Enrichment of HP1a on Drosophila chromosome 4 genes creates an alternate chromatin structure critical for regulation in this heterochromatic domain

Abstract

Chromatin environments differ greatly within a eukaryotic genome, depending on expression state, chromosomal location, and nuclear position. In genomic regions characterized by high repeat content and high gene density, chromatin structure must silence transposable elements but permit expression of embedded genes. We have investigated one such region, chromosome 4 of Drosophila melanogaster. Using chromatin-immunoprecipitation followed by microarray (ChIP-chip) analysis, we examined enrichment patterns of 20 histone modifications and 25 chromosomal proteins in S2 and BG3 cells, as well as the changes in several marks resulting from mutations in key proteins. Active genes on chromosome 4 are distinct from those in euchromatin or pericentric heterochromatin: while there is a depletion of silencing marks at the transcription start sites (TSSs), HP1a and H3K9me3, but not H3K9me2, are enriched strongly over gene bodies. Intriguingly, genes on chromosome 4 are less frequently associated with paused polymerase. However, when the chromatin is altered by depleting HP1a or POF, the RNA pol II enrichment patterns of many chromosome 4 genes shift, showing a significant decrease over gene bodies but not at TSSs, accompanied by lower expression of those genes. Chromosome 4 genes have a low incidence of TRL/GAGA factor binding sites and a low T(m) downstream of the TSS, characteristics that could contribute to a low incidence of RNA polymerase pausing. Our data also indicate that EGG and POF jointly regulate H3K9 methylation and promote HP1a binding over gene bodies, while HP1a targeting and H3K9 methylation are maintained at the repeats by an independent mechanism. The HP1a-enriched, POF-associated chromatin structure over the gene bodies may represent one type of adaptation for genes embedded in repetitive DNA.

Figure 1. The chromatin composition of D. melanogaster chromosome 4 shows distinct patterns of enrichment.

A. Enrichment levels for all histone marks/chromosomal proteins (green asterisks indicate newly reported marks) are shown for the five main combinatorial chromatin states within heterochromatin as defined in Riddle et al . Panel 1- histone marks; panel 2- chromosomal proteins; panel 3- repeat enrichment and expression status; panel 4- the fraction of chromatin within each state associated with various structural features of genes; panel 5- enrichment/depletion of states for each chromosome arm, with the numbers to the right reflecting the percentage of the heterochromatin assigned to each state. Chromatin source: BG3 cells. B. Karyotype view of the assembled heterochromatic domains defined by the five combinatorial chromatin states in A. The distal 1.2 Mb region of chromosome 4 exhibits a higher density of transcriptionally active genes (states B, C, D) and polycomb-dominated domains (state E). Chromatin source: BG3 cells. C. hsp70-white transgenes leading to a red eye phenotype are preferentially found in state E. Triangles mapped onto the expanded chromosome 4 indicate the insertion sites for hsp70-white transgenes exhibiting a red (red triangle) or variegating (dotted triangle) eye phenotype . D. Browser shot illustrating the relationship between histone marks and chromosomal proteins on chromosome 4. The silent gene CG1909 shows correlations typical of pericentric heterochromatin (dashed blue box; enriched H3K9me2, H3K9me3, HP1a; state A), while the expressed gene Ephrin shows the patterns typical for chromosome 4 active genes (dashed red box; upstream promoter regions, depleted for H1 and H4, state D; regions immediately downstream from TSSs, enriched for H3K4me2/3 and depleted of H3K9me2/3, state C; and regions across the body of the gene, enriched for H3K36me3, H3K9me3, HP1a and POF, state B). Silent gene zfh shows the chromatin pattern typical for genes under the regulation of Polycomb system, enriched in H3K27me3 and depleted of HP1a.

Figure 2. The relationship between marks of classical heterochromatin and gene expression are altered on chromosome 4.

