Typical Transcriptional Activator

A typical activator has two essential functions: DNA binding and transcriptional activation (Ptashne, 1988). Many activators have separate protein domains to confer these two functions. The DNA binding domain of an activator enables the protein to recognize specific DNA sites located within enhancers. There are different families of DNA binding domains that form distinct structures to recognize DNA (Garvie and Wolberger, 2001). These domains tend to bear names that depict their structural and/or functional properties or follow their founding member's names. For example, a zinc-finger DNA binding domain uses zinc to maintain its three-dimensional structure required for DNA recognition. A basic region-leucine zipper (bZIP) domain contains a basic region (that contacts DNA) and a leucine zipper (that forms dimers). A homeodomain is a conserved 60-aa DNA binding domain initially identified in proteins encoded by Drosophila homeotic genes, which play critical roles in specifying segment identity. A Rel homology domain is a DNA binding domain that bears the name of its founding member Rel.

Most DNA binding domains, including all the examples mentioned above, recognize short, specific DNA sequences by making elaborate contacts with the bases in the major groove of the DNA double helix (Garvie and Wolberger, 2001; Patikoglou and Burley, 1997). Others, e.g., the high-mobility-group (HMG) domain, recognize DNA sequences by interacting with the minor groove (Travers, 2000).2 While most DNA binding domains recognize DNA sites as dimers (e.g., bZIP and Rel family members), others can bind as monomers (e.g., some homeodomain proteins). For proteins that bind DNA as dimers, many can form both homodimers and heterodimers with other family members. Heterodimer formation can increase the repertoire of DNA sequences recognized by a given family of transcription factors. Many activators can bind DNA cooperatively with one another, which can increase the stability of the protein complexes formed at the enhancers (Adams and Workman, 1995; Ma et al., 1996).

The activation domain of an activator plays critical roles in stimulating transcription. Unlike DNA binding domains that require elaborate structures for DNA recognition, activation domains tend to be short sequences often with very limited sequence complexity. There are different types of activation domains, which are named after their sequence characteristics, such as acidic, glutamine-rich, proline-rich, and alanine-rich. For the acidic class of activating sequences, it was estimated that 1% of the peptides encoded by random DNA sequences (from the E.coli genome) can activate transcription when fused to a DNA binding domain (Ma and Ptashne, 1987c). This finding further highlights the "relaxed" specificity between activation sequences and their targets (Ma, 2004), a feature that contrasts the interaction mode between DNA binding domains and DNA sites.

One important finding in understanding activator functions was the demonstration of the modular nature of activators, i.e., the DNA binding and activation functions are provided by separable domains (Brent and Ptashne, 1985; Keegan et al., 1986). This finding suggested that DNA binding per se was insufficient for activation in eukaryotes (Brent, 2004; Ptashne, 2004). Subsequent demonstration that activation sequences are short, simple peptides (Hope and Struhl, 1986; Ma and Ptashne, 1987b; Ma and Ptashne, 1987c) further supported the notion that activation domains achieved their functions by touching other proteins (also see below). The demonstration of activators' modular nature has also enabled researchers to determine easily whether a

TBP, a GTF that can bind specific DNA sequences, also makes contacts with the minor groove (Kim et al., 1993; Nikolov et al., 1992).

particular transcription factor has an activation function, through experiments of assaying the activity of hybrid proteins containing the factor's fragments fused to a heterologous DNA binding domain.

The Recruitment Model

What does an activator do to stimulate transcription? As discussed above, an essential domain of an activator is its DNA binding domain, which brings the activator to DNA sites in an enhancer. But the DNA binding domain itself is insufficient to activate transcription; an activation domain is required for activation. Activation domains have been shown to have the ability to interact with a wide array of proteins, many of which are components of the transcription machinery, including the GTFs (e.g., TBP, TFIIB, TFIIE, and TAFs), co-factors and chromatin modifying/remodeling complexes (Malik and Roeder, 2000; Naar et al, 2001; Narlikar et al., 2002; Orphanides et al., 1996; Peterson and Workman, 2000; Ptashne and Gann, 1990). All these (and other) interactions lead to a unified final outcome: increased level of transcription. According to a well-established recruitment model, the ultimate and only goal of these interactions is to bring the transcription machinery, in particular RNAP, to the promoter (Ptashne and Gann, 1997; Stargell and Struhl, 1996). During the activation process, a DNA loop may be formed as a result of the interaction between the activator bound at the enhancer and the transcription machinery at the core promoter (Ptashne, 1986).

