TASP1-mediated cleavage of REV3L enhances the activity of DNA polymerase ζ in mammalian cells Authors: Jordane Goulas, Basile Guignier, Barbara Ben Yamin, Caroline Pouvelle, Andrey Yurchenko, Emmanuelle Despras, Luciana de Oliveira, Pierre Laplante, Dorian Bochaton, Damien Plassard, Wei Yang, François-Xavier Theillet, Xavier Veaute, Richard D Wood, Sergey Nikolaev, Agnès M Cordonnier, Patricia L Kannouche Published: 2026 ABSTRACT AbstractDNA polymerase ζ (Polζ) is essential for replication across damaged DNA, yet the mechanisms governing its regulation in mammalian cells remain incompletely understood. The catalytic subunit REV3L is nearly twice the size of its yeast counterpart owing to a large vertebrate-specific insertion, and recent studies reported that REV3L undergoes proteolytic cleavage by the endopeptidase TASP1. Here, we show that REV3L cleavage generates two stable fragments, Nter-p70 and Cter-p300, which reassociate into a long-lived heterodimeric complex with enhanced polymerase activity. This interaction is stabilized by key residues within the catalytically inactive EXO domain. The cleaved Polζ complex associates with chromatin, binds nascent DNA, and undergoes ATR-dependent phosphorylation in response to replication stress. Importantly, preventing REV3L cleavage markedly impairs Polζ activity, reduces cisplatin-induced mutagenesis, and increases RAD51 nuclear foci and sister chromatid exchange, consistent with a compensatory shift toward homologous recombination–mediated repair. Together, our findings suggest that the vertebrate-specific expansion of REV3L necessitated cleavage to restore structural organization and optimize catalytic function, representing an evolutionary adaptation in DNA damage tolerance pathways.Subject terms: Translesion synthesis, Translesion synthesis Introduction Genomic instability is a double-edged sword, serving as both a source of genetic diversity that drives evolution and a major contributor to tumorigenesis. Understanding the mechanisms underlying mutation generation in eukaryotes is crucial for addressing human health concerns, including cancer. Mutations frequently result from assaults by endogenous and environmental agents that induce DNA damage1. While cells have evolved sophisticated repair systems to eliminate such damage, unresolved DNA lesions persisting in the genome during replication can cause replication blocks, fork collapse, double-strand breaks (DSBs), and genomic rearrangements2. To mitigate these outcomes, specialized DNA polymerases, including Y-family members (Polη, Polι, Polκ, and REV1) and the B-family DNA polymerase zeta (Polζ), enable cells to tolerate DNA damage. These polymerases perform translesion synthesis (TLS) by replicating across lesions or filling single-stranded gaps left post-replication. However, TLS comes at the cost of increased mutagenesis, underscoring its dual role in maintaining genome integrity while contributing to mutation burden3,4. In eukaryotic cells, Polζ plays a pivotal role in mutagenic TLS. Initially identified in yeast as a heterodimer5, Polζ is actually a four-subunit enzyme comprising the catalytic subunit Rev3 and the accessory subunits Rev7, Pol31/POLD2, and Pol32/POLD36–9. The latter two subunits are shared with the replicative polymerase Polδ. Despite being a member of the B-family DNA polymerases, Polζ lacks intrinsic 3'–5' exonuclease activity, making it inherently error-prone with a spontaneous mutation rate 10 to 100-fold higher than that of replicative DNA polymerases10–12. Polζ is essential for the bypass of most lesions, predominantly acting as an extender of aberrant primer termini at sites of DNA damage13–15. Accordingly, DNA damage-induced and spontaneous mutagenesis are significantly reduced in rev3Δ and rev7Δ strains16–19. In the yeast S. cerevisiae, Rev3 consists of 1504 amino acids. Although Rev3 is evolutionarily conserved from yeast to higher eukaryotes, its structure has significantly diverged in higher organisms. In humans and mice, the Rev3-like (REV3L) protein contains a large intermediate domain, making it twice the size of yeast Rev320. Unlike yeast REV3, which is not essential for viability21, disruption of the Rev3l gene results in lethality in mouse embryos and in primary cells22–24. This reveals an essential role for Polζ in supporting cell proliferation and maintaining genomic stability, even in the absence of exogenous DNA damage25. In line with this, Polζ also facilitates the replication of naturally occurring, hard-to-replicate DNA sequences in mammalian genomes26–28. However, it is important to note that although REV3L is indispensable for the proliferation of primary mammalian cells, certain immortalized or cancer-derived cell lines can tolerate REV3L loss29. Despite its well-established roles in lesion bypass, DNA replication, and repair, the mechanisms regulating Polζ in mammalian cells remain poorly understood. Investigations into mammalian REV3L protein are particularly challenging due to difficulties in expressing and detecting this large polypeptide (3130 residues). This is significant, as Polζ is a critical factor for tumor resistance to chemotherapeutic agents30,31, and represents a promising target for cancer therapy. Understanding how Polζ is regulated under normal conditions and in response to DNA damage is therefore crucial for advancing cancer biology and therapeutic strategies. Site-specific proteolysis plays a crucial role in regulating various physiological and cellular processes by either eliminating regulatory proteins when they are no longer needed or activating dormant proteins into their biologically active forms32. The endopeptidase threonine aspartase 1 (TASP1) is a 50-kDa protease belonging to a family of hydrolases characterized by an asparaginase 2 homology domain33. TASP1 was initially identified as the protease responsible for cleaving mixed-lineage leukemia 1 (MLL1) protein, a crucial step for the proper regulation of HOX gene expression34. Genuine TASP1 substrates include MLL1, MLL2 (KMT2B), TFIIAα−β (GTF2A1), ALFα−β (TFIIA-like factor, GTF2A1L), and all TASP1 substrates are translated as α–β precursors that undergo proteolysis to form mature α/β heterodimers, like TASP1 itself34. Interestingly, a TASP1 consensus motif within REV3L was identified by Bier and colleagues, suggesting that this atypically large DNA polymerase might also be regulated by proteolysis35. Subsequently, Wang and colleagues demonstrated that TASP1-dependent cleavage stabilizes Polζ by shielding REV3L from ubiquitin-mediated proteasomal degradation, thereby facilitating tolerance to bulky DNA lesions36. In a later study, the same group advanced a more elaborate model in which REV3L is maintained in an autoinhibited state; replication stress-induced ATR-dependent phosphorylation was shown to relieve this inhibition, concomitantly coupling catalytic activation to caspase-mediated turnover and thereby dynamically modulating TLS37. In this study, we further investigate the functional significance of TASP1-mediated REV3L cleavage using genetically engineered mouse embryonic fibroblasts (MEFs) and knock-in mouse models. We define the biochemical and structural interplay between the N- and C-terminal REV3L fragments and determine how their association promotes the assembly of an active Polζ holoenzyme. In addition, we assess how proteolytic processing regulates polymerase activity in vitro, chromatin recruitment, and cisplatin-induced, TLS-dependent mutagenesis in cells. Finally, we examine the physiological consequences of cleavage deficiency in vivo. Collectively, our findings reveal a mechanism whereby heterodimerization of REV3L products enhances Polζ functionality, providing further evidence that the vertebrate-specific expansion of REV3L evolve to preserve optimal catalytic activity. Results REV3L heterodimerization relies on a defined NTD–EXO interaction surface REV3L has been reported to be a substrate of TASP1, which cleaves the protein at a conserved QLDGTAD motif (residues 523–527), generating two polypeptides (Supplementary Fig. 1a)35,36. This motif closely resembles the TASP1 cleavage sites identified in canonical substrates, such as MLL1 and TFIIA, supporting a conserved proteolytic mechanism (Supplementary Fig. 1b). To confirm that this motif is both necessary and specific for TASP1-mediated maturation of REV3L, we generated an uncleavable (uc) REV3L mutant by altering the QLDGTAD sequence to QLAATAD fused to GFP (Supplementary Fig. 1c, top panel). Disruption of this recognition sequence completely abolished REV3L processing, leading to accumulation of noncleaved protein and loss of N-terminal cleavage products compared with the wild-type protein (Supplementary Fig. 1c, lower panel). Consistently, the C-terminal cleaved fragment was detected exclusively in cells expressing wild-type REV3L, but not in the uc mutant. These data validate that the QLDGTAD motif constitutes the bona fide TASP1 recognition site and that its integrity is strictly required for REV3L processing. Moreover, the detection of both N- and C-terminal fragments in the wild-type context indicates that TASP1-mediated proteolysis generates stable cleavage products rather than transient degradation intermediates. The TASP1 cleavage site is embedded within a sequence stretch that was acquired during vertebrate evolution. Comparative phylogenetic analysis revealed that the QLDGTAD motif is highly conserved across vertebrate REV3L orthologs, indicating that REV3L cleavage represents an evolutionarily conserved regulatory mechanism that emerged in higher eukaryotes (Supplementary Fig. 1d). Because all known TASP1 substrates are synthesized as α–β precursors that undergo proteolytic cleavage to yield mature α/β heterodimers, we investigated whether the N-terminal and C-terminal fragments of REV3L generated by TASP1 processing could similarly associate to form a heterodimer. To test this, we fused the N-terminal and C-terminal portions of REV3L with Flag- and GFP-tags, respectively (Fig. 1a, top panel). Using GFP pull-down assays, we observed that the N-terminal domain of REV3L1–525 interacts with the C-terminal fragment REV3L761–3130. However, no interaction was observed with the constructs REV3L761–1029, REV3L761–1876, or REV3L761–2001, indicating that the interaction region resides within the C-terminal fragment between residues 2001 and 3130 (Fig. 1a). To delineate more precisely the C-terminal region of REV3L required for interaction with its N-terminal fragment, we performed fine-mapping experiments (Fig. 1b, top panel). These analyses showed that the interaction depends on the integrity of the EXO-domain (residues 2291–2528) of REV3L: a fragment spanning residues 2296–3130 retained the ability to bind the N-terminal portion of REV3L, whereas a shorter fragment (2401–3130) lost this interaction (Fig. 1b). Consistent with this observation, the N-terminal domain (NTD) of yeast Rev3 has been reported to bridge the exonuclease and fingers domains38, suggesting that proteolytic cleavage of the large mammalian REV3L may be necessary to reproduce the structural organization observed in yeast Rev3. REV3L heterodimerization relies on a defined NTD–EXO interaction surface. a, b Schematic of REV3L constructs (top). HEK293T cells were co-transfected with Flag-REV3L1–525 and GFP-tagged C-terminal fragments or vice versa. Interactions were assessed by GFP-Trap and WB with anti-Flag and anti-GFP antibodies (representative from 2 independent experiments for (a, b). c Model of the REV3L N- and C-terminal fragment interaction predicted by AF2: