1、Combinatorial function of transcription factors andcofactorsFranziska Reiter*,Sebastian Wienerroither*and Alexander StarkDifferential gene expression gives rise to the many cell types ofcomplex organisms.Enhancers regulate transcription bybinding transcription factors(TFs),which in turn recruitcofac
2、tors to activate RNA Polymerase II at core promoters.Transcriptional regulation is typically mediated by distinctcombinations of TFs,enabling a relatively small number of TFsto generate a large diversity of cell types.However,how TFsachieve combinatorial enhancer control and how enhancers,enhancer-b
3、ound TFs,and the cofactors they recruit regulateRNA Polymerase II activity is not entirely clear.Here,we reviewhow TF synergy is mediated at the level of DNA binding andafter binding,the role of cofactors and the post-translationalmodifications they catalyze,and discuss different models ofenhancerco
4、re-promoter communication.AddressResearch Institute of Molecular Pathology(IMP),Vienna Biocenter(VBC),Campus-Vienna-Biocenter 1,1030 Vienna,AustriaCorresponding author:Stark,Alexander(starkstarklab.org)*These authors contributed equally to this work.Current Opinion in Genetics&Development 2017,43:73
5、81This review comes from a themed issue on Genome architecture andexpressionEdited by Bart Deplancke and Charles SagerstromFor a complete overview see the Issue and the EditorialAvailable online 19th January 2017http:/dx.doi.org/10.1016/j.gde.2016.12.0070959-437X/2017 Elsevier Ltd.All rights reserve
6、d.IntroductionGene regulation is central to development and cellulardifferentiation 1,2,and erroneous gene expression islinked to many diseases including cancer 3,4.Generegulatory information is encoded in the DNA sequencesof genomic cis-regulatory elements called enhancers 5,which activate or repre
7、ss transcription from their targetgenes core-promoters 2.Different transcription factors(TFs)bind to short recognition sites within enhancers thus essentially reading the regulatory information con-tained in the enhancer sequence and recruit cofactors(COFs),such as the Mediator complex or the acetyl
8、-transferase CBP/p300.Together,these regulatory pro-teins mediate RNA polymerase II(Pol II)recruitmentand activation at core-promoters(Figure 1;reviewed inRefs.2,6).Examples from cellular reprogramming(e.g.,7,8)oranimal development(e.g.,9,10,11?,12)suggest that generegulation and cell type determina
9、tion typically dependon multiple TFs that function in a combinatorial manner.In the early Drosophila embryo for example,the even-skipped(eve)muscle and heart enhancer(MHE)exem-plifies how different developmental signals are integratedwith cell type-specific gene regulation by TF coopera-tivity.The M
10、HE is activated by the Wnt,Dpp/TGF-band RTK pathways,yet its activity relies on additionalinput from the mesodermal TFs Twist and Tinman 12.Similarly,in Ciona intestinalis,the Orthodenticle homeobox(Otx)-a enhancer integrates FGF-signaling with GATA-factor activities to achieve specific activation i
11、n the ante-rior neural plate and the dorsal nerve chord 11?.Thisdependency of signaling pathway TFs on combinatorialinput from partner TFs has been termed activatorinsufficiency 13 and is a common property of signalingpathways that allows their employment in different cel-lular contexts in which the
12、y typically regulate differenttarget genes via distinct partner TFs.The combinatorial function of different TFs is typicallyreflected in the enhancers,which contain short bindingsite sequences for the respective TFs.Many studiesdemonstrated the importance of these binding sites forenhancer function,
13、including for example the eve stripe2 enhancer 9 or sparkling 10 in Drosophila,the Otx-aenhancer in Ciona 11?,or the human interferon-betaenhancer 14.Typically,sites for several different TFsare required and none of the individual TFs are sufficient,highlighting the importance of combinatorial regul
14、ation15.However,how TFs function and how their com-bined regulatory cues might be integrated at enhancers tosynergistically activate transcription has remained a fas-cinating open question that we discuss in this review.TF cooperativity at the level of DNA bindingMultiple lines of evidence suggest t
15、hat cooperativitybetween TFs can be established at the level of DNAbinding.For example,TF binding between differentspecies or between individuals within one species isinfluenced not only by mutations in the TFs own recog-nition sequences(also called TF motif matches or instances)but also in those of
16、 partner TFs 1619(recentlyreviewed in Ref.20).Furthermore,experimentally dis-rupting the recognition sequences of some TFs orAvailable online at ScienceD Current Opinion in Genetics&Development 2017,43:7381depleting the corresponding TF proteins can cause loss ofbinding of other TFs 2124,25?,26?.Giv
17、en that TFs recognize consensus motifs that aretypically only 48 bp long,and therefore occur with highfrequency in long DNA sequences simply by chance,combinatorial TF binding not only enables higher speci-ficity in binding,but also the cell-type-specific redirectionof TFs to different binding sites
