Histone Deacetylase--A Regulator of Transcription

Histones are DNA binding proteins that bind without regard to the sequence of the DNA and help determine the three-dimensional organization of the chromatin into chromosomes. A paper in this week's issue (Taunton et al., p. 408) reports the cloning of histone deacetylase, an ezyme that modifies histones by removing an acetyl group. In his Perspective, Wolffe explains how the action of this enzyme can lead to the regulation of the transcription of certain genes.

steroids termed brassinosteroids, and that these steroids may play a role in the regulation of gene expression by light.
A conclusive demonstration of this role is provided by Szekeres et al. (4). These authors show that a pleiotropic phenotype arising from mutation of the Arabidopsis CPD gene, which includes disregulation of light-responsive genes, dwarfism, male sterility, and activation of stress-response genes, can also be traced to a defect in brassinolide synthesis. The DNA sequence of the CPD gene reveals it to be a member of the cytochrome P450 class of enzymes, with distinct sequence identity to P450s involved in mammalian steroid hormone biosynthesis. Because steroid intermediates in the brassinolide pathway with a 23-hydroxyl group successfully reverse the phenotypes arising from mutation of the CPD gene, whereas those that lack this modification do not, it seems likely that the CPD enzyme catalyzes the 23-hydroxylation of cathasterone to form teasterone, which is thereafter converted into brassinolide. Szekeres et al. (4) go on to show that brassinolide will overcome the pathologies associated with at least five other Arabidopsis mutations (detl, cop], several fus mutations, dim], and axr2), each of which lead to disruptions in light-mediated development. These results implicate brassinolide as a major player in the growth and differentiation of Arabidopsis.
The studies of Li et al. (1) and Szekeres et al. (4) reveal the tip of an iceberg that will affect disciplines ranging from botany to biochemistry to endocrinology. We can look forward to the elucidation of several processes including the mechanism of intracellular signaling through which brassinosteroids work, the role of steroid reductases and P450 monooxygenases in the biosynthesis of this widely distributed class of compounds, and the cross-talk that must occur between light-activated regulatory systems and those that respond to steroid hormones, and perhaps we will also gain insight into nonreceptor-based actions of steroid hormones, as it is rumored that plants lack DNA sequences common to mammalian steroid hormone receptors (5). Finally, the development of therapeutic inhibitors of steroid 5a-reductase is a thriving area in the pharmaceutical industry (3). At least some of these inhibitors come from plant sources (6), and it is conceivable that by acting on DET2-like enzymes, or other enzymes in the pathway, they serve as endogenous regulators of brassinosteroid synthesis. The seeds of the Arabidopsis findings will sprout much future research! Histones, nuclear proteins that interact with DNA to form nucleosomes, are essential for both the regulation of transcription and the packaging of DNA within chromosomes. The stability and positioning of chromatin structures determine whether nucleosomes repress (1) or activate transcription (2). Mutation of individual histones in vivo alters the general organization of chromatin throughout the eukaryotic nucleus, yet the concomitant changes in gene expression are highly selective (3). This selectivity appears to depend on the assembly of specific nucleoprotein architectures. Within these structures, transcriptional regulators make direct contact with specific domains of individual core histones (4,5). Targeting of these "chromatin organizers" to particular genes, through protein-protein or protein-DNA interactions, can account for the selective impact of histone mutations on gene expression. This conceptual framework only provides a static image of gene regulation. In reality, events are much more dynamic, as highlighted in a report in this week's issue of Science (6) describing the cloning of one of these organizers, histone deacetylase. A major source of dynamic variation in chromatin structure in vivo is histone acetylation. Acetylated lysine residues in the NH2-terminal tail domains of nucleosomal histones serve as landmarks for transcriptionally active chromatin within the chromosome (7). Hyperacetylation of the histones reduces their ability to constrain the path of DNA within chromatin, resulting in allosteric changes in nucleosomal conformation (see figure, part A) (8), destabilization of intemucleosomal contacts (9), and an increase in the accessibility of nucleosomal DNA to transcription factors (10). The elimination of histone acetylation is correlated with transcriptional silencing ( 11). The amount of histone acetylation is determined by an equilibrium between histone acetyltransferases and deacetylases. Thus, chromatin structure could be reversibly modulated to activate or silence transcription by targeting histone acetyltransferases or deacetylases to a particular gene. This model now receives strong support from the recent purification and molecular characterization of a human deacetylase with remarkable identity (-60%) to the Saccharomyces cerevisiae transcriptional regulator Rpd3p (6).
