Downregulation of Lnc-Spry1 mediates TGF-β-induced epithelial-mesenchymal transition by transcriptional and posttranscriptional regulatory mechanisms

Downregulation of Lnc-Spry1 mediates TGF-β-induced epithelial–mesenchymal transition by transcriptional and posttranscriptional regulatory mechanisms Cristina Rodríguez-Mateo 1 , Belén Torres 1 , Gabriel Gutiérrez 2 and José A Pintor-Toro* ,1 Long non-coding RNAs (lncRNAs) are a class of regulatory genes that participate in a wide range of biological processes, including proliferation, differentiation and development, as well as in a broad spectrum of diseases. Although the role of lncRNAs in TGF-β-induced epithelial-to-mesenchymal transition (EMT) has been well established, little is known about the role of lncRNAs as immediate-early regulators of EMT. Here lnc-Spry1 is identified as an immediate-early regulator of EMT that is downregulated by TGF-β. It is also found that knockdown of lnc-Spry1 promotes a mesenchymal-like phenotype and results in increased cell migration and invasion. In addition, it is shown that lnc-Spry1 depletion preferentially affects the expression of TGF-β-regulated gene targets. Moreover, lnc-Spry1 associates with U2AF65 splicing factor, suggesting a role in alternative splicing. Depletion of lnc-Spry1 induces, as TGF-β, isoform switching of fibroblast growth factor receptors, resulting in FGF-2-sensitive cells. Taken together, these results show that lnc-Spry1 could act as an early mediator of TGF-βsignaling and reveal different roles for a lncRNA in modulating transcriptional and posttranscriptional gene expression. Cell Death and Differentiation (2017) 24, 785–797; doi:10.1038/cdd.2017.9; published online 10 February 2017 EMT is a basic cellular process in which epithelial cells lose their epithelial characteristics and take on properties of mesenchymal cells. During EMT, epithelial cells lose their cell–cell junctions and the epithelial apical–basal polarity. The actin cytoskeleton is reorganized and cells acquire migratory and invasive properties. 1–3 This process is essential in the generation of tissues and organs during embryogenesis, during wound healing and is associated with pathologies, such as fibrosis and cancer. In carcinomas, cancer cells can undergo EMT to escape the primary tumor, invade surrounding tissues and colonize remote sites via blood or lymphatic routes generating metastases. 4–6 The EMT program can be activated efficiently and rapidly in epithelial cells in response to soluble factors or cytokines, including epidermal growth factor, fibroblast growth factors (FGFs) and transforming growth factors (TGF-βs). TGF-βmay induce EMT through distinct signaling mechanisms causing substantial changes in gene expression. These changes involve three families of transcription factors: the zinc finger Snail, Zeb, and basic helix–loop–helix families. 7 Beyond the well-established transcriptional reprogramming during EMT, posttranscriptional mechanisms, such as regulation by alternative pre-mRNA splicing and regulation by noncoding RNA, have an important role and provide an additional layer of complexity on the gene regulation during EMT. 8–12 The mammalian genome encodes many thousands of lncRNAs, a class of transcripts 4200 nucleotides with limited protein-coding potential. 13 They can act as molecular signals, tethers, decoys, guides or scaffolds at every level of gene regulation in a wide range of biological processes. 