The primary role of chromatin assembly is the compaction of an otherwise very long DNA molecule into discrete structures that fit into the nucleus. However, this compaction cannot be achieved at the expense of a cells ability to transcribe and replicate DNA. Recent findings have highlighted several properties intrinsic to the assembly and arrangement of chromosomes that retain access to genomic DNA and contribute to the regulation of gene expression.
Despite 20,000 genes being present in each haploid nucleus, the number of transcription foci is limited to around 2000. These transcription foci, also known as transcriptional factories are distinct submicron nuclear regions that are associated with nascent RNA production and are enriched in RNA polymerase II (RNA pol II) complexes . The limitation in the ratio of transcriptional foci per genes to be transcribed is overcome by a phenomenon where groups of genes share the same transcription machinery. These genes are considered to be co-transcribed. Co-transcribed genes are, however, arranged on a linear DNA molecule, and not necessarily adjacent to each other. In order for these distant genes to be transcribed by the same transcription factory, they must first co-localize with this machinery. Indeed, various dynamic properties, and molecular processes, have been identified that promote this very phenomenon. For example, chromosome dynamics and the spatial distribution of genes may lead to the active redistribution of genes and chromosomal regions. In other cases, post translational modification of the chromatin, or active remodeling of chromatin by ATP powered proteins, facilitates this process.
The spatial organization of chromatin within the 3-dimensional space of a chromosome territory enables the co-localization of co-transcribed genes and their transcriptional foci. Many gene positioning studies have shown that individual genes often loop out of their chromosomal territory to co-localize with transcription factories. This often leads to interchromosomal compartments becoming enriched with intermingling chromatin loops, either from the same chromosome, or different chromosomes. It has been suggested that this repositioning can occur upon transcriptional activation .
Measuring the diffusional motion of chromatin by sub-micrometer single-particle tracking, a characteristic confinement radius (significantly smaller than the size of the nucleus) can be determined for each locus. This demonstrates that, at least in yeast, centromers and telomers have a radii of confinement approximately twice as small as the rest of the chromosomal sites . The authors also showed that in yeasts and drosophila, chromatin constantly undergoes diffusive Brownian motion, constrained by confinement regions of gene loci, which rarely exceed 0.3 µm .
Importantly, the repositioning of genes through chromosome territory dynamics is not always random, and the spatial redistribution of genes may involve specific nuclear structures or landmarks. This may have a significant effect on gene expression . Local compaction dynamics, long-distance interactions with alternative sections of DNA, and interactions with nuclear scaffolds  all play a role in the control of gene redistribution. Where interactions between DNA and nuclear scaffolds occur, anchor points , known as matrix attachment regions (MARs), are formed.
Long-distance chromatin interactions may either involve the establishment of physical contacts between two sequence elements that are not adjacent to each other, but are present on the same linear chromosome (as is the case when enhancers interact with promoters), or between loci on different chromosomes . Importantly, with most interactions between loci occurring in only a small fraction of cells at any given time, long-range contacts are considered to be, at least in part, random, and are therefore difficult to predict . Despite these difficulties, it has been suggested that the nucleoskeleton is involved in the regulation of long-distance contacts. For example, Chuang et al.  observed a fast (0.1-0.9 µm/min) long-range (1-5 µm) directional movement of transgenic chromatin arrays. Nuclear actin together with myosin – two important components of the nucleoskeleton – were proposed to serve as molecular motors that direct the movement of chromatin towards a given target region . This was supported when the movement of actin arrays was blocked by the expression of mutant nuclear myosin I or mutant actin. Moreover, with the actin mutant unable to polymerize, the looping of U2 snRNA genes towards coiled bodies was also abolished .
Interactions between genes and nuclear landmarks also affect gene transcription. These landmarks, which are distinct nuclear regions, include the nuclear lamina (NL), nuclear pore complexes (NPCs) and the nucleolus (reviewed in ).
