关于真核生物关于真核生物mRNA降解及代谢过程的计降解及代谢过程的计算机模拟算机模拟曹丹亚力桑那大学分子细胞生物学系 DanCaoandRoyParkerHowardHughesMedicalInstitute&DepartmentofMolecularCellularBiologyUniversityofArizonaComputational modeling of mRNA turnoverDNAmRNAProteinTranscription,Splicing,transporttranslationdegradationdegradationThe Central DogmaWhy want to model this process?1. IntegrationWhy want to model this process(Contd)? 2.An explanatory/descriptive tool1)correctionofmisconceptions2)helpinterpretresultcorrectly3)advicesuitableexperiment3.Predictionfromdiscrepanciesbetweensimulationresultandexperimentalresult4.An ideal system to start knowledge-driven simulationPlentyofdataavailabletoestimatetheratesfordistinctsteps,includingsteadystatedistribution,half-lives,effectsofvariousmutantsMajor assumptions1.Transcription is a zeroth order process, all other steps are first order processes2.All steps are irreversible3.No feedback loopsrates for all steps Steady state level of each intermediatekinetics over certainexperimental time courseVirtual northern gelfor direct comparisonTranscriptionalshut-offexp.(decayfromsteadystate)Transcriptionalpulsechaseexp.AtsteadystateInhibittranscriptionAttheendofashort“pulse”InhibittranscriptionA sample screenshotinputDetermine the fitness of the model with the in vivo decay network: MFA2pG & PGK1pGTheyrepresentaunstable(MFA2)andstable(PGK1)mRNAinyeastTheirdegradationhasbeenextensivelycharacterized,withlotsofdataavailabletoestimatetherates.E.g:transcriptionalpulsechasegelcangiveinformationfortheratesofdeadenylationanddecapping.Strongpoly(G)structureatthe3UTRcantrapthedecayintermediates,giveadditionalinformationaboutin vivoprocess.E.g:ratesofterminaldeadenylationand3to5decayComparison of modeling results with experimental observationsThe simulated steady state poly (A) distribution, pulse chase gel pattern,previously characterized mutants (transport, decapping, 3 to 5 decay) are also consistent with exp. observations.Modeling of MFA2pG in transcriptional pulse chaseObservedComputedModeling of PGK1pG in transcriptional pulse chaseObservedComputedThe fact that we can reproduce the experimental results by modeling suggests that our model is quite accurate, and we have a relatively robust understanding of the in vivo process.The obtaining of a good model for both MFA2pG and PGK1pG allows us to further analyze the whole network.We have used our model to examine how transcripts respond to a variety of perturbations by performing a series of in silico experiments. E.g.: increase or decrease the rate of transport, deadenylation, decapping, 5 to 3 exonucleolytic decay, 3 to 5 exonucleolytic decay, see how the transcript level, half-life, steady state distribution are affected. What have we learned?Comparison of in silico mutantsDeadenylation is a key step in controlling mRNA turnover.Provide explanation for why many decay elements identified affecting deadenylation. 3to5decayrateforfulllength.Obtainedbymatchingthecalculatedtwiththeobservedtindcp1mutant.The calculated 3 to 5 decay rates Implication: 3 to 5 decay by exosome shows mRNA specific degradation rates that are dependent on the 5 structure of the mRNA3to5decayrateforfragment.Obtainedbymatchingthecalculatedtoffragmentwiththeobservedtwhendecappingisblocked.Half-life.Measuredbytranscriptionalshut-offexperiment.Atthepointwheretranscriptreacheshalfofitsinitiallevel.tisthoughttorepresenthowlongthemRNAispersistinthecell.Averagelifespan.Calculatedfromthesimulationfortranscriptionalpulsechaseexperiment.MoreaccuraterepresentationoftheaveragetimethemRNAispresentinthecell.View on half-lifeHalf-life Average life span. The traditional way of measuring t1/2 may underestimate the life span of an mRNA. The difference is due to the distribution of mRNA at steady