![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
European Molecular Biology Laboratory, Gene Expression Programme, 69117 Heidelberg, Germany
 |
Abstract |
|---|
 Top Abstract ResultsDiscussionMaterials and methodsAcknowledgmentsReferences |
|---|
[Keywords: PcG/trxG; Polycomb response element; histone methylation; quantitative chromatin immunoprecipitation; HOX gene; Drosophila]
Received March 26, 2006; revised version accepted May 24, 2006.
). The classic example for this kind of regulation is the maintenance of HOX gene expression in Drosophila, where PcG and trxG regulators were first identified through genetic studies (for review, see Kennison 1995). PcG proteins are defined as regulators required for keeping Drosophila HOX genes repressed in cells in which they must stay inactive, whereas trxG proteins are defined as proteins required to keep HOX genes active in cells in which their products are needed.
Biochemical studies revealed that PcG and trxG proteins exist in distinct multimeric protein complexes that bind to and modify chromatin in in vitro assays (for review, see Ringrose and Paro 2004). To date, three different Drosophila PcG protein complexes have been purified: PhoRC, PRC1, and PRC2. PhoRC contains the DNA-binding PcG protein Pho that targets the complex to PREs, and dSfmbt, a novel PcG protein that selectively binds to histone H3 and H4 tail peptides that are mono- or dimethylated at H3-K9 or H4-K20 (H3-K9me1/2 and H4-K20me1/2, respectively) (Klymenko et al. 2006). PRC1 contains the PcG proteins Ph, Psc, Sce/Ring, and Pc (Shao et al. 1999). PRC1 inhibits nucleosome remodeling and transcription in in-vitro assays (Francis et al. 2001; King et al. 2005) and its subunit Pc specifically binds to trimethylated K27 in histone H3 (H3-K27me3) (Fischle et al. 2003; Min et al. 2003). PRC2, finally, contains the PcG proteins E(z), Su(z)12, and Esc as well as Nurf55, and this complex functions as a histone methyltransferase (HMTase) that specifically methylates K27 in histone H3 (H3-K27) in nucleosomes (Cao et al. 2002; Czermin et al. 2002; Kuzmichev et al. 2002; Müller et al. 2002).
trxG proteins also exist in several different multiprotein complexes and, intriguingly, two of those also possess HMTase activity. Notably, TAC1 contains Trithorax (Trx), an HMTase with specificity for H3-K4 (Petruk et al. 2001; Smith et al. 2004), whereas Ash1, an HMTase reported to methylate H3-K4, H3-K9, and H4-K20 (Beisel et al. 2002; Byrd and Shearn 2003), exists in a different, currently uncharacterized protein assembly (Papoulas et al. 1998). Thus, HMTases that target distinct histone lysine residues and proteins that bind methylated histone lysine residues appear to play a central role in transcriptional control by the PcG/trxG system.
Regulation of HOX genes by PcG proteins requires specific cis-regulatory sequences that are typically several hundred base pairs in length and are called Polycomb response elements (PREs) (Simon et al. 1993; Chan et al. 1994). In reporter gene assays in Drosophila embryos and larvae, PREs act as potent general silencers that prevent transcriptional activation by a variety of different transcription factors (Chan et al. 1994; Zink and Paro 1995; Dellino et al. 2004; Sengupta et al. 2004). Consistent with the finding that silencing by PREs requires PcG gene function, chromatin immunoprecipitation (X-ChIP) analyses showed that PhoRC, PRC1, and PRC2 are all specifically associated with PREs in Drosophila embryos and larvae (e.g., Orlando et al. 1998; Cao et al. 2002; Klymenko et al. 2006). Recent X-ChIP studies reported that H3-K27 methylation is specifically enriched at PREs (Cao et al. 2002; Ringrose et al. 2004; Wang et al. 2004). This led to a model where the role of H3-K27 methylation is to recruit PRC1 to PREs (Cao et al. 2002; Ringrose et al. 2004; Wang et al. 2004). Finally, although PREs act as silencers in all reporter gene assays described to date, several studies provided evidence that the Trx protein binds to and acts through DNA sequences that are either identical or closely linked to PRE sequences (Orlando et al. 1998; Tillib et al. 1999).
But how do the different chromatin-modifying and chromatin-binding activities encoded by PcG and trxG regulators generate transcriptional ON and OFF states at HOX promoters? Genetic studies suggest that there is a tightly balanced antagonism between PcG and trxG regulators. Specifically, in ash1 or trx mutants, expression of HOX genes is lost in their normal expression domains, but HOX expression is restored in ash1 and trx mutants that also lack PcG gene function (Klymenko and Müller 2004). Moreover, such trxG/PcG double mutants also show severe misexpression of HOX genes, comparable to PcG single mutants (Klymenko and Müller 2004). These observations suggest that Ash1 and Trx HMTases are not needed for transcriptional activation but are required to prevent "default" PcG silencing of HOX genes in cells in which these genes must stay active (Klymenko and Müller 2004). However, the molecular mechanisms by which these two trxG HMTases prevent PcG silencing have remained elusive.
