Suppressor of variegation 205 (here termed Heterochromatin protein 1, or HP1) is a chromosomally bound protein. It is not considered one of the originally described Polycomb group members, because it does not alter segment identity. Nevertheless, it shares many other properties with proteins of the Polycomb group.

HP1 is a nonhistone chromosomal protein primarily associated with the pericentric heterochromatin and telomeres in Drosophila. The molecular mechanism by which HP1 specifically recognizes and binds to chromatin is unknown. A study tested whether HP1 can bind directly to nucleosomes. It turns out that HP1 can bind both nucleosome core particles and naked DNA. HP1-DNA complex formation is length-dependent and cooperative but relatively sequence-independent. Histone H4 amino-terminal peptides bind to monomeric and dimeric HP1 in vitro. Acetylation of lysine residues has no significant effect on in vitro binding. The C-terminal chromo shadow domain of HP1 specifically binds H4 N-terminal peptide. Neither the chromo domain nor chromo shadow domain alone binds DNA; intact native HP1 is required for such interactions. Together, these observations suggest that HP1 may serve as a cross-linker in chromatin, linking nucleosomal DNA and nonhistone protein complexes to form higher order chromatin structures (Zhao, 2000).

To test whether HP1 can bind to nucleosomes, an electrophoretic mobility shift was performed assay using purified recombinant HP1 protein (rHP1) and labeled reconstituted nucleosome core particles. rHP1 significantly retards the electrophoretic mobility of reconstituted nucleosome core particles in a dose-dependent manner, indicating an HP1-nucleosome association. Significant effects on core particle mobility are obvious at 1 µM HP1 or at a molar ratio of 100:1 of HP1 over core particles. Prominent features of the nucleosome surface that could provide substrates for HP1 binding include the N-terminal histone tails and DNA. To test whether this binding depends on the unstructured histone tails that extend from the nucleosome core, the nucleosome electrophoretic mobility assay was repeated using trypsinized nucleosomes. Limited trypsin digestion of nucleosome cores selectively digests the N-terminal tails of core histones. Under these conditions, a significant fraction of the labeled DNA dissociates from the histone cores and migrates as free DNA. Interestingly, after trypsin digestion, a distinctive electrophoretic behavior is observed. The trypsinized core fraction contains somewhat more free DNA, and this material disappears with increasing HP1, while a distinctive slow mobility complex appears. The trypsinized core particles begin to disappear at higher HP1 concentrations than required for comparable shifts of intact cores, suggesting that trypsin digestion of core particles makes HP1 binding to cores less efficient. In addition, the disappearance of the free DNA band with increasing HP1 concentration suggests that HP1 binds free DNA as well as nucleosome cores. A direct comparison of intact and trypsinized cores shows that the trypsinized cores have significantly higher mobility than intact core particles, consistent with the expected loss of N-terminal tail peptides from the particles (Zhao, 2000).

To confirm HP1-DNA interaction, the ability of rHP1 alone, in the absence of nucleosomes, to shift the Cy5-labeled 146-bp DNA alone was tested. A mobility shift of the DNA fragment occures with increasing amounts of HP1 protein, indicating DNA binding. Furthermore, the abrupt appearance of complexes between 5 and 10 µM HP1 (between 500- and 1000-fold molar excess of HP1) suggests cooperativity in complex formation. To test for sequence or structural specificity for DNA in HP1-DNA binding, circular BlueScript plasmid DNA (pBS) or linear DNA fragments produced by AluI digestion of the pBS plasmid were included in the gel shift assay. Both the circular plasmid and the linear DNA fragments are able to eliminate the band shift of the Cy5-labeled 146-bp 5 S rDNA fragment. Thus, the HP1-DNA interaction is not sequence-specific, and HP1 binding does not require free DNA ends. HP1 complexes form with increasing efficiency as DNA length increases. While HP1-DNA interactions could substantially account for HP1-nucleosome binding, there appears to be some contribution of histone (and/or DNA conformation imposed by nucleosome wrapping) at high salt concentration (Zhao, 2000).

Side by side comparison of the mobility shifts observed with DNA and with nucleosome cores in the presence of HP1 emphasizes the distinct electrophoretic behaviors of these complexes. When the naked DNA and the nucleosome are run side by side in the gel shift assay, the pattern of band shift is different. While a small amount of material in the core fraction co-migrates with the HP1-DNA complex, reflecting the presence of histone-free DNA after core reconstitution, most of the HP1-dependent core complex migrates more slowly. This difference indicates that HP1 binds to nucleosome core particles instead of just stripping off the DNA from the nucleosome (Zhao, 2000).

