Polycomb group (PcG) proteins repress homeotic genes in cells where these genes must remain inactive during development. This repression requires cis-acting silencers, also called PcG response elements. Currently, these silencers are ill-defined sequences and it is not known how PcG proteins associate with DNA. Here, we show that the Drosophila PcG protein Pleiohomeotic binds to specific sites in a silencer of the homeotic gene Ultrabithorax. In an Ultrabithorax reporter gene, point mutations in these Pleiohomeotic binding sites abolish PcG repression in vivo. Hence, DNA-bound Pleiohomeotic protein may function in the recruitment of other non-DNA-binding PcG proteins to homeotic gene silencers.

The body plan of higher eukaryotes depends on spatially restricted expression of homeotic genes (Lewis, 1978; Duboule and Dollé, 1989; Graham et al., 1989, McGinnis and Krumlauf, 1992; Salser and Kenyon, 1994). Cis-regulatory sequences controlling homeotic gene expression have been characterised in Drosophila and in vertebrates (Simon et al., 1990; Müller and Bienz, 1991; Qian et al., 1991; Zink et al., 1991; Püschel et al., 1991; Gérard et al., 1993; Sharpe et al., 1998). In Drosophila homeotic genes, two types of regulatory sequences have been identified by reporter gene assays in transformed animals: enhancers and silencers (for review see Bienz and Müller, 1995). Enhancers fall into two broad classes: early enhancers, which transiently direct transcription at the blastoderm stage, and late enhancers, which become active only after gastrulation and direct expression in embryonic or imaginal tissues. Early enhancers activate transcription exclusively within the limits of the corresponding homeotic gene expression domain, whereas most late enhancers – if linked individually to a reporter gene – are active not only within but also outside of homeotic gene expression domains. Such misexpression is suppressed if certain DNA fragments, called silencers, are linked to late enhancers in reporter gene constructs (Müller and Bienz, 1991; Busturia and Bienz, 1993; Simon et al., 1993; Chan et al., 1994; Christen and Bienz, 1994). Since this silencing depends on PcG gene function (Müller and Bienz, 1991; Busturia and Bienz, 1993; Simon et al., 1993; Chan et al., 1994; Christen and Bienz, 1994), these silencers were called PcG response elements (Simon et al., 1993). Most known PcG proteins do not bind to DNA directly but bind to the chromatin of homeotic genes (Zink and Paro, 1989; Zink et al., 1991; DeCamillis et al., 1992). PcG protein binding regions in chromatin have been mapped by formaldehyde cross-linking PcG proteins to DNA (Orlando and Paro, 1993). This approach identified the same DNA fragments that were previously found to act as silencers in functional assays (Strutt et al., 1997; Orlando et al., 1998). Currently, PcG response elements are only poorly defined sequences, several hundred base pairs in length and, despite extensive cross-linking studies it is not known how PcG proteins bind to DNA (Orlando and Paro, 1993; Strutt and Paro, 1997; Strutt et al., 1997; Orlando et al., 1998).

The product of the PcG gene pleiohomeotic (pho), a zinc finger protein related to the mammalian transcription factor YY1, was recently identified as a factor that binds to a silencing fragment from the engrailed gene (Brown et al., 1998). The role of this DNA fragment in the regulation of engrailed expression is not known but it functions as a pairing- sensitive silencer if linked to a mini-white reporter gene (Kassis, 1994). This pairing-sensitive silencing of mini-white requires an intact PHO protein binding site and is partially impaired in pho mutants (Brown et al., 1998); however, the effects of mutations in PcG proteins on pairing-sensitive silencing are variable and highly dependent on the chromosomal insertion site (Kassis, 1994 and J. K., unpublished data). Thus, it remains unclear which PcG proteins actually mediate the pairing-sensitive silencing. Here we analysed the requirement for PHO protein binding sites in a bona fide PcG response element of a homeotic gene. One of the best-studied PcG response elements is a 1.6 kb fragment from the Ultrabithorax (Ubx) gene (Chan et al., 1994). This fragment was named PRE (Chan et al., 1994) and we shall refer to it as PRE to distinguish it from the term PRE, the generally used abbreviation for PcG response elements (Simon et al., 1993). We show here that PHO binds to PRE and that the PHO binding sites are essential for PcG repression in vivo. These experiments establish a direct link between a PcG protein and its target site in a homeotic gene.

Drosophila transformation, mutant strains and staining procedures

Transformants were generated as previously described (Bienz et al., 1988). Mutations used in this study were PcXT109, a protein null allele (Franke et al., 1995), and pho1 and phob alleles, which contain stop codons upstream of the DNA-binding domain and are presumed nulls (Brown et al., 1998). pho homozygotes were identified either by their misexpression phenotype or by using a 4th chromosome marked with ci-lacZ. A TM6B Tb chromosome was used to identify Pc heterozygotes. We found that pho1 and phob homozygotes are indistinguishable with respect to misexpression of reporter genes and Ubx. X-gal stainings were done as described (Christen and Bienz, 1994) and antibody stainings using fluorescence were done following standard protocols.

Plasmid constructions

The basic IDE-Ubx-lacZ reporter gene (based on a Carnegie 20 transformation vector) containing the 2.8 kb PBX-41 fragment called IDE linked to the proximal Ubx promoter and lacZ-coding region has been described (Christen and Bienz, 1994). This construct was modified to contain the 0.6 kb embryonic PBX enhancer pbxSB (Zhang et al., 1991) upstream of IDE. Unique KpnI and XbaI sites were engineered between PBX and IDE to insert the various PRE subfragments. Transformants carrying the PBX-IDE-Ubx-lacZ and the PBX-PRE1.6-IDE-Ubx-lacZ construct were kindly provided by M. Bienz and G. Struhl (personal communication); in these two cases, the basic lacZ reporter gene had been further modified to contain a nuclear localisation signal N-terminal to the β-gal coding region and, in addition, the transformation vector backbone carried the yellow gene rather than rosy as selective marker.

