Three classes of general cofactors are typically involved in gene activation to facilitate the communication between gene-specific transcription factors and components of the general transcription machinery. These general cofactors include TBP-associated factors (TAFs) found in TFIID, Mediator associated with the RNA polymerase II carboxy-terminal domain, and upstream stimulatory activity (USA)-derived positive cofactors (PCI, PC2, PC3, and PC4) and negative cofactor 1 (NCI). Among these general cofactors, only TFIID has core promoter-binding activity. While TBP and some TAF components of TFIID are able to recognize distinct core promoter elements, the largest subunit of TFIID, TAF1, also possesses enzymatic activities, which can posttranslationally modify transcriptional components and histone proteins. The enzymatic activities of TAFs constitute a unique feature of TFIID in core promoter recognition and in activator-dependent transcription, distinct from that of TBP. TFIID also interacts directly with a variety of activators and facilitates the recruitment of RNA polymerase II to the promoter region. Similar to TAFs, Mediator and USA-derived components are all capable of repressing basal transcription when activators are absent, and stimulating transcription in the presence of activators. In this chapter, we will discuss mainly in the human system how TAFs, Mediator, USA-derived cofactors, and a unique TBP-interacting negative cofactor 2 (NC2) regulate the initiation step of transcription, following the action of histone acetyltransferases and chromatin remodeling factors to expose the promoter region.

Transcription Components Modulating PIC Assembly

The nucleation pathway for preinitiation complex (PIC) formation involves TFIID binding to the core promoter, which specifies the site for the initiation of transcription, mainly through direct contact with the TATA box, initiator element (Inr), or downstream promoter element (DPE) by specific subunits of TFIID. Following promoter recognition by TFIID, the remaining general transcription factors (GTFs) and RNA polymerase II (pol II) then assemble on the TFIID-bound promoter either via a defined order of entry: TFIIA, TFIIB, pol II/TFIIF, TFIIE, and TFIIH, or via a preassembled pol II holoenzyme pathway (see Chapter by Hou and Chiang). Assembly of this promoter-bound complex is sufficient for a basal level of transcription observed in the absence of transcriptional activators. However, in vivo, transcriptional activators, which typically contain a DNA-binding domain (DBD) for gene targeting and an activation domain (AD) for contacting promoter-bound factors, are necessary for modulating the efficiency of PIC assembly. These activator-dependent transcription events often require general cofactors, such as TAFs, Mediator and USA-derived components, functioning as bridging molecules to transmit regulatory signals to the general transcription machinery (Fig.4.1).

Corresponding Author: Cheng-Ming Chiang, Tel: (216) 368-8550, Fax: (216) 368-3419, E-mail: [email protected]

Fig.4.1 General cofactors serve as molecular bridges in activator-dependent transcription. General cofactors (TAFs, Mediator, USA) are required for transducing signals between gene-specific activators and components of the general transcription machinery. An activator normally contains a DNA-binding domain (DBD) contacting specific DNA sequences and an activation domain (AD) interacting with general cofactors or with components of the general transcription machinery. It is of note that TAFs normally function as an integral part of TFIID, not as a free entity in mammalian cells as drawn here.

Fig.4.1 General cofactors serve as molecular bridges in activator-dependent transcription. General cofactors (TAFs, Mediator, USA) are required for transducing signals between gene-specific activators and components of the general transcription machinery. An activator normally contains a DNA-binding domain (DBD) contacting specific DNA sequences and an activation domain (AD) interacting with general cofactors or with components of the general transcription machinery. It is of note that TAFs normally function as an integral part of TFIID, not as a free entity in mammalian cells as drawn here.

TFIID Recognition of Core Promoter Elements

TFIID is a multiprotein complex comprised of TATA-binding protein (TBP) and approximately a dozen TBP-associated factors (TAFs). That TBP and some TAF components of TFIID bind distinct core promoter elements classifies TFIID as a core promoter-binding factor. The TBP subunit of TFIID contacts the TATA box allowing TFIID to recognize TATA-containing promoters, and the interaction between TAF-Inr and TAF-DPE also confer TFIID the ability to recognize TATA-less promoters.

