• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of embojLink to Publisher's site
EMBO J. Oct 6, 2010; 29(19): 3286–3300.
Published online Aug 13, 2010. doi:  10.1038/emboj.2010.195
PMCID: PMC2957203

An ESRP-regulated splicing programme is abrogated during the epithelial–mesenchymal transition

Abstract

Alternative splicing achieves coordinated changes in post-transcriptional gene expression programmes through the activities of diverse RNA-binding proteins. Epithelial splicing regulatory proteins 1 and 2 (ESRP1 and ESRP2) are cell-type-specific regulators of transcripts that switch splicing during the epithelial–mesenchymal transition (EMT). To define a comprehensive programme of alternative splicing that is regulated during the EMT, we identified an extensive ESRP-regulated splicing network of hundreds of alternative splicing events within numerous genes with functions in cell–cell adhesion, polarity, and migration. Loss of this global ESRP-regulated epithelial splicing programme induces the phenotypic changes in cell morphology that are observed during the EMT. Components of this splicing signature provide novel molecular markers that can be used to characterize the EMT. Bioinformatics and experimental approaches revealed a high-affinity ESRP-binding motif and a predictive RNA map that governs their activity. This work establishes the ESRPs as coordinators of a complex alternative splicing network that adds an important post-transcriptional layer to the changes in gene expression that underlie epithelial–mesenchymal transitions during development and disease.

Keywords: alternative splicing, epithelial–mesenchymal transition, epithelial splicing regulatory proteins

Introduction

Alternative splicing vastly expands transcriptomic diversity as evidenced by recent studies demonstrating that nearly all multi-exon human genes undergo alternative splicing (Pan et al, 2008; Wang et al, 2008). Regulated alternative splicing is achieved through the interplay of multiple RNA-binding proteins that associate with auxiliary cis-elements within pre-mRNA transcripts that can positively or negatively influence splicing of regulated exons and are designated as intronic or exonic splicing enhancers and silencers. The combinatorial assembly of RNA-binding proteins with these cis-elements determines whether an exon is included or skipped in the mature transcript. Combinatorial control of splicing involves contributions by the generally ubiquitous SR and hnRNP families of proteins in cooperation with tissue-specific regulators of splicing, such as Nova, nPTB, Fox, nSR100, MBNL, and CELF proteins (Chen and Manley, 2009; Hartmann and Valcarcel, 2009; Nilsen and Graveley, 2010). With the advent of novel cross-linking and high-throughput sequencing approaches, binding sites for several splicing regulators have been determined on a genome-wide scale and demonstrated position-dependent effects of binding relative to regulated exons that determine the splicing outcome (Licatalosi et al, 2008; Xue et al, 2009; Yeo et al, 2009). These RNA maps are thus predictive of genome-wide effects of a given splicing factor and the combined contributions of numerous regulators are likely to define a splicing code that governs diverse splicing outcomes in different cellular milieus (Barash et al, 2010).

The well-characterized neuron-specific Nova proteins and the neural/muscle-enriched Fox proteins co-regulate numerous splicing events as part of functionally coherent splicing regulatory networks (SRNs) (Licatalosi et al, 2008; Zhang et al, 2008; Venables et al, 2009). Just as sets of genes that are co-regulated at the transcriptional level have been shown to function in common processes, complexes, and pathways, increasing evidence suggests that genes co-regulated post-transcriptionally as part of SRNs also have functionally related properties. However, genes that are co-regulated transcriptionally in different cells and tissues do not overlap significantly with genes that are regulated by alternative splicing in the same cells and tissues (Pan et al, 2004). These observations indicate that identification of genes co-regulated within SRNs will uncover sets of functionally related gene products in addition to those uncovered by analysis of total mRNA levels alone.

The epithelial-to-mesenchymal transition (EMT) is a critical process during the development and differentiation of multiple organs and is implicated in tissue fibrosis and tumour metastasis (Thiery et al, 2009). The EMT is a complex process involving reorganization of the actin cytoskeleton, loss of cell–cell adhesion, and the acquisition of increased cell motility. The changes in the transcriptional profile of epithelial cells as they undergo an EMT have been well studied, including the downregulation of E-cadherin and upregulation of vimentin and N-cadherin (LaGamba et al, 2005). However, the extent to which coordinated changes in splicing might contribute an additional layer of gene regulation that impacts the EMT at the post-transcriptional level has not been systematically investigated. We recently identified epithelial splicing regulatory proteins 1 and 2 (ESRP1 and ESRP2) as epithelial cell-type-specific regulators of FGFR2 splicing as well as several other transcripts with epithelial-specific splice variants (Warzecha et al, 2009a, 2009b). We determined that the ESRPs are a component of an epithelial gene signature and demonstrated a downregulation of the ESRPs in cells that undergo an EMT, which corresponded with switches in splicing of several ESRP-regulated exons. We hypothesized that a global epithelial SRN regulated by the ESRPs constitutes an essential post-transcriptional gene expression programme with broad developmental and patho-physiological functions. We further predicted that the different protein isoforms encoded by these co-regulated transcript variants would have distinct functions in epithelial and mesenchymal cells that contribute to the unique features and biological functions of each cell type.

In this study, we used a next generation splicing sensitive microarray platform to characterize an ESRP-regulated SRN and identified hundreds of novel ESRP-regulated splicing events. Computational and experimental approaches defined an ESRP-binding motif that defined position-dependent functions on exon splicing, thereby deriving an RNA map for ESRP-mediated splicing regulation. Importantly, we demonstrated that a global reversion of splicing from the epithelial-to-mesenchymal patterns through sustained knockdown of ESRP1 and ESRP2 in mammary epithelial cells induced cellular changes indicative of an EMT. These results establish that the ESRPs orchestrate an epithelial cell-type-specific SRN that regulates mRNAs encoding proteins with isoform-specific functions in epithelial versus mesenchymal cells. This study also highlights the importance of coordinated changes in alternative splicing, in addition to changes in transcription, to promote dynamic changes in proteomic complexity during the EMT.

Results

Identification of an extensive ESRP-regulated epithelial SRN using splicing sensitive exon junction microarrays

Our previous observations demonstrated several splicing switches in ESRP-regulated gene transcripts during the EMT, suggesting that these splicing factors direct a large-scale splicing network of genes with functions in this process. Although no systematic analysis of global changes in alternative splicing during the EMT has previously been performed, we hypothesized that numerous ESRP-regulated transcripts undergo splicing switches during the EMT. To comprehensively determine a global ESRP-regulated alternative splicing network, we used affymetrix human exon junction arrays (HJAY) to profile splicing changes in response to ectopic expression of ESRP1 in MDA-MB-231 mesenchymal cells and knockdown of ESRP1 and ESRP2 in PNT2 epithelial cells (Figure 1A–C). To identify ESRP-regulated changes in splicing, we used the microarray analysis of differential splicing plus (MADS+) computational pipeline (Shen et al, 2010). The HJAY arrays consist of eight tiled probes per probeset that target 315 137 exons and 260 488 exon–exon junctions, covering most human alternative splicing events supported by mRNA and/or EST evidence. To detect splicing changes between samples, we selected events that displayed reciprocal changes in the probeset targeting the exon skipping events and at least one of three exon inclusion probesets. This approach predicted 417 and 540 changes in cassette exon splicing upon ectopic expression of ESRP1 and combined ESRP knockdown, respectively (Figure 1D; Supplementary Table S1). We noted numerous gene transcripts that contained more than one predicted ESRP-regulated exon and examples in which the ESRPs were predicted to promote either exon inclusion or exon skipping. There was a significant positive correlation of the two data sets consisting of 104 cassette exons showing opposite changes in splicing (P-value<4.0e−23) (Figure 1D; Supplementary Table S1). A related analysis detected reciprocal changes in probesets corresponding to alternative 3′ and 5′ splice sites as well as mutually exclusive exons (Supplementary Table S1).

