Chromatin and Transcription Factors : Who's on First?

Chromatin and transcription factors : who’s on first? Recent results suggest that nucleosomes and transcription factors interact dynamically to determine the transcriptional state of eukaryotic genes.

Chromatin is the physiologically relevant template for receptor, but not NFl, can bind when a nucleosome is po- transcription: chromatin structure and transcriptional ini sitioned on the promoter [ $41. It is unclear whether this tiation must therefore be mechanistically linked. Nuc- dilference in binding reflects the location of the nucleo- cleosome coating of the DNA severely restricts the some with respect to the binding sites or distinct prop- access of transcription factors to promoters. Nucleo- erties of the proteins. However, the transcription factors somal templates are transcribed much less efficiently GAL4 and HSF (heat shock factor) appear to be mecha- than naked DNA templates in vitro, and gross dis- nistically distinct, because only GAL4 can bind randomly ruption of chromatin structure in vivo by histone positioned or rotationally phased nucleosomes [ 51. The loss results in increased transcription. More generally, GAL4nucleosome complex is metastable; it dissociates changes in chromatin structure are very likely to af into nucleosomes and GAL&iaked DNA complexes upon feet the access and/or the function of transcription fac- addition of competitor DNA [6]. tors, thereby altering patterns of gene regulation. Con- versely, the binding of transcription factors to DNA Histone acetylation can increase the access to nucleo- must perturb chromatin structure by affecting nuc- somal DNA of at least one transcription factor, TFIIIA, leosome conformation, positioning, stability or removal. without influencing the extent of histone binding or the For these reasons, correlations between chromatin struc- DNA helical repeat [7]. TFIIIA also binds more efficiently ture and transcription, though easy to find, are not very to modified nucleosomes lacking the amino-terminal tails informative. Furthermore, activator proteins can stimulate of the histones. Thus, the histone tails inhibit the access transcription from naked DNA templates in vitro, leading of transcription factors, possibly through the presence of some to view chromatin structure as simply a mech- basic residues that interact with the phosphodiester back- anism for compacting DNA, with only a passive, though bone of DNA in the nucleosome. Acetylation might pre- generally inhibitory, role in transcription. However, work vent the histone tail from associating with DNA and/or over the past few years has provided increasing evidence alter the path of the DNA in the nucleosome. These in mat chromatin structure plays a more active role in the vitro results may be relevant in vivo, because acetylated transcription process. nucleosomes are associated with transcriptionally active chromatin. The access of transcription factors to promoters does not inherently require a special chromatin state, such as might The binding of one protein to nucleosomal DNA can fa- occur during DNA replication or mitosis. Although nucleo- cilitate binding by others. GAL4 derivatives bind coopera- somes constitute a sign&ant barrier, transcription factors tively to multiple sites in nucleosomal but not naked DNA, can bind DNA at essentially any time in the cell cycle and and HSF can bind nucleosomes in the presence, but not rapidly cause changes in chromatin structure that can ex- the absence, of TFIID [5]. In these examples, coopera tend over several nucleosomes [ 1,2]. In the promoter of tive binding occurs only on nucleosomes, and hence is the rat TATgene, hormone-activated binding of the gluco- unlikely to involve the conventional mechanism of direct corticoid receptor alters chromatin structure to allow a interactions between the relevant proteins. Instead, it is distinct protein, HNFS, to bind to the same sequence likely that binding by the first protein disrupts, but does [l ] As the glucocorticoid receptor and HNF5 cannot not displace, nucleosomes so that the second protein can simultaneously occupy this sequence, the induced struc- then bind more easily. Because this nucleosome-disrup- tural change must persist long enough to permit HNF5 tion mechanism of cooperative binding does not require access to DNA. Chromatin structural changes often pre- specific protein-protein interactions, it may explain why cede transcriptional activation during the development of transcriptional activation is often synergistic and promis- multicellular organisms. cuous - promoters containing multiple binding sites for unrelated proteins are generally much more active than Because chromatin structure in vivo is assayed by the in- promoters with single sites. In this regard, transcriptional direct method of nuclease digestion, it is unclear whether synergy depends on the number of proteins bound to the nucleosome disruption by transcription factors reflects promoter, not the number of activation domains [S] nucleosome removal. In vitro, it is clear that some transcription factors can bind to nucleosomal DNA. Of In vitro, nucleosomes severely repress basal TATA-depen- the proteins with binding sites in the promoter of the dent transcription, but only modestly affect transcription mouse mammary tumor virus (MMTV), the glucocorticoid activated by GAL4-vp16 chimeric molecules, consisting