To gain insight into PcG/trxG transcriptional control, we compared the chromatin of the HOX gene Ultrabithorax (Ubx) in its ON and OFF state in imaginal disc cells in developing Drosophila larvae. Our quantitative analysis of X-ChIP reveals that most PcG and trxG proteins are constitutively bound at PREs, independently of the transcriptional state. However, we find that differential histone trimethylation at the Ubx promoter and coding region, generated by PcG and trxG HMTases, is critical for maintaining transcriptional ON and OFF states.
|
Results |
|---|
|
TopAbstractResultsDiscussionMaterials and methodsAcknowledgmentsReferences |
|---|
; Christen and Bienz 1994), F4 and F5 are located within the core of the bxd PRE (Chan et al. 1994; Fritsch et al. 1999), F7 is located within the BXD enhancer that activates expression in the embryo but is inactive in discs (Müller and Bienz 1991; Castelli-Gair et al. 1992), F9 encompasses the Ubx transcription start site (Saari and Bienz 1987), F13 corresponds to the bx PRE (Chiang et al. 1995; Orlando et al. 1998), and F17 is located in the Ubx 3' untranslated region. The following regions located outside of the Ubx locus served as controls. A first control, F18, is located within the iab-7 PRE of the HOX gene Abd-B (Fig. 1B; Mihaly et al. 1997). Since Abd-B is repressed in both wing and haltere/third-leg imaginal discs, X-ChIP signals at the iab-7 PRE are expected to be comparable, if not identical, in wing and haltere/third-leg disc cells. Three additional control sequences (F19–F21) are located in intergenic regions in euchromatin, whereas F22 is located in heterochromatin (Fig. 1B).
|
We first analyzed the binding of PhoRC, PRC1, and PRC2 in UbxON and UbxOFF chromatin, using antibodies against different subunits of each complex. For PhoRC, we used antibodies against Pho and dSfmbt (Klymenko et al. 2006). For PRC1, we used antibodies against Psc, Sce/Ring, Pc, and Ph (Francis et al. 2001). For PRC2, we used antibodies against Su(z)12 (Czermin et al. 2002; Müller et al. 2002) as well as an antibody against Pcl, a protein that has been reported to be associated with a fraction of PRC2 in Drosophila embryos (O’Connell et al. 2001; Tie et al. 2003; M. Nekrasov and J. Müller, unpubl.). We observe high-level binding of PhoRC, PRC1, and PRC2 at the bxd, bx, and iab-7 PREs (Fig. 2). Surprisingly, however, we measured comparable X-ChIP signals in UbxOFF and UbxON chromatin, not only at the iab-7 PRE where this is expected, but also at the bxd and bx PREs (Fig. 2). Thus, all three PcG protein complexes are bound at Ubx PREs in both the OFF and ON state.
|
Constitutive binding of Trx, but differential bindingof Ash1 in the ON and OFF state
We next analyzed binding of the trxG proteins Trx and Ash1. Trx is specifically bound at the bxd, bx, and iab-7 PREs and at the Ubx promoter but, as in the case of PcG proteins, Trx is bound to all of these sequences at comparable levels in UbxOFF and in UbxON chromatin (Fig. 3). In sharp contrast, Ash1 is only bound to Ubx in UbxON chromatin, where it is localized 
1 kb downstream of the Ubx transcription start site (Fig. 3, F10). We note that Ash1 signals at PREs are at background levels (Fig. 3, F4/F5, F13, F18), a result that is in disagreement with the recently reported binding of Ash1 to the bxd PRE (Sanchez-Elsner et al. 2006). Taken together, our experiments show that Trx and Ash1 show very distinct binding patterns, consistent with them being part of distinct protein complexes (Papoulas et al. 1998; Petruk et al. 2001). Trx behaves like PcG proteins, it is constitutively bound at PREs in the ON and OFF state, whereas Ash1 is bound downstream of the transcription start site, but only in the ON state (see Fig. 7 below, for summary).
|
We next wanted to analyze the relationship between the binding of PcG and trxG proteins and the status of histone lysine methylation in Ubx chromatin in the ON and OFF state. As a first step toward this experiment, we performed X-ChIP analyses with antibodies that recognize different portions of unmodified H3, H4, or the linker histone H1. With each of these antibodies, we detect comparable levels of histones at all regions except at the PREs and at the Ubx promoter. Specifically, we observe a sixfold to 10-fold reduction of histone signals in the region of the Ubx PREs in both UbxOFF and UbxON chromatin (Fig. 4A, F4, F5, F13; note, we also find similar results for H2A; data not shown). Histone signals are also reduced at the Ubx promoter, where the reduction is consistently more pronounced in UbxON chromatin (sixfold to 10-fold) than in UbxOFF chromatin (twofold) (Fig. 4A, F9). The observation that lower levels of core and linker histones are detected at PREs and at the promoter could have different reasons. One possibility would be that the histone epitopes at PREs and at the promoter are masked because of binding of PcG, trxG, or other protein complexes, and that nucleosomes present in these regions can therefore not be detected by immunoprecipitation. However, it is important to note that the same observations were made with different antibodies that recognize different epitopes on H3, H4, H2A, and H1. Another possible explanation for the reduced histone signals at PREs and the Ubx promoter could thus be that these regions are depleted of nucleosomes.
|
We next performed immunoprecipitations with antibodies against methylated H3-K4. Recent studies showed that H3-K4me3 is highly localized in the vicinity of transcription start sites of active genes (e.g., Ng et al. 2003; Bernstein et al. 2005; Wirbelauer et al. 2005). X-ChIP analysis with antibodies against H3-K4me3 (Fig. 5A) and H3-K4me2 (data not shown) shows that both of these modifications are localized at and 1 kb downstream of the transcription start site in UbxON chromatin, but not in UbxOFF chromatin (Fig. 5A,wt, F9–F10; data not shown). H3-K4me3 signals thus correlate with transcriptional activity at the Ubx promoter. Note that the relative H3-K4 trimethylation at the transcription start site (F9) and in the 5' coding region (F10) is at comparable levels if normalized to H3 signals.
|
; Wang et al. 2004). In contrast to those studies, we find widespread H3-K27 trimethylation across the whole Ubx gene (Fig. 5A, wt). In particular, in UbxOFF chromatin, H3-K27me3 signals are almost uniform across the whole gene (Fig. 5A, wt). The extent of reduction of H3-K27me3 signals at PREs and at the Ubx promoter correlates very well with the reduction of H3 signals in these regions, (cf. Figs. 5A [wt] and 4).
