The splicing factor XAB2 interacts with ERCC1-XPF and XPG for R-loop processing

RNA splicing, transcription and the DNA damage response are intriguingly linked in mammals but the underlying mechanisms remain poorly understood. Using an in vivo biotinylation tagging approach in mice, we show that the splicing factor XAB2 interacts with the core spliceosome and that it binds to spliceosomal U4 and U6 snRNAs and pre-mRNAs in developing livers. XAB2 depletion leads to aberrant intron retention, R-loop formation and DNA damage in cells. Studies in illudin S-treated cells and Csbm/m developing livers reveal that transcription-blocking DNA lesions trigger the release of XAB2 from all RNA targets tested. Immunoprecipitation studies reveal that XAB2 interacts with ERCC1-XPF and XPG endonucleases outside nucleotide excision repair and that the trimeric protein complex binds RNA:DNA hybrids under conditions that favor the formation of R-loops. Thus, XAB2 functionally links the spliceosomal response to DNA damage with R-loop processing with important ramifications for transcription-coupled DNA repair disorders.

T he spliceosome is a highly dynamic, ribonucleoprotein complex that comprises five small nuclear (sn) RNAs (the U1, U2, U4, U5, and U6 snRNAs) and a growing number of associated splicing factors that enable the selective intron excision of nascent pre-mRNA transcripts prior to mRNA translation 1 . Splicing initiates with the binding of the U1 snRNP to the GU sequence at the 5′ splice site of an intron and the zinc finger protein splicing factor 1 (SF1) that binds to the intron branch point sequence. The U2 snRNP displaces SF1 and binds to the branch point sequence followed by ATP hydrolysis. The snRNPs U2 and U4/U6 position the 5′ end and the branch point in proximity. The latter stimulates snRNP U5 and the 3′ end of the intron is brought into proximity and joined to the 5′ end. The resulting lariat is released with U2, U5, and U6 bound to it 2 . Pre-mRNA splicing is functionally coupled to transcription 3,4 . The process of mRNA synthesis is known to regulate constitutive and alternative splicing (AS) through the physical association of splicing factors to the elongating RNAPII 5 , the 3D chromatin structure 6,7 or through the fine-tuning of RNA polymerase II (RNAPII) elongation rates 8 . Recent findings support the notion that DNA damage is inherently linked to splicing [9][10][11] . The presence of irreparable DNA lesions may influence RNA splicing through the selective nuclear transport of splicing factors 12,13 , the inhibition of the elongating RNAPII 9 , the interaction of RNAPIIand spliceosome-associated factors 14 or the displacement of the spliceosome when RNAPII is stalled at DNA damage sites 11 .
XPA-binding protein (XAB)-2 is the human homologue of the yeast pre-mRNA splicing factor Syf1 that contains 15 half a tetratricopeptide (HAT) repeats shared by proteins involved in RNA processing 15,16 . Disruption of the Xab2 gene results in preimplantation lethality in mice 17 . Thus, there are currently no available data to address the functional role of XAB2 in transcript maturation and the DNA damage response (DDR) in vivo. Using a yeast two-hybrid screen and in vitro pull-down assays, XAB2 was shown to interact with the nucleotide excision repair (NER) factors XPA, CSA and CSB proteins and with RNAPII 18 . The yeast homolog of XAB2, Syf1 was also shown to be a TREX (transcription and export complex)-interacting factor 19 . Microinjection of antibodies raised against XAB2 inhibited the recovery of RNA synthesis after UV irradiation and normal RNA synthesis in fibroblasts 18 . Follow-up work showed that XAB2 interacts with PRP19 20 an ubiquitin-protein ligase involved in pre-mRNA splicing and DNA repair 21,22 and Aquarius (AQR) an intron-binding spliceosomal factor that links pre-mRNA splicing to small nucleolar ribonucleoprotein biogenesis 23 . Moreover, XAB2 is functionally involved in homology-directed repair and single-strand annealing 24 and it was recently shown to be directly involved in pre-mRNA splicing of many genes in vitro, including POLR2A 25 .
To dissect the functional contribution of XAB2 in mRNA synthesis and DNA repair, we established an in vivo biotinylation tagging approach in mice and primary cells. Our findings provide evidence that XAB2 is essential for NER, the pre-mRNA splicing and for R-loop processing. We show that XAB2 is part of a core spliceosome complex that associates with the UsnRNAs and that transcription-blocking DNA damage triggers the release of XAB2 from all RNA targets tested. RNAi-mediated knockdown of XAB2 leads to decreased RNA synthesis, aberrant intron retention, Rloop accumulation and DNA damage. A series of immunoprecipitation strategies revealed that XAB2 interacts with ERCC1-XPF and XPG endonucleases and that the XAB2 complex is recruited to RNA:DNA hybrids under conditions that favor the accumulation of R-loops. We propose that XAB2 functionally links persistent DNA damage with the core spliceosome and the processing of R-loops highlighting the causal contribution of transcription-blocking DNA lesions to the progeroid and developmental defects associated with TC-NER disorders [26][27][28][29] .

Results
Generation of biotin-tagged XAB2 mice. Ablation of Xab2 in the murine germline results in preimplantation lethality, indicating an essential role for XAB2 during mouse embryogenesis 17 . To delineate the functional role of XAB2, we generated knock-in animals expressing XAB2 fused with a tandem affinity purification tag comprising two affinity moieties, a 1X FLAG tag and a 15aa Avitag sequence, separated by a Tobacco Etch Virus (TEV) site for easy tag removal. The tag was inserted before the stop codon of the last exon 19 (Fig. 1a). The targeting vector was transfected to 129/SV ES cells expressing the Protamine 1-Cre recombinase transgene, which efficiently excises the neomycin cassette in the male germ line 30 . After selecting properly targeted clones ( Fig. 1b; as indicated), two independently transfected clones were used to generate germ line-transmitting chimeras. Avitag-fused heterozygous males (aviXab2 +/− mice) were backcrossed and maintained in a C57/BL6 background. Homozygous aviXab2 +/+ knock-in animals were then crossed with mice ubiquitously expressing the HA-tagged BirA biotin ligase transgene 31 . BirA is a bacterial ligase that specifically recognizes and successfully biotinylates the short 15aa biotinylation Avitag, thus creating a high affinity "handle" for the in vivo isolation of XAB2-bound protein complexes in aviXab2 +/+ ; birA (from now on designated as bXAB2) mice by binding to streptavidin. Importantly, biotinylation of the inserted Avitag sequence does not interfere with murine development as bXAB2 animals are born at the expected Mendelian frequency (Fig. 1c), grow normally (Fig. 1d), and show no developmental defects or other pathological features (Fig. 1e). Nuclear extracts of P15 livers from aviXAB2 or bXAB2 mice were probed with streptavidin-HRP (stp-HRP) confirming the biotinylation of bXAB2 in vivo (Fig. 1f). The use of anti-FLAG and anti-HA confirmed the presence of the knock-in allele and the BirA transgene, respectively. Pulldowns with 0.6 mg of nuclear extracts from livers of P15 bXAB2 +/+ mice revealed the percentage of biotinylated XAB2 in the pulldown (90%) and flow through (fth; 10%) fraction (Fig. 1g). Previous findings suggest that XAB2 is involved in the transcription-coupled repair (TC-NER) sub-pathway of NER 18 and during the end resection step of homologous recombination repair 24 . To test whether the Avitag sequence interferes with the putative function of XAB2 in DNA repair in vivo, we further exposed mouse embryonic fibroblasts (MEFs) to ultraviolet irradiation (UV) as well as to mitomycin C (MMC). MMC is a potent genotoxic agent which inhibits replication and transcription of DNA by introducing DNA inter-strand crosslinks that prevent dissociation of the strands. We find a similar percentage of UV-induced unscheduled DNA synthesis in bXAB2 cells compared to wild-type (wt.) controls (Fig. 1h) and that bXAB2 MEFs are not hypersensitive to UV irradiation (Fig. 1i) or to MMC (Fig. 1j). Thus, bXAB2 animals develop normally to adulthood and are proficient at repairing UV-induced cyclobutane pyrimidine dimers (CPDs) and DNA interstrand cross-links (ICLs).
