Novel Drosophila Heterochromatin Protein 1 (HP1)/Origin Recognition Complex-associated Protein (HOAP) Repeat Motif in HP1/HOAP Interactions and Chromocenter Associations*

Association of the highly conserved heterochromatin protein, HP1, with the specialized chromatin of centromeres and telomeres requires binding to a specific histone H3 modification of methylation on lysine 9. This modification is catalyzed by the Drosophila Su(var)3-9 gene product and its homologues. Specific DNA binding activities are also likely to be required for targeting this activity along with HP1 to specific chromosomal regions. The Drosophila HOAP protein is a DNA-binding protein that was identified as a component of a multiprotein complex of HP1 containing Drosophila origin recognition complex (ORC) subunits in the early Drosophila embryo. Here we show direct physical interactions between the HOAP protein and HP1 and specific ORC subunits. Two additional HP1-like proteins (HP1b and HP1c) were recently identified in Drosophila, and the unique chromosomal distribution of each isoform is determined by two independently acting HP1 domains (hinge and chromoshadow domain) (47). We find heterochromatin protein 1/origin recognition complex-associated protein (HOAP) to interact specifically with the originally described predominantly heterochromatic HP1a protein. Both the hinge and chromoshadow domains of HP1a are required for its interaction with HOAP, and a novel peptide repeat located in the carboxyl terminus of the HOAP protein is required for the interaction with the HP1 hinge domain. Peptides that interfere with HP1a/HOAP interactions in co-precipitation experiments also displace HP1 from the heterochromatic chromocenter of polytene chromosomes in larval salivary glands. A mutant for the HOAP protein also suppresses centric heterochromatin-induced silencing, supporting a role for HOAP in centric heterochromatin.

Association of the highly conserved heterochromatin protein, HP1, with the specialized chromatin of centromeres and telomeres requires binding to a specific histone H3 modification of methylation on lysine 9. This modification is catalyzed by the Drosophila Su(var)3-9 gene product and its homologues. Specific DNA binding activities are also likely to be required for targeting this activity along with HP1 to specific chromosomal regions. The Drosophila HOAP protein is a DNA-binding protein that was identified as a component of a multiprotein complex of HP1 containing Drosophila origin recognition complex (ORC) subunits in the early Drosophila embryo. Here we show direct physical interactions between the HOAP protein and HP1 and specific ORC subunits. Two additional HP1-like proteins (HP1b and HP1c) were recently identified in Drosophila, and the unique chromosomal distribution of each isoform is determined by two independently acting HP1 domains (hinge and chromoshadow domain) (47). We find heterochromatin protein 1/origin recognition complex-associated protein (HOAP) to interact specifically with the originally described predominantly heterochromatic HP1a protein. Both the hinge and chromoshadow domains of HP1a are required for its interaction with HOAP, and a novel peptide repeat located in the carboxyl terminus of the HOAP protein is required for the interaction with the HP1 hinge domain. Peptides that interfere with HP1a/HOAP interactions in co-precipitation experiments also displace HP1 from the heterochromatic chromocenter of polytene chromosomes in larval salivary glands. A mutant for the HOAP protein also suppresses centric heterochromatin-induced silencing, supporting a role for HOAP in centric heterochromatin.
Early microscopic studies revealed the eukaryotic nucleus to have a heterogeneous morphology. The bulk of eukaryotic chromatin has a decondensed amorphous appearance during interphase. However, certain chromosomal regions retain the compact appearance of metaphase chromatin throughout the cell cycle and, according, have been termed "heterochromatin" (1). The distinct cytological properties of heterochromatin translate into distinct functional properties as well; for example, it is typically replicated later in S phase and is transcriptionally inert relative to the more typical "euchromatin" (2,3). In Dro-sophila, heterochromatin has been found to induce silencing of euchromatic genes that become juxtaposed to it by a chromosomal rearrangement, a phenomenon known as position effect variegation (4). Ironically, the relatively few genes that normally reside within heterochromatin suffer a similar fate when translocated to a euchromatic chromosomal region (5)(6)(7).
These distinct cytological and functional properties of heterochromatin are thought to reflect its unique nucleoprotein composition. For example, euchromatic genes that have undergone heterochromatin-induced silencing adopt a more highly ordered nucleosomal array (8). In some cases, the silenced gene is also recruited into the heterochromatin compartment of the nucleus, where it is thought to be sequestered from the enzymatic machinery for a variety of DNA metabolic processes including transcription.
One conserved feature of heterochromatin is its non-coding repetitive DNA sequence content, although little conservation is observed between these repeats at the primary sequence level (9 -11). Another conserved feature of heterochromatin is the heterochromatin-associated protein, heterochromatin protein 1 (HP1). 1 This protein, first described in Drosophila, is enriched in the heterochromatin of species ranging from fission yeast to humans (12)(13)(14). The Drosophila protein (HP1a) is encoded by the Suppressor of variegation (Su(var)) 2-5 gene, which belongs to a group of Su(var) genes with mutant phenotypes of reversing the heterochromatin-induced silencing of euchromatic genes (15)(16)(17)(18).
