TEX264 drives selective autophagy of DNA lesions to promote DNA repair and cell survival

2-fold

In brief TEX264, a conserved and cancer-relevant autophagy receptor, orchestrates selective autophagy of DNA lesions induced by topoisomerase 1 cleavage complexes to ensure genome stability and cell survival.

INTRODUCTION
2][3] Topoisomerase 1 (TOP1) is crucial for resolving torsional DNA stress. 4,5TOP1 cleaves one strand of supercoiled DNA structures ahead of DNA replication or transcription to release torsional stress on the DNA, facilitating uninterrupted DNA synthesis and transcription.
As part of its enzymatic cycle, TOP1 forms a transient phosphotyrosine covalent link between the 3 0 -broken DNA ends and a tyrosine in its active DNA-binding domain. 6This linkage represents a typical 3'-DNA-protein crosslink, a deleterious DNA lesion commonly known as a TOP1 cleavage complex (TOP1cc). 7ypically, TOP1cc is transient and reversible.Various types of endogenous DNA damage, such as abasic sites, oxidized nucleotides, ribonucleotides, or alkylated bases, prevent the release of TOP1. 6,8Stabilized TOP1cc hinders the progression of replication forks, making this DNA damage a primary endogenous source of genomic instability and cell lethality. 91][12] Irinotecan, topotecan, belotecan, and trastuzumab deruxtecan are CPT analogs approved for treating colorectal, small-cell lung, ovarian, and breast cancer.However, the development of resistance to TOP1 poisons, particularly in colorectal cancer patients, poses a significant clinical challenge. 13,14Thus, understanding the molecular mechanisms of TOP1cc repair is essential for understanding genome stability and critical for improving clinical outcomes and stratification for patients treated with TOP1 poisons.
TOP1cc repair involves restoration of DNA integrity and clearance of damaged TOP1.DNA repair mainly relies on either tyrosyl DNA phosphodiesterase 1 (TDP1) excision or nucleases such as MRE11. 1,46][17][18] The remaining 3 0 phosphoryl-nicked DNA is repaired by base excision repair (BER). 19lternatively, nucleases such as MRE11 cleave and resect DNA in a 3 0 to 5 0 direction on one strand. 18,20fficient TOP1cc repair tightly correlates with the upstream degradation of the proteinaceous part of TOP1cc, making the enzyme-DNA bond accessible for repair machinery.2][23][24][25] However, this process is predominantly described with doses above a certain threshold, corresponding to micromolar (mM) CPT doses.Because these are unachievable in clinical settings, [26][27][28] the physiological relevance of this model is called into question.Recently, the debulking of TOP1cc at clinically relevant doses of CPT (low-dose CPT, nanomolar [nM]) has been shown to depend on TOP1 unfolding by the ATPase p97, and its cofactor TEX264. 291][32] Strikingly, TEX264 has also recently been identified as a main receptor for autophagy. 33,346][37][38] Lysosomes are highly acidic vesicles, rich in proteases, hydrolases, nucleases, and lipases, ensuring the fragmentation of their contents into single amino acids, nucleotides, carbohydrates, and lipid structures. 39,40elective removal of organelles or substrates by autophagy is achieved through the action of receptors, commonly characterized by an ATG8/LC3 (LC3)-interacting region (LIR). 41,42LC3 links the receptors to the growing autophagic membranes to sequester substrates.The E1-like activating enzyme ATG7 governs selective autophagy by mediating the lipidation of LC3, which is essential for anchoring to the autophagic membranes. 43Under starvation, TEX264 is responsible for 50% of autophagy-dependent endoplasmic reticulum turnover (ER-phagy). 33,34Specifically, TEX264 is a single transmembrane protein embedded in the ER and nuclear envelope. 29,33,34It has a LIR domain and mediates the remodeling and processing of the ER through autophagy in response to starvation. 44,45oss of autophagy impairs DNA damage repair and increases cell death in response to genotoxic stress, [46][47][48][49][50] highlighting its importance in maintaining genomic stability.It has been shown that autophagy maintains cellular energy levels, provides nucleotides, and controls the level of DNA repair proteins required for accurate DNA replication and repair. 51,527][58][59] In response to oncogenic stress, Lamin B1 binds to LC3 and is delivered from the nucleus to lysosomes.The process of lysosomal degradation of nuclear Lamin B1 induces cell senescence. 56,57Further, the survival of mammalian cells suffering from laminopathies depends on the clearance of aberrant cytosolic DNA by autophagy. 60Moreover, defects in the lysosomal nucleases lead to improper clearance of nuclear DNA found in the cytoplasm and, thus, to numerous human diseases. 61herefore, mammalian cells rely on autophagy to clear cytosolic DNA fragments under various chronic stress conditions.However, whether autophagy directly processes nuclear DNA lesions and how this process is orchestrated is unknown.So far, no evidence exists that selective autophagy directly processes DNA lesions to regulate DNA repair for genome stability and cell survival.
Given that TEX264 is involved in two separate cellular processes, DNA damage repair 29 and ER-phagy, 33,34 in the nucleus and in the ER, respectively, 62 we investigated whether TEX264 orchestrates TOP1cc repair via autophagy.We employed biochemical and cell biological methods coupled with mass spectrometry, next-generation DNA sequencing (NGS), live imaging, and electron microscopy to investigate the role of autophagy in TOP1cc repair in human cells.We used zebrafish and colorectal cancer patient data to examine the relevance of selective autophagy of DNA lesions at the organismal level and in response to chemotherapy.Our data demonstrate TEX264orchestrated selective autophagy of nuclear TOP1cc DNA lesions in vertebrates.

Crosstalk between autophagy and DNA replication in response to CPT treatment
Based on the recently discovered separate roles of TEX264 in ER-phagy and TOP1cc repair, 29,33,34 we hypothesized that TEX264 might be an autophagy receptor that directly processes DNA lesions induced by stabilization of TOP1 covalently bound to nuclear DNA.Immunoprecipitation of TEX264-WT-V5 demonstrated that TEX264 forms a complex with DNA replication, such as the proliferating cell nuclear antigen (PCNA), and autophagy proteins in unsynchronized and S-phasesynchronized cells (Figure 1A).Similar results were obtained by co-immunoprecipitating endogenous TEX264.Interestingly, ATG7 regulated the interaction of endogenous TEX264 with TOP1, the autophagy protein LC3, and the DNA replication component MCM7 (Figure S1A).These results confirmed that TEX264 is indeed involved in autophagy and DNA replication.
4][65][66] The data were collected from 5 independent studies and analyzed for 640 characterized autophagy factors (Figures 1B and S1B; Table S1).Interestingly, autophagy factors represented around 5% of all identified proteins on or around DNA replication sites.Proteins involved in every stage of the autophagy pathway were found at/around sites of DNA replication forks (Figure 1C).Upon CPT treatment, key autophagy proteins were significantly enriched, including factors regulating the initiation of autophagy, such as USP13, a regulator of Beclin-1; receptors, including TEX264; and proteins involved in vesicle maturation, such as CHMP7 and SNAP29.Also, several resident proteins of lysosomes were enriched at the replication fork, including 10 V-ATPases, 8 chaperones, the cathepsin D protease, and RAB39A (Table S1).For validation, we performed isolation of proteins on nascent DNA (iPOND) at both low-dose and high-dose CPT (Figure 1D).Autophagy proteins, such as Beclin-1, TEX264, CHMP7, ATG7, SNAP29, and LC3, were enriched at/around DNA replication forks after CPT treatment.Clinically relevant doses of CPT (low-dose CPT, nM) [26][27][28] induce a higher level of recruitment to stalled replication forks than highdose CPT (mM).We concluded that autophagy proteins are actively recruited at/around sites of DNA replication forks and enriched in response to TOP1cc stabilization.
To further support our observations, we created stable HeLa, HCT116, and hTERT RPE-1 cell lines expressing the tagged lysosomal resident protein TMEM192-hemagglutinin (HA) to isolate intact lysosomes by immunoprecipitation (IP), a method known as LysoIP (Figure 1E). 67The isolated lysosomes were clean from contamination by other organelles, and immunofluorescence staining showed proper colocalization of the tagged lysosomal resident protein TMEM192-HA and late endosome marker LAMP1 68 (Figures S1C and S1D).The content of the intact lysosomes was isolated by LysoIP in HeLa and analyzed by quantitative label-free mass spectrometry.Consistent with previous results, 67 resident proteins of lysosomes, such as numerous cathepsins, V-ATPases, and hydrolyzes, were isolated.Interestingly, CPT treatment induced the lysosomal uptake of many DNA replication and nuclear proteins, such as DNA polymerases, MCM complex proteins, and nuclear pore components (Figure 1F; Table S2).Validation by immunoblotting in S-phase-synchronized cells confirmed the lysosomal uptake of core DNA replication proteins in response to CPT treatment (Figure 1G).TOP1 was also a substrate of lysosomes in HCT116 and hTERT RPE-1 upon low-dose CPT treatment (Figures S1E and  S1F).Other TOP1 inhibitors, such as irinotecan and its active metabolite SN-38, induced TOP1 lysosomal uptake (Figure S1G).Altogether, we conclude that crosstalk between autophagy and DNA replication is enhanced as a response to the stabilization of cytotoxic TOP1cc.

