Soft-matter properties of multilayer chromosomes

In eukaryotic cells, each chromosome contains a single genomic DNA molecule, which is associated with histone proteins and forms enormously long chromatin filaments filled with many nucleosomes. The nucleosome is composed of a flat cylindrical core particle formed by ∼146 bp of DNA wrapped around an octamer of histones (H2A₂, H2B₂, H3₂, H4₂) and linker DNA that connects adjacent core particles and is associated with histone H1 [1]. During mitosis, to avoid any damage of the genomic DNA when it is transferred to the daughter cells, the chromatin filament is highly packed within condensed metaphase chromosomes [2–6]. In interphase, chromatin is not so densely packed to allow DNA replication and gene expression, and it is organized into topologically associating domains (TADs) and larger compartments [7–10].

There is a growing interest in the mechanisms underlying the formation of intracellular condensates and several studies suggested that they can be self-assembled by liquid–liquid phase separation [11–15]. In agreement with the condition of fluidity of the resulting liquid condensates, chromatin in the active interphase nucleus, and even in heterochromatin regions and metaphase chromosomes is locally dynamic like a fluid [16, 17]. To study chromosome structure, it must be taken into account that associative polymers such as chromatin can give rise to complex network fluids [15]. It was previously reviewed [18] that the chromatin filament forms multilayer plates in chromosomes. In this perspective, the structural and dynamic properties of multilayered chromosomes will be discussed with the objective to show that chromatin has the typical properties observed for different soft-matter systems [19], which are very interesting because of their emergent behavior leading to spontaneous pattern formation. It will be shown that the relatively large size of nucleosomes and the weak energy of interaction between them make chromatin easily deformable, but at the same time the interplay between these interactions and thermal fluctuations leads to self-organizing multilayer structures. Furthermore, these soft-matter assemblies are dynamic because they can be highly altered by different factors, which in the case of chromatin are the changes in Mg²⁺ concentration and in nonhistone protein composition. The possible functional implications of these dynamic properties will also be discussed.

Chromatin filaments behave as associative polymers. In vitro studies performed with diluted chromatin fragments (obtained by micrococcal nuclease digestion of cell nuclei) and nucleosome arrays showed that the interaction between nucleosomes gives rise to fibers and aggregated structures having different compaction degrees depending on the concentration of cations [20–23]. Completely extended filaments with individual nucleosomes clearly visible are only observed in the absence of cations; in the presence of relatively low concentrations of divalent cations (∼2 mM Mg²⁺), nucleosomes in the filament interact with each other and form compact 30 nm fibers in which nucleosomes cannot be distinguished as separate units [24, 25].

Early studies showed fibers emanating from mitotic chromosomes suspended in water without cations and it had been generally assumed that the 30 nm fiber is the basic structural element for the packaging of chromatin in chromosomes [20]. During mitosis, there is an increase in the concentration of Mg²⁺ in the nucleus [26, 27] and chromatin filaments are condensed in chromosomes. The concentrations of Mg²⁺ and chromatin are high in metaphase chromosomes, respectively, 5–22 mM [26] and ∼0.34 g ml⁻¹ [28]. Cryo-electron microscopy images of vitrified sections of mitotic chromosomes showed that they have a dense and uniform 10 nm granular morphology that is not compatible with a widespread presence of 30 nm fibers [29]. These results suggested that chromatin in condensed chromosomes forms a highly disordered 10 nm fiber that behaves like a polymer melt. The structural relevance of the 30 nm fiber in vivo has been questioned by several authors [30–34].

