Content Detail

• Content META

  Title	Abstract
  ID	6961803
  External ID	7550486981
  Author	Unknown
  Published Date	Feb. 14, 2025, 4:14 a.m.
  Summary	Ewing sarcomas (ESs) are biologically aggressive tumours of bone and soft tissues caused by chromosomal translocations yielding in-frame fusion proteins driving the neoplastic transformation. The DNA/
  Tags	None
  URL	https://academic.oup.com/nar/article/53/4/gkaf068/8015312
  Sentiment	Negative: 0.02155    Neutral: 0.86708    Positive: 0.11137
  Relevance	0.00000

• Content Details

  Content ID	4345042
  Key	dd32b281-4d79-46dc-8704-ea80a9fabd28
  Content	Ewing sarcomas (ESs) are biologically aggressive tumours of bone and soft tissues caused by chromosomal translocations yielding in-frame fusion proteins driving the neoplastic transformation. The DNA/RNA helicase DHX9 is an important regulator of cellular processes often deregulated in cancer. Using transcriptome profiling, our study reveals cancer-relevant genes whose splicing is modulated by DHX9. Immunodepletion experiments demonstrate that DHX9 impacts on the recruitment of U2 small nuclear RNP (snRNP) onto the pre-mRNA. Analysis of structure and sequence features of DHX9 target exons reveal that DHX9-sensitive exons display shorter flanking introns and contain HNRNPC and TIA1 consensus motifs. A prominent target of DHX9 is exon 11 in the Cortactin ( CTTN ) gene, which is alternatively spliced to generate isoforms with different activities in cell migration and tumour invasion. Alternative inclusion of the exon 11 in CTTN gene is one of the most recurrent isoform switches in multiple cancer types, thus highlighting the pivotal role of DHX9 in defining the tumour phenotype. Biochemical analyses reveal that DHX9 binding promotes the recruitment of U2snRNP, SF3B1, and SF3A2 to the splice sites flanking exon 11. These findings uncover a new role of DHX9 in the control of co-transcriptional splicing in ES, which may represent a new druggable target to counteract ES malignancy. Ewing sarcomas (ESs) are biologically aggressive tumours of bone and soft tissues caused by different chromosomal translocations yielding in-frame fusion proteins that activate a specific oncogenic program and direct neoplastic transformation [ 1 ]. Approximately 85% of ES tumours harbour the cytogenetic hallmark t(11;22) (q24;q12) translocation, which creates a fusion between the Ewing sarcoma breakpoint Region 1 gene ( EWSR1 ) in chromosome 22 and the Friend leukaemia virus Integration site 1 gene ( FLI1 ) in chromosome 11 [ 2 ]. Despite recent therapeutic advances, clinical outcome of most patients remains relatively poor [ 1, 3 ]. Thus, discovery of new molecular markers for early diagnosis and of new molecular mechanisms to develop more efficacious targeted therapies are urgently needed to improve prognosis and survival of ES patients. EWS-FLI1 oncoprotein modulates gene expression and splicing of both protein-coding [ 4–6 ] and noncoding transcripts [ 7 ], and this regulation likely determines its oncogenic function. The transcriptional and splicing activity of EWS-FLI1 is modulated by its direct interaction with the DNA/RNA helicase DHX9 in ES cells [ 8–10 ], thus leading to the expression of splice isoforms involved in oncogenesis [ 11 ]. Remarkably, DHX9 expression in mouse embryonic fibroblasts stably transfected with EWS-FLI1 enhances anchorage-independent growth and oncogenic properties [ 8 ], whereas DHX9 downregulation in xenograft mouse models of ES reduces tumour growth in vivo [ 12 ]. These findings strongly indicate that DHX9 binding to EWS-FLI1 is crucial for its oncogenic function and for ES pathogenesis. Therefore, we and others have reported that knockdown of DHX9 expression in ES cells critically reduces their viability and sensitizes them to genotoxic stress [ 9, 12–15 ], while no decrease in cell viability was detected in cancers in which chromosomal translocations do not involve ETS genes [ 9 ]. These observations strongly support a key role of DHX9 in ES. Accordingly, the small molecule YK-4-279, which inhibits the interaction between EWS-FLI1 and DHX9, displays strong antitumour activity by specifically reducing the transcriptional activity of EWS-FLI1 in ES cells [ 9 ]. Furthermore, the clinical derivative of YK-4-279, named TK-216, is currently being tested in a clinical trial with relapsed or refractory ES patients (NCT02657005), with the aim to obtain information about the safety and efficacy of TK-216, administered alone or in combination with vincristine [ 16–18 ]. DHX9 is a DEAH box RNA/DNA helicase involved in all the essential steps of RNA metabolism, such as transcription, pre-mRNA alternative splicing, translation, and ribosome biogenesis [ 10 ]. The DHX9 gene encodes a 140 kDa protein [ 10 ]: the N-terminal part of the protein is characterized by two repeats of double-stranded RNA-binding domain (dsRBD I and dsRBD II) and by the minimal transactivation domain; the central part of the protein contains a conserved ATPase-dependent helicase domain, required for ATP binding, hydrolysis, nucleic acid binding, and unwinding; the C-terminus of DHX9 is formed by repeated arginine and glycine (RG) residues (RG-rich domain), cooperating with other domains to achieve nucleic acid affinity and involved in RNA-based binding to G quadruplex structures [ 10 ]. Nevertheless, whether the function of DHX9 in alternative splicing relates to oncogenic transformation is currently unknown. Alternative splicing is the molecular process that generates multiple messenger RNA (mRNA) variants from single eukaryotic genes through variable assortment of their exons [ 19 ]. This process amplifies the coding potential of genomes and represents a plastic device for the regulation of gene expression. Splicing is catalyzed by the spliceosome, which forms by the ordered association of the U1, U2, U4/U6, and U5 spliceosomal small nuclear RNPs (snRNPs) and numerous non-snRNP proteins with the pre-mRNA [ 20 ]. Distinct spliceosome assembly/functional intermediates can be isolated including the E, A, B, B*, and C complexes. Upon catalytic activation, the B complex is converted to the activated B* complex, which lacks the U1 and U4 snRNPs. Complex C is formed after the first catalytic step of splicing. After the second step, the spliced mRNA and the excised intron are released [ 20 ]. Mass spectrometry (MS) analyses provided a comprehensive inventory of the protein composition of