Figure 1
An exon junction microarray detects hundreds of novel ESRP-regulated alternative splicing events. (A) Outline of the experimental systems and array platform used to identify ESRP-regulated exons. (B) Western blot using a monoclonal antibody that recognizes ...

The ESRPs are phylogenetically conserved splicing factors and we therefore expected that many of the regulated exons and genes would be highly conserved between humans and mice, as well as among broader metazoans. Consistent with this prediction, we found a significant number of alternative conserved exons (ACEs) among the HJAY-detected cassette exons as predicted by the computational ACEScan tool (Yeo et al, 2005). These ACEScan exons are highly enriched for biologically relevant alternative splicing events that are conserved between humans and mice. Among the 828 non-redundant exons from both ESRP-regulated cassette exon data sets, 200 (24.2%) were ACEScan exons, whereas only 9.0% of the background set of cassette exons on the array that passed an expression filter in either experiment were ACEScan exons (P-value<2.2e−16) (Supplementary Table S2). ESRP-regulated exons were also enriched in a data set of exons with strong evidence of more extensive phylogenetic conservation of regulated splicing. Analysis of RNA selection pressure (RSPR) across 28 vertebrate genomes recently identified 345 human exons with strong evidence of conserved alternative splicing (Lu et al, 2009). Of the predicted ESRP-target exons, 49 (5.9%) of these were among the 345 highly conserved alternative exons with high RSPR compared with 220 (1.9%) of the background set (P-value<5.1e−11) (Supplementary Table S2). Finally, ESRP-regulated exons from both sets were more likely to be a multiple of three in nucleotide (nt) length thus maintaining the reading frame, further evidence of enrichment for functionally relevant splicing events (PNT2: 58 versus 47% in HJAY background; MDA-MB-231: 59 versus 46% in background) (Resch et al, 2004).

Validation of ESRP-regulated exons reveals numerous large switches in exon splicing

To independently validate a large set of splicing changes predicted by the HJAY analysis, we predominantly used high-throughput capillary reverse transcription PCR (HT-RT–PCR) to validate nearly all simple cassette exons with constitutive flanking exons identified by the HJAYarrays (Venables et al, 2009). In the ESRP1 ectopic expression system, 158 of the 173 exons with successful HT-RT–PCR reactions demonstrated the HJAY predicted changes in exon inclusion of which 134 showed a higher confidence change of >5% in percent exon inclusion (Supplementary Table S3). In PNT2 knockdowns, 115 out of 183 successful HT-RT–PCR reactions yielded a lower overall validation rate, of which 71 exons had a >5% change in exon inclusion (Supplementary Table S4). We also used the HT-RT–PCR format to test 20 alternative 3′ or 5′ splice sites that were identified in the knockdown experiments and 11 validated the predicted change with a high confidence of >5% change between competing 5′ or 3′ splice sites (Supplementary Table S4). We also validated an additional 13 HJAY predicted splicing changes using traditional RT–PCR, including cases that involved predicted mutually exclusive exons (Supplementary Figure S1). In total, we validated 210 predicted changes in splicing with a >5% change in splice isoforms.

Exons showing the overall largest validated changes in exon inclusion identified in each experimental approach are shown in Table I. Among the validated events, were 22 ‘switch-like' changes in exon splicing in which knockdown or ectopic expression induced a switch from predominant exon inclusion (>66%) to mostly exon skipping (<34% inclusion), or vice versa as defined in Xing and Lee (2005). Also noteworthy were 19 examples in which one splice variant was predominant at baseline in each cell type, but where knockdown or ectopic ESRP expression induced expression of a second splice variant. Even though such changes were less complete, they nonetheless predicted expression of numerous epithelial- and mesenchymal-specific variants (e.g. ARFGAP2, ARHGEF11). RALGPS2 and ITGA6 demonstrated large changes in ESRP-mediated exon inclusion in both experimental systems (Figure 2A and B). In MAGI1, ESRP expression promoted exon skipping in both the ESRP overexpression and knockdown analysis (Figure 2C). In addition, we noted that two different exons in MAP3K7 (also known as TAK1) transcripts were either enhanced or silenced by the ESRPs (Figure 2D). Of note, regulated exons in ITGA6 and MAP3K7 are alternative penultimate exons that generate different C-terminal domains in the resulting protein isoforms. Similar to the ITGA6 exon, we identified splicing switches in a number of other penultimate exons containing stop codons (e.g. in MPRIP, CD46, STX2, and STX3 transcripts). As most of these stop codons are <50–55 nt upstream of the terminal intron, these transcripts are predicted to elude RNA degradation through the nonsense-mediated decay pathway and this type of exon switch provides an effective means of changing the C-terminal domain. Many of these splicing switches occur in transmembrane proteins, indicating that this might be a common mechanism of altering the intracellular signalling pathways of integral membrane receptors and matrix binding proteins. To further verify that changes in splicing were the result of ESRP knockdown, we also validated a subset of splicing switches using a second combination of siRNAs directed against different sequences in ESRP1 and ESRP2. We also performed a ‘rescue' using the RNAi resistant mouse Esrp1 cDNA. This analysis confirmed that these splicing switches are specifically due to ESRP1 and ESRP2 knockdown and not because of off-target effects (Figure 2E and F).

Figure 2
Robust examples of validated ESRP-induced enhancement and silencing of microarray predicted target exons. (A) Validation of an ESRP-enhanced penultimate exon in ITGA6. The UCSC browser view of the transcripts that contain or skip the exon is shown at ...
Table 1
Exons with the largest validated positive and negative change in splicing upon ESRP knockdown or ectopic expression

Identification of an ESRP-binding motif within ESRP-regulated exons and intronic flanking sequences

