MEN1 is a regulator of alternative splicing and prevents R-loop-induced genome instability through suppression of RNA polymerase II elongation

Abstract The fidelity of alternative splicing (AS) patterns is essential for growth development and cell fate determination. However, the scope of the molecular switches that regulate AS remains largely unexplored. Here we show that MEN1 is a previously unknown splicing regulatory factor. MEN1 deletion resulted in reprogramming of AS patterns in mouse lung tissue and human lung cancer cells, suggesting that MEN1 has a general function in regulating alternative precursor mRNA splicing. MEN1 altered exon skipping and the abundance of mRNA splicing isoforms of certain genes with suboptimal splice sites. Chromatin immunoprecipitation and chromosome walking assays revealed that MEN1 favored the accumulation of RNA polymerase II (Pol II) in regions encoding variant exons. Our data suggest that MEN1 regulates AS by slowing the Pol II elongation rate and that defects in these processes trigger R-loop formation, DNA damage accumulation and genome instability. Furthermore, we identified 28 MEN1-regulated exon-skipping events in lung cancer cells that were closely correlated with survival in patients with lung adenocarcinoma, and MEN1 deficiency sensitized lung cancer cells to splicing inhibitors. Collectively, these findings led to the identification of a novel biological role for menin in maintaining AS homeostasis and link this role to the regulation of cancer cell behavior.

( 1 ). The discovery of AS led to the deciphering of how the metazoan proteome can come to possess different biological functions and tremendous complexity from a small number of genes. Nearly 90% of human pre-mRNAs undergo AS, generating mRNA isoforms that tend to be differentially expressed across distinct cells or tissues and de v elopmental stages ( 2 ). The AS process is highly contr olled thr ough a variety of mechanisms, including the interaction of splicing factors with cis -acting elements in the pre-mRNA sequence, the rate and pausing of RN A pol ymerase II (Pol II) elongation and epigenetic modification of template chromatin ( 3 ). Dysregulation in the processing of alternati v ely spliced pre-mRN A has been closel y linked to various human diseases, especially cancers. A comprehensi v e analysis re v ealed that nearly all cancer tissues exhibit abnormal AS profiles relati v e to their healthy tissue counterparts ( 4 ). An increasing number of studies have shown that tumor-specific splicing variants participate in the regulation of cellular processes, including proliferation ( 5 ), invasion ( 6 , 7 ), apoptosis ( 8 ), drug resistance ( 9 ) and metabolism ( 10 ). Thus, identifying tumor-specific AS and elucidating oncogenic mechanisms provide new opportunities for the de v elopment of cancer therapeutics.
Alternati v e pre-mRNA splicing depends on the precise recognition of splice sites and selecti v e intron e xcision, processes that are mediated by the spliceosome, a dynamic ribonucleoprotein (RNP) complex consisting of five small nuclear RNAs (snRNAs; U1, U2, U4, U5 and U6) and a wide variety of proteins ( 11 ). The U1 and U2 small nuclear RNPs (snRNPs) are critical for the identification of splice sites, facilitating the recruitment of the U4 / U5 / U6 snRNP complex and various non-snRNPs. When U1 / U4 snRNPs are subsequently displaced, the conformational changes and two splicing transesterification reactions are mediated through a catal yticall y acti v e spliceosome consisting of U2, U5 and U6 snRNPs ( 12 ). Altered expression or disorderly assembly of spliceosome components can lead to pathological changes in AS in response to external dama ge. DNA dama ge has been proposed to impact the choice of splice site by disrupting spliceosome mobilization ( 13 ), and abnormal AS induced by DNA damage has been attributed to changes in the Pol II elongation rate or disrupted interactions between Pol II and splicing factors ( 14 ). Dysregulated recruitment of splicing factors to Pol II triggers DNA double-strand breaks (DSBs) ( 15 ). W hen aberrant regula tors of splicing hinder the repair of DN A damage, carcino genesis and tumor pro gression can result ( 16 ). Observations of aberrant AS processes and human diseases, once rare, have increased, yet understanding of the molecular mechanisms and biological significance of AS regulation in tumorigenesis remains in its infancy.
The tumor suppressor gene MEN1 ( Men1 in mice) encodes the nuclear protein menin whose inactivation causes the de v elopment of multiple endocrine neoplasia type 1 (MEN1) ( 17 , 18 ). Functional studies have shown that MEN1 is e xtensi v ely involv ed in the regulation of genome stability ( 19 , 20 ), the DNA damage response (DDR) ( 21 ) and multiple cellular behaviors ( 22 ). Menin binds a variety of DNA structures, including Y-structures, branched structures and f our-wa y junction structures, in a sequence-independent manner ( 23 ). Mechanistic analyses have demonstrated that menin is a bona fide transcription factor that binds promoter regions to activate the transcription of multiple genes ( 22 ); in fact, transcription factors recruited to promoter sequences influence the AS process ( 24 ). In line with its potential role in RNA splicing, menin localizes to SRSF2 nuclear speckles ( 25 ), which are involved in the regulation of alternati v e and constituti v e splicing (26)(27)(28). These observations suggest a connection (that remains mechanistically obscure) between MEN1 and the pre-mRNA splicing process.
Her e, we r eport a pr eviousl y una ppreciated function of MEN1 in regulating alternati v e pre-mRNA splicing. MEN1 -regula ted alterna ti v ely spliced genes contain weak 5 or 3 splice sites. In this study, mechanistically speaking, MEN1 altered exon skipping and the abundance of splicing isoforms by slowing the rate of Pol II elongation. MEN1 deletion led to R-loop formation and DNA damage accumulation, thereby inducing genome instability. Our results suggest that splicing inhibitors may be thera peuticall y beneficial for MEN1 -inactivated cancers.

Generation of men1 knockout mice
Mice with whole-body expression of Ubc-Cre were crossed with mice harboring floxed alleles of Men1 to obtain conditional tamoxifen (TAM)-inducible Men1 flo x / flo x ( Men1 f / f ); Ubc-Cre mice. Homozygous floxed Men1 f / f mice without Cre were used as controls. TAM was dissolved in corn oil containing 10% ethanol. At 4 weeks of age, the Men1 f / f ; Ubc-Cre and Men1 f / f mice recei v ed intraperitoneal (i.p.) administration of 100 mg / kg TAM (T5648-1G, Sigma) once a day for 5 days to generate Men1 KO ( Men1 / ) mice. Polymerase chain reaction (PCR) was used to verify the genotype. Mice were housed under standard conditions with a light / dark cycle of 12 h and free access to food and water. Animal experiments were performed in accordance with animal welfare and American Veterinary Medical Association (AVMA) Guidelines for the Euthanasia of Animals (2020), and the procedures were approved by the Institutional Animal Care Committee of Guizhou Medical Uni v ersity (no. 2100027).
Nucleic Acids Research, 2023, Vol. 51, No. 15 7953 RNA sequencing and ASE analysis RNA-seq was performed by Wuhan IGENEBOOK Bio-Tech Co., Ltd. Total RNA was extracted from lung tissue from 4-month-old Men1 f / f and Men1 / mice ( n = 3 biological replicates per group) or MEN1 -WT and MEN1 -KO NCI-H460 cells ( n = 2 biological replicates per group) using an RN A pr ep Pur e Kit (DP432, Tiangen) following the manufacturer's instructions. RNA was assessed for quality and yield with a Qsep1 instrument (BiOptic). A 1 g aliquot of total RNA was purified with poly(A) oligo-attached magnetic beads and fragmented to ∼150 nt in size, and cDNA was synthesized with random hexamer primers. RNA-seq was performed on an Illumina No-vaSeq6000 platform with an average depth of ∼30 million, 150 nucleotide paired-end reads per sample. With base calling, raw image data files were converted into sequenced r eads, r eferr ed to as raw data or raw reads. The adapter and low-quality reads were filtered in cutadapter (version 1.11) to generate clean reads. The clean reads were aligned to the mm9 mouse or GRCh38 human r efer ence genome with STAR (version 3.2.5) ( 29 ). Transcripts were assembled in StringTie (version 2.0.4) ( 30 ), followed by estimates of raw gene counts in featureCounts (version v1.6.0) ( 31 ) and FPKM (fragments per kilobase of transcript per million mapped reads) normalization ( 32 ). Differentially expressed genes (DEGs) were identified with DESeq2 with a filter threshold of adjusted P < 0.05 and fold change ≥ 1.5.
Next, we conducted AS analysis using multivariate analysis of transcript splicing (rMATS) (version 3.2.5) ( 29 ). We used the sorted BAM files generated by STAR to run rMATS using default unpaired procedures. To identify significant differential splicing e v ents, we set the following cut-of fs: false discovery ra te (FDR) < 0.05, | percent spliced-in ( PSI)| ≥ 0.1 and average junction reads per e v ent of each replicate ≥ 20. Modeling alternati v e junction inclusion quantification (MAJIQ) and Voila ( https:// biociphers.bitbucket.io/majiq/index.html ) were used to determine, quantify and visualize local splicing variants (LSVs) from the RNA-seq data ( 33 ). Briefly, the MAJIQ build tool used the BAM alignment files from STAR along with a gene annotation file to define splice diagrams, and MAJIQ PSI and PSI were used to calcula te rela ti v e le vels (PSI) of LSVs and changes in relati v e LSV abundance ( PSI) between the two conditions. For the MAJIQ data analysis, Voila TSV software was used to create a tabdelimited text file and then analyze specific LSVs or spliced genes of interest. Gene Ontology (GO) or Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis was performed on the AS e v ents (ASEs) identified as having differential splicing changes using clusterProfiler.