Because early studies failed to demonstrate DNA binding activity for HP1, the mechanism for its association with heterochromatin has remained a mystery until the recent discovery that a specific covalent modification (methylation on lysine 9) of histone H3 provides a chromatin binding site for it in species ranging from fission yeast to humans (19 -23). Interestingly, this histone modification is catalyzed by the product of the Drosophila Su(var)3-9 gene and its homologues in other species, and recognition of this histone binding site by HP1 requires the conserved chromodomain of HP1.
The mechanism used by the cell to target HP1 and the Su(var)3-9 H3 methyltransferase activity specifically to heterochromatin is not as well understood. A number of HP1-interacting proteins have been identified that directly or indirectly bind DNA. Some of these (e.g. TIF1 proteins and SP100) are capable of acting as HP1-dependent transcriptional co-repressors when tethered to heterologous DNA binding domains (24 -28). The DNA binding activities of TIF1-interacting zinc finger Kruppel repressor proteins (KRAB-ZFPs) are thought to target the transcriptional repressing activity of HP1 through TIF1␤ (29 -31). These proteins are found in complexes with lysine 9 histone H3 methyltransferase activities (31). The retinoblastoma protein targets both HP1 and the mammalian Su(var)3-9 homologue (SUV39H) to the mammalian cyclin E promoter, presumably through the DNA binding activity of the E2F transcription factor (32,33). These data support a role for specific DNA binding activities in recruiting HP1-and histone-modifying activities to mammalian euchromatic genes. Mammals contain three different isoforms of HP1 that differ in their euchromatic and heterochromatic localizations, and it has even been proposed that HP1 heterodimers may play a role in recruiting euchromatic genes to the heterochromatin compartment (30,34,35).
DNA binding activities may similarly play a role in recruiting HP1 to heterochromatic regions. For example, fission yeast mutants for homologues of the mammalian ␣-satellite sequence binding centromere protein B (CENP-B) have reduced levels of lysine 9-methylated histone H3 and the HP1 homologue, Swi 6, at centromeres (36). The double-stranded interference RNAprocessing machinery has also been found to have a role in Swi 6 targeting to centromeres and the silent mating type loci of Schizosaccharomyces pombe (37). In this paper, we examine the role of a DNA binding activity that we identified as a component of a maternally loaded complex of HP1 in the early Drosophila embryo (38,39). This complex also contained subunits of the Drosophila DNA replication origin recognition complex (DmORC); thus, the unidentified component was designated as HP1/ORC-Associated Protein (HOAP). The DmORC2 subunit is enriched in centric heterochromatin of early embryos, and mutants for this protein also suppress heterochromatin-induced silencing and display defects in HP1 localization in centric heterochromatin of diploid nuclei (38,40). These data support a role for the ORC in localizing HP1 to heterochromatin. The amino terminus of the HOAP protein contains similarity to the HMG box of sequence-specific HMG proteins and binds specific double-stranded AT-rich satellite sequences of Drosophila melanogaster in vitro (39). It is also localized in heterochromatin, predominantly at telomeres, but weaker enrichment is also observed in pericentric heterochromatin (39). Mutants for the HOAP-encoding gene, caravaggio, display a telomere fusion phenotype that is associated with a defect in HP1 localization at telomeres (41). Here we report the molecular parameters that specify interactions between HOAP and the predominantly heterochromatin-localized HP1a isoform (commonly known as Drosophila HP1). We also present evidence that these interactions play a role in the association of this HP1 protein with pericentric heterochromatin.

EXPERIMENTAL PROCEDURES
Protein Expression-Hexahistidine-tagged recombinant HOAP proteins were expressed from a cDNA of the anon fe 1G5 gene cloned into the pET20b expression vector (39). FLAG-tagged recombinant HP1 proteins were expressed from a cDNA of the HP1a gene cloned into the PQE30 (Qiagen) expression vector. The HP1a cDNA was obtained by PCR amplification using EST clone LD10408 and a forward primercontaining DNA sequence encoding the FLAG tag (5Ј-GCGCGCGGAA-TTCATGGACTATAAAGACGATGACAAAGGCAAGAAAATCGACAA-CCCT-3Ј) and reverse primer (5Ј-GCGCGCGTCTAGAATCTTCATTAT-CAGAGTACCAGGATAG-3Ј). Truncated HOAP and HP1 proteins were expressed from deletion derivatives of these expression vectors created by the introduction of restriction sites into the HOAP-and HP1-coding sequences by site-directed mutagenesis (Stratagene QuikChange sitedirected mutagenesis kit, catalog number 200518). All cDNA clones used to express recombinant proteins were sequenced before their use for expressing proteins in BL21 (DE 3) Escherichia coli strain. Hexahistidine-tagged proteins were purified by nickel nitrilotriacetic acidagarose (Qiagen catalog number 30210) chromatography using imidazole for protein elution, and FLAG-tagged proteins were purified by FLAG M2 (Sigma A-1205) chromatography using FLAG peptide (Sigma F3290) for protein elution. 35 S-Labeled ORC, HP1a, HP1b, and HP1c proteins were synthesized in vitro using a coupled transcription/translation reaction system (TNT Quick Coupled Transcription/Translation System, Promega, catalog number L1170) and cDNA clones LD11626 (ORC1), GH 13824 (ORC2), GM 14657 (ORC3), LD 43280 (ORC4), LP 12153 (ORC5), RE 52740 (ORC6), LD 10408 (HP1a), RE 72354 (HP1b), and RE 28447 (HP1c) produced by the Berkeley Drosophila Genome Project and distributed by Research Genetics (Invitrogen).
