Session T06: Biomolecular Phase Separation II - Active Droplets

Label-free composition determination for biomolecular condensates with an arbitrary number of components

Many cellular compartments are protein-rich biomolecular condensates demixed from the cyto- or nucleoplasm. Although condensates are defined by their molecular composition, traditional approaches to measure composition are inefficient or require confounding labels, and are typically limited to few components. Here, we describe a label-free method to measure the shape and composition of micron-sized condensates based on quantitative phase microscopy and the physics of sessile droplets. This method has a precision better than 2%, requires 1000-fold less material than bulk techniques, and exposes systematic errors as large as 50-fold in common approaches based on fluorescence intensity ratios. Further, we show that this method can be used in combination with standard dilute-solution detection methods to determine the composition of condensates containing an arbitrary number of molecular components. We demonstrate this explicitly by measuring the tie-lines and binodals for a ternary mixture containing RNA and the full-length RNA-binding protein FUS. In addition to recovering the expected re-entrant behavior in the dilute binodal branch with increasing total RNA concentration, our measurements reveal an unexpected kink in the condensed binodal branch above which the condensed phase maintains a constant polymer volume fraction over an extended range. Interestingly, this density is comparable to that of many condensates in vivo and significantly lower than that typical of protein condensates in binary systems, underscoring the ability of multi-component reconstitutions to more faithfully reflect in vivo conditions. Finally, we discuss the fundamental role of composition in controlling condensate mechanical, dielectric and surface properties, which are expected to collectively underly biological functionality.   

A Background Enzymatic Active Bath Affects Liquid-Liquid Phase Separation of Proteins

The cell interior is an active bath driven by a myriad of enzymes. It is an open problem as to how this background activity can affect physical processes in the cell including liquid-liquid phase separation. We seek to experimentally reconstitute a model system for an active bath of enzymes to determine the effects on the liquid phase separation of a model condensate protein. We will use urease, an exothermic and kinetically fast enzyme that converts urea to carbon dioxide and ammonia, as the background enzyme. We will use ubiquilin-2 (UBQLN2), a protein that phase separates when high salt is added. We have scanned both the salt concentration to form condensates, and the urea concentration to control urease activity and observed droplets using fluorescence microscopy. Preliminarily, we find that the enzyme activity increases both the number and size of the droplets of UBQLN2.   

Liquid-liquid phase separation and coarsening in an active fluid

 Recent studies have revealed the existence of non-membrane bound organelles which form via liquid-liquid phase separation in cytoplasmic and nucleic cellular environments. These liquid-like bodies undergo a coalesce-based coarsening process which increases their average size over time. Nucleic organelle motion and coarsening rate have been well studied, however it is not clear how the out-of-equilibrium nature of the cell cytoplasm contributes to organelle coarsening. We investigate the impact of cytoplasmic flows on liquid-liquid phase separation by immersing phase separating DNA structures in microtubule-based active fluids. In my talk, I will show that active flows yield short-time superdiffusive motion and long-time diffusive motion, and that this motion accelerates droplet coarsening compared to a passive network.  I will also show that the coarsening rate of actively mixed condensates monotonically increases with increasing activity.  Together, these results could provide a better understanding how forces generated by the cytoskeleton impact the coarsening of membranless organelles.   

Effect of ATP on Phase Separation in FUS Condensates

The fused-in-sarcoma (FUS) protein is known to phase separate from the cellular medium into liquid-like condensate droplets. Aberrant FUS phase separation is seen in diseases such as ALS and frontotemporal dementia. Adenosine triphosphate (ATP), the common cellular energy carrier, was recently observed to promote FUS condensation at low ATP concentrations and dissolve FUS condensates at high ATP concentrations. However, the mechanism by which this occurs has not yet been determined. Here, we used all-atom simulations to elucidate, at the microscopic level, how ATP affects FUS phase separation. In our simulations, two FUS proteins were placed in electrolyte solution and simulated, with and without ATP, in the presence of harmonic restraints that probed local and global inter-FUS forces. Analysis of the simulation trajectories reveals changes in the effective force between FUS molecules, number of inter-FUS contacts, and FUS radius of gyration mediated by the formation of ATP-FUS contacts and ATP bridges between FUS molecules, charting a microscopic mechanism by which ATP may affect phase separation.   

