Cells routinely integrate mechanical (stretch, compression, or shear) and biochemical signals (cytokines and growth factors) to regulate gene expression programs and bring about major cell fate decisions such as differentiation, division, migration, or apoptosis [1, 2, 3, 4, 5, 6]. Collectively, cells need to coordinate their cell fate processes with one another and the extracellular matrix (ECM) so as to maintain tissue homeostasis or to carry out tissue repair and regeneration. Towards this, the same genetic material must be read and expressed in drastically different ways to give rise to the different cell types. In this context, the nucleus plays a critical role in this process by organizing the genome within to bring about tissue-specific genetic programs and regulating the genome by interpreting signals from the intracellular and extracellular microenvironment. These signals trigger biochemical pathways within the cell, ultimately causing the activation of transcription factors leading to the transcription of specific parts of the genetic material. In this review, we briefly discuss the coupling between mechanical forces and 3D genome to regulate gene expression programs and highlight some recent developments and potential new directions.

Cells are subject to multiple types of mechanical forces from their microenvironment. Certain classes of proteins have been identified as potential mechanosensors. Some of these include ion channels (such as Piezo channels), gap junctions, integrins, cadherins, G protein-coupled receptors (GPCRs) and junctional connections eventually signaling to the cell nucleus [7]. Mechanical sensing by cells is dependent on the stress caused by a balance of extracellular and intracellular forces acting at the junctions of contact between the cell and the ECM, or between neighboring cells. Formation of such adhesive structures involves the actomyosin structure within the cell generating tension on the junction, so as to adapt to stiffness of the local microenvironment [8]. To establish this feedback, focal adhesions transmit forces from the ECM to the cytoskeletal actin network and vice versa, via adhesion receptors (e.g. integrins). Similarly, adheren junctions between cells transmit tensile and compressive forces between actin cytoskeletal networks [9]. For cells to maintain their mechanical homeostasis, they need to constantly tune their cytoskeletal forces to counter external stress signals. Recent studies also highlight the importance of microtubules and intermediate filaments as critical regulators to maintain cellular mechanical homeostasis [10,11].

Studies of the molecular structure of the nucleoskeleton and cytoskeleton have built up a growing body of knowledge about the mechanical links between the two and its impact on cellular identity [12]. The LINC (linkers of nucleoskeleton and cytoskeleton) complexes are posited to be made up of two types of protein complexes - SUN and KASH, interacting with each other to span the nuclear membrane. SUN domain proteins are found in the inner nuclear membrane, while KASH domain proteins are largely associated with the outer membrane. Nesprins, KASH domain proteins, bind to cytoskeletal proteins such as actin and microtubules, either directly or with adapter proteins. Through interactions with actin and other cytoskeletal proteins, nesprins mediate an extensive mechanotransduction network between the cell's plasma membrane and its nuclear envelope to regulate chromatin and gene expression [13]. The intranuclear and extranuclear forces on the nucleus result in opposing tensions that hold the nucleus in a certain conformation. Actin exerts a tensile force on the nucleus, while microtubules act in an antagonistic manner to actin by exerting a compressive force [14]. Clearly, the stability of the nucleus is maintained by a balance of forces exerted by the viscoelastic cytoskeletal networks outside the nucleus and the chromatin and its related components within the nucleus. A simple spring-dashpot model (referred as Kelvin-Voigt model) is depicted in Figure 1 to represent the nature of the collective viscoelastic behavior of cytoskeletal networks, although this requires more rigorous theoretical framework. Given the importance of the nuclear membrane in linking extracellular information to the chromatin structures, a major challenge going forward is to dissect the molecular bridges and adapters. While recent progress in super-resolution microscopy and cryo-electron tomography has advanced our knowledge [15,16], missing in this picture is the integration of molecular structures on the nuclear membrane and how they are modulated in space and time during nuclear mechanotransduction. Correlative imaging methods integrating light and EM combined with progress in expansion microscopy and structure prediction of molecular complexes by AI-based methods, such as Alpha fold [17, 18, 19, 20], could provide novel insights on our understanding of how forces are transmitted via the cytoskeletal to nuclear to chromatin links.

