|28 May 2001, Volume 20, Number 24, Pages 3055-3066|
|Table of contents Previous Article Next [PDF]
|Chromatin silencing and activation by Polycomb and trithorax group proteins|
|Tokameh Mahmoudi and C Peter Verrijzer|
Department of Molecular Cell Biology, MGC Centre for Biomedical Genetics, Leiden University Medical Centre, PO Box 9503, 2300 RA Leiden, The Netherlands
Correspondence to: C P Verrijzer, Department of Molecular Cell Biology, MGC Centre for Biomedical Genetics, Leiden University Medical Centre, PO Box 9503, 2300 RA Leiden, The Netherlands
The Polycomb group (PcG) of repressors and the trithorax group (trxG) of activators maintain the correct expression of several key developmental regulators, including the homeotic genes. PcG and trxG proteins function in distinct multiprotein complexes that are believed to control transcription by changing the structure of chromatin, organizing it into either a 'closed' or an 'open' conformation. The hallmark of gene regulation by PcG/trxG proteins is that it can lead to a mitotically stable pattern of gene expression, often referred to as epigenetic regulation. Although much remains to be learned, recent studies have provided insights into how this epigenetic switch is set, how PcG/trxG proteins might be linked to cis-acting DNA elements and what potential mechanisms underlie stable inheritance of gene expression status over multiple cell divisions. Finally, the study of the evolutionarily conserved PcG/trxG factors has recently gained additional urgency with the realization that they play a pertinent role in certain human cancers. Oncogene (2001) 20, 3055-3066.
polycomb; trithorax; chromatin; epigenetics; transcription
Chromatin is a highly dynamic structure that is remodeled and modified in response to physiological and developmental signals and has a critical function in the control of gene expression (Kadonaga, 1998; Workman and Kingston, 1998; Kornberg and Lorch, 1999; Wolffe and Matzke, 1999). The packaging of DNA into chromatin generates a barrier to processes such as transcription, which require access to the DNA. To overcome nucleosomal repression, several distinct biochemical mechanisms have co-evolved to modulate the structure of chromatin. The two main strategies by which cells alleviate chromatin-mediated repression are through the action of ATP-dependent chromatin remodeling complexes and histone acetyltransferases (HATs) (Grunstein, 1997; Armstrong and Emerson, 1998; Kadonaga, 1998; Workman and Kingston, 1998; Varga-Weisz and Becker, 1998; Kingston and Narlikar, 1999; Wade and Wolffe, 1999; Peterson and Workman, 2000; Brown et al., 2000; Cheung et al., 2000). The energy consuming chromatin remodelling factors (remodelers) restructure chromatin and partially disrupt the contacts between DNA and histones. Acetylation of the amino-terminal histone tails is believed to loosen up the interactions between neighboring nucleosomes and result in a more open and accessible chromatin structure. Thus, a productive way to think about HATs and remodelers is as molecular machines that open up the chromatin and provide access to the DNA for transcription factors.
Chromatin structure also plays a key role during epigenetic control of developmental regulators such as the homeotic genes. Epigenetic regulation of transcription leads to the mitotically stable propagation of differential expression of genetic information (Weiler and Wakimoto, 1995; Pirrotta, 1998; Wolffe and Matzke, 1999; Lyko and Paro, 1999). Thus, an epigenetic mark (i.e. that a gene is 'on' or 'off') is inherited by the progeny of a cell during development and ensures that the appropriate gene expression pattern and cell phenotype is maintained over many generations. This process is often referred to as cellular memory. An increasing number of genes encoding components of epigenetic gene regulatory complexes are found to be mutated or translocated in human cancers, underscoring the importance of the correct maintenance of cellular memory of developmental fate (Jacobson and Pillus, 1999). Understanding the molecular mechanisms by which chromatin modulating factors control gene expression will be central to our understanding of associated cancers.
In this review we will focus on the Polycomb and trithorax groups (PcG/trxG) of transcriptional regulators. These factors are part of a conserved cellular memory system in animals that maintains the active or inactive state of many developmental regulators. Evidence has accumulated that these proteins specifically alter the structure of chromatin in order to 'freeze' the expression status of a target gene. Here we will discuss parallels between the PcG/trxG system and heterochromatin silencing and recent findings concerning targeting of PcG/trxG proteins, how an epigenetic decision may be made and the potential mechanisms underlying mitotically stable gene expression.
Heterochromatin silencing and position effect variegation (PEV)
Heterochromatin was first defined cytologically as a deeply stained, relatively minor portion of eukaryotic chromatin that remains condensed throughout the cell cycle. In contrast, euchromatin, which comprises the majority of chromatin, is decondensed and less densely stained during interphase. Heterochromatin, while also detected at some interstitial sites, is mainly concentrated at the centromeric and telomeric regions of chromosomes (Karpen, 1994; Weiler and Wakimoto, 1995; Pillus and Grunstein, 1995; Elgin, 1996; Grunstein, 1998; Wakimoto, 1998; Wallrath, 1998). Heterochromatin has several functional characteristics which distinguish it from euchromatin; it tends to be 'gene-poor', contains highly repetitive DNA sequences, is often located at the nuclear periphery and replicates late in S phase. Moreover, the DNA in heterochromatin appears to be less accessible to transcription factors and other DNA binding proteins while its associated histones are often underacetylated. Heterochromatin plays a critical role both in centromeric and telomeric function and chromosomal pairing during meiosis. In addition, several essential genes reside in heterochromatic regions (Karpen, 1994; Weiler and Wakimoto, 1995). Thus, rather than merely representing a wasteland of junk DNA, heterochromatin plays an essential role in distinct cellular processes.
When, due to a chromosomal rearrangement, a gene originally located in euchromatin is placed near or within heterochromatin, its transcription is frequently repressed. The decision as to whether a translocated gene is repressed or expressed in a given cell appears to be random. However, once this binary decision has been made, the active or inactive state remains mitotically stable and clonally inherited. During the development of an animal or growth of a yeast colony, patches of progeny cells then appear in which the affected gene has become either repressed or expressed. Since the mosaic expression pattern of the translocated gene is determined by its position within a global chromosomal context, this phenomenon was named Position Effect Variegation (PEV) (Weiler and Wakimoto, 1995; Pillus and Grunstein, 1995; Elgin, 1996; Wakimoto, 1998; Wallrath, 1998). PEV has been observed in yeast, Drosophila and mammals. Polytene chromosome analysis of fruit fly larvae revealed that a repressed translocated gene is packaged into heterochromatin, whereas if the gene is in an active state, there is no spreading of heterochromatin (Wallrath and Elgin, 1995). These observations provide a strong correlation between heterochromatic packaging and transcriptional repression in PEV. Nevertheless, the molecular nature of heterochromatin structure, often referred to as 'higher order', remains poorly understood.
In yeast, heterochromatin is found at telomeres as well as the silent mating type (HM) loci (Pillus and Grunstein, 1995; Grunstein, 1998). Genetic screens have identified the Silent Information Regulator (SIR) genes required for silencing at the HM loci and telomeres. The Sir proteins do not bind DNA directly but are recruited to the telomeres and HM loci by sequence-specific DNA binding proteins such as Rap1p, Abf1p and ORC (Grunstein, 1998). It is pertinent to note that at other locations these tethering factors perform distinct functions not related to heterochromatin silencing. From the initial complex, the SIR proteins spread along the chromosome via direct interactions with the amino-terminal histone tails and block accessibility to the transcription machinery (Pillus and Grunstein, 1995; Hecht et al., 1996; Grunstein, 1998). Spreading is, at least in part, restricted by the availability of SIR proteins in the cell and subject to dosage effects.
Modifiers of heterochromatic silencing
In Drosophila, heterochromatin is present at the subtelomeric and centromeric regions of chromosomes. The genetic accessibility of the fruit fly has allowed for powerful genetic screens that led to the identification of modifiers of heterochromatic silencing (Weiler and Wakimoto, 1995; Elgin, 1996; Wallrath, 1998). One frequently used strategy involves variegated expression of the white gene, which is essential for the red eye color of Drosophila. When a chromosomal inversion places the normally euchromatic white locus adjacent to heterochromatin, it becomes repressed in many but not all cells in which it is ordinarily expressed. The expression status of the white gene, either 'on' or 'off', is stably inherited upon cell division. As a result, the Drosophila eye will show patches of pigmentation (variegation). The eye facets are red if the white gene product is present and white if it is not expressed. When silencing is strong, the compound eye will be mostly white, whereas when silencing is weaker there will be more red facets. These characteristics are easily observable and have been used in genetic screens for suppressors (Su(var)) or enhancers (E(var)) of variegation, which respectively result in weaker or stronger repression of white.
