The graph shows the number of cells with nuclei containing PS bodies, normalized to the Control condition (i.e., no nuclease treatment; ≥100 cells per repeat, means ± SD, n = 3). ***p = 0.0001 as determined by One-sample two-tailed t-test. Source data are provided as a Source Data file.

Surprisingly, the removal of ATM, DNA-PKcs, ATR, PARP1 or TCP1β by siRNA did not influence the number of PS bodies (Supplementary Fig. 4F, G). Moreover, PS bodies formed to a similar extent in DNA-PKcs^-/-^ CHO cells (Supplementary Fig. 4H), in ATM^-/-^ RPE-1 cells (Supplementary Fig. 4I), and following co-depletion of DNA-PKcs and ATM (Supplementary Fig. 4F, Supplementary Fig. 4J, K). This indicates that the role of ATM and DNA-PKcs in PS body assembly can be compensated by other factors in the absence of these proteins, and that the influence of ATMi and DNA-PKi on PS bodies require the presence of these enzymes.

Next, the role of LLPS in PS body assembly and in the concomitant accumulation of repair factors was investigated. Live-cell imaging of Cy3-labeled ASOs showed that PS bodies fused into larger droplets (Supplementary Video 1). Photobleaching of regions containing PS bodies was followed by a rapid recovery of fluorescence and the reappearance of the bleached PS bodies (around 50% recovery after 10 s, plateauing at 60% within 1 min; Fig. 4B), consistent with these bodies exhibiting liquid-like properties. To further test for LLPS characteristics, ASO-treated cells were exposed to 1,6-hexanediol (4% for up to 15 min, disrupting hydrophobic interactions), or high salt (50 mM ammonium acetate for 3 min, disrupting electrostatic interactions). Both treatments removed PS bodies (marked by TCP1β) in around half of all cells, and particularly 1,6-hexanediol reduced the size of remaining PS bodies (Fig. 4C, D). We conclude that PS bodies are LLPS condensates, mediated by both hydrophobic and electrostatic interactions.

Examination of pATM in the same samples (Fig. 4C, D) showed that high salt removed pATM in PS bodies to a similar extent as TCP1β (Fig. 4E). In contrast, 1,6-hexanediol caused additional loss of pATM from remaining bodies (Fig. 4E). Although this loss could be explained by a less efficient detection of pATM at these sites, it may reflect a difference in the phase separation properties of pATM, compared to TCP1β. Expression levels of TCP1β and ATM were unchanged by either 1,6-hexanediol or 50 mM ammonium acetate treatment (Supplementary Fig. 4L), excluding such indirect effects.

To investigate the dependency of PS bodies on RNA or DNA, cells were exposed to RNase A or DNase I which resulted in a significant loss of PS bodies by RNase A treatment, and a non-significant reduction by DNase I (Fig. 4F). Altogether, these findings demonstrate that PS bodies are LLPS condensates that are dependent on endogenous RNA and on ATM activity.

Activation of repair enzymes by ASOs is linked to chromatin-association and the microenvironment of PS bodies

Employing an in vitro kinase assay of DNA-PK (consisting of recombinant DNA-PKcs and Ku70/Ku80), we tested the ability of ASO to directly activate this enzyme. While double-stranded DNA resulted in robust activation and autophosphorylation of DNA-PKcs at S2056 and T2609, ASO failed to induce activation (Fig. 5A). Similar results were obtained for PARP1, using an in vitro PARylation assay, which showed activation by DNA but not by ASO (Fig. 5B). At high concentrations, ASOs even prevented the activation of DNA-PK and PARP1 by DNA (Fig. 5C, D). We conclude that ASOs are unable to directly stimulate the catalytic activities of DNA-PKcs and PARP1, at least in vitro, indicating another mode of activation of these enzymes in PS bodies.Fig. 5. Activation of repair enzymes by ASOs is indirect but linked to chromatin-association and the microenvironment of PS bodies.A In vitro assay of DNA-PK kinase activity employing recombinant human DNA-PK and double-stranded DNA, ASO or PO ASO, detected by Western blotting. This experiment was performed three times with similar results. The asterisks (*) next to the blot indicate degradation products and/or a smaller isoform of phosphorylated DNA-PKcs, as determined by MS analysis^11^. B In vitro assay of PARP1 PARylation activity employing recombinant human PARP1 and double-stranded DNA, ASO or PO ASO, detected by Western blotting. This experiment was performed three times with similar results. C In vitro assay of DNA-PK kinase activity in the presence of double-stranded DNA and 1-200 ng ASO, detected by Western blotting. This experiment was performed three times with similar results. D In vitro assay of PARP1 PARylation activity in the presence of double-stranded DNA and 1-200 ng ASO, detected by Western blotting. This experiment was performed three times with similar results. E Western blot of indicated DNA repair proteins extracted from chromatin fractions of U2OS control cells (No ASO) or from cells transfected with ASO 25 nM or 50 nM for 24 h. The graphs show the densitometric quantifications of phosphorylated proteins, after background subtraction, normalized to their total counterparts, or in the case of ATR normalized to β-actin, and to No ASO (means ± SD, n = 3-5 independent experiments). **(pDNA-PK^S2056^/DNA-PK) p = 0.0064, *(pDNA-PK^S2056^/DNA-PK) p = 0.0363, *(pATM^S1981^/ATM) p = 0.0210, *(pATR^T1989^/ β-actin) p = 0.0481, ns = not significant as determined by One-sample two-tailed t-test. F Immunostaining of γH2AX in U2OS cells transfected with ASO-FAM 50 nM for 24 h. Scale bar (white line) = 5 μm. Intensity line scan shows enrichment of γH2AX in ASO-positive PS bodies (gray zones), n > 3. G Immunostaining of indicated repair proteins in U2OS cells transfected with ASO-FAM 50 nM for 1 h, n > 3. Scale bar (white line) = 5 μm. Source data are provided as a Source Data file.

We next tested whether ASOs activate DNA-PKcs by disrupting its interaction with inhibitory RNAs. Cross-linking and immunoprecipitation (CLIP) of DNA-PKcs showed no change in DNA-PKcs binding to scaRNA2, an endogenous inhibitor of this enzyme^34^ or snoRNA U3, another partner RNA of DNA-PKcs^11^ (Supplementary Fig. 5A), ruling out this mechanism.

Given that prolonged chromatin association can activate ATM or ATR/ATRIP in the absence of DNA damage^9,10^, we examined chromatin of ASO-treated cells. The levels of (p)DNA-PKcs, (p)ATM and ATR were increased in chromatin, which occurred in an ASO dose-dependent manner (Fig. 5E), in accordance with PS body formation (Fig. 1E). We hypothesized that the association of PS bodies with chromatin along with the accumulation of DNA-PKcs, ATM and ATR within these bodies could explain this elevation. Supporting this, PS bodies remained in cells following pre-extraction of soluble cellular factors with detergent (Fig. 1B) but were lost following chromatin stripping (Supplementary Fig. 5B), indicating their chromatin association. Moreover, H2AX, a core chromatin component and substrate of DNA-PKcs, ATM, and ATR, was phosphorylated (at S139 and referred to as γH2AX) at sites of PS bodies (Fig. 5F). We conclude that PS bodies associate with chromatin, supporting the idea that chromatin association can contribute to enzyme activation.