The nuclear periphery is often found to preferentially interact with transcriptionally silent chromatin, which is characterized by a low gene density. It has been proposed that the nuclear periphery itself creates a specific environment that favors histone deacetylation and gene silencing . Indeed, in yeast cells (which generally lack nuclear lamin), gene silencing resulted from the tethering of a gene locus to the periphery . This suggests that in mammals, the nuclear periphery is heterogeneous, with microdomains of different compositions having different effects on genome function . Takizawa et al.  proposed that simply being near the periphery without physically associating with the NL is not enough to induce gene repression.
Nuclear pore complexes represent another distinct microenvironment; however, in contrast to the nuclear periphery this landmark is often associated with gene activation . In yeast cells, the highly transcribed ribosomal protein (RP) is connected with NPCs via the actin-related protein, Arp6 . Interestingly, interactions between chromatin and NPCs may take place both in and away from the nuclear periphery, making the dynamic movement of lamins and nuclear pore proteins integral to gene regulation . The exact dynamics that drive these interactions in the nucleoplasm remains poorly understood.
The nucleolus may also anchor specific chromatin loci. In addition to rRNA genes, it often harbors large genomic regions (median size 750 kb) that are enriched in centromeric satellite repeats and inactive gene clusters. Centromeric regions are also found associated with nuclear lamina, suggesting that centromeres are distributed between the nuclear lamina and nucleoli .
Despite the current evidence highlighting the influence of chromosome territory dynamics in the regulation of gene expression, little is known on the mechanisms behind these processes. For example, it is unclear how and why activated genes are translocated, or loop, from one chromosome territory to another. One plausible scenario involves the transduction of mechanical stimuli to the nucleus directly via the cytoskeleton. In such cases, cytoskeletal forces induce nuclear deformation (i.e., elongation or squeeze) and subsequently, alter chromosome topology and gene expression . The nucleoskeleton (i.e., nuclear actin and myosin) provides another mechanism to control the long-distance directional movement of genes , but this is likely to be restricted to specific genes, e.g. U2 snRNA . Currently, the extent to which genes move through active guidance as opposed to diffusion, remains unclear.
While chromosome territory dynamics is believed to regulate gene expression through the redistribution of genes and the subsequent co-localization of these genes with transcription machinery, changes are also commonly made to the chromosome structure at a ‘local’ level. Although these changes do not necessarily involve the redistribution of genes, they do have a significant influence on gene regulation.
Alterations to the chromatin structure occur as a consequence of the limitations acquired through its condensation. As chromatin is condensed into the primary nucleosome structure, DNA becomes less accessible for transcription factors. With the loosening of this chromatin structure, however, transcription machinery is better able to access the genomic DNA, and transcription is thus promoted. Hence, nucleosome organization and dynamics are regularly modified by the combined influence of covalent post-translational modifications (PTMs), histone chaperones, ATP-dependent nucleosome remodelers and histone variants .
To date, over 100 distinct posttranslational modifications (PTMs) of histone have been described . These often take place at the N-terminal of the histone tails , and may include, but are not limited, to various acetylation, methylation and phosphorylation reactions. Although not all PTMs can be correlated with an increase or decrease in the expression of a given gene, several trends have been identified. Firstly, acetylation, phosphorylation and ADP-ribosylation, generally weaken charge-dependent interactions between histones and DNA, thus increasing the accessibility of genetic material to transcription machinery . Lysine methylation, on the other hand, tends to increase nucleosomal stability and promotes heterochromatin formation , thereby reducing the accessibility of DNA. Of course, histone formation is dynamic, and the rate of histone turnover can occur rapidly . It has therefore been proposed that mechanisms exist to maintain specific PTMs even in the face of ongoing nucleosome turnover and DNA replication. Although the mechanisms behind this are unclear, it has been suggested that that some histone-modifying enzymes (e.g., HDACs, methyltransferases Suv39h, SETDB1, SetD8 and G9a) remain associated with chromatin during its turnover. This would allow them to modify their cognate residues immediately following the deposition of new histones .