Over 50 modifiers of heterochromatic silencing have been identified in the fruit fly (Weiler and Wakimoto, 1995; Elgin, 1996; Wallrath, 1998). Partial deletions of the tandem array of histone genes lead to a Su(var) phenotype, revealing a direct role of histone dosage in heterochromatin formation. The best characterized Su(var) is Heterochromatin Protein 1 (HP1) that is evolutionarily conserved from yeast to man. HP1 contains a highly conserved 'chromo' domain also present in the developmental regulator Polycomb (PC), and is involved in directing HP1 to heterochromatin (Cavalli and Paro, 1998a). Another prominent modifier of PEV, SUV39H1 the human homologue of Drosophila Su(var)3-9, belongs to the family of SET domain proteins (Jenuwein et al., 1998). The SET domain is present in a diverse group of chromosomal proteins, including the trxG protein trithorax (TRX). An important recent finding is that SUV3gH1 encodes a histone methylase that selectively methylates lysine 9 of the amino terminal tail of histone H3 (Rea et al., 2000). Although histone methylation requires the SET domain of SUV3gH1, not all SET proteins are histone methylases. Modification of the histone tails by acetylation, phosphorylation and methylation is believed to generate a histone code, which may control the organization of the chromatin fiber into higher order structures (Paro, 2000; Strahl and Allis, 2000). Although heterochromatin silencing does not seem to play a role during normal development, it displays many parallels with the mechanisms by which PcG and trxG proteins control developmental gene expression.
Maintenance of developmental gene expression patterns by PcG and trxG proteins
Development of the fly embryo is controlled by an intricate cascade of genes, many of which encode transcription factors that first divide the embryo into broad domains and subsequently lead to increasingly fine subdivisions (St Johnston and Nusslein-Volhard, 1992). The identity of the body parts is determined by transcriptional activators encoded by the homeotic genes. The Drosophila homeotic genes are clustered in two main gene complexes: the Antennapedia complex (ANT-C) and the Bithorax complex (BX-C) (McGinnis and Krumlauf, 1992). Together, these two gene clusters are referred to as the homeotic complex (HOM-C). Each of the eight genes in the HOM-C is expressed in a restricted region of the developing fly and determines the identity of that region. Incorrect expression of any of these genes causes, the sometimes spectacular, homeotic transformation of one segment into the likeness of another. In vertebrates, the highly conserved homeotic (Hox) genes are organized in four different chromosomal complexes believed to have evolved by a process of duplication and divergence from a common single ancestral cluster (McGinnis and Krumlauf, 1992). Similar to Drosophila, in vertebrates, the Hox genes are expressed in specific domains along the anterior-posterior axis and play a central role in determination of regional identity of the body plan.
What regulates the expression of the homeotic genes? In Drosophila, the expression domains of the homeotic genes are established early in development by activators and repressors encoded by the gap and pair-rule genes. Pair-rule gene products activate transcription whereas the gap proteins repress the homeotic genes outside their appropriate expression domain. However, these transcription factors are only transiently present and disappear later during development. The faithful inheritance of the previously established expression pattern of the homeotic genes now requires the activities of the PcG and trxG proteins (Kennison, 1995; Pirrotta, 1998; Lyko and Paro, 1999). PcG proteins maintain repression of homeotic genes outside their normal expression domains whereas trxG proteins are needed to sustain expression. Although the molecular mechanisms are still unclear, PcG/trxG proteins are believed to control gene expression by modulating the structure of chromatin. In addition to the maintenance of homeotic gene expression patterns, PcG/trxG proteins are also involved in the control of many other genes (Phillips and Shearn, 1990; Pelegri and Lehmann, 1994; Breen, 1999; Brizuela and Kennison, 1997; Jacobs et al., 1999). The PcG/trxG genes constitute an evolutionarily highly conserved cellular memory system that is present in plants, animals and humans (Schumacher and Magnuson, 1997; Gould, 1997; Satijn and Otte, 1999). Furthermore, some trxG proteins are related to yeast regulators of gene expression. Like their Drosophila counterparts, the mammalian PcG/trxG proteins maintain the correct expression patterns of the Hox genes. In addition, evidence has accumulated that PcG/trxG proteins play a critical role in certain human cancers.
Silencing by the Polycomb-group proteins
Genetic screens based on homeotic derepression has so far identified 13 Drosophila PcG genes while up to 30 more PcG members may exist (Jurgens, 1985; Kennison, 1995). Typically, embryos bearing a mutation in one of the PcG genes display transformations of larval segments towards the identity of posterior ones. The derepression of distinct HOM-C genes in PcG mutants agrees well with the observed homeotic transformations (Kennison, 1995). Protein sequence comparison and molecular cloning of PcG genes revealed that their primary structures are quite heterogeneous. Of interest, several structural motifs such as the chromo and SET domains present in some PcG members are shared with proteins involved in PEV (Jenuwein et al., 1998; Cavalli and Paro, 1998a). Although the PcG genes are essential for viability, maternal contributions are believed to be responsible for the weak phentoypes of some PcG mutant embryos.
The PcG proteins do not appear to be required for the initiation of HOM-C gene repression as this task is performed mainly by the products of the gap genes. The initial pattern of expression of the HOM-C genes is normal in embryos mutant for a PcG gene. However, later during development, repression of HOM-C genes is no longer maintained in the absence of normal amounts of PcG proteins (Jurgens, 1985; Struhl and Akam, 1985). Derepression often sets in after the blastoderm stage when expression of the initial repressors of the HOM-C genes ceases and PcG proteins become critical for maintenance of the repressed state. Once established, silencing by the PcG proteins is mitotically stable and maintained over many cell divisions. Since PcG proteins are widely expressed in the developing embryo, their mere presence or absence does not convey positional information and specific silencing.
Many vertebrate homologues to Drosophila PcG genes have been isolated (Schumacher and Magnuson, 1997; Gould, 1997; Satijn and Otte, 1999). Although there may be differences in the establishment of the homeotic gene expression patterns in mammals and flies, the mechanism of maintenance involving PcG/trxG proteins seems to be largely conserved. Indeed, disruption or over-expression of PcG genes in mice leads to homeotic transformations. Furthermore, the phenotypes of the mutant mice emphasize that the activities of the PcG proteins are not limited to regulation of the Hox genes. This notion is highlighted by the involvement of mammalian PcG proteins in tumorigenesis. The mouse Bmi-1 gene, related to Drosophila Posterior sex comb (Psc) and Suppressor of Zeste 2 (Su(z)2), was first identified by proviral activation as an oncogene that co-operates with c-Myc in mouse lymphomagenesis (Haupt et al., 1991; van Lohuizen et al., 1991). Recently, it has been shown that Bmi-1 controls cell proliferation and senescence through down regulation of the tumor suppressors p16 and p19Arf, encoded by the ink4a locus (Jacobs et al., 1999). In summary, the PcG proteins form a highly conserved family of silencing factors, which not only determine the transcription pattern of a group of developmental regulators, but also regulate the expression of key cell cycle control factors. Aberrant regulation of the latter function of PcG genes may represent a critical step in tumorigenesis.
