Alternative Splicing Regulatory Networks Funct.pdf

Molecular CellReviewAlternative Splicing Regulatory Networks:Functions,Mechanisms,and EvolutionJernej Ule1,2,*and Benjamin J.Blencowe3,4,*1The Francis Crick Institute,London NW1 1AT,UK2Department of Neuromuscular Diseases,UCL Queen Square Institute of Neurology,Queen Square,London WC1N 3BG,UK3Donnelly Centre,University of Toronto,Toronto,ON M5S 3E1,Canada4Department of Molecular Genetics,University of Toronto,Toronto,ON,M5S 3E1,Canada*Correspondence:jernej.ulecrick.ac.uk(J.U.),b.blencoweutoronto.ca(B.J.B.)https:/doi.org/10.1016/j.molcel.2019.09.017High-throughput sequencing-based methods and their applications in the study of transcriptomes haverevolutionized our understanding of alternative splicing.Networks of functionally coordinated and biologi-cally important alternative splicing events continue to be discovered in an ever-increasing diversity of celltypes in the context of physiologically normal and disease states.These studies have been complementedby efforts directed at defining sequence codes governing splicing and their cognate trans-acting factors,which have illuminated important combinatorial principles of regulation.Additional studies have revealedcritical roles of position-dependent,multivalent protein-RNA interactions that direct splicing outcomes.In-vestigations of evolutionary changes in RNA binding proteins,splice variants,and associated cis elementshavefurther shed lighton theemergence,mechanisms,andfunctionsofsplicing networks.Progressintheseareas has emphasized the need for a coordinated,community-based effort to systematically address thefunctions of individual splice variants associated with normal and disease biology.Transcripts from nearly all human protein-coding genes undergoone or more forms of alternative splicing,such as inclusion orskipping of individual cassette exons,switching betweenalternative 50and 30splice sites,differential retention of introns,mutually exclusive splicing of adjacent exons,and other,morecomplex patterns of splice site selection(Pan et al.,2008;Wang et al.,2008).All of these forms of splicing require thespliceosome,a megadalton machine that catalyzes splicing re-actions(Wahletal.,2009).Spliceosomeformation entailsacom-plex interplay of trans-acting factors,including small nuclearribonucleoprotein particles(snRNPs,U1,U2,U4/U6,and U5)comprising small nuclear RNAs(snRNAs)and associated pro-teins,together with?150 additional proteins.The formation ofspliceosomes and their mechanism of action has been illumi-nated in remarkable detail by a series of recent cryoelectronmicroscopy structures,work that has been reviewed elsewhere(Kastner et al.,2019;Plaschka et al.,2019;Yan et al.,2019).Binding of snRNPs to pre-mRNA is typically stabilized bymutual definition interactions across introns and exons(DeConti et al.,2013;Figure 1).Intron definition interactions pre-dominate when introns are relatively short(e.g.,in the range ofup to a few hundred nucleotides),as is the case in yeast andmost invertebrate species.In contrast,exon definition interac-tions predominate in vertebrates(Robberson et al.,1990),whereintrons have a median length of approximately 1 kb(Hong et al.,2006).In either case,the principles governing splice site recog-nition and pairing are thought to be similar.For example,currentmodels posit that U1 snRNP binds to the 50splice site and com-municates via bridging interactions with splicing factor 1(SF1)and the U2 snRNP auxiliary factor(a heterodimer of U2AF1and U2AF2)bound to the 30splice site and its adjacent polypyr-imidine tract(Abovich and Rosbash,1997;De Conti et al.,2013).Additional interactions that contribute to exon and intron defini-tion are mediated by members of the RNA recognition motif(RRM)-containing SR family of proteins(referred to below asSR proteins)and SR-related proteins,each of which containsone or more intrinsically disordered region(IDR)rich in alter-nating arginine and serine residues,referred to as the RSdomain(Figure 1).For example,it has been proposed thatwhen SR proteins bind to exonic enhancer sequences,theirRS domains interact with the RS domains of the U1 snRNP-spe-cific 70-kDa protein(SNRNP70)and U2AF1 to promote exondefinition(Wu and Maniatis,1993).In S.pombe,it has beenshown that intron definition is promoted by interactions betweentheRSdomainsofRsd1andPrp5(orthologuesofhumanRBM38and DDX46),which interact with U1 and U2 snRNP,respectively(Shao et al.,2012).Numerous additional interactions come into play to forgeintron and exon definition interactions.For example,the SR-related proteins SRRM1 and SRRM2 can bridge interactionsbetween snRNPs bound at splice sites and SR proteins boundat exonic enhancers(Eldridge et al.,1999).Moreover,phos-phorylated RS domains have been reported to bind double-stranded RNA,which can promote base-pairing betweensnRNPs and pre-mRNA(Shen and Green,2006).Collectively,these and additional early interactions,some of which aredescribed later,facilitate the stable