We previously reported that TDP1 is required for the repair of TOP1-induced DSBs in quiescent hTERT-immortalized retinal pigment epithelial cells (RPE-1) cells [23]. To evaluate potential defects in TOP1-induced DSB repair in TDP1-associated neurological disease SCAN1, we conducted complementation experiments in TDP1 RPE-1 cells using an empty vector (EV), with a wild-type TDP1 (WT), with a histidine-to-alanine mutant in the catalytic histidine 263 of TDP1 (hereafter TDP1), reported to be inactive [5, 25], or with the SCAN1-associated mutation H493R (hereafter TDP1), which were under the control of a doxycycline-inducible promoter (for details, see methods). To study replication-independent TOP1-associated DSBs, we synchronized cells in G0/G1 by confluency and serum starvation resulting in over 97% of RPE-1 cells arrested in G0/G1 (Supplementary Fig. 1) (for details, see methods) [26]. Twenty-four hours of doxycycline treatment in RPE-1 quiescent cells induced equivalent levels of wild-type TDP1 and TDP1, which were overexpressed compared to endogenous TDP1 (Fig. 1a). TDP1 expression was significantly lower than that of wild-type TDP1 and TDP1, likely due to the previously reported toxicity of TDP1 (Fig. 1a) [27,28,29]. Next, we treated quiescent cells with CPT, which selectively induces abortive TOP1ccs [30]. Notably, exposure to CPT rapidly induced 53BP1 and H2AX serine 139 phosphorylation (hereafter γH2AX) immunofoci, common surrogate markers of DSBs [31] (Fig. 1b). RNA polymerase II elongation inhibition with 5,6-dichloro-1-β-D-ribofuranosylbenzimidazole (DRB) before TOP1 poisoning resulted in ≃80% reduction in 53BP1 foci in all cell lines, in agreement with TOP1-induced DSBs in quiescent RPE-1 cells being dependent on gene transcription [23] (Fig. 1b). Importantly, expression of wild-type TDP1 suppressed the accumulation of CPT-induced DSBs to wild-type levels (TDP1) and to a much greater extent than TDP1, which remained close to the levels observed with the EV (Fig. 1b). TDP1 expression resulted in similar number of DSBs compared to EV and TDP1(Fig. 1b). These results suggest that, similarly to TDP1 deficiency, both TDP1 and TDP1 lead to the accumulation of transcription-associated CPT-induced DSBs in RPE-1 quiescent cells. Notably, the overexpression of TDP1 mutants did not alter global transcription levels (Supplementary Fig. 2a). Furthermore, we did not observe significant changes in TOP1 levels, which, as previously shown, underwent significant degradation upon CPT treatment [19] (Supplementary Fig. 2b). Degradation was uniform among the mutants, indicating that they did not influence the formation or degradation of abortive TOP1ccs, consistent with the established role of TDP1 (Supplementary Fig. 2b).
Next, we measured TOP1-induced DSB repair rates by following the kinetics of 53BP1 foci after CPT removal in quiescent cells. Wild-type TDP1 expression suppressed the repair defect observed in EV-complemented cells, completing repair three hours after CPT removal (Fig. 1c). In contrast, TDP1-expressing cells exhibited a significant repair defect compared to wild-type TDP1, similar to EV-complemented cells, suggesting that TDP1 overexpression might slow down TOP1-induced DSB repair. Notably, TDP1-expressing cells exhibited a very strong repair defect that extended for six- and twenty-four hours post-treatment, suggesting that, not only TDP1 is unable to repair transcription-associated TOP1-induced DSBs in quiescent cells, but it completely blocks the repair process (Fig. 1c). Importantly, we obtained similar results when the doxycycline concentration was reduced, resulting in lower levels of TDP1, TDP1 and TDP1 expression, comparable to endogenous TDP1 (TDP1), suggesting that the strong repair defect observed in TDP1 cells was not due to TDP1 overexpression (Fig. 1d).