14–16 Although transcription factors, microRNAs and signaling pathways that control EMT have been studied intensively, lncRNAs that may regulate this process in an immediate-early phase have not been identified. Several lncRNAs have been reported to be involved in EMT; 17–21 however, these lncRNAs have been identified in cells treated with TGF-βfor days or in cells stably expressing an EMT-inducer transcription factor. Here we profiled immediate-early changes of lncRNAs in NMuMG cells following TGF-βtreatment. This study reports the identification of a lncRNA located adjacent to the mouse Spry1 gene, which was named lnc-Spry1, early downregulated by TGF-β. Decreased lnc-Spry1 levels induce a mesenchymal-like phenotype leading to increased cell migration and invasion. Lnc-Spry1 was found to interact with U2AF65 RNA splicing regulator, suggesting a posttranscriptional regulation of EMT-associated genes. Taken together, our findings indicate that lnc-Spry1 modulates early transcriptional expression and alternative splicing (AS) of genes involved in the EMT process. Results TGF-β-mediated EMT regulates the expression of lncRNAs. Transcription factors regulating EMT are induced early. We hypothesized that lncRNAs with a differential expression during early EMT could represent putative EMT regulators. To identify such lncRNAs, transcriptional profiling was performed in untreated NMuMG mammary epithelial 1 Department of Cell Signaling, Centro Andaluz de Biología Molecular y Medicina Regenerativa (CABIMER-CSIC), Avda Américo Vespucio s/n, Seville 41092, Spain and 2 Department of Genetics, Seville University, Seville 41080, Spain *Corresponding author: JA Pintor-Toro, Department of Cell Signaling Centro Andaluz de BiologÃa Molecular y Medicina Regenerativa (CABIMER-CSIC), Avda Americo Vespucio s/n Isla de la Cartuja, Sevilla 41092, Spain. Tel: +34 954 467 995; Fax: +34 954 46 16 64; E-mail: [email protected] Received 16.6.16; revised 16.12.16; accepted 11.1.17; Edited by X Lu; published online 10.2.2017 Cell Death and Differentiation (2017) 24, 785–797 & 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved 1350-9047/17 www.nature.com/cdd cells, a widely employed model to study TGF-β-induced EMT, and NMuMG cells treated with TGF-βfor 2 h using chips that cover 31 423 annotated lncRNAs and also 25 376 proteincoding transcripts. Fold change values 42 and P-values o0.05 were set as the filtering criteria to analyze differentially expressed RNAs (Supplementary Figure S1). At this time, microarray results indicated that TGF-βtreatment for 2 h led to EMT gene expression signature 22 (Supplementary Lnc-Spry1 mediates EMT C Rodríguez-Mateo et al 786 Cell Death and Differentiation Figure S1e). Six hundred genes were found to be upregulated and 490 downregulated in treated cells compared with untreated cells, including 215 downregulated and 220 upregulated lncRNAs (Supplementary Table S1). Expression of several candidates was confirmed by real-time PCR. Among them, we focused on AK040898 located on mouse chromosome 3, because of its consistent early downregulation (Figure 1a). We termed this candidate lnc-Spry1 based on the name of the neighboring gene. Given the proximity of lnc-Spry1 to the Spry1 gene (Figure 1b), and the fact that both are transcribed in the same orientation, we sought to ensure that the lnc-Spry1 transcript was distinct from that of the Spry1 gene. We determined by 5′and 3′rapid amplification of cDNA ends (RACE) that lnc- Spry1 is a 2851 nucleotides transcript without introns, 472 nucleotides shorter at the 5′end than the ENCODE AK040898 transcript annotation (Figure 1c and Supplementary Figure S1). We next sought to detect trimethylation of histone H3 at lysine 4 (H3K4me3) enrichment at the 5′regulatory region of lnc-Spry1. Chromatin immunoprecipitation assays showed a remarkable H3K4me3-enriched region within 800 bp of the lnc-Spry1 transcription start site, enrichment not detected in TGF-β-treated cells (Figure 1d). These data indicate that Spry1 mRNA and lnc-Spry1 RNA are transcribed from independent promoters. Sequence analysis using ORF Finder (NCBI) predicted open reading frames o70 amino acids without a valid Kozak sequence. In addition, a low coding potential score (−0.924) and a low coding probability (0.032) returned by the Coding Potential Calculator software 23 and Coding-Potential Assessment Tool, 24 respectively, confirmed a negligible protein-coding potential for lnc-Spry1. Sequence homology searches and 5′RACE analysis revealed no clear orthologous lnc-Spry1 sequence or transcripts in syntenic regions of the human genome, suggesting that lnc-Spry1 is a mouse-specific lncRNA. Smad4 is a central mediator of TGF-β intracellular signaling. 