Session L12: Macromolecular Phase Separation III

Live

The lipid raft hypothesis states that cellular membrane keep their composition close to a critical point. Most detractors of this hypothesis have raised an important point: maintaining such a composition for the hundreds of components in cell membranes would require extensive feedback mechanism that are still elusive. In this talk, I will show that by carefully extending the 2D Ising model to N states, finite-size effects drive the system to adopt a critical composition. Similar to biological cells, the overall composition can slowly drift along a critical manifold.   

Live

Many signaling cascades take place in the dynamic yet structured plasma membrane. Recent evidence suggests that proximity to thermodynamic phase separation through a critical point may underlie some of these membrane domains. These liquid domains, sometimes termed rafts, have been implicated in the functioning of various signaling cascades, influencing function by sorting the interactions between components. Here we present a model and Monte-Carlo simulation framework for proteins coupled to their surrounding lipid membrane using a modified 2D Ising model. In addition to simulations, we introduce schematic diagrams that predict the effects of reduced temperature on the outcomes of signaling cascades in which components have domain preference. Our model naturally explains how changes in the domain size, arising from perturbations to membrane criticality can lead to changes in the rates of interactions among signaling components, eventually leading to altered signaling outcomes. We predict that near the critical point many different cascades will have their outcomes depend sensitively on perturbations that influence the critical behavior.   

Live

It has been observed that the membrane-localized molecular assemblies of LAT:Grb2:SOS, a protein complex relevant for T-cell signaling, form condensates across a broad range of conditions. We have developed a particle based reaction diffusion model of LAT that incorporates multivalent bonding and diffusive dynamics to study the kinetics and thermodynamics of condensate formation. We find that LAT undergoes a gel transition, with significant slowing down of monomer diffusion and condensate coarsening with a small change in temperature. The mechanism of condensate formation depends senstively on the relative timescales of protein complexation and membrane diffusion, in some cases resulting in finite domains over long time scales.   

Live

Biomolecular condensates formed via liquid-liquid phase separation (LLPS) typically display multi-layered structuring. These structures are understood as LLPS of components into distinct phases and their subsequent spatial rearrangement into layered morphologies. Using a minimalistic system, comprising of a Prion-like disordered polypeptide (PDP), an Arginine-rich disordered polypeptide (RDP), and RNA, we investigate the thermodynamics of multicomponent intracellular phase behavior. We show that the ternary system forms two stable co-existing condensates, namely, PDP condensates and RDP-RNA condensates. The morphology of these condensates sensitively depend on the mixture stoichiometry and the amino acid sequence composition. Experiments and simulations indicate that these morphological changes are due to variable intermolecular interactions at the liquid-liquid interface. Predictive control of these interactions enabled us to design a diverse set of morphologies, such as completely engulfed, partially-engulfed, completely-detached and Janus-like condensates. Our findings illuminate a relation between molecular-level interactions and interactions between liquid phases at the mesoscale, highlighting a plausible mechanism for spatial organization of multiphasic condensates.   

Live

Membrane-assisted misfolding and aberrant self-assembly of amyloid beta peptides has been associated with pathogenesis of Alzheimer's disease. While several reports have suggested that an increased membrane curvature can support faster peptide aggregation, a mechanistic explanation of this relationship is missing. In this work, we have explored the effects of membrane curvature on Aβ 16–22 aggregation, using physics-based coarse-grained molecular simulations of model membranes composed of phosphatidylcholine lipids. Our simulations agree with experimental observation of a positive correlation between increased peptide aggregation and membrane curvature. The initial competition between peptide-peptide and peptide-membrane interactions results in three regimes of peptide aggregation behavior, with low curvatures promoting peptides aggregation in solution and high curvatures on membranes. In addition, the membranes with high curvature have higher defects in lipid packing that can engage peptide’s hydrophobic groups and initiate an ordered aggregation into beta-sheet rich structures. Higher curvatures can also promote faster rearrangement of lipid molecules to increase local solvent accessible hydrophobic surface area, that is necessary for membrane associated peptide assemblies.   