Nucleotide-dependent Dynamics and Material Properties of a Model Biomolecular Condensate

Biomolecular condensates are biochemical organizing centers within cells made up of complex mixtures of protein and RNA. The dynamics and material states of these bodies are often central to their function, controlling condensate formation and dissolution, how condensates interact with other cellular structures, and the mobility of their constituents. Identifying specific biomolecular players capable of controlling material state and uncovering fundamental principles surrounding dynamics within biomolecular condensates is crucial to our understanding of their regulation in biological systems. This information will also be key to understanding how these bodies may be leveraged for bioengineering and biomedical purposes. By studying DEAD-box helicases, a family of proteins commonly found in eukaryotic condensates, we are uncovering how nucleotide-dependent protein-RNA interactions structure biomolecular condensates at the microscopic scale, coupling measurements of biomolecular mobility to their consequences for condensate material properties. We identify two distinct modes for nucleotide-dependent regulation of condensate material state and demonstrate tunable dynamics dependent on nucleotide concentration.   

Non-coding RNAs could fine-tune transcription of proximal genes by modulating dynamics of transcriptional condensates

Bio-molecular factories called transcriptional condensates transcribe several genes associated with cell identity and cancer. They are dense assemblies of transcriptional proteins that form at genomic regions with a high density of protein-binding DNA called super-enhancers (SE). Some of these proteins have positively charged disordered domains that can interact via screened electrostatic interactions with the negatively charged phosphate backbone of RNAs. Using a free energy functional that takes into account the interactions between the different molecular species, we modeled the dynamics of concentration fields using Model B dynamics for the stable protein and non-coding RNA coupled to a reaction-diffusion equation for the transcribed mRNA that pushes the system far away from equilibrium. In the absence of active mRNA transcription, we predict that non-coding RNAs localized near SEs can aid the recruitment of transcriptional proteins by (i) serving as an attractive well to nucleate the condensate and (ii) by jumping over to the condensate after its formation to help it recruit more protein. When there is active mRNA transcription, we predict that non-coding RNAs localized nearby can initially “accelerate” protein recruitment and mRNA transcription at the SE. As more mRNA is produced, the unfavorable electrostatic repulsion between the RNA species “slows down” protein recruitment and mRNA transcription. The tussle between these two effects determines the total amount of mRNA transcribed over time. The consequences of this mechanism could explain some puzzles related to gene regulation by non-coding RNAs such as the sequence-independent effect of certain non-coding RNAs on gene expression and why non-coding RNAs promote gene expression in certain cases and repress gene expression in others.   

Physical Principles Underlying Regulation of Size Distributions of Intracellular Condensates

Condensates play crucial roles in driving biochemical reactions in the cell, and many of these functions depend on condensate size. While previous work has demonstrated that droplet growth is primarily driven by coalescence following fast quench of the system, the size distribution produced by this mechanism has not been described. We first demonstrate that this mechanism produces exponential cluster-size distributions in both optogenetic live-cell experiments and Monte Carlo simulations.  Then, to elucidate the relative effects of quench and coalescence dynamics on size distribution, we analyze Huntingtin polyQ protein aggregation in the cytoplasm, finding that it demonstrates a power-law cluster-size distribution. A similar power-law distribution was recovered by adapting our Monte Carlo simulations to model the case of slow production of condensate components. We demonstrate that these power laws are due to a preferential attachment or “rich get richer” effect. We then describe the transition between a coalescence-limited, exponential regime and a production-limited power-law regime, which can be tuned by the biologically relevant parameters of subdiffusion and material production. Finally, we apply this analysis to endogenous organelles to infer their dynamics.   

Non-equilibrium phase separation in mixtures of catalytically active particles: size dispersity and screening effects

Biomolecular condensates in cells are often rich in catalytically active enzymes. This is particularly true in the case of the large enzymatic complexes known as metabolons, which contain different enzymes that participate in the same catalytic pathway. One possible explanation for this self-organization is the combination of the catalytic activity of the enzymes and a chemotactic response to gradients of their substrate, which leads to a substrate-mediated effective interaction between enzymes. These interactions constitute a purely non-equilibrium effect and show exotic features such as non-reciprocity. Here, we analytically study a model describing the phase separation of a mixture of such catalytically active particles. We show that a Michaelis–Menten-like dependence of the particles' activities manifests itself as a screening of the interactions, and that a mixture of two differently sized active species can exhibit phase separation with transient oscillations. We also derive a rich stability phase diagram for a mixture of two species with both concentration-dependent activity and size dispersity.   

