Journal of Cell Science 102, 1-5 (1992)Printed in Great Britain © The Company of Biologists Limited 1992

COMMENTARY

Structure and dynamics of transcriptionally active chromatin

JUAN AUSIO

Department of Biochemistry & Microbiology, University of Victoria, Victoria, British Columbia, Canada V8W 3P6

Introduction

Chromatin is a highly dynamic macromolecular com-plex that undergoes continuous structural modificationduring the various stages of genetic activity. Suchdynamic behavior is the result of a complex andperfectly balanced network of interactions involvinghistones, DNA and ions in an aqueous environment. Inthe past eighteen years since the discovery of thenucleosome particle, a great deal of information hasbeen obtained on interactions responsible for maintain-ing the static three-dimensional structure of inactivechromatin (see Van Holde, 1988, for a recent review onchromatin). More recently, structural studies ofchromatin have focused on the analysis of the molecularevents involved in chromatin activation. A variety ofmodels have been proposed, primarily to explain thestructural transitions undergone by chromatin duringtranscription. In the following sections, I will brieflyreview what is known about the structure of transcrip-tionally active chromatin and propose a model for"activation" that would account for the chemical andionic interactions that are possibly involved. I hope thatthis model will stimulate design of new experimentalapproaches with which to understand better thestructure-function relationship in chromatin.

Active chromatin

Early studies carried out by Weintraub and Groudine(1976) demonstrated that active genes exhibited analtered chromatin structure with enhanced sensitivity toDNase I. However, the relationship if any, betweenthese two phenomena remained unclear for manyyears. In the interim, the compositional characteriz-ation of active chromatin was further elucidated. Non-histone proteins HMG-14 and HMG-17 were found tobe preferentially associated with active genes andinitially held responsible for the DNase hypersensitivity(Weisbrod and Weintraub, 1979). Acetylated histoneswere also shown to be preferentially associated with theactive chromatin template (Davie and Candido, 1978).However, it was subsequently shown that selectiveremoval of the linker histones (histones of the HIfamily), as well as HMG-14 and HMG-17 from activechromatin regions, failed to abolish DNase I sensitivity

(Goodwin et al., 1985). A radical turn in this directionwas taken with the demonstration that torsional stress,due to topoisomerase activity, was ultimately respon-sible for the nuclease hypersensitivity of transcription-ally active chromatin (Villaponteau et al., 1984).Following this discovery, it was clear that geneactivation involved dynamic structural changes inchromatin with distinctive compositional differencesbetween the active and inactive state. Unfortunatelythis finding neither explained nor defineed the nature ofthe altered chromatin structure initially observed byWeintraub and Groudine.

Weintraub and coworkers (Weintraub et al., 1976)have suggested that the structure of active chromatininvolves an altered "unfolded" nucleosomal organiz-ation with changes primarily in the histone-histoneinteraction patterns. More recently, an unfolded nu-cleosome particle "lexosome" has been proposed (Prioret al., 1983) with an altered "more accessible" dispo-sition of the SH groups of histone H3 (Chen andAllfrey, 1987). U-shaped particles possessing 20% lessprotein mass than normal globular nucleosomes (mostlikely as a result of the loss of an H2A-H2B dimer, seebelow) have also been recently observed in transcrip-tionally active chromatin (Lociclear et al., 1990).

At the supranucleosomal level, reduced compactionof the active chromatin seems to be associated with apartial depletion of linker histones (Kamakaka andThomas, 1990). This transient release of the linkerhistones as well as the histone octamer (Lorch et al.,1987; Pfaffle et al., 1990; Clark and Felsenfeld, 1991)and/or one histone H2A-H2B dimer (Baer and Rhodes,1983) as a result of the RNA polymerase activity mayalso contribute to the altered conformation of chroma-tin within actively transcribed regions. Such RNApolymerase-mediated changes in chromatin structurecould involve displacement of histones from their"native" positions and possibly binding to adjacent orneighboring secondary sites. Indeed, the association ofnucleosome core particles with additional core histonesat intermediate ionic strengths (Voordow and Eisen-berg, 1978) and under physiological conditions (Stein etal., 1985) has been well documented. The extent ofthese compositional changes and thus the level ofchromatin unfolding would ultimately depend on the

Key words: chromatin, transcription, histones.

