Transcriptional Repressors and Repression Mechanisms

Lorena Perrone, Hitoshi Aihara and Yutaka Nibu

Department of Cell and Developmental Biology, Weill Medical College of Cornell University, 1300 York Avenue, Box 60, A308, New York, NY 10021, USA

Key Words: transcriptional repressors, transcriptional repression, corepressor, short-range repression, longrange repression


A harmonious balance between transcriptional activation and repression in eukaryotes is necessary for a variety of biological phenomena, such as pattern formation, tissue differentiation, and normal development. In this chapter, we will use well-understood cases to provide an overview of the molecular mechanisms by which transcription factors mediate repression.


Drosophila melanogaster (fruit fly) has 13,379 protein-coding genes and of which 700 encodings transcription factors (Adams et al., 2000; Celniker and Rubin, 2003; Misra et al., 2002). Of the 15,832 protein-coding genes identified in the primitive chordate, Ciona intestinalis (ascidian), 400 genes are transcription factors (Dehal et al., 2002; Imai et al., 2004). In genomes of other species, such as Arabidopsis thaliana (plant), Caenorhabditis elegans (worm), and Saccharomyces cerevisiae (yeast), 3-6% of the protein-coding genes are transcription factors (Riechmann et al., 2000). Thus, genes encoding transcription factors are a relevant fraction of the genome, perhaps reflecting their crucial role in several biological processes.

Obviously, not all of the protein-coding genes are transcribed in any specific cell. For instance, 70% of the Drosophila genes tested are transcribed tissue-specifically during a wide range of embryonic stages (Tomancak et al., 2002). Conversely, these genes are not transcribed in some tissues. In addition, 20% of the fly genes are maternally expressed, while 15% are not expressed during the entire embryogenesis. In Ciona intestinalis tailbud embryos, 30% of the genes tested are specifically expressed in only one tissue, such as epidermis, nervous system, endoderm, mesenchyme, notochord, and muscle (Satou et al., 2001). Another 30% of the genes are expressed in multiple tissues and expression of 12% of the genes is not detected by in situ hybridization. How is such a large number of genes expressed (or not) at the right moment and at the right place? A well-balanced program of gene expression is mostly regulated at the level of transcription. Ultimately, the coordinate expression of different sets of genes directs normal development, differentiation, and morphogenesis.

For example, cooperation of transcriptional repression and activation is essential for establishing localized stripes, bands, and tissue-specific patterns of gene expression in the early Drosophila embryo (Gray and Levine, 1996b; Ip and Hemavathy, 1997; Jackie et al., 1992; Mannervik et al., 1999). The initial formation of both the anterior-posterior and dorsal-ventral axes during early Drosophila embryogenesis depends on broadly distributed activators that are maternally expressed as well as localized sequence-specific repressors. In brief, after fertilization, the maternal activators begin turning on a set of zygotic genes called "gap genes" that mostly encode DNA-binding repressors. The same maternal activators also activate expression of a second set of downstream genes, called "pair-rule genes" that are expressed in patterns of seven stripes. The seven stripes are evenly spaced along the anterior-posterior axis. The borders of individual stripes are formed by the localized repressors (gap gene products)

Corresponding Author: Yutaka Nibu, Tel: +1-(212) 746-6184, Fax: +1-(212) 746-8175, E-mail: [email protected]

which turn off transcription via a concentration threshold mechanism. Thus, the pair-rule genes define the initial metameric layout of the body plan in Drosophila.

Transcription of protein-coding genes driven by RNA polymerase II can be repressed or modulated at a local or at a genomic level. For instance, expression of some genes is locally repressed by DNA-binding transcriptional repressors. This process is the major focus of this chapter. On the other hand, some genes are silenced along with large regions of the genome, genes located near centromeres and telomeres, nearly all genes on one of the two mammalian female X chromosomes, and Hox genes, are repressed.

Transcriptional Repressors

In general, transcriptional repressors affecting RNA polymerase II dependent transcription can be classified into two categories, DNA-binding and non-DNA-binding factors (Burke and Baniahmad, 2000; Gaston and Jayaraman, 2003; Hanna-Rose and Hansen, 1996). DNA-binding repressors bind sequence-specifically to DNA via DNA-binding domains, such as zinc finger, homeodomain, basic helix-loop-helix (bHLH), basic leucine zipper (bZip), and others. In addition to the DNA-binding domain, these factors usually have repression domains that mediate "active" repression. It has been shown in the early Drosophila embryo that sequence-specific DNA-binding repressors can be classified as either short-range or long-range repressors, depending upon their range of action (Courey and Jia, 2001; Gray and Levine, 1996b; Mannervik et al., 1999; Nibu et al., 1998a). Non-DNA-binding factors are typically termed corepressors and can either be recruited to DNA by DNA-binding repressors or directly interact with the components of the preinitiation complex.

