A major limitation of RNA sequencing (RNA-seq) analysis of alternative splicing is its reliance on high sequencing coverage. We report DARTS (https://github.com/Xinglab/DARTS), a computational framework that integrates deep-learning-based predictions with empirical RNA-seq evidence to infer differential alternative splicing between biological samples. DARTS leverages public RNA-seq big data to provide a knowledge base of splicing regulation via deep learning, thereby helping researchers better characterize alternative splicing using RNA-seq datasets even with modest coverage.

The glomerular podocyte is a highly specialized and polarized kidney cell type that contains major processes and foot processes that extend from the cell body. Foot processes from adjacent podocytes form interdigitations with those of adjacent cells, thereby creating an essential intercellular junctional domain of the renal filtration barrier known as the slit diaphragm. Interesting parallels have been drawn between the slit diaphragm and other sites of cell-cell contact by polarized cells. Notably mutations in several genes encoding proteins localized to the foot processes can lead to proteinuria and kidney failure. Mutations in the Wilm's tumor gene (WT1) can also lead to kidney disease and one isoform of WT1, WT1(+KTS), has been proposed to regulate gene expression post-transcriptionally. We originally sought to identify mRNAs associated with WT1(+KTS) through an RNA immunoprecipitation and microarray approach, hypothesizing that the proteins encoded by these mRNAs might be important for podocyte morphology and function. We identified a subset of mRNAs that were remarkably enriched for transcripts encoding actin-binding proteins and other cytoskeletal proteins including several that are localized at or near the slit diaphragm. Interestingly, these mRNAs included those of alpha-actinin-4 and non-muscle myosin IIA that are mutated in genetic forms of kidney disease. However, isolation of the mRNAs occurred independently of the expression of WT1, suggesting that the identified mRNAs were serendipitously co-purified on the basis of co-association in a common subcellular fraction. Mass spectroscopy revealed that other components of the actin cytoskeleton co-purified with these mRNAs, namely actin, tubulin, and elongation factor 1alpha. We propose that these mRNAs encode a number of proteins that comprise a highly specialized protein interactome underlying the slit diaphragm. Collectively, these gene products and their interactions may prove to be important for the structural integrity of the actin cytoskeleton in podocytes as well as other polarized cell types.

Metastatic breast cancer is a leading cause of cancer-related deaths in women worldwide. Patients with triple negative breast cancer (TNBCs), a highly aggressive tumor subtype, have a particularly poor prognosis. Multiple reports demonstrate that altered content of the multicopy mitochondrial genome (mtDNA) in primary breast tumors correlates with poor prognosis. We earlier reported that mtDNA copy number reduction in breast cancer cell lines induces an epithelial-mesenchymal transition associated with metastasis. However, it is unknown whether the breast tumor subtypes (TNBC, Luminal and HER2+) differ in the nature and amount of mitochondrial defects and if mitochondrial defects can be used as a marker to identify tumors at risk for metastasis. By analyzing human primary tumors, cell lines and the TCGA dataset, we demonstrate a high degree of variability in mitochondrial defects among the tumor subtypes and TNBCs, in particular, exhibit higher frequency of mitochondrial defects, including reduced mtDNA content, mtDNA sequence imbalance (mtRNR1:ND4), impaired mitochondrial respiration and metabolic switch to glycolysis which is associated with tumorigenicity. We identified that genes involved in maintenance of mitochondrial structural and functional integrity are differentially expressed in TNBCs compared to non-TNBC tumors. Furthermore, we identified a subset of TNBC tumors that contain lower expression of epithelial splicing regulatory protein (ESRP)-1, typical of metastasizing cells. The overall impact of our findings reported here is that mitochondrial heterogeneity among TNBCs can be used to identify TNBC patients at risk of metastasis and the altered metabolism and metabolic genes can be targeted to improve chemotherapeutic response.