Autophagy repairs TOP1cc and promotes cell survival to camptothecin
To assess whether autophagy has a functional role in the processing of TOP1cc, we analyzed TOP1cc level recovery after CPT treatment with a specific TOP1cc antibody (Figures 2A, S2A, and S2B). 69,70Repair of TOP1cc foci was significantly faster in cells treated with the mTOR inhibitor Torin, a well-known booster of autophagy. 71In contrast, the inactivation of autophagy by inducible depletion of ATG7, an essential protein for selective autophagy, 43 completely blocked the repair of TOP1cc foci.Autophagy impairment by ATG7 inactivation showed a similar effect on TOP1cc levels when assessed by RADAR assay, a specialized method to isolate proteins covalently bound to DNA 70,72 (Figures 2B and S2C).Interestingly, ATG7 inactivation strongly prevented TOP1 uptake in lysosomes by LysoIP (Figure 2C), suggesting that TOP1cc repair is mediated by selective autophagy.
To further confirm the role of autophagy in TOP1cc repair, we inactivated two other selective autophagy factors in HeLa cells: syntaxin 17 (STX17) and RB1CC1 (also known as FIP200).Syntaxin 17 is a SNARE protein involved in the fusion of mature autophagosomes to lysosomes, 73 and RB1CC1 is a member of the ULK complex and the master regulator of autophagy initiation. 74s expected, depletion of syntaxin 17 caused the accumulation of autophagosome double-membrane vesicles (Figure S2D).RB1CC1 and syntaxin 17 depletion abolished TOP1 uptake in lysosomes (Figure S2E).Syntaxin 17 inactivation completely prevented TOP1cc repair after CPT treatment (Figure S2F).Strikingly, both ATG7 and syntaxin 17 knockdown strongly impaired cell survival in response to CPT treatment (Figure 2D).Correspondingly, inhibition of lysosomal acidification and function by bafilomycin A1 (BAF) caused increased sensitivity to CPT.However, boosting autophagy by Torin enhanced HeLa cell survival to CPT treatment (Figure S2G).These data suggest that autophagy positively regulates TOP1cc repair and cellular survival to CPT.
To further demonstrate that TOP1 translocates to lysosomes as a response to CPT treatment, we modified a well-established mCherry-GFP autophagy reporter assay to follow TOP1. 75The mCherry-TOP1-GFP reporter can be detected in red and green channels, except when transported inside the highly acidic lysosomes, where GFP fluorescence is quenched while mCherry remains unaffected (Figure 2E).As expected, untreated cells  S1 and S2.(legend continued on next page) exhibit a robust nuclear signal for both colors and a limited cytoplasmic signal, consistent with the expression and localization of TOP1 in the nucleus (Figure 2F).Notably, the tagged TOP1 enzyme conserved its original DNA-binding activity, as CPT can stabilize mCherry-TOP1-GFP on DNA as it would for the endogenous TOP1 enzyme (Figure S2H).Upon treatment with low-dose CPT, but not high-dose CPT, we observed cytoplasmic red-only puncta, quantified by a decrease in the cytoplasmic GFP/mCherry colocalization ratio (Figure 2F).Inhibiting lysosomal acidification with bafilomycin A1 prevented GFP quenching and restored the GFP/mCherry colocalization ratio, demonstrating the specificity of this assay toward detecting proteins inside lysosomes.Similarly, inactivating either ATG7 or syntaxin 17 abolished the mCherry-TOP1-GFP translocation inside lysosomes in response to CPT treatment (Figure S2I), further suggesting that the autophagy pathway regulates TOP1cc repair and TOP1cc uptake into lysosomes.
Autophagy processes TOP1 during DNA replication stress TOP1 processing by lysosomes is intriguing, as TOP1cc degradation by the proteasome has been extensively described.2][23][24][25] Notably, treatment with high-dose CPT did not induce translocation of the mCherry-TOP1-GFP reporter to lysosomes (Figure 2F), nor did we observe increased lysosomal uptake of endogenous TOP1 assessed by LysoIP (Figure 3A).This shows that TOP1 is a substrate of lysosomes in response to clinically relevant doses of CPT only (low-dose CPT, nM).
To investigate the co-dependency between autophagy and the proteasome in TOP1cc repair, we used low-dose (nM) or high-dose (mM) CPT and monitored cell response.First, we tested the lysosomal uptake of TOP1.As expected, low-dose CPT stimulated TOP1 delivery to lysosomes (Figure 3A).However, this stimulation was not observed with high-dose CPT, even when the proteasome was blocked with bortezomib.
Second, we monitored cell survival in response to low-and high-dose CPT in cells inactivated for TEX264 or RNF4, independently or simultaneously.TEX264 inactivation is a specific way to prevent TOP1 processing by autophagy (see Figure 6).As RNF4 was reported to be essential for TOP1cc proteasomal degradation at high-dose CPT, 25,76,77 we used RNF4 inactivation as a specific way to inhibit TOP1cc processing by the proteasome.Also, we monitored TOP1 lysosomal uptake in response to low-dose CPT, and RNF4 inactivation did not severely impact it (Figure S6E).TEX264 inactivation specifically hypersensitized cells to low-dose CPT, whereas only RNF4inactivation hypersensitized to high-dose CPT (Figures 3B  and S3A).The co-inactivation of both factors did not further increase sensitivity, confirming the independence of the two degradation pathways.
Finally, we monitored the degradation of TOP1 in total cell extract (Figure S3B).Cells were treated with bafilomycin A1 or MG132 to inhibit lysosome function or the proteasome, respectively.Both low-and high-dose CPT induced TOP1 degradation.Still, only autophagy inhibition could rescue TOP1 levels in total cell extract upon low-dose CPT treatment.In contrast, only proteasome inhibition rescued the TOP1 level after high-dose CPT treatment.In conclusion, although autophagy and the proteasome can degrade TOP1cc, there is no co-dependency between the pathways.
Previous research analysis showed that mM and nM doses of CPT induce different types of secondary DNA damage. 28Lowdose CPT induces DNA replication stress, as evidenced by the slow progression of DNA replication forks and activation of g-H2AX, which do not localize with 53BP1 foci, a marker of DNA double-strand breaks (DSBs). 78In contrast, high-dose CPT leads to a nearly complete arrest of DNA replication forks and the formation of DSBs.Monitoring the effect of low-and highdose CPT on DNA damage by immunofluorescence, we observed that low-dose CPT only induced activation of g-H2AX but not 53BP1 foci, in contrast to high-dose CPT that induced focal activation of both g-H2AX and 53BP1 (Figure S3C).Based on these results, we hypothesized that DNA replication stress signaling could trigger TOP1cc processing by autophagy.Indeed, inhibition of ATR, the central kinase involved in the signaling of replication stress, 79,80 strongly abolished the delivery of TOP1cc to lysosomes (Figure 3C).In contrast, inhibition of ATM, the central kinase involved in DSB repair, 79,80 did not impact TOP1cc delivery to lysosomes.Moreover, ATR inhibition prevents the increased binding of p97 to TOP1 in response to low-dose CPT (Figure 3D), which happens through the p97 cofactor TEX264. 29e concluded that DNA replication stress, induced by lowdose CPT, drives TOP1 processing by autophagy.This pathway is uncoupled from the known RNF4-dependent proteasomal TOP1cc degradation pathway in response to DSB formation induced by high-dose CPT.
Low-dose CPT induces the formation of protein aggregates Selective autophagy was first described as a pathway for clearing protein aggregates in cells. 81To investigate the role of lysosomal degradation of TOP1 under replication stress, we monitored the formation of protein aggregates after low-dose CPT.In brief, the Proteostat dye becomes fluorescent upon binding to the amyloid-type b-sheet tertiary structure of protein aggregates. 82Inhibition of autophagy combined with low-dose CPT treatment elicited a drastic accumulation of protein aggregates, mainly colocalizing with lysosomes and late endosome marker LAMP1 (Figure 3E). 68Biochemical purification of protein aggregates using detergents confirmed the formation of TOP1 aggregates at low-dose CPT when lysosome function was (C) LysoIP in S-phase HeLa cells after 3 h of 50 nM CPT and 1 mM ATR inhibitor VE-822 (ATRi) or 1 mM ATM inhibitor KU-55933 (ATMi) (n = 3).(D) Co-immunoprecipitation of p97-Strep in S-phase HEK293T cells after 1 h of 50 nM CPT (nM) or 1 mM CPT (mM) or 1 mM ATR inhibitor VE-822 (ATRi) (n = 2).(E) Immunofluorescence of protein aggregates using the Proteostat dye after 24 h of 50 nM CPT and quantification (n = 3).Scale bar, 10 mm.Error bar, SD.Two-way ANOVA.**p < 0.005; ***p < 0.0005; ns, not significant.See also Figure S3.(legend continued on next page) inhibited by bafilomycin A1 (Figure S3D).In contrast, high-dose CPT did not induce the formation of TOP1 aggregates.This observation further supports our earlier finding that TOP1 degradation in response to high-dose CPT, where DSBs are formed, is independent of lysosomes.These results show that TOP1cc repair at low-dose CPT depends on autophagy to prevent proteotoxic stress caused by aggregate formation.
Lysosomes directly process TOP1cc DNA lesions At a nM dose of CPT, full-length TOP1 becomes a direct substrate for autophagy, as shown by the mCherry-TOP1-GFP reporter assay (Figure 2F) and LysoIP assays (Figures 1G, 2C, 3A,  and 3C).Focusing on low-dose CPT, we hypothesized that lysosomes process the entire DNA lesion, including TOP1 and its associated DNA fragment.If so, TOP1cc should undergo cleavage by DNA nucleases to allow its export to lysosomes.
To directly assess whether lysosomes process DNA fragments associated with TOP1cc lesions, we quantified and characterized DNA purified from intact lysosomes.DNA fragments of less than 500 bp were observed in lysosomes after combined treatment with CPT and bafilomycin A1 (Figures 4C, S4A, and  S4B).This suggests that the complete DNA lesion, TOP1 and its associated DNA fragment, is transported to lysosomes for degradation.
NGS showed that over 98% of DNA fragments isolated from intact lysosomes were of nuclear origin (Figure 4D).TOP1 does not have a consensus binding region on DNA and is found at supercoiled regions ahead of active transcription sites, such as introns 87,88 or replicative centromeric regions. 28,89Strikingly, the sequence analysis of isolated lysosomal DNA fragments represented mainly intronic and centromeric regions (Figure 4E).To investigate whether the DNA fragments isolated from lysosomes are likely to be covalently linked to TOP1, we overlapped three published TOP1 chromatin immunoprecipitation sequencing (ChIP-seq) results 87,88 with our LysoIP-seq data.The sequenced DNA fragments isolated from lysosomes overlapped 88.6% with the three known TOP1/TOP1cc ChIP-seq datasets (Figures 4F, 4G, and S4C).The overlap between the three independent TOP1 ChIP-seq datasets was similar (around 85%).These data prove that nuclear DNA regions covalently bound to TOP1 are translocated to lysosomes in response to low-dose CPT.
Immunofluorescence using the specific monoclonal antibody against TOP1cc, but not TOP1 alone, 69,70 confirmed the presence of TOP1cc in the cytoplasm in response to low-dose CPT but not to high-dose CPT (Figures 4H and S4D).Altogether, we conclude that lysosomes actively process entire TOP1cc lesions, including their proteinaceous part and the associated DNA fragments.

Lysosomal degradation of TOP1cc occurs in interphase cells
As low-dose CPT induces the direct recruitment of autophagy proteins at sites of stressed DNA replication forks (Figures 1B-1D) and the lysosomal uptake of replication proteins (Figures 1F  and 1G), we hypothesized that TOP1cc repair by autophagy occurs during S-phase.To test this hypothesis, cells were synchronized in the G1/S-phase using a thymidine block released and treated with low-dose CPT for 6 h to stabilize TOP1cc during S-phase.Indeed, TOP1 uptake in lysosomes was increased in S-phase (Figure 5A).
To investigate whether TOP1 can exit the nucleus during interphase, we followed the mCherry-TOP1-GFP reporter in live cells stained with LysoView, a pH-sensitive dye specific for lysosomes (Figure 5B; Videos S1 and S2).Within 30 min of CPT treatment, a mCherry/GFP-positive protrusion emerged from the nucleus, eventually forming a new mCherry/GFP vesicle fused with lysosomes and turned mostly red due to GFP quenching.Finally, the red-only vesicle disappeared as the digestion process was completed.This live imaging, together with the quantification of the red-only cytoplasmic puncta in fixed cells (Figures 2F  and S2I), demonstrated that TOP1cc degradation through autophagy occurs during interphase when the nuclear envelope is intact.
To investigate the state of the nuclear envelope during interphase in response to CPT, we monitored the lamina structure following the GFP-tagged Lamin A/C by live-cell imaging (Figure 5C; Video S3).Nuclear envelope rupturing during interphase (NERDI) is a transient loss of permeability in the barrier, documented in several cancer cell lines under severe stress conditions. 90,91NERDI is associated with the tendency to form nuclear envelope protrusions that can lack a lamina structure. 92,93trikingly, the lamina undergoes transient and localized reshaping within 30 min of CPT exposure.Cell rendering reveals a thinner lamina structure, primarily observed at contact sites with lysosomes.This suggests that transient and restricted rupture of the nuclear envelope occurs in response to low-dose CPT.
We used transmission volume electron microscopy to visualize the nuclear envelope after treatment with CPT (Figures 5D and  S5A; Video S4).The distance between the inner nuclear membrane (INM) and the outer nuclear membrane (ONM) under physiological conditions is about 50 nm, 94 as reproduced in our control (D) Distribution of the LysoIP-purified DNA, mapped reads sequenced by NGS.(E) Genomic distribution of the LysoIP-purified mapped reads sequenced by NGS.(F) Proportional Eulerr representation of the overlap among the different ChIP-seq of TOP1cc, TOP1cc-seq 1 (MCF7-GEO: GSE135808), 87 TOP1cc-seq 2 (LNCAP-GEO: GSE135808), 87 and TOP1-seq (HCT116-GEO: GSE57628). 88All conditions were treated with CPT.(G) Genomic browser alignment on genome hg38 of the different TOP1cc ChIP-seq and LysoIP-seq at RAD23A gene loci as sequence tags per million (TPM).(H) Immunofluorescence after 4 h of 50 nM CPT (nM) or 1 mM CPT (mM) and quantification (n = 3).Scale bar, 10 mm.Two-way ANOVA.*p < 0.05; ns, not significant.See also Figure S4.(legend continued on next page) cells.However, treatment with CPT revealed abnormal enlargement of this space to more than 150 nm, forming blister-like structures protruding toward the cytoplasm.To exclude any technical artifacts due to the fixation process on the membrane structure, we confirmed the formation of the blister-like structure in response to CPT by using high-pressure freezing (Figure S5B).Cryo-focused ion beam (FIB) milling and cryo-fluorescence correlation images with the LysoTracker dye showed the proximity of lysosomes to the blister-like structures (Figures 5E and S5D).
Interestingly, similar blister-like structures were displayed in the nuclear envelope under ER stress. 95These alterations are resolved by asymmetric autophagic digestion of the ONM.This was consistent with the enrichment of nuclear envelope proteins in intact lysosomes as identified by LysoIP mass spectrometry upon CPT treatment (Figures 1F and 1G).Moreover, inhibition of the major nuclear pore transport receptor, exportin-1, by leptomycin B, which is involved in the export of nucleic acids, 58,96,97 did not impact TOP1cc cytoplasmic localization in response to CPT (Figure S5C).
In summary, the nuclear protrusions observed by live imaging and electron microscopy, as well as the alterations observed in the lamina structure, suggest that TOP1cc processing by lysosomes occurs directly through the nuclear envelope.