Mechanisms of Repression

In the past two decades, by the virtue of the dramatic advancement of experimental techniques and molecular tools, several molecular mechanisms have been proposed to explain how transcriptional repression is achieved (Burke and Baniahmad, 2000; Gaston and Jayaraman, 2003; Hanna-Rose and Hansen, 1996; Johnson, 1995; Levine and Manley, 1989). Here we describe the molecular mechanisms of how eukaryotic protein-coding genes are repressed.

A: Inhibition of TBP and General Transcription Factors (GTFs)

Recruitment of TATA-binding protein (TBP) to eukaryotic promoters is an essential step for the initiation of transcription. Extensive analyses of TBP have shown that the function of TBP can be inhibited by direct binding of several factors, such as human BTAF1 (TAF-172) and its yeast ortholog Motlp, the Drl/Drapl complex (also known as NC2), and TAF1 (Drosophila TAFII230) (Burley and Roeder, 1998; Lee and Young, 1998; Pugh, 2000).

Human BTAF1 and yeast Motlp are members of the evolutionarily conserved SWI/SNF family, a DNA-dependent ATPase (Pereira et al., 2003). BTAF1/Motlp are large proteins (210-kDa) and Motlp is essential for yeast cell viability. BTAFl/Motlp directly interact with TBP and are able to remove TBP from the TATA box using ATPase activity (Fig. 9.1 A) (Auble et al., 1994; Chicca et al., 1998). Hence, Motlp negatively regulates transcription by impeding the binding of TBP to DNA. In contrast, it has been shown that, in some cases, Motlp can also activate a few genes (Andrau et al., 2002; Dasgupta et al., 2002; Geisberg et al., 2002; Prelich, 1997). Microarray analysis revealed that 178 genes (3% of yeast genes) are repressed by Motlp, while 6 genes are activated (Dasgupta et al., 2002). Both northern blot and microarray analyses have demonstrated that transcription of BNA1, URA1, and YDR539W genes is decreased in moll mutant yeast, while the INOl gene is activated. An ATPase-defective Motlp introduced in the motl mutant yeast could not rescue their expression levels. Hence, the ATPase activity of Motlp is essential for both repression and activation.

The Drl/Drapl complex consists of two subunits, Drapl (NC2alpha) and Drl (NC2beta), and is conserved among eukaryotes and yeast (Lee and Young, 1998). In vitro studies suggested that mammalian Drapl heterodimerizes with Drl through the histone fold motifs at their amino-terminal ends, and increases the repression activity of Drl (Fig. 9.IB) (Goppelt et al., 1996; Mermelstein et al., 1996). The Drl subunit directly interacts with the basic repeat domain of TBP bound to the TATA box. This interaction blocks the subsequent recruitment of TFIIA and TFIIB to the core promoter, thereby resulting in repression. X-ray structural studies also support this mechanism (Kamada et al., 2001). However, additional studies on Drl and Drap 1 indicate alternative modes of action. For example, mice lacking Drapl exhibit severe gastrulation defects, likely due to an increased expression of Nodal (Iratni et al., 2002). In this case, mouse Drapl, but not Drl, is sufficient to prevent DNA binding of the FoxHl activator that regulates the Nodal gene. Moreover, the Drl/Drapl complex can be purified from postdiauxic yeast, however, Drapl does not associate with Drl in growing yeast (Creton et al, 2002). In chromatin immunoprecipitation assays, both yeast Drap 1 and TBP associate with active promoters, while the Drl /Drapl complex is found on repressed promoters. Finally, in vitro transcription assays, the Drosophila Drl /Drapl complex purified from embryonic nuclear extract represses TATA-box containing promoters, but it activates promoters containing the downstream promoter element (DPE) (Willy et al., 2000).

Fig. 9.1 Inhibition of TBP and general transcription factors (GTFs). TBP function can be inhibited by BTAFl/Motlp (A), Drapl/Drl (B), and TAF1 (C). (D) DNA-binding repressors (R) inhibit TBP and GTFs. The black bars represent double-stranded DNA.

Recent studies suggest that human Drapl interacts with the central region of BTAF1 in vitro and in yeast (Klejman et al., 2004). The physical interaction between Drapl and BTAF1 may account for the regulation of the same genes by these two factors in yeast (Lemaire et al., 2000; Prelich, 1997).