Several lines of evidence support the recruitment model. First, for many genes, the GTFs and RNAP are absent from their promoters unless the genes are turned on by activators (Chatterjee and Struhl, 1995; Klein and Struhl, 1994; Li et al, 1999). Second, activators are known to interact with components of the transcription machinery; as noted above, one property common to the activation domains is that they tend to have the ability to interact with multiple target proteins (Bryant and Ptashne, 2003; Ma, 2004; Ptashne and Gann, 1990). Finally, in a set of "artificial recruitment" experiments that provided pivotal support to this model, it was shown that transcription can be elicited by artificially attaching components of the transcription machinery to a DNA binding domain (Chatterjee and Struhl, 1995; Farrell et al, 1996; Gonzalez-Gouto et al, 1997; Nevado et al, 1999; Xiao et al, 1995). In these artificial recruitment experiments, the requirement of a classical activator is bypassed, i.e., the activator is no longer needed for transcription. This suggested that, at least for the promoters tested, all the functions that are provided by the activators could be substituted by physically bringing the RNAP holoenzyme to the promoter. It should be noted that, since the eukaryotic DNA is wrapped in nucleosomes, the recruitment model may also cover situations in which the chromatin modifying or remodeling factors recruited by activators facilitate the assembly of the transcription machinery by increasing the accessibility of promoter DNA. As discussed below, the recruitment model, though attractive due to its simplicity and experimental support, does not exclude other possibilities of how activators may stimulate transcription.

Now with this broad description of what a typical activator looks like and how it may activation transcription according to one model, we will discuss several additional issues to further our understanding of activator functions and activation mechanisms. Readers should refer to other chapters in this volume that discuss specific examples of activators in greater details.

Composite Activators

Although a typical activator contains both an activation domain and a DNA binding domain, sometimes these two domains can reside on separate proteins. For example, the herpes simplex virus (HSV) activator VP 16 dos not bind to DNA, but rather, it is brought to DNA by interacting with other DNA-binding proteins (Triezenberg et al, 1988). The activation domain of VP 16 can also activate transcription when directly linked to a DNA binding domain (Sadowski et al, 1988). This finding demonstrated that an activation domain can be brought to DNA by distinct, but interchangeable, means, either directly binding to DNA (through its linked DNA binding domain) or interacting with other DNA-binding proteins. This concept was further demonstrated by the creation of an artificial composite activator (Ma and Ptashne, 1988). The yeast repressor protein GAL80 inhibits the activation function of GAL4 by interacting with and masking its activation domain (Johnston et al, 1987; Lue et al., 1987; Ma and Ptashne, 1987a). When GAL80 was attached to an activation domain, the hybrid GAL80 protein, which itself cannot bind DNA, gained an ability to activate transcription, but only through a GAL4 derivative that could interact with both DNA and GAL80 (Ma and Ptashne, 1988). The concept that an activation domain can be brought to DNA through protein-protein interactions led to the proposal of the yeast two-hybrid system (Fields and Song, 1989). This powerful genetic system has allowed researchers to dissect proteinprotein interactions and to identify proteins' interacting partners (Bai and Elledge, 1996; Fields and Sternglanz, 1994; Ma, 2000).

Transcriptional activators that do not bind DNA but interact with other DNA-binding proteins are sometimes also referred to as co-activators. But it may be useful to make a distinction between these non-DNA binding activators and the "true" co-activators that play more general roles in transcription. Unlike non-DNA binding activators, which are gene-specific, co-activators of the latter class (e.g., CBP and Swi-Snf complexes) play important roles in facilitating the actions of many activators. Some of these general co-activators are components of the RNAP holoenzyme (Myers and Kornberg, 2000; Ranish and Hahn, 1996).

The concept that an activation domain can be brought to its action site, the vicinity of a gene's promoter, through multiple means can be further extended to activators that bind to RNA sequences. One such example is the HfV activator Tat, which is discussed in further detail in another chapter of this volume. Relevant to this discussion is the finding that the RNA-binding activator Tat can also activate transcription from DNA sites when fused to a DNA binding domain (Southgate and Green, 1991), further illustrating that an activation domain can be brought to the vicinity of a promoter through distinct, but interchangeable, mechanisms.

