Folded genome as a platform for the functional compartmentalization of the eukaryotic cell nucleus

In a number of recent studies a tight interconnection between the spatial organization of the eukaryotic genome and its functioning has been demonstrated. Moreover, it is becoming evident that the folded DNA by itself constitutes an important, if not the key, factor supporting the internal nuclear organization. In this review, we will discuss the current state of chromatin research with the special attention focused on chromosome territories, chromatin folding and dynamics, chromatin domains, transcription and replication factories. Based on this analysis we will show how interphase chromosomes define the assembly of different nuclear compartments and underlie the spatial compartmentalization of the cell nucleus.

Introduction.For many years, the eukaryotic cell nucleus was considered as a test tube where different processes related to the genome activity were proceeding in solution.Now it is clear that this point of view is very far from reality.It has been demonstrated that the nucleus is subdivided into a number of functional compartments adapted to perform various functional processes [1][2][3][4].The list of characterized nuclear compartments includes replication and transcription factories, splicing speckles, Cajal bodies, PML bodies, polycomb bodies and is constantly expanding.As a matter of fact, staining the nuclei with antibodies raised against a number of different proteins reveals some kind of foci, which argues for the existence of numerous functional compartments within the cell nucleus.Taking into account all these observations one may pose a question as to what constitutes a structural basis for the assembly of various nuclear compartments.It has been proposed that all functional compart-ments are assembled on the nuclear matrix [5,6].Indeed, replication and transcription foci as well as other nuclear compartments could be observed on isolated nuclear matrices after removal of a bulk of chromatin [6][7][8][9].
However, the nature of the nuclear matrix itself remains unclear.Numerous attempts to identify and characterize a proteinous network within the cell nucleus (i.e. something similar to the cytoskeleton) resulted in controversial observations and conclusions [10][11][12].Presently, most of the scientists agree that the so-called internal nuclear matrix does not exist in living cell (for a review see [9,13]).
The nuclear lamina appears to be the only skeletal element of the cell nucleus.However, most of the functional compartments are located far from the nuclear lamina inside nucleus.Thus, nuclear lamina cannot, at least directly, support positioning of nuclear compartments.Here we propose that folded DNA itself fulfills the scaffolding function for the nuclear compartments.

Spatial organization of interphase chromosomes.
Chromosomes become visible in mitosis.However, the fate of chromosomes in interphase remained enigmatic until chromosome-specific probes and confocal microscopy were developed.Using FISH with chromosome-specific probes followed by the analysis under confocal microscope, Cremer and collaborators demonstrated that, in interphase, individual chromosomes occupy relatively compact non-overlapping regions that were named «chromosomal territories» [14,15].Chromosomal territories appeared to be separated by some space that was relatively free of chromatin.This space was termed «InterChromatin Domain», ICD, and was proposed to serve for the transport of different compounds across nuclei [14].Later it was found that the interchromatin domain spanned also chromosomal territories so that internal regions were also easily accessible for different compounds including enzymes involved in DNA and RNA metabolism [3,15,16].The interchromatin domain also allows fast transport of RNA (RNP particles) from transcription factories to the nuclear periphery and finally to the cytoplasm [17].High-resolution microscopic analysis performed using combined chromatin staining and immunostaining of different nuclear compartments demonstrated that splicing speckles, PML bodies and several other functional compartments are situated in the interchromatin domain [18].It will thus be correct to say that the specificity of interphase chromosome folding delimits the portion of the nuclear space that can be used for the assembly of the above-mentioned compartments.It should be noted that chromosome territories are