By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Due to essential maintenance the subscribe/renew pages will be unavailable on Wednesday 26 October between 02:00- 08:00 BST/ 09:00 – 15:00 SGT/ 21:00- 03:00 EDT. Apologies for the inconvenience.
Heterochromatin protein 1 (HP1) is a conserved nonhistone chromosomal protein that localizes to centric and telomeric heterochromatin (James et al., 1989). HP1 associates with centric regions through an interaction between the amino terminal chromo domain of HP1 and methylated lysine nine of histone H3 (mK9 H3), an epigenetic mark generated by the SU(VAR)3-9 histone methyltransferase (Bannister et al., 2001). Within centric heterochromatin, HP1 plays a role in chromatin packaging and chromosome segregation (Kellum, 2003).
In Drosophila, HP1 is encoded by the gene Su(var)2-5, a suppressor of position effect variegation (Eissenberg et al., 1990). Mutations in Su(var) genes lead to suppression of silencing, which can be observed when genes are brought into juxtaposition with heterochromatin. Silencing correlates with HP1-dependent alterations in chromatin structure (Cryderman et al., 1998). Whereas these data suggest a role for HP1 in gene silencing, the silenced genes are in an unnatural chromatin context, where they have been positioned next to centric chromatin. Thus, the role of HP1 in the regulation of genes in their natural euchromatic locations is not well understood.
A global role for HP1 in gene regulation is suggested by localization studies in tissue culture cell lines (Greil et al., 2003) and by the pattern of HP1 localization on Drosophila polytene chromosomes (Fanti et al., 2003). In addition to association at centric and telomeric regions, HP1 localizes in a banded pattern along the fourth chromosome and to approximately 200 sites throughout euchromatin (James et al., 1989; Fanti et al., 2003). Cytological region 31 on the left arm of the second chromosome contains eight bands of intense staining with HP1 antibodies. This pattern is conserved among Drosophila strains and related species (Fanti et al., 2003). One hypothesis for the localization pattern of HP1 at euchromatic sites such as region 31 is that these regions share features with centric heterochromatin. Such features might include the presence of specific repetitive sequences or histone modifications that attract HP1. In contrast to this hypothesis, euchromatic sites containing HP1 neither consistently correspond to locations of repetitive elements (Fanti et al., 2003) nor locations of mK9 H3, an epigenetic mark of heterochromatin (Cowell et al., 2002; Li et al., 2002). These findings suggest that HP1 associates with euchromatic sites that have features distinct from those of centric heterochromatin.
A possible role for HP1 at euchromatic sites is to silence genes in a manner similar to that operating within centric heterochromatin. To determine the role of HP1 in gene regulation, a gene expression analysis was performed using RNA isolated from wild-type and Su(var)2-5 larvae with reduced levels of HP1. The results indicate that HP1 regulates several hundred genes throughout the genome. Genes within cytological region 31 were examined in greater detail. A total of 10 genes within this region show altered expression in the Su(var)2-5 mutant; 3 of the 10 genes were found to associate with HP1 based on chromatin immunoprecipitation experiments. Surprisingly, these three genes require HP1 for expression, implicating a role in activation rather than silencing. Only one of the genes associates with mK9 H3, suggesting that HP1 localizes to a subset of sites within euchromatin using a mechanism similar to that used in heterochromatin. Association of HP1 with the other two genes appears to occur in the absence of mK9 H3, suggesting an alternate mechanism of localization. Genes silenced by heterochromatin are typically sensitive to changes in Su(var)3-9 dosage, however, the expression of the three genes associated with HP1 was unaffected in a Su(var)3-9 null mutant. Collectively, these data implicate HP1 in the expression of genes within euchromatin and demonstrate that HP1 uses multiple mechanisms for localization to euchromatic genes that are not regulated by SU(VAR)3-9.
