Phase Transitions in Polymeric and Protein Systems

Over recent decades, model lipid membranes have become standard systems for examining the phase diagrams of domains of the cell membrane. Interrogation of lipid domains within membranes is generally done with fluorescence microscopy via exogenous chemical probes. However, fluorophores for membrane imaging have a series of hurdles, e.g. limited domain partitioning tunability and potential interference in the phase diagram. I will present recent developments for fast and sensitive chemical imaging of phase-separated lipid membranes using the Raman scattering effect: a label-free, chemically quantitative method with high spatial resolution. These recent developments have potential to interrogate other phase-separated systems, such as proteins and RNA liquid-liquid transitions.

Emulsions are phase separated mixtures where a condensed phase coexists with a liquid solvent phase. We discuss two examples in which both chemical transitions and phase separation play a relevant role. First we consider a fuel driven chemical network which results in anomalous droplet ripening. Second we discuss the phase behaviour of a DNA mixture in which DNA can undergo conformational transitions. In both systems the phase equilibrium competes with the equilibrium related to the chemical transition giving rise to either an anomalous fast coarsening kinetics or a transition driven phase transition.

During a first-order phase transition, an interfacial layer is formed between the coexisting phases and kinetically limits homogeneous nucleation of the new phase in the original phase. This inhibition is commonly alleviated by presence of impurities, often of unknown origin, that serve as heterogeneous nucleation sites for the transition. Living systems present theoretical opportunity: The regulated structure of living systems allows modeling of the impurities, enabling quantitative analysis and comparison between homogeneous and heterogeneous nucleation mechanisms, usually a difficult task. Here, we formulate an analytical model of heterogeneous nucleation of holes in the nuclear lamina, a phenomenon with implications in cancer metastasis, aging and additional diseases. Then, we present measurements of hole nucleation in the lamina of nuclei migrating through controlled constrictions and fit the experimental data to our heterogeneous nucleation model as well as a homogeneous model. Surprisingly, we find that different mechanisms dominate depending on the density of filaments that comprise the nuclear lamina.

Living cells organize a plethora of different biochemical reactions by creating compartments or membrane-bound organelles, which are distinct chemical environment. In addition to membrane-bound organelles, there are intracellular organelles that are devoid of an enclosed membrane and are primarily composed of proteins and RNA. These membrane-less organelles play diverse roles in a variety of biological processes and are also implicated in protein aggregation. An increasing body of work suggests that membrane-less intracellular organelles are multicomponent viscous liquid droplets composed of intrinsically disordered proteins (IDPs) as well as regions (IDRs). This class of proteins/regions have an intrinsic preference for conformational disorder and are often characterized by low complexity (LC) domains. It is reported that several IDPs/IDRs and LC regions in proteins phase separate as a response to stress conditions and the phase-separated state predispose the protein toward the formation of aggregates. However, the fundamental molecular drivers and the sequence of events that govern the intracellular phase transition into liquid-like droplets and ordered aggregates remain poorly understood. Using a multidisciplinary approach involving a combination of biophysical, biochemical, molecular biology and imaging tools, I have embarked upon studies aimed at characterizing the role of conformational dynamics, heterogeneity and intermolecular association that dictates liquid-liquid phase separation (LLPS) of an IDR. I will discuss our recent results on the LLPS of an IDR derived from a functional amyloid-forming protein that phase separates into liquid-like droplets which can further promote the formation of fibrous aggregates.

Intrinsically disordered proteins differ from globular proteins in their high dynamic secondary and tertiary structure and their ability to undergo liquid-liquid phase separation (LLPS) (1). For these reasons, conventional structure-based drug discovery cannot be applied to IDPs and other strategies need to be sought (2). In this purpose, we investigated how to inhibit an IDP by modulating its LLPS behaviour with small-molecules. Androgen Receptor (AR) is a hormone-activated transcription factor. AR N-terminal domain (AR-NTD) is intrinsically disordered. AR over-activation leads to prostate cancer and, eventually, castration-resistant prostate cancer (CRPC) for which there is currently no treatment (3). EPI-001 is the only small molecule inhibitor of the AR-NTD and was identified by phenotypic screening (4). A derivative of EPI-001 entered clinical trials for CRPC treatment although not much was known about its mechanism of action. Our studies show that EPI-001 interacts with a region of the AR-NTD called Transactivation Unit-5 with very low affinity (5). In my poster I will present how EPI-001 modulates AR-NTD phase separation boundaries. Our results help understand the mode of action of this experimental drug and exploit the LLPS process as a new avenue for targeting proteins rich in intrinsic disorder such as transcription factors. 1 Shin, Y. et al. Science. 2017, 357, 1253–1264. 2 Metallo, S.J. Curr. Opin. Chem. Biol. 2010, 14, 481–488. 3 Gelmann, E.P. J. Clin. Oncol. 2002, 3001–3015. 4 Andersen, R.J., et al. Cancer Cell. 2010, 17(6), 535–546. 5 De Mol, E. et al. ACS Chem. Biol., 2016, 11 (9), 2499–2505.