Immunoprecipitation Experiments-Immunoprecipitation experiments were carried out as described by Huang et al. (38). An anti-HOAP immunoaffinity resin (39) was used to immunoprecipitate recombinant FLAG-tagged HP1 proteins, 35 S-labeled ORC subunits, or HP1a, HP1b, HP1c proteins with recombinant hexahistidine-tagged HOAP. M2 resin (Sigma A-1205) was used in co-precipitation experiments of hexahistidine-tagged HOAP proteins with recombinant FLAG-tagged HP1 proteins. All immunoprecipitation reactions were carried out with equimolar concentrations of co-precipitating proteins. Peptide competition experiments were carried out with a 10-and 100-fold molar excess of competing PRMVI, PETEMNE, PGETEMNE, GETEMNE, HP1a hinge (KSKRTTDAEZDTIPVSGST), HP1b hinge (RSKRKSFLEDDT-EEQKKLI), and HP1c hinge (KKRGEKKPKCEEIQKLR) peptide synthesized by Research Genetics (Invitrogen). Immunoprecipitation reactions were incubated with rotation for 1 h at 4°C in Buffer A (50 mM Hepes, pH 7.6, 10% glycerol (w/v), 1 mM sodium metabisulfite, 100 mM phenylmethylsulfonyl fluoride, 200 mM benzamidine, and a 1:100 dilution of protease inhibitor mixture (1.6 mg/ml benzamidine and 1.0 mg/ml each phenanthroline, aprotinin, leupeptin, and pepstatin)) containing 100 mM KCl. Immunoprecipitations were washed 3 times (15 min each) in the same buffer followed by 1 wash in Buffer A containing 0.5 M KCl and 1 wash in Buffer A containing 1.0 M KCl before elution of the bound protein with 100 mM glycine, pH 2.0. HP1 and HOAP immunoblotting was performed on 2% of the input sample, 2% of the unbound supernatant fraction, and 25% of the bound pellet fraction. Immunoblotting signals were detected by enhanced chemiluminescence detection or by autoradiography of 35 S-labeled in vitro translated protein. Adobe Photoshop 7.0 software was used to scan and process all digital images.
Gel Filtration-Purified recombinant HOAP and HP1 proteins were combined at stoichiometries of 1:1, 1:2, and 1:4 and loaded onto a Sephacryl S-200 gel filtration column that had been pre-equilibrated in Buffer A. Trichloroacetic acid precipitates of 1-ml fractions from the column were analyzed by Coomassie staining on SDS-polyacrylamide gels. Native protein molecular weight standards from Bio-Rad (catalog number 151-1901) were used.
Peptide Challenge Assays-The effect of a peptide on HP1 association with the insoluble chromatin fraction was determined by incubating five salivary glands in 50 l of Cohen's permeabilization buffer (42) (10 min) containing a 6 M concentration of the challenging peptide (di-MeK9hisH3 (Upstate Biotechnology catalog number 12-430), PETEMNE, GETEMNE, PGTEMNE, PRMVI, or HP1a, -b, and -c hinge peptide, as described above) in a 500-l microcentrifuge tube. The glands were then pelleted by centrifugation for 2 min at 8,000 ϫ g. HP1 and HOAP immunoblotting was then performed on a trichloroacetic acid precipitate of the entire solubilized supernatant fraction and the entire solubilized pellet fraction from each set of treated glands. Equivalent loading of protein samples was monitored by Ponceau S staining of protein transferred to nitrocellulose before immunoblotting. Polytene chromosome immunostaining was used to determine the effect of each peptide on HP1 association with the chromocenter. After peptide extraction as described above, salivary glands were fixed in formaldehyde and used for polytene chromosome squash preparations. Chromosome squash preparations and immunostaining were carried out as previously described (39) using rat anti-HOAP (1:200) and rabbit anti-HP1 (1:1000) primary antibodies and fluorescein-labeled anti-rabbit IgG and rhodamine-labeled anti-rat IgG secondary antibodies (1:1000). The effect of overexpressing HOAP on HP1 chromocenter displacement by peptide treatment was determined using salivary glands dissected from a heat shock-inducible HOAP transgenic line (39) that had been subjected to 37°C heat shock treatment for 30 min followed by a 30-min recovery at room temperature. Similar treatments of control animals of the same genotype lacking the HOAP transgene were carried out in parallel. A Photometrics CoolSnap cooled high speed digital color camera and MetaView imaging software were used to acquire images at equivalent exposure settings for all specimens.