PcG proteins function in distinct multi-protein complexes
The synergistic effects of distinct PcG mutations first suggested that they act in concert to repress target genes (Kennison, 1995). This notion was reinforced by the frequent co-localization of different PcG proteins on many sites on Drosophila polytene chromosomes (Zink and Paro, 1989; DeCamillis et al., 1992; Rastelli et al., 1993; Lonie et al., 1994; Strutt and Paro, 1997; Orlando et al., 1998). However, on certain loci, different sets of PcG proteins have been detected (Strutt and Paro, 1997). Direct evidence for the existence of multi-protein PcG complexes was provided by biochemical experiments in flies and mammals. PcG proteins associate into large complexes via a plethora of protein-protein interactions that frequently involve conserved domains (Franke et al., 1992; Alkema et al., 1997; Strutt et al., 1997; Kyba and Brock, 1998; Satijn and Otte, 1999; Shao et al., 1999). Recently, a Drosophila Polycomb complex, named Polycomb Repressive Complex 1 (PRC1), was purified containing the products of the PcG genes Pc, Psc, polyhomeotic (ph), Sex combs on midleg (Scm) as well as several other uncharacterized polypeptides (Shao et al., 1999). Only when added to a nucleosomal array prior to SWI/SNF, PRC1 can block the accessibility of these arrays to remodeling by SWI/SNF. This 'order of addition' effect is likely to play a role in the stable locking of a target gene in either the active or silenced state.
Two other Drosophila PcG proteins, Enhancer of Zeste (E(Z)) and extra Sex Combs (ESC), associate in a separate complex that lacks PC or PH (Jones et al., 1998; Tie et al., 1998; Shao et al., 1999). Likewise, in mammalian cells, the homologous Embryonic Ectoderm Development (EED) and Enhancer of Zeste 2 (EZH2) proteins assemble into a complex that lacks the mammalian homologues of PC, PH and PSC (Sewalt et al., 1998; van Lohuizen et al., 1998). Interestingly, EED directed transcriptional repression is mediated via its interaction with histone deacetylase (HDAC) proteins (van der Vlag and Otte, 1999). In contrast to EED, repression by a human homologue of PC (HPC2) is resistant to treatment with histone deacetylase inhibitors. Moreover, PRC1 repression of chromatin remodeling by SWI/SNF does not depend on the presence of histone tails, arguing against a role for HDACs (or HATs) in repression by this PC containing complex (Shao et al., 1999). Taken together, these data reveal that there are at least two distinct PcG complexes that repress transcription via different molecular mechanisms (Figure 1). The PRC1 complex induces a SWI/SNF resistant chromatin structure whereas the EED containing complex deacetylates the histone tails.
Activation by trxG proteins: trithorax and mixed lineage leukemia
Whereas maintenance of repression requires the PcG proteins, the trxG proteins are needed to ensure continued expression of the homeotic genes. trxG genes have either been identified by mutations that mimic homeotic phenotypes (e.g. trx, ash-1, ash-2) or by mutations that suppress the phenotypes of PcG mutants (e.g. brm, mor, Osa) (Kennison, 1995). The trxG constitutes a heterogeneous set of proteins that are involved in gene activation, and some members of this group act as enhancers of PEV, consistent with a general role in counteracting chromatin mediated gene repression.
The essential trx gene is the founding member of the trxG of genes and was identified as a mutation that mimics a homeotic loss-of-function mutation (Ingham and Whittle, 1980). It encodes for two large protein isoforms of approximately 400 kDa that contain protein domains such as PHD fingers and the SET domain, typical for chromatin associated factors (Sedkov et al., 1994; Kuzin et al., 1994; Stassen et al., 1995). The trx gene products are required from early embryogenesis through late larval stages and control normal activity of many homeotic genes and several other genes (Ingham, 1983; Breen and Harte, 1993; Kuzin et al., 1994; Breen, 1999).
Chromosomal translocations associated with infant leukemias led to the discovery of the human Mixed Lineage Leukemia (mll) gene that shows a striking similarity to Drosophila trx (Waring and Cleary, 1997). Studies on mll deficient mice support the idea that Drosophila TRX and MLL are not only structurally related but also perform homologous functions during development. Heterozygous mll deficient mice are growth retarded and display bidirectional homeotic transformations of the axial skeleton as well as sternal abnormalities (Yu et al., 1995, 1998). Additionally, some hematopoietic abnormalities were observed in these animals. Homozygous mll deficiency results in abolished Hox gene expression and embryonic lethality. Thus, the mll and trx genes are part of a conserved regulatory network that is required for correct expression of the homeotic genes and determination of segment identity in both mammals and Drosophila. An important issue, which has so far remained unclear, is the molecular mechanisms by which TRX and MLL control gene activation.
The trxG proteins brahma, osa and moira are components of a SWI/SNF chromatin remodeling complex
The Drosophila brahma (brm) gene was identified by a screen for dominant suppressors of Pc mutations. In a Pc mutant background, mutations in brm prevent the homeotic transformations caused by misexpression of HOM-C genes (Kennison and Tamkun, 1988; Tamkun et al., 1992). brm encodes the Drosophila homologue of the yeast SWI2/SNF2 ATPase that forms the motor of the SWI/SNF chromatin remodeling complex. Two other trxG genes that were identified in a screen for suppressors of Pc, Osa and Moira (mor) were later found to encode components of the BRM complex (Crosby et al., 1999; Collins et al., 1999; Kal et al., 2000). The majority of BRM associated proteins are not encoded by trxG genes and several other trxG proteins have been found to be part of separate protein complexes (Papoulas et al., 1998). What is the function of the BRM complex during Drosophila development? brm homozygous mutants die prior to hatching but show little defect in patterning, most likely due to maternal contributions (Tamkun et al., 1992; Brizuela et al., 1994). The progeny of females with weak brm alleles die early in embryogenesis and display severe development abnormalities (Brizuela et al., 1994). The occurrence of very early developmental defects suggest that BRM function is needed for normal expression of a significant number of developmental regulators that act prior to the HOM-C genes. Pc mutations do not seem to bypass the requirement for brm in the imaginal tissues and an increased brm gene dosage does not affect the Pc mutant homeotic phenotype. These findings are consistent with the idea that BRM does not function by specifically preventing PC repression but is an activator of the HOM-C genes (Kennison, 1995).
BRG-1 and hBRM are two mammalian homologues of Drosophila BRM that are mutually exclusive subunits of human SWI/SNF complexes (Workman and Kingston, 1998; Kingston and Narlikar, 1999). Mammalian SWI/SNF complexes contain, in addition to common subunits, tissue-specific components as well (Wang et al., 1996; Workman and Kingston, 1998; Kingston and Narlikar, 1999). The human SWI/SNF complex has been implicated in gene activation by a variety of activators, suggesting a broad function in gene expression control (Armstrong and Emerson, 1998). Furthermore, a number of recent studies revealed a role for hSWI/SNF in tumorigenesis. hSNF5/INI1, a common core subunit of the hSWI/SNF complexes, is encoded by a tumor suppressor gene that is inactivated in Malignant Rhabdoid Tumors (MRT) (Versteege et al., 1998; Sevenet et al., 1999). Mice that are haploinsufficient for hSnf5 are predisposed to tumors, consistent with MRT (Roberts et al., 2000; Klochendler-Yeivin et al., 2000). Homozygous hSnf5 knockout mice do not develop beyond embryonic day 7, demonstrating that the hSWI/SNF complex is essential during early development.
A clue to the underlying mechanism for tumorigenesis in the absence of functional hSNF5 might be the role for hSWI/SNF in Rb mediated repression of genes such as cyclin A and E, which control cell cycle progression (Dunaief et al., 1994; Trouche et al., 1997; Strobeck et al., 2000; Zhang et al., 2000). In addition, hSNF5 has been implicated in transcriptional activation by the human oncogene c-myc (Cheng et al., 1999). Two other interactions suggest a role for hSNF5 in human cancers. First, the product of another trxG gene, mll, which is frequently translocated in infant leukemias, directly binds hSNF5 via its SET domain (Rozenblatt-Rosen et al., 1998). Second, purification of the endogenous breast cancer associated tumor suppressor BRCA1 revealed that it was associated with a hSWI/SNF-related complex (Bochar et al., 2000). Taken together, these results show that hSWI/SNF complexes are essential for viability and development. In addition, inactivation of hSnf5 can lead to tumor formation in humans. It will be of great interest to identify the relevant target genes and decipher the pathways that can lead to uncontrolled cell growth.