Mechanotransduction to the nucleus can take place either by biochemical signaling and/or by physical transduction through the cytoskeletal network [21]. Biochemical signals rely on diffusive transport or directed transport through molecular motors for their transmission. The spread of the signal is controlled by the concentration of the molecules along the chemical gradient and is either diffusion-limited or regulated by molecular motor dynamics for directed transport. By comparison, the transmission of mechanical signals is independent of concentration gradients. A force applied at one point can be rapidly transmitted at acoustic modes along any number of physical connections to another point and can easily reach sites a long distance away from the source. In addition, recent studies have also begun to address the general principles underlying cytoplasmic to nuclear transport triggered by mechanical forces [22∗, 23, 24]. Collectively, cells coordinate cell–matrix and cell–cell junction mechanosensing via both physical and biochemical transduction pathways to the nucleus in a context-dependent manner to activate transcription factors and chromatin remodeling enzymes for downstream regulatory programs [1, 2, 3, 4, 5, 6].

Transcription factor activation in response to extracellular microenvironmental signals is characterized by three general types of intracellular signaling, based on the location of the factors in the cell membrane, cytosol, or nucleus. For example, transcription factors closely associated with the cell membrane contain LIM domains and nuclear localization signals (NLS) [25]. Upon activation, the transcription factor is phosphorylated, causing its dissociation from its protein complex at the cell membrane and subsequent diffusion to the nucleus. The second group of transcription factors are localized in the cytosol, and their activation and their crosstalk are dependent on upstream cell membrane mechanosensing signals. This group of transcription factors includes cofactors, e.g. myocardin-related transcription factors (MRTFs) and transcription factors, e.g. YAP-TAZ [26,27]. The third group of transcription factors is always localized in the nucleus. The regulators of such transcription factors are found outside the nucleus, and only enter upon activation by extracellular stress signals. An example is the well-studied tumor suppressor p53 [28]. In addition, cytoskeletal filaments such as actin have been found to play a crucial regulatory role in the spatial compartmentalization of chromatin remodeling enzymes [29]. For example, chromatin remodeling enzymes, such as HDAC3, are located outside the nucleus and only enter in response to extracellular signals and are regulated by acto-myosin contractility [29]. Much of our understanding of how mechanical forces impact cellular function stops at the analysis of the activation of transcription factors which is then correlated with the downstream gene expression of its target genes. However, missing in this picture is the spatiotemporal organization of genes, and gene clusters that are part of the elaborate 3D genome organization within the cell nucleus, discussed in the next section. In addition, the mechanical forces that regulate cytoskeletal structure have a profound influence on nuclear mechanics and the 3D genome organization [5,14].

Eukaryotic DNA is packed into a nucleus with a mere few μm3 volume, depending on the cell type [30]. At the smallest scale, the compaction of the DNA strand involves winding it around multiple histone proteins, giving rise to the familiar “beads-on-a-string” chromatin structure. Additional folding and bending of this fiber is necessary to pack the DNA into higher-ordered chromosome structures. A cell spends most of its life cycle in interphase, where it carries out a series of genomic transactions necessary for maintaining its homeostasis. The flurry of activity in the cell would mean that the genome needs to be constantly kept in a partially decondensed state. The cell achieves this balance via a complex set of chemical modifications of the histones bound to the DNA [31]. The regulation of histone modifications also directs the recruitment of various non-histone proteins to the histone tails, to bring about modular genetic functions. Electron microscopy images visualizing the density of chromatin, i.e. euchromatin (less dense with increased histone acetylation) and heterochromatin (denser with increased histone methylation), show that in cells with high transcription activity, there is a relatively higher proportion of euchromatin present in the nucleus, and vice versa [32]. This correlation between transcription activity and euchromatin levels is also reflected in the differential chromatin condensation patterns in different cell types [33].

While it is known that changes in DNA-histone interactions affect the extent of DNA condensation, the mechanistic principles involved remain unclear [34,35]. A likely mechanism could be facilitated by a family of proteins known as heterochromatin proteins which are essential for the formation of heterochromatin. Advanced optical imaging methods have observed that genomic regions associated with HP1α proteins can aggregate/coalesce/cluster with each other [36, 37, 38]. Such clustering of heterochromatin regions by HP1α potentially due to a phase separation mechanism would result in the spatial segregation of the genomic material into heterochromatin and non-heterochromatin regions. From a mechanical perspective, the DNA coil is exerting an outward entropic force on the nucleus as it is being physically constrained within the nucleus' boundaries. Conversely, DNA experiences an inward enthalpic force, provided by DNA-histone and other non-histone proteins including HP1α interactions that act against the entropic forces by condensing the DNA. These results show that the compaction of chromatin plays an important role in the maintenance of nuclear structural stability [39]. It also illustrates the fact that the balance of forces on the nucleus makes it a very mechanically stressed container; a concept that is often neglected when the functions of the nucleus are studied using biochemical methods [14,39].