the A, B, B*, and C spliceosomal complexes [ 21–24 ] and revealed that a large number of proteins associate with the pre-mRNA in a highly dynamic manner. DHX9 was initially copurified with the human spliceosome in two distinct large scale proteomic analyses [ 25, 26 ]. Moreover, MS analysis of the spliceosome components associated with AdML and β−globin pre-mRNA, revealed the presence of DHX9 only in the first one, suggesting that it might be associated with only a subset of RNP species, or that it is loosely associated and, therefore, readily lost in more stringent condition that dissociate less stably associated components [ 27 ]. Subsequently, DHX9 was copurified with the human spliceosomal B complex, which is the spliceosome assembly intermediate that is structurally remodelled during catalytic activation [ 24 ]. In this study, we show that depletion of DHX9 displays different effects on the ability of ES cells to proliferate. Using transcriptome profiling, we identified exons, including many implicated in tumour formation and progression, whose splicing is modulated by DHX9. Immuno-depletion experiments demonstrated that DHX9 reduction impacts on the recruitment of U2 snRNP on the CTTN pre-mRNA, used as a paradigm of DHX9 splicing regulation. CTTN gene encodes the Cortactin protein, which is one of the major F-actin-binding proteins that regulates cytoskeletal dynamics by activating the Arp2/3 complex [ 28 ]. By dynamically assembling and disassembling their filamentous actin (F-actin) cytoskeletal network, eukaryotic cells can alter their shape and migrate. This process occurs both in physiological process, such as embryogenesis, and in pathological conditions, during tumour cell migration and metastasis [ 29 ]. Alternative inclusion of the exon 11 in the cortactin gene CTTN is one of the most recurrent isoform switches in multiple cancer types [ 30–32 ], thus highlighting the pivotal role of DHX9 in defining the tumour phenotype. Human TC-71 ( RRID:CVCL_2213 ) were maintained in Iscove’s modified Dulbecco’s medium (IMDM, GIBCO), supplemented with 10% fetal bovine serum (GIBCO) and penicillin/streptomycin 1% (GIBCO). Cell lines were obtained from DSMZ and tested negative for mycoplasma contamination. Transfection of Ewing sarcoma cell line was performed using RNAimax (Invitrogen) according to manufacter’s instructions. Cells were transfected with control siRNA (Sigma–Aldrich) and siRNA for DHX9 (Sigma Aldrich) at the final concentration of 25 nM. To obtain an additional DHX9 silencing, TC-71 cells were transfected with pGFP-V-RS vector control (shCTRL) or pGFP-V-RS_DHX9 vector (shDHX9) (Origene, Cat# TR30013 and Cat# TG316515, respectively), using Lipofectamine 2000 for 72 h. For the overexpression experiments, cells were transiently transfected with either GFP or GFP-DHX9, using Lipofectamine 2000 (Thermo Fisher Scientific), according to the manufacturer’s instructions and harvested after 48 h. When appropriate, TC-71 cells were treated with DHX9 inhibitor (DHX9-IN-1 Medchem Express, HY-156782) at the indicated concentrations for 48 h. For extract preparation, cells were washed twice with ice-cold phosphate buffered saline (PBS), resuspended in RIPA lysis buffer [150 mM NaCl, 50 mM Tris–HCl, pH 7.5, 2 mM EDTA, 0.1% in sodium dodecyl sulfate (SDS), 0.5% sodium deoxycolate, 1 mM dithiothreitol, 0.5 mM Na-ortovanadate, 1%, 10 mM B-glycerolphosphate, 10 mM sodium fluoride, 1% NP-40, and Protease-Inhibitor Cocktail (Sigma–Aldrich), and kept on ice for 10 min. Soluble protein extracts were separated by centrifugation at 12 000 rpm for 10 min and diluted in SDS sample buffer. Cell lysates were resolved by SDS-PAGE and transferred to PVDF membranes (GE, Healthcare). The membranes were blocked in TTBS (TBS with 0.1% Tween 20), containing 5% milk. Primary antibody incubations were performed in TTBS with either 5% BSA, overnight at 4°C. After washing, the membranes were incubated with the appropriate secondary peroxidase conjugated antibody for 1 h in TTBS. Proteins were visualized by chemiluminescence detection system (Clarity Western ECL Substrate). Following antibodies were used: anti-DHX9 (Santa Cruz Biotechnology Cat# sc-137232, RRID:AB_2261698 ), anti-GAPDH (sc-365062, RRID:AB_10847862 ), anti-CTTN (sc-55579, RRID:AB_831187 ), anti-ACTIN (sc-47778, RRID:AB_2714189 ), anti-β-TUBULIN (sc-5274, RRID:AB_2288090), anti-GFP (sc-9996), anti-SAM68 (Bethyl, Cat# A302-110A; RRID:AB_1604289 ), anti-TIA1 (C20 SC-1751), anti-PTB (sc-56701, RRID:AB_783822), anti-hnRNPC1/C2 (sc-32308, RRID:AB_627731 ), anti-U2AF65 (sc-53942, RRID:AB_831787), anti-U2AF35 (sc-514459), anti-SF3B1 (sc-514655), anti-SF3A2 (sc-390444), anti-SF3A3 (sc-393673), and anti-U1-70K (06-1297 Sigma–Aldrich). For RNA-seq analysis, Ewing sarcoma TC-71 cells were silenced by using siCtr or siDHX9 oligonucleotides. Forty-eight hours after transfection, RNA was extracted and DNAse treated using the miRNEasy extraction kit (Qiagen) according to manufacturer’s instruction. RNA sequencing was performed by IGA Technology services (Udine, Italy). PolyA plus RNA-seq libraries were constructed according to Illumina’s protocol and sequenced using a 100 bp single-end format on an Illumina HiSeq 2000. RNA-Seq data analysis was performed by GenoSplice technology ( https://www.genosplice.com/ ). Sequencing, data quality, reads repartition (e.g. for potential ribosomal contamination), and insert size estimation were performed using FastQC, Picard-Tools, Samtools, and rseqc. Reads were mapped using STARv2.4.0 [ 33 ]. Genes were considered as expressed if their rpkm value was >97.5% of the background rpkm value based on intergenic regions. Results were considered statistically significant for P -values ≤0.05 and fold-changes ≥1.5. Analysis at the splicing level was performed as previously described [ 34 ]: using the FAST DB splicing patterns annotation. All types of alternative events can be analyzed: alternative first exons, alternative terminal exons, cassette exon, mutually exclusive exons, alternative 5′ donor splice site, alternative 3′ acceptor splice sites, intron retention, internal exon deletion, and complex events corresponding to mix of several alternative event categories). ‘EXON’ and ‘PATTERN’ analyses were based on the splicing-index calculation as previously described [ 35 ]. Results were considered statistically significant for P -values ≤0.05 and fold-changes ≥1.5 for ‘PATTERN’ and P -values ≤0.05 and