Chromosomes in buffers containing the cation concentrations corresponding to metaphase are highly electron-opaque. This precludes the direct transmission electron microscopy (TEM) analysis of the chromatin structure inside the metaphase chromatids. To overcome this problem, several procedures were developed to disrupt chromosomes maintaining the ionic conditions of metaphase. Chromosomes were spread on carbon-coated grids and, before glutaraldehyde crosslinking, they were incubated at 37° C with different buffers containing metaphase cation concentrations [35]. Using this procedure, emanations having a multilayered plate-like structure rather than a fibrillar structure were observed surrounding partially denatured chromosomes. In other experiments [36], multilayered plates were observed in TEM images of chromosomes mechanically disrupted by passage through a syringe needle. This procedure can be criticized because the shear forces generated could favor the formation of planar chromatin. However, as can be seen in the TEM images presented in figures 1(A) and (B), when chromosomes are disassembled using a much softer procedure (see details in the legend of this figure), multilayered plates are also observed [37]. These planar structures are formed by the chromatin filament, which can be seen surrounding plates denatured by treatment with the divalent cation chelator EDTA [36], but chromatin is densely packed in native plates and the path of the filament cannot be determined from the images.

Zoom In Zoom Out Reset image size

Cryo-electron tomography (cryo-ET) studies [6] confirmed these findings and demonstrated that frozen-hydrated chromatin (uncrosslinked and unstained, and not adsorbed to the flat carbon films used in TEM [38]) emanated from metaphase chromosomes is planar and forms multilayered plates (figures 1(C)–(E)). The tomographic three-dimensional reconstructions showed that plates with a single layer have a thickness of ∼7.5 nm, corresponding to a sheet of slightly tilted nucleosomes. However, when two layers are in contact, they have a thickness of only ∼13 nm, which is thinner than the sum of two independent layers, suggesting that the nucleosomes in the layers interact and interdigitate. The images always show plates having different sizes and shapes. This is because plates have a large surface area but are very thin and can be easily deformed and broken during the preparation and deposition procedures. However, in some samples (figure 1(C) [6]) the cryo-tomograms contained large plates with many stacked layers, which probably correspond to a relatively large part of a chromosome. Chromatin layers are only clearly visible in regions where chromosomes are distorted; in native condensed chromosomes, layers are interdigitated, space is completely filled by nucleosomes and there is no visible higher-order organization. It was argued [6] that this high chromatin compaction could explain why layers were not distinguished as differentiated structural units in the cryo-sections of native mitotic chromosomes (see above), which led to the proposal of a highly disordered organization of the 10 nm fibers in chromosomes [29]. These observations support the compact thin-plate model for chromatin folding in metaphase chromosomes (schematized in the inset of figure 1(C)), consisting of many chromatin layers stacked along the chromosome axis.

As represented in figure 2(C), according to the images and thickness values obtained in TEM and cryo-ET experiments and to the core particle dimensions, chromatin plates consist of stacked mononucleosome layers with core particles slightly tilted with respect to the layer surface. The distance between layers (∼6 nm) is the same as the distance between two core particles associated thought their faces (figure 2(A)). This repetitive distance is consistent with the dominant peak at ∼6 nm observed in small-angle x-ray scattering experiments performed with whole metaphase chromosomes (figure 2(B) [6]). The association between stacked layers is due to multiple face-to-face interactions between nucleosomes. This is in agreement with in vitro studies showing that purified nucleosome core particles associate through their faces and form long columns [39–41]; these lateral interactions are also produced between nucleosomes in compact chromatin fibers [24, 25, 42–44]. Nucleosomes in the layers are represented with different orientations in figure 2(C) because chromosomes are not birefringent [37], and it is known that parallel columns of stacked nucleosomes with the same orientation (see figure 3(A), right panel) are birefringent [39]. Furthermore, it has been found that face-to-face association can be produced between nucleosomes having diverse relative orientations [45–51].