A proteomics strategy reveals XAB2-bound protein partners involved in RNA processing and DNA repair. To isolate and characterize XAB2-associated protein complexes during postnatal hepatic development, we combined the in vivo biotinylation tagging approach 31 with a hypothesis-free, high-throughput proteomics strategy. To do this, we prepared nuclear extracts from livers of P15 bXAB2 and BirA mice using high-salt extraction conditions. Nuclear extracts from BirA mice were used to identify proteins non-specifically biotinylated by the BirA ligase (Fig. 2a). The liver is a relatively homogeneous organ that accurately depicts the growth status of the developing animal.
Nuclear extracts were treated with benzonase and RNase A, to ensure that DNA or RNA does not mediate the identified protein interactions. Nuclear extracts were further incubated with streptavidin-coated beads and bound proteins were eluted. To confirm the successful pulldown of XAB2, eluates were subjected to Western blotting for FLAG and for the pre-mRNA-processing factor 19 (PRP19), a known interactor of XAB2 32 (Fig. 2b). Next, the proteome was separated by 1D SDS-PAGE (Fig. 2c) S1A) containing a significantly disproportionate number of proteins relative to the murine proteome were flagged as significantly over-represented (FDR < 0.05). At this confidence level, the over-represented biological processes and pathways involved 372 out of the initial 636 XAB2-bound core proteins; the latter set of proteins also showed a significantly higher number of known protein interactions (i.e., 958 interactions) than expected by chance (i.e., 531 interactions; Fig. 2f) indicating a functionally relevant and highly interconnected protein network. Using this dataset, we were able to discern four major XAB2-associated protein complexes involved in (i) pre-mRNA splicing (p ≤ 3.2 × 10 −38 , e.g., AQR, BCAS2, PRP19, PRP8, SNRP40), (ii) RNA transport (p ≤ 1.8 × 10 −7 , e.g., NUP107, NUP133, NUP153, NUP160, NUP205, NUP210, NUP50, NUPl2), (iii) Cell cycle (p ≤ 7.1 × 10 −6 , e.g., ANAPC1, ANAPC 2, ANAPC 4, ANAPC 5, ANAPC 7, CDC23, CDC27, FZR1, RAD21, SMC1a, SMC3, STAG1, STAG2) and iv. Ribosome biogenesis (p ≤ 4.8 × 10 −6 , e.g., BMS1, GNL3, MDN1, NOP58, UTP15, WDR36, WDR43, WDR75, XRN2). Together, these findings indicate that the great majority of XAB2-bound protein partners are functionally involved in RNA processing and genome utilization processes. The role of XAB2 in splicing in vitro is already well-documented. In support, spliceosomal factors were highly enriched in the bXAB2-bound proteome of P15 livers. Nevertheless, as splicing factors are often nonspecifically enriched in affinity enrichment-mass spectrometry approaches 33 , we first challenged the specificity of XAB2 interaction with the pre-mRNA splicing machinery by comparing the bXAB2 MS liver proteome to that derived previously from P15 biotin-tagged xeroderma pigmentosum complementation F (bXPF) livers under similar experimental conditions 22 . XPF is the obligatory interacting partner in the NER structure-specific ERCC1-XPF endonuclease complex 34 . Unlike bXPF, we find that bXAB2 interacts with several core spliceosome factors and more than half of them are proteins involved in the PRP19 and PRP19-related complex (nine out of 16 pre-mRNA splicing factors) ( Fig. 2g; colored in red).
Depletion of XAB2 leads to defective NER. Previous findings revealed that XAB2 is involved in the transcription-coupled subpathway of NER (TC-NER) 18,33 . To test for the functional role of XAB2 in NER, we carried a series of pulldown assays in nuclear extracts of UV-irradiated (10 J/m 2 ) and control bXAB2 primary MEFs. We find that bXAB2 interacts with the DNA damage-binding protein-1 (DDB1) involved in UV-induced DNA damage recognition 34 , and with a portion of xeroderma pigmentosum, complementation group A protein (XPA) known to assemble the NER incision complex at sites of DNA damage 35 , in control and UV-irradiated cells (Fig. S1B). Confocal microscopy in UVirradiated HEPA cells revealed that XAB2 does not co-localize with the UV-induced CPDs (Fig. S1C). The great majority of XAB2 is distributed throughout the nucleoplasm but is excluded from the denser DAPI-stained heterochromatin regions or the nucleolus. In line with previous findings 24 , we also find that XAB2 forms punctate staining, that is distinctively adjacent to the DNA damage marker γH2Ax (Fig. S1D). Knockdown of XAB2 in siXab2 HEPA cells (Fig. S1E) results in increased cell death (Fig. S1F), reduced proliferative capacity (Fig. S1G) at 72 h posttransfection, and in the presence of irreparable UV-induced CPDs at 24 h post-UV irradiation (Fig. 3a). The role of XAB2 in transcription-coupled repair has been previously documented 18 .
In the present work, we find that siXab2 cells manifest a noticeable decrease in unscheduled DNA synthesis at 2.5 h post-UV irradiation (measuring global genome NER; Fig. 3b) 36 . siXab2 cells also present with substantial delay in the recovery of RNA synthesis (assessing TC-NER; Fig. 3c) as evaluated by the fluorescent detection of bromouridine (BrU) labeled nascent RNA at 4 h post-UV irradiation. In agreement with previous findings 25 , knockdown of XAB2 reduced RNAPII ( Fig. S1H) but did not affect the protein levels of NER factors DDB1, XPA and ERCC1, (Fig. S1I). Next, we investigated whether DDB1 and XPA are efficiently recruited at DNA damage sites in UV-irradiated siXab2 HEPA cells. We find that DDB1 co-localizes competently with UV-induced CPDs in siXab2 cells (Fig. 3d). In contrast, XPA fails to form visible foci and properly recruit on UV-induced CPDs in these cells (Fig. 3e). Finally, we find that treatment with Iso-Ginkgetin (IsoG), a pre-mRNA splicing inhibitor leads to severe splicing defects as documented by the accumulation of pre-mRNAs and the decrease of corresponding mRNA levels, (Fig. S1J-K) but has no significant impact on the repair of UVinduced CPDs in HEPA cells indicating that splicing per se does not affect NER.
