Immunity, Vol. 2, 31 l-319, April, 1995, Copyright 0 1995 by Cell Press Combinatorial Regulation of Review I: General Aspects of Transcriptional Control

Patricia Ernst and Stephen T. Smaie Next month, we will describe our current understanding Howard Hughes Medical institute of transcriptional control mechanisms for the immunoglob- institute ulin u (Igu) heavy chain (Ernst and Smale, 1995 [next Department of Microbiology and immunology issue of /mmun~]). Transcriptional regulatory strategies University of California, Los Angeles have been studied in greater detail for this gene than for School of Medicine any other gene specific to the immune system, providing Los Angeles, California 99695-1662 some of the best experimental support for the combinato- rial theory. Several recent reviews have presented de- tailed descriptions of the control elements and binding The ceils that constitute the mammalian immune system proteins that regulate the Igu gene (Staudt and Lenardo, are derived from a multipotential stem cell through a com- 1991; Nelsen and Sen, 1992; Kadesch, 1992; Ghosh and plex and highly ordered sequence of events. After these Calame, 1995) and other expressed in the immune cells reach maturity, they carry out a wide variety of func- system (Leiden and Thompson, 1994; Winoto, 1991; Hag- tions when exposed to stimuli that elicit an immune re- man and Grosschedl, 1994; Clevers et al., 1993). We will sponse. Strict transcriptional regulation of numerous focus instead on conceptual advances in understanding genes is required throughout these developmental and how cell type-specific expression of the lgu gene is functional pathways. Thus, to define the molecular events achieved by the combined action of multiple regulatory associated with both normal and abnormal cellular pro- events. cesses, the mechanisms by which these genes are differ- entially transcribed must be elucidated. Overview To understand transcriptional regulation in the immune Appropriate transcriptional regulation of a given gene de- system, as in any complex cell system, a principle objec- pends on contributions from a variety of factors. Control tive must be to determine how a tremendous array of ex- elements within the core , regulatory promoter, pression patterns is dictated by a genome of limited size. and enhancers interact with DNA-binding proteins that di- If every gene with a unique expression pattern were regu- rectly influence the ability of RNA polymerase II (pol II) to lated by adedicated protein, a large fraction of the genome initiate RNA synthesis from a specific location and at a would be utilized solely for regulating transcription in a defined frequency. Coactivators and , which single cell lineage. Researchers began to consider this carry out protein-protein interactions with the DNA- issue about 25 years ago, shortly after Jacob and Monod binding proteins, appearto modulate interactions between (1961) first argued that the expression of protein-coding regulatory proteins and the general transcription machin- genes must be controlled by the interaction of tram+acting ery. Chromatin structure and more specialized elements regulatory factors with cis-acting DNA sequence ele- like silencers, locus control regions (LCRs), insulators, and ments. The resulting “combinatorial” theory of gene regu- matrix attachment regions (MARS) play crucial, but poorly lation (Gierer, 1973; Britten and Davidson, 1969; Geor- understood, roles in transcriptional regulation, in some giev, 1969) proposed that, to minimize the number of cases by modulating accessibility of transcription factors proteins required for gene regulation in an animal, most and pol II to the DNA. DNA methylation may play a further genes must be regulated by multiple proteins, each of role in regulation, by modulating either chromatin structure which plays a role in controlling the transcription of a vari- or the binding of specific transcription factors to DNA. A ety of genes with widely different expression patterns. The basic introduction to each of these areas follows, with re- expression of a given gene would therefore depend on cent review articles often cited, rather than primary re- the simultaneous interaction of a specific combination of search articles. regulatory proteins with the gene’s control elements. This review, which will be presented in two installments, Core Promoters will introduce our current knowledge of combinatorial con- We will consider as the core promoter the DNA sequence trol mechanisms to a general audience of immunologists. elements residing between approximately -35 to +35 rela- Here, we will provide an overview of fundamental issues tive to a transcription start site (Kollmar and Farnham, in the analysis of basal and regulated transcription in eu- 1993; Zawel and Reinberg, 1993; Hori and Carey, 1994; karyotes. Many of the issues presented here have not Smale, 1994). This region contains the sequences most been studied in detail for genes expressed in the immune likely to be contacted directly by pol II or other components system. Therefore, much of our attention focuses on clas- of the general transcrjption machinery. The primary roles sic examples from studies in other mammalian cell types of core promoter elementsare to participate in determining or in yeast. When appropriate, these examples are com- the precise location of the transcription start site and, in pared briefly with recent studies in the immune system. some cases, the’dlrection of transcription; to direct the By employing a diverse array of examples, we hope to formation of the preinltiation complex containing pol II and provide an appreciation of the types of control regions several general transcription factors; and to mediate tran- and regulatory events that are likely to contribute to gene scriptional activation byZsurrounding regulatory proteins. regulation in the immune system. In most genes, the core promoter elements responsible for the above activities do not play a direct role in regulated APyPy. The basal Inr appears to be recognized by two transcription, but in some genes regulatory factors appear independent proteins: a TAF within the TFIID complex to bind control elements within the core promoter region (Kaufmann and Smale, 1994; Purnell et al., 1994) and pot (e.g., Eichbaum et al., 1994). II (Carcamo et al., 1991). A plausible model for the initiation The DNA sequence element most commonly found of transcription from a TATA-less promoter containing an within the core promoters of mammalian protein-coding Inr isasfollows. ATAFwithin theTFIIDcomplexfirst recog- genes is the TATA box, located 25-30 bp upstream of the nizes the Inr. This recognition event directs the TBP sub transcription start site (Breathnach and Chambon, 1981; unit of TFIID to associate with the -30 region of the pro- Kollmar and Farnham, 1993; Smale, 1994). The TATA box moter in a sequence-independent manner. Following the is capable of independently directing a low level of tran- stable binding of TFIID to the core promoter, the remaining scription by pol II and is sufficient for directing activated steps leading to formation of a functional preinitiation com- transcription when an activator protein binds to an up plex and transcription initiation proceed by a similar mech- stream control element. These activities result from the anism and require a similar set of general