Conformational Changes

The artificial recruitment experiments mentioned above support the notion that the ultimate and only function of activators is to bring RNAP to the promoter. It is known that the preinitiation complex undergoes several conformational changes before RNAP actually initiates transcription (Carey and Smale, 2000). For example, the promoter DNA is significantly bent and unwound upon TBP binding (Kim et al, 1993; Nikolov et al., 1992). In addition, the DNA double helix at the transcription start site becomes unpaired, or melted, to form a bubble prior to transcription initiation by RNAP (Giardina and Lis, 1993; Wang et al., 1992). In one study, it was shown that activators can change the conformation of the TFIIA-TFIID-TATA complex and such a conformational change is necessary and sufficient for activation in an in vitro system (Chi and Carey, 1996). Thus, conformational changes of the transcription machinery represent potential steps that can also be targeted by transcriptional activators.

Initiation vs. Elongation

Although transcription initiation is a critical step that can be stimulated by many activators, other steps of transcription, such as elongation, can also be activated. For example, the Drosophila heat shock gene hsp 70 already has the transcription machinery loaded at its promoter even before heat shock (induction) (Rougvie and Lis, 1988). In fact, RNAP is able to transcribe the 5' region of the gene prior to induction, but it fails to transcribe through the gene (Rasmussen and Lis, 1993; Rasmussen and Lis, 1995). Upon induction, the transcriptional activator HSF stimulates elongation by RNAP, enabling it to complete transcription through the gene.

Many proteins (or complexes) have been identified that play important roles in facilitating transcription elongation, and some of these factors represent targets for activators (Sims et al., 2004a). For example, experiments in Drosophila suggested that the elongation factor P-TEFb is recruited (likely by the activator HSF) to the heat shock loci to facilitate transcription elongation upon heat shock induction (Lis et al., 2000). In addition, in vitro experiments using the human hsp70 gene demonstrated that the Swi-Snf complex was recruited by the human activator HSF1 to facilitate transcription elongation through the chromatin template of the gene (Brown et al., 1996). As discussed elsewhere in this volume, the HIV Tat activator stimulates transcription elongation by recruiting the elongation factor P-TEFb (Mancebo et al., 1997; Zhou et al., 1998; Zhu et al., 1997). Together, these examples highlight the importance of the elongation step in transcriptional activation.

In this context, it should be noted that recent studies in both yeast and Drosophila suggest that transcription silencing can work at a step after the assembly of the transcription machinery (Breiling et al., 2001; Dellino et al, 2004; Sekinger and Gross, 2001). In other words, the mere presence of the transcription machinery (including the RNAP itself) assembled at the promoter does not necessarily equate to productive transcription of the gene.


One of the characteristic features of transcriptional activation is synergism. Synergy refers to the situation where the transcription level achieved by multiple activators is higher than the sum of the levels by individual factors separately. Synergy can arise from different mechanisms. In the simplest case, it can be due to cooperative binding of activators to multiple sites in the enhancer. This is obvious particularly if the activators are at limiting (sub-saturating) concentrations. An enhanceosome model has been proposed that further emphasizes the role of multiple activators for activation (Merika and Thanos, 2001; Thanos and Maniatis, 1995). According to this model, different activators, including those that play architectural roles, are together required to form a stable complex at the enhancer for efficient transcriptional activation.

Synergy can also be achieved even when activators are at saturating levels. This particular form of synergy, which can be demonstrated readily under in vitro conditions, provides useful insights into how activators work. It suggests that activators can contact multiple targets in the transcription machinery (Ptashne and Gann, 1998). If all activator molecules contacted the same target through the same surface, then increasing the number of activator molecules should only increase transcription level in an additive, rather than synergistic, manner.

Studies to compare the roles of different activators suggest that synergy may represent a consequence of combinatorial actions of activators that work on distinct steps of transcription (Blau et al., 1996). By comparing the RNAP density along a gene, it is possible to gain information about which step, initiation or elongation, an activator may stimulate. Using this and other analyses, Blau et al. concluded that, while some activators (e.g., Spl and CTF) work primarily on the initiation step, others (e.g., Tat) work primarily on the elongation step (Blau et al., 1996). Another class of activators (e.g., VP16, p53 and E2F1) can work on both initiation and elongation steps. An analysis of these activators revealed that synergy was only achieved between those that work on different steps of transcription (Blau et al., 1996).