A: TBP Recognition of the TATA Box

The TATA box, with a consensus sequence TATA(A/T)A(A/T), is recognized by the C-terminal region of TBP, which is phylogenetically conserved and is made up of approximately 180 amino acids (Hernandez, 1993; Nikolov and Burley, 1994; Burley and Roeder, 1996). The crystal structures of yeast TBP in complex with the TATA box of the yeast CYC1 promoter (Kim et al., 1993b) as well as Arabidopsis

TBP (Kim et al., 1993a) and human TBP (Nikolov et al., 1996) bound to the TATA sequence of the adenovirus major late promoter, all indicate that the DNA is severely distorted upon TBP binding. In the TATA-bound state, TBP resembles a molecular "saddle" with a pair of "stirrups" flanking the DNA-binding surface that helps bend the DNA. This saddle-shaped TBP molecule is composed of four a-helices and ten (3-strands (Nikolov et al., 1992), which are organized into a bipartite DNA-binding surface with each half consisting of five antiparallel P-strands located on the concave underside of TBP that straddles the DNA and onea-helix forming the side stirrup and the other a-helix situated at the convex upperside that interacts with other transcription factors (Kim et al., 1993a; Kim et al., 1993b; Nikolov et al., 1996). This concave surface of human TBP, with two phenylalanine residues (Phe-284 and Phe-193) situated at its outermost edges, binds the TATA box in the minor groove of the DNA double helix and induces sharp kinks (more than 80°) into the DNA via intercalation of Phe-284 at the 5' end and Phe-193 at the 3' end of the TATA box (Nikolov et al., 1996). Moreover, amino acid residues 190 to 194 and 281 to

285 of human TBP, which make up the two side stirrup loops, further accentuate the bending of the DNA (Juo et al., 1996; Nikolov et al., 1996). These structural studies revealing the detailed molecular interactions leading to a widening of the minor groove while compressing the major groove in order to bend the DNA helix backward are consistent with the observation that recombinant human and yeast TBP are able to induce bending of the TATA element derived from the adenovirus major late promoter in gel mobility shift assays using permuted DNA fragments (Horikoshi et al., 1992).

In contrast, the N-terminal region of human TBP, which is highly divergent among species and appears dispensable for the assembly of TFIID complexes (Zhou et al., 1993), possesses a contiguous stretch of 38 glutamine residues (amino acids 58-95) and three imperfect Pro-Met-Thr (PMT) repeats (amino acids 142-150) preceding the C-terminal core region of TBP. Although not directly involved in activator-dependent transcription by pol II (Zhou et al., 1993), this N-terminal region of human TBP is necessary for the recruitment of small nuclear RNA-activating protein complex (SNAPc) to the U6 promoter transcribed by RNA polymerase III (pol III) and seems to inhibit TATA binding by the C-terminal region of TBP on both pol III- (Mittal and Hernandez, 1997) and pol II-dependent promoters (Zhao and Herr, 2002). Since TBP binding to the TATA box is in fact a two-step process involving an initial binding of TBP to the TATA box without bending the DNA and followed by a slow transition into a more stable bent TATA-TBP complex, it is interesting to find that deletion of the glutamine-rich domain and PMT repeats in the N-terminal inhibitory region promotes formation of the stable bent TBP-DNA complex and that TFIIB's ability to enhance TBP binding to the TATA box is likely due to induced stabilization of the bent TATA-TBP complex following TFIIB binding to the solvent-exposed surface on the convex side of the TBP core (Zhao and Herr, 2002).

B: A utoregulation of TFIID-Promoter Complex Formation

As commonly observed with many DNA-binding proteins, TBP also exhibits nonspecific DNA-binding activity sometimes leading to the formation of nonproductive transcription complexes on scattered AT-rich sequences. To prevent spurious transcription events initiating at nonpromoter sequence elements, TBP may form a homodimer that is inactive in DNA binding or associate with TAFs to reduce nonspecific DNA-binding activity of TBP and concurrently increase its specificity toward TATA-containing promoters.

Formation of TBP homodimers was initially observed in the crystal structure of Arabidopsis TBP, in the absence of DNA (Nikolov et al., 1992), and later confirmed by biochemical analysis using gel filtration/glycerol gradient profile of mouse TBP (Kato et al., 1994) and chemical cross-linking studies with human TBP (Coleman et al., 1995) and yeast TBP (Jackson-Fisher et al., 1999). Dimerization of TBP is mediated through extensive contacts between the concave surfaces of each TBP monomer, thereby masking the DNA-binding domain also located in the concave region. Clearly only the monomeric form of TBP can bind to the DNA, as evidenced by the structural analysis of the TBP-TATA complex (Kim et al., 1993a; Kim et al., 1993b; Nikolov et al., 1996). In yeast, these contacts are located in the deepest part of the concave surface, including amino acid residues N69, V71, V122, T124, N159, V161, V213, and T215, since substitutions of these amino acids individually with a bulky charged arginine residue result in destabilization of TBP dimers (Kou et al., 2003) and, as shown with the V161R mutant, also significantly reduce the half-life of TBP in vivo (Jackson-Fisher et al., 1999). Not surprisingly, the ability of TBP to form dimers also leads to the formation of TFIID homodimers, which can be detected by size exclusion column chromatography following chemical cross-linking and appears to be inactive in TATA recognition (Taggart and Pugh, 1-996).