not randomly positioned within the nucleus.It has been mentioned that homologous chromosomes are positioned far from each other.Gene-dense chromosomes tend to be located in the central part of the nucleus and gene-poor chromosomes tend to the nuclear periphery [14,[19][20][21][22].If chromosome possesses generich and gene-poor regions, it acquires such a configuration that gene-rich regions are located closer to the nuclear center and gene-poor regions -closer to the nuclear periphery [23].One particular example is a centromeric region that is gene-poor and is located in the part of the chromosomal territory that is closer to the nuclear periphery [24].The modern model of the interphase chromosome organization is based on the results obtained using the so-called C-methods [25,26].Of particular importance are the data obtained using the Hi-C protocol which permits to study general principles of chromatin folding in interphase chromosomes [27].Hi-C analysis has demonstrated that chromosomes are partitioned into topologically-associated domains (TADs) of various sizes (from several hundred Kb to several Mb) [28][29][30].This organization is characterized by high frequency of intra-TAD contacts of remote genomic elements and low frequency of inter-TAD contacts of remote genomic elements.Modelling experiments suggest that a TAD can be considered as a globule.However, the path of the chromatin fiber within the globule is not clear.There are some data in favor of the fractal globule and some data in favor of the molten globule [31,32].The situation is complicated by the fact that the actual mode of chromatin packaging within the cell nucleus is likely to differ drastically from the one predicted based on in vitro experiments.For years, it has been generally accepted that chromatin is organized in a more or less regular hierarchical set of folded structures starting from the 10 nm fiber («beads on the string») that is further packed in the 30 nm fiber arranged in ~100 Kb looped domains [33,34].Recent studies performed using electron cryotomography, and electron spectroscopic imaging (ESI) combined with electron tomography provided no evidence for the existence of the 30 nm fibers within the non-disturbed cell nucleus [35][36][37].Both euchromatic and heterochromatic regions appeared to be composed of tightly packed 10 nm fibers [36].This less regular organization should allow more flexible positioning of nucleosomes and nucleosomal fibers suggesting there is a way to suppress interactions between neighboring fibers.Here the modification of histones and the removal of architectural proteins may play an essential role.Based on the hierarchical model of chromatin folding, it was reasonable to assume that progressive expansion of chromatin domains occurs along the chromatin chain.Within a globule composed of irregularly (stochastically) folded and closely spaced 10 nm chromatin fibers, the distances between nucleosomes located on the same and neighboring 10 nm fibers are likely to be comparable if not the same [35][36][37].Correspondingly, the expansion of a chromatin domain mediated by the progressive modification of histones is to be considered as a process that occurs in all possible directions within the 3D nuclear space.Consequently, a three-dimensional, at first appro-ximation, globular chromatin domain will be formed.This domain may be mosaic i. e. be composed of segments of chromosomes that do not necessarily neighbor each other on a linear DNA chain.In case of spreading active signals, the loosely packed and thus expanded 3D domain will easily accommodate a transcription factory or other functional compartments (reconfiguration of chromatin occurs also in connection with DNA repair).
It is necessary to underline that all the above-mentioned organization is highly dynamic.The local mobility (range of fluctuations) of nucleosomes in nuclear chromatin was estimated experimentally and found to constitute ~50 nm/30 ms [38].Higher-order organization of chromosomes is also dynamic as it shows cell to cell variations.Although C-methods give an integral picture based on a statistical analysis, FISH data clearly demonstrate that the mutual positions of different genomic regions in individual cells present in a population are variable [39][40][41].
The same applies to the mutual positions of chromosomal territories [15,24,42].Furthermore, it was demonstrated that in living cells chromosomal territories constantly move [43,44].