XAB2 is part of a core spliceosome complex that binds on UsnRNAs. Follow-up pulldown experiments confirmed the mass spectrometry interactions and revealed that the endogenous bXAB2 interacts with AQR, PRP19, and BCAS2, an integral component of the spliceosome that is required for activating pre-mRNA splicing 37 in P15 livers (Fig. 4a). The interaction of bXAB2 with core spliceosome factors in P15 livers prompted us to test whether XAB2 is in complex with RNA in vivo. To do so, we carried out a series of RNA immunoprecipitation (RIP) Fig. 1 Generation of biotin-tagged XAB2 animals. a Schematic representation of knock-in mice expressing XAB2 fused with the 1xFLAG tag, a 15aa Avi tag sequence and a Tobacco Etch Virus (TEV) site separating the two tags. EcoRI and HindIII sites are synthetic; b: Flag-Tev-AVI fragment. b Two independently transfected ES clones (marked with arrow heads) were used to generate germline transmitting chimeras that were backcrossed with C57Bl/ 6 mice to generate aviXab2 +/pups. Homozygous aviXab2 +/+ knock-in animals (aviXAB2) were then crossed with mice ubiquitously expressing the HAtagged BirA biotin ligase transgene (aviXab2 +/+ ;birA, designated as bXAB2). c Average number of wt (n = 185 pups/24 birth events) and bXAB2 (n = 319 pups/44 birth events) pups born per birth event. d Weight (grams; g) of wt. and bXAB2 animals (n = 8) at the indicated time points. e A photograph of P20 wt. and bXAB2 animals. f In vivo biotinylation of the short 15aa Avi-tag in bXAB2 animals. Nuclear extracts from P15 livers of mice expressing either only aviXAB2 (XAB2) or aviXAB2 and BirA biotin ligase (bXAB2) were tested by Western blot. The blot was probed with streptavidin-HRP (stp-HRP), anti-FLAG and anti-HA. experiments in P15 bXAB2 liver nuclear extracts incubated with streptavidin-coated beads (for bXAB2). The co-precipitated RNA was eluted, reversely transcribed, and quantified for spliceosomal UsnRNAs by quantitative (q)PCR. We find that bXAB2 associates with U4, U5 and U6 small nuclear RNAs (snRNAs) but shows low to no specificity for U1, U2, or U3 snRNA (Fig. 4b). Unlike with XAB2 and AQR, RIP assays revealed that BCAS2 is not recruited to U4 and U6 snRNAs (Fig. 4c). To test whether XAB2 binds directly to UsnRNAs, we performed a series of gel mobility shift assays using a biotinylated RNA oligo that was designed based on the sequence similarities of U4, U5, and U6 snRNAs with the minimal RNA binding sequence of HCF107, a chloroplast-localized protein with repeats of HAT motifs (Fig. S1L). This strategy reveals that the XAB2 IP eluent from nuclear extracts of MEFs consumes the RNA oligo but not its mutant version (Fig. S1M) indicating that the XAB2 complex a.  Data 4) in siXab2 mESCs, with a striking overrepresentation of ribosomal protein genes. Using this dataset, we next tested whether knockdown of Xab2 has an effect on splicing decisions, i.e., exon skipping (ES), intron retention (IR), and alternative selection of 5′ or 3′ splice sites (Fig. 4d). Of the 126335 and the 56677 alternative spliced (AS) events detected in HEPA cells and mESCs respectively, 774 and 1134 were found differentially spliced in siXab2 cells (probability of differential splicing >0.8 and an absolute change in percent spliced-in (ΔPSI) >5%). We find that the great majority (i.e., 805 out of 849, 94.82%, p = 10 −16 ) of IR events with a significant probability of differential splicing are retained in siXab2 HEPA and mESCs cells further confirming previous findings in HeLa cells 25  To investigate which biological processes are mostly affected by XAB2 knockdown-mediated AS, we subjected the AS events found to GO classification. At the confidence level set in our analysis (FDR ≤ 0.05), we find that biological processes related, among others, to cell cycle, RNAPII-mediated transcription, DNA repair, and RNA processing comprise a disproportional number of transcripts in both mESCs and HEPA cells relative to the murine transcriptome (Figs. S3A, S3B). RIP in P15 livers revealed that XAB2 binds to the pre-mRNAs of several genes involved in e.g. DNA repair and the DDR (Fig. S3C) as well as to the pre-mRNAs of genes known to be highly transcribed during postnatal hepatic development (e.g., Ghr, Igf1) (Fig. S3D). Transcripts with retained introns often contain premature stop codons (PTCs), which leads to their cytoplasmic degradation by the nonsense-mediated RNA decay machinery 38,39 or, as in the case of XAB2-mediated intron retained transcripts to aberrant RNA-decay pathways 25 . Otherwise, a subset of these transcripts may be stably retained within the nucleus for later processing 40 . Beginning at 48 h post-transfection, we find that the pre-mRNA levels of transcripts with retained introns gradually accumulate in HEPA cells depleted for XAB2 with two distinct siRNAs ( Fig. 4i and Fig. S3E) whereas their corresponding mRNA levels progressively decrease (Fig. 4j) indicating a defect in splicing rather than IR regulation in siXab2 cells. Thus, XAB2 is part of the core spliceosome complex and has a functional role in pre-mRNA splicing; it is recruited on UsnRNAs and pre-mRNAs and it is required for proper mRNA splicing.