Real-time quantitative PCR (qPCR) and reversetranscription PCR (RT-PCR)
RNA preparation was conducted with TRIzol reagent (10606ES60, Yeasen), and cDNA synthesis was performed according to the instructions on the re v erse transcription kit (11141ES60, Yeasen). qPCR was carried out with TB Gr een ® Pr emix Ex T aq ™ (RR820A, T aKaRa) on a CFX Connect ™ Real-Time System (CFXConne, Bio-Rad). qPCRs were performed in at least three independent experiments, and relati v e gene e xpression was quantified based on the 2 − Ct method with r efer ence genes as a normalization indicated control. For RT-PCR assays, based on ASE analysis, primers for detecting exon skipping were designed in Primer Premier 5 for validation of RNA splicing. RT-PCR was performed with the following program: 95 • C for 2 min, 36 cycles of dena tura tion a t 95 • C for 30 s, annealing a t 56 • C for 30 s and elongation at 72 • C for 1 min, and a final elonga tion tempera ture of 72 • C for 5 min and 4 • C holding temperatur e. PCR products wer e run in 2% agarose gels stained with 0.1 ‰ ethidium bromide (EB) dye (C14141868, Macklin). Bands were visualized with a GleUV system (Baygene Biotech). Band intensities were quantified in ImageJ software.

Chromatin immunoprecipitation (ChIP)-qPCR assay
ChIP assays were performed according to the instructions for the Simple ChIP Kit (9003S, CST). Briefly, 1 × 10 6 cells were cross-linked with 1% (v / v) formaldehyde for 10 min at room temperature. Cell pellets were incubated in Buffer A for 10 min a t 4 • C . Pellet nuclei were collected by centrifugation at 2000 g and digested in Buffer B containing 25 U of micrococcal nuclease (MNase) (10011, CST) per immunoprecipitation (IP) for 20 min at 37 • C, followed by pulsed ultrasonication to shear cellular DNA, and harvested by centrifuga tion a t 12 000 g for 10 min. After quantification of chromatin DNA, equal amounts of chromatin were incubated overnight at 4 • C with the specific antibodies or nonimmune IgG as a negati v e contr ol. Pr otein A / G magnetic beads were used to couple the immunoprecipitated complexes for 2 h a t 4 • C , and then these bound complexes were rinsed e xtensi v ely with washing buffer. Next, DNA pulled down by the antibodies was purified on spin columns, and purified DNA was quantified by qPCR with TB Green Premix Ex Taq ™ on a CFX Connect Real-Time System. Values were calcula ted rela ti v e to the input and enriched relati v e to the signal obtained for IgG (set to 1). The following primary antibodies were adopted: menin (ab31902, Abcam, 6 g / IP), histone H3 (4620, CST, 3 g / IP), Pol II (ab264350, Abcam, 8 g / IP), p-Pol II (Ser2) (ab5095, Abcam, 5 g / IP), p-Pol II (Ser5) (2629, CST, 5 g / IP), U2AF65 (sc-53942, Santa Cruz, 5 g / IP) and SNRPA (10212-1-AP, Proteintech, 5 g / IP).

DN A-RN A immunoprecipitation (DRIP) assay
DRIP assays were performed according to a previously established protocol ( 34 ). Briefly, MEN1 -WT and MEN1 -KO NCI-H460 cells were rinsed three times in phosphatebuffered saline (PBS), resuspended in TE buffer (5 mM EDTA and 50 mM Tris-HCl, pH 8.0) containing 5 l of proteinase K, and lysed overnight at 37 • C by the addition of sodium dodecylsulfate (SDS) to a final concentration of 0.5%. Genomic DNA was extracted with the phenol / chlor oform pr ocedure and enzymatically digested overnight at 37 • C with a subset of restriction enzymes (20 U of EcoRI, 20 U of HindIII, 20 U of XbaI, 25 U of Sspl and 10 U of BsrGI) in buffer. Digested DNA was purified by standard phenol / chloroform extraction and ethanol precipitation as described previously. As a negative control, digested DNA was incubated with 10 U / ml RNase H (New England Biolabs) overnight at 37 • C. For DRIP experiments, 10 g of digested DNA was immunoprecipitated with 10 g of S9.6 antibody (MABE1095, Millipore) overnight at 4 • C in 1 × DRIP buffer (100 mM NaH 2 PO 4 , 1.4 M NaCl and 0.5% Triton X-100). A 20 l aliquot of protein G Dynabeads was pre-washed with 1 × DRIP buffer for 20 min at room temperature, and then these Dynabeads were added to couple the IP complexes for 2 h at 4 • C. Next, immunopr ecipitates wer e washed thr ee times with 1 × DRIP buf fer a t room tempera ture. After the last w ash, DNA w as eluted by addition of elution buffer containing 50 mM Tris-HCl (pH 8.0), 10 mM EDTA and 0.5% SDS, and incubated at 65 • C for 45 min. Finall y, DN A was purified using a Universal DNA Purification Kit (DP214-03, Tiangen) according to the manufacturer's instructions. Locus-specific DRIP signals were analyzed by qPCR, and the IP rate was expressed as the input percentage. Relati v e values with respect to MEN1 -WT cells without RNase H were also determined and plotted.
For RN A isolation, RN A was extracted from the chromatin pellet samples with TRIzol reagent according to the manufacturer's instructions. Briefly, these samples (500 l) wer e mix ed with 1 ml of TRIzol solution and extracted for 5 min by adding 200 l of chloroform. Then, the upper aqueous phase containing RNA was separated from the samples by centrifugation (12 000 g , 10 min, 4 • C) and transferred to a new 1.5 ml centrifuge tube for follow-up isopropanol precipitation and RN A isolation. Finall y, isolated RN A was pr e-tr eated with DNase I (9003-98-9, Merck), and 5 ng of RNA was re v erse-transcribed to synthesize template DNA. qPCR was conducted with primers complementary to human or mouse snRNAs or the chroma tin-associa ted HotAir ncRNA (control for data normalization) with TB Green Premix Ex Taq ™ on the CFX Connect Real-Time System. The absence of contaminating genomic DNA was validated by the lack of amplified products for all sample / primer sets by the inclusion of control re v erse transcription reactions in which enzyme was not added.

Lentivirus-mediated targeted gene short hairpin RNA (shRNA) knockdown and o ver expr ession
Lentiviral particles were generated by the transient transfection of HEK-293T cells following a standard protocol as described previously ( 36 ). Briefly, the MEN1 knockdown, fulllength MEN1 and RNase H1 sequence were constructed by inserting the corresponding sh MEN1 , MEN1 and RNase H1 sequences into the pLVX-CMV-ZsGreen-Puro vector and pCDH-CMV-MCS-EF1-Puro v ector, respecti v ely. Subsequently , Chemi-T rans ™ FectinBor DNA Transfection Reagent (T008, GeneCodex) was used to transfect the sh MEN1 , MEN1 and RNase H1 ov ere xpression, lenti viral skeleton and helper plasmids (pMD2.G, psPAX2) into HEK-293T cells. Lentiviral supernatant was harvested 72 h after transfection, and concentrated lentiviral medium containing 2 mg / ml polybrene was used to culture A549 cells or NCI-H446 cells. Infected cells were selected with 1 g / ml puromycin for 2 weeks to obtain stable cell lines. The same cell lines were infected with a lentiviral vector with the gr een fluor escent protein (GFP) gene as a control. The shRNA sequence specifically targeting MEN1 is 5 -GGAACCTGGCA GATCTA GA-3 .

