Getting gene expression “just right” Authors: Paul C. Whitford Published: 2025 The cell is a chaotic environment where seemingly random events give rise to robust emergent properties. For this, there must be organizing principles that govern the dynamics across scales, ranging from individual molecules to communities of organisms. The most well characterized example is protein folding, where decades of effort have provided a quantitative bridge between physical concepts and empirical measurements. These studies have shown that the energy landscapes must be smooth, or funnel-like (1), in order for folding to occur on physiological timescales. While this has provided a comprehensive view of dynamics in naturally occurring proteins, many biomolecules do not have well-defined folded/native structures. For example, “intrinsically disordered” molecules adopt dispersed ensembles of conformations, where biological function often requires transitions between ordered and disordered states (2). In recent years, there has been much interest in how these disordered molecules/fragments can form condensates, which facilitate membrane-less compartmentalization in the cell (3). Chromosomes provide additional examples for how a balance between order and disorder is necessary for biological function. These meter-long molecules are able to organize sufficiently well that they can remain confined to micrometer dimensions in the nucleus. While this compaction is fascinating on its own, it is even more impressive that this form of DNA allows for efficient control of gene expression. That is, a “folded” chromosome is not a static structure. Rather, it is constantly undergoing reorganization at many scales that is driven by both active (energy consuming) and passive (e.g. diffusive) processes (4). In PNAS, Zhu et al. (5) present a theoretical model that illustrates how disorder, biomolecular interactions, condensation, and DNA reorganization contribute to gene expression. Chromosomes contain hierarchical levels of organization that provide protection to the genome, while allowing each cell to precisely access and express individual genes (6). At the smallest scale, short stretches of DNA (∼150 base pairs) wind tightly around bundles of proteins called histones (Fig. 1). Formation of these compact units (nucleosomes; ∼10 nm) contracts the DNA in length, but not nearly enough for it to fit within the nucleus. Accordingly, the volume is then reduced through internucleosomal organization (∼50 nm), where further ordering leads to a structured chromosome (∼1 to 10 μm). Together, these levels of organization result in a six order-of-magnitude compaction of the DNA molecule. Hierarchical order in a chromosome. In order to package into the nucleus, DNA (blue) is tightly wrapped around histone proteins (pink), together forming a nucleosome [Left panel; RCSB entry 7kbf (7) shown]. Nucleosomes can then locally arrange in various forms (Middle panel). In the study of Zhu et al., a model is developed where a single “particle” represents roughly 20 nucleosomes (Right panel). With this model, they are able to identify factors that allow for promoter and enhancer regions (red, yellow) to efficiently come into contact, a process that is essential for the control of gene expression. Within this hierarchy exist stretches of DNA known as enhancers and promotors. When specialized proteins (i.e. transcription factors, or TFs) bind to a promoter, it can be brought into proximity of an enhancer. As the name would suggest, this enhances/amplifies the level of gene expression. Intuitively, one would expect enhancer–promoter pairs will rapidly encounter each other, by chance, if they are close in sequence (i.e. at short genomic distances; ∼104 base pairs). However, it is less obvious how these pairs can effectively find one another when they are separated by long genomic distances (e.g. ∼106 base pairs). In addition, the crowded nuclear environment may impede the search process. One way that nature can overcome these challenges is by forming chromatin loops (8). Motor proteins pull on the DNA, similar to how one would tie their shoelaces. While this can help aid enhancer–promoter pairing, these chromosomal loops (called Topologically Associated Domains, or TADs) can be extremely large (9), suggesting that additional physical factors contribute to the dynamics. In the study of Zhu et al., the authors consider the length scales associated with chromatin dynamics and propose a theoretical model that can be used to explore the impact of TF interactions on the encounter probability of enhancer–promoter pairs. In their model, they use a “beads-on-a-string” representation of chromatin, where each bead represents ∼3,000 base pairs (Fig. 1). They additionally introduce pseudoparticles that represent the transcription machinery and related cofactors. Rather than aim to describe specific protein–DNA interactions, these TF particles provide an idealized representation that accounts for TF–DNA affinity and the possibility of TF condensation. Within this framework, they show that when the TF–TF interactions are weak, the frequency with which enhancer–promoter pairs will form is correlated with the size of the TF condensate. However, for stronger TF–TF interactions, this relationship is nonmonotonic, where formation of larger condensates can actually impede the search process and reduce the chance that the promoter will find the enhancer. Thus, the model indicates there is a “Goldilocks” regime, where the condensate is not too small, nor too large, but is instead “just right.” In PNAS, Zhu et al. (5) present a theoretical model that illustrates how disorder, biomolecular interactions, condensation, and DNA reorganization contribute to gene expression. In the search for general principles in biology, it is interesting to note some parallels between the identified Goldilocks regime for chromatin reorganization and aspects of protein folding. The funnel-like energy landscapes of protein folding (1) inspired development of simplified “structure-based” force fields (10), where native contacts (i.e. those present in the folded structure) are assumed to dominate the energetics. While idealized descriptions have been useful for many questions, coarse-grained variants of these models (11, 12) have also included nonnative energetic roughness. These extended models showed that a small degree of roughness can actually increase the folding rate, though excessive roughness will arrest the dynamics. In the study of Zhu et al., enhancer–promoter association is viewed as a search problem of a polymer-like chain. Similar to the effect of nonnative interactions on protein folding, the current study finds that TF clustering increases the frequency of enhancer–promoter association, up to a point. If the propensity to condense/cluster is too great (analogous to increased roughness in folding), the search process becomes prohibitively slow. Another parallel to protein folding dynamics is found in the study of molecular crowding. Through a combination of coarse-grained models and experimental measurements (13), it was observed that protein folding rates can depend nonmonotonically on crowder concentrations. At low concentrations, crowders can help localize the chain, whereas at higher concentrations, they impede the search, similar to the predicted effects of TF clustering on chromatin reorganization. A strength of this study is its simplicity. By stripping away the details and focusing on a set of core physical attributes, the work explores broad relationships in these intricate biological systems. This exemplifies the spirit of physics-driven inquiry, where idealized models are used to understand classes of thermodynamic and kinetic phenomena. A classic example is the Ising model (14), where a magnetic material is described in terms of evenly spaced particles, each with a discrete spin state that only interacts with its nearest neighbors. For nearly a century, this has been the textbook example for introducing the properties of phase transitions. Later, by adding frustration (i.e. energetic roughness) to the model (15), it was possible to analytically derive conditions under which a glass phase can emerge. While these approaches are necessarily abstract, they can have applicability in the study of real physical systems. For example, so-called “random energy models” provided insights into condensed matter phenomena, while also establishing a foundation for describing the energy landscapes of biomolecules (16). These efforts naturally led to early theoretical models of folding in which a protein was described in terms of coarse-grained beads that can move across a discrete lattice (1, 17). In those studies, as well as in the current, the strategy was to consider simplified models that have sufficient physical detail, in order to explore general principles in complex systems. The idealized nature of the present study also complements the range of experimental and theoretical approaches that are available to probe chromosome dynamics and gene regulation. For example, the integrated use of Hi-C measurements and statistical-mechanical models recently identified cocompartmentalization of enhancer–promoter pairs in human cells (18). At a finer level of detail, explicit-nucleosome models have helped identify the molecular interactions that enable formation of local structure in chromatin (19). At even higher resolution, models that explicitly treat the interactions between transcription factors and DNA have revealed how individual TFs are able to effectively search the DNA through diffusion along the DNA and “hopping” between regions that are distant in sequence (20). Moving forward, it will be exciting to see how a combination of theoretical, computational, and experimental techniques can shed light on the many facets of chromosome dynamics. In this context, the theoretical analysis presented by Zhu et al. brings the field closer to fully understanding the physical principles that enable cellular life.