PRE1.6 corresponds to 2212StR1.6 (Chan et al., 1994) and subfragments of it were obtained by subcloning or by PCR. The sequence of PRE1.6 can be obtained from GenBank (acc. no. L32205). For simplicity, the sequence of the ten first and ten last nucleotides of each fragment preceded and followed by the nucleotide position in L32205 are given here:

PRE1.6 (1.56 kb): 33106gaattcaaaa… agcgccaagg34667;

PREa (0.58 kb): 33106gaattcaaaa… tgataaggtc33683;

PREb (0.96 kb): 33519atatgcaaccc… aagagcgtgc34479;

PREc (0.48 kb): 34184gctccgtcgc… agcgccaagg34667;

PRED (0.57 kb): 33683ccataatctt… ctcataatcg34249.

Sequencing of the PRED fragment used in this study suggests that there are several deviations from the database sequence, two of these putative polymorphisms affect PHO-binding sites. The first deviation generates the sequence GCCATCTC that corresponds to site 3 in our PREd fragment while the database sequence is GCCTTCTC; this latter sequence probably would not constitute a PHO protein binding site. The second deviation is at site 5: in our PRED fragment it has the sequence ACCATTAC, the database sequence is GCCATTAC, which binds PHO even stonger than site 4 (see below).

PRED Pho mut was obtained by substituting two to three nucleotides (shown in bold) in the conserved core of all six PHO sites using site-directed mutagenesis; see Fig. 3 for wild-type sequence.

site 1: CACGGAAGCACGAACGGCAG

site 2: CGCAGCTGTTAGCATGCGCG (note that in this case we also mutated a second potential PHO binding site on the opposite strand)

site 3: ACGGTTAGATATCTCGCTCG

site 4: CTCCGTCGCACGAACTGTCG

site 5: TAAAACGATCGGTACGAACG

site 6: TTATGAGGCACGCTCAGTCG

The distal end of PRED starts with CCATA and is preceded by polylinker sequences, this sequence was altered to CCGCG in PREDPho mut.

The complete pho coding region from the pho 12a cDNA (Brown et al., 1998) was subcloned downstream of the β-globin 5’ UTR and ATG in pT7link (provided by R. Treisman) to obtain PHO2-520pT7.

Detailed maps of plasmids are available on request.

Electromobility shift assays

Radiolabeled double-stranded probes of the sequences listed in Fig. 3B were generated by annealing the corresponding single-stranded oligonucleotides containing a (dG)3 overhang at the 5 end. The ends were then filled with [32P]-dCTP using Klenow polymerase followed by phenol extraction and separation of the probe from unincorporated nucleotides over a G-25 column. [14C]-Leu-labeled full-length PHO protein was in vivo translated from the PHO2-520pT7 template using the TNT reticulocyte lysate system (Promega); integrity of the labeled protein was checked by SDS-page followed by autoradiography. For binding tests, 2-5 fmol DNA probe and 2 μl of the in vivo translation reaction were incubated for 20 minutes on ice in a 20 μl reaction (100 mM KCl, 35 mM Hepes 7.9, 1 mM DTT, 50 μΜ ZnCl2, 12% glycerol, 2 mM spermidine, 1 mg/ml BSA, 0.1 mg/ml dI:dC). DNA-protein complexes were resolved at 4°C on a native 4% polyacrylamide gel at 10 V/cm using 0.5× TB as running buffer; the gel was pre-run for 1 hour prior to loading. The gel was fixed, dried and exposed for autoradiography.

In addition to the oligos shown in Fig. 2, we also tested additional oligos using the same gel-shift assay. Of three further double-stranded oligos with CCAT motifs from the PBX region, the oligos hb2/3 (5'ATAATTTTTTGCCATGGCTAATAAAA3) and hb6 (5'ACGGGA-ATGCGCCATAAAAAATGTGT3) were not bound by PHO protein whereas hb4 (5'AGAGCCGTCGGCCATT-AAAAAAGGTG3) was bound by PHO. We also note that PHO sites 1, 2, 3, 4 and 6 in PRED all have a G 5’ to the CCAT motif whereas PHO site 5 lacks such a G. In our PRED fragment, PHO site 5 has the sequence ACCATTAC whereas the database sequence is GCCATTAC. We found that changing the 5’ A to a G in the PHO site 5 oligo increases its affinity for PHO protein binding; this altered site binds even stronger than site 4 (C. F. and J. M., data not shown). Conversely, although oligos ‘D’ from PREd (see Fig. 2 for sequence) as well as oligos hb2/3 and hb6 from PBX do contain a GCCAT motif, they still failed to bind in our assay. Thus, it appears that, besides the critical CCAT core motif, a 5’ G is required but this G can be compensated for by appropriate nucleotides 3’ to the CCAT motif, i.e. the sequence TAC as in the case of PHO site 5 from PRED.

Fig. 1.

PRE-mediated silencing in imaginal discs. (A) PRE subfragments (black bars) were inserted upstream of IDE into the PBX-IDE-Ubx-lacZ reporter gene (top) to avoid unspecific blocking of IDE by PRE. The subfragments are drawn to scale (see Materials and Methods). (B) β-gal expression in wing and haltere discs of transformant lines carrying the reporter gene indicated on the left was visualized by X-gal staining. In all cases, the anterior compartment of the disc is to the left. Transformants carrying the PBX-IDE reporter gene without a PRE insert (no PRE) and PREA transformants show β-gal expression throughout ps 4, 5 and 6 although in wing discs typically small patches without β-gal expression are present; the wing disc of the no PRE line shown here lacks such unstained patches. In PRE1.6, PREB and PRED transformants β-gal expression is restricted to the posterior compartment of the haltere disc (ps 6) and no staining is present in wing and anterior haltere discs (ps 4 and 5) due to silencing by the PRE fragment. Partial silencing is observed in several PREC lines; shown here are discs from the weakly silenced line PREC line 33.1. (C) Loss of PRE-mediated silencing in PcG mutants. In Pc heterozygotes, the PRED reporter gene is misexpressed in small patches in the wing and anterior haltere disc; such misexpression was never observed in wild-type discs (see above). Note that these animals still carry one wild-type copy of the Pc gene. pho homozygotes show extensive misexpression of the reporter gene in wing and haltere discs. See Table 1 for more details.