To further explore the relationship between PS bodies and repair enzyme activation, we examined newly formed PS bodies. Notably, within one hour of ASO transfection, and concomitant with initial PS body detection, these structures already contained active forms of DNA-PKcs, ATM and ATR (Fig. 5G), suggesting an early role in repair enzyme activation. Since phase-separated condensates can act as kinase activators by stimulating their confirmational changes and/or their oligomerization^35,36^, these findings support a model in which ASOs activate repair enzymes indirectly - through chromatin-association and/or the LLPS environment of PS bodies.

ASO-mediated activation of DNA damage response triggers cell cycle arrest by ATM

To determine which of DNA-PKcs, ATM, and ATR are responsible for H2AX phosphorylation in PS bodies, inhibition of the enzymes was combined with ASO treatment, followed by a measurement of nuclear γH2AX intensity. Inhibition of DNA-PK or ATM lowered γH2AX intensity by 30-40%, while ATRi had no significant effect (Fig. 6A). As expected, PARPi had no effect γH2AX levels (Fig. 6A).Fig. 6ASOs trigger activation of DNA damage response, and ATM-mediated cell cycle arrest.A Quantification of γH2AX by immunostaining of U2OS cells transfected with ASO 50 nM and co-treated with indicated inhibitors for 24 h. The graph shows the nuclear γH2AX intensity as determined by CellProfiler and normalized to background and the No inhibitor condition (i.e. only ASO; > 50 cells per experiment, means ± SD, n = 3). The background was defined as the area between 10 and 25 pixels outside of each nucleus. **p = 0.0021, *p = 0.0436 as determined by One-sample two-tailed t-test. B Illustration of downstream DNA repair proteins enriched in PS bodies. Immunofluorescence images are shown in Supplementary Fig. 6A. C Western blot of proteins extracted from U2OS control cells (No ASO) or transfected with ASO (24 h). The graph shows the densitometric quantification of phosphorylated CDK substrates, after background subtraction, normalized to β-actin levels, and to No ASO (means ± SD, n = 5 independent experiments). **p = 0.0043, p = 0.0215 as determined by One-sample two-tailed t-test. In the CDK substrate motif, X denotes any one of the 20 amino acids and S is the phosphorylation site. D Assessment of cell cycle distribution of U2OS control cells (No ASO) or transfected with ASO (50 nM, 24 h), pulsed with EdU (30 min) followed by EdU detection (click chemistry) and Cyclin A2 immunostaining. The graph shows percentage of cells in G1, S and G2 phases as determined by CellProfiler ( > 50 cells per experiment, means ± SD, n = 6). ***left to right p = 0.0005, p = 0.0008, ns = not significant as determined by two-way ANOVA and Šidák’s multiple comparisons test. E Quantification as for (D) but after co-treatments with ASO and inhibitors for 24 h. The graph shows the percentage of cells in G2 phase ( > 50 cells per experiment, means ± SD, n = 3-9). ***p = 0.0006, *p = 0.0141, ns = not significant as determined by two-way ANOVA and Šidák’s multiple comparisons test. All phases are shown in Supplementary Fig. 6E. F tSNE-NN maps of RNA-seq samples obtained from U2OS treated as indicated for 24 h. Maps are coloured based on each sample’s Infomap cluster identity (left) or treatment (right). The bar graph shows the overlap enrichment of each treatment group within the four infomap clusters. G Graph showing up- or downregulated genes (determined by DESeq2) in U2OS cells treated with 12.5–75 nM ASO for 24 h and compared to No ASO or PO ASO-treated cells. H Graphs showing GO term fold enrichment of genes up- or downregulated in the RNA-seq samples described above: ASOs 12.5–75 nM ASO over No ASO samples. P-values < 0.05 are shown and were determined using PANTHER Fisher’s exact overrepresentation test (without correction). Hashed bars indicate that zero genes in that group matched the gene ontology term. Source data are provided as a Source Data file.

To extend these observations, we assessed signaling downstream of the PIKK enzymes. Phosphorylated CHK1^S317^ (ATR target), and CHK2 (ATM target) showed enrichment in PS bodies, which resembled their natural assembly in repair foci (Fig. 6B, Supplementary Fig. 6A, B). pRPA32^S4/S8^ (target of DNA-PK) was enriched in PS bodies (Fig. 6B, Supplementary Fig. 6A), while total RPA32 was not (Supplementary Fig. 6C), indicating phosphorylation of chromatin-associated RPA, rather than recruitment of free/soluble RPA into these bodies. Downstream repair factors BRCA2, RAD51, XRCC4 and DNA ligase 4 were also detected in PS bodies (Fig. 6B, Supplementary Fig. 6A). XRCC1, a PAR-binding protein, was also enriched in PS bodies while other PAR-associated proteins (FUS, EWSR1, TAF15, TDP-43) were absent (Fig. 6B, Supplementary Fig. 6A, C). Thus, the activation of DNA-PK, ATM, ATR and PARP1 in PS bodies is functional and elicits DNA damage signaling.

Activation of ATM, ATR, and DNA-PK can all induce checkpoint signaling (i.e., G1/S, intra-S and G2/M), to stop cell cycle progression until DNA is properly repaired, involving inhibition of cyclin-dependent kinases (CDKs)^1^. Western blotting, employing an antibody against the phospho-serine substrate motif of CDKs, revealed that ASO-treatment reduced CDK activity in a dose-dependent manner (Fig. 6C).

Assessment of cell cycle progression through measurements of EdU incorporation (S-phase marker) and Cyclin A2 levels (S/G2-phase marker) showed reduced S phase entry and accumulation in G2 phase (EdU-negative and Cyclin A2-positive), following ASO treatment (Fig. 6D), consistent with G2/M checkpoint activation. Moreover, since the number of cells in G1 phase (EdU and Cyclin A2-negative) was not reduced after ASO treatment - as would be expected if they were arrested in G2 phase and could proceed normally into S phase - the G1/S checkpoint was likely activated as well (Fig. 6D). Notably, inhibition of ATM, but not of DNA-PK or ATR, prevented ASO-induced G2 arrest (Fig. 6E), identifying ATM as the key mediator of this effect. ATMi alone did not alter G2 phase levels (Supplementary Fig. 6D) and the reduction in S phase could not be rescued by inhibiting any of the PIKK enzymes (Supplementary Fig. 6E). Together these findings demonstrate that ASO-seeded PS bodies elicit DNA damage signaling via PIKK enzymes, leading to checkpoint activation and ATM-dependent cell cycle arrest.

Dysregulation of genes involved in DNA damage signaling and repair following ASO treatment

To investigate the global cellular responses to ASO treatment, we performed RNA sequencing analysis on U2OS cells treated with different ASO variants and concentrations for 24 hours. We initially mapped all samples based on the most variably expressed genes across the dataset (using a t-stochastic neighbor embedding-nearest neighbor (tSNE-NN) analysis), which demonstrated dose-dependent spatial clustering of the samples (Fig. 6F). While all No ASO and PO ASO samples clustered in Cluster 1, cells treated with increasing doses of ASOs formed Clusters 2 and 3 (Fig. 6F, Supplementary Fig. 6F). Interestingly, cells treated with ASO#2, composed of a different sequence than ASO, formed a separate Cluster 4 (Fig. 6F), which, although positioned near Cluster 3, underscores the sequence-specific effect of ASOs on the transcriptome. Assessment of gene-expression changes, using DESeq2, showed that ASO treatment induced up- and downregulation of genes in a dose-dependent manner, both in comparison to No ASO and PO ASO samples, with more pronounced changes observed at higher concentrations (Fig. 6G, Supplementary Fig. 6G).