25 We investigated the requirement of Smad4 in lnc-Spry1 downregulation by TGF-β. Two different commercial Dicer-substrate short interfering RNAs were used for silencing Smad4 and the knockdown efficiencies were tested by western blotting. Lnc-Spry1 downregulation by TGF-βtreatment was notably affected in Smad4-suppressed cells, indicating TGF-β-Smad signaling dependency of this TGF-βresponse (Figure 1f). Smad-mediated signaling pathway, extracellular signal-regulated kinases (ERKs) and p38 kinases are activated very rapidly in TGF-β-treated NMuMG cells. 26 However, lnc-Spry1 downregulation appeared to uniquely rely on the Smad-mediated pathway, as preincubation of NMuMG cells with the MEK inhibitor U0126 and the p38 inhibitor SB203580 had no effect on TGF-β-downregulated lnc-Spry1 (Figures 1e and g). Lnc-Spry1 downregulation detected within 15 min of TGF-βtreatment (Figure 4b) suggested an active role in the start of EMT program. To determine whether lnc-Spry1 was a direct target of TGF-β- Smad pathway, we analyzed the levels of lnc-Spry1 in cells treated with TGF-βin the presence or absence of cycloheximide, an inhibitor of protein synthesis. Cycloheximide treatment did not inhibit downregulation of lnc-Spry1, whereas upregulation of SIP1, an indirect target control of the TGF-β- Smad pathway, 27 was blocked (Figure 1h). Together, these findings strongly suggest that lnc-Spry1 is downregulated by TGF-β-Smad signaling and it is a direct target of this pathway. Lnc-Spry1 downregulation promotes a mesenchymal-like phenotype. To investigate whether the reduced expression of lnc-Spry1 could promote the process of EMT, we generated short hairpin RNAs (shRNA)-mediated knockdown stable clones using three lnc-Spry1-targeting shRNA expression constructs. Expression of lnc-Spry1 was reduced from 60 to 70% using these different shRNAs (Figure 2a). A consistent epithelial-to-mesenchymal-like phenotype was observed for all three shRNAs used (Figure 2b), indicating that this change was specific and was not likely due to off-targeting. Furthermore, immunofluorescence analysis revealed a decrease in the epithelial marker E-cadherin expression and the remaining protein was dislocated from the membrane to the cytoplasm. In the absence of TGF-β, phalloidin-labeled F-actin was predominantly organized in thin cortical bundles tightly associated with cell–cell adhesions. 28 In contrast, after 48 h of treatment with TGF-β, F-actin was assembled into thick parallel bundles, or actin stress fibers, traversing the ventral cell surface. We also observed a notable remodelling of the cytoskeleton from cortical actin to stress fibers in lnc-Spry1-depleted cells Figure 1 Lnc-Spry1 expression in TGF-β-treated NMuMG cells. (a) Expression of lnc-Spry1 measured by quantitative reverse transcriptase-PCR (RT-qPCR) at representative time points shows decreased levels in TGF-β-treated cells. Experiments were performed in triplicate, and error bars represent S.D. **Po0.01 by two-tailed Student’st-test. (b) Schematic illustration of the mouse lnc-Spry1 gene locus chromosomal location. Arrowheads indicate the transcription orientation. (c) Top, a representative image of PCR products from the 5′-RACE and 3′-RACE procedure. Bottom, sequencing of PCR products revealed the boundary between the adaptor primers and lnc-Spry1 sequences. (d) Chromatin immunoprecipitation analysis were performed to study H3K4me3 enrichment at the lnc-Spry1 5′regulatory region of NMuMG