fold-changes ≥2.0 for ‘EXON’. Analysis for enriched GO functional clusters among IR and other AS regulated genes was performed using DAVID Functional annotation Tool (v6.8), using the expressed genes as background [ 36 ]. The data generated in this study are publicly available in Gene Expression Omnibus (GEO) GSE214257. Analysis of 3′ and 5′ splice-sites strengths and GC context of regulated exon cassette was performed as previously described [ 37, 38 ]. Polypyrimidine tract length/score and distance of branchpoint from acceptor splice-site were evaluated using SVM-BPfinder [ 39 ]. SVM-BPfinder ( https://bitbucket.org/regulatorygenomicsupf/svm-bpfinder/src/master/ ) was used to estimate branchpoint and polypyrimidine tract strength and other statistics. Scores were estimated using the sequence of introns to the 3′end of exon between 20–500 nt. The results were filtered according to the instructions in the official website of SVM-BPfinder ( https://bitbucket.org/regulatorygenomicsupf/svm-bpfinder/src/master/ ). Search for enriched motifs within regulated cassette exons and their comparison with the compendium of RNA-binding motif from [ 40 ] were performed using DREME and Tomtom Motif comparison tools from the MEME Suite Collection [ 41, 42 ]. Analysis for enriched GO functional clusters among AS regulated genes was performed as previously described [ 37 ]. Clusters were considered as enriched if fold enrichment ≥2.0 and P -value ≤0.05. For splicing feature analysis in order to calculate the following values: MaxEnt Score Acceptor, MaxEnt Score Donor, Delta MaxEnt Score Acceptor-Donor, the tool called MaxEntScan [ 43 ] was used to estimate the strength of 3′ and 5′ splice sites. 5′ splice site strength was assessed using a sequence including 3 nt of the exon and 6 nt of the adjacent intron ( http://hollywood.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq.html ). 3′ splice site strength was assessed using a sequence including—20 nt of the flanking intron and 3 nt of the exon ( http://hollywood.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq.html ). The pcDNA3-CTTN exon 10–exon 13 minigene was generated using the forward primer 5′-AGGAATTCGGTGTGCAGACAGACAGACAAGACA-3′ and the reverse primer 5′-TCCTCGAGCTTATCCATCCGATCCTTCTGCACC-3′ from TC-71 cell genomic DNA and cloned into the EcoRI/ XhoI restriction sites of pcDNA3.1(−) vector. Subsequently, the construct has been analyzed by direct sequencing (Eurofins Genomics). The template for RNA synthesis was generated using PCR product from pcDNA3.1- CTTN exon 10–exon 13 minigene, amplified with forward primer containing T7 promoter sequence at the 5′-end upstream from sequence in intron-10 of CTTN and reverse primer complementary to the 3′-end of intron-11. Amplified band was purified from gel and used as template (500 ng) for in vitro RNA transcription using the MEGAscript T7 Transcription Kit (ThermoFisher Scientific) in the presence of biotinilated nucleotides. RNA was purified using Illustra MicroSpin G-25 columns (GE Healthcare) and diluted in RNAse-free water (Sigma–Aldrich) for subsequent biotin-RNA pulldown with immunodepleted nuclear extracts experiments. TC-71 cells were collected in PBS and resuspended in hypotonic buffer A (10 mM Tris–HCl, pH 7.9, 1.5 mM MgCl 2, 10 mM KCl, 0.05% NP-40). Cells were manually homogenized by 10 iterations of up and down passes of a pestle. After incubation on ice for 10 min, samples were centrifuged at 1000 g for 7 min at 4°C. Supernatant (cytoplasmic fraction) was re-centrifuged at 1000 g for 3 min at 4°C to pellet debris. Pellets were washed with buffer A without detergent and lysed with nuclear extraction buffer B (10 mM Tris–HCl, pH 7.9, 1.5 mM MgCl2, 300 mM KCl, 0.5% NP-40, 10% glycerol) for 30 min at 4°C on a rotating platform. The insoluble and soluble nuclear fractions were separated by centrifugation at 12 000 g for 30 min. Nuclear extracts were diluted in buffer B without KCl to halve salt concentration. Two microgram anti-DHX9 monoclonal antibody (Santa Cruz Biotechnology Cat# sc-137232, RRID:AB_2261698 ) was incubated overnight with 60 μl of 50% slurry protein G Dynabeads (Invitrogen) in PBS supplemented with protease inhibitor cocktail (Roche) at 4°C with continuous mixing. As a control, 60 μl of 50% slurry protein G Dynabeads were coupled with 2 μg IgG from mouse serum (Sigma–Aldrich Cat# I5381, RRID:AB_1163670 ). The protein G Dynabeads were then recovered with a magnet and three washes with PBS to remove excess anti-DHX9 antibody. Beads were then incubated overnight with 800 μg of TC-71 nuclear extract at 4°C with continuous mixing. The supernatant was used as DHX9-depleted TC-71 nuclear extract. Elution of DHX9 protein from the immune-absorbed beads was made by adding 20 μl of 4 × SDS sample buffer, followed by boiling 10 min. The eluent and aliquot of the DHX9-depleted TC-71 nuclear extracts were subjected to gradient SDS-PAGE gel and Western blot analysis. Biotin-RNA pulldown experiments were carried out using TC-71 immunodepleted nuclear extracts and the CTTN biotinylated RNA oligonucleotide prepared as described above. Nuclear extracts were diluted in pulldown buffer (10 mM Tris–HCl, pH 7.4, 2.5 mM MgCl 2, 100 mM KCl, 0.5% NP-40) plus protease inhibitor cocktail (Sigma–Aldrich) and pre-cleared for 1 hour on protein-A–Sepharose beads in the presence of 0.25 mg/ml yeast tRNA and 0.01% BSA, to reduce nonspecific protein-RNA interactions. In contemporary, streptavidin–Sepharose beads were pre-incubated for 1 hour with 0.01% BSA and the biotinylated RNA oligonucleotide in pulldown buffer. Incubation was performed for 1 hour at 4°C under rotation. Beads bound to biotinylated RNA were washed once with pulldown buffer plus protease inhibitor cocktail and RNAse inhibitor and incubated with pre-cleared extracts for 45 min at 30°C in a shaking incubator, in presence of ATP. Bound complexes were cross-linked on ice by irradiation (400 mJ) in a Stratalinker 2400 at 254 nm and rinsed three times with pulldown buffer plus protease inhibitors. Half of the complexed beads were eluted in an SDS–PAGE sample buffer for analysis by Western Blot, and the other half were subjected to phenol/chloroform RNA extraction and ethanol precipitation according to standard procedures, for a subsequent RT-qPCR analysis. Nuclear cell lysates, obtained as previously described, were incubated with anti-DHX9/Sam68/TIA1/PTB or HNRNPC1/C2 antibodies or purified mouse/rabbit IgGs (Thermo Fisher Scientific) and protein G Dynabeads (Thermo Fisher Scientific) at 4°C, overnight. Where indicated, TC-71 extracts were treated with Benzonase Nuclease (Millipore, #70746) at 50 U/ml for 2 hours at 4°C. After stringent washes, isolated immunocomplexes were analysed by Western blot analysis, as previously described [ 44 ]. Cells cultured on coverslips were fixed with 4% paraformaldehyde for 20 min, permeabilized with 0.2% Triton-X-100, for 7 min and blockage with 5% BSA in PBS, for 1 hour. Antibody against CTTN was used followed by the anti-mouse Alexa-Fluor-594 secondary antibody (Thermo Fisher Scientific). Actin cytoskeleton was visualized using Alexa Fluor 488-phalloidin (Thermo Fisher Scientific). Nuclei were stained using 4′,6′-diamidino-2-phenykindole (DAPI). Coverslips were mounted with Vectashield mounting medium for fluorescence. 