Condensates and Cell States: A New Paradigm for Understanding Tumor Biology

Abstract

Traditionally, cells have been classified by their type. Identifying cell types was deemed vital for understanding biological processes. More recently, cell type classification has been recognized as not good enough. Cells have many transient states that depend on their spatial environment and vary over time. The current cell states description recognizes that a cell is dynamic, varying over developmental time, location, senescence, and disease. While cell states refer to the functional behavior of the cells, biomacromolecular condensates are now recognized as the membraneless structures within them, which by concentrating functionally related proteins, make the function happen. Here, we (i) clarify the molecular basis of the current separation between cell types and states and point to the merit of the “cell states”, which make the classical “cell types” distinction expendable; (ii) discuss how fundamental physical principles evolved the functionally specific, conformationally biased biomacromolecular condensates; and (iii) consider the pharmacology of cell states and condensates. Recent reports highlighted condensates as drug targets. However, drugs designed to dismantle condensates can be toxic due to lack of specificity–unlike cell states, which are controlled by targetable specific epigenetics players. Especially (iv), we consider transitioning primary cancer cells, linking cell states and condensates to tumor proliferation. We suggest that original cancer cells transitioning to other tissues are primarily influenced by cell states, supported by their multi-condensates structuring. The high genetic homogeneity of untreated cancers—within primary tumors and metastases—underscores the significance of the transient nature of cell states.

INTRODUCTION

We focus on cell types, cell states, and biomolecular condensates. Multiple works refer to cell types and states [e.g., (1–14)]. Here we suggest replacing the classical, decades-old view of morphology, and function-specific cell types, by dynamic, transient cell states (14–16). Cells do not execute precisely the same functions over their lifetime (17,18). Cells are not oblivious to their changing spatial environment and organismal life (19). Biological processes are dynamic. They do not have distinct starting and end points. They do not even necessarily share their developmental history, as can be seen in cells transiting and evolving to take on the behavior of different tissues, nor do they share similar molecular composition (20–22). They are classified for convenience—reminiscent of the hierarchical sorting of proteins into secondary and tertiary structures. Proteins fold not recognizing these distinctions, which do not exist in nature (23,24). They are regulated, influenced by multiple parameters that vary with the environment and time. This description implicitly conveys the reality that cells transition, responding to their changing conditions, which the classical definition of cell types does not capture. It is not only that a skin cell differs from a pancreatic cell—but also a blistered skin cell differing from a healthy skin cell, a fatty liver cell differing from an un-symptomatic liver cell, and transplanted cells from one tissue to another taking on altered functions to replace damaged tissues. The classical cell type narrative is unable to explain how a melanoma skin cell can transit and thrive in brain tissue. At its root, a cell type label does not relate to dynamic, heterogeneous populations. As such, it cannot capture cell evolution, nor provide insight into crucial questions such as the cell’s proliferative state, or predisposition to cancer phenotypic presentation. In contrast, cell states and their transitions refer to their transcriptomes and protein conformational ensembles, which are the hallmark of dynamic functions.

Cells have been viewed as shaped by hierarchical organization (25,26). Classically, cells are described as encompassing well-defined, membrane-enveloped, structured organelles. Faced with being highly crowded and heterogeneous, transient cell states adopt a tantalizing dynamic solution to efficiency: they harness membraneless biomolecular condensates, an inherent physics-based phenomenon formed by phase separation (27–32). Biomacromolecular condensation takes place through liquid-liquid phase separation, where the biomolecules spontaneously assemble (and disassemble) based on their chemical and physical properties and their shifting surroundings (32–38). Whereas cell states can be assessed by their transcriptomes (19,22,39,40), which are controlled by their epigenetics (41–44), efficient regulation in the crowded and heterogeneous environment is driven by their transient biomolecular condensates (31,45).

