|Home | About | Journals | Submit | Contact Us | Français|
Two chromatin compartments are present in most mammalian cells; the first contains primarily euchromatic, early replicating chromatin and the second, primarily late-replicating heterochromatin, which is the subject of this review. Heterochromatin is concentrated in three intranuclear regions: the nuclear periphery, the perinucleolar space and in pericentromeric bodies. We review recent evidence demonstrating that the heterochromatic compartment is critically involved in global nuclear organization and the maintenance of genome stability, and discuss models regarding how this compartment is formed and maintained. We also evaluate our understanding of how heterochromatic sequences (herein named heterochromatic associated regions (HADs)) might be tethered within these regions and review experiments that reveal the stochastic nature of individual HAD positioning within the compartment. These investigations suggest a substantial level of functional redundancy within the heterochromatic compartment.
A readily evident heterochromatic compartment is present in most differentiated cell nuclei. Chromatin in this compartment is generally more compacted than euchromatin, and although many sequences may be transcribed in heterochromatin, this transcription usually maintains gene silencing, rather than giving rise to protein-coding mRNAs (see [1–3] for reviews). Heterochromatin is commonly described as constitutive (primarily silenced repeat sequences) or facultative (unexpressed developmentally specific genes), but both types of heterochromatin are usually concentrated together in the heterochromatic compartment [2,4]. Spatially, heterochromatin in the compartment is distributed within three intranuclear regions: the nuclear periphery (PH), the perinucleolar region (PNH) and pericentromeric bodies (PCH), (Figure 1A–1B) [2,5]. The size and composition of the compartment often changes dramatically during cell differentiation. For example, the compaction and amount of heterochromatin increases significantly during erythroid differentiation as non-erythroid genes are silenced (Figure 1C–1D). In some vertebrate classes this culminates in an erythrocyte nucleus in which the entire genome is highly condensed chromatin. In contrast, naïve stem cells and very early embryos have relatively little heterochromatin (Figure 1E–1F)[7–9]. In this review, we explore recent advances, primarily in mammalian cells, that help to define the role of the heterochromatic compartment in nuclear organization and function. We focus on the composition of the heterochromatin associated domains (HADs), the effect of their sequestration on nuclear organization and function, and finally, on the redundancy of the heterochromatic compartment and how this redundancy impacts studies on nuclear organization and function.
The major chromatin marks that define HADs are the constitutive heterochromatin marks, 5-mC, H3K9me and H4K20me (in PCH especially) and the facultative mark, H3K27me3, although there are other marks also involved (see [3,4] for reviews). Condensation and compaction is mediated by the linker histone H1, the H3K9me binding heterochromatin protein 1 (HP1α and β), and the DNA and histone methyl transferases (DNMTs and HMTs). A number of other heterochromatin-binding proteins (HBPs) and histone variants are also involved, but their direct roles in compaction remain less well defined (see [2,3,10] for reviews).
Highly and moderately repetitive HADs, such as centromeric and pericentromeric sequences (satellite DNA), telomeric sequences, endogenous retrotransposons and a portion of the rDNA repeats are commonly constitutively silenced in differentiated cells. Unexpressed developmentally specific HADs are usually subject to facultative silencing via polycomb-mediated H3K27 methylation. If heterochromatic marks are removed experimentally [11,12], or are lost during aging [13,14] or cancer [15,16], inappropriate gene expression (e.g. of retrotransposons [17,18]) and spontaneous recombination between newly exposed repeat sequences (especially rDNA [19,20]) result in genome instability and eventual cell death. It is of no surprise then that most organisms possess redundant silencing pathways to maintain heterochromatin features and thus genome stability [1,2,17,21].