1.0 × 105 cells were dispensed in the cell culture transwell inserts and were left to migrate for 16 hours at 37°C. After the incubation period, the medium was removed from inserts, and the migrated cells were fixed with 70% ethanol, at room temperature for 10 min. Then, transwell membranes were stained for 7 min with 0.2% of crystal violet diluted in water. Transwell inserts with 8 μm pore size were washed twice with water to remove the excess crystal violet and allowed to dry and analysed with EVOS M5000 imaging system. Animal procedures were performed according to European Guidelines (2010/63/EU) and Italian law requirements (D.lgs. 26/2014) and were approved by the National Ethics Committee for Animal Experimentation of the Italian Ministry of Health (Authorization N° 1170/2020-PR). Briefly, 4-weeks old female NOD/SCID mice from Charles River Laboratories (Cat# 005 557 RRID:IMSR JAX:005 557) were used. Animal sample size was calculated with the G* Power software. Mice were maintained under housing and feeding conditions according to the standard animal facility procedures and the ethics committee of the Fondazione Santa Lucia. Mice were housed with a 12 hours light/dark cycle, under controlled temperatures (20–22 °C) and received water and food ad libitum. 1.5 × 106 TC-71_shCNT or TC-71 _sh DHX9 cells were injected IM in the gastrocnemius muscle of anesthetized mice. Tumour mass formation was evaluated and measured by a caliper. Tumour volume was calculated by the formula: V = 1 2 (Length × Width2 ). At day 28 after tumour injection, mice were euthanized by an overdose of anesthetic and tumours were collected, weighted and frozen for immunohistochemical analyses. Immunohistochemical analyses were performed as previously described [ 12 ]. Briefly, samples were fixed in 4% formaldehyde buffered with Phosphate Buffer for 24 hours at 4°C and processed by Leica HistoCore Processor. Samples included in Paraffin Blocks were cut by Leica SM 2000R microtome in 3 μm thickness sections. Histological Haematox/Eos staining was performed after timely dewaxing and rehydrating procedures. IHC was performed after antigen retrieval procedure by citrate buffer, pH 6.0, at high temperature using microwaves horn for 5 min 550 W (2 cycles) and peroxidases blocking with Vector Bloxall solution. Saturation of nonspecific sites was performed with Vector 2.5% Normal Horse Serum, also used for antibody dilution. Cortactin antibody was used at 1:200 dilution at 4°C, overnight. BA-1000 Biotinylated secondary anti-mouse antibody (1:200) was incubated for 1–2 hours at RT and Vectastain Elite ABC Kit (in PBS) was incubated for 1 hour at RT. Sections were counterstained with Hematoxylin and observed with Zeiss Axiophot Microscope. Statistical analyses were conducted with the GraphPad Prism (GraphPad Prism, RRID:SCR_002798 ) and the values represent mean ± SD of three independent experiments. The statistical significance of the differences was determined by the Student’s t test or analysis of variance test. To evaluate the impact of DHX9 on the transcriptome of Ewing sarcoma cells, we performed RNA-seq analysis of TC-71 cells depleted or not of DHX9. SiRNA-mediated knockdown of DHX9 yielded significant depletion (∼65%) of DHX9 at both RNA and protein level (Fig. 1A and B ). Transcriptome analysis revealed a widespread regulation elicited by DHX9 downregulation. Particularly, we identified 648 genes regulated at the expression level, with 58% downregulated and 42% upregulated genes ( Supplementary Fig. S1A and Supplementary Table S1 ) [ 12 ]. Comparative bioinformatics analyses also identified 426 genes regulated at splicing level, with a significant but limited overlap between splicing and expression regulated genes (Fig. 1C ; Supplementary Tables S1 – S2 ). These results suggest a multi-layered impact of DHX9 on the transcriptome of TC-71 cells. Among the 524 splicing events that are susceptible to DHX9 expression ( Supplementary Fig. S1B ), exon cassettes (16.9%) and alternative terminal exons (12.7%) were the most represented splicing patterns (Fig. 1D ). We also detected a large fraction of alternative first exon events (10.2%), which likely reflect differential promoter choice. Among the splicing events affected by DHX9 knockdown, we found several exons that were not previously annotated as alternative (indicated as unknown) (Fig. 1D ), suggesting that depletion of DHX9 leads to generation of aberrant RNA variants, possibly due to decreased fidelity of splice-site recognition by the spliceosome. DHX9 impacts on alternative splicing choices in Ewing sarcoma cells. ( A ) RT-qPCR showing the levels of DHX9 transcript in TC-71 ES cells after 48 h of transfection with either siControl (siCtrl) or siDHX9 oligonucleotides. Values are the mean ± SD of three independent experiments, each performed in triplicate. Statistical analysis was performed by Student’s t -test. ( B ) Western blot showing the DHX9 protein levels in TC-71 cells after 48 h of transfection with either control siRNA or siDHX9, normalized to GAPDH expression. Values are the mean ± SD of three independent experiments, each performed in triplicate. Statistical analysis was performed by Student’s t -test. ( C ) Venn diagram showing overlap between transcriptional- and splicing- regulated genes, identified by comparison of siDHX9 versus siCtrl TC-71 ES cells ( P < 5.053e-06). ( D ) Bar graph showing percentages of AS pattern differentially regulated between siDHX9 and siCtrl. ( E ) Bar graph showing Fold enrichment of Gene Ontology categories (GO) of differentially splicing-regulated genes