Biomolecular condensates’ makeup, spatial location in the cell, organization, and status, are crucial factors. They provide temporal functional specificity, linking atomic features and physical forces to condensate function, and transient cell states (33,46–53). Spatial environments and time frames influence the nature of the condensates, as exemplified by condensates of a melanoma skin cell varying as the cell transitions to brain tissue. Within their phase-separated droplets, condensate assemblies exhibit unique conformational dynamics, constrained by a tighter, higher density environment. Driven by exposed nonpolar protein surface areas, this environment results in hydrophobic interactions at their interfaces and defines the mechanics of intrinsically disordered proteins (IDPs). Lacking hydrophobic cores, disordered proteins undergo folding upon binding scenarios (54–57). The nonpolar-induced protein condensation repels water, resulting in low dilution (58,59), culminating in condensates’ conformational ensembles that could be significantly altered compared to those in the diluted cytoplasmic state. While the hydrophobic effect can be a major driving force in condensation, gelation is further enhanced by charges in dynamic disordered complexes between IDPs and folded domains, or proteins (60). Since function is largely expressed by the cell biomolecular condensates, altered conformational distributions could impact function (34).

Stable protein–protein interactions are the bedrock of condensate nucleation, extension, and efficient signaling (61–64). They are the linchpin of functional specificity and effective allostery (65,66). Detailed mechanistic knowledge, on the atomic level, is crucial for elucidation of exactly how function works in the condensate environment. Phase-dependent conformational changes can disfavor interactions favored in the dilute phase and expose pockets which are hidden, or ‘cryptic’ in the dense phase. This requires reconsideration of (i) the energy landscape and conformational distributions under lower dilution environment, (ii) spatiotemporal function-related proteomic organization in the condensed space, and (iii), constructing the condensate-specific interactome. Transitioning cell states provide a conceptual framework. Biomacromolecular condensates are specific structures that organize them, making the cell states work.

Below, we first discuss ‘classical’ cell types and cell states, the nature of membraneless biomacromolecular condensates organelles within them, consider the pharmacology of cell states and condensates, and comment on transitioning primary cancer cells, linking cell states and condensates to tumor proliferation. The high genetic homogeneity of untreated cancers within primary tumors and metastases—while presenting phenotypic heterogeneity—emphasizes cell states. This phenotypic cellular variation—while the cells harbor practically identical genetic information—highlight the critical role of the tumorigenic epigenome. Single cell transcriptomic over transitioning time can help establish drug targets (67).

CELL TYPES AND CELL STATES

Classically, cell types are inherited and stable whereas cell states are transient, dynamic, with phenotypes adopted in response to their changing environment or functional needs (3,68). Rather than a view of cell types as evolving through specific developmental trajectories, a cell states perspective recognizes that while cells have a common origin, their environments are dynamic, making them adapt and evolve through dynamic transcriptomics, which is largely controlled by epigenetic changes. These different views are important. As a clinically related example, we suggested that a cell state-based perspective can help decipher pediatric low grade brain tumors and neurodevelopmental disorders, which share proteins, signaling pathways, and networks, and germline mutations (19). However, they differ in their spatial location, and timing of gene expression during brain development, pointing to diverged cell states impairing prenatal differentiation. A rigid cell-type outlook may struggle to explain the different pediatric brain pathologies. High expression levels and potent activation drive cancer, while lower expression and weaker activation may result in neurodevelopmental disorders (69).

However, perhaps the most glaring rationale for why the “cell types” classification may fall out of favor comes from cell transplantation and cancer metastases (70). As to the first, in one example, cells are harvested from a patient, expanded in culture, seeded onto an appropriate scaffold, and then transplanted to the defect site, often in a different tissue (71). Cell transplantation is now hailed as having the potential to revolutionize the way we think about personalized medicine (72). In metastases, cancer cells detach and spread from their original location, traveling through the bloodstream or lymphatic system, to other tissues, commonly in other organs. Under the microscope, and in biochemical tests, their features resemble those of the primary cancer—not those where they settled—facilitating the identification of their origin. This, however, is not always the case (73,74). Cancer cells may spread in the body and form metastatic tumors, with the primary cancer unknown. In both cases, even if identifiable, melding into a different organ or tissue points to cell alterations (75). The premise can be understood. Even though during cancer development somatic mutations accumulate, including gene deletion and amplification, the genome is largely unchanged, which is also the case in cell transplantation. The differences between the tissue of origin and target rest in the patterns of expression, which relate to chromatin packing and epigenetic codes (76–81). While not considered in the classical cell type phenotypic distinctions, transitions in chromatin shapes, influenced by the spatial microenvironment over developmental time and organismal life, precipitate the stabilization (destabilization) of specific cell states (82–85). The resulting shifts in the landscapes of cell states is the mechanism through which transplanted and metastatic cells meld into their new environments. Through widespread, local feedback loops, and epigenetic codes, the remodeled chromatin shapes transform cell states (86,87). Stabilized forms spend more time in their acquired states, becoming more abundant—until another event transforms the population. Waddington imaged these processes, portraying them as ‘canalization’ (88–90), where balls roll down sloping valleys (91,92) (Fig. 1). To transition between canals perturbed states need be able to hop over their separating hills, which can be triggered by an epigenetic change (93,94). In our cell type scenarios above, the cell of origin has (almost) the same genetic makeup as its new environment does. However, the cell states landscape is transformed by the balls jumping over the barriers separating the canals, driven by cancerous epigenetic processes, such as those promoting over- (under-) expression of oncogenes (tumor suppressors). The shifted landscape reflects the now-stabilized resistant states. With the resulting altered transcriptomes and cell signaling, cell functions can be transformed.