To directly evaluate TOP1-induced DSB formation and repair, we performed a neutral comet assay, a technique that specifically detects DSBs [32], confirming that TDP1-deficient and TDP1-complemented cells accumulate CPT-induced DSBs and that TDP1 is unable to resume repair after twenty-four hours (Fig. 1e). Intriguingly, TDP1 expression also resulted in a significant TOP1-induced DSBs repair defect, even more severe than in TDP1 cells (Fig. 1e). These results were somehow contradictory to repair kinetics and suggest that TDP1 blocks TOP1-induced DSBs, although these excess breaks are not detected by 53BP1 foci immunofluorescence or by γH2AX foci which colocalize with 53BP1 (Fig. 1c). To further confirm this observation, we directly visualize DSB in quiescent cells using premature chromosome condensation (see materials and methods and Supplementary Fig. 2c for details). After CPT treatment and repair, quiescent cells were fused to HeLa mitotic cells, promoting the condensation and thus the visualization of single chromatid chromosomes in quiescent cells. TOP1-induced DSB formation was directly measured by visualizing small chromosomal fragments upon CPT treatment and by Giemsa staining, confirming that TDP1 cells accumulate CPT-induced DSBs after twenty-four hours of repair (Fig. 1f).
Finally, to examine whether TDP1 and TDP1 expression also affect the repair rate of SSBs, we quantified nuclear poly(ADP-ribose) (hereafter PAR) levels in cells upon DNA damage by immunofluorescence, which provides an indirect measure of SSB levels [33]. Notably, we detected much higher PAR signal in TDP1 and in TDP1 than in wild-type TDP1-expressing cells upon CPT treatment (Supplementary Fig. 3). Nevertheless, although the decrease in PAR signal was clearly slower in TDP1 and TDP1 than in wild-type TDP1-expressing cells, almost all PAR signal had disappeared three hours after repair (Supplementary Fig. 3). These results demonstrate that the blockage of TOP1-induced DSB repair by TDP1 and TDP1 is much more prolonged in time than the blockage of SSB repair.
To elucidate the mechanism underlying the blockage of TOP1-induced DSB repair, we measured DNA-protein covalent complexes by in vivo complex of enzyme (ICE) assay [34]. Our cellular model provides a highly useful system for detecting abortive TOP1cc and DNA-TDP1 covalent complexes, as previously shown by Ghosh and collaborators in the study of mitochondrial TOP1 [35]. For the detection of TOP1ccs, we employed an antibody raised against a peptide corresponding to the active site of the TOP1 with a phosphorylated Tyr723 residue [36]. In agreement with previous reports, TDP1-deficient quiescent RPE-1 cells accumulated TOP1ccs upon CPT treatment (Fig. 2a). Notably, the expression of wild-type TDP1, but not of TDP1 or TDP1, reduced the accumulation of TOP1cc upon CPT treatment in TDP1 cells, suggesting that TDP1 and TDP1 largely prevent the removal of abortive TOP1ccs (Fig. 2a). We next employed an antibody to detect the complementing FLAG-tagged TDP1 variants. Strikingly, we detected a significant accumulation of covalent DNA-TDP1 complexes in TDP1, but not in TDP1 or TDP1-expressing cells, demonstrating that H493R mutation covalently traps TDP1 on DNA in quiescent RPE-1 cells upon TOP1 poisoning (Fig. 2b). Importantly, RNA polymerase II elongation inhibition with DRB prior to TOP1 poisoning significantly reduced TOP1cc accumulation (Fig. 2c) and ablated DNA-TDP1 covalent trapping (Fig. 2d), in agreement with abortive activity of TOP1 in quiescent RPE-1 cells being dependent on gene transcription.