structural model of the N-terminal fragment (REV3L residues 1–525) bound to the C-terminal fragment (REV3L residues 2289–3130). A close-up view highlights the critical interface residues K49 (N-terminal) and D2397, D2399, R2431 (C-terminal, within the EXO-domain), which are predicted to mediate stable contacts between the two fragments. A rotation along y-axis displays the N-terminal cleavage site (right panel). d HEK293T cells were co-transfected with plasmids expressing Flag-tagged C-terminal REV3L (residues 2296–3130) and GFP-tagged N-terminal REV3L (residues 1–525). GFP pull-down assays were performed with either wild-type or point-mutated constructs (K49A, D2397A-D2399A, and R2431A). Interactions were assessed by GFP-trap and WB with anti-Flag and anti-GFP antibodies (representative from 3 independent experiments). Source data are provided as a Source data file. We next sought to elucidate the structural determinants of the interaction driving REV3L heterodimer formation. Using AlphaFold2 (AF2)39, we predicted residues likely to contribute to this interface (Fig. 1c). The AF2 model highlighted two acidic residues (D2397 and D2399) and one basic residue (R2431) within the EXO- domain of REV3L, as well as a basic residue (K49) within the N-terminal domain, as potential mediators of charge–charge interactions at the binding surface. To evaluate their functional relevance, we performed site-directed mutagenesis, individually substituting each residue with alanine to disrupt side-chain charge and assess its contribution to heterodimerization (Fig. 1d). GFP pull-down assays revealed that the R2431A mutation completely abolished interaction with the N-terminal fragment of REV3L, as did the double D2397A/D2399A mutation. In contrast, the K49A mutant displayed only a partial reduction in binding to the C-terminal fragment of REV3L. Strikingly, substituting K49 with an acidic residue (K49D) resulted in complete loss of interaction with the C-terminal fragment (Supplementary Fig. 1e). As a negative control, mutation of a residue not predicted to contribute to the interface (A2390V) had no detectable effect on the interaction. These findings indicate that specific residues within both the EXO and N-terminal domains are essential for REV3L heterodimerization and suggest a defined and selective interface between the N- and C-terminal fragments. We then explored the regulation of native REV3L protein in mammalian cells. For this, we employed a knock-in MEF Flag-REV3L model, in which a triple-Flag epitope was inserted at the 5' end of the endogenous Rev3l locus40. This strategy enables specific detection of the native Flag-REV3L protein in mammalian cells27. Importantly, the knock-in did not alter Rev3l mRNA expression levels (Supplementary Fig. 1f). Immunoblot analysis of asynchronous MEF Flag-REV3L cell extracts using an anti-Flag antibody revealed, in addition to full-length REV3L (p360), a distinct ~70 kDa species. This band was markedly reduced upon Rev3l siRNA-mediated depletion and was absent in control MEFs lacking the Flag insertion (Supplementary Fig. 1g). Based on its molecular weight and TASP1 cleavage mapping, we interpret this species as the endogenous N-terminal cleavage fragment (p70) generated by TASP1-mediated proteolysis as previously reported in human cells36. If REV3L is physiologically processed by TASP1 into an active p70/p300 heterodimer, the accessory subunit REV7 should associate with this complex. Because the REV7 interaction motifs map to residues 1877–2003 within the C-terminal region of REV3L (Supplementary Fig. 1a)41, REV7 is expected to bind the p300 fragment. To determine whether the N-terminal fragment could be recovered in a REV7-associated complex, we performed co-immunoprecipitation using anti-REV7 antibodies in MEF Flag-REV3L cells. Both full-length REV3L (p360) and the cleaved N-terminal fragment (p70) co-immunoprecipitated with REV7 (Supplementary Fig. 1h), indicating that REV7 interacts with both uncleaved REV3L and the processed p70/p300 heterodimer. REV3L cleavage stabilizes p70 while p360 undergoes rapid turnover and UV-induced transcriptional inhibition To gain insight into the regulation of REV3L, we compared the steady-state stability of the full-length (p360) and cleaved (p70) species. MEF Flag-REV3L cells were treated with cycloheximide (CHX) to inhibit de novo protein synthesis, and protein levels were monitored over a 6-h time course (Fig. 2a, b). Full-length REV3L (p360) declined rapidly, with an apparent half-life of ~90 min, indicating a short steady-state persistence in MEFs. In contrast, the N-terminal cleavage fragment (p70) displayed markedly greater stability, with a half-life exceeding 8 h (Fig. 2b and Supplementary Fig. 2a). These results suggest that the rapid decline of p360 predominantly reflects active TASP1-mediated proteolytic processing. Thus, REV3L cleavage likely constitutes a regulatory maturation step required for the formation of a stable and functional p70/p300 heterodimeric complex. Transcriptional and post-translational regulation of REV3L following UV-induced genotoxic stress. a Protein stability of REV3L was assessed in MEF Flag-REV3L cells treated with cycloheximide (CHX, 10 μg/mL). WB with anti-Flag antibody detected REV3L p360, and anti-p125 antibody was used as loading control. Quantification and regression curves represent mean ± SD (n = 3 independent experiments). b Same as (a) to analyze REV3L p70 levels, n = 2 independent experiments. c Time-course analysis of REV3L following UV-C irradiation (20 J/m²) over 6 h (n = 4 independent experiments). d WB analysis of REV3L p360 levels in untreated cells (NT), UV-C-irradiated cells (20 J/m²), or aphidicolin-treated cells (APH, 3 μM, 4 h) (n = 2 independent experiments). Dose–response (e) and time-course (f) analyses of REV3L p360 following UV-C exposure (n = 3 and n = 2 independent experiments, respectively). g WB analysis of REV3L p70 after UV-C irradiation (20 J/m²), detected by anti-Flag WB; β-catenin served as loading control (n = 3 independent experiments). h Subcellular fractionation of whole-cell extracts (WCE) and chromatin-bound fractions (Chr) following UV-C exposure (20 J/m²). REV3L was detected by anti-Flag WB. Histone H4 and GAPDH were used as positive and negative controls for chromatin fractions, respectively (n = 2 independent experiments). i REV3L phosphorylation status was assessed by λ-phosphatase treatment of WCE and Chr fractions harvested 6 h post UV-C. REV3L p70 was analyzed by WB (n = 3 independent experiments). j REV3L response to replication stress and DNA damage was evaluated following 6 h treatment with aphidicolin (APH), hydroxyurea (HU), camptothecin (CPT), formaldehyde (FA), or UV-C (20 J/m²) (n = 2 independent experiments). k Cells pretreated with kinase inhibitors (30 min) were UV-C irradiated (10 J/m²) and analyzed 1 h later for REV3L p70 regulation (n = 2 independent experiments). l Co-immunoprecipitation of REV7 complexes from untreated, HU- or UV-treated cells, probed for REV7 and REV3L (n = 3 independent experiments). m iPOND assay schematic (top). EdU-labeled nascent DNA-associated proteins were purified from cells immediately after pulse, following thymidine chase, or after HU treatment. REV3L and REV7 association with replication forks was analyzed by WB; PCNA and histone H3 served as controls (n = 2 independent experiments). Source data are provided as a Source data file. Since Polζ is essential for resistance to various DNA-damaging agents, such as UV radiation, cisplatin, and aflatoxin29, we examined whether the amount of REV3L p360 is increased following genotoxic stress. Unexpectedly, p360 levels declined markedly within 4 h of UV irradiation (Fig. 2c). This decrease was not observed after treatment with aphidicolin, a DNA polymerase inhibitor (Fig. 2d), indicating that the effect is specific to UV-induced cellular damage rather than a consequence of DNA replication inhibition. The reduction in p360 was dose-dependent and reproducible, reaching a minimum at 6–8 h post-irradiation, before returning to basal levels by 16 h (Fig. 2e, f). Given the apparent short half-life of REV3L p360, it is likely that the decrease of REV3L p360 upon UV irradiation is linked to changes in Rev3l gene expression. Therefore, we measured Rev3l mRNA half-life in MEF Flag-REV3L cells using the Click-iT Nascent RNA Capture approach (Supplementary Fig. 2b). Linear regression analysis revealed a half-life of ~3.5 h (Supplementary Fig. 2c), shorter than the median mRNA half-life (~7 h) in embryonic stem cells42. This indicates that Rev3l transcription is tightly regulated in mammalian cells, making it sensitive to stimuli affecting transcriptional machinery. To test this hypothesis, we monitored Rev3l mRNA levels in MEF Flag-REV3L cells following UV irradiation (20 J/m²), as UV exposure is known to transiently inhibit transcription43. Consistent with protein-level observations, Rev3l mRNA levels reached a minimum at 6–8 h post-irradiation, and correlated with the UV dose (Supplementary Fig. 2d, e). Similar results were obtained in various human cell lines, including primary, transformed, and tumor cells (Supplementary Fig. 2f–h). Given that Polζ is critical for DNA damage tolerance and genome stability, these findings on REV3L p360 levels are counterintuitive. However, the long half-life of REV3L p70 prompted us to investigate how the cleaved REV3L responds to UV irradiation and other genotoxic stresses. REV3L p70 binds on chromatin at replication forks and undergoes ATR-dependent phosphorylation in response to replication stress We characterized the behavior of REV3L p70 following UV-induced genotoxic stress. Immunoblot analysis revealed the rapid appearance of slower-migrating p70 species within 1 h post-irradiation, persisting for more than 20 h (Supplementary Fig. 2i and Fig. 3g). Moreover, subcellular fractionation demonstrated that both basal and modified p70 species were predominantly enriched in the chromatin fraction (Fig. 2h), supporting an active role of REV3L p70 in DNA processing under normal and stressed conditions. Molecular and cellular characterization of sgTasp1 cells. a MEF Flag-REV3L sgTasp1 and sgRev3l cell lines were generated using CRISPR/Cas9. REV3L protein levels were analyzed by WB using anti-Flag antibody, while TASP1 inactivation was confirmed using anti-TASP1 antibody. p125 and β-actin served as loading controls (n = 2 independent experiments). b Protein stability of REV3L was assessed in WT and sgTasp1 clones (#32, #33) treated with cycloheximide (CHX, 10 μg/mL). Cells were harvested at indicated time points and analyzed by WB using anti-Flag antibody; p125 was used as loading control (n = 2 independent experiments). c Subcellular fractionation of WT and sgTasp1 clone #33 cells following no treatment (NT), hydroxyurea (HU, 2 mM, 2 h), or UV-C irradiation (20 J/m², 2 h). REV3L detection in whole-cell extracts (WCE) and chromatin (Chr) fractions was analyzed by WB. MCM2 and GAPDH were used as positive and negative controls for chromatin fractions, respectively (n = 2 independent experiments). d WT and sgTasp1 clone #33 cells were pretreated with ATR inhibitor (ATRi, 1 h) and subsequently exposed to HU (2 mM) for 2 h. Whole-cell extracts were treated ± λ-phosphatase prior to WB analysis. REV3L was detected using anti-Flag antibody, and β-catenin served as loading control (n = 2 independent experiments). e Cell survival following UV-C exposure was assessed in WT, sgRev3l clone #13, and sgTasp1 clones (#32, #33). Cells were stained with crystal violet 72 h post-treatment (n = 2 