Fig. 1.

PRE-mediated silencing in imaginal discs. (A) PRE subfragments (black bars) were inserted upstream of IDE into the PBX-IDE-Ubx-lacZ reporter gene (top) to avoid unspecific blocking of IDE by PRE. The subfragments are drawn to scale (see Materials and Methods). (B) β-gal expression in wing and haltere discs of transformant lines carrying the reporter gene indicated on the left was visualized by X-gal staining. In all cases, the anterior compartment of the disc is to the left. Transformants carrying the PBX-IDE reporter gene without a PRE insert (no PRE) and PREA transformants show β-gal expression throughout ps 4, 5 and 6 although in wing discs typically small patches without β-gal expression are present; the wing disc of the no PRE line shown here lacks such unstained patches. In PRE1.6, PREB and PRED transformants β-gal expression is restricted to the posterior compartment of the haltere disc (ps 6) and no staining is present in wing and anterior haltere discs (ps 4 and 5) due to silencing by the PRE fragment. Partial silencing is observed in several PREC lines; shown here are discs from the weakly silenced line PREC line 33.1. (C) Loss of PRE-mediated silencing in PcG mutants. In Pc heterozygotes, the PRED reporter gene is misexpressed in small patches in the wing and anterior haltere disc; such misexpression was never observed in wild-type discs (see above). Note that these animals still carry one wild-type copy of the Pc gene. pho homozygotes show extensive misexpression of the reporter gene in wing and haltere discs. See Table 1 for more details.

Fig. 2.

PHO protein binds to PRED in vivo. (A) EMSA with radiolabeled PRED subfragments 1-6 and A-D. Labeled DNA probes were incubated with in vivo translated PHO (‘PHO’ lanes) or with reticulocyte lysate from a mock translation reaction (‘lysate’ lanes). In the presence of Pho, probes 1-6 formed specific complexes (white dots) that were not observed with lysate alone. Note that a 3-base pair substitution in the strongest binding site 4 abolishes binding of PHO protein (4mut). No stable PHO-specific complexes were observed with probes A-D; it is unclear why in these cases complex formation by other binding activities in the lysate is reduced in the presence of PHO. (B) Sequences of PREd subfragments used for EMSA. A consensus sequence derived from fragments 1-6 and from the binding site in engrailed originally used to isolate PHO (Brown et al., 1998) suggests GCCATTAC is the optimal PHO binding site (see Materials and Methods for further discussion of this). Below, arrows, drawn to scale, mark the position, length and orientation (5'>3’) of these subfragments within PREd (long thin line); black ovals indicate PHO protein. (C-F) Binding of PHO protein to BXC and PREd in vivo. PHO protein bound to polytene chromosomes was visualized by histochemical staining. (C) A strong PHO signal (arrow) is present at band 89E, the location of the BXC. (D) In situ hybridization of a lacZ probe to polytene chromosomes from a PREd transformant line shows a chromosomal insertion site of 46D (arrow). (E) This reporter gene generates a new PHO protein binding site (arrow) that is not present (open arrowhead) in control animals (F); two endogenous chromosomal PHO binding sites are marked for reference (dot and asterisk, respectively). The dot corresponds to the engrailed locus (48A).

Fig. 2.

PHO protein binds to PRED in vivo. (A) EMSA with radiolabeled PRED subfragments 1-6 and A-D. Labeled DNA probes were incubated with in vivo translated PHO (‘PHO’ lanes) or with reticulocyte lysate from a mock translation reaction (‘lysate’ lanes). In the presence of Pho, probes 1-6 formed specific complexes (white dots) that were not observed with lysate alone. Note that a 3-base pair substitution in the strongest binding site 4 abolishes binding of PHO protein (4mut). No stable PHO-specific complexes were observed with probes A-D; it is unclear why in these cases complex formation by other binding activities in the lysate is reduced in the presence of PHO. (B) Sequences of PREd subfragments used for EMSA. A consensus sequence derived from fragments 1-6 and from the binding site in engrailed originally used to isolate PHO (Brown et al., 1998) suggests GCCATTAC is the optimal PHO binding site (see Materials and Methods for further discussion of this). Below, arrows, drawn to scale, mark the position, length and orientation (5'>3’) of these subfragments within PREd (long thin line); black ovals indicate PHO protein. (C-F) Binding of PHO protein to BXC and PREd in vivo. PHO protein bound to polytene chromosomes was visualized by histochemical staining. (C) A strong PHO signal (arrow) is present at band 89E, the location of the BXC. (D) In situ hybridization of a lacZ probe to polytene chromosomes from a PREd transformant line shows a chromosomal insertion site of 46D (arrow). (E) This reporter gene generates a new PHO protein binding site (arrow) that is not present (open arrowhead) in control animals (F); two endogenous chromosomal PHO binding sites are marked for reference (dot and asterisk, respectively). The dot corresponds to the engrailed locus (48A).

Fig. 3.

PHO binding sites in PRED are essential for PRE function in vivo. β-gal expression in wild-type imaginal discs of PRED and PRED pho mut transformants visualized by X-gal staining. Point mutations in PHO binding sites 1-6 in PRED abolish silencing function; the PRED pho mut reporter gene is expressed in wing and haltere discs comparable to the PBX-IDE reporter gene without a PRE insert (no PRE,Fig. 1).

Fig. 3.

PHO binding sites in PRED are essential for PRE function in vivo. β-gal expression in wild-type imaginal discs of PRED and PRED pho mut transformants visualized by X-gal staining. Point mutations in PHO binding sites 1-6 in PRED abolish silencing function; the PRED pho mut reporter gene is expressed in wing and haltere discs comparable to the PBX-IDE reporter gene without a PRE insert (no PRE,Fig. 1).