TEX264 orchestrates TOP1cc repair by autophagy
To ensure cell survival and allow proliferation, the lysosomal degradation of genetic material and alterations of the nuclear envelope must be tightly regulated.So far, our data show that TOP1cc processing by lysosomes depends on selective autophagy mediated by ATG7, syntaxin 17, and RB1CC1 (Figures 2  and S2).Selective autophagy is tightly regulated and orchestrated by specific autophagy receptors that bind to cargo proteins.Among the 42 known autophagy receptors, 41,98,99 6 were enriched in intact lysosomes in response to CPT treatment (Figure 6A; Table S3).Five of these six identified autophagy receptors in lysosomes were also present at the DNA replication fork, identified by NCC-SILAC proteomics 63 including TEX264.Given that TEX264 was recently recognized as essential for TOP1cc repair 29 and cell survival to low-dose CPT (Figure 3B), we asked whether TEX264 is the autophagy receptor that mediates TOP1cc repair by autophagy.
TEX264-knockout HeLa cells had defective TOP1cc repair and were hypersensitive to CPT compared with wild-type (WT) cells (Figures 3B, S6A, and S6B).Complementation of TEX264-knockout cells with TEX264-V5-WT fully restored cell survival and TOP1cc repair (Figures S6A and S6B).Structural analysis and previous interaction studies revealed defined domains in TEX264, such as TOP1-binding site (TBS), LIR, p97binding motif (SHP), SUMO-interacting motif (SIM), and a phosphorylation site 29,33,34,100 (Figure 6B).To directly investigate the role of TEX264 in the autophagic degradation of TOP1cc, we created stable TEX264-V5 HeLa cell lines expressing either TEX264-WT or TEX264 variants in a TEX264-knockout background.Engineered cells expressing TEX264 variants were validated and displayed mild expression and proper cellular localization at the nuclear periphery (Figures 6C, S6C,  and S6D).
Inactivation of TEX264 binding to TOP1, LC3, or p97 by mutations in its TBS, LIR, or SHP domains, respectively, abolished the TOP1cc repair compared with the TEX264-WT expression (Figure 6C).Furthermore, cells expressing TEX264 variants, which prevented their binding to TOP1, LC3, p97, or SUMO1, were defective in delivering TOP1 to lysosomes (Figure 6D).However, a mutation in the phosphorylation sites of TEX264, close to its LIR domain, had no impact on the delivery of TOP1 to lysosomes.As phosphorylation on two serines juxtaposed to the LIR region of TEX264 tunes its activity as a receptor for ER-phagy under starvation, 100 we concluded that these phosphorylation sites regulate its role in ER-phagy but not autophagy of TOP1cc.
In contrast, a mutation in the SIM domain of TEX264 strongly reduced TOP1 uptake in lysosomes (Figure 6D).Consistently, TOP1cc delivery to lysosomes was entirely blocked by using a specific E1-SUMO-activating enzyme inhibitor (ML792, Figure S6E).These data suggest that TOP1cc processing by lysosomes is dependent on SUMO.Because the E3-SUMO ligase PIAS4 and the SUMO-targeted ubiquitin ligase RNF4 are essential to direct TOP1cc for proteasomal degradation, 101 we investigated their role in regulating TOP1 delivery to lysosomes.Interestingly, neither PIAS4 nor RNF4 depletion prevented TOP1 uptake to the level of the specific E1-SUMO-activating enzyme inhibitor (Figure S6E).This implies that SUMOylation regulates TOP1cc lysosomal processing but in a PIAS4-independent manner.Lastly, TEX264 variants defective for binding to TOP1, LC3, p97, and SUMO were hypersensitive to CPT treatment compared with TEX264-WT cells (Figure 6E).
ATG7 regulates TEX264 binding to the replisome and autophagy machinery, MCM7 and LC3, respectively (Figure S1A).Therefore, we investigated where the TEX264-autophagy cascade starts in the cell using proximity ligation assay (PLA) between TEX264-V5 and LC3-GFP (Figure 6F).As established before, TEX264-LC3 interaction exists in the cytoplasm. 33,34owever, PLA demonstrated that TEX264 also interacts with LC3 inside the nucleus, near the nuclear periphery.Importantly, TEX264/LC3 interaction was abolished when the TEX264-LIR domain was mutated.Based on TEX264-LC3 interaction in the nucleus, the dependency on the LIR motif to deliver TOP1 to lysosomes (Figure 6D), and the recruitment of these two proteins at sites of DNA replication forks in response to CPT treatment (Figures 1C and 1D), we concluded that TEX264 is the receptor for TOP1cc degradation by autophagy within the nucleus at sites of stalled DNA replication forks.(legend continued on next page) The TEX264-p97 axis orchestrates TOP1cc delivery to lysosomes To investigate the role of p97 in TOP1cc processing by autophagy, we used the specific p97 inhibitor, CB-5083, which does not impact autophagy flux. 102,103Similarly to the TEX264-SHP variant, p97 inhibition abolished TOP1 transfer to lysosomes and TOP1cc repair (Figures S6F and S6G).SPRTN was recently identified as a partner of p97 and TEX264 in the repair of TOP1cc, 29 so we investigated the role of SPRTN in the autophagy of TOP1cc.Interestingly, SPRTN inactivation did not affect TOP1cc delivery to lysosomes (Figure S6H) but caused the formation of cytoplasmic aggregates when the lysosomal function was blocked (Figure S7A).We concluded that SPRTN processes endogenous TOP1cc during S-phase, 29,30 and its inactivation combined with the lysosomal inactivation causes increased aggregate formation.These results suggest that the p97-TEX264 complex orchestrates TOP1cc delivery to lysosomes independently of SPRTN protease function.

TEX264 role in TOP1cc repair is evolutionarily conserved and relevant for chemotherapy response in the clinic
To investigate the role of TEX264 at the organismal level, we used a zebrafish model.Specific gene silencing of tex264 in zebrafish embryos was achieved using morpholino oligonucleotides (tex264MO) and then complemented with either tex264-WT or LIR* sequence via co-injection into 1-to 4-cell-stage embryos.Despite robust tex264 silencing and proper tex264 re-expression, we did not observe any phenotypic changes in the zebrafish 2-day-old embryos (Figure 7A).Total genomic DNA was isolated from 2-day-old embryos, and TOP1cc level was analyzed by RADAR assay (Figure 7B).Inactivation of tex264 caused an almost 2-fold increase in endogenous TOP1cc compared with the WT control embryos.The expression of tex264-WT, but not the autophagy-pathway-defective variant tex264-LIR*, prevented TOP1cc accumulation in tex264-depleted background.These data suggest that TEX264 and its autophagy role in TOP1cc repair are evolutionarily conserved and relevant at the organismal level.
Finally, to explore the relevance of TEX264 in a clinical setting, we analyzed TEX264 expression levels in 361 colorectal cancer primary tissues from patients treated with different chemotherapy regimens.We observed a strong positive association between TEX264 expression and the progression-free survival (PFS) of colorectal cancer patients in response to TOP1 inhibitor irinotecan (FOLFIRI: folinic acid, 5-fluorouracil [5-FU], and irinotecan) but not to 5-FU and folinic acid alone.Patients whose primary colorectal cancers express high levels of TEX264 had a 50% increase in PFS (9.1 months vs. 6.1 months, p value = 0.003) when treated with FOLFIRI compared with 5-FU and folinic acid (Figure 7C).In sum, TEX264-orchestrated autophagic repair of TOP1cc is highly relevant for response to irinotecanbased chemotherapy in the clinic and might prevent the increased genomic instability in colorectal cancers by ensuring proper DNA damage repair.
Indeed, depletion of TEX264 correlates with higher genomic instability measured by TOP1cc accumulation in cells (Figure S6A), the formation of Bloom syndrome protein (BLM)-positive ultrafine chromosomal bridges (Figure S7B), and DNA breaks measured by alkaline comet assay (Figure S7C).Moreover, TEX264 inactivation triggered a different mutational burden after long-term low-dose CPT treatment than in HeLa WT parental cells (Figures S7D-S7F).Whole-genome sequencing showed a shift to thymine-to-adenine (T>A) mutational profile compared with WT, corresponding to SBS25 [104][105][106] and SBS34. 107Finally, TEX264 inactivation combined with lowdose CPT induced the high formation of g-H2AX/53BP1 foci, a recognized DSB marker (Figure S7G). 78This effect is similar to the inactivation of ATR combined with the same dose of CPT, suggesting that TEX264 protects DNA replication fork collapse in response to TOP1cc.Based on all this evidence, we concluded that TEX264 mediates autophagic degradation of TOP1cc to protect DNA replication forks and, consequently, genome stability.[110][111]