Since HP1-nucleosome complexes appear somewhat more salt-resistant than HP1-DNA complexes and since the binding of HP1 to trypsinized cores appears less efficient than to intact cores, a role for histone tail binding in the HP1-nucleosome interaction was considered. The histone H4 N-terminal tails have been implicated in nucleosome assembly and the regulation of template accessibility in chromatin. They extend out from the nucleosome core and contain four acetylatable lysines (Lys5, Lys8, Lys12, and Lys16). In Drosophila, euchromatic histone H4 can be modified at any of these four lysines, whereas heterochromatic H4 is modified primarily at Lys12. The distinctive distribution of acetylated histone H4 isoforms suggests that HP1 may recognize and bind to heterochromatin through histone H4 tails specifically acetylated at K12. To test whether HP1 interacts with the H4 N-terminal tail and whether acetylation regulates this interaction, five synthetic 20-amino acid peptides based on the conserved H4 tail sequence were challenged to bind to HP1 in vitro. One peptide was nonacetylated, and the other four peptides were acetylated at one of four different lysines (Lys5, Lys8, Lys12, Lys16). These peptides were labeled with the Cy5 at their C-terminal cysteine residues. rHP1 was mixed with Cy5-labeled peptides, subjected to chemical cross-linking, and cross-linked proteins were resolved by SDS-PAGE. All five peptides are cross-linked to rHP1 with similar efficiency, suggesting that, although histone tails bind to rHP1, acetylation does not affect this interaction (Zhao, 2000).

In these experiments, peptide cross-linking was also observed to a protein of slower mobility with twice the apparent molecular weight of HP1. Since the rHP1 preparation is >95% pure, this was inferred to be dimeric HP1. Drosophila HP1 chromo shadow domain peptides can dimerize in solution, supporting this inference. To confirm that full-length HP1 also dimerizes in vitro, cross-linking was performed on purified rHP1 protein and the cross-linked products were analyzed by SDS-PAGE and Western blot with an HP1 antibody. Intact rHP1 indeed forms dimers and trimers in vitro. To identify which region of HP1 binds to the H4 peptides, an rHP1 sample was used in which the protein had been cleaved into two peptides. The identities of these two peptides as the N-terminal chromo domain and the C-terminal chromo shadow domain were established using a chromo domain-specific antibody. In the cross-linking assay, H4 peptides were found to be associated with the C-terminal chromo shadow domain (Zhao, 2000).

To test the domain specificity of DNA binding, three peptides from bacterial HP1 were used: an N-terminal peptide that contains the chromo domain, a C-terminal peptide that contains the chromo shadow domain, and an 84-amino acid peptide in the middle of the protein that links the two domains. None of these three peptides binds to DNA in vitro, indicating that an intact HP1 protein is required for this activity. Intact HP1 still binds DNA in the presence of chromo domain peptides or the central peptide, indicating that peptides do not compete with intact HP1 binding and that there is no contaminating inhibitor in the peptide fractions (Zhao, 2000).

The cross-linking assays provide clear evidence for an interaction between H4 N-terminal tail peptides and HP1. The cross-linking of histone H4 tails to HP1 dimers suggests that H4 tail binding occurs on a surface of the chromo shadow domain, distinct from that mediating chromo shadow domain dimerization. Gel mobility shift assays demonstrate binding by HP1 of both nucleosome core particles and naked DNA. Although the experiments cannot apportion the relative contributions of HP1-histone and HP1-DNA interactions, it appears that both contribute to the HP1-nucleosome association. The histone tails are apparently relatively unstructured in the nucleosome core and are thought to extend far enough beyond the core particle to permit contacts with the DNA wrapped on the surface and/or contacts with adjacent nucleosomes. The effect of trypsin digestion is consistent with a role for histone tails in HP1 binding to nucleosome cores. However, the overall similarity in salt sensitivity of the HP1 core particle and HP1-DNA interactions suggests that HP1-DNA interaction plays an important role in HP1-nucleosome binding (Zhao, 2000).

Binding of nucleosomes requires rHP1 concentrations in the low micromolar range. A rough estimate of HP1 concentration in Drosophila suggests that there is about one HP1 molecule per 15 nucleosomes in a third instar larval nucleus. Thus, normal nuclear HP1 levels should be in the low micromolar range as well. This suggests that the interactions measured are biologically relevant for a relatively abundant chromatin-binding protein whose binding is dynamic during the cell cycle (Zhao, 2000 and references therein).