Live

How diffusivity (e.g., of proteins in the plane of biomembranes) is impacted by obstruction has been explored using Lattice Monte Carlo methods for random and periodic obstacle configurations. However real systems are neither periodic nor totally random. We present a study of transient and steady-state molecular diffusion in two-dimensional ”Fuzzy” systems of immobile obstacles, i.e., systems which bridge the gap between the ideal periodic and random limits. In particular, we examine whether there are ”diffusion phase transitions”, i.e., abrupt quantitative and/or qualitative changes at some critical degree of disorder. For instance, while the crossover length r* (describing the transition from anomalous to normal diffusion) decreases when the concentration of obstacles, phi, increases in a periodic system, it is the opposite for random systems. Interestingly, r* can become a very weak function of phi in some fuzzy systems. We investigate several ways of creating tunable disorder leading to different behaviour, and introduce a parameter describing how the disorder in fuzzy systems impacts diffusion. Furthermore we introduce a new relation between the crossover length r*, and the properties of the (anomalous) transient regime, including the excess diffusivity that it generates.   

Live

Permeation of antibiotics through the Gram-negative bacterial cell envelope is a complex process. Specifically, the outer membrane serves as an effective permeability barrier. To shed light on the energetics of antibiotic permeation across the outer membrane, we employed a molecular dynamics (MD) simulation approach to obtain the free energy profiles as various clinically important antibiotics were pulled across an Escherichia coli outer membrane model. Here, we deliver the first reported free energy estimates of erythromycin, gentamicin, novobiocin, rifampicin, and tetracycline as they cross the asymmetric outer membrane. While all antibiotics free energy curves have similar trends, the relative free energy barriers could be significantly different when each antibiotic permeates the outer membrane. Our results allow ranking of the outer membrane permeability of various antibiotics based on their free energy and diffusion coefficient values along with different segments of the asymmetric outer membrane. These detailed findings of drug/membrane interactions provide critical insights for understanding drug permeation and efficacy against Gram-negative bacteria such as E. coli.   

Live

Cellular membranes are lipid bilayers which are composed of two asymmetric leaflets and play an important role in regulation. In particular, the coupling and inductions between leaflets can help to colocalize proteins in the inner and outer leaflets, and thereby contribute to signaling processes. In experiments with model membranes, it was shown that in asymmetric bilayers domain formation in one leaflet may suppress or induce domain formation in the other leaflet, ultimately leading to three-phase coexistence in the membrane. This finding of three-phase coexistence hints to the fact that asymmetric membranes exhibit a tricritical point with special thermodynamic properties. To investigate the conditions under which such a tricritical point occurs, we consider an Ising-type lattice model with two coupled layers representing the two leaflets of cell membranes. Combining Monte-Carlo simulations with a Landau free energy approach, we find a line of tricritical points requiring both a coupling between the leaflets and leaflet asymmetry. Moving forward, we hope to suggest specific experiments to test our model predictions.   

Live

Graphene-based nano-technology is the future of biomedical devices. As cellular membrane is the first target of any foreign molecule to interact with living organism, here, the membrane is mimicked by stack of lipid bilayers and its structures have been probed by x-ray reflectivity and grazing incidence diffraction techniques. Lipid-GO complex has exhibited two sets of lamellar diffraction peaks illustrating a GO-rich micro-domain present in the matrix of phospholipid bilayers (GO-poor phase). In GO-rich phase, the flakes are observed to penetrate into the hydrophobic core of the membrane altering the bilayer thickness and electron density while in GO-poor phase they are destined to remain in hydrophilic head group region [1]. To understand the role of electrostatics, a single layer of lipids formed at air-water interface with varying charge entity has been. The surface pressure-area isotherm shows an interaction of GO with positively charged lipid while it shows a weak interaction with the zwitterionic one. The interaction with a negatively charged lipid was almost negligible. X-ray data show that the flakes bring a discerning effect in the out-of-plane ordering of the positively charged lipid layer, whereas it does not alter the in-plane structure.   