Immunostaining of polytene chromosomes

Rabbit polyclonal PHO antibodies were generated against a gel-isolated HIS-tag/PHO full-length fusion protein. The production, affinity purification and characterization of the antibody will be presented elsewhere (J. L. B. and J. A. K., unpublished data). The specificity of the antibody was demonstrated by inhibition or supershifting of PHO/DNA complexes in the gel-shift assay, by the detection of a single protein species on westerns of 0-12 hour nuclear embryonic extracts and by the absence of PHO staining on polytene chromosomes of pho1 mutant larvae. Affinity-purified and crude PHO antisera give the same banding pattern on polytene chromosomes.

Antibody staining of polytene chromosomes

Polytene chromosomes were fixed in 2% formaldehyde, 40% glacial acidic acid for 2.5 minutes. Under these fixation conditions, approximately 35 PHO bands are observed. Slides were washed in phosphate-buffered saline for 30 minutes, incubated in blocking buffer (PBS, 0.5% BSA, 0.1% Tween 20) for 30 minutes and then incubated with a 1/200 dilution of the crude PHO serum overnight at 4°C. Signals were developed using the secondary antibody and HRP detection system from the Vectastain ABC Elite kit (Vector Laboratories). The signal was enhanced by the inclusion of 0.008% NiCl2 and 0.008% CoCl2 in the HRP reaction.

To dissect the 1.6 kb Ubx PRE, we used a Ubx-lacZ reporter gene to monitor silencing capacity of PRE subfragments. We previously identified an embryonic enhancer, called PBX and an imaginal disc enhancer, called IDE, which are both located about 30 kb upstream of the Ubx transcription start site (Müller and Bienz, 1991; Castelli-Gair et al., 1992). PBX directs expression in early embryos in a pattern similar to Ubx with a sharp anterior boundary in parasegment 6 (ps 6) (Müller and Bienz, 1991). In contrast, if IDE is linked to a reporter gene it activates transcription not only in haltere discs where endogenous Ubx is expressed but also in wing discs where Ubx is not expressed (Castelli-Gair et al., 1992; Chan et al., 1994; Christen and Bienz, 1994; White and Wilcox, 1984; Beachy et al., 1985). A PBX-IDE reporter gene is thus active within Ubx expression boundaries in early embryos but is later expressed also outside of the Ubx domain, i.e. in the wing disc. We therefore tested whether PRE or subfragments thereof would silence this misexpression if inserted into the PBX-IDE reporter gene.

First, we inserted the 1.6 kb PRE (Chan et al., 1994) between the PBX and IDE enhancers and introduced this reporter gene (PRE1.6) into flies (Fig. 1). Whereas PBX-IDE transformants without the PRE fragment show nearly uniform β-galactosidase (β-gal) expression in wing and haltere discs, β-gal expression in PRE1.6 transformants is confined to the posterior compartment of haltere discs (Fig. 1; Table 1). The boundary between β-gal-positive and β-gal-negative cells runs through the middle of the haltere disc and apparently coincides with the ps 6 compartment boundary. Thus, IDE activity is completely suppressed anterior to ps 6 but is unaffected in ps 6 itself. This suggests that PRE1.6 silences the reporter gene anterior to ps 6 and thereby preserves the anterior expression boundary delimited by PBX in the embryo. We note that the expression pattern directed by PBX in the embryo is not silenced by PRE1.6 (data not shown, see also Discussion). After we intiated these experiments, an independent study by Pirrotta and co-workers showed that the 1.6 kb PRE is able to silence misexpression of a reporter construct that is very similar to our PBX-IDE reporter gene (Poux et al., 1996).

Table 1.

Transformant lines and their expression pattern in individual compartments of wing and haltere discs

Transformant lines and their expression pattern in individual compartments of wing and haltere discs
Transformant lines and their expression pattern in individual compartments of wing and haltere discs

We next tested subfragments of the 1.6 kb PRE for silencing function. Silencing anterior to ps 6 was also observed in imaginal discs of two PREB lines and in all four PRED lines (Fig. 1; Table 1). In contrast, only one of five PREC lines showed substantial silencing anterior to ps 6; two lines showed partial silencing and two lines showed no silencing at all (Fig. 1; Table 1). None of the PREA lines showed silencing; these transformants showed β-gal staining in imaginal discs similar to transformants carrying the PBX-IDE reporter gene without PRE (Fig. 1; Table 1). Taken together, these data suggest that the PRE silencer is contained in the central 567 bp PRED fragment.

We asked whether the silencing mediated by the PRE fragments depends on PcG gene function. We tested all PRED and several PRE1.6 lines in Pc heterozygotes and found in each case small patches of β-gal staining in the wing disc and in the anterior part of the haltere disc (Fig. 1; Table 1). Thus, a reduction in Pc gene dosage leads to a partial loss of silencing; the extent of the observed misexpression is comparable to the misexpression of the endogenous Ubx gene in Pc heterozygotes (Fig. 4). We then examined the patterns of PRED lines in larvae homozygous for a pho mutation. We found in each case that pho mutant wing and haltere discs show an extensive loss of silencing (Fig. 1). These results demonstrate that silencing by PRED requires PcG gene function.

Fig. 4.

pho represses the endogenous Ubx gene in wing discs. Confocal images of wing discs stained with Ubx antibody. In the wild type, no Ubx expression is detected in the wing disc proper; the labeled cells discernible here and out of focus in the pictures below are part of the peripodial membrane and/or trachea (cf. White and Wilcox, 1984; Beachy et al., 1985). pho1 homozygotes alone and PcXT109 heterozygotes alone show misexpression of Ubx in the wing disc, but only in a few cells (white arrowheads). In pho1 homozygotes that are also heterozygous for PcXT109 silencing of Ubx is almost completely lost and Ubx protein is strongly expressed in most wing disc cells.