DNA elements that mediate PcG/trxG action
How do PcG/trxG proteins act in a gene-specific manner and what are the cis-acting DNA sequences that direct regulation by these proteins? Although there is little data from mammals, studies in Drosophila have been fruitful in addressing these questions. Polycomb Response Elements or PREs were identified by their silencing effect on reporter genes in transgenic flies as well as their ability to induce white gene variegation (Muller and Bienz, 1991; Zink et al., 1991; Fauvarque and Dura, 1993; Busturia and Bienz, 1993; Simon et al., 1993; Chan et al., 1994; Kassis, 1994; Pirrotta and Rastelli, 1994; Zink and Paro, 1995; Gindhart and Kaufmann, 1995). PRE mediated silencing is dependent on the PcG genes and formaldehyde cross-linking immunoprecipitation experiments have shown that the binding sites for PcG/trxG proteins coincide with PRE elements identified in functional assays (Orlando and Paro, 1993; Strutt and Paro, 1997; Strutt et al., 1997; Orlando et al., 1998). Furthermore, transposable P-elements that contain a PRE can create a new binding locus for PcG proteins at the site of insertion as detected by immunostaining of polytene chromosomes (Chan et al., 1994; Zink and Paro, 1995; Chiang et al., 1995).
PREs are found in the Drosophila homeotic gene cluster and work over large distances of up to several tens of kilobases to repress expression of their target gene. What does a PRE look like? PREs are much larger than typical enhancers, ranging from hundreds to even a few thousand basepairs in length and lacking clear cut borders or a well defined common core. Moreover, multimerization of DNA sequences with only weak PRE activity can create a strong PRE. Thus, a picture emerges of a large, rather diffuse control element whose function seems to depend on multiple distinct DNA elements that co-operate (Pirrotta, 1998; Lyko and Paro, 1999). PREs function not only as silencers; several have also been shown to mediate gene activation by trxG proteins and are sometimes called Trithorax Response Elements (TREs) (Rastelli et al., 1993; Chan et al., 1994; Chang et al., 1995; Chinwalla et al., 1995; Orlando et al., 1998; Tillib et al., 1999; Cavalli and Paro, 1999). Although recent deletion studies indicate that functional sub-elements can be distinguished (Tillib et al., 1999), it seems that generally TREs and PREs are part of an integrated module. This dual function of directing mitotically stable gene activation as well as repression is reflected by the name Cellular Memory Module (CMM) as proposed by Cavalli and Paro (1998b).
The majority of PcG/trxG proteins appear to lack sequence specific DNA binding activity, raising the question of how they assemble onto a PRE. One attractive group of candidate tethering factors comprises the PcG protein Pleiohomeotic (PHO) and the trxG proteins GAGA and Zeste. These factors are sequence-specific DNA binding transcription factors. Support for the notion that they are involved in recruitment to PREs is the identification of binding sites for these factors in some PREs. Below we will discuss these three PcG/trxG proteins.
Pleiohomeotic is a sequence-specific DNA binding PcG protein
The PcG gene pleiohomeotic (pho) encodes a zinc finger protein related to the mammalian transcription factor YY1 (Brown et al., 1998). Pho mutants that lack a maternal contribution display severe pleiotropic defects indicating that PHO executes important functions during early development in addition to its role in regulation of homeotic genes. PHO is a sequence-specific DNA binding protein that binds a silencer in the engrailed locus as well as to the Ubx PRE. Moreover, sequence elements that correspond to PHO binding elements have been identified in a number of PREs (Brown et al., 1998; Mihaly et al., 1998; Fritsch et al., 1999). Importantly, recent experiments demonstrated that the PHO sites in the Ubx PRE1.6 play a critical role in PcG silencing during development (Fritsch et al., 1999). Thus, an attractive and simple model would be that PHO provides a link between the DNA and non-DNA binding PcG proteins. However, biochemical experiments demonstrating that PHO can directly recruit other PcG proteins are still lacking. In addition, other experiments have shown that PHO binding sites alone are insufficient for pairing-sensitive silencing of a mini-white reporter gene (Brown et al., 1998). Furthermore, parts of the en silencer which do not appear to bind PHO are also critical for repression, suggesting that PHO may not work alone but co-operates with other factors.
Zeste recruits the BRM complex to activate transcription
The non-essential Zeste gene encodes a transcription factor that performs a variety of distinct, sometimes peculiar, functions during chromatin-directed gene regulation. First, Zeste is a DNA binding activator of homeotic and other genes (Biggin et al., 1988; Laney and Biggin, 1992) and is found associated with over 60 different sites on polytene chromosomes (Pirrotta et al., 1998; Rastelli et al., 1993). Zeste forms very large homo-oligomers that bind co-operatively to multiple sites present in its natural response elements (Chen and Pirrotta, 1993). Second, in a process called transvection, Zeste is able to activate a promoter on a paired homologous chromosome (Pirrotta, 1999). Third, Zeste loss-of-function mutations are enhancers of PEV, suggesting that Zeste counteracts heterochromatin-induced silencing (Judd, 1995). Fourth, particular gain-of-function mutations turn Zeste into a pairing-dependent repressor of gene expression (Rosen et al., 1998). Finally, Zeste shows positive as well as negative allele-specific genetic interactions with several PcG genes (Phillips and Shearn, 1990; Pelegri and Lehmann, 1994). Thus, Zeste may perform a dual regulatory function and cooperate with PcG as well as with trxG proteins.
Recent biochemical experiments have shown that chromatin directed gene activation by Zeste is critically dependent on the chromatin remodeling BRM complex (Kal et al., 2000). In contrast, the related ISWI containing remodeling factors are dispensable for Zeste activation, revealing functional specialization of remodeling factors. Zeste directly recruits the BRM complex to a promoter by binding to two subunits of the BRM complex that also belong to the trxG, MOR and OSA (Figure 1). These studies provide an example of tethering of a trxG complex by a sequence-specific DNA binding protein and a biochemical basis for genetic studies which have indicated that MOR, OSA, BRM and Zeste share at least some target genes (Biggin et al., 1988; Laney and Biggin, 1992; Tamkun et al., 1992; Brizuela and Kennison, 1997; Vazquez et al., 1999).
GAGA plays multiple roles in chromosome dynamics
The GAGA transcription factor was first identified as an activator of in vitro transcription of the en and Ubx genes and by its association with the promoters of heat shock and histone genes (Soeller et al., 1988; Biggin and Tjian, 1988; Gilmour et al., 1989; Granok et al., 1995; Wilkins and Lis, 1997). GAGA is encoded by the essential Trithorax-like (Trl) gene (Farkas et al., 1994). Analysis of mutant flies emphasized the in vivo importance of GAGA for the expression of several patterning genes including en, ubx and ftz (Farkas et al., 1994; Bhat et al., 1996). Immunofluorescent staining of polytene chromosomes revealed that GAGA is bound to many hundreds of euchromatic sites, highlighting its involvement in the regulation of a large number of genes (Tsukiyama et al., 1994; Benyajati et al., 1997).
GAGA binds co-operatively to multiple sites in its natural response elements as a large oligomer whose formation is mediated by the conserved POZ domain (Espinas et al., 1999; Katsani et al., 1999). GAGA is involved in the formation of DNaseI hypersensitive sites associated with gene activity and may function by directly counteracting the repressive effects of histones (Croston et al., 1991). The ISWI ATPase containing chromatin remodeling complex, NURF, was purified as a factor required for chromatin remodeling by GAGA (Tsukiyama et al., 1995). However, analysis of flies mutant for ISWI indicated that although ISWI containing remodelers play a general role in chromatin dynamics, they do not appear to be essential for homeotic gene activation by GAGA (Deuring et al., 2000). Furthermore, there is no obvious co-localization of GAGA and ISWI on the larval salivary gland polytene chromosomes. Thus, the molecular mechanism by which GAGA activates transcription remains an open question. In addition to its involvement in activation, GAGA is associated with some repressive functions as well. Mutations in GAGA were shown to diminish Fab-7 PRE silencing and GAGA may help the recruitment of PcG complexes (Hagstrom et al., 1997; Horard et al., 2000). Finally, an interaction between GAGA and SAP18, a Sin3-HDAC associated protein has been described, suggesting that at certain genes GAGA might recruit this co-repressor complex to mediate silencing (Espinas et al., 2000).
