The genome-wide enrichment profiles of 20 histone modifications and 25 chromosomal proteins demonstrate the distinct nature of chromatin on Drosophila chromosome 4. As anticipated based on the behavior of transgene reporters 
, we found that chromosome 4 sequences are almost ubiquitously packaged with marks commonly associated with heterochromatin, H3K9me2, H3K9me3, HP1a, and HP2 (). The TSSs of active genes are depleted for these marks (). Surprisingly, “permissive” domains, which allow full expression of reporter genes, were found not to resemble euchromatin, but to show evidence of Polycomb
regulation (associated with H3K27me3 and PC in some cell types) (). The association with Polycomb
marks is cell-type specific; thus, some genes on chromosome 4 appear to be able to switch between the two main silencing systems in what appears to be a developmentally regulated process. We do not know the state of the Polycomb
regulated domains in the cells of the eye imaginal disc, where white
reporter expression is required to result in a red eye phenotype. It is possible that in these cells the Polycomb
regulated domains are associated with its activating antagonist, trithorax
, and its partners. However, packaging in the PC state, which appears to exclude HP1a and H3K9me2/3 in this situation, is sufficient to allow DNase1 hypersensitive site (DH site) formation at the genes in these domains in BG3 cells, while such sites are not evident when the same genes are packaged with HP1a and H3K9me2/3 in S2 cells (modENCODE data tracks; www.modENCODE.org
). Given that loss of DH site formation has been observed for the variegating reporter 
, a domain that permits DH site formation may be sufficient for reporter expression.
Due to the unusual chromatin environment, chromosome 4 genes experience a unique regulatory system and display decreased polymerase pausing (). Mutant analysis indicates that the RNA pol II distribution pattern is dependent on HP1a (). In mutants lacking HP1a or POF, enrichment of RNA pol II decreases in the gene bodies, leading to an increase in PI due to the now strongly TSS-biased RNA pol II distribution (). This shift in RNA pol II seen in pof
mutants is potentially an indirect effect due to the inability to recruit HP1a to active gene bodies in the absence of POF. Alternatively, both HP1a and POF together might be required for the wildtype RNA pol II distribution – and the decrease of polymerase pausing - normally seen on chromosome 4. How HP1a, and possibly POF, influence polymerase distribution is still unknown. This influence might occur at various steps of RNA pol II regulation, either by interfering with the establishment, maintenance, or resolution of the paused polymerase, or by promoting elongation. For example, the Positive Transcription Elongation Factor b (P-TEFb) and PAF1C act by promoting elongation 
. On the other hand, Min and colleagues found that in mouse embryonic stem cells, “bivalent” genes associated with PRC1 and PRC2 display low levels of polymerase pausing, possibly due to their chromatin structure 
. POF's influence could be mediated by its RNA recognition motif 
and its ability to interact with RNA transcripts 
, leading to a positive effect on gene expression 
. Our data, however, indicate that POF alone is insufficient to determine the RNA pol II distribution on chromosome 4 genes, and that HP1a is vital for their regulation. Thus, further work is needed to elucidate the exact mechanism of interaction between POF, HP1a, and the polymerase.
Another protein to consider in the regulation of chromatin structure and RNA polymerase distribution on chromosome 4 genes is JIL-1, which is enriched on chromosome 4 (). JIL-1 is an H3S10 kinase; it limits heterochromatin extent, as in its absence, HP1a and H3K9me2 spread to new genomic regions 
. Depletion of JIL-1 overall has little effect on gene expression 
, with the major effect being on the X chromosome, with approximately 10% of the genes affected, based on our analysis. In contrast, ~5% of the chromosome 4 genes are affected, less than the percentage of X chromosome genes but slightly more than seen in the remainder of the genome. As in HP1a and POF mutants, the expression of the affected genes decreases. However, given the small number of genes affected by JIL-1 depletion, the impacts of HP1a/POF depletion are unlikely to be dependent on JIL-1. This interpretation is supported by the genetic interaction analysis of JIL-1 and HP1a, which indicates that their mutations counteract each other's effects, and that the spread of H3K9me2 triggered by Jil-1 mutations is not dependent on HP1a 
While HP1a is best known for its role in heterochromatin formation and silencing, several reports have also linked HP1a to regulation of transcriptional activity of both heterochromatic and some euchromatic genes 
. Heterochromatic genes light
are reported to be dependent on a heterochromatic environment, and specifically on HP1a, for optimal expression 
, and we find that the majority of the chromosome 4 genes show a similar dependence (). The distribution of H3K9me2/me3 at several active heterochromatic genes shows depletion at the TSS 
, as reported here for chromosome 4 genes. However, it has recently been reported that two chromosome 4 genes, CAPS
, lose DNase accessibility at the 5′ DH site in the absence of HP1a 
. Thus, while absence of HP1a and other silencing marks from the TSS is associated with gene expression in heterochromatic and chromosome 4 genes, the presence of HP1a in the domain as a whole appears to be required for DH site formation at these genes. In contrast, HP1a domains are prohibitive for DH site formation at the TSSs of eukaryotic reporter genes inserted into these regions 
In euchromatin, we have found HP1a associated with a number of TSSs, a finding that is supported by the detection of small amounts of HP1a in chromosome arms of polytene chromosomes 
. Others have identified HP1a as a positive regulator of more than 100 genes, associating with the transcript and apparently facilitating elongation 
. HP1a has been reported to interact with dKDM4, an H3K36 demethylase, 
, whose yeast homologs promote transcript elongation 
. Thus, there are precedents for an “activating” role for HP1a, and an interaction with dKDM4 provides an attractive model for how HP1a might influence RNA pol II processivity and pausing. However, what remains to be determined is why polymerase pausing would be affected specifically on chromosome 4 rather than also affecting genes in pericentric heterochromatin. We note that while the overall pericentromeric domains are strongly enriched for HP1a, one does not see the increase over the gene body observed for the chromosome 4, and hence these genes do not exhibit the same contrast between TSS and gene body observed for chromosome 4 genes (). POF may play a role in enhancing HP1a presence at active genes on chromosome 4.
