The non-equilibrium genome - where Physics meets Biology

Stem-cell fate decisions require the coordinated dynamics of transcription factors (TFs) like NANOG and SOX2, yet their biophysical properties within nucleus, such as interactions with DNA and spatial clustering, remain unclear. Here, we used single-molecule light-field microscopy to achieve 3D whole-nucleus tracking of these TFs at high density. Analysis using our published temporal pipelines show that they adopt different DNA binding strategies: while both TFs undergo transient non-specific binding, NANOG exhibits more stable specific binding than SOX2 (~25s vs ~16s). Furthermore, we developed two novel spatial analysis pipelines. Our first pipeline based on Delaunay triangulation demonstrates that chromatin-bound TFs form spatial clusters. Our second pipeline based on convex hull statistics reveals that freely diffusing TFs exhibit confinement by nuclear domains, with NANOG forming smaller domains than SOX2 (~390nm vs ~450nm). Our work establishes new quantitative pipelines widely applicable for dissecting spatiotemporal nuclear dynamics using single-molecule microscopy.

DNA nicks caused by single-strand breaks (SSBs) are repaired efficiently by dedicated enzymatic machinery, yet how cells physically protect DNA nicks from expanding further remains unclear. We find that Barrier-to-Autointegration Factor (BAF) prevents SSBs from escalating into double-strand breaks (DSBs) through a kinetic trapping mechanism. Using optical tweezers coupled with confocal microscopy, we show that peeling initiates normally from SSBs but is subsequently arrested in the presence of BAF. We describe a molecular mechanism where BAF binds peeled DNA and crosslinks it to the intact primary molecule of DNA, thus preventing nick expansion. siRNA-mediated depletion of BAF in HCT116 cells results in significantly elevated DSB levels following oxidative stress, in line with the described mechanism of BAF suppressing the conversion of SSBs into DSBs. Together, these findings uncover a physical mechanism that stabilizes nicked DNA, and expands the functional repertoire of BAF to include the physical suppression of damage escalation to preserve genome integrity.

Chromatin modification and structure are crucial for establishing cellular identities, yet understanding the causal regulatory relationships between these factors remains challenging. In particular, studies of how modification patterns and solution environments affect gene-scale chromatin structure and function have been limited by the lack of methods to reconstitute and characterize long chromatin arrays with defined modification patterns. Here, we present an in vitro method for reconstituting and characterizing gene-scale chromatin arrays containing 96 nucleosomes with precisely controlled modification patterns at a 12-nucleosome resolution. Using these arrays, we characterized the effects of histone H4 hyperacetylation on chromatin structure using single-molecule microscopy and an in vitro Hi-C method developed for arrays reconstituted with the Widom 601 sequence. Single-molecule measurements of the end-to-end distance fluctuations revealed that the fluctuation amplitude and timescale depend on global acetylation density. This relationship between the amplitude and timescale aligns with predictions from a polymer model that incorporates hydrodynamic interactions. The diffusion coefficients of 12-nucleosome arrays with different linker DNA lengths further supported the presence of these hydrodynamic interactions. Moreover, in vitro Hi-C analysis revealed that acetylation decreases nucleosome contact frequency, resulting in the formation of distinct structural domains within heterogeneously modified arrays. Together, these results illustrate the physical principles by which histone modifications modulate chromatin architecture through altered nucleosome-nucleosome interactions.

