The relationship between nucleosome positioning and higher-order genome folding

Eukaryotic genomes are organized into 3D structures, which range from small-scale nucleosome arrays to large-scale chromatin domains. These structures have an important role in the regulation of transcription and other nuclear processes. Despite advances in our understanding of the properties, functions, and underlying mechanisms of genome structures, there are many open questions about the interplay between these structures across scales. In particular, it is not well understood if and how 1D features of nucleosome arrays influence large-scale 3D genome folding patterns. In this review, we discuss recent studies that address these questions and summarize our current understanding of the relationship be-tween nucleosome positioning and higher-order genome folding.


Introduction
Eukaryotic genomes are organized into 3D structures across different scales.This allows large genomes to be efficiently compacted in order to fit in micron-sized cell nuclei.The spatial organization of eukaryotic genomes is also thought to have a key role in regulating nuclear processes, including transcription, replication, DNA repair, and chromosome segregation.At the largest scale, chromosomes occupy relatively distinct territories within the nucleus [1].At an intermediate level, chromosomes are further organized into compartments and Topologically Associating Domains (TADs).Compartments reflect separation of euchromatin and heterochromatin, whereas TADs are local domains formed by an active process of loop extrusion [2].On the smallest scale, eukaryotic genomes are organized into nucleosome core particles, which consist of 147 base pairs of DNA wrapped around a histone octamer [3].Nucleosome core particles are connected by short DNA linkers in nucleosome arrays, which form the characteristic w10 nm "beads-on-a-string" structures [4].
Since the presence of a nucleosome at gene promoters prevents transcription initiation, nucleosome positioning is important for the regulation of transcription [5].In line with this regulatory role, the positioning of nucleosomes along the genome is not random.For example, the transcription start sites of genes are characterized by a nucleosome-free region (NFR), which is flanked by regularly spaced and phased nucleosome arrays that extend into the gene body [4,6,7].A similar pattern can be observed at binding sites of regulatory proteins, such as the CCCTC-binding factor (CTCF) [8].The positioning of nucleosomes is regulated by ATP-dependent chromatin remodelers, which facilitate nucleosome sliding, exchange, and removal [9,10].The action of chromatin remodelers and histone-modifying enzymes results in a large diversity of nucleosome arrays, with variations in DNA linker length, incorporation of histone variants and linker histone H1, and histone modifications [11].
Our understanding of the mechanisms that drive 1D features of nucleosome arrays, such as the positioning and properties of nucleosomes, has advanced substantially in the last decades.In addition, we have an increasingly detailed understanding of the higher-order 3D structures into which eukaryotic genomes are organized and the underlying molecular mechanisms.However, the interplay between the processes that organize eukaryotic genomes across different scales is less well studied and understood.In this review, we summarize recent progress in this area.We discuss how nucleosomes fold into small-scale structures and how their features influence the formation of basic chromatin domains that form across eukaryotes, as well as larger-scale compartments and TADs that characterize the genomes of higher eukaryotes.

Heterogeneity of nucleosome structures
Early in vitro studies of the arrangement of w10 nm nucleosome arrays into higher-order structures have described compact, highly organized w30 nm chromatin fibers [12e14].These studies were based on structural analysis of reconstituted chromatin with strong, repetitive nucleosome-positioning DNA sequences.More recent analyses have revealed that such highly organized chromatin fibers encompassing oligo-nucleosome arrays do not exist in vivo [15e17].Ordered nucleosome structures appear to be more dominant on a smaller scale.In particular, specific tri-and tetra-nucleosome structures have been described as a basic folding motif in higher-order structures [18e20].Structural analyses of tetra-nucleosomes have shown that they can adopt different stacking configurations, which are dependent on the DNA linker length [18,21,22].Sequencing-based methods have identified more diverse tri-and tetranucleosome motifs, including both zigzag and solenoid geometries [20,23,24].The organization of chromatin at the level of tetra-nucleosomes may have a regulatory function.It has been suggested that more compact zigzag configurations are enriched in inactive regions of chromatin, whereas less compact solenoid-like structures predominantly form in active chromatin regions [23].Moreover, experiments based on single-molecule force spectroscopy indicate that tetra-nucleosomes form relatively stable structures that inhibit transcription and require remodeling by the histone chaperone Facilitates chromatin transcription (FACT) to enable productive transcription [25].
At a larger scale, nucleosome structures are thought to be less regular and more heterogeneous.Electron tomography data have shown that nucleosomes form flexible chains with diameters between 5 and 24 nm [26].Similar structures have been identified in superresolution microscopy experiments and described as "nucleosome clutches", which correspond to clusters of nucleosomes of various sizes and densities [27].Heterochromatin regions are associated with relatively large and dense clutches enriched in histone H1, whereas euchromatin regions are characterized by smaller and less compact clutches.Clutches are separated by NFRs, which have been shown to be important for regulating the size of nucleosome clutches [28].The specific conformation of chromatin structures at this scale is thought to be further influenced by variations in the length and bending of DNA linkers, the density of linker histones, and histone acetylation levels, which all contribute to their heterogeneity [28e30] (Figure 1).At a yet larger scale, super-resolution microscopy data have described dynamic chromatin structures with a 160 nm diameter [31], which may result from progressive clustering of nucleosome chains and clutches.

