Dispatch R341

Transcriptional regulation: Kamikaze activatorsDominique Thomas* and Mike Tyers†

Transcription factors are often targeted for rapiddegradation by the ubiquitin–proteasome system.Recent evidence points to a correlation between thepotency and instability of transcriptional activators,suggesting a possible direct role for ubiquitin-dependent proteolysis in transcriptional activation.

Addresses: *Centre de Génétique Moléculaire, Centre National de laRecherche Scientifique, 91198 Gif-sur-Yvettte, France. †Programme inMolecular Biology and Cancer, Samuel Lunenfeld Research Institute,Mount Sinai Hospital, Toronto M5G 1X5, Canada.E-mail: thomas@cgm.cnrs-gif.fr; tyers@mshri.on.ca

Current Biology 2000, 10:R341–R343

0960-9822/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved.

The ubiquitin proteolytic system often attenuates cellularresponses to extracellular cues by rapidly degrading keysignalling proteins, from cell surface receptors to nucleartranscription factors. Ubiquitin is covalently attached tosubstrate proteins by a conserved enzymatic cascade,E1→E2→E3, which assembles a polyubiquitin chain onthe substrate, leading to its capture and rapid degradationby the 26S proteasome [1]. The crucial step in thepathway is substrate recognition, carried out by the E3ubiquitin ligases, which bind to a specific region on thesubstrate termed a degron. Although the ubiquitin-depen-dent elimination of numerous transcription factors is welldocumented, recent studies suggest that the most potenttranscriptional activators are highly unstable, perhaps as avery consequence of their efficient interaction with thetranscriptional machinery.

In eukaryotes, activation of transcription by RNApolymerase II requires the coordinated assembly of acomplex protein machinery, composed of the RNApolymerase II holoenzyme, general transcription factors,chromatin remodeling activities, histone acetylases andgene-specific transcriptional activators [2]. Activators have amodular structure, typically comprising a DNA bindingdomain that recognizes enhancer sequence elements in thetarget gene promoter, and one or more activation domainsthat stimulate RNA polymerase II-dependent gene tran-scription. Domain swap experiments have revealed thatactivation domains are functionally autonomous units ableto stimulate RNA polymerase II activity when fused to aheterologous DNA binding domain. Activation domains areloosely classified by amino acid composition as acidic, gluta-mine-rich, proline-rich and isoleucine-rich. In extremecases, a stretch of six to eight amino acids is sufficient fortranscriptional activation. Activation domains make direct

protein–protein contacts with RNA polymerase II-associ-ated factors and are believed to stimulate transcription bydirecting the recruitment of the RNA polymerase IIholoenzyme to promoter regions on DNA [3].

An initial link between activator degradation andtranscriptional potency arose from a study of the Myctranscription factor, a powerful regulator of cell prolifera-tion that is normally held in check by ubiquitin-depen-dent proteolysis. Deletion analysis of Myc showed thatits activation domain and degron regions are tightlysuper-imposed [4]. A subsequent literature surveyrevealed overlapping activation domains and degrons innumerous other transcription factors [5]. To evaluate thepossible functional significance of the destabilizingregions in activators, Salghetti et al. [5] undertook a sys-tematic analysis of various activation domains fused tothe same Gal4 DNA binding domain. Independently,while in the course of dissecting transcription factorstructure/function relationships, Molinari et al. [6] exam-ined a similar panel of chimeric activators. Both studiesunveiled a striking coincidence of activator potency andinstability. In particular, the most potent class of transcrip-tional activators, those rich in acidic residues, are by farthe most unstable. In contrast, transcription factors withweak proline-rich or glutamine-rich activation domains aremuch more stable.

The relationship between activator potency and instabil-ity was corroborated by a detailed analysis of syntheticactivation domains, consisting of multiple tandem copiesof a short acidic fragment derived from the potent activa-tion domain of the HSV viral protein VP16 [5,6]. As thecopy number of the VP16 fragment is increased, so istranscriptional activation and, concomitantly, instabilityof the chimeric activator. Importantly, single point muta-tions in the VP16 sequence that abolish transcriptionalactivation strongly stabilize the chimeric activators. Inaddition, recruitment of the activator to DNA appearsnecessary for proficient degradation [6]. The variouschimeric activators are targeted by the ubiquitin system,as shown by the accumulation of ubiquitinated forms ofthe activators in cells treated with proteasome inhibitors.Moreover, the extent of activator ubiquitination is pro-portional to the number of VP16 repeats [5]. Intrigu-ingly, in pursuing the activation domain–degronconnection, Salghetti et al. found that PEST-rich degrondomains derived from yeast G1 cyclins, which do notnormally participate in transcriptional activation, alsofunction as activation domains when fused to the Gal4DNA binding domain [5].