The strength of correlation between marks is illustrated in this diagram by the color intensity (red - positive correlation; blue - negative correlation). In pericentric heterochromatin, the black outline demarcates the strong correlation structure observed between H3K9me2, H3K9me3, and HP1a (right). This strong correlation is not present on chromosome 4; HP1a and H3K9me3 instead are positively correlated with H3K36me3, a mark of elongation, and the chromosome 4-specific protein POF (left).

Figure 3. Metagene analysis shows a unique distribution of chromosomal proteins and histone marks on chromosome 4 genes.

Enrichment (averaged smoothed M-values, Y-axis) for select chromosomal proteins and histone marks is plotted for a 3 kb scaled metagene (bp, X-axis). The enrichment is examined in three genomic domains: Chromosome 4 (top); pericentric heterochromatin (middle); and euchromatin (bottom) for active genes (left column) and repressed genes (right). Active genes on chromosome 4 have distinctive signatures of HP1a, POF, and H3K9me3 with highest enrichment levels across gene bodies. The number of genes included is demonstrated at the right corner for each figure. Data from BG3 cells.

Figure 4. Chromosome 4 has a very low incidence of polymerase pausing identified by GRO-seq data.

A. The bar graph shows the percentage of transcripts associated with RNA polymerase (grey) and the percentage of those RNA polymerase-associated genes exhibiting significant pausing (black) for euchromatin, pericentric heterochromatin, and chromosome 4. A pausing index (PI) threshold value of 10 was used. B. The low frequency of polymerase pausing observed for chromosome 4 is independent of the PI selected. C. The average T_m of 9-mers downstream of the TSS for chromosome 4 genes tends to be lower than that of other genes. T_m (Y-axis) of 9-mers in the 100 bp downstream of the TSS (bp, X-axis) is compared for genes on chromosome 4 (red) to genes classified as either paused (black) or not-paused (grey) by GRO-seq analysis. Data from S2 cells .

Figure 5. Lack of HP1a or POF shifts the enrichment pattern of RNA pol II on chromosome 4.

A. RNA pol II distribution of expressed genes on chromosome 4, comparing wildtype (WT) to HP1a (red) and POF (blue) mutants. Chromosome 4 genes are compared to genes with similar expression levels randomly chosen from the euchromatin on chromosome arm 3R (N = 58 for both). Note the dramatic loss of RNA pol II enrichment in the gene body of the mutants for chromosome 4 genes, but not for the control euchromatic genes. The TSS-proximal RNA pol II peak also changes in the mutants compared to wildtype, moving approximately 68 bp and 50 bp downstream, respectively. B. Scatter plot showing enrichment of RNA pol II at the promoter region (+/−500 bp of TSS, top panel) and gene body (bottom panel). In both HP1a and POF mutants, the RNA pol II enrichment decreases significantly only in the gene body. C. Pausing index (PI) changes specifically on chromosome 4 in HP1a and POF mutants. In both mutants, PI for chromosome 4 genes increases to a range similar to the other chromosomes. D. Examples of the shift of RNA pol II in four chromosome 4 gene regions in the mutants. In Pur-alpha, no probes were available from 581 kb–584 kb. E. Fold changes of expression level (FPKM) of chromosome 4 genes in the HP1a mutant compared to the wildtype. Data from third instar larvae.

Figure 6. Lack of POF leads to large-scale changes in HP1a and H3K9me2/3 and demonstrates that HP1a on chromosome 4 consists of POF-dependent and -independent pools.

