Hierarchies in eukaryotic genome organization: Insights from polymer theory and simulations
© Iyer et al; licensee BioMed Central Ltd. 2011
Received: 7 January 2011
Accepted: 15 April 2011
Published: 15 April 2011
Skip to main content
© Iyer et al; licensee BioMed Central Ltd. 2011
Received: 7 January 2011
Accepted: 15 April 2011
Published: 15 April 2011
Eukaryotic genomes possess an elaborate and dynamic higher-order structure within the limiting confines of the cell nucleus. Knowledge of the physical principles and the molecular machinery that govern the 3D organization of this structure and its regulation are key to understanding the relationship between genome structure and function. Elegant microscopy and chromosome conformation capture techniques supported by analysis based on polymer models are important steps in this direction. Here, we review results from these efforts and provide some additional insights that elucidate the relationship between structure and function at different hierarchical levels of genome organization.
The term "genome" refers to the complete linear DNA sequence containing all of the hereditary material possessed by an organism. One of the goals of the human genome project was to determine the sequence of the 3 billion base pairs that constitute human DNA. Despite the wealth of information this tour-de-force of scientific enterprise has generated, it is increasingly becoming clear that the cellular function of the human genome is not merely determined by the linear ordering of its DNA base pairs. In fact, many of the functional aspects of the genome are governed by its three dimensional (3D) structure, which involves meters long DNA packaged into the limiting space of a micrometer sized cell nucleus.
The classic image of double stranded DNA (dsDNA) is that of a naked double helix. However, in eukaryotes, dsDNA seldom occurs in its naked form. Instead, it is packaged through a ubiquitous hierarchical process involving specialized proteins called histones. This packaging serves two purposes. First, it compacts the DNA allowing it to fit into the confines of the cell nucleus, and second, it controls the accessibility of DNA to cellular machinery for transcription, regulation, repair and recombination .
The primary structure of the eukaryotic genome consists of DNA wrapping ~1.7 times around histone octamers comprising of two copies of the four histone proteins H2A, H2B, H3, and H4 [4–7]. The combined histone octamer-DNA complex is called the nucleosome. This nucleosomal organization of DNA is considered to be the primary determinant to accessibility of genetic information. Although the atomic structure of the nucleosome has been resolved through X-ray crystallography [6, 7], the dynamics of nucleosomes remains far from fully understood. However, theoretical modeling and simulations are beginning to provide new insights into nucleosomal dynamics by answering questions related to: (1) how histone post-translational modifications and histone variants affect nucleosome structure and intra/inter-nucleosome interactions and (2) how nucleosomes undergo spontaneous conformation transitions between fully- and partially-wrapped states.
Recently, we have  employed all-atom molecular dynamics (MD) simulations and molecular docking to examine the structure of the H4 histone tail and its interactions with the acidic patch of nucleosomes. The H4 tail was found to exhibit a propensity for an α-helical conformation within a stretch of six residues encompassing the lysine 16 (K16) residue. Interestingly, this α-helical region interacted very strongly with the acidic patch; K16, in particular, mediated strong electrostatic interactions with the negatively-charged residues of the acidic patch. Acetylation of K16 diminished these binding interactions, suggesting a plausible mechanism by which post-translational acetylation of K16 could trigger chromatin unfolding . To examine nucleosomal dynamics beyond the time scales accessible to all-atom simulations, Sharma et al.  and Voltz et al.  have developed coarse-grained models of the nucleosome. The former has helped in the identification of important histone residues, termed "cold sites", that maintain the stability of the histone octamer and the latter has been used to compute long wavelength fluctuations of the nucleosome. Some effort has also been devoted to elucidating the energy barriers and the kinetic rate constants associated with the accessibility of DNA in the nucleosome  and explaining specific features of force-induced nucleosome unraveling observed in single-molecule experiments [12–14].
Beyond the single nucleosome level, contiguous nucleosomes separated by short sections of naked DNA called linker DNAs yield the classic beads-on-a-string structure depicted in Figure 2. The secondary structure of chromatin involves folding of this beads-on-a-string motif into a ~30-nm thick fiber called chromatin. While, in vivo imaging of cell nuclei has yielded little information on the secondary structure of chromatin, and has even brought into question the very existence of a 30-nm fiber [15, 16], electron microscopy of isolated nucleosomal arrays have been successful in imaging the transition from 10-nm beads-on-a-string structures to 30-nm condensed structures with increasing salt concentrations [15, 17]. Regardless of the gaps in our knowledge of the secondary structure of chromatin in vivo, this structure is expected to dictate the accessibility of DNA sequences for interactions with nuclear machinery and is also likely to play a role in the recruitment of histone modifying and remodeling factors to specific regions of the genome. Local packing of nucleosomes within the chromatin fiber could also potentially affect its interactions with distant portions of the fiber or other chromatin fibers through processes like interdigitation .