In the cell nucleus, the chromosomal DNA is organized into higher-ordered 3D structures forming distinct functional regions. FISH-based methods to probe chromatin architecture have revealed that chromosomes are arranged in a non-random fashion in the nucleus, as chromosome territories [40]. In addition, gene-rich domains are concentrated on the surface of chromosomes, while the interior of chromosomes is made up of largely gene-poor sequences [41]. However, this may vary for chromosomes closer to the nuclear periphery [42]. Furthermore, live-cell imaging experiments have shown that chromosomes are spatially confined to their territories through much of interphase. Although chromosomes localize to their own territories, they have also been found to intermingle with one another. Several experiments and modeling have suggested that the intermingling domains are directly linked to the transcriptional activity of genes within such domains in a cell-type-specific manner [43,44]. Importantly, chromosome conformation capture techniques have enabled the creation of genome-wide 3D chromosome contact maps, revealing that the genome is organized into what are called topologically associated domains (TADs) which comprise of self-interacting DNA sequences [45]. Proteins associated with TADs include the cohesin, condensins, and CTCF-binding proteins and these proteins could serve as anchor points to differentially loop out long segments of DNA [46]. In addition, the parts of the chromatin that contact the nuclear lamina are known as lamin-associated domains (LADs), which can be detached in functional contexts [47]. LAD regions show lower gene density, consist of more inactive chromatin regions, and show histone modifications that silence chromatin. A final step towards efficient transcription control involves the co-localization of active genes, transcription factors and co-factors, chromatin remodeling enzymes, and the RNA polymerase within intra-chromosomal or inter-chromosomal regions [48, 49, 50, 51]. Growing evidence suggests that transcription factories or hubs or condensates could be such sites of transcription control and these hubs could themselves generate mechanical forces [52, 53∗, 54, 55∗].

To respond promptly to mechanical forces, cells will need to remodel their chromosomes to activate gene expression programs and protect the genome [56, 57, 58, 59]. In this context, the spatial 3D reorganization of chromosomes has been directly observed in the nuclei of cells grown on anisotropic (stretched cell states) and isotropic (relaxed cell states) substrates [60]. Besides the spatial rearrangement of chromosomes, their orientation, size, radial distance, and intermingling fractions have all shown distinct variations depending on ECM mechanical constraints [60]. An important aspect of the spatial clustering of active genes is their co-localization in their respective intra-chromosomal or inter-chromosomal regions. For example, it has been found that the combination of certain intermingling “partner” chromosomes changes in response to cell relaxation and stretching [60]. For instance, in cells that adhere poorly to their substrate, inflammatory response transcription factors such as NFkB are transduced to the nucleus. In contrast, when cells are strongly adhered to their substrate, serum response transcription factors are activated via the translocation of MRTFs to the nucleus. Consistent with these findings, it would be important to identify if pathway-specific genes that are expressed upon mechanical forces are indeed spatially co-clustered within the nucleus for their co-regulation. Such studies suggest that there likely exists a tight coupling between cellular perception of microenvironmental signals, intracellular signaling pathways, and the spatial organization of genetic material to bring about context-dependent gene expression [29,56,60].

Collectively, the interplay between mechanical forces and the 3D genome is a multi-scale dynamic process. Firstly, the activation of signaling cascades for cytoplasmic to nuclear mechanotransduction by post-translational modifications of specific protein messengers. Secondly, the structural remodeling of the cytoskeleton, the nuclear lamina, and the 3D genome by physical and biochemical signals. Finally, the co-localization, in space and time within the nucleus, of various genome regulatory factors for optimal gene regulation. While high-resolution imaging and molecular biology methods have provided important insights on genome packing and function, major questions remain. Importantly, the role of mechanical forces governing chromosome packing and its inherent cell-to-cell heterogeneity, the spatiotemporal organization of chromosome territories and hubs of gene clusters and how these functional hubs are actively maintained and memorized through cell division, and its impact on cell-state transitions are still unknown. Since the 3D genome is also a functional integrator of extracellular mechano-chemical signals and genome function, AI-based imaging methods could also provide powerful biomarkers of cell states and their transitions [61].