Position Effect Variegation Modifier Assays-The w m4 and BL1-hsLacZ reporter genes were used to assay a mutant for HOAP (cav) (41) for phenotypes associated with modifying pericentric heterochromatininduced silencing. Crosses were carried out between w 1 /Y;cav/TM3Sb males and w m4 females and between w 1 ;cav/TM3Sb males and BL1-hsLacZ/TM3Sb females. Eye variegation phenotypes of progeny of the cross between w 1 ;cav/TM3Sb and w m4 were visually scored as described in Table I. ␤-Galactosidase activity was quantitated in protein extracts prepared from female progeny of the cross between cav/TM3Sb and BL1-hsLacZ/TM3Sb after a 30-min 37°C heat shock followed by 15 min recovery using chlorophenol red-␤-D-galactopyranoside as substrate (43).

HOAP Interacts
Directly with HP1-Co-precipitation experiments were carried out with bacterially expressed recombinant HOAP and HP1 proteins to determine whether HOAP is capable of directly interacting with HP1 (Fig. 1). An anti-FLAG M2 resin was used to precipitate hexahistidine-tagged HOAP protein with FLAG-tagged HP1. The full-length HOAP protein (full) co-precipitated with HP1 in these experiments (Fig. 1A). The domain of HOAP that is responsible for this binding to HP1 was then mapped to the carboxyl-terminal half of the protein. Truncated versions of the FLAG-tagged HP1 protein were then used to map the HP1 domain responsible for binding the HOAP carboxyl terminus. An anti-HOAP immunoresin was used to precipitate carboxyl-terminal HOAP protein and associated HP1 fragments (Fig. 1B). The carboxyl-terminal chromoshadow domain (CSD) and the hinge (H) domain located between the conserved chromo and chromoshadow domains of HP1 were each found to independently bind the HOAP carboxyl terminus. The amino-terminal HP1 chromodomain (CD) failed to bind the HOAP carboxyl terminus.
A pentapeptide motif (PRMVI) located in the HP1 chromoshadow domain and also in a number of HP1-interacting proteins (p150 subunit of CAF-1, TIF1 proteins, and the Su-(var)3-7 protein) has been shown or implicated to mediate the association of these proteins with HP1 (44 -46). Mutations in the pentapeptide motif of mouse HP1␤ (mMOD1) also prevent it from forming homodimers (44). Although HOAP lacks a canonical pentapeptide motif, its interaction with the HP1 chromoshadow domain prompted us to determine whether this motif is required for the HOAP/HP1 interaction. We used a molar excess of a competing PRMVI peptide in the co-precipitation assay to examine the role of this motif in this interaction (Fig. 1C, ϩPRMVI). A 10-fold molar excess of this peptide was found to interfere with the interaction between the HOAP carboxyl terminus and the HP1 chromoshadow domain. In contrast, the interaction of the HOAP carboxyl terminus with the HP1 hinge domain was not even inhibited by a 100-fold molar excess of the peptide. A mutant form of the HP1 protein containing a substitution of glutamic acid for valine at position 190 of the PRMVI motif also failed to bind the HOAP carboxyl terminus (Fig. 1D, V190E).
The carboxyl terminus of the HOAP protein contains three copies of a novel proline-containing repeat (PETEM/INE) that could also have a role in HP1 binding (39). A synthetic peptide for this sequence was also used in competitive binding experiments with the HOAP carboxyl terminus and each of its HP1 interaction domains (Fig. 1C, ϩPETEMNE). A 100-fold molar excess of the peptide was able to inhibit the interaction between the HOAP carboxyl terminus and the HP1 hinge domain but had no effect on the interaction between the HOAP carboxyl terminus and the HP1 chromoshadow domain. A 10-fold molar excess of PETEMNE peptide also partially interfered with the interaction between the HOAP carboxyl terminus and the HP1 hinge domain (data not shown). Mutant forms of the peptide (Fig. 1C, ϩPGETEMNE and ϩGETEMNE) did not inhibit binding of the HP1 hinge domain to the HOAP carboxyl terminus. Also, a P290E substitution in the third PETEMNE motif of the HOAP carboxyl terminus prevented it from binding the HP1 hinge domain but had no effect on its binding to the chromoshadow domain (Fig. 1D, P290E).
HOAP Interacts with a Dimer of HP1-The ability of the PRMVI peptide to interfere with binding of the HOAP carboxyl terminus to the HP1 chromoshadow domain suggests some role for this motif in the HOAP/HP1 interaction. The peptide interference could result from direct competition by the peptide for a site of interaction between carboxyl-terminal HOAP and the PRMVI motif of HP1. Alternatively, the interference could be an indirect consequence of the peptide impeding HP1 homodimerization. The peptide might then indirectly interfere with an HP1/HOAP interaction that requires HP1 in dimeric form.