To gain a deeper understanding of the cellular processes influenced by ASO treatment, gene ontology (GO) analysis was performed on the differentially expressed genes. In line with our findings that ASOs impact the DNA damage response, several genes associated with DNA damage signaling and repair pathways were significantly altered - both upregulated and downregulated - following ASO treatment (Fig. 6H). Moreover, ASO treatment led to significant downregulation of genes involved in cell cycle regulation, consistent with the observed ASO-induced alterations in cell cycle progression. Furthermore, an upregulation of genes associated with protein unfolding (Fig. 6H) and additional changes across other cellular processes were observed as well (Supplementary Fig. 6H). Altogether, these findings demonstrate that ASOs modulate gene expression in a dose- and sequence-dependent manner, including genes related to DNA damage signaling, repair and cellular stress responses, indicating that ASOs could affect endogenous DNA repair.

ASO-mediated activation of DNA damage response impairs homologous recombination and sensitizes cells to IR

To explore whether ASOs alter repair of double-strand breaks, we first examined the clearance of 53BP1 repair foci after IR. 53BP1 was chosen as a marker of DNA damage, because this protein does not accumulate in PS bodies, which could be mistaken for repair foci (Fig. 3A). While most damage had been resolved within 24 hours in control cells, a significant amount of residual damage remained in cells pre-treated with ASO (Fig. 7A). In addition, in non-irradiated conditions, the number of 53BP1 foci was elevated after ASO treatment (Fig. 7A): a sign of accumulation of spontaneous DNA damage, which was confirmed by comet assay (Fig. 7B).Fig. 7ASO treatment impairs HR repair, reduces cell survival, and sensitizes cells to IR.A Quantification of 53BP1 repair foci by immunostaining of U2OS control cells (No ASO) or transfected with ASO (50 nM, 24 h), followed either by irradiation (2 Gy) and 24 h recovery (residual damage) or no IR and instead immediate fixation (spontaneous damage). The graph shows the number of cells with nuclei containing >10 53BP1 foci ( ≥ 100 cells per experiment, means ± SD, n = 3-4). ***p = 0.0004, *p = 0.0136 as determined by two-way ANOVA and Šidák’s multiple comparisons test. B Fluorescent images of alkaline comet assay of U2OS control cells (No ASO) or transfected with ASO (50 nM, 48 h). The graph shows the tail moment (tail DNA percent multiplied by tail length; ≥ 100 tails per experiment, n = 3). The center line represents the median, the boxes the 25^th^ to 75^th^ percentile, the whiskers the 10^th^ to 90^th^ percentile, and the plotted values are outside this range. ****p < 0.0001 as determined by Welch’s two-tailed t-test. C Immunostaining of indicated repair proteins in U2OS control cells (No ASO) or transfected with ASO (50 nM, 24 h), irradiated with 2 Gy and 1 h recovery. No permeabilization prior to fixation. IR-induced foci quantification as in (A) (n = 3). ***p = 0.0002, ns = not significant as determined by unpaired two-tailed t-test. D Immunostaining of indicated repair proteins in U2OS FokI control cells (No ASO) or transfected with ASO (50 nM, 24 h), and FokI-mediated DNA break induction for another 4 h. No permeabilization prior to fixation. The graph shows the number of cells in which the repair protein co-localizes with the FokI break site ( ≥ 50 cells per experiment, means ± SD, n = 5). *left to right p = 0.0301, p = 0.0401, p = 0.0231, ns = not significant as determined by unpaired two-tailed t-test. E Assessment of HR repair efficiency in U2OS DR-GFP cells treated as indicated and determined by flow cytometry. The graph shows values normalized to the corresponding control (i.e., No ASO for ASOs or siControl for siRAD51; means ± SD, n = 3-6 independent experiments). ****p < 0.0001, **p = 0.0020, ns = not significant as determined by One-sample two-tailed t-test. F Assessment of viability of U2OS cells treated as indicated (ASO 0-100 nM, 72 h) and determined by CellTiter-Glo^®^ 2 assay. The graph shows values normalized to 0 nM (means ± SD, n = 3 independent experiments). ****p < 0.0001, ***p = 0.0001, *p = 0.0278 as determined by two-way ANOVA and Šidák’s multiple comparisons test. G Assessment of viability of U2OS cells: No ASO or ASO transfected (25-50 nM, 6 h), followed by irradiation and 66 h recovery and determined by CellTiter-Glo^®^ 2 assay. The graph shows values normalized to the No IR condition (means ± SD, n = 3 independent experiments). ****p < 0.0001 as determined by two-way ANOVA and Šidák’s multiple comparisons test. H Schematic model: ASOs with PS backbones bind to DNA repair proteins DNA-PKcs, ATM and PARP1, which triggers LLPS and formation of PS bodies containing ASOs and these proteins. Activation of the repair enzymes in PS bodies elicits DNA damage signaling independent of DNA lesions. ASOs interfere with endogenous DNA repair by HR and impair recruitment of the BRCA2/RAD51 complex to RPA-covered resected DNA. The dysregulation of DNA damage signaling and repair by ASOs result in cell cycle arrest, radiosensitivity, and ultimately cell death. Source data are provided as a Source Data file.

To gain further insight into the effects of ASOs on endogenous DNA repair, recruitment of repair proteins to sites of damage was assessed. One hour post-IR exposure, foci containing γH2AX, pATM and 53BP1 formed to the same extent in control and in ASO-treated cells (Fig. 7C). However, RAD51 foci, essential for HR repair, were significantly reduced (Fig. 7C). This impairment of RAD51 assembly by ASO treatment was also observed at double-strand breaks generated by the FokI endonuclease (Fig. 7D), and at sites of spontaneous damage (in non-irradiated conditions), with more severe impairments at higher ASO concentrations (Supplementary Fig. 7A). Concurrently with ASO treatment leading to reduced RAD51 levels (Supplementary Fig. 7B), we also observed decreased assembly of the upstream protein BRCA2 in ASO-treated cells (Fig. 7D), despite unchanged BRCA2 expression (Supplementary Fig. 7B). Since BRCA2 is responsible for loading RAD51 onto resected DNA^2^, this suggest that impaired RAD51 recruitment may result from disrupted BRCA2 function rather than reduced expression. Supporting this, siRNA-mediated RAD51 depletion did not significantly affect BRCA2 assembly at FokI break sites (Supplementary Fig. 7C), despite reducing RAD51 levels more efficiently than ASO treatment (Supplementary Fig. 7D). Additionally, BRCA1 accumulation, known to cooperate with RAD51 and BRCA2 during HR, remained unchanged at breaks sites in both ASO-treated and RAD51-depleted cells (Fig. 7D, Supplementary Fig. 7D). Thus, ASOs appear to interfere specifically with BRCA2-dependent RAD51 loading.

In contrast to RAD51, RPA32 (marker of end-resected DNA) was elevated at damage sites in cells treated with ASOs (Fig. 7D). This may reflect increased resection - as would be expected by the higher proportion of ASO-treated cells in G2 phase (Fig. 6D) where HR repair is active - and/or defective RPA32-to-RAD51 replacement (mediated by BRCA2), also pinpointing BRCA2 as the disrupted step in HR at which ASOs interferes.

To determine whether ASOs impair actual repair of DNA breaks by HR or only the associated signaling, we measured the rate of this repair pathway in U2OS HR reporter cells containing a direct repeated (DR)-GFP sequence^37^. In these cells, expression of I-SceI endonuclease introduces a single double-strand break, the repair of which by HR generates a functional coding sequence for GFP. Notably, ASO treatment reduced HR efficiency by 60%, comparable to RAD51 depletion (positive control; Fig. 7E, Supplementary Fig. 7E). PO-modified ASOs had no effect on HR repair (Fig. 7E), highlighting the PS backbone as the cause of HR disruption.