Given the importance of heterochromatin to genome stability, it had been difficult to understand how heterochromatin-poor early embryos and naïve stem cells accommodate active retrotransposons, for example, but recent work suggests that piwi-based RNAi mechanisms [22,23] and/or the deposition of the histone variant, H3.3, and specific histone chaperones [24–26], silences retrotransposons until heterochromatic marks are deposited. Notably, however, some early retrotransposon activity, such as that resulting in the presence of the long non-coding RNA (ncRNA), HERV, in primate cells, is crucial in early embryos and naïve stem cells [27,28]. In aging or cancerous cells, where piRNAs are not prevalent, retrotransposon activity can wreak havoc on chromatin organization [15,18]. This does not preclude the fact, however, that many repeats, such as those in PCH, are transcribed at low levels in differentiated cells and some of these ncRNAs complex with HBPs and serve as adaptors to help target their binding (e.g. [3,29,30]).
An intriguing aspect of the transition from pluripotency involves the rDNA repeats. In naïve mouse embryonic stem cells (mESCs), all or almost all rDNA cassettes are active, while in differentiated cells, approximately half are inactive and condensed into heterochromatin . The silencing is specifically mediated by pRNA, which is transcribed from the rDNA promoter region and binds TIP5 within the nucleolar remodeling complex . Interestingly, this rDNA heterochromatization has recently been shown to be a necessary step for exiting pluripotency. If pRNA is removed, cells remain pluripotent, with very low levels of heterochromatin . Earlier work had shown that knockdown of the pRNA partner, TIP5, disrupts rDNA heterochromatization in fibroblasts, resulting in disorganization and unfolding of PNH, including the non-rDNA sequences localized there [32,34], and that deletion of rDNA cassettes decreases the levels of global heterochromatin . Notably, knockdown of the linker histone H1 has recently been shown to disrupt nucleolar structure as well. Unexpectedly, H1 not only interacted with silencing enzymes in the nucleolus but also with many proteins involved in early ribosomal RNA transcription and processing, suggesting H1 may be involved in maintaining the balance between active and inactive rDNA loci . Taken together, these findings further buttress an attractive model which postulates that silenced rDNA seeds the formation of the heterochromatic compartment [37–39]. Consistent with this model, PCH accumulates around pre-nucleolar bodies at the mid-zygote stage .
Although silenced rDNA may well seed heterochromatin formation, heterochromatin accumulates not only in the PNC, but also the PH and in spatially distinct PCH bodies in some cells. There are two main mechanisms at work in this compartmentalization: first, self-association of repetitive chromatin guided by biophysical principles, and second, tethering of these chromatin domains within the heterochromatic compartment.
Assembly of nuclear compartments within the membrane-free nucleoplasm can be modeled using self-organization principles based on molecular self-association, volume exclusion and phase partitioning [41–44] and in the case of chromatin, polymer physics and the behavior of fractal globules [45–47]. Simply put, early in G1, repetitive sequences, including rDNA, centromeric and telomeric sequences, are present at much higher concentrations than unique sequences and thus encounter one another more often, favoring self-association (the so-called “birds of feather stick together” model . The resulting volume exclusion effects [47–49] drive nuclear organization in what has been termed a dog-on-a-leash mechanism, where strong associations between repetitive sequences cause clustering and adjacent sequences are dragged along on the “leash”. In this way, highly repetitive HADs may act as dominant seeds to probabilistically concentrate any nearby heterochromatin into the PNH, PCH and PH during reorganization after each mitosis. Consistent with a self-assembly model, it has been known for years that the relative compaction of heterochromatin can be reversibly manipulated simply by changes in osmotic strength [51–53].
Crowding and clustering in turn favors concentration of HBPs at their sites of action. Proteins move with fractal kinetics through the compacted chromatin , with intermittent corralling within micro-regions (giving rise to anomalous intranuclear diffusion . This increased residence time increases binding efficiency at a particular site since proteins have time to sample most binding sites within the corral. For example, as mentioned above, HP1α and β have been long known to be important global compactors of heterochromatin via their affinity with H3K9me sites and their propensity for oligomerization, which drives heterochromatic spreading . Oligomerization would be favored in corrals with high monomer concentration.
