Evolving Models of Heterochromatin: from Foci to Liquid Droplets

Two recent papers (Larson et al., 2017Larson A.G. Elnatan D. Keenen M.M. Trnka M.J. Johnston J.B. Burlingame A.L. Agard D.A. Redding S. Narlikar G.J. Nature. 2017; 547: 236-240Crossref PubMed Scopus (875) Google Scholar, Strom et al., 2017Strom A.R. Emelyanov A.V. Mir M. Fyodorov D.V. Darzacq X. Karpen G.H. Nature. 2017; 547: 241-245Crossref PubMed Scopus (934) Google Scholar) in Nature propose that heterochromatic domains are organized into phase-separated liquid compartments. Here we highlight the main findings that support the liquid-like nature of HP1 domains and discuss their functional implications in gene silencing and genome organization. Two recent papers (Larson et al., 2017Larson A.G. Elnatan D. Keenen M.M. Trnka M.J. Johnston J.B. Burlingame A.L. Agard D.A. Redding S. Narlikar G.J. Nature. 2017; 547: 236-240Crossref PubMed Scopus (875) Google Scholar, Strom et al., 2017Strom A.R. Emelyanov A.V. Mir M. Fyodorov D.V. Darzacq X. Karpen G.H. Nature. 2017; 547: 241-245Crossref PubMed Scopus (934) Google Scholar) in Nature propose that heterochromatic domains are organized into phase-separated liquid compartments. Here we highlight the main findings that support the liquid-like nature of HP1 domains and discuss their functional implications in gene silencing and genome organization. Heterochromatin is a conserved feature of eukaryotic chromosomes with important roles in epigenetic regulation of gene expression, maintenance of genome stability, and proper chromosome segregation (Wang et al., 2016Wang J. Jia S.T. Jia S. Trends Genet. 2016; 32: 284-294Abstract Full Text Full Text PDF PubMed Scopus (84) Google Scholar). Heterochromatic domains, which are morphologically distinct and transcriptionally silent, comprise nearly half of the genome in some eukaryotes. These domains are decorated by histone H3 lysine 9 methylation (H3K9me), which provides binding sites for the highly conserved heterochromatin protein 1 (HP1) regulators. HP1 proteins associate with H3K9me using their N-terminal chromodomain and self-associate through their C-terminal chromo shadow domain. Their association with homologs of the human histone H3K9 methyltransferase SUV39H1, coupled to sequential cycles of modification and binding, results in spreading of H3K9 methylation and HP1 along nucleosomal DNA. HP1 serves as a platform for recruitment of other factors, leading to formation of domains that silence DNA transactions such as recombination and transcription. This silencing is thought to result from specific changes in histone post-translational modifications, such as deacetylation, and/or steric hindrance, but how HP1 domains cluster together to form nuclear foci and how silencing is achieved are not fully understood. In two recent papers in Nature, Strom et al., 2017Strom A.R. Emelyanov A.V. Mir M. Fyodorov D.V. Darzacq X. Karpen G.H. Nature. 2017; 547: 241-245Crossref PubMed Scopus (934) Google Scholar and Larson et al., 2017Larson A.G. Elnatan D. Keenen M.M. Trnka M.J. Johnston J.B. Burlingame A.L. Agard D.A. Redding S. Narlikar G.J. Nature. 2017; 547: 236-240Crossref PubMed Scopus (875) Google Scholar propose that heterochromatin formation involves phase separation of HP1-bound chromatin domains into liquid-like foci with distinct physical properties that are critical for silencing. The two studies show that purified Drosophila HP1a and human HP1α proteins undergo phase separation in vitro and spontaneously form droplets. The so-called intrinsically disordered regions (IDRs) in proteins drive the assembly of liquid droplets, which also often involves interactions with RNA (Hyman et al., 2014Hyman A.A. Weber C.A. Jülicher F. Annu. Rev. Cell Dev. Biol. 2014; 30: 39-58Crossref PubMed Scopus (1451) Google Scholar). Similarly, the N-terminal and hinge domains of HP1a and HP1α proteins contain IDRs, which may explain their ability to form liquid droplets. Strom and coworkers also analyzed early stages of heterochromatin formation in Drosophila embryos and observed that HP1a fused with GFP forms highly spherical foci that grow and frequently fuse together. Although such HP1 foci have been observed for a long time (Cheutin et al., 2003Cheutin T. McNairn A.J. Jenuwein T. Gilbert D.M. Singh P.B. Misteli T. Science. 