To determine the role of HP1 in euchromatic gene regulation, a gene expression analysis was performed on total RNA isolated from larvae with wild-type and reduced levels of HP1. Mutations in Su(var)2-5, encoding HP1, are homozygous lethal. However, maternal supplies of HP1 allow development through the late third larval instar stage (Fanti et al., 1998; Lu et al., 2000). Therefore, RNA was isolated from heteroallelic Su(var)2-5 mutants at the early third larval instar stage, just when HP1 protein levels are no longer apparent by western analysis and by polytene chromosome staining (data not shown). Affymetrix microarray analysis revealed that 284 genes show a twofold or greater increase in expression and 261 genes show a twofold or greater decrease in expression in the Su(var)2-5 mutant compared with wild-type (Supplementary Table S1, which can be viewed at http://www.interscience.wiley.com/jpages/1058-8388/suppmat ). These 545 genes map throughout the genome, suggesting HP1 acts globally to regulate expression. A wide range in the magnitude of change in expression was observed, with 62% of the genes showing a 2- to 5-fold change and 4% showing a greater than 30-fold change in expression.
Gene ontology analysis (Liu et al., 2003) showed that the misregulated genes encode a variety of structural and enzymatic proteins (Supplementary Table S2). Overall, 17% of the categorized genes that are expressed during the third larval instar stage showed altered expression in the Su(var)2-5 mutant. Categories where at least 50% of the genes were misexpressed in the Su(var)2-5 mutant include oxygen binding, defense/immunity, heat shock regulation, receptor binding, puparial glue formation, and cuticle structure. These changes in gene expression might be related to impending death from the loss of HP1; however, only one of 58 genes known to be involved in cell death showed altered expression in the mutant. It is difficult to ascertain whether changes in particular mRNA levels are due to a primary consequence of the loss of HP1, as in the case of direct transcriptional regulation, or due to indirect secondary consequences. Given that the expression of at least five transcriptional regulators is altered in the Su(var)2-5 mutant, indirect effects are likely (Supplementary Table S2). In support of direct transcriptional regulation, 78 genes showing a twofold or greater increase in expression and 58 genes showing a twofold or greater decrease in expression map to sites of HP1 localization at the cytological level (Fig. 1). These genes represent candidates for direct targets of HP1.
HP1 Is Associated With Three Genes Within Region 31
Cytological region 31 was selected for detailed analysis due to the fact that it contains eight bands that stain intensely with antibodies against HP1. This region encompasses 550 kb of DNA containing 121 genes, 50 that are expressed during the third larval instar stage. Of the 50 genes, 10 show altered expression in the Su(var)2-5 mutant. Clearly, the gene density within region 31 is typical of euchromatin; however, a BLAST search was preformed to determine whether sequences within this region resemble a euchromatic context. The analysis revealed that region 31 contains 154 copies of repetitive elements, 80 copies of retroviral elements, 27 copies of Line-like elements, and 20 IR elements. Except for two retrotransposons and one Line-like element, all of the repetitive sequences are partial elements that show no evidence of clustering. For comparison purposes, a similar analysis was performed on a 550-kb region within cytological region 36F10–37C7; this region shows no HP1 localization. Within this region there are 108 genes, 45 that are expressed during the third instar larval stage and 6 that are misregulated in the Su(var)2-5 mutant. BLAST analysis revealed 219 repetitive elements, 50 copies of retroviral elements, 8 Line-like elements, and 31 IR elements. Except for a single retrotransposon, all are partial elements that show no evidence of clustering. This comparative analysis strongly suggests that region 31 has sequence complexity similar to that of another euchromatic region, one that does not associate with HP1. Furthermore, the association of HP1 within region 31 does not appear to correlate with the location of particular repetitive elements, a conclusion supported by cytological studies of HP1 distribution throughout the genome (Fanti et al., 2003). Thus, DNA sequence analysis of region 31 did not reveal insights on the mechanism of HP1 localization.
Ten genes within region 31 showed a twofold or greater change in expression in the Su(var)2-5 mutant compared with that in wild-type larvae (Table 1). These changes were confirmed by northern and/or reverse transcriptase-polymerase chain reaction (RT-PCR) experiments, and values for fold-change in expression were in agreement with the values obtained by microarray analysis (Table 1; Fig. 2). These 10 genes are not clustered and map to locations throughout region 31, similar to the localization of HP1. There is no correlation between the location of any particular type of repetitive sequence and proximity to these 10 genes, once again suggesting that sequence context is not a predominant feature for HP1 localization.