Alternative pre-mRNA splicing is a prevalent mechanism in mammals that promotes proteomic diversity, including expression of cell-type specific protein isoforms. We characterized a role for RBM38 (RNPC1) in regulation of alternative splicing during late erythroid differentiation. We used an Affymetrix human exon junction (HJAY) splicing microarray to identify a panel of RBM38-regulated alternatively spliced transcripts. Using microarray databases, we noted high RBM38 expression levels in CD71(+) erythroid cells and thus chose to examine RBM38 expression during erythroid differentiation of human hematopoietic stem cells, detecting enhanced RBM38 expression during late erythroid differentiation. In differentiated erythroid cells, we validated a subset of RBM38-regulated splicing events and determined that RBM38 regulates activation of Protein 4.1R (EPB41) exon 16 during late erythroid differentiation. Using Epb41 minigenes, Rbm38 was found to be a robust activator of exon 16 splicing. To further address the mechanism of RBM38-regulated alternative splicing, a novel mammalian protein expression system, followed by SELEX-Seq, was used to identify a GU-rich RBM38 binding motif. Lastly, using a tethering assay, we determined that RBM38 can directly activate splicing when recruited to a downstream intron. Together, our data support the role of RBM38 in regulating alternative splicing during erythroid differentiation.

During liver regeneration, most new hepatocytes arise via self-duplication; yet, the underlying mechanisms that drive hepatocyte proliferation following injury remain poorly defined. By combining high-resolution transcriptome and polysome profiling of hepatocytes purified from quiescent and toxin-injured mouse livers, we uncover pervasive alterations in messenger RNA translation of metabolic and RNA-processing factors, which modulate the protein levels of a set of splicing regulators. Specifically, downregulation of the splicing regulator ESRP2 activates a neonatal alternative splicing program that rewires the Hippo signaling pathway in regenerating hepatocytes. We show that production of neonatal splice isoforms attenuates Hippo signaling, enables greater transcriptional activation of downstream target genes, and facilitates liver regeneration. We further demonstrate that ESRP2 deletion in mice causes excessive hepatocyte proliferation upon injury, whereas forced expression of ESRP2 inhibits proliferation by suppressing the expression of neonatal Hippo pathway isoforms. Thus, our findings reveal an alternative splicing axis that supports regeneration following chronic liver injury.

Alternative splicing contributes to gene expression dynamics in many tissues, yet its role in auditory development remains unclear. We performed whole-exome sequencing in individuals with sensorineural hearing loss (SNHL) and identified pathogenic mutations in Epithelial Splicing-Regulatory Protein 1 (ESRP1). Patient-derived induced pluripotent stem cells showed alternative splicing defects that were restored upon repair of an ESRP1 mutant allele. To determine how ESRP1 mutations cause hearing loss, we evaluated Esrp1-/- mouse embryos and uncovered alterations in cochlear morphogenesis, auditory hair cell differentiation, and cell fate specification. Transcriptome analysis revealed impaired expression and splicing of genes with essential roles in cochlea development and auditory function. Aberrant splicing of Fgfr2 blocked stria vascularis formation due to erroneous ligand usage, which was corrected by reducing Fgf9 gene dosage. These findings implicate mutations in ESRP1 as a cause of SNHL and demonstrate the complex interplay between alternative splicing, inner ear development, and auditory function.

Non-syndromic cleft lip with or without cleft palate (NS-CL/P) is one of the most common human birth defects and is generally considered a complex trait. Despite numerous loci identified by genome-wide association studies, the effect sizes of common variants are relatively small, with much of the presumed genetic contribution remaining elusive. We report exome-sequencing results in 209 people from 72 multi-affected families with pedigree structures consistent with autosomal-dominant inheritance and variable penetrance. Herein, pathogenic variants are described in four genes encoding components of the p120-catenin complex (CTNND1, PLEKHA7, PLEKHA5) and an epithelial splicing regulator (ESRP2), in addition to the known CL/P-associated gene, CDH1, which encodes E-cadherin. The findings were also validated in a second cohort of 497 people with NS-CL/P, comprising small families and singletons with pathogenic variants in these genes identified in 14% of multi-affected families and 2% of the replication cohort of smaller families. Enriched expression of each gene/protein in human and mouse embryonic oro-palatal epithelia, demonstration of functional impact of CTNND1 and ESRP2 variants, and recapitulation of the CL/P spectrum in Ctnnd1 knockout mice support a causative role in CL/P pathogenesis. These data show that primary defects in regulators of epithelial cell adhesion are the most significant contributors to NS-CL/P identified to date and that inherited and de novo single gene variants explain a substantial proportion of NS-CL/P.