The DNA-binding activity of TBP is likewise subject to negative regulation by TAF1, the largest subunit of TFIID (Kokubo et al., 1993). At the N-terminus of Drosophila TAF1 are two separate regions able to contact TBP: amino acid residues 11-77 (TAND1, for TAF N-terminal domain 1) interacting with the concave underside of TBP to block TATA recognition (Liu et al., 1998) and amino acid residues 82-156 (TAND2) binding to the convex surface of TBP to compete with TFIIA that would otherwise facilitate TBP-TATA complex formation (Kokubo et al., 1998). Interestingly, the solution structure of Drosophila TAND1 bound with yeast TBP shows a remarkable resemblance to the structure of the TBP-TATA complex (Liu et al., 1998), as TAND1 exhibits an arch-shaped surface contacting the concave underside of TBP through both hydrophobic and electrostatic interactions. The negatively charged side chains of TAF1 (Asp-29, Glu-31, Glu-51, Glu-70, and Asp-73) interact with the conserved lysine and arginine residues of TBP that are also used to contact the phosphate backbone of the TATA sequence. Competition for the same binding surface underlies the mechanism for TAF 1-mediated inhibition of TBP binding to the TATA box (Kokubo et al., 1998), which may also account for reduced TATA-binding and transcription activity of TFIID in comparison with that of TBP in vivo and in vitro (Ozer et al., 1998; Wu and Chiang, 1998; Wu et al., 1998). Likewise, TAND2-mediated inhibition of TFIIA binding to TBP is due to competition between TAF1 and TFIIA for the same conserved positive charge residues on the convex surface of TBP (Kokubo et al., 1998). This TAF1-mediated effect on TBP-TFIIA and TBP-TATA interactions appear to be functionally conserved, as the same observation is also seen with experiments performed with homologous yeast and human proteins (Kokubo et al., 1998; Ozer etal., 1998; Banik et al., 2001).

C: The Dual Function of BTAF1 in Regulating TBP-TATA Complex Formation

Other than interacting with TAFs, TBP also forms a distinct complex with a functional property similar to that of TFIID. This TFIID-like protein complex, found in the PI 1 0.3 M KC1 (or "B") fraction (see Chapter by Hou and Chiang), is named B-TFIID which is consisted of TBP and BTAF1 (Timmers et al., 1992). As observed with TAF1, human and yeast BTAF1, formerly named TAFn170 in humans (Timmers et al., 1992; van der Knapp et al., 1997), Motl in yeast (Poon et al., 1994) and 89B helicase in Drosophila (Goldman-Levi et al.,

1996), also bind the concave and convex surfaces of TBP likely in a reversible manner (Pereira et al., 2001). This concave-binding region, located at the N-terminal amino acid residues 290-381 of BTAF1, not only hinders TBP-DNA complex formation, but also blocks TAF1 interaction with the concave underside of TBP (Pereira et al., 2001). While the N-terminus of BTAF1 interacts with TBP, the C-terminus allows BTAF1 to induce the dissociation of TBP-TATA complexes in an ATP-dependent manner (Chicca et al., 1998). The ATPase domain, which resides in the carboxy-terminus of BTAF1, has a conserved signature DEGH box within the Walker A motif, which classifies BTAF1 as a member of the DNA-dependent SWI2/SNF2 ATPase family (Pereira et al, 2003). This ATPase activity of human BTAF1 is strongly stimulated in the presence of both TBP and DNA (Chicca et al., 1998). Moreover, while ATP is needed for TBP-TATA dissociation, it is not necessary for BTAF1 binding to TBP (Auble et al.,