Table 1. Summary of Gene Expression and HP1 Localization Dataa
Fold Δ microarray (l)
Fold Δ secondary assay+ (l)
Fold Δ RT-PCR (sg)
Cytological colocalization on polytenes (sg)
ChIP; % of input (sg)
p-value for ChIP
HP1, heterochromatin protein 1; l, larval; sg, salivary gland; N, Northern blot used as secondary assay; R, reverse transcriptase-polymerase chain reaction used as secondary assay; nd, not determined.
Asterisks indicate statistically significant differences between HP1 association at a given gene and rp49.
Two approaches were taken to determine whether the 10 genes within region 31 associate with HP1. First, a cytological approach was performed on polytene chromosomes that were simultaneously stained with antibodies to HP1 and hybridized with cDNAs for each gene tested (Fig. 3A, and data not shown). Three of the genes (CG5676, CG5096, and cdc2) colocalize with HP1, two genes (Pros35 and CG5322) map adjacent to an HP1 band, and four genes (Pen, CG13135, Myo31DF, and CG6113) clearly do not colocalize with HP1 (Table 1). Based on the genomic position of CG5734 (14 kb proximal of Pros35) and the cytological staining pattern of HP1, it can be deduced that this gene does not colocalize with HP1. In summary, 5 of the 10 genes assayed localize within or are adjacent to an HP1 band, indicating a role for HP1 in their regulation. However, this type of cytological analysis does not provide the level of resolution needed to assign HP1 localization to a given gene.
As a second approach that offers increased resolution, chromatin immunoprecipitations using HP1 antibodies were performed to determine the association of HP1 with each candidate target gene. An hsp26 transgene inserted within centric heterochromatin was used as a positive control. This transgene shows HP1-dependent silencing and alterations in chromatin structure, strongly suggesting an association with HP1 (Cryderman et al., 1998). The rp49 gene, which does not change expression in the HP1 mutant and does not colocalize with HP1, was used as a negative control.
Chromatin immunoprecipitation experiments revealed that sequences for Pros35, CG5676, and cdc2 show enrichment for HP1 relative to rp49 (p values were 0.009, 0.014, and 0.002, respectively). Association ranged on average from 1.2 ± 0.1 to 1.8 ± 0.2% of input compared with 0.5 ± 0.1% of input for the rp49 negative control and 4.0 ± 0.6% input for the centric transgene (Fig. 3B,C; Table 1). Myo31DF showed 0.8 ± 0.1% of input with a p value of 0.071 and, therefore, is insignificant. These data are consistent with the observation that Myo31DF maps distal to the last HP1 band in the region 31 cluster. Pen, CG13135, and CG6113 reside outside of the HP1 staining within region 31 and did not show an association with HP1 (0.5 ± 0.1, 0.5 ± 0.2, and 0.7 ± 0.1% of input, respectively). Two genes, CG5096 and CG5322, mapping within or adjacent to an HP1 band, do not show association by chromatin immunoprecipitation (0.6 ± 0.0% and 0.5 ± 0.1% of input, respectively). With the exception of CG5096 that is located immediately adjacent to cdc2, the chromatin immunoprecipitation results were in good agreement with the cytological studies (Fig. 3; Table 1). All three genes, Pros35, CG5676, and cdc2, found to associate with HP1 localize within or adjacent to an HP1 band. Surprisingly, these three genes show decreased expression in the Su(var)2-5 mutant (Table 1), suggesting a new role for HP1 in gene activation at euchromatic sites.
Given that these gene expression studies were performed on whole larvae and the chromatin immunoprecipitation experiments had to be performed using salivary glands due to limitations of the antibody (see Experimental Procedures section), it was possible that HP1 could be regulating these genes differently in salivary gland versus other larval tissues. To address possible tissue-specific difference in gene regulation by HP1, RT-PCR was performed using RNA isolated from salivary glands of wild-type and Su(var)2-5 mutants. All three HP1 target genes (Pros35, CG5676, and cdc2) exhibited a twofold or greater decrease in gene expression (−2.1 ± 0.2, −5.3 ± 1.2 and −4.2 ± 1.0, respectively, Table 1) in Su(var)2-5 mutant salivary glands compared with expression in wild-type salivary glands, completely consistent with the gene expression data derived from whole larvae and supporting a role for HP1 in their activation. Thus, in this cell type, HP1 association correlates with increased gene expression.