During embryonic development sexually reproducing species rely on the segregation of germ granules as one characteristic to specify their germ line. In $\textit{C. elegans}$, P granules, a kind of germ granule, have been found to behave as liquid-like protein condensates. The underlying biochemical control of the segregation has been described as an mRNA competition mechanism. Furthermore, it has been suggested that this drives segregation via spatially defined changes in the phase separation behavior of the condensates. Using physical principles underlying phase-separation, we are able to rescue the asymmetric localization of P granules in mutants with defective segregation $\textit{in vivo}$. We replace biochemical control with a localized temperature gradient that mimics its physical mechanism. Furthermore, with this approach, we are able to invert the endogenous spatial distribution of P granules in zygotes. This enables us to study the dynamics of $\textit{in vivo}$ phase separation via controlled physical perturbations. In this study we conclude, that P granule segregation is a spatially tuned, diffusive-flux dependent, dissolution-condensation phenomenon.

CPEB4 is an RNA-binding protein that regulates the translation of CPE-containing mRNAs. Phosphorylated CPEB4 is in a monomeric state and active for cytoplasmic polyadenylation, while unphosphorylated CPEB4 is inactive and undergoes liquid-liquid phase separation. These liquid droplets are formed through intermolecular interactions between residues in the intrinsically disordered N-terminal domain (Guillén-Boixet, J. et al. Elife, 2016). It has recently been discovered that there’s a neuron specific microexon which is mis-spliced in Autism Spectrum Disorder (Parras, A. et al. Nature, 2018). The aim of our work is to characterize in detail the liquid-liquid phase separation process undergone by CPEB4, to determine the inter-molecular interactions that stabilize the droplets and to study the effect of post-translational modifications. It is also of great relevance to study the LLPS of different isoforms of the protein of great biomedical relevance, as mis-splicing of CPEB4 gives rise to autism-like phenotype. In this communication we will present evidence that the latter severely alters the phase diagram of CPEB4 and will propose possible rationales for this behavior.

Recently emerged reports have established that in eukaryotic cells, many biochemical processes take place in temporary membrane-less compartments which are formed by the association of certain RNA binding proteins (RBPs). These compartments e.g. nucleolus, P bodies etc have liquid like properties and are formed by a process known as liquid-liquid phase separation (LLPS). The RBPs that undergo LLPS harbour a disordered prion-like domain that is rich is polar amino acid and under certain conditions undergo a reversible LLPS. Once the biochemical process is accomplished, these compartments dissolve back into the bulk solution (nucleoplasm or cytoplasm). Under certain conditions such as in mutations, reversibility of LLPS compromise and RBPs form stable pathological aggregates which are causative agents of diseases like ALS, FTD etc. In the present study, TDP-43 has been chosen as representative RBP, aggregation of whose is implicated in ALS and FTD. Owing to the difficulty in purification, most studies on TDP-43 have been carried out on its truncated versions or short peptides. We purified full length TDP-43 through oxidative refolding and subjected to various buffer conditions in which different polyol osmolytes were present as macromolecular crowder. The recombinantly purified TDP-43 underwent LLPS and droplets were visualized under the microscope and the extent of droplet formation was quantified through light scattering. We observed that with increasing crowder concentration, the propensity of TDP-43 undergoing LLPS increased. Ours is the first of very few attempts to study the role of osmolytes in modulation of LLPS process. This investigation has a potential to unravel the role of osmolytes in modulation of LLPS behaviour and thus a possible clue in phase separation could be gleaned by studying their mechanism of action.

Emulsions are phase separated mixtures where a condensed phase coexists with a liquid solvent phase. We discuss two examples in which both chemical transitions and phase separation play a relevant role. First we consider a fuel driven chemical network which results in anomalous droplet ripening. Second we discuss the phase behaviour of a DNA mixture in which DNA can undergo conformational transitions. In both systems the phase equilibrium competes with the equilibrium related to the chemical transition giving rise to either an anomalous fast coarsening kinetics or a transition driven phase transition.

