Genotoxic stress triggers the activation of IRE1α-dependent RNA decay to modulate the DNA damage response

The molecular connections between homeostatic systems that maintain both genome integrity and proteostasis are poorly understood. Here we identify the selective activation of the unfolded protein response transducer IRE1α under genotoxic stress to modulate repair programs and sustain cell survival. DNA damage engages IRE1α signaling in the absence of an endoplasmic reticulum (ER) stress signature, leading to the exclusive activation of regulated IRE1α-dependent decay (RIDD) without activating its canonical output mediated by the transcription factor XBP1. IRE1α endoribonuclease activity controls the stability of mRNAs involved in the DNA damage response, impacting DNA repair, cell cycle arrest and apoptosis. The activation of the c-Abl kinase by DNA damage triggers the oligomerization of IRE1α to catalyze RIDD. The protective role of IRE1α under genotoxic stress is conserved in fly and mouse. Altogether, our results uncover an important intersection between the molecular pathways that sustain genome stability and proteostasis.

T he integrity of the genome is constantly threatened by endogenously produced toxic metabolites, physical, and chemical insults, resulting in a variety of DNA lesions. Inefficient DNA repair translates into cellular dysfunction and death, but also into the propagation of somatic mutations and malignant transformation. To limit genome instability, cells engage the DNA damage response (DDR) and activate repair mechanisms to reverse or minimize alterations in DNA integrity 1 . The DDR pathway involves the interconnection of complex signaling networks that enforce cell cycle arrest and DNA repair. The failure of this adaptive mechanism is detrimental for the cell, resulting in an irreversible cell cycle arrest (senescence) or the activation of different types of regulated death programs 1 . Accordingly, perturbations in the DDR largely contribute to oncogenesis, tumor progression, and the resistance to irradiation and chemotherapy with genotoxic drugs. The accumulation of synonymous mutations, aneuploidy, as well as the activation of oncogenes, deregulate proteostasis 2 . The endoplasmic reticulum (ER) is the main subcellular compartment involved in protein folding and quality control 3 , representing a central node of the proteostasis network. The unfolded protein response (UPR) is a specialized mechanism to cope with ER stress 4,5 , that also influences most hallmarks of cancer 6 . Nevertheless, the possible involvement of the UPR in the surveillance and maintenance of genome integrity remains elusive.
Inositol requiring enzyme 1 alpha (known as ERN1, referred to as IRE1α hereafter) controls the most evolutionary conserved UPR signaling branch, regulating ER proteostasis and cell survival through distinct functional outputs 4 . IRE1α is a serine/threonine protein kinase and endoribonuclease that catalyzes the unconventional splicing of the mRNA encoding X-Box binding protein 1 (Xbp1), generating an active transcription factor that enforces adaptive programs 7 . IRE1α also degrades a subset of mRNAs and microRNAs through a process known as regulated IRE1αdependent decay of RNA (RIDD), impacting various biological processes, including cell death and inflammation [8][9][10][11] . A screen aiming to define the universe of XBP1-target genes under ER stress identified a cluster of DDR-related components 12 , and suboptimal DNA repair may trigger ER stress 2 . Together, these observations suggest a link between DNA damage and ER proteostasis. Here we investigate the possible contribution of IRE1α to the DDR. Surprisingly, we observed that genotoxic stress engages IRE1α signaling in the absence of ER stress markers. In fibroblasts undergoing DNA damage, IRE1α activation results in the selective activation of RIDD in the absence of XBP1 mRNA splicing, impacting genome stability, cell survival and cell cycle control. At the molecular level, we identify specific RIDD mRNA substrates as possible effectors of the phenotypes triggered by IRE1α deficiency. We also validated the significance of IRE1α signaling to the DDR in vivo using genetic manipulation in mouse and fly models. Our results suggest that IRE1α has an alternative function in cells undergoing genotoxic stress, where it serves to amplify and sustain an efficient DDR to maintain genome stability and cell survival.

Results
DNA damage selectively induces IRE1α signaling toward RIDD. Upon ER stress, IRE1α dimerization leads to its autotransphosphorylation and the formation of large clusters that are needed for optimal signaling 13 . Exposure of mouse embryonic fibroblasts (MEF) to the DNA damaging agent etoposide, a topoisomerase II inhibitor, triggers mild IRE1α phosphorylation (Fig. 1a) and formation of IRE1α clusters, as revealed using an IRE1α-GFP reporter (Fig. 1b). Similar results were obtained in cells exposed to γ-irradiation ( Supplementary Fig. 1a). Unexpectedly, MEF cells stimulated with etoposide or γ-irradiation failed to engage Xbp1 mRNA splicing, as determined by two independent PCR-based assays (Fig. 1c, d) or western blot analysis ( Supplementary Fig. 1b). Moreover, no signs of ER stress were observed in cells undergoing DNA damage when we assessed canonical markers of UPR activation, including the expression of CHOP, ATF4, BiP, as well as ATF6 processing and the phosphorylation of both PERK and eIF2α ( Supplementary Fig. 1c, d). As positive controls of DNA damage, we monitored the levels of phosphorylation of the histone H2AX (γ-H2AX) or the upregulation of the cyclin-dependent kinase inhibitor CDKN1A (also known as p21) ( Supplementary Fig. 1b, d). Unexpectedly, classical RIDD mRNAs substrates such as Bloc1s1 and Sparc 8,9 decayed upon exposure to DNA damaging agents (Fig. 1e). Importantly, this decrease in Bloc1s1 and Sparc mRNAs did not occur in IRE1α-deficient cells (Fig. 1e), nor upon pharmacological inhibition of the RNase activity of IRE1α with MKC-8866 ( Supplementary Fig. 1e, f), confirming the occurrence of RIDD. These results suggest that DNA damage selectively stimulates IRE1α activity toward RIDD and not Xbp1 mRNA splicing in the absence of global ER stress markers.