Despite decades of research, the internal structure of the 30-nm chromatin fiber remains controversial. This controversy arises because of two main reasons: (1) in vivo chromatin is too "messy" to be visualized with even the most advanced microscopy techniques  and (2) in vitro reconstituted nucleosome arrays are too large and flexible to be crystallized, and too compact at physiological conditions for their linkers to be fully resolved through microscopy. These limitations have led researchers to adopt indirect ways of deducing the internal structure of chromatin, using chemical cross-linking and single-molecule pulling techniques, with varying degrees of success. Based on the data, two types of models for the chromatin structure have been proposed that differ mainly in the location and configuration of the linkers. In the one-start solenoid model, linkers exhibit a strongly bent configuration and reside at the fiber interior [17, 19, 20]. In the two-start helix, linkers exhibit a straight or gently bent configuration, and they could reside at the periphery of the fiber in one version of the model  or inside the fiber (close to its axis) in a zigzag manner in another version [22–24].
Computational modeling has played an integral role in providing new insights into chromatin architecture. The main challenge in modeling chromatin is the vast degrees of freedom possessed by even small segments of the fiber (e.g., an array of 50 nucleosomes contains >1 million atoms). Hence, all-atom approaches demand prohibitive amounts of computational resources to converge to equilibrium structures. Further, the larger and the more flexible the system, the less adequate a description by one single equilibrium conformation because of the large variation in possible structures around thermal equilibrium. In other words, the systems become "fuzzy" and can only be described in terms of statistical averages. The advantage of this situation is that atomic-detail resolution may easily be abandoned for vast gains in computational speed in a coarse-grained model. To this end, a major focus of computational modeling has been on developing lower-resolution models or "force fields" of nucleosome arrays that still account for the energetic interactions and constraints between the different components of the arrays. A large effort in computational modeling also lies in developing efficient procedures for "sampling" low-energy conformations of the nucleosome array subject to appropriate force fields.
The E2A model is an extension of the two angle model  that also accounts for the cylindrical shape of the nucleosomes . Simulations based on the E2A model are ideally suited for studying long nucleosomal arrays as they overcome the difficulty of lack of knowledge of actual interaction potentials between nucleosomes by using model parameters obtained from experimental data. In particular, the E2A model has been used for studying the effects of variability in linker lengths and variability in linker histone occupancy . In combination with high-resolution light microscopy experiments these simulations promise a way to capture local chromatin structure in the range of 10 - 40 nm resolution .
The next class of models, in addition to including the features of the two-angle model, include the mechanics and electrostatics of the linker DNAs along with a simple treatment of inter-nucleosome interactions [34–36]. The linker DNA is treated as a discretized wormlike chain, possessing energy terms for stretching and bending of the bead-chain with an additional energy term for the relative twist angle between adjacent beads to account for the twisting rigidity of DNA  (Figure 3b). The linker beads are also assigned suitable effective charges that interact with each other via the Debye-Hückel potential . The inter-nucleosome interactions have thus far been treated as hard spheres, interacting with a hard-core attraction  or as cylinders interacting with Gay-Berne [35, 36] and Zewdie potentials . The nucleosome array conformations and packing ratios obtained using Monte Carlo simulations [35, 36] based on these models generally resemble those obtained by electron microscopy for moderately folded chromatin. These models  have also provided valuable insights into the stretching behavior of chromatin and yielded first estimates of the strength of internucleosome interactions by matching simulated force-extension curves to those obtained experimentally . Importantly, the models have revealed the sensitivity of the chromatin structure to the internucleosome interaction strength and linker lengths . Furthermore, it was observed that nucleosome arrays exhibit a zigzag structure without the linker histone and a more solenoidal structure for linker histone-bound arrays .