Rpd3p is a global transcriptional regulator required for target genes to achieve maximal transcriptional efficiency (12). Without Rpd3p, both the activation and 371 Histone Deacetylase: A Regulator of Transcription Alan P. Wolffe repression of regulated genes is less effective. Thus, the range of transcriptional regulation achieved in vivo is amplified by Rpd3p. Although they display dramatic evolutionary conservation, neither the human deacetylase nor Rpd3p has recognizable functional motifs. Rpd3p does not bind to DNA, nor does it have biochemically defined interactions with other proteins. However, genetic experiments do indicate a close functional relation with another global transcriptional regulator, Sin3p (12,13). RPD3 and SIN3 participate in the same transcriptional regulatory functions and appear to be components of one pathway. Both genes are required for the regulation of inducible genes responding to extemal signals (PH05), cell differentiation (SPO1 1 and SP013), and cell type (HO, TY2, and STE6). Sin3p is not itself a DNA binding protein but contains paired amphipathic helices and tetratricopeptide repeats that are known motifs for protein-protein interactions. Sin3p is proposed to interact with bona fide yeast DNA binding proteins (14 geting of Sin3p by fusion to a DNA 1 domain will direct transcriptional sion (15). Thus, Rpd3p and Sin3p ft altemately as components of co-actim co-repressor complexes that are targ particular genes by DNA binding tra tion factors. How might these cont functional requirements be realized? An attractive model builds c known consequences for chromatin ture of histone acetylation. For t i majority of genes regulated by Sin' regulator acts as a repressor (15). Rpd3p could augment the actiN Sin3p by directing the equilibrium 1 histone acetylation toward the deace state with the concomitant stabiliza nucleosomal and higher order structu figure,  1). Tar-also facilitate the association of other rebinding pressive transcriptional regulators within a repres-chromosomal domain (4). inction The role of Rpd3p in transcriptional acvator or tivation is more difficult to explain; how-'eted to ever, recognition of the architectural role of inscrip-histones provides a solution. Once trantrasting scription factors are recruited to a particular promoter, the specificity of their interac-Dn the tions might depend on a tightly constrained 1 struc-chromatin environment (2,5). It is also he vast possible that the influence of Rpd3p might 3p, the be more indirect: Rpd3p might be required . Thus, for the general packaging of DNA into revity of stricted chromatin environments. Chromaevels of tin structure likely limits the number of tylated sites a regulatory molecule has to search beition of fore a productive interaction occurs (16). In ire (see the absence of histone deacetylation, a :n may larger number of nonproductive interaccontains Hypemotes, which were created by the cyberlibrarians of Stanford University's HighWire Press and which take the reader directly to WWW sites of immediate relevance to the topic under discussion. The editors of Science invite you to examine the Enhanced Perspectives, available at http://sciencemag. aaas.org/science/content/current/ e-perspectives/. tions likely occurs, diverting transcription factors and RNA polymerase from the gene of interest. Thus, histone deacetylation within the bulk of chromatin that contains many inactive genes could facilitate the selective recruitment of the transcriptional machinery to a specific active gene. A key point emphasized by these new results is that chromatin 9 structure modulates gene activity. As many in vitro biochemical experiments on naked DNA demonstrate, genes can be regulated in the absence of chromatin. Howie struc-ever, the range of regulation the core achieved in this way is much ctions of smaller than that in the chromoial struc-somal environment. RPD3 is not estails with sential for transcriptional activathe core tion or repression, but it has a sigptionally nificant impact on regulation of desr(yel) particular genes by amplifying actianscripvation and repression mechanisms.

the core
The molecular characterization ther with of the histone deacetylase (6) ity of the opens a door to understanding e). Tranhow dynamic chromatin structural transitions can be targeted to particular chromosomal regions. Such targeting reflects the functional differentiation of the chromosome. We might anticipate that other enzymatic activities that modify chromatin structure-such as histone acetyltransferases and kinases-will exhibit a comparable functional localization. Communication between architectural proteins, such as the histones, and the transcriptional machinery is an essential component of life in the nucleus.