IRE1α regulates DDR signaling under genotoxic stress. To evaluate the significance of IRE1α expression to the adaptive capacity of cells undergoing DNA damage, we compared the viability of IRE1α-deficient and control (wild type, WT) cells after exposure to various agents that induce distinct types of DNA lesions, including etoposide, 5-hydroxyurea, 5-fluorouracil, and γ-irradiation. Remarkably, IRE1α deficiency sensitized cells to all types of genotoxic stress, thus increasing the incidence of cell death ( Fig. 1f and Supplementary Fig. 2a). We confirmed these results by measuring caspase-3 activation, a marker of apoptosis (Fig. 1g). We then stably reconstituted IRE1α null cells with an HA-tagged version of IRE1α (IRE1α-HA) that expresses levels similar to endogenous (described in refs. 14,15 ). Importantly, the hypersensitivity of IRE1α deficient cells to DNA damage was partially reverted by expressing IRE1α-HA, suggesting that the phenotypes observed in IRE1α-deficient cells under genotoxic stress are a primary phenotype, and are not due to clonal effects or compensatory changes ( Supplementary Fig. 2b). In sharp contrast, IRE1α knockout MEFs did not reveal any differential sensitivity to the ER stress inducer tunicamycin, in line with prior results based on pharmacological inhibition of IRE1α 16 . Also, the pharmacological inhibition of IRE1α RNase activity with MKC-8866 increases the susceptibility to cell death after prolonged genotoxic treatments (Fig. 1h, i). These findings highlight the importance of IRE1α in the adaptation response to DNA damage.
The downstream effects of the DDR are mediated by the activity of check point kinases CHK1 and CHK2, engaging the tumor suppressor protein P53 to induce cell-cycle arrest and the transcription of DNA damage-responsive genes, or to trigger apoptotic cell death 17 . DDR signaling translates into the phosphorylation of histone H2AX (γ-H2AX), and the rate of γ-H2AX decay after DNA injury is a sign of DNA repair. Thus, we monitored the kinetics of H2AX (de)phosphorylation after exposing cells to a pulse of etoposide. IRE1α null cells exhibited a lower response and faster attenuation of γ-H2AX phosphorylation compared to cells in which IRE1α was reintroduced (Fig. 2a). Similar results were obtained when cells where exposed to a shorter pulse of a higher concentration of etoposide ( Supplementary Fig. 2c). Analysis of nuclear γ-H2AX foci by immunofluorescence and confocal microscopy also indicated reduced γ-H2AX phosphorylation in IRE1α deficient cells (Fig. 2b). To corroborate these results, we performed the comet assay to directly visualize the DNA damage observing that IRE1α knockout cells exposed to etoposide exhibited increased DNA damage (Fig. 2c). We also performed a cytokinesis-block micronucleus cytome (CBMN) assay as an indirect measure of DNA breaks. Cells were incubated with etoposide for 3 h, followed by the administration of cytochalasin-B (CytB) for an additional 24 h. IRE1α deficiency resulted in a higher percentage of cells with binucleated nuclei, micronuclei or nuclear buds (Fig. 2d). Altogether, these results suggest that IRE1α null cells have reduced engagement of DNA repair programs.
A direct consequence of DNA damage is cell cycle arrest. No significant differences were observed in cell proliferation ( Supplementary Fig. 3a) or cell cycle progression between control and IRE1α null cells in normal conditions. Nevertheless, the fraction of cells arrested in the S and G 2 /M phases was reduced in IRE1α knockout MEFs exposed to etoposide (Fig. 2e). Similar results were obtained when we reintroduced IRE1α in knockout cells ( Supplementary Fig. 3b). At the molecular level, the ablation of IRE1α expression resulted in a strong attenuation in the phosphorylation of CHK1 and CHK2 in DNA-damaged cells (Fig. 2f). In contrast, the phosphorylation of the apical kinase ataxia telangiectasia mutated kinase (ATM; an upstream sensor of the DDR) was not altered in IRE1α knockout cells exposed to etoposide ( Fig. 2f and Supplementary Fig. 3c). Consistent with these results, IRE1α knockout and WT MEF cells showed similar phosphorylation of SMC1 18 and KAP1 19 , two direct substrates of ATM ( Supplementary Fig. 3d), suggesting that the defects triggered by IRE1α deficiency occur downstream of ATM activation regulating CHK1 and CHK2 activation. Moreover, IRE1α deficient cells showed a sustained kinetic of activation of P53 ( Supplementary Fig. 3e). Taken together, these results suggest that IRE1α deficiency deregulates DDR signaling, thus affecting cell cycle progression, DNA repair and cell survival under genotoxic stress.
Genotoxic stress triggers RIDD to regulate DDR signaling. To obtain mechanistic insights, we attempted to identify RIDD target mRNAs that might connect IRE1α signaling to the DDR. Previously, a global in vitro screening uncovered a cluster of mRNAs containing consensus sequences cleaved by IRE1α that are associated to a stem-loop structurally similar to the Xbp1 mRNA splicing site 20 . Among the 13 top hits, two DDR-related genes were identified as possible RIDD substrates: PPP2CA-scaffolding A subunit (Ppp2r1a) and RuvB like AAA ATPase1 (Ruvbl1) mRNAs 21 (Fig. 3a). PPP2R1A encodes the scaffold A subunit of protein phosphatase 2 catalytic subunit alpha (PPP2CA, also known as PP2A), which dephosphorylates check point kinases, reversing the G2/M arrest, and directly catalyzing the decay of γ-H2AX phosphorylation and foci 22 . RUVBL1 (also known as Pontin), participates in chromatin remodeling and modulates the stability of DDR protein complexes, thus influencing the dephosphorylation of γ-H2AX 23 . All panels data is shown as mean ± s.e.m.; *p < 0.05, **p < 0.01, and ***p < 0.001, based on b two-tailed unpaired t-Student's test, (c, d). One-way ANOVA followed Tukey's test (e-g, i), two-way ANOVA followed Bonferroni's test. Data is provided as a Source Data file. Quantification of Ppp2r1a and Ruvbl1 mRNA levels in cells treated with etoposide demonstrated a decay that was dependent on IRE1α expression (Fig. 3b). These effects on mRNA levels translated into reduced protein expression of PP2A and RUVBL1 only in wild-type cells exposed to etoposide and the basal upregulation in IRE1α null cells (Fig. 3c). In a cell-free assay, recombinant IRE1α directly cleaves a fragment of the Ppp2r1a mRNA that contains the RIDD consensus site (spanning nucleotides 1336-1865), but not an adjacent fragment (Fig. 3d). Similarly, IRE1α exhibited RNase activity on Ruvbl1 mRNA, thus cleaving this substrate as efficiently as it's known targets Bloc1s1 mRNA and Xbp1 mRNA (Fig. 3d). This reaction was suppressed by the IRE1α inhibitor 4µ8C (Fig. 3d).
The lack of Xbp1 mRNA splicing under DNA damage conditions might involve inhibitory signals, for example mediated by the downregulation of the tRNA ligase RTCB, the targeting of the Xbp1 mRNA to the ER membrane, or the activity of other regulatory components that are part of IRE1α clusters and component associated with them 24 . Analysis of RTCB levels revealed no changes in IRE1a knockout cells undergoing DNA damage ( Supplementary Fig. 4a). To test if DNA damage inhibits Xbp1 mRNA splicing, we pre-treated cells with tunicamycin for 2 h and then added etoposide at different time points. Remarkably, etoposide failed to interfere with Xbp1 mRNA splicing induced by tunicamycin (Fig. 3e). Virtually identical results were obtained when a pulse of etoposide was performed followed by the stimulation of ER stress (Fig. 3g). In contrast, an additive effect was observed on the decay of Bloc1s1 and Sparc mRNAs when ER stress and DNA damaging agents were combined (Fig. 3f, h). These results indicate that DNA damage selectively engages RIDD yet does not cause active suppression of Xbp1 mRNA splicing.
Considering that PP2A and Pontin are upregulated in IRE1α null cells under genotoxic stress and are involved in the DDR, we attempted to reverse the phenotype of those cells by depleting Ppp2r1a or Ruvbl1 mRNA with suitable short hairpin RNAs (shRNAs) (Supplementary Fig. 4b). Remarkably, knocking down Ppp2r1a or Ruvbl1 in IRE1α null cells augmented the levels of phosphorylated γ-H2AX foci after etoposide treatment during the recovery phase (Fig. 3i). Similar results were obtained when phosphorylated γ-H2AX was monitored by immunoblot (Supplementary Fig. 4c). Moreover, knocking-down Ppp2r1a and Ruvbl1 expression in IRE1α null cells reestablished normal levels of CHK1 phosphorylation ( Fig. 3j and Supplementary Fig. 4d, e) and increased population of cells in S/G2M after etoposide treatment ( Supplementary Fig. 4f). Taken together, these experiments suggest that the regulation of PP2A and Pontin by IRE1α contributes to DDR signaling under genotoxic treatment. . In all panels, data is shown as mean ± s.e.m.; *p < 0.05, **p < 0.01, and ***p < 0.001, based on (a, c, d, f) two-way ANOVA followed Bonferroni's test, b two-way ANOVA followed Tukey's test. Data is provided as a Source Data file.