Very detailed models of local interactions within oligonucleosomes have been developed by Schlick and coworkers [41–49] (Figure 3c). The earliest model [41, 42] treated the nucleosome as a rigid cylinder (along with a small protrusion to represent the H3 histone tail) with hundreds of "pseudo" charges scattered uniformly on the surface. The magnitude of the charges were optimized to reproduce as closely as possible the electric field in the vicinity of the nucleosome. This approach allowed one to account for the salt-dependence in the inter-nucleosome interactions. The model was later refined by employing an irregular-shaped representation of the nucleosome  that was based on a more recent nucleosome crystal structure with all histone tails fully resolved . The model reproduced the experimentally observed [15, 17] compaction of the arrays with increasing salt concentration and indicated that the arrays maintain a zigzag morphology under monovalent salt conditions. Further, the simulations demonstrated that reduced electrostatic repulsion between the linkers is the main mechanism responsible for the folding of arrays at high salt. These models are also being used to study the dynamics of chromatin arrays, especially under different kinds of forces including torsional stresses .
Recently, this model was further improved by accounting for histone tail flexibility , linker histone binding , and effects of divalent ions . The tails were treated as coarse-grained bead-chains, where each bead represented five amino acid residues. The stretching, bending, and the electrostatic terms in the bead-chain were parametrized using an iterative procedure. The linker histone was coarse-grained as three charged beads rigidly bound at the nucleosome dyad with the magnitude of the charges optimized to reproduce the electric field of the atomic linker histone. Divalent ions were treated phenomenologically in terms of their effect on flexibility and electrostatic screening of the linker DNAs. A configurational-bias Monte Carlo approach was used to sample the tail configurations and translation, rotation and pivot moves were used to sample the global array configurations . This model helped elucidate the role of each histone tail, the linker histone and physiological salt condition in chromatin folding [45, 49]. Specifically, the H4 tails were found to mediate the strongest internucleosome interactions, the H3 tails mediated strong internucleosome interactions and screened electrostatic repulsion between the entering/exiting linkers, and the H2A and H2B tails mediated inter-fiber interactions . The linker histones constricted the linker entry/exit angle to bring alternate nucleosomes together. Divalent ions were also found to facilitate tight packing of nucleosomes by allowing a fraction of the linkers to bend and by strongly screening the linker repulsion at the fiber axis. Moreover, the model, in conjunction with sophisticated cross-linking experiments, confirmed the existence of a heteromorphic fiber containing both zigzag and solenoid conformations in the presence of additional divalent cations and linker histones . This model has also recently been used to reproduce the linker length dependence in the observed chromatin structures .
In summary, mesoscale models at varying levels of sophistication such as those discussed above are proving to be valuable tools for examining chromatin structure. The choice of which model to use is dictated by the amount of detail required and the amount of computational resources available. For instance, the detailed models accounting for histone tail flexibility and nucleosome geometry may provide the most accurate representation of short nucleosome arrays. For long arrays containing hundreds of nucleosomes, these models rapidly become computationally intractable and the intermediate-resolution models like the E2A model  become more suitable.
In the previous section, we discussed the secondary structure of chromatin from a static perspective. In reality, the chromatin fiber within the cell nucleus is present in a dynamic state -it is flexible over lengths much larger than the fiber diameter  and it is constantly being subjected to various kinds of remodeling activities, including histone modifications [54–56], sliding and depletion of nucleosomes [57–64], and incorporation of histone variants . There is strong evidence from light microscopy studies indicating that at the gene locus level the chromatin fiber is organized into loops . Studies on several multigene clusters conceive such loops as instrumental in bringing together distant enhancer and promoter regions crucial for gene activation, regulation and recombination [67–69]. Although the detailed mechanism of looping is not fully understood, there is no question that the "intrinsic" bending rigidity of the chromatin fiber dictates to a considerable degree the loop size-dependent statistical probability of two distant regions of the fiber coming into close proximity to form a loop.
Emerging evidence suggests that the relationship between flexibility and looping probability may be utilized by cells for gene regulation. Specifically, the idea that modulation of flexibility through remodeling processes like acetylation of histones alters the probability of interactions between a remote enhancer and cognate gene by means of looping has been successfully applied to explain gene regulation of the Hoxd gene cluster and the β-globin locus . Strong support for this idea also comes from recent experiments in the Murre lab that show large-scale conformational changes in the IgH-locus during B-Cell development accompanying genome-wide and locus-specific histone modifications and nucleosome depletion events [66, 71, 72]. Thus, it seems that chromatin remodeling events, apart from modulating the local structure of chromatin and DNA accessibility, could lead to changes in the higher-order folding of chromatin through its effects on macroscopic properties of the fiber such as its flexibility.