We previously demonstrated that ESRP1 binds the FGFR2 auxiliary cis-element ISE/ISS-3, located in the intron between mutually exclusive exons IIIb and IIIc, which implied that the ESRPs promoted splicing of the upstream exon IIIb while silencing the downstream exon IIIc (Warzecha et al, 2009a). To define a consensus ESRP-binding motif, we searched for conserved hexamers in the introns flanking validated ESRP-regulated cassette exons as well as in the exons themselves. The top five motifs in each position relative to enhanced or silenced exons are presented in Figure 3A (complete motif analysis is shown in Supplementary Table S5). We noted a remarkably enriched motif downstream of ESRP-enhanced exons consisting of hexamers that contained repeats of UGG or GGU; similar to the FGFR2-binding site and consistent with a model in which ESRP-binding downstream of an exon promotes its inclusion. Similar motifs were enriched within ESRP-silenced exons. As UGG was a core of the most enriched motifs and to account for some degeneracy in the binding site, we mapped the 5-mer motifs UGGUG, GGUGG, and GUGGU across the set of validated ESRP-regulated exons within the windows used to derive the motifs (Supplementary Figure S2A and B). We noted that clusters of these sequences were enriched downstream of enhanced exons. In contrast, we observed enrichment of these motifs within ESRP-silenced exons, whereas these motifs were largely absent in ESRP-enhanced exons. We also noted that these 5-mers were present in greater numbers upstream of silenced exon than the enhanced exons. Therefore, the trend of these data was consistent with a ‘map' in which ESRP-binding downstream of a regulated exon promotes exon inclusion and binding within the exon, and possibly the upstream intron, promotes exon skipping. We also note that although there are a number of ESRP-enhanced exons in which the motif was not identified within the 250-nt window used here, we often noted UGG-rich motifs in conserved sequence tracks located further downstream of these exons (e.g. ENAH). To determine whether the UGG-rich motifs are direct binding targets for the ESRPs, we performed RNA electrophoretic mobility shift assay (EMSAs) using recombinant ESRP1 protein. We tested whether the protein bound to RNA sequences containing repeats of UGG compared with UCG repeats. These experiments validated sequence-specific binding to the UGG repeat motifs, but not to those in which the motif is disrupted (Figure 3B). EMSA titration analysis to determine binding affinities revealed a dissociation constant (Kd) of 31.7 nM, indicating that this motif represents a bona fide high-affinity target for the ESRPs (Figure 3C).

Figure 3
UGG motifs define a high-affinity ESRP-binding site with position-dependent functions. (A) Motif enrichment analysis in the exon and flanking 250 nt of intronic sequence, excluding the splice sites. The top five most enriched motifs ranked by Fisher ...

We noted that the Fox-binding site UGCAUG was one of the top conserved hexamer motifs downstream of both ESRP-enhanced and ESRP-silenced exons. As the presence of this motif downstream of an exon predicts Fox-mediated enhanced exon inclusion, this observation suggests that the Fox proteins and the ESRPs can function both cooperatively and antagonistically in the regulation of exon splicing, although further studies are needed to investigate functional interactions between these regulators (Zhang et al, 2008). However, we did note a number of ESRP-regulated exons that were previously identified as Fox-regulated exons and these included examples in which these splicing regulators promoted the same as well as opposite effects on exon splicing (Supplementary Table S6).

The position of ESRP-binding motifs define an epithelial splicing regulatory ‘RNA map'

To validate that the ESRPs can directly regulate splicing by binding to UGG-rich motifs, we constructed minigenes corresponding to two alternative exons in the ITGA6 and RALGPS2 transcripts that are robustly enhanced by ESRP expression. These exons and conserved flanking intronic sequences were inserted between two adenoviral exons. Both gene transcripts contain a highly conserved sequence element downstream of the exon that contains several putative ESRP-binding sites (Figure 4A and B). We co-transfected these minigenes in 293T cells, which do not express ESRP, with a plasmid directing expression of ESRP1 or empty vector control. Exon inclusion was significantly increased in both minigenes when ESRP1 was co-expressed (Figure 4C). We introduced point mutations in both minigenes to disrupt conserved UGG motifs (Figure 4A and B). In ITGA6 mutations in the UGG motifs abolished the ability of ESRP1 to promote exon inclusion, whereas mutations outside of these motifs did not. In the case of RALGPS2 splicing of the exon containing the mutated sequence motifs was reduced even in the control empty vector co-transfections. This could be due either to reduced binding of other GU-rich binding proteins that may also promote exon inclusion, or through the creation of a silencing element. Nonetheless, these mutations similarly abrogated the ability of ESRP1 to promote exon inclusion (Figure 4C). We validated that these ITGA6 and RALGPS2 sequence elements bind ESRP1 directly by EMSA using RNA probes from the sequences presented in Figure 4A and B. Mutations in the ITGA6 element targeting UGG motifs nearly abolished ESRP1 binding, whereas the mutations in the RALGPS2 element disrupted binding, albeit to a lesser extent (Figure 4D).

Figure 4
The ESRPs regulate splicing through direct binding to UGG-rich motifs. (A) ITGA6 exon 25 and flanking intronic sequences were inserted into the intron of a minigene reporter. A conserved UGG-rich element downstream of the exon is indicated by a grey box ...

We next investigated whether the ESRPs promote exon skipping through binding to UGG-containing motifs in the upstream introns or within the exon itself as predicted by the motif analysis. We constructed a minigene corresponding to an alternative exon in the OSBPL3 transcript that is silenced by the ESRPs. There are putative ESRP-binding sites both directly upstream of and within the alternative exon (Figure 4E). Point mutations were introduced to disrupt the UGG motifs in the upstream intron, in the exon, or in both regions. Exon inclusion was greatly reduced when ESRP1 was co-expressed with the wild-type minigene (Figure 4F). ESRP1 was able to silence exon inclusion when the exonic motifs were mutated, but to a lesser extent compared with wild type. Interestingly, mutation of the intronic motifs alone had a negligible effect on the ability of ESRP1 to promote exon silencing, but when both sets of mutations are made in the minigene, ESRP1-mediated silencing of this exon is nearly abolished. These results suggest that silencing of OSBPL3 exon 9 by the ESRPs is promoted by binding to both sites, although binding at the exonic sites has a predominant function. To further examine whether the ESRPs can silence splicing through binding to upstream intronic sites alone, we constructed a minigene corresponding to the alternative exon in MAGI1 that is strongly silenced by ESRP expression. Here, we noted a highly conserved UGG-containing cis-element upstream of the alternative exon, whereas no clear candidate motifs were present in the exon itself (Figure 4G). Co-transfections revealed that ESRP1 significantly reduced exon inclusion levels in the wild-type minigene but that the UGG mutations essentially abolished ESRP1-mediated silencing activity (Figure 4H). Collectively, these results provide evidence that the ESRPs directly regulate splicing of these exons and strongly implicate UGG-containing motifs as components of ESRP-binding sites that mediate position-dependent splicing regulation.