control cells and NMuMG stimulated with TGF-βfor 2 h. Transcription star site (TSS) and the enrichment regions analyzed are indicated in the figure. Enrichment values are represented as the percentage of recovered input. Control cell data are the mean of at least n=9 qPCR reactions from three independent experiments. Background enrichment with normal rabbit IgG was o0.05% of the total input chromatin. TGF-β-treated cell data are the mean of n=6 qPCR reactions from two independent experiments. Background enrichment with normal rabbit IgG was o0.015% of the total input chromatin. (e) Effect of MAP kinase inhibitor, U0126, on lnc-Spry1 downregulation. NMuMG cells were pretreated with 10 μM inhibitor for 1 h prior to addition of TGF-βfor 2 h where indicated. Lnc-Spry1 levels were measured by RT-qPCR (top); p-ERK1/2 were assessed by western blotting and total ERK1/2 were used as protein loading control (bottom). (f) NMuMG cells were transfected with DsiRNAs against Smad4 (DsiSmad4) or scramble DsiRNA (SRC) and treated with TGF-βfor 2 h. Levels of lnc-Spry1 were measured by RT-qPCR analysis (top). Smad4 was assessed by western blotting (bottom); β-actin was used as protein loading control. (g) Effect of p38 kinase inhibitor, SB203580, on lnc-Spry1 downregulation. Cellswere pretreated with 10 μM inhibitor for 1 h prior to addition of TGF-βfor 2 h where indicated. Levels of lnc-Spry1 were measured by RT-qPCR analysis (top). Activation of p38 kinase was detected via phosphorylated p85 ribosomal kinase (p85). A representative immunoblot analysis for p85 is shown (bottom); β-actin was used as a protein loading control. (h) Requirement of de novo protein synthesis for lnc-Spry1 downregulation by TGF-β. NMuMG cells pretreated with 5 μM of cycloheximide (CHX) for 1 h were stimulated with TGF-βfor 2 h (top) or 24 h (bottom). Expression levels of lnc-Spry1 (top) and SIP1 (bottom) were measured by RT-qPCR analysis Lnc-Spry1 mediates EMT C Rodríguez-Mateo et al 787 Cell Death and Differentiation (Figure 2c). Immunoblotting analysis revealed a remarkable increase in the expression of mesenchymal markers, such as vimentin and Snail1, and a decrease in the expression of the epithelial marker E-cadherin in lnc-Spry1-depleted clones compared with control clones (Figure 2d). Also RT-qPCR analysis showed that mesenchymal marker SIP1 was slightly increased in lnc-Spry1-depleted clones (Figure 2e). As lnc-Spry downregulation mimics TGF-βsignaling in terms of EMT induction, we wondered whether lnc-Spry1 downregulation induced TGF-βexpression and this could cause an autocrine TGF-βsignaling. Quantification of TGF-βmRNA levels in lnc-Spry1-depleted shRNA1 and shRNA2 clonal cell lines showed that these clones contained TGF-βmRNA levels 1.4 and 2.2 times higher, respectively, than control cells. To test a possible role of autocrine TGF-βsignaling in lnc-Spry1- depleted clones, we treated them with TGF-βtype I receptor Lnc-Spry1 mediates EMT C Rodríguez-Mateo et al 788 Cell Death and Differentiation (ALK5) kinase inhibitor SB431542. 29 Clones were minimally affected by treatment with the inhibitor and showed mesenchymal-like phenotype (Supplementary Figure S2a), and EMT-related molecules analyzed were not significantly affected (Supplementary Figures S2b–e). These results indicate that downregulation of lnc-Spry1 promotes per se a mesenchymal-like phenotype. As cell migration and invasion are altered in EMT, we assessed the migratory capacity and invasiveness of lnc-Spry1-silenced cells. Wound-healing assays revealed that shlnc-Spry1 NMuMG cells migrated much faster than control cells (Figures 3a–cand Supplementary Movie S1). Furthermore, the effect of lnc- Spry1 downregulation on cell invasion was determined in invasion assays in matrigel-coated transwell chambers. Lnc- Spry1 silencing resulted in a highly significant increase in the invasive potential of NMuMG cells (Figures 3d and e). However, monitoring of NMuMG real-time cell proliferation showed no differences between control and lnc-Spry1- depleted cells (Figure 3f). These results demonstrate a role of lnc-Spry1 in remodeling the cell cytoskeleton and the effect of its depletion on cell migration and invasion. Lnc-Spry1 depletion alters gene expression. LncRNAs have been shown to regulate the expression of neighboring genes in cis or in trans by diverse mechanisms. 