So far, little effort has been devoted to investigating the intrinsic flexibility of chromatin and the associated looping probability as a function of loop size; especially how the flexibility correlates with external conditions like monovalent/divalent salt concentration and system parameters like nucleosome repeat length, DNA wrapping angle, histone variants, histone modifications and presence/absence of linker histones. Chromatin flexibility as characterized by its persistence length, L p , seems to exhibit large variations, depending on the experimental method used for its determination and the conditions and type of chromatin investigated. For example, single-molecule pulling of nucleosome arrays at low salt  and in vivo looping driven cross-linking/recombination assays [53, 73] measure L p as low as 30-50 nm while analysis of fluorescent markers in the genome of erythrocytes measure L p of 100-200 nm [74, 75]. However, it is important to review these results in light of a recent fundamental study  indicating that standard definitions of persistence length as used in these studies may not describe the local intrinsic flexibility of chromatin.
Aumann et al.  recently examined L p of nucleosome arrays using Monte Carlo simulations of a mesoscale model of chromatin. The persistence length was obtained from the decay in the correlation of the tangent vector representing the local fiber axis. It was found that L p decreased strongly with increasing nucleosome repeat length and increasing entry/exit angle, and that binding of the linker histone led to an increase in L p , consistent with experimental observations of linker histone deficient and inclusive chromatin. A comparison between the magnitude of the bending and elastic rigidity suggested that chromatin is much easier to bend than stretch, leading to the interesting hypothesis that it may be easier for the cells to pack chromatin via tight loops rather than by linear compression of the fiber.
We have recently begun to examine the mechanisms behind the conformational collapse in the IgH - locus observed by Murre and coworkers during B-cell development . We hypothesize that chromatin remodeling events (nucleosome depletion and histone modifications [71, 72]) introduce flexible "hinges" within the chromatin fiber causing it to collapse (Figure 4). By treating the chromatin fiber as a worm-like chain (WLC) with fixed contour length L c and variable persistence length L p , we showed that the compaction, as characterized by the ratio of the final to initial mean square end-to-end distances, is given by the ratio of the final to initial persistence length in the ideal chain limit (see Appendix for complete derivation). Using the initial and final L p observed by Heermann and coworkers , this ratio becomes 1/2, indicating that the gene locus compacts by ≈30%. Interestingly, 3D distance measurements in the IgH-locus during progressive stages of B-cell development show a similar ~30% decrease in locus dimensions in response to remodeling events . To examine the functional implications of this collapse, we used a simple Flory type argument to show that the collapse leads to an 8-fold increase in the number of binary interactions n b within the gene locus. Assuming that the frequency of promoter-enhancer (P-E) interactions is proportional to n b , we conclude that the conformational collapse facilitates transcriptional regulation and/or recombination by allowing for higher probability of P-E interactions. Thus, the chromatin remodeling induced changes in the persistence length offers one possible mechanism for governing the conformational state of the gene locus and consequent modification of the functional state of the cell (Figure 4).
Interestingly, the condensed higher-order structures of chromatin, namely chromosomes, were observed as distinct entities during mitosis as early as the 19th century, much before the primary and secondary structures of chromatin were known to exist. However, disappearance of this condensed structure during interphase and the underlying chromatin organization remained a mystery till the late 20th century . In the beginning two different models were proposed for the interphase chromatin organization: (1) random organization akin to a bowl of spaghetti without any apparent structure and (2) organization in territories that later condense during mitosis to form distinct chromosomes . The earliest clues to deciding between the two models came from experiments shining a laser light onto a specific volume of the cell nucleus and observing the effects of the consequent damage on replication . It was found that only a few chromosomes were affected by the laser light, indicating existence of distinct chromosome territories within the cell nucleus. This conjectural evidence has now been confirmed beyond doubt by advances in imaging techniques like fluorescent in situ hybridization (FISH) [82–84] and chromosome painting [82, 83].
While there is concrete evidence for the existence of territories, the internal architecture of chromatin within chromosome territories, the interaction between territories, and the organizational principles governing their formation remain poorly understood . Results from MC simulations of confined polymer chains suggest that territory formation could arise from simple, non-specific entropic forces and from segregation of long chains attempting to conserve their topological state while undergoing confined Brownian motion [86, 87]. Increasingly, insights from such coarse-grained polymer models coupled with experiments are being used to examine the internal architecture of chromosome territories.