1997). Thus, it is apparent that both N-terminal TBP interaction domain and C-terminal ATPase domain of BTAF1 are necessary for inhibiting TBP-TATA complex formation. Besides preventing the free form of TBP from binding to promoter and nonpromoter AT-rich sequences, BTAF1 also plays an active role in dissociating preformed TBP-TATA complexes, likely by acting as a molecular motor translocating across the DNA after loading through some uncharacterized DNA elements (Darst et al., 2001). This nonpromoter-dissociating activity of BTAF1 may account for the coactivating activity of BTAF1 in enhancing both basal and activator-dependent transcription by redistributing TBP to correct promoter regions (Collart, 1996; Li et al., 1999; Muldrow et al., 1999; Andrau et al., 2002; Geisberg et al., 2002), consistent with the finding that the ATPase activity of BTAF1, originally implicated in transcriptional repression, also contributes to BTAF1-mediated transcriptional activation (Dasgupta et al., 2002). Clearly, BTAF1 has a dual role in transcription. Under normal conditions, BTAF1-TBP seems to exist as an inactive promoter-bound complex, through nonconcave interaction, in yeast and is later activated by associating with other GTFs and pol II following environmental stress (Geisberg and Struhl, 2004). This is also in agreement with genome-wide transcriptional profiling indicating that 10% and 5% of yeast genes are respectively upregulated and downregulated by BTAF1 (Geisberg et al., 2002).

D: NC2 Regulation of TBP-TATA Complex Formation

A third TBP-containing complex regulating promoter activity contains negative cofactor 2 (NC2) in association with TBP. NC2, consisting of NC2a (DRAP1) and NC2ß (Drl) that interact with each other through histone fold motifs (Goppelt et al., 1996), normally functions as a repressor in inhibiting transcription from TATA-containing promoters (Meisterernst and Roeder, 1991; Inostroza et al., 1992) and as a coactivator in stimulating transcription through DPE-driven TATA-less promoters (Willy et al., 2000). Recent structural analysis of NC2-TBP-TATA ternary complex, revealed by X-ray crystallography at 2.6 Ä resolution, shows that the N-terminal regions of both NC2 subunits bind DNA on the underside of the preformed TBP-TATA complex and the C-terminus of NC2ß additionally contacts the convex surface of TBP, hence giving NC2 the appearance of a molecular "clamp" that grips both upper and lower surfaces of the TBP-TATA complex (Kamada et al., 2001). Obviously, this molecular clamp is able to block PIC assembly by inhibiting TFIIA and TFIIB binding to the upper side of TBP, an observation consistent with gel-shift and in vitro transcription assays performed with recombinant proteins (Inostroza et al., 1992; Kim et al., 1995; Goppelt et al., 1996). Indeed, some transcription factors, such as hypoxia-inducible factor la (HIF-la), appear to inhibit transcription by inducing NC2-mediated blocking of PIC assembly (Denko et al., 2003).

In contrast to our understanding regarding the repression activity of NC2, very little is known about the coactivating function of NC2. It is likely that NC2 may work in conjunction with TFIID to enhance TATA-less gene transcription, as both proteins are implicated in DPE function (Burke and Kadonaga, 1997; Willy et al., 2000). Alternatively, NC2 may positively regulate transcription by targeting events downstream of the initiation step (e.g., elongation), since NC2 association with the hyperphosphorylated form of pol II seems necessary for Gal4-VP16-mediated activation from the TATA-containing HIV-1 promoter (Castaño et al., 2000). However, the exact mechanism underlying the coactivating function of NC2 during the transcription process remains to be elucidated.

In addition to forming a heterodimeric NC2 complex, NC2a and NC2p are also individually involved in the control of specific gene transcription. It seems that free forms of NC2a and NC2P are mainly found in exponentially growing yeast cells, whereas stable NC2a and NC2P complex is only detected after glucose depletion (Creton et al., 2002). The free entity of NC2a allows it to interact with BTAF1 and in turn stimulates BTAF1 association with the convex surface of TBP on the DNA (Klejman et al., 2004). Interestingly, NC2a appears to coreside with TBP at transcriptionally active promoters and NC2P is mainly associated with TBP-bound promoters at transcriptionally repressed genes (Creton et al., 2002). This is not surprising considering the fact that NC2P binding to the upper surface of TBP blocks TFIIB association and the available NC2a structure shows only binding to the underside of the TBP-TATA complex presumably without affecting PIC assembly (Kamada et al., 2001).

E: Positive Regulators that Promote TBP-TATA Complex Formation

For TBP to nucleate the assembly of a functional PIC, it must overcome molecular impediments that prevent TBP-TATA complex formation. Some of the impediments include dimerization of TBP and inhibition of TBP binding to the TATA box exerted by TAF1, BTAF1, and NC2, as mentioned in the previous sections. Alleviation of these inhibitory activities can be achieved by transcriptional regulators, which may function by enhancing TBP binding to the TATA box, antagonizing repressor binding for the overlapping surfaces, stabilizing the bent TATA-TBP complex, or by modifying the chromatin structure surrounding the promoter region.