To address this possibility, gel filtration experiments were used to determine the stoichiometry of HP1 and HOAP in the HP1⅐HOAP complex. Mixtures of the two recombinant proteins at molar ratios of 1:1, 1:2, and 1:4 (HOAP:HP1) were fractionated over a Sephacryl S200 gel filtration column. The fractionation profile of each mixture was then analyzed by Coomassie staining on SDS-polyacrylamide gels (Fig. 2). The fractionation profiles for the mixtures of HOAP and HP1 at 1:1 and 1:4 (HOAP:HP1) molar ratios each contained a fractionation peak for either HOAP or HP1 that was distinct from the peak containing both proteins (ϳ300 kDa). A separate peak for HOAP (ϳ100 kDa) was observed in the fractionation profile of the 1:1 molar ratio mixture of HOAP:HP1. A separate peak for HP1 alone (ϳ44 kDa) was observed in the fractionation profile for the 1:4 molar ratio mixture of HOAP:HP1. The fractionation profile of the mixture of proteins at a 1:2 (HOAP:HP1) stoichiometry, by contrast, lacked separate peaks for either protein.
These data indicate a stoichiometry of two HP1 molecules for every molecule of HOAP in the HOAP⅐HP1 complex. Each protein behaved as a species with twice its known molecular weight in these experiments, possibly indicating the presence of two HOAP molecules and four HP1 in the HOAP⅐HP1 complex or that both proteins are non-globular in shape.
HOAP Also Interacts with Specific DmORC Subunits-Because the HOAP protein was identified as a component of an HP1 complex that also contains ORC subunits, we also assessed the ability of the HOAP protein to directly bind individual ORC subunits. Binding studies were carried out with bacterially expressed recombinant HOAP and S 35 -labeled ORC proteins synthesized in an in vitro transcription/translation system. All binding reactions were carried out with an equimolar concentration of recombinant HOAP protein and each individual in vitro translated ORC protein. DmORC subunits 1, 3, and 6 were each quantitatively depleted from the input and co-precipitated with HOAP in the binding reaction (Fig. 3). The DmORC4 subunit also co-precipitated with the HOAP protein but was only partially depleted from the binding reaction. The partial binding of DmORC4 in this assay could indicate weaker or non-stoichiometric binding to the HOAP protein or could reflect unintended non-stoichiometric concentrations of the two proteins in the binding reaction. Full-length HOAP protein was required for binding to DmORC subunits 3, 6, and 4 (data not shown), whereas the HOAP carboxyl terminus (Fig. 3, C-HOAP) was sufficient for binding the DmORC1 subunit. Unlike the interaction between the HOAP carboxyl terminus and the HP1 hinge domain, the interaction between the HOAP carboxyl terminus and DmORC1 was not inhibited by the PETEMNE peptide (data not shown).
HOAP Specifically Binds Drosophila HP1a Protein-Two other HP1-like proteins have recently emerged in sequence analyses of the Drosophila genome (47). Like the mammalian HP1 isoforms, each of these proteins displays a distinct localization pattern. HP1b and HP1c are predominantly found in euchromatin, whereas the originally described HP1a is enriched in centric and telomeric heterochromatin. The hinge and chromoshadow domains of each of these proteins imparts distinct euchromatin-and heterochromatin-specific targeting activities. This was of interest to us with regards to our finding that HOAP interacts with each of these HP1 domains independently. We wished to determine whether the interaction we have identified between HOAP and HP1a is specific to this Drosophila HP1 isoform. We used S 35 labeled in vitro translated HP1a, HP1b, and HP1c proteins in co-immunoprecipitation experiments with recombinant HOAP protein (Fig. 4A). HP1a was the only HP1-like protein found to co-precipitate with HOAP in these experiment.
To more precisely define the sequence within the HP1a hinge domain that is responsible for this interaction with HOAP, we synthesized HP1 isoform-specific peptides corresponding to the region of greatest similarity between the three proteins within the otherwise poorly conserved hinge domain (Fig. 4B). Each of these peptides was then used as a competitor in co-precipitation experiments with the HOAP carboxyl terminal domain and the full HP1a hinge domain (Fig. 4C). A 10-fold molar excess of the HP1a-specific peptide was able to compete with binding of the HOAP carboxyl terminal domain to the HP1a hinge domain, whereas the HP1b and HP1c peptides were unable to compete even at a 100-fold molar excess. This result implicates this region of the HP1a hinge domain as well as the PETEMNE sequence of HOAP in the interaction between the hinge domain of HP1a and the HOAP carboxyl terminus.