XAB2 is released from UsnRNAs and pre-mRNAs upon DNA damage and transcription blockage. UV-induced DNA lesions trigger the displacement of co-transcriptional mature catalytically active spliceosome (U2, U5, and U6) 11 . In agreement, we find that the chromatin association of U4 and U6 snRNAs is reduced 2 h post-UV exposure and is restored within 6 h after UV irradiation (Fig. S3F). RIP assays revealed that PrP3, a U4/U6 snRNP is released from U4 and U6 snRNAs within 2 h post-UV irradiation and re-associates with the snRNAs 6 h post-UV irradiation (Fig. S3G). XAB2 is released from U4, U5, and U6 snRNAs and from all pre-mRNAs tested 2 h post-UV irradiation and its association with UsnRNAs is restored 24 h after UV exposure ( Fig. 5a, b). Release from UsnRNAs was also evident for AQR, which persists for 6 h after UV irradiation (Fig. S3H). Importantly, the release of XAB2 from UsnRNAs is not accompanied by its disassociation from other protein components of the spliceosome machinery, i.e., PRP8, PRP19, BCAS2, and SNRP40 ( Fig. 5c) and is consistent with the substantial increase in intron retention for several transcripts tested in UV-irradiated cells (Fig. S4A). In contrast to the ATM-dependent UV-induced disassociation of mature spliceosome from nascent transcripts 11 , selective inhibition of ATM (ATMi) with KU-55933 inhibitor 41 or of ATR (ATRi) with NU6027 in cells treated with UV did not rescue the release of XAB2 from U4 and U6 snRNAs (Fig. 5d). These findings indicate that the DNA damage-driven release of XAB2 from RNA is not mediated through active DDR signaling. Next, we reasoned that DNA lesions altering the DNA double helix itself and/or interfering with transcription in cis are causal to the aberrant release of XAB2 from UsnRNAs or the pre-mRNAs. To test this, we first treated UV-irradiated cells with BrU; the latter revealed that RNA synthesis recovery is delayed 2 h post-UV irradiation and resumes 6 h-post-UV exposure to levels similar to those seen in untreated wt. controls (Fig. S4B). Next, we made use of MEFs derived from CPD photolyase transgenic mice 42 . Photolyases directly revert photolesions into undamaged bases using visible light energy (i.e. interacts with protein complexes involved in transcript maturation. a Schematic representation of the isolation of bXAB2 complexes derived from livers of P15 bXAB2 animals expressing the BirA transgene and BirA transgenic control mice. b bXAB2 pulldowns (PD) and western blot with anti-FLAG (for XAB2) and anti-PRP19 in nuclear extracts derived from P15 bXab2 and BirA livers (n = 3 livers). c A representative SDS-PAGE gel of proteins extracts derived from P15 XAB2 and BirA transgenic mice. d Venn's diagram of bXAB2-bound protein factors from three independent pull-downs (PD) and subsequent MS analyses; % depicts the percentage of XAB2-bound proteins involved in the indicated biological process over the total number of XAB2bound proteins (e). List of significantly over-represented pathways (KEGG). The −log(FDR: False Detection Rate), which is calculated by Fisher's exact test right-tailed (one-sided), sorts the biological processes. The red-dotted line marks the threshold of significance at FDR ≤ 0.05. Count (G.C.): the number of identified XAB2-bound protein factors involved in the indicated biological process. F.E. (Fold Enrichment), represents the ratio of XAB2-bound proteins involved in a process (sample frequency) to the total genes involved in the process (background frequency). f Number of observed (obs.) and expected (exp.) known protein interactions within the core 372 XAB2-bound protein set and schematic representation of the four significantly over-represented XAB2-bound protein complexes based on experimental evidence. g Heat map representation of bXAB2-and bXPF-bound proteins involved in the core spliceosome in P15 bXAB2 and bXPF livers. Constitutive factors of snRNPs are green-colored; the PRP19 and the PRP19-related complex are blue-colored. Protein components of the TREX and exon junction complex are black-colored. SPF27 (BCAS2) is marked with an "*" as it has been identified in two out of the 3 MS runs in P15 bXAB2 livers but has been further confirmed in downstream immunoprecipitation/pulldown assays. The images shown in Fig. 2b and c are representative of experiments that were repeated three times. All scatter and bar blots in this manuscript are presented as mean ± SEM. P values were calculated by two-tailed Student's t test, unless otherwise indicated. Source data provided as a Source Data file.
photoreactivation; PR) and display substrate specificity for UVinduced 6-4PPs or CPDs 43 . Using this system, CPD photolyase transgenic MEFs were exposed to UV irradiation and were subsequently kept in the dark (UV-induced CPDs remain in the mammalian genome) or exposed to PR light (UV-induced CPDs are repaired) for 60 min post-UV irradiation (Fig. 5e). Importantly, we find that the exposure of UV-irradiated MEFs to PR light substantially rescues the XAB2 RIP signals in U4 and U6 snRNAs (Fig. 5f) and in pre-mRNAs tested (Fig. 5g) indicating that the presence of DNA helix-distorting lesions inadvertently affects the process of pre-mRNA splicing in the mammalian genome. In line, inhibition of splicing in IsoGtreated cells or of transcription elongation blockage in DRBtreated cells triggers the substantial release of XAB2 from U4 snRNA (Fig. S4C-D). XAB2 and AQR are also released from U6 snRNA (Fig. S4D) and all pre-mRNAs tested after DRB treatment (Fig. S4E). Similar findings, albeit to a lesser magnitude, were observed upon inhibition of transcription initiation in TPL-treated cells ( Fig. S4D-E). Likewise, we find the aberrant release of XAB2 from UsnRNAs in MEFs treated with illudin S known to induce DNA lesions that are ignored by global-genome NER and are exclusively processed by transcription-and replication-coupled repair pathways (Fig. 5h) 44 . The latter prompted us to test whether XAB2 binding to U4 and U6 snRNAs is also altered in Csb m/m mice carrying an inborn defect in TC-NER 45 . Using P15 bXab2/Csb m/m livers, we find that the XAB2 RIP signals on U4 and U6 snRNAs as well as on pre-mRNAs Cdk7, Ercc1 and the highly transcribed Ghr and Igf1 genes are substantially decreased in the P15 livers of bXab2/Csb m/ m animals ( Fig. 5i and Fig. S4F).
Abrogation of XAB2 promotes R-loop formation triggering DNA damage. R-loops are generated during transcription when nascent RNA exits RNA polymerase and pairs with its complementary DNA template to form a stable RNA-DNA hybrid that displaces single-stranded DNA (ssDNA) 46 . Increased formation of R-loops is frequently observed in cells that are defective in DNA repair or RNA processing [47][48][49] . A DNA-RNA immunoprecipitation (DRIP) approach followed by treatment with RNase H (RNH; it digests RNA in DNA-RNA hybrids eliminating R-loops) revealed that, upon UV irradiation, R-loops accumulate in Cdk7 and Ercc1 genes whose pre-mRNAs were previously shown to associate with XAB2 (Fig. 6a). Inhibition of ATM led to a detectable yet mild decrease in the formation of Rloops indicating that defective DDR signaling alone cannot fully justify the observed accumulation of R-loops in UV-irradiated cells (Fig. 6a). Consistent with our DRIP analysis on UVirradiated cells (Fig. 6a), confocal microscopy studies revealed that UV irradiation triggers the formation of R-loops in wt. cells that further accumulate when Xab2 is silenced (Fig. S4G); the latter leads to the substantial increase in the number of γΗ2Αx (+) 53BP1 (+) cells (Fig. S4H). This and our finding that R-loops accumulate substantially in MEFs treated with IsoG (Fig. 6b,  Fig. S4I) prompted us to further explore the functional role of XAB2 in R-loop processing. Similar to UV irradiation or IsoG treatment in MEFs (Fig. 6a, b), we find that knockdown of XAB2 triggers the formation of R-loops in cells; importantly, treatment of cells with RNase H post-fixation led to a substantial decrease in R-loops confirming the validity of these findings (Fig. 6c,   Fig. S5A). Treatment of cells with IsoG also led to the substantial increase of γΗ2Αx (+) 53BP1 (+) cells (Fig. 6d) as well as to higher γH2Ax protein levels compared to untreated controls ( Fig. 6e), indicating that a defect in splicing leads to persistent DNA damage accumulation. Likewise, RNAi-mediated ablation of XAB2 with two distinct siRNAs increased substantially the number of γΗ2Αx (+) 53BP1 (+) cells (Fig. 6f). As R-loops often lead to persistent DNA breaks, we next tested whether genome instability in siXab2 cells is a result of R-loop accumulation. Immunofluorescence studies revealed that protein transfection of RNase H in siXab2 cells leads to a decrease in R-loops (Fig. 6g) and in the number of γΗ2Αx (+) 53BP1 (+) cells (Fig. 6h).