FIGURE 1.

Waddington’s epigenetic landscape of cell fate illustrates differentiation during normal development of cells. Normal cell development progresses from pluripotent stem cells to multipotent stem cells and finally to progenitor cells. Pluripotent stem cells can develop into any type of cell in the body. Multipotent stem cells are restricted to developing into cells of a specific tissue or germ layer. Progenitor cells can differentiate into a limited number of specialized cell types. The final stage is terminal differentiation, in which progenitor cells become specific, mature cell types with defined structures and functions. Transdifferentiation is the process by which a cell switches directly from one differentiated cell type to another, bypassing the intermediate stem cell stage. Reprogramming is the process of converting one cell type into another by reverting a specialized cell to a more primitive, stem cell-like state.

A functional cell type has been thought of as controlling homeostasis and differentiation through circuits among functionally specific genes (95). In 1969, Kauffman suggested that following multiple transitions even randomly constructed gene regulatory networks can converge toward this (stable) state, and that stability can be encoded by combinations of connected transcription factors that control specific genes (96–98). Subsequent work proposed that rather than random network wiring, functional complexity requires it to be highly organized and modular (99–103), with master regulators being tightly regulated through feedback loops (104), a proposition that was later validated by multiple works [e.g., (98,105–111)]. However, with (almost) identical genomes in organismal cells, the root cause controlling homeostasis and differentiation rests with chromatin structure—the transcriptionally available genes and their epigenetic regulation. That is, the linchpin is the transcriptome landscape—not only its available states—but as depicted in the Waddington diagrams—the transitions, captured by Waddington’s hills, making the genes transcriptionally available and compliant (19,112–115). Thus, it is not the cell type that captures cell life. Instead, it is the cell states. Within this emerging complex cell states fabric, the intracellular spatial environment of the proteome landscape is segregated by atomic features and chemistry that selectively match into function-specific and efficient membraneless condensates.

Taken together, cell types were viewed as long-lasting, having specific functions and morphologies, whereas cell states as temporary, responding to the environmental stimuli. Cell types have been associated with specific cell lineages that cells follow. However, while recognized as undergoing developmental and pathological changes, such as proliferation, the phenotypic descriptive framework made it challenging to explain and quantitate. This is not the case for single-cell transcriptomic [and genomic (3)] landscapes of cell states over time, which can provide a high-resolution view of transitions between cell states. These provide the mechanism, illuminating in detail the metaphor Waddington imagined over 60 years ago and enduring to date.

BIOMACROMOLECULAR CONDENSATES ARE CRITICAL FUNCTIONAL ELEMENTS OF CELL STATES

Much is still not understood about how molecular condensates accomplish function in living cells (116,117). Functional output depends on allostery, which necessitates molecular proximity. It requires protein interactions (64,65,118–120) and a favorable environment for the molecules to adopt binding-favored conformations (64,65,121,122) (Fig. 2). Function needs a series of coordinated interaction events, further requiring that these proteins be sufficiently close to efficiently promote such events (64,123–126). Function involves regulation. It needs to retain a homeostatically stable internal environment despite changes in the external milieu. Since regulation relies on feedback loops, the involved proteins and their regulators need to be nearby. This description illustrates the problem that the cell faces: despite high crowding, it must abide by functional necessities (28,30,127–136). To maximize productivity with limited resources the cell must optimize its spatial organization. One solution is formation of membraneless condensates, a purely physical phenomenon.