Interestingly, in UbxON chromatin, H3-K27me3 signals in the upstream control region are comparable to those seen in UbxOFF chromatin (Fig. 5A, wt, F1–F7). However, H3-K27me3 signals are reduced to background levels at the promoter and in the 5' coding region (F9–F12) in the ON state, suggesting that lack of H3-K27 trimethylation in these regions may be critical to permit transcription of the Ubx gene. Previous studies (Ebert et al. 2004) and our immunological analysis of imaginal disc cell clones that are homozygous for E(z) or Su(z)12-null mutations (SOM2) show that all H3-K27 methylation in Drosophila is generated by PRC2. These results therefore provide strong evidence that PRE-tethered PRC2 does not only methylate nucleosomes in the vicinity of PREs, but that it methylates extended stretches of flanking chromatin.
We also tested for the presence of H3-K9me3 (Fig. 5A) and H4-K20me3 (Fig. 5B), two histone methylation marks that are characteristic for pericentric heterochromatin (Ebert et al. 2004; Schotta et al. 2004). No H3-K9me3 and only very low H4-K20me3 signals are detected in the euchromatic control regions F19–F21, but we find that H3-K9me3 is strongly enriched in the heterochromatic control region F22 (see Fig. 5A, wt, for H3-K9me3, and B for H4-K20me3). At the Ubx gene, the H3-K9me3 and H4-K20me3 profiles are strikingly similar to the H3-K27me3 profile (Fig. 5A,B). First, in UbxOFF chromatin, H3-K9me3 and H4-K20me3 are present at high levels across the whole Ubx locus (Fig. 5A,B). Second, in UbxON chromatin, H3-K9me3 and H4-K20me3 signals are high in the upstream control region but, as in the case of H3-K27me3, they are reduced to low levels in the promoter and 5' coding region (Fig. 5A,B).
Thus, we find that in the OFF state, three "repressive" histone lysine modifications—H3-K27me3, H3-K9me3, and H4-K20me3—are present throughout the Ubx gene, while in the ON state all three marks are restricted to the upstream control region and are specifically absent at the Ubx promoter and in the coding region. The localization of these modifications is summarized in Figure 7, below. It is intriguing that high levels of H3-K27me3, H3-K9me3, and H4-K20me3 trimethylation are present at the PBX enhancer (F1) in the ON state. PBX is strictly required for Ubx transcription in imaginal discs (White and Wilcox 1985). These results therefore suggest that high levels of H3-K27, H3-K9, and H4-K20 trimethylation at the PBX enhancer do not interfere with the ability of this enhancer to promote transcriptional activation in the ON state.
Loss of H3-K4me3 and gain of repressive histone trimethylation in ash1 mutants
To investigate the role of trxG HMTases, we next performed X-ChIP analysis in larvae that are homozygous for ash122, an ash1-null mutation (Tripoulas et al. 1996). ash122 homozygotes develop into pupal stages and show loss of Ubx expression in most cells in the haltere and third-leg imaginal disc, while a small fraction of cells still maintain apparently undiminished expression of Ubx (Fig. 5A; cf. LaJeunesse and Shearn 1995; Klymenko and Müller 2004). In such ash122 mutant haltere and third-leg discs, H3-K4me3 signals in the promoter and 5' coding region are fivefold reduced (Fig. 5A, ash1–). Complementing this observation, we find that in ash122 mutant haltere and third-leg discs, the "repressing" H3-K27 and H3-K9 trimethylation in the Ubx promoter and coding region is increased to levels that are comparable to those observed in wing discs (Fig. 5A, ash1–). This suggests that Ash1 function prevents Pc-repressive H3-K27 and H3-K9 trimethylation in the promoter and coding region in the ON state. The reduction of H3-K4me3 signal in ash1 mutants is consistent with Ash1 trimethylating H3-K4 as reported by earlier studies (Beisel et al. 2002; Byrd and Shearn 2003; Sanchez-Elsner et al. 2006).
We note that studies from the Sauer lab reported that Ash1 also methylates H3-K9 and H4-K20 (Beisel et al. 2002) and, according to Sanchez-Elsner et al. (2006), Ash1 tri-methylates H3-K4, H3-K9, and H4-K20 in the region of the bxd PRE and at the Ubx promoter in haltere/third-leg discs. Our results are at odds with several of the data reported by Sanchez-Elsner et al. (2006). For example, we do not find H3-K4me3 at the bxd PRE and we do not detect H3-K9me3 and H4-K20me3 at the Ubx promoter in wild-type haltere/third-leg discs. Moreover, our observation that ash1 mutant haltere/third-leg discs show a gain of H3-K9 trimethylation at the Ubx promoter is difficult to reconcile with the proposed H3-K9 HMTase activity of Ash1.