independent experiments). f Cell viability following cisplatin treatment (72 h) was measured in triplicates of WT, sgRev3l, and sgTasp1 clones (#32, #33) using CellTiter-Glo assay. Data were normalized to untreated controls. A representative experiment is shown with values corresponding to the mean of three technical replicates ± SD. n = 3 biological replicates. g Sensitivity to mitomycin C (MMC) was assessed similarly. Two independent datasets are presented: experiment #1 (WT, sgRev3l, sgTasp1 #33; n = 3 biological replicates) and experiment #2 (WT, sgTasp1 #32; n = 2). A representative experiment is shown with values corresponding to the mean of three technical replicates ± SD. Source data are provided as a Source data file. A previous report described UV-induced phosphorylation of the REV3L N-terminal fragment37. To validate this modification in our cellular model, extracts from UV-treated cells were incubated with λ-phosphatase prior to immunoblotting. Phosphatase treatment completely abolished the slower-migrating p70 species in both whole-cell and chromatin fractions (Fig. 2i), confirming that these forms correspond to phosphorylated REV3L p70. Importantly, p70 phosphorylation was not restricted to UV damage. Replication stress induced by aphidicolin (APH), hydroxyurea (HU), camptothecin (CPT), or formaldehyde (FA) similarly triggered distinct phosphorylation patterns (Fig. 2j), suggesting that REV3L p70 modifications are tailored to the type of DNA damage. Regarding REV3L p360 levels, a decrease was restricted to UV and FA treatment, likely due to global transcription inhibition impacting Rev3l mRNA levels. Pharmacological inhibition assays further showed that REV3L p70 phosphorylation is ATR-dependent, whereas inhibition of ATM, DNA-PK, CHK1/CHK2, or PLK1 had no detectable effect (Fig. 2k), in agreement with previous work37. Comparable ATR dependency was observed following HU and cisplatin treatment (Supplementary Fig. 2j), indicating that ATR represents a central upstream regulator of REV3L p70 phosphorylation across multiple forms of replication stress. Whether ATR directly phosphorylates p70 or acts through downstream effector kinases remains to be determined. To confirm that REV7 protein still forms a complex with REV3L p70 after replication stress, MEF Flag-REV3L cells were exposed to UV or HU treatment, then protein complexes enriched after REV7 immunoprecipitation were analyzed. Co-immunoprecipitation experiments revealed that REV7 remains associated with phosphorylated p70 after replication stress (Fig. 2I), supporting the hypothesis that the p70/p300 heterodimer and REV7 form an active Polζ complex. We previously revealed that Polζ contributes to heterochromatin replication during unperturbed S phase27, suggesting that this polymerase can be recruited to replication forks in the absence of DNA damage. To assess the biological relevance of the REV3L heterodimer, we investigated the association of cleaved REV3L p70 with replication forks using iPOND approach. In unstressed cells, REV3L p70 was retrieved immediately after thymidine pulse labeling but absent after a chase (Fig. 2m). This behavior is similar to that of the Polζ subunit REV7, as well as to one of the known replisome components, PCNA. This suggests that REV3L p70, likely with p300, and REV7, is engaged at replication forks during normal S phase. Following HU treatment, phosphorylated p70, together with REV7, was retained at replication forks, emphasizing its role in response to replication stress. As our experiments thus far had been conducted in SV40-transformed MEFs, we next sought to confirm that REV3L undergoes proteolytic cleavage in primary MEFs expressing Flag-REV3L. In these cells, the cleaved N-terminal fragment (p70) was detected as well as the full-length REV3L (p360) (Supplementary Fig. 3a). We also observed that REV3L p70 underwent phosphorylation in response to various replication stresses (Supplementary Fig. 3b, c). To assess REV3L expression and processing in vivo, we examined tissues from Flag-REV3L knock-in mice40. The cleaved REV3L p70 was detected in multiple organs, with particularly high levels in heart, ovary, and testis (Supplementary Fig. 3d). In contrast, REV3L p360 was exclusively detected in testis among all tissues analyzed, suggesting that Rev3l expression and/or stability may be particularly elevated in this tissue. REV3L full-length accumulates in TASP1-deficient cells To further dissect the functional impact of REV3L cleavage in our cellular system, we inactivated the Tasp1 gene in MEF Flag-REV3L cells using a CRISPR/Cas9-based strategy and established independent clones. In parallel, we generated Rev3l knock-out clones as control. Western blot analysis revealed the disappearance of the REV3L p70 fragment, accompanied by a marked accumulation of the full-length REV3L p360 in sgTasp1 clones (Fig. 3a). In contrast, both the p70 and p360 forms were completely absent in the sgRev3l clone. These results provide definitive evidence that REV3L undergoes physiological cleavage by TASP1, with p70 representing its N-terminal cleavage product. To confirm that the short apparent half-life of full-length REV3L (p360) reflects its active proteolytic processing by TASP1, we examined the half-life of REV3L p360 in two independent sgTasp1 clones by treating cells with cycloheximide (CHX) (Fig. 3b). In wild-type cells, p360 exhibited rapid turnover kinetics, whereas in sgTasp1 cells it was markedly stabilized, demonstrating that its apparent instability primarily results from continuous TASP1-mediated cleavage rather than proteasomal degradation per se. Thus, TASP1 activity is the dominant determinant of p360 turnover under basal conditions. Subcellular fractionation further revealed that REV3L p360 in sgTasp1 cells remains efficiently associated with the chromatin fraction and undergoes ATR-dependent phosphorylation upon replication stress (Fig. 3c, d). These results indicate that phosphorylation of the N-terminal region of REV3L occurs independently of proteolytic cleavage, while cleavage may define the functional context in which phosphorylated p70 operates. Disruption of REV3L cleavage does not compromise resistance to genotoxic stress or genome stability We next investigated the functional significance of REV3L cleavage by TASP1 in murine cells. Since previous studies demonstrated that loss of REV3L impairs proliferation of primary cells40,44, we first examined whether blocking its cleavage influences growth. In agreement with previous work36, proliferation assays in MEF Flag-REV3L lines revealed that sgTasp1 cells grow at rates comparable to WT cells, whereas sgRev3l cells proliferated more slowly (Supplementary Fig. 4a). These results indicate that the absence of TASP1, and thus the lack of REV3L cleavage, barely affect cell proliferation in this immortalized cell model. We then asked whether cleavage impairment alters cellular responses to genotoxic stress. Cells were exposed to increasing doses of UV radiation, and proliferation was measured 72 h later by crystal violet staining. Surprisingly, sgTasp1 cells displayed UV resistance comparable to WT cells (Fig. 3e), suggesting that the accumulation of full-length REV3L (p360) upon TASP1 inactivation is sufficient to protect against UV-induced damage. As expected, sgRev3l cells were hypersensitive to UV45. In line with these findings, sgTasp1 cells exhibited WT-like viability following cisplatin exposure (Fig. 3f) and no increased sensitivity to the DNA crosslinking agent mitomycin C (MMC), in sharp contrast to the pronounced hypersensitivity observed in sgRev3l cells (Fig. 3g). We next assessed the impact of REV3L cleavage on the DNA damage response (DDR). Following low-dose UV irradiation (2 J/m²), sgRev3l cells exhibited a pronounced and persistent activation of DDR markers as previously reported27, including strong phosphorylation of KAP1 and CHK1 lasting up to 24 h, and phosphorylation of RPA32 at serine 33, peaking at 6 h post-irradiation (Supplementary Fig. 4b). In contrast, sgTasp1 clones displayed DDR activation comparable to WT cells, even when exposed to higher UV doses (10 J/m²) (Supplementary Fig. 4c). Thus, unlike complete REV3L deficiency, the absence of REV3L cleavage does not trigger aberrant DDR signaling after genotoxic stress. We further analyzed cell cycle progression after UV exposure. Both WT and sgTasp1 cells transiently accumulated in S phase 6 h post-irradiation and returned to a normal distribution by 24 h (Supplementary Fig. 4d). In contrast, sgRev3l cells showed a sustained accumulation in S–G2 phases, highlighting the essential role of REV3L in response to UV-induced replication-blocking lesions. Because REV3L loss is known to cause spontaneous DSBs and chromosomal instability45, we next examined whether similar defects occur in cells lacking REV3L cleavage. Immunofluorescence analysis of the DSB marker 53BP1 revealed a significant accumulation of spontaneous foci in sgRev3l but not in sgTasp1 cells, which displayed levels comparable to WT. Following UV irradiation, sgRev3l cells exhibited a strong and persistent increase in 53BP1 foci, whereas both WT and sgTasp1 cells showed only a transient increase at 6 h that resolved by 24 h (Fig. 4a, b). Loss of REV3L cleavage does not compromise genome stability. a, b MEF Flag-REV3L WT, sgRev3l, and sgTasp1 cells were irradiated (2 J/m² UV-C) or left untreated, and 53BP1 foci were quantified by immunofluorescence (IF) at indicated time points. Representative images (scale bar = 10 μm) (a) and quantification (b) are shown. For each condition, >150 nuclei were analyzed. Each dot represents one nucleus; median and interquartile range (IQR) are indicated. Statistical analysis: Kruskal–Wallis with Dunn’s post hoc test. n = 3 biological replicates. WT and sgTasp1 cells treated ± cisplatin (2 μM, 24 h) were analyzed for RAD51 foci (c, d) and γH2AX intensity (e) by IF. Representative images (scale bar = 10 μm) (c) and quantifications (d, e) are shown. >250 nuclei per condition were analyzed. Each dot corresponds to a single nucleus; median and IQR are indicated. Statistical analysis: Kruskal–Wallis with Dunn’s post hoc test, n = 3. f–h Sister chromatid exchange (SCE) analysis was performed in MEFs labeled with BrdU (10 μM, 24 h), with or without UV-C (2 J/m²). Representative metaphases (f) show SCE events (red stars). Quantification of SCE frequency (per chromosome per metaphase) is shown for untreated cells (g, n = 2) and UV-treated cells (h, n = 1). Thirty metaphases were analyzed per condition. In g, dots represent individual metaphases, grey diamonds denote replicate means, and the black line indicates overall mean ± SD. In h, dots represent individual metaphases; mean ± SD is shown. Statistical analysis: unpaired two-sided t-test. i Micronuclei were quantified from experiments in (a) using DAPI staining (>150 cells per condition). Histograms represent the percentage of cells with ≥1 micronucleus. Black dots indicate means from four biological replicates; error bars show SD. Statistical analysis: one-way ANOVA with Tukey’s post hoc test. j Chromosomal instability was assessed by metaphase spreads following UV-C (2 J/m², 24 h) or no treatment (n = 1). Representative images show chromosomal breaks/gaps (red stars). Quantification (per chromosome per metaphase) was performed on 30 metaphases per condition. Mean ± SD is shown. Statistical analysis: Kruskal–Wallis with Dunn’s post hoc test. Source data are provided as a Source data file. We then investigated homologous recombination (HR) activity by quantifying RAD51 nuclear foci