Sustained ESRP knockdown in mammary epithelial cells induces changes in epithelial–mesenchymal cell markers and morphology consistent with an EMT

To determine whether prolonged disruption of this broad ESRP-regulated splicing programme might induce changes in cell morphology or phenotype indicative of an EMT programme, we transduced human mammary epithelial cells (HMECs) with lentiviruses expressing shRNAs targeting ESRP1, ESRP2, or both proteins together. These shRNAs effectively reduced the levels of each protein (Figure 5A). As expected, prolonged knockdown of the ESRPs resulted in changes of exon inclusion in ESRP-target transcripts (Figure 5B). We noted little or no change in splicing upon knockdown of ESRP2 alone, whereas ESRP1 knockdown was associated with some changes in splicing and combined knockdown generally achieved the largest change, suggesting that ESRP1 is a more robust splicing regulator. To confirm that these changes in splicing were specific for the defined set of ESRP targets, we also tested several unrelated alternatively spliced exons that showed no change in exon inclusion levels (Supplementary Figure S3). Strikingly, HMECs in which ESRP1 alone or both ESRPs were depleted displayed altered cellular morphology, with a loss of typical epithelial cobblestone appearance and acquisition of an elongated spindle shape compared with control cells (Figure 5C). HMECs expressing the ESRP2 shRNA retained the morphology of the control cells (data not shown). To assess whether these morphological changes were indicative of an EMT, we analysed for changes in expression of several prototypical EMT markers. Immunoblotting showed that there was a corresponding induction of protein expression for the mesenchymal markers vimentin, fibronectin, N-cadherin, and α-SMA with ESRP1 or combined ESRP1 and ESRP2 knockdown (Figure 5D). We also observed increased matrix metalloprotein-2 activity upon ESRP1 or combined knockdown by in gel zymography; further evidence of invasive mesenchymal cell properties (Figure 5E) (Liu et al, 2009). However, there was negligible if any reduction in E-cadherin protein expression or that of other epithelial cell-specific markers (Figure 5D and data not shown). Nonetheless, immunofluorescence analysis of E-cadherin protein showed that ESRP knockdown induced a generalized loss of E-cadherin localization from the plasma membrane at sites of cell–cell junctions and relocalization in a diffuse cytoplasmic pattern, consistent with changes in E-cadherin during the EMT in other model systems (Figure 5F). To investigate the phenotypic consequences of ESRP depletion, we used the scratch wound assay, which demonstrated that a loss of the ESRP-splicing programme promoted a modest, but significant increase in cell motility, consistent with the acquisition of mesenchymal cell properties (Figure 5G and H). Together, these results demonstrate that disruption of the epithelial-splicing programme, regulated predominantly by ESRP1, is sufficient to induce epithelial cells to undergo some of the changes in gene expression and morphology that are associated with the EMT. We note that although ESRP loss is associated with an increase in expression of several mesenchymal markers, there was not an overt decrease in expression of epithelial markers such as E-cadherin. These findings thus indicate that even in the absence of a complete switch in the transcription of several known EMT markers, these global changes in splicing can induce some of the transitions in cell behaviour that are observed during the EMT. These studies also support our hypothesis that the ESRP-regulated splicing programme is an essential feature of the epithelial phenotype and that many changes in cell behaviour associated with the EMT are due to alterations in the functions of proteins that undergo isoform switching during this process.

Figure 5
Identification of a discrete ESRP-regulated splicing signature that can distinguish epithelial from mesenchymal cell types. (A) Western blot using a monoclonal antibody that recognizes both ESRPs to validate specific shRNA-mediated depletion in the nucleus ...

A splicing signature that can be used to define epithelial and mesenchymal cell types and the EMT

Using the validated set of ESRP-regulated exons with large changes in exon inclusion, we chose a panel of alternative exons to derive a select set of splicing assays that might distinguish epithelial from mesenchymal cell types and that might also serve as markers that change during the EMT. We used HT-RT–PCR to assay splicing of 13 ESRP-enhanced exons and seven ESRP-silenced exons across six epithelial cell lines and three mesenchymal cell lines. These cell lines have previously been characterized based on traditional epithelial and mesenchymal markers such as E-cadherin and vimentin, and we showed that ESRP1 and ESRP2 mRNA expression is limited to the epithelial cell lines (Warzecha et al, 2009a). A distinct epithelial ‘splicing signature' of 10 exons that are specifically included or skipped in epithelial cells emerged from this analysis (Figure 5I; Supplementary Figure S4A and B). ESRP expression is strongly downregulated during the EMT induced in HMECs by the transcription factor Twist. We therefore also assayed alternative splicing following Twist-induced EMT for changes in splicing of this subset of alternative splicing events. This analysis revealed a switch from epithelial-to-mesenchymal exon inclusion for transcripts derived from this splicing signature (Figure 5J). These data, thus, clearly establish that distinct splicing switches that occur during the EMT are regulated by the ESRPs and that limited components of this splicing signature hold promise for use as specific markers that can be used to define the EMT as well as in the classification of cell types.

It is noteworthy that among the cell lines we used to profile this splicing signature were several breast cancer cell lines that have been categorized as belonging to cancer subtypes designated luminal (MCF7, SKBR3, T47D, and ZR75) or basal B (MDA-MB-231, MDA-MB-435, and BT549). In contrast to luminal or basal A cell lines, the basal B cell lines are representative of cancer cells that have undergone the EMT and have greater metastatic potential, invasive properties, and resistance to several cancer therapies (Neve et al, 2006; Blick et al, 2008; Sarrio et al, 2008). Consistent with a loss of an ESRP-regulated splicing signature in the cell lines analysed here, analysis of the expression data from the large set of cell lines used to define these cancer cell line subtypes revealed a very strong association of low or absent ESRP1 and ESRP2 expression with basal B-cell classification (Supplementary Figure S4C). Furthermore, low ESRP1 and ESRP2 mRNA expression was also a component of a gene signature in breast cancer specimens classified as claudin-low and metaplastic, which are also more aggressive and show EMT-like features (Hennessy et al, 2009). It is thus likely that this splicing signature is broadly associated with cancer cells with EMT features and might therefore have future clinical applications.

Discussion

We defined an extensive ESRP-regulated splicing network and provided strong evidence that alternative splicing provides an additional layer at which gene expression is regulated during the EMT. Furthermore, our data demonstrate that this post-transcriptional gene expression programme is distinct from the transcriptional gene expression programme that is altered during this process; thereby pointing out the degree to which alternative splicing contributes to the global transcriptomic complexity that is implicated in the EMT. Our data demonstrating that the loss of this splicing programme alone can induce some of the phenotypic changes that occur during the EMT support our hypothesis that the ESRP-regulated splicing programme is an essential feature of the epithelial phenotype and that many changes in cell behaviour associated with the EMT are due to alterations in functions of proteins that undergo isoform switching during this process. We suggest that the ESRPs are among the most essential epithelial-specific genes that, like E-cadherin, must be downregulated directly or indirectly by mesenchymal transcription factors such as Twist, Snail/Slug, or Zeb1/2 in order to achieve a complete phenotypic cellular switch in the EMT.