Loss of repressive histone trimethylation in E(z) mutants
We next explored how removal of the PcG HMTase E(z) affects histone methylation. E(z)-null mutants die as larvae with extremely poorly developed imaginal discs, and we therefore performed X-ChIP analysis in larvae that are homozygous for the temperature-sensitive allele E(z)61. When E(z)61 mutant embryos were allowed to complete embryogenesis at the permissive temperature and were then shifted to the restrictive temperature during larval development (96 h at 30°C), they show high-level misexpression of Ubx in 20% of wing disc cells (Fig. 5A; Jones and Gelbart 1990). HOX misexpression in E(z)61 mutant larvae at the restrictive temperature is thus less severe than in clones of E(z) protein-null mutant cells (Müller et al. 2002), suggesting that the E(z)61 protein still retains some E(z)+ function. Even with this residual activity, we find almost complete loss of H3-K27 trimethylation at the promoter and in the 5' coding region (F9–F11) in E(z)61 mutant wing discs; H3-K27me3 signals in these regions drop to the levels of H3-K27me3 signals seen in haltere/third-leg discs; i.e., in UbxON chromatin [Fig. 5A, E(z)–]. However, H3-K27me3 signals are only slightly reduced in the 3' coding region in wing discs and they are largely unchanged in the upstream control region in both wing and haltere/third-leg discs [Fig. 5A, E(z)–]. Thus, the promoter and 5' coding region show the most severe loss of H3-K27me3 signals in E(z)61 mutants. It is possible that H3-K27me3 loss in this region is more dramatic than in other regions because F9, F10, and F11 are located far away from PRE-tethered PRC2. However, distance is unlikely to be the only explanation, since F16, a region as distant from the bx PRE as F9 and F10, still shows considerable levels of H3-K27 trimethylation in E(z)61 mutant wing discs. Perhaps the loss of H3-K27me3 at the promoter and 5' coding region is acerbated by exchange of histones due to initiation of (abortive) transcription, a possibility that we have not explored.
Interestingly, E(z)61 mutant wing discs also show localized loss of H3-K9me3 signals in the same regions where H3-K27me3 signals are lost (F9–F11) but, again, the levels of H3-K9 trimethylation in the upstream control region in wing and haltere/third-leg discs is largely unchanged and only slightly reduced in the 3' coding region in wing discs [Fig. 5A, E(z)–].
In vitro, Drosophila PRC2 preferentially methylates H3-K27 in nucleosomes and it only poorly methylates H3-K9 (Müller et al. 2002). We cannot exclude the possibility that, in vivo, PRC2 methylates both H3-K27 and H3-K9, but we think it is equally possible that a different, presently unidentified HMTase is responsible for H3-K9 methylation at Ubx. For example, H3-K27 trimethylation of HOX gene chromatin by PRC2 may be a prerequisite for subsequent trimethylation by a H3-K9-specific HMTase. We also note that E(z)61 mutant wing discs show significant H3-K4me3 signals in the promoter and 5' coding region, consistent with the observed transcriptional activation in E(z)61 mutant cells [Fig. 5A, E(z)–].
Finally, we also analyzed the binding of the PRC1 subunits Pc and Ph in E(z)61 mutant discs. We find that in both wing and haltere/third-leg discs, Pc and Ph binding at PREs is unaffected (SOM3) (data not shown).
The repressed Ubx gene is extensively trimethylated at H3-K27, H3-K9, and H4-K20
The results described above establish that trimethylation at H3-K27, H3-K9, and H4-K20 at the Ubx promoter and coding region tightly correlates with transcriptional repression. We next wanted to explore to what extent these histone lysine residues are in the tri-, di- or monomethylated state. We performed X-ChIP in wild-type animals using antibodies specific for dimethylated H3-K9 and for di- or monomethylated H4-K20 and H3-K27, respectively.
For H3-K9me2 we find strong signal in the heterochromatic control region F22, but no signals are detected at the Ubx gene in either UbxOFF or UbxON chromatin (Fig. 5C). In the case of H4-K20, we find that H4-K20me1 is strongly enriched in the transcribed region (F10–F17) in UbxON chromatin, but that in UbxOFF chromatin, signals are at low levels throughout the gene (Fig. 5C). Similarly, H4-K20me2 signals are also low in all regions (data not shown). With antibodies against H3-K27me2 and against H3-K27me1, we find that signals at the Ubx gene in UbxOFF chromatin are even lower than in the control regions F19–F22 (data not shown).
Taken together, these observations support the idea that in the OFF state, the whole Ubx gene is extensively trimethylated at H3-K27, H3-K9, and H4-K20 and that only a minor fraction of nucleosomes may be mono- or dimethylated at these sites. In the ON state, the promoter and coding region lack nucleosomes that are trimethylated at H3-K27, H3-K9, and H4-K20 and nucleosomes of lower methylated states—e.g., H4-K20me1—are present instead (see Fig. 7, below, for summary).
General transcription factors and factors involvedin transcriptional elongation are differentially boundin the ON and OFF state
Finally, we examined how the recruitment and localization of general transcription factors correlates with the histone methylation state at the active and inactive Ubx gene. We first analyzed the binding of TATA-binding protein (TBP) and the elongation factor Spt5 in wild-type animals. TBP is bound at the Ubx promoter in UbxON chromatin and, albeit at two- to threefold lower levels, also in UbxOFF chromatin (Fig. 6, F9). This is consistent with earlier studies that showed that TBP is bound to repressed HOX promoters in Drosophila tissue culture cells (Breiling et al. 2001, 2004). In agreement with earlier studies that reported that TBP is bound to the bxd PRE in embryos (Dellino et al. 2004) we find that TBP is bound to the bxd, bx, and iab-7 PREs in both UbxON and UbxOFF chromatin (Fig. 6, F4, F5, F13, F18).