recruitment of U2 snRNPto the pre-mRNA branch site,followed by addition of U4/U6and U5 snRNPs in the form of a tri-snRNP particle.The actionsof many RNA helicases then promote rearrangements ofsnRNPinteractions and establishment of acatalyticallycompe-tentspliceosomethatcarriesoutthetwotrans-esterificationre-actions that lead to lariat formation,intron removal,and exonligation(Wahl et al.,2009).Molecular Cell 76,October 17,2019 Crown Copyright 2019 Published by Elsevier Inc.329ManytypesofRNA-bindingproteins(RBPs)canregulatealter-native splicing.In addition to SR proteins,these include the het-erogeneous ribonucleoprotein(hnRNP)family of proteins as wellas RBPs containing RRM,K homology domain(KH),zinc-finger,or other domains(Lunde et al.,2007).The full set of proteins thatcontrol alternative splicing is not known,although recent large-scale screens employing systematic RNAi or CRISPR-Cas-mediated ablation of genes have revealed repertoires involvinga few hundred proteins that act directly or indirectly to regulatespecific alternative exons(Gonatopoulos-Pournatzis et al.,2018;Han et al.,2017;Papasaikas et al.,2015;Tejedor et al.,2015).Among other unexpected factors,these studies havehighlighted previously annotated DNA-binding proteins as hav-ing potential direct roles in RNA binding and splicing regulation.RBPs bind cis elements in introns and exons and regulate splicesite selection largely by promoting or repressing definition inter-actions(De Conti et al.,2013;Fu and Ares,2014).Thus,theymainlyactattheearlystagesofspliceosomeformation,althoughregulation can also be imparted at later stages of assembly(Wahl et al.,2009).In this review,we highlight recent advances in the identifica-tion and characterization of networks of splicing regulation,including significant strides that have been made in the system-atic analysis of RBPs and associated regulatory mechanismsthrough application of in vitro binding(Dominguez et al.,2018)and invivo cross-linking andimmunoprecipitation(CLIP)methods(Lee and Ule,2018;Van Nostrand et al.,2018),prote-omics(Hentze et al.,2018),functional genomics(Gonatopou-los-Pournatzis et al.,2018),and increasingly powerful computa-tional approaches(Baeza-Centurion et al.,2019;Jaganathanet al.,2019).We review how these and other complementary ap-proaches are further providing unprecedented new insights intothe evolution of mechanisms governing alternative splicing aswell as how disruption of these mechanisms causes or contrib-utes to human diseases and disorders.Finally,we conclude bydiscussing challenges for the field that lie ahead.Biological Significance of Alternative SplicingRegulatory NetworksThe development and application of custom microarrays and,later,high-throughput RNA sequencing(RNA-seq)methods,re-vealed the extraordinary complexity of regulated alternativesplicing in metazoans,particularly in vertebrate species(Blen-cowe,2015;FuandAres,2014).Recenttranscriptomesequencing efforts involving both short-and long-read technol-ogies are increasingly focusing on specialized cell types andA(eg.GAA)Y.YNYAGYURAYGGURAGcis-actingmotifs:exonintrontrans-acting factors:GGURAGexon definitioninteractionsU1snRNPU1snRNPSR proteins(eg.SRSF1)RSU2AFRSRSRSRSRSintron definitioninteractionsSF1BU2 snRNPSF3cis-acting motifs:Y.Y UCUC.UGC YAGYURAYGGURAGmicroexondefinitioninteractionsSRRM4SRSF11RNPS1U2AF2U2AF1U1 snRNPU2 snRNPRSeMICRSRSRSRSSF1Figure 1.Exon and Intron Definition Interactions(A)Schematic of spliceosomal components and regulatory proteins that participate in exon and intron definition and interactions between them.Trans-actingsplicing factors are shown as blue shapes,and their names are shown next to the shape.The RS domain is marked by RS.Blue arrows denote intron or exondefinition interactions,many of which are mediated by the RS domain.Exons are represented as gray boxes,intronic RNA and snRNAs as gray lines,andcis-actingmotifsascolored lines,withtheconsensussequencesofthesemotifsshown underneath.Thepairing ofU1snRNAwiththe50splicesiteisindicated byblack lines.(B)Schematic of microexon definition,shown in the same manner as described in(A).330Molecular Cell 76,October 17,2019Molecular CellReviewindividual cells from different organs.To date,dynamic alterna-tive splicing networks have been detected in embryonic stemand precursor cells,during differentiation or reprogramming ofvarious cell lineages as well as epithelial-mesenchymal transi-tions,and in adult organs such as the brain,heart,skeletal mus-cle,liver,kidney,adipose tissue,and testes and in the immunesystem(Baralle and Giudice,2017;Bhate et al.,2015;Gabutet al.,2011;Han et al.,2013;Irimia et al.,2014;Kalsotra andCooper,2011;Licatalosi and Darnell,2010;Mallory et al.,2015;Tapial et al.,2017;Wong et al.,2013;Zhang et al.,2016).Additional regulated alternative splicing networks havebeen detected in association with specific physiologic states ofcells,such as thermal regulation and the stress response(Boutzet al.,2015;Gotic et al.,2016;Low et al.,2008;Preuner