The regulatory roles of GAGA are not restricted to euchromatic gene loci. GAGA is also a dominant enhancer of PEV and plays an important role in global chromosome dynamics (Farkas et al., 1994). Bhat and colleagues used a maternal effect allele (Trl13C) to study the function of GAGA during embryogenesis (Bhat et al., 1996). The mutant embryos display profound defects in chromosome condensation and segregation as well as chromosome fragmentation and asynchrony in their cleavage cycles. Analysis of the mutant embryos suggests that these abnormalities most likely reflect a direct requirement for GAGA. Finally, GAGA's chromosomal localization is cell cycle regulated. During interphase, GAGA is bound to hundreds of euchromatic sites and not to the heterochromatic chromocenter. However, at mitosis, a massive migration of GAGA from the euchromatic sites to the GA-rich satellites takes place, suggesting it may act as a mitosis-specific reservoir for GAGA (Platero et al., 1998). In summary, GAGA plays an important and diverse role during nuclear division, in chromosome organization and in the control of gene expression.
Formation of PcG/trxG complexes
How is the assembly of PcG/trxG complexes onto chromosomes nucleated? The sequence-specific DNA binding members of the PcG/trxG protein family, PHO, Zeste and GAGA, are good candidate tethering factors. These proteins may act through a mechanism that is reminiscent of initiation of silencing in yeast by Rap1p or ORC. However, there is a second group of regulators that may recruit the PcG/trxG complexes. This group comprises the transiently expressed gap and pair-rule proteins that establish the expression patterns of the homeotic genes. This notion is supported by the finding that the repression domain of the gap protein Hunchback (HB) binds dMi-2, a core subunit of a histone deacetylase and remodelling complex (Kehle et al., 1998). In vivo experiments revealed that dMi-2 participates in both HB and PC repression, providing a link between the establishment and maintenance of repression. The dMi-2 complex may act directly by binding PcG proteins or indirectly by generating a local chromatin structure favored by PcG proteins but not by trxG proteins. An obvious advantage of an involvement of gap and pair-rule proteins in the recruitment of PcG/trxG complexes is that it would ensure a smooth transition from establishment to maintenance of gene expression patterns. However, studies so far have indicated that there is no single factor solely responsible for tethering PcG/trxG complexes (Poux et al., 1996; Brown et al., 1998). It seems more likely that the recruitment of either repressive or activating PcG/trxG complexes is the result of extensive cross-talk between early regulators, chromatin modifying enzymes and chromatin binding PcG/trxG proteins such as PHO, GAGA and Zeste.
The need for a natural DNA anchor can be bypassed by linking a PcG protein to a heterologous DNA binding domain. Artificial recruitment of Drosophila PC or murine BMI-1 protein fused to the yeast Gal4p DNA binding domain, can lead to silencing of a reporter gene in mammalian cells or flies (Bunker and Kingston, 1994; Muller, 1995). Importantly, stable repression requires endogenous PcG proteins as repression is abolished by PcG mutations (Muller, 1995). This suggests that the tethered GAL4-PC protein recruits other PcG proteins to nucleate the formation of a silencing complex. Strikingly, after formation of a repressive PcG complex, the GAL4-PC protein could be removed without compromising the silencing effect for several subsequent cell divisions. This phenomenon was dependent on flanking (non PRE) sequences of the ubx gene. These observations show that the initial tethering factor does not need to be permanently present and that transiently expressed early regulators may, in principle, initiate the recruitment of PcG/trxG complexes. Subsequently, these repressive or activating complexes remain stably associated with the chromatin in a process that involves flanking sequences.
Spreading and looping may direct PcG/trxG function to target genes
How to the activating trxG or silencing PcG proteins communicate with their target genes? Spreading and looping present two simple, not necessarily mutually exclusive, mechanisms by which a PRE may be linked to its target. Although PcG proteins aggregate at PREs, in vivo crosslinking followed by immunoprecipitation (X-CHIP) experiments indicated that there are lower levels of PcG binding over up to a few kbs of DNA surrounding the PREs (Orlando and Paro, 1993; Strutt et al., 1997; Orlando et al., 1998; Cavalli and Paro, 1998b). Combined with the observation that PcG mediated silencing is sensitive to gene dosage effects, these results suggest that PcG complexes nucleate at a PRE and subsequently spread out in a linear fashion along the chromosome. The low binding specificity of PcG proteins becomes evident after strong over-expression, when the polytene chromosomes can be largely covered by the over-expressed PcG protein (Sharp et al., 1997). These results indicate that there may be mechanistic parallels between PcG silencing in animals and telomeric silencing in yeast cells.
By analogy to yeast SIR proteins, it is of interest to examine whether interactions with nucleosomes or histone tails play a role in PcG silencing. The repression domain of the PC protein is capable of binding to nucleosome core particles in vitro (Breiling et al., 1999). However, this interaction does not strictly depend on the histone tails. Likewise, the PRC1 complex does not require histone tails to block SWI/SNF mediated chromatin remodeling. These results demonstrate a difference between the way SIR and PcG proteins interact with chromosomes, but do not completely rule out that the histone tails may play a role in spreading of PcG complexes. It is intriguing to note that many PcG/trxG members contain motifs, such as AT-hooks, HMG-boxes, ARID, and SANT domains, that are known or suspected to bind chromosomal DNA in a largely sequence independent manner. These interactions with the DNA together with histone binding are likely to mediate spreading of PcG/trxG proteins along the chromatin template. In addition, the acetylation pattern of histone tails may also determine the affinity for certain PcG or trxG proteins. For example, deacetylation of the histones by the Mi-2 HDAC complex (Kehle et al., 1998) may promote binding of PcG proteins and inhibit spreading of trxG proteins. Conversely, the Bromo domain, which preferentially binds acetylated histone tails (Jacobson et al., 2000), may mediate binding of the BRM complex to hyperacetylated areas.
There is also evidence that PREs collaborate with each other via a looping mechanism, which may explain how PREs can act over very large distances. P-elements harbouring a PRE frequently insert near chromosomal PREs, suggesting that DNA bound PcG complexes associate with each other resulting in a preferential insertion at PcG binding sites (Fauvarque and Dura, 1993; Kassis, 1994). Pairing of homologous chromosomes can lead to strong co-operation between paired PREs, whereas disruption of chromosomal pairing by chromosome rearrangements leads to reduced silencing (Fauvarque and Dujra, 1993; Chan et al., 1994; Kassis, 1994; Gindhart and Kaufman, 1995). Many promoters controlled by PcG/trxG proteins contain binding sites for GAGA that may function to direct PREs to certain target promoters (Granok et al., 1995). Indeed, GAGA mediates the enhancer blocking activity of an insulator in the eve promoter, suggesting a mechanism by which GAGA traps distal regulatory elements to direct enhancer-promoter interactions within complex loci (Ohtsuki and Levine, 1998). Other PcG/trxG proteins are likely to be involved in the function of insulators since mutations in trxG genes enhance effects of the Drosophila gypsy insulator whereas mutations in the PcG have opposite results (Gerasimova and Corces, 1998).
Taken together, the experiments discussed above are consistent with a model in which PcG complexes first assemble on PREs. Co-operative binding might be mediated by looping that brings distant PREs together and facilitates binding to individually weak PcG binding sites and thereby connects them with their target gene. Next, PcG complexes may spread further out along the chromosome and create a repressive chromatin structure. Conversely, trxG proteins may create an open chromatin structure, conducive to transcription. It is pertinent to note that there is not an all or nothing competition between PcG versus trxG proteins. Both PcG and trxG proteins are bound to PREs that are either locked into an active or repressive mode. The molecular nature of either a repressed or activated chromatin structure remains largely unknown. Moreover, PREs may act similar to insulators and affect enhancer-promoter interactions.
Although PcG proteins are able to somehow block RNA polymerase II from transcribing a target gene, they do not simply form an impenetrable barrier for any protein. In contrast to the pol II transcription machinery, T7 RNA polymerase is unimpeded by the repressive PcG protein complexes (McCall and Bender, 1996). Likewise, probing of chromatin with restriction enzymes did not reveal major changes in accessibility accompanying PEV or gene silencing by the PcG (Schlossherr et al., 1994). Thus, there may be a difference between PcG silenced chromatin and heterochromatin structure, which appears to provide a more general barrier for DNA binding proteins (Wakimoto, 1998; Wallrath, 1998).