transcription binding to the TATA box of a sequence-specific DNA bind- factors as with TATA-containing promoters. ing protein called TATA-binding protein (TBP; Hernandez, Several factors other than the TFIID complex and pol II 1993; Zawel and Reinberg, 1993; Hori and Carey, 1994). have been reported to interact with specific Inr sequences. Within a cell, the TBP involved in pol II transcription is a For example, the YYl protein binds to and activates tran- subunit of a multiprotein complex commonly called tran- scription from a subset of Inr elements containing a CCAT scription factor IID (TFIID), which contains several addi- sequence (Usheva and Shenk, 1994). TFII-I binds to other tional subunits known as TBP-associated factors (TAFs). Inr elements and recruits TBP to the core promoter (Roy The binding of TFIID to the TATA box initiates an ordered et al., 1993). Further experiments are needed todetermine cascade of events that leads to the formation of a func- how these activators function when bound to transcription tional preinitiation complex containing the general tran- start site sequences, which subsequently are incorporated scription factors TFIIA, TFIIB, TFIIE, TFIIF, TFIIH, TFIIJ, into a preinitiation complex containing the numerous poly- and pol II Qawel and Reinberg, 1993; Hori and Carey, peptides of the general transcription machinery. 1994). All but one of these general factors contain multiple Although numerouscorepromoterscontainaTATAbox, subunits, providing numerous potential targets for interac- an Inr, or both, many core promoters contain neither of tions with transcriptional regulatory proteins. Most of the these elements. The important components of these core basal factors essential for the formation of a preinitiation promoters are poorly understood, as are the mechanisms complex have been purified, and many of their genes have by which they lead to accurate transcription initiation. One been isolated. In addition, the approximate order by which example of a core promoter that may lack both TATA and the general factors assemble onto the promoter has been Inr elements is found in the FcyRlb gene, which is ex- determined. However, the precise functions of many of pressed primarily in cells of the myeloid lineage (Eichbaum the proteins during the initiation of RNA synthesis remain et al., 1994). The critical control element in this core pro- to be elucidated. moter is located 20 bp upstream of the transcription start In many promoters, the sequence of the TATA box is site and includes a recognition site for the PU.l protein, TATAAA. However, it appears that most A/T-rich se- which is preferentially expressed in myeloid and B cells. quences of 6 or more bp are capable of imparting TATA Another example is a class of promoters that contain sev- activity (Kollmar and Farnham, 1993; Smale, 1994). The eral transcription initiation sites, a high G/C content, and flexible sequence requirements result from the binding of multiple binding sites for ubiquitous TBP to the minor groove of DNA, where the protein primar- Spl (Smale, 1994). In these promoters, which often are ily contacts the phosphate backbone, resulting in an un- associated with ‘housekeeping” genes, Spl appears to usual bent DNA conformation (Kim et al., 1993a, 1993b). direct the formation of preinitiation complexes to a window This loose sequence requirement suggests that the activ- 40-100 bp downstream of its binding sites. Within that ity of a TATA box is strongly dependent on its proximity window, TFIID may direct preinitiation complex formation to other control elements. In other words, an A/T-rich se- at the DNA sequences that most closely resemble TATA quence at a random location within the genome is unlikely or Inr elements. to direct significant transcription, although the sequence One issue that remains unresolved is why core promot- may function as a strong TATA box in the context of a ers have evolved to contain widely varying structures, es- natural promoter. pecially since the mechanisms of initiation appear to be A second type of core promoter element that appears fairly similar. An explanation for core promoter heteroge- to be functionally analogous to the TATA box is called an neity may emerge from studies that have revealed a re- initiator (Inr; Kollmar and Farnham, 1993; Smale, 1994). quirement for specific core promoter structures during Although this element carries out the same functions as transcriptional regulation. For example, the p53 tumor- TATA in directing the formation of a preinitiation complex, suppressor protein directly represses transcription from in determining the location of the start site, and in mediat- TATAcontaining core promoters, but not from Inr-contain- ing activation by upstream activator proteins, it directly ing core promoters (Mack et al., 1993). In addition, the overlaps the transcription start site, rather than being lo- activity of the lymphocyte-specific terminal transferase cated 30 bp upstream. Functional Inr activity depends on a promoter depends on its Inr element, as the promoter can- loose consensus sequence of approximately PyPyA+,NT/ not function if a TATA box is inserted and the Inr elimi- Review 313

nated. Thus, the specific core promoter structure found in binds with highest affinity. In other families, however, in- a given gene is likely to play a critical role in transcriptional cluding the zinc finger family, there is little similarity be- regulation, not by interacting with cell type-specific pro- tween recognition sites for the different family members, teins in most cases, but by providing an appropriate con- largely because the key recognition amino acids are highly text for efficient activation or repression. variable among family members. The term “transcriptional activation domain” has been Regulatory Promotere and Enhancers used to refer to a wide variety of protein domains that An enormous number of studies have revealed that a ma- directly interact with components of the general transcrip jor contribution to precise transcriptional regulation is im- tion machinery, with proteins bound to nearby sites, or parted by the binding of sequence-specific DNA binding with or chromatin proteins. The mechanisms proteins to regulatory promoters and enhancers. For the by which activation domains stimulate transcription is a purposes of this review, the regulatory promoter will be major focus of current research (Hori and Carey, 1994; defined as the control region surrounding the core pro- Tjian and Maniatis, 1994). Although most activator pro- moter and within a few hundred base pairs of the transcrip- teins rely on specific activation domains, some proteins tion start site. An will be defined as a control also contribute to promoter or enhancer activity by bend- region found at a greater distance from the transcription ing DNA, which may modulate adjacent protein-protein start site, either upstream or downstream of the gene, or interactions (Grosschedl et al., 1994). Specific examples within an intron. Many enhancers can function from any of of transcriptional activation mechanisms will be described these locations and in either orientation. The distinctions for the lgu gene in the second review, next month. between regulatory promoters and enhancers have be- An important feature of many transcriptional regulatory come blurred because many proteins that