as a surrogate marker of ongoing recombination events. sgTasp1 cells displayed elevated spontaneous RAD51 foci compared to WT, with a further increase after cisplatin treatment (Fig. 4c–e). This was accompanied by a modest increase in γH2AX intensity, suggesting an enhanced HR activity when REV3L cleavage is impaired. To confirm that HR is functionally upregulated in this context, we performed sister chromatid exchange (SCE) assays, a cytogenetic readout of crossover events arising from HR-dependent repair (Fig. 4f). Quantification of spontaneous SCEs revealed a significant elevation in sgTasp1 cells relative to WT (Fig. 4g), supporting increased recombination activity under unchallenged conditions. Following UVC irradiation (2 J/m2), SCE frequencies increased in both cell lines, reflecting damage-induced HR activation; however, sgTasp1 cells consistently displayed a higher SCE burden, although the difference did not reach the same magnitude as under basal conditions (Fig. 4h). These data suggest that loss of REV3L cleavage shifts the balance of DNA damage tolerance toward enhanced HR engagement, both at steady state and following genotoxic stress. Finally, we assessed genomic instability by quantifying micronucleus formation. As expected, sgRev3l cells exhibited a marked fourfold increase in micronuclei under basal conditions, which was further exacerbated following UV irradiation (Fig. 4i). In contrast, sgTasp1 cells did not display a significant increase compared with WT cells. Consistently, cytogenetic analysis of metaphase spreads revealed comparable frequencies of chromosomal breaks and gaps in sgTasp1 and WT cells (Fig. 4j), indicating that loss of TASP1-mediated REV3L cleavage does not induce apparent chromosomal instability under these conditions. Altogether, these results demonstrate that inhibition of REV3L cleavage does not compromise MEF viability, cell cycle progression, or genome stability upon genotoxic stress. However, HR activity is elevated even under basal conditions in these cells. The absence of overt sensibility phenotype in sgTasp1 cells, therefore, likely reflects either the preserved functionality of the uncleaved REV3L protein or a compensatory upregulation of HR, which may buffer the cellular response to DNA damage and sustain resistance to genotoxic agents. Uncleaved REV3L is catalytically weakened and supports reduced DNA damage-induced mutagenesis An important question arising from this study is whether TASP1-mediated cleavage of REV3L affects DNA damage-induced mutagenesis. To address this, we tested whether the Polζ-dependent mutagenic TLS pathway remains functional in the absence of TASP1 by analyzing mutational profiles following cisplatin treatment. Cisplatin generates DNA lesions, including mono-adducts, intrastrand crosslinks, and interstrand crosslinks, that require Polζ for bypass and give rise to characteristic cisplatin-specific mutations30,46,47. MEF Flag-REV3L WT and sgTasp1 cells were exposed to four sequential cycles of cisplatin treatment (3 days on/4 days off, repeated 4 times) (Fig. 5a). Before and after treatment, single-cell clones were expanded, and DNA from four independent clones per condition was subjected to whole-genome sequencing. Loss of REV3L cleavage mainly prevents cisplatin-induced DBS mutations. a Experimental design. MEF Flag-REV3L WT and two independent sgTasp1 clones were subjected or not to four cisplatin treatment cycles at a dose of 0.8 μg/ml, a concentration corresponding to the IC20 for both cell lines, followed by clonal expansion and WGS. Created in BioRender. Kannouche, P. (2026) https://BioRender.com/64s4phb. b Representative mutational spectra of a WT clone before and after cisplatin. Single base substitutions (SBS, top) and double base substitutions (DBS, bottom) counts are shown. Quantification of SBS (c) and cisplatin-associated DBS (d) across clones. Means ± SD; statistical analysis was performed using ordinary one-way ANOVA followed by Turkey’s post-hoc test. e Heatmap showing the relative contribution of COSMIC SBS signatures in each treated group. As reported previously48, cisplatin mutational signatures are dominated by single-base substitutions (SBS), notably C>T (CCC>CTC, CCT>CTT), T>A (CT>CA), and C>A substitutions, and by an even stronger enrichment of double-base substitutions (DBS) in CC, CT, TC, and TG contexts. Consistent with this, parental MEF Flag-REV3L WT cells displayed cisplatin-induced mutational signatures, with a 3-fold increase in SBS and a 46-fold increase in DBS after treatment relative to untreated clones (Fig. 5b–d). Interestingly, sgTasp1 clones showed a trend toward reduced SBS levels after cisplatin, although the difference was not statistically significant (Fig. 5c). Focusing on DBSs which are highly specific for cisplatin mutagenesis, we observed that cisplatin-treated sgTasp1 cells displayed a 66.3% reduction compared to WT (Fig. 5d). Analysis of SBS mutational signatures revealed SBS8 across all samples (Fig. 5e and Supplementary Fig. 5), consistent with its widespread but poorly understood presence in multiple cancer types49. In contrast, the platinum-associated SBS35 was detected in WT and, to a lesser extent, sgTasp1 clones. Collectively, these findings revealed that TASP1-mediated cleavage of REV3L is required for the full efficiency of the mutagenic TLS pathway dependent on Polζ. To further examine the catalytic properties of REV3L, we performed biochemical assays comparing the DNA synthesis activity of cleaved versus uncleaved REV3L in complex with REV7. For these experiments, we used the “mini” human REV3L construct (TR5, 2358 amino acids) previously developed by Yang’s lab9. TR5 contains the four highly conserved domains of vertebrate REV3L—the N-terminal domain (NTD), the REV7-binding domain (R7B), the inactive exonuclease domain (EXO-), and the C-terminal polymerase domain (Pol)—as well as a positively charged domain (PCD), while excluding predicted disordered regions to improve expression (Fig. 6a). To generate an uncleaved form, we mutated the TASP1 cleavage site (DG525–526 → AA), yielding TR5-REV3LDG-AA. Both wild-type and mutant REV3L constructs were co-expressed with MBP-tagged REV7 in HEK293 cells, and the resulting REV3L–REV7 complexes (Polζ2) were partially purified by amylose affinity chromatography. Coomassie stain analysis revealed that the wild-type TR5 protein produced both uncleaved and cleaved fragments of REV3L, whereas the TR5-REV3LDG-AA mutant yielded only the uncleaved form (Fig. 6b). REV3L was detected by WB using specific REV3L antibody recognizing an epitope in N-terminal portion of REV3L (Fig. 6c). Because mammalian expression systems can assemble Polζ with accessory subunits, we tested for POLD2, POLD3, and SCAI by WB and detected their presence at comparable levels in both complexes (Fig. 6d). These findings suggest that Polζ naturally exists in cells as both cleaved and uncleaved complexes. Cleaved REV3L heterodimer exhibits higher catalytic activity than uncleaved REV3L. a Schematic representation of human REV3L and deletion construct TR5, aligned. Three predicted unstructured portions were deleted and replaced by two amino acids as indicated9. In the TR5DG-AA mutant, the TASP1 cleavage site is mutated into a non-cleavable sequence. Purification of Polζ2 complexes: SDS-PAGE stained with Coomassie (b) and WB for detecting REV3L using an anti-REV3L antibody against a REV3L epitope located in the N-terminal region (c) (n = 2 independent experiments). d Increasing amount of each Polζ2 complex were analyzed for REV3L partners by WB. e Measurement of initial velocities of dATP incorporation opposite a dT template using variable concentrations of dATP (from 0.2 to 100 μM). The values obtained for equimolar concentrations of wild-type (REV3L TR5) or mutant Polζ (REV3L TR5DG-AA) are shown (n = 3 independent experiments). Nonlinear regression analyses were performed in R using the drc package. Data were fitted to a Michaelis–Menten model, and parameters were estimated by maximum likelihood assuming normally distributed residuals. Standard errors, confidence intervals, and statistical significance were derived from Wald statistics. f Primer extension assay using a dilution series of wild-type or mutated Polζ2 (126, 63, 25.2, 12.6, and 6.3 ng/μl). Radiolabeled products were subjected to electrophoresis through an 8% denaturing polyacrylamide gel. The quantification of the full-length extension products and of the primer used are indicated below each lane (n = 2 independent experiments). Long exp: long exposure. Source data are provided as a Source data file. We next evaluated the catalytic activity of the partially purified Polζ2 complexes (Fig. 6e). At equimolar concentrations, wild-type Polζ2 displayed a Vmax approximately fourfold higher than that of the uncleaved mutant, demonstrating that TASP1 cleavage enhances enzymatic efficiency. In parallel, the mutant Polζ2 complex exhibits a significantly increased apparent Km for dNTP substrates compared with the wild-type enzyme. The concomitant increase in Km and decrease in Vmax indicate that REV3L cleavage contributes both to efficient dNTP binding and to optimal chemical catalysis. Primer extension assays corroborated this finding: substantially higher amounts of uncleaved Polζ2 were required to generate comparable levels of extended DNA products relative to cleaved Polζ2 (compare lane 3 obtained with 25.2 ng/ml of wild-type Polζ2 with lane 6 obtained with 126 ng/ml of mutant Polζ2; Fig. 6f). Using a relevant TLS substrate containing a site-specific cisplatin intrastrand crosslink (Pt-AG), we compared the catalytic properties of wild-type Polζ2 and the non-cleavable mutant Polζ2 complex (Supplementary Fig. 6a). Four distinct primer configurations were used to interrogate nucleotide incorporation opposite the lesion and extension beyond the adduct. Consistent with previous biochemical analyses9, the presence of the Pt-AG lesion significantly reduced the bypass efficiency of wild-type Polζ2, with a particularly marked decrease in nucleotide incorporation opposite the 3'G of the cross-link, which represents the most structurally constraining position of the adduct (Supplementary Fig. 6b). Comparison between wild-type and mutant Polζ2 revealed that cleavage deficiency compromises TLS efficiency across the Pt-AG lesion, both at the insertion step opposite the damaged base and during extension past the lesion. Collectively, these results establish that TASP1-dependent cleavage is critical for optimal Polζ catalytic function. Since we identified key residues mediating the interaction between the N-terminal and C-terminal fragments of REV3L (Fig. 1c, d), we next evaluated their functional relevance using the TR5-hREV3L construct. To monitor heterodimer formation from the precursor protein, we engineered a Flag tag at the N-terminus of the NTD and a Myc tag at the N-terminus of the cleaved C-terminal fragment (Supplementary Fig. 7). Site-directed mutagenesis was then performed to substitute arginine 2431 with alanine (R2431A) and assess its contribution to TR5-hREV3L heterodimerization. Flag and Myc pull-down assays revealed that the R2431A mutation markedly reduced the interaction between the cleaved fragments and concomitantly destabilized the heterodimer. In contrast, the uncleaved precursor protein remained as stable as the wild-type TR5-hREV3L. These findings suggest that disruption of the heterodimeric interface compromises the stability of the cleaved REV3L fragments. Loss of REV3L cleavage is associated with subtle embryonic developmental defects