Fig. 4.

pho represses the endogenous Ubx gene in wing discs. Confocal images of wing discs stained with Ubx antibody. In the wild type, no Ubx expression is detected in the wing disc proper; the labeled cells discernible here and out of focus in the pictures below are part of the peripodial membrane and/or trachea (cf. White and Wilcox, 1984; Beachy et al., 1985). pho1 homozygotes alone and PcXT109 heterozygotes alone show misexpression of Ubx in the wing disc, but only in a few cells (white arrowheads). In pho1 homozygotes that are also heterozygous for PcXT109 silencing of Ubx is almost completely lost and Ubx protein is strongly expressed in most wing disc cells.

We next examined whether PHO protein binds directly to PRED. PHO contains a DNA-binding domain with very high similarity to the DNA-binding domain of YY1, which is known to bind to the sequence G/t C/t/a CATN T/a T/g/c (Hyde-DeRuyscher et al., 1995). The PRED fragment contains several motifs that match versions of this YY1 protein binding site. Oligos spanning each of these motifs were tested for PHO binding in gel-shift assays (Fig. 2). We found that PHO protein formed a specific complex with six of the ten tested oligos (Fig. 2A). These and additional binding tests with other oligos suggest GCCATTAC as an optimal binding site for PHO (see Fig. 2B and Material and Methods for further details). To test whether PHO protein binds to the PRED construct in vivo, we generated antibodies against the PHO protein. On polytene chromosomes from salivary glands, PHO antibodies bind to approximately 35 different loci. The strongest signal was found at the location of the Bithorax-Complex (BXC), suggesting that PHO protein is bound to the BXC genes (Fig. 2C). Furthermore, on polytene chromosomes of a PRED transformant line, we found a strong additional signal at the transposon insertion site (Fig. 2D-F). These data suggest that PHO protein binds directly to PREd in vitro and in vivo.

We then tested whether PHO protein binding sites are needed for silencing in imaginal discs. We mutated all six PHO binding sites in the PRED fragment by altering two or three nucleotides in each CCAT core motif (see Materials and Methods). The introduced base changes abolish binding of PHO protein in vitro (Fig. 2). The mutated PRED fragment was inserted into the PBX-IDE reporter gene to obtain PRED pho mut. PRED pho mut transformants show uniform β-gal staining in wing and haltere discs comparable to transformants carrying the reporter gene without PRE (Fig. 3). Thus, mutations in the PHO binding sites abolish PRE function. Note that PHO protein binding sites 4-6 are also present in the PREC fragment that overlaps PRED (Fig. 1 and Materials and Methods). It is possible that these sites are responsible for the partial silencing observed in some of the PREC transformant lines (Fig. 1; Table 1). Taken together, these experiments provide strong evidence that PHO protein binds directly to PRE and is required for silencing.

Finally, we analysed expression of the endogenous Ubx gene in imaginal discs of pho mutants. Animals that are homozygous for pho null mutations develop into pharate adults with only relatively mild homeotic transformations (Gehring, 1970; Breen and Duncan, 1986; Girton and Jeon, 1994). Consistent with this, we found that pho1 and phob homozygotes show only slight misexpression of Ubx in wing and antennal discs (Fig. 4 and data not shown). The observed misexpression is comparable to the misexpression of Ubx in Pc heterozygotes (Fig. 4). We note that, in pho mutants, the PRED reporter gene shows substantially more misexpression than the endogenous Ubx gene (Fig. 1). Thus, silencing of the reporter gene is more sensitive to the lack of pho product than the native Ubx gene. Animals that are mutant for two different PcG mutations often show more severe misexpression of homeotic genes and consequently enhanced homeotic transformations compared to the single mutants by themselves (Jurgens, 1985). pho homozygotes that are also heterozygous for Pc show very dramatic misexpression of Ubx in wing and other discs (Fig. 4). Thus, in this genetically sensitized background due to only one rather than two copies of Pc, pho is required to repress Ubx in all imaginal disc cells.

Previous studies using a formaldehyde cross-linking assay showed that Polycomb protein is specifically associated with chromatin encompassing PRE1.6 (Strutt et al., 1997; Orlando et al., 1998). However, no direct link has been made between Pc or any other PcG protein and this DNA. Our experiments here establish a direct physical link between the PcG protein PHO and PRE1.6 and demonstrate that PHO binding is essential for repression. In a database search, sequence motifs resembling PHO binding sites have been noted in cis-regulatory regions from many Drosophila homeotic genes (Mihaly et al., 1998). Our functional data suggest that PHO protein binding sites might constitute an essential, integral part of PcG response elements. The results presented here support the view that PHO may act to recruit and anchor PcG proteins to the DNA (see below).

Early function of pho

Most pho mutant embryos, which lack maternal wild-type pho product, fail to develop altogether and the rare putatively paternally rescued embryos which do develop die with segmentation defects and homeotic transformations (Breen and Duncan, 1986). In contrast, if maternal pho product is present, pho homozygotes survive to pharate adults. This suggests that pho function is particularly important in the very early embryo. Here, we found that mutation of the PHO binding sites in the PREd pho mut reporter gene abolish silencing in all disc cells. Thus, it appears that if PHO protein is prevented from binding to PRE, i.e. in the PREd pho mut reporter gene, silencing is probably never established. Conversely, silencing of the PRED reporter gene is only partially lost in larvae homozygous for a pho null mutation (compare Figs 1 and 3). Thus, in pho homozygous embryos (which contain maternal PHO protein) silencing of the PRED reporter is probably established but is subsequently lost in imaginal discs. In summary, these observations strongly suggest that maternally deposited PHO protein is crucial for the establishment of silencing but that zygotic PHO protein is required for complete silencing.