TAF1 (formerly named TAF250 in human and TAF230 in Drosophila) is the largest subunit of TFIID complex (Pugh, 2000). TAF1 contains multiple enzymatic activities, a histone acetyltransferase, a serine/threonine kinase, and a histone-specific ubiquitin-activating/ conjugating enzyme (Dikstein et al., 1996; Mizzen et al., 1996; Pham and Sauer, 2000). TAF1 has bromodomains that bind to acetylated histone H4 (Jacobson et al., 2000). In vitro studies displayed that the 80-residue N-terminal region of Drosophila TAF1, containing three alpha helices and a beta hairpin, binds directly to TBP and inhibits the TBP function (Fig. 9.1C) (Kokubo et al, 1994; Nishikawa et al., 1997). Subsequent NMR structural studies revealed that the structure of the

N-terminal region of Drosophila TAF1 is similar to the minor groove surface of the TATA box DNA sequence and that the TATA box-binding domain of TBP binds to this region of TAF1 (Liu et al., 1998). Thus, TAF1, by mimicking the TATA box, can prevent TBP from binding to DNA. Additional studies in yeast suggested the following two-step hand off model (Kotani et al, 2000). The interaction between TAF1 and TBP can be compromised by activators containing acidic activation domains that also interact with the TATA box-binding domain of TBP. The intermediate complex containing TBP and the activator can not bind the TATA box. Finally, TBP is released from the activator and can bind the TATA box to initiate transcription.

It is conceivable that TBP may need to be in an inactive state before activators turn on transcription, and/or that TBP may need to be removed from DNA to turn off transcription after the genes receive the signal leading to transcriptional repression.

In addition to the factors mentioned above, DNA-binding repressors also directly interact with TBP or GTFs to inhibit transcription (Fig. 9. ID). This mechanism is called "direct repression". For instance, Even-skipped (Eve), a Drosophila homeodomain protein, interacts with TBP in vitro and in tissue culture (Austin and Biggin, 1995; Li and Manley, 1998; Um et al, 1995). A repression domain of Eve directly contacts the C-terminal region of TBP, leading to inhibition of TFIID binding to the promoter. This repression is independent of the distance between the promoter and the Eve binding sites in vitro transcription assays. Unliganded thyroid hormone receptor (TR) also interacts with TBP and inhibits the formation of preinitiation complex in vitro, however TR bound to its ligand activates transcription by interacting with TFIIB (Baniahmad et al, 1993; Fondell et al, 1996; Fondell et al, 1993). Similarly, TFIIEbeta binds to a dimeric form of Krüppel in vitro, a Drosophila zinc finger protein, through its C-terminal domain (Sauer et al, 1995). This direct association is sufficient to cause repression and hence this is a corepressor-independent mechanism. On the contrary, a monomer form of Krüppel acts as an activator that interacts with TFflB in vitro (Sauer et al, 1995). It should be noted that both Krüppel and Eve interact with corepressors: dCtBP, dRpd3, Groucho, and Atrophin, in vitro and/or in yeast, and that the corepressors are required for their repression activities in Drosophila genetic assays (Kobayashi et al, 2001; Mannervik and Levine, 1999; Nibu et al, 2003; Nibu et al, 1998a; Zhang et al, 2002b).

B: Short-range Repression

The DNA-binding repressors expressed in the

Fig. 9.1 Inhibition of TBP and general transcription factors (GTFs). TBP function can be inhibited by BTAFl/Motlp (A), Drapl/Drl (B), and TAF1 (C). (D) DNA-binding repressors (R) inhibit TBP and GTFs. The black bars represent double-stranded DNA.

Drosophila embryo appear to fall into two categories, short-range or long-range repressors, depending upon their range of action (Courey and Jia, 2001; Gray and Levine, 1996b; Mannervik et al., 1999). Short-range repression takes place through three mechanisms, quenching (Bl), direct repression (B2), and competition (B3).

Bl: Quenching

An activator bound to DNA can be inactivated by an adjacent repressor and this mechanism is called quenching (Fig. 9.2B). Short-range repressors, such as Krüppel (Cys2His2 zinc fingers), Knirps (nuclear receptor), Snail (Cys2His2 zinc fingers), and Giant (bZip), work over distances of less than 100 bp to inhibit adjacent activators in the Drosophila embryo (Arnosti et al., 1996b; Gray and Levine, 1996a; Gray et al., 1994; Hewitt et al., 1999; Nibu and Levine, 2001). When binding sites for these short-range repressors are located within lOObp from activator binding sites, these repressors inhibit these adjacent activators in transgenic Drosophila embryos. However, repression is lost when the repressor sites are moved more than 100 bp away from the activator sites.