30 The Spry1 gene is located 480 bp upstream from lnc-Spry1 and, therefore, could be a candidate to be regulated in cis.In fact, lnc-Spry1-depleted clones markedly reduced the Spry1 mRNA levels (Figure 4a). This effect was specific for Spry1 gene, as changes were not detected in other protein-coding genes surrounding lnc-Spry1 (Supplementary Figure S3). A detailed time course of lnc-Spry1 and Spry1 mRNA expression after TGF-βtreatment for short periods of time showed a strong correlation between downregulation of both genes (Figure 4b). Similar to the effect of cycloheximide on lnc-Spry downregulation in TGF-β-treated cells, downregulation of Spry1 mRNA levels was not affected by cycloheximide, indicating that Spry1 is a direct target (Figure 4c). Moreover, forced expression of lnc-Spry1 in lnc-Spry1-depleted cells did not rescue the expression of Spry1 gene (Figures 4d and e), as would be expected from a lncRNA whose primary function was exerted locally. 31 Also, overexpression of Spry1 did not lead to any noticeable morphological change in lnc-Spry1- depleted cells (Supplementary Figure S4). To determine whether the expression of lnc-Spry1 and Spry1 genes were linked, we examined their relative expression across the same panel of different mouse tissues (Figure 4f). Spry1 mRNA was found to be much more highly expressed than lnc-Spry1 in all the tissues analyzed. Both RNAs were more abundant in the heart, lung and muscle. Despite the variation in the expression levels of both transcripts and the different expression in the tissues analyzed, the expression profiles of Spry1 and lnc-Spry1 genes were closely correlated (Pearson correlation coefficient R=0.98). Altogether, we might suggest that lnc-Spry1 could act locally to regulate the Spry1 expression and that they both could regulate similar biological processes. Overexpression of lnc-Spry1 did not seem to perturb normal EMT by TGF-β. Immnunofluorescence analysis of TGF-β-treated control cells and TGF-β- treated lnc-Spry1-overexpressing cells did not reveal apparent morphological changes. Also, immunoblot analysis of E-cadherin, vimentin, Snail1 and Spry1 proteins did not show any significant change in the expression levels of these proteins (Supplementary Figure S5). This result would be consistent with the fact that EMT relies on a complex signaling network that involves multiple transcription factors and signaling proteins, in some instances, with redundant functions. As Snail1 transcription factor is a major regulator of EMT and it is a primary responder to TGF-βtreatment, 32,33 we tested whether ectopic expression of Snail1 affected the levels of lnc-Spry1 and Spry1 mRNA. Expression of both genes was unaffected by overexpression of Snail1, indicating that Snail1 is not required for lnc-Spry1 inhibition upon TGF-β treatment (Supplementary Figure S6). We next reasoned that analysis of global gene expression pattern in lnc-Spry1- depleted cells could reveal further clues toward elucidating its function. Changes in RNA levels were analyzed by DNA microarray analysis both in stable lnc-Spry1-depleted cells and control cells. We found 181 misregulated genes in the lnc-Spry1-depleted cells (lineal change 42-fold and P-valueo0.05) among which 110 were downregulated and 71 were upregulated (Figure 4g and Supplementary Table S2). RT-qPCR analysis confirmed microarray results for 10 of the 12 selected genes (83%), including