Gene expression control in early embryonic development involves contacts between genes and enhancers over unusually long ranges, supposedly via simultaneous association with transcriptional clusters with liquid-phase properties. Capillary effects were proposed as an underlying mechanism for such long-range contacts, but a specific role in embryonic gene regulation remains elusive. Here, we show that transient elastocapillary adhesion to transcriptional condensates formed on super-enhancers hyper-activates an epigenetically distinct subset of zygotic zebrafish genes. We obtained a microscopy dataset of 8 zygotically regulated genes over 5 stages of early development, in which some genes approached transcriptional clusters as close as 200 nm during specific developmental stages. In a linear model fitted to this entire data set, gene-cluster proximity emerges as a variable that is fully explicable by a combination of RNA polymerase II (Pol II) recruitment and transcription level ($R^2$=0.99995). To uncover the underlying biophysical mechanism, we performed molecular dynamics simulations representing two chromosome segments at 10-kilobase-pair resolution. Hyper-activation occurred via capillarity-driven adhesion between a “super-enhancer block” on one segment and a “promoter block” on the other segment. This adhesion proceeded via a Pol II condensate shared by the super-enhancer and the promoter block and emerged only when a promoter block of extended length was introduced. Analysis of genome-wide sequencing data confirmed that putative extended promoter regions with high H3K27ac coverage and occupancy by phase-separation-prone pluripotency factors (Nanog, Pou5f3, and Sox19b) occur specifically for zygotically expressed genes with increased Hi-C contact to Pol II peaks. Taken together, our findings suggest that promoter-proximal, enhancer-like chromatin features enable long-range, capillarity-driven adhesion of specific zygotic genes to super-enhancer-associated transcriptional condensates.

Bacteria traverse diverse environments, many of which can be detrimental, and to survive these fluctuations they rely on internal regulatory programs shaped by multiple layers of regulatory topologies, from global chromosomal structure to local protein flexibility. While individual mechanisms have been studied in detail, how these topological layers collectively constrain and shape genome-wide stress responses remains poorly understood. Using Escherichia coli as a model system, we examined how regulatory topology shapes genome-wide adaptation during bacterial stress response. We first investigated operon topology and transcriptional responses to antibiotic and starvation stresses, by combining transcriptomics, flow cytometry, and synthetic promoter constructs, we show that internal promoters mitigate transcriptional losses caused by premature termination, which are further amplified by topological constraints such as DNA supercoiling and collisions between elongating and promoter-bound RNA polymerases, with similar operon-level responses conserved across evolutionarily distant bacteria. We next explored transcription factor network dynamics using fluorescent reporters for 16 global regulators, revealing single-cell variability and enabling quantitative mapping of network hierarchy, timing, and activation under diverse stress conditions. Finally, by integrating genome-wide expression analysis with molecular dynamics simulations, we show that under combined temperature and antibiotic stress transcriptomes collapse toward temperature-defined states, arising from topological constraints on nucleoid organization, ATP-dependent enzymatic activity, regulatory networks, and protein conformational flexibility. Together, these results demonstrate how physical constraints and topological organization govern bacterial stress adaptation across scales.

Swimmers moving in the same direction form an ordered state of living matter. However, this ordered state is not always stable to ambient disturbances. This may lead to chaotic flows characterized by the pres- ence of topological defects, a phenomenon known as active turbulence. The ordered state of microswimmers can be destroyed by an instabil- ity created by their swimming stresses. For slightly larger swimmers, where viscous and inertial forces are comparable, an instability due to the fluctuations in the concentration of swimmers destroys the order [1]. In this talk, I will discuss about the instabilities and turbulence in weakly inertial suspensions of extensile swimmers, where the defect turbulent state transitions to the concentration-wave turbulent state [2]. These findings reveal new ways in which living matter may get orga- nized in nature. [1] P. Jain et. al., Phys. Rev. Lett. 133, 158302 (2024). [2] P. Jain et. al., Phys. Rev. Fluids 10, 114602 (2025).

Cells must process fluctuating signals to make reliable decisions. At the molecular scale, this is mediated by epigenetic modifications along the DNA (1d genome), and by dynamic changes in chromatin organization (4d genome). Here, we show that the interplay between epigenetic modifications and the dynamic conformation of chromatin generically implements a low-pass filter for cellular signals. We formulate a field theory for a one-dimensional epigenetic field coupled to four-dimensional spatio-temporal conformation of a polymer and, using linear-response theory, derive spatiotemporal response functions. We find that this genome architecture enhances sensitivity to slow, biologically meaningful signals but not rapid fluctuation compared with epigenetic dynamics alone. Spatially modulated signals are further enhances on the domain scale. We validate these predictions with molecular-dynamics simulations. Our results demonstrate that the intrinsic biophysical coupling between genome folding and epigenetic state is sufficient to endow chromatin with signal-processing capabilities, providing a minimal mechanism by which cells can discriminate molecular noise from informative cues.