CRISPR / Cas9-mediated MEN1 gene knockout
MEN1 -KO NCI-H460 cell lines were established using the CRISPR / Cas9 [clustered regularly interspaced palindromic repeats (CRISPR) / CRISPR-associated protein 9] system as previously described ( 37 ). Briefly, a single-guide RN A (sgRN A) sequence targeting exon 3 of the MEN1 locus (sg MEN1 sequence: 5'-CAAATT GGACAGCTCCGGT GT GG-3') was designed on the basis of http://crispr.mit.edu , synthesized from Sangon Biotech and annealed. Subsequently, sg MEN1 was digested using BsmBI (ER0452, Thermo Fisher Scientific) and ligated to the pX458 plasmid using a normal ligation r eaction as r ecommended by the manufactur er (2011A, TaKaRa). The pX458-sg MEN1 plasmid was transfected into NCI-H460 cells using electrotransfection. After 48 h of transfection, the cells were diluted by stepwise dilution to obtain single colonies. A GFP-positi v e cell cluster was selected to conduct genomic DNA sequencing, and cells displaying a single sequencing peak with a gap were considered candidate knockout cells.

Isolation of chromatin-associated proteins
After three washes in ice-cold PBS, cells (1 × 10 7 ) were resuspended in 200 l of Buffer A containing 10 mM HEPES (pH 7.9), 10 mM KCl, 1.5 mM MgCl 2 , 0.34 M sucrose, 10% glycerol, 1 mM dithiothreitol (DTT), 10 mM NaF, 1 mM Na 2 VO 3 and 1 × protease inhibitor cocktail. They were incubated for 10 min at 4 • C by the addition of Triton X-100 to a final concentration of 0.1%, followed by centrifugation at 1500 g for 5 min to separate the cytoplasmic proteins from the nuclei. Then, the extracted nuclei were lysed in 200 l of Nucleic Acids Research, 2023, Vol. 51, No. 15 7955 Buffer B containing 3 mM EDTA, 0.2 mM EGTA, 1 mM DTT and protease inhibitor cocktail. Insoluble chromatin was gathered by centrifugation at 2000 g for 5 min, rinsed once with Buffer B and centrifuged at 12 000 g for 5 min. For the release of chroma tin-associa ted proteins by MNase treatment, cell nuclei were resuspended in solution A and incubated with 0.25 U of MNase for 5 min at 37 • C. The nuclease reaction was stopped by the addition of 1 mM EGTA and 1 mM EDTA. Nuclei were collected by centrifugation at 1200 g for 5 min, and protein abundance was detected by immunoblotting.

Isolation of native non-cross-linked chromatin
Non-cr oss-linked chr omatin was pr epar ed as pr eviously described ( 38 ), except MNase pr e-tr eatment was substituted by a combination of DNase I digestion as recommended by the manufacturer and sonication for fiv e cy cles of 10 s on a JY92-IIN Ultrasonc Homogenizer (SCIETZ).

Cell proliferation assay
The indicated cells were seeded onto 96-well pla tes a t a density of 1500 cells per well.

Colony formation assay
The indicated cells were seeded onto 6-well plates (1.5 × 10 3 cells per well) for 24 h and treated with optimized concentrations of Cisplatin, ETO, MMC, Mad or Iso. The cells wer e cultur ed f or ∼9 da ys and then stained with 0.5% crystal violet in methanol solution to determine colony formation efficiency. The cluster of stained cells was considered a colony at > 50 cells.

EdU incorporation and detection
A Click ™ EdU-488 Cell Proliferation Detection Kit for Imaging (C0071S, Beyotime) was used to assay cell proliferation through EdU (5-ethynyl-2 -deoxyuridine) incorporation. Briefly, cells were seeded onto 96-well plates (1.0 × 10 3 cells per well) and treated with dimethylsulfoxide (DMSO), Mad or Iso for 48 h. The cells were incubated with 10 M EdU for 30 min. Then, samples were fixed and permeabilized, and the Click ™ reaction was performed according to the manufacturer's instructions. Nuclei were stained with Hoechst 33342 for 15 min at room temperature in the dark. Plates were imaged with ImageXpress Micro-4 (Molecular Devices) or Nikon Ar1 (Nikon) microscopes. MetaMorph was used for image analysis. Data from nine different fields in each condition were quantified per experiment. The EdU entire population nuclear intensity and the percentage of cells incorporating EdU were measured. EdU intensity was also detected only for those cells that incorporated EdU.

Flow cytometry for apoptosis detection
Cell apoptosis was determined by using an Annexin V-FITC / PI Apoptosis Detection Kit according to the manufacturer's protocol (40302ES60, Yeasen). Briefly, Mad-, Isoand MNNG-treated cells were washed with ice-cold PBS and harvested by centrifugation at 1000 g for 5 min. Cells wer e r esuspended in 100 l of Binding Buffer and stained by adding 5 l of Annexin V-FITC and 10 l of propidium iodide (PI) Staining Solution for 15 min at room temperature in the dark before cell a poptosis anal ysis by flow cytometry (Cyto FLEX S, Beckman).

S9.6 dot blot
Genomic DNA was extracted with a TIANamp Genomic DNA Kit (DP304-03, Tiangen), except that samples were treated without RNase A according to the manufacturer's instructions. After quantification by a UV / VIS spectrophotometer (UV5Nano, Mettler Toledo), 1 mg of DNA from each sample was blotted on a nylon membrane (Amersham) with a dot blot apparatus and vacuum suction. For RNase H treatment, 1 mg of DNA was incubated with RNase H at 37 • C ov ernight, then e xtracted with a TIANamp Genomic DNA Kit. The membrane was denatured for 10 min in a solution containing 0.5 M NaOH and 1.5 M NaCl, and neutralized in a solution containing 1 M NaCl and 0.5 M Tris-HCl pH 7.0 for another 10 min. Before being completely dried, the membranes were cross-linked with UV (1200 mJ / cm 2 ) and stained in 1% methylene blue solution (G1303, Solarbio) for 10 min. Subsequently, the membranes were washed with TBST buffer, blocked in 5% (w / v) milk in TBST buffer for 1 h at room temperature and incubated with S9.6 antibody (1:500) overnight at 4 • C. After se v eral washes in TBST buffer, the blots were incubated with IgG secondary antibody (1:1000) in TBST buffer for 1 h at room temperatur e. The blots wer e visualized using chemiluminescence on a Tanon Imaging System and quantified by ImageJ software.

Immunohistochemistry (IHC)
Human lung cancer tissue or nude mouse tumor tissue was fixed in 4% paraformaldehyde solution, embedded in paraffin and sectioned (5 m) onto glass slides. IHC staining was conducted as described previously ( 21 ) using primary antibodies against menin (A300-105A, Bethyl Laboratories, 1:4000), ␥ H2AX (80312, CST, 1:1000) and Ki67 (9129, CST, 1:400). Negati v e controls wer e tr eated identically except no primary antibody was added. Pictures were captured with an Olympus VS200 SLIDEVIEW microscope with panoramic scan. For the quantification of IHC staining, six immunostained images with a final ×400 magnitude were randomly obtained for each section sample. ImageJ Pro Plus was used to calculate the le v el of immunostaining. The images were converted into 8-bit grayscale. Image r egions wer e selected for the measurement of area and integrated density, and background intensity was measured by selecting three distinct areas in the background with no staining. The corrected optical density (COD) was determined as follows: COD = ID -(A × MGV), where ID is the integrated density of the selected image region, A is the area of the selected image region and MGV is the mean gray value of the background readings.

Co-immunoprecipitation (co-IP) assay
Plasmids WT Menin-Flag and mutant (Mut) Menin-Flag were subcloned into the lentivirus plasmids. The substitution of amino acid residue M278W in menin was performed using a site-directed mutagenesis kit (200518, Stratagene) according to the manufacturer's protocol. After more washes with ice-cold PBS, cells were lysed for 30 min at 4 • C in NP-40 lysis buffer (ST2045, Beyotime) containing 1 g / ml PMSF, phosphatase inhibitor and 1 × protease inhibitor cocktail. The lysates were collected by centrifugation (12 000 g , at 4 • C). After quantification of protein concentrations, 1 mg of protein was loaded onto pre-washed protein A / G magnetic beads for 30 min at 4 • C. Then, the l ysates were imm unoprecipitated with the indicated antibodies overnight at 4 • C. Next, the protein A / G magnetic beads were added and incubated for 2 h at 4 • C to recover the IP complexes. The beads were washed fiv e times with NP-40 buffer using a magnetic separator. The bound proteins were eluted with 2 × SDS buffer and subjected to immunoblotting.