FIGURE 2.

Biomolecular condensates in cell. Biomolecular condensates are membraneless organelles that perform specialized functions within cells: (i) They promote proteins to be in close proximity, (ii) allowing for coordinated protein-protein interactions (PPIs). (iii) These PPIs involve allostery, (iv) which is necessary for accomplishing cellular function. In eukaryotic cells, condensates can be divided into two groups: cytoplasmic and nuclear condensates. Cytoplasmic condensates include membrane cluster, signalosome, stress granule, TIS granule, P-body, and U-body. Nuclear condensates include nucleolus, Cajal body, nuclear speckle, heterochromatin, and superenhancer. In particular, the membrane cluster and the signalosome are supramolecular assemblies of various sizes. Examples of membrane clusters include biomolecules that participate in the Ras/Raf/MEK/ERK and PI3K/AKT/mTOR pathways. Examples of membrane cluster include Ras/Raf/MEK/ERK and PI3K/AKT/mTORC2. Examples of signalosomes include clusters of biomolecules formed by mTORC1 and lysosomal proteins.

To date, the free energy landscapes of proteins in condensates have not been fully worked out. Accurately capturing them under conditions prevailing in condensates is challenging, because the higher concentration constrains molecular motions and conformational transitions. Furthermore, these large assemblies may challenge nuclear magnetic resonance (NMR), and their low (40% volume fraction) dilution likely influences molecular dynamics (MD) simulation force fields, especially where interactions between molecules are significant. Variations in dilution alter the effective concentration of interacting molecules, affecting the strength and range of intermolecular forces, while for the cell, these conditions can imply tighter interactions, more efficient cell signaling, and catalysis. Favorable binding accomplished at short intermolecular distances often increases reaction rates, but crowding, which restricts conformational dynamics and reduces diffusion, can hinder it (137).

Conformational ensembles in the condensates’ environment

Conformational ensembles underlie all protein functions (21,121,138–141) and their distributions differ between aqueous solutions and crowded condensates, impacting their conformational preferences and, consequently, their interactions (116,142). Low water activity and reduced space —both consequences of high local protein concentration—influence the kinetics and relative stabilities of the ensembles and their transitions, shifting them toward the native state (64,65,143). Increased macromolecular crowding influences single proteins (144) and favors binding interactions, especially of nonpolar surfaces (136,145,146). Crowding diminishes available solvent, making molecular interactions energetically favorable by reducing their exposed surface area to the solvent. This is especially true for the less soluble nonpolar surfaces, resulting in a reduction of their solvent contact. Intrinsically disordered proteins, or protein segments, which lack a hydrophobic core, fold upon binding via hydrophobic interactions. While unfolded on their own, upon binding they often form large nonpolar buried interfaces resulting in stable protein complexes (147–149). They also contribute to electrostatic interactions and hydrogen bonds, creating molecular meshes in the condensates.

Condensates can anchor the assembly in favorable spatial location, as exemplified by the Ras proteins, whose C-terminal tails are key components that anchor them to the cell membrane, near receptor tyrosine kinase (RTK) clusters (150–153) (Fig. 2). This anchoring is crucial for Ras to function as a signaling molecule. RNA contributes to the formation of condensates through electrostatic interactions. Disordered protein segments are often decorated by bulky posttranslational modifications (PTMs), contributing to their disordered nature. Bulky, negatively charged phosphates are one example; lipids are another, such as palmitoylation and farnesylation in N-Ras and H-Ras tails. Further, small molecules, in particular ATP, could have significant functional effects on condensate growth, with kinases playing a key role (154–159). Phosphorylation can promote (or inhibit) the formation and growth of biomolecular condensates, influencing the assembly and disassembly of condensates. The denser environment also supports their intra-condensate actions. The phosphoinositide 3-kinase (PI3K)/protein kinase B (AKT)/mammalian target of rapamycin complex 1 (mTORC1) kinase pathway increases ribosome synthesis. While AKT and mTORC1 activation enhance protein synthesis, including ribosomal proteins, they can also suppress autophagy resulting in increased molecular crowding (137,160).