Chemical reaction kinetics in phase separated systems

The theory of phase separation determines how condensates form in living cells. These condensates act as compartments that partition molecules differently in each of the phases. This partitioning leads to different physicochemical properties of both phases that can alter the kinetics of chemical reactions. The mechanisms which determine how chemical reactions are affected by coexisting phases are not clearly understood. Here, we derive the kinetic theory for chemical reactions at phase equillibrium. We show that the condition of phase equilibrium leads to a fundamental relationship that relates chemical reaction fluxes and partitioning of reacting components. A key finding is that for chemical reactions that can relax to thermodynamic equilibrium, differences in chemical flux between phases solely stem from phase dependent reaction rate coefficients. In such cases, the average compositions of reacting components are affected by phase coexistence. In comparison, phase coexistence can alter the stationary compositions of reacting components more strongly when maintaining chemical reactions away from equilibrium. As an example of such reactions we study phosphorylation processes. Our studies exemplify the enormous potential of phase separated compartments as biochemical reactors in living cells. Understanding the control of biochemical reactions via compartments is key to elucidate the functionality of stress granules for the cell and is also crucial for biochemical communication among synthetic cells and RNA catalysis in coacervate protocells.   

Active Droplets supplied with Energy and Matter

Chemically active droplets provide simple models for cell-like systems that can grow and divide. They are maintained away from thermodynamic equilibrium and host chemical reaction cycles, which correspond to a simple metabolism. We consider two scenarios of non-equilibrium driving. First, droplets are driven via the boundaries of the system by external reservoirs supplying nutrients and removing waste. Second, droplets are driven by chemical energy provided by a fuel in the bulk. We discuss the conservation of energy and matter as well as the balance of entropy. We use conserved and non-conserved fields to discuss the rules governing energy and matter supply. Using an effective droplet model, we explore droplet stability and instabilities leading to droplet division. A key finding is that that droplet division occurs quite generally in active droplet systems. Our work reveals that life-like processes such as metabolism and division can emerge in simple non-equilibrium systems that combine the physics of phase separation and chemical reactions.   

Molecular Assembly Lines in Active Droplets

Large protein complexes are assembled from protein subunits to form a specific structure. In our theoretic work, we propose that assembly into the correct structure could be reliably achieved through an assembly line with a specific sequence of assembly steps.  Using droplet interfaces to position compartment boundaries, we show that an assembly line can be self organized by active droplets.  As a consequence, assembly steps can be arranged spatially so that a specific order of assembly is achieved and incorrect assembly is strongly suppressed.   

Quantitative Reaction-Diffusion Dynamics in Liquid Condensates

Liquid droplets have been hypothesized as reaction compartments in cell biology, which can, for example, locally increase or decrease biochemical reaction rates. Providing direct evidence of spatially varying reaction rates provides a formidable challenge in biophysics. We recently introduced a precise way of measuring intra-droplet diffusion rates based on a spatially resolved phase separation model. By extending this framework to chemical reactions, we enable data analysis of phase-separating reaction-diffusion systems. We showcase the applicability of our theoretical method by measuring reaction rates in a system with an enzymatic reaction. 

To eventually enable measuring reaction-diffusion rates in vivo, where droplets are small and approaches based on fully bleaching a condensate are set to fail, we show different fluorescence recovery after photo-bleaching (FRAP) geometries can be used to quantify molecular dynamics in small condensates reliably.   

Mesoscale properties of non-equilibrium liquid-liquid phase separation with molecular production and degradation

The role of liquid-liquid phase separation (LLPS) in the formation of cellular, membraneless organelles and chromatin organization is only starting to be elucidated by a large number of experimental studies. However, the theoretical understanding of the phenomenon is largely based on soft matter research of systems that are in thermodynamic equilibrium, which is not the case in many biological systems. Motivated by cellular LLPS, we present a theoretical framework for the analysis of out-of-equilibrium LLPS, where the phase-separating molecules are constantly synthesized and degraded. We show that while the concentrations of solutes in the coexisting domains remain the same as in equilibrium systems, the non-equilibrium nature of biological systems may modify the large-scale properties of the phase separated domains. These include the shapes and number of phase separated domains, which do not undergo Ostwald ripening, the fluctuations of the interface, and effective interactions of the ribosomes (responsible for production). For each of these properties, our theoretical model predicts the deviations from the equilibrium case and their magnitude, which may be significant in some biological scenarios.