Neutral comet assay of DNA double-strand breaks
We performed the neutral comet assay using a Trevigen CometAssay ® Kit as recommended by the manufacturer with 1 × TBE running buffer. Briefly, ∼2 × 10 5 treated cells wer e r esuspended in 0.1% low-melting point agarose in PBS. A 50 l aliquot of the cell suspension was pipetted onto the indicated regions of CometSlides ® and allowed to solidify for 15 min at 4 • C. The following operations were performed under dark conditions. Cells were lysed for 1 h at 4 • C in CometAssay Lysis Solution and then immersed in neutral electrophoresis buffer (NEB; Tris-acetate pH 9.0) for 30 min a t 4 • C . Slides wer e transferr ed to an electrophor esis tank filled with NEB and electrophoresed with a constant current of 1.0 V / cm for 45 min a t 4 • C . The slides were removed fr om the electr ophoresis tank and placed flat in DNA precipita tion buf fer (1 M NH 4 Ac in 95% ethanol) for 30 min at room temperature. Subsequently, the slides were transferred into 70% ethanol for an additional 30 min and then dried overnight at room temperature. Next, sample DNA was stained with 1 × EB diluted 1:10 000 in Tris-EDTA buffer pH 7.5. Comet images were captured at ×10 magnification with an IX71 epifluorescence microscope (Olympus, Japan). The percentage tail moments of > 150 nuclei per sample were measured with OpenComet Assay software ( 39 ). In box and whisker plots, box and whiskers indicate the 25th-75th and 10th-90th per centiles, r especti v ely, with lines r epr esenting median values.

Alternative splicing assay
MEN1 -WT and MEN1 -KO NCI-H460 cells were transfected with the pMTE1A plasmid using Chemi-Trans ™ FectinBor DNA Transfection Reagent following the manufacturer's protocol. After 48 h of transfection, total RNA was extracted using TRIzol reagent. pMTE1A AS was analyzed by RT-PCR with 0.5 g of RNA using the high-ca pacity cDN A re v erse transcription Kit (11141ES60, Yeasen). PCR was carried out with the exon 1 forward PCR products were run in 2% agarose gels stained with 0.1 ‰ EB dye. Bands were visualized using a GleUV system. Band intensities were quantified using ImageJ software.

Splice site motif analysis
Splice site motif scor es wer e calculated using the matrices for splice sites available in ESEfinder version 3.0 ( http:// rulai.cshl.edu/tools/ESE/ ). The matrices for splice sites were deri v ed from constituti v e e xons, and the thr esholds corr espond to the first quantile of all splice site scores.

Lucifer ase / ␤-gal double-r eporter assay
Cells were seeded onto 12-well plates (2 × 10 5 cells per well) and transfected with the pTN24 splicing reporter plasmid with a constituti v ely e xpr essed ␤-galactosidase ( ␤-gal) r eporter for transfection normalization and a luciferase (Luc) reporter that was conditional on the excision of a translational stop codon by splicing. The cells were collected 48 h after transfection, and the reporter activity was measured with a Dual-Light Reporter Assay System (Applied Biosystems) and determined by calculating the ratio of Luc to ␤gal activity.

Inhibition and re-initiation of transcription
Cells wer e cultur ed in 60 mm dishes to 60-70% confluency and then treated with 100 M 5,6-dichlorobenzimidazole 1-␤-D -ribofuranoside (DRB) (C4798, APExBIO) in complete medium for 5 h. The cells were washed three times with PBS to remove the DRB and cultured again in fresh complete medium for different amounts of time. The treated cells were harvested at 5 min intervals after the removal of DRB, and total RNA was extracted with TRIzol reagent according to the manufacturer's instructions. The expression of utrophin pre-mRNA was detected by qPCR using the primers spanning exon-intron junctions. The elongation rate of Pol II transcription was determined as described previously ( 40 ).

Human lung cancer specimens
All specimens used in this study were obtained with informed consent according to protocols approved by the Human Ethics Committee of Guizhou Medical Uni v ersity (no. 2021-53). Human r esear ch procedur es wer e performed in strict accordance with the Declaration of Helsinki. We collected 52 lung cancer samples: 42 lung adenocarcinoma (LUAD) samples, 8 squamous carcinoma samples, 1 small cell lung cancer (SCLC) sample and 1 large cell neuroendocrine carcinoma sample. We also collected corresponding adjacent non-cancerous specimens ( > 2 cm from the cancer ous tissue) fr om patients who had undergone resection. The patients were diagnosed with LUAD, squamous carcinoma and SCL C b y pathologists at the Affiliated Hospital of Guizhou Medical Uni v ersity.

LUAD patient survival analysis
mRNA splicing pa ttern da ta and clinical parameters of LUAD cohorts were downloaded from the TCGA database portal ( http://bioinformatics.mdanderson.org/ TCGASpliceSeq and https://tcga-data.nci.nih.gov/tcga/ , respecti v ely). A total of 502 LUAD patients with fully characterized tumors were included in this study. To determine the correlation between le v els of gene splicing isoforms and LUAD patients' overall survival (OS), we divided the LUAD patients into high-PSI and low-PSI groups by the median cut-off and then conducted Kaplan-Meier survival analysis. We determined the correlation between gene expression and LUAD patient survival using a microarray RNA-seq dataset from a LUAD cohort ( n = 502) by setting the online ( http://ualcan.path.uab.edu/ ) Kaplan-Meier plotter tool for the optimal cut-off for the automatic separa tion of pa tients into high-and low-gene expression groups.

Xenograft experiment in vivo
MEN1 -WT or MEN1 -KO NCI-H460 cells (4 × 10 6 cells per mouse) were inoculated subcutaneously into the axillae of 6-week-old male BALB / C nude mice (Beijing HFK Bioscience Co., Ltd, Beijing, China), and tumor volumes were monitored e v ery day with a caliper. Tumor volumes were calculated with the modified ellipsoidal formula as 1 / 2 × (longest diameter) × (shortest diameter) 2 . Mad (4 mg / kg) in 90% corn oil containing 10% DMSO was injected i.p. once daily for 12 days after the tumors had grown to ∼200 mm 3 . The mice were humanely killed after 2 weeks of Mad tr eatment. The x enograft tumors were dissected, and their w eights w er e measur ed.

Statistical analysis and reproducibility
Data were analyzed using the two-tailed Student's t -test for pairwise comparison, one-way analysis of variance (ANOVA) for multiple comparisons or log-rank tests for K aplan −Meier survi val analysis. W hen da ta wer e expr essed as scatter plots, the Mann-Whitney U -test was performed. Test details are indicated in the figure legends. Statistical analyses were performed using the GraphPad Prism 8 package. Data ar e r epr esented as means ± standar d de viation (SD) or standard error of the mean (SEM) of at least three independent experiments. n values indicate biolo gicall y independent samples and experiments, and a P -value < 0.05 was considered statistically significant (with ** P < 0.01, *** P < 0.001 and **** P < 0.0001 indicating higher le v els of significance). Statistical parameters can be found in the figure legends.