Our observation of changes in cell morphology and behaviour through abrogation of an epithelial-splicing programme suggests that the ESRPs regulate transcripts that, while expressed in both epithelial and mesenchymal cells, encode protein isoforms specific to each cell type with isoform-specific functions that impact the EMT. A prototypical example of such a protein is p120-catenin, which has been extensively documented to have a function in the EMT and undergo isoform switches during this process. The ESRPs regulate the splicing of exons that yield distinct epithelial versus mesenchymal p120-catenin isoforms that oppositely promote cell–cell adhesion in epithelial cells and cell migration and invasion in mesenchymal cells (Yanagisawa et al, 2008; Warzecha et al, 2009a). Among the ESRP targets identified in this study are numerous examples of transcripts that encode proteins with functions in processes that underlie the EMT such as regulation of the actin cytoskeleton, cell–cell and cell–matrix adhesion, migration, and cell polarity. For example, NUMB has been shown to have an essential function in the maintenance of cell polarity and cell–cell adhesion through binding to E-cadherin through an N-terminal phosphotyrosine-binding (PTB) domain. Abrogation of this interaction during the EMT leads to loss of cell–cell adhesion and increased cell migration (Wang et al, 2009). The 33 nt epithelial-specific exon in NUMB encodes an 11 amino acid insert in this PTB domain that is likely to promote the NUMB-E-cadherin interaction. Consistent with this possibility, isoforms that contain this insert are predominantly associated with the plasma membrane, whereas isoforms that exclude it are diffusely cytoplasmic (Dho et al, 1999). Relocalization of β-catenin from adherens junctions to the nucleus occurs when E-cadherin is downregulated, and in association with TCF/LEF transcription family members (including TCF7L2) induces changes in gene expression that are associated with the EMT (Medici et al, 2008). The ESRPs promote inclusion of exon 4 in TCF7L2 (also known as TCF4) transcripts, whereas this exon is predominantly skipped in mesenchymal cells. Isoforms that include exon 4 were recently shown to have reduced transactivation activity of several Wnt/β-catenin-target genes, suggesting the interesting possibility that a partial switch towards the skipped isoform during the EMT might further enhance the transcriptional activation of β-catenin-target genes upon its translocation to the nucleus (Weise et al, 2010). These examples thus not only highlight several of the ESRP-regulated transcripts with EMT-related functions, but also cases in which there is evidence suggesting that the resulting epithelial versus mesenchymal isoforms have distinct and potentially even opposing functions in the maintenance of cell adhesion versus cell migration.

Another notable ESRP target encoding alternative protein isoforms that have been functionally characterized with implications for the function of splicing in cancer is Bin1. Bin1 binds Myc and inhibits its oncogenic activity, but isoforms that lack the domain encoded by exon 12A lack this tumour suppressor activity (Ge et al, 1999). A previous study showed that overexpression of the splicing factor SRSF1 caused increased inclusion of exon 12A, one of several alterations it induced in alternative splicing suggesting that this SR protein functions as a proto-oncogene (Karni et al, 2007). Conversely, the ESRPs promote skipping of exon 12A, thereby promoting expression of the ‘tumour suppressor' isoform and the hypothesis that ESRP1 itself is a tumour suppressor was put forth prior to our finding that it is a splicing regulator (Leontieva and Ionov, 2009). Although most of the changes in splicing of ESRP-regulated exons have not been functionally characterized, several additional examples of ESRP-regulated transcripts in which splicing can change the functions of the resulting isoforms, or affect domains predicted to change their functions in ways that might be relevant for the EMT are described in Supplementary Table S7. This study therefore sets the stage for further cell biological investigations into the functional consequences of the changes in splicing in numerous transcripts as a result of ESRP downregulation during the EMT. As more studies begin to catalog the functions of these proteins, it will likely become clear that functional annotations of gene products will need to increasingly incorporate isoform-specific properties into functional genomic databases in order to fully capture gene-function associations.

In the published literature, we also identified examples that support the hypothesis that the ESRPs co-regulate transcripts that function in common processes and pathways. One example of functionally related ESRP targets with functions in the EMT includes p38 (MAPK14), p38IP (FAM48A), and MAP4K4 (NIK). NIK has been shown to act upstream to activate p38 pathways that require p38IP to repress E-cadherin expression at the protein level during the EMT (Zohn et al, 2006). Thus, changes in the splicing of these transcripts may influence the dynamics of this pathway during the EMT. Another example that illustrates this concept is a coherent set of ESRP targets encoding proteins that function in vesicle-mediated protein trafficking pathways that have essential functions in the maintenance of cell polarity and cell–cell adhesion (Mellman and Nelson, 2008). The exocyst is a hetero-octameric complex that is essential for maintenance of epithelial polarity through the targeting of secretory and membrane-bound proteins to specific sites at the plasma membrane (He and Guo, 2009). In contrast, in cells that lack epithelial E-cadherin, the exocyst is localized to the leading edge and focal complexes of migrating cells (Spiczka and Yeaman, 2008). We identified two ESRP-regulated epithelial-specific exons in two of the eight exocyst proteins, EXOC1 (Sec3) and EXOC7 (Exo70) (Figure 6). These exocyst proteins are the two membrane proximal components of the complex that are essential for vesicle targeting and Rho GTPase recruitment and are therefore at a critical interface in regulating polarized exocytosis. RalA is required upstream for exocyst assembly and function and the RalA-specific activating guanine nucleotide exchange factor (GEF) RALGPS2 displayed nearly complete switches in splicing (Cascone et al, 2008). The epithelial-specific exon in RALGPS2 is an in-frame exon located between a PxxP motif and pleckstrin homology (PH) domain, a region of the protein previously proposed to regulate its GEF activity (Ceriani et al, 2007). It therefore will be of interest to determine functional differences in activity of RALGPS2, EXOC1, and EXOC7 protein isoforms that may differentially impact the diverse activities of exocyst complex. We also noted several ESRP targets among SNARE protein complexes that mediate downstream vesicle fusion with the plasma membrane, including Syntaxin 2 (STX2; epimorphin) and Syntaxin 3 (STX3) as well as syntaxin-binding proteins STXBP2 (Hunc18b), STXBP5 (tomosyn), and STXBP6 (amisyn) (Kauppi et al, 2002; Sharma et al, 2006). Taken together, the multiple ESRP-regulated splicing events that connect upstream and downstream functions of the exocyst complex suggest that the different isoforms carry out functions that differentially impact exocytosis to either maintain cell polarity in epithelial cells or regulate focal complex turnover and cell migration in mesenchymal cells (Figure 6).

Figure 6
ESRP-regulated gene transcripts encode multiple components and regulators of the vesicular transport system. (A) Schematic illustrating ESRP-target transcripts that function in vesicular transport events mediated through exocyst function. Exocyst components ...

We also performed Gene Ontology analysis using the DAVID functional annotation tool to characterize processes and functions that were enriched among the ESRP-target genes (Dennis et al, 2003). Consistent with our hypothesis that the ESRPs regulate gene products with EMT-related functions, we observed enrichment for genes involved in maintenance of cell–cell adhesion, cell motility, and cell–matrix adhesion (Supplementary Table S8). There was also enrichment for regulators and effectors of small GTPases, including the Rho GTPases that are well characterized to have functions in the EMT as well as exocyst function (Iden and Collard, 2008; He and Guo, 2009). Another enriched category was protein serine/threonine kinases, including components of ERK/MAPK signalling pathways. Analysis of KEGG pathways showed enriched participation of these targets in regulation of the actin cytoskeleton and maintenance of adherens junctions.