18、 in different cell types.This becomes possible if the recognition sequences of agiven TF at different sites in the genome occur with therecognition sequences of different partner TFs,thusenabling binding only in cell types in which the respec-tive partners are present.Such differential TF bindinghas
19、 indeed been seen for example for Twist,which bindsto sites co-bound by Zelda during early Drosophilaembryogenesis,but to different sites co-bound by differ-ent partners at later stages 22,27.Similarly,in themammalian hematopoietic lineage,the TF PU.1 forexample functions in either B-cell or macroph
20、age devel-opment depending on its partner TFs 21,23.Several mechanisms can explain combinatorial binding ofTFs in the context of chromatin.Chromatin at inactiveenhancers is typically closed such that the TF bindingsites are occupied by nucleosomes,which can function asgatekeepers for TF binding(see
21、e.g.,Ref.23).Whileindividual TFs might not be able to effectively competewith nucleosomes for DNA binding 28,29,multiple TFsthat recognize closely spaced binding sites within enhan-cers might together evict nucleosomes by mass actionthus enabling cooperative binding.Such a passive form ofcooperativi
22、ty,recently called collaborative binding 20,would not require additional TF or cofactor functions nordirect proteinprotein-interactions(PPIs)between theTFs,and could rely solely on the individual TFsDNA affinities(Figure 2a)30,31.This mechanism isconsistent with a billboard model for enhancers 32,wh
23、ich highlights the apparent flexibility of recognitionsite-arrangements between different enhancers withinone species and across orthologous enhancers(reviewedin 33).Nucleosome competition by the collaborativebinding of different TFs is further consistent with theenrichment of cell-type-specific TF
24、binding sequences atsites of accessible DNA(e.g.,34)and could also explainhow accessible sites are re-established after replication35?or maintained during mitosis by TFs bound tomitotic chromosomes 36?.Cooperativity during binding can be enhanced by directPPIs between TFs.While such interactions rel
25、y ondefined interaction interfaces and compatible motif spac-ing,they can confer high specificity and DNA affinity(Figure 2a).Many examples of homodimeric or hetero-dimeric TF binding are known,including Fos that bindsto DNA as a homodimer or a Jun/Fos heterodimer 37.Interestingly,for some TF pairs,
26、the combined bindingpreference is composed of the two individual TF motifs,while other pairs have binding preferences different fromthe ones of the individual TFs 38?(for a review,seeInukai,Kock and Bulyk,in this issue).PPIs that favor co-binding or stabilize TFs at activeenhancers are not restricte
27、d to TF dimers,but can includeinteractions between TFs and COFs or higher orderprotein complexes(for a review,see Ref.6).An extremeexample for cooperative binding is the interferon-betaenhanceosome,which contains binding sites for severalTFs(Figure 2a).These TFs,the architectural proteinHMG I(Y)and
28、the co-activator p300/CBP create stronglysynergistic activating cues that critically rely on the fixedarrangement of the TF binding sites and potentiallyinclude the formation of a defined multi-protein complex14,39.The discussed mechanisms are not mutually exclusivebut can all occur at different enh
29、ancers(e.g.,billboard-type developmental enhancers and the enhanceosome)oreven within a single enhancer(e.g.,collaborative bindingof individual TFs plus cooperative binding of interactingTF pairs;Figure 2a).The comparison of enhancersequences between closely related species however sug-gest that bin
30、ding site-arrangements in enhancers aretypically flexible and that rigid enhanceosome-likeenhancer architectures are rare(reviewed in 33).A different model for TF binding assumes sequentialrather than simultaneous binding of specialized pioneerTFs(Figure 2b)such as FoxA or PU.1 in mammals 23 orZelda
31、 in Drosophila 24,25?,26?,40.According to theiroriginal definition,pioneer TFs possess distinct biochem-ical properties that enable binding to nucleosomal DNAwithin closed chromatin and facilitate the subsequentbinding of additional TFs(reviewed in Ref.41).Inaddition to FoxA in liver or PU.1 in the
32、hematopoieticlineage,Sox2 for example enables subsequent Oct474 Genome architecture and expressionFigure 1enhancerTFTFTFCOFsPol IIcore-promoterCurrent Opinion in Genetics&DevelopmentTranscriptional regulation and its main players.Enhancers contain short sequence motifs that can be recognized bytrans
33、cription factors(TFs).TFs in turn recruit transcriptional cofactors(COFs),which recruit and activate RNA polymerase II(Pol II)at core-promoters(short sequences surrounding the transcriptional start site)to enable transcription.Current Opinion in Genetics&Development 2017,43:7381 binding in ES cells