Eto Tm We then explored possible signaling events that may link the DDR with the induction of RIDD. Interestingly, a recent report indicated that the non-receptor c-Abl tyrosine kinase physically interacts with IRE1α under metabolic stress, allosterically inducing its oligomerization into a conformation that is more likely to catalyze RIDD than Xbp1 mRNA splicing 26 . Of note, c-Abl has been extensively associated to the DDR, regulating cell-cycle arrest and apoptosis 27 . We confirmed the activation of c-Abl under genotoxic stress ( Supplementary Fig. 5b). Treatment of cells with the c-Abl inhibitor imatinib reduced the decay of Ruvbl1 mRNA in cells exposed to tunicamycin or etoposide ( Supplementary  Fig. 5c, d). Furthermore, knocking down the expression of c-Abl using shRNAs ( Supplementary Fig. 5e) had no effects on the levels of Xbp1 mRNA splicing (Fig. 4a), but fully prevented the decay of Ruvbl1 and Ppp2r1a mRNAs under ER stress or DNA damage (Fig. 4b), phenocopying the consequences of IRE1α deficiency. Consistent with these results, imatinib treatment attenuated the generation of IRE1α-GFP positive clusters in HEK293 cells undergoing genotoxic stress (Fig. 4c, d). Moreover, we also generated a set of c-Abl null MEF cells using the CRISPR/CAS9 technology (Fig. 4e). The deletion of c-Abl prevented the decay of Ruvbl1, Ppp2r1a (Fig. 4f) and Bloc1s1 mRNAs, without an effect on the levels of Xbp1 mRNA splicing (Supplementary Fig. 5f-g). These observations correlated with the formation of a protein complex between IRE1α and c-Abl as monitored using coimmunoprecipitation in 293T HEK cells overexpressing the proteins (Fig. 4g). We confirmed these results using Proximity Ligation Assay of endogenous c-Abl and IRE1-HA at basal conditions ( Fig. 4h) or after exposure to etoposide ( Supplementary  Fig. 5h, i). Furthermore, using a cell free system, we assessed the effects of recombinant c-Abl on the oligomerization status of purified cytosolic domain of IRE1α. Incubation of purified IRE1α at 37°C induced its spontaneous oligomerization, which was further enhanced when c-Abl was present in the reaction (Fig. 4i). Taken together, these results suggest that the activation of c-Abl in cells undergoing DNA damage contributes to the selective engagement of RIDD, possibly through a direct interaction with IRE1α.
IRE1α protect flies against genotoxic stress. To test the possible role of IRE1α in the maintenance of genome integrity in vivo, we took advantage of D. melanogaster as a model organism. The GAL4/UAS system was employed to knockdown the fly orthologue of IRE1 (dIre1) using RNAi transgenic animals (Supplementary Fig. 6a). Etoposide treatment failed to trigger an increase in the levels of dXbp1s mRNA in larval tissue, as monitored by real time PCR (Fig. 5a). However, larvae fed with etoposide or tunicamycin exhibited a similar reduction in the mRNA levels of dSparc and dMys, two well-known RIDD targets in flies 28 , in addition to dPontin, fly orthologue of Pontin (Fig. 5b). dIre1 depletion ablated the downregulation of dSparc, dMys and dPpp2r1a, confirming the occurrence of RIDD (Fig. 5b). Moreover, we then determined the impact of dIre1 on the survival of animals under genotoxic conditions and quantified the number of larvae reaching adulthood. Knock-down of dIre1 generated a hypersensitivity phenotype, meaning that most etoposide-treated animals died before reaching maturity (Fig. 5c). Next, we determined the participation of dIre1 in the maintenance of genome integrity. To this end, we performed the somatic mutation and recombination test (SMART). This assay is based on the induction of mutant spots (clones) that arise from loss of heterozygosity in cells of developing animals, which are heterozygous for a recessive wing cell marker mutation ( Supplementary Fig. 6b) generating a multiple wing hair (mwh) phenotype (Fig. 5d, left panel). We expressed a dIre1 RNAi construct in the wing imaginal disc using a Nubbin-Gal4 driver (Nub-Gal4). Exposure to doxorubicin increased the number of mutant spots in the fly wing, and this phenotype was exacerbated upon depletion of dIre1 (Fig. 5d, right panel), suggesting compromised DNA repair. Doxorubicin also caused higher rates of apoptosis-associated caspase-3 activation upon dIre1 knockdown (Fig. 5e). Next, we developed a mosaic analysis to ablate dIre1 expression with a repressible cell marker (MARCM), a strategy that allows the comparison of wild-type and mutant cells in the same tissue by assessing GFP expression ( Supplementary Fig. 6c). Using this mosaic technology, we generated mutant clones for dIre1 in the eye-antenna imaginal disc and determined the frequency of GFPpositive (dIre1 null cells) and negative cells (WT cells) that persist in the tissue after etoposide treatment. While dIre1 expressing cells maintained their viability after exposure to etoposide (Fig. 5f, left panels), dIre1 null cells proved highly susceptible to this genotoxic agent (Fig. 5f, right panels). Taken together, these results indicate that the fly orthologue of IRE1α protects against genotoxic stress in vivo.