Our identification of a splicing signature of ESRP-regulated exons provides a novel set of biomarkers for the EMT that can facilitate analysis of cancers in which analysis of EMT-like changes may predict prognosis or therapeutic responses. As many of the splicing-target genes we identified are expressed in both epithelial and mesenchymal cells, these splicing switches provide a set of gene expression changes distinct from those that are regulated at the level of transcription. Current markers used to characterize the EMT rely on collections of transcript or protein biomarkers that have utility in different subtypes of EMT (Zeisberg and Neilson, 2009). However, the utility of these standard markers is often compromised in tissue specimens that can often have significant protein or RNA degradation. In contrast to the use of whole transcript profiling of RNA, the determination of changes in splicing provides a tool that can define epithelial and mesenchymal cell types even when RNA quality is compromised, provided that products can be obtained by RT–PCR. We are currently testing the utility of this signature in an expanded set of cell lines and tissues in order to determine a splicing metric that can be applied to quantitatively define the EMT in diverse tissues and pathologic specimens.

Sequence enrichment analysis in ESRP-regulated exons and flanking introns was used to derive a candidate-binding motif that was validated as a direct ESRP-binding site as well as an ESRP-responsive element in minigene reporters. Mapping of this motif relative to ESRP-regulated exons revealed a trend suggestive of an RNA map of position-dependent ESRP binding in which binding sites in the downstream intron promote exon inclusion and binding within or upstream of an exon promote exon skipping. This RNA map is similar to that previously characterized for position-dependent functions of the Nova- and Fox-splicing regulators, with the possible exception that the ESRPs would appear to more commonly regulate exon silencing through direct binding within exons (Licatalosi et al, 2008; Zhang et al, 2008). However, some exceptions were observed that might be due to more degenerate UG-rich ESRP-binding sites or presence of binding sites in more distal intron regions or flanking exons. Alternatively, some of the ESRP-induced splicing changes we observed may be indirect as we identified a number of known splicing factors among the ESRP targets. It is therefore possible that some of the splicing changes may come about through modifications in the activity of these splicing regulators. Nonetheless, our results suggest that many if not most of the splicing changes we observed are likely to involve direct affects on splicing of these targets and that the identification of ESRP-binding sites can predict de novo targets and the splicing outcome based on the position of binding. Experiments to map genome-wide ESRP-binding sites using cross-linking and immunoprecipitation are underway and will further refine the ‘RNA map' that predicts the functions of these proteins in alternative splicing, including more complex types of alternative splicing. We note, however, that an RNA-binding map for a single-splicing regulator alone is not sufficient to determine a comprehensive splicing code. Given the overlap of this ESRP-SRN with the Fox-splicing network, the predictive capabilities of an ‘RNA map' derived from a single regulator is limited when applied to splicing outcomes in different cellular milieus. Such observations further confirm that global combinatorial control of alternative splicing involves large and overlapping networks of splicing regulators with different activities in different cell types that together determine the transcriptome within any given cell. However, through compilation of RNA-binding maps and splicing events regulated by expanding numbers of splicing factors we will not only unveil a more definitive splicing code, but also begin to uncover the molecular mechanisms by which these regulators function.

Materials and methods

Cell culture, transfection, and transduction

PNT2, MDA-MB-231, HMEC (HMLE), and 293T cells were maintained, transfected, and retrovirally transduced as described (Warzecha et al, 2009a). Permanent ESRP1 and 2 knockdown in HMLE cells was achieved with miR-30-based shRNA lentiviral plasmids with puromycin and blasticidin selection cassettes and selection in 0.5 μg/ml of puromycin and 10 μg/ml of blasticidin S.

HJAY microarray target preparation and hybridization

RNA purification, preparation of biotinylated sense-strand DNA, and hybridization was carried out as described for exon array analysis (Warzecha et al, 2009b). The CEL file data sets can be accessed at NCBI GEO repository through accession number GSE22937.

RT–PCR and HT-RT–PCR

Quantification of splicing by standard RT–PCR and high-throughput RT–PCR at the Université de Sherbrooke was performed as described (Warzecha et al, 2009b). For complex splicing events, restriction digestion and/or DNA sequencing was used to validate PCR product identity. Complete HT-RT–PCR data can be accessed at http://palace.lgfus.ca.

Antibodies and immunoblotting

Cell lysates were resolved on 4–12% NuPAGE Bis-tris gels (Invitrogen) and immunoblotted as described (Warzecha et al, 2009a). Antibodies used were anti-β actin (Sigma AC-15, 1:1000), anti-Flag (Stratagene, 1:1000), anti-Vimentin (Thermo Scientific Ab2 Clone V9, 1:150), anti-Fibronectin (BD Transduction, 1:10 000), anti-smooth muscle actin (Sigma, 1:300), anti-N-cadherin (BD Transduction, 1:250), and anti-E-cadherin (Cell Signaling Technology, 1:1000). ESRP1 antibodies that specifically detect ESRP1 (clone 27H12) and both ESRP1 and ESRP2 (clone 23A7) were obtained as custom monoclonal antibodies against our Glutathione-s-transferase (GST)-Esrp1 recombinant protein as prepared by Rockland Immunochemicals, Inc. (Gilbertsville, PA). These antibodies were shown to be specific with no detectable cross-reactivity against the closest RNA-binding protein homologs GRSF1, hnRNPF, or hnRNPH1.

Recombinant protein production and RNA EMSA analysis

Production of recombinant GST-Esrp1 fusion protein from plasmid GST-Esrp1-FL was performed using standard techniques and purified using Glutathione sepharose beads (GE healthcare). Sequences for in vitro transcription were inserted into the ClaI-Xho restriction sites in plasmid pDP19RC-ΔEE and transcription by T7 polymerase (Ambion) after XhoI digestion was carried out as described (Hovhannisyan and Carstens, 2007). [32P]UTP-radiolabelled RNAs were used at a specific activity of 2.4 × 108 CPM/μg and 50 000 CPM (~8–13 fmol) of radiolabelled RNA and GST-Esrp1 were incubated in a 10-μl binding reaction (16 mM Hepes pH 7.9, 95 mM KCl, 2 mM MgCl2, 60 μM EDTA, 1 mM DTT, 6% glycerol, 0.1 mM heparin (Sigma #H-3393), and 0.15 mM PMSF). Binding reactions were incubated at 30°C for 30 min and loaded on 4% non-denaturing Tris-borate acrylamide gels and electrophoresed at 4°C for 1.5 h at 200 V. Quantification for Kd determination was performed using a PhosphorImager (Molecular Dynamics) and the % RNA bound was determined by calculating the ratio of [(shifted complexes)/(free RNA+shifted complexes)]. Data were analysed using the dose response with variable slope regression equation in Prism software and the Kd was determined using the EC50 value corresponding to the concentration of ESRP1 at 50% binding saturation.

In gel zymography

Cells were cultured in serum-free MEGM (Lonza) for 24 h. Cell culture media was lyophilized to dryness and rehydrated with SDS loading buffer (10% SDS, 50% glycerol, 0.4 M Tris, pH 6.8, and 0.1% bromophenol blue) and separated on 8% polyacrylamide/0.3% gelatin gels. Gels were then washed in 0.01% Triton X-100, 1 h rocking gently, and incubated in reaction buffer (50 mM Tris, pH 8.0, and 5 mM CaCl2) at 37°C for 24 h. After the reaction, gels were stained with staining buffer (0.12% Coomassie blue R-250, 50% methanol, and 20% acetic acid) for 1 h and destained overnight with destaining buffer (22% methanol and 10% acetic acid). Gel loadings were normalized to total protein measured with a Bio-Rad Protein Assay (Richmond, CA).