DISCUSSION
Our findings demonstrate the direct involvement of selective autophagy in the repair of a specific type of nuclear DNA lesion, TOP1cc, in vertebrates (Figure 7D).As selective autophagy processes the nuclear DNA lesion, we refer to this process as nucleophagy.The mechanism of nucleophagy described here processes both endogenous and chemotherapy-relevant TOP1cc lesions that are excised from DNA replication forks by the nuclease MRE11.TOP1 and its associated DNA fragment are extruded from the nucleus to lysosomes through transient and restricted alteration of the nuclear envelope.Interestingly, besides TOP1cc, we also observed that replisome components were present in lysosomes after CPT-induced DNA damage.This suggests that, very likely, the replisome stalled at TOP1cc is excised and removed by nucleophagy.We focused our research on the mechanistic details of TOP1cc nucleophagy.However, the reason for the replisome or its components to be transported to lysosomes is unknown, and it will be essential to address in the future.
By comparing low-and high-dose CPT, we were able to elucidate that nucleophagy of TOP1cc is linked to stalled DNA replication.This process is inactivated when DNA replication forks collapse and are converted to DSBs.2][23][24][25] We hypothesize that the choice of the TOP1cc degradation pathway in the cell depends on the topological accessibility of the lesion.DSBs create more flexibility around the DNA damage 112,113 and could potentially facilitate the recruitment of the proteasome and downstream DNA-damage-repair proteins compared with a tighter conformation when the DNA replication fork is only stalled.Interestingly, a similar threshold concept for activating selective autophagy was also observed in the degradation of damaged mitochondria (mitophagy). 114In response to mitochondrial stress, autophagy regulates the removal of defective mitochondria.However, chronic stress leads to a complete cessation of mitophagy to prevent the potential depletion of mitochondria within the cell.Specifically, we identified the autophagy factor TEX264 as a receptor for nucleophagy of TOP1cc.TEX264-mediated nucleophagy promotes DNA replication fork stability, genome stability, and cell survival response to DNA lesions induced by CPT.TEX264 bridges TOP1cc, stalled DNA replication and lysosomes, and orchestrates the export of DNA lesions to lysosomes in an ATR-but not ATM-dependent manner, further supporting the model that nucleophagy of TOP1cc is linked to DNA replication.
1]115 Inactivation of proteasome activity causes a pleiotropic effect on many cellular functions, including depletion of free nuclear ubiquitin. 116,117By inactivation of the E3-ubiquitin ligase RNF4, which promotes proteasomal degradation of TOP1cc, 25,76,77 we overcame this pleiotropic effect of proteasomal inhibition and demonstrated that human cells were not sensitive to RNF4 inactivation when exposed to low-dose CPT that induces DNA replication stress but not DSBs.Our experiments also showed that the proteasome does not affect total TOP1 degradation after low-dose CPT.We concluded that the proteasome does not have a significant role in degrading TOP1cc around sites of DNA replication.On the contrary, the SPRTN protease was shown to process TOP1cc at/around sites of DNA replication [29][30][31]115 and SPRTN was shown to form a physical complex with TEX264 and p97 to process replication-related TOP1cc lesions. 29 Hwever, the SPRTN protease cleaves specifically soluble proteins, including TOP1cc, during DNA replication fork progression and it works upstream of TDP1.This enzyme removes the remnant peptide from DNA after SPRTN proteolysis.[29][30][31]115,118 Neither SPRTN nor TDP1 inactivation prevented the delivery of TOP1 to lysosomes, but SPRTN inactivation increased the formation of aggregates.Based on these facts and previous knowledge of the role of SPRTN, we concluded that the p97-SPRTN complex removes soluble TOP1cc during DNA replication fork progression.
This work provides evidence that the p97-TEX264 pathway also operates during DNA replication and removes insoluble or aggregated TOP1cc by selective nucleophagy.This pathway depends on MRE11 nucleolytic cleavage of DNA lesions and is independent of SPRTN-TDP1 processing.Our conclusion is further supported by the fact that low-dose, but not high-dose, CPT induces the formation of aggregates.
4][55] However, no homolog for the yeast receptor atg39 that recognizes a substrate in the nucleus and delivers it to the autophagosome has been identified yet. 54,55In mammals, nucleophagic clearance of abnormal nuclear content is observed in cells with laminar defects.In the context of laminopathies, inhibiting autophagy reduces cell viability, suggesting a beneficial role of autophagy in clearing leaking nuclear content. 60In cancer cell lines and migrating pri-mary cells, nuclear content is frequently transiently exposed and degraded by autophagy.This temporary loss of barrier integrity is often due to mechanical stress or weaknesses in the nuclear envelope structure. 91However, how nucleophagy is orchestrated and how nucleophagy contributes to DNA repair and genome stability was not clear.
By discovering the TEX264-orchestrated nucleophagy pathway, we directly prove the existence of nucleophagy as a specialized DNA repair pathway in vertebrates.Whether this pathway also operates on other DNA-protein adducts remains to be further investigated.The increased T>A mutational profile in TEX264-inactivated cells treated with low-dose CPT led to the mutational signature resembling SBS25 [104][105][106] and SBS34, 107 both of unknown etiology.This finding reinforces our statement that the TEX264-orchestrated TOP1cc nucleophagy pathway is a DNA repair mechanism that causes specific single base substitution (SBS) signatures when inactivated and does not belong to the previously described DNA repair pathway.
In summary, we reported TEX264-orchestrated selective nucleophagy of TOP1cc lesions as an evolutionarily conserved DNA repair pathway in vertebrates that operates at/around sites of DNA replication.This clinically relevant pathway promotes 50% better PFS in colorectal patients treated with the TOP1 inhibitor irinotecan.We believe this work opens avenues of research in biology, cancer therapy, and aging.

Limitations of the study
We suggest that the nucleophagy of TOP1cc is a DNA repair process that occurs at/around sites of DNA replication located at the nuclear periphery.Even though we showed that autophagy and the proteasome play a role in TOP1cc repair based on DNA structures induced by low-dose (stalled DNA replication) and high-dose (DSB) of CPT, respectively, it is still unknown how the different TOP1cc repair pathways cooperate at low-dose CPT.Interestingly, high-dose CPT also recruits autophagy factors at/around sites of DNA replication but to a lesser extent than low-dose CPT.The crosstalk between stalled or broken replication forks (converted to DSBs) and lysosomes should be investigated further.Finally, how the transient alteration of the nuclear envelope is formed and allows the transport of DNA lesions to lysosomes has yet to be discovered.Further work is needed to address the questions above and potentially translate this concept to the clinic.

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled upon reasonable request by the lead contact, Kristijan Ramadan (kristijan.ramadan@ntu.edu.sg).

Materials availability
All unique/stable reagents generated in this study are available from the lead contact with a completed Materials Transfer Agreement.

Data and code availability
Both NGS datasets, LysoIP-Seq and WGS, are deposited to NCBI: GSE242298 and SRA: PRJNA1127335.The mass spectrometry proteomics data is deposited to the ProteomeXchange Consortium, PRIDE: PXD046037.Raw data are deposited on Mendeley Data: https://doi.org/10.17632/x6sv8zrjsv.1.The control arm of the FOCUS transcriptome is publicly available in GEO under accession number NCBI: GSE156915.The transcriptome of the irinotecan arm and additional S:CORT data is available to all academic researchers on submission of a data request to the data access committee.For commercial agencies, the data will be made available through Cancer Research Horizons on behalf of the funders and consortium members.This study did not generate a new unique code.Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Cloning procedure
Cloning of the mCherry-TOP1-GFP construct for the reporter assay was performed by inserting the TOP1 DNA sequence 29 into the backbone mCherry-GFP-pEGFP-N1 (Addgene).Due to the challenging cloning procedure to insert TOP1 into a lentivirus backbone, transient expression was performed with the above construct mCherry-TOP1-GFP-pEGFP-N1. It was expressed transiently for 24 hr before imaging.The backbone was linearised by KpnI and BamHI.Phusion Hot Start II High-Fidelity DNA Polymerase (Thermofisher) was used to amplify the TOP1 sequence with primers (KpnI_TOP1_F and BamHI_TOP1_R).Ligation was performed using NEBuilderâ HiFi DNA Assembly Cloning Kit (NEB) following the manufacturer's instructions.Cloning of the TEX264-V5-pLX313 plasmids was performed using TEX264 sequences generated previously 29,100 including WT or TEX264 E194A (TBS*), F273A (LIR*), G280R, G282R, L284A (SHP*), I141A, S143A, W145A (SIM*) or S271A, S272A (P*).The pLX313-TP53-WT plasmid (Addgene) was used as the backbone for all the TEX264-V5-pLX313 plasmids.The backbone was digested by NHeI-HF and EcoRV-HF to remove TP53.TEX264 sequences were amplified by the Phusion Hot Start II High-Fidelity DNA Polymerase (Thermofisher) with primers (TEX-pLX313_Fwd and TEX-pLX313_Rev).NEBuilderâ HiFi DNA Assembly Cloning Kit (NEB, E5520S) was used to ligate the TEX264 sequences into the pLX313 backbone according to the manufacturer's protocol.Sequencing of the correct insertion was performed using the EF1a_Fwd and WPRE_Rev primers (Invitrogen).All plasmids were sequenced by Source BioScience, Oxford, UK.
Zebrafish assays were performed using tex264 complementation.Full-length tex264-WT was amplified with primers (tex264_F and tex264_R) from the cDNA of 2 days post fertilisation embryos and cloned into a pCS2+HisMyc expression vector (Addgene) using XhoI and XbaI restriction sites with the In-Fusionâ Snap Assembly Master Mix (Takara Bio USA, Inc.).The tex264-pCS2+HisMyc plasmid was modified using the QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Agilent) with the primer (tex264_F283A LIR mutagenesis) to create tex264 F283A LIR* mutant.The plasmids were linearized using NotI and transcribed in vitro using the HiScribe SP6 RNA kit (NEB) together with the ARCA kit (NEB) to cap the resulting mRNA.The mRNA was purified with the Monarchâ RNA cleanup kit (NEB) for subsequent injections.

Transfection and generation of stable cell
Transient plasmid transfections were performed using polyethylenimine (PEI) for biochemistry application or FuGENEâ for microscopy usage, following the manufacturer's instructions.All siRNA transfections were carried out using LipofectamineÔ RNAiMAX, according to the manufacturer's protocol and assayed after 72 h.
Lentiviruses were produced by transfecting HEK-293T with the transfer plasmid containing the insert of interest flanked by the viral LTR and combined with the CMV-pAmphoR envelope and CMV-D8.2Rpackaging plasmids.Transfection was performed using PEI (ratio DNA:PEI; 3 mg:1 ml).Fifty hours after transfection, the virus-containing supernatant was collected and centrifuged at 1,000 x g to remove cells and filtered with 0.45 mm PVDF filters.A stable cell line was produced by seeding 500,000 cells in 6-well plates in 250 ml of DMEM with 10% FBS, 8 mg/mL polybrene and 750 ml of virus-containing media.Sixteen hours later, selection with antibiotics was started.After clonal expansion, clones obtained were tested for correct insertion by immunoblotting and immunofluorescence.The pLJC5-TMEM192-3xHA construct was used to produce cell lines stably expressing TMEM192-3xHA in HeLa, HCT116 and hTERT RPE-1.The pLJC6-3XHA-TMEM192 construct was used for LysoIP in HeLa TEX264 KO as the cells were already puromycin resistant.
To induce the depletion of ATG7, 1 mg/ml of doxycycline was added to the cell 4 days and again 2 days before the experiment.To induce the expression of p97-Myc/Strep, 1 mg/ml of doxycycline was added to the cell 24 hours prior to the experiment.Stable cell lines expressing TEX264 WT and the different mutants were generated in a TEX264 KO background and used for clonogenic assay and immunofluorescence experiments.Transient transfection with the same plasmids was used for LysoIP, with plasmid validation by immunoprecipitation and a representative image of the immunofluorescence experiment.LysoIP in freshly made TEX264 KO HeLa without complementation or transient siTEX264-depletion still displays TOP1 uptake in lysosomes.Once the cell settled after about 10 passages, TOP1 uptake was prevented.

Zebrafish model
The AB strain of zebrafish (Danio rerio) was obtained from the European Zebrafish Resource Centre (EZRC) in Karlsruhe, Germany, and its housing and care were carried out following ethical guidelines (EU Directive 86/609/EEC, Croatian Federal Act on Animal Protection) under project licence HR-POK-023.According to established protocols, the zebrafish were kept at 28 C and a 14-hour light and 10-hour dark cycle. 138Zebrafish embryos of both gender were cultured in E3 media (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, and 0.33 mM MgSO4) at the same temperature and light cycle until they reached 2 days post-fertilization (dpf).Before starting the experiments, the embryos were manually dechorionated.Phenotypes were observed and recorded at 6 hpf, 30 hpf and 48 hpf.Images were captured using a Samsung 13-megapixel camera with an f/1.9 aperture applied to the ocular of the binocular Motic SMZ-171.

MRC FOCUS trial
The transcriptome of 361 formalin-fixed paraffin-embedded primary resections from advanced colorectal cancer patients enrolled in the randomised controlled MRC FOCUS trial 139 were analysed.All data concerning the patients characteristics, consent, and study design are reported here. 139Patients were required to have a colorectal adenocarcinoma diagnosis confirmed by histopathology, with inoperable metastatic or locoregional disease.These patients, starting first-line treatment with non-curative intent, were randomly assigned in the trial to either single agent 5-fluorouracil (given with levofolinate over 48 hours every two weeks -286 patients) or 5-fluorouracil plus irinotecan (FOLFIRI -75 patients) until progression.The study was registered as an International Standard Randomised trial (ISRCTN 79877428).