Two predominant themes recur during discussions of higher-order structures of interphase chromosomes: (1) formation of loop structures and (2) confined fractal organization. Interestingly, each of these themes derive their support from different kind of experiments. The loop structure theme is predominantly supported by light microscopy experiments [75, 88–90] and the fractal theme by chromosome conformation capture , small angle neutron scattering  and tracer diffusion  experiments. Although the two themes are not mutually exclusive, it is often possible to classify the polymer models based on them. Here, we follow this classification in discussing the polymer models and their central features.
The proposed loop models show a power-law behavior for geometric distances versus contour length at scales <10 Mbp, with scaling exponents ranging from ν = 1/3-1/2. This suggests two different internal structures at work: (1) equilibrium/globular state at short scales and (2) random looped structure at large scales. The transition between these structures is not clearly understood. In contrast, the confined fractals do not encounter these difficulties as they propose a single size exponent at all scales. Recent advances in chromosome conformation capture  have enabled genome-wide measurements of probability of formation of loops as a function of the size of the loops in the human genome (Figure 6). Comparison of the probability measurements with confined freely jointed chain (FJC) simulations indicate that the human genome may be organized as a fractal globule with size scaling exponent ν = 1/3 . However, we note that the agreement of the probability measurements is observed only in the 1-10 Mbp range (Figure 6).
The simple arguments we used for the persistence length calculations inherently assume that the scaling exponent ν = 1/3 is universal within the human genome. However, simulations investigating properties of compact polymers have pointed out that the chromatin fiber, although in a state that resembles the fractal globular state of the compact polymers with ν = 1/3, shows deviations from these type of polymers . Specifically, the moment ratios calculated from intrachain distance distributions for chromatin are observed to be different from that observed for globular state . Instead, the intrachain distributions are observed to be reasonably approximated by the random loop model . However, the random loop model also shows that the distances become independent of genomic separations, i.e., ν → 0. The anomaly in these results can possibly be explained by the argument that in the presence of loops it is inappropriate to use the intrachain distances to determine the scaling exponent ν. Instead, the radius of gyration, a measure of size for closed structures like rings, may be a more appropriate quantity for determining the exponent.
We end this section with a perspective on combined use of the information at the gene locus level and the chromosomal level to infer qualitative aspects governing the functional state of the cell. In the previous section, we have shown that the decrease in persistence length L p causes a collapse and consequent increase in the concentration (density) of segments in a gene locus, via Eqs. (2) and (3). In contrast, in this section, we have shown using Figure 7 that a decrease in L p is accompanied by a decrease in the linear mass density c (since c and Φ are inversely related). The paradoxical nature of this result indicates that chromatin remodeling events, which lower the effective persistence length of chromatin, could have opposing effects on gene-related activities like transcription. On one hand, locus-wide remodeling could open up regions for transcription and lead to increased frequency of interactions between promoters and enhancers; while on the other hand, genome-wide remodeling events could cause genome folding leading to formation of inaccessible, dense regions and potential suppression of transcription. Thus, nucleosome depletion, in addition to its established role in gene activation, may also lead to repression. Such opposing effects are in fact observed when the nucleosomal content is reduced through H4 depletion in the yeast genome, resulting in increased expression in 15% and reduced expression in 10% of genes .
Understanding genome organization within the cell nucleus and its functional implications are fundamental issues in modern cell biology. In this article, we have reviewed the incisive role played by polymer theory and simulations in providing new insights into the structure of the genome at various levels of organization, ranging from packing of nucleosomes into the 30-nm chromatin fiber to the higher-order folding of chromatin via looping into chromosomes. Despite enormous progress in both experimental and computational fronts, a comprehensive model of genome organization at the chromatin and chromosome level is still lacking. Below we list four main challenges that computational models would have to overcome to tackle this highly intriguing and important problem.
First, genome organization is heterogeneous. At the chromatin level, this heterogeneity arises from variations in the nucleosome repeat length, nucleosome composition (histone modifications, histone variants, and linker histones), and potential entrapment of chromatin in metastable states. Most models of chromatin, including those developed in our group, tend to assume uniform nucleosome positioning and composition. Only recently have studies begun to examine the effects of chromatin heterogeneity; preliminary work in this direction has already yielded some very interesting results. At the chromosome level, the size, shape and location of chromatin loops is not fixed. Thus, a key aspect towards examining heterogeneity in chromosome models would be through more realistic modeling of the interactions across chromatin fibers, e.g., allowing loops to dynamically form and break according to the associated energetics of loop formation in chromosome models, as opposed to "fixing" looping points and loop sizes.