PETEMNE Peptide Displaces HP1 from the Chromocenter-We next used a peptide challenge assay to test the relevance of the HOAP-HP1 protein interaction domains identified in the biochemical experiments in the association of HP1 with the heterochromatic chromocenter (Fig. 4). The assay we used is an adaptation of one used by Bannister et al. (19) to assess the role of dimethylated lysine 9 containing histone H3 (di-MeK9) in the association of HP1 with chromatin of human U2OS cells. In the experiments of Bannister et al. (19), mammalian HP1 ␣ and ␤ isoforms were displaced from chromatin and released into the soluble fraction when permeabilized cells or nuclei were incubated in the presence of challenging di-MeK9 peptide. We have designed a similar peptide challenge assay using polytene nuclei from third instar larval salivary glands. Two methods were used to monitor HP1 displacement from the chromosomes. We first used HP1 immunoblotting of the soluble and pellet fractions from the peptide-challenged glands to monitor displacement of HP1 from chromatin. The second method was HP1 immunostaining of polytene chromosome squashes from the peptide-challenged salivary glands to precisely identify the regions of HP1 displacement from the chromosomes.
Salivary glands were first placed in microcentrifuge tubes and incubated with the challenging peptide in a buffer routinely used to permeabilize polytene chromosomes in squashing protocols. The glands were then pelleted by centrifugation to separate the insoluble chromatin and solubilized protein fractions. HP1 immunoblotting of the pellet and supernatant fractions in their entireties was then used to assess the ability of each peptide to displace HP1 from chromatin (Fig. 5A). HP1 was equally divided between the pellet and supernatant fractions from glands incubated in permeabilization buffer alone. By contrast, HP1 was virtually depleted from the insoluble chromatin fraction and enriched in the solubilized supernatant fraction from glands incubated with the di-MeK9 peptide. It was similarly depleted from the pellet fraction and enriched in the supernatant fraction from glands incubated with the PETEMNE peptide. The PRMVI peptide did not cause a similar release of HP1 from the chromatin pellet fraction nor did any of the HP1 hinge domain peptides.
HP1-and HOAP-immunostaining of the di-MeK9 and PETEMNE peptide-extracted polytene chromosomes revealed that HP1 was displaced from the chromocenter as well as throughout the polytene chromosomes (Fig. 5B). HP1 immunostaining of untreated chromosomes shows enrichment of the protein in the pericentric heterochromatin of the chromocenter (large arrow) (Fig. 5B, panel a) (12, 13). The protein is also observed at telomeres (asterisk) and at several reproducible euchromatic sites (e.g. region 31, small arrow). HOAP displays prominent immunostaining at telomeres (asterisk), with less pronounced enrichment also in pericentric heterochromatin (large arrow) and a number of reproducible euchromatic sites (small arrow) (39). To determine the distribution of both HP1 and HOAP on polytene chromosomes from the peptide-treated salivary glands, an anti-HOAP antibody that was produced in rat was used in combination with an anti-HP1 antibody produced in rabbit. The immunostaining pattern for the rat anti-HOAP antibody (C, red) precisely overlaps that obtained with the previously published anti-HOAP antibody produced in rabbit (B, green) except that euchromatic and centric heterochromatin sites are more prominently labeled with the rat anti-HOAP antibody (Fig. 6).
The introduction of di-MeK9 peptide to the permeabilization buffer eliminated HP1 immunostaining throughout the chromosomes but had little effect on HOAP immunostaining (Fig.  5B, panel b). The PETEMNE peptide had a similar disrupting effect on the chromatin association of HP1 (Fig. 5B, panel c). Similar to what was observed with the di-MeK9 peptide, the PETEMNE peptide caused very little disruption in the chromatin association of the HOAP protein. The mutant PETEMNE peptides carrying glycine substitutions for proline at amino acid 1 or glutamic acid at amino acid 2 that failed to interfere with HOAP/HP1 interactions in vitro also had no effect on the chromatin association of HP1 (Fig. 5B, panel d).
Each of these peptide treatments also left the HOAP immunostaining pattern relatively unperturbed.
HOAP Overexpression Reduces HP1 Sensitivity to PETEMNE Peptide Extraction-The PETEMNE peptide appears not only to be a novel HP1 interaction motif but also to be unique to the HOAP protein. We are unable to identify other Drosophila proteins containing similarity to this motif when the HOAP carboxyl-terminal sequence is used to search the data base of proteins predicted from the Drosophila genome sequence. We have also been unable to identify other proteins containing this motif when using the PETEMNE sequence to carry out a BLAST search for short nearly exact matches. This indicates a limited number of potential cellular targets for the PETEMNE peptide in the displacement of HP1 from the chromocenter.