Formation of droplets by phase separation plays an important role in the spatio-temporal organization of matter in biological cells. Precise control over the droplets' properties is necessary to regulate intracellular processes. To achieve this, cells drive chemical reactions affecting the droplet material. By varying the reaction rates, the size and growth of such active droplets can be controlled. We study the dynamics of active droplets by numerically solving a modified Cahn-Hilliard equation and comparing the results with analytical predictions. In the first project, we aim to understand the dynamics of active droplets in heterogeneous environments. We start by considering a single droplet in an external chemical gradient. This helps us to examine two and more active droplets in close vicinity. In the second project, we study pattern formation of active droplets. In particular we investigate the two-dimensional case, where hexagonal patterns are formed. Exploiting an analogy with equilibrium systems with long-range interactions, we determine droplet sizes and spacings. Both these projects will help us to understand the dynamics of active droplets and how cells could use chemical reactions to organize intracellular matter.

The processes that contribute to protein expression are subject to stochastic fluctuations and are affected by the environment in which they operate. As a result, concentration of a given protein can vary greatly between organisms, cells, as well as in time. Since many biological processes demand a tight control over protein concentration, cells have evolved various mechanisms to control the degree of concentration variability often referred to as noise. The best studied mechanisms for buffering protein expression levels rely on feedback through transcriptional regulation. Such regulation systems are slow and can reduce the expression noise only to a certain level. Here we explore the potential for a phase separated organelle to buffer the noise in protein concentration at the post-translational level. Based on a simple thermodynamic model, we predict that liquid droplets function as dynamic reservoirs, which can buffer variations in a highly effective and near-optimal manner. Using an engineered fluorescent protein that forms liquid droplets in the nucleus of Hela cells we show that phase separation attenuates variations in protein concentrations by up to a 100-fold. We propose that phase separation could be a common strategy for achieving extremely stable protein concentrations in cells.

Tau is an intrinsically-disordered protein, which diffuses on microtubules. In neurodegenerative diseases, collectively termed tauopathies, tau malfunction and its detachment from axonal microtubules is correlated with microtubule degradation. Using in vitro reconstitution, we show that tau molecules on microtubules cooperatively form cohesive islands which are kinetically more stable than molecules diffusing individually. Dependent on the flux between diffusive and stable phase of tau, islands reversibly grow or shrink by addition or release of molecules at their boundaries. We observe such flow in bulk measurements as well as in single molecule experiments, where we see individual tau molecules switching between states. Furthermore, we show that while the diffusive state is of hydrophilic nature, the stable, cooperative state is more hydrophobic in nature.

presented on behalf of Paolo Arosio Membraneless compartments resulting from liquid-liquid phase separation (LLPS) exhibit rapid internal mixing, fast exchange with the cytoplasm, and transform rapidly via a constant flux of molecules. In this context, kinetics, in addition to thermodynamics, define the system’s behavior. Despite the essential role of dynamics in living organisms, the mechanisms of assembly and disassembly have remained essentially unexplored, largely due to the lack of suitable tools to access the biologically relevant timescales and volumes of the dynamics of the phase transition. Here, we leverage microfluidics technology and develop microfluidic mixers and reactors to reproduce the liquid-liquid phase separation at the microscale. Specifically, we build on the ability of microfluidics to control multiphase flow and generate well-defined microcompartments, which allow us to rapidly mix reagents via chaotic advection and observe nucleation and growth events in volumes comparable to cells (pL). We demonstrate the power of this approach and analyze the microscopic processes underlying the phase separation of with the model protein DEAD-box ATPase Dhh1, which is associated to the formation of Processing bodies and undergoes phase transition in the presence of ATP and RNA. Moreover, we demonstrate the possibility to mimic these mechanisms and induce similar behaviours in soluble proteins by conjugating low complexity domains to soluble globular regions. We show that these biologically derived molecular adhesives enable
 the self-assembly of these proteins into supramolecular
 architectures via a
 multistep process. This multistep pathway involves an initial liquid−liquid phase transition, which creates protein-rich droplets that mature into protein aggregates over time. These protein aggregates consist of permeable structures that maintain activity and release active soluble proteins. We further demonstrate that this feature, together with the dynamic state of the initial dense liquid phase, allows one to directly assemble different globular domains within the same architecture, thereby enabling the generation of both static multifunctional biomaterials and dynamic microscale bioreactors. References Faltova L., Küffner A. et al, “Multifunctional Protein Materials and Microreactors using Low Complexity Domains as Molecular Adhesives”, ACS Nano, 2018, 12, