Spatial and topological organization of DNA chains induced by gene co-localization

Transcriptional activity has been shown to relate to the organization of chromosomes in the eukaryotic nucleus and in the bacterial nucleoid. In particular, highly transcribed genes, RNA polymerases and transcription factors gather into discrete spatial foci called transcription factories. However, the mechanisms underlying the formation of these foci and the resulting topological order of the chromosome remain to be elucidated. Here we consider a thermodynamic framework based on a worm-like chain model of chromosomes where sparse designated sites along the DNA are able to interact whenever they are spatially close-by. This is motivated by recurrent evidence that there exists physical interactions between genes that operate together. Three important results come out of this simple framework. First, the resulting formation of transcription foci can be viewed as a micro-phase separation of the interacting sites from the rest of the DNA. In this respect, a thermodynamic analysis suggests transcription factors to be appropriate candidates for mediating the physical interactions between genes. Next, numerical simulations of the polymer reveal a rich variety of phases that are associated with different topological orderings, each providing a way to increase the local concentrations of the interacting sites. Finally, the numerical results show that both one-dimensional clustering and periodic location of the binding sites along the DNA, which have been observed in several organisms, make the spatial co-localization of multiple families of genes particularly efficient.

💡 Research Summary

The paper tackles a fundamental question in genome biology: how the spatial organization of DNA contributes to the formation of transcription factories—discrete foci where highly transcribed genes, RNA polymerases, and transcription factors concentrate. The authors adopt a minimalist yet powerful thermodynamic framework in which a chromosome is represented as a worm‑like chain (WLC), a semiflexible polymer characterized by its persistence length. Along this polymer they embed a sparse set of “designated sites,” which correspond to genomic loci that are known to cooperate functionally (e.g., promoters, enhancers, or binding sites for a common transcription factor). The key hypothesis is that whenever two designated sites become spatially proximate, they experience a short‑range attractive interaction, mimicking the physical tethering mediated by transcription factors or co‑activators.

Thermodynamic analysis
Starting from the WLC Hamiltonian, the authors add a pairwise interaction term that is non‑zero only when the Euclidean distance between two designated sites falls below a cutoff. By integrating over all polymer conformations, they derive an effective free‑energy functional that depends on the density of interacting sites, the strength of the attraction (ε), and the concentration of mediating factors (transcription factors). Minimizing this functional reveals a micro‑phase separation: the interacting sites segregate into dense clusters embedded in a dilute matrix of non‑interacting DNA. This result provides a physical justification for the emergence of transcription factories without invoking active transport or motor‑driven clustering; the factories arise spontaneously when the product of site density and interaction strength exceeds a critical value. The analysis also predicts that transcription factors, by providing the attractive energy, are natural candidates for the mediators of this phase separation.

Numerical simulations
To explore the richness of the model beyond mean‑field theory, the authors perform extensive Monte‑Carlo and molecular‑dynamics simulations of a discretized WLC with varying parameters: chain stiffness (persistence length), spacing of designated sites, interaction strength, and the number of mediators. Four distinct structural regimes emerge:

1. Dispersed regime – interacting sites are scattered throughout the polymer; no clusters form, corresponding to low transcriptional activity.
2. Clustered regime – a few large aggregates appear, each representing a transcription factory. This regime is favored when the average linear distance between designated sites is small and ε is moderate.
3. Periodic/Ring regime – designated sites are arranged quasi‑periodically along the contour, leading to a series of equally spaced clusters. The polymer adopts a helical or ring‑like conformation, allowing multiple factories to be positioned in a regular pattern.
4. Network regime – multiple clusters interconnect, forming a complex topological network. This state is promoted by high supercoiling, knotting, or strong bending constraints, which bring distant segments into proximity.

Transitions between these regimes are sharp, reflecting underlying phase transitions in the free‑energy landscape. Importantly, the simulations reveal that two organizational principles observed in real genomes dramatically enhance clustering efficiency. First, one‑dimensional clustering—where designated sites are grouped in contiguous blocks along the genome—reduces the contour distance that must be bridged, so even weak attractions can nucleate stable factories. Second, periodic placement—regular spacing of interacting loci—creates a scaffold that naturally partitions the polymer into repeat units, each capable of hosting a factory. Both patterns are documented in bacterial operons, yeast gene clusters, and mammalian topologically associating domains (TADs).

Topological considerations
The model also incorporates topological constraints such as knots and supercoils. Simulations show that knots act as “attraction hotspots”: designated sites that happen to lie near a knot are more likely to co‑localize, effectively seeding a factory. Supercoiling compresses the polymer, decreasing the average Euclidean distance between distant loci and thereby lowering the critical ε required for phase separation. These findings suggest that the mechanical state of the chromosome (e.g., torsional stress, nucleoid compaction) can modulate transcription factory formation in vivo.

Biological implications
From a cellular perspective, the framework predicts that cells can regulate the number, size, and spatial distribution of transcription factories by tuning transcription‑factor concentrations or by altering chromatin stiffness (through histone modifications, nucleosome positioning, or architectural proteins like CTCF). Because the model links a quantitative parameter (ε·ρ, where ρ is the linear density of designated sites) to the emergence of factories, it offers a testable hypothesis: perturbing the expression of a specific transcription factor should shift the system across the predicted phase boundary, leading to observable changes in factory number or gene‑expression patterns. Moreover, the authors speculate that disease states characterized by aberrant transcription factor levels (e.g., certain cancers) may correspond to a mis‑tuned phase, resulting in either hyper‑clustering (excessive transcription) or factory dissolution (global transcriptional repression).

Conclusions
The study delivers three major insights: (1) transcription factories can be understood as a thermodynamically driven micro‑phase separation of interacting genomic loci; (2) the polymer model exhibits a rich set of topological phases, each providing a distinct route to increase local concentration of co‑regulated genes; and (3) one‑dimensional clustering and periodic placement of binding sites—features observed across diverse organisms—greatly facilitate efficient spatial co‑localization. By bridging polymer physics, statistical thermodynamics, and genome organization, the work establishes a quantitative platform for future experimental validation and for designing synthetic gene circuits that exploit spatial clustering to achieve precise transcriptional control.