IRE1α deficiency impairs the DDR in mice. We then moved forward and investigated the significance of IRE1α expression to the DDR in vivo and deleted the RNase domain of IRE1α in the liver and bone marrow using a conditional knockout (cKO) system controlled by the Mx-Cre system 29 . Poly[I:C] was injected to induce Cre expression, and three weeks later animals were treated with a single dose of either etoposide or tunicamycin, followed by the analysis of liver tissue. A well-established mammalian model of ER stress consists in the intraperitoneal injection Fig. 3 IRE1α controls the stability of mRNAs involved in the DRR. a Putative IRE1α cleavage sites on the Ppp2r1a and Ruvbl1 mRNAs (blue arrows). b WT and IRE1α KO MEF cells were treated with 10 μM etoposide (Eto). Ppp2r1a and Ruvbl1 mRNA levels were monitored by real-time-PCR. Treatment with 500 ng/mL tunicamicyn (Tm) as positive control (n = 3). c Cells were treated with 10 μM Eto (16 h) and PP2A and Pontin expression were monitored by western blot (n = 3). d In vitro RNA cleavage assay was performed using mRNA fragments of human Ppp2r1a and Ruvbl1, incubated in the presence or absence of recombinant cytosolic portion of IRE1α (IRE1α-ΔN) protein (30 min). Experiments were performed in presence or absence of IRE1α inhibitor 4μ8C. Blos1c1 and Xbp1 mRNA were used as positive controls. e Experimental setup (upper panel): MEF cells were pretreated with 100 ng/mL Tm for 2 h and then treated with 10 μM Eto. Xbp1 mRNA splicing was monitored by RT-PCR (bottom panel). f RIDD activity was monitored in samples described in e (n = 3). g Experimental setup (upper panel): MEF WT cells were pretreated with 10 μM Eto for 2 h and then treated with 100 ng/mL Tm. Xbp1 mRNA splicing was monitored by RT-PCR (bottom panel). h RIDD activity was monitored in samples described in g (n = 3). i IRE1α KO MEF cells were transduced with lentiviruses expressing shRNAs against Ppp2r1a (shPpp2r1a), Ruvbl1 (shRuvbl1) or luciferase (shLuc). Cells were incubated with 1 μM Eto (16 h), washed three times with PBS and fresh media was added. P-H2AX levels were monitored by immunofluorescence after 4 h. P-H2AX foci quantification is shown (Bottom panel) (>200 cells, n = 4-5). j WT, IRE1α KO and reconstituted with IRE1α-HA, expressing shRuvbl1, shPpp2r1a or shLuc cells were treated with 5 μM Eto for 8 h and P-CHK1 and P-ATM monitored by western blot. P-CHK1 quantification is shown (bottom panel) (n = 3). All panels data is shown as mean ± s.e.m.; *p < 0.05, **p < 0.01, and ***p < 0.001, based on b two-way ANOVA followed Bonferroni's test, (f, h-j) One-way ANOVA followed Tukey's test. Data is provided as a Source Data file.
of tunicamycin, which elicits a rapid stress response in the liver. Although evident signs of DNA damage were observed in both control and IRE1α cKO animals (indicated by a rise in p21 mRNA) (Fig. 6a upper panel and Supplementary Fig. 7a), no Xbp1 mRNA splicing was detected in the etoposide treated group (Fig. 6a,  bottom panel). In sharp contrast, a clear down-regulation of Ppp2r1a and Bloc1s1 mRNA levels occurred in the livers (Fig. 6b) and bone marrows (Supplementary Fig. 7b)  i Recombinant IRE1α and c-Abl proteins were incubated at indicated time points and assess its possible interaction by western blot. All data represents the mean ± s.e.m. of three independent experiments, except for co-IP that were performed twice. *p < 0.05, **p < 0.01, and ***p < 0.001, based on (b, f) two-way ANOVA followed Bonferroni's test, (d, h) One-way ANOVA followed Tukey's test. Data is provided as a Source Data file. IRE1α-deficient) animals injected with etoposide. Again, no Xbp1 mRNA splicing was detected in the etoposide treated group, whereas exposure of animals to tunicamycin triggered a very mild response in bone marrow tissue ( Supplementary Fig. 7c). In addition, IRE1α deficiency in the liver altered the DDR, reflected in reduced phosphorylation of CHK1 in animals injected with etoposide (Fig. 6c). Importantly, ablation of IRE1α resulted in enhanced susceptibility of liver cells to apoptosis measured as enhanced caspase-3 activation (Fig. 6d). Finally, to assess the significance of IRE1α to the DDR on an unbiased manner, we performed a gene expression profile analysis of liver tissue derived from mice exposed to etoposide or tunicamycin. Pathway enrichment analysis indicated that IRE1α deficiency attenuated the establishment of a global DDR, delayed the expression of cell cycle arrest genes and activated proapoptotic pathways (Fig. 6e and Supplementary Fig. 8a). As expected, under ER stress induced by tunicamycin, IRE1α deficiency in the liver significantly impacted the expression of genes involved in proteostasis control in the secretory pathway (trafficking, folding and quality control) ( Supplementary  Fig. 8b, c). Taken together, these findings demonstrate that IRE1α signaling contributes to maintaining the stability of the genome when cells face DNA damage.