Expression of HP1 Target Genes Is Not Altered by Loss of SU(VAR)3-9
In centric heterochromatin, HP1 plays a role in silencing euchromatic genes translocated near heterochromatin, a process that involves the methylation at K9 of histone H3 by the SET domain histone methyltransferase SU(VAR)3-9 (Bannister et al., 2001). We examined the modification status of K9 H3 for genes within region 31 and tested for effects of Su(var)3-9 on expression of the genes affected by HP1 dosage. In vitro, HP1 associates with both di- and trimethylated K9 H3 (Jacobs and Khorasanizadeh, 2002). Therefore, polytene chromosomes were co-stained with methylated K9 H3 and HP1 antibodies. As anticipated from genetic studies with Su(var)3-9, strong colocalization of both HP1 and di-mK9 H3 is observed in the chromocenter (Fig. 4A,B). In contrast, several sites of di-mK9 H3 in euchromatin did not colocalize with HP1. Likewise, several euchromatic sites of HP1 did not show colocalization with di-mK9 H3, including a subset of HP1 sites within region 31 (Fig. 4A). To test for colocalization with tri-mK9 H3, polytene chromosomes were stained with antibodies that specifically recognize trimethylated K9 H3 and HP1. Whereas strong colocalization was observed at the chromocenter, little staining with this antibody was observed within region 31 (Fig. 4B). Taken together, these data suggest that HP1 might use different mechanisms for heterochromatic and euchromatic localization.
Given the presence of di-mK9 H3 within region 31, chromatin immunoprecipitation experiments were performed using antibodies specific to di-mK9 H3 to determine whether this histone modification is present at genes associated with HP1. An hsp26 transgene inserted within centric heterochromatin was used again as a positive control. This transgene exhibits a reduction in silencing in a Su(var)3-9 mutant background (data not shown), suggesting packaging with mK9 H3. The rp49 gene, located in a chromosomal region that does not stain with di-mK9 H3, served as a negative control. Chromatin immunoprecipitation experiments revealed that only cdc2 showed enrichment for di-mK9 H3 (7.4 ± 1.1% of input; Fig. 4C,D). These data show that of the three genes associated with HP1, cdc2 is the only gene in which localization might occur through an interaction with di-mK9 H3 produced by SU(VAR)3-9 as it does in centric heterochromatin.
To determine whether genes within region 31 are regulated by SU(VAR)3-9, their expression was examined in a Su(var)3-906 homozygous null mutant (Schotta et al., 2002). No effect on mRNA levels of Pros35, CG5676, or cdc2 was observed (Fig. 5), demonstrating that SU(VAR)3-9 was not required for their regulation. This finding is particularly interesting for cdc2 in which di-mK9 H3 is observed by chromatin immunoprecipitation.
The microarray data in combination with the HP1 localization studies strongly suggest that HP1 regulates euchromatic gene expression; hundreds of genes are up- and down-regulated in the Su(var)2-5 mutant (Supplementary Table S1; Fig. 1). Given the classic role of HP1 in gene silencing, it was anticipated that HP1 would be associated with silenced euchromatic genes. However, our focused study on genes within cytological region 31 has provided an alternative view. The three genes, Pros35 (encoding the 35-kd subunit of the proteasome), cdc2 (encoding a cell cycle dependent protein kinase), and CG5676 (encoding a predicted protein of uncharacterized function), found to associate with HP1 are down-regulated in the Su(var)2-5 mutant. These data implicate a positive role for HP1 in euchromatic gene regulation.
Several studies have suggested a role for HP1 in euchromatic gene regulation. It has been reported that genes within region 31 are silenced by HP1 based on gene expression studies alone; in particular, CG13135 was shown to be up-regulated in the Su(var)2-5 mutant (Hwang et al., 2001). Consistent with this finding, our gene expression analysis also showed an up-regulation of CG13135 in a Su(var)2-5 mutant (Table 1). However, the HP1 localization data presented here (both cytological studies and chromatin immunoprecipitation data), failed to detect enrichment of HP1 at CG13135 on salivary gland chromosomes (Table 1). These data suggest that the altered expression of CG13135 in the Su(var)2-5 mutant is either due to an indirect effect or that this gene is associated with HP1 in other tissues.