To investigate the physiological role of REV3L cleavage, we generated Rev3lDG-AA knock-in (KI) mice by CRISPR/Cas9-mediated editing of two-cell stage embryos, substituting the endogenous TASP1 cleavage site (QLDGTAD) with a non-cleavable variant (QLAATAD) (Fig. 7a). Targeted mutations were verified by Sanger sequencing in blastocysts, and intercrosses of Rev3lDG-AA/WT animals produced Rev3lDG-AA/DG-AA offspring at the expected Mendelian ratio (Supplementary Table 1). Mutant mice appeared grossly normal and showed no overt developmental abnormalities, indicating that REV3L cleavage is not strictly required for embryonic viability or postnatal development (Fig. 7b). Consistently, Rev3lDG-AA/KO mice, carrying one non-cleavable and one null allele, were viable. Lifespan monitoring over two years revealed no significant difference between wild-type and mutant cohorts, although a subset of Rev3lDG-AA/DG-AA animals exhibited a reduced longevity (Fig. 7c). In vivo consequences of disrupting REV3L cleavage. a Strategy for generating homozygous Rev3lDG-AA mice with non-cleavable REV3L using CRISPR/Cas9 approach in blastocysts. Right: Sanger sequencing showing point mutations. b Representative male mice of indicated genotypes (16–19 weeks old). c Kaplan–Meier survival curves of Rev3lWT/WT (n = 8), Rev3lWT /DG-AA (n = 11) and Rev3lDG-AA /DG-AA (n = 12) mice monitored for 2 years. d PCA of RNA-seq samples. Two-dimensional scatterplot of PC1 (53%) and PC2 (22%) showing biological replicates for Rev3lWT/WT (WT, blue) and Rev3lDG-AA/DG-AA (KI, orange) samples. Dashed lines indicate mean-centered axes. e Volcano plot showing upregulated (red) and downregulated (blue) genes identified by DESeq2, using a Wald test (two-sided) based on a negative binomial generalized linear model, with p-values adjusted by the Benjamini–Hochberg (BH) method. Genes with Padj < 0.01 are highlighted; grey points denote genes below this threshold. The top 20 DEGs are labeled by gene symbol. f GO biological process enrichment analysis of DEGs performed using one-sided hypergeometric test and Benjamini–Hochberg correction. Bar chart displays the top 20 enriched terms (Padj < 0.01) for upregulated (red) and downregulated (blue) genes. The x-axis represents fold enrichment; color intensity reflects –log10(Padj). Closer examination of reproductive outcomes, however, revealed defects in embryonic fitness. Only 50% of Rev3lDG-AA/DG-AA × Rev3lDG-AA/DG-AA crosses yielded litters, compared with a 100% success rate for Rev3lWT/WT × Rev3lWT/WT matings, which averaged of 6.5 and 7.5 pups per litter, respectively (Supplementary Tables 2 and 3). As expected, Rev3lWT/KO × Rev3lWT/KO crosses produced smaller litters (average 5.7 pups), consistent with the embryonic lethality of Rev3lKO/KO mice23. In contrast, Rev3lWT/DG-AA and Rev3lKO/DG-AA crosses generated eight litters with a drastic reduced average of only three pups. Another mating produced two litters in which all offspring were stillborn (Supplementary Tables 4 and 5). Similar outcomes were obtained in crosses between Rev3lWT/KO and Rev3l DG-AA/DG-AA animals (Supplementary Table 6). Together, these results reveal an unexpected developmental bottleneck: although REV3L cleavage is dispensable for gross viability under homeostatic conditions, its absence compromises reproductive success and embryonic fitness in specific genetic contexts. REV3L loss has previously been linked to transcriptional alterations in MEFs, notably within developmental and imprinting programs27. To assess whether disruption of REV3L cleavage produces similar deregulation, we isolated primary MEFs from WT (n = 3) and KI (n = 5) embryos at stage E13.5 and performed genome-wide RNA-seq profiling. Principal component analysis (PCA) revealed a clear segregation between WT and KI samples along PC1, which accounted for 53% of total variance, while PC2 explained an additional 22%, resulting in 75% cumulative variance (Fig. 7d). This unsupervised analysis indicates that genotype is the primary driver of transcriptional variability. Differential expression analysis using stringent criteria (adjusted p-value < 0.01, Benjamini–Hochberg correction; |log₂FC| > 0.5) identified a robust set of significantly deregulated genes, visualized in the volcano plot, highlighting the top 20 DEGs (Fig. 7e). Gene Ontology (GO) enrichment analysis of biological process categories revealed 81 significantly enriched terms. The top 12 enriched categories among upregulated and downregulated genes are shown in Fig. 7f and predominantly relate to developmental and morphogenetic programs. These signatures overlap with transcriptional profiles previously reported in Rev3l-/- MEFs27 and Tasp1-/- cells50, suggesting convergent pathway perturbations. Together, these findings suggest that impaired REV3L cleavage perturbs transcriptional networks important for embryogenesis, supporting a mechanistic link between cleavage-dependent Polζ regulation and developmental fitness. Discussion DNA polymerase ζ has long been recognized as a central player in DNA damage tolerance20, yet the regulatory mechanisms governing its catalytic subunit, REV3L, in mammalian cells have remained incompletely understood. Previous studies identified REV3L as a substrate of the endopeptidase TASP1, revealing that the protein undergoes site-specific proteolytic cleavage35,36. TASP1-dependent processing was proposed to protect REV3L from ubiquitin-mediated proteasomal degradation, thereby facilitating tolerance to bulky DNA lesions36. Here, we demonstrate that the TASP1-mediated cleavage also constitutes a fundamental regulatory mechanism that structurally and functionally reconfigures REV3L into a bipartite p70/p300 complex with enhanced stability and catalytic activity. This proteolytic maturation not only endows Polζ with increased mutagenic potential in response to cisplatin but also integrates with ATR-dependent phosphorylation pathways37 and transcriptional programs to establish a multilayered control system. These findings reveal that REV3L activity is dynamically regulated through an evolutionary innovation that fine-tunes the balance between DNA replication, tolerance to genotoxic stress, and mutagenesis. TASP1 cleavage activates REV3L by restoring structural organization Our biochemical and structural analyses indicate that the NTD of REV3L p70 directly interacts with the inactive EXO domain of REV3L p300, promoting the formation and stabilization of an active heterodimeric complex. Disruption of this interface through targeted mutagenesis abolishes heterodimerization and compromises heterodimer stability, underscoring the functional importance of this interaction. Interestingly, the critical residues within the EXO-domain (D2397, D2399, and R2431) that mediate this binding are conserved in yeast Rev3, highlighting the evolutionary conservation of this structural module. This observation is consistent with a general conserved feature of B-family DNA polymerases51,52. Our biochemical analyses also revealed that heterodimerization of the cleaved REV3L products markedly enhances the catalytic efficiency of the Polζ complex. In contrast, the uncleaved mutant exhibits both a reduced Vmax and an increased apparent Km for dNTPs, indicating impaired substrate binding and decreased catalytic turnover. Consistently, primer extension assays show that substantially higher amounts of the uncleaved Polζ2 complex are required to achieve comparable DNA synthesis. This catalytic defect also extends to translesion synthesis across a cisplatin-induced Pt-AG intrastrand crosslink, where cleavage deficiency reduces both nucleotide insertion opposite the lesion and extension beyond it. Thus, the catalytic impairment of uncleaved REV3L affects not only basal DNA synthesis but also translesion synthesis across a clinically relevant DNA lesion. The requirement for TASP1-mediated cleavage appears to be a distinctive feature of higher eukaryotes. Compared with yeast Rev3, mammalian REV3L has nearly doubled in size, primarily due to the insertion of a vertebrate-specific exon that separates the NTD from the polymerase core29. This large additional sequence (∼1386 amino acids) may have arisen through retrotransposition in a common vertebrate ancestor, introducing substantial folding constraints that could destabilize the architecture of the full-length protein and necessitate proteolytic processing to restore functional organization. We therefore propose that TASP1-mediated cleavage evolved as an adaptive mechanism to relieve these constraints, thereby promoting proper folding and optimal enzymatic activity of Polζ (see hypothetical model in Supplementary Fig. 8). Crosslinking mass spectrometry (XL-MS) would provide a powerful experimental strategy to directly test this model. Applied to REV3L, XL-MS could define long-range intramolecular contacts between the N- and C-terminal regions and determine whether TASP1-mediated proteolysis induces conformational rearrangements that reorganize domain architecture and stabilize the active polymerase complex. Consistent with this hypothesis, the strong conservation of the TASP1 recognition motif across vertebrate species suggests that proteolytic processing of REV3L represents an evolutionary adaptation that enabled the maintenance of Polζ structural integrity and catalytic efficiency during vertebrate evolution. REV3L levels are tightly regulated at transcriptional and protein levels REV3L regulation extends beyond proteolysis. We showed that Rev3l mRNA has a relatively short half-life (~3.5 h), and the full-length REV3L p360 protein is an apparent short half-life (~90 min), in contrast to the higher stability of the cleaved N-terminal fragment p70. Consequently, REV3L p360 levels are highly sensitive to transcriptional perturbations, as exemplified by UV or formaldehyde exposure. These findings indicate that cellular resistance to genotoxic stress relies predominantly on the stable p70/p300 heterodimer rather than on uncleaved REV3L. Moreover, the short half-life of REV3L p360 provides a plausible explanation for the strikingly low abundance of full-length REV3L in mammalian cells and suggests that fluctuations in transcriptional output can impact Polζ availability. Such regulation has potential clinical relevance. CDK9 inhibitors and other agents targeting transcriptional elongation53 could indirectly downregulate REV3L expression, and sustained suppression may exacerbate chromosomal instability, given the essential role of Polζ in safeguarding genome integrity45. Post-translational control integrates Polζ into the DNA damage response We further demonstrate that ATR-dependent phosphorylation of REV3L p70 is context-dependent and varies according to the nature of the genotoxic insult. Phosphorylated REV3L p70 is enriched on chromatin and associated with nascent DNA, indicating that this post-translational modification occurs at active replication forks under conditions of replication stress. These observations support a model in which ATR-mediated phosphorylation dynamically regulates Polζ function in situ, potentially by reshaping its interaction network or stabilizing its engagement at stalled forks. Although the catalytic core resides within the REV3L p300 product, phosphorylation of the N-terminal p70 fragment may influence higher-order complex assembly, protein–protein interactions, or chromatin retention. Thus, TASP1-dependent proteolytic maturation and ATR-dependent phosphorylation appear to cooperate in fine-tuning Polζ activity. This work, therefore, establishes an additional regulatory layer connecting