|
). In particular, studies on the Drosophila hsp70 gene showed that Spt5 is already bound to the hsp70 promoter before heat-shock induction and has a role in restricting elongation of paused RNA Pol II but that, after heat shock, Spt5 is required for efficient elongation of the released RNA Pol II (Saunders et al. 2003; Wu et al. 2003). Importantly, biochemical analyses suggest that both steps involve physical association of Spt5 with RNA Pol II (Krogan et al. 2002; Lindstrom et al. 2003). Presence of Spt5 in the coding region of genes thus serves as a marker for paused as well as for elongating RNA Pol II. At the Ubx transcription start site, we find that Spt5 is bound not only in UbxON chromatin but also in UbxOFF chromatin, but, as in the case of TBP, Spt5 levels are about threefold lower at the inactive Ubx promoter (Fig. 6; F9). However, across the coding region we observe significant Spt5 signals only in UbxON chromatin but not in UbxOFF chromatin (Fig. 6; F10–F17). These observations are consistent with a model where Spt5 is associated with a stalled RNA Pol II at the Ubx promoter in the OFF state, but that elongating RNA Pol II complexes with Spt5 are only present in UbxON chromatin. Unexpectedly, we also observe strong binding of Spt5 at the bxd, bd, and iab-7 PREs at comparable levels in both UbxON and UbxOFF chromatin (Fig. 6; F4, F5, F13, F18).
Finally, we analyzed binding of Kismet (Kis), a CHD family ATPase (Daubresse et al. 1999). Kis was originally classified as a trxG protein because of its role in HOX gene regulation, but recent studies suggest that it plays a more general role in transcription by facilitating an early step in transcriptional elongation (Srinivasan et al. 2005). Consistent with this role, we find that Kis protein is bound to the Ubx promoter region only in UbxON chromatin, but not in UbxOFF chromatin (Fig. 6). Nevertheless, we also find Kis protein bound at the bxd, bx, and iab-7 PREs, where it is bound at comparable levels in UbxOFF and UbxON chromatin (Fig. 6).
The observation that TBP, Spt5, and Kis are all bound not only at the promoter but also at PREs is puzzling. It is tempting to speculate that PREs do not only serve as binding platforms for PcG protein complexes but may also constitute sites for assembly of different transcription factor complexes. However, it should be emphasized that TBP, Spt5, and Kis signals at PREs are comparable in UbxON and UbxOFF chromatin; the binding at PREs therefore does not correlate with the differential binding that these proteins show at the promoter (Fig. 6). Binding of TBP, Spt5, and Kis at PREs may therefore reflect a distinct function of these proteins at PREs. In this context, it should also be pointed out that the trxG protein Ash1 is not bound at PREs (Fig. 3), supporting the idea that not all factors participating in HOX gene regulation bind to PREs.
Taken together, we find that TBP and Spt5 are bound to the Ubx promoter in the ON and OFF state, whereas binding of Kis at the transcription start site and binding of Spt5 in the coding region are only observed in the ON state. The similarity with the situation at the inactive hsp70 promoter (Saunders et al. 2003; Wu et al. 2003) is striking. The presence of TBP and Spt5 at the inactive Ubx promoter suggests that Ubx transcription is blocked at a late step of transcriptional initation, and the absence of Kis suggests that the block occurs prior to the transition to transcriptional elongation. Repressive histone methylation at H3-K27, H3-K9, and H4-K20 may directly or indirectly participate in generating this block.
|
Discussion |
|---|
|
TopAbstractResultsDiscussionMaterials and methodsAcknowledgmentsReferences |
|---|
; Beuchle et al. 2001; Birve et al. 2001; Müller et al. 2002; Klymenko and Müller 2004; Klymenko et al. 2006). The following conclusions can be drawn from the analyses reported in this study and are schematically summarized in Figure 7. (1) The PcG protein complexes PhoRC, PRC1, and PRC2 and the Trx protein are all highly localized at PREs, but they are all constitutively bound at comparable levels in the OFF and ON state. (2) The trxG protein Ash1 is only bound in the ON state, where it is specifically localized 1 kb downstream of the transcription start site. (3) In the OFF state, PRC2 and other unknown HMTases trimethylate H3-K27, H3-K9, and H4-K20 over an extended 100-kb domain that spans the whole Ubx gene. (4) In the ON state, comparable H3-K27, H3-K9, and H4-K20 trimethylation is restricted to the upstream control regions and Ash1 selectively prevents this trimethylation in the promoter and coding region. (5) Repressed Ubx chromatin is extensively tri- but not di- or monomethylated at H3-K27, H3-K9, and H4-K20. (6) Trimethylation of H3-K27, H3-K9, and H4-K20 at imaginal disc enhancers in the upstream control region does not impair the function of these enhancers in the ON state. (7) TBP and Spt5 are bound at the Ubx transcription start site in the ON and OFF state, but Kis is only bound in the ON state. This suggests that in the OFF state, transcription is blocked at a late step of transcriptional initiation, prior to the transition to elongation.
|
, Ringrose et al. (2004), and Wang et al. (2004) reported that H3-K27 trimethylation is localized at PREs and this led to the model that recruitment of PRC1 to PREs occurs through H3-K27me3 (i.e., via the Pc chromodomain). In contrast, our study and the study by Vincent Pirrotta and colleagues found H3-K27 trimethylation to be present across the whole inactive Ubx gene, both in wing discs (this study) and in S2 cells (V. Pirrotta, pers. comm.). We have not been able to detect specific enrichment of H3-K27 trimethylation at PREs; rather, we observed a reduction of H3-K27me3 signals at PREs, consistent with the reduced signals of H3 that are detected at these sites. Consistent with our results, genome-wide analyses of PcG protein binding and H3-K27me3 profiles in S2 cells revealed that, at most PcG-binding sites in the genome, PcG proteins are tightly localized, whereas H3-K27 trimethylation is typically present across an extended domain that often spans the whole coding region (Schwartz et al. 2006). How could the differences between our study and the earlier studies be explained? It should be noted that in contrast to the qPCR analysis used in our study here, the studies by Cao et al. (2002), Ringrose et al. (2004), and Wang et al. (2004) all relied on nonquantitative end-point PCR after 36 or more cycles to assess the X-ChIP results. It is possible that these experimental differences account for the discrepancies. In the following we shall base our discussion on the findings reported in this study.