MEN1 deficiency disrupts global alternative pre-mRNA splicing profiles
To determine whether MEN1 plays a role in AS, we profiled the transcriptome in whole-lung tissue from Men1 f / f and Men1 / mice by deep RNA-seq and then performed PSI analyses of the AS using rMATS ( 29 ). Gene expression analysis led to the identification of 444 DEGs ( P < 0.05 and fold change ≥ 1.5), of which 210 (47.3%) were up-regulated and 234 (52.7%) were down-regulated in the Men1 / mice compared with the Men1 f / f mice (Supplementary Figure  S1A; Supplementary Data S1). It is intriguing that the loss of Men1 resulted in notable changes in the AS profiles (Figure 1 A). Through an rMATS analysis, we identified 2459 ASEs that belonged to 2135 genes in the lung tissues of Men1 / mice (FDR < 0.05 and | PSI| ≥ 0.1) (Figure 1 Figure S1D), which suggests that altera tions in AS pa tterns in the lung tissue of the Men1 / mice were unlikely to have been caused by changes in transcript le v els. These results demonstra te tha t the loss of Men1 results in an aberrant AS profile in mouse lung tissue.
GO analysis re v ealed tha t Men1 -regula ted ASEs were involved in RNA metabolism (such as mRNA processing, RN A splicing and RN A binding), nuclear speckles and mRNA splice site selection. Importantly, the terms enriched with these ASE-carrying genes included regulation of mRNA splicing via spliceosome, U2-type prespliceosome, U1 snRNP and spliceosomal complex assembly ( Figure 1 D; Supplementary Figure S1E-G). To further probe global AS affected by Men1 loss during the RNA metabolic process, we reanalyzed RNA-seq data using MA-JIQ, another AS analysis tool, to calculate both typical binary ASEs and intricate local splicing variants ( 33 ). We identified Men1 -regulated splicing alterations totaling 1779 ASEs in the lung tissue of the Men1 / mice (Figure 1 E). Venn diagrams generated via rMATS and MAJIQ verified with high confidence that the differential splicing of 299 genes was regulated by Men1 deletion (Figure 1 E). We also confirmed the aberrant splicing of a few genes, as exemplified by Prpf40b , Rbm27 , Polm , Fance and Prpf39 , by visualization using Sashimi plots ( Figure 1 F; Supplementary  Figure S1H). Some of the aberrantly spliced genes in the lung tissue of Men1 / mice were validated via RT-PCR assays (Figure 1 G, H). Similar findings were observed in the Men1 f / f and Men1 / MEFs (Supplementary Figure S1I).
Ne xt, we e valuated whether AS is impacted by MEN1 expression. To this end, we transiently transfected the adenovirus E1A minigene reporter plasmid pMTE1A carrying multiple 5' splice sites into the Men1 f / f and Men1 / MEFs. This minigene reporter can produce fiv e mRNAs, sizes 13S, 12S , 11S , 10S and 9S , by AS (Figure 1 I) ( 41 ). RT-PCR displayed that in the Men1 / MEFs, AS of the E1A reporter was impaired, as indicated by an increase in the 11S, 10S and 9S isoform le v els compared with those in the Men1 f / f MEFs (Figure 1 J, K). Furthermore, we carried out another splicing assay with NCI-H460 cells following the transduction of a double-reporter plasmid pTN24 expressing ␤-gal, through which luciferase was expressed only when suitable splicing excised an upstream intron sequence with translational stop codons (Figure 1 L) ( 41 ). MEN1 -KO dramatically promoted the ratio of spliced pTN24, as indicated by incr eased Luc / ␤-gal activity, wher eas the r econstituted expression of wild-type MEN1 (r MEN1 ) in the MEN1 -KO cells re v ersed the promotion of MEN1 -KO on pTN24 splicing efficiency (Figure 1 L).
Finally, we measured the pre-mRNA splicing of the ra pidl y inducible gene IRF1. Primer sets were designed to detect the intron-containing pre-mRNA (unspliced primer set) and the total mRNA (spliced primer set) that is produced upon IRF1 induction (Figure 1 M). Transcripts of the IRF1 gene were detected when using the spliced primer set, and IRF1 mRNA le v els wer e monitor ed 0, 30, 45 and 60 min after the addition of interferon-␥ (IFN-␥ ) (Figure 1 N).
To determine pre-mRNA splicing, we considered the ratio of unspliced to total mRNA (splicing efficiency). Under normal conditions, the ratio of unspliced to total mRNA was highest 30 min after IRF1 induction, which indicates that splicing is a slow step relati v e to transcription and increases over time during the generation of mRNA. We further analyzed the relati v e splicing efficiency of IRF1 in the MEN1 -KO cells compared with the MEN1 -WT cells. It is striking that 30 min after the induction of IRF1, we observed a 1.8-fold decrease in the ratio of unspliced to total mRNA in the MEN1 -KO cells compared with the MEN1 -WT cells. The effect on splicing was diminished 45 min after induction of IRF1, and no effect was observed after 60 min of induction (Figure 1 O). These results demonstrate that MEN1 deficiency enhances the splicing efficiency of IRF1 pre-mRNA. Taken together, these findings clearly demonstra te tha t MEN1 is a key regula tor of alterna ti v e RNA splicing and that its deficiency disrupts the homeostasis of the global splicing network.

MEN1 alters e x on skipping and the abundance of RNA splicing isoforms
The generalizability of the effect of MEN1 on AS patterns was str ongly corr obora ted by repea ted RNA-seq and dif ferential exon usage analyses with the MEN1 -WT and MEN1 -KO NCI-H460 cells (Supplementary Data S3 and S4). Venn diagram analysis identified 399 conserved MEN1 -regulated ASEs from the mouse lung tissue and lung cancer cells, 164 of which were SE e v ents (Supplementary Figure S2A). We further confirmed the differential splicing of se v eral conserved genes, such as SENP6 , EMSY , YTHDF3 and GGPS1 , by visualization using Sashimi plots (Supplementary Figure S2B). Because SE e v ents represent the vast majority of MEN1 deletion-related ASEs (57% in mouse lung tissue and 62.2% in NCI-H460 cells) (Figure 1 B; Supplementary Data S4), we investigated the underlying mechanism by which menin regulates exon skipping.
By comparing splice site scores of responsi v e and unresponsi v e SE-forming e v ents, we found that both MEN1enhanced and MEN1 -r epr essed exons had weaker 5' splice sites (5SSs) or 3SSs than those observed in unresponsive SEs (Figure 2 A, B). We analyzed the features of the splice sites of MEN1 -regulated and MEN1 -independent genes using the matrices for splice sites available in ESEfinder ( 42 ). We were surprised to find tha t MEN1 -regula ted SE genes,  but not all differentially spliced genes, harbored a weak 5SS or 3SS. Some of these genes were associated with motif scores that were below the threshold score for classic constituti v e splice sites; more intriguing is that some splice site scores, such as those for the TFDP1 , ZNF18 and SLTM genes, were similar to those for standard constituti v e splice sites but corresponded to bispecific splice sites (Supplementary Figure S2C, right), which may confuse the splicing components because they may be recognized as either 5SS or 3SS ( 43 ). We designed se v eral PCR primer pairs targeting two neighboring constituti v e e xons in a cluster of MEN1 -regulated genes with SEs. Knock-ing out MEN1 in NCI-H460 cells resulted in a marked increase in SENP6 , MYL6, TFDP1 , PHF7 and SLC4A7 exon skipping, whereas it inhibited exon skipping in EMSY , YTHDF3 and GGPS1 genes (Figure 2 C, D). In agreement with this finding, shRNA-mediated genetic knockdown of MEN1 (sh MEN1 ) in A549 cells yielded similar results (Supplementary Figure S2C, left). In contrast, ov ere xpression of MEN1 reduced the isoform abundance of SENP6 , ZNF18 , PHF7 , CDH1 and TFDP1 but promoted SLTM isoform generation in NCI-H446 cells ( Supplementary Figure S2D, E). Our data suggest that the presence of suboptimal 5SSs or 3SSs may render splicing of these genes largely The human CD44 gene consists of 10 constituti v e e xons (C1-C5 and C16-C19) and 9 clustered variable exons (V2-V10), and produces multiple splice variants, some of which are involved in the development and progression of various tumors ( 44 ). We found that knockout of MEN1 promoted the generation of constant and variant CD44 exons (Figure 2 E); treatment with the phorbol ester PMA (phorbol 12-myristate 13-acetate), an activator of the protein kinase C (PKC) pathway, enhanced the MEN1 KOinduced accumulation of CD44 V6, V8 and V10, and did so in a time-dependent manner, as determined by qPCR using primers specific for constant or variab le e xons (Figure 2 F). Alterations in the abundance of RNA splicing isoforms can be caused by changes in AS processes or differential isoform degradation rates. Her e, our r esults showed that MEN1 affects the AS process of CD44 genes, not their degr adation, as inhibiting tr anscription with actinomycin D (Act.D) did not alter the effect of MEN1 (Supplementary Figure S2F, G).
Ne xt, we inv estiga ted whether MEN1 regula tes AS in a manner dependent on mixed lineage leukemia protein 1 (MLL1), a histone H3 lysine 4 methyltr ansfer ase that interacts with menin ( 45 ). qPCR results showed that treatment with MI-3, a specific inhibitor of the menin-MLL1 interaction ( 46 ), strikingly enhanced the abundance of C5, C19 and variable CD44 exons (V2-V10) in NCI-H446 cells treated with or without PMA (Supplementary Figure S2H, I). To consolidate this finding, we stably transfected plasmids expressing WT-Menin or mutant Menin Met278Trp (Mut-Menin) into MEN1 -KO NCI-H460 cells. Consistent with the pr evious r eports ( 45 ), Co-IP da ta indica ted tha t a menin Met278Trp substitution completely disrupted the interaction of menin with MLL1 but not the menin-LEDGF (lens epithelium-deri v ed growth factor) interaction (Figure 2 G). As expected, ectopic expression of WT-Menin, but not of Mut-Menin, profoundly pre v ented the generation of variable CD44 exons (Figure 2 H). These results suggest that the MLL1-dependent effects of MEN1 contribute to its regulation of AS.