The nuclear envelope protects the genome from mechanical stress during processes such as migration, division, and compression, but how it buffers forces at the scale of DNA remains unclear. In this work, we utilize optical tweezers to show that a multivalent protein–DNA co-condensate containing the nuclear envelope protein LEM2 and the DNA-binding protein BAF shield DNA beyond its melting point at 65 pN. Under load, their collective assembly induces an unconventional DNA stiffening effect that provides mechanical reinforcement, dependent on the intrinsically disordered region (IDR) of LEM2. At the nuclear surface, these components form an elastic surface hydrogel in which LEM2 IDR-IDR interactions contract the surface hydrogel relative to its relaxed state, introducing a pre-stress in the lamin network. Inside cells, this surface hydrogel model can recapitulate elastic properties of the nuclear envelope measured via AFM indentation experiments as well as nuclear morphology, using parameters obtained at the molecular scale by use of optical tweezers. Disruption of the surface hydrogel increases DNA damage and micronuclei formation during nuclear deformation. These findings reveal a load-bearing, mesoscale surface hydrogel that reinforces the nucleus and expands the functional repertoire of biomolecular condensates to include DNA protection under mechanical stress.

The E. coli chromosome participates in essential biological processes such as replication and transcription, which rely on specific interactions between proteins and chromosomal segments. Along the chromosome, regions called rrn operons contain genes that are transcribed into ribosomal RNA (rRNA). These operons are among the most transcriptionally active sites in the genome [1], and they are known to spatially colocalize within the E. coli cell [2]. During transcription, RNA polymerase (RNAP) binds to these genetic sites along the chromosome, and forms dense clusters in the cell [3]. Recent experimental evidence suggests that liquid–liquid phase separation (LLPS) contributes to cluster formation of RNAP proteins where an antitermination factor (NusA) plays a key role[4]. We present a simulation model to investigate the mechanism underlying the formation of these biomolecular condensates. In our model the mutual attraction of NusA, which display a micibilty gap at higher concentration, drives condensate formation via the polymer-assisted condensation pathway [5]. We further establish how these condensates drive the colocalization of rrn operons. Our results reconcile seemingly disparate experimental observations regarding chromosomal organization reported in [2] and [6]. [1] Hans Bremer and Patrick P. Dennis. Modulation of chemical composition and other parameters of the cell at different exponential growth rates. EcoSal Plus, 3(1), January 2008. [2] Tamas Gaal, Benjamin P. Bratton, Patricia Sanchez-Vazquez, Alexander Sliwicki, Kristine Sliwicki, Andrew Vegel, Rachel Pannu, and Richard L. Gourse. Colocalization of distant chromosomal loci in space in e. coli: a bacterial nucleolus. Genes & Development, 30(20):2272–2285, October 2016. [3] Jun Fan, Hafez El Sayyed, Oliver J Pambos, Mathew Stracy, Jingwen Kyropoulos, and Achillefs N Kapanidis. RNA polymerase redistribution supports growth in E. coli strains with a minimal number of rRNA operons. Nucleic Acids Research, 51(15):8085–8101, 06 2023. [4] Anne-Marie Ladouceur, Baljyot Singh Parmar, Stefan Biedzinski, James Wall, S. Graydon Tope, David Cohn, Albright Kim, Nicolas Soubry, Rodrigo Reyes-Lamothe, and Stephanie C. Weber. Clusters of bacterial rna polymerase are biomolecular condensates that assemble through liquid–liquid phase separation. Proceedings of the National Academy of Sciences, 117(31):18540–18549, July 2020. [5] Jens-Uwe Sommer, Holger Merlitz, and Helmut Schiessel. Polymer-assisted condensation: A mechanism for hetero-chromatin formation and epigenetic memory. Macromolecules, 55(11):4841–4851, May 2022. [6] Virginia S. Lioy, Axel Cournac, Martial Marbouty, Stephane Duigou, Julien Mozziconacci, Olivier Espeli, Frederic Boccard, and Romain Koszul. Multiscale structuring of the e. coli chromosome by nucleoid-associated and condensin proteins. Cell, 172(4):771–783.e18, February 2018.