interact with proteins is their ability to synergize (have multiplicative promoters or enhancers can function in either location. effects) in transcriptional activation. The mechanisms by Control elements found in an enhancer can often function which multiple regulatory proteins integrate their effects in the context of a promoter. Conversely, individual pro- are largely unknown. Some regulatory proteins may en- moter elements can often impart enhancer activity if hance the binding of another protein to an adjacent site, multimers of that element are inserted at a more distant either by clearing from that location or by location. At this time, the specific properties that allow a directly stabilizing binding. Alternatively, the various acti- control region to function from a great distance have not vators each may stimulate transcription by interacting with been determined. specific members of the general transcription machinery; one factor may interact with pol II, while another may inter- TranscriptIonal Activators act with TFIIB, etcetera. Multiple targets for the activators Proteins that activate transcription by binding to regulatory could lead to influences on multiple rate-limiting steps, promoters and enhancers generally contain modular which could result in synergistic activation. Another model structures, with distinct domains for DNA binding and tran- proposes that the activator may form a large multiprotein scriptional activation (Johnson and McKnight, 1989). complex that then transmits a single integrated signal to These proteins are grouped into families according to the the general transcription machinery at the core promoter, structure of their DNA-binding domain. Several classes inducing either the formation of a preinitiation complex or of DNA-binding domains have been described in higher the initiation of transcription. eukaryotes (Pabo and Sauer, 1992) including zinc finger In studying transcriptional activators, considerable at- and steroid receptor (Coleman, 1992) basic leucine zipper tention has been focused, not only on their mechanisms (Kerppola and Curran, 1991) helix-turn-helix (Harrison of action, but also on the mechanisms by which their func- and Aggarwal, 1990; Gehring et al., 1994) basic helix- tions are modulated. Although many activators are regu- loop-helix (Kadesch, 1992; Murre et al., 1994) Rel homol- lated at the transcriptional level, many other activators are ogy (Grilli et al., 1993) Ets homology (Wasylyk et al., regulated by posttranslational mechanisms. As exempli- 1993) Myb homology (Graf, 1992), high mobility group fied by the NF-xB and NFAT complexes, the subcellular (Grosschedl et al., 1994) MADS(Kaushal et al., 1994), and localization of an activator can be modulated, resulting in paired (Czerny et al., 1993) domains. Some DNA-binding sequestration from target genes (Liou and Baltimore, proteins do not fit into any of the defined families, including 1993; Israel, 1994). Furthermore, phosphorylation can af- several zinc-containing proteins that do not form struc- fect the DNA binding activity of a transcription factor either tures similar to those found in zinc finger proteins and positively or negatively. Alternatively, a protein may bind steroid receptors (see Vallee et al., 1991). Some of these to DNA, but be unable to activate transcription unless mod- broad classes have been further subdivided. For example, ified through phosphorylation (see Hunter and Karin, the homeodomain proteins define the predominant eu- 1992). Posttranslational regulation of the Rel family of tran- karyotic helix-turn-helix family; a subtype of the homeo- scription factors will be described in detail in an upcoming domain family is the POU homeodomain family (Herr et /mmunify review (S. Ghosh, submitted). al., 1988; Rosenfeld, 1991). Not surprisingly, members of some protein families bind Coactivatore to similar DNA sequences. For example, most members For the purposes of this review, coactivators (or adaptors) of the Ets family bind to a core sequence of GGA, with are proteins that are brought to a promoter or enhancer the surrounding sequences determining which protein(s) primarily by binding to DNA-binding proteins rather than by Immunity 314

binding directly to DNA. These proteins appear to bridge gene. A classic example of this type of repression is de- interactions between activators and basal factors or to scribed below, and other examples of transcriptional re- contribute activation domains that have additional func- pression will be described in the second review. tions. By definition, coactivators are necessary for acti- Repression through a sequence-specific DNA-binding vated transcription, but not for basal transcription. The protein has been studied in great detail in Saccharomyces TAFs within the TFIID complex are therefore prototypical cerevisiae, which exists as two haploid cell types, a and coactivators, in that they are brought to the promoter by a. In a cells, the cell type identity is maintained by the binding to TBP and are required specifically for activated repression of a-specific genes by a protein called a2 (or transcription (see Tjian and Maniatis, 1994). MATa2; see Herschbach and Johnson, 1993; Komachi et In addition to the TAFs, a few other cellular coactivators al., 1994; Cooper et al., 1994). The a2 protein, which is recently have been identified (e.g., Arany et al., 1994; Luo a member of the homeodomain family, binds cooperatively et al., 1992; Gstaiger et al., 1995; Strubin et al., 1995), with an activator, Mcml, to a control element found up including a coactivator that is important for lgu transcrip- stream of the TATA box in a variety of coordinately re- tion (see Ernst and Smale, 1995 [next issue of Immunity]). pressed genes. Genetic data have revealed that the re- However, a classic example of a protein that could be pression once ascribed to a2 depends on the activities of considered as a coactivator is the herpes simplex virus two corepressors, Tupl and Ssn6. Corepressors can be transactivator, VP16. This protein is recruited to specific defined as proteins that mediate repression when brought promoters primarily through protein-protein interactions to a promoter by protein-protein interactions. Tupl ap with the Ott-1 DNA-binding protein. At these promoters, pears to be the primary mediator of repression, as it can VP16 contributes potent activation domains that increase repress transcription when artificially brought to a pro- the eff iciency of preinitiation complex formation by binding moter via fusion to a heterologous DNA-binding domain to TFIIB and TBP and by recruiting TFIIB into the TBP- (Tzamarias and Struhl, 1994). Recent biochemical studies promoter complex (Stringer et al., 1990; Roberts et al., have shown that Tupl and Ssn6 associate with each other 1993). VP16 interacts not only with TBP and TFIIB, but and that Tupl couples the complex to the a2 protein (see also with TFIIH (Xiao et al., 1994) and a TAF (Goodrich Komachi et al., 1994; Tzamarias and Struhl, 1994). The et al., 1993). In this example, the sole function of Ott-1 precise mechanism of repression by this complex is not may be to mediate sequence-specific DNA binding, while known, but the targets of repression appear to be either VP16 contributes the activation functions. In general, co- a general transcription factor or a chromatin component activators have