XAB2 interacts with R-loop processing factors and is recruited to RNA:DNA hybrids under conditions that favor R-loop formation. Highly transcribed loci are prone to R-loop formation 50,51 . To explore the possibility of a direct association of XAB2 with R-loops and R-loop processing factors, independently of its role in splicing, primary MEFs were treated with alltrans-retinoic acid (tRA), a pleiotropic factor known to activate transcription during cell differentiation and embryonic development 52 . We find that transcription activation in tRAtreated MEFs leads to the substantial accumulation of R-loops in these cells, which becomes further pronounced when cells are treated with tRA and illudin S (Fig. 7a). DRIP analysis with primers spanning the promoter or 3′ end regions of genes ( Fig. 7b) where the frequency of RNA-DNA hybrids is expected to be higher revealed the significant accumulation of R-loops in tRA-induced gene targets i.e. Rarb2, Stra6 and in the intronless tRA-induced gene Fibin, but not in the tRA non-inducible Chordc gene (Fig. 7b, c). Further DRIP analysis revealed that R-loops accumulate in the promoter regions of tRA regulated genes when cells are treated with tRA or with tRA and either illudin S or UV irradiation or DRB (Fig. S5B). Instead, we find that the levels of R-loops in cells decrease after tRA/DRB-treatment or tRA/UV irradiation in the gene bodies of tRA-induced Rarb2 and Stra6 genes (Fig. S5C). To confirm that XAB2 and R-loops co-occupy the tRA-responsive Stra6 gene and the tRA non-inducible Chordc gene, we next employed native bioXAB2 pulldown approach followed by S9.6 DRIP in tRA-and tRA/illudin S-treated MEFs. This strategy revealed that bXAB2 is recruited preferentially on the 3′ end region of Stra6 gene compared to the tRA noninducible Chordc gene in tRA-treated MEFs and that bioXAB2/ DRIP signals are higher when cells are treated with illudin S (Fig. 7d). Native ChIP signals prior to S9.6 DRIP revealed that bXAB2 is recruited on the coding region of the tRA-induced Stra6 gene but not on the coding region of the tRA non-inducible Chordc1 gene (Fig. S5D). Similar to the DNA damage-driven release of XAB2 from RNA targets (Fig. 5a, b), XAB2 ChIP signals are substantially reduced when tRA-treated MEFs are also exposed to illudin S accumulating transcription-blocking DNA lesions (Fig. S5D). In agreement with the role of XAB2 in NER (Fig. 3a-e), we find that the association of XAB2 with R-loops is significantly reduced in the promoter regions of tRA-responsive genes Stra6 and the intronless Fibin and Sstr4 genes in tRA/UVtreated cells when compared to non-irradiated, tRA-treated control cells (Fig. S5E-F). Instead, the association of XAB2 with R-loops remained unaltered in the tRA non-inducible Chordc gene (Fig. S5E). Importantly, the decrease in the association of XAB2 with R-loops on tRA/UV-treated cells was not due to a    (Fig. S5B). Further work revealed that XAB2 dissociates from R-loops in tRA-treated bXAB2/Csb m/m and bXAB2/Xpc −/− MEFs exposed to UV irradiation (Fig. S5G); in both genotypes, the levels of R-loops were comparable between tRA/UV-treated cells and tRA-treated, nonirradiated controls (Fig. S6A). Lastly, we find that transfection of illudin S-treated MEFs with recombinant RNaseH1 restores the release of XAB2 from UsnRNAs (Fig. S6B). R-loops are known to be actively processed by the NER structure-specific endonucleases XPF and XPG 53,54 . PRP19 may act as a sensor of RPA-ssDNA moieties that emerge naturally when R-loops are formed during ongoing transcription 21 . Using a series of bXAB2 pulldown under basal transcription conditions or upon tRA treatment in primary MEFs, we find that XAB2 interacts with XPF, XPG, and PRP19 proteins (Fig. 7e, Fig. S6C). Unlike with PRP19, we find that the interaction of XAB2 with XPG or XPF is abolished when cells are treated with RNase H, indicating that the XAB2-XPF-XPG complex requires the presence of R-loops (Fig. 7e, Fig. S6C). The interaction of XAB2 with bXPF and XPG independently of DNA was also evident with bXPF pulldowns in tRA-treated cells albeit to a lesser extent (Fig. S6D). Instead, we find that the interaction of XAB2 with XPF and XPG is lost in tRA-treated cells exposed to UV irradiation (Figs. S6E, S6F). S9.6 DRIP-Western analysis under basal transcription conditions or upon tRA treatment reveals that the XAB2 complex is recruited on R-loops (Fig. 7f, Fig. S6G). A similar analysis reveals that the XAB2 complex is released from R-loops in UV-irradiated MEFs (Fig. 7g) as well as in tRA/UVtreated MEFs (Fig. S6H) indicating the high affinity of XPF and XPG endonucleases for UV-induced DNA lesions. Importantly, neither XPF nor XPG associate with R-loops when XAB2 is silenced (Fig. 7h). Since the recruitment of XPF and XPG on Rloops requires XAB2 and both endonucleases were previously shown to process R-loops into DNA breaks 54,55 , it is surprising that Xab2 knockdown leads to an increase in the number of γH2AX + 53BP1+ cells. Alternatively, these findings could reflect the known role of XAB2 in HR 24 or the possibility that the accumulation of R-loops in siXab2 cells could spontaneously lead to DNA breaks due to transcription-DNA replication conflicts in dividing cells 56 . In support, we find that the number of γH2AX + 53BP1+ cells (due to R-loops alone), is reduced in serum-starved MEFs compared to proliferating corresponding control cells (Fig. 7i, Fig. S6I, Fig. S7A). To test whether XAB2 interaction with XPF and XPG on R-loops requires functional NER, we also performed S9.6 DRIP in wt. and totally NER-defective Xpa −/− MEFs (Fig. S7B-C; as indicated) followed by western blotting for XPA (for wt cells), XPF and XPG. We find that XPA does not associate with R-loops and that recruitment of NER endonucleases on R-loops does not require XPA. Nevertheless, we find that R-loops accumulate in tRA-treated Xpa −/− MEFs compared to wt. controls (Fig. S7D) likely reflecting the contributing role of irreparable DNA lesions (owing to the NER defect in Xpa −/− cells) on R-loop accumulation. In support, we find an enrichment of R-loops on the promoters of highly transcribed Igf1 and Ghr genes in P15 Csb m/m livers compared to littermate controls (Fig. S7E). Similar to the reduction of XAB2 ChIP signals in illudin S-treated primary MEFs (Fig. S5D) we find that bXAB2 ChIP signals are substantially reduced on Ghr and Igf1 gene promoters in P15 bXAB2/Csb m/m livers compared to age-matched bXAB2 control livers (Fig. S7F). Finally, a sequential native bXAB2 pulldown followed by S9.6 DRIP in P15 bXAB2 and bXAB2/Csb m/m livers revealed that XAB2 interacts with Rloops on the promoters of Igf1 and Ghr genes in vivo (Fig. 7j). Thus, XAB2 interacts with XPF and XPG endonucleases outside NER and is recruited to RNA:DNA hybrids under conditions that promote R-loop formation (Fig. 7k).