METHOD DETAILS Total cell extract & immunoblot
For the total cell extract, cells were lysed using RIPA buffer supplemented with protease and phosphatase inhibitors and sonicated with a Bioruptor Plus sonicator (30 s ON, 30 s OFF for 10 cycles).The supernatant obtained from centrifugation at 21,000 x g for 15 min at 4 C was quantified with a BCA assay (ThermoFisher Scientific).
For immunoblot, samples were denatured by boiling for 5 min at 95 C in Laemmli buffer, then separated by SDS-PAGE and transferred (wet transfer) to PVDF membrane.Membranes were blocked in 5% milk in TBS-T for 1 hour at room temperature.Membranes were incubated overnight with primary antibodies (5% BSA in TBS-T), washed three times with TBS-T and incubated with secondary antibodies coupled to HRP for 1 hour at room temperature.After three additional washes with TBS-T, membranes were detected with ECL-based chemiluminescence on the iBright imaging system.The immunoblot images were taken using the iBright system and contrast was adjusted using the iBright analysis software.The images presented correctly represent the original data.The contrast and brightness adjustment were equally applied to an entire membrane.
Isolation of proteins on nascent DNA -iPOND iPOND was performed as described previously. 140Briefly, HEK293T cells were synchronised by a single thymidine pulse for 17 hr, then released for 3 hr into the S phase.Approximately 100 million cells per condition were labelled with 10 mM EdU for 15 min or treated with Camptothecin (CPT) at 100 nM or 1 mM for 40 min and labelled with EdU for the final 20 min.A longer EdU pulse accounted for the reduced EdU incorporation rates in CPT-treated cells. 63Cells were fixed with 1% formaldehyde for 20 min and quenched with 0.125 M glycine.After permeabilization with 0.25% Triton, the click reaction was performed at room temperature for 1 hr in a buffer containing 10 mM biotin azide, 10 mM sodium L-ascorbate and 2 mM CuSO4.For negative control samples, DMSO was used instead of biotin azide.Cells were then lysed in 1% SDS supplemented with phosphatase and protease inhibitors, and chromatin was fragmented into 50-300 bp fragments by sonication with a Bioruptor Pico sonicator (Diagenone, 30 s ON, 30 s OFF, 50 cycles).After centrifugation at 16,100 x g for 20 min, a 5% lysate volume was set aside as input.Biotin-labelled EdU was captured by incubating samples with streptavidin-coupled agarose beads (Novagen) for 16 hr.Beads were washed once with 1% SDS, once with 1 M NaCl and twice with NTN buffer (100 mM NaCl, 20 mM Tris HCl pH 7.4, 0.05% IGEPAL).Crosslink reversal and elution of immunoprecipitated proteins were achieved by heating beads in 2x Laemmli buffer at 95 C for 5 min.

LysoIP
LysoIP was performed in HeLa expressing TMEM192-33HA as described previously. 67All conditions were treated with Bafilomycin A1 except for the proteomics analysis samples.For immunoblotting, 8 million cells of HeLa or HCT116 or 20 million cells of hTERT RPE-1 were collected in KPBS (136 mM KCl, 10 mM KH2PO4, adjusted pH 7.25 with KOH).For the input, 2.5% of the cells were lysed for total cell extract analysis.Cells for LysoIP were homogenized with 20 strokes of a 2 ml homogenizer.The homogenate was then centrifuged at 1000 x g for 2 min at 4 C.The supernatant containing the cellular organelles, including lysosomes, was incubated with anti-HA magnetic beads on a rotator shaker for 15 min at 4 C. Beads were washed four times in KPBS and eluted in 2 X Laemmli buffer, then boiled 10 min at 95 C.

Sample preparation
LysoIP was performed from 60 million HeLa cells per condition on four biological replicates.Lysosome content proteins (LysoIP) of non-treated and CPT-treated cells (50 nM for 6 hr) were obtained using 125 ml of beads per condition.Beads were washed 5 times and tubes containing the beads were changed twice.Elution was done in 60 ml Milli-Q water and sonication with the Bioruptor Plus sonicator (30 s ON, 30 s OFF for 10 cycles).17% of the sample was used to assess the quality of the sample (immunoblotting and Coomassie staining), and the remaining 83% was digested using the S-TrapÔ micro columns following manufacturer's protocol (ProTifi).Briefly, SDS was added to the samples (3% final concentration) before they were sequentially reduced with DTT (20 mM final concentration) and alkylated with Iodoacetamide (40 mM final concentration) for 30 min in the dark at room temperature.Samples were then acidified with phosphoric acid (1.2% final concentration) and proteins precipitated by adding 90% methanol in 100 mM TEAB buffer (1 to 7, sample: buffer ratio).Samples were loaded into the S-TrapÔ micro-column cartridges and washed four times with 90% methanol in 100 mM TEAB buffer.One microgram of trypsin in 50 mM TEAB buffer was added into the S-TrapÔ micro columns and incubated overnight at 37 C. Finally, tryptic peptides were sequentially eluted from the S-TrapÔ micro columns with 50 mM TEAB, 0.2% formic acid and 0.2% formic acid in 50% acetonitrile solution.Tryptic peptides were dried using a vacuum concentrator and reconstituted in 20 ml of 2% acetonitrile, 0.1% trifluoracetic acid.LC_MS/MS Peptides were analysed by reverse phase chromatography using a Vanquish Neo UHPLC system (operated in Trap and Elute mode) connected to an Orbitrap Ascend (Thermo Fisher Scientific).1.5% of tryptic peptides were loaded onto a trap column (AcclaimÔ PepMapÔ 100 C18) and separated on a 50 cm-long EASY-SprayÔ HPLC column using a 60 min linear gradient from 2% to 35% of buffer B (0.1% forsmic acid in acetonitrile) at 250 nl/min flow rate.Eluted peptides were then analysed on an Orbitrap Ascend operated in data-dependent mode, with advanced peak detection (APD) enabled.Survey scans were acquired in the Orbitrap at 120 k resolution over a m/z range of 400 -1500 and S-lens RF of 30.MS2 scans were obtained in the Ion trap (rapid scan mode) with a Quad isolation window of 1.6, 40% AGC target and a maximum injection time at auto, with HCD activation and 28% collision energy.

Data analysis
Mass spectrometry raw files were analysed in Fragpipe (v19.1) using the Label-Free Quantitation with Match Between Runs workflow (LFQ-MBR) with minor changes to the default settings.Briefly, data were searched against the reviewed human UniProt-Swissprot database (downloaded Jul 2022, containing 20386 sequences), selecting trypsin as proteolytic enzyme (maximum 2 missed cleavages), carbamidomethylation (C) as fixed modification and oxidation (M), acetylation (K, N-terminal) and phosphorylation (STY) as variable modifications.MS1 IonQuant was selected to calculate LFQ with MBR ion FDR at 1% and to report MaxLFQ.In addition, the top N ion was set up to 10 to get an iBAQ equivalent output.Fragpipe outputs were further analysed in Perseus (v1.6.2.2).In brief, the top 10N intensities were log 2-transformed, filtered by three valid numbers out of the four in at least one group and normalised by median subtraction.Missing values were then imputed (following the normal distribution).A two-sample student t-test combined with a Permutation -FDR correction (5%) was applied.

LysoIP-Seq
NGS sequencing of the DNA fragments purified by LysoIP was performed by GenScript for two biological replicates.After the capture of lysosomes by LysoIP from 40 million Hela cells, beads were washed four times in KPBS and eluted in 50 mM Tris pH 7.4, 150 mM NaCl and 0.2% Triton.For Picogreen quantification, samples were digested with proteinase K for 1 hr at 55 C before quantification according to the manufacturer protocol.For the library preparation, samples were denatured at 95 C to convert (ds)DNA into (ss) DNA.Adapters were added to both ends of the ssDNA fragments using the T4 polynucleotide kinase and the T4 ligase.Subsequently, a digestion enzyme was used to remove excess adapters, adapter dimers and unligated ssDNA samples.PCR amplification of the library was performed by the KAPA HiFi HotStart (Roche).The library was sequenced on the Illumina NovaSeq platform.
Data analysis was performed using Trimmomatic, to remove adapters and low-quality reads, Pandaseq, to merge the clean data, and BWA-MEM, to align the merge sequence to Hg38 reference genome.Samtools was used to extract the Mitochondrial DNA and Nuclear DNA mapping reads and assess the enrichment in centromeric regions.Genome distributions were obtained using the Homer annotates peaks tool.The bedtool multiple intersection was used to identify overlapping picks in the sequencing data, and the Venn diagram was produced using Eulerr online tool.Bigwig files were generated for visualisation in the Genomic browser IGV.

Whole genome sequencing
Genomic extraction was performed in HeLa WT or TEX264 KO , either untreated or treated with 10 nM CPT for 7 days continuously.Treatment was changed every 48 hr, and cells were split on day 5 to avoid overconfluency.When collected, a fraction was kept for immunoblotting.After harvesting, genomic DNA was purified using the Genomic DNA Buffer Set (Qiagen).
Whole genome sequencing was performed GenScript.Genomic DNA quality was assessed using a Qubitâ 3.0 Fluorometer.Whole genome DNA libraries were prepared using Hieff NGSâ OnePot Pro DNA Library Prep Kit V2 (Yeasen) and VAHTS Dual UMI UDI Adapters Set 1 -Set 4 for Illumina (Vazyme).The quality of the libraries produced was controlled using Qsep 100 Analyzer (BiOptic lnc.) and Qubitâ 3.0 Fluorometer.Each genome DNA library was sequenced using 150bp paired end reads on the Illumina NovaSeq6000 platform.All samples satisfied the minimum 150G data with an average 50x coverage.Quality control on raw Illumina fastq reads was done to detect contamination and assess the quality of reads.Reads were cleaned by removing low-quality reads and adapters using Trimmomatic (v0.3).Somatic variant calling was carried out using the Sarek workflow 141 v3.3.2, with reads aligned using BWA-MEM v0.7.17-r1188 and single base substitutions (SBS) called using Mutect2 142 v4.4.0.0 and Strelka 143 v2.9.10.SBS mutational signatures were assigned based on the union of Mutect2 and Strelka variant calls, using SigProfilerAssignment 144 v0.0.31 against COSMIC 145 signatures set v3.3.