Next, we measured TOP1cc and DNA-TDP1 covalent complexes during repair in TDP1-expressing cells. We observed a fast disappearance of TOP1ccs within the first hour (Fig. 2e). This is consistent with the fact that ICE signal comes from abortive and non-abortive TOP1cc but, considering the difference upon induction observed in wild-type TDP1 overexpressing cells, it also suggests that most of abortive TOP1ccs rapidly diminish following CPT removal (Fig. 2e). Contrary, the levels of DNA-TDP1 covalent complexes remained unchanged for one hour and started to decay after three hours (Fig. 2f). These results indicate that DNA-TDP1 covalent complexes persist but that, eventually, are either reverted, removed, or degraded. To directly test the persistence of trapped TDP1 we increased the DNA load in the assay. Importantly, twenty-four hour after CPT, significant amounts of DNA-TDP1 covalent complexes, but not of abortive TOP1cc, remained compared to wild-type TDP1 (Fig. 2g, h). Altogether these results demonstrate that TDP1 trapping is not completely irreversible in vivo but, at least in some cases, is persistent.
We next compared post-repair TOP1cc accumulation among all mutants. Notably, after one hour of repair, TOP1ccs were significantly more persistent in TDP1 than in EV, wild-type and TDP1-expressing cells, indicating that TDP1 is unable to efficiently repair abortive TOP1cc (Fig. 2i). More importantly, after twenty-four hours, TDP1-expressing cells still retained significant TOP1cc signal compared to wild-type TDP1 (Fig. 2j). These results suggest that, unlike TDP1-deficient and TDP1-expressing cells, TDP1-expressing cells accumulate abortive TOP1ccs in vivo.
The blockage of TOP1-induced DSB repair in TDP1-expressing cells suggests a potential dominant-negative effect of the H493R mutation. To determine whether this DSB repair blockage is compatible with SCAN1, which is an autosomal recessive syndrome, we overexpressed wild-type TDP1 and TDP1 in wild-type RPE-1 cells (hereafter TDP1) (Fig. 3a). The overexpression levels were similar to those observed in TDP1 cells (Supplementary Fig. 4a). We did not observe significant changes in TOP1 levels, which, as previously shown, underwent uniform degradation upon CPT treatment (Supplementary Fig. 4b). Following CPT treatment, wild-type TDP1 overexpression significantly reduced the accumulation of TOP1-induced DSBs, suggesting that endogenous TDP1 might be limiting for the repair of TOP1-induced SSBs and DSBs under the conditions used in this study (Fig. 3b). Notably, TDP1-overexpressing cells exhibited very similar levels of DSBs to those of EV-complemented cells when treated with CPT, indicating that TDP1 expression in wild-type cells does not lead to a large accumulation of TOP1-induced DSBs (Fig. 3b). More importantly, TDP1-overexpressing cells generally exhibited no significant defects in TOP1-induced DSB repair either, demonstrating that TDP1 does not block repair when present in heterozygous state (Fig. 3c). Since we had previously associated abortive TOP1cc accumulation and TDP1 trapping with TOP1-induced DSB repair blockage, we analyzed TOP1cc and TDP1 covalent complexes by ICE assay. Strikingly, TDP1 expression in wild-type cells did not result in either abortive TOP1cc accumulation nor in TDP1 trapping (Fig. 3d, e). Given the strong overexpression of TDP1, together with our previous observations, these results suggest that the reason of TDP1 trapping and DSB blockage are not dominant is dependent on the presence of wild-type TDP1.
To study the dominant or recessive nature of H263A, we also expressed TDP1 in wild-type RPE-1 cells (Fig. 3a). TDP1 expression was significantly lower than that of wild-type TDP1 and TDP1 (Fig. 3a). Notably, TDP1-overexpressing cells accumulated more DSBs than EV-complemented cells when treated with CPT, indicating that TDP1 expression in heterozygous state can promote the accumulation of TOP1-induced DSBs upon TOP1 poisoning (Fig. 3b). TDP1-expressing cells did not show significant repair defects compared to EV, wild-type TDP1 or TDP1-overexpressing cells (Fig. 3c). More importantly, TDP1 overexpression did not result in a significant increase in abortive TOP1ccs or in DNA-TDP1 complexes (Fig. 3d, e). Despite the observed increase in the accumulation of DSBs upon CPT treatment, these results suggest the recessive nature of TDP1.