/. Ausio

particular kind of gene being transcribed (Van Holde,1988).

Histone acetylation: the unsolved puzzle

Without a doubt, the structural motif most stronglycorrelated with chromatin activity is acetylation of thecore histones (see Csordas, 1990, for a recent review).The biological relevance of this post-translationalmodification of histones, which involves charge neutral-ization of some of the lysine residues within the N-terminal regions (tails) of the histones, was alreadyenvisaged from the early days of its discovery. Becauseof the chemical nature of this phenomenon suchmodifications have been generally thought to partici-pate in the loosening of the chromatin structure. Inspite of the extensive experimental work carried out tosupport this hypothesis, most of this effort has failed toprovide evidence of any significant structural changes.In the absence of any other factors, a histoneacetylation-dependent unfolding of either nucleosomesor higher-order chromatin structure has not beenobserved. At the first level of chromatin organization,highly hyperacetylated nucleosomes remain essentiallyfolded in solution at physiological ionic strength (Ausioand Van Holde, 1986). This finding is not surprising,however, since complete removal of the N-terminalregions of the histones by immobilized trypsin also failsto affect the overall structure of the chromatin subunit(Ausio et al., 1989). However, these observations donot preclude a partial loss of stability of the particle inresponse to mechanical forces, resulting from theweakening or disappearance of some of the histone-DNA interactions in these regions. Indeed, a selectiveunfolding of highly hyperacetylated nucleosome coreparticles has been observed in the electron microscopeas a result of the additional shearing mechanical stressintroduced by this technique (Oliva et al., 1990). Invivo, such stress may be introduced into the chromatintemplate by topoisomerase II activity or RNA polym-erase passage (Pfaffle et al., 1990; Clark and Felsenfeld,1991). At the next level of chromatin organization,histone hyperacetylation seems to have very Little effecton the condensation state of the 30 run fiber (McGhee etal., 1983). This observation is in contrast with theapparent complete absence of folding exhibited bytrypsinized chromatin that had been further reconsti-tuted with native histone HI (Allan et al., 1982). It isimportant to emphasize, however, that the chromatinsamples used in the acetylation experiments wereobtained from HeLa cells grown in the presence ofsodium butyrate in order to inhibit histone deacetylaseactivity (Riggs et al., 1977). Indeed butyrate treatmentalso enhances the expression of histone HI0 (Chabanaset al., 1985). The presence of high levels of Hl° couldaccount for the condensed chromatin structure ob-served in these chemically perturbed cells. Therefore,results obtained under these conditions must beinterpreted with caution.

A solution to the apparent lack of correlation

between acetylation and chromatin unfolding may beascertained from a careful analysis of the structural dataalready available. We observed that, during thepreparation of nucleosome particles, the linker regionsof hyperacetylated chromatin, which had been strippedof histone HI, were digested more readily by micrococ-cal nuclease than in the non-acetylated control (Liber-tini et al., 1988). In addition, only minor acetylation-induced structural changes were detected at the level ofthe nucleosome core particle (Ausio and Van Holde,1986), while more extensive differences have beenreported on nucleosome monomers containing undi-gested linker DNA (Bode et al., 1983). The conclusionfrom these studies is that in the absence of histone HIthe major structural effect of acetylation originates atthe linker DNA component of chromatin. The findingthat in Hl-stripped chromatin the linker DNA is moreaccessible to nuclease attack in the hyperacetylatedfractions clearly indicates that this DNA region has amore accessible conformation. Indeed it has beenshown that histone acetylation reduces the nucleosomecore-particle linking number change (Norton et al.,1989) and that it alters the capacity of exogenouslyadded histones HI to precipitate transcriptionallyactive/competent chromatin (Ridsdale et al., 1990). It isthus possible that upon neutralization (or removal bytrypsinization) of the charges of the histone tails, anddisplacement of histone HI by non-histone proteins, orother unknown effects, the exit and entry trajectories ofthe flanking DNA regions of the nucleosomes becomesubstantially altered. This could in turn temporarilyprevent histone HI from binding to its native site and incombination with torsional stress may be responsiblefor maintaining the extended (partially depleted ofhistone HI) chromatin conformation necessary for thepassage of RNA polymerase.