RNA in the condensates’ environment

Exactly how RNA structure contributes to condensate formation is not entirely clear. RNA sequence and structure were suggested to control assembly and function of RNA condensates lacking proteins (161). RNA properties such as composition, length, structure, modifications and expression level were explored, and proposed to work by modulating the biophysical features of native condensates (162). Structures that were discussed include hairpin loops in the presence of magnesium ions (Mg²⁺) to which proteins can bind. A study focusing on the influence of mRNA structure in phase separation using the Whi3 protein and a native mRNA binding partner, CLN3, suggested that RNA structure controls the composition and network dynamics of condensates (163). However, detailed favored RNA conformations under the condensates’ dilution conditions constrained by crowding are still missing, despite the condensates being the functional hubs (164).

Kinetics in the condensate life

As to the molecular (kinetic) basis of biomacromolecular condensate nucleation, growth and dissolution, the effective K_d was observed to be reduced (affinity is increased), while the unbinding rate (k_off) did not change. Based on in vivo experiments and simulations, it was concluded that the decrease in effective K_d can be largely due to higher effective association rate constant k_on, resulting from the increase in effective concentration (116). That is, molecular condensation reduces the effective K_d for chemical bond formation, but also restrains condensate growth by limiting molecular diffusion, hindering condensate mergers across the cell (116).

The hydrophobic effect, complementary protein surfaces, and allostery in the condensates

Biomacromolecular condensates result from phase separation, often driven by the nonpolar surfaces of proteins and their interactions. Their underlying basis is the hydrophobic effect. Favored interactions of specific proteins with complementary surfaces can trigger the allosteric effect, permitting phosphorylation by kinases—the core of signaling pathways—and signal transduction. Boosting these effects are multivalent interactions, which drive phase separation including π-π interactions, cation-anion interactions, dipole-dipole interactions, hydrophobic interactions, and electrostatic interactions (165–170). Dynamic conformational transitions, which are cardinal for efficient signaling, and the functionally critical condensate organization are also favored within a certain dilution range. Key to viable functional biomolecular condensates is their location. They are not randomly distributed. Instead, they are preferentially attached to the cytoskeleton, which is dynamic and whose network spans the cell, and to the cell membrane, organelles, and more, making them viscoelastic. Possessing viscous and elastic characteristics make them able to deform under stress like a liquid and recover their original shape like a solid. Their attachments allow efficient signaling response to external cues via membrane-crossing receptors activated by external chemicals or hormones, and transducing the signals in the cell, down to the nucleus, and other organelles. Small biomolecular condensates are widespread across the cell. They merge and work by fusion, forming larger, more complex structures through migration from smaller to larger condensates to reduce overall surface energy, through specific interactions between condensates.

Physics-based condensates match the physics-based nature of cell states

With condensates being the vehicle carrying cell function and sensitive to signals from the environment and nutrients, their compositions and abundance can be expected to fluctuate across transient cell states. The dynamic nature of cell states, which transition with changing conditions, demonstrates why the properties of physics-based condensates are relevant to cell life.

ORIGINAL CANCER CELLS TRANSITIONING TO OTHER TISSUES ARE PRIMARILY INFLUENCED BY CELL STATES

Cancer cells emerge when a normal cell state transitions to an oncogenic one. In cancer evolution, one cancerous state transitions to another largely through changes in its transcriptomics, triggered by mutations. Key mechanisms include epithelial-to-mesenchymal transition (EMT) for detachment, and mesenchymal-to-epithelial transition (MET) for colonization, collectively enabling metabolic adaptation, treatment resistance, and motility (171,172).