In general, TFIIA and TFIIB enhance TBP binding to the TATA box and further stabilize the TBP-promoter complex via distinct mechanisms. TFIIA increases the likelihood of precise promoter recognition by TBP via promoting the dissociation of TBP dimers and thus facilitating the loading of monomeric TBP onto the TATA box (Coleman et al., 1999). TFIIA also competes with the inhibitory domain of TAF1 for overlapping binding regions on TBP, hence alleviating TAF1-mediated inhibition of TATA recognition (Kokuba et al., 1998). Furthermore, incorporation of TFIIA into the TBP-TATA complex renders the ternary complex resistant to BTAF1-mediated dissociation of the TBP-TATA complex (Auble and Hahn, 1993). Similarly, TFIIB can enhance TBP binding to the TATA box (Imbalzano et al., 1994b) and also stabilize the bent TBP-TATA complex (Zhao and Herr, 2002), thereby reducing the dissociation rate of TBP from the promoter region (Wolner and Gralla, 2001). Clearly, TFIIA and TFIIB are often needed for stable TBP-TATA complex formation, yet both protein factors may not enhance TBP binding to the TATA box when the promoter is in a nucleosome configuration.

Considering that approximately 146 base pairs of DNA is wrapped around each core histone octamer in vivo, the promoter region may be incorporated into a nucleosome, rendering the TATA box inaccessible for TBP binding thus suppressing basal transcription (Imbalzano et al., 1994a). Undoubtedly, additional protein factors are needed to alleviate transcriptional repression from nucleosome-embedded promoters. In this regard, chromatin-modifying enzymes play a key role in enhancing TBP access to its recognition sequence by altering chromatin structure surrounding the TATA box (Martinez-Campa et al., 2004). Chromatin modifiers may reconfigure nucleosome structure by covalently modifying histones to reduce protein-DNA interactions (Fischle et al., 2003) or by using the energy of ATP hydrolysis to alter histone-DNA contacts in a process known as chromatin remodeling (Narlikarei al., 2002). Since the interaction between the TATA box and nucleosomal core histones are regulated by acetylation of lysine residues at the N-terminal tails of core histones, acetylation of nucleosomal core histones at the promoter region creates a less compact chromatin structure allowing TBP to bind the TATA box (Sewack et al., 2001). SWI/SNF is an example of an ATP-dependent chromatin remodeler that enhances access to the TATA box and subsequent binding by TBP and TFIIA in vitro (Imbalzano et al., 1994a). The action of SWI/SNF following the acetylation of histones by GCN5 histone acetyltransferase is seen in vivo as well, where SWI/SNF facilitates TBP binding to the p-interferon promoter (Agalioti et al., 2000). Interestingly, loading of TBP onto both TATA-containing and TATA-less promoters can be further enhanced by TBP association with other cellular proteins to form multiprotein complexes, such as TFIID and Spt-Ada-Gcn5 acetyltransferase (SAGA), both of which contain acetyltransferase activity (Zanton and Pugh, 2004).

F: TAF Nomenclature

By using antibodies raised against TBP, Tjian's group initially identified a set of polypeptides tightly associated with TBP, known as TAFs (Dynlacht et al., 1991). Later studies indicated that many of these polypeptides are highly conserved based on their characterization in S. cerevisiae, S. pombe, C. elegans, D. melanogaster, and H. sapiens (Burley and Roeder, 1996; Albright and Tjian, 2000). Since individual TAFs were originally named based on their apparent molecular weights, which often differ among species, a common name for each highly conserved TAF was proposed (Tora, 2002). This unified nomenclature facilitates cross-species comparisons of TAFs, designated as TAF1 to TAF 15 for TFIID and a unique TBP-associated factor found in B-TFIID as BTAF1 (Table 4.1).

G: Core Promoter Recognition by Distinct TAF Components of TFIID

DNase I footprinting revealed that TFIID generates extended footprints beyond the TATA box on the adenovirus major late promoter when compared to that of TBP (Zhou et al., 1992; Chiang et al., 1993), indicating that TAFs may provide additional contacts with core promoter sequences (Fig.4.2) (Oelgeschlager et al., 1996; Verrijzer and Tjian, 1996). In a study using random DNA-binding site selection to identify TAF-targeted DNA sequences, a TAF1-TAF2 complex was shown to bind preferentially to sequences that matched the initiator consensus (Chalkley and Verrijzer, 1999). TFIID binding to TATA-less core promoter elements was also characterized by DNase I footprinting and photocrosslinking experiments which revealed that

Table 4.1 Nomenclature of TAFs involved in RNA polymerase II-mediated transcription.

New Name

H. sapiens

D. elanogaster

S. cerevisiae

S. pombe

C. elegans New Name Previous Name

0 0

Post a comment