Interplay Between DNA Binding and Activation Functions

It is well established that, in general, the DNA binding and transcriptional activation domains of an activator are physically separable. It should be emphasized, however, that the functions provided by these two domains are interconnected. First, an activation domain cannot exert its activating effect unless it is brought to DNA, either through a physical link to a DNA binding domain or through other means such as interacting with another DNA-binding protein. Second, some activators, e.g., the glucocorticoid receptor and MyoD, have DNA binding and activation functions conferred by single protein domains (Davis et al., 1990; Schena et al., 1989). Furthermore, several studies have suggested that the DNA binding properties of an activator can be influenced by its activation function (Bunker and Kingston, 1996; Tanaka, 1996). In particular, it was observed that activators that stimulated transcription strongly bound DNA better than those that activated weakly. It has been proposed that the interaction between the activation domain and the transcription machinery can help the binding of not only the transcription machinery to the promoter but also the activator to its DNA sites. For activators that work by recruiting chromatin remodeling or modifying complexes, the increased DNA accessibility is beneficial to not only GTFs but also activators themselves. The connection between the activation and DNA binding functions may also contribute to how activator gradients, such as Drosophila Bicoid, stimulate transcription in a concentration-dependent manner (Driever et al., 1989; Fu et al., 2003; Zhao et al., 2003; Zhao et al., 2002).

Activation vs. De-repression

One of the major differences between eukaryotes and prokaryotes is that eukaryotic genomes are packaged into nucleosomes. Nucleosomes can impede DNA binding of transcription factors and GTFs, thus repressing transcription (Narlikar et al., 2002; Peterson and Workman, 2000; Wu, 1997; Wu and Grunstein, 2000). One of the questions regarding mechanisms of eukaryotic transcription activation is how much is due to de-repression and how much is due to "real" activation. It is well established that genes can be de-repressed when histones are depleted from cells (Han and Grunstein, 1988). To further obtain insights into the role of histones in gene activation, Wyrick et al. used the microarray strategy to determine the profiles of gene expression upon hi stone H4 depletion (Wyrick et al., 1999). The authors found that 15% of the genes exhibited de-repressed (increased) expression in response to the removal of histone H4. Genes that are located near telomeres tend to be more sensitive to histone depletion than genes located elsewhere. These results show that depletion of histone can lead to gene-specific de-repression. The authors also found that histones did not play a generally repressive role for all genes, since the majority (75%) of genes appeared to be insensitive to histone depletion. Interestingly, 10% of the genes in yeast had reduced expression upon histone depletion, suggesting that histones and nucleosomes may also play positive roles in transcription (Wyrick et al., 1999).

Activation and Cellular Memory

In some cases the effect of transcriptional activators can be maintained or inherited even after the activators themselves are no longer present. One such example is homeotic gene expression in Drosophila (Levine et al, 2004; Orlando, 2003). During early embryogenesis the homeotic genes respond to transcription factors that are encoded by gap and pair rule genes. The active and silent states of these genes are subsequently maintained by proteins encoded by the trithorax group (trxG) and Polycomb group (PcG) genes, respectively. trxG and PcG proteins form co-factor complexes that work through DNA elements called Polycomb response elements (PREs). In an elegant study, an isolated PRE, Fab-7, was shown to be able to maintain the active state of a linked reporter construct that had been activated by a transiently expressed activator (Cavalli and Paro, 1998). In other words, the reporter gene remained on even after the activator itself was no longer present. Intriguingly, such memory can be transmitted in an activator-independent manner to subsequent generations through female (but not male) germline. Recent studies show that components in both PcG and trxG complexes contain histone methyltransferase (HMT) activities with different specificities/preferences for different lysine residues in histone tails (Levine et al., 2004; Lund and van Lohuizen, 2004; Sims et al., 2003). It is thought that distinct histone methylation patterns represent cell memory systems to maintain the active and silent states of homeotic genes (Orlando, 2003; Simsei al., 2003).

In yeast, genes that are transcribed recently are also marked by a specific pattern of histone methylation (Hampsey and Reinberg, 2003). This is achieved by the recruitment of the HMT Setl to the genes by the elongating RNAP (Ng et al., 2003). Interestingly, the Setl-mediated histone methylation pattern persists for some time even after the genes are no longer transcribed. Unlike the long-term memory of active genes mediated by trxG proteins in Drosophila (which can last for several generations), Setl-mediated marking of recently active genes in yeast is only short term (up to several hours). In addition, while the consequence of the trxG-mediated marking is to maintain the genes on, the yeast Setl-mediated system only marks the recently transcribed genes without actually keeping them on. Interestingly, yeast Setl is also involved in the long-term memory of gene silencing (Bryk et al., 2002; Krogan et al., 2002).