A structural model for chromatin activation

From what has been described above, it is clear thatacetylation by itself cannot directly account for thestructural changes observed in active/competentchromatin. Rather, I think that the role of acetylation isto poise chromatin for activation. This may be carriedout in two ways (see Fig. 1). First of all, the finding thathistone acetylation occurs sequentially throughout mostof the chromatin (Perry and Chalkley, 1982) suggests arole in partial decondensation, unshielding some of theDNA sequences that may become available for proteinrecognition by regulatory proteins such as transcriptionfactors and/or RNA polymerase itself (see Fig. 1A).Alternatively, acetylation may be required for furtherunfolding of the otherwise condensed 30 nm chromatinfiber. Such unfolding could be induced by the introduc-tion of torsional stress mediated by the presence oftopoisomerases or by polymerase advance duringtranscription (Pfaffle et al., 1990) or both (see Fig. IB).Interestingly, topoisomerase II is found associated withnuclear matrix attachment regions (MARs) at the baseof chromosomal loops (Cockerill and Garrard, 1986).

Transcriptionally active chromatin

30nm

B RNA C RNA polymerase /?E

Fig. 1. A model for chromatin activation. (A) Acetylation of a chromatin domain (shaded region) relaxes slightly thehigher-order structure within this region and unshields the DNA sequences for protein recognition by regulatory proteins.Chromatin is thus "poised" for activation. (B) Upon induction of torsional stress, the "poised" domain may partiallyunfold, releasing histone HI. The flanking DNA regions and the linker DNA of the nucleosomes involved become moreextended, thus temporarily preventing HI from binding to its primary nucleosomal site. The passage of RNA polymerasemay induce a further unfolding of the nucleosome structure (shown here) with the displacement of an H2A-H2B dimer(see Fig. 2) or complete dissociation of the histone octamer depending on the class of gene being transcribed (van Holde,1988). (C) Upon removal of the acetyl groups by deacetylases, and relaxation of the torsional stress (in the wake of RNApolymerase), histone HI is able to recognize and bind again to its native binding sites and fold the nucleosome fiber backinto its 30 nm higher-order inactive structure.

The activity of topoisomerase II at these sites could actin targeting specific DNA loci for expression. Regu-lation of the level of supercoiling of entire loops bytopoisomerase II and other trans-acting factors (Grossand Garrard, 1987) in conjunction with acetylationwould thus provide a mechanism for recognition of theDNA sequences to be transcribed. The resultingextended structure of this active chromatin confor-mation may promote local release or displacement ofhistone HI molecules from their native binding sites.Non-histone proteins such as HMG 14 or HMG 17 mayplay an important role in stabilizing this transientlyunfolded structure. In transcribed regions, the negativesupercoiling of DNA in the wake of RNA polymerase(Pfaffle et al., 1990) and deacetylation of histoneswould restore the initial inactive chromatin structure(Fig. 1C).

The finding that histone residues targeted for acety-

lation occur non-randomly and to a different extent inresponse to physiological situations (i.e. in transcriptioncompared to replication; Csordas, 1990) adds an extralevel of complexity in the model proposed. However,beyond a critical number of about 12 acetyl groups pernucleosome particle, the structural changes observed atthis level seem to be independent of the extent ofacetylation (Ausio and Van Holde, 1986). Thus thefunction-dependent specific pattern of acetylation mayplay an additional regulatory role in the transitionsinvolved (Loidl, 1988). Such distinctive patterns ofchromatin acetylation may be important for recognitionby deacetylases or in controlling the rate of removal ofthe acetyl groups by these enzymes.

At the nucleosome level, the combination of tor-sional stress and histone acetylation could also beresponsible for the transient unfolding of the nucleo-some particle (see Fig. 2), inducing U-shaped "lexo-

J. Ausio

torsional stress

Fig. 2. Possible unfolding of the nucleosome core particle as a result of chromatin activation. Upon acetylation, thenucleosome core particle remains essentially in its folded native (unshaded particles) conformation. Yet, it adopts a slightlylooser conformation (shaded particles) probably arising from the release of the N-terminal regions (tails) of the histoneoctamer. In the presence of torsional stress these particles may reversibly unfold into lexosome-like structures. It is possiblethat during this process some rearrangement of the core histones takes place.

some"-like structures such as those observed in tran-scriptionally active chromatin (Locklear et al., 1990).The requirement for torsional stress could explain theelusive nature of this structure, which would be highlyunstable (i.e. upon removal of torsional constraints bymicrococcal nuclease cleavage at the adjacent DNAlinker regions). It is noteworthy that the only "lexo-somes" characterized thus far contained high levels ofboth non-histone protein (Prior et al., 1983) andacetylated histones (Boffa et al., 1990).