Intra- and inter- tumoral heterogeneity can be genetic and non-genetic (173). Vogelstein and his colleagues observed that untreated primary tumors and their metastases are genetically homogeneous, largely documenting the same mutations (174). However, their expression differs. Whereas some mutations were expressed in all tumor cells, others were observed only in certain tumor cells, resulting in different tumor phenotypes. This pointed to nongenetic effects (174), including small copy number alterations (175–180), and larger events as in the case of ERBB2 or EGFR gene amplifications. As to genetic, single point mutations, some are activating drivers, others are passengers or latent drivers. Latent drivers behave like passengers. They do not confer an observable cancer hallmark. However, coupled with other mutations, they promote cancer development as drivers do (181–183).

Increased rate of phenotypic variation points to emergence of new competing subclones bestowing a selective advantage, some with epigenetic dysregulation, including promoter hypermethylation (silencing of tumor suppressor genes), hypomethylation (promoting genomic instability and activate oncogenes), dysregulated enhancer activity (particularly super-enhancers), and alterations in chromatin structures affecting gene accessibility (influencing how easily DNA can be accessed by transcription factors and other regulatory proteins), ultimately affecting gene expression (173,184) and gene expression interaction networks. Copy number amplification of the EGFR involves lysine demethylases and methyltransferases acting on histone H3 lysine 4 (H3K4), H3K9 and H3K27 (185). Increased enhancer expression is fairly common in aggressive cancers (115), coupled with somatic copy number alterations (186). The transcriptome is also frequently aberrant (187–191), enabling using it to predict clinical outcomes (173,192).

The relatively low genetic heterogeneity in untreated cancers, intratumorally and across metastases (174), coupled with the observations that the phenotypic outcome of cancer mutations is not homogeneous over tumor cells, and that the major role in phenotypic transformation results from epigenetic alterations, supports our premise that cell states are the central players in cancer cells transitioning to other tissues. This conclusion aligns with Waddington’s metaphor (94,193): the ball, or lineage, rolling down the evolution canals, is controlled by epigenetic changes. The changes guiding and dominating cell differentiation and transformation largely rest with dynamic transitions between chromatin configurations. These changes, governed by epigenetics, define cell states, and permit cancer cells ‘adapting’ to the tissues that they are transiting into. Adaptation triggers transitioning of the cell states, ultimately peaking in the selection of cells that are more fit for survival in the specific target tissue. Thus, epigenetics, temporally modulated by the spatial (micro)environment contribute to the heterogeneity of cancer cells (70).

Do condensates contribute to cancer cells transitioning to other tissues? Our thesis is that they do—but in a different way than cell states. Condensates execute function. They do this by concentrating function-related proteins, in dynamic and conformationally favored restricted states, which promote productive allosteric communication and catalysis. In cancer, they can act as platforms for proliferative signaling, with over-/under-expressed proteins offsetting regulation of physiological pathways. Epigenetics is primarily at the cell states level. However, as to their impact on chromatin organization and regulation, while data is scant and not at high resolution, they may involve transient clusters of topologically associated domains (TADs) (194), influencing their interactions, thus cooperative allosteric behavior and cell fate transitions (195). While stable across cell types (thus states), altering phase separation can control cell fate transitions (196). Their boundaries were noted for their high transcriptional activity, spatial clustering, and interchromosomal interactions between transcription units (197).

A cell state contains multiple condensates. Chattaraj and Shakhnovich suggested multi-condensate states as functional strategy to optimize the cell signaling output (198). We support this view. As we noted above, evolution made use of purely physical phenomena.

PHARMACOLOGY OF CELL STATES AND OF BIOMACROMOLECULAR CONDENSATES

We consider the pharmacology of cell states and condensates. Condensates were discussed as drug targets, with the underlying premise that they offer a gateway to concentrated cellular components like enzymes, transcription factors, and key regulatory proteins, presenting a conducive environment for therapeutic interventions. Disrupting the condensate assembly has been deemed powerful, making currently undruggable proteins accessible, targeting multiple proteins related to the disease, and potentially reducing drug resistance (199).

However, drugs designed to dismantle condensates can be toxic due to lack of specificity. While drugs can partition cancer-associated condensates, the concentrations of the drug and target in the condensates are not uniform, and the interactions in the different condensates can vary, influencing effectiveness. Biomolecular condensates are not homogeneous across different cells in the tumor, as their expression is cell state dependent. They often exhibit spatial inhomogeneities. Since their content is function-related, the outcome of their interactions with target-specific drugs can however be powerful.