ATR checkpoint signaling to TLS control, consistent with previous studies linking DNA damage response pathways to the modulation of translesion synthesis37,54–57. Functional consequences: reduced mutagenesis but preserved survival Surprisingly, in our cellular system, abrogation of TASP1-mediated REV3L cleavage does not result in hypersensitivity to cisplatin or UV irradiation. Instead, TASP1-deficient cells retain survival capacity but display a 2- to 3-fold reduction in cisplatin-induced mutagenesis. This phenotype likely reflects the diminished catalytic efficiency of uncleaved REV3L, as observed in vitro, coupled with a compensatory engagement of HR, consistent with the increased accumulation of RAD51 foci and SCE levels. A plausible scenario is that cisplatin-induced fork stalling enhances repriming events, thereby generating excessive single-stranded DNA gaps behind replication forks58. Such gaps can be processed either by TLS or HR59,60. In sgTasp1 cells, where uncleaved REV3L is less proficient than the cleaved heterodimer in gap filling, the balance of pathway usage may shift toward HR, thereby limiting mutagenesis without compromising overall cell survival. Thus, while Polζ cleavage is dispensable for short-term survival, it is critical for mutagenic TLS. In contrast to our findings, Wang and colleagues reported that uncleaved REV3L is unstable and targeted for proteasomal degradation36. They further observed that cellular responses to UV- and cisplatin-induced DNA lesions are markedly impaired in cells expressing a cleavage-defective REV3L mutant. These discrepancies may arise from differences in the cellular models employed. While our study was conducted in MEFs, their experiments were primarily performed in mismatch repair–deficient HCT116 cells, a context characterized by hypermutation and compromised DNA repair networks61,62. In such a genetic background, the reliance on REV3L function may be accentuated, and the absence of TASP1-mediated cleavage could have more severe consequences on DNA damage tolerance than in a repair-proficient setting. Our in vivo analyses further revealed that, although REV3L cleavage is dispensable for gross viability under homeostatic conditions, its absence compromises reproductive success and embryonic fitness in specific genetic contexts. Consistent with this, transcriptomic analyses pointed to deregulation of developmental and morphogenetic programs, in line with previous reports in Rev3l-/- MEFs27 and Tasp1-/- mice50. These findings suggest that TASP1-mediated cleavage fine-tunes Polζ activity, possibly to accommodate the high demands of genome duplication during embryogenesis, with incomplete compensation by alternative repair pathways. Polζ at replication forks: roles beyond lesion bypass A key finding is that cleaved REV3L fragments, together with REV7, localize at replication forks even in unchallenged cells. This observation is consistent with Polζ’s contribution to heterochromatin replication27 and spontaneous mutagenesis63. Moreover, REV3L phosphorylation and fork enrichment following HU suggest that Polζ supports replication under nucleotide depletion, aligning with yeast studies that revealed Polζ-mediated synthesis at HU-stalled forks64, contributing to increased mutagenesis65. Recent work implicating Polζ in replication through AT-rich hairpins28 and centromeric DNA sequences66 further broadens its role in managing difficult templates. Collectively, these findings indicate that Polζ is not restricted to lesion bypass but contributes to the replication of challenging genomic regions, expanding its mutagenic potential. Therapeutic implications of targeting REV3L cleavage Our findings raise the intriguing possibility of exploiting TASP1-mediated REV3L cleavage as a therapeutic vulnerability. Inhibiting cleavage greatly reduces Polζ mutagenesis without compromising cell viability, suggesting that TASP1 inhibitors could reduce chemotherapy-induced resistance while preserving short-term DNA damage tolerance. TASP1 inhibitors are already under development67. Combined with DNA-damaging chemotherapy, such agents may deliver a “dual benefit”: enhanced cytotoxicity and reduced tumor adaptability. In conclusion, our study identifies heterodimerization of the cleaved REV3L products as a key regulatory mechanism that enhances the catalytic competence of the mammalian Polζ complex. This cleavage integrates with transcriptional regulation and ATR-dependent phosphorylation37 to fine-tune Polζ activity. By enabling Polζ to balance its dual roles in lesion bypass, replication of difficult templates, and mutagenesis, this multilayered control ensures genome stability while preserving adaptability. Evolutionarily, TASP1 cleavage represents a structural innovation necessitated by the expansion of REV3L in vertebrates. Therapeutically, modulating this cleavage may offer a new avenue to reduce mutagenesis and chemoresistance in cancer during therapy. Methods Cell culture SV40-transformed MEF Flag-REV3L27,40, human osteosarcoma cell line U2OS, and primary MEF Flag-REV3L and SV40-transformed MEF wt were cultivated in D-MEM (Dulbecco’s Modified Eagle medium-Gibco) containing 10% FBS, 1% penicillin-streptomycin. HEK293T human cell line was cultivated in D-MEM containing 10% FBS and 1% penicillin-streptomycin, 1% sodium pyruvate, and 1% non-essential amino acids. Primary human fibroblasts (405-VI) and SV40-immortalized MRC5-VI were cultivated in MEM containing 10% FBS 1% penicillin-streptomycin. FreeStyle 293-F cells were cultivated in Gibco FreeStyle medium (ThermoFisher, cat. #11550426) and 1% penicillin-streptomycin in suspension. All cultures were maintained at 37 °C in a humidified atmosphere with 5% CO2, except the FreeStyle 293-F cells, which were incubated in 8% CO2. Cells were regularly tested for Mycoplasma contamination using MycoStrip kit (Invivogen, cat. #rep-mysnc-100). Drug treatments For ultraviolet-C irradiation (254 nm), cells were plated 24 h prior treatment. Before treatment, the culture medium was removed, and cells were washed once with PBS and then exposed to a certain dose (specified for each experiment) without any liquid. After irradiation, cells were grown in culture medium at 37 °C for indicated time and then scraped for protein or RNA extraction. For treatment using chemical agents, cells were plated 24 h prior treatment, then treated either with 3 μM aphidicolin (APH), 1–2 mM hydroxyurea (HU), 1 μM camptothecin (CPT), 200 μM formaldehyde (FA), or 6 µM cisplatin. After treatment, cells were grown at 37 °C for indicated time, then harvested for protein extraction. For determining protein half-life, MEF Flag-REV3L were treated with cycloheximide (CHX) at the final concentration of 10 μg/ml, and cells were harvested at different time points. All chemical products were purchased from Sigma-Aldrich. Cell extracts and Western blot Cells were lysed with NET-N buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1% NP-40, protease and phosphatase inhibitors-Roche, cat. #04906845001) for 30 min at 4 °C. After sonication, protein extracts were centrifuged and kept in −20 °C or loaded on SDS-PAGE gel for Western blot analysis. For cell fractionation, cells were incubated with CSK100 buffer (100 mM NaCl, 300 mM sucrose, 3 mM MgCl2, 10 mM pipes, pH 6, 8, 1 mM EGTA, pH 8, 0.2% Triton) + protease inhibitor cocktail (Roche, cat. #11836170001) for 5 min incubation on ice, then centrifuged at 3000 × g for 5 min at 4 °C. Supernatant was kept as soluble fraction, then pellet was resuspended in lysis buffer (10 mM Tris-HCl, pH 7.5, 20 mM NaCl, 0.4% SDS, protease inhibitor cocktail) supplemented with 150 U/ml benzonase (Millipore) at room temperature for 10 min. For Western blot analysis, proteins were denatured in Laemmli buffer and resolved by SDS-PAGE. Proteins were then transferred on PVDF membrane, which was blocked in 5% milk in PBS-1% Tween and incubated with primary antibodies overnight at 4 °C. Following washing and secondary antibody incubation, protein detection was carried out using ECL reagents (Thermo Scientific). Lambda phosphatase treatment Whole-cell extracts were prepared using NET-N buffer as described above, then protein extracts were treated with lambda phosphatase according to the following instructions: lysates were incubated with 500 U of lambda phosphatase in NEB buffer supplemented with 1 mM MnCl2 (NEB, cat. #P0753) for 30 min at room temperature. Dephosphorylated proteins were then analyzed by Western blot. Treatment with kinase inhibitors Cells were pre-treated with the following kinase inhibitors: 250 nM CHIR-124 (CHK1 inhibitor, MCE cat. #HY-13263), 25 µM PV-1019 (CHK2 inhibitor, MCE cat. #HY-125203), 10 μM KU-55933 (ATM inhibitor, MCE cat.#HY-12016), 1 µM AZD6738 (ATR inhibitor, MCE cat. #HY-19323), 10 μM NU7441 (DNA-PK inhibitor, MCE cat. #HY-11006), 600 nM CDK1/2 inhibitor III (MCE cat. #HY-112462), 10 µM XL413 (CDC7 inhibitor, MCE cat. #HY-15260) or 200 nM BI2536 (PLK1 inhibitor, MCE cat. #HY-50698) independently for 30 min at 37 °C before treatment (10 J/m² UVC, 2 mM HU or 3 µM cisplatin). Cells were then harvested at the indicated time, and protein extracts were analyzed by Western blot. Co-immunoprecipitation, GFP-trap, and Western blot For immunoprecipitation, cells were lysed in NETN buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.5% NP-40, protease inhibitor cocktail, and anti-phosphatases) on ice with gentle shaking. After sonication, protein extracts were centrifuged, and 10% of the extracts was used as input, and the remaining suspension was incubated overnight at 4 °C with 2 μg anti-Mad2L2/REV7 antibody (Abcam, cat. #ab180579) for REV7 co-immunoprecipitation. Samples were incubated 2 h with Protein G Sepharose beads (Cytiva, cat. # 17-0618-01) at 4 °C, then immunoprecipited proteins with beads were extensively washed in NET-N buffer before to be resuspended in Laemmli buffer, boiled, and analyzed by Western blot. For GFP-trap, Flag-trap, or Myc-trap, 25 μl of specific Nanobody/VHH coupled to agarose beads (Proteintech, cat. #gta for GFP; cat. #ffa for Flag and #yta for Myc) were used to pull-down protein complexes. After washing, immunoprecipitated proteins were resuspended in Laemmli buffer and boiled before being loaded. Antibodies α-GFP (ab290 Abcam 1:2000); α-Flag-M2 (F1804 Sigma-Aldrich 1:2000); αRev3 Nter (GTX02533 GeneTex 1:500); α-MAD2L2 (ab180579 Abcam 1:1000); α-TASP1 (Proteintech 1:500); α-p-KAP1(S824) (A300-767A-M Bethyl 1:1000); α-KAP1 (ab10483 Abcam 1:2000); α-p-CHK1(S345) (2348 Cell Signaling 1:1000); α-CHK1 (sc8408 Santa Cruz 1:1000); α-p-RPA32(S33) (A300-264A Bethyl 1:5000); α-RPA32 (A300-244A Bethyl 1:5000); α-p125 (sc8797 Santa Cruz 1:2000); α-actin (A5441 Sigma-Aldrich 1:10000); α-β-catenin (610153 BD Transduct 1:10000); α-MCM2 (ab4461 Abcam 1:5000); α-H4 (93269 Active Motif 1:1000); α-GAPDH (ab9484 Abcam 1:2000). Secondary antibodies (goat polyclonal anti-mouse HRP, goat polyclonal anti-rabbit HRP, donkey polyclonal anti-goat HRP; DAKO) were used at 1:5000 and incubated 45 min at RT before being revealed using ECL Pierce substrate. Immunofluorescence Cells were plated on poly-D-lysine-coated coverslips 24 h before treatment. Following treatment, cells were collected at various time points. The coverslips were first washed twice with cold PBS, then pre-extracted with CSK100 buffer (100 mM NaCl, 300 mM sucrose, 3 mM MgCl2, 10 mM pipes pH 6.8, 1 mM EGTA pH 8, 0.2% Triton) containing a protease inhibitor cocktail, to remove soluble proteins then cells were fixed with 4% paraformaldehyde for 20 min, permeabilized with 0.5% Triton