Localized binding of PcG protein complexesand widespread histone trimethylation
PhoRC, PRC1, and PRC2 are all tightly localized at PREs but they are all constitutively bound at the inactive and active Ubx gene. This suggests that recruitment of PcG complexes to PREs occurs by default. Although all three complexes are bound at comparable levels to the bxd PRE in the inactive and active state and PhoRC is also bound at comparable levels at the bx PRE, it should be pointed out that the levels of PRC1 and PRC2 binding at the bx PRE are about twofold reduced in the active Ubx gene compared with the inactive Ubx gene. Even though there is still high-level binding of PRC1 and PRC2 at the bx PRE, we cannot exclude that the observed reduction in binding helps to prevent default PcG repression of the active Ubx gene. It is possible that transcription through the bx PRE reduces PRC1 and PRC2 binding at this PRE. Transcription through PREs was recently proposed to serve as an "anti-silencing" mechanism that prevents default silencing of active genes by PREs (Schmitt et al. 2005).
The highly localized binding of all three PcG protein complexes at PREs, together with earlier studies on PRE targeting of PcG protein complexes (Mohd-Sarip et al. 2002, 2005; Wang et al. 2004; Klymenko et al. 2006) supports the idea that not only PhoRC but also PRC1 and PRC2 are targeted to PRE DNA through interactions with Pho and/or other sequence-specific DNA-binding proteins. In the case of trxG proteins, the binding modes are more diverse. In particular, recruitment of Trx protein to PREs and to the promoter is also constitutive in both states but recruitment of Ash1 to the coding region is clearly only observed at the active Ubx gene. At present, we do not know how Trx or Ash1 are targeted to these sites. It is possible that a transcription-coupled process recruits Ash1 to the position 1 kb downstream of the transcription start site.
In contrast to the localized and constitutive binding of PcG protein complexes and the Trx protein, we find that the patterns of histone trimethylation are very distinct in the active and inactive Ubx gene. Our results also suggest that the locally bound PcG and trxG HMTases act across different distances to methylate chromatin. For example, H3-K4 trimethylation is confined to the first kilobase of the Ubx coding region where Ash1 and Trx are bound, whereas H3-K27 trimethylation is present across an extended 100-kb domain of chromatin that spans the whole Ubx gene. This suggests that PRE-tethered PRC2 is able to trimethylate H3-K27 in nucleosomes that are as far as 30 kb away from the bxd or bx PREs. Unexpectedly, we found that the H3-K9me3 and H4-K20me3 profiles closely match the H3-K27me3 profile. At present it is not known which HMTases are responsible for H3-K9 and H4-K20 trimethylation, but our analysis of E(z) mutants indicate that these modifications may be generated in a sequential manner, following H3-K27 trimethylation by PRC2. The molecular mechanisms that permit locally tethered HMTases such as PRE-bound PRC2 to maintain such extended chromatin stretches in a trimethylated state are only poorly understood. However, a recent study showed that the PhoRC subunit dSfmbt selectively binds to mono- and dimethylated H3-K9 and H4-K20 in peptide-binding assays (Klymenko et al. 2006). One possibility would be that dSfmbt participates in the process that ensures that repressed Ubx chromatin is trimethylated at H3-K27, H3-K9, and H4-K20. For example, dSfmbt, tethered to PREs by Pho, may interact with nucleosomes of lower methylated states (i.e., H3-K9me1/2 or H4-K20me1/2) in the flanking chromatin and thereby bring them into the vicinity of PRE-anchored HMTases that will hypermethylate them to the trimethylated state.
Repressive and antirepressive HMTases target the promoter and coding region
Our analyses suggest that H3-K27, H3-K9, and H4-K20 trimethylation in the promoter and coding region is critical for Polycomb repression. First, although H3-K27, H3-K9, and H4-K20 trimethylation is present at the inactive and active Ubx gene, it is specifically depleted in the promoter and coding region in the active Ubx gene. Second, misexpression of Ubx in wing discs with impaired E(z) activity correlates well with loss of H3-K27 and H3-K9 trimethylation at the promoter and 5' coding region. It is possible that the persisting H3-K27 and H3-K9 trimethylation in the 3' coding region is responsible for maintenance of repression in those E(z) mutant wing discs cells that do not show misexpression of Ubx. Third, in haltere and third-leg discs of ash1 mutants, the promoter and coding region become extensively trimethylated at H3-K27 and H3-K9, and this correlates with loss of Ubx expression. Previous studies showed that Ubx expression is restored in ash1 mutants cells that also lack E(z) function (Klymenko and Müller 2004). Together, these findings therefore provide strong evidence that Ash1 is required to prevent PRC2 and other HMTases from trimethylating the promoter and coding region at H3-K27 and H3-K9. The loss of H3-K4 trimethylation in ash1 mutants is formally consistent with the idea that Ash1 exerts its antirepressor function by trimethylating H3-K4 in nucleosomes in the promoter and 5' coding region, but other explanations are possible.