MEN1 regulates AS by slowing the Pol II elongation rate
The aforementioned findings prompted us to continue to explore how MEN1 regula tes alterna ti v e RNA splicing. Because menin did not interact with the splicing factors evaluated, we ruled out the possibility that MEN1 -affected AS is mediated through protein-protein interactions (Supplementary Figure S3A). We speculated that MEN1 might impact the recruitment of splicing factors to Pol II, which in turn would change the AS process. Co-IP showed that Pol II was associated with splicing factors, including SRSF2 and U2AF65, but not with SNRPA or hnRNPA1 in the MEN1 -WT cells (Figure 3 A). It is important to note that the interaction of SRSF2 or U2AF65 with Pol II was greatly reduced in the MEN1 -KO cells (Figure 3 A). These results suggest that menin facilitates the recruitment of key spliceosomal factors, such as SRSF2 and U2AF65, to elongate Pol II complexes during transcription.
The slowing of the Pol II elonga tion ra te pre v ented e xon skipping by increasing the chances for recruitment and recognition of the splicing machinery to weak splice sites ( 47 ). The r equir ement for simultaneous transcription to induce an effect of MEN1 on AS drew our attention to the abnormal AS patterns induced by MEN1 deletion on Pol II elongation. ChIP walking assays with NCI-H460 cells showed that menin was present not only in the promoter regions of the CD44 gene but also in exon and intron regions; this protein appeared to be pr efer entially positioned at variant exons, particularly in V5, V6, V7 and V8, relati v e to constant exons or their neighboring regions (Figure  3 B). PMA treatment significantly increased the abundance of menin inside the CD44 V3, V4, V5, V9, i15 and i16 regions (Figure 3 B) Figure S3B). The accumulation of p-Pol II (Ser2) induced by MEN1 deletion was not further augmented by PMA, presumably because 100% abundance was reached at these regions (Figure 3 D). Immunoblotting showed that MEN1 -KO notably decreased the enrichment of chromatin with Pol II, p-Pol II (Ser5) and total Rbp1 C-terminal domain (CTD; phosphorylated and unphosphorylated Pol II forms), and increased p-Pol II (Ser2) accumulation, although treatment with Cisplatin reduced the chromatin association of these Pol II proteins in a time-dependent manner (Supplementary Figure S3C). Importantly, the deletion of MEN1 reduced recruitment of U2AF65 to the variant exons but not to the constant exon and intron regions (Figure 3 F), and moderately increased the recruitment of the U1 snRNP protein SNRPA to the CD44 gene regions (Supplementary Figure S3D). These findings suggest that MEN1 regulates AS by slowing the Pol II elonga tion ra te and by facilita ting recruitment of splicing machinery either directly or through Pol II.
An independent evaluation of the impacts of MEN1 on the elonga tion ra te of RN A pol ymerase involved anal yzing the distribution of Pol II and p-Pol II (Ser2) along the c-Myc gene ( 49 ). Protein occupancy was detected at 17 positions within the c-Myc gene and its flanking sequences in the MEN1 -WT and MEN1 -KO NCI-H460 cells. We observed that Pol II occupancy was elevated at the transcription start site and decreased in other gene regions (Supplementary Figure S3E), whereas the high occupancy of p-Pol II (Ser2) in the c-Myc gene was noted in a region from +4828 to +7028 bp of the poly(A) site (Supplementary Figure  S3F). Notably, MEN1 deletion markedly reduced the occupancy of Pol II throughout the gene while increasing the enrichment of p-Pol II (Ser2) (Supplementary Figure S3E, F). To corroborate the evidence for the regulation of Pol II elongation by MEN1 , we made use of an observation by others suggesting that transcription mediated by a slow mutant human Pol II led to an increased pr omoter-pr oximal to promoter-distal pre-mRNA ratio ( 47 ). We isolated pre-mRNAs from the nuclei of MEN1 -WT and MEN1 -KO cells and performed qPCR with primer sets loca ted a t each end of the CD44 gene (intron 1 and intron 18). Knocking out MEN1 resulted in a 1.5-fold decrease in the ratio of intron 1 to intron 18, which was re v ersed by re-e xpression of MEN1 (r MEN1 ) (Figure 3 G).
We validated this fast Pol II progression induced by MEN1 deletion by using the DRB release method to calculate the rate of in situ Pol II elongation on the utrophin gene . Transcription of the exon 1 region of the utrophin gene was able to recover within minutes of DRB release in both cell types, and pr e-mRNA expr ession in MEN1 -KO cells was higher than that in MEN1 -WT cells (Figure 3 H). Reciprocally, in MEN1 -WT NCI-H460 cells, the recovery of expression of the exon 2 region was delayed until 35 min after drug removal, which is consistent with previous data ( 40 ); howe v er, the delay in transcription of this gene was ∼20 min in MEN1 -KO cells (Figure 3 H). This is consistent with a transcriptional lag due to the genomic distance between the first two exons (the utrophin gene consists of 74 exons and 73 introns, and the first intron is 110 kb long). These data suggest that in MEN1 -WT NCI-H460 cells, Pol II transcribed the 110 kb region of the utrophin gene in ∼30 min at a rate of 3.2 kb / min, whereas in the absence of MEN1 , Pol II transcribed the same region of the utrophin gene within 20 min at a rate of 4.4 kb / min (Figure 3 H).
A deficiency of MEN1 expedited the processivity rate of RN A pol ymerase, w hich compelled us to examine the retention of nascent pre-mRNA on template chromatin. We e xtracted nati v e non-cr oss-linked chr oma tin and sonica ted it at a low frequency to obtain a fragment of relati v ely large RN A with transcriptionall y acti v e chromatin ( 38 ). qPCR was used to quantify pre-mRNA le v els at different positions throughout the CD44 gene. We found that corresponding exons or introns encoded greater amounts of CD44 pre-mRNA in the MEN1 -KO cells than in the MEN1 -WT cells (Figure 3 I). This finding was congruent with the results obtained after the depletion of P ol II and p-P ol II (Ser5) and the accumulation of p-Pol II (Ser2) observed in these gene regions (Figure 3 C-E); all these results indicate that MEN1 -KO accelerates the progression of Pol II elongation, which results in the retention of nascent unspliced transcripts on the DNA template. Similar to MEN1 -regulated SE genes, most exons throughout the entire CD44 gene possess a weak 5SS or 3SS (Supplementary Figure S3G), which renders these exons increasingly prone to skipping and missplicing when MEN1 is dysfunctional. Indeed, we observed that MEN1 knockdown induced the accumulation of pre-mRNA splicing isoforms encoded by corresponding constant and variant exons (Supplementary Figure S3H) and that these effects were PMA inducible and time dependent (Supplementary Figure S3I, J). Furthermore, we designed PCR primers targeting certain exon-intron (5SS) or intronexon (3SS) junctions within the CD44 gene, confirming that MEN1 deletion resulted in the marked accumulation of pre-mRNA splicing isoforms at most splice sites (Figure 3 J). Taken together, these experiments demonstra te tha t MEN1 decreases the rate of RN A pol ymerase elongation and favors the use of the weak splice sites present in the CD44 gene, thereby inducing the generation of variant exons or splicing isoforms.