X-100 for 5 min, and washed three times for 5 min each with PBS. Blocking was performed with PBS containing 3% BSA for 30 min at room temperature. Coverslips were incubated overnight at 4 °C or 1 h at room temperature with primary antibody (α-53BP1 (ab21083 Abcam 1:500); α-RAD51 (PC130 Millipore 1:1000); α-γH2AX (05-636 Millipore 1:500). After three washes of 5 min with PBS, cells were incubated with a secondary antibody conjugated to Alexa Fluor 488 or 594 (Life Technologies) diluted 1:1000 for 30 min at room temperature. The coverslips were then washed three more times with PBS, and nuclei were counterstained with DAPI-containing Vectashield. Imaging was performed on a Zeiss Imager Z1 epifluorescence microscope. Image quantification and processing were done using FIJI software on more than 150 cells, and the number of micronucleated cells was manually counted on DAPI images. Statistical analyses were performed using Prism v9.0 (GraphPad). Kruskal–Wallis with post-hoc Dunn test were used for comparing foci numbers. Cell proliferation assay Cells were seeded in triplicates in 96-well plates, and the plate was placed in an Incucyte® incubator (Sartorius S3), set to capture bright-field images of the wells every 6 h for 120 h. Then analyzing mask parameters were adapted to detect the cells and determine well confluency. Standard deviation is calculated by the Incucyte software. Results were plotted using Prism v9.0 (GraphPad). Cell viability assays For drug sensitivity assays, cells were seeded in triplicate in 96-well plates, 24 h prior to addition of the drug. After 72 h of cell growth the luminescent CellTiter-Glo® 2.0 Assay (Promega) was used according to the manufacturer’s instructions, for determining the number of viable cells in each well by quantifying ATP, which indicates the presence of metabolically active cells. The values were normalized with non-treated to obtain relative cell viability in %. All further statistical analysis were performed using Prism v9.0 (GraphPad). To visualize UV sensitivity, the crystal violet assay was used. Cells were seeded in 6-well plates 24 h prior irradiation, and then allowed to grow for 72 h. Plates were washed twice with PBS before the addition of crystal violet solution (0,5%) for 20 min incubation. Flow cytometry Cells were seeded in 60 mm dishes and irradiated 24 h later with 2 J/m², or left untreated. They were then harvested 6 or 24 h after UV exposure, with non-irradiated cells being retrieved at the 6-h time point. The media was collected into a polystyrene round-bottom Falcon tube, and the dish was washed once with PBS, also collected. The cells were trypsinized and retrieved into the tube, which was centrifuged at 300 × g for 5 min at 4 °C. The cell pellet was washed with PBS, followed by another centrifugation and removal of PBS. For fixation, the cell pellet was resuspended in 500 µL cold PBS, and 2 mL of 100% ice-cold ethanol were added dropwise while continuously vortexing. The samples remained at least one night at −20 °C. On the day of analysis, cells were centrifugated at 805 × g for 5 min at 4 °C before being filtered and resuspended in Propidium Iodide (PI) staining solution (PI diluted in PBS at a final concentration of 25 µg/mL, and RNase A at 50 µg/mL) for 30 min incubation at room temperature. Then each sample was processed through a CytoFLEX (Beckman Coulter) device, and data were analyzed using FlowJo v10.9.0 software. Sister chromatin exchange Cells were plated in 100-mm dishes and, 24 h later, irradiated with UV-C (2 J/m²) or left untreated before incubation with BrdU (10 µM). After 20 h, colcemid (10 µg/ml) was added for 4 h. Cells were harvested, subjected to hypotonic treatment (75 mM KCl, 20 min, 37 °C), and fixed in methanol:acetic acid (3:1) overnight at −20 °C. Metaphase spreads were prepared on glass slides, incubated overnight at 65 °C, stained with Hoechst 33258 (20 µg/ml), exposed to UV-A for 25 min in 2× SSC, and incubated in 2× SSC at 60 °C for 20 min before Giemsa staining. Images were acquired using a Zeiss Observer microscope (63× oil objective). Thirty metaphases per condition were analyzed using Fiji (RRID: SCR_002285) to quantify SCE per metaphase. Metaphase spread–Giemsa staining Cells were plated in 100 mm dishes. Twenty-four hours later, cells were irradiated or not with UV-C (2 J/m²) and then incubated in fresh media. After 20 h, colcemid (10 µg/ml) was added for 4 h to arrest cells in metaphase. Cells were harvested, subjected to hypotonic treatment (75 mM KCl, 20 min, 37 °C), and fixed in methanol:acetic acid (3:1) overnight at −20 °C. Metaphase spreads were prepared on glass slides, air-dried, and stained with Giemsa (1:20 in PBS; Sigma). Images were acquired using a Zeiss Observer microscope (63× oil objective). Thirty metaphases per condition were analyzed with Fiji (RRID: SCR_002285). iPOND MEF Flag-REV3L cells were pulse-labeled with 10 μM EdU (Invitrogen) for 10 min. Labeling was either terminated immediately (pulse condition) or followed by a 1-h chase in fresh medium supplemented with 10 μM thymidine (Sigma). For HU condition, 3 mM HU was added for 1 h after the initial 10-min EdU labeling period. Cells were crosslinked with 1% formaldehyde (Sigma) in PBS for 15 min at room temperature under gentle agitation, and crosslinking was quenched with 125 mM glycine for 5 min. Cells were harvested by scraping, washed with ice-cold PBS, and permeabilized in PBS containing 0.5% Triton X-100 for 30 min at room temperature. EdU was conjugated to biotin-azide (Molecular Probes) via click chemistry for 2 h in reaction buffer (10 mM sodium ascorbate, 10 μM biotin-azide, 2 mM CuSO4 in PBS). Cells were lysed in iPOND buffer (10 mM HEPES-NaOH, pH 7.9, 100 mM NaCl, 2 mM EDTA, 1 mM EGTA, 1 mM PMSF, 0.2% SDS, 0.1% sarkosyl, protease inhibitors) and sonicated using a Bioruptor (30 cycles, 30 s on/30 s off, high power). Solubilized chromatin was clarified by centrifugation (16,000 × g, 10 min), and supernatants were incubated overnight with streptavidin-coated magnetic beads (Dynabeads MyOne Streptavidin C1, Invitrogen). Beads were washed sequentially in lysis buffer, 500 mM NaCl, and lysis buffer, and bound proteins were eluted by boiling in 1× Laemmli buffer at 95 °C for 30 min. Purification of Polζ2 WT and mutant pLEXm vectors expressing TR5-hREV3L wt or mutant (DG-AA) with His8-MBP-tagged REV7 were transiently co-transfected at a ratio 20:1 in FreeStyle 293-F cells in suspension (800 ml) using polyethylenimine (PEI, Polysciences). Forty-eight hours later, cells were centrifuged, washed twice with cold PBS, and cell pellets were immediately frozen in liquid nitrogen before purification being proceeded. Polζ2 complexes were purified by using affinity pull-down, ion-exchange, and size-exclusion chromatography as previously described9. DNA synthesis assay DNA polymerase activities were assayed at 37 °C using the reaction mixture containing 20 mM Tris (pH 8), 20 mM KCl, 50 mM NaCl, 5 mM MgCl2, 0.5 mM DTT, 3% glycerol, dATP (variable concentration) and α−32P-dATP (0.2 μM), 220 nM annealed DNA (dT44/dA25), and 29 ng/μl or 126 ng/μl of wild-type or mutant Polζ, respectively. Fractions aliquots of the reactions were acid-precipitated at 2, 3, and 5 min of incubation. The quantification of dATP incorporation into acid-insoluble product was determined using the ImageQuantTM TL software after phosphorimaging (GE Healthcare). The Vmax and Km were determined by fitting the reaction rate over dNTP concentrations in R using DRC library and the Michaelis–Menten equation68. For the primer extension assay, the 61/30 mer template was obtained by annealing the 5' end radiolabeled 30 mer primer (5'-TCACACAGGCAGGACAGCTTAGACCATGAT-3') with the 61 mer (5'-GGGTACCGAGCTCCAGCAACTAGTTCGTAATCATGGTCTAAGCTGTCCTGCCTGTGTGAAA-3'). The reaction mixture (5 μl) containing 60 fmoles of primed template DNA and different amount of Polζ2 was incubated at 37 °C in 20 mM Tris (pH 8), 100 mM NaCl, 5 mM MgCl2, 0.5 mM DTT, 100 μM of dNTPs, 3% glycerol. At the indicated timepoints, reactions were stopped by addition of 1 vol of stop solution (95% formamide, 20 mM EDTA, 0.025% Xylene cyanol, 0.025% bromophenol blue), heated to 95 °C for 5 min, and analyzed on an 8% denaturing polyacrylamide gel. The quantification of the labeled products was determined using the ImageQuantTM TL software after phosphorimaging (GE Healthcare). The percentage of extended product was calculated as the ratio of the intensity of the bands of elongation products to the sum of the intensity of the products and primer bands. Preparation of cisplatin-adducted DNA substrates The 61-nt oligonucleotide template strand (5'-GGGTACCGAGCTCCAGCAACTAGTTCGTAATCATGGTCTAAGCTGTCCTGCCTGTGTGAAA-3') containing a 1,2-intrastrand d(AG)-cisplatin crosslink (Pt-AG) as indicated by the underlined bold letters was generated. Radiolabeled primer strands for insertion steps (P1 or P2) or extension steps (P3 or P4) (Supplementary Fig. 6a) were annealed to either the undamaged or cisplatin-adducted template strand and processed as described above. Long term mutagenesis experiment One population from each cell line underwent four cycles of cisplatin treatment, each cycle consisting of 3 days with cisplatin diluted in the cell medium and 4 days of drug-free medium. MEF Flag-REV3L WT and sgTasp1 clones #32 and #33 were treated with 0.8 μg/ml of cisplatin, a concentration corresponding to the IC20 for both cell lines. Immediately before or after the long-term treatment, cells from each population were plated at low density to generate independent clones. Several clones were expanded, and for each clone a cell pellet was retrieved to carry out DNA extraction with the NucleoSpin Tissue (Macherey-Nagel) kit, according to the manufacturer’s instructions. For each cell line, 4 non-treated (NT) and 4 treated (T) clones were sent for Whole Genome Sequencing. WGS analysis The murine genomes were sequenced in BGI, Shenzhen, China, with BGISEQ-500 sequencer according to the manufacturer's protocols (BGI Tech Solutions, Hong Kong, Co., Ltd). The reads were mapped to the mm9 mouse reference genome using BWA-MEM software69, and then the standard GATK best practice pipeline was used70 to process the samples and call somatic mutations. PCR duplicates were removed, and base quality scores were recalibrated using MarkDuplicates and BaseRecalibrator tools (GATK package;71. Quality control of FASTQ and mapping was done with samtools72, GATK HSmetrics, and multiqc73. All the processing steps were combined into a pipeline built with snakemake74. The called somatic mutations with the PASS flag from GATK Mutect2 were additionally filtered to be supported by at least 2 reads from each strand and at least 5 reads in total, and have variant allele frequency higher than 0.1. To remove possible leaking germline variants or common nonunique mutations accumulating during the experiments, we removed all mutations present in more than one sample with bcftools isec software75. Mutational matrixes were generated separately for SBS and other types of mutations using SigProfilerMatrixGenerator76. The contribution of the COSMIC mutational signatures was assessed using MutationalPatterns package77 with strict refitting (max_delta = 0.004). RNA extraction for RT-qPCR Total RNA extraction was performed using Maxwell RSC simply RNA cell kit (Promega) according to the manufacturer’s instructions. One μg RNA was reverse transcribed to cDNA using RevertAid H minus Reverse Transcriptase (Thermo Scientific, Cat. #EP051) according to the according to the manufacturer’s instructions. RT-qPCR was conducted