Irf6 and Esrp1 are important for palate development across vertebrates. In zebrafish, we found that irf6 regulates the expression of esrp1 We detailed overlapping Irf6 and Esrp1/2 expression in mouse orofacial epithelium. In zebrafish, irf6 and esrp1/2 share expression in periderm, frontonasal ectoderm and oral epithelium. Genetic disruption of irf6 and esrp1/2 in zebrafish resulted in cleft of the anterior neurocranium. The esrp1/2 mutant also developed cleft of the mouth opening. Lineage tracing of cranial neural crest cells revealed that the cleft resulted not from migration defect, but from impaired chondrogenesis. Analysis of aberrant cells within the cleft revealed expression of sox10, col1a1 and irf6, and these cells were adjacent to krt4+ and krt5+ cells. Breeding of mouse Irf6; Esrp1; Esrp2 compound mutants suggested genetic interaction, as the triple homozygote and the Irf6; Esrp1 double homozygote were not observed. Further, Irf6 heterozygosity reduced Esrp1/2 cleft severity. These studies highlight the complementary analysis of Irf6 and Esrp1/2 in mouse and zebrafish, and identify a unique aberrant cell population in zebrafish expressing sox10, col1a1 and irf6 Future work characterizing this cell population will yield additional insight into cleft pathogenesis.

Abnormalities in ureteric bud (UB) branching morphogenesis lead to congenital anomalies of the kidney and reduced nephron numbers associated with chronic kidney disease (CKD) and hypertension. Previous studies showed that the epithelial fibroblast growth factor receptor 2 (Fgfr2) IIIb splice variant supports ureteric morphogenesis in response to ligands from the metanephric mesenchyme during renal organogenesis. The epithelial-specific splicing regulator Esrp1 is required for expression of Fgfr2-IIIb and other epithelial-specific splice variants. Our objective was to determine whether Esrp1 is required for normal kidney development.

Ultra-deep RNA sequencing has become a powerful approach for genome-wide analysis of pre-mRNA alternative splicing. We develop MATS (multivariate analysis of transcript splicing), a bayesian statistical framework for flexible hypothesis testing of differential alternative splicing patterns on RNA-Seq data. MATS uses a multivariate uniform prior to model the between-sample correlation in exon splicing patterns, and a Markov chain Monte Carlo (MCMC) method coupled with a simulation-based adaptive sampling procedure to calculate the P-value and false discovery rate (FDR) of differential alternative splicing. Importantly, the MATS approach is applicable to almost any type of null hypotheses of interest, providing the flexibility to identify differential alternative splicing events that match a given user-defined pattern. We evaluated the performance of MATS using simulated and real RNA-Seq data sets. In the RNA-Seq analysis of alternative splicing events regulated by the epithelial-specific splicing factor ESRP1, we obtained a high RT-PCR validation rate of 86% for differential exon skipping events with a MATS FDR of <10%. Additionally, over the full list of RT-PCR tested exons, the MATS FDR estimates matched well with the experimental validation rate. Our results demonstrate that MATS is an effective and flexible approach for detecting differential alternative splicing from RNA-Seq data.

Epithelial-mesenchymal transition (EMT) is an important developmental process hijacked by cancer cells for their dissemination. Here, we show that Exo70, a component of the exocyst complex, undergoes isoform switching mediated by ESRP1, a pre-mRNA splicing factor that regulates EMT. Expression of the epithelial isoform of Exo70 affects the levels of key EMT transcriptional regulators such as Snail and ZEB2 and is sufficient to drive the transition to epithelial phenotypes. Differential Exo70 isoform expression in human tumors correlates with cancer progression, and increased expression of the epithelial isoform of Exo70 inhibits tumor metastasis in mice. At the molecular level, the mesenchymal-but not the epithelial-isoform of Exo70 interacts with the Arp2/3 complex and stimulates actin polymerization for tumor invasion. Our findings provide a mechanism by which the exocyst function and actin dynamics are modulated for EMT and tumor invasion.