PcG and trxG proteins constitute a memory of cell fate
What is the switch that sets a PRE in either the active or inactive state? Using a transgene carrying the Fab-7 PRE upstream of a GAL4 driven lacZ reporter and mini-white gene, Cavalli and Paro (1998b, 1999) showed that a single embryonic pulse of transcription by induction of the GAL4 activator could switch Fab-7 from a potent silencer into a mitotically stable active state. The maintenance of repression was dependent on PcG proteins whereas maintenance of the active state required the trxG proteins. Amazingly, the activated state of Fab-7 could also be transmitted through female meiosis. Importantly, it appeared that early in development, at the embryonic stage, there is a window of opportunity in which the switch can be set, whereas later, during larval stages, this becomes increasingly difficult. Without a Gal4p pulse, Fab-7 is a strong silencer, suggesting that repression is the default state. Gene expression is not simply the consequence of the loss of silencing. Instead, Fab-7 appears to play an active role in the maintenance of gene activation that requires the trxG proteins (Cavalli and Paro, 1999). It should be noted that this transient and not permanent requirement for the Gal4p activator distinguishes Fab-7 from a classical enhancer. A similar 'hit-and-run' mechanism has recently been described for activation of the yeast HO endonuclease gene (Cosma et al., 1999). The activator Swi5p binds the promoter only very transiently and recruits chromatin remodeling complexes that, subsequently persist at the HO promoter to allow sustained expression.
The results with the Fab-7 reporter lines indicate that transcriptional activation may play an important role in setting the switch, maintained later by the ubiquitously expressed PcG/trxG proteins. Active transcription through chromatin may prevent either spreading of PcG proteins from the PRE or looping and stable association of a distal PRE with its target promoter (Poux et al., 1996). Alternatively, transcription may promote association with trxG proteins that lock chromatin into an open conformation. In addition, PcG/trxG proteins may target a gene into a nuclear compartment that promotes either silencing or expression (Lemon and Tjian, 2000). Finally, in complex loci such as HOM-C, a contribution from the homeotic gene products themselves in maintenance of positional information cannot be excluded and would add another level of control.
After the transcriptional status of a gene has been stabilized by the PcG and trxG proteins, how is it faithfully propagated through cell divisions? Although the early regulators act transiently, it appears that a silencer is continuously required for maintenance of the repressed state throughout development (Busturia et al., 1997). These results argue against a simple templating mechanism but rather indicate an active role of the silencer for maintenance. Possibly, some PcG proteins remain associated or rapidly re-associate with the silencer, but not with the remainder of the gene during replication and mitosis. Since core histones distribute equally to each new DNA strand during DNA replication they, or tightly associated factors, would be ideal epigenetic markers. Indeed, histone H4 hyperacetylation is associated with Fab-7 after activation and may act as an epigenetic tag (Cavalli and Paro, 1999). This notion is further supported by the finding that repression by particular PcG proteins involves histone deacetylation (van der Vlag and Otte, 1999). Other modifications of the histones such as phosphorylation or methylation are also candidates for such a heritable marker. Alternatively, PC, which binds both nucleosomes and core histones, may co-segregate with the histones and function as a tag for re-assembly of repressive complexes (Breiling et al., 1999).
In this review we have discussed recent advances in our understanding of epigenetic regulation by the PcG/trxG proteins. We anticipate that the powerful combination of genetic and biochemical approaches will further unravel the mechanism of chromatin mediated epigenetic gene control during development. Furthermore, these studies will be instrumental to understanding the molecular basis of human diseases that involve chromatin modulating factors.
We apologize to authors whose work was not cited because of space limitations. We thank Josephine Dorsman, AG Jochemsen, Eric Kalkhoven, Robert Vries and members of our lab for critical reading of the manuscript.
Alkema MJ, Bronk M, Verhoeven E, Otte A, van't Veer LJ, Berns A, van Lohuizen M. (1997). Genes Dev. 11, 226-240. MEDLINE
Armstrong JA, Emerson BM. (1998). Curr. Opin. Genet. Dev. 8, 165-172. MEDLINE
Benyajati C, Mueller L, Xu N, Pappano M, Gao J, Mosammaparast M, Conklin D, Granok H, Craig C, Elgin S. (1997). Nucleic Acids Res. 25, 3345-3353. Article MEDLINE
Bhat KM, Farkas G, Karch F, Gyurkovics H, Gausz J, Schedl P. (1996). Development 122, 1113-1124. MEDLINE
Biggin MD, Bickel S, Benson M, Pirrotta V, Tjian R. (1988). Cell 53, 713-722. MEDLINE
Biggin MD, Tjian R. (1988). Cell 53, 699-711. MEDLINE
Bochar DA, Wang L, Beniya H, Kinev A, Xue Y, Lane WS, Wang W, Kashanchi F, Shiekhattar R. (2000). Cell 102, 257-265. MEDLINE
Breen TR. (1999). Genetics 152, 319-344. MEDLINE
Breen TR, Harte PJ. (1993). Development 117, 119-134. MEDLINE
Breiling A, Bone E, Ferrari S, Becker PB, Paro R. (1999). Mol. Cell Biol. 19, 8451-8460. MEDLINE
Brizuela BJ, Elfring L, Ballard J, Tamkun JW, Kennison JA. (1994). Genetics 137, 803-813. MEDLINE
Brizuela BJ, Kennison JA. (1997). Mech. Dev. 65, 209-220. MEDLINE
Brown CE, Lechner T, Howe L, Workman JL. (2000). Trends Biochem. Sci. 25, 15-19. Article MEDLINE
Brown JL, Mucci D, Whiteley M, Dirksen ML, Kassis JA. (1998). Mol. Cell 1, 1057-1064. MEDLINE
Bunker CA, Kingston RE. (1994). Mol. Cell Biol. 14, 1721-1732. MEDLINE
Busturia A, Bienz M. (1993). EMBO J. 12, 1415-1425. MEDLINE
Busturia A, Wightman CD, Sakonju S. (1997). Development 124, 4343-4350. MEDLINE
Cavalli G, Paro R. (1998a). Curr. Opin. Cell Biol. 10, 354-360. MEDLINE
Cavalli G, Paro R. (1998b). Cell 93, 505-518. MEDLINE
Cavalli G, Paro R. (1999). Science 286, 955-958. Article MEDLINE
Chan CS, Rastelli L, Pirrotta V. (1994). EMBO J. 13, 2553-2564. MEDLINE
Chang YL, King BO, O'Connor M, Mazo A, Huang DH. (1995). Mol. Cell Biol. 15, 6601-6612. MEDLINE
Chen JD, Pirrotta V. (1993). EMBO J. 12, 2075-2083. MEDLINE
Cheng SW, Davies KP, Yung E, Beltran RJ, Yu J, Kalpana GV. (1999). Nat. Genet. 22, 102-105. Article MEDLINE
Cheung P, Allis CD, Sassone-Corsi P. (2000). Cell 103, 263-271. MEDLINE
Chiang A, O'Connor MB, Paro R, Simon J, Bender W. (1995). Development 121, 1681-1689. MEDLINE
Chinwalla V, Jane EP, Harte PJ. (1995). EMBO J. 14, 2056-2065. MEDLINE
Collins RT, Furukawa T, Tanese N, Treisman JE. (1999). EMBO J. 18, 7029-7040. Article MEDLINE
Cosma MP, Tanaka T, Nasmyth K. (1999). Cell 97, 299-311. MEDLINE
Crosby MA, Miller C, Alon T, Watson KL, Verrijzer CP, Goldman-Levi R, Zak NB. (1999). Mol. Cell Biol. 19, 1159-1170. MEDLINE
Croston GE, Kerrigan LA, Lira LM, Marshak DR, Kadonaga JT. (1991). Science 251, 643-649. MEDLINE
DeCamillis M, Cheng NS, Pierre D, Brock HW. (1992). Genes Dev. 6, 223-232. MEDLINE
Deuring R, Fanti L, Armstrong JA, Sarte M, Papoulas O, Prestel M, Daubresse G, Verardo M, Moseley SL, Berloco M, Tsukiyama T, Wu C, Pimpinelli S, Tamkun JW. (2000). Mol. Cell 5, 355-365. MEDLINE
Dunaief JL, Strober BE, Guha S, Khavari PA, Alin K, Luban J, Begemann M, Crabtree GR, Goff SP. (1994). Cell 79, 119-130. MEDLINE
Elgin SC. (1996). Curr. Opin. Genet. Dev. 6, 193-202. MEDLINE
Espinas ML, Canudas S, Fanti L, Pimpinelli S, Casanova J, Azorin F. (2000). EMBO J. 1, 253-259.