HR is critical for cell survival and cellular recovery from both endogenous and exogenous DNA damage. ASO treatment reduced cell viability in a concentration- and PS backbone-dependent manner (Fig. 7F). Moreover, pre-treatment with ASOs sensitized cells to IR, and enhanced IR-induced cell death (Fig. 7G), consistent with a compromised ability to repair DNA double-strand breaks. Taken together, these results demonstrate that ASOs impair recruitment of key HR factors to DNA breaks, disrupt repair, reduce cell survival, and sensitize cells to genotoxic stress.

Discussion

The current investigation reveals previously unknown effects of ASOs on cellular DNA damage signaling and repair. Following transfection of cells with ASOs, these molecules bind to DNA-PKcs, ATM, and PARP1, triggering LLPS and the formation of nuclear bodies in which ASOs and repair enzymes are enriched. Assembly of DNA-PKcs, ATM, ATR, and PARP1 in ASO bodies and/or the association of these bodies with chromatin, appears sufficient to activate the repair enzymes and erroneously elicit the DNA damage response in the absence of DNA damage. Moreover, in ASO-treated cells, HR factors are not recruited properly to damaged chromatin, leading to an impairment in associated repair, accumulation of DNA breaks, sensitization to DNA damage, and reduced cell viability (Fig. 7H).

The local concentration of repair factors in PS bodies bears a striking resemblance to the natural formation of repair foci at DNA lesions. Given that ASOs are negatively charged single-stranded nucleic acids, the mechanisms involved in assembly of PS bodies could elucidate the role of endogenous nucleotide polymers in DNA repair. Our findings demonstrate that assembly of PS bodies is driven by LLPS and occurs when the ASO-protein interaction network reaches a threshold concentration. The amount of ASO required for this threshold effect may differ based on the properties of the ASO, but all PS ASOs tested in this study formed PS bodies at nanomolar concentration (50 nM) and condensate assembly followed a concentration-dependent behavior. Proteins of this ASO-associated network include the DNA repair enzymes DNA-PKcs, ATM, ATR, and PARP1, all of which are activated in PS bodies. This activation does not appear to result from DNA lesions, or to be mediated by the ASO directly. Rather, condensation of these proteins by ASOs, and/or their prolonged association with chromatin, could be underlying triggers, and similar activation mechanism(s) might be employed by natural nucleotide polymers. Although this remains to be proven, phase-separated condensates are increasingly being recognized as platforms for kinase activation^35,38^.

Moreover, the presence of the TCP1ß chaperone in PS bodies could be indicative of protein (mis)folding and/or aggrephagy at these sites, the latter being a process that removes toxic protein aggregates and which is regulated by TCP1ß monomers^39^. This idea is further supported by our observation that genes involved in the cellular response to unfolded protein were upregulated following ASO treatment (Fig. 6H).

Kinases can function as co-drivers or scaffolds of LLPS due to their intrinsic tendency to undergo LLPS^35^. In the ASO-associated protein network, ATM and DNA-PKcs appear to play more dominant roles and function as molecular switches that either trigger or limit ASO-dependent LLPS, respectively. This is based on the observation that ATM and DNA-PKcs bind abundantly to ASOs, even at high salt concentrations, indicative of high affinity. Moreover, catalytic inhibition of ATM results in the disappearance of these condensates in more than half of the cells, while after inhibition of DNA-PKcs more cells form PS bodies. Notably, these effects occur only when ATM or DNA-PKcs is catalytically inactivated, and not following depletion of these enzymes. Such enhanced effects by kinase inhibition, compared to depletion, are known for these factors and are associated with trapping of enzymatically inhibited proteins on chromatin in various complexes or in certain conformational states, all of which may interfere with potential compensatory mechanisms^40^. Our finding that pDNA-PKcs, but not total DNA-PKcs, is enriched in PS bodies, suggest that PS bodies assemble at sites of DNA-PKcs-ASO interactions. This would be analogous to H2AX phosphorylation, one of the earliest events in DNA damage signaling, occurring at sites of DNA damage where repair foci later form.

Our finding that phosphorylation by ATM stimulates formation of PS bodies is in agreement with the previous observation that these bodies require activities and/or components of living cells for formation^26^. It could involve both phospho-dependent recruitment of proteins to these condensates and/or fine-tuning of their LLPS properties. Moreover, this suggests that phosphorylation at DNA lesions, in addition to stimulation of protein recruitment, might also function to regulate LLPS of repair condensates.

Regarding the properties of ASO that influence its binding to DNA-PKcs, ATM, and PARP1, we found that the PS backbone of ASO plays a key role. This is consistent with this modification promoting protein binding, which has been shown to require at least 10 PS moieties in ASOs^27^, and to be mediated by the sulfur substitution lengthening the bond, and spreading the negative charge compared to oxygen, making nonpolar interactions more favorable^18,19^. Interestingly, PS modifications exist naturally on both DNA and RNA^41,42^, raising the possibility that these modifications may be involved in DNA repair. Our findings that PS-modified DNA and RNA ASOs trigger PS body formation in cells further supports that if present naturally, this modification could aid in repair factor condensate formation. In eukaryotes, PS modifications were only recently identified, with unknown function(s)^41^, whereas in prokaryotes, they are widespread on DNA and have multiple functions, such as protection from invading foreign DNA^43^, control of gene expression^43^, regulation of cellular redox homeostasis, and involvement in the oxidative stress response^43–45^. Moreover, the susceptibility of PS modification to oxidation can result in DNA break formation and genomic instability^46^.

The enrichment of DNA-PKcs, ATM, ATR, and/or PARP1 in PS bodies, opens the possibility that specialized sub-condensates containing these factors may form also at DNA lesions. Several kinds of condensates have been described at sites of DNA damage, seeded by different types of polymers. This includes LLPS of the repair protein 53BP1^47^, which can be seeded by RNA transcribed from break sites^48^, and PAR chains that engage RNA-binding FET proteins to form condensates that exclude 53BP1^49^. Moreover, nuclear microtubule filaments can trigger LLPS of RAD52, and clustering of DNA breaks^50^. In the case of PARP1, recent work demonstrates that this protein can form condensates together with DNA and thus mediate the synapsis of broken DNA ends^51^. What endogenous nucleic acid polymers could trigger LLPS of DNA-PKcs, ATM, and ATR in cells remain to be shown.

Although ASOs bind to DNA-PKcs, ATM and PARP1, accumulation of repair factors occurs only in PS bodies, not at other ASO-enriched sites, such as nuclear filaments. The assembly of ASOs at these sites are likely mediated through interaction with different kinds of proteins. At least 80 intracellular proteins have previously been determined to bind to ASOs with a PS backbone^19^, including NONO and nucleolin, which are present in filaments and nucleoli. In addition, we find that ASOs are not enriched at DNA break sites, but instead excluded, even though the concentration of repair factors is high at these sites. As endogenous RNAs can inhibit repair enzymes in the presence of DNA damage, with pronounced consequences not only for DNA repair, but also for cell differentiation and stem cell renewal^34,52^, such exclusion of ASO from break sites could be a means of protection. This is highly relevant for both DNA-PK and PARP1, which are known to be inhibited by scaRNA2^34^ and SNORA73A/B^52^, respectively, and shown here also to be inhibited by ASOs at high concentrations. Alternatively, ASO-bound repair proteins may not be perceived as functional, and non-ASO bound counterparts are therefore enriched at break sites.