The nature of nuclear compartments and the role of folded chromosomes in their positioning.The term «nuclear compartment» is not well defined.One can consider euchromatin and heterochromatin as functional compartments.Respectively, perilamellar, perinucleolar and pericentromeric regions should be considered as special compartments because heterochromatin is located preferentially in these areas.It is hardly necessary to prove that the folded genome plays an essential, if not crucial, role in the organization of these compartments.Indeed, it has been shown that the localization of chromosomal regions in respect to the nuclear lamina is nonrandom.Segments of the interphase chromosomes that are located in vicinity of the nuclear lamina and nucleolus (Lamina Associated Domains LADs [45,46] and Nucleollar Associated Domains, NADs [47]) have been identified and characterized.Genomic data demonstrate that LADs and NADs significantly overlap.It looks likely that in each cell the pattern of association of inactive chromatin regions with either the nuclear lamina or nucleolus is established de novo after mitosis and that the partitioning of inactive genomic domains between the perinuclear (perilamellar) and perinucleolar layers occurs at random [32,47].Based on these observations one may suggest that the assembly of inactive domains is controlled by the internal signals (ex.: profiles of histone modifications) and their positioning in the vicinity of either nuclear lamina or nucleolus occurs in a passive fashion in order to make room for the proper spatial organization of active chromosomal segments.This point of view does not contradict the observations demonstrating that the forced repositioning of a gene to the nuclear periphery causes inactivation of this gene [48][49][50][51].This is due to the presence of heterochromatin in this region rather than the direct interaction with the nuclear lamina.All chromatin compartments are highly dynamic.Heterochromatin is constantly disassembled and assembled.Indeed, the exchange rate of structural components of heterochromatin (HP1 and PcG proteins) constitutes just seconds [52,53].The equilibrium is, however, moved in the direction of heterochromatin assembly due to the high local concentration of histone deacetylases and other proteins involved in the heterochromatin assembly.These enzymes will be attracted to the area just by the high concentration of nucleosomes bearing relatively stable histone modifications typical of heterochromatin (ex.: H3K9-trimet).It is not surprising that any chromatin fragment placed in such a locality will become heterochromatic.The conversion of euchromatin to heterochromatin in this case will simply follow the principle of spatial expansion of a heterochromatin domain explained in the previous section.
The most important functional compartments of the genome are transcription and replication factories [54][55][56] as they are directly involved in the transfer of genetic information.Replication factories were first observed as replication foci, focal places where DNA synthesis occurred as followed from the analysis of spatial distribution of the places of BUdR incorporation into nascent DNA [57].It was then shown that the same foci could be visualized in G1 cells by immunostaining with antibodies against replication factor A [58].Although this observation may be interpreted in terms of the model postulating that replication factories are special assemblies of replication enzymes that exist in the absence of replication, the simplest explanation is that the replication machinery is assembled at replication origins before the onset of replication.Several replication origins are likely to be located close to each other in the nuclear space.Upon simultaneous activation of these replication origins replication factories containing several replication forks located in close spatial proximity will be formed.Although it has been proposed that mutual positions of replication origins are locked by special links (for example, by association with the nuclear matrix [6,59]) there are not much data supporting this model.We think that the location of several replication origins in spatial proximity may be provided just by the way of chromatin fiber folding.With the assembly of replication forks, the protein complexes attached to DNA become large enough to be gathered together and held in a complex (replication factory) by the depletion attraction force [60][61][62].Different procedures used for the nuclear matrix isolation are likely to stabilize these dynamic assemblies so that they can be observed in chromatin-depleted nuclei [63].The relation of replication clusters to TADs is not clear at the moment.Based on the sizes of replicons (~100 kb on the average [64,65]) one may speculate that TADs correspond to replicon clusters identified by immunostaining [66].Under certain conditions, these clusters can be unfolded into a rosette-like structure.
Transcription factories have gained much attention in resent researches [54,55,67].It has been proposed that they exist in the absence of transcription [68] and that to be transcribed genes should be moved to the preexisting transcription factories [54,55].This may indeed happen in the case of activation of tissue-specific gene transcription which is likely to occur at preexisting transcription factories mediating transcription of house-keeping genes.