Discussion
Inborn defects in RNA processing or DNA repair associate with progeria 27,28,57 , metabolic defects 58-63 , neurodegeneration [64][65][66][67] , and cancer [68][69][70][71][72] , arguing for functional links between genome maintenance and the splicing machinery in development or disease. Using an in vivo biotinylation tagging approach in mice, we find that XAB2 interacts with components of the core spliceosome as well as with DDB1, involved in UV-induced DNA damage recognition and XPA that plays a prominent role in the assembly of the NER incision complex at sites of DNA damage 67,73 . As DDB1 is successfully recruited to UV-induced CPDs in siXab2 cells, the implication of XAB2:DDB1 interaction remains elusive. However, in siXab2 cells, XPA fails to form visible foci and to be recruited to UV-induced DNA lesions. In support of a defect in NER, siXab2 cells have decreased UDS and delayed RNA synthesis recovery following UV-induced DNA damage. Consistently, UV-induced CPDs remain unrepaired even 24 h post-UV irradiation in siXab2 cells. Arguably, the defect in splicing does not interfere with the DNA repair defect in siXab2 cells because exposure of cells to a pre-mRNA splicing inhibitor has no detectable impact on the repair kinetics of UV-induced CPDs. We find that XAB2 is recruited to UsnRNAs and pre-mRNAs and that it is released from all RNA targets tested when cells are treated with UV-induced DNA damage or with the genotoxin illudin S. Intriguingly, the release of XAB2 from RNA following DNA damage is not rescued when ATM or ATR are inhibited but only when UV-induced CPDs are efficiently removed through PR-dependent repair. The latter supports previous observations showing that DNA damaging agents affect the subcellular localization of splicing factors or their association with Fig. 4 XAB2 is part of a core spliceosome complex. a bXAB2 pulldowns and western blot analysis of AQR, FLAG-tagged XAB2 (FLAG), PRP19, and BCAS2 proteins in P15 bXAB2 livers. b RNA pulldowns with bXAB2 and BirA (control) on U1-U6 UsnRNAs (as indicated). U6 p-value calculated by one-tailed Student's t-test. c RNA immunoprecipitation with XAB2, Aquarius and BCAS2 on U4 and U6 snRNAs (as indicated). d Schematic representation of splicing decisions, i.e., exon skipping, intron retention, and alternate (Altern.) selection of 5′ or 3′ splice sites. e, g Graph depicting the differences in Percent Spliced-In (PSI) between siXab2 and siScrb HEPA and mESCs cells respectively (ΔPSI > 0: increased alternative exon/intron inclusion in siXab2 and ΔPSI < 0: decreased alternative exon/intron inclusion) in siXab2 compared to siScrb cells. f, h Distribution of the direction of alternative splicing changes between siXab2 and siScrb HEPA and mESCs cells, respectively. Represented is the observed proportion of positive (ΔPSI > 0, in gray) values for each event type, with the error bar reflecting its 95% confidence interval, based on a Pearson's chi-squared proportion test as explained in "Methods" (i). Pre-mRNA levels and (j) mRNA levels of transcripts with retained introns in siXab2 and siScrb HEPA cells at 48 and 72 h post-transfection (as indicated). The images shown in Fig. 4a, e, g are representative of experiments that were repeated three times. The RIP signals are shown as fold enrichment (F.E) of % input antibody or bXAB2 over % input control antibody (IgG) or BirA. All scatter and bar blots in this manuscript are presented as mean ± SEM. P values were calculated by two-tailed Student's t test, unless otherwise indicated. Error bars indicate SEM among three biological replicates in all cases. Source data provided as a Source Data file. nascent transcripts [74][75][76][77] and that ATM (and likely ATR) function downstream of the spliceosome displacement 11 . Nascent pre-mRNAs that associate with transcription complexes blocked at DNA lesions are released from late-spliceosome components and hybridize to the template strand of the melted DNA to form RNA:DNA duplexes (R-loops), a structure that slows transcription and activates DDR 77 . Increased formation of R-loops is frequently observed in cells that are defective in DNA repair or RNA processing [47][48][49] leading to DNA breaks 54 , transcription stalling, and DDR activation 77 . In support, we find that R-loops accumulate in primary MEFs following DNA damage or upon treatment with a pre-mRNA splicing inhibitor. Knockdown of Xab2 leads to the substantial increase in R-loop formation and to an R-loop-dependent accumulation of γΗ2ΑX and 53BP1 in   cells. Indeed, the number of γΗ2ΑX (+) 53BP1 (+) cells is significantly reduced when siXab2 cells are protein transfected with RNaseH1 known to eliminate R-loops. Using a series of IP strategies, we find that XAB2 interacts with XPG and XPF endonucleases and that the XAB2 is required for the trimeric protein complex to be recruited on RNA:DNA hybrids under conditions that favor R-loop formation. The latter supports previous observations indicating that the NER endonucleases XPF and XPG actively process unscheduled R-loops into DNA breaks 54,78 .
The interaction of XAB2 with XPG or XPF as well as the recruitment of the XAB2 complex on RNA:DNA hybrids are also observed in NER-defective Xpa −/− MEFs, indicating that R-loop processing does not require functional NER. Similar to RNA, transcription-blocking DNA lesions trigger the release of XAB2 from DNA sites. However, we find that bXAB2 ChIP signals followed by DRIP are substantially higher on the 3′ end region of the tRA-inducible Stra6 gene and increase even further when cells are also treated with illudin S. Likewise, DRIP signals followed by bXAB2 ChIP are higher on the R-loop-enriched promoter regions of Igf1 and Ghr genes in P15 Csb m/m livers indicating that XAB2 preferentially binds on DNA sites that associate with RNA. Taken together, our findings suggest that the splicing factor XAB2 is released from UsnRNAs and pre-mRNAs and together with XPG and XPF, it is recruited on RNA:DNA hybrids under conditions that promote R-loop formation, e.g., transcription induction or DNA damage (Fig. 7k). The finding that XAB2 is released from RNA targets in progeroid Csb m/m developing livers makes it attractive to test whether DNA damage-driven changes in RNA processing are causal to the premature onset of age-related pathological symptoms seen in TC-NER progeroid syndromes and during natural aging.

Methods
Generation of biotin-tagged XAB2 animals. To generate the targeting vector for the insertion/knock-in of the Avi tag cassette before the stop codon of the last exon of the XAB2 gene for the generation of the avXAB2 knock-in mice, PCR products were first amplified using Phusion High-Fidelity DNA Polymerase (NEB Offspring were screened by PCR for neo-deletion using primers F1 and R1 (Fig. 1a, supplementary Table 1). Expression of Protamine-1 Cre transgene in the male germline resulted in the deletion of the floxed neomycin gene in all the first pups born, leaving behind a single loxP site after the avi tag cassette. The Cre recombinase transgene, derived from the PC3 embryonic stem cell background, was bred out in the process of backcrossing to C57BL/6 mice. Biotin-tag XAB2 knock-in mice were further crossed to transgenic BirA transgenic mice 31 . Mice were kept on a regular diet and housed at the IMBB animal house, which operates in compliance with the 'Animal Welfare Act' of the Greek government, using the 'Guide for the Care and Use of Laboratory Animals' as its standard. As required by Greek law, formal permission to generate and use genetically modified animals was obtained from the responsible local and national authorities. The independent Animal Ethical Committee at FORTH approved all animal studies.