Transcription factors (TFs) regulate gene expression by binding to specific genomic loci determined by DNA sequence. Their sequence specificity is commonly summarized by a consensus binding motif. However, eukaryotic genomes contain billions of low-affinity DNA sequences to which TFs associate with a sequence-dependent binding energy. We currently lack insight into how the genomic sequence defines this spectrum of binding energies and the resulting pattern of TF localization. Here, we set out to obtain a quantitative understanding of sequence-dependent TF binding to both motif and non-motif sequences. Using in vitro measurements of equilibrium binding energies of the human TF Klf4 to a library of short DNA molecules, we reveal a highly non-linear sequence-dependence of Klf4 binding. However, we show that this non-linearity can be captured by combining a linear model of binding energies with an Ising model of the coupled recognition of nucleotides by a TF. Strikingly, we find that this equilibrium model parametrized by our \textit{in vitro} measurements quantitatively captures Klf4 occupancy statistics in human cell lines without additional fit parameters.

Understanding how epigenetic patterns emerge and are maintained in chromatin, and how they couple to its three-dimensional organization, is a central open problem in biophysics. In this work, we introduce one-mark and two-mark chromatin-like magnetic polymer models, where monomers can be either active/marked (endowed with a spin, having either one or two states) or passive/neutral (only steric repulsion). For both models, using mean-field theory and Monte Carlo simulations, we show that, when the concentration of neutral states is kept fixed, the filament phase-separates into uniformly marked globules coexisting with extended disordered sections. We then examine the similarities and distinctions between the two models, particularly concerning phase diagrams, and make connections with Hi-C and FISH experiments and potential epigenetic memory.

Transcription regulation is one of the most intriguing and complex processes in molecular biology, yet it remains only partially understood. In prokaryotes, extensive theoretical and experimental work has mapped the quantitative landscape of gene regulation, revealing mechanistic principles by which transcription factors (TFs) control gene expression. Extending these insights to more complex eukaryotic systems remains challenging because of the strong context dependence of TF binding. Eukaryotic genomes typically contain far more potential binding sites than TF molecules, leading to competition among sites for limited TF resources—a phenomenon increasingly evident in experiments, but still lacking a comprehensive quantitative and biophysical characterisation across the genome. To address this gap, this work develops a simple mechanistic biophysical model, grounded in statistical mechanics, for transcription factor binding in a eukaryotic genomic context. The model quantitatively characterises the hallmarks of binding-site competition across biologically realistic parameter ranges. Coupling it to deep learning–based inference enables in silico interrogation of TF redistribution and the extraction of informative constraints that refine the mechanistic description.

Traditional epigenomic analyses rely on population-averaged methylation "Beta-values," which collapse molecule-specific patterns essential for understanding the structural regulation of the genome. Here, we present a framework to bridge the gap between 1D chemical modifications and 3D chromatin architecture by integrating Dam Assisted Fluorescent Tagging of Chromatin Accessibility (DAFCA) with long-read Nanopore sequencing. We utilize DAFCA to label accessible DNA regions within the cell nucleus, which are subsequently visualized via single-molecule optical mapping in silicon nanochannel arrays. This approach allows for the characterization of long-range structural variations and the direct observation of chromatin compaction patterns along megabase-scale DNA molecules. Simultaneously, long-read Nanopore sequencing provides the high-resolution methylation patterns necessary to calculate local methylation entropy based on the correlation of neighboring CpG sites along individual molecules. Our framework explores the potential of these long-read metrics to serve as a predictive link between epigenetic "disorder" and physical genomic packaging. We hypothesize that regions of high local methylation entropy (1D) correspond to specific structural "hinges" or zones of differential compaction (3D) that define cellular identity. By resolving these patterns on individual molecules rather than ensemble averages, this approach offers a unique vantage point to study the interplay between chemical stochasticity and the mechanical constraints of the genome, providing a quantitative path to assess how 1D information encodes 3D structural plasticity.