The epithelial-specific splicing regulators Esrp1 and Esrp2 are required for mammalian development, including establishment of epidermal barrier functions. However, the mechanisms by which Esrp ablation causes defects in epithelial barriers remain undefined. We determined that the ablation of Esrp1 and Esrp2 impairs epithelial tight junction (TJ) integrity through loss of the epithelial isoform of Rho GTP exchange factor Arhgef11. Arhgef11 is required for the maintenance of TJs via RhoA activation and myosin light chain (MLC) phosphorylation. Ablation or depletion of Esrp1/2 or Arhgef11 inhibits MLC phosphorylation and only the epithelial Arhgef11 isoform rescues MLC phosphorylation in Arhgef11 KO epithelial cells. Mesenchymal Arhgef11 transcripts contain a C-terminal exon that binds to PAK4 and inhibits RhoA activation byArhgef11. Deletion of the mesenchymal-specific Arhgef11 exon in Esrp1/2 KO epithelial cells using CRISPR/Cas9 restored TJ function, illustrating how splicing alterations can be mechanistically linked to disease phenotypes that result from impaired functions of splicing regulators.

Alternative splicing of fibroblast growth factor receptor-2 (FGFR2) mutually exclusive exons IIIb and IIIc results in highly cell-type-specific expression of functionally distinct receptors, FGFR2-IIIb and FGFR2-IIIc. We previously identified an RNA cis-element, ISE/ISS-3, that enhanced exon IIIb splicing and silenced exon IIIc splicing. Here, we have performed comprehensive mutational analysis to define critical sequence motifs within this element that independently either enhance splicing of upstream exons or repress splicing of downstream exons. Such analysis included use of a novel fluorescence-based splicing reporter assay that allowed quantitative determination of relative functional activity of ISE/ISS-3 mutants using flow cytometric analysis of live cells. We determined that specific sequences within this element that mediate splicing enhancement also mediate splicing repression, depending on their position relative to a regulated exon. Thus, factors that bind the element are likely to be coordinately involved in mediating both aspects of splicing regulation. Exon IIIc silencing is dependent upon a suboptimal branchpoint sequence containing a guanine branchpoint nucleotide. Previous studies of exon IIIc splicing in HeLa nuclear extracts demonstrated that this guanine branchsite primarily impaired the second step of splicing suggesting that ISE/ISS-3 may block exon IIIc inclusion at this step. However, results presented here that include use of newly developed in vitro splicing assays of FGFR2 using extracts from a cell line expressing FGFR2-IIIb strongly suggest that cell-type-specific silencing of exon IIIc occurs at or prior to the first step of splicing.

Control over gene expression is exerted, in multiple stages of spermatogenesis, at the post-transcriptional level by RNA binding proteins (RBPs). We identify here an essential role in mammalian spermatogenesis and male fertility for 'RNA binding protein 46' (RBM46). A highly evolutionarily conserved gene, Rbm46 is also essential for fertility in both flies and fish. We found Rbm46 expression was restricted to the mouse germline, detectable in males in the cytoplasm of premeiotic spermatogonia and meiotic spermatocytes. To define its requirement for spermatogenesis, we generated Rbm46 knockout (KO, Rbm46-/-) mice; although male Rbm46-/- mice were viable and appeared grossly normal, they were infertile. Testes from adult Rbm46-/- mice were small, with seminiferous tubules containing only Sertoli cells and few undifferentiated spermatogonia. Using genome-wide unbiased high throughput assays RNA-seq and 'enhanced crosslinking immunoprecipitation' coupled with RNA-seq (eCLIP-seq), we discovered RBM46 could bind, via a U-rich conserved consensus sequence, to a cohort of mRNAs encoding proteins required for completion of differentiation and subsequent meiotic initiation. In summary, our studies support an essential role for RBM46 in regulating target mRNAs during spermatogonia differentiation prior to the commitment to meiosis in mice.