A short-range repressor bound to one enhancer is not able to interfere with activators bound to a neighboring enhancer. For instance, the pair-rule eve gene carries five enhancers located 5' and 3' of the transcription unit and each of them controls one or two stripes (Clyde et al., 2003; Fujioka et al., 1999; Small et al., 1996; Small et al., 1991). These enhancers are typically 300 bp to 1 kb in length and contain clustered binding sites for both activators and repressors (Berman et al., 2002). Even though expression of Krüppel and eve stripe 3 overlap, the binding of Krüppel to the eve stripe 2 enhancer to form the posterior border does not interfere with stripe 3 expression (Gray and Levine, 1996b; Small et al., 1993; Small et al., 1992; Small et al., 1991). Similarly, the anterior border of stripe 2 is defined by the binding of Giant repressor to the stripe 2 enhancer, while Giant expression overlaps with stripe 1 (Arnosti et al., 1996a; Small et al., 1992; Small et al., 1991). The maternal D-Stat activator turns on the expression of eve stripe 3 and 7, while Knirps establishes the borders of eve stripe 3/7 and 4/6 in a concentration dependent manner (Clyde et al., 2003; Struffi et al., 2004). However, binding of Knirps to the eve stripe 3/7 and 4/6 enhancers does not inhibit expression of eve stripe 5. Quenching is also widely employed for establishment of the dorsal-ventral axis. The maternal Dorsal (rel domain) nuclear gradient activates rhomboid, short gastrulation, singleminded, and ventral nervous system defective genes, in both ventral and lateral regions of early embryos, but the Snail repressor keeps these genes off in the ventral mesoderm (Cowden and Levine, 2003; Ip et al., 1992; Nibu et al., 1998a; Stathopoulos and Levine, 2002).

Interestingly, removal of the corepressor Drosophila CtBP (dCtBP) that interacts with Krüppel, Knirps, and Snail, through the conserved PxDLS motif, impairs the activity of these repressors in the Drosophila embryo (Nibu et al., 1998a; Nibu et al., 1998b). Thus, quenching is a corepressor-dependent repression.

What is the molecular mechanism by which dCtBP operates? dCtBP is similar to NAD - dependent D-isomer-specific 2-hydroxy acid dehydrogenases, which are metabolic enzymes such as pyruvate dehydrogenase (Chinnadurai, 2002; Turner and Crossley, 2001). Human CtBPl (hCtBPl) has been shown to be a functional dehydrogenase (Balasubramanian et al., 2003; Kumar et al., 2002; Shi et al., 2003). It is known that NAD+/NADH binding to hCtBPl enhances the interaction with the adenovirus El A oncoprotein containing the PxDLS motif and then facilitates the oligomerization of hCtBPl itself (Balasubramanian et al., 2003; Kumar et al., 2002; Zhang et al., 2002a). Additional studies indicated that the dehydrogenase domain of hCtBPl is essential for its repression activity in tissue culture (Kumar et al., 2002). However, it has been argued that a mutation of the catalytic histidine residue does not alter the repression activity of mouse CtBP2, dCtBP, or hCtBPl in transient reporter assays (Grooteclaes et al., 2003; Phippen et al., 2000; Sutrias-Grau and Arnosti, 2004; Turner and Crossley, 1998). A complex containing hCtBPl tagged with both the Flag and haemagglutinin (HA) epitopes has been purified from HeLa cells. The purified fraction contains histone methyltransferases (HMT) and histone deacetylases (HDAC) (Shi et al., 2003). These results suggest that the CtBP complex represses transcription by directing deacetylation and methylation of histones through HDACs and HMTs. However, it should be pointed out that, in Drosophila S2 cells, the HDAC inhibitor TSA did not inhibit dCtBP-mediated repression, but blocked the repression activity of Groucho which interacts with the histone deacetylase dRpd3 (Ryu and Arnosti, 2003).

B2:Short-range Direct Repression

When the binding sites for the short-range repressors (Krüppel, Knirps, Snail, and Giant) are located within 100 bp from the core promoter, these factors can dominantly shut down the promoter activity regulated by multiple enhancers in the transgenic Drosophila embryo (Fig. 9.2C) (Arnosti et al., 1996b; Gray and Levine, 1996a; Gray et al., 1994; Hewitt et al., 1999). In this case, these repressors do not directly inhibit the enhancers, since the enhancers can actually activate the other linked promoter which lacks the repressor sites. In addition, when these sites are moved away from the core promoter, repression is lost. The direct repression activity of Krüppel and Knirps, as monitored by transgenes, is diminished in the dCtBP mutant embryo, indicating that short-range direct repression is corepressor-dependent (Nibu et al., 2003). In contrast, Krüppelmediated repression detected in vitro is corepressor-independent, as mentioned earlier.

B3: Competition

Some repressors compete with activators for identical or overlapping DNA sequences (Fig. 9.2D). This is often termed "passive" repression, in contrast to the corepressor-dependent repression, which is called "active" repression.

0 0

Post a comment