We investigated publicly available microscopy data for two loci of the human Chr21 obtained from multiplexed fluorescence in situ hybridization (mFISH) methods for different cell lines and treatments. Inspired by polymer physics models, our analysis centers around distance distributions between different tags with the aim being to unravel the chromatin conformational arrangements. We show that for any specific genomic site, there are two different conformational arrangements of chromatin, implying coexisting distinct topologies which we refer to as phase $\alpha$ and phase $\beta$. These two phases show different scaling behaviors: the former is consistent with a crumpled globule, while the latter indicates a confined, but more extended conformation, such as a looped domain. We show that a simple heterogeneous random walk model captures the main behavior observed in experiments and brings considerable insights into chromosomal structure.

We consider single-ring and multiple-ring systems and compare their static and dynamical properties, by using Molecular Dynamics computer simulations with explicit solvent. By switching from good- to bad-solvent conditions, we show that the two types of systems react quite differently, in particular systems with multiple chains tend to be dynamically stuck. We argue that the reported behavior is caused by ring-ring interpenetrations, and characterize this behavior by a detailed analysis of static and dynamic properties.

Enhancers regulate promoters across genomic distances ranging from a few kilobases to several megabases. Whether and how this depends on physical encounters and on the loop-extrusion activity of cohesin remains unclear, with loop extrusion being essential for enhancer function at large genomic distances, yet dispensable for nearby enhancers. Here, combining polymer simulations with high- resolution live-cell microscopy, we discover that loop extrusion creates long-lived encounters between genomic loci. These events are rare but nevertheless enable promoter activation by long-range enhancers. Nearby enhancers rely instead on similarly long, yet extrusion-independent encounters. This new paradigm reconciles apparently contradictory results on the differential requirement of cohesin for enhancer activity across genomic distances, and offers a unifying model for distal regulation of gene expression.

Topologically constrained genome-like polymers often double-fold into tree-like configurations, which can be modeled on the level of the folded (ring) polymer or on the level of the underlying random tree. For both descriptions, we have recently obtained expressions for the configurational entropy in ensembles with controlled branching activity. In my presentation, I will demonstrate that they are equivalent up to a contribution originating from the number of distinct wrappings of a single tree and obtain an exact mapping between the two ensembles for models, where excluded volume interactions are treated consistently on the tree and on the ring level. I will also present a grand canonical version of the Amoeba Monte Carlo algorithm for simulating lattice trees with a statistical weight corresponding to the number of distinct wrappings. While the results are in excellent agreement with data from Monte Carlo simulations of the corresponding elastic lattice model for tightly double-folded rings, the new algorithm is ${\mathcal O}(N)$ faster. In the last part of my talk, I will show how this algorithm is currently being used to model bacterial DNA at scales larger than a few hundred basepairs.

We investigated publicly available microscopy data for two loci of the human Chr21 obtained from multiplexed fluorescence in situ hybridization (mFISH) methods for different cell lines and treatments. Inspired by polymer physics models, our analysis centers around distance distributions between different tags with the aim being to unravel the chromatin conformational arrangements. We show that for any specific genomic site, there are two different conformational arrangements of chromatin, implying coexisting distinct topologies which we refer to as phase $\alpha$ and phase $\beta$. These two phases show different scaling behaviors: the former is consistent with a crumpled globule, while the latter indicates a confined, but more extended conformation, such as a looped domain. We show that a simple heterogeneous random walk model captures the main behavior observed in experiments and brings considerable insights into chromosomal structure.