MEN1 prevents R-loop-induced accumulation of DNA damage and genome instability
Disrupting the recruitment of splicing factors to Pol II prolongs the association of naked nascent RNA with single-stranded template DNA, ther eby r esulting in threestranded nucleic acid structures known as R-loops ( 50 ). Unscheduled R-loops lead to DNA damage and genome instability after the depletion of splicing factors, which in turn contribute to cancer (51)(52)(53). In this study, we first monitored genomic DNA extracted from MEN1 -WT and MEN1 -KO NCI-H460 cells with the monoclonal antibody S9.6, w hich specificall y detects DN A −RN A hybrids ( 54 ). We observed a significant increase in the enrichment of DN A −RN A hybrids in the MEN1 -KO cells that was abolished by treating the DNA with RNase H, which binds and hydrolyzes the RNA strand of DNA −RNA duplexes ( Figure 4 A; Supplementary Figure S4A) ( 55 ), validating the specificity of the S9.6 antibody. It is interesting that treating NCI-H460 cells with Mad, a splicing inhibitor that disrupts earl y-stage spliceosome assembl y and arrests at spliceosome complex A ( 56 ), also resulted in abundant R-loop formation, which was enhanced by the deletion of MEN1 (Figure  4 B; Supplementary Figure S4B). IF staining showed that the S9.6 signal was present in both the cytoplasm and nucleus; the MEN1 -deleted cells were significantly enriched in DN A −RN A hybrids, in particular in the nucleus, relati v e to the MEN1 -WT cells (Figure 4 C, D). The deletion of MEN1 promoted Mad-induced R-loop formation (Figure 4 C, D), which was consistent with the S9.6 dot blot results. Similar findings wer e r e v ealed for Men1 f / f and Men1 / MEFs, which showed that in the absence of MEN1 , MEFs exhibited conspicuous accumulation of nucleolar DN A −RN A hybrids, which we quantified after subtraction of the cytoplasmic signal (Supplementary Figure S4C, D). MEN1 -KO moderately increased the nuclear RNase H1 signal and the co-localization of RNase H1 with S9.6 in the nucleolus (Supplementary Figure S4E), which suggests that the deletion of MEN1 leads to the accumulation of R-loops, w hich a ppear to recruit RNase H1 into nucleoli. Next, we confirmed the IF results using the more accurate method of DRIP followed by qPCR (DRIP-qPCR). DN A −RN A hybrids accumulated up to at least 1.5-fold more in the MEN1 -KO cells than in the MEN1 -WT cells in all analyzed genes; these gene regions have been shown to existence of abundant R-loops ( 34 , 57 ). The hybrid signals were eliminated by RNase H treatment as a confirmation of the specificity of the assay (Figure 4 E). Immunoblotting also indica ted tha t MEN1 -KO modestly augmented the chroma tin abundance of RNase H1 but reduced the enrichment of MBNL1 [a splicing factor that binds a stem-loop structure within cardiac troponin T pre-mRNA ( 58 )] and DDX17, an RNA helicase that unwinds DNA-RNA hybrids ( 59 ) in NCI-H460 cells with or without Mad treatment (Supplementary Figure S4F). These results suggest that MEN1 precludes the formation of R-loops presumably via an RNA helicase-dependent mechanism in addition to affecting the n ucleolar-n ucleoplasmic distribution of RNase H. Nucleic Acids Research, 2023, Vol. 51, No. 15 7965 in MEN1 -KO cells (Figure 4 H, I), which suggests that the increase in DNA damage in MEN1 -deleted cells is R-loop dependent. Immunoblotting confirmed that MEN1 deletion markedly enhanced chromatin levels of ␥ H2AX and phosphorylated A TM (p-A TM), a canonical DDR pathway (Supplementary Figure S4F) ( 63 ). In contrast, ov ere xpression of MEN1 in A549 cells markedly a ttenua ted ␥ H2AX activation induced by treatment with APH compared with in the vector cells (Supplementary Figure S4I, J). To test whether the inactivation of MEN1 generated DSBs or activ ated DDR signaling b y some alternati v e pathway, we carried out a neutral comet assay and observed that the comet tail moment in MEN1 -KO cells treated with or without APH was significantly increased, and this effect was suppressed by RNase H1 ov ere xpression ( Figure 4 J; Supplemental Figure S4K), which provides direct evidence that MEN1 deficiency ultimately contributes to DSB accumulation and confirms that DNA damage generated in MEN1deleted cells is R-loop dependent.
Our previous studies showed that the loss of MEN1 gave rise to DDR activation and lung tumorigenesis ( 21 ). To elucidate the correlation between menin expression and Rloop-mediated DNA damage and genome instability in human lung cancer samples, we randomly collected 52 lung cancer specimens, histolo gicall y classified samples from 42 patients with LUAD, 8 patients with squamous carcinoma and 2 patients with lung neuroendocrine carcinoma, and corresponding adjacent non-cancerous specimens (Supplementary Table S1). Our results showed that lung cancer tissues possessed dramatically reduced le v els of menin expression and increased ␥ H2AX expression compared with the adjacent non-cancerous tissue (Supplementary Figure S4L-N). These human lung cancer tissues consistently showed abundant accumulation of R-loops, as confirmed by increased IF staining for the S9.6 antibody (Supplementary Figure S4O, P). Moreover, we observed that chromosomal instability markers, such as nuclear buds (NBUDs), nucleoplasmic bridges (NPBs) and micronuclei (MNi), were increased in Men1 / MEFs (Figure 4 K, L). Taken together, these findings indicate that MEN1 protects cells from Rloop-mediated DNA damage and chromosomal instability during lung tumorigenesis.

MEN1 -regulated DNA damage-mediated AS is involved in lung cancer
We next evaluated whether MEN1 dysfunction causes aberrant AS patterns through R-loop-induced activation of DN A damage. We reanal yzed RN A-seq in MEN1 -WT and MEN1 -KO NCI-H460 cells after Cisplatin treatment. As expected, Cispla tin trea tment resulted in changes in gene expression profiles, in which 168 genes were down-regulated and 461 genes were up-regulated (Supplementary Figure  S5A Figure S5B, C). We identified 2241 ASEs (corresponding to 1845 genes) that wer e differ entially affected by MEN1 deletion (Supplementary Figure S5D). The distribution of the ASE types is illustrated in Supplementary Figure S5E; most of these ASEs were SEs (62.2%). We also found that ∼22% of DEGs (408 genes) underwent two or more types of AS; one gene may e v en hav e had up to four types of AS affected by MEN1 depletion (Supplementary Figure S5F). Moreover, we identified 1228 DEGs in the MEN1 -KO cells compared with the MEN1 -WT cells, and only 9.2% of these DEGs (113 genes) were found in the ASEs (Supplementary Data S3); among them, 34.5% (39 genes) were up-regulated and 65.5% (74 genes) were downregulated. Howe v er, no significant correlation between gene expression and their PSI was f ound f or an y of the 113 genes (data not shown), which further suggests that the alteration in AS patterns caused by MEN1 deletion in mouse lung tissue or lung cancer cells is not a direct result of changes in transcription.
Unexpectedly  Figure S5G). These findings are consistent with previous studies demonstrating that DNA damage triggered splicing pattern reprogramming of transcripts of genes pivotal for DDR and the regulation of genome stability ( 64 ). The incidence of MEN1 deletionrelated ASEs was not further increased by Cisplatin treatment (1655 differential splicing e v ents were detected in the Cispla tin-trea ted MEN1 -WT and MEN1 -KO cells) (Supplementary Data S7). These analyses demonstrated that inactivation of MEN1 induced concerted reprogramming of AS that is partially dependent on the DNA damage activation.
Gi v en that MEN1 plays an important role in the regulation of RNA metabolism and genome stability, we next investigated the biological implications of MEN1 -regulated ASEs in human lung cancer. By analyzing overlapping ASEs between rMATS and MAJIQ in NCI-H460 cells, we identified 640 high-confidence differentially spliced genes that wer e r egulated upon MEN1 deletion (Figure 5 D). KEGG pathway enrichment analysis confirmed that these differentially spliced genes were mainly involved in various human diseases, especially cancers, and cellular processes (cell growth and death and cell motility) (    Figure S5I). Furthermore, we took advantage of data analyzed by others who generated AS profiles in LUAD and identified a total of 3691 survival-associated ASEs by conducting uni variate survi val analyses for OS ( 65 ). Using Venn diagram analysis, we found that 137 out of 640 MEN1 -related ASEs identified in NCI-H460 cells overlapped with LUAD survival-associated ASEs ( Figure  5 G). A pproximatel y 37% of the ASEs (234 e v ents) were SE e v ents, of which 28 SE e v ents were found among LUAD survi val-associated SE e v ents (Figure 5 H); 71% (20 / 28) of the SE e v ents wer e validated to be corr elated with the OS of LUAD patients (of these SE e v ents, 9 predicted significantly worse outcomes and 11 predicted good outcomes) (Figure 5 I). In addition, we downloaded clinical parameters of the LUAD cohort from the TCGA database and divided the patients into high-and low-PSI groups on the basis of the median PSI of 28 MEN1 -regulated SE genes. A survival analysis showed that 17 out of 28 MEN1 -regulated SE e v ents also correlated with patient OS, of which 8 SE e v ents Nucleic Acids Research, 2023, Vol. 51, No. 15 7967 showed a negati v e correlation and 9 SE e v ents displayed a positi v e corr elation (Figur e 5 J). Inter estingly, howe v er, the mRNA expression of these genes did not correlate with the OS of patients (Supplementary Figure S5J), which suggests tha t MEN1-regula ted gene AS, ra ther than their gene expression, is dramatically correlated with LUAD patient survival. Indeed, RT-PCR results indicated that the deletion of MEN1 in lung cancer cells increased the instances of short splicing isoforms in the FAM189B , NUBP2 , ENDOV and TARBP2 genes, but inhibited generation of the CPSF7 , MRRF and LETMD1 splicing isoforms (Supplementary Figure S5K). Altogether, these findings further support the claim that MEN1 -regulated ASEs have functional significance in lung cancer.