The role of nucleosome positioning in the formation of small-scale chromatin domains
Chromosome Conformation Capture (3C) methods, which use proximity ligation coupled with highthroughput sequencing to analyze the higher-order conformation of chromatin [32,33], have shown that eukaryotic genomes are organized into self-interacting chromatin domains that span a wide range of sizes [2].Several processes have been implicated in the formation of these domains, including nucleosomeenucleosome interactions, transcription, loop extrusion, and phase separation [2].However, since these processes are interconnected, it is difficult to disentangle their independent effects on chromatin organization and to define the fundamental principles that drive the formation of chromatin domains.Reductionist approaches based on in vitro reconstitution experiments provide a useful tool to identify cause-consequence relationships that are difficult to disentangle in the complex cellular milieu [34].Saccharomyces cerevisiae offers a suitable model system for reconstitution studies due to the relatively small size of its genome and chromatin domains, which are w2e10 kb in size [35].By combining salt gradient dialysis of a genome-wide DNA template and purified histones with incubation of transcription factors and ATP-dependent chromatin remodelers, it is possible to reconstitute in-vivo-like chromatin, characterized by regularly spaced and phased nucleosome arrays [36,37].Recently this reconstitution approach has been coupled with high-resolution 3C analysis to study the requirements for chromatin domain formation [38].Surprisingly, these analyses have revealed that the reconstitution of regular, in-vivo-like nucleosome arrays is sufficient for the formation of S. cerevisiae chromatin domains.Consistent with the above-mentioned nucleosome clutches [27,28], the boundaries of the reconstituted domains correspond to NFRs at transcription factor binding sites.Moreover, the width of the NFRs correlates positively with insulation strength.Comparison of remodelers with and without regular nucleosome spacing activity has revealed that domain formation does not only require a NFR, but also regularly spaced and phased arrays surrounding the NFR (Figure 2).These findings are consistent with previous computational models, which have shown that nucleosome positioning data is sufficient to predict in vivo patterns of higherorder chromatin folding in S. cerevisiae [39,40].In addition, more recent simulations have shown that binding of transcription factors and associated changes in nucleosome positioning can drive the formation of microdomains in mammalian chromatin fibers, suggesting a conserved role for nucleosome positioning in domain formation [41].
The compaction of chromatin domains is thought to depend on several factors.It has been shown that longer DNA linkers, which are enriched in heterochromatin, are associated with more compact domains compared to short DNA linkers, which are more frequently found at active genes [42e44].A role for DNA linker length in domain compaction is consistent with recent modeling [41] and reconstitution approaches [38].The latter have shown that incubation of reconstituted chromatin with different remodelers, which each set a distinct linker length [37], results in domains with distinct compaction levels; remodelers that set longer linkers, form more compact domains [38].In addition to DNA linker length, histone modifications, especially acetylation, and the incorporation of H1 linker histones, are thought to modulate chromatin domain compaction [41,43,44].
The observation that regular nucleosome positioning suffices for the formation of small domains in S. cerevisiae demonstrates that transcription and loop extrusion are not per se required for domain formation at a relatively small scale.This indicates that the formation of these domains is dependent on nucleosomeenucleosome interactions, likely mediated via histone tails [45,46].In this context, the formation of a domain boundary at NFRs could be explained by the stiffness of nucleosome-free DNA [47], which may therefore act as a rigid spacer over which nucleosome cannot interact, thus separating two regions of interacting nucleosomes into separate domains.Notably, it has recently been suggested that the intrinsic property of nucleosome arrays to undergo liquideliquid phase separation contributes to the formation of chromatin domains.A recent study based on cryo-ET and deep-learning-based 3D reconstruction has provided a model for the stepwise formation of phase-separated chromatin structures.This model involves the initial phase separation of tetranucleosomes into relatively loosely packed, irregular condensates, which subsequently form more compact, spherical condensates, in a process that is strongly catalyzed by the presence of H1 histone [48].Interestingly, a recent report based on imaging and modeling experiments suggests that at the scale of 150 nm The role of nucleosome positioning in chromatin domain formation.In S. cerevisiae, the formation of regularly spaced and phased nucleosome arrays is sufficient for the formation of chromatin domains, of which insulation strength scales with the width of the NFR at the domain boundary.NFR = nucleosome-free region.
domains, chromatin behaves like a liquid, while it has more solid-like properties at a larger micrometer scale [49].This configuration may offer a mechanism for the balance between chromatin flexibility, which is required for various dynamic nuclear processes, and chromatin stability, which supports genome integrity.