2003; 299: 721-725Crossref PubMed Scopus (486) Google Scholar, Kellum et al., 1995Kellum R. Raff J.W. Alberts B.M. J. Cell Sci. 1995; 108: 1407-1418PubMed Google Scholar), this study shows that they display key features of phase-separated compartments (Strom et al., 2017Strom A.R. Emelyanov A.V. Mir M. Fyodorov D.V. Darzacq X. Karpen G.H. Nature. 2017; 547: 241-245Crossref PubMed Scopus (934) Google Scholar). In addition to their spherical shape and their ability to fuse, these liquid-like foci are sensitive to the presence of an aliphatic alcohol that disrupts weak hydrophobic interactions. Moreover, by measuring the diffusion rates of a yellow fluorescent protein and fluorescence fluctuation correlations of HP1a and H2A fused to fluorescent proteins, Strom et al., 2017Strom A.R. Emelyanov A.V. Mir M. Fyodorov D.V. Darzacq X. Karpen G.H. Nature. 2017; 547: 241-245Crossref PubMed Scopus (934) Google Scholar conclude that exclusion at the surface of HP1a droplets, rather than physical compaction of chromatin, underlies limited access to heterochromatin domains. In the second study, Larson and colleagues show that the human HP1α protein, but not HP1β and HP1γ paralogs, can form phase-separated droplets (Larson et al., 2017Larson A.G. Elnatan D. Keenen M.M. Trnka M.J. Johnston J.B. Burlingame A.L. Agard D.A. Redding S. Narlikar G.J. Nature. 2017; 547: 236-240Crossref PubMed Scopus (875) Google Scholar). Unlike the case with the Drosophila HP1a, phosphorylation of HP1α N-terminal extension domain or binding to DNA was required for the formation of phase-separated droplets. Using a DNA curtain assay, Larson et al., 2017Larson A.G. Elnatan D. Keenen M.M. Trnka M.J. Johnston J.B. Burlingame A.L. Agard D.A. Redding S. Narlikar G.J. Nature. 2017; 547: 236-240Crossref PubMed Scopus (875) Google Scholar observed that both wild-type unphosphorylated HP1α and HP1α phosphorylated on the N-terminal extension domain formed a single or multiple fluorescent puncta, respectively, suggesting that phosphorylation may facilitate droplet formation in this assay, which uses DNA rather than H3K9-methylated nucleosomes on DNA curtains. Furthermore, Larson et al., 2017Larson A.G. Elnatan D. Keenen M.M. Trnka M.J. Johnston J.B. Burlingame A.L. Agard D.A. Redding S. Narlikar G.J. Nature. 2017; 547: 236-240Crossref PubMed Scopus (875) Google Scholar observed that nucleosomes and DNA, but not transcription factor TFIIB, preferentially enter the HP1α liquid droplets, supporting the idea that the droplets exclude factors involved in transcription. Previous studies have documented transient binding and dynamic exchange of HP1 proteins in stable heterochromatin domains, and labeling of telomeric and satellite repeat DNA by CRISPR/Cas9-based strategies has revealed the dynamic movement of these silent regions in human cells (Chen et al., 2013Chen B. Gilbert L.A. Cimini B.A. Schnitzbauer J. Zhang W. Li G.W. Park J. Blackburn E.H. Weissman J.S. Qi L.S. Huang B. Cell. 2013; 155: 1479-1491Abstract Full Text Full Text PDF PubMed Scopus (1259) Google Scholar, Cheutin et al., 2003Cheutin T. McNairn A.J. Jenuwein T. Gilbert D.M. Singh P.B. Misteli T. Science. 2003; 299: 721-725Crossref PubMed Scopus (486) Google Scholar). The liquid-like behavior of heterochromatic domains proposed by Strom et al., 2017Strom A.R. Emelyanov A.V. Mir M. Fyodorov D.V. Darzacq X. Karpen G.H. Nature. 2017; 547: 241-245Crossref PubMed Scopus (934) Google Scholar and Larson et al., 2017Larson A.G. Elnatan D. Keenen M.M. Trnka M.J. Johnston J.B. Burlingame A.L. Agard D.A. Redding S. Narlikar G.J. Nature. 2017; 547: 236-240Crossref PubMed Scopus (875) Google Scholar provides an alternative explanation for the dynamically stable nature of heterochromatin. Moreover, liquid fusion properties of heterochromatin domains may provide a mechanism for long-range chromosome interactions by facilitating looping of HP1-associated chromosome domains. However, in contrast to the liquid-like properties described in the present studies, Cheutin et al., 2003Cheutin T. McNairn A.J. Jenuwein T. Gilbert D.M. Singh P.B. Misteli T. Science. 2003; 299: 721-725Crossref PubMed Scopus (486) Google Scholar observed that the human HP1α, HP1β, and HP1γ foci were positionally stable and showed very little displacement over periods ranging from several minutes to 2 hr in Chinese hamster ovarian (CHO) and other cells. The extent to which HP1 droplets occur in various differentiated cell types therefore remains to be determined. Strom et al., 2017Strom A.R. Emelyanov A.V. Mir M. Fyodorov D.V. Darzacq X. Karpen G.H. Nature. 