Peripheral tethering of chromatin (so-called lamin-associated domains or LADs ) requires the presence of heterochromatic-specific histone marks and is thought to occur through both direct and indirect interactions with the lamina [4,56]. Somewhat surprisingly, the lamins themselves are not required for the peripheral localization of LADs in mESCs [57,58], but their presence is necessary for peripheral heterochromatin localization in differentiated cells . For example, the loss of lamin B1 in aging cells disrupts heterochromatin organization [59,60] and terminally differentiated retinal cells display an “inside-out” heterochromatin arrangement (where heterochromatin occupies the center of the nucleus and euchromatin and nucleoli the periphery) primarily because neither lamin A/C nor Lamin B Receptor (LBR) is expressed in these cells . Both of these latter proteins have been implicated in heterochromatin tethering: the transmembrane LBR interacts with methylated histones and HP1 and lamin A/C is associated with LADs in murine embryonic fibroblasts and pro-B cells . Notably, Kind et al.  have shown that lamin A/C (and Barrier to Autointegration Factor (BAF)) associates with both LADs and nucleolar-associated-domains (NADs) in human fibrosarcoma cells. This indicates that lamin A and BAF are not PH specific tethers in human and it will be interesting to determine whether this is true in mouse as well. On the other hand, like LBR, the inner nuclear membrane protein LAP2β has so far only been implicated in PH tethering , and it, like BAF, binds specific regions of chromatin during mitosis . Other potentially specific PH tethering candidates include nuclear envelope transmembrane proteins and the proline-rich protein 14 . Nucleoli were not labeled in the latter study, however. Thus, even though much work has been done, only a few proteins that tether heterochromatin specifically to the PH have been identified.
Similarly, few candidates for direct binding of NADs  specifically to nucleolar components have been identified. One encouraging finding is that a complex of CTCF and a nucleoplasmin-like protein has been shown to mediate binding of centromeres to the nucleolar protein nucleolin (Modulo) in Drosophila . CTCF also was shown in earlier studies to mediate binding of a transgene to the nucleolus, perhaps through nucleophosmin (aka B23) . One reason few specific tethers have been identified could be that many repetitive sequences are not specifically tethered to one region within the compartment, but rather are able to move among the three regions, as discussed below in “Redundancy of the heterochromatic compartment”. Additionally, most past studies have been focused on identification of protein tethers, and it is only very recently that studies have begun to identify ncRNAs that act as specific adaptors for tethering. A clear example of this is the long ncRNA Firre. This ncRNA is transcribed from a macrosatellite repeat on the murine X chromosome and can organize 3D chromatin topology in trans . Intriguingly, Firre, and also a previously identified CTCF-site-associated long ncRNA, Dxz4, have now been implicated as important mediators in the interaction between CTCF sites on the inactive X with the PNH, maybe in a complex with cohesin and nucleophosmin . The fact that these ncRNAs are transcribed from the inactive X and then employed in tethering it within the heterochromatic compartment (and maintaining its silencing) suggests this may be a more general mechanism for tethering. pRNA, active in the silencing of rDNA discussed under “Formation of the heterochromatin compartment”, is another example of this mechanism, as it is transcribed from the rDNA promotor and is a necessary component of the nucleolar remodeling complex. Much exciting work remains to be done in this area.
A number of experiments have now shown that many sequences within the heterochromatic compartment can be stochastically shuffled between the PN, PNH and PCH (Figure 2). For example, “mother” LADs become distributed throughout the heterochromatic compartment in daughter cells without affecting cell function (Figure 2A, ), a finding supported by earlier photoactivation and photobleaching experiments . The distribution patterns are different from mother to daughter and between daughter nuclei, suggesting that location of a given LAD within a particular heterochromatic region is not necessary. In further support of this interpretation, late-replicating (type B compartment, primarily silent) DNA sequences shown to interact with one another via Hi-C  can be found, by using in situ hybridization, in either the PH, PNH or PCH in different lymphoblastoid cells within the same culture (Figure 2B) . In many cases these sequences correspond to LADs, supporting the live cell relocalization experiments. Multiple laboratories have also now shown that many LADs overlap with NAD sequences (e.g. [62,69,73,74]), and in situ hybridization studies have long shown that PCH can overlap with PH or PNH in many tissue types. Taken together, these observations are consistent with a self-assembly model, where inactive sequences may be sequestered into any region of the heterochromatin compartment and this may suffice to induce/maintain silencing (Figure 2C)[37,72].