Live

Our work probes the behavior of complex, soft matter formulations–often comprised of macromolecules–by leveraging the strengths of both a particle and a field representation. Mesostructured polymeric solutions are difficult to study using traditional particle-explicit approaches (e.g., molecular dynamics) due to the disparate time and length scales, while the predictive capability of field theories is hampered by the need to specify emergent parameters (e.g., chi parameters) with nonobvious connections to molecular architecture. To overcome the weaknesses of both, we use small-scale, atomistic simulations to parameterize the statistical field theory models. Subsequently, field-theoretic simulations can probe behavior at larger length scales in these complex solutions efficiently while maintaining a rigorous connection to the underlying chemistries. This synergistic computational approach opens the door to explore–de-novo–the phase behavior of a wide variety of industrially relevant formulations, e.g., emulsions, coacervates, micelle assemblies, colloidal suspensions, and block copolymers. We demonstrate the capability of this approach by predicting the aqueous, PEO phase diagram, and the composition dependence of Pluronic® microphases.   

Live

Proteins, RNA, and DNA phase-separate into liquid-like droplets in the 3D cytoplasm and nucleus. In 2D, lipids forming the plasma membrane are suggested to lie near a miscibility critical point below which they separate into coexisting liquid phases. Some phase-separated structures in the neuronal and immune synapse exclusively localize to the plasma memebrane. These structures are enriched in components that strongly partition into one of the membrane phases, and polymer domains often co-localize with particular membrane components. Here we explore the physical underpinnings of these domains. Using lattice Monte-Carlo simulations and a minimal Landau theory we find that surface-localized protein droplets are best described as prewet – a 2D phase of both 2D and 3D components which is stable outside of 3D coexistence regions. Criticality in the membrane greatly expands the prewet region of the phase diagram, within which lipid and protein components phase separate together. The underlying phase diagram exhibits coexistence of up to 3 unique surface phases at the membrane. Our theoretical results make concrete predictions for experiments in synthetic systems and explain a range of previous observations across biology   

Live

Liquid-liquid phase separation has emerged as a mechanism of biological organization, particularly within the cell nucleus. However, the impact of chromatin on the dynamics of phase separation is poorly understood. Here, we utilize a powerful optogenetic strategy to examine the interplay of droplet coarsening with the surrounding chromatin network. We demonstrate that droplet growth dynamics are inhibited by the viscoelastic environment, giving rise to a slow coarsening exponent, β≈0.12, contrasting with the classical prediction of β=1/3. Using scaling arguments and simulations, we show how this arrested growth can arise due to subdiffusion of individual condensates, predicting β=α/3, where α is the diffusive exponent. Tracking condensates within chromatin reveals α≈0.5, explaining the anomalous coarsening behavior. Our work has implications for the regulation of the size and shape of biomolecular condensates and suggests that condensate emulsions can be used to probe the viscoelastic environment within living cells.   

Live

Ribonucleoprotein granules, such as stress granules, P bodies, and the nucleolus, are fluid-like sub-cellular condensates formed through liquid-liquid phase separation of multivalent proteins in association with cellular nucleic acids. Here we study how variations in RNA-to-protein stoichiometry modulate the phase behavior and condensed phase fluid dynamics. We find that RNA mediates a protein reentrant phase transition that is marked by a condensation and a subsequent de-condensation process in response to the change in RNA-to-protein ratio at a fixed protein concentration. Employing particle tracking using video microscopy and optical-tweezer induced active droplet fusion, we show that the viscosity and surface tension of protein-RNA condensates are not fixed, but vary across the reentrant phase space. In parallel, we use fluorescence correlation spectroscopy (FCS) to quantify protein and nucleic acid diffusion in the condensed phase. Results of our experiments reveal that the mixture composition plays a fundamental role not only in the formation and dissolution of multi-component heterotypic condensates, but also in controlling their material properties.