Although major genetic networks controlling early liver specification and morphogenesis are known, the mechanisms responsible for postnatal hepatic maturation are poorly understood. Here we employ global analyses of the mouse liver transcriptome to demonstrate that postnatal remodelling of the liver is accompanied by large-scale transcriptional and post-transcriptional transitions that are cell-type-specific and temporally coordinated. Combining detailed expression analyses with gain- and loss-of-function studies, we identify epithelial splicing regulatory protein 2 (ESRP2) as a conserved regulatory factor that controls the neonatal-to-adult switch of ∼20% of splice isoforms in mouse and human hepatocytes. The normal shift in splicing coincides tightly with dramatic postnatal induction of ESRP2 in hepatocytes. We further demonstrate that forced expression of ESRP2 in immature mouse and human hepatocytes is sufficient to drive a reciprocal shift in splicing and causes various physiological abnormalities. These findings define a direct role for ESRP2 in the generation of conserved repertoires of adult splice isoforms that facilitate terminal differentiation and maturation of hepatocytes.

Tissue- and cell-type-specific regulators of alternative splicing (AS) are essential components of posttranscriptional gene regulation, necessary for normal cellular function, patterning, and development. Mice with ablation of Epithelial splicing regulatory protein (Esrp1) develop cleft lip and palate. Loss of both Esrp1 and its paralog Esrp2 results in widespread developmental defects with broad implications to human disease. Deletion of the Esrps in the epidermis revealed their requirement for establishing a proper skin barrier, a primary function of epithelial cells comprising the epidermis. We profiled the global Esrp-mediated splicing regulatory program in epidermis, which revealed large-scale programs of epithelial cell-type-specific splicing required for epithelial cell functions. These mice represent a valuable model for evaluating the essential role for AS in development and function of epithelial cells, which play essential roles in tissue homeostasis in numerous organs, and provide a genetic tool to evaluate important functional properties of epithelial-specific splice variants in vivo.

The epithelial splicing regulatory proteins, ESRP1 and ESRP2, are essential for mammalian development through the regulation of a global program of alternative splicing of genes involved in the maintenance of epithelial cell function. To further inform our understanding of the molecular functions of ESRP1, we performed enhanced crosslinking immunoprecipitation coupled with high-throughput sequencing (eCLIP) in epithelial cells of mouse epidermis. The genome-wide binding sites of ESRP1 were integrated with RNA-Seq analysis of alterations in splicing and total gene expression that result from epidermal ablation of Esrp1 and Esrp2. These studies demonstrated that ESRP1 functions in splicing regulation occur primarily through direct binding in a position-dependent manner to promote either exon inclusion or skipping. In addition, we also identified widespread binding of ESRP1 in 3' and 5' untranslated regions (UTRs) of genes involved in epithelial cell function, suggesting that its post-transcriptional functions extend beyond splicing regulation.

Epithelial-Splicing-Regulatory-Protein 1 (Esrp1) is a cell-type specific RNA-binding protein (RBP) that is essential for mammalian development through maintenance of epithelial cell properties including barrier function. Esrp1 also regulates splicing during the epithelial to mesenchymal transition (EMT). It contains three highly conserved RNA recognition motifs (RRMs) in the absence of other clearly defined protein domains. Esrp1 itself is also alternatively spliced to produce multiple protein isoforms. Here we determined that two competing alternative 5' splice sites in exon 12 yield Esrp1 isoforms with differential nucleocytoplasmic localization. We carried out a detailed characterization of the Esrp1 peptide that is sufficient to confer nuclear localization. Furthermore, we identified splice variants encoding distinct nuclear and cytoplasmic isoforms of fusilli, the D. Melanogaster Esrp1 ortholog. Our observations demonstrate that the production of both nuclear and cytoplasmic Esrp1 isoforms through alternative splicing is phylogenetically conserved; strongly suggesting it is biologically significant. Thus, while previous studies have described extensive regulation by nuclear Esrp1 to promote epithelial specific splicing, it will be of great interest to study the contribution of cytoplasmic Esrp1 in maintenance of epithelial cell functions.