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The interactions between nucleosomes are anisotropic. The lateral electrostatic association of nucleosomes is mediated by the positively charged N-terminal tail of histone H4 and an acidic surface formed by histones H2A and H2B [33, 53]. As can be seen in figure 2(A), the energy corresponding to this face-to-face interaction (top-on-top stacking) is higher than those corresponding to other nucleosome orientations. In simplified chromatin systems (nucleosome core particles in figure 3(A) [39] and tetranucleosomes with fixed linker lengths in figure 3(B) [54]), this anisotropic interaction leads to large face-to-face stacked structures. In the crystal structure shown in figure 3(B), the orientation of the nucleosomes is constant, but presumably in the stacked layers of chromosomes (schematized in figure 2(C)) nucleosomes have diverse orientations due to variability in linker lengths and thermal fluctuations. According to results obtained using optical tweezers [55, 56], a DNA origami-based force spectrometer [57], and computer modeling [58–61], the energy required for breaking a single face-to-face interaction is between 2.7 and 14k_B T (k_B T = 4.1 × 10⁻²¹ J at room temperature). The interactions that stabilize the two-dimensional network in the chromosome layers might be stronger than the interactions between adjacent layers because the nucleosomes in each layer are connected by the covalent backbone of linker DNA, and histone H1 and the core histone tails [62] not involved in the face-to-face nucleosome associations could also contribute to the stabilization of the layers. Furthermore, the interactions that stabilize two turns of DNA in each nucleosome (∼40k_B T [1, 63, 64]) could avoid the disruption of the layers and the breakage of the covalent backbone of DNA when chromosomes are deformed by large stretching forces (see section 4). Therefore, in the model in figure 2(C), it is considered that the stabilizing energy ε_intralayer per nucleosome corresponding to the interactions within a layer is higher than the interlayer energy ε_interlayer. The relatively weak association of adjacent layers is consistent with the observation that many images of plates emanated from chromosomes show a relative sliding between the successive layers (see examples in figure 1(B) and in references [36, 37]).

The discovery of multilayered chromatin was unexpected, but this structure should not be considered so surprising because many natural materials have a multilaminar structure [18]. As observed with chromatin plates, all these materials have stronger bonds within layers than between adjacent layers. There are multilayered structures formed by DNA. In dinoflagellates, which lack histones and do not form nucleosomes [65], genomic DNA is packed within chromosomes as a multilayered liquid crystal formed by many DNA layers oriented perpendicular to the chromosome axis [66–68]. In vitro experiments showed that purified DNA of different sizes can self-assemble into multilayer structures [69, 70], and it was found that multilayer chromatin plates can be self-assembled from small chromatin fragments dialyzed against buffers containing the Mg²⁺ concentration corresponding to metaphase (figure 4 [71]). This later finding suggested the possibility that the whole chromosome could be a self-organizing supramolecular structure [72]. Although, in principle, it seems unlikely that large structures such as human chromosomes (with lengths of several micrometers) can be self-organized, numerous laboratories have demonstrated that different kinds of soft-matter building blocks having nanometer-scale sizes can spontaneously form micrometric structures [73–75]. For instance, the self-association of DNA origami elements (∼50 nm) in buffers containing Mg²⁺ gives rise to large multilayered cylinders, having micrometer-scale dimensions [76]. According to the knowledge of different soft-matter systems [19, 77], the fact that the thermal energy (which is responsible of the Brownian motion and structural fluctuations) is comparable in size to the energy of the interaction between large building blocks justifies the spontaneous self-assembly of large structures having a minimum energy. Since the relatively large size of the nucleosome building blocks of chromatin and the low energy of interaction between them corresponds to typical soft-mater structures, it can be proposed that chromosome self-organization by layering may be produced by the interplay between thermal energy and weak interactions between nucleosomes in the chromatin filament. In addition, other structural factors such as condensins are essential for the complete self-organization of mitotic chromosomes (see below).

Zoom In Zoom Out Reset image size

Metaphase chromosome volume depends on the size of the DNA molecule that it contains, but the DNA density is about the same (∼166 Mb μm⁻³) for diverse plant and animal chromosomes [78]. Chromosomes are elongated cylinders with a smooth surface, and relatively similar shape proportions were observed for plant and animal chromosomes of different sizes [72]. These structural similarities of metaphase chromosomes of diverse species strengthen the possibility that chromosomes are self-organizing soft-matter structures built following the same pattern in many species. If the chromatin filament is folded according to the repetitive association pattern shown in figure 2(C), it is possible to explain some aspects of the morphology of the chromosomes [72]. Nucleosomes in the periphery of the chromosome are in contact with the medium, they cannot fully interact with nucleosomes within the chromosome, and this generates a surface energy that destabilizes the structure. Presumably, chromosomes are smooth cylinders because this geometry has a lower surface area, and consequently a lower surface energy than cylinders with irregular surfaces. As it will be discussed below, condensin activity is also required for the building of the characteristic morphology of mitotic chromosomes.