Mass spectrometry (MS) studies. For MS studies, nuclear protein extracts derived from livers of 15-day-old BirA or bXAB2 mice (n = 3) were prepared using the high-salt extraction method (10 mM HEPES-KOH pH 7.9, 380 mM KCl, 3 mM MgCl2, 0.2 mM EDTA, 20% glycerol and protease inhibitors). Nuclear extracts derived from BirA animals were used as negative controls. Nuclear lysates were diluted threefold with HENG buffer and digested O/N with benzonase. For pulldown assays used in MS, 0.6-0.7 mg of nuclear extracts were incubated with 40 μl M-280 paramagnetic streptavidin beads (Invitrogen); streptavidin interacts with the biotin moiety that is added by the BirA ligase on the short 15aa biotinylation Avitag of XAB2. Proteins eluted from the beads were separated by SDS/PAGE electrophoresis on an 10% polyacrylamide gel and stained with Colloidal blue silver (ThermoFisher Scientific, USA 79 ); SDS-PAGE gel lanes were cut into 2-mm slices. Gel areas that were very dense in protein content or contained a significant amount of a particular protein were cut in smaller slices. Gel slices were subjected to in-gel reduction with dithiothreitol, alkylation with iodoacetamide and digested with trypsin (sequencing grade; Promega) 80,81 ; the 'in gel" strategy allows the possibility to verify the molecular weight of the identified proteins, thereby minimizing any false positive hits. Nanoflow liquid chromatography-tandem mass spectrometry (nLC-MS/MS) was performed on an EASY-nLC coupled to an Orbitrap Fusion Tribid mass spectrometer (Thermo) operating in positive mode. Peptides were separated on a ReproSil-C18 reversed-phase column (Dr. Maisch; 15 cm × 50 μm) using a linear gradient of 0-80% acetonitrile (in 0.1% formic acid) during 90 min at a rate of 200 nl/min. The elution was sprayed into the electrospray ionization (ESI) Fig. 5 XAB2 is released from U snRNPs and pre-mRNAs upon DNA damage. a RNA immunoprecipitation of XAB2 on U4, U5, and U6 snRNAs at 2, 6, and 24 h post-UV irradiation. b RNA immunoprecipitation of XAB2 on pre-mRNAs in untreated and UV-irradiated HEPA cells (as indicated). c bXAB2 pulldowns followed by western blot analysis with FLAG (for bXAB2), PRP8, PRP19, BCAS2, and SNRP40 in UV-irradiated (10 J/m 2 ) and control MEFs. The images shown on Fig. 5c are representative of experiments that were repeated three times. d RNA immunoprecipitation of XAB2 on UsnRNAs in untreated, UV-irradiated cells and UV-irradiated cells pre-treated with inhibitors against ATM (ATMi) and ATR (ATRi; as indicated). e Immunofluorescence detection of UV-induced CPDs in UV-irradiated photolyase transgenic MEFs exposed to 1 h of PR light (CPDs are repaired) or dark (CPDs remain) as indicated. The images shown are representative of experiments that were repeated three times. f RNA immunoprecipitation (RIP) of XAB2 on U4 and U6 snRNAs in UVirradiated photolyase transgenic MEFs exposed to 1 hour of PR light (CPDs are repaired) or dark (CPDs remain); g RNA immunoprecipitation (RIP) of XAB2 on pre-mRNAs in UV-irradiated photolyase transgenic MEFs exposed to 1 h of PR light (CPDs are repaired) or dark (CPDs remain); h bXAB2 RNA pull downs on U4 and U6 snRNAs in untreated or illudin S-treated BirA and bXAB2 MEFs; i bXAB2 RNA pull downs on U4 and U6 snRNAs in P15 bXAB2 or bXAB2;Csb m/m mouse livers. RIP signals are shown as fold enrichment (F.E) of % input of antibody or bXAB2 over % input of control antibody (IgG) or BirA. All scatter and bar blots in this manuscript are presented as mean ± SEM. P values were calculated by two-tailed Student's t test, unless otherwise indicated. Error bars indicate SEM among three biological replicates in all cases. Source data provided as a Source Data file.
source of the mass spectrometer. Spectra were acquired in continuum mode; fragmentation of the peptides was performed in data-dependent mode by HCD.
LC-MS/MS data analysis. Raw mass spectrometry data were analyzed with the MaxQuant software suite 82 (version 1.6.0.16) 81 with the additional options 'LFQ' and 'iBAQ' selected. A false discovery rate of 0.01 for proteins and peptides and a minimum peptide length of seven amino acids were used to discern the significant data obtained from bXAB2 and BirA mouse livers. The Andromeda search engine was used to search the MS/MS spectra against the Uniprot database (taxonomy: Mus musculus, release September 2017), concatenated with the reversed versions of all sequences. A maximum of two missed cleavages was allowed. The peptide tolerance was set to 10 ppm and the fragment ion tolerance was set to 0.6 Da for HCD spectra. The enzyme specificity was set to trypsin and cysteine a.
h. carbamidomethylation was set as a fixed modification. In case the identified peptides of two proteins were the same or the identified peptides of one protein included all peptides of another protein, these proteins were combined by Max-Quant and reported as one protein group. The generated 'proteingroups.txt' table was filtered for contaminants and reverse hits. For bXAB2 interactor identification, a t-test-based statistics was applied on LFQ of proteome derived from bXAB2 and BirA livers. First, the logarithm (log 2) of the LFQ values were taken, resulting in a Gaussian distribution of the data. This allowed imputation of missing values by a normal distribution (width = 0.3, shift = 1.8), assuming these proteins were close to the detection limit. Statistical outliers for the XAB2 samples compared to BirA control were then determined using a two-tailed t-test. Multiple testing correction was applied by using a permutation-based false discovery rate (FDR) method in Perseus. Immunofluorescence, antibodies, westerns blots, and FACS. Immunofluorescence experiments were performed as previously described 31,60 . Briefly, cells were fixed in 4% formaldehyde, permeabilized with 0.5% Triton-X and blocked with 1% BSA. After 1-h incubation with primary antibodies, secondary fluorescent antibodies were added and DAPI was used for nuclear counterstaining. Samples were imaged with SP8 confocal microscope (Leica). For local DNA damage infliction, cells were UV-irradiated (60 J m −2 ) through isopore polycarbonate membranes containing 3-μm-diameter pores (Millipore Differential AS analysis. Global quality of FASTQ files with raw RNA-seq reads was analyzed using fastqc v.0.11.5 (https://www.bioinformatics.babraham.ac.uk/ projects/fastqc/). Vast-tools 83,84 aligning and read processing software was used for quantification of alternative sequence inclusion levels from FASTQ files using VASTD-DB annotation 85 for mouse genome assembly mm9. To ensure sufficient read coverage for AS quantification, we considered only vast-tools' events with a minimum mapability-corrected read coverage score of "VLOW" across all samples. The relative inclusion of an alternative sequence (exon or intron), henceforth called AS event, is based on the number of junction reads supporting inclusion (#inc) and exclusion (#exc), used to quantify percent spliced-in (PSI) 86 values. The beta distribution (conjugate prior probability distribution for the binomial), constrained in the] 0,1 [interval and characterized by a mean value of α/(α + β) from the distribution's shape parameters α and β, was exploited in modeling the precision of each PSI from its supporting coverage (#inc and #exc). We used R function rbeta to emit, for each sample, 500 values from a beta distribution with α = #inc + 1 and β = #exc + 1, where 1 is added to ensure that both the shape parameters, α and β, are different from zero. Since beta distributions get narrower with increasing shape parameters (for the same PSI value), this parametrization allows the scattering of emitted values to serve as a surrogate for that PSI's dispersion (given the original read coverage supporting it), while the distribution's mean value, (#inc + 1)/(#inc h.