But how might H3-K27, H3-K9, and H4-K20 trimethylation in the promoter and coding region repress transcription? Our observation that TBP and Spt5 are also bound to the promoter in the OFF state suggests that these methylation marks do not prevent assembly of the basic transcription apparatus at the promoter. However, the nucleosome remodeling factor Kis is not recruited in the OFF state, and transcription thus appears to be blocked at a late step of transcriptional intiation prior to elongation. We found that the low-level binding of Pc in the coding region correlates with the presence of H3-K27 trimethylation; i.e., Pc and H3-K27me3 are both present in the OFF state, but are absent in the ON state. One possible scenario would thus be that H3-K27 trimethylation in the promoter and coding region permits direct recruitment of PRC1. According to this view, locally recruited PRC1 would then repress transcription; e.g., by inhibiting nucleosome remodeling in the promoter region (Shao et al. 1999; King et al. 2005). However, several observations are not easily reconciled with such a simple "recruitment-by-methylation" model. First, peak levels of all PRC1 components are present at PREs and, apart from Pc, very little binding is observed outside of PREs. Second, excision of PRE sequences from a PRE reporter gene during development leads to a rapid loss of silencing (Sengupta et al. 2004), suggesting that transcriptional repression requires the continuous presence of PREs and the proteins that are bound to them. A second, in our opinion more plausible scenario would therefore be that DNA-binding factors first target PcG protein complexes to PREs, and that these PRE-tethered complexes then interact with trimethylated nucleosomes in the flanking chromatin in order to repress transcription. For example, it is possible that bridging interactions between the Pc chromodomain in PRE-tethered PRC1 and H3-K27me3-marked chromatin in the promoter or coding region permit other PRE-tethered PcG proteins to recognize the chromatin interval across which they should act, e.g., to inhibit nucleosome remodeling in the case of PRC1 or to trimethylate H3-K27 at hypomethylated nucleosomes in the case of PRC2.
Concluding remarks
The analysis of a HOX gene in developing Drosophila suggests that histone trimethylation at H3-K27, H3-K9, and H4-K20 in the promoter and coding region plays a central role in generating and maintaining of a PcG-repressed state. Contrary to previous reports, our findings provide no evidence that H3-K27 trimethylation is specifically localized at PREs and could thus recruit PRC1 to PREs; we find widespread H3-K27 trimethylation across the whole transcription unit. The data presented here provide evidence that PREs serve as assembly platforms for PcG protein complexes such as PRC2 that act over considerable distances to trimethylate H3-K27 across long stretches of chromatin. The presence of this trimethylation mark in the chromatin that flanks PREs may in turn serve as a signal to define the chromatin interval that is targeted by other PRE-tethered PcG protein complexes such as PRC1. The results reported here also provide a molecular explanation for the previously reported antirepressor function of trxG HMTases; selective binding of Ash1 to the active HOX gene blocks PcG repression by preventing PRC2 from trimethylating the promoter and coding region. It is possible that the extended domain of combined H3-K27, H3-K9, and H4-K20 trimethylation creates not only the necessary stability for transcriptional repression, but that it also provides the molecular marks that permits PcG repression to be heritably maintained through cell division.
|
Materials and methods |
|---|
|
TopAbstractResultsDiscussionMaterials and methodsAcknowledgmentsReferences |
|---|
For this study, antibodies against Pc15–78 and against Pho324–520 were raised in rabbits. Epitopes for antibody production were expressed as 6xHis-tagged fusion proteins in Escherichia coli. Antibodies against Pc15–78 and Su(z)12448–802 (Müller et al. 2002) were affinity-purified on the same epitope used for immunization.
Immunostaining of discs
Staining of imaginal discs was performed as described (Beuchle et al. 2001).
X-ChIP from Drosophila imaginal discs using qPCR analysis
The immunoprecipitation protocol is a modified version of the X-ChIP protocol described in Orlando et al. (1997) and is described in SOM4. Details about the quantitative analysis of immunoprecipitated material are also described under SOM4. The antibodies used in this study are listed in Table 1.
|
|
Acknowledgments |
|---|
|
TopAbstractResultsDiscussionMaterials and methodsAcknowledgmentsReferences |
|---|
|
Footnotes |
|---|
E-MAIL Juerg.Mueller{at}embl.de; FAX 49-6221-387518. 
Article is online at http://www.genesdev.org/cgi/doi/10.1101/gad.388706
|
References |
|---|
|
TopAbstractResultsDiscussionMaterials and methodsAcknowledgmentsReferences |
|---|
Beisel C., Imhof A., Greene J., Kremmer E., Sauer F. 2002. Histone methylation by the Drosophila epigenetic transcriptional regulator Ash1. Nature 419: 857–862.[CrossRef][Medline]
Bernstein B.E., Kamal M., Lindblad-Toh K., Bekiranov S., Bailey D.K., Huebert D.J., McMahon S., Karlsson E.K., Kulbokas E.J. III, Gingeras T.R. et al. 2005. Genomic maps and comparative analysis of histone modifications in human and mouse. Cell 120: 169–181.[CrossRef][Medline]
Beuchle D., Struhl G., Müller J. 2001. Polycomb group proteins and heritable silencing of Drosophila Hox genes. Development 128: 993–1004.[Abstract]
Birve A., Sengupta A.K., Beuchle D., Larsson J., Kennison J.A., Rasmuson-Lestander A., Müller J. 2001. Su(z)12, a novel Drosophila Polycomb group gene that is conserved in vertebrates and plants. Development 128: 3371–3379.