Some plates observed in the TEM images of disassembled chromosomes are very large. There are plates with a surface area larger than the cross-section of a chromatid (∼0.3 μm² considering a chromatid diameter of ∼0.6 μm); for instance, the big plates shown in figure 1(B) [37] have a surface area of ∼1 μm². Since plates are very thin and are easily broken in the microscope preparations (see above), the observation of large plates suggests the possibility that they could form a continuous structure within native chromosomes. Considering that early studies indicated that mitotic chromosomes have a helical shape [79–81], this led to the suggestion that the successive chromatin layers are connected between them forming a helicoid [72], in which each chromatin layer is equivalent to a helicoidal turn. Nanotechnology research has shown that the self-assembly of rod-shaped viruses and other materials can form large helical structures [74, 82]. A continuous chromatin helicoid gives a structural solution for the homogeneous protection of the entire genomic DNA molecule in each chromosome. Furthermore, a helicoid surface has a zero mean curvature [83] and this allows a full contact between adjacent layers as proposed in the model in figure 2(C). Since the whole structure is crosslinked by the covalent backbone of the DNA, chromosomes are hydrogels, which at the same time have an internal multilaminar liquid-crystal order [72]. The high water content of the chromosome (one-third of its volume is occupied by water [84]) provides fluidity to this lyotropic liquid crystal, which can exist only in the presence of relatively high concentrations of Mg²⁺ and when the concentration of chromatin is high (see above). All the observations considered in this section suggest that condensed chromosomes are self-organized membraneless organelles.

This model based on interactions between the fundamental building blocks of chromatin is not complete. Other studies showed that cohesin and condensins are involved in the definition of the chromosome architecture during interphase and mitosis [85–87]. It was suggested that condensins can produce chromatin folding by loop extrusion [88, 89]. More recently the loop extrusion produced by condensins was directly observed [90–92]. In Xenopus cell-free egg extracts almost completely depleted of nucleosomes, condensins are able to assemble cylindrical chromosome-like structures [93]. Taken together these studies indicate that loop extrusion can generate cylindrical chromatin brush structures. Results obtained using chromosome conformation capture techniques (see section 5) showed that condensins mediate the formation of a helical array of nested loops in metaphase chromosomes [5]. This study also showed that in the absence of condensins the morphology of chromosomes is altered but they achieve a normal degree of chromatin volume compaction in mitosis. This is in agreement with previous studies showing that chromosomes are compacted laterally by condensins [94], and that chromatin volume is uncoupled from chromosome architecture in mitosis [95, 96]. From these observations, it was suggested that chromosome volume compaction is mediated by the intrinsic properties of chromatin [96], which can be modulated by histone post-translational modifications. These results offer a possibility to reconcile fibrillar models with a multilayer organization of chromatin. In this regard, it was argued [6] that the long nested loops (∼0.5 Mb) proposed in chromosome conformation capture experiments [5] must be highly packed to achieve the high chromatin concentration of metaphase chromosomes (see above) and it was suggested that they could be compacted into chromatin layers. Other authors [3, 97] suggested a hierarchical layering of loops to reconcile the models based on a highly disordered chromatin filaments [2, 29] and results indicating the multilayer organization of chromosomes. It would be very interesting if future studies shed light on the interaction of nonhistone chromosomal proteins with chromatin plates.