34、42,and Oct4,Sox2,and Klf4 enablebinding of cMyc during reprogramming 43.In the earlyfly embryo,the maternally deposited TF Zelda 40facilitates TF binding and early gene expression,andtherefore plays a central role in early zygotic genomeactivation 22,24,25?,26?.We find it interesting to spec-ulate t
35、o which extent distinct biochemical properties orrather these TFs nuclear concentrations and the result-ing mass action(see above)play a role in nucleosomeremoval and pioneering activity.The latter is for examplecompatible with recent observations that even the bind-ing of pioneer TFs can be depende
36、nt on other TFs(recently reviewed in Ref.20).TF cooperativity after DNA bindingThe existence of pioneer factors as a class of specializedTFs prompts the question whether more functionallydistinct types of TFs exist.Traditionally,TFs have beenviewed as activators or repressors that additively con-tri
37、bute positive or negative cues to the overall transcrip-tional output 44.Indeed,computational approaches topredict enhancer activities typically model the TFs reg-ulatory contributions using positive or negative weights4547.While some approaches also take TF cooperativ-ity into account,they typicall
38、y do not model qualitativedifferences between TFs,such as distinct and potentiallycomplementary molecular functions.However,evidence exists that combinations of differentTFs are more effective in activating transcription,sug-gesting that different TFs might have different molecularfunctions and cont
39、ribute distinct regulatory cues.Forexample,heterotypic clusters of binding sites conferredstronger activation than homotopic clusters in transcrip-tional reporter assays,indicating that different TFs aretypically needed to activate transcription robustly48,49,50?.Consistently,homotypic arrays of bin
40、dingsite sequences for individual TFs display complex spa-tio-temporal activity in Drosophila embryos that do notrecapitulate the respective TFs expression patterns51?.The regulatory activities of different TF proteins can bemost directly assessed and compared by tethering theTFs to transcriptional
41、reporter constructs via heterolo-gous DNA-binding domains(DBDs),such as the DBD ofthe yeast TF Gal4 52,53?or engineered DNA-bindingproteins like zinc-fingers 54?.As TF binding is stan-dardized,any difference in reporter expression betweentwo TFs would indicate that the TFs possess differentregulator
42、y functions.A recent analysis of the regulatoryCombinatorial function of transcription factors and cofactors Reiter,Wienerroither and Stark 75Figure 2passivecooperativity(collaborative binding)pioneer factorsTF-TFinteractionsenhanceosomemodel(b)closed chromatinclosed chromatin(a)Current Opinion in G
43、enetics&DevelopmentTF cooperativity at the level of DNA binding.(a)Different mechanisms of passive and active cooperativity between TFs at the level of DNA binding.Nucleosome-bound DNA is typically notaccessible to TFs(left),which have to compete with nucleosomes for DNA binding.Several TFs(rather t
44、han individual ones)can displacenucleosomes even in the absence of direct proteinprotein-interactions(passive cooperativity or collaborative binding,top).Direct proteinprotein-interactions,which for example occur between TFs in homodimers or heterodimers,can favor DNA binding(middle).An extreme case
45、 ofTF-TF and higher order proteinprotein interactions at enhancers is the enhanceosome(bottom).The formation of a rigid protein complex drivesbinding to enhancers with highly specifically arranged binding sites.(b)Pioneer factors are thought to be specialized TFs that can bind to theirpartly occlude
46、d recognition site in nucleosomal DNA and initiate chromatin remodeling to facilitate further TFs Current Opinion in Genetics&Development 2017,43:7381activities for 474 Drosophila TFs indeed suggests thatdifferent types of TFs with qualitatively different regu-latory functions exist 53?:the maybe mo
47、st obviousdifference was found between two types of TFs thatpreferentially activated transcription from the core-pro-moters of housekeeping genes(e.g.,the TF DREF)orfrom the core-promoters of developmentally regulatedgenes(e.g.,the TF Trithorax-like(Trl/GAGA),respec-tively(Figure 3a)(53?,recently re
48、viewed in 55).These factors might constitute the protein machineryof a dedicated program and could explain enhancercore-promoter specificities that separate developmentaland housekeeping gene regulation 55,56?.In yeast,different types of promoters that function predominantlyvia TFII D versus SAGA ha
49、ve also been described 57,suggesting that this mechanism exists more widely andacross different species.The observed core-promoter-specificity emphasizes that TFs can have qualitativelydifferent gene regulatory functions,which presumablydifferentially affect communication between TFs,cofac-tors and
50、Pol II 55.More interesting with regard to combinatorial enhancercontrol are however TFs that cannot activate transcrip-tion on their own but are able to complement,that is,re-activate,enhancers that were inactivated by mutating thebinding sites of individual TFs 53?(Figure 3b).Thissuggests that thes