enriched in siDHX9 TC-71 cells. Histograms represent the fold enrichment score and the -log 10( P- values). GO analysis of splicing-regulated genes highlighted a specific enrichment in functional categories associated with cancer cell migration and proliferation, such as cell-substrate junction assembly and cell cycle (Fig. 1E ). Analysis of eight randomly chosen events by RT-PCR confirmed the splicing changes identified by RNA-sequencing (RNA-seq) in TC-71 cells (Fig. 2 ). Moreover, overexpression of GFP-DHX9 confirmed the regulation of the identified events by the helicase ( Supplementary Fig. S2 ). These observations suggest that DHX9 expression elicits an impact on the transcriptome of Ewing sarcoma cells by affecting splicing choices. DHX9 affects alternative splicing choices of ES cells. ( A–H ) Representative images of the PCR analyses for the indicated alternative splicing events differentially regulated between TC-71 cells transfected with either siCtrl or siDHX9. Schematic representation for each event analyzed is depicted above the representative agarose gel. Green and red boxes indicate the regulated exons in siDHX9 compared with control TC-71 cells, either skipped (green) or included (red). Black arrows in the scheme indicate primers used for the PCR analysis of EPB41L2, TNIP1, SPAG9, C2CD5, CTTN, CLSTN1, ABCB6, and KIAA0753 event. The graphs show the densitometric analysis of the ratio between isoforms with included and skipped exons. P -values were determined by Student’s t -test * P ≤ 0.05, ** P ≤ 0.01, and *** P ≤ 0.001 ( n = 3; mean ± SD). To investigate the mechanism of splicing regulation by DHX9, we first explored the structural and sequence features of the regulated cassette exons. DHX9-sensitive cassette exons were mostly downregulated (79% downregulated upon DHX9 knockdown versus 21% upregulated; Supplementary Fig. S1B ) and displayed regular exon length when compared to both reference alternative and constitutive exons (Fig. 3A ). Furthermore, DHX9-sensitive downregulated exon cassettes displayed a shorter upstream intron compared to reference cassette exons (Fig. 3B ), whereas the upregulated cassette exons showed a shorter downstream intron compared to reference (Fig. 3C ). Other sequence features, including the distance between the branch point and the polypyrimidine tract, the length and score of the polypyrimidine tract, the distance between the branch point and the acceptor site, and the percentage of pyrimidine between the branch point and the acceptor site, were not significantly different with respect to those of reference alternative exons ( Supplementary Fig. S3 ). Moreover, the strength of the acceptor and donor splice sites obtained by using the maximum entropy scores [ 43 ] was also not different in DHX9-sensitive cassette exons (Fig. 3D – F ). Interestingly, however, DHX9-sensitive upregulated exons showed a weaker branch point score lower than constitutive exons ( Supplementary Fig. S3E ). Structural features of DHX9-sensitive exons. Box plots representing comparison between up- regulated, down-regulated, reference (not-regulated in our experiment), and constitutive exons for size ( A ), previous intron size ( B ), following intron size ( C ), MaxEnt Donor Score ( D ), MaxEnt Acceptor Score ( E ), and MaxEnt Acceptor -Donor Score ( F ). P -values are measured by Fisher test. To identify RBPs possibly involved in the splicing modulatory function of DHX9, we searched for RNA motifs enriched in the DHX9-regulated exons and flanking introns compared to both non-regulated cassette exons and constitutive exons ( Supplementary Tables S3 – S4 ). For all regulated exons, the associated enriched pentamers were always more recurrent within the exon cassette than in the flanking introns (Fig. 4A ). Several sequence motifs were significantly enriched in the regions nearby or within DHX9-downregulated exons ( Supplementary Table S4 ; Fig. 4B ), with AAUUU and UUUUG being the most significant (Adjusted P -value <0.05) sequence elements in down-regulated exons. Next, by using the Tomtom Motif Comparison tool, ( https://meme-suite.org/meme/tools/tomtom ) we screened a compendium of known RBPs for their potential affinity for the identified motifs [ 40 ]. Notably, the AAUUU and UUUUG motifs displayed highest similarity to the binding sites for splicing factors that are known to act in a position-dependent manner, such as HNRNPC, HNRNPCL1, and TIA1 (Fig. 4B ). Next, the Z score was calculated in two different ways: by counting the number of exons with an enriched pentamer and by counting the number of pentamers in each exon ( Supplementary Table S5 ). Z-score calculation confirmed binding motifs for TIA1 and HNRNPC as the most represented in DHX9-sensitive cassette exons, even though significant enrichment of other pentamers was also identified, compatible with the binding specificity of other RBPs, such as PTBP1 and KHDRBS1 (Sam68) (Fig. 4C and Supplementary Table S5 ). The distribution of the identified motifs was different in exons regulated in opposite manner, suggestive of a positional regulatory effect of the cognate splicing factor ( Supplementary Table S5 ). In particular, the AAUUU motif was mainly exonic in the downregulated exons, whereas the UUUUG motif was mainly localized in the downstream intron of downregulated exons ( Supplementary Table S5 ). Sequence features of DHX9-sensitive exons. ( A ) Number of enriched pentamers in the regulated exons or in the nearby regions (downstream and upstream). ( B ) Most significant sequence motifs in the regions nearby or within DHX9-sensitive exons (adjusted P -value <0.05). Tomtom Motif Comparison tool, ( https://meme-suite.org/meme/tools/tomtom ) identified the AAUUU and UUUUG motifs as binding sites for HNRNPC, HNRNPCL1, and TIA1. ( C ) Z score was calculated by counting the number of exons with a pentamer (upper part) and by counting the number of pentamers in each exon. ( D ) Immunoprecipitation (IP) experiments were performed to evaluate the interaction between DHX9 and the identified splicing factors in panel (C). Immunoprecipitated proteins were separated by Western blot analysis. The interaction of HNRNPC, Sam68, PTB, and TIA1 with DHX9 was evaluated. Given the potential involvement of HNRNPC, Sam68, PTB, and TIA1 in the regulation of DHX9-sensitive exons, we verified their interaction with DHX9 in IP experiments. Purified immunocomplexes confirmed that DHX9 interacts with