Immunofluorescence

Transduced HMEC cells were grown on coverslips, washed with phosphate-buffered saline (PBS), fixed in 4% paraformaldehyde at room temperature for 12 min, washed, permeabilized for 5 min with PBS+0.2% Triton X-100, and blocked for 10 min with 10% goat serum in PBS. The coverslips were incubated sequentially with anti-E-cadherin primary antibody (1:150) and Texas Red-conjugated anti-Rabbit secondary antibody (1:4000) and counterstained with DAPI blue.

In vitro scratch assay

Cells were cultured to confluence on six-well plates, and scratched with a 200-μl pipette tip. Detached cells were removed, and the remaining cells were incubated with MEGM (Lonza). Scratched areas for each sample were marked and photographed immediately or at the indicated time points. For quantification, the same analysis was performed with nine replicate scratches for the control and combined ESRP1 and ESRP2 knockdown cells. Each scratch was then visually observed until the time point at which complete closure of the wound occurred.

Supplementary Material

Supplementary Data:
Supplementary Table S1:
Supplementary Table S2:
Supplementary Table S3:
Supplementary Table S4:
Supplementary Table S5:
Supplementary Table S6:
Supplementary Table S8:
Review Process File:

Acknowledgments

We thank Don Baldwin, Yi Li, and John Hogenesch for technical support and discussion; Kristen Lynch for review of the manuscript; and the University of Iowa ICTS for computer support. This research was supported by NIH grants R01-CA093769 and R01-GM088809 (RPC), R01-HG004634 and Edward Mallinckrodt, Jr Foundation (YX), R01-GM085146 (WG), and T32-GM08216 (CCW).

Footnotes

The authors declare that they have no conflict of interest.