Immunofluorescence
For standard immunofluorescence, cells were fixed in 4% formaldehyde in PBS for 15 min at room temperature.Then washed with PBS and permeabilised in 0.5% Triton X-100 in PBS for 15 min at 4 C.After blocking in 5% BSA/PBS for 1 hr at 37 C, cells were incubated with primary antibody (1:250) diluted in 2.5% BSA/PBS solution for 1 hr at room temperature.Then, coverslips were washed with PBS and incubated with secondary antibodies (1:500) and (1:500) for 1 hr at room temperature.Coverslips were mounted onto slides using Fluoromount G (ThermoFisher).Images were taken using a Zeiss 710 LSM microscope utilising a Plan-apochromat 63 x lens with a 1.4 NA and oil immersion.Images were collected sequentially to avoid any overlap between dyes, ensuring the same MBS filterset was maintained for all acquisitions.Images were gathered in 1024 x 1024-pixel format at approximately 55% Nyquist sampling, with full Nyquist sampling not being appropriate for this experiment.Images were acquired at 12-bit with 4x averaging being utilised to help with spurious noise within the images.Analysis was carried out using ImageJ Fiji and CellProfiler.
For imaging the reporter mCherry-TOP1-GFP protein, cell were transfected transiently and seeded on coverslips.After treatment, cell were fixed in PBS containing 2% FBS, 3.2% formaldehyde.Then stained with DAPI, mount and imaged as standard.
For imaging of the protein aggregates, cells were fixed in 4% formaldehyde, permeabilised in 0.5% Triton X-100 and 3 mM EDTA in PBS for 15 min at 4 C. Blocking was performed in 5% BSA, 3 mM EDTA in PBS for 1 hr at 37 C. EDTA was used to avoid staining of membranes, nuclei or lipid droplets. 82Staining with Proteostat at 1:2,000 diluted in 2.5% BSA in PBS was done in the dark for 1 hr.Then, the dye was washed five times with 0.1% BSA, 0.2% Tween and 0.3 mM EDTA in PBS and twice with PBS only.Cells were incubated with primary and secondary antibodies and DAPI as described for standard immunofluorescence if needed.Imaging of the Proteostat dye was done using the Rhodamine settings as standard.
For imaging of TOP1cc foci, the protocol was previously described. 69Briefly, cells were fixed in 4% formaldehyde, permeabilised in 0.5% Triton X-100, and blocked with 5% BSA/PBS for 1 hr at 37 C. To render the DNA-protein crosslinks more accessible to the antibody, the coverslips wee incubated in 1% SDS/PBS at RT for 5 min.Then washed five times with wash buffer (0.1% BSA, 0.1% Triton X-100 in PBS) and twice with PBS.Incubation with TOP1cc primary antibody in 2.5% BSA/PBS was performed for 1 hr at RT. Before subsequent incubation, cells were washed once with a wash buffer for 3 minutes with gentle shaking and twice with PBS.Cells were incubated with another primary antibody if needed, then secondary antibodies, DAPI, and imaged as described for standard immunofluorescence.
For imaging of Anaphase Ultrafine Bridge (UFB), cells were fixed and permeabilised with the UFB pre-extraction/fixation buffer (4% PFA in 20 mM PIPES at pH 6.8, 1 mM MgCl2, 10 mM EGTA, and 0.2% Triton X-100) for 10 minutes at room temperature and washed with PBS containing 0.2% Triton for 5 minutes before being blocked with blocking buffer.Primary antibodies were incubated overnight and then washed using 0.2% Triton in PBS three times.After 2 hours of incubation with secondary antibodies, coverslips were washed with 0.2% Triton in PBS three times before being mounted onto a glass slide using VECTASHIELD antifade mounting medium (Eurobio Scientific) supplemented with 5 mg/ml DAPI.
Pre-extraction was used to better visualise g-H2AX and 53BP1 foci in the experiment with TEX264 KO and ATR inhibition.Before fixation, cells were washed with ice-cold PBS and incubated with pre-extraction buffer on ice for 2 min (HEPES 25 mM pH 7.4, 50 mM NaCl, 1 mM EDTA, 3 mM MgCl2, 0.5% Triton X-100).Then, it was washed in a pre-extraction buffer without Triton for 2 minutes on ice before fixation.Images were taken using the Leica DMi8 SP8 FALCON (see below).

Proximity Ligation Assay
Proximity ligation assay was performed according to the manufacturer's protocol of the Duolinkâ In Situ PLAâ kits (Sigma-Aldrich).HeLa cells stably expressing V5-tagged TEX264-WT were transfected with LC3-pEGFP-C2 for 24 hours before fixing and permeabilising as described for standard immunofluorescence.After blocking in the provided blocking reagent, coverslips were washed with wash buffer A (150 mM NaCl, 10 mM Tris pH 7.4, 0.05% Tween 20) before incubation with anti-GFP and anti-V5 primary antibodies (dilution 1:500).After 3 washes in buffer A, coverslips were incubated with PLA PLUS and MINUS probes.Coverslips were incubated with ligase followed by polymerase, with washes in buffer A between each step.Coverslips were washed twice for 10 minutes with wash buffer B (100 mM NaCl, 250 mM Tris pH 7.5) before staining with DAPI (1:1000) for 10 minutes.Finally, coverslips were washed twice with buffer A and once with 0.01 x buffer B before mounting using ProLongÔ Glass Antifade Mountant (Invitrogen).Images were taken using the Leica DMi8 SP8 FALCON laser scanning confocal microscope with a 63x lens with a 1.2 NA water immersion objective lens and a confocal pinhole size set to 111.4 mm.Images were acquired in 2048 x 2048-pixel format at 8-bit with a pixel size of 90nm.Analysis was carried out using ImageJ and CellProfiler.

Alkaline Comet Assay
Around 50,000 cells per condition were collected with Trypsin and resuspended in PBS 1% low melting point agarose (BioRad).Cell were added on slides already prepared with dried 1% agarose and lysed in lysis buffer (2.5 M NaCl, 100 mM EDTA, 10 mM Tris-HCl at pH 10.5, 1% DMSO and 1% Triton X-100) overnight at 4 C. DNA denaturation was performed in electrophoresis buffer (300 mM NaOH, 1 mM EDTA, 1% DMSO, pH > 13) for 30 min at 4 C. Then electrophoresis was performed for 25 min at 25 V. Neutralisation was performed with 500 mM Tris-HCl, pH 8 for 5 min twice and staining with SYBRÔ Gold Nucleic Acid Gel Stain (1:10 000, Invitrogen) for 30 min.Images were acquired using a Nikon Ni-E (Nikon, Japan) microscope utilising an Andor Zyla 4.2 Plus sCMOS camera (Oxford Instruments, United Kingdom) and a CoolLED PE-4000 light source (CoolLED, United Kingdom).For comet assay images a 10x magnification with 0.45 NA Plan Apo Lamba objective lens (Nikon, Japan) was used.Images were acquired using the standard FITC Nikon filter set with excitation set to 10% on the 470 nm line.The camera was exposed for 6 ms at gain 4 to allow a 12-bit image to be acquired.Olive moment was obtained using the OpenComet plugin on ImageJ software.

Live cell microscopy
Live imaging of the Lamin A/C -GFP assay was performed using the custom-made spinning disc confocal setup based on the UltraVIEW VoX (Perkin Elmer, Waltham, Massachusetts, USA) with the scan head (CSU-X1, Yokogawa) and a Nikon Eclipse Ti-E microscope with a Plan APO olLED PE-4000 li DIC N2 N / 0,17 WD 0.13 OFN 25 and APO TIRF 100x / 1.49 DIC N2 N / 0,13-0,20 WD 0.12 objectives.The illumination was provided by a diode pumped solid state 488nm/50mW laser diode and a cobalt solid state 561nm/50mW laser diode.405/488/561/640 dichroic mirror and multi-color channels were used, and images were acquired using an EMCCD iXon897 camera (Andor Technologies) and Volocity software (Improvision) every 20 s.Bleaching correction and rendering were done using ImageJ and the package TrackMate.
Live imaging of the reporter mCherry-TOP1-GFP assay was performed using an Olympus/Evident SpinSR SoRa spinning disc with Yokogawa CSU-W1 in spinning disc mode (50 mm pinhole disc).Imaging was carried out using a 60 x 1.3 NA silicone oil objective.Detection was carried out in dual camera mode with 2x Hamamatsu ORCA Fusion BT cameras set to 480 MHz read speed.A 561 LP dichroic mirror sent blue/green emission light to camera 2 and red/far-red emission to camera 1.The channel combinations were designed in 2 groups.Chanel group 1 was used at 561 nm 3% laser with only camera 1 at a 617/73 nm filter and 30 ms exposure.Chanel group 2 was used at 640 nm 3% laser and 488 nm 3% laser with camera 1 at a filter of 685/40 nm and 30 ms exposure and camera 2 at a filter of 525/50 nm and 30 ms exposure.Seven Z sections were captured at 0.4 mm spacing every across 5 independent xy positions every 20 s for 2 hr.A humidified 5% CO2 atmosphere and 37 C temperature control ensured sample viability.Postacquisition images were deconvolved using Evident/Olympus Cellense constrained iterative deconvolution using an advanced maximum likelihood algorithm with 5 processing iterations.The included adaptive PSF was used, and noise reduction was applied.Images and videos displayed are single z-slices.Repeats of the live imaging of the reporter mCherry-TOP1-GFP assay were performed using the custom-made confocal spinning disc.

Electron Microscopy
Cells were fixed in 2.5% glutaraldehyde and 4% PFA in 0.1 M PIPES buffer and embedded using the modified NCMIR Method for Serial Block Face SEM sample preparation. 146Following embedding, the blocks were sectioned (90 nm sections on 200 mesh Cu grids) and were imaged without further staining using a JEOL 1400 TEM operated at 120 kV with a Gatan Rio CMOS camera for a first check.The tip of the resin block containing the cell was then sawed off and remounted onto a 3View pin using conductive epoxy glue and polymerised overnight.The surface was then polished using a diamond knife and sputter coated with $15 nm gold/palladium.Volume EM datasets were acquired using a Zeiss Merlin Compact SEM fitted with a Gatan 3View system and OnPoint Detector.The parameters were 1.8 kV accelerating voltage, 30 mm aperture, and Focal Charge Compensation at 100%.The voxel size is 7.7 nm (x) x 7.7 nm (y) x 100 nm (z).At least 740 slices were acquired for each dataset.Manual characterisation and quantification of the blister-like structure were done using ImageJ.Further segmentation and 3D rendering of the volume EM was done using Arivis Vision 4D.STX17 KD were validated using a LEO 912AB transmission electron microscope with an Omega energy filter operated at 120 kV.
To preserve membrane and structure integrity, growing cells were fixed using high-pressure freezing fixation.Cells attached to 3 mm coverslips were put in a small volume of cryoprotectant solution (20% BSA in PIPES buffer) and high-pressure frozen using a Leica EM ICE.Frozen samples were then transferred under liquid nitrogen to a Leica EM AFS2 freeze substitution unit and processed as described 147 with the following alterations: the freeze substitution medium was 0.2% Uranyl acetate and 5% water in acetone, the resin used was Lowicryl HM20 Monostep (PolySciences) and the UV polymerisation schedule was 24 hr at -45 C, warm to 0 C over 12 hr, then 36 hr at 0 C. 90 nm sections were transferred to 50 mesh formvar-coated Cu grids and post-stained for 5 min with Reynolds lead citrate, washed and air dried, and then imaged on the JEOL 1400 TEM as above.
Cryo-Focused Ion Beam (FIB) milling QuantifoilÔ R 2/2 on 200 gold mesh grids (Jenna Bioscience) were glow discharged at air atmosphere for 45s on high power using a Harrick plasma cleaner.The grids were then immersed in a solution of fibronectin at 20 ng/ml, deposited in a 6-well plate and incubated in a droplet of fibronectin for an extra 30 min.Grids were washed twice with PBS, and 300 000 HeLa cells in media were added on top of the grids and left overnight.Cells were incubated with 50 nM of LysoTrackerÔ Deep Red (Thermo Fisher Scientific) for 2 hr, then treated with 50 nM CPT for 25 to 35 min before fixing by being plunge frozen using a Leica GP2.Vitrified grids were then clipped into autogrids and subsequently stored under liquid nitrogen.
Autogrid-clipped TEM grids were inserted into a cassette and loaded into the Plasma FIB Arctis microscope (Thermo Fisher Scientific).Identification of lamella sites and milling was then carried out using webUI (Sagio Development LLC).Briefly, an SEM tile set of the grid was used to identify cells suitable for milling.A reflection and far-red fluorescent z-stack were acquired for each identified cell using the in-chamber fluorescent microscope across a +/-7.5 mm distance from the focus position in 0.5 mm increments.The fluorescent z-stack was used to guide the position of the lamella template.The grid was then coated with organo-platinum using the gas injection system (GIS) for 50 and then sputter coated with platinum metal for 120.For each site, electron centring of the fluorescent target, followed by eucentric height and milling angle search (12 target), was performed before positioning a 15 mm width lamella milling template.Milling was performed at 30 kV using Argon gas to generate plasma and a targeted final lamella thickness of 150 nm.For each site, stress relief cuts away on each side of the intended lamella were first milled using a 0.74 nA ion beam followed by three milling steps at 0.74 nA, 0.2 nA and 60 pA to obtain a $400 nm thick lamella.Once all three milling steps had been completed for all the lamella sites, these thicker lamellae were successively ''polished'' using 20 pA down to a nominal software target thickness of 150 nm.A final reflection and far-red fluorescent z-stack was acquired using the in-chamber fluorescent microscope across a +/-7.5 mm distance from the focus position, in 0.5 mm increment for each of the polished lamellas followed by 5 s sputter coating with platinum metal before retrieving the autogrid into the cassette.
Grids were directly moved from the Arctis into the Titan Krios for data collection.Cryogenic electron tomography (cryo-ET) data were acquired on a G3i Titan Krios equipped with a Falcon 4i and Selectris-X energy filter.The PACEtomo_targetsFromMontage.py script 148 was used in SerialEM to add targets from medium mag montages acquired at 15,000x magnification.The PACEtomo_mea-sureOffset.py script 148 was used to measure and calculate the tilt axis offset value optimized for the z-movement of the sample.For the correlation, medium magnification montages were saved as single images and imported into MAPS for manual correlation.After the setup, the PACEtomo.pyscript 148 was run to collect selected targets in parallel, with beam tilt compensation.A dose-symmetrical tilt scheme covering the +/-54 range with 3 increments starting at a -12 degrees tilt angle to compensate for the lamellae pretilt was used.Tilt-series were recorded at a nominal magnification of 42.000x, resulting in a pixel size of 2.95 A ˚, with 3.36 e-/A ˚2 per tilt, across 8 frames, leading to a total dose of 121 e -/A ˚2.
Data were pre-processed using RELION, MRC files were motion-corrected using the RELION motion correction algorithm, half maps were created for downstream denoising, and CTF was estimated with CTFFIND4.Tomograms for initial visualization were reconstructed using automated reconstruction in AreTomo, while reconstructions from RELION were used for denoising using CARE.Membrane segmentation was carried out on denoised tomograms using the standard Membrain-Seg protocol. 149Segmentations were trimmed using IMOD before visualisation in ArtiaX in ChimeraX.
Rapid approach to DNA adducts recovery -RADAR A modified RADAR assay was used to assess the level of TOP1cc. 72Around 20 million HeLa cells were lysed in M buffer (6 M guanidine thiocyanate, 10 mM Tris-HCl pH 6.8, 20 mM EDTA, 4% Triton X-100, 1% N-lauroylsarcosine and 1% dithiothreitol).DNA was precipitated by adding 100% ethanol, then washed three times in DNA wash buffer (20 mM Tris-HCl pH 7.4, 50 mM NaCl, 1 mM EDTA, and 50% ethanol), and solubilised in 8 mM NaOH.DNA concentrations were quantified by Picogreen and confirmed by slot blot analysis on a Hybond N + membrane, followed by detection with an anti-(ds)DNA antibody.For TOP1cc, samples were digested with 100 U/mL of Benzonase nuclease for 30 min at 37 C and analysed by slot blotting on a Nitrocellulose membrane.
Isolation and detection of TOP1cc from zebrafish embryos was performed using a modified RADAR protocol 118 optimised for zebrafish embryos.DPCs were isolated from wild-type (WT) and tex264-silenced embryos at two days post fertilisation, as well as embryos overexpressing recombinant zebrafish tex264-WT or tex264-LIR mutant (30 embryos per condition were used for isolation).The main change from previously published RADAR isolations was snap-freezing the samples in liquid nitrogen and overnight lyophilization using a FreeZone 2.5 lyophilizer (Labconco, USA) instead of TCA precipitation.DPC isolates were dissolved in 50 mL of SDS loading buffer (4 M urea, 62.5 mM Tris-HCl (pH 6.8), 1 mM EDTA, and 2% SDS).For the detection of TOP1cc, an equivalent of 1-2 mg of DNA-normalized DPCs were dissolved in 200 mL TBST buffer (10 mM Tris-HCl (pH 7.5), 15 mM NaCl, 0.02% Tween 20), and applied to a nitrocellulose membrane using the vacuum aspiration and dot blot system (Bio-Dot Microfiltration System, BioRad).To validate the DNA quantifications, dot blot analysis was performed using a nylon membrane (RPN303B, GE Healthcare). 2 ng of DNA was applied to the nylon membrane, and DNA was detected with an anti-(ds)DNA antibody.