Discussion
The current study supports a conserved function for IRE1α as a signaling module of the DDR that differs from its canonical role as an UPR mediator. We propose that IRE1α is part of a key decision-making node in a complex interplay between cell survival and DNA repair upon genotoxic stress. In this context, IRE1α regulates the levels of PP2A and RUVBL1 through the selective engagement of RIDD, controlling the kinetics and amplitude of γ-H2AX phosphorylation. The contribution of IRE1α to genome stability is conserved in evolution from insects to mammals and impacts whole animal survival as demonstrated using flies. Our results suggest a regulatory mechanism in which the RNase domain of IRE1α is selectively regulated to specifically engage RIDD, presumably upon interaction with c-Abl (Fig. 4g,  h). This view is consistent with recent studies that connected using unbiased approaches the pathways involved in maintenance of genome integrity and proteostasis, showing that dysregulation of the DDR resulted in protein aggregation and autophagy induction 30,31 . Moreover, previous work demonstrated that the function and structure of the ER is drastically affected by DNA damaging agents used in chemotherapy 32,33 . Other recent reports suggested that chronic ER stress suppresses DNA repair and sensitizes cancer cells to ionizing radiation and chemotherapy [34][35][36][37] , in addition to enhancing oxidative damage to the DNA 38 . Interestingly, a recent study also reported that XBP1u, the protein encoded by the unspliced version of Xbp1 mRNA, regulates the stability of TP53, suggesting alternative connections between the UPR and the DDR under resting conditions 39 . Our results suggest that IRE1α specifically affects signaling events regulating the DDR, and not the DNA damage sensing process. IRE1α operates as an amplification loop, impacting the sustained activation of CHK1/2 and the phosphorylation of γ-H2X through the control of the RIDD targets Ppp2r1a and Ruvbl1, leading to cell cycle arrest and improved DNA repair and as a consequence maintenance of cell survival (see working model in Fig. 6f).
Although RIDD is proposed to be necessary for the maintenance of ER homeostasis 8,10 and to contribute to the pathogenesis of diabetes 5 , cancer 40,41 , and inflammatory conditions [42][43][44] , most of the available evidence is difficult to interpret due to the concomitant existence of Xbp1 mRNA splicing. Our study supports a fundamental biological function for RIDD in the maintenance of . Quantification of the ratio clone size/disc size is presented (right panel) (n = 10 clones). In all panels, data is shown as mean ± s.e.m.; *p < 0.05, **p < 0.01, and ***p < 0.001, based on a one-way ANOVA followed Tukey's test, b-f two-way ANOVA followed Bonferroni's test. Data is provided as a Source Data file.   genome integrity, representing a unique example for a selective and specific activation of RIDD with clear physiological implications. IRE1α is frequently affected by loss-of-function mutations in cancer 2,45 , contrasting with the notion that cancer cells require IRE1α to survive in hypoxic conditions 3,6 . Our present results support the idea that the genetic alterations of IRE1α observed in cancer may synergize with oncogenes to promote genomic instability due to inefficient DNA repair. Altogether, we uncovered a previously unanticipated function of a major UPR signal transducer as an integral component of the DDR, revealing an intimate connection between the pathways that assure the integrity of the proteome and the genome.

Methods
Reagents. Etoposide, doxorubicin, 5-fluorouracil, hidroxyurea, imatinib and 4μ8C were purchased from Sigma Aldrich. Tunicamycin was obtained from Calbiochem EMB Bioscience Inc. IRE1 inhibitor MKC-8866 was provided by Dr. John Patterson (Fosun Orinove). Cell culture media, fetal calf serum, and antibiotics were obtained from Life Technologies and Sigma Aldrich. Fluorescent probes and secondary antibodies coupled to fluorescent markers were purchased from Molecular Probes, Invitrogen. All other reagents were obtained from Sigma or the highest grade available.
Cell culture and DNA constructs. All MEF and HEK cells used here were maintained in Dulbecco's modified Eagles medium supplemented with 5% fetal bovine serum, non-essential amino acids and grown at 37°C and 5% CO 2 . IRE1α deficient cells were previously described 25 . The production of amphotropic retroviruses using the HEK293GPG packing cell line was performed as described in ref. 46 . Retroviral plasmids were transfected using Efectene (Qiagen, Valencia, CA, USA) according to the manufacturer's protocols. IRE1α-HA expressing retroviruses were previously described in the pMSCV-Hygro vector 46 , where IRE1α contains two tandem HA sequences at the C-terminal domain and a precision enzyme site before the HA tag.
RNA isolation, RT-PCR and real time PCR. Total RNA was prepared from cells and tissues using Trizol (Invitrogen, Carlsbad, CA, USA), and cDNA was synthesized with SuperScript III (Invitrogen) using random primers p(dN)6 (Roche). Quantitative real-time PCR reactions were performed using SYBRgreen fluorescent reagent and/or EvaGreen TM using a Stratagene Mx3000P system (Agilent Technologies, Santa Clara, CA 95051, United States). The relative amounts of mRNAs were calculated from the values of comparative threshold cycle by using Actin as a control and Rpl19 or Rpl32 for RIDD substrates in mouse or D. melanogaster samples, respectively. All methods for Xbp1 mRNA splicing assay were previously described in ref. 47 using primers described in Supplementary Table 1.
Immunoprecipitations. HEK-293T cells reconstituted with IRE1α-HA and c-Abl-GFP and IRE1α deficient MEF cells stably transduced with retroviral expression vectors for IRE1α-HA or empty vector were incubated in presence or absence of tunicamycin (500 ng/mL for 4 h) or etoposide (10 μM for 16 h). Cell lysates were prepared for immunoprecipitations in lysis buffer (0.5% NP-40, 50 mM NaCl, 50 mM Tris pH 7.6, 50 mM NaF, 1 mM Na 3 VO 4 and protease inhibitors). Immunoprecipitations (IP) were performed as described 48 . In brief, to IP HAtagged IRE1α, protein extracts were incubated with anti-HA antibody-agarose complexes (Roche), for 4 h at 4°C, and then washed three times with 1 mL of lysis buffer and then one time in lysis buffer with 350 mM NaCl. Beads were dried and resuspended in Sample Buffer 2x. Samples were heated for 5 min at 95°C and resolved by SDS-PAGE 8% followed by western blot analysis.  Acquisition was performed by means of a 20X PlanApo objective (Nikon, Tokyo, Japan). Minium 9 views fields per well for 96-wells plate were acquired. MetaX-press® was utilised to segment cells into nuclear area (based on Hoechst 33342 signal). Cell-like objects were segmented and divided into cytoplasmic and nuclear regions as previously reported 49 .