the potential to add another layer of combi- (Herschbach and Johnson, 1993; Cooper et al., 1994). natorial control to the simple scenario of cell type-specific For the repression described above, at least four differ- DNA-binding proteins determining the expression pattern ent proteins are required, with only one of the components of a gene. (a2) expressed in a cell type-specific manner. It is note- worthy that in addition to the a-specific genes, Tupl and and Corepressors Ssn6 are known to be required for repression of at least is often regulated by activators and co- four other sets of genes (Herschbach and Johnson, 1993) activators, but repressors and corepressors are likely to suggesting that these corepressors may interact with a make equally important contributions. Unfortunately, re- variety of sequence-specific DNA-binding proteins. Simi- pression mechanisms are even less well understood than lar mechanisms may be employed for transcriptional re- activation mechanisms. Several thorough reviews discuss pression in higher eukaryotes. transcriptional repression in eukaryotes and propose dif- The properties of the corepresaors described above, ferent conceptual and mechanistic classes (Levine and combined with the properties of coactivators discussed Manley, 1969; Renkawitz, 1990; Herschbach and John- earlier, are interesting to consider with regard to combina- son, 1993). In general, transcriptional repression can be torial gene regulation. By modulating the expression pat- divided into three broad categories. First, repression can terns of corepressors and coactivators, as well as of the occur by inactivation of an activator protein, which can be DNA-binding proteins that interact with them, a limited set accomplished by several distinct mechanisms: posttrans- of proteins may contribute to diverse expression patterns. lational modification of the activator, dimerization of the Furthermore, if differentially expressed coactivators and activator with a nonfunctional partner, competition for the corepressors interact with the same DNA-binding protein, binding site of the activator, or a direct -activator a single control element could mediate both activation and interaction that results in masking of the function of the repression of a given gene. activator. Second, repression can be mediated by proteins that tightly associate with general transcription factors and thereby inhibit the formation of a preinitiation complex Chromatin (Drapkin et al., 1993). These global repressors may play Most studies of gene regulation have focused on the mech- an integral role in the transcriptional activation process, anisms by which sequence-specific DNA-binding proteins in that some activators may stimulate the formation of a directly modulate transcription initiation by pol II. These preinitiation complex by displacing a global repressor. The studies have largely ignored the fact that, in eukaryotic third category of repression is mediated by a specific DNA cells, genes are assembled into chromatin. During the past element and DNA-binding protein, which act dominantly few years, genetic and biochemical studies have begun to repress both activated and basal transcription of a given to reveal the critical role that chromatin plays in modulating transcription (Grunstein, 1990; Felsenfeld, 1992; Lewin, sequences wrapped in nucleosomes (Kwon et al., 1994; 1994; Paranjape et al., 1994). lmbalzano et al., 1994); whether this is, in fact, the function In a mammalian interphase nucleus, DNA is incorpo- of the SWI-SNF complex in vivo remains to be deter- rated into a 10 nm diameter beads-on-a-string nucleoso- mined. Other biochemical studies in cell extracts from mal fiber, with each containing a core higher eukaryotes have begun to elucidate the precise octamer (with two molecules of l-MA, H2B, H3, and H4) nature of the interplay between transcription factors and and one linker histone, Hl (van Holde, 1989). At many nucleosomes (Paranjape et al., 1994; Lewin, 1994). Nota- loci, nucleosomes appear to be precisely positioned along bly, in vitro studies have suggested that some activator the DNA strand so that the same DNA sequences consis- proteins may play dual roles of disrupting chromatin and tently face either towards or away from the nucleosomal of direct transcriptional activation (Paranjape et al., 1994). core (thereby influencing the accessibility of a subset of Finally, many studies of the role of chromatin in mamma- control elements to transcription factors). The 10 nm lian gene regulation have focused on specialized control nucleosome fiber is incorporated into a higher order struc- elements, like silencers and LCRs, which, as described ture, referred to as the 30 nm filament. In addition to the below, appear to influence chromatin structure. histone proteins, several abundant nonhistone proteins, including various high mobility group proteins, are com- Silencers and LCRs monly associated with chromatin. Correlative evidence Silencers and LCRs are cis-acting regulatory regions that strongly suggests that genes within the “condensed” 30 appear, at least in some instances, to modulate transcrip nm chromatin filament must be’decondensed” to be com- tion by influencing chromatin structure through an ex- petent for transcription. The status of decondensed DNA tended DNA region. The best characterized examples of associated wtth nucleosomes is less clear, but a growing silencing arefound in S. cerevisiae (see Brand et al., 1985; body of evidence suggests that transcription initiation may Laurenson and Rine, 1992; Braunstein et al., 1993; Hecht require reconfiguration, rather than removal, of nucleo- et al., 1995) and Drosophila (Paro, 1993). S. cerevisiae somes (Grunstein, 1990; Lewin, 1994). strains typically possess copies of the two mating-type Studies in S. cerevisiae have provided the most convinc- genes, a and a, in repressed (silenced) loci termed HMLa ing evidence for a dynamic role of chromatin in gene regu- and HMRa (reviewed by Laurenson and Rine, 1992). A lation. Nucleosome depletion in vivo has been found to mating-type gene is activated only if the entire gene, in- result in transcriptional activation in a general manner cluding sequences required for transcriptional activation, (Grunstein, 1990). In addition, a multiprotein complex is transposed to the MAT locus during mating-type switch- called SWI-SNF has been defined through genetic and ing. The DNA sequence elements required for silencing biochemical approaches as one critical link between tran- flank the HML and HMR loci and contain an autonomous scriptional control and chromatin. This complex is required replication sequence and binding sites for two proteins, for activated transcription from a variety of genes and has one of which is called RAPl. Silencing of the HML and been shown to disrupt or reconfigure nucleosomes and HMR loci (and also of nearby telomeric heterochromatin) to stabilize the binding of transcription factors to nucleoso- depends on multiple trans-acting factors that do not di- mal DNA (Richard-Foy, 1994). Independent genetic stud- rectly bind to DNA, including SIR2, SIR3, and SIR4. ies in yeast have revealed that the N termini of Several studies have pointed to a model in which si- H3 and H4 play important roles in transcriptional activation lencer binding proteins and SIR proteins direct the forma- of a subset of