downregulation of Id2, Sim2, Atoh8, Cxcl5, Ccl20 and upregulation of Sox9 (Supplementary Figure S7). Gene ontology analysis of misregulated genes in lnc-Spry1-depleted cells showed a notable enrichment in those genes involved in chemokine and cytokine activity (Ccl2, Ccl20, Ccl7, Cxcl1, Cxcl16, Bmp2, Grem1) (Figure 4h). Strikingly, 30 and 28% of downregulated and upregulated genes, respectively, were also downregulated and upregulated in TGF-βtreated Figure 2 Downregulation of lnc-Spry1 promotes a mesenchymal-like phenotype. (a) Lnc-Spry1-downregulated cell lines were generated by transduction of shRNA expression lentivirus against either lnc-Spry1 or a non-targeting control. Lnc-Spry1 RNA levelswere determined by quantitative reverse transcriptase-PCR (RT-qPCR) in NMuMG cells treated with TGF-βfor 48 h, in a clonal cell line transduced by empty lentivirus (pLKO), three clonal cell lines expressing reduced levels of lnc-Spry1 and a control cell line. Experiments were performed in triplicate, and error bars represent S.D. within one representative experiment. Results were analyzed by one-way analysis of variance, followed by Bonferroni posttest. ***Po0.001, NS, not significant. (b) Downregulationof lnc-Spry1 induces morphological changes in NMuMGcells.Phase-contrast imagesillustrate that cells depleted of lnc-Spry1 undergo EMT-like changes in morphology, including a reduction in cell–cell contacts and cobblestone appearance. (c) Immunofluorescence microscopic analysis of changes in location and the expression of marker proteins. Cells were stained with antibodies against the epithelial marker E-cadherin and with tetramethylrhodamine isothiocyanate (TRITC)-conjugated phalloidin and 4,6-diamidino-2-phenylindole to visualize the actin cytoskeleton and nuclei, respectively. Scale bar, 25 μm. (d) Immunoblotting of Snail1, E-cadherin and vimentin in NMuMG cells, NMuMG treated with TGF-βfor 48 h and lnc-Spry1-downregulated NMuMG clones generated by transduction of shRNAs expression lentivirus against lnc-Spry1. β-Actin was used as a loading control. (e) Expression levels of SIP1 mRNA were detrmined by RT-qPCR in NMuMG control cells, treated with TGF-βfor 48 h and three clonal cell lines expressing reduced levels of lnc-Spry1. Error bars represent S.D. ***Po0.001, **Po0.01, *Po0.05 by two-tailed Student’st-test Lnc-Spry1 mediates EMT C Rodríguez-Mateo et al 789 Cell Death and Differentiation cells (Figure 4i and Supplementary Table S3) (P-value =4.7 ×10 −48 and P-value =6.9 × 10 −20 , hypergeometric distribution), strongly suggesting that lnc-Spry1 could participate directly and indirectly in TGF-β-dependent transcriptional response. Lnc-Spry1 interacts with the U2AF65 spliceosome component. To investigate the mechanism by which lnc- Spry1 mediates gene regulation, we searched for proteins that were associated with lnc-Spry1 using in vitro biotin-RNA pull-down. In vitro-transcribed biotinylated lnc-Spry1 and a non-relevant RNA as a negative control were incubated with total cellular extracts and coprecipitated proteins were isolated using streptavidin-coated beads. RNA-associated proteins were resolved on a SDS-PAGE gel, and specific bands to the lnc-Spy1 probe were visualized using SYPRO Ruby staining and cut out for analysis by mass spectrometry (Figure 5a), identifying U2AF65 as a protein specifically associated with lnc-Spry1 (Supplementary Figure S8). These interactions were confirmed by western blotting analysis Figure 3 Lnc-Spry1-depleted cells acquire migratory and invasive properties. (a) The migratory capacity of lnc-Spry1-downregulated NMuMG cells were tested in woundhealing assays. Cells were imaged at 20 min intervals for 23 h. The frames of the movie at 9, 12 and 23 h are shown. (b) The frames at 9 and 12 h are used to estimate the percentage of surface covered by the cells. Values represent the average (%) of wound