ARTICLE
of each ΔPSI was set as the ratio between the number of differences that are greater than zero and the total number of differences, reflecting the probability of PSI Xab2 siRNA being greater than PSI Control . Differentially spliced events were considered as those with a probability of a |ΔPSI| > 0 > 0.8 and an absolute ΔPSI g > 5% (Table EV3) RNA-Seq and quantitative PCR studies. Total RNA was isolated from cells using a Total RNA isolation kit (Qiagen) as described by the manufacturer. For RNA-Seq studies, libraries were prepared using the Illumina® TruSeq® mRNA stranded sample preparation Kit for HEPA cells and for mESCs. Library preparation started with 1 µg total RNA. After poly-A selection (using poly-T oligo-attached magnetic beads), mRNA was purified and fragmented using divalent cations under elevated temperature. The RNA fragments underwent reverse transcription using random primers. This is followed by second-strand cDNA synthesis with DNA polymerase I and RNase H. After end repair and A-tailing, indexing adapters were ligated. The products were then purified and amplified to create the final cDNA libraries. After library validation and quantification (Agilent 2100 Bioanalyzer), equimolar amounts of all 12 libraries were pooled. The pool was quantified by using the Peqlab KAPA Library Quantification Kit and the Applied Biosystems 7900HT Sequence Detection System. The pool was sequenced by using an Illumina HiSeq 4000 sequencer with a paired-end (2 × 75 cycles) protocol. Quantitative PCR (Q-PCR) was performed with a Biorad 1000-series thermal cycler according to the instructions of the manufacturer (Biorad) as previously described 31 . All relevant data and primer sequences for the genes tested by qPCR are available upon request.
DRIP. DRIP analysis was based on ChIP analysis with some modifications. DRIP analysis was performed without a cross-linking step. Nuclei were isolated using 0.5% NP-40 buffer. Isolated nuclei were resuspended in TE buffer supplemented with 0.5%SDS and 100mg proteinase K. Genomic DNA was isolated after the addition of potassium acetate (1 M) and isopropanol precipitation. DNA was sonicated on ice 3 min using Covaris S220 Focused ultrasonicator. Samples were treated with RNase H (10 units/5 μg DNA) at 37°C overnight. Samples were immunoprecipitated with S9.6 antibodies (8 μg antibody/5 μg DNA) overnight at 4°C followed by incubation for 3 h with protein G-Sepharose beads (Millipore) and washed sequentially. The complexes were eluted and purified DNA fragments were analyzed by qPCR using sets of primers targeting different regions of related genes.
Sequential native ChIP analysis. Non-cross-linked cells were lysed using 0.5% NP-40 buffer. Chromatin was digested with MNase (50 Units/0.5mgDNA) at 37°C for 10 min. S1 chromatin fraction was isolated. S2 chromatin fraction was dialyzed against Tris-EDTA buffer for 2 h and isolated. S1 and S2 chromatin fractions were used for pulldown using M280 paramagnetic streptavidin beads (Invitrogen). After sequential washes complexes were eluted. A fraction of them was kept for qPCR analysis and the rest was used for S9.6 IP as described above.
Gel mobility shift assays. The XAB2 complex or recombinant Xab2 (0,1-1-5-10 nmoles; Abnova) was incubated with the biotinylated RNA probes (100 pmoles) in binding buffer (10 μl) containing 10 mm Tris-HCl (pH 8.0), 50 mm KCl, 10% Fig. 7 High transcription promotes XAB2 interaction with R-loops. a Immunofluorescence detection of R loops in untreated, tRA-treated (4 h) and illudin S-treated (1.5 h) MEFs (untreated ∓RNH: n = 975/705 cells, tRA treated ∓RNH: n = 1488/750 cells, illudin ∓RNH: n = 1500/730 cells/4 independent experiments.). b-c DRIP analysis of tRA-responsive (Rarb2, Fibin, and Stra6) and non-responsive (ChordC) gene promoters (P) or coding regions (c) w/o RNHin untreated and tRA-treated MEFs. p values between (−) and (+) RNH treated were calculated with one-tailed Student's t test. d Native ChIP followed by DRIP on the tRA-induced Stra6 and the tRA-non induced Chordc1 coding regions in tRA and illudin S/tRA-treated BirA and bXAB2 MEFs. e bXAB2 pulldowns (PD) and western blotting for XPF, XPG, and PRP19 in tRA-treated bXAB2 and BirA MEF nuclear extracts w/o RNH. f-h S9.6 immunoprecipitation followed by WB for XAB2, XPG, XPF and PRP19 in (f) tRA-treated or (g) UV-irradiated (10 J/m 2 ) MEF nuclear extracts w/o RNH and in (h) tRA-treated siScrb or si1Xab2 MEF nuclear extracts. i Immunofluorescence detection of γH2AX and 53BP1 (white arrowheads) in wt. MEFs cultured w/o fetal bovine serum (FBS) and transfected (RNH+) or not (RNH-) with RNaseH1. (+FBS ∓RNH: n = 1070/774 cells, −FBS ∓RNH: n = 785/831 cells/3 independent experiments) (j). Native ChIP followed by DRIP on the Ghr and Igf1 promoters in P15 wt. and Csb m/m livers. k DNA damage inhibits RNA synthesis. XAB2 is released from RNA targets leading to aberrant intron retention and R-loop accumulation. XAB2 interacts with ERCC1-XPF and XPG and the complex is recruited on RNA:DNA hybrids, linking R-loop processing with the spliceosomal response to DNA lesions. DRIP signals are shown, as fold enrichment (F.E) of % input of antibody over % input of control antibody (IgG). The images shown in Fig. 7e-h are representative of experiments that were repeated three times. All scatter and bar blots in this manuscript are presented as mean ± SEM. p values were calculated by two-tailed Student's t test, unless otherwise indicated. Error bars indicate SEM among three biological replicates in all cases, unless otherwise stated. Source data are provided as a Source Data file. glycerol, 0.5 mm EDTA, 0.5 mm dithiothreitol, 1 μg/ml BSA and 6 mm MgCl 2 for 30 min at 4°C as previously described 20 . After the incubation, the binding reaction mixtures are separated on a non-denatured 4% polyacrylamide gel in 0.5× TBE buffer, pH:7,5. Complexes and free oligos are transferred on a nylon membrane and UV cross-linked. The presence of biotin was detected with stp-HRP.
Primer sequences. Primer sequences are provided in Supplementary Table 1.
Quantification and statistical analysis. A two-tailed t-test was used to extract the statistically significant data by means of the IBM SPSS Statistics 19 (IBM) and the R software for statistical computing (www.r-project.org). Significant overrepresentation of pathways and gene networks was determined by GO (http:// geneontology.org/). Data analysis is discussed also in the Method Details section. Experiments were repeated at least three times. The data exhibited normal distribution (where applicable). There was no estimation of group variation before experiments. Error bars indicate standard deviation unless stated otherwise (standard error of the mean; s.e.m.). For animal studies, each biological replicate consists of 3-5 mouse tissues or cell cultures per genotype per time point or treatment. No statistical method was used to predetermine sample size. None of the samples or animals was excluded from the experiment. The animals or the experiments were non-randomized. The investigators were not blinded to allocation during experiments and outcome assessment.