Some studies have shown that tyrosyl-DNA phosphodiesterase 2 (TDP2), a TOP2cc debulking enzyme involved in TOP2-induced DSB repair [37,38,39], can process TOP1-DNA adducts in vitro. Additionally, TDP2 deficiency has been shown to increase CPT hypersensitivity in TDP1-deficient avian, murine and human cells [40, 41]. Recently, Geraud and colleagues demonstrated that TDP2 can back up TOP1-induced DSB repair in G1 U2OS cells [24]. To investigate the participation of TDP2 in TOP1-induced DSB repair in quiescent RPE-1 cells, we depleted TDP2 in both TDP1 and TDP1 cells (Fig. 4a). As previously reported, TOP1 levels remained unchanged, undergoing uniform degradation upon CPT treatment (Supplementary Fig. 5a). While TDP2 depletion did not significantly promote the accumulation of TOP1-induced DSBs in TDP1 cells, it led to a mild increase compared to TDP1-deficient cells (Fig. 4b). Consistently, TDP2 depletion only caused a minor defect (3 h) in TOP1-induced DSB repair, suggesting that TDP2 is dispensable for repairing these breaks in wild-type cells (Fig. 4c). However, strikingly, TDP2 depletion in TDP1 cells exacerbated the DSB repair defect observed in the latter, supporting the idea that TDP2 plays a key role in facilitating the residual DSB repair detected in TDP1 cells (Fig. 4c). This repair defect was suppressed by wild-type TDP1 overexpression (Supplementary Fig. 5b).
Next, we asked whether TDP2 could hydrolyze TOP1ccs during TOP1-induced DSB repair, similar to TDP1. To ensure an appropriate time window, we measured TOP1ccs one hour after CPT removal, when the TDP1 debulking defect in TDP1 cells is barely detectable (Fig. 2i). Notably, TDP2 depletion in TDP1 cells further highlighted this defect (Fig. 4d). This repair defect was suppressed by TDP1 overexpression (Supplementary Fig. 5c). Altogether, these results indicate that TDP2 is not required to release TOP1ccs from DNA ends in quiescent RPE-1 cells unless TDP1 is absent.
Since the H263A mutation does not trap TDP1 on DNA and its repair kinetics differ from those of H493R, we reasoned that the mechanism of TOP1-induced DSB blockage might be different from that of H493R but somehow linked to the recruitment of TDP1 to abortive TOP1ccs. To explore this possibility we analyzed the chromatin recruitment of wild-type TDP1, TDP1 and TDP1 upon CPT-induced DSB formation and repair (Supplementary Fig. 6a). However, we did not observe significant changes in chromatin recruitment for any TDP1 forms, either upon induction or after twenty-four hours of repair (Supplementary Fig. 6b). Notably, TDP1 presented two different electrophoretic mobilities (Supplementary Fig. 6b, c). Strikingly, we observed a very strong increase in the low-mobility form of TDP1 on chromatin upon CPT treatment (Supplementary Fig. 6c).
To finely explore whether the H263A mutation might provoke a prolonged non-covalent association of TDP1H263A to abortive TOP1ccs, we analyzed the co-localization of TDP1 and TOP1cc by proximity ligation assays (PLA). Upon CPT treatment, both TDP1 and TDP1 exhibited a slightly increased colocalization with colocalization with TOP1ccs compared to wild-type TDP1 (Supplementary Fig. 6d). Notably, while most of PLA signal decayed during repair, TDP1 retained higher colocalization with TOP1ccs than wild-type TDP1 and TDP1, whose levels returned to those observed under untreated conditions (Supplementary Fig. 6d). Together with SSB and DSB repair kinetics previously shown in TDP1 (Fig. 1c-e & Supplementary Fig. 3), these results suggest that TDP1 may persist associated to TOP1-induced DSBs.