The relationship between nucleosome positioning and large-scale chromatin domains
The genomes of higher eukaryotes are organized into larger chromatin domains, including compartments and TADs.Compartments span a wide range of sizes, extending from a few kb to several Mb.The formation of compartments is thought to depend on (micro-)phase separation, possibly via the two-step mechanism described in the previous section.The specific associations between regions of euchromatin and heterochromatin into A and B compartments, respectively, depends on molecular affinity between the factors that associate with these distinct chromatin regions; these associations subsequently result in the relative separation of A and B compartments in the nucleus [50].
TADs are usually between w0.1 and 1 Mb in size and are formed by an active process of loop extrusion, during which Cohesin complexes reel in the chromatin fiber and thereby form progressively larger loops, until they encounter CTCF molecules.This allows Cohesin to mediate interactions between chromatin regions located within, but not beyond CTCF-binding sites (CBSs), and thereby results in the formation of relatively insulated domains that are demarcated by CBSs [51,52].The interaction between Cohesin and CTCF has been shown to depend on a specific CTCF motif that is required for efficient stalling of extruding Cohesin molecules [53].Although this provides a satisfactory explanation for the formation of TADs, it is of interest that in addition to its ability to specifically interact with Cohesin, CTCF also has a very strong effect on nucleosome positioning, which is more prominent than at transcription start sites, and extends to up to 20 nucleosomes [8].The nucleosomes that flank CBSs are characterized by a relatively short DNA linker length, which is anti-correlated with the binding strength of CTCF [54].The positioning of nucleosomes at CBSs is mediated by the activity of ISWI chromatin remodelers.Several reports have shown that perturbation of SNF2H (SMARCA5), the ATPase of ISWI complexes, affects nucleosome positioning and protein binding at CBSs [55e57].In addition, a recent report suggests that FACT modulates the accessibility and local insulation at CBSs in transcribed regions [58].
Although it has been shown that perturbation of the above-mentioned remodelers affects insulation at TAD boundaries, it remains unclear whether this reflects a direct role for nucleosome positioning in TAD insulation or reduced CTCF binding.It has recently been reported that deletion of the ISWI accessory subunit BPTF impacts chromatin accessibility at CBSs, while only modestly impacting binding of CTCF.Interestingly, perturbation of BPTF results in reduced Cohesin occupancy and insulation at CBSs [59].Although the described effects are relatively modest and only apply to a subset of CBSs, these findings suggest that the accessibility of CBSs has an independent role in TAD insulation.In addition to a possible role in Cohesin stalling, nucleosome positioning might also shape TAD structures by influencing the recruitment of Cohesin to chromatin.Since it has been shown that Cohesin loading in yeast is dependent on chromatin remodeling by RSC [60], it is conceivable that the arrangement of nucleosomes influences the recruitment and subsequent extrusion patterns of Cohesin (Figure 3).

Concluding remarks and perspectives
Accumulating evidence suggests that the configuration of nucleosomes at the scale of a few nanometers The influence of nucleosome positioning on loop extrusion.The positioning of nucleosomes at CTCF-bound TAD borders affects CTCF binding levels and may also have a direct role in regulating loop extrusion.In addition, nucleosome positioning may influence the recruitment and extrusion trajectories of Cohesin molecules.TAD = topologically associating domain; CBS = CTCF-binding site.influences higher-order genome folding into chromatin domains at the scale of hundreds of nanometers.However, many questions remain regarding the interplay between the mechanisms involved and their functional roles in nuclear processes, including whether and how nucleosome positioning directs Cohesin movement and stalling on chromatin and how nucleosome arrangements influence compartmentalization.As described, recent progress involves the development of innovative methods to study 3D genome organization in everincreasing detail based on high-resolution microscopy, live-cell imaging, genomics and biochemistry.Integration of these techniques provides a powerful approach to address open questions and obtain a more holistic understanding of the relationship between genome structure and function across scales.

Figure 1 The
Figure 1