Genes within region 31 have been reported as targets of HP1 based on localization studies alone. To identify HP1 target genes, the Escherichia coli DNA adenine methyltransferase (Dam)-HP1 fusion protein was expressed in Drosophila tissue culture cells (derived from embryos; van Steensel et al., 2001; Greil et al., 2003). Target genes show an enriched hybridization to methylated genomic fragments; this technique is referred to as Dam ID (van Steensel et al., 2001). Unlike the enrichment of HP1 within region 31 observed on salivary gland chromosomes, no enrichment within this region was observed using Dam ID. Of the 50 genes tested within region 31, three genes, CG5734, CG5694, and mRpS, showed association with HP1 using Dam ID. For example, CG5734 showed an enrichment compared with controls using Dam ID, whereas, no enrichment using chromatin immunoprecipitation was observed here (Table 1; Fig. 3B,C). Discrepancies between the Dam ID results and this study might be attributed to differences in sensitivity of detection between the techniques or tissue/stage specific differences in HP1 localization. In summary, the study reported here is the first to compare both gene expression and HP1 localization in the same type of cells at the same stage of development. These experiments identified Pros35, cdc2, and CG5676 as down-regulated in the Su(var)2-5 mutant and as associated with HP1 during transcriptional activation, strongly indicating a requirement for HP1 in gene expression.
A role for HP1 in the positive regulation of euchromatic gene expression has been recently reported (Piacentini et al., 2003); however, the mechanism appears to be different from that reported here. HP1 was found in association with genes that are strongly induced, such as those activated during heat shock (Piacentini et al., 2003). These genes exhibit chromosomal “puffing” on Drosophila polytene chromosomes, a morphological feature associated with high levels of transcription. Association with HP1 at sites of puffing was sensitive to RNase treatment, suggesting an interaction with RNA, possibly transcripts of the gene itself. Consistent with this idea, HP1 was found in association with the coding region, but not the promoter region, of induced genes. These findings are in contrast to those reported here. HP1 was found in association with fragments that spanned the promoter regions of Pros35, cdc2, and CG5676. Furthermore, HP1 association is retained at region 31 after treatment with RNase (Piacentini et al., 2003). RT-PCR analysis of the expression of Pros35, cdc2, and CG5676 (using RNA isolated from whole larvae or salivary glands) demonstrated that these genes are not induced by heat shock (data not shown). Furthermore, region 31 does not show regions of puffing upon heat shock or during larval development. In fact, the entire region is relatively restricted. Taken together, these data show that HP1 associates with euchromatic genes through both RNA-dependent and -independent mechanisms and that both contribute to positive regulation of euchromatic genes.
Measures were taken to perform micorarray analysis on RNA from larvae of similar genetic backgrounds and stages of development. The y1,w67c23 stock was used as the wild-type. Two stocks in which a Su(var)2-5 mutation had been introduced into y1,w67c23 and maintained over a Curly (Cy) marked balancer containing sequences encoding green fluorescent protein (GFP; Flybase) were used. These stocks were generated through a series of crosses using balancers such that all chromosomes, with the exception of the second chromosomes, were from the y1,w67c23 stock. The two Su(var)2-5 alleles, Su(var)2-504 and Su(var)2-505, used here should not produce a functional protein (Fanti et al., 1998). When used in heteroallelic combination, there is little concern regarding the homozygous state for second site mutations on the second chromosome present on the Su(var)2-5-containing chromosome. A cross between Su(var)2-504/Cy,GFP and Su(var)2-505/Cy,GFP results in several classes of progeny. Because the balancer chromosome is homozygous embryonic lethal, all non-GFP progeny are trans-heterozygous for the Su(var) alleles, representing the desired class for RNA analysis. Heterozygous Su(var)2-5/Cy,GFP siblings were not used as controls, because this genotype influences expression of euchromatic genes placed in heterochromatin. The following procedure provided synchronization for developmental staging of larvae used for RNA isolation. Genetic crosses to generate the heteroallelic Su(var)2-5 mutants and the control stock were established by seeding the bottles with 50 adults (25 virgin females and 25 males) and placed at room temperature. Adults were cleared from the bottles on day 7 and early third instar larvae of the same size were harvested from the food. For microarray analysis, 10 μg of total RNA from two independent preparations of 50 larvae for each genotype was prepared using standard Affymetrix protocols (University of Iowa DNA Core). Labeled cRNA representing each genotype was hybridized to Affymetrix Drosophila gene chip (version 1). Comparison of the gene expression profile for the two y1,w67c23 samples showed 89% agreement; comparison of the two Su(var)2-5 mutant samples showed 92% agreement. These data demonstrate consistency between preparations. The average of all four comparisons were used for the data shown in Supplementary Table S1. Data from the microarray experiments were analyzed using Affymetrix Microarray Suite version 4 and Microsoft Excel two-sample unequal variance Student's t-test.