References

  • Barash Y, Calarco JA, Gao W, Pan Q, Wang X, Shai O, Blencowe BJ, Frey BJ (2010) Deciphering the splicing code. Nature 465: 53–59 [PubMed]
  • Blick T, Widodo E, Hugo H, Waltham M, Lenburg ME, Neve RM, Thompson EW (2008) Epithelial mesenchymal transition traits in human breast cancer cell lines. Clin Exp Metastasis 25: 629–642 [PubMed]
  • Cascone I, Selimoglu R, Ozdemir C, Del Nery E, Yeaman C, White M, Camonis J (2008) Distinct roles of RalA and RalB in the progression of cytokinesis are supported by distinct RalGEFs. EMBO J 27: 2375–2387 [PMC free article] [PubMed]
  • Ceriani M, Scandiuzzi C, Amigoni L, Tisi R, Berruti G, Martegani E (2007) Functional analysis of RalGPS2, a murine guanine nucleotide exchange factor for RalA GTPase. Exp Cell Res 313: 2293–2307 [PubMed]
  • Chen M, Manley JL (2009) Mechanisms of alternative splicing regulation: insights from molecular and genomics approaches. Nat Rev Mol Cell Biol 10: 741–754 [PMC free article] [PubMed]
  • Dennis G Jr, Sherman BT, Hosack DA, Yang J, Gao W, Lane HC, Lempicki RA (2003) DAVID: database for annotation, visualization, and integrated discovery. Genome Biol 4: P3. [PMC free article] [PubMed]
  • Dho SE, French MB, Woods SA, McGlade CJ (1999) Characterization of four mammalian numb protein isoforms. Identification of cytoplasmic and membrane-associated variants of the phosphotyrosine binding domain. J Biol Chem 274: 33097–33104 [PubMed]
  • Ge K, DuHadaway J, Du W, Herlyn M, Rodeck U, Prendergast GC (1999) Mechanism for elimination of a tumor suppressor: aberrant splicing of a brain-specific exon causes loss of function of Bin1 in melanoma. Proc Natl Acad Sci USA 96: 9689–9694 [PMC free article] [PubMed]
  • Hartmann B, Valcarcel J (2009) Decrypting the genome's alternative messages. Curr Opin Cell Biol 21: 377–386 [PubMed]
  • He B, Guo W (2009) The exocyst complex in polarized exocytosis. Curr Opin Cell Biol 21: 537–542 [PMC free article] [PubMed]
  • Hennessy BT, Gonzalez-Angulo AM, Stemke-Hale K, Gilcrease MZ, Krishnamurthy S, Lee JS, Fridlyand J, Sahin A, Agarwal R, Joy C, Liu W, Stivers D, Baggerly K, Carey M, Lluch A, Monteagudo C, He X, Weigman V, Fan C, Palazzo J et al. (2009) Characterization of a naturally occurring breast cancer subset enriched in epithelial-to-mesenchymal transition and stem cell characteristics. Cancer Res 69: 4116–4124 [PMC free article] [PubMed]
  • Hovhannisyan RH, Carstens RP (2007) Heterogeneous ribonucleoprotein m is a splicing regulatory protein that can enhance or silence splicing of alternatively spliced exons. J Biol Chem 282: 36265–36274 [PubMed]
  • Iden S, Collard JG (2008) Crosstalk between small GTPases and polarity proteins in cell polarization. Nat Rev Mol Cell Biol 9: 846–859 [PubMed]
  • Karni R, de Stanchina E, Lowe SW, Sinha R, Mu D, Krainer AR (2007) The gene encoding the splicing factor SF2/ASF is a proto-oncogene. Nat Struct Mol Biol 14: 185–193 [PubMed]
  • Kauppi M, Wohlfahrt G, Olkkonen VM (2002) Analysis of the Munc18b-syntaxin binding interface. Use of a mutant Munc18b to dissect the functions of syntaxins 2 and 3. J Biol Chem 277: 43973–43979 [PubMed]
  • LaGamba D, Nawshad A, Hay ED (2005) Microarray analysis of gene expression during epithelial-mesenchymal transformation. Dev Dyn 234: 132–142 [PubMed]
  • Leontieva OV, Ionov Y (2009) RNA-binding motif protein 35A is a novel tumor suppressor for colorectal cancer. Cell Cycle 8: 490–497 [PubMed]
  • Licatalosi DD, Mele A, Fak JJ, Ule J, Kayikci M, Chi SW, Clark TA, Schweitzer AC, Blume JE, Wang X, Darnell JC, Darnell RB (2008) HITS-CLIP yields genome-wide insights into brain alternative RNA processing. Nature 456: 464–469 [PMC free article] [PubMed]
  • Liu J, Yue P, Artym VV, Mueller SC, Guo W (2009) The role of the exocyst in matrix metalloproteinase secretion and actin dynamics during tumor cell invadopodia formation. Mol Biol Cell 20: 3763–3771 [PMC free article] [PubMed]
  • Lu H, Lin L, Sato S, Xing Y, Lee CJ (2009) Predicting functional alternative splicing by measuring RNA selection pressure from multigenome alignments. PLoS Comput Biol 5: e1000608. [PMC free article] [PubMed]
  • Medici D, Hay ED, Olsen BR (2008) Snail and Slug promote epithelial-mesenchymal transition through beta-catenin-T-cell factor-4-dependent expression of transforming growth factor-beta3. Mol Biol Cell 19: 4875–4887 [PMC free article] [PubMed]
  • Mellman I, Nelson WJ (2008) Coordinated protein sorting, targeting and distribution in polarized cells. Nat Rev Mol Cell Biol 9: 833–845 [PMC free article] [PubMed]
  • Neve RM, Chin K, Fridlyand J, Yeh J, Baehner FL, Fevr T, Clark L, Bayani N, Coppe JP, Tong F, Speed T, Spellman PT, DeVries S, Lapuk A, Wang NJ, Kuo WL, Stilwell JL, Pinkel D, Albertson DG, Waldman FM et al. (2006) A collection of breast cancer cell lines for the study of functionally distinct cancer subtypes. Cancer Cell 10: 515–527 [PMC free article] [PubMed]
  • Nilsen TW, Graveley BR (2010) Expansion of the eukaryotic proteome by alternative splicing. Nature 463: 457–463 [PMC free article] [PubMed]
  • Pan Q, Shai O, Lee LJ, Frey BJ, Blencowe BJ (2008) Deep surveying of alternative splicing complexity in the human transcriptome by high-throughput sequencing. Nat Genet 40: 1413–1415 [PubMed]
  • Pan Q, Shai O, Misquitta C, Zhang W, Saltzman AL, Mohammad N, Babak T, Siu H, Hughes TR, Morris QD, Frey BJ, Blencowe BJ (2004) Revealing global regulatory features of Mammalian alternative splicing using a quantitative microarray platform. Mol Cell 16: 929–941 [PubMed]
  • Resch A, Xing Y, Alekseyenko A, Modrek B, Lee C (2004) Evidence for a subpopulation of conserved alternative splicing events under selection pressure for protein reading frame preservation. Nucleic Acids Res 32: 1261–1269 [PMC free article] [PubMed]
  • Sarrio D, Rodriguez-Pinilla SM, Hardisson D, Cano A, Moreno-Bueno G, Palacios J (2008) Epithelial-mesenchymal transition in breast cancer relates to the basal-like phenotype. Cancer Res 68: 989–997 [PubMed]
  • Sharma N, Low SH, Misra S, Pallavi B, Weimbs T (2006) Apical targeting of syntaxin 3 is essential for epithelial cell polarity. J Cell Biol 173: 937–948 [PMC free article] [PubMed]
  • Shen S, Warzecha CC, Carstens RP, Xing Y (2010) MADS+: discovery of differential splicing events from Affymetrix exon junction array data. Bioinformatics 26: 268–269 [PMC free article] [PubMed]
  • Spiczka KS, Yeaman C (2008) Ral-regulated interaction between Sec5 and paxillin targets Exocyst to focal complexes during cell migration. J Cell Sci 121: 2880–2891 [PubMed]
  • Thiery JP, Acloque H, Huang RY, Nieto MA (2009) Epithelial-mesenchymal transitions in development and disease. Cell 139: 871–890 [PubMed]
  • Venables JP, Klinck R, Koh C, Gervais-Bird J, Bramard A, Inkel L, Durand M, Couture S, Froehlich U, Lapointe E, Lucier JF, Thibault P, Rancourt C, Tremblay K, Prinos P, Chabot B, Elela SA (2009) Cancer-associated regulation of alternative splicing. Nat Struct Mol Biol 16: 670–676 [PubMed]
  • Wang ET, Sandberg R, Luo S, Khrebtukova I, Zhang L, Mayr C, Kingsmore SF, Schroth GP, Burge CB (2008) Alternative isoform regulation in human tissue transcriptomes. Nature 456: 470–476 [PMC free article] [PubMed]
  • Wang Z, Sandiford S, Wu C, Li SS (2009) Numb regulates cell-cell adhesion and polarity in response to tyrosine kinase signalling. EMBO J 28: 2360–2373 [PMC free article] [PubMed]
  • Warzecha CC, Sato TK, Nabet B, Hogenesch JB, Carstens RP (2009a) ESRP1 and ESRP2 are epithelial cell-type-specific regulators of FGFR2 splicing. Mol Cell 33: 591–601 [PMC free article] [PubMed]
  • Warzecha CC, Shen S, Xing Y, Carstens RP (2009b) The epithelial splicing factors ESRP1 and ESRP2 positively and negatively regulate diverse types of alternative splicing events. RNA Biol 6: 546–562 [PMC free article] [PubMed]
  • Weise A, Bruser K, Elfert S, Wallmen B, Wittel Y, Wohrle S, Hecht A (2010) Alternative splicing of Tcf7l2 transcripts generates protein variants with differential promoter-binding and transcriptional activation properties at Wnt/{beta}-catenin targets. Nucleic Acids Res 38: 1964–1981 [PMC free article] [PubMed]
  • Xing Y, Lee CJ (2005) Protein modularity of alternatively spliced exons is associated with tissue-specific regulation of alternative splicing. PLoS Genet 1: e34. [PMC free article] [PubMed]
  • Xue Y, Zhou Y, Wu T, Zhu T, Ji X, Kwon YS, Zhang C, Yeo G, Black DL, Sun H, Fu XD, Zhang Y (2009) Genome-wide analysis of PTB-RNA interactions reveals a strategy used by the general splicing repressor to modulate exon inclusion or skipping. Mol Cell 36: 996–1006 [PMC free article] [PubMed]
  • Yanagisawa M, Huveldt D, Kreinest P, Lohse CM, Cheville JC, Parker AS, Copland JA, Anastasiadis PZ (2008) A p120 catenin isoform switch affects Rho activity, induces tumor cell invasion, and predicts metastatic disease. J Biol Chem 283: 18344–18354 [PMC free article] [PubMed]
  • Yeo GW, Coufal NG, Liang TY, Peng GE, Fu XD, Gage FH (2009) An RNA code for the FOX2 splicing regulator revealed by mapping RNA-protein interactions in stem cells. Nat Struct Mol Biol 16: 130–137 [PMC free article] [PubMed]
  • Yeo GW, Van Nostrand E, Holste D, Poggio T, Burge CB (2005) Identification and analysis of alternative splicing events conserved in human and mouse. Proc Natl Acad Sci USA 102: 2850–2855 [PMC free article] [PubMed]
  • Zeisberg M, Neilson EG (2009) Biomarkers for epithelial-mesenchymal transitions. J Clin Invest 119: 1429–1437 [PMC free article] [PubMed]
  • Zhang C, Zhang Z, Castle J, Sun S, Johnson J, Krainer AR, Zhang MQ (2008) Defining the regulatory network of the tissue-specific splicing factors Fox-1 and Fox-2. Genes Dev 22: 2550–2563 [PMC free article] [PubMed]
  • Zohn IE, Li Y, Skolnik EY, Anderson KV, Han J, Niswander L (2006) p38 and a p38-interacting protein are critical for downregulation of E-cadherin during mouse gastrulation. Cell 125: 957–969 [PubMed]

Articles from The EMBO Journal are provided here courtesy of The European Molecular Biology Organization
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...