Repression by PHO

How does PHO repress transcription? One possible model is that PHO protein recruits other non-DNA-binding PcG proteins to DNA to form ‘silencing complexes’ (Bienz and Müller, 1995). We imagine that such silencing complexes interact with proteins at the proximal promoter to prevent recruitment or formation of active RNA-Polymerase II complexes (Bienz and Müller, 1995). However, PHO protein binds to DNA in a sequence-specific manner and pho RNA and protein are expressed in all cells throughout embryonic development (Brown et al., 1998 and J. L. B. and J. A. K., unpublished observations). Why then does PHO protein not repress Ubx in all cells? One possibility is that PHO protein only has access to PRE DNA in cells where Ubx is repressed. A more likely scenario is that PHO protein also binds to PRE in cells where Ubx is expressed but that PHO is unable to repress on its own, e.g. because it cannot recruit functional PcG complexes by itself. Previous studies showed that repression by the gap protein Hunchback (HB) determines where Ubx is turned off in the early embryo (White and Lehmann, 1986; Irish et al., 1989; Qian et al., 1991; Müller and Bienz, 1992; Zhang and Bienz, 1992). dMi-2, the fly homologue of a mammalian histone deacetylase subunit, was recently identified as a HB-interacting protein that is needed for repression of Ubx by HB (Kehle et al., 1998). It is possible that chromatin modifying activities of a HB:dMi-2 complex are a prerequisite for DNA-bound PHO to recruit other PcG proteins in the embryo. A different scenario would be that assembly of repressive PcG complexes occurs by default unless the linked homeotic gene promoter is transcriptionally activated in the early embryo (Pirrotta et al., 1995; Poux et al., 1996). This model is based on previous studies on the 1.6 kb PRE which revealed an unusually potent silencing property of this fragment (Pirrotta et al., 1995; Poux et al., 1996). In particular, the 1.6 kb PRE fragment could silence linked imaginal disc enhancers even within the Ubx expression domain in some reporter constructs (Pirrotta et al., 1995; Poux et al., 1995). Since this unusual silencing occurred only in reporter genes lacking an early embryonic enhancer (e.g. PBX), this lead to the suggestion that early transcriptional activation is needed to prevent ubiquitous silencing by the 1.6 kb PRE fragment. According to this view, assembly of silencing complexes (e.g. recruited by PHO) and consequent repression would occur by default unless the reporter gene is transcriptionally activated in the early embryo (Poux et al., 1996; discussed in Bienz and Müller, 1995). It should be noted that at present none of the other identified silencers thought to contain PcG response elements has been found to indiscriminately prevent transcriptional activation by late-acting enhancers (Busturia and Bienz, 1993; Christen and Bienz, 1994). Furthermore, studies on a reporter gene that is inserted in the endogenous Ubx gene suggest that early transcriptional activation of this reporter gene alone is insufficient to prevent its subsequent repression by PcG proteins (McCall and Bender, 1996). Thus, it remains to be seen to what extent in the endogenous Ubx gene PcG-mediated repression occurs by default and to what extent establishment of this repression requires direct cooperation from gap proteins.

Maintenance function of PHO

Previous studies on PcG protein function suggest that both PcG proteins and PcG response elements are required throughout development to maintain silencing. In particular, experiments on a PcG response element from the Drosophila homeotic gene Abd-B showed that the silencer DNA itself is continuously required for PcG repression (Busturia et al., 1997). Furthermore, analysis of Pc mutant clones showed that Pc protein is required throughout development to silence homeotic genes (Busturia and Morata, 1988). The simplest model to explain these observations is that sequence-specific DNA- binding proteins are required for anchoring PcG proteins to the DNA throughout development. Since maternal PHO is not sufficient for complete silencing of the PRED reporter gene or of Ubx (see Figs 1, 4), we suggest that PHO may play a role in continuously anchoring PcG proteins to the DNA. Why then is the derepression of Ubx in pho homozygotes not more severe? It is unlikely that maternally deposited PHO is responsible for the residual silencing in pho mutant discs for two reasons. First, in embryos homozygous for a pho protein- null mutation, maternal PHO protein becomes undetectable by 9 hours of development (J. L. B. and J. A. K., unpublished data). Second, the extensive dilution during postembryonic cell divisions would virtually eliminate any persisting PHO protein molecules in imaginal disc cells of pho homozygotes. Thus, we imagine that PHO protein is required to continuously anchor PcG protein complexes on DNA but suggest that other, currently unidentified DNA-binding proteins can partially substitute for PHO protein in pho mutant larvae. We note that other PcG genes are present in duplicate in the Drosophila genome and can partially substitute for each other (Dura et al., 1987; Brunk et al., 1991; van Lohuizen et al., 1991; DeCamillis et al., 1992; Soto et al., 1995). Although we have not found another YY1- or PHO-related sequence by database searches (J. M., unpublished observations), there remains the possiblity that there is a second pho-like gene in Drosophila.

In summary, we have presented strong evidence that PHO is required for the activity of a Polycomb group response element from the Ubx gene. We have also shown that this element responds to Pc. Further, our observation that Pc-/+; pho-/- animals show nearly ubiquitous misexpression of Ubx in third instar larvae shows that PHO protein strongly synergises with Pc. However, we have no evidence that PHO and Pc interact and it remains to be seen whether recruitment of Pc to PRE occurs by PHO protein itself or through another DNA-binding protein.

We thank G. Struhl and Mariann Bienz for the gift of unpublished transformant lines 206 and 205 and Aidan Peterson and Jeffrey Simon for the His-tagged PHO fusion protein used to generate the PHO antibody and Tom Kornberg for ci-lacZ flies. We thank Christiane Nusslein-Volhard for encouragement and support and we thank her and Ralf Sommer for comments on the manuscript. J. L. B. and J. A. K. were supported by a grant from the Cystic Fibrosis foundation and by internal funds from the FDA and thank Phil Noguchi for his enthusiastic support of our work.