Biochemical purification of protein aggregates
The isolation of aggregates was adapted from a protocol previously described. 150Briefly, 8 million HeLa cells were treated for 12 hr.Lysis was performed using 50 mM Tris-Cl (pH 8), 150 mM NaCl, 0.2% Triton X-100 and 100 U/mL of Benzonase nuclease for 30 min with shaking at 4 C. Centrifugation was conducted at 21,000 x g for 30 min at 4 C. Supernatant was collected and used as the soluble fraction.The pellet was washed twice with 1.5% SDS in 50 mM Tris-Cl (pH 8) and 150 mM NaCl and centrifuged at 21,000 x g for 30 min at 16 C.A higher temperature was used to avoid SDS precipitation.The supernatant was collected and used as the SDS soluble fraction.The final pellet was solubilised in 100% formic acid by sonication with a Bioruptor Plus sonicator (30 s ON, 30 s OFF for 10 cycles).After at least 3 hr at RT, samples were centrifuged at 1,000 x g for 2 min, and then formic acid was evaporated using a Concentrator 5301 (Eppendorf) at 60 C for approximately 30 min.Once dried, the pellet was resuspended in water and 5 X Laemmli buffer.The pH was adjusted using 7.5 ml of 2 M Tris Base.Immunoprecipitation TEX264 KO HeLa cells transiently transfected with WT or TEX264 mutants were seeded at 20 million cells per condition.Doxycyclineinducible HEK293 Flp-In TRex cells expressing p97-Myc/Strep were seeded at 20 million cells per condition.When indicated, cells were synchronised in the S phase by a single Thymidine pulse for 17 hr, followed by a 3-hr released step.Cells were lysed in IP lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.2% Triton X-100, supplemented with protease and phosphatase inhibitors) and incubated on a rotator at 4 C for 10 min.Chromatin was then pelleted by centrifugation for 5 min at 1000 x g and then digested with 100 U/mL Benzonase in Benzonase buffer (2 mM MgCl2, 50 mM Tris, pH 7.4, 150 mM NaCl).A fraction of the lysates representing 1% of the elution was taken for the input.For recombinant TEX264-V5 or p97-Strep IP, lysates were supplemented with ethidium bromide (50 mg/mL) and incubated with anti-V5 agarose beads (ChromoTek) or anti-Strep-Tactinâ Sepharoseâ resin (iba LifeSciences) for 2 hr on a rotator at 4 C, washed four times with IP wash buffer (150 mM NaCl, 50 mM Tris-HCl, 0.06% Triton X-100), and resuspended in 2 X Laemmli buffer.For endogenous TEX264 IP, the lysates were incubated with 5 mg of TEX264 or IgG control antibodies (both homemade) for 2 hours on a rotator at 4 .Then, the lysates and antibodies mix was incubated with Protein A magnetic beads (New England BioLabs) overnight on a rotator at 4 C, washed and eluted as previously described.

Clonogenic assay
For the S-phase study, cells were seeded at 1,000 HeLa cells per well and synchronised by a single Thymidine pulse for 16 hr, then released for 3 hr into S-phase.Cells were then treated with the indicated autophagy drug and indicated doses of CPT for 6 hr.For the study comparing TEX264 KO to RNF4 inactivation, all cell lines used were seeded 72 hr after RNF4 siRNA depletion for both low-and high-dose CPT treatment at the same time.Low-dose CPT study was treated for 8 hr with the indicated doses of CPT (nM), and highdose CPT study for 45 min (mM).For the study of TEX264 mutant, stable cell lines expressing the WT or the mutant TEX264 were treated for 8 hr with the indicated doses of CPT.Seven days after treatment, colonies were fixed in 100% methanol for 10 min and stained using 0.5% crystal violet and 20% methanol for 10 min.The number of colonies was counted using the automated GelCount colony counter (Oxford Optronix).The surviving fraction was determined by normalizing the number of colonies in each condition to the CPT-untreated control.All experiments were conducted in triplicate across technical replicates.

Cell Viability Assay
Cells were seeded in 96-well plates (3,000 cells/well), synchronised by a single Thymidine pulse for 20 hr, and then released for 3 hr into the S phase.Treatment was performed for 48 hr with indicated doses of CPT.Cell viability was measured using the resazurin assay 4 hr after adding 30 mg/ml of resazurin (Sigma-Aldrich) to the cell.Reading was done using a microplate reader (BMG LABTECH CLARIOstar Plus,Serial no.430-3000) excitation at 560-15 nm and emission at 590-20 nm.

Zebrafish assay
To silence zebrafish tex264 (ENSDARG00000092154), a specific antisense morpholino oligonucleotide (tex264MO-splice) was designed to bind to the exon1 -intron 1 boundary, 151 thereby preventing splicing.Another tex264 morpholino was designed to inhibit translation (tex264MO-ATG) and was combined with the splice morpholino.To verify the specificity of the tex264 morpholino, we performed rescue experiments by co-injecting tex264MO and mRNA encoding the full-length coding sequence of either WT tex264 or mutant tex264 (LIR mutation, F283A) into embryos at the 1-4 cell stage.Zebrafish embryos were injected with a 1 nl injection mix between the one and the four-cell stage.The injection mix contained splice MO (100 mM), ATG-MO (100 mM), mRNA (250ng/ul) and 0,02 % phenol red in 300 mM KCl solution.
RNA isolation from zebrafish embryos was performed using the Monarch Total RNA Miniprep Kit (NEB).At two days post-fertilisation (dpf), five embryos per condition were collected and pooled to measure tex264 expression levels.Reverse transcription was performed with the ProtoScript II First Strand cDNA Synthesis Kit (NEB) using random primer and oligo-dT primer mix, according to the manufacturer's instructions.Analysis of RT-qPCR was performed using GoTaq qPCR mix (Promega), and primer pairs were designed to span exon-exon boundaries, thus ensuring specific amplification without interference with genomic DNA.Normalisation was done using the housekeeping gene atp50 (ENSDARG00000001788). Quantification was performed using the Qgene method. 152ene expression levels were calculated as Mean Normalized Expression (MNE) as previously described. 153

MRC FOCUS trial analysis
The transcriptome of the primary resections from advanced colorectal cancer patients enrolled in the MRC FOCUS trial 139 were analysed.Samples were profiled as part of the S:CORT consortium by Xcel microarray (Almac).The control arm of the MRC FOCUS transcriptome is publicly available in GEO as part of GSE156915, 119 while the FOLFIRI arm was profiled at a later stage.CEL files from both arms underwent robust multiarray average normalization with the 'affy' R package, 154 and batch effects were corrected at the probeset level by ComBat with the 'sva' R package.TEX264 expression was represented by one single probeset (ADXEC.3184.C1_s_at).
The primary endpoint was progression-free survival (PFS) censored at 15 months as previously done 119 and agreed by the investigators prior to any inspection of the data or analyses.This endpoint is biologically relevant and spans the timeframe of differential effects of first-line FOLFIRI on survival.Tumour assessment was done at baseline (%4 weeks before treatment), then every 12 weeks, with either CT or MRI, and scored according to RECIST 1.0.criteria.
Patients were divided into two groups, TEX264 high and TEX264 low, according to the median of their TEX264 RNA expression.Kaplan-Meier curves were produced according to the treatment received by the patients, either 5-fluorouracil alone or FOLFIRI.Univariate Cox regression determined hazard ratios (HR), confidence intervals (95% CI) and p-values with no addition of confounding factors.Interaction analysis between TEX264 vs Treatments (FOLFIRI/5-Fluorouracil) was also performed, providing a significant p-value.

QUANTIFICATION AND STATISTICAL ANALYSIS
Experiments were conducted in at least two biological replicates (n) and statistical analysis was performed using Prism 10 (GraphPad Software); significance is labelled as follows * p < 0.05; ** p < 0.005; *** p < 0.0005; ns, not significant.Statistical tests, biological replicates and details of the experiment can be found in the figure legend.Error are displayed as standard deviation (SD), unless stated otherwise in figure legend.Normality (Gaussian) distribution was tested using the Shapiro-Wilk test to determine whether the data met the assumptions of the statistical approach used (Ordinary Two-way ANOVA with main effect only and Dunnett's multiple comparisons tests, with a single pool variance).