The chromatin structure analysis we present from mutants lacking POF, HP1a, and EGG is mostly in agreement with previously published results based on polytene chromosome analysis. On chromosome 4, lack of POF induces loss of HP1a 
, H3K9me2 
, and H3K9me3 (our study). However, our higher resolution analysis reveals a pool of HP1a, H3K9me2, and H3K9me3 associated with repeated sequences on chromosome 4 that is independent of POF. Also in contrast to prior findings 
, our results indicate that POF is maintained on chromosome 4 independent of HP1a, as mutants lacking HP1a still show a normal POF enrichment pattern. Note that HP1a depletion was accomplished here by a heteroallelic cross; thus HP1a was present during the initial assembly of heterochromatin.
It has been postulated that POF is recruited to chromosome 4 from a site close to the centromere of the chromosome, based on translocation studies 
. However, the affinity of POF for transcribed genes leads to an enrichment pattern that changes from cell type to cell type, arguing against a simple recruitment and spreading model (comparison of modENCODE data from Bg3 and S2 cells). Our analysis of mutants (resulting in depletion) of Su(var)205
, and pof
products instead suggests a model where there is a simultaneous requirement for EGG and POF, which together create conditions to recruit HP1a to active gene bodies on chromosome 4, presumably utilizing H3K9 methylation by EGG. EGG and POF are reported to physically interact with each other in vivo
, providing a mechanism for this process. How the complex is targeted to chromosome 4 active genes remains to be established.
An interesting aspect of our study is that on chromosome 4, the association between HP1a, H3K9me2, and H3K9me3 is substantially different from what is observed in pericentric heterochromatin (). The loss of the strong correlation between H3K9me2 and H3K9me3 on chromosome 4 is likely due to differences in H3K9 HMTs. While little is known about G9a, both EGG and SU(VAR)3-9 have been examined in our study and by others 
. Both EGG and SU(VAR)3-9 are found on chromosome 4, but the predominant H3K9 methylation signal depends on EGG 
. Our analysis suggests that H3K9me2 and H3K9me3 enrichment on chromosome 4 reflects both HP1a-dependent and HP1a-independent mechanisms. H3K9me2 and H3K9me3 enrichment on chromosome 4 reflects an EGG-dependent mechanism to modify the histone over the body of the genes, and an EGG-independent mechanism to modify the histone associated with repeat sequences (). Presumably the latter reaction is carried out by SU(VAR)3-9. We suggest that this activity of SU(VAR)3-9 was missed in the polytene chromosome studies of Su(var)3-9
, as EGG appears to be responsible for ~80% of the H3K9me2/me3 signal on chromosome 4 in our analysis. Currently, it is unclear how the HMT activities on chromosome 4 are coordinated. In HeLa cells, several H3K9 HMTs interact with each other 
, thus providing potential mechanisms for coordination. However, how the enzymes on Drosophila chromosome 4 produce the H3K9me2 and H3K9me3 enrichment pattern as well as the active gene-specific increase in H3K9me3 remains to be discovered.
The available data suggest the following model for the assembly of chromatin on chromosome 4 and regulation of the genes in this domain (). Two mechanisms recruit HP1a to chromosome 4, one dependent on POF and EGG, the other independent of these components. POF is required for the recruitment of HP1a and H3K9 methylation in gene bodies of actively transcribed genes, and EGG appears to be required for significant recruitment or stabilization of POF. POF in turn interacts with the nascent transcript, positively affecting transcript output. Neither POF nor EGG is required for the recruitment of HP1a and the presence of H3K9me2/me3 at repeat clusters (and silent genes) on the chromosome 4. These findings suggest that the same general mechanisms that result in heterochromatic packaging of repetitious, TE-derived DNA in pericentric heterochromatin are at work here as well. Studies in plants and some fungi suggest that small RNAs play a role in targeting heterochromatin formation, and there is growing evidence for such a mechanism establishing heterochromatin patterns in the germline and early embryo of Drosophila 
. However, direct targeting of one of the heterochromatin components by other means (such as direct DNA recognition) remains a possibility. The analysis above clearly shows that chromosome 4 is a mosaic of HP1a-associated domains, with each of the two modes of assembly detected here potentially impacting gene expression.
A model illustrating the two mechanisms proposed for HP1a assembly on chromosome 4.