It has not been clear whether this is true for all HADs, however, since, for example, particular centromeric sequences have classically been observed to cluster preferentially in either the PNH or PH, at least in certain cell types [2,5]. Interestingly, a recent paper goes some distance toward shedding light on this preferential localization, at least in human lymphoblastoid cells . These authors showed that PH centromeric sequences in non-proliferating cells became more closely associated with the PNC upon proliferation. The location of the centromeric-bearing chromosome territories did not change substantially, but rather the position of the centromeric sequence within each territory was altered so that it was now closer to nucleoli than the periphery. Thus, at least in this system, it appears centromeric sequences can move between the PH and PNC also, just as LADs and NADs do. It would be interesting to determine whether lamin A/lamin B1 ratios increase when the cells proceed toward proliferation, since Kind et al. (discussed above ) have shown that changing these ratios can affect the PNH/PH distribution of HADs.
Another aspect of preferential location to either the PH or PNH is exemplified by experiments performed in B cells from mice of different strains. In the standard laboratory mouse strain, C57BL/6J, chromosomes 12 and 15 possess acrocentric rDNA arrays (aka nucleolar organizing centers or NORs) and thus localize preferentially near nucleoli. However, in other laboratory (and wild) mouse strains, one or both of these chromosomes do not contain an NOR (presumably due to the increased recombination rate between mouse acrocentric sequences) and in these strains, chromosomal territories 12 and 15 associate more frequently with the periphery (Figure 2D) . It is unlikely that lamin A/lamin B1 ratios affect the proximity of an NOR bearing chromosome to a nucleolus, as can be proposed for the centromeric sequences above. One model that explains both these sets of results supposes that association with the PH is a default state, while association with the PNH requires an active tether. The nucleolar tether could be nucleophosmin, nucleolin, lamin A, or BAF, perhaps associated with adaptor RNAs specific to the sequence tethered, or in other cases, an NOR.
This gives rise to a model in which HADs are sequestered into the PH, PNH and PCH in a largely stochastic way (Figure 3). The compartmentalization is driven by biophysical forces, and enforced by tethering proteins and ncRNAs adaptors transcribed from the HADs. As exemplified by lamin A, the model supposes that some tethering proteins are present at multiple locations within the heterochromatic compartment, enforcing probabilistic distribution. We also speculate that “promiscuous” transcription from different regions of a particular HAD would yield a constellation of adaptor ncRNAs that could promote binding to different tethering proteins within the PH, PNH or PCH. Affinity for a particular tethering protein could vary, depending on which transcribed sequences were present in higher quantities at the time. This mechanism also would enforce stochasticity.
In addition to the probabilistic distribution of HADs, some HADs still associate preferentially with a particular region within the heterochromatic compartment. This finding is more solid for PNH than PH, as in many PH studies, NADs are not characterized. Thus, we only include specific PNH- tethering proteins (blue) and ncRNAs (Firre) in this model. Nevertheless, it is early days in this field, so we may find that ncRNAs serve as specific adaptors at the lamina also (“?” in Figure 3). It should be stressed, however, that even in the cases where nucleolar tethering is observed, localizations are probabilistic, and a sizable fraction of cells still show PH localization.
In general, current data indicates that functional cells can contain a given HAD in the PH, PNC or PCH, suggesting that the three regions within the heterochromatic compartment are redundant with respect to function. This redundancy results in a stochastic distribution of HADs throughout a significant fraction of the nuclear volume and helps explain why, although the spatial positioning of chromosome territories in a given cell type is not random, it is probabilistic rather than fixed [70,77]. It will be important to better define the degree and nature of redundancy in future experiments, which should include characterization of HACs in the PH, PNC and PCH. Live cell experiments in primary cell lines maintained in defined medium are the Holy Grail, but further technical advances are necessary before these types of experiments become standard.
We thank Thoru Pederson and Tobias Ragoczy for helpful comments on the manuscript. We apologize to authors whose work we could not cite given space considerations in this minireview. The authors were supported by NIH R01 HL65440 and NIH R37 DK44746.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.