We recently described that TDP1 suppresses chromosomal translocations and cell death induced by abortive TOP1 activity during gene transcription [23]. For a better understanding of the relevance of the removal of the TOP1cc adduct we analyzed potential TOP1-induced DSB repair defects in other SSB repair factors downstream of TDP1. We focused on PNKP, the following enzymatic activity working on TOP1cc repair. We achieved more than 90% depletion of PNKP with CRISPR-Cas9 by a single guide RNA (sgRNA) in two independent clones (Supplementary Fig. 7a). PNKP depletion resulted in a small but significant defect in TOP1-induced DSB repair, similar to that observed in TDP1 cells. These results suggest that restoration of DNA end polarity is relevant in TOP1-induced DSB repair. Next, we studied whether PNKP depletion influenced the formation of transcription-associated chromosomal reorganizations in TDP1 complemented cells upon CPT treatment. Cells were treated and maintained in G0/G1 during six hours after treatment with CPT and then released for the isolation of metaphases [23] (Fig. 5a). However, PNKP-depleted cells did not provoke a significant increase in chromosomal translocations (Supplementary Fig. 7b). These results indicate that while PNKP participates in TOP1-induced DSB repair, it is not essential, being dispensable for the suppression of genome instability. This also suggest that removal of TOP1cc from 3'-DNA termini must be a key step to prevent genome instability induced by TOP1-induced DSBs.
Given that TDP2 backs up TDP1 loss in TOP1-induced DSB repair, we hypothesized that TDP2 may be relevant to suppress genome instability in quiescent TDP1-lacking cells. To test this, we examined the influence of TDP2 in the formation of TOP1-induced chromosomal translocations. Notably, TDP2 depletion in TDP1 cells further increased CPT-induced translocations compared to TDP1-deficient cells alone, whereas no effect was observed in TDP1 cells (Fig. 5a). Altogether, these results suggest that TOP1cc removal is a key step to prevent genome instability generated by TOP1-induced DSBs in quiescent cells.
Next, to estimate the physiological relevance of TDP1 and TDP1 blockage of TOP1-induced DSB repair, we studied the formation of chromosomal reorganizations in TDP1 complemented cells. While wild-type TDP1 suppressed chromosomal translocations induced by CPT, TDP1 did not (Fig. 5b). TDP1 showed an intermediate phenotype (Fig. 5b). Notably, TDP1-expressing cells showed a higher accumulation of chromosomal translocations than both EV and TDP1-expressing cells, being the latter the one that accumulated the least (Fig. 5b). These translocations correlated with relative accumulation of TOP1-induced DSB upon CPT treatment (Fig. 1b), reinforcing the link between TOP1-induced DSBs and chromosomal translocations.
Finally, to study the contribution of TDP1 and TDP1 to CPT-induced cytotoxicity in quiescent cells, we performed clonogenic survival in RPE-1 cells that had been treated with CPT, allowed to repair while quiescent and finally transferred to serum-containing medium (Fig. 5c). As we had previously shown, TDP1 cells exhibited a high sensitivity to CPT (Fig. 5c and Supplementary S5d). This hypersensitivity was significantly exacerbated by TDP2 depletion, in agreement to our results on DSB repair kinetics and chromosomal translocations, and further confirming that TDP2 backs up TDP1 in RPE-1 quiescent cells (Fig. 5c and Supplementary S5d). Next, we analyzed clonogenic survival in TDP1 and TDP1-complemented TDP1 cells. Strikingly, both mutants failed to rescue CPT sensitivity (Fig. 5d). This effect was limited to TDP1-deficient cells, since TDP1 and TDP1 expression in TDP1 cells did not promote any significant defect (Supplementary Fig. 8). We obtained similar results with reduced doxycycline concentrations, ruling out potential effects due to overexpression of TDP1 mutants (Fig. 5e). Importantly, RNA polymerase II elongation inhibition with DRB prior to TOP1 poisoning suppressed CPT sensitivity in EV, TDP1 and TDP1-expressing cells (Fig. 5e). Altogether, these results demonstrate that TDP1 and TDP1 promote toxicity by TOP1-induced DSBs associated with gene transcription in quiescent cells. Furthermore, the H263A mutation appears to be more deleterious than both TDP1 loss and the H493R mutation.