Second, the chromatin fiber and chromosomes are highly dynamic. Nucleosomes are constantly being displaced, modified, dissolved and reformed through various mechanisms. At the higher scale, chromatin loops are continuously being broken and reformed. In addition, a range of nuclear proteins including transcription factors and architectural proteins dynamically bind and dissociate from chromatin . Such binding/dissociation events could affect chromatin structure and in turn may be affected by chromatin structure. Most models currently look at the genome from a static perspective. Incorporation of the above mentioned kinetic features into polymer models coupled with structural heterogeneity would be a crucial step in building more comprehensive models of the genome.
Third, development of accurate, coarse-grained models of chromosomes poses a grand challenge. One potential route is via "multi-scale" approaches, where each level represents coarse-graining of the previous, higher-resolution level. Such an approach generally involves the use of potential of mean forces (PMFs) for treating the interactions between coarse-grained subunits. The PMFs account for effects of the degrees of freedom that are "averaged out" during each coarse-graining step. The challenge is to maintain self-consistency from one level to another, as the PMFs are valid only when there is a clear separation of length and time scales across the levels . When there is no clear separation of time/length scales, one might need to resort to "multi-resolution" approaches. In this approach, different portions of the system are treated at different resolutions, depending on their relative importance to the phenomena being investigated (in the same spirit as the QM/MM method  employed in enzyme catalysis). The challenge here is to identify the portions of the system (or degrees of freedom) important enough to be treated at higher resolution as opposed to those regions that can be treated at lower resolution. Another challenge is developing suitable potentials for linking low- and high-resolution regions.
Fourth, little modeling effort has been invested in determining structure-function relationships in genome organization. A definitive connection between structure and function comes from the observation that gene density is high in largely decondensed euchromatin and low in highly condensed heterochromatin . Further, gene-rich and gene-poor regions are found to be physically separated from each other [105, 106]. Specifically, in human lymphocytes gene-rich regions are positioned in the nuclear interior while the gene-poor regions are positioned towards its periphery . Apart from gene density patterns, the molecular species responsible for nuclear processes and gene expression are connected through a complex and extensively coupled network giving rise to functional coupling between macromolecular structures and compartments [108, 109]. Though many aspects of these studies are been driven by the experiments, polymer theory and simulations could contribute to better understanding of such structure-function relationships.
To study the generic effects of changes in the persistence length of the chromatin fiber on the conformation of the gene locus, we have treated the chromatin fiber as a worm-like chain (WLC) with fixed contour length L c . The WLC is the standard model used to describe semi-flexible polymer chains, and it has been successfully used to explain the conformational properties of double-stranded DNA and other biopolymers . Though this model is too idealized to represent in vivo chromatin (as it neglects fiber-fiber interactions, existence of stable loops and heterogeneous flexibility), it can nonetheless provide qualitative insights into the magnitude of the conformational changes.
Ratio of mean-square radius of gyration of chains
In the case of the IgH locus, this translates to an 8-fold increase in n b .
where ν is a scaling exponent that connects the size of the chain to its number of repeat units. Note that the fractal dimension, d f , is the exponent that connects the mass of the chain M to its size via and that M is also proportional to the number of repeat units M ~ N r ~ (C s )1/ν (from Eq. 5). Hence, ν is the inverse of the fractal dimension ν = 1/d f and determines the organization of the FJC at all length scales.
Since the contour length L c = d/c, where d is the genomic distance in bp and c is the linear mass density in bp/nm, Φ increases with decreasing linear mass density.
The contour length within the FJC framework can be partitioned arbitrarily, i.e., if the contour length of a structure is 100 μm then it can be partitioned as, say, 100 repeat units each of 1 μm length or 1000 repeat units of length 0.1 μm. The flexibility of the contour length partitioning allows for choice of either number of repeat units N r or the persistence length L p as a variable in Eq. (6). However, the contour length L c according to Eq. (6) is determined by the product of the folding index Φ and size C s . This constraint has to be satisfied when L p is chosen as the variable and the contour length in the framework changes with the change in the folding index. Thus Φ can be used like a lens of variable resolution to view genome organization at different length scales.
We thank Prof. Cornelis Murre of the Division of Biological Sciences at UCSD for many delightful and insightful discussions on organization of chromosomes.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.