2017; 547: 241-245Crossref PubMed Scopus (934) Google Scholar and Larson et al., 2017Larson A.G. Elnatan D. Keenen M.M. Trnka M.J. Johnston J.B. Burlingame A.L. Agard D.A. Redding S. Narlikar G.J. Nature. 2017; 547: 236-240Crossref PubMed Scopus (875) Google Scholar propose exciting models for the formation and function of heterochromatin domains based on their findings. The two studies propose that HP1 droplet formation depends on the self-association of multiple molecules of HP1 facilitated by local increases in concentration due to the association of HP1 with chromatin and, possibly, to a lesser extent, by a chromatin-independent mechanism (Figure 1). Liquid-like fusion of HP1a droplets then leads to the formation of more “compact” heterochromatic domains that exclude other types of molecules. In this regard, a recent study of stress granules, mRNA-protein complexes that form liquid droplets in yeast and mammalian cells, revealed that their formation in cells involves core interactions that precede the formation of liquid droplets (Wheeler et al., 2016Wheeler J.R. Matheny T. Jain S. Abrisch R. Parker R. eLife. 2016; 5: e18413Crossref PubMed Scopus (380) Google Scholar). However, in contrast to HP1 foci, human and yeast stress granules are not sensitive to aliphatic alcohols, demonstrating that sensitivity to such environmental effectors could not be used as a criteria for liquid droplets in vivo (Wheeler et al., 2016Wheeler J.R. Matheny T. Jain S. Abrisch R. Parker R. eLife. 2016; 5: e18413Crossref PubMed Scopus (380) Google Scholar). Further studies are needed to determine how HP1 droplets are first formed and whether the observed physical properties of HP1 proteins are important for heterochromatin functions such as transcriptional gene silencing or protection of heterochromatin from large fluctuations in local HP1 concentration (Figure 1). Such studies will require the generation of mutations that specifically disrupt the ability of HP1 proteins to form liquid droplets. These findings (Strom et al., 2017Strom A.R. Emelyanov A.V. Mir M. Fyodorov D.V. Darzacq X. Karpen G.H. Nature. 2017; 547: 241-245Crossref PubMed Scopus (934) Google Scholar, Larson et al., 2017Larson A.G. Elnatan D. Keenen M.M. Trnka M.J. Johnston J.B. Burlingame A.L. Agard D.A. Redding S. Narlikar G.J. Nature. 2017; 547: 236-240Crossref PubMed Scopus (875) Google Scholar) open new areas of investigation about the function of liquid-like HP1 compartments in gene silencing and genome organization. Future studies are likely to reveal whether inhibiting the formation or promoting disassembly of heterochromatic droplets affects gene silencing and genome stability. Since the ability of HP1a/α to form liquid droplets is at least partially attributed to the less conserved, unstructured regions of the protein (N and C termini, hinge domain), we might expect to discover recently evolved ancillary functions for HP1 droplets that are unrelated to core heterochromatin functions. Cooperative interactions between proteins and nucleic acids leading to phase separation have been previously suggested to mediate the formation of other membrane-less cellular compartments, such as the P granules, nucleoli, and stress granules, and play a critical role in the function of receptor-mediated signaling pathways (Hyman et al., 2014Hyman A.A. Weber C.A. Jülicher F. Annu. Rev. Cell Dev. Biol. 2014; 30: 39-58Crossref PubMed Scopus (1451) Google Scholar). These studies all suggest new functions for phase separation and the formation of membrane-less compartments in regulation of specific biological reactions. As noted above, what this potentially exciting area now needs are strategies that can test the functional importance of phase separation phenomena in cells. Other explanations for the function of cellular bodies or foci, which are not mutually exclusive with phase separation, also need to be considered. Pioneering studies by Gasser and colleagues on the organization of heterochromatin in yeast suggested that the foci promote more efficient heterochromatin assembly by creating areas in which the concentration of heterochromatin building blocks is high due to multivalent interactions (Gasser et al., 2004Gasser S.M. Hediger F. Taddei A. Neumann F.R. Gartenberg M.R. Cold Spring Harb. Symp. Quant. Biol. 2004; 69: 327-337Crossref PubMed Scopus (35) Google Scholar). Thus, promoting efficient and specific assembly, rather than a downstream effect on silencing, may be the primary function of heterochromatic foci.