220 0 Current Biology 1993, Vol 3 No 4 DISPATCH 221 of the GAL4 DNA-binding domain linked to the potent will differ among activators and promoters. For exam ~~16 activation domain [9]. A similar phenomenon is ple, the DEDl upstream element can stimulate transcrip- observed on DNA templates coated with histone Hl in- tion by T7 RNA polymerase in yeast cells presumably stead of nucleosome cores 1101. This relief of nucleo- by increasing access to the promoter, whereas the GAL some repression reiquires the activation domain and is enhancer cannot [ 151. not due to local disruption caused by DNA binding. Al- leviation of repression also may require RNA, perhaps Perhaps the strongest argument for an active role of chro- to accept histones that otherwise would be present on matin structure comes from histone mutations that cause the transcription template [lo]. Other proteins, such as specific transcriptional effects in yeast cells. Deletions SP-1 and the GAGA-factor, relieve Hl and nucleosome of the histone H4 amino terminus and mutations that repression, but the GAGA-factor is unable to activate prevent H4 acetylation severely compromise transcrip- transcription on naked DNA. These observations have tional enhancement by several activator proteins [16]. led to the suggestion that activator proteins may stimu- This effect on activation is specific to H4: it is not ob- late transcription by relieving nucleosome repression, in served in strains containing amino-terminal deletions of addition to directly stimulating the transcription machin- the other histones. In fact, amino-terminally deleted and ery. Clear interpretations are dilficult, however, because non-acetylatable derivatives of H3 increase activation \ 171. the experimental conditions may not be physiologically A sub-region of the histone H4 amino terminus, distinct relevant and because the contributions of the nucleo- from that involved in acetylation and activation, plays somal template and the activation domains are cleanly a specific role in transcriptional silencing of the yeast separated. mating-type genes [18] and of genes near chromosomal telomeres [ 19). Mutations in the I14 amino terminus that Transcriptional activation domains can disrupt nucleo- abolish silencing also disrupt nucleosomes positioned by some structure in viva. Analysis of estrogen receptor the a2 repressor in minichromosomes [ 201. These *and derivatives in yeast indicates that the degree of chromatin other results suggest that silencing is likely to involve a disruption is related to the strength of the transcriptional repressed chromatin state that can be propagated through activation domain [ll]. This disruption of chromatin many cell division cycles. structure requires multiple estrogen receptor binding sites and correlates with tr*anscriptional competence; hence it For nearly two decades, opinions about chromatin strut is difficult to distinguish the cause-and-effect relationship ture have oscillated between it being the key determinant between transcriptional activity and chromatin structure. or a passive bystander in transcriptional regulation. It is GAL4 disrupts a nucleosome in the GAL promoter region now clear that activation and repression of specific genes in a manner dependent on the activation domain [la]. results from the dynamic interactions between nucleo- Importantly, nucleosome disruption in the GAL promoter somes and transcription factors with DNA and with each is caused by the activation domain and is not a conse- other. The molecular mechanisms are just beginning to quence of transcriptional activity: it is observed even when be understood. transcription is blocked by mutational inactivation of the TATA element. References The yeast SNF and SWI proteins, which do not bind DNA but are important for transcriptional enhancement by a va- 1. RICA~III G, ROLIX J, 1’1c’r‘li-r K, GRANGE: T: In vizw footprinting of riety of DNA-bound activators, may activate transcription rat TAT gene: Dynamic interplay between the glucocorticoid receptor and a liver-specific factor. cell 1991, 67:977-c)% by affecting chromatin structure. Several lines of evidence 2. SCHMII) A, FAXHER K-D, HORZ W: Nucleosome disruption at suggest that these proteins function together, possibly as the yeast PHO5 promoter upon PHO5 induction occurs in a large complex that possesses a transcription activation the absence of DNA replication. Cell 1992, 71:853~864. domain(s ). SW3 interacts directlywith DNA-bound activa- 3. PINK R, BRIIGGEMEIER U, REATO M: Nucleosome positioning tors and is necessary for transcriptional activation in r&-o modulates accessibility of regulatory proteins to the mouse at an early stage of the reaction [ 131. SNF2 and SNF5 af- mammary tumor virus promoter. cell 1990, 60:7 19~731. feet chromatin structure in a manner that is influenced by 4. ARCIIER TK, COlu)INGI.I:Y MG, WOIFOIU) RG, HXXK GL; Transcrip- tion factor access is mediated by accurately positioned nuclc- histone dosage but is independent of the transcriptional osomes on the mouse mammary tumor virus promoter. ilk~l status of the promoter [14]. Cell Bid 1991, 1 1:688--698. 5. ‘l‘AYl.OR ICA, WORJMAN JL, fXiiE’l’Z TJ, I(IMXl’ON RE; kCi,itaXd The observations above suggest that activation domains binding of GAL4 and heat shock factor to nucleosomal templates: differential function of DNA-binding domains. (;erw.s might stimulate transcription by altering chromatin strut Der, 1991. 5:1X&1298. ture. However, DNA-bound activator proteins and the 6. WORKMAN Jr., KINGSTON R’S: Nucleosomc core displacement in putative SW-SIN complex, which may associate with vitro via a metastable transcription factor-nucleosome com- such activators, function in zdro on non-nucleosomal plex. Scimce 1992, 258:1780-1784. DNA templates. Thus, disruption of chromatin structure, 7, LEI: DY, HAYES JJ, PRUSS D, WOII’FE AI-‘: A positive role for his- though likely to be important for transcriptional active tone acetylation in transcription factor access to nucleosomal DNA. cell 1993. 72.73-84. tion, is clearly not sufficient to explain the full process. It 8. OLMERO s, STRUHL K: Synergistic transcriptional enhancement is also likely that the relative contributions of chromatin does not depend on the number of acidic activation do- disruption and stimulation of the transcription machinery mains bound to the promoter. Proc Nat1 Acad Sci liSA 1991, 222 Current Biology 1993, Vol 3 No 4