On the other hand, chromosome images obtained from typical banded karyotypes and from different multicolor cytogenetic analyses were used to gain information about the internal structure of chromosomes [98]. It was proposed that the observed transverse orientation of the cytogenetic bands is due to the orthogonal orientation of the chromatin layers with respect to the chromosome axis. The number of layers in a specific band is dependent on the amount of DNA contained in the band. A multilayered structure with weak interactions between adjacent layers can explain the splitting of broad bands (formed by several layers) observed in chromosome stretching experiments [99], and the maintenance of the orthogonal orientation of the resulting split bands. The analysis of the human genome showed that the thinnest bands (which are also orthogonal to the chromosome axis) correspond to short sequences of ∼1 Mb [100, 101], indicating that within metaphase chromosomes chromatin containing relatively short stretches of DNA fills completely the cross-section of each chromatid. This observation is compatible with a multilayered organization of chromosomes because each layer of a human chromosome contains ∼0.5 Mb of DNA. Furthermore, the multilayered structure of chromatin is compatible with the orthogonal orientation and planar structure of the connection surfaces seen in sister chromatid exchanges and in the chromosome translocations observed in cancer cells [102].

Monolayer plates are stable at room temperature in aqueous solution in the presence of Mg²⁺ and can be scanned by atomic force microscopy (AFM) (see selected topographic images in figures 1(F) and (G) [36]). When DNA was cleaved with micrococcal nuclease, plates were irreversible denatured and a high increase in the friction coefficient was observed in AFM-based friction force measurements [103]. Consistent with these observations, it was found by other authors [104] that whole metaphase chromosomes digested with nucleases lost their mechanical integrity. In contrast, the structure of undigested monolayer plates was completely recovered after AFM scanning, indicating that chromatin in native monolayers forms an elastic two-dimensional network that protect the covalent backbone of the DNA. This protection is observed even when the scanning force (∼5 nN) is higher than the force that can produce the breakage of naked DNA (∼1 nN). Presumably, this remarkable elasticity is due to a reversible nucleosome unwrapping during the AFM scanning.

Hydrogels containing covalent and ionic bonds are more stretchable and have a larger fracture toughness than typical hydrogels stabilized exclusively by covalent crosslinks [105]. After strong deformations, these hydrogels have the capacity of self-healing through the regeneration of the broken electrostatic interactions. As demonstrated in stretching experiments performed with micropipettes [106–108], metaphase chromosomes have good elastic properties. They show a viscoelastic behavior and reversibly recover their initial shape and dimensions after repeated extension-relaxation cycles up to 5 times their native length. This elasticity can be explained considering that during the extensions the face-to-face interactions between nucleosomes in adjacent layers are broken and that these electrostatic interactions are then regenerated allowing the recovery of the initial chromosome length [72]. The work W_stretch done by the stretching forces to produce different chromosome elongations was calculated from the extension curves obtained experimentally for mitotic chromosomes with two sister chromatids. The work required for fivefold reversible extensions is dependent on the size of the chromosomes: W_stretch is ∼0.1 pJ for the large N. Viridescents chromosomes (∼3400 Mb per chromatid) and ∼0.01 pJ for the relatively small human chromosomes (∼150 Mb per chromatid). Since the estimated total energy of face-to-face interactions between nucleosomes for N. Viridescents and human chromosomes is, respectively, 0.4–2 and 0.02–0.09 pJ (considering that the energy of a single face-to-face interaction is between 2.7 and 14k_B T [55–61]; see section 2), W_stretch for fivefold extensions can be justified by the breakage of a significant part of the face-to-face nucleosome interactions. Furthermore, it was observed that chromosomes are irreversibly deformed (but are capable of a partial recovery of the initial length) in extensions beyond the elastic limit. The high amount of work required for fiftyfold extensions (W_stretch ≈ 10 pJ for N. Viridescents chromosomes with two chromatids) can be justified by the breakage of all the nucleosome–nucleosome interactions between layers (0.4–2 pJ), large deformations of chromatin layers involving nucleosome unwrapping (∼6 pJ, estimated considering that the energy required for the unwrapping of a single nucleosome is ∼40k_B T [1, 63, 64]; see section 2), and the breakage of additional stabilizing interactions involving nonhistone proteins such as condensins (see the preceding section). These mechanisms may act as energy dissipators to protect the covalent continuity of genomic DNA during the pulling of chromosomes toward the spindle poles in anaphase.