Proximity ligation assay (PLA). Cells were seeded in 12 mm cover slips. After the indicated treatments, cells were fixed for 20 min at RT with 4% paraformaldehyde and permeabilized 0.5% NP-40 in PBS containing 0.5% BSA (Bovine serum albumin) for 10 min. After blocking for 1 h with 10% FBS in PBS containing 0.5% BSA, cells were incubated with the indicated antibodies: Anti-HA (Cat: 901514, Biolegend or Cat: 9110, Abcam) and anti-Abl 1:1000 (Sigma, Cat. A5844) overnight at 4°C following by Duolink manufacturer´s instructions (Duolink®, Sigma-Aldrich). Images were acquired by confocal microscopy (Nikon C2 plus) using a 60X oil objective lens stacking the images every 0.5 µm to cover all the image of interest. Stack images were deconvoluted using Huygens and ImageJ. Stack deconvolved images were reduced to one dimension by sumslices function (ImageJ). Colocalization was performed in thresholded and masked images were used to determine Manders/Pearson's index was calculated with ImageJ plugin.
Comet assay. The comet assay was performed as previously described 50 . Briefly, agarose-slides were prepared with 1% low-gelling-temperature agarose and 2 × 10 4 cells/ml and submerged in lysis solution (1.2 M NaCl, 100 mM Na 2 -EDTA, 0.1% sodium lauryl sarcosinate, 0.26 M NaOH (pH > 13)) for 18 h at 4°C in the dark. Then, carefully slides removed and submerged in room temperature (18 − 25°C) in rinse solution (0.03 M NaOH, 2 mM Na 2 EDTA (pH ∼12.3)) for 20 min. electrophoresis was conducted in the same solution for 25 min at a voltage of 0.6 V/cm. Finally, slides were stained in a solution containing 2.5 µg/ml of propidium iodide in distilled water for 20 min and observed in a epifluorescence microscopy. Images were analyzed using Comet Assay IV software.
Cytokinesis-block micronucleus assay. Cytokinesis-block micronucleus (CBMN) assay was performed as previously described 51 . In brief, cells were treated with 5 etoposide (5 μM for 3 h). Then, cells were washed three times with PBS and incubated for 24 h with fresh media with 5 μM of Cytochalasin-B. Cells were fixed, stained with Hoeschst. Binucleated cells (BN) with micronucleus (MN), Nuclear buds (Nbuds) or nucleoplasmid bridges (NPB) were detected and quantified using epifluorescence microscopy.
In vitro RNA cleavage assay. Bloc1s1 (NM_001487.3), Ppp2r1a (NM_014225) and RuvbL1 (NM_00370) cDNA were obtained from MGC cDNA library (Dharmacon). Long sense and antisense oligonucleotides containing a minimal T7 RNA polymerase promoter (5′-TAATACGACTCACTATAGG-3′) fused upstream of the sequence containing different fragments of the genes Ppp2r1a, RuvbL1 and Bloc1s1 harboring 5′-EcoRI and 3′-BamHI overhangs were annealed and ligated into the cognate restriction sites of pUC19 (Invitrogen, Life Technologies). Oligonucleotides sequence to clone Ppp2r1a and RuvbL1 fragments were described previously 20 . In vitro transcription reactions were performed with T7 RNA polymerase using the HiScribe T7 high-yield RNA synthesis kit (New England Biolabs) following the manufacturer's recommendations. The transcribed RNA were treated for 20 min with DNase and purified by urea-polyacrylamide gel electrophoresis (urea-PAGE). RNAs recovered from gel fragments by the crush-and-soak method, were precipitated with 300 mM NaOAc and 1 volume of isopropanol. No coprecipitants were employed. The precipitated RNA pellets were desalted by two washes with 70% ice-cold ethanol, air-dried and re-suspended in an appropriate volume of either nuclease-free water or RNA resuspension buffer (20 mM HEPES, 100 mM NaCl, 1 mM Mg(OAc) 2 ). The oligonucleotides sequence used are listed in the Supplementary Table 2.
The cytosolic kinase/ribonuclease domain construct of IRE1α (KR43) was expressed and purified as described previously 52 . In vitro transcribed, PAGEpurified, refolded RNAs (50 ng) were incubated with 0.5 μM IRE1α-KR43 for the indicated times in RNA cleavage buffer (20 mM HEPES pH 7.5, 70 mM NaCl, 2 mM Mg(OAc) 2 , 1 mM TCEP, and 5% glycerol). Stop solution (10 M urea, 0.1% SDS, 1 mM EDTA, 0.05% xylene cyanol, 0.05% bromophenol blue) was added at five-fold excess to stop the reactions followed by heating at 80°C for 3 min. The denatured samples were then loaded on 6% TBE-urea gels (Invitrogen, Life Technologies) and the gels stained with SYBR Gold nucleic acid stain (Invitrogen, Life Technologies). As a negative control we utilized the IRE1α inhibitor 4μ8C to a final concentration of 5 μM.