closure±S.E. ***Po0.001 by two-tailed Student’st-test. (c) Wound density quantitation of the time-lapse video frames from control and two lnc-Spry1-depleted clonal cell lines. (d) Representative images of invasion in a clonal cell line expressing reduced levels of lnc- Spry1 and in a control cell line analyzed in a transwell invasion assay. After 24 h, cells were fixed and stained with crystal violet. (e) Histograms represent the mean±S.D. of the number of cells per field counted in three different inserts. **Po0.01 by two-tailed Student’st-test. (f) Real-time monitoring of in vitro proliferation of two clonal cell lines expressing reduced levels of lnc-Spry1 and a control cell line using the eCELLigence system. The system measures electrical impedance across interdigitated microelectrodes integrated on the bottom of tissue culture plates Lnc-Spry1 mediates EMT C Rodríguez-Mateo et al 790 Cell Death and Differentiation (Figure 5b) and RNA immunoprecipitation (RIP). Enrichment of lnc-Spry1 RNA was observed using specific U2AF65 antibody compared with non-specific IgG antibody (Figure 5c). Additionally, deletion-mapping experiments combined with western blotting analysis showed that the different truncated Figure 4 Gene expression is affected in lnc-Spry1-depleted cells. (a) Relative Spry1 mRNA levels determined by quantitative reverse transcriptase-PCR (RT-qPCR) in control cells, clonal cell line transduced by empty lentivirus (pLKO), cells treated with TGF-βfor 48 h and three independent shRNA-mediated knockdown of lnc-Spry1 stable clones. Results were analyzed by one-way analysis of variance, followed by Bonferroni posttest. ***Po0.001. (b) Relative lnc-Spry1 and Spry1 mRNA levels determined by RT-qPCR in NMuMG control cells treated with TGF-βfor 15, 30, 60 and 120 min. Error bars represent S.D. (c) Expression levels of Spry1 mRNA in NMMG cells pretreated with 5 μM cycloheximide (CHX) for 1 h and stimulated with TGF-βfor 2 h. (d) Real-time PCR analysis of lnc-Spry1 levels in cells infected at MOI 2 with pHRSIN-DUAL lentivirus control (pSD) or lentivirus carrying full-length lnc-Spry1 cDNA. Error bars represent S.D. within one representative experiment. (e) Relative levels of Spry1 mRNA in cells overexpressing lnc-Spry1. (f) Relative lnc-Spry1 and Spry1 mRNA levels determined by RT-qPCR in a broad range of adult mouse tissues. Results are presented relative to the average value of HPRT reference gene and a fixed reference sample (bone marrow). (g) Heat map showing statistically significant changes of 181 genes in lnc-Spry1-depleted cells. Color intensities are shown on a log2 scale. (h) Significant Gene Ontology annotation enrichments of lnc-Spry1-regulated genes (5% FDR). (i) Venn diagram showing the intersection of genes affected in TGF-β-treated or lnc-Spry1-depleted cells. All misregulated genes in the upper panel. Lower left and right panels: upregulated and downregulated genes, respectively. Probability of overlapping based on hypergeometric distribution is provided Lnc-Spry1 mediates EMT C Rodríguez-Mateo et al 791 Cell Death and Differentiation versions of lnc-Spry1 associated with U2AF65. Lnc-Spry1 sequence contained several disseminated U2AF65-binding sites (Figure 5d) and association between U2AF65 and lnc- Spry1 decreased drastically after mutagenesis of some of these binding sites (Figure 5e). Next, lnc-Spry1 expression was measured in nuclear and cytosolic fractions from NMuMG cells by RT-qPCR. The differential enrichments of HPRT, Xist and U3 RNAs were used as fractionation indicators. A considerable increase in lnc-Spry1 level was observed in the nucleus of NMuMG cells (Figure 5f), indicating that lnc-Spry1 may have a major regulatory function. Lnc-Spry1 modulates EMT-associated AS. A strong link between EMT and AS has been demonstrated in the case of FGF receptors (FGFRs). A functional FGFR consists of three