Cell states are controlled by specific epigenetics players. Epigenetic players (enzymes, readers, writers, erasers) can be targeted with small molecule drugs (200,201) and gene-editing tools like CRISPR (202). Targeting epigenetic pathways has also been considered (203). Identifying cell states for drug targeting in cancer is challenging, yet critical since cells in different states depend on distinct signaling pathways for survival. Analyzing patient datasets to reconstruct evolutionary trends is one such strategy. Single-cell technologies (like scRNA-seq) and computational methods (AI, machine learning) can help map functional profiles (proliferative, invasive, stem-like) within tumors, revealing intratumoral heterogeneity and plasticity linked to therapy resistance, and tumor evolution (13,204,205). Such approaches move beyond genetic mutations to understand the dynamic, multi-faceted nature of cancer.

Even within the same cell state, condensates composition and distributions vary, which we argue is how cell states functional adaptation happens. The different distributions can modulate biochemical reaction rates, enabling specific cellular responses. They can allow cells to tune their metabolic and signaling activities to survive changing environments (35,198).

Regarding the impact on drug sensitivities, our thesis is that differing protein compositions in condensates between cell states—such as in tumor versus healthy cells, or among tumor sub-populations—can lead to varied drug efficacy. This can result from the altered extent and residence time of the drug interactions with its intended protein targets, potentially driving resistance. Because condensates concentrate specific, functionally related proteins, their unique composition and protein-protein interactions determine target concentration and accessibility; consequently, a drug can be highly effective in one cell state but not another. In addition to protein composition in condensates, localized drug enrichment can also impact the therapeutic outcome. The aim is primary tumor-specific selection of target protein combinations (206–208). Furthermore, modeling mutant dynamics can uncover mutant-specific targets (209,210).

CONCLUSIONS

Our thesis is that cell states are fundamental to the understanding of biological processes and pathologies involving proliferation and differentiation, since they are rooted in epigenetics. With identical genomes, a single cell transforms into cells optimally customized to execute an enormous range of functions. Understanding exactly how—is the challenge that biophysics is facing. Membraneless macromolecular condensates can provide an important pointer. Even within the same cell state their composition and distributions vary, and these condensates are the primary drivers of cellular function.

In living cells, biomacromolecular condensates result from phase separation, driven by the hydrophobic effect and furthered by multivalent interactions (165–170). Key to condensates’ functional advantage is that these complementary, tightly bound proteins can transduce allosteric signals (65). Allostery is the hallmark of cell regulation. Condensates, which assemble at specific spatial locations, such as adjacent to clusters of membrane-spanning receptors, can flawlessly transmit extracellular signals down to the nucleus and serve as a key vehicle in making cell fate decisions. Condensates’ composition and interactions, their physical forces and environment, especially including their dilution state, and the mode of their extension and growth, support cell function. Modeling conformational behavior in the condensates’ environment is challenging and still needed. Further complexity arises from their fluctuations in the already highly dynamic cell states.

Collectively, the fundamental biophysics of macromolecular condensation promotes functionally related proteins to concentrate, constraining motion and conformational heterogeneity, synchronizing allostery, thus promoting signaling and regulation. Biomolecular condensates amplify cell states furthering their output, helping to clarify how complex cell decisions are accomplished. As brain studies have shown (2), much remains to be understood.

Experimentally, combining single-cell transcriptomics (scRNA-seq) with spatial proteomics could reveal the relationship between cell state dynamics and condensate composition during tumor progression. This would involve mapping functional cellular states (assessed by transcriptomics) onto their spatial location (211).

SIGNIFICANCE.

The cell type definition fails to capture evolving functions under different conditions and spatial microenvironment. The emerging cell state perception contrasts with this classical static cell type perspective. It recognizes that cell function varies with cell migration in cancer, in cell transplants, in healing, and broadly different environments. While their genomes are identical, in the cell states perspective the continuously changing epigenetics adapts the cell function to its dynamic conditions. Whereas the distinction of cell types and cell states has been described, our perspective proceeds to the next level of membraneless condensates. Even within the cell states their composition and distributions vary, and this is how cell states functional adaptation happens. We discuss implications for tumor proliferation and drug sensitivities.