The chromatin structure may also be altered by ATP-dependent chromatin-remodeling enzymes which use the energy of ATP hydrolysis to mobilize nucleosomes along DNA, evict histones off DNA or promote the exchange of histone variants (reviewed in ). Four families of chromatin-remodeling ATPases have been described. These are classified based on their domain structures and include the SWI/SNF, (switching defective/sucrose non-fermenting) family, the ISWI (imitation SWI) family, the NuRD (nucleosome remodeling and deacetylation)/Mi-2/CHD (chromodomain, helicase, DNA binding) family and the INO80 inositol requiring 80) family .
The SWI/SFN family, and ISWI family, are the two most-studied and best-characterized classes of chromatin-remodeling complexes. Both of them can bind to either DNA or nucleosomes with various degree of affinity. After binding to their target, these complexes utilize ATP hydrolysis to destabilize histone-DNA contacts (reviewed in . SWI/SNF-based remodelers presumably act by rotating DNA along its axis which subsequently generates positive supercoils. This may lead to disruption of histone octamer and transfer of histones to different DNA locations, as well as to formation of altered dimeric nucleosomes with enhanced access to DNA regulatory regions . ISWI ATP-dependent remodeling, on the other hand, involves the formation of loops or bulges in the DNA that can propagate throughout the DNA on the octamer surface. Similarly to SWI/SNF, this ultimately leads to destabilization of octamer-DNA complexes and nucleosome remodeling  . Finally, after histone displacement, the remodeling complexes have to be released from their targets. The actual mechanism of this release and the potential role of ATP hydrolysis in this process remains poorly understood.
Clearly, the regulation of gene transcription is influenced by, and possibly influences, chromosome structure and chromosome territory dynamics. The impact of mechanical stimuli on the nucleus is evident when considering the forces the nucleus and its contents are subjected to. An outward ‘pulling’ force is generated within the actin filament network and this opposes the inward pulling force generated during chromatin condensation. With each nuclear landmark and gene redistribution process subject to the same forces, it is likely that mechanical signals and external forces influence gene expression, albeit indirectly .
Pluripotent stem cells can potentially differentiate into any given cell type. They commonly exhibit mechanically softer nuclei , lack type-A lamins  and have a poorly defined cytoskeleton .
Embryonic stem cells are pluripotent in early organism development, but gradually undergo lineage restriction and transform into the stem cells with limited differentiation capacities (e.g., hematopoietic stem cells, neural stem cells). These stem cells divide to give rise to specialized cells and tissues, but half of the new generation does not differentiate, retaining their stem cell identity. This type of division cycles is possible due to both epigenetic inheritance (i.e., DNA methylation, chromatin modifiers, histone variants) and chromatin plasticity of stem cells (reviewed in ).
Generally, stem cell chromatin is less compact and more transcription-permissive when compared to differentiated cells. Fluorescent recovery after photobleaching (FRAP) experiments showed high exchange rate of several chromatin-associated proteins (e.g., histones H2B and H3, heterochromatin-associated protein HP1 and linker histone H1) in embryonic stem cells . However, a search for molecular “signature” of stem cells failed to retrieve much: only Oct4, Sox2 and Nanog were identified as commonly expressed embryonic stem cell-specific genes (reviewed in ). In terms of localization, a region of 12p chromosome that contains clustered pluripotency genes (including Nanog) is confined to more central nuclear regions in human embryonic stem cells as compared to differentiated cells . Furthermore, Oct4 locus was found to loop out from its chromosome territory (chromosome 6p) in human embryonic stem cells . The same study demonstrated that centromeric regions of chromosomes in stem cells are not as much confined to the nuclear periphery as they are in differentiated cells .
To describe the transcriptional regulation in embryonic stem cells, Meshorer and Misteli proposed the following models :