Even if the model presented here is not entirelyaccurate, it seems clear that, the "tails" of the histoneoctamer play a more dynamic role in the modulation ofchromatin structure than was initially envisaged.Through modification of the net charge distribution inthese regions it is possible to modulate the functionalstate of chromatin (see also Earnshaw, 1987).

In their native fully charged conformation, the role ofthe tails may be to enhance the stability of thenucleosome core particle and provide the proper DNAconformation at the exit and entry points of the coreparticle, allowing for the proper binding of the linkerhistones. At the same time they may provide thenecessary charge shielding on the DNA required for thefolding and compaction of the chromatin fiber from itsextended 10 run configuration to its 30 nm higher-orderstructure. By simply altering this ionic balance it ispossible to reverse these functions and thus poisechromatin for transcriptional activation.

I am very grateful to Nik Veldhoen for carefully reading themanuscript and also for his suggestions. I also thank MsCheryl Gonnason for typing the manuscript.

References

Allan, J., Harbome, N., Rau, D.C. and Gould, H. (1982).

Participation of core histone "tails" in the stabilization ofchromatin solenoid. J. Cell Biol. 93, 285-297.

Ausio, J., Dong, F. and Van Holde, K. E. (1989). Use of selectivelytrypsinized core particles to analyze the role of histone "tails" inthe stabilization of the nucleosome. J. Mol. Biol. 206, 451-463.

Ausio, J. and Van Holde, K. E. (1986). Histone hyperacetylation: Itseffects on nucleosomes conformation and stability. Biochemistry25, 1421-1428.

Baer, B. W. and Rhodes, D. (1983). Eukaryotic RNA polymerase IIbinds to nucleosome cores from transcribed genes. Nature 301,482-488.

Bode, J., Gomez-Lira, M. and Schrfiter, H. (1983). Nucleosomalparticles open as the histone core becomes hyperacetylated. Eur. J.Biochcm. 130, 437-445.

Boffa, L. C , Walker, J., Chen, T. A., Sterner, R., Mariani, M. R.and Allfrey, V. G. (1990). Factors affecting nucleosome structure intranscriptionally active chromatin. Eur. J. Biochcm. 194, 811-823.

Chabanas, A., Khoury, E., Goeltz, P., Gjerset, R., Dod, B., Eisen,HI. and Lawrence, J. J. (1985). Effects of butyric acid on cell cycleregulation and induction of histone HI degree in mouse cells andtissue culture. J. Mol. Biol. 183, 141-151.

Chen, T. A. and Allfrey, V. G. (1987). Rapid and reversible changesin nucleosome structure accompany the activation, repression andsuperinduction of murine fibroblast protooncogenes. Proc. Nat.Acad. Sci. U.S.A. 84, 423-431.

Clark, D. and Felsenfeld, G. (1991). Formation of nucleosomes onpositively supercoiled DNA. EMBO J. 10, 387-395.

Cockerill, P. N. and Garrard, W. T. (1986). Chromosomal loopanchorage of the kappa immunoglobulin gene occurs next to anenhancer in a region containing topoisomerase II sites. Cell44,273-282.

Csordas, A. (1990). On the biological role of histone acetylation.Biochem. J. 265, 23-38.

Davie, J. R. and Candido, P. M. (1978). Acetylated histone H4 ispreferentially associated with template-active chromatin. Proc.Nat. Acad. Sci. U.S.A. 75, 3574-3577.

Earnshaw, W. C. (1987). Anionic regions in nuclear proteins. J. CellBiol. 105, 1479-1482.

Goodwin, G. H., Nicolas, R. H., Cockerill, P. N. and Wright, C. A.(1985). The effect of salt extraction on the structure oftranscriptionally active genes; evidence for a DNAase-I sensitivestructure which could be dependent on chromatin structure atlevels higher than 30 nm fiber. Nud. Acids Res. 13, 3561-3579.