genes (Grunstein, 1990; see below). Further- tion and maintenance of an inaccessible heterochromatin more, a strong correlation has been established between configuration. RAP1 appears to recruit the SIR proteins transcriptionally active genes and acetylation of N-ter- to the loci by direct protein-protein interactions (Moretti minal lysine residues in histone H4 (Hebbes et al., 1988; et al., 1994). Moreover, the N termini of histones H3 and Paranjape et al., 1994). H4 are needed for silencing and recently have been shown Integrating the above results, a model for transcriptional to associate with specific SIR proteins (Hecht et al., 1995). activation of a chromatin-associated locus is that a specific Silencing also has been strictly correlated with deacetyla- protein-DNA interaction within the locus leads to histone tion of core histones (Braunstein et al., 1993). Finally, in acetylation, which provides a tag for decondensation of cells and in isolated nuclei, the silenced loci are relatively the 30 nm filament. Precisely positioned nucleosomes as- inaccessible to DNA methylation and to nucleases (Thomp sociated with the decondensed chromatin are then desta- son et al., 1993; Loo and Rine, 1994). Together, these bilized or reconfigured by sequence-specific DNA-binding findings suggest that RAP1 recruits SIR proteins to the proteins that bind to the DNA in association with the SWI- loci, where the SIR proteins directly interact with histones. SNFcomplex. This latter event may make the locus acces- SIR-histone complexes may then be propagated through- sible toother transcription factors whose binding sites orig out the loci, leading to an inaccessible chromatin configu- inally faced the nucleosomal core, resulting in the direct ration. activation of transcription. activities have been detected in several genes Studies of the relationship between chromatin and tran- expressed in cells of the immune system (Wmoto and Balti- scription initiation in higher eukaryotic cells have lagged more, 1989; Sawada et al., 1994; Siu et al., 1994 and behind the studies in S. cerevisiae. Recently, however, a references therein). A recent example is a silencer found mammalian SWI-SNF complex has been found to facili- in an intron of the murine CD4 gene. In experiments em- tate the binding of activators and TBP to their cognate ployingtransgenicmice, Sawadaetal. (1994)andSiuet al. Immunity 316

(1994) found that the CD4 gene contains a T cell-specific sensitive sites are present in both a9 and y5 cells, the enhancer and promoter, but that those elements direct LCR most likely enhances transcription of both the a and transcription in both CD4+ and CD8’ subsets of mature 5 genes. Presumably, the LCR contributes to the expres- T cells. When a DNA fragment from the first intron was sion of each gene by functioning in concert with promoters, included in the reporter construct, expression was re- enhancers, and silencers, which are responsible for the stricted to the CD4+ subset. Thus, this intronic fragment differential regulation. contains a silencing activity that appears to play a critical rote in developmental regulation in T cells. Further studies are needed to determine whether silencing found in mam- Insulators and Matrix Attachment Regions malian cells occurs by a mechanism similar to that found As described above, control regions like enhancers, in yeast, or by a more direct repression mechanism. LCRs, and silencers are capable of influencing gene ex- LCRs, like silencers, appear to influence accessibility pression over long distances. Considering these long of agenetic locus, but LCRs appearto induce and maintain range effects, one issue that must be considered is how accessibility rather than inaccessibility (Felsenfeld, 1992; these control regions are prevented from influencing tran- Crossley and Orkin, 1993). The actual mechanism of LCR scription of adjacent loci. Two different classes of ragula- function remains poorly understood. LCRs were first iden- tory regions have been described that may be involved in tified during studies of the Bglobin locus in transgenic this function: insulators and MARS. mice (Felsenfeld, 1992; Crossley and Orkin, 1993). Like Insulators have been described flanking a Drosophila many transgenes, j3globin transgenes typically were ex- heat-shock locus (Kellum and Schedl, 1991,1992) and a pressed at low levels and the expression level was strongly chicken Bglobin locus (Chung et al., 1993). In the heat influenced by the site of insertion into the chromosome. shock locus, the regions are located at a consid- However, when the transgene contained a specific DNA erable distance both upstream and downstream of the lo- fragment from the distal end of the globin locus, which is cus, with each containing a pair of nuclease-hypersen- now known to encompass the LCR, high levels of position- sitive sites that surround a 250-350 bp nuclease-resistant independent (or integration site-independent) expression core. The insulators appear to be in the vicinity of junctions were observed. (In addition to high levels of position- between decondensed and condensed chromatin, which independent expression, some but not all LCRs impose presumably correspond to junctions between active and expression levels that are roughly proportional to the num- inactive loci. Like LCRs, the insulators were found to im- ber of integrated copies of the transgene.) Further analysis part position-independent expression to a heterologous of the j3globin LCR fragment revealed that it influenced gene. However, insulators appear to be distinct from LCRs chromatin structure (based on DNase I sensitivity) through because they do not enhance transcription and, in the heat a 199 kb region. In general, LCRs share features with shock locus, insulators are needed both upstream and cell-specific enhancers, in that they coincide with cell- downstream of the locus for their activity. Furthermore, specific DNase I hypersensitive sites and bind to typical the Bglobin insulator could prevent an LCR from activating transcription factors. Despite these similarities, LCRs are transcription if the insulator was placed between the LCR clearly distinct from enhancers because, in many in- and the promoter. Thus, the insulator appears to provide stances, they enhance transcription only when integrated a functional boundary for accessible or inaccessible chro- into a chromosome. Moreover, classical enhancers do not matin structures. impart high levels of position-independent expression in The study of MARS initiated with the idea that, within a transgenic mice. Further studies are needed to address eukaryotic nucleus, chromosomes are incorporated into the precise mechanism of LCR action, including the possi- higher order looped structures, with the loops fastened to bility that LCRs function in a manner analogous to the the intranuclear matrix (Gasser and Laemmli, 1987). yeast mating-type silencers, by modulating the ability of a These loops have been proposed to influence gene ex- specific SIR-like factor to interact with histones throughout pression by separating a chromosome into individual regu- the adjacent locus. latory domains. Numerous MARS have been isolated as LCRs appear to bs important for gene expression in the DNA fragments capable of associating with the nuclear immune system, having been identified in the macro- structures that