bb10i09.qxd 10/5/00 9:04 am Page R341

A recent study of the Microphthalmia (Mi) transcriptionfactor by Wu et al. [7] lends further support to the idea thatpotent activation domains can simultaneously serve assignals for proteolysis. Mi is an essential effector of thec-Kit receptor tyrosine kinase during melanocyte develop-ment. Activation of c-Kit by its ligand Steel results inphosphorylation of Mi by two downstream kinases in thesignaling pathway, mitogen activated protein (MAP)kinase and Rsk. In turn, phosphorylated Mi recruits a tran-scriptional coactivator, p300/CBP, to target gene promot-ers, thereby activating transcription. Once phosphorylated,Mi is also rapidly degraded by the ubiquitin–proteasomesystem. Mutation of the two relevant serine residues onMi that are normally phosphorylated abolishes Mi-depen-dent transcription without affecting the ability of Mi tobind DNA. And correspondingly, the mutant version ofMi is completely refractory to signal-induced degradation.

Taken together, these results argue that transcriptionalactivation and activator degradation are closely coupledevents. It is worth emphasizing that this relationship is theexact converse of that observed in numerous other circum-stances, in which inhibition of activator degradation leadsto an increase in transcription. A notable example of thelatter is the stabilization of p53 and concomitant activationof p53-dependent transcription during the DNA damageresponse [8]. How might these superficially disparate find-ings be reconciled? A key difference between the twoprocesses lies in the context of the degradation pathways.On the one hand, the many dedicated pathways thatsignal turnover of specific activators do so independently

of transcriptional status, while on the other hand, the insta-bility of strong activators apparently requires a productiveinteraction with the RNA polymerase II machinery. Thatis, in addition to an intact activation domain, proper recruit-ment of the activator to the promoter is necessary for effi-cient degradation [6]. In perhaps the simplest model, theRNA polymerase II holoenzyme and/or its cofactors mightbridge the activator to the ubiquitination machinery(Figure 1). In a formal sense, the RNA polymerase IImachinery would serve as the substrate recognition compo-nent of an E3 ubiquitin ligase.

If this notion is correct, then it should be possible to detectan activator-specific ubiquitin ligase activity associatedwith the RNA polymerase II holoenzyme or a coactivatorcomplex. A number of components of the ubiquitin–pro-teasome pathway have been detected in association withthe RNA polymerase II machinery, including E3 ubiquitinligases, ubiquitin hydrolases and subunits of the protea-some (see citations in [5–7]). Indeed, an age-old puzzle isthe association of mono-ubiquitinated forms of histoneH2A with actively transcribed regions of the genome [9].In a recent intriguing example, the HECT domain E3enzyme Tom1 was found to be required for efficient tran-scription of several genes in yeast and physically associateswith SAGA, a multisubunit histone acetylase [10]. Perhapsnot by chance, the SAGA complex is preferentiallyrecruited by acidic activation domains [11]. Whether it isTom1 or some other E3 that catalyzes ubiquitination ofacidic activators remains to be determined. Delineation ofthe degradation pathway will also await the identification

R342 Current Biology Vol 10 No 9

Figure 1

Ubiquitin-dependent proteolysis of potent transcriptional activators.(a) A strong acidic activation domain (AD) binds to an enhancerelement (ENH) via a DNA binding domain (DBD). The activationdomain is envisaged to recruit the basal transcription apparatus,consisting of the RNA polymerase II holoenzyme (RNA pol II) andassociated general transcription factors, including TFIID, whichbinds to the TATA element in the core promoter. It is not known atwhat point or by what means the ubiquitination machinery

(E1/E2/E3) is recruited to the promoter. (b) Assembly of thetranscriptional initiation complex leads to ubiquitination of theactivator and its rapid degradation by the 26S proteasome. Possiblecoupling of activator degradation to elongation is indicated byphosphorylation of the CTD and/or the activator by various CTDkinases. (c) Speculative role for activator degradation in elongationand possible requirement for continuous reloading of newlysynthesized activators.

bb10i09.qxd 10/5/00 9:04 am Page R342

of the relevant E2 enzyme, and other factors that mightregulate the dynamics of ubiquitin conjugation to activa-tors, such as ubiquitin hydrolases. Finally, it will be of con-siderable interest to determine if components of thegeneral transcriptional machinery are also degraded inconcert with intense transcriptional activity.

But why destroy the very factors dedicated in the firstplace to high level transcription? The obvious possibilityis to provide a fail-safe mechanism against unbridled, andhence potentially deleterious, levels of transcription.Similarly, the intrinsic instability of potent activatorsmight allow the transcriptional apparatus to be quickly re-programmed upon a change in cellular state without theneed for degradation pathways tailored to each and everyactivator. But perhaps the effect has deeper origins.Another unique property of potent acidic activators is theirability to stimulate transcriptional elongation and down-stream mRNA processing events [12]. Might this coinci-dence reflect an unsuspected liaison between efficientelongation and transcription factor degradation?