ADDITIONAL RESOURCES
Published mass spectrometry proteomics was obtained from the ProteomeXchange Consortium via the PRIDE: PXD018092 and PRIDE: PXD011727 (Nakamura et al. 63 and Srivastava et al. 66 ) and from the source data of the publications (Dungrawala et al. 64 and Ribeyre et al. 65 ).All proteins found by proteomics were compared to a list of known autophagy-related proteins created from the human UniProt-Swissprot database, the Gene Ontology (GO) database and the Kyoto Encyclopedia of Genes and Genomes (KEGG) database.Category identification for the 277 autophagy proteins identified by IPOND proteomics was done manually.Ridge regression plot was performed using Anaconda with Python packages.Published ChIP-seq data were obtained from the Gene Expression Omnibus database for TOP1cc and TOP1 under accession numbers NCBI: GSE135808 and NCBI: GSE57628 ( Tan et al. 88 and Baranello et al. 87 ).TEX264 structure was predicted by AlphaFold.(legend continued on next page) Figure S3.Autophagy promotes TOP1 degradation during replication stress to prevent protein aggregation, related to Figure 3 (A) Immunoblot showing TEX264 and RNF4 depletion in cells used for clonogenic assay.(B) Immunoblot of TOP1 level in total cell extract (n = 4).Treatment with 50 nM CPT was performed for 24 h, and 2 mM MG132 (MG) or 50 nM BAF was added in the last 8 h.Treatment with 1 mM CPT was performed for 3 h with MG132 or BAF.Quantification of normalized TOP1 protein level.One-way ANOVA, SEM represented; UT, untreated.(C) Immunofluorescence of g-H2AX and 53BP1 foci after 1 h of 50 nM CPT or 1 mM CPT (n = 3).Scale bar, 10 mm.Quantification of positive nuclei for colocalized foci of both g-H2AX and 53BP1.Two-way ANOVA.Error bar, SD. (D) Method for isolation of aggregated protein by sequential lysis and denaturation step.The aggregates fraction is insoluble in lysis buffer and 1.5% SDS but solubilized in 100% formic acid to be loaded on the gel.Isolation of the aggregates performed after 12 h of treatment with 50 nM CPT or 1 mM CPT (n = 3); UT, untreated.*p < 0.05; ns, not significant.(legend continued on next page) (C) Alkaline comet assay performed on untreated cells and compared with 50 nM CPT for 1 h treatment (n = 3).WT HeLa cells were compared with TEX264 KO cells or indicated TEX264 complementation in TEX264 KO background.The quantification of Olive moment normalized against WT untreated; the median is represented.Scale bar, 20 mm.One-way ANOVA.(D) Mutational profiles of WT and TEX264 KO comparing untreated to treated cell population assessed by whole-genome sequencing.was performed with 10 nM CPT for 7 days continuously.Single base substitutions (SBSs) are represented as counts in each trinucleotide context.(E) Proportion of SBS mutational signatures found in the mutational profiles of WT and TEX264 KO cells.SigProfilerAssignments used to assign mutations to the nearest SBS mutational signatures present in the COSMIC v.3.3 signature set. 145F) Immunoblot showing the poly (ADP-ribose) polymerase (PARP) full length as a loading control and to monitor potential apoptotic effect in WT and TEX264 KO cells used for whole-genome sequencing after treatment with 10 nM CPT performed for 7 days continuously.(G) Immunofluorescence of g-H2AX and 53BP1 foci after 1 h of 50 nM CPT and 1 h recovery (n = 4).ATR inhibition was performed using 1 mM ATR inhibitor VE-822 (ATRi) during 1 h with 50 nM CPT and 1 h recovery.Scale bar, 10 mm.Quantification of nuclei positive for colocalized foci of both g-H2AX and 53BP1.Two-way ANOVA.Error bar, SD. *p < 0.05; **p < 0.005; ***p < 0.0005; ns, not significant.

Figure 1 .
Figure 1.Crosstalk between autophagy and DNA replication upon replication fork stalling by CPT (A) Co-immunoprecipitation of TEX264-WT-V5 in HeLa TEX264 KO background in asynchronized or S-phase-synchronized cells (n = 3).(B) Strategy for replisome proteomics analysis.Venn diagram of the replisome found by nascent chromatin capture (NCC) coupled to stable isotope labeling by amino acids in cell culture (SILAC) and mass spectrometry (MS) analysis, NCC-SILAC-MS 63 overlapped with all known autophagy factors.(C) Ridge regression plot of known autophagy factors at replication fork identified by SILAC-NCC-MS. 63(D) iPOND performed after 40 min of 100 nM CPT (0.1 mM) or 1 mM CPT. Representative blots from different biological repeats and quantification (n = 3).Error bar, SD. (E) Strategy for purification of intact lysosomes; LysoIP.(F) Proteome profiling of lysosomes purified by LysoIP after 6 h of 50 nM CPT.Proteins differentially expressed (À10log p >1.301) shown with full dark circles.(G) LysoIP performed in S-phase-synchronized HeLa after 6 h of 50 nM CPT.See also Figure S1 and TablesS1 and S2.

Figure 4 .
Figure 4. TOP1cc (TOP1 and its bound DNA fragment) are processed by lysosomes (A) LysoIP performed after 5 h of 50 nM CPT, and quantification graph (n = 3).Error bar, SD. (B) LysoIP performed after 5 h of 50 nM CPT (n = 3).(C) Quantification of DNA by Picogreen assay to detect DNA purified by LysoIP after 6 h of treatment (n = 3).Two-way ANOVA.Error bar, SD.Around 450 ng of DNA fragments of less than 1 kb were purified from 20 million cells by LysoIP after treatment with 50 nM CPT and 50 nM BAF.

Figure 5 .
Figure 5. Lysosomal degradation of TOP1cc occurs in interphase cells through the nuclear envelope (A) LysoIP performed after 6 h of 50 nM CPT in unsynchronized or S-phase HeLa cells (n = 3).SE, short exposure; LE, long exposure.(B) Live-cell imaging started 3 min after treatment with 50 nM CPT.Normalized fluorescence intensity measured along the dashed line for both mCherry and GFP.Scale bar, 5 mm; zoom panel, 1 mm.(C) Live-cell imaging.Arrows pointing at the alteration of Lamin A/C-GFP integrity at lysosomal contact sites (n = 3).Rendering by TrackMate of lysosomes and nuclear envelope integrity.Scale bar, 10 mm.

( D )FFigure 6 .
Figure 6.TEX264 is the receptor for TOP1cc degradation by selective autophagy (A) Overlap of known autophagy receptors at replication fork by NCC-SILAC-MS 63 and in lysosomes upon 50 nM CPT by LysoIP-MS.(B) AlphaFold prediction of TEX264 structure, with indicated domains.(C) Immunofluorescence after 2 h of recovery after 2 h of 50 nM CPT and quantification (n = 3).Two-way ANOVA.Error bar, SD.
All subjects provided written informed consent for further research on their samples to the clinical trial.Both the original clinical trial (FOCUS Ref: 79877428) and S:CORT study (ref 15/ EE/0241) were approved by the National Research Ethics Service in the United Kingdom.

(
A) Co-immunoprecipitation of endogenous TEX264 in WT cells or in the context of ATG7 depletion (n = 2).Unspecific immunoglobulin G (IgG) was used instead of TEX264 antibody as a control.(B) Overlap of the protein set identified by immunoprecipitation of replication forks interactome by MS[64][65][66] and known autophagy-related proteins from the human UniProt-Swissprot database, the Gene Ontology (GO) database, and the Kyoto Encyclopedia of Genes and Genomes (KEGG) database.Replisomes were identified by iPOND, or Bio-ID of PCNA, in different cell lines after treatment with different replication stress inducers.HU, hydroxyurea; UV, ultraviolet.(C) LysoIP was performed in HeLa WT or cells modified for LysoIP experiments by expression of the resident lysosomal tagged protein TMEM192-HA (n = 3).The LysoIP cell line allows the specific pull-down of lysosome vesicles without contamination from other cellular organelles.(D) Immunofluorescence of HeLa WT cells and TMEM192-HA cells used for LysoIP, using anti-HA antibody and anti-LAMP1 marker of the late endosome and lysosome vesicles (n = 3).Scale bar, 10 mm.(E) LysoIP performed after 6 h of treatment with 50 nM CPT in S-phase-synchronized HCT116 cells (n = 3).All conditions were treated with 50 nM BAF.(F) LysoIP performed after 6 h of treatment with 50 nM CPT in S-phase-synchronized cells in hTERT RPE-1 (n = 3).All conditions were treated with 50 nM BAF.(G) LysoIP performed in HeLa released from a single thymidine block into the S-phase and treated for 3 h with 50 nM CPT, 5 mM irinotecan, or 50 nM SN38 (n = 3).All conditions are treated with 50 nM BAF.

Figure S2 .
Figure S2.Autophagy regulates TOP1 degradation to promote DNA damage repair and cell survival, related to Figure 2 (A) Immunofluorescence of TOP1cc foci in WT cells, after induction of shATG7 (ATG7 KD ), 4 h of treatment with 250 nM Torin, or 2 h of treatment with 50 nM CPT.Quantification of positive nuclei for TOP1cc foci (n = 3).Two-way ANOVA, error bar, SD; comparison against untreated.(B) Immunoblot showing ATG7 depletion in cells used for immunofluorescence and RADAR.(C) RADAR assay to assess TOP1cc level in WT and ATG7 KD cells unchallenged (n = 2).Double-stranded (ds) DNA is used as a loading control.

Figure S5 .
Figure S5.CPT induces nuclear envelope blistering and TOP1cc exit to the cytoplasm, related to Figure 5 (A) Representative electron microscopy images for a distinct set of cells.Treatment was conducted for 30 min with 50 nM CPT.Scale bar, 500 nm.Arrows pointing at blister structures.Nu, nucleoplasm; Cy, cytoplasm; A, autophagosome.(B) Representative electron tomographic slices fixed by high-pressure freezing showing nuclear envelope blister in HeLa.Treatment was conducted for 30 min with 50 nM CPT.Scale bar, 2 mm (left); scale bar, 1 mm (zoom, right).Arrows pointing at blister structures.(C) Immunofluorescence of TOP1cc foci and cyclin B1 after 3 h of treatment with 50 nM CPT (n = 3).Cyclin B1 is used as a control for the inhibition of exportin by leptomycin B (LTB).Scale bar, 10 mm.(Left) Quantification of positive cells for cytoplasmic TOP1cc foci.One-way ANOVA.(Right) Cyclin B1 intensity per nuclei (quantification for a representative experiment).t test.Error bar, SD. *p < 0.05; ***p < 0.0005; ns, not significant.

Figure
Figure S6.TEX264 with p97 and SUMO acts as the receptor for TOP1cc degradation by autophagy, related to Figure6 Figure S7.Autophagy of TOP1cc is independent of SPRTN, and TEX264 inactivation promotes genomic instability, related to Figure7 (A) Immunofluorescence of protein aggregates using the Proteostat dye and the late endosome and lysosome marker LAMP1 after 16 h of treatment with 50 nM CPT in WT or D-SPRTN (partial knockout [KO]).Quantification of cells positive for aggregates (n = 3).Scale bar, 10 mm.Two-way ANOVA.Error bar, SD. (B) Anaphase ultrafine bridge detection in cells treated for 2 h with 50 nM CPT and allowed to recover for 24 h (n = 3).Positive BLM and RPA32 anaphase bridges are labeled with arrows.Scale bar, 10 mm.Quantification of anaphase cell presenting ultrafine bridge.t test.Error bar, SD.