Recombinant HP1 binds to eukaryotic (5 S rDNA) and prokaryotic DNA (pBS plasmid and lac operon cAMP receptor protein binding site) sequences, suggesting a lack of a strong sequence preference in HP1 binding to DNA and nucleosomes. While it makes sense that HP1 could interact with DNA and nucleosomes, a general and sequence-nonspecific affinity for DNA or nucleosomes cannot explain the relatively restricted chromosomal distribution of HP1 in vivo. Presumably, there are other HP1-interacting factors that confer additional specificity on HP1 binding, serving to target or stabilize HP1 binding in heterochromatin. In addition, hyperphosphorylation of HP1 correlates with heterochromatin assembly, and phosphorylation by casein kinase II is required for efficient heterochromatin targeting. These observations, considered in light of the results reported here, suggest that HP1 phosphorylation could act by regulating HP1-nucleosome binding (Zhao, 2000).

The apparent cooperativity in HP1-DNA complex formation is consistent with a model that heterochromatin complexes spread from an 'initiator' in a cooperative and sequence-independent mechanism. When this spreading occurs across a rearrangement breakpoint, it can result in heterochromatic position effect silencing. Virtually any euchromatic gene can be silenced by heterochromatin, and genes lying hundreds of kilobases from a breakpoint may be silenced, arguing that heterochromatin complexes are relatively efficient and promiscuous in their ability to form higher order chromatin structures. The apparent sequence neutrality and cooperativity of HP1-DNA complexes observed here may underlie the in vivo properties of the HP1-dependent heterochromatin (Zhao, 2000).

Neither the chromo domain, the chromo shadow domain, nor the middle region of HP1 is sufficient for DNA binding, even when they are mixed together. The overall negative surface charge of the chromo domain of the mouse HP1 family protein MOD1 makes it unlikely that the chromo domain alone binds to DNA or RNA. The findings presented here are consistent with this prediction. It seems that in order to bind DNA, the intact HP1 protein is absolutely required. A straightforward model suggests that several different regions of HP1 are required for DNA binding. The binding activity is weak for each individual region and can not be detected in gel shift assay, and the regions have to be in a specific spatial relationship in order to bind to DNA with high affinity. By this model, each chromo domain may contribute to DNA binding, and both of them have to be present in the same molecule in order to 'clip' HP1 to DNA. Another possibility is that HP1 self-association is required for DNA binding: if the self-association and DNA binding domains in HP1 belong to different regions of the protein, both would have to be present to see DNA binding activity. It cannot be ruled out, however, that truncation of full-length HP1 causes misfolding of a single, discrete DNA binding domain (Zhao, 2000).

The archial chromosomal protein Sac7d possesses a folded structure similar to the chromo domain fold. Sac7d protein binds the minor groove of DNA as a monomer through hydrophobic interactions between bases and sugars in the DNA and a triple-stranded-sheet in Sac7d. If HP1 chromo and chromo shadow domains were to bind weakly to DNA in an analogous fashion, the inability of each domain to bind separately could be rationalized as a requirement for cooperativity conferred by tandem chromo domains. Cooperative binding would also be enhanced through dimerization. Dimerization of a mouse chromo shadow domain occurs through the C-terminal-helix of the chromo domain fold. If Drosophila HP1 dimerizes by a similar mechanism, this would leave the triple-stranded-sheet of each dimer partner free to interact with DNA. Such interactions could explain both the delocalized DNA binding sites in HP1 and the cooperativity observe for DNA binding in solution. In vivo, such cooperativity could facilitate the condensation of nucleosomal DNA in heterochromatin (Zhao, 2000 and references therein).

The data presented here demonstrate that Drosophila HP1 binds DNA and nucleosomes directly. A human HP1 family protein, HP1hs, has also been shown to bind to DNA directly. HP1 is implicated as a fundamental component of the protein-DNA complex in heterochromatin. This inference is consistent with cytological evidence that cis-spreading of heterochromatic silencing in chromosome rearrangements is accompanied by the spreading of HP1-associated chromatin into the silenced chromosomal region (Zhao, 2000 and references therein).

A recent estimate of HP1 concentration in Drosophila suggests that there are roughly three molecules of HP1 per nucleosome within polytene heterochromatin. While this estimate is crude, it is consistent with approximately stoichiometric binding of HP1 dimers to nucleosomes throughout heterochromatin. Such binding could interfere with nucleosome sliding and transient nucleosome displacement required for gene activation or with enhancer-promoter communication in euchromatic genes. This model can account for the reported regularization of nucleosome arrays in a heat shock gene silenced by heterochromatin (Zhao, 2000).