et al.,2017).Many regulatory RBPs function in a cell-,tissue-,or con-dition-specific manner and are capable of coordinately regu-lating functionally coherent networks of exons and introns(Braunschweig et al.,2013;Licatalosi and Darnell,2010).Thus,our understanding of the repertoires of detected splice variantsas well as other forms of transcript variation across cellularconditions in the context of normal and disease physiologycontinues to rise dramatically.Notably,regulated alternative exons that overlap protein-cod-ing sequences are often located within predicted IDRs that arecoincident with sites of post-translational modifications and pro-tein-protein interactions,and the role of alternative splicing indiversifying protein interaction capabilities has been demon-strated experimentally(Buljan et al.,2012;Ellis et al.,2012;Yang et al.,2016).An important and likely general function ofalternative splicing networks is therefore to control protein-pro-tein interactions to impart important cell-,tissue-,and condi-tion-specific functions of widely expressed genes.In additionto remodeling the IDRs,a smaller number of conserved eventsin alternative splicing networks directly overlap critical modularprotein domains and affect various important protein functions,such as those involved in control of transcription and chromatin,as reviewed elsewhere(Irimia and Blencowe,2012;Kelemenet al.,2013;Porter et al.,2018).The identification and character-ization of such events highlights the capacity of alternativesplicing networks to have a broad effect on physiology throughtheir ability to cross-talk with orthogonal gene regulatory layers.A Neuronal Microexon NetworkComparative vertebrate transcriptomics of a common set of or-gans from fish to human revealed that brain-specific alternativesplicing events are among the most evolutionarily conserved(Barbosa-Morais et al.,2012;Merkin et al.,2012).Particularlystriking in this regard is a network of a few hundred 3-to 27-ntneuronal microexons that represent the most highly conservedclass of alternative splicing events discovered to date(Irimiaet al.,2014;Li et al.,2015).These microexons predominantlypreserve open reading frames and insert one to several aminoacids in proteins associated with neurogenesis,axon guidance,and synaptic functions(Irimia et al.,2014;Johnson et al.,2019;Quesnel-Vallie res et al.,2019;Ustianenko et al.,2017).Similarto longer exons,individual neuronal microexons can also affectorthogonal regulatory layers;for example,they alter the functionof the lysine-specific histone demethylase 1A(KDM1A,alsoknown as Lsd1),control activation domains of Mef2 family tran-scriptionfactors,andregulatetheactivityofthecytoplasmicpol-yadenylation element binding protein 4(CPEB4),which controlspoly(A)tail length and translation(Ebert and Greenberg,2013;Parras et al.,2018;Rusconi et al.,2017).Notably,misregulationofactivity-dependentsplicingofmicro-exons likely plays a causative role in autism,at least in part as aconsequence of disrupted expression of the major trans-actingregulator of microexons,the neuronal SR-related proteinnSR100/SRRM4(Irimia et al.,2014;Quesnel-Vallie res et al.,2015,2016).SRRM4 activates splicing of microexons by bindingspecialized upstream intronic enhancer elements together withthe SR proteins SRSF11 and RNPS1(Gonatopoulos-Pournatziset al.,2018;Raj et al.,2014;Figure 1B).The C-terminal IDR ofSRRM4 contains an enhancer of microexons(eMIC)domain,which interacts with the branchpoint-binding protein SF1 andU2AF to promote recruitment of U2 snRNP and,thus,activatesthe earliest stages of splicing complex formation(Torres-Me n-dez et al.,2019).The neuronal inclusion of microexons can befurther modulated by additional proteins;for example,theneuronal enriched Nova proteins repress microexons in Robo1and Robo2 genes during the later stages of neuronal develop-ment to control navigation of commissural axons(Johnsonet al.,2019).In contrast to longer neurally regulated exons thatare enriched within IDRs,as described above,neural microex-ons significantly overlap or are adjacent to modular domainsthat function in mediating protein and other ligand interactions(Irimia et al.,2014).Many microexon-regulated proteins areknown to form physical interactions with one another,suggest-ing that the coordinated inclusion of dozens of microexons likelyremodels large protein interaction networks in neurons that aredisrupted in autism.Mechanisms Underlying the Regulation of AlternativeSplicing NetworksMultivalency and RNP Condensation in SplicingRegulationKey to a more general understanding of the function and regu-lation of alternative splicing networks is the systematic identifi-cation of corresponding splicing regulators and their cognatecis-acting binding sites.Most RNA-binding domains of RPBsrecognize short(i.e.,34 nt)and degenerate motifs,and thecapacity of such individual motifs to predict splicing outcomesis low.Instead,mutagenesis of minigene reporters as well astranscriptomics studi