remain following a stringent extraction and phage-specific lysozyme locus, the Ig heavy chain locus, wash with high salt or detergent (Phi-Van and Strtltling, the CD2 locus, and the T cell receptor (TCR) a15 locus 1999). MARS are A/T-rich sequences that often are located (Diax et al., 1994; Madisen and Groudine, 1994; Forrester at the boundaries of transcription units or in the vicinity et al., 1994 and references therein). In the TCR a/5 locus, of transcriptional enhancers. Several unique proteins as- the a and 5 gene segments are interspersed, yet the a and sociate with MARS (Gasser and Laemmli, 1987; Dickinson 5 genes are expressed in different cell types and during et al., 1992; Scheuermann and Chen, 1989). However, different stages of development. Diaz et al. (1994) identi- because MARS are defined by physical rather than func- fied a series of Tcell-specific DNase I hypersensitive sites tional characteristics, their actual functions may be hetero- at the 3’ end of the locus, which function as an LCR. A geneous. Indeed, in some cases, MARS have been found fragment containing these hypersensitive sites does not to function similarly to insulators or LCRs (McKnight et function as an enhancer in a transient transfection assay, al., 1992 and references therein). However, at least one but it imparts integration site-independent expression to MAR was shown to be distinct from the Drosophila heat a linked gene in transgenic mice. Since the DNase I hyper- shock insulator, in that it was not capable of insulating a Review 317

gene from the action of an upstream enhancer (Kellum PersPective and Schedl, 1992). The identification of MARS adjacent The information presented in this review reveals that multi- to intronic enhancers (e.g., Cockerill et al., 1987) further ple types of control regions and controlling events may demonstrates that MARS cannot always act as locus contribute to the precise regulation of a given gene. The boundaries. Additional studies are needed to establish the multiplicity of regulatory strategies is consistent with and, precise relationship between MARS, insulators, and LCRs. in fact, predicted by the combinatorial theory of gene regu- lation (Gierer, 1973; Britten and Davidson, 1989; Geor- giev, 1989). It is important to note, however, that most of Methylatlon these regulatory strategies (e.g., regulation by a silencer, In mammalian somatic cells, the CpG dinucleotides found LCR, MAR, coactivator, or DNA methylation) have been in tissue-specific genes often contain methylated cytosine studied in depth for only one or a few genes. Moreover, bases when transcriptionally inactive and unmethylated it is unlikely that any mammalian gene has been studied cytosines when transcriptionaliy active. This correlation in sufficient detail to reveal every mode of regulation for has suggested that methylation and demethylation may that gene. Thus, the number of regulatory strategies in- play critical roles in the regulation of tissue-specific genes, volved in the expression of a gene remains to be estab- with methylation inhibiting transcription either through the lished. binding of methyl-CpG-binding proteins or through the While focusing on the multiple types of regulatory re- inhibition of specific transcriptional activators (for review gions and regulatory events that are capable of influencing see Bird, 1992). The actual importance of methylation, gene expression, we have discussed only briefly a second however, remains unknown because it has been difficult level of combinatorial control: the complex interactions to distinguish cause from effect. In other words, does the between the multiple regulatory proteins that act on a sin- natural methylation state of a gene dictate its expression gle control region, such as a promoter or enhancer. Stud- level, or is the methylation state a result of the expression ies of transcriptional regulation in the immune system, level? Supporting the former possibility are studies show- beginning with the analysis of immunoglobulin gene regu- ing that in vitro DNA methylation of various genes can lation in the early 1989s have provided some of the best result in stable repression, and that demethylation, follow- experimental support for this level of combinatorial con- ing addition of Bazacytidine, can result in transcriptional trol. Our second review article will focus on studies of the activation. In contrast, the latter possibility is supported Igu heavy chain gene todescribe in detail the mechanisms by studies demonstrating that gene inactivation can lead by which multiple proteins can contribute to gene regula- to the passive acquisition of DNA methylation and that tion through interactions with an Igu promoter and en- transcriptional activation can lead to demethylation. The hancer. critical issue of whether de novo DNA methylation and demethylation of natural genes within an animal play im- portant causative roles in transcriptional regulation has Acknowledgmento not been adequately addressed. This topic is further com- We are extremely grateful to Ft. Grosschsdl, M. Grunstein, C. Miceli, plicated by the finding that the methylation state of a tis- C. Murre, S. Plevy, R. Sen, and A. Winoto for valuable comments. sue-specific gene in cultured cells may not accurately re- We apologize to those whose work could not be included because of flect its methylation state in the animal (Bird, 1992). space limitations. Work in our laboratory is supported by Pubflc Health Services grant DK43728. P. E. was supported by Public Health Ser- A better understanding of the role of DNA methylation vices training grants GM07185 and CA09120 and by a Unlvemtty of has emerged from studies of the lgx gene by Bergman California Cffice of the President Fellowship. S. T. S. is an Assistant and colleagues (Lichtenstein et al., 1994). This study be- Investigator with the Howard Hughes Medical Institute. gan with the observation that an in vitro methylated plas- Referencea mid containing a rearranged K gene was demethylated upon transfection into B cell lines where the gene is active, Arany, Z., Sellers, W. Ft., Livingston, D. M., and Eckner, R. (1994). but remained methylated following transfection into non- ElA-associated ~300 and CREB-associated CBP belong to a con- lymphoid cells. A subsequent mutant analysis found that sewed family of coactivatom. Cell 77, 799-800. the K intronic enhancer was necessary, but not sufficient Bird, A. (1992). The essentials of DNA methylatton. Cell 70, 5-9. for the tissue-specific demethylation; additional regions of Brand, A. H., Breeden, L., Abraham, J., Sternglanz, R., and Nasmyth, the intron were also required. Furthermore, these intronic K. (1985). Characterization of a %ilencer’ in yeast: a DNA sequence with properties opposite to those of a transcriptional enhancer. Cell sequences induced demethylation in the apparent ab- 41.41-49. sence of activated transcription. These findings, therefore, Braunstein, M., Rose, A. B., Holmes, S. G., Allis, C. D., and Broach, distinguish the control elements needed for demethylation J. R. (1993). Transcriptional silencing in yeast is associated with rs- from the control elements needed for transcriptional en- duced nuclsosome acetylation. Genes Dev. 7,592-804. hancement, strongly suggesting that demethytation is not Breathnach, R., and Chambon. P. (1991). Organization and expres- simply a result of transcriptional activation. Further analy- sion of eukaryotic split genes coding for proteins. Annu. Rev. Biochem. 50,349-383. sis of the mechanism of demethylation and of the role Britten, R. J., and Davidson, E. H. (1989). Gene regulation for higher of the important K intronic sequences during murine B cells: a theory. Science 165, 349-357. cell development should provide valuable information re- Carcamo, J.. Buckbinder, L., and Reinberg, D. (1991). The initiator garding the role of the DNA methylation in transcriptional directs the assembiy of a transcription factor lIDdependent transcrip- control. tion complex. Proc. Natl. Acad. Sci. USA 88, 9052-8058. Immunity 310