There is a good deal of evidence indicating that elongationrequires phosphorylation of the extensive carboxy-terminalheptad repeat (CTD) domain of the large subunit of RNApolymerase II, a complex event carried out by several resi-dent cyclin-dependent kinases of the transcriptionalmachinery [13]. It is possible that activator degradationmight arise either as a consequence of CTD phosphoryla-tion, perhaps through recruitment of E3 activity to the RNApolymerase II holoenzyme (for example [14]), or even as adirect result of activator phosphorylation by one or moreRNA polymerase II associated kinases (for example [15]).The ubiquitination machinery often specifically targetsphosphorylated substrates and in fact, the yeast G1 cyclindegrons that act as synthetic activation domains when fusedto a DNA binding domain are normally targeted for degra-dation by phosphorylation [5]. It is just conceivable thatsuch degrons might recruit their cognate E3 enzymes toectopically stimulate the transcriptional machinery.

What specific role could activator degradation play intranscriptional elongation? As an alternative to models inwhich elongation depends on transactivator-RNApolymerase II holoenzyme dissociation, one might envisageproteolysis of the promoter-bound activator as a way ofdecoupling the engaged RNA polymerase II complex fromthe promoter (Figure 1). In this scenario, on-going transcrip-tion would entail continuous reloading of newly synthesizedtranscription factors, a possibility that has yet to be exam-ined in vivo. A rapid turnover of promoter-bound transcrip-tion factors would certainly lend an appealing dynamic edgeto the process of RNA polymerase II holoenzyme recruit-ment. Flights of fancy aside, ubiquitination seems set toclaim a place alongside acetylation, methylation and phos-phorylation on the list of the biochemical events intimately

associated with RNA polymerase II-dependent transcrip-tion. So much the worse for the potent activators that areindeed the divine wind of high-level transcription.

AcknowledgementsWe thank our colleagues in the transcription field for stimulating (and oftenheated!) discussions.

References1. Hershko A, Ciechanover A: The ubiquitin system. Annu Rev Biochem

1998, 67:425-479.2. Kornberg RD: Eukaryotic transcriptional control. Trends Cell Biol

1999, 9:M46-M49.3. Stargell LA, Struhl K: Mechanisms of transcriptional activation

in vivo: two steps forward. Trends Genet 1996, 12:311-315.4. Salghetti SE, Kim SY, Tansey WP: Destruction of Myc by ubiquitin-

mediated proteolysis: cancer-associated and transformingmutations stabilize Myc. EMBO J 1999, 18:717-726.

5. Salghetti SE, Muratani M, Wijnen H, Futcher B, Tansey WP:Functional overlap of sequences that activate transcription andsignal ubiquitin-mediated proteolysis. Proc Natl Acad Sci USA2000, 97:3118-3123.

6. Molinari E, Gilman M, Natesan S: Proteasome-mediated degradationof transcriptional activators correlates with activation domainpotency in vivo. EMBO J 1999, 18:6439-6447.

7. Wu M, Hemesath TJ, Takemoto CM, Horstmann MA, Wells AG,Price ER, Fisher DZ, Fisher DE: c-Kit triggers dual phosphorylations,which couple activation and degradation of the essentialmelanocyte factor Mi. Genes Dev 2000, 14:301-312.

8. Carr AM: Piecing together the p53 puzzle. Science 2000,287:1765-1766.

9. Levinger L, Varshavsky A: Selective arrangement of ubiquitinatedand D1 protein-containing nucleosomes within the Drosophilagenome. Cell 1982, 28:375-385.

10. Saleh A, Collart M, Martens JA, Genereaux J, Allard S, Cote J, BrandlCJ: TOM1p, a yeast hect-domain protein which mediatestranscriptional regulation through the activation domainA/SAGAcoactivator complexes. J Mol Biol 1998, 282:933-946.

11. Ikeda K, Steger DJ, Eberharter A, Workman JL: Activation domain-specific and general transcription stimulation by native histoneacetyltransferase complexes. Mol Cell Biol 1999, 19:855-863.

12. Bentley DL: Regulation of transcriptional elongation by RNApolymerase II. Curr Opin Genet Dev 1995, 5:210-216.

13. Yankulov K, Bentley D: Transcriptional control: Tat cofactors andtranscriptional elongation. Curr Biol 1998, 8:R447-R449.

14. Mitsui A, Sharp PA: Ubiquitination of RNA polymerase II largesubunit signaled by phosphorylation of carboxyl-terminal domain.Proc Natl Acad Sci USA 1999, 96:6054-6059.

15. Vandel L, Kouzarides T: Residues phosphorylated by TFIIH arerequired for E2F-1 degradation during S-phase. EMBO J 1999,18:4280-4291.

Dispatch R343

bb10i09.qxd 10/5/00 9:04 am Page R343