Gross, D. S. and Garrard, W. T. (1987). Poising chromatin fortranscription. Trends Biochem. Sci. 12, 293-297.

Transcriptionally active chromatin

Kamakaka, R. T. and Thomas, J. O. (1990). Chiomatin structure oftranscriptionally competent and repressed genes. EMBOJ. 9,3997-4006.

Liberttnl, L. J., Ausk), J., Van Holde, K. E. and Small, E. (1988).Histone hyperacetylation: Its effects on nucleosome core particletransition. Biophys. J. 53, 477-487.

Locklear, L., Ridsdale, A. J., Bazett-Jones, D. P. and Davle, J. R.(1990). Ultrastructure of transcriptionally competent chromatin.Nucl. Acids Res. 18, 7015-7024.

Loidl, P. (1988). Towards an understanding of the biological functionof histone acetylation. FEBS Lett. 227, 91-95.

Lorch, Y., LaPolnte, J. W. and Kornberg, R. D. (1987). Nucleosomesinhibit the initiation of transcription but allow chain elongationwith the displacement of histones. Cell 49, 203-210.

McGhee, J. D., Nickol, J. M., Felsenfeld, G. and Ran, D. C. (1983).Histone hyperacetylation has little effect on the higher orderfolding of chromatin. Nucl. Acids Res. 11, 4065-4075.

Norton, V. G., Imal, B. S., Yau, P. and Bradbury, E. M. (1989).Histone acetylation reduces nucleosome core particle linkingnumber change. Cell 57, 449^457.

OUva, R., Bazett-Jones, D. P., Locklear, L. and Dixon, G. H. (1990).Histone hyperacetylation can induce unfolding of the nucleosomecore particle. Nucl. Acids Res. 18, 2739-2747.

Perry, M. and Chalkley, R. (1982). Histone acetylation increases thesolubility of chromatin and occurs sequentially over most of thechromatin. / . Biol. Chem. 257, 7336-7347.

Pfaffle, P., Gerlach, V., Bnnzel, L. and Jackson, V. (1990). In vitroevidence that transcription-induced stress causes nucleosomedissolution and regeneration. J. Biol. Chem. 265, 16830-16840.

Prior, C. P., Cantor, C. R., Johnson, E. M., Littan, V. C. and Allfrey,

V. G. (1983). Reversible changes in nucleosome structure andHistone H3 accessibility in transcriptionally active and inactivestates of rDNA chromatin. Cell 34, 1033-1042.

Ridsdale, A. J., Hendzel, M. J., Decluve, P. G. and Davie, J. R.(1990). Histone acetylation alters the capacity of HI histones tocondense transcriptionally active/competent chromatin. / . Biol.Chem. 265, 5150-5156.

Riggs, M. G., Whltaker, R. G., Neumann, J. R. and Ingram, V. M.(1977). n-butyrate causes histones modification in HeLa and Frienderythroleukaemia cells. Nature 268, 462^164.

Stein, A., Holley, K., Zeiiff, J. and Townsend, T. (1985). Interactionsbetween core histones and chromatin at physiological ionicstrength. Biochemistry 24, 1783-1790.

Van Holde, K. E. (1988). Chromatin (ed. Rich A.). Springer-Verlag,Berlin, New York, Heidelberg, London, Paris, Tokyo.

Vulaponteau, B., Lundell, M. and Martinson, H. (1984). Torsionalstress promotes the DNAase I sensitivity of active genes Cell 39,469^78.

Voordow, G. and Eisenberg, H. (1978). Binding of additional histonesto chromatin core particles. Nature 273, 446-448.

Weintraub, H. and Groudlne, M. (1976). Chromosomal subunits inactive genes have an altered conformation. Science 193, 848-856

Weintraub, H., Worcel, A. and Alberts, B. (1976). A model ofchromatin based upon two symmetrically paired half-nucleosomes.Cell 9, 409-417.

Weisbrod, S. and Weintraub, H. (1979). Isolation of a subclass ofnuclear proteins responsible for conferring a DNAase I sensitivestructure on globin chromatin. Proc. Nat. Acad. Sci U.S.A. 76,630-634.