Chung, J. H., Whit&y, M., and Felsenfeld, G. (1993). A 5’ element Grunstein, M. (1999). Histone function in transcription. Annu. Rev. of the chicken 5-globin domain serves as an insulator in human ery- Cell. Biol. 6, 843-879. throid cells and protects against position effect in Drosophila. Cell 74, Gstaiger, M., Knoepfel, L., Georgiev, O., Schaffner, W., and Hovens, 595-514. C. M. (1995). A B-cell coacttvator of octamer-binding transcription fao Clevers, H. C., Oosterwegel, M. A., and Georgopoulos, K. (1993). tors. Nature 373, 389-382. Transcription factors in early T-cell development. Immunol. Today 14, Hagman, J., and Grosschedl, R. (1994). Regulation of gene expression 591596. at early stages of B-cell differentiation. Curr. Opin. Immunol. 6, 221% Cockerill, P. N., Yuen, M. H., and Garrard, W. T. (1987). The enhancer 230. of the immunoglobulin heavy chain locus is flanked by presumptive Harrison, S. C., and Aggarwal, A. K. (1990). DNA recognkion by pro chromosomal loop anchorage elements. J. Biol. Chem. 282, 5394- teins with the helix-turn-helix motif. Annu. Rev. B&hem. 59, 933- 5397. 969. Coleman, J. E. (1992). Zinc proteins: , storage proteins, tran- Hebbes, T. R., Thorne, A. W., and Crane-Robinson, C. (1988). Adirect scription factors, and replication proteins. Annu. Rev. Biochem. 67, link between core histone acetytation and transcriptionally active chro- 097-846. matin. EMBO J. 7, 1395-1402. Cooper, J. P., Roth, S. Y., and Simpson, R. T. (1994). The global Hecht, A., Laroche, T., St&l-Bolsinger, S., Gasser, S. M., and transcriptional regulators, SsnB and Tupl, play distinct roles in the Grunstein, M. (1995). Histone H3 and H4 N termini interact with SIR3 establishment of a repressive chromatin structure. Genes Dev. 8, and SIR4 proteins: a molecular model for the formation of heterochro- 1499-1410. matin in yeast. Cell 88. 583-592. Crossley, M., and Orkin, S. H. (1993). Regulation of the beta globin Hernandez, N. (1993). TBP, a universal fac- locus. Curr. Opin. Genet. Dev. 3, 232-237. tor? Genes Dev. 7, 1291-1399. Czerny, T., Schaffner, G.. and Busslinger, M. (1993). DNA sequence Herr, W., Sturm, R. A., Clerc, R. 0.. Corcoran, L. M., Baltimore, D., recognition by Pax proteins: bipartite structure of the paired domain Sharp, P. A., Ingraham, H. A., Rosenfeld, M. G., Finney, M., Ruvkun, and its binding site. Genes Dev. 7, 2949-2061. G., and Horvitz, H. R. (1988). The POU domain: a large conserved Diaz, P., Cado, D., and Winoto, A. (1994). A in region in the mammalianpit-7, octl, ocf-2, and Ceeno~ase/egans the T cell receptor c/S locus. Immunity 1,207-217. uric-86 gene products. Genes Dev. 2, 1513-1516. Dickinson, L. A., Joh, T., Kohwi, Y., and KohwiShigematsu, T. (1992). Herschbach, B. M., and Johnson, A. D. (1993). Transcriptional repres- A tissue-specific MAWSAR DNA-binding protein with unusual binding sion in eukaryotes. Annu. Rev. Cell. Biol. 9, 479-509. site recognition. Cell 70, 831-845. Hori, R., and Carey, M. (1994). The role of activators in assembly of Drapkin, R., Merino, A., and Reinberg, D. (1993). Regulation of RNA RNA polymerase II transcription complexes. Curr. Opin. Genet. Dev. polymerase II transcription. Curr. Opin. Cell Biol. 5, 489-476. 4,238-244. Eichbaum, Q. G., lyer, R., Raveh, D. P., Mathieu, C., Gekowitz, R. A. Hunter, T., and Karin, M. (1992). The regulation of transcription by (1994). Restriction of interferon gamma responsiveness and basal ex- phosphorylation. Cell 70, 375-397. pression of the myeloid human Fc gamma RI b gene is mediated by Imbalzano, A., Kwon, H., Green, M. R., and Kingston, R. E. (1994). a functional PU.l site and a transcription initiator consensus. J. Exp. Facilitated binding of TATA-binding protein to nucleosomeal DNA. Na- Med. 179,1985-1996. ture 370, 481-485. Felsenfeld, G. (1992). Chromatin as an essential part of the transcrip Israel, A. (1994). NF-AT comes under control. Nature 369, 443-444. tion machinery. Nature 355, 219-223. Jacob, F., and Monod, J. (1981). Genetic regulatory mechanisms in Forrester, W. C., van Genderen, C., Jenuwein, T., and Grosschedl, the synthesis of proteins. J. Mol. Biol. 3, 318-358. R. (1994). Dependence of enhancer-mediated transcription of the im- Johnson, P. F., and McKnight, S. L. (1999). Eukaryotic transcriptional munoglobulin mu geneon nuclear matrix attachment regions. Science regulatory proteins. Annu. Rev. Biochem. 58799-839. 285, 1221-1225. Kadesch, T. (1992). Helix-loop-helix proteins in the regulation of im- Gasser, S. M., and Laemmli, U. K. (1997). A glimpse at chromosomal munoglobulin gene transcription. Immunol. Today 73, 31-39. order. Trends Genet. 3, 18-22. Kaufmann, J., and Smale, S. T. (1994). Direct recognition of initiator Gehring, W. J., Qian, Y. Q., Billeter, M., Furukubo-Tokunaga, K., elements by a component of the transcription factor IID complex. Schier, A. F., Resendez-Perez, D., Affolter, M., Otting, G., and Genes Dev. 8,821-9X Wuthrich, K. (1994). Homeodomain-DNA recognition. Cell 78, 21 l- 223. Kaushal, S., Schneider, J. W., NadalGinard, B., and Mahdavi, V. (1994). Activation of the myogenic lineage by MEF2A, a factor that Georgiev, 0. P. (1989). On the structural organization of and induces and cooperates with MyoD. Science 266, 1238-1249 the regulation of RNA synthesis in animal cells. J. Theor. Biol. 25, 473-499. Kellum, R., and Schedl, P. (1991). A positiin~ffect assay for bound- aries of higher order chromosomal domains. Cell 64, 941-959. Ghosh, S., and Calarne, K. (1995). Factors regulating immunoglobulin transcription. In lmmunoglobulin Genes, T. Honjo, F. Alt, and T. H. Kellum, R., and Schedl, P. (1992). A group of scs elements function Rabbits, eds. (London: Academic Press), in press. as domain boundaries in an enhancer-blocking assay. Mol. Cell. Biol. 12,2424-2431. Gierer, A. (1973). Molecular models and combinatorial principles in cell differentiation and morphogenesis. Cold Spring Harbor Symp. Quant. Kerppola, T. K., and Curran, T. (1991). Transcription factor interac- Biol. 38, 951-981. tions: basics on zippers. Curr. Opin. Struct. Biol. 7, 71-79. Goodrich, J. A., Hoey, T., Thut, C., Admon, A., and Tjian, R. (1993). Kim, J. L., Nikolov, D. B., and Burley, S. K. (1993a).Cocrystalstructure Drosophila TAFIWO with both a VP16 activation domain and basal of TBP recognizing the minor groove of a TATA element. Nature 385, transcription factor TFIIB. Cell 75, 519-530. 520427. Graf, T. (1992). Myb: a transcriptional activator linking proliferation Kim, Y., Geiger, J. H., Hahn, S., and Sigler, P. B. (1993b). Crystal and differentiation in hematopoietic cells. Curr. Opin. Genet. Dev. 2, structure of a yeast TBPiTATA-lmx complex. Nature 385, 512-529 249-255. Kollmar, R., and Farnham, P. J. (1993). Site-specific initiation of tran- Grilli. M., Chiu, J. J., and Lenardo, M. J. (1993). NF-kappa B and Rel: scription by RNA polyrnerase II. Proc. Sot. Exp. Biol. Med. 203. 127- participants in a multiform transcriptional regulatory system. Int. Rev. 139. Cytol. 743, l-82. Komachi, K., Redd, M. J., and Johnson, A. D. (1994). The WD repeats Grosschedl, R., Giese, K., and Pagel, J. (1994). HMG domain proteins: of Tupl interact with the homeo domain protein a2. Genes Dev. 8, architectural elements in the assembly of nucleoprotein structures. 2857-2867. Trends Genet. 10, 94-l 99. Kwon, H., Imbalzano, A., Khavari, P. A., Kingston, R. E., and Green, Review 319