Another case of activator-induced memory is noteworthy in this context. This is an extremely short-term memory, which lasts only through the initial activation process itself. Under some in vitro conditions, transcriptional activators can induce conformational changes of the preinitiation complex. Interestingly, one study demonstrated that an activator-induced conformational change persisted, and led to the completion of the transcription process, even after the activator itself was removed (Chi and Carey, 1996). This result further illustrates the importance of conformational changes in transcriptional activation.

Activator-repressor Switches

Although this chapter deals primarily with activators and activation mechanisms, it should be noted that many transcription factors can often work as either activators or repressors in a context-dependent manner (Ma, 2005). For example, many transcription factors that mediate signal transduction processes work as repressors in the absence of the signals but as activators in the presence of the signals. In addition, the concentrations and posttranslational modifications of a transcription factor can affect its ability to either activate or repress transcription. The presence of other nearby DNA binding proteins on DNA, as well as the availability and concentration of co-factors, can also influence the behavior of a transcription factor. See a recent review article for further details (Ma, 2005).

Short Distance and Long Distance Actions

One of the questions in eukaryotic gene activation concerns actions at long distances. Enhancers in higher eukaryotes have the ability to exert their effects even when they are located many kilobases away from the promoters (Blackwood and Kadonaga, 1998). There are no specific definitions of short distance vs. long distance, but for our discussion we can consider short distance as anything up to a few hundred base pairs and long distance greater than one kilobase (Blackwood and Kadonaga, 1998; Dorsett, 1999). The mechanisms for activation at short or long distances may be fundamentally similar in that they are both achieved through a network of protein-protein interactions and alterations of chromatin structure. But activation at a long distance (e.g., 50-60 kilobases) faces two additional challenges that are less relevant to activation at a short distance (e.g., 100-200 bp). First, how can promoters and enhancers communicate through such long distances? Second, how does an enhancer "choose" to activate one promoter, but not another one that is also within its reach?

Proteins called facilitators have been proposed to promote the interaction between enhancers and promoters that are separated by long distances (Bulger and Groudine, 1999; Dorsett, 1999). One such example is a Drosophila protein called Chip (Morcillo et al., 1997; Torigoi et al., 2000). It is thought Chip can interact with proteins that may bind throughout the genome, such as homeodomain proteins, thus bringing enhancers closer to promoters through the formation of a series of loops (Dorsett, 1999).

Recent studies suggest that the efficiency (and specificity) of the communication between enhancers and promoters can also be augmented by DNA elements located near the core promoters. These elements have been called tethering elements (Bertolino and Singh, 2002; Calhoun et al., 2002). In one study, it was shown that the POU domain of Oct-1 bound to DNA sites near a promoter enables the promoter to respond to a distant enhancer (Bertolino and Singh, 2002). Interestingly, the POU domain itself does not work as a classical activator because it cannot activate transcription. It was suggested that the POU domain of Oct-1 recruits the TFIID complex to the promoter, so that the promoter becomes poised for activation by an enhancer at a distance (Bertolino and Singh, 2002).

The specificity of long-distance communication between enhancers and promoters can be regulated by different mechanisms (Blackwood and Kadonaga, 1998). First, the tethering elements mentioned above can selectively facilitate the communication between a promoter and one, but not another, enhancer (Calhoun et al., 2002). Second, in some cases promoters can compete with each other for an enhancer, thus the enhancer preferentially communicates with the strong promoter, while ignoring the weak promoter (Foley and Engel, 1992; Sharpe et al., 1998). Finally, insulator elements can prevent "unwanted" communications between enhancers and promoters thus encouraging "wanted" interactions; an insulator is a DNA element that can block the communication between an enhancer (or a silencer) and a promoter when the insulator is located between them, but not when it is located outside the enhancer-promoter unit (Kuhn and Geyer, 2003; West et al., 2002).

Recent studies reveal that the Drosophila genome contains organized domains—some as large as 200 kilobases — that contain many genes with similar expression profiles (Spellman and Rubin, 2002). How genes within these large domains are coordinately regulated is currently unclear. It is proposed that each of these domains may contain some higher order control elements (Calhoun and Levine, 2003), such as the recently discovered global control region (GCR) for the mouse HoxD complex (Spitz et al., 2003; Zuniga et al.,

2004). In this context it is noted that UASs in yeast generally do not work at distances greater than several hundred base pairs (also see de Bruin et al., 2001). It is evident metazoans have evolved mechanisms to facilitate long-distance enhancer-promoter communications and accommodate the increased complexity of gene regulation.

0 0

Post a comment