Espinas ML, Jimenez-Garcia E, Vaquero A, Canudas S, Bernues J, Azorin F. (1999). J. Biol. Chem. 274, 16461-16469. Article MEDLINE
Farkas G, Gausz J, Galloni M, Reuter G, Gyurkovics H, Karch F. (1994). Nature 371, 806-808. MEDLINE
Fauvarque MO, Dura JM. (1993). Genes Dev. 7, 1508-1520. MEDLINE
Franke A, DeCamillis M, Zink D, Cheng N, Brock HW, Paro R. (1992). EMBO J. 11, 2941-2950. MEDLINE
Fritsch C, Brown JL, Kassis JA, Muller J. (1999). Development 126, 3905-3913. MEDLINE
Gerasimova TI, Corces VG. (1998). Cell 92, 511-521. MEDLINE
Gilmour DS, Thomas GH, Elgin SC. (1989). Science 245, 1487-1490. MEDLINE
Gindhart JG, Kaufman TC. (1995). Genetics 139, 797-814. MEDLINE
Gould A. (1997). Curr. Opin. Genet. Dev. 7, 488-494. MEDLINE
Granok H, Leibovitch BA, Shaffer CD, Elgin SC. (1995). Curr. Biol. 5, 238-241. MEDLINE
Grunstein M. (1997). Nature 389, 349-352. Article MEDLINE
Grunstein M. (1998). Cell 93, 325-328. MEDLINE
Hagstrom K, Muller M, Schedl P. (1997). Genetics 146, 1365-1380. MEDLINE
Haupt Y, Alexander WS, Barri G, Klinken SP, Adams JM. (1991). Cell 65, 753-763. MEDLINE
Hecht A, Strahl-Bolsinger S, Grunstein M. (1996). Nature 383, 92-96. MEDLINE
Horard B, Tatout C, Poux S, Pirrotta V. (2000). Mol. Cell Biol. 20, 3187-3197. MEDLINE
Ingham PW. (1983). Nature 306, 591-593.
Ingham PW, Whittle R. (1980). Mol. Gen. Genet. 179, 607-614.
Jacobs JJ, Kieboom K, Marino S, DePinho RA, van Lohuizen M. (1999). Nature 397, 164-168. Article MEDLINE
Jacobson RH, Ladurner AG, King DS, Tjian R. (2000). Science 288, 1422-1425. Article MEDLINE
Jacobson S, Pillus L. (1999). Curr. Opin. Genet. Dev. 9, 175-184. MEDLINE
Jenuwein T, Laible G, Dorn R, Reuter G. (1998). Cell Mol. Life Sci. 54, 80-93. Article MEDLINE
Jones CA, Ng J, Peterson AJ, Morgan K, Simon J, Jones RS. (1998). Mol. Cell Biol. 18, 2825-2834. MEDLINE
Judd BH. (1995). Genetics 141, 245-253. MEDLINE
Jurgens G. (1985). Nature 316, 153-155.
Kadonaga JT. (1998). Cell 92, 307-313. MEDLINE
Kal AJ, Mahmoudi T, Zak NB, Verrijzer CP. (2000). Genes Dev. 14, 1058-1071. MEDLINE
Karpen GH. (1994). Curr. Opin. Genet. Dev. 4, 281-291. MEDLINE
Kassis JA. (1994). Genetics 136, 1025-1038. MEDLINE
Katsani KR, Hajibagheri MA, Verrijzer CP. (1999). EMBO J. 18, 698-708. MEDLINE
Kehle J, Beuchle D, Treuheit S, Christen B, Kennison JA, Bienz M, Muller J. (1998). Science 282, 1897-1900. Article MEDLINE
Kennison JA. (1995). Annu. Rev. Genet. 29, 289-303. MEDLINE
Kennison JA, Tamkum JW. (1988). Proc. Natl. Acad. Sci. USA 85, 8136-8140. MEDLINE
Kingston RE, Narlikar GJ. (1999). Genes Dev. 13, 2339-2352. Article MEDLINE
Klochendler-Yeivin A, Fiette L, Barra J, Muchardt C, Babinet C, Yaniv M. (2000). EMBO J. 1, 500-506.
Kornberg RD, Lorch Y. (1999). Curr. Opin. Genet. Dev. 9, 148-151. Article MEDLINE
Kuzin B, Tillib S, Sedkov Y, Mizrokhi L, Mazo A. (1994). Genes Dev. 8, 2478-2490. MEDLINE
Kyba M, Brock HW. (1998). Mol. Cell Biol. 18, 2712-2720. MEDLINE
Laney JD, Biggin MD. (1992). Genes Dev. 6, 1531-1541. MEDLINE
Lemon B, Tjian R. (2000). Genes Dev. 14, 2551-2569. Article MEDLINE
Lonie A, D'Andrea R, Paro R, Saint R. (1994). Development 120, 2629-2636. MEDLINE
Lyko F, Paro R. (1999). Bioessays 21, 824-832. MEDLINE
McCall K, Bender W. (1996). EMBO J. 15, 569-580. MEDLINE
McGinnis W, Krumlauf R. (1992). Cell 68, 283-302. MEDLINE
Mihaly J, Mishra RK, Karch F. (1998). Mol. Cell 1, 1065-1066. MEDLINE
Muller J. (1995). EMBO J. 14, 1209-1220. MEDLINE
Muller J, Bienz M. (1991). EMBO J. 10, 3147-3155. MEDLINE
Ohtsuki S, Levine M. (1998). Genes Dev. 12, 3325-3330. MEDLINE
Orlando V, Jane EP, Chinwalla V, Harte PJ, Paro R. (1998). EMBO J. 17, 5141-5150. Article MEDLINE
Orlando V, Paro R. (1993). Cell 75, 1187-1198. MEDLINE
Papoulas O, Beek SJ, Moseley SL, McCallum CM, Sarte M, Shearn A, Tamkun JW. (1998). Development 125, 3955-3966. MEDLINE
Paro R. (2000). Nature 406, 579-580. Article MEDLINE
Pelegri F, Lehmann R. (1994). Genetics 136, 1341-1353. MEDLINE
Peterson CL, Workman JL. (2000). Curr. Opin. Genet. Dev. 10, 187-192. Article MEDLINE
Phillips MD, Shearn A. (1990). Genetics 125, 91-101. MEDLINE
Pillus L, Grunstein M. (1995). Chromatin Structure and Gene Expression Elgin S (ed). Oxford University Press, pp. 123-146.