M. R. (1994). Nucleosome disruption and enhancement of activator Sawada, S., Scarborough, J. D., Killeen, N., and Littman, D. R. (1994). binding by a human SWllSNF complex. Nature 370, 477481. A lineage-specific transcriptional silencer regulates CD4 gene expres- Laurenson, P., and Rine, J. (1992). Silencers, silencing, and heritable sion during T lymphocyte development. Cell 77, 917-929. transcriptional states. Microbial. Rev. 58, 543-589. Scheuermann, R. H.. and Chen, U. (1989). A developmental-specific Leiden, J. M., and Thompson, C. B. (1994). Transcriptional regulation factor binds to suppressor sites flanking the immunoglobulin heavy- of T-ceil genes during T-cell development. Curr. Opin. Immunol. 8, chain enhancer. Genes Dev. 3, 1255-1288. 231-237. Siu, G., Wurster, A. L., Duncan, D. D., Soliman. T. M., and Hedrick, Levine, hf., and Manley, J. L. (1989). Transcriptional repression of S. M. (1994). A transcriptional silencer controls the developmental eukaryotic promoters. Cell 59, 405499. expression of the CD4 gene. EMBO J. 73.357-79. Lewin, B. (lQQ4). Chromatin and gene expression: constant questions, Smale, S. T. (1 Q94). Core promoter architecture for eukaryotic protein- but changing answers. Cell 79. 397498. coding genes. In Transcription: Mechanisms and Regulation, R. C. Conaway and J. W. Conaway, eds. (New York: Raven Press), pp. 83- Lichtenstein, M., Keini. G., Cedar, H., and Bergman, Y. (1994). B cell- 81. specific demethylation: a novel role for the intronic K chain enhancer sequence. Cell 78,913-923. Staudt, L. M., and Lenardo. M. J. (1991). lmmunoglobulin gene tran- scription. Annu. Rev. Immunol. 9, 373-398. Liou, H. C., and Baltimore, D. (1993). Regulation of the NF-kappa B/rel transcriptfon factor and I kappa B inhibitor system. Curr. Opin. Stringer, K. F., Ingles, C. J., and Greenblatt, J. (199g). Direct and Cell Bioi. 5, 477487. selective binding of an acidic transcriptional activation to the TATA-box factor TFIID. Nature 345, 783-798. Loo, S., and Rine, J. (1994). Silencers and domains of generalized repression. Science 284, 1788-1771. Strubin, M., Newell, J. W., and Matthias, P. (lQ95). OBF-I, a novel B cell-specific coactivator that stimulates immunoglobulin promoter Luo, Y., Fujii, H., Gerster, T., and Roeder, R. G. (1992). A novel B activity through association with octamer-binding proteins. Cell 80, cell-derived coactivator potentiates the activation of immunoglobulin 497-808. promoters byoctamer-binding transcription factors. Cell 71,231-241. Thompson, J. S.. Hecht, A.. and Grunstein, M. (1993). Histones and Mack, D. H., Vartlkar. J., Pipas, J. M., and Laimins, L. A. (1993). the regulation of heterochromatin in yeast. Cold Spring Harbor Symp. Specific repression of TATA-mediated but not initiator-mediated tran- Quant. Biol. 58, 247-258. scription by wild-type p53. Nature 383, 281-293. Tjian, R.. and Maniatis, T. (1994). Transcriptional activation: acomplex Madisen, L., and Groudine, M. (1994). Identification of a locus control puzzle with few easy pieces. Cell 77, 5-8. region in the immunoglobulin heavy-chain locus that deregulates c-myc expression in plasmacytoma and BurkitYs lymphoma cells. Tzamarias, D., and Struhl, K. (1994). Functional dissection of the yeast Genes Dev. 8,2212-2228. Cyc8-Tupl transcriptional co-repressor complex. Nature 389, 758- 781. McKnight, R. A., Shamay, A., Sankaran, L., Wall, R. J., and Hennig hausen, L. (IQQ2). Matrix-attachment regions can impart position- Usheva. A. and Shenk, T. (1994). TATA-binding protein-independent independent regulation of a tissue-specific gene in transgenic mice. initiation: YYl, TFIIB, and RNA polymerase II direct basal transcription Proc. Natt. Acad. Sci. USA 8Q, 8943-8947. on supercoiled template DNA. Cell 78, 1115-l 121. Moretti, P., Freeman, K., Coodly, L., and Shore, D. (1994). Evidence Valiee, 8. L., Coleman, J. E., and Auld, D. S. (1991). Zinc fingers, zinc that a complex of SIR proteins interacts with the silencer and telomere clusters, and zinc twists in DNA-binding protein domains. Proc. Natl. binding protein RAPI. Genes Dev. 8, 2257-2289. Acad. Sci. USA 88, 999-l 003. Murre, C., Bain, G., van Dijk, M. A., Engel, I., Furnari, 8. A., Massari, van Holde. K. E. (1989). Chromatin. (New York: Springer-Verlag). M. E., Matthews, J. R., Quong, M. W., Rivera, R. R., and Stuiver, M. H. Wasyfyk. B., Hahn, S. L., and Giovane, A. (1993). The Ets family of (1994). Structure and function of helix-loop-helix proteins. Biochem. transcription factors. Eur. J. Biochem. 277, 7-18. Biophys. Acta 7218, 129-135. Winoto, A. (1991). Regulation of the early stagesof T-cell development. Nelsen, B., and Sen, R. (1992). Regulation of immunoglobulin gene Curr. Opin. Immunol. 3, 199-293. transcription. Int. Rev. Cytol. 733, 121-149. Wfnoto, A., and Baltimore, D. (1989). ap lineage-specific expression Pabo, C. 0.. and Sauer, R. T..(l992). Transcription factors: structural of the a T cell receptor gene by nearby silencers. Cell 59, 849-855. families and principles of DNA recoghltion. Annu. Rev. Biochem. 81. Xiao, H., Pearson, A., Coulombe, B., Truant, R., Zhang, S., Regier, 1953-1095. J. L., Treizenberg, S. J., Reinberg, D., Flares, 0.. Ingles, C. J., and Paranjape, S. M., Kamakaka, R. T., and Kadonaga, J. T. (1994). Role Greenblatt, J. (1994). Binding of basal transcription factor TFIIH to the of chromatin structure in the regulation of transcription by RNA poly- acidic activation domains of VP18 and ~53. Mol. Cell. Biol. 74,7013- merase II. Annu. Rev. Biochem. 83,285-297. 7024. Pam, R. (1993). Mechanisms of heritable gene repression during de Zawel, L., and Reinberg, D. (1993). Initiation of transcription by RNA velopment of Drosophila. Curr. Opin. Cell Biol. 5, 999-1005. polymerase II: a multi-step process. Prog. Nucl. Acid Res. Mol. Biol. Phi-Van, L., and Str&ling, W. H. (19Q9). Association of DNA with nu- 44.87-108. clear matrix. Prog. Mol. Subcell. Biol. 77, l-l I. Purneil, B. A., Emanuel, P. A., and Gilmour, D. S. (1994). TFIID se- quence recognition of the initiator and sequences farther downstream in Drosophila class II genes. Genes Dev. 8, 839-842. Renkawitz, R. (1999). Transcriptional repression in eukaryotes. Trends Genet. 8, 192-197. Richard-Foy, H. (1994). A way into the packaging. Nature 370. 417- 418. Roberts, S. G. E., Ha, I., Matdonado, E., Reinberg, D., and Green, M. R. (1993). Interaction between an acidic activator and transcription factor IIB is required for transcriptional activation. Nature 383, 741- 744. Rose&Id, M. G. (1991). POUdomaln transcription factors: pou-er-ful developmental regulators. Genes Dev. 5.897-907. Roy, A. L.. Malik, S., Meisterernst. M., and Roeder, R. G. (1993). An alternative pathway for transcription initiation involving TFII-I. Nature 385, 355-359.