HNRNPC, Sam68, and TIA1 in TC-71 ES cells (Fig. 4D ), whereas no specific interaction with PTBP1 was detected. Next, we sought to investigate the possible mechanism underlying DHX9 regulation of alternative splicing. Spliceosome assembly occurs through the sequential association and release of snRNPs on the pre-mRNA [ 20 ]. The ATP-independent binding of U1snRNP through base-pairing with the 5′ splice site and the recruitment of SF1 and U2AF65 to, respectively, the branchpoint and the polypyrimidine tract at the 3′ splice site represent the earliest steps in spliceosome assembly and they correspond to the E or commitment complex. After formation of the E-complex, U2 snRNP stably associates with the branchpoint region in an ATP-dependent manner, which involves base-pairing interactions between U2 snRNP and sequences flanking the branchpoint adenosine, leading to the formation of the A complex [ 20 ]. Previous observations suggested a role for DHX9 in association with the spliceosomal A complex and with complex B [ 21, 24, 27, 45 ]. In particular, DHX9 was initially purified with the A complex under low stringency condition [ 45 ] and confirmed by tobramycin affinity selection [ 21 ], in line with a possible interaction with U2 components in the earliest steps of spliceosome assembly. To test this hypothesis, co-IP assays were performed and showed that DHX9 does not form complexes with U1 70K and U2AF in ES cells (Fig. 5A ). On the contrary, we found that DHX9 interacts with SF3A2 and SF3B1 (Fig. 5A ) and benzonase (50 U/ml) treatment did not affect this interaction (Fig. 5B ), suggesting a role for DHX9 in the early U2 recruitment onto the pre-mRNA. DHX9 immuno-depletion affects U2snRNP recruitment on CTTN pre-mRNA. ( A ) Western blot analysis of SF3B, U1-70K, SF3A2, SF3A3, U2AF65, U2AF35, and DHX9 after IP of endogenous proteins from TC-71 nuclear extracts with anti-DHX9 or control mouse IgGs. The experiments were performed at least three times; statistical analysis was performed by Student’s t- test ( P < 0.01 **). ( B ) IP experiments were performed to evaluate the interaction between DHX9 and the splicing factors identified in A after Benzonase treatment (50 U/ml, for 2 h). ( C ) TC-71 cells were treated with DHX9 inhibitor for 48 hours at the indicated concentrations and PCR analyses for the CTTN alternative splicing event was performed. The graphs show the densitometric analysis of the ratio between isoforms with included and skipped exons. P -values were determined by Student’s t -test * P ≤ 0.05, ** P ≤ 0.01, and *** P ≤ 0.001 ( n = 3; mean ± SD). ( D ) DHX9 expression has been evaluated by RT-qPCR. ( E ) In the upper part, scheme of the affinity chromatography assays from TC-71 nuclear extracts using biotinylated RNA corresponding to CTTN exon 11 pre-mRNA and flanking intronic sequences. Pull-down assays of TC-71 nuclear extracts immunodepleted using either purified rabbit IgGs or anti-DHX9 antibody. The RNA pull-down was performed using in vitro transcribed biotinylated RNA corresponding to CTTN exon 6 pre-mRNA and flanking intronic sequences spanning from −79 to +72 nucleotides as a bait. After extensive washes, pulled-down proteins were analyzed by Western blot using antibodies against DHX9, U170K, U2AF65, U2AF35, SF3B1, and SF3A2. On the right, quantification of pull-down results by densitometric analysis. Bars indicate mean values ± SD of three independent experiments. Significance: binding to CTTN pre- mRNA sequences was evaluated using Student’s t -test:∗ P < 0.001. ( F ) Quantification of pull-down of U1, U2, U4, and U6 snRNPs by RT-qPCR. Bars indicate mean values ± SD of three independent experiments. Significance: cross-link to CTTN pre- mRNA sequences was evaluated using Student’s t -test:∗ To further test our hypothesis, we selected CTTN splicing event as a paradigm of DHX9 splicing regulation, because it was downregulated upon DHX9 knockdown and displayed a short upstream intron, according to the features of DHX9 regulated exons. To corroborate our findings, we confirmed the downregulation of the CTTN alternative exon 11 by using a different set of oligonucleotides to decrease DHX9 expression ( Supplementary Fig. S4A and B ). Moreover, we performed overexpression experiments of either GFP or GFP-DHX9 together with a minigene spanning from exon 10 to exon 12 of CTTN genomic region. PCR analyses confirmed the effect of DHX9, leading to the inclusion of the alternative exon in DHX9-overexpressing cells ( Supplementary Fig. S4C and D ). Conversely, treatment with the DHX9 inhibitor (DHX9-IN-1) impaired exon 11 inclusion in a dose dependent fashion (Fig. 5C ), without affecting DHX9 expression (Fig. 5D ). In line with the features identified in DHX9-regulated exons, CTTN exon 11 and the upstream intron 10 contain binding sites for HNRNPC, Sam68, and TIA1. Furthermore, knockdown of Sam68, TIA1, and HNRNPC in TC-71 affects splicing of CTTN transcript ( Supplementary Fig. S5 ), thus supporting the hypothesis of a cooperative action of DHX9 with these RBPs. Next, we employed an affinity chromatography approach using the CTTN exon 11 pre-mRNA (−79 to +74) as a bait (Fig. 5E, upper scheme) to monitor the recruitment of spliceosomal proteins onto the pre-mRNA [ 46 ]. Biotinylated RNAs were synthesized in vitro and incubated with TC-71 nuclear extracts, previously immunodepleted with either purified rabbit IgGs or anti-DHX9 antibody, and the complexes associated with the RNAs were pulled down using streptavidin beads. Proteins retained after stringent washes were eluted and analysed by SDS-PAGE and Western blot. We observed that DHX9 is recruited on the CTTN pre-mRNA together with U170K, U2AF65, and U2AF35 (Fig. 5E ). While DHX9 immunodepletion did not affect the association of U170K, U2AF65, and U2AF35 with the CTTN pre-mRNA, it led to a significant decrease in the recruitment of SF3A2 and SF3B1 (Fig. 5E ). Next, RT-qPCR analysis was performed on the RNA retained on streptavidine beads to monitor snRNA recruitment. DHX9 immunodepletion caused a small but significant reduction in the amount of U1 spliceosomal RNA (∼25%) that associated with the beads, whereas U2 snRNA recruitment showed 90% decrease (Fig. 5F ), in line with the reduced recruitment of SF3B1 and SF3A2, that are crucial for the correct base-pairing of U2 snRNA on the branchpoint site [ 47 ]. U4 and U6 recruitment was also reduced by 70% and 80% respectively, paralleling the effect on U2snRNP (Fig. 5F ) and further suggesting that DHX9 is important for the assembly of a catalytically active spliceosome. Collectively, these results suggest that DHX9-sensitive exons display impaired spliceosome assembly due to defective recruitment of SF3A2, SF3B1, and U2 snRNA at the 3′splice site. CTTN gene encodes the cortactin protein, which is required for the formation of invadopodia and filopodia-like overhangs, associated with matrix proteolytic degradation and cell invasion [ 48, 49 ]. Inclusion of exon 11 in cortactin transcript was reported to enhance the migratory and invasive behaviour of cancer cells [ 50, 51 ]. Thus, we asked whether in ES cells, the inclusion of exon 11 in CTTN transcript induced upon DHX9 silencing (Fig. 2E ) might affect the shape and migration capability. As first, Western blot analysis confirmed the expression of a smaller protein isoform detected by the anti-cortactin antibody, corresponding to Δ11 cortactin splice variant, induced in cells depleted for DHX9 (Fig. 6A ). In line with the RT-PCR results, also showing no changes in the steady state levels of the CTTN mRNAs by using with primers in a common region ( Supplementary Fig. S6A ), DHX9 silencing was associated with an increment of ∼50% of Δ11 cortactin protein isoform (Fig. 6A ). As a control of the antibody specificity, both isoforms disappeared when CTTN knockdown was performed in ES cells with oligonucleotides targeting a constitutive common exon, as reported in Supplementary Fig. S6B ( Supplementary Table S6 ). Remarkably, co-staining for cortactin and the actin cytoskeleton (phalloidin) indicated that DHX9 depleted cells display an altered morphology (Fig. 6B ), in line with previous reports documenting that the skipping of exon 11 in cortactin impacts on F-actin polymerization activity [ 50 ] and F-actin bundle formation in filopodia [ 52 ]. These results suggest that in siDHX9 cells there is a diminished migration. To test this hypothesis, cell migration assay was performed, showing that siDHX9 cells displayed a significant reduction in their migratory capacity, in comparison with control cells (Fig. 6C and D). To strengthen our observations, control and DHX9-depleted TC-71 cells were injected into NSG mice at the level of the gastrocnemius muscle and tumour volume was evaluated in the following 28 days. DHX9 knockdown significantly reduced the rate of tumour growth in mice [ 12 ], as well as tumour volume and weight (Fig. 6E and F). Moreover, immunohistochemical analysis of cortactin indicated a different staining in DHX9-depleted tumours (Fig. 6G ), paralleling the increase of the Δ11 protein isoform (Fig. 6H ). Thus, downregulation of DHX9 expression in ES cells is sufficient to impair their oncogenic properties in vitro and to reduce tumour growth in vivo. DHX9 knockdown modulates cellular morphology via CTTN alternative splicing. ( A ) Western blot analysis showing CTTN protein isoforms upon DHX9 silencing. Ten micrograms of total protein extracts were loaded in each lane. β-actin content was used as loading control. ( B ) Representative pictures showing cellular morphology after transfection with either siCTRL or siDHX9 oligonucleotides. Phalloidin staining was used to detect filipodia, DAPI was used for DNA staining. ( C ) TC-71 cells were transfected as in panel (A) and migration assay was performed. The crystal violet-stained migrating cells were photographed and counted. ( D ) Values are the mean ± SD of three independent experiments, each performed in triplicate, considering the migration of siCtrl as 100%. Statistical analyses were performed by Student’s t -test. * P ≤ 0.05; ** P ≤0.01; and *** P ≤ 0.001. ( E ) Measurement of tumour xenograft from shCTRL and shDHX9 mice at 28 days after injection. Statistical analysis was performed by Student’s t -test, P- values: **** P ≤ 0.0001. ( F ) Tumour xenografts were collected from shCTRL and shDHX9 mice ( n = 8 for each group) and weighed. Graph represents the mean of the weights ± SD. Statistical analysis was performed by Student’s t -test, P- values: **** P ≤ 0.0001. ( G ) Representative IHC images of Cortactin staining in tumour xenografts originated from shCTRL and shDHX9 cells. Scale bar: 10 μm. ( H ) Western blot analysis for the expression of Cortactin and Tubulin from xenografts formed by shCTRL and shDHX9 TC-71 cells. On the right, densitometric analysis to evaluate full length and Δexon 11 isoforms of Cortactin. Statistical analysis was performed by Student’s t -test, P- values: ** P ≤ 0.01. Our results also suggest that splicing regulation of DHX9, in addition to its role in transcriptional regulation [ 12 ] is crucial for the cytoskeletal architecture and the aggressiveness phenotype of ES cells. To further test the role of CTTN splicing in cell migration, TC-71 ES cells were transfected with oligonucleotides directed against exon 11, to selectively downregulate the main CTTN isoform, without affecting the Δ11 isoform (Fig. 7A ). PCR analysis confirmed a significant reduction of the main isoform of CTTN, without any perturbation of the Δ11 isoform (Fig. 7A ). Migration assay showed 25% reduction of cell migration (Fig. 7 B and C) confirming the role played by CTTN splicing in the migration capacity of ES cells. CTTN alternative splicing affects cell migration. ( A ) PCR analysis to monitor CTTN splicing from TC-71 cells transfected with either control or si CTTN oligonucleotides directed against the isoform 1 of CTTN containing exon 11. ( B ) Migration assay of siCtrl and siCTTN. Crystal violet-stained migrating cells were photographed and counted. ( C ) Histograms represent the mean ± SD of three independent experiments, each performed in triplicate, considering the migration of siCtrl as 100%. Statistical analysis was performed by Student’s t -test. * P ≤ 0.05; ** P ≤0.01; and *** P ≤ 0.001. ( D ) Hypothetical model of DHX9 splicing regulation (upper part) and its impact on cancer cell migration and invasion (lower part). In this study, we demonstrate that the RNA/DNA helicase DHX9 writes a distinct transcriptome signature in Ewing sarcoma cells, impacting the proliferation capacity and acquisition of the tumorigenic phenotype. This specific signature is attributable, at least in part, to its impact on splicing regulation. Using transcriptome profiling, we identified exons, including many implicated in tumour formation and progression, whose splicing is modulated by DHX9. Employment of transcriptome