We show that ASOs impair DNA repair by HR and connect this to reduced assembly of BRCA2 and RAD51 at break sites. Such impairment could have multiple causes, for example result from the sequestration of RAD51 and BRCA2 in PS bodies or from the downregulation of genes involved in double-strand break repair - particularly those associated with HR repair - as revealed by our RNA-seq analysis. However, the exact mechanism remains to be determined. In accordance with defective DNA repair, ASO treatment results in accumulation of DNA breaks (both endogenous and IR-induced), sensitization to IR, and enhanced cell death. Thus, dysregulation of DNA repair can be added to the list of factors involved in ASO toxicity.

It is important to note the discrepancy between ASO delivery via transfection and via methods used clinically, as transfection results in significantly higher concentrations of ASOs in the nucleus. Nonetheless, even at lower concentrations achieved through clinically relevant delivery, ASOs seem sufficient to potentially activate DNA damage signaling, despite not inducing PS body formation. This suggests that ASOs may affect the DNA damage response through multiple mechanisms, including pathways independent of PS bodies. Such mechanisms could involve ASO binding to DNA repair enzymes and inhibition of their DNA-dependent activation (Fig. 5C, D) and transcriptional changes associated with ASO treatment (Fig. 6F–H). While further validation in larger clinical cohorts is warranted, it is important to consider the therapeutic implications of our findings at this stage, given that several ASO drugs are already approved, and many more are in late-stage clinical development^17^. Moreover, as the efficacy of ASO therapeutics increasingly depend on enhanced cellular uptake and overcoming endosomal entrapment, the risk for interference with DNA repair is likely to increase. Our discovery that ASOs modulate DNA damage signaling and repair is therefore of considerable value, both to the academic community and for the development of better ASO-based therapies.

Methods

Cells, culture conditions and treatments

The cells and culture conditions used in this study are listed in Supplementary Table 2. All culture media was supplemented with 10% fetal bovine serum (Thermo Fisher Scientific). Cells were grown and maintained at 37 °C and under 5% CO_2_ in humidified incubators. Cells were plated one day prior to treatment at a concentration of 40 000 cells/ml, unless otherwise stated. When indicated, inhibitors were used at the following concentrations: DNA-PKi (AZD-7648; used unless otherwise stated) 10 μM, DNA-PKi (AZD-7648) 10 μM, DNA-PKi (NU7441) 4 μM, ATMi (KU55933; used unless otherwise stated) 10 μM, ATMi (KU-60019) 10 μM, ATRi (VE-821) 5 μM, PARPi (Olaparib) 10 μM, with treatment duration of 24 h, unless otherwise stated. 5-Bromo-2′-deoxyuridine (BrdU) was used at a concentration of 10 μM for 48 h. Conditions and treatments specific for certain methods are stated in each section below.

ASO/siRNA transfections

ASOs and siRNAs employed in this study are listed in Supplementary Table 1. ASO concentrations and transfection times are indicated in figure legends or specified in each method section. siRNAs were transfected at a concentration of 15 nM for 48 h. Both ASOs and siRNAs were transfected into cells using 3 μl/ml HiPerFect (Qiagen) in accordance with the manufacturer’s recommendations. “ASO” without a specified number refers to unlabeled ASO#1 and “No ASO” refers to treatment with HiPerFect alone.

Oligonucleotide synthesis and purification

All Cy3-labeled oligonucleotides were synthesized by phosphoramidite solid-phase synthesis using a K&A H8-SE synthesizer. Universal solid support (CUTAG CPG, 1000 Å, Sigma-Aldrich) and all DNA, RNA, and LNA phosphoramidites, along with capping reagents, oxidation reagent, deblocking solution, and dry acetonitrile, were obtained from commercial sources (Sigma-Aldrich, Thermo Fisher Scientific, and Glen Research). To obtain thiophosphates, the oxidation step was replaced with a sulfurization step using 0.05 M N,N-dimethyl-N’-(3-thioxo-3H-1,2,4-dithiazole-5-yl)formimidamide (Sulfurizing Reagent II). The sulfurization reaction was extended to 6 min to ensure satisfactory yields. For 5’-end labeling, a Cy3 group was introduced using MMT-Cy3-phosphoramidite (Glen Research). Following synthesis, the oligonucleotides were cleaved from the support and deprotected. The oligonucleotides were treated with excess 25% aqueous ammonia at 55 °C for 16 h, after which the solid support was filtered off, and the solution was purged with argon to remove residual ammonia. The full-length oligonucleotides were purified by reverse-phase high-performance liquid chromatography (RP-HPLC) on a C18 column (XBridge® Prep C18, 5 μm, OBD, 19 × 100 mm, Waters). A gradient elution was applied in 50 mM triethylammonium acetate buffer (pH 7.0), starting with 10% acetonitrile and increasing to 70% over 8 min, followed by a further increase to 100% acetonitrile over the next 7 min. The flow rate was maintained at 8 mL/min, with UV detection at 260 nm and mass analysis performed using electrospray ionization mass spectrometry (ESI-MS). Purified oligonucleotides were freeze-dried, redissolved in 1.5 mL of Milli-Q water, and desalted using NAP columns (NAP™−10, Cytiva) according to the standard desalting protocol. Final products were quantified by measuring absorbance at 260 nm, analyzed by liquid chromatography-mass spectrometry (LC-MS, Q-TOF), and concentrated again by freeze-drying. The sequence and composition of all synthesized oligonucleotides are presented in Supplementary Table 1.

Immunofluorescence microscopy

Cells were grown on sterilized coverslips and thereafter either fixed directly with 4% paraformaldehyde (PFA) for 15 min, followed by permeabilization with 0.1% Triton X-100 in phosphate-buffered saline (PBS) for 5 min, or alternatively, if soluble proteins were to be removed, cells were pretreated with cytoskeleton buffer (CSK; 10 mM pipes (pH 7.0; Sigma-Aldrich), 100 mM NaCl, 300 mM sucrose, 3 mM MgCl_2_, 0.7% Triton X-100 (Thermo Fisher Scientific)) for 3 min and then fixed with 4% PFA in PBS, all steps at room temperature. CSK buffer was employed unless otherwise stated. The coverslips were treated with blocking buffer (2% bovine serum albumin (BSA; Sigma-Aldrich), 5% glycerol (Sigma-Aldrich), 0.2% Tween20 (Sigma-Aldrich), 0.1% NaN_3_ (Sigma-Aldrich)) for 30 min at room temperature, and incubated with primary antibodies overnight at 4 °C. Coverslips were washed with PBS and incubated with Alexa fluor-conjugated secondary antibody in blocking buffer for 1 h at room temperature. All antibodies utilized are listed in Supplementary Table 3. Coverslips were incubated with 0.1 μg/ml 4’,6-diamidino-2-phenylindole (DAPI) for 20 min at room temperature, washed with PBS, and mounted with Mowiol 4-88 (CSH protocols) mounting media. Images were acquired with a LSM700 confocal microscope (Zeiss), mounted on Axio observer.Z1 (Zeiss) equipped with a Plan-Apochromat 63x/1.4 oil immersion lens, and then processed with the Zen 2012 Black software (Zeiss). Images were analyzed using CellProfiler (Version 4.2.6)^53^ and ImageJ (Version 1.53t)^54^.