88:22&228. in viva Cdl 1991, 65:10231031. 9. WORKMAN JI TAYIIX ICA, KINGSTON RE; Activation domains of 17. MANN RR, GRUNSTKIN M: Histone H3 N-terminal mutations al- stably bound GAL4 derivatives akviate repression of promot- low hyperactivation of the yeast GAL1 gene in vivo. .!?hfBO ers by nucleosomes. cell 1991, 643533544. J 1992, 11:3297-3306. 10. CROS~ON GE, IAYBOURN PJ, PARANJAPE SM, KADONAGA JT: Mech- 18. JOHNSON LM, FISHER-ADAMS G, GRUNSTEIN M: Identification of anism of transcriptional antirepression by GAL4-VP16. Genes a non-basic domain in the histone H4 N-terminus required DCZV 1992, 6~227~2281. for repression of the yeast silent mating loci. EMBO J 1992, 11. PHAM TA, HWUNG Y-P, MCDONNEU. DP, O’MAWZY Bw: Transacti- 11:2201-2209. vation timctions facilitate the disruption of chromatin struc- 19. APARICIO OM, BILLINGTON BI, GOTIXHLING DE: Modifiers of po- ture by estrogen receptor derivatives in viva J Biol Cbem sition effect are shared between telomeric and silent mating- 1991, 266:1817%18187. type loci in S. cerevisiae Cell 1991, 66:1279-1287. 12. AKEIROD JD, MAJOB J: GAL4 disrupts a nucleosome to activate 20. ROTH SY, S~uuzu M, JOHNSON L, GRUNSTEIN M, SIMPSON RT: Sta- GAL1 tmnscription in viva Genes h, in press. ble nucleosome positioning and complete repression by the 13. YOSHINAGA SK, PETERSON CI+ HErSKOWITZ 1, YAMAMOTO RF? Roles yeast a2 repressor are disrupted by ammo-terminal mutations of WI, swI2, and SWI3 proteins for tranxriptional enhance- in histone H4. Genes Dev 1992, 6:411-425. ment by steroid receptors. Science 1992, 258:159k%1604. 14. HIRXHHORN JN, BKOWN SA, CLUUC CD, WINSTON F: Evidence tht SNF2/SWI2 and SNF5 activate transcription in yeast by akering chromatin structure. Genes Da, 1992, 6:2288-2298. 15. Cm W, TABOR S, STRUHL K: Distinguishing between mechanisms of eukaryotic transaiptional activation with bacterio- phage l7 RNA polymerax. Cell 1987, 50:1047-1055. Kevin Struhl, Department of Biological Chemistry and 16. DWWN LK, MANN RK, KAYNE PS, GRUNSTEIN M: Yeast histone Molecular Pharmacology, Harvard Medical School, Boston, H4 N-terminal sequence is required for promoter activation Massachusetts 02115, USA

COMMING UP IN THE APRIL ISSUE OF CURRENT OPINION IN GENETICS AND DEVELOPMENT

Ueli Schibler and Steve M&night will edit the following reviews on Gene expression and differentiation:

Trawacting factors involved in adipogenic differentiation by Mireille Vasseur-Cognet and M. Daniel Lane Regulating the HO endonuclease in yeast by Kim Nasmyth RNA polymerase II transcription cycles by Jeffrey L. Corden Lessons from lethal albino mice by Gavin Kelsey and Giinter Schlitz Circadian-clock regulation of gene expression by Joseph S. Takahashi Regulation of the P-globin locus by Merlin Crossley and Stuart H.Orkin Histones, nucleosomes and transcription by John Svaren and Wolfram H&-z Transcription regulatory proteins In higher plants by Alan N. Brunelle and Nam-Hai Chua Regulation of the HNF-1 homeodomain proteins by DCoH by Linda P. Hansen and Gerald R. Crabtree Signal transduction in Bacillus subtZZfs sporulation by Mark A. Strauch and James A. Hoch Skeletal myogenesis: revelations of genetics and embryology by Charles P. Emerson, Jr Interaction of coiled-coils in transcription factors: where is the specificity? by Andreas D. Baxevanis and Charles R. Vinson Homeobox genes in CaenorhabdWs elegant by Martin Chalfie Effects of DNA methylation on DNA-binding proteins and gene expression by Adrian P. Bird and Peri H. Tate Gene regulation: translation initiation by internal ribosome binding by Soo-Kyung OH and Peter San-row