profiling by RNA-seq analyses has represented a striking breakthrough in cancer research [ 53 ]. Comprehensive expression profiling of cancer cells has highlighted hallmarks of cancer previously unexplored and revealed that tumours display considerably wider transcriptome diversity than corresponding normal tissues [ 54 ]. Such increased complexity is mainly caused by widespread dysregulation of alternative splicing, yielding oncogenic splice variants that promote tumorigenesis and chemoresistance [ 55, 56 ]. However, aberrant splice variants also represent a vulnerability of tumour cells that can be exploited therapeutically [ 57, 58 ]. Our study discloses the impact of DHX9 on U2snRNP recruitment onto the pre-mRNA. These results are in line with seminal studies by the Luhrmann laboratory devoted to the proteomic characterization of spliceosome complexes [ 21, 23, 24, 27, 45 ], which revealed DHX9 interaction with spliceosomal complexes. Spliceosome assembly is initiated by the ATP-independent interaction of the U1 snRNP with the conserved 5′ splice site of the pre-mRNA, forming the E complex. At this stage, the U2 snRNP is loosely associated with the pre-mRNA [ 20 ]. In a subsequent step requiring ATP, the U2 snRNP stably interacts with the pre-mRNA’s branch site, leading to formation of the A complex (also called the pre-spliceosome). Spliceosome assembly culminates with the formation of the spliceosomal B complex, during which the preformed U4/U6.U5 tri-snRNP particle interacts with the A complex. The B complex thus contains a full set of U snRNAs in a precatalytic state. It subsequently undergoes major rearrangements, including destabilization or loss of the U1 and U4 snRNPs, leading to catalytic activation and the formation of the so-called activated spliceosome (B* complex). DHX9 was initially purified with the A and B spliceosomal complexes [ 21, 45 ]. These observations are in line with our results documenting DHX9 interaction with the U2snRNP components SF3B1 and SF3A2. Our data also reveal that DHX9-sensitive exons display a shorter previous intron than reference alternative exons, very similar to constitutive exons. Exon-intron structure plays a role in the recognition of exons by the splicing machinery [ 59 ]. I n vitro and in vivo studies in flies and vertebrates have demonstrated that intron length negatively affects exon inclusion levels, and alternatively spliced exons are flanked by longer introns compared with those flanking constitutively spliced exons [ 59 ]. However, splicing kinetics seems to be more affected by intron length in lower eukaryotes, such as yeast and flies, where the removal of the relatively shorter introns presumably occurs via intron definition, while mammalian exons flanked by longer introns are mainly identified via exon definition [ 60, 61 ], and the rate of intron removal from the pre-mRNA seems unrelated to intron length [ 62 ]. U2snRNP plays essential role in intron definition and assemblies dynamically in a multistep process [ 63, 64 ]. The 12S U2 snRNP and SF3b associate to form a particle of 15S; addition of SF3a to the 15S U2 snRNP results in the assembly of the 17S particle [ 63 ]. U2 snRNP functions in splicing by converting the early complex E into pre-splicing complex A [ 65 ]. Complex E is formed by binding of U1 snRNP to the 5′ splice site and interaction of splicing factors U2AF and SF1 with the polypyrimidine tract upstream of the 3′ splice site and the branch site, respectively. The formation of complex A occurs by subsequent binding of U2 snRNP to the pre-mRNA, which involves the formation of a short helix between a single-stranded region in U2 snRNA and the intron branch site, which requires the 12S U2 snRNP plus SF3a and SF3b [ 66 ] or the isolated 17S U2 snRNP [ 67 ], while the 12S U2 snRNP alone is inactive in complex assembly. Thus, SF3a and SF3b are essential for pre-spliceosome assembly in combination with the 12S U2 snRNP, and three RNA-helicases, DDX42, DDX46, and DHX15, are functionally associated with U2 snRNP [ 68 ]. Since these interactions are thought to stabilize the U2 snRNA/branch site helix and may also aid to incorporate the U2 snRNP into the spliceosome, it is possible that this function is carried out by DHX9 in the DHX9-sensitive exons. As inferred by the GO analysis, several transcripts affected by DHX9 knockdown are involved in cell adhesion and migration [ 69, 70 ]. We focused our attention on CTTN gene, encoding the F-actin‐binding protein cortactin, responsible for F-actin polymerization and cytoskeletal rearrangements at cell–cell and cell-substratum interaction sites of migrating cells [ 71, 72 ]. We found that DHX9 depletion promotes the Δ11 isoform, previously shown to be involved in cell migration and invasion [ 50 ]. Notably, both RBFOX2 and PTBP1 have been shown to regulate CTTN splicing [ 31, 32, 51 ], thus affecting tumour cell migration and invasion. This splicing isoform lacks the 6th repeats in the F-actin binding domain of cortactin. Although both splice variants behave similarly with respect to subcellular localization and tyrosine phosphorylation, they differ significantly in their ability to induce cell migration, to bind and cross-link F-actin, and to promote Arp2/3-mediated actin polymerization [ 50 ]. Thus, alternative splicing of cortactin represents a mechanism to modulate actin dynamics and cell migration of ES cells (Fig. 7D ). In summary, our results indicate that DHX9 is important in establishing the invasive phenotype of Ewing sarcoma cells through its regulation of an oncogenic alternative splicing program. Further studies are needed to clarify the role of DHX9 during spliceosome assembly and complex A formation on target exon. The authors wish to thank Prof. Claudio Sette for sharing of expertise and reagents. Author contributions: The execution and analyses of most of the experiments were performed by VF, VR, LC, and RP. VF and MPP participated in the analysis of gene expression data. D.B. performed the immunohistochemical analysis. MPP supervised and guided the entire project. VF and MPP wrote the manuscript.
  Content Reference URL	None

• Publisher

  None

• Data Source

  Data Source Name	Quantexa News (#5)
  Data Source Type	news
  Data Source URL	https://app.aylien.com/dashboard
  Description	Welcome to Quantexa News API — the most powerful way to aggregate, filter, and integrate global news data into your apps and models.
  Weight	10