Phase separation experiments: cells on coverslips were treated with CSK buffer for 3 min at room temperature followed by treatment with 4% 1,6-hexanediol (Sigma-Aldrich) for 5 or 15 min, or treatment with 50 mM ammonium acetate (Sigma-Aldrich) for 3 min, also at room temperature. Thereafter the coverslips were fixed and further processed as described above.

Nuclease experiments: cells on coverslips were treated with CSK buffer for 5 min at room temperature, washed and then treated either with 300 μg/ml RNase A (Thermo Fisher Scientific) in PBS for 10 min or with 50 U/ml DNase I (Sigma-Aldrich) in PBS for 20 min, both at 37 °C. Thereafter the coverslips were fixed and further processed as described above.

Chromatin stripping experiments: cells were plated on coverslips at a concentration of 150 000 cells/ml, and 24 h later transfected with 100 nM ASO-FAM for 24 h. Coverslips were treated with CSK buffer for 1 min, followed by high-salt CSK buffer (10 mM pipes (pH 7.0), 500 mM NaCl, 300 mM sucrose, 3 mM MgCl_2_, 0.1% Triton X-100) for 1 min, two washes with DNase I reaction buffer (Sigma-Aldrich) in PBS, and incubation with 50 U/ml DNase I in PBS for 20 min at 37 °C. The coverslips were then treated again with high-salt CSK buffer for 1 min, washed with PBS, and fixed and further processed as described above.

EdU and TUNEL labeling

Cells were grown on sterilized coverslips for both 5-ethynyl 2´-deoxyuridine (EdU) labeling and Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL). EdU labeling: cells were pulsed with 10 μM EdU (Sigma-Aldrich) for 30 min, fixed with 4% PFA for 15 min at room temperature and washed twice with 3% BSA in PBS. Coverslips were immersed in 0.5% Triton X-100 in PBS for 20 min at room temperature, washed twice with 3% BSA in PBS and incubated with EdU labeling solution (100 mM Tris-HCl (pH 6.8; Thermo Fisher Scientific), 1 mM CuSO_4_ (Sigma-Aldrich), 10 mM ascorbic acid (Sigma Aldrich), and Alexa Fluor™ 488 Azide (Thermo Fisher Scientific)), for 30 min at room temperature and protected from light. Coverslips were washed twice with 3% BSA in PBS and thereafter immersed in blocking buffer and further staining was performed according to “Immunofluorescence microscopy” above. TUNEL labeling: cells were fixed with 4% PFA for 15 min at room temperature and the Click-iT™ TUNEL Alexa Fluor Imaging Assays for Microscopy & HCS kit (Thermo Fisher Scientific) was utilized, in accordance with the manufacturer’s instructions. Thereafter, the coverslips were immersed in blocking buffer and further staining was performed according to “Immunofluorescence microscopy” above.

The FokI system

The U2OS FokI cells contain a stably integrated LacO array and continuously express the mCherry-LacI-FokI protein fused to a destabilization domain (DD) and a modified estrogen receptor (ER; ER-mCherry-LacI-FokI-DD)^55^. This enables inducible nuclear expression of ER-mCherry-LacI-FokI-DD upon exposure of these cells for 4 h to 1 μM Shield-1 (Takara Bio) ligand (which stabilizes the DD domain and thereby induces expression of ER-mCherry-LacI-FokI-DD) and 1 μM 4-hydroxytamoxifen (Sigma-Aldrich; which results in translocation of the protein to the nucleus). Immunofluorescence staining was performed according to “Immunofluorescence microscopy” above.

The AsiSI system

The AsiSI U2OS cells stably express the AsiSI restriction enzyme fused to an ER that relocalize to the nucleus upon treatment with 300 nM 4-hydroxytamoxifen for 4 h. The AsiSI enzyme generates double-strand breaks at sequences 5′-GCGATCGC-3’. Immunofluorescence staining was performed according to “Immunofluorescence microscopy” above.

Ionizing irradiation

Cells were exposed to ionizing irradiation (IR; approximately 1.3 Gray (Gy)/min) from a CIX2 Xstrahl X-ray machine (Xstrahl; through X-ray Irradiation Core Facility at Karolinska Institutet), utilizing a 3 mm aluminum filter, a focus-to-specimen distance (FSD) of 40 cm and the settings of 195 kV and 10 mA. The exact dose is indicated for each experiment.

Live-cell imaging

U2OS cells were plated in polymer microscopy dishes (ibidi) at a concentration of 100 000 cells/ml and 24 h later transfected with 100 nM Cy3-ASO in FluroBrite™ DMEM (Thermo Fisher Scientific), supplemented with 10% fetal bovine serum, 1 mM sodium puruvate (Thermo Fisher Scientific), and 2 mM L-Glutamine (Thermo Fisher Scientific), for 24 h. During imaging cells were kept at 37 °C and 5% CO_2_. Images were acquired each minute during 1 h with a Nikon CrEST X-Light V3 Spinning Disk Confocal Microscope equipped with a 60x oil objective, and then processed with the NIS-Elements software. Images were acquired as Z-stacks, were denoised by the NIS-Elements software and processed as maximum intensity projections by ImageJ (Version 1.53t)^54^.

Fluorescence recovery after photobleaching (FRAP)

U2OS cells were plated in polymer microscopy dishes at a concentration of 150 000 cells/ml and 8 h later transfected with 100 nM Cy3-ASO in supplemented FluroBrite™ DMEM for 16 h. Images were acquired using a LSM980-Airy (Zeiss) equipped with a 63x/1.4 oil immersion lens and Zen Blue (v3.3) software and 100% transmission of the 561 nm laser was used for bleaching. Images were acquired every second from 10 s before bleaching to 60 s after bleaching. Images were processed by ImageJ (Version 1.53t)^54^, after which background intensity was subtracted and the intensity was normalized to pre-bleached intensity and to bleaching depth.

Sub-cellular fractionation

Nuclear fractionation: cells were harvested by scraping and lysed in Buffer A (5 mM pipes, 85 mM KCl, 0.5% NP40 (Thermo Fisher Scientific)) for 10 min on ice, followed by centrifugation at 2352 x g (5000 rpm) for 5 min at 4 °C. The supernatant (cytoplasmic fraction) was discarded, and the remaining nuclear pellet was washed once in Buffer A without NP40 (5 mM pipes, 85 mM KCl), then resuspended in NP40 buffer high salt (400 mM NaCl, 50 mM Tris-HCl (pH 8.0; Thermo Fisher Scientific), and 1% NP40, 1% protease inhibitor cocktail (Sigma-Aldrich)). Chromatin fractionation: after the washing step with Buffer A without NP40 described for nuclear fractionation, the nuclear pellet was instead resuspended in NP40 buffer 137 mM (137 mM NaCl, 50 mM Tris-HCl (pH 8.0), and 1% NP40, 1% protease inhibitor cocktail), passed through a 25 G needle four times, centrifuged at 1505 x g (4000 rpm) for 10 min at 4 °C and resuspended in NP40 buffer high salt. Both nuclear or chromatin fractionation: DNA was sheared by sonication for 5 min at medium intensity (30 s ON/30 s OFF cycles) using a Bioruptor (Diagenode), followed by centrifugation at 15900 x g (13000 rpm) for 10 min at 4 °C, and the resulting supernatant was saved as the nuclear/chromatin fraction.