The structural relationship between cytogenetic bands and the multilayered organization of chromatin in metaphase chromosomes is summarized at the end of section 3. The typical banding procedures were developed for the analysis of mitotic chromosomes [109], but results obtained with microdissection-based multicolor banding showed that the band pattern of metaphase chromosomes does not disappear during interphase [110–114]. This result suggests that the multilayered structure of metaphase chromosomes is conserved in the chromosome territories [115] observed in interphase. This possibility is strengthened by results showing that, in buffers containing the Mg²⁺ concentration corresponding to interphase (2–4 mM [26]), chromatin emanated from mechanically disrupted nuclei is planar [116]. This planar structure was observed in all the stages of the interphase, and it was also found that the chromatin fragments produced by micrococcal nuclease digestion of G1, S, and G2 nuclei self-assemble into plate-like structures. In contrast to the thick multilayered plates frequently observed in the preparations of metaphase chromosomes (see above), the TEM images of chromatin from interphase nuclei often show thin monolayers [116], indicating that planar interphase chromatin has a lower tendency to form stacked structures than mitotic chromatin. These structural changes, which may be related to the dissociation of condensins from chromosomes at the end of mitosis [87], can be interpreted as a phase transition that gives a higher fluidity to chromatin during interphase. The transformation of condensed chromosomes into a more fluid phase may have functional implications that will be discussed in a speculative way in this section.

Genome-wide chromosome conformation capture (Hi-C) studies have shown that in interphase there are many contacts between DNA sequences corresponding to topologically associated domains (TADs), which are considered to be the functional subunits of chromatin [7–10]. Since the size of a TAD is similar to the amount of DNA in a chromatin layer in metaphase chromosomes, it was suggested that each layer may correspond to a TAD [116], and that the enhancer–promoter interactions required for gene expression could be produced by the folding of the chromatin filament in the layers. It was proposed that TADs are formed by active extrusion of chromatin loops mediated by cohesin [117, 118]; DNA loop extrusion performed by cohesin was directly observed in vitro [92, 119, 120]. Furthermore, the association of cohesin with the insulator protein CTCF creates boundaries between TADs [121, 122] and it was suggested that these proteins could insulate layers from each other to generate well-defined regions for gene expression [18]. The most frequent contacts detected in S and G2 cells, which correspond to TADs (⩽1 Mb), are replaced by dominant contacts at much larger distances (∼10 Mb) in mitotic and early-G1 cells [123, 124]. These results were interpreted considering that the multilayer organization of chromosomes has different compaction degrees during the cell cycle [116]. In interphase cells, the layers have a low tendency to be stacked (see above) and the dominant contacts are produced at short distances within the layers, but the close stacking of chromatin layers in mitotic cells favors long-distance inter-layer contacts. It was hypothesized that during mitosis the tightly stacked chromatin layers inhibits transcription, and that chromatin accessibility increases when the layers become unstacked in interphase [18].

The easy sliding of layers in chromatin plates (see above) indicates that there is no topological entanglement between adjacent layers. This well-defined topology could justify recent super-resolution FISH [125] and multi-contact chromosome conformation capture [126] results showing a lack of entanglements within and between chromosomal domains during interphase. In a previous study [127], it was proposed that chromatin in interphase is folded like a Peano curve that forms a fractal globule without entanglements. Note that a multilayered organization of chromatin can be idealized as a Peano curve forming planes that are not entangled between them. Since the thickness of a single chromatin layer corresponds to a monolayer of nucleosomes (figure 2(C)), the chromatin filament is readily accessible from both sides of unstacked layers. The total surface area of planar chromatin in a diploid human cell is ∼8 × 10³ μm², which is an enormous value, comparable to the surface area of the plasma membrane plus the endoplasmic reticulum of a 10 μm-diameter cell [128]. Nevertheless, taking into account that there is a switching between active and inactive compartments during cell differentiation [129], it can be suggested that only specific clusters of layers are fully unstacked and active in the different stages of the development. Presumably, chromatin layers are stacked in heterochromatin regions, but in active compartments unstacked layers can interact with transcription factors, mediators, and other proteins related with gene expression. The local chromatin fluctuations and the mobility of proteins driven by thermal energy [16, 17] could facilitate the association of all the elements of the transcription machinery with active chromatin regions, but the lower local mobility [130] and relatively high stiffness [131] of heterochromatin do not favor these interactions.