A. Mutations in POF alter H3K9me2, H3K9me3 and HP1a enrichment on chromosome 4. Enrichment levels (M-values) are shown for H3K9me2, H3K9me3, and HP1a on chromosome 4 (left) and in pericentric heterochromatin (right) in wildtype (dark color) and pof^D119 homozygous mutant (light color) third instar larvae. Error bars: Standard error of the mean (SEM). B. Browser shots illustrating the loss of HP1a on chromosome 4 (top) in pof^D119 homozygous mutant third instar larvae and the retention of high levels of HP1a in pericentric heterochromatin (bottom panel). The M-value scale (Y-axis) is identical for wildtype and mutant but differs between marks (0–3 for H3K9me2 and H3K9me3; 0–4 for HP1a). C. Metagene plots showing H3K9me3 and H3K9me2 levels are reduced mainly over active genes on chromosome 4 in the mutant. Genes on chromosome 4 were divided into transcriptionally active (left column) and transcriptionally silent (right) based on RNA-seq data. D. The changes in H3K9me2, H3K9me3, and HP1a enrichment induced by the pof mutation correlate with gene features on chromosome 4. Changes in H3K9me2/me3 and HP1a enrichment (Y-axis: smoothed M-values) are examined separately for TSSs of actively transcribed genes, gene bodies of active genes, and silent regions on chromosome 4. Error bars: SEM.

Figure 7. Lack of HP1a does not lead to a loss of POF from chromosome 4.

A. H3K9me2 and H3K9me3 levels decrease in HP1a mutants, while POF enrichment is not reduced. The smoothed M-value (Y-axis) is shown for pericentric heterochromatin (right) and chromosome 4 (left) comparing wildtype (dark color) and trans-heterozygous Su(var)205⁰⁴/Su(var)205⁰⁵ mutants. Error bars: SEM. B. Browser shot illustrating the retention of POF enrichment on chromosome 4 in HP1a mutants (top panel) and depletion of H3K9me2 and H3K9me3 both in chromosome 4 (top) and pericentric heterochromatin (bottom). The M-value scale (Y-axis) is identical for wildtype and mutant ranging from 0 to 3. C. Changes in H3K9me2, H3K9me3, and HP1a enrichment at TSSs of actively transcribed genes, over gene bodies of active genes, and in silent regions on chromosome 4 in HP1a mutant. Error bars: SEM.

Figure 8. Lack of EGG leads to large-scale changes in POF, HP1a, and H3K9 methylation specifically on chromosome 4.

A. Depletion of EGG alters H3K9me2/3, HP1a, and POF enrichment. Scaled enrichment is shown on chromosome 4 (left) and in pericentric heterochromatin (right) comparing wildtype (dark color) and egg 10.1-1a homozygous mutant (light color). Error bars: SEM. B. Browser shots showing the reduction of POF, HP1a, and H3K9me2/3 on chromosome 4 (top panel) and the relatively small change in pericentric heterochromatin observed in egg mutants. The M-value scale (Y-axis) is identical for wildtype and mutant ranging from 0 to 3. C. The 70 kb proximal region on chromosome 4 shows minimal changes in POF, HP1a, and H3K9me2/3 levels in egg mutants, distinct from the alterations in the remainder of the chromosome illustrated in B. The M-value scale (Y-axis) is identical for wildtype and mutant ranging from 0 to 3. D. Changes in H3K9me2/me3, HP1a, and POF enrichment (Y-axis: smoothed M-values) are examined separately for TSS of actively transcribed genes, their gene bodies, and silent regions on chromosome 4. Error bars: SEM. Data from third instar larvae.

Figure 9. A model illustrating the two mechanisms proposed for HP1a assembly on chromosome 4.

In active transcribed regions, POF and EGG are recruited, which leads to high levels of HP1a across the gene body (left). Details are unknown; POF is not reported to interact directly with HP1a, so other protein partners may be involved, in addition to HP1a binding to H3K9me2/3. The chromatin complex containing POF, HP1a and other partners has a general positive effect on RNA pol II – possibly by affecting transcription elongation - specifically on chromosome 4. In silent, repeat-rich regions, which lack POF enrichment, H3K9me marks deposited by a second histone methyltransferase (presumably SU(VAR)3-9) lead to a POF-independent assembly of HP1a–containing chromatin (right). Note that neither EGG, POF, nor SU(VAR)3-9 are known to interact directly with DNA; the binding described occurs in a chromatin context.