MEN1 deficiency sensitizes human lung cancer cells to splicing inhibitors
We further investigated whether splicing inhibitors can reverse the malignant behaviors of MEN1 -deficient lung cancer cells. We tested two splicing inhibitors (Mad and Iso), three anticancer drugs (Cisplatin, ETO and MMC) and MNNG for selecti v e lethality in MEN1 -WT and MEN1 -KO NCI-H460 cells. The six tested chemical drugs tested inhibited the proliferation of NCI-H460 cells ( Figure 6 A; Supplementary Figure S6A, B). Interestingly, once a certain concentration was reached, both Mad and Iso, but not Cispla tin, ETO or MMC , suppressed MEN1 -KO cells to a greater extent than MEN1 -WT cells and did so in a dosedependent manner; MNNG, to a lesser extent, also exhibited selecti v e lethality in MEN1 -KO cells (Figure 6 A, B). A similar phenomenon was found in Mad-and Iso-treated To delineate the selecti v e suppression of splicing inhibitors in MEN1 -deficient lung cancer growth in vivo , we subcutaneously transplanted MEN1 -WT and MEN1 -KO NCI-H460 cells into nude mice. The deletion of MEN1 profoundly promoted tumor growth for the duration of the observation period (Figure 6 F). As expected, treatment with Mad inhibited the growth of MEN1 -deficient tumors more efficiently than WT-MEN1 -expressing tumors (Figure 6 F). The mice were sacrificed on day 14, and the tumor weight and Ki67 index of the MEN1 -KO-vehicle xenograft tumors were dramatically augmented compared with these measures in the MEN1 -WT-vehicle xenograft tumors; treatment with Mad reduced the tumor weight and Ki67 index, and showed higher selectivity for the MEN1 -KO xenograft tumors ( Figure 6 F-J). These findings indica te tha t splicing inhibitors such as Mad and Iso predominantly suppress the growth of lung tumors in which MEN1 is dysfunctional.
Finally, we sought to define the modality of MEN1 deletion-rela ted cell dea th induced by splicing inhibitors. Annexin-V / FITC staining and flow cytometry showed that both Mad and Iso induced cell apoptosis but with no selecti v e lethality in the MEN1 -WT and MEN1 -KO cells, whereas MNNG-induced cell apoptosis exhibited exceptional selectivity for the MEN1 -KO cells ( Figure 6 K; Supplementary Figure S6E). These data suggest that Mad and Iso selecti v ely inhibit the proliferation of MEN1 -KO cells independent of the apoptosis pathway. We were surprised to find that MEN1 deletion significantly suppressed ferroptosis, a non-apoptotic form of cell death ( 66 ), as confirmed by increased GPX4 expression and decreased COX2 expression, ef fects tha t wer e r escued by Mad or Iso tr eatment ( Taken together, these results demonstra te tha t Mad and Iso e xhibit selecti v e growth inhibition in MEN1 -deficient lung tumors, which may be explained by ferroptosis rather than increased apoptosis.

DISCUSSION
Dysregulation of AS substantially contributes to an e v erincreasing number of human diseases. Howe v er, the regulation of alternati v e pre-mRNA splicing processes is poorly understood. In the present study, we provide multiple lines of evidence supporting MEN1 as an important modulator of the AS process, the most important of which is that the deletion of MEN1 leads to aberrant AS profiles in mouse lung tissue, human lung cancer cells and human breast cancer cells (data not shown). Our subsequent data indica te tha t MEN1 slows the ra te of Pol II elonga tion and is r equir ed to pr e v ent of R-loop formation and the accum ulation of DN A damage. In this connection, because of their augmented chromosomal instability and correlation with LUAD patient survival, lung cancers harboring MEN1 inactivation may be dependent on a fast Pol II elonga tion ra te to genera te R-loops, abundantl y accum ulating DSBs, thereby supporting tumor growth, which to some extent explains the selective lethality of lung cancer cells to splicing inhibitors. In accordance with this hypothesis, deficiency in MEN1 pro vok ed the splicing isoform generation, genome instability and cell prolifera tion. This observa tion is in line with previous studies showing that highly processive transcription fav ors ex on skipping, w hereas DN A damageinhibited Pol II elongation pro vok es exon inclusion and tumorigenesis ( 14 ).
Because menin lacks motifs that are homologous to known proteins, it is challenging to discover its biochemical function. Menin has been widely characterized as an important scaffold protein that regulates gene transcription by interacting with multiple proteins with di v erse functions ( 22 ). Consistent with this conception, menin activates gene transcription by binding to the transcription activator MLL1; in contr ast, the menin-JUND inter action suppresses JUND-induced transcription ( 45 ). Here we show that, in , analyzed by one-way ANOVA; * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001; ns, not significant. addition to having this quantitati v e effect, MEN1 also impacts the quality of transcripts by altering exon skipping in the pre-mRNAs of some genes. We and others have previously demonstrated that MEN1 inactivation or deletion leads to various human diseases, such as renal fibrosis ( 67 ), diabetes ( 68 ) and lung cancers ( 21 ). Since the AS pro gram tightl y controls the quantity and quality of posttranscriptional gene expression, which plays a crucial role in di v erse cellular processes, including cell prolifera tion, dif ferentiation and death, we ther efor e r eason that these defects are caused in part by abnormal AS programs upon MEN1 dysfunction. The following evidence supports this assumption: (i) abundant differential AS e v ents, instead of a small number of DEGs, are strongly influenced upon MEN1 knockout in mouse lung tissue and human cancer cells; (ii) menin e xtensi v ely binds to CD44 variant e xons and reduces their abundance by slowing Pol II elongation; and (iii) in MEN1 -regulated SE e v ents, RNA splicing isoform le v els, rather than their corresponding transcription le v els, correla te with LUAD pa tient survival (Figure 5 J; Supplementary Figure S5J).
On a mechanistic le v el, MEN1 maintains homeostasis of the AS network in part by slo wing do wn the rate of RNA polymerase elonga tion. Alterna ti v e RNA splicing processes are tightly coupled to Pol II transcription ( 69 ). This is an important mechanism for controlling the AS of the vast majority of genes. In this study, we showed that menin affects AS patterns by decreasing the rate of Pol II elongation. Under these circumstances, slowing down Pol II elongation favors the use of weak or suboptimal splice sites by increasing the time window opportunity for their recognition by splicing factors before downstream stronger splice sites are synthesized ( 47 ). Our findings propose that menin facilitates pausing of Pol II elongation on the DNA template, as exemplified by the oncogene CD44 or c-Myc , which may lead to Pol II interaction with key splicing factors through the CTD of the large subunit of Pol II; its exact mechanisms ar e curr ently being investigated. Mor eover, although menin has been shown to be a DNA-binding protein that plays a significant role in gene expression ( 23 ), whether it dir ectly r egulates AS by binding to target RNA sequences has not been explored. Our unpublished RIP-seq data analysis showed that menin is capable of directly binding to a subset of RNAs. Considering the fundamental role of RNAbinding proteins in regulating the choice of splice site during the AS process ( 26 , 70 ), we propose that menin directly regulates RNA splicing through its direct interplay with specific RNAs under both physiological conditions and exposure to external stimuli. Ho wever, ho w the function of the menin RNA-binding protein impacts alternati v e pre-mRNA splicing remains to be determined. All these findings indica te tha t menin is a crucial splicing regulatory factor that equilibrates the AS network through indirect Pol II elongation-dependent and direct RNA binding-dependent mechanisms.
In conclusion, we re v eal a complex relationship among MEN1 dysfunction, a bnormal AS , R-loop formation, the accumulation of DNA damage and genome instability during lung tumorigenesis. MEN1 inactivation accelerates the rate of Pol II elongation, producing abundant nascent pre-mRNAs. This enables unspliced nascent transcripts to be retained on template chromatin and triggers R-loop formation and DSB accumulation, thereby leading to genome instability and lung cancer progression. Gi v en the roles of these changed MEN1 -related exon skipping events in the progression of lung cancer, their potential as therapeutic targets specifically against MEN1 -deficient cancer should be investigated in the future.

DA T A A V AILABILITY
The antibodies , primers , cell lines , lung cancer samples, mice , plasmids , chemicals , drugs and commercial kits used in this study are listed in Supplementary Tables S2 and  S3. RNA-seq data in this study were deposited in the National Center for Biotechnology Information (NCBI) under the accession numbers PRINA888804 for the Men1 f / f and Men1 / mouse lung tissue and PRINA888457 for the MEN1 -WT and MEN1 -KO NCI-H460 cells. Information on the lung cancer patients is available as Supplementary Da ta S8. Da tasets for the 3691 LUAD survival-associated AS e v ents are availab le from pub lished r efer ences ( 65 ). All other data supporting the findings of this study are available from the corresponding author on reasonable request. Source data are provided with the paper.