Pirrotta V. (1998). Cell 93, 333-336. MEDLINE
Pirrotta V. (1999). Biochim. Biophys. Acta 1424, M1-M8. MEDLINE
Pirrotta V, Bickel S, Mariani C. (1988). Genes Dev. 2, 1839-1850. MEDLINE
Pirrotta V, Rastelli L. (1994). Bioessays 16, 549-556. MEDLINE
Platero JS, Csink AK, Quintanilla A, Henikoff S. (1998). J. Cell Biol. 140, 1297-1306. MEDLINE
Poux S, Kostic C, Pirrotta V. (1996). EMBO J. 15, 4713-4722. MEDLINE
Rastelli L, Chan CS, Pirrotta V. (1993). EMBO J. 12, 1513-1522. MEDLINE
Rea S, Eisenhaber F, O'Carroll D, Strahl BD, Sun ZW, Schmid M, Opravil S, Mechtler K, Ponting CP, Allis CD, Jenuwein T. (2000). Nature 406, 593-599. Article MEDLINE
Roberts CW, Galusha SA, McMenamin ME, Fletcher CD, Orkin SH. (2000). Proc. Natl. Acad. Sci. USA 97, 13796-13800. MEDLINE
Rosen C, Dorsett D, Jack J. (1998). Genetics 148, 1865-1874. MEDLINE
Rozenblatt-Rosen O, Rozovskaia T, Burakov D, Sedkov Y, Tillib S, Blechman J, Nakamura T, Croce CM, Mazo A, Canaani E. (1998). Proc. Natl. Acad. Sci. USA 95, 4152-4157. Article MEDLINE
Satijn DP, Otte AP. (1999). Biochim. Biophys. Acta 1447, 1-16. MEDLINE
Schlossherr J, Eggert H, Paro R, Cremer S, Jack RS. (1994). Mol. Gen. Genet. 243, 453-462. MEDLINE
Schumacher A, Magnuson T. (1997). Trends Genet. 13, 167-170. Article MEDLINE
Sedkov Y, Tillib S, Mizrokhi L, Mazo A. (1994). Development 120, 1907-1917. MEDLINE
Sevenet N, Sheridan E, Amram D, Schneider P, Handgretinger R, Delattre O. (1999). Am. J. Hum. Genet. 65, 1342-1348. Article MEDLINE
Sewalt RG, Van DV, Gunster MJ, Hamer KM, den Blaauwen JL, Satijn DP, Hendrix T, van Driel R, Otte AP. (1998). Mol. Cell. Biol. 18, 3586-3595. MEDLINE
Shao Z, Raible F, Mollaaghababa R, Guyon JR, Wu CT, Bender W, Kingston RE. (1999). Cell 98, 37-46. MEDLINE
Sharp EJ, Abramova NA, Park WJ, Adler PN. (1997). Chromosoma 106, 70-80. MEDLINE
Simon J, Chiang A, Bender W, Shimell MJ, O'Connor M. (1993). Dev. Biol. 158, 131-144. Article MEDLINE
Soeller WC, Poole SJ, Kornberg T. (1988). Genes Dev. 2, 68-81. MEDLINE
St Johnston D, Nusslein-Volhard C. (1992). Cell 68, 201-219. MEDLINE
Stassen MJ, Bailey D, Nelson S, Chinwalla V, Harte PJ. (1995). Mech. Dev. 52, 209-223. Article MEDLINE
Strahl BD, Allis CD. (2000). Nature 403, 41-45. Article MEDLINE
Strobeck MW, Knudsen KE, Fribourg AF, DeCristofaro MF, Weissman BE, Imbalzano AN, Knudsen ES. (2000). Proc. Natl. Acad. Sci. USA 97, 7748-7753. Article MEDLINE
Struhl G, Akam M. (1985). EMBO J. 4, 3259-3264. MEDLINE
Strutt H, Cavalli G, Paro R. (1997). EMBO J. 16, 3621-3632. Article MEDLINE
Strutt H, Paro R. (1997). Mol. Cell Biol. 17, 6773-6783. MEDLINE
Tamkun JW, Deuring R, Scott MP, Kissinger M, Pattatucci AM, Kaufman TC, Kennison JA. (1992). Cell 68, 561-572. MEDLINE
Tie F, Furuyama T, Harte PJ. (1998). Development 125, 3483-3496. MEDLINE
Tillib S, Petruk S, Sedkov Y, Kuzin A, Fujioka M, Goto T, Mazo A. (1999). Mol. Cell Biol. 19, 5189-5202. MEDLINE
Trouche D, Le Chalony C, Muchardt C, Yaniv M, Kouzarides T. (1997). Proc. Natl. Acad. Sci. USA 94, 11268-11273. Article MEDLINE
Tsukiyama T, Becker PB, Wu C. (1994). Nature 367, 525-532. MEDLINE
Tsukiyama T, Daniel C, Tamkun J, Wu C. (1995). Cell 83, 1021-1026. MEDLINE
van Lohuizen M, Tijms M, Voncken JW, Schumacher A, Magnuson T, Wientjens E. (1998). Mol. Cell Biol. 18, 3572-3579. MEDLINE
van Lohuizen M, Verbeek S, Scheijen B, Wientjens E, van der GH, Berns A. (1991). Cell 65, 737-752. MEDLINE
van der Vlag J, Otte AP. (1999). Nat. Genet. 23, 474-478. Article MEDLINE
Varga-Weisz PD, Becker PB. (1998). Curr. Opin. Cell Biol. 10, 346-353. MEDLINE
Vazquez M, Moore L, Kennison JA. (1999). Development 126, 733-742. MEDLINE
Versteege I, Sevenet N, Lange J, Rousseau-Merck MF, Ambros P, Handgretinger R, Aurias A, Delattre O. (1998). Nature 394, 203-206. Article MEDLINE
Wade PA, Wolffe AP. (1999). Curr. Biol. 9, R221-R224. MEDLINE
Wakimoto BT. (1998). Cell 93, 321-324. MEDLINE
Wallrath LL. (1998). Curr. Opin. Genet. Dev. 8, 147-153. MEDLINE
Wallrath LL, Elgin SC. (1995). Genes Dev. 9, 1263-1277. MEDLINE
Wang W, Xue Y, Zhou S, Kuo A, Cairns BR, Crabtree GR. (1996). Genes Dev. 10, 2117-2130. MEDLINE
Waring PM, Cleary ML. (1997). Curr. Top. Microbiol. Immunol. 220, 1-23. MEDLINE
Weiler KS, Wakimoto BT. (1995). Annu. Rev. Genet. 29, 577-605. MEDLINE
Wilkins RC, Lis JT. (1997). Nucleic Acids Res. 25, 3963-3968. MEDLINE
Wolffe AP, Matzke MA. (1999). Science 286, 481-486. Article MEDLINE
Workman JL, Kingston RE. (1998). Annu. Rev. Biochem. 67, 545-579. MEDLINE
Yu BD, Hanson RD, Hess JL, Horning SE, Korsmeyer SJ. (1998). Proc. Natl. Acad. Sci. USA 95, 10632-10636. MEDLINE
Yu BD, Hess JL, Horning SE, Brown GA, Korsmeyer SJ. (1995). Nature 378, 505-508. MEDLINE
Zhang HS, Gavin M, Dahiya A, Postigo AA, Ma D, Luo RX, Harbour JW, Dean DC. (2000). Cell 101, 79-89. MEDLINE
Zink B, Engstrom Y, Gehring WJ, Paro R. (1991). EMBO J. 10, 153-162. MEDLINE
Zink B, Paro R. (1989). Nature 337, 468-471. MEDLINE
Zink D, Paro R. (1995). EMBO J. 14, 5660-5671. MEDLINE
Figure 1 A summary of the interactions involved in gene regulation by PcG/trxG proteins as discussed in the text. PcG proteins seem to utilize at least two distinct mechanisms to repress gene expression. The PC containing PRC1 complex renders the nucleosomal array resistant to remodeling by SWI/SNF remodelers. PC can directly bind nucleosomes and may promote a closed chromatin structure that is less accessible to other transcription factors. The action of PRC1 does not involve HDAC activity, and neither binding of PC to nucleosomes nor the competition between PRC1 and SWI/SNF depends on the histone tails. In contrast, the PcG protein EED, which is part of a distinct complex, mediates repression via association with a HDAC. It is still unclear how PcG/trxG complexes are recruited to their targets but this process may involve PHO, Zeste and GAGA, PcG/trxG proteins, which possess sequence-specific DNA binding activity. Although normally thought of as activators, the trxG proteins Zeste and GAGA may also be involved in recruitment of PcG complexes. Moreover, GAGA can associate with a HDAC complex. The transiently expressed gap protein HB may play a pivotal role in the transition between establishment and maintenance of a silenced state. HB recruits the remodeling and HDAC activity containing dMi-2 complex that could, directly or indirectly, recruit PcG complexes. Zeste activates transcription by recruitment of the BRM complex through direct binding to the OSA and MOR subunits that are also trxG proteins. The mechanisms of activation by TRX and GAGA remain unclear. However, TRX associates with PREs and interacts with the SNF5 subunit of the BRM complex. It should be noted that GAGA and Zeste also bind promoters and may function to anchor distant regulatory elements to the correct target genes
|28 May 2001, Volume 20, Number 24, Pages 3055-3066|
|Table of contents Previous Article Next [PDF]