Breiling A.B., Turner M., Bianchi M.E., Orlando V. 2001. General transcription factors bind promoters repressed by Polycomb group proteins. Nature 412: 651–655.[CrossRef][Medline]
Breiling A., O’Neill L.P., D’Eliseo D., Turner B.M., Orlando V. 2004. Epigenome changes in active and inactive Polycomb-group controlled regions. EMBO Rep. 5: 976–982.[CrossRef][Medline]
Byrd K.N. and Shearn A. 2003. ASH1, a Drosophila trithorax group protein, is required for methylation of lysine 4 residues on histone H3. Proc. Natl. Acad. Sci. 100: 11535–11540.
Cao R., Wang L., Wang H., Xia L., Erdjument-Bromage H., Tempst P., Jones R.S., Zhang Y. 2002. Role of histone H3 lysine 27 methylation in Polycomb-group silencing. Science 298: 1039–1043.
Castelli-Gair J., Müller J., Bienz M. 1992. Function of an Ultrabithorax minigene in imaginal cells. Development 114: 877–886.[Abstract]
Chan C.S., Rastelli L., Pirrotta V. 1994. A Polycomb response element in the Ubx gene that determines an epigenetically inherited state of repression. EMBO J. 13: 2553–2564.[Medline]
Chiang A., O’Connor M.B., Paro R., Simon J., Bender W. 1995. Discrete Polycomb-binding sites in each parasegmental domain of the bithorax complex. Development 121: 1681–1689.[Abstract]
Christen B. and Bienz M. 1994. Imaginal disc silencers from Ultrabithorax: Evidence for Polycomb response elements. Mech. Dev. 48: 255–266.[CrossRef][Medline]
Czermin B., Melfi R., McCabe D., Seitz V., Imhof A., Pirrotta V. 2002. Drosophila enhancer of Zeste/ESC complexes have a histone H3 methyltransferase activity that marks chromosomal Polycomb sites. Cell 111: 185–196.[CrossRef][Medline]
Daubresse G., Deuring R., Moore L., Papoulas O., Zakrajsek I., Waldrip W.R., Scott M.P., Kennison J.A., Tamkun J.W. 1999. The Drosophila kismet gene is related to chromatin-remodeling factors and is required for both segmentation and segment identity. Development 126: 1175–1187.[Abstract]
DeCamillis M., Cheng N.S., Pierre D., Brock H.W. 1992. The polyhomeotic gene of Drosophila encodes a chromatin protein that shared polytene chromosome-binding sites with Polycomb. Genes & Dev. 6: 223–232.
Dellino G.I., Schwartz Y.B., Farkas G., McCabe D., Elgin S.C.R., Pirrotta V. 2004. Polycomb silencing blocks transcription initiation. Mol. Cell 13: 887–893.[CrossRef][Medline]
Ebert A., Schotta G., Lein S., Kubicek S., Krauss V., Jenuwein T., Reuter G. 2004. Su(var) genes regulate the balance between euchromatin and heterochromatin in Drosophila.. Genes & Dev. 18: 2973–2983.
Fischle W., Wang Y., Jacobs S.A., Kim Y., Allis C.D., Khorasanizadeh S. 2003. Molecular basis for the discrimination of repressive methyl-lysine marks in histone H3 by Polycomb and HP1 chromodomains. Genes & Dev. 17: 1870–1881.
Francis N.J., Saurin A.J., Shao Z., Kingston R.E. 2001. Reconstitution of a functional core polycomb repressive complex. Mol. Cell 8: 545–556.[CrossRef][Medline]
Fritsch C., Brown J.L., Kassis J.A., Müller J. 1999. The DNA-binding polycomb group protein pleiohomeotic mediates silencing of a Drosophila homeotic gene. Development 126: 3905–3913.[Abstract]
Jones R.S. and Gelbart W.M. 1990. Genetic analysis of the Enhancer of zeste locus and its role in gene-regulation in Drosophila melanogaster.. Genetics 126: 185–199.[Abstract]
Kennison J.A. 1995. The Polycomb and trithorax group proteins of Drosophila: Trans-regulators of homeotic gene function. Annu. Rev. Genet. 29: 289–303.[CrossRef][Medline]
King I.F., Emmons R.B., Francis N.J., Wild B., Müller J., Kingston R.E., Wu C.T. 2005. Analysis of a polycomb group protein defines regions that link repressive activity on nucleosomal templates to in vivo function. Mol. Cell. Biol. 25: 6578–6591.
Klymenko T. and Müller J. 2004. The histone methyltransferases Trithorax and Ash1 prevent transcriptional silencing by Polycomb group proteins. EMBO Rep. 5: 373–377.[CrossRef][Medline]
Klymenko T., Papp B., Fischle W., Köcher T., Schelder M., Fritsch C., Wild B., Wilm M., Müller J. 2006. A Polycomb group protein complex with sequence-specific DNA-binding and selective methyl-lysine-binding activities. Genes & Dev. 1110–1122.
Krogan N.J., Kim M., Ahn S.H., Zhong G., Kobor M.S., Cagney G., Emili A., Shilatifard A., Buratowski S., Greenblatt J.F. 2002. RNA polymerase II elongation factors of Saccharomyces cerevisiae:A targeted proteomics approach. Mol. Cell. Biol. 22: 6979–6992.
Kuzmichev A., Nishioka K., Erdjument-Bromage H., Tempst P., Reinberg D. 2002. Histone methyltransferase activity associated with a human multiprotein complex containing the Enhancer of Zeste protein. Genes & Dev. 16: 2893–2905.