extracellular immunoglobulin domains (Ig-I, Ig-II and IgIII), a single-transmembrane domain, and a cytoplasmic tyrosine kinase domain. AS in the second half of the third extracellular immunoglobulin-like domain is encoded by one of the two mutually exclusive alternative exons: IIIb or IIIc. The expression of either isoform is highly cell-type specific, with FGFR2IIIb expression predominantly expressed in epithelial cells and FGFR1IIIc and FGFR2IIIc mainly expressed in mesenchymal cells. Given the interaction between U2AF65 and lnc-Spry1 and the documented role of U2AF65 in splicing regulation, we hypothesized that lnc-Spry1 could have a trans-mediated effect in AS by leading or fitting the pool of U2AF65. Analysis by RT-qPCR of FGFR1 and FGFR2 mRNA levels showed that FGFR1 mRNA levels increased 60-fold in Figure 5 Lnc-Spry1 interactswith U2AF65 splicing factor. (a) A representative Sypro ruby-stained gel picture showing the proteins bound to lnc-Spry1 (left lane) or unrelated RNA (right lane). The highlighted region was analyzed by mass spectrometry. U2AF65 was identified as lnc-Spry1-binding partner. (b) Western blotting analysis of the specific interaction between U2AF65 and lnc-Spry1. Protein loading normalization was performed by measuring total protein directly on the membrane using the criterion stain-free gel imaging system. (c) Physical association of lnc-Spry1 and U2AF65 determined by RNA immunoprecipitation (RIP). (Bottom) Representative western blotting of U2AF in the corresponding samples. Results are the median of three technical replicates; bars represent S.D. (d) Deletion analysis of lnc-Spry1 transcript. Motifs for U2AF65 binding in lnc- Spry1 are shown. Associated U2AF65 with different RNA fragments was detected by western blotting. (e) Associated U2AF65 with the 1580 nt lnc-Spry1 fragment after mutagenesis of U2AF65-binding sites. (f) Lnc-Spry1 subcellular localization. Expression of lnc-Spry1, U3, Xist and Hprt in nuclear and cytoplasmic fractions of NMuMG cells. Error bars represent S.D. Lnc-Spry1 mediates EMT C Rodríguez-Mateo et al 792 Cell Death and Differentiation cells stimulated with TGF-βfor 48 h and from 11- to 20-fold in lnc-Spry1-depleted cells, whereas FGFR2 mRNA levels decreased in both conditions (Figures 6a and b). Using isoform-specific primers, we determined that the FGFR1 isoform upregulated by TGF-βor by lnc-Spry1 depletion was the mesenchymal isoform FGFR1IIIc (Figure 6c), whereas Figure 6 Isoform switching of FGFRs in TGF-β-treated cells and lnc-Spry1-depleted cells. (aand b) Expression levels of FGFR1 and FGFR2 in NMuMG control cells, treated with TGF-βfor 48 h, a clonal cell line transduced by empty lentivirus (pLKO) and three clonal cell lines expressing reduced levels of lnc-Spry1. Error bars represent S.D. ***Po0.001 by two-tailed Student’st-test. (c–e) Expression of the alternatively spliced forms of FGFR1 and FGFR2 in NMuMG control cells, treated with TGF-βfor 48 h, in a clonal cell line transduced by empty lentivirus (pLKO) and in three clonal cell lines expressing reduced levels of lnc-Spry1. Error bars represent S.D. ***Po0.001 by two-tailed Student’st-test. (fand g) Representative western blotting results for ERK1/2 and p-ERK1/2 protein expression from NMuMG control cells, a clonal cell line transduced by empty lentivirus (pLKO) and a clonal cell line expressing reduced levels of lnc-Spry1 (Clone shRNA1). NMuMG cells were preincubated with (f) or without (g) TGF-βfor 2 days and then stimulated with 30 ng/ml FGF-2 for 15 min. A clonal cell line expressing reduced levels of Spry1 mRNA (Clone shSpry1) is included in panel (g). (h) Immunofluorescence analysis of actin cytoskeleton reorganization in NMuMG control and lnc-Spry1-depleted cells cultured in the presence of 30 ng/ml FGF-2 for 48 h. Scale bar, 25 μm Lnc-Spry1 mediates EMT C Rodríguez-Mateo et al 793 Cell Death and Differentiation