Nevertheless, to more directly test the role of the HP1/HOAP interaction as a target of the peptide in this displacement, we determined if overexpressing the HOAP protein would reduce the sensitivity of HP1 to displacement from the chromocenter by the PETEMNE peptide. A heat shock-inducible HOAP transgene was used to this end. Third instar larvae carrying this transgene were subjected to heat shock before dissecting their salivary glands for the peptide extraction assay (Fig. 7). The salivary glands from the hs-HOAP transgenic line and those from control heat shocked larvae of the same genotype but lacking the HOAP transgene were each incubated with the PETEMNE peptide and used to prepare polytene chromosome squashes as described above. HP1 immunostaining of the polytene chromosomes from heat-shocked larvae lacking the HOAP transgene revealed a complete displacement of HP1 from the chromocenters by the PETEMNE peptide (Fig. 7B). In contrast, the same concentration of peptide failed to displace HP1 from the chromocenters of larvae carrying the heat shock-inducible HOAP transgene (Fig. 7D). The heat shock-induced expression of the HOAP protein resulted in enhanced HOAP immunostaining throughout the chromosomes (Fig. 7, C and D), and this association with the chromosomes was unperturbed by the incubation with PETEMNE peptide.
Mutants for HOAP Suppress Centric Heterochromatin-induced Silencing-This dramatic effect of the PETEMNE peptide in the peptide challenge immunostaining assay was somewhat unexpected. The prominent localization of HOAP at telomeres might be considered evidence for an exclusive role for it in localizing HP1 at telomeres. As demonstrated in Figs. 5 and 6 we also reproducibly observed HOAP immunostaining in pericentric heterochromatin as well as a number of euchromatic sites using immunosera from multiple animals immunized with HOAP protein (39). We previously showed a deletion removing the HOAP-encoding gene (along with ϳ60 other genes) to suppress variegation of reporter genes juxtaposed to pericentric heterochromatin. To determine whether a point mutation for the HOAP-encoding cav gene also modifies varie- FIG. 7. HOAP overexpression reduces HP1 sensitivity to chromocenter displacement by PETEMNE peptide. HP1 (green) and HOAP (red) immunostaining of DAPI-stained polytene chromosome squashes from salivary glands from heat shocked wild type larvae after incubation in the presence of no peptide (A) and 6 M PETEMNE peptide (B). HP1 (green) and HOAP (red) immunostaining of DAPIstained polytene chromosome squashes from salivary glands from heatshocked hs-HOAP transgenic larvae after incubation in the presence of no peptide (C) and 6 M PETEMNE peptide (D).
FIG. 6. HOAP immunostaining with multiple immunosera shows localization at telomeres, regions of centric heterochromatin, and numerous euchromatic sites. A, DAPI-staining of Drosophila larval salivary gland polytene chromosome squash immunostained with previously published anti-HOAP antibody produced in rabbit (green) (B), anti-HOAP antibody produced in rat (red) (C), and merge of immunostaining patterns for rabbit (green) and rat (red) anti-HOAP antibodies (D). Short arrows denote telomeres; long arrows denote centric heterochromatin (prominent staining in centric heterochromatin of 3L indicated); asterisks denote euchromatic sites. gation, we determined the effect of this mutation on the variegation of two different reporter genes that have undergone heterochromatin-induced silencing. One reporter is the white gene that has been translocated next to centric heterochromatin of the X chromosome (w m4 ), and the other is the heat shock-inducible hs70-LacZ transgenic reporter that has been translocated to centric heterochromatin of chromosome 3L (BL1) ( Table I). The cav mutation was found to mildly suppress the variegated phenotype of both reporters. A smaller fraction of animals carrying the cav mutation (w 1 /w m4 ; cav/ϩ) displayed a strongly variegated phenotype (Ͻ20% pigmented cells), whereas a larger fraction displayed moderate to weak variegated phenotypes (Ͼ50% pigmented cells) relative to their sibling controls (w 1 /w m4 ; TM3Sb/ϩ). The hs70-lacZ reporter located in the centric heterochromatin of chromosome 3L was also affected by the cav mutation. The ␤-galactosidase activity in animals carrying the cav mutation (BL1/cav) was ϳ50% higher than that in their sibling controls (BL1/TM3Sb).

DISCUSSION
The HOAP protein was first purified as a component of an ORC-containing multiprotein complex of HP1 in the cytoplasm of early Drosophila embryos. The association of HOAP with both HP1 and ORC subunits in this complex suggested a role for it in HP1 targeting to heterochromatin and perhaps also in determining the unique replicating properties of heterochromatin (38,40). In this paper we show the protein to directly interact both with HP1 and Drosophila ORC subunits. The interaction with HP1 is specific to the originally described predominantly heterochromatin-localized HP1a protein. Coprecipitation experiments with peptide competitors and mutant recombinant proteins allowed us to identify specific domains of HP1a and HOAP that are required for this interaction. The HOAP protein contains a novel PETEM/INE repeat motif in its carboxyl terminus (39). This domain of the HOAP protein was capable of independently interacting with both the chromoshadow and hinge domains of HP1a. The PETEMNE motif of the HOAP carboxyl terminus was required for the interaction of this HOAP domain with the HP1 hinge domain but not for its interaction with the HP1 chromodomain. A peptide sequence located within the most highly conserved portion of the HP1 hinge domain that is unique to the HP1a protein also interfered with this interaction. The PRMVI peptide motif of the HP1 chromoshadow domain, which has been shown to function in a number of HP1 protein/protein interactions including HP1 homo-dimerization, is also required for the interaction between the HP1 chromoshadow domain and the HOAP carboxyl terminal domain. Unlike most HP1-interacting proteins that require this motif for interaction with HP1, however, the HOAP protein lacks a canonical pentapeptide motif.