Western blotting

Cells were lysed in NP40 buffer high salt (400 mM NaCl, 50 mM Tris-HCl (pH 8.0), 1% NP40, and 1% protease inhibitor cocktail) for 15 min on ice. DNA was sheared by sonication for 5 min at medium intensity (30 s ON/30 s OFF cycles) using a Bioruptor (Diagenode) followed by centrifugation at 15900 x g (13000 rpm) for 10 min at 4 °C. The supernatant was acquired, and protein concentration determined using the Bradford assay (Bio-Rad). Protein samples were mixed with NuPAGE^TM^ Sample Reducing Agent and NuPAGE^TM^ LDS Sample Buffer, resolved on NuPAGE^TM^ precast gels, and transferred onto nitrocellulose membranes (all from Thermo Fisher Scientific). Membranes were incubated with primary antibodies (Supplementary Table 3) overnight at 4 °C, followed by incubation with horseradish peroxidase-conjugated secondary antibodies (Supplementary Table 3) for 1 h at room temperature. Antibodies were diluted in 5% milk containing 0.1% PBS-Tween20. The Western blots were developed using SuperSignal West Femto maximum sensitivity substrate (Thermo Fisher Scientific) according to the manufacturer’s instructions. Densitometric quantifications were performed using ImageJ (Version 1.53t)^54^ and are based on separate experiments (not on multiple exposures of the same experiment).

Phase separation experiments: cells were plated at a concentration of 50 000 cells/ml and 24 h later transfected with 50 nM ASO for 24 h. Protein was harvested as described above, and extracts were treated with 4% 1,6-hexanediol for 15 min or 50 mM ammonium acetate for 3 min, at 37 °C, and thereafter prepared for Western blotting as described above.

Pull-down of biotinylated ASO

Pull-down from nuclear lysates: nuclear fractions were obtained as described above for sub-cellular fractionation. 100-200 μg protein lysate was incubated with 80 pmol 3’biotinylated ASO or 3’biotinylated PO ASO (Supplementary Table 1) and 0.1 μg/μl yeast tRNA (Thermo Fisher Scientific) in a total volume of 500 μl in binding buffer high salt (300 mM KCl, 1.5 mM MgCL_2_, 20 mM Tris-HCl (pH 7.5; Thermo Fisher Scientific), 0.2 mM EDTA (Thermo Fisher Scientific), and 1% protease inhibitor cocktail) or binding buffer low salt (137 mM KCl, 1.5 mM MgCL_2_, 20 mM Tris-HCl (pH 7.5), 0.2 mM EDTA, and 1% protease inhibitor cocktail) with rotation overnight at 4 °C. The day after, 20 μl Dynabeads™ M-280 Streptavidin (Thermo Fisher Scientific), pre-washed once in H_2_O and twice in binding buffer, were added to each sample. After incubation for 90 min with rotation at room temperature, beads were washed four times in binding buffer, with rotation for 5 min at room temperature between each wash, and were prepared for Western blotting.

Pull-down after transfection: cells were plated with a concentration of 80 000 cells/ml and were transfected with 100 nM 3’biotinylated ASO or PO ASO for 24 h, harvested by scraping and lysates were prepared as described for Western blotting. Cells were plated at this higher density to obtain more material and consequently transfected with matching, higher concentration of ASO. Dynabeads™ M-280 Streptavidin were pre-washed once in H_2_O, twice in binding buffer and thereafter blocked overnight in 0.1 μg/μl yeast tRNA, 1% BSA, and binding buffer in a total volume of 500 ul, with rotation at 4 °C. 100 μg protein lysate was incubated with 10 μl Dynabeads™ M-280 Streptavidin in binding buffer low salt in a total volume of 500 μl, with rotation for 90 min at room temperature. The beads were washed four times in binding buffer, with rotation for 5 min at room temperature between each wash, and prepared for Western blotting.

DNA-PK in vitro assay

The DNA-PK in vitro assay was performed in 1x kinase buffer A (Thermo Fisher Scientific) containing 200 μM ATP, 80 μg/ml BSA and 25-80 U recombinant DNA-PK holoenzyme (Promega), in a final volume of 30 μl. When indicated, 10 ng linearized DNA plasmid, 240 ng ASO, or 240 ng PO ASO was added. All reactions proceeded for 60 min at 30 °C after which samples were further processed according to “Western blotting” above.

PARP1 in vitro assay

The PARP1 in vitro assay was performed in ADP-Ribosylation buffer (50 mM Tris-HCl (pH 7.5), 12.5 mM MgCl_2_, 125 mM NaCl), containing 50 ng (or 1-2 U) recombinant PARP1 (Abcam) and 100 uM NAD^+^ (New England Biolabs). When indicated 10 ng linearized DNA plasmid or 1 ng sonicated salmon sperm DNA (Thermo Fisher Scientific), 240 ng ASO, or 240 ng PO ASO was added. All reactions proceeded for 15-30 min at room temperature after which samples were further processed according to “Western blotting” above.

RNA extraction, reverse transcription, and RT-qPCR

Total RNA was extracted with TRIzol reagent (Thermo Fisher Scientific), according to the manufacturer’s instructions and using 5PRIME Phase Lock Gel heavy tubes (Quantabio). cDNA was generated with the Superscript IV reverse transcriptase, random hexamer primers, 10 nM dNTPs Mix and RNaseOUT (all from Thermo Fisher Scientific), in accordance with the manufacturer’s instructions. The levels of RNA were determined by RT-qPCR in a 96-well format utilizing a 7500 Fast Real-Time PCR System (Thermo Fisher Scientific) and 2x SYBR Green qPCR Master Mix (Selleckchem). The primers employed are listed in Supplementary Table 1.

Cross-linking immunoprecipitation (CLIP)

Cells were treated with 100 μM 4-Thiouridine (Sigma-Aldrich) for 16 h and thereafter cross-linked by exposure to UV light (365 nm, 2×400 mJ/cm^2^ in a CL-1000 Ultraviolet Crosslinker). Cells were harvested and lysed as described for Western blotting. 100 μg of total protein was incubated together with 5 μl Dynabeads^TM^ Protein G (Thermo Fisher Scientific) and 1 μg IgG or DNA-PKcs antibody (Supplementary Table 3) in a total of 600 μl in NP40 buffer (100 mM Tris-HCl (pH 8), 150 mM NaCl, 1% NP40, and 1% protease inhibitor cocktail), with rotation overnight at 4 °C. Thereafter, the beads were washed twice with cold high salt buffer (1 M NaCl, 50 mM Tris-HCl (pH 7.5), 1 mM EDTA, 1% NP40, 0.10% SDS, 0.50% sodium deoxycholate) and twice with cold CLIP wash buffer (10 mM MgCl2, 20 mM Tris- HCl (pH 7.5), 0.20% Tween 20), each wash for 10 min with rotation at 4 °C, and finally the beads were washed once more with cold NP40 buffer. For isolation of RNA, the beads were incubated with 4 U Proteinase K (Molecular Biology Grade, New England Biolabs) diluted in Proteinase K buffer (10 mM Tris-HCl (pH 7.5), 0.1 M NaCl, 1 mM EDTA, 0.20% SDS) in a final volume of 100 μl, for 60 min at 50 °C and while shaking at 1100 rpm. This was followed by addition of TRIzol reagent (Thermo Fisher Scientific) and subsequent extraction of RNA as described in “RNA extraction, reverse transcription and RT-qPCR”.