state. Certain % of mRNAs has already passed several decay steps.Why Average life span half-life?Decay from steady statet 1/2Transcriptional pulse chaseAverage Life Span The measurement of measure a half-life can predominantly different steps in the decay networkDifferent mRNAs will be affected differentially by certain change on specific stepNeed to be very cautious when interpreting mRNA specific effects.Short-livedmRNAs(MFA2)aremoreresponsivetochangesontransportratethanlong-livedmRNAs(PGK1).Possible factors that disrupt the correlation between mRNA and protein levelThe increase of transcript level in the transport mutant solelycomes from the increase in the nuclear pool.The increase of transcript level in the 5 to 3 exo mutants solely comes from the increase in the cap- speciesWeareabletosimulateMFA2andPGK1,whichsuggeststhatwehavearelativelyrobustunderstandingaboutyeastmRNAturnover.ThisprogramcanbeadaptabletoothereukaryoticmRNAsthatfollowthesamedegradationscheme.Thisisauseful,explanatorytoolforquantitativeanalysisoftheprocessandregulationofmRNAturnoverineukaryoticcells. Some In silicoexperimentsperformedmightbeabletosuggestthebestexp.foraparticularpurpose.E.g:decappingmutants.Discrepanciesbetweenin silico resultsandrealresultsmightleadtonewinsightsforthein vivonetwork.ComputationExperimentationNonsenseMediatedmRNADecay(NMD)AAAAAAA70DNAtranscriptionAUG UAA UAAm7GpppNormaldecay30NonsenseDecay3Normal Decay and NMDModelingPredictionExp.testingKnowledgeandhypothesisbasedmodeling.Mighthavemultiplemodels:model1,2,3n.Allmodelsshouldfitwithcurrentdata.Makepredicationsbyanalyzingthemodelsandperformingin silicoexperiments.Designandconductcriticalexperimentstotestimportantpredictionsanddistinguishthemodels.Somemodelsmaygetinvalidated.Othersmayneedtoberefinedtofitwellwithnewexperimentaldata.Mayleadtonewhypotheses.ThemostfaithfulmodelofNMDandnewinsightsaboutthisprocess.Illustrationoftheiterativeapproach.Advantage:increasetherateofhypothesisformingandtestingModel1.1Model1.2Multiple models or different combinations of rate constants within the same model can be obtained to fit with current observations.Model 1Model 1.1 and Model 1.2Model1.1TheentryintoPASisinhibitedbymorethan100fold,therateofDIDcontributestothet1/2ofA70NuclearmRNATransport(slow,52folddown,3mint1/2)CytoplasmicmRNA(A70)PASDID(fast)NormaldecayNMDModel1.2TheentryintoPASisnotinhibited,therateoftransport(ormaybeotherunknownstepsupstreamofthebifurcation)issloweddown,whichisusuallyveryfastfornormaltranscript.NuclearmRNATransport(fast)CytoplasmicmRNA(A70)PASDID(slow,3mint1/2)NormaldecayNMDIfentryintoPASisnotinhibited,willseesamedecreaseonsteadystatelevel,butthehalf-lifedoesnotchangemuch.Distinguish Model 1.1 and Model 1.2RelativelevelofnucleartranscriptModel1.1predicts2%nuclearModel1.2predicts90%nuclearAssumption:therate-limitingstepabovethebifurcationisindeedtransportDecappingmutantsdcp1,dcp2Current data is consistent with model 1.2, in which the entry into PAS is not inhibited, the rate of DID is much higher than PAS, and there might be a rate limiting step upstream of the bifurcationin silico biologyComputationalmodelingofcellularnetworks,signalingpathways,metabolicpathwayspathways,cells,tissuesanddiseases.in vivo,in vitro,in silicoItisarelativelynewerbranchofbioinformatics,attractinggreatattentioninpost-genomicera.Examples:Academics:Virtualcell(Ecell),Evirus,DrosophilalegdevelopmentCompanies:Entelos,Physiomesciences,pharmaceuticalcompaniesworkingonhumandiseasemodeling.providesguidancefortargetselection.improvedtargetprioritizationcomparedwithrelyingonempiricalresearchaloneRequireverydeepunderstandingofbiology,xlntmathandcomputerskills,workcloselywithbiologist.Come on, lets work on it!AcknowledgementsAcknowledgementsQuestions?Thank you!