Total RNA for Northern analysis was isolated from third instar larvae (staged as described above) using Trizol (Invitrogen); 30 μg was used for each lane in the Northern analysis. cDNAs representing each gene were labeled with 32P-dATP (Amersham) using random priming (Roche); Small ribonucleoprotein particle B, SmB, that does not show a change expression in the Su(var)2-5 mutant by microarray analysis, was used as a negative control. The results were quantitated using an Instant Imager (Packard) using ribosomal protein gene rp49 as a loading control. Northern analyses were performed on three independent RNA preparations for each genotype.
RNA for RT-PCR analysis was isolated from 70 larval salivary glands from whole larvae (staged as described above) using Trizol (Invitrogen). After DNase I (Invitrogen) treatment, 3.5 μg of total RNA was used to make cDNA for subsequent PCR reactions according to the SuperScript First-Strand Synthesis System for RT-PCR (Invitrogen). Linear PCR was performed on the cDNA using primers specific for each gene being analyzed. The resulting products were quantitated using a CCD camera and the Labworks 4 image software system (UVP Laboratory Products). Differences in the amount of PCR material loaded in each lane was corrected using the rp49 or SmB signal intensity; both genes do not change in expression in the Su(var)2-5 mutant.
Third instar larval salivary glands were fixed, squashed, and stained (Platero et al., 1995), with antibodies to HP1 (C1A9; gift of S. Elgin), di-mK9 H3 (gift of C.D. Allis), and tri-mK9 H3 (gift of P. Singh). For immunostaining and sequential DNA in situ hybridization, polytene chromosomes were fixed, squashed, and stained with C1A9, and a fluorescently labeled cDNA represented each gene tested (Berloco et al., 2001).
Chromatin immunoprecipitation experiments were essentially preformed as described (Danzer and Wallrath, 2004). Here, salivary glands (50 pairs) from third instar larvae were used for cross-linking and subsequent precipitation with HP1 antibodies (#PRB-291C, Covance) or di-mK9 H3 antibodies (#07-030, Upstate Biotechnology). Salivary glands were used due to limitations associated with the HP1 polyclonal antibody. In addition to HP1, this antibody recognizes several other proteins on Western blots of total larval extract (see Danzer and Wallrath, 2004, for additional details). Polyclonal GFP antibodies (Molecular Probes) were used as a nonspecific antibody control. Linear PCR was performed on the immunoprecipitated DNA using primers specific for sequences corresponding to each gene tested (Supplementary Table S3). Enrichment of the gene sequences in the immunoprecipitated material relative to their abundance in the input fraction is stated as “percent of input.” Primers for rp49 were used as a negative control. Primers for an hsp26 transgene (Wallrath and Elgin, 1995) in centric heterochromatin were used as a positive control. Chromatin immunoprecipitation experiments were performed at least three times using independent starting material for each assay. A Student's t-test was performed to determine whether the amount of precipitated material was statistically significant from that obtained for rp49. p values less than 0.05 were considered significant.
We thank S.C.R. Elgin for the C1A9 antibodies, C.D. Allis for di-mK9 H3 antibodies, P. Sing for the tri-mK9 H3 antibodies, G. Reuter for the Su(var)3-906 mutants, the University of Iowa DNA Core for assistance with the Affymetrix microarrays, and members of the Wallrath lab for comments on the manuscript. L.L.W. was funded by the National Institutes of Health, and S.K.G. was supported by an American Heart Association predoctoral fellowship.