Comet assay

U2OS cells were plated at a concentration of 40 000 cells/ml and 24 h later transfected with ASO 50 nM for 48 h. Cells were harvested by trypsinization, pelleted at 500 g for 5 min at 4 °C, and resuspended in PBS at a concentration of 10^6^ cell per ml. Cell suspension was mixed 1:10 with 0.75% agarose (Thermo Fisher Scientific), at 37 °C, and added to microscope slides pre-coated with 1% agarose. To spread the mix, a coverslip was added on top. After solidification of the agarose on ice, the coverslip was removed, and the slides were incubated in lysis buffer (pH 10, 2.5 M NaCl, 100 mM EDTA, 10 mM Tris-HCl, 10% DMSO, 1% Triton X-100, 200 mM NaOH) overnight at 4 °C. Slides were incubated with cold alkaline electrophoresis buffer (300 mM NaOH, 1 mM EDTA) for 20 min, subjected to electrophoresis at 14 V, 300 mA for 35 min, followed by three washes for 2 min with cold water, and fixation in cold 70% ethanol for 5 min. Slides with cells were protected from light during all steps. After air-drying the slides, they were stained with Vista Green DNA Dye (Cell Biolabs Inc.) at room temperature for 15 min. Slides were imaged on an Invitrogen EVOS M7000 imaging system with a 10x objective. Comets were analysed using AutoComet software (version 1.0)^56^. At least 100 comets per sample were scored. Tail moment was calculated as tail DNA percent multiplied by tail length.

HR reporter assay

U2OS DR-GFP HR reporter cells^37^ were plated at a concentration of 120 000 cells/ml. Cells were first transfected with ASO or PO ASO 50 nM and 2 h later transfected with an I-SceI vector (Supplementary Table 1; 500 ng/ml) using 1.5 μl/ml Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer’s instructions. In the case of siRNAs, cells were plated at a concentration of 100 000 cells/ml and were first transfected with siRNA (siControl or siRAD51) and 6 h later transfected with the I-SceI vector as described above. 48 h after the initial transfection (of ASO, siRNA) cells were harvested using Trypsin (Thermo Fisher Scientific), followed by centrifugation at 272 x g (1700 rpm) for 10 min at 4 °C, one wash with PBS, and fixation in 1% PFA (added slowly while vortexing) for 20 min at room temperature, protected from light. Cells were centrifuged at 857 x g (3000 rpm) for 5 min at 4 °C and then resuspended in PBS. The GFP signal, resulting from HR repair, was measured by flow cytometry on a ID7000™ Spectral Cell Analyzer (Sony Biotechnology), with quantification analyzed using ID7000™ software.

Viability assay

The CellTiterGlo^®^ 2.0 Cell Viability assay (Promega) was performed in 96-well plate format according to the manufacturer’s instructions. Cells were transfected 72 h before analysis and, when indicated, irradiated 6 h after transfection with recovery for 66 h at 37 °C. Luminescence was measured in a FLUOstar Omega Microplate Reader (BMG Labtech).

Animal studies

All the animal experiments were performed according to ethical permit number 13849-2020 (Samir EL Andaloussi), approved by The Swedish Board of Agriculture (Jordbruksverket).

ASO treatment: For intracerebroventricular (ICV) delivery of Cy3-labeled ASO#7 into mouse brain, NMRI mice were anesthetized with isoflurane and placed on a stereotaxic frame with a prewarmed heating pad to maintain the body temperature at 37 °C. The eyes of mice were protected with eye ointment and mice were injected subcutaneously with pain relief (buprenorphine, 0,1 mg/kg) prior the procedure. Next, the head was shaved and vertical incision made to expose bregma to determine the coordinates for skull drilling and subsequent injection (anteroposterior 0.3 mm, mediolateral 1 mm, dorsoventral 3 mm). Mice were injected with 5 µL of 10 μg Cy3-labeled ASO#7 at the injection rate 1 μL/min and the Neuros syringe was left inserted for another 2 min before slow withdrawal. The bone was closed with the bone wax and the incision sutured. Mice were sacrificed 48 h post-injection and the whole brain was fixed in 4% PFA overnight before subjected to paraffin embedding and sectioning for further immunofluorescence analysis.

ASO and DNA damage signaling assessment: Slides were incubated 1 h at 65 °C before being immersed in: 1) Xylene (Sigma-Aldrich) 20 min, 2) Xylene 10 min, 3) 100% ethanol 5 min, 4) 100% ethanol 3 min, 5) 95% ethanol 2 min, 6) 80% ethanol 2 min, 7) 70% ethanol 2 min, 8) deionized water rinse, 9) PBS 5 min, 10) PBS 5 min, 11) PBS 5 min. Thereafter the slides were boiled in IHC Select Citrate Buffer pH 6.0 (Sigma-Aldrich) for 15 min and were left to cool to room temperature. Slides were washed twice in PBS for 5 min and incubated with SuperBlock (TBS) Blocking Buffer (Thermo Fisher Scientific) for 1 h followed by incubation with primary antibodies (Supplementary Table 3), diluted 1:25 in the same blocking buffer, overnight at 4 °C. Slides were washed three times in PBS for 5 min and incubated with secondary antibodies (Supplementary Table 3), diluted 1:300 in 0.1% Triton X-100 in PBS, for 1 h at room temperature, followed by three washes with PBS for 5 min. Slides were incubated with 1 μg/ml DAPI for 20 min at room temperature, were rinsed twice in PBS for 5 min, and mounted using ProLong Antifade Mountant (Thermo fisher Scientific). Images were acquired with a LSM700 confocal microscope (Zeiss), mounted on Axio observer.Z1 (Zeiss) equipped with a Plan-Apochromat 63x/1.4 oil immersion lens, and then processed with the Zen 2012 Black software (Zeiss). Images were analyzed in CellProfiler (Version 4.2.6)^53^ and ImageJ (Version 1.53t)^54^.

RNA sequencing

U2OS cells were plated at a concentration of 40 000 cells/ml, and was 24 h later transfected with indicated ASOs for 24 h. Total RNA was extracted from samples using the SimlyRNA Cells kit and Maxwell automated system (Promega). RNA-sequencing libraries were prepared using Smart-seq3^57^ and sequenced on an Illumina Nextseq X to generate 150 bp paired-end reads. Reads were aligned and count tables prepared using zUMIs^57^ and normalized counts were calculated using DESeq2^58^.

RNA sequencing data processing

For tSNE-NN, the R package Rtsne 0.17 was used to generate weighted PCA scores, which were projected into five dimensions. Euclidean distances were then calculated between each sample in the data set and edges were formed between each sample and its ten nearest neighbours. The weighted adjacency matrix was then visualized using the R package igraph 2.0.3 to generate a force directed map of the relationships between all samples in the data set^59^. The cor and hclust functions in R were used together with gplot heatmap.2 to generate the heatmap of all sample-sample correlations and the hierarchical clustering tree. Differential expression was performed using Deseq2 1.38.2, with all up or downregulated genes differentially expressed padj <0.01 used for subsequent analysis. Gene ontology was analyzed using panther.org complete biological processes statistical overrepresentation test ontology database release 2024-08-07. A control group of all up- and downregulated genes graphed together was calculated such that fold enrichments displayed for each individual group were divided by the enrichment in the control group. Overlap enrichment was performed as: (# samples in cluster X)/(# samples in sample group)*(# total samples in cluster X)^60^.

Statistics and reproducibility

Unless otherwise indicated, all values presented are the means ± standard deviations (SD) and n = 3 refer to three biologically independent experiments. The Prism GraphPad software (Version 10.5.0) was used for all analyses. p < 0.05 was considered significant and the p-values are provided as follows: ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05, ns = not significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Video 1 Reporting Summary Transparent Peer Review file

Source data

Source Data