using the CFX96 Real-time system (Biorad) with 1 μl cDNA in Maxima SYBR Green/Rox qPCR Mastermix (Thermo Scientific, Cat. #K0221). Quantities of Rev3l mRNA was calculated by dCt method with CFX Maestro software and normalized to GAPDH levels. The following primers were used for qPCR (Eurofins Genomics): mGAPDH: F-AGTGCCAGCCTCGTCCCGTAGACAAAATG and R-AAGTGGGCCCCGGCCTTCTCCAT; mRev3L: F-ACAGCGGAGAAGAAACAGGA and R-ACTCTTGGGATCCATTGCTG; hRev3L: F-ACCCCCGGAGTACCACTTATCCAGC and R-CCGGAGATATGGTGCCTTGGACA. hGAPDH: F-ACCACAGTCCATGCCATCAC and R-TCCACCACCCTGTTGCTGTA. Click-iT nascent RNA capture MEF Flag-REV3L incubated for 24 h with 0.1 mM of 5'-ethynyluridine (EU) in culture medium were washed three times with PBS to stop the incorporation, then incubated in fresh culture medium for indicated times. Cells were harvested, and total RNA extraction was carried out as described above. EU-RNAs were biotinylated by Click reaction, and biotinylated RNAs were immunoprecipitated using Dynabeads MyOne Streptavidin T1 magnetic beads according to the manufacturer’s instructions (Click-It Nascent RNA Capture kit, Molecular Probes). qRT-PCR was performed as described above. Plasmid constructs and siRNA transfection The human Rev3l full-length cDNA was kindly provided by C. Lawrence (University of Rochester Medical Center, Rochester, NY, USA). To generate the series of GFP-REV3L plasmids, constructs were obtained by polymerase chain reaction (PCR) amplification from Rev3l cDNA as a XhoI–EcoRI fragment and cloned into peGFP-C3 expression vector (Clontech). Alternatively, Rev3l cDNA fragments were cloned into pFLAG-CMV-4 (Addgene, #E7158) using EcoRI/KpnI restriction sites. pLEXM-TR5-hREV5L and pLEX-His-MBP-REV7 have been previously described9. Point mutants were generated by PCR-based methods using a QuikChange® site-directed mutagenesis kit (Agilent, cat. #200518), according to the manufacturer’s instructions. All constructs were verified by DNA sequencing. Plasmids were transfected using jetPEI (Polyplus), according to the manufacturer’s instructions, and cells were processed 24–48 h later. For siRNA transfection, cells were transfected with 10 nM of siRNAs using Interferin (Polyplus) according to the manufacturer’s instructions, and were processed 72 h later. All siRNAs were purchased from Dharmacon (Horizon Discovery): siRNA mouse Rev3l (ON-TARGETplus SMART pool): GAAAUGGAGUGGACCGAUA; AGUCAAUGAUGUUGCGUAU; CGACAUAUCCCGUGUGUUU; ACCCAGAAGUUGUAUAAU, siRNA mouse TASP1 (ON-TARGETplus Smart pool): CUUAGAUGGGUGAGCGGAA; CUGAAGAGAAGACGACAAU; GAGAAUGUUCACACGCUUU; CGAGCUUGUCAGAAGGCAA, non-target siRNA (ON-TARGETplus non targeting siRNA •1): UGGUUUACAUGUCGACUAA. Generation of MEF Flag-REV3L TASP1 knock-out using CRISPR/Cas9 technique Single guide RNA (sgRNA) was designed using appropriate software (IDT) and two single-stranded complementary oligos (Eurofins Genomics): F-CACCGGCCCTGTAAGGAGAAACGAG and R-AAACCTCGTTTCTCCTTACAGGGCC were annealed, cloned into pX458 plasmid (Addgene, #48138). Plasmid was then electroporated in MEF Flag-REV3L using MEF Nucleofection kit II (Lonza), then 48 h later, GFP-positive cells were sorted by FACS and plated at low density to generate independent clones, which were expanded and assessed for TASP1 loss by western blot and Sanger sequencing. Similar strategy was used to knock-out mRev3l in MEF Flag-REV3L using the following sgRNA TCTGCAGTTCCATCTAACTG AGG. Generation of Rev3lDG-AA/DG-AA mouse model Rev3lDG-AA/DG-AA mice were generated at the LEAT Transgenesis Platform (Imagine Institute, Paris). A guide RNA targeting exon 13 of the murine Rev3l gene, corresponding to the nucleotide sequence encoding the TASP1 cleavage site, was selected for CRISPR/Cas9 editing. A single-stranded oligodeoxynucleotide (ssODN; IDT) was designed as a repair template to introduce point mutations abolishing the TASP1 cleavage site (underlined nucleotides indicate mutated nucleotides): C*G*AAGAAGATGACTCGTCCTCAGAAGAGGAAATGGAGTGGACCGATAATAGTTTGCTTTTTGCCAATCTCTCTATACCTCAGTTAGCTGCAACTGCAGATGAAAATAGTGGTAAGAAAACAAACCTATTGTATTTGGTTATTTAGTGACTCAAAATGTATTTTATTTGTGACATATTT*G*T. Zygotes at pronuclear stage were microinjected with the CRISPR/Cas9 components, including the Rev3l sgRNA, ssODN repair template, and purified Cas9 protein, following the protocol of Mashiko et al.78 with minor modifications. Genotyping of F0 animals was performed by PCR amplification of the targeted locus followed by Sanger sequencing. Founder mice were crossed to C57BL/6J to segregate the CRISPR/Cas9-generated alleles. To eliminate potential off-target effects, each line was subsequently backcrossed for up to five generations onto the C57BL/6J background. For genetic crosses, Rev3lflag/flag knock-in mice44 and heterozygous Rev3l+/- mice23, were bred with Rev3lDG-AA/DG-AA mice as required. All experimental procedures were conducted in accordance with French government and institutional animal welfare regulations, and approved by the French Animal Experimentation Ethics Committee No. 26 (protocol #02005.02, APAFIS #33215-202109241556796 v3). Every effort was made to minimize animal suffering and to reduce the number of animals used. Sex was not considered in the study design. Isolation of primary MEFs from mouse embryos Primary MEFs from either Rev3lWT/WT, Rev3lDG-AA/DG-AA, or Flag-REV3L embryos at stage E13.5 were isolated as previously described79, then MEFs were cultured for approximately 10 passages, during which the experiments were performed. Protein extraction from mouse organs Organs were harvested from euthanized mice and flash-freezed in liquid nitrogen, then conserved at −80 °C. For protein extraction, small samples of each organ were placed in specialized tubes designed for tissue homogenization (Ozyme Lysing Kit M (Mix), 1.4 & 2.8 mm ceramic beads). T-PER tissue protein extraction reagent (Thermo Scientific) was added to each sample, which were then homogenized using PreCellys 24 (Ozyme) device (3000 g, 2 × 15 s, break of 10 s), under constant dry ice refrigeration. Tubes were then centrifuged for 10 min at 12,000 × g at 4 °C to eliminate organ debris, and processed for Western blot analysis. RNA extraction, library preparation, RNA sequencing, and analysis Total RNA was extracted from primary MEF using Qiagen column (RNeasy kit) according to the manufacturer’s protocol. RNA-seq libraries were prepared from 350 to 500 ng total RNA using the Illumina Stranded Total RNA Prep with Ribo-Zero Plus kit and IDT for Illumina RNA UD Indexes (Illumina), following the manufacturer’s protocol. DNA libraries were amplified using 13 cycles of PCR, and excess primers were removed by two successive purifications with SPRIselect beads (Beckman Coulter). Library quality and size distribution were assessed using a Bioanalyzer 2100 (Agilent Technologies). Sequencing was performed as 100-bp paired-end reads on an Illumina NextSeq 2000. Image processing and base calling were conducted using RTA (v4.12.2) and BCL Convert (v4.2.7). Reads were aligned to the Mus musculus reference genome (mm10) using STAR v2.5.3a80, and only uniquely mapped reads were retained. Differential expression analysis was performed in R (v4.4.3) using DESeq2 (v1.46.0)81, which normalized read counts from the HTseq tool82 and modeled gene expression using a negative binomial distribution81. Differentially expressed genes (DEGs) were defined by an adjusted p-value (P_adj) < 0.01 and Log_2FC > 0.5. Functional enrichment analyses were performed using clusterProfiler (v4.14.0) with gene annotations from org.Mm.eg.db (v3.20.0) mapped to the mm39 (GRCm39) genome build. Statistical analysis Statistical analyses were performed using GraphPad Prism 9-10 software. Data are presented as either mean ± SD or median with IQR (as indicated in the figure legends). The significance of difference in protein intensity or foci number assessed by immunofluorescence was determined using Kruskal–Wallis test, followed by post-hoc Dunn’s multiple comparisons test. The significance of mRNA abundance assessed by qPCR was determined using a two-way ANOVA, followed by post-hoc Turkey’s multiple comparisons test. A p < 0.05 was considered as significant. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer Review File Source data Source Data Acknowledgements The authors thank Patrick Gonin, Karine Ser-Le Roux, and Mélanie Polrot from the Preclinical Evaluation Platform (PFEP) for their valuable contribution. We are grateful to the Gustave Roussy core facilities for technical support, the LEAT transgenesis platform (Imagine, Paris) for having generated the Rev3lDG-AA mice, and to the GenomEast Platform, a member of the France Génomique consortium (ANR-10-INBS-0009), for RNA sequencing. For particular constructs, plasmid cloning and protein purification were carried out at the CIGEX core facility (CEA, Fontenay-aux-Roses), whose assistance is gratefully acknowledged. We also thank O. Gavet (UMR9019, Gustave Roussy) for insightful advice on kinase inhibitors and reagents, as well as all members of the Kannouche team for stimulating discussions. This work was supported by Fondation ARC (Programme labellisé, ARCPGA2019120001055_1578) and Institut National du Cancer (INCa, PLBio 2019-152). J.G. was supported by La Ligue Nationale Contre le Cancer and Fondation pour la Recherche Médicale. B.B.Y., P.L., and B.G. received support from the Ministère de l’Enseignement Supérieur et de la Recherche. E.D. was supported by INCa (PLBio 2019-152). Author contributions P.L.K. conceived and supervised the study. P.L.K., A.Y., and S.N. designed the experimental strategy for the analysis of the mutational landscape in cell lines. C.P. and D.B. generated and characterized cell lines derived from MEF Flag-REV3L WT. C.P. also performed several molecular and cellular experiments. E.D. carried out iPOND experiments. B.B.Y. initiated the project, and J.G. performed an extensive set of experiments. B.G. also conducted a large set of experiments, generated structural models, provided expertise for immunofluorescence analyses, and contributed actively to the editing of the revised manuscript. W.Y. and R.D.W. contributed technical guidance on Polζ complex purification and MEF Flag-REV3L systems, respectively. F.X.T. and X.V. carried out partial purification of Polζ₂ complexes. Bioinformatic analyses of the mutational landscape were performed by A.Y. and S.N., while RNA-seq analyses were conducted by D.P., P.L., and L.d.O. A.M.C. performed primer extension assays and additional biochemical experiments. J.G. and P.L.K. wrote the manuscript, and all authors reviewed, edited, and approved the final version. Peer review Peer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Data availability RNA-seq and WGS datasets for this study have been deposited in the European Nucleotide Archive (ENA) at EMBL-EBI under accession number PRJEB110854. Cell lines and reagents used in this study are available from the corresponding author upon request. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Jordane Goulas, Basile Guignier, Barbara Ben Yamin. Supplementary information The online version contains Supplementary material available at 10.1038/s41467-026-74271-x. References Associated Data Supplementary Materials Supplementary Information Reporting Summary Transparent Peer Review File Source Data Data Availability Statement RNA-seq and WGS datasets for this study have been deposited in the European Nucleotide Archive (ENA) at EMBL-EBI under accession number PRJEB110854. Cell lines and reagents used in this study are available from the corresponding author upon request. Source data are provided with this paper.