Gel filtration studies of the HOAP⅐HP1 complex indicate that this requirement for the PRMVI motif in HP1 reflects a requirement for HP1 dimerization in HOAP binding.
HOAP/HP1 Interaction in HP1 Chromocenter Association-A peptide challenge assay similar to one used to show tethering of HP1 to chromatin of human cells through recognition of histone H3 containing methylation on lysine 9 (19) was used to assess the role of HOAP in the association of HP1 with the heterochromatic chromocenter of salivary gland polytene chromosomes. The PETEMNE peptide that interfered with the interaction between the HOAP carboxyl terminus and HP1 hinge domain was also found to displace HP1 from the chromocenter in this assay. In view of the predominantly telomeric localization of HOAP on polytene chromosomes, this dramatic effect at the chromocenter was somewhat surprising. Perhaps it is not so surprising, however, because HOAP is also observed in pericentric heterochromatin of the chromocenter as well as a number of euchromatic sites (Ref. 39 and Figs. 5 and 6). The non-telomeric sites of HOAP localization are particularly pronounced when using an anti-HOAP antibody that was produced in rat as shown in Fig. 6. The relevance of the HOAP PETEMNE motif in the association of HP1 with telomeres has been previously demonstrated by the HP1 telomere-targeting phenotype of the cav mutant for HOAP that lacks the third of three PETEMNE repeats (41). We similarly observed a complete perturbation in the in vitro binding of HP1 to a HOAP protein containing a point mutation in a single PETEMNE repeat in this study. The effect of the PETEMNE peptide on HP1 association with the chromocenter indicates a similar role for this interaction motif in the association of HP1 with the chromocenter.
The ability of the cav mutant to suppress centric heterochromatin-induced silencing of euchromatic reporter genes also supports a role for HOAP in the association of HP1 with the chromocenter. However, we do not observe a major disruption in HP1 association with the chromocenter in this mutant (data not shown). This limited effect of the cav mutation on HP1 association with the chromocenter could reflect a role for the protein exclusively in an initiation phase of heterochromatin assembly. The perdurance of the maternal supply of HOAP protein in the cav mutant until the late larval stages only allows us to examine the requirement of the protein for maintenance of heterochromatin structure. Mutants for proteins required exclusively for heterochromatin initiation in S. pombe also fail to display HP1 (Swi6)-targeting phenotypes unless the mutant cells are treated with histone deacetylase inhibitors, thus necessitating the function of the proteins to reinitiate heterochromatin assembly (48). The insensitivity of some Drosophila histone deacetylases to drug inhibition (49) prevents us from using a similar approach to determine the role of HOAP and other proteins in the initiation of Drosophila heterochromatin assembly.
The peptide extraction experiments of this study provide an alternative mechanism for addressing the roles of specific protein/protein interactions in maintaining heterochromatin structure. The approach, admittedly, is limited by the specificity of the competing peptide used. The uniqueness of the PETEMNE motif to the HOAP protein indicates a very limited number of potential cellular targets for this peptide. The effects of the PETEMNE peptide and its mutant derivatives in the biochemical assays exactly mirrored their effects in the peptide extraction assay. This, combined with the reduced sensitivity of cells overexpressing the HOAP protein to HP1 displacement by the peptide, point to the HOAP/HP1 interaction as the most likely target of the PETEMNE peptide in the displacement of HP1 from the chromocenter. a Variegation is scored as % pigmented cells: Ͻ20% ϭ strong; Ϸ50% ϭ moderate; Ͼ80% ϭ weak. b ␤-Galactosidase activity from expression of hsLacZ reporter is given in units/mg of protein.
Two HP1a Domains Are Required for Heterochromatin Targeting-Previous studies by Smothers and Henikoff (47) showed the hinge and chromoshadow domains of HP1 to impart distinct heterochromatin targeting activities to HP1a. We found the HOAP protein to interact only with HP1a and to interact independently with each of these targeting domains. These findings suggest a role for HOAP in targeting this specific isoform to the heterochromatin domain. To date, only one other HP1-interacting protein has been shown to interact with HP1 through its hinge domain (50), and to our knowledge HOAP is the only protein capable of independent interactions with both the hinge and chromoshadow domain. An abundance of AT-rich repetitive sequences that are potential binding sites both for HOAP and ORC in the heterochromatin compartment may contribute to specific targeting of the HP1a protein to heterochromatin. It is hoped that the HP1 chromocenter association assay of this study can be used to test the roles of a variety of protein/protein associations in this association. Moreover, this assay might provide an alternative strategy for studying heterochromatin assembly that circumvents some of the problems associated with more traditional genetic assays.