However, there is no solid evidence for the existence of transcription factories in the absence of transcription.The conditions used to study this question ensured transcription arrest but not disassembly of elongating Pol II complexes [68].The force that could bring inactive gene to a transcription factory remains enigmatic.We have proposed that the Pol II pre-initiation complex is assembled on a gene located at a distance from the transcription factory [67].Later this gene can be brought to the transcription factory by the depletion attraction force [69,70].The same force may bring together several genes with assembled Pol II pre-initiation complexes.In this way a new transcription factory will be formed [67,71].Recent analysis of the mobility of Pol II in living cells strongly supports the model of dynamic transcription factories [72].If indeed assembly of Pol II complexes into factories is mediated by the depletion attraction mechanism, the main factor determining the specificity of gene association in transcription factories should be their spatial proximity.Although it was reported that in erythroid cells there is a preference of assembly of erythroid-specific genes into common transcription factories [39], the presented data did not strongly support this conclusion.Indeed, although the probability to find two erythroid-specific genes in the same transcription factory was a little higher than expected based on the occasional distribution of transcribed genes between all transcription factories, the probability that three or more erythroid-specific genes shared the same transcription factory was already less than expected based on the random distribution [39].
Other data including our own observations made in the chicken erythroid cells (unpublished) strongly support the supposition that genes located close to each other (at least within the same TAD) tend to share transcription factories.Of course, there are well documented cases when genes from different chromosomal territories share the same transcription factory, for example IgH and c-myc in B cells [40].Apparently, this is due to some peculiarities in the folding of chromatin in both territories.
In most eukaryotic cells, there are two types of chromatin domains -Constitutive (mostly centromeric) heterochromatin organized via H3K9 tri-methylation and recruitment of HP1 [73,74] and facultative heterochromatin organized via H3K27 tri-methylation and recruitment of Polycomb group proteins (PcG) [75].Genomic regions repressed via recruitment of PcG tend to associate to form the so-called Polycomb bodies (PC bodies).This process is best studied in Drosophila cells.Assembly of PcG targets into repressive chromatin compartments appears to enhance the level of the PcG-mediated transcriptional silencing [76], apparently due to the increase of repressive factor local concentration.The integrity of the repressive compartment is likely supported by the interactions between insulator proteins.Whatever is the precise mechanism of the PC body formation, it is clear that repression by PcG proteins involve reconfiguration of chromatin in a relatively large genomic area(s).
Summarizing one may conclude that many nuclear compartments are assembled starting from nucleation centers located on a chromatin fiber.The positioning of these compartments is mediated by folding the chromatin fibers.Other compartments (such as splicing speckles) are located in interchromatin domain.Already for this reason the way of chromatin folding will partially determine their positions.However, it is possible that these compartments are indirectly linked to some particular genomic regions.Indeed, it has been demonstrated that, at least in some cases, active genes (and thus, transcription factories) are located at the surface of the speckles [77], and transcribed RNA is moved to the speckles [78,79].The mechanism of the speckle assembly is not clear at the moment.Some authors consider speckles as places where non-used components of the splicing machinery are temporarily deposited.Others suggest that splicing occurs within speckles.If the latter is true, the speckles may expand and collapse depending on functional necessity.Obviously, their positions in this case will be determined by positions of active genomic regions.
Conclusions.For many years the spatial organization of interphase chromosomes and functional compartmentalization of the eukaryotic cell nucleus have been studied independently.Now it is becoming increasingly evident that they are tightly interconnected.It is likely that folded chromosomes underlie the spatial compartmentalization of the eukaryotic cell nucleus constituting a structural milieu for the assembly of functional compartments.One thing that should be stresses is that both the chromosome folding and functional compartmentalization of the cell nucleus are highly dynamic.Chromosomes can adopt numerous alternative configurations as follows from FISH data [39][40][41], and nuclear compartments are constantly assembled and disassembled as follows from the high exchange rates of their constituents [80].The equilibrium between all these processes can be modulated by a number of factors.Apparent order in the eukaryotic cell nucleus is in a way illusive.It originates from disorder via continuous choices between different spatial arrangements aimed to adapt the functioning of the genome to the current conditions.