It would be very interesting to know the structural relationships between epigenetic elements (DNA methylation, histone post-translational modifications, HP1 and polycomb proteins, noncoding RNAs [132, 133]) and the modulation of accessibility of chromatin layers in different cell types. In this regard, there are studies suggesting epigenetic mechanisms that could disrupt face-to-face nucleosome interactions between layers to favor gene expression. It was reported that the binding of different proteins to the acidic nucleosome surface formed by histones H2A and H2B (see above) prevents its association with the histone H4 tails of other nucleosomes [134]. Furthermore, computer modeling showed that the acetylation of histone tails inhibits internucleosome interactions [135], and it has been demonstrated experimentally that the acetylation of the histone H4 tail disrupts nucleosome stacking mediated by face-to-face interactions [41, 136]. Other mechanisms could strengthen the association of adjacent layers. In particular, the cryo-electron microscopy results showing that the protein HP1 can bridge two nucleosomes containing a trimethylated histone H3 (H3K9me3) [137] suggest that this heterochromatin protein could favor nucleosome interactions between adjacent layers to silence gene expression. The binding of HP1 to nucleosome arrays induce the formation of compact droplets via liquid–liquid phase separation [138]. It would be very interesting to investigate whether planar chromatin layers bridged by this protein can form phase-separated condensates.

The results reviewed in this work show that chromatin has properties that are typical of soft matter (figure 5(A)). This provides tools for a better understanding of chromosome structure and function and for the design of future research. As summarized in figure 5(B): (1) the interplay of the weak anisotropic interactions between nucleosomes and thermal energy is probably partially responsible of the self-organization of metaphase chromosomes; (2) changes in the conditions of the medium (decrease of cation concentration [26, 27]) and dissociation of condensins from chromosomes [87] produce a phase transition from condensed metaphase chromosomes to fluid interphase chromatin; (3) the local structural fluctuations (typical of soft-matter structures) driven by thermal energy may facilitate gene expression in interphase chromatin; and (4) soft-matter structures such as metaphase chromosomes are easily deformable and consequently their native conformation is difficult to be studied experimentally.

Zoom In Zoom Out Reset image size

As described in the text, multilayer planar chromatin have been observed in vitro under physiological conditions using a wide variety of microscopy techniques. In a recent electron diffraction study [139], the authors have found a repetitive structure oriented perpendicular to the chromosome axis that could be directly related to the multilayer structure of chromosomes considered in this perspective. The relatively high thickness (100–200 nm) of these transverse structures suggests that they could correspond to clusters of stacked chromatin layers. In another recent study [140], dynamic simulations using a near-atomistic model have shown that tetranucleosomes form relatively stable co-planar stuctures. As discussed above, it was proposed [3, 6, 97] that the chromosome models based on irregularly folded fibers and chromatin looping [2, 4, 5, 29] can be made compatible with a multilayer organization of chromosomes if chromatin fibers are compacted into layers. Multilayer chromatin explains many structural, topological, mechanical, cytogenetic, and functional properties of chromosomes. According to the basic properties of different soft-matter systems and the results discussed in this perspective, multilayer chromosomes can be considered emergent structures generated by anisotropic interactions between nucleosomes in the chromatin filament. Furthermore, it can be suggested that histones were selected very early in the evolution as the fundamental elements to pack DNA in plant and animal cells [141, 142], because they have unique structural properties that can produce complex emergent patterns at higher levels of chromatin organization. New structural techniques and experimental approaches will be necessary to analyze at high resolution the pristine structure of whole chromosomes and the exact path of the chromatin filament in the layers.

The author thanks Curt A Davey for useful comments about planar stacked layers of nucleosomes observed in different crystal structures studied in his laboratory.

No new data were created or analysed in this study.