Viability assay. In all, 2.0 × 10 4 cells were seeded in 48-well plate and the maintained by 24 h in DMEM cell culture media supplemented with 5% bovine fetal serum and non-essential amino acids. Genotoxic and ER stress were induced by adding genotoxic and ER stress agents to the cells at different concentrations, and maintained for 24 h. Then, cell viability was monitored using propidium iodide staining and flow cytometry (BD FACS Canto, Biosciences).
Mouse model. Ern1 floxed mice were previously described 53 and crossed with Mx1-cre transgenic mice to generate a conditional KO animal (IRE1α cKO ). Deletion was induced by injection of polyinosinic-polycytidylic acid (Poly (I:C)) which efficiently delete the floxed gene in the liver and bone marrow 29 . I all, 5-6-weeksold mice were intraperitoneally (i.p) injected three times with 250 µg of poly(I:C) each time with 2 day intervals to induce the Cre expression. Mice were used for experiments at least 2-3 weeks after the final poly (I:C) injection. DMSO or 50 mg/ Kg etoposide or 1 mg/Kg tunicamycin or were i.p. injected and 6 h later mice were sacrificed as reported 54 . The liver and bone marrow were frozen at −80°C for biochemical analysis and the right major lobe of the liver was placed in a petri dish (on ice). Liver tissue was washed in PBS to remove the blood and then, it was fixed in 4% paraformaldehyde (72 h) for histological analyses. The animals' works and care was in accordance with institutional guidelines. Institutional Committee for Animal Care and Handling, University of Chile (Protocol CICUA-CBA-0833).
Fly studies. Flies were kept at 25°C on standard medium with a 12-12 dark-light cycle. Drug administration protocol for all experiments is as follows: Larvae were grown in standard fly medium until day 3 after egg lying (AEL). Then they were transferred to fly instant medium (Carolina Biological Supply 2700 York Burlington, NC, USA) complemented with the appropriate drug for the different treatments. Larvae were fed with the corresponding drug-supplemented media until dissection or adulthood.
Survival curve and biochemical analysis of fly tissue. Survival analyses were carried out growing 20 experimental or control larvae in 500 μM etoposide and the number of living animals was quantified at different time points. For real time RT-PCR analysis, total RNA was extracted from third instar larvae (same treatment as the survival analysis animals) using Trizol (Invitrogen, Carlsbad, CA, USA), and cDNA was synthesized with SuperScript III (Invitrogen) 56,57 .
For immunohistochemistry, larvae were fed with 0.125 mg/mL doxorubicin. Then third instar larvae were dissected and fixed as described previously 58 . Larvae carcasses were incubated with Anti-caspase-3 (1:100, Cell Signaling) overnight in PBT supplemented with 0.5% BSA at 4°C, washed four times with PBT for 15 min and stained with Alexa-conjugated secondary antibody (1:200, Molecular Probes) and TO-PRO3 (1 μM, Invitrogen). VectaShield (Vector Laboratories, Burlingame, California, USA) was used as a mounting media. The number of active Caspase-3 positive cells was quantified in 10 wing imaginal discs (N = 3). Images were taken using a Zeiss LSM510 confocal microscope and analyzed with ImageJ software.
Microarray analyses. Affymetrix gene expression data were pre-processed using Transcriptome Analysis Console (TAC) (v4.0, ThermoFisher Scientific). Custom mouse Brainarray chip definition (v22) was used to further annotate the DE files with Entrez and gene symbol IDs. For further analysis, just gene transcript with FDR (a = 0.05) correction and 1.5 > fold change was considerated. For pathway enrichment analysis, data obtained from untreated wild type or IRE1α cKO after poly I:C treatment liver mice tissues were used as reference for tunicamycin (Tm) and etoposide (Eto) treatment comparisons, to further input them in ClueGO (v2.3.2) software using Reactome pathway enrichment database (v09.11.2016). In addition, to visualize patterns in the gene expression and pathway enrichment scores for specific ontologies, heatmaps were generated using RStudio (v0.99.489, R 3.4.1) based on KEGG pathway database gene lists. Genes which show change of the ratios higher (or lower) than 1.5-fold in the arrays at any comparison have been considered as up or downregulated and subjected for functional enrichment analysis. The Bioconductor package 'clusterProfiler' was applied to perform functional enrichment analysis using the following repositories: GO (Gene Ontology), KEGG (Kyoto Encyclopedia of Genes and Genomes), and Reactome Pathways. GEO Dataset ID: GSE130952.
Statistical analysis. Results were statistically compared using the Ordinary Oneway ANOVA and Two-way ANOVA followed by different multiples comparison post-tests (Tukey's Multiple Comparison Test or Bonferroni's Multiple Comparison Test). When pertinent, Student's t-test was performed for unpaired or paired groups. In all plots p values are show as indicated: *: p < 0.05, **: p < 0.01, ***: p < 0.001 and ****: p < 0.0001 and were considered significant. n.s: nonsignificant.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.