Accepted 11 June; published on WWW 5 August 1999

Beachy
,
P. A.
,
Helfand
,
S. L.
and
Hogness
,
D. S.
(
1985
).
Segmental distribution of bithorax complex proteins during Drosophila development
.
Nature
313
,
545
551
.
Bienz
,
M.
and
Müller
,
J.
(
1995
).
Transcriptional silencing of homeotic genes in Drosophila
.
BioEssays
17
,
775
784
.
Bienz
,
M.
,
Saari
,
G.
,
Tremml
,
G.
,
Müller
,
J.
,
Züst
,
B.
and
Lawrence
,
P. A.
(
1988
).
Differential regulation of Ultrabithorax in two germ layers of Drosophila
.
Cell
53
,
567
576
.
Breen
,
T. R.
and
Duncan
,
I. M.
(
1986
).
Maternal expression of genes that regulate the Bithorax complex of Drosophila melanogaster
.
Dev. Biol.
118
,
442
456
.
Brown
,
J. L.
,
Mucci
,
D.
,
Whiteley
,
M.
,
Dirksen
,
M.-L.
and
Kassis
,
J. A.
(
1998
).
The Drosophila Polycomb group gene pleiohomeotic encodes a sequence-specific DNA binding protein with homology to the multifunctional mammalian transcription factor YY1
.
Molecular Cell
1
,
1057
1064
.
Brunk
,
B. P.
,
Martin
E. C.
and
Adler
,
P. N.
(
1991
).
Drosophila genes Posterior sex combs and Suppressor two of zeste encode proteins with homology to the murine bmi-1 oncogene
.
Nature
353
,
351
353
.
Busturia
,
A.
and
Bienz
,
M.
(
1993
).
Silencers in Abdominal-B, a homeotic Drosopohila gene
.
EMBO J.
12
,
1415
1425
.
Busturia
,
A.
and
Morata
,
G.
(
1988
).
Ectopic expression of homeotic genes caused by the elimination of the Polycomb gene in Drosophila imaginal epidermis
.
Development
104
,
713
720
.
Busturia
,
A.
,
Wightman
,
C. D.
and
Sakonju
,
S.
(
1997
).
A silencer is required for maintenance of transcriptional repression throughout Drosophila development
.
Development
124
,
4343
4350
.
Castelli-Gair
,
J.
,
Müller
,
J.
and
Bienz
,
M.
(
1992
).
Function of an Ultrabithorax minigene in imaginal cells
.
Development
114
,
877
886
.
Chan
,
C.-S.
,
Rastelli
,
L.
and
Pirrotta
,
V.
(
1994
).
A Polycomb response element in the Ubx gene that determines an epigenetically inherited state of repression
.
EMBO J.
13
,
2553
2564
.
Christen
,
B.
and
Bienz
,
M.
(
1994
).
Imaginal disc silencers from Ultrabithorax: evidence for Polycomb response elements
.
Mech. Dev.
48
,
255
266
.
DeCamillis
,
M.
,
Chen
,
N.
,
Pierre
,
D.
and
Brock
,
H. W.
(
1992
).
The polyhomeotic gene of Drosophila encodes a chromatin protein that shares polytene chromosome-binding sites with Polycomb
.
Genes & Dev.
6
,
223
232
.
Duboule
,
D.
and
Dolle
,
P.
(
1989
).
The structural and functional organization of the murine HOX gene family resembles that of Drosophila
.
EMBO J.
8
,
1497
1505
.
Dura
,
J.-M.
,
Randsholt
,
N. B.
,
Deatrick
,
J.
,
Erk
,
I.
,
Santamaria
,
P.
,
Freeman
,
J. D.
,
Freeman
,
S. J.
,
Weddell
,
D.
and
Brock
,
H. W.
(
1987
).
A complex genetic locus, polyhomeotic, is required for segmental specification and epidermal development in D. melanogaster
.
Cell
51
,
829
839
.
Franke
,
A.
,
Messmer
,
S.
and
Paro
,
R.
(
1995
).
Mapping functional domains of the Polycomb protein of Drosophila melanogaster
.
Chromosome Res.
3
,
351
360
.
Gehring
,
W. J.
(
1970
).
A recessive lethal (l(4)29) with a homeotic effect in D. melanogaster
.
Dros. Inform. Serv
45
,
103
.
Gérard
,
M.
,
Duboule
,
D.
and
Zakany
,
J.
(
1993
).
Structure and activity of regulatory elements involved in the activation of the Hox-11 gene during late gastrulation
.
EMBO J.
12
,
3539
3550
.
Girton
,
J. R.
and
Jeon
,
S. H.
(
1994
).
Novel embryonic and adult homeotic phenotypes are produced by pleiohomeotic mutations in Drosophila
.
Dev. Biol.
161
,
393
407
.
Graham
,
A.
,
Papalopulu
,
N.
and
Krumlauf
,
R.
(
1989
).
The murine and Drosophila homeobox gene complexes have common features of organization and expression
.
Cell
57
,
367
378
.
Hyde-DeRuyscher
,
R. P.
,
Jennings
,
E.
and
Shenk
,
T.
(
1995
).
DNA binding sites for the transcriptional activator/repressor YY1
.
Nucleic Acids Res.
23
,
4457
4465
.
Irish
,
V. F.
,
Martinez Arias
,
A.
and
Akam
,
M.
(
1989
).
Spatial regulation of the Antennapedia and Ultrabithorax homeotic genes during Drosophila early development
.
EMBO J.
8
,
1527
1537
.
Jurgens
,
G.
(
1985
).
A group of genes controlling the spatial expression of the bithorax complex in Drosophila
.
Nature
316
,
153
155
.
Kassis
,
J. A.
(
1994
).
Unusual properties of regulatory DNA from the Drosophila engrailed gene: three ‘pairing-sensitive’ sites within a 1. 6 kb region
.
Genetics
136
,
1025
1038
.
Kehle
,
J.
,
Beuchle
,
D.
,
Treuheit
,
S.
,
Christen
B.
,
Kennison
,
J. A.
,
Bienz
,
M.
and
Müller
,
J.
(
1998
).
dMi-2, a Hunchback-interacting protein that functions in Polycomb repression
.
Science
282
,
1897
1900
.
Lewis
,
E. B.
(
1978
).
A gene complex controlling segmentation in Drosophila
.
Nature
276
,
565
570
.
McCall
,
K.
and
Bender
,
W.
(
1996
).
Probes for chromatin accessibility in the Drosophila bithorax complex respond differently to Polycomb-mediated represssion
.
EMBO J.
15
,
569
580
.
McGinnis
,
W.
and
Krumlauf
,
R
(
1992
).
Homeobox genes and axial patterning
.
Cell
68
,
283
302
.
Mihaly
,
J.
Mishra
,
R. K.
and
Karch
,
F.
(
1998
).
A conserved sequence motif in Polycomb-response elements. Molec
.
Cell
1
,
1065
1066
.
Müller
,
J.
and
Bienz
,
M.
(
1991
).
Long range repression conferring boundaries of Ultrabithorax expression in the Drosophila embryo
.
EMBO J.
10
,
3147
3155
.
Müller
,
J.
and
Bienz
,
M.
(
1992
).
Sharp anterior boundary of homeotic gene expression conferred by the fushi tarazu protein
.
EMBO J.
11
,
3653
3661
.
Orlando
,
V.
and
Paro
,
R.
(
1993
).
Mapping Polycomb-repressed domains in the bithorax comples using in vivo formaldehyde cross-linked chromatin
.
Cell
75
,
1187
1198
.
Orlando
,
V.
,
Jane
,
E. P.
,
Chinwalla
,
V
,
Harte
,
P. J.
and
Paro
,
R.
(
1998
).
Binding of Trithorax and Polycomb proteins to the bithorax complex: dynamic changes during early Drosophila embryogenesis
.
EJMBO J.
17
,
5141
5150
.
Pirrotta
,
V.
,
Chan C. S.
McCabe
,
D.
and Qian
, S. (
1995
).
Distinct parasegmental and imaginal enhancers and the establishment of the expression pattern of the Ubx gene
.
Genetics
141
,
1439
1450
.
Poux
,
S.
,
Kostic
,
C.
and
Pirrotta
,
V.
(
1996
).
Hunchback-independent silencing of late Ubx enhancers by a Polycomb group response element
.
EMBO J.
15
,
4713
4722
.
Püschel
,
A.
,
Balling
,
R.
and
Gruss
,
P.
(
1991
).
Separate elements cause lineage restriction and specify boundaries of Hox 1. 1 expression
.
Development
112
,
279
287
.
Qian
,
S.
,
Capovilla
,
M.
and
Pirrotta
,
V.
(
1991
).
The bx region enhancer, a distant cis-control element of the Drosophila Ubx gene and its regulation by hunchback and other segmentation genes
.
EJMBO J.
10
,
1415
1425
.
Salser
,
S. J.
and
Kenyon
,
C.
(
1994
).
Patterning C. elegans: homeotic cluster genes, cell fates and cell migrations
.
Trends Genet.
10
,
159
164
.
Sharpe
,
J.
,
Nonchev
,
S.
,
Gould
,
A.
,
Whiting
,
J.
and
Krumlauf
,
R.
(
1998
).
Selectivity, sharing and competitive interactions in the regulation of Hoxb genes
.
EMBO J.
16
,
1788
1798
.
Simon
,
J.
,
Peifer
,
M.
,
Bender
,
W.
and
O’Connor
,
M.
(
1990
).
Regulatory elements of the bithorax complex that control expression along the antero-posterior axis
.
EMBO J.
9
,
3945
3956
.
Simon
,
J.
,
Chiang
,
A.
,
Bender
,
W.
,
Shimell
,
M. J.
and
O’Connor
,
M.
(
1993
).
Elements of the Drosophila bithorax complex that mediate repression by Polycomb group products
.
Dev. Biol.
158
,
131
144
.
Soto
,
M. C.
,
Chou
,
T.-B.
and
Bender
,
W.
(
1995
).
Comparison of germline mosaics of genes in the Polycomb group of Drosophila melanogaster
.
Genetics
140
,
231
243
.
Strutt
,
H.
and
Paro
,
R.
(
1997
).
The Polycomb group protein complex of Drosophila melanogaster has different compositions at different target genes
.
Mol. Cell. Biol.
17
,
6773
6783
.
Strutt
,
H.
,
Cavalli
,
G.
and
Paro
,
R.
(
1997
).
Co-localization of Polycomb protein and GAGA factor on regulatory elements responsible for the maintenance of homeotic gene expression
.
EJMBO J.
16
,
3621
3632
.
van Lohuizen
,
M.
,
Frasch
,
M.
,
Wientjens
,
E.
and
Berns
,
A.
(
1991
).
Sequence similarity between the mammalian bmi-1 proto-oncogene and the Drosophila regulatory genes Psc and Su(2)z
.
Nature
353
,
353
355
.
White
,
R. A. H.
and
Lehmann
,
R.
(
1986
).
A gap gene, hunchback, regulates the spatial expression of Ultrabithorax
.
Cell
47
,
311
321
.
White
,
R. A. H.
and
Wilcox
,
M.
(
1984
).
Protein products of the bithorax complex in Drosophila
.
Cell
39
,
163
171
.
Zhang
,
C.-C.
and
Bienz
,
M.
(
1992
).
Segmental determination in Drosophila conferred by hunchback (hb), a repressor of the homeotic gene Ultrabithorax (Ubx)
.
Proc. Natl. Acad. Sci.
89
,
7511
7515
.
Zhang
,
C. -C.
,
Müller
,
J.
,
Hoch
,
M.
,
Jäckle
,
H.
and
Bienz
,
M.
(
1991
).
Target sequences for hunchback in a control region conferring Ultrabithorax expression boundaries
.
Development
113
,
1171
1179
.
Zink
,
B.
,
Engström
,
Y.
,
Gehring
,
W. J.
and
Paro
,
R.
(
1991
).
Direct interaction of the Polycomb protein with Antennapedia regulatory sequences in polytene chromosomes of Drosophila melanogaster
.
EMBO J.
10
,
153
162
.
Zink
,
B.
and
Paro
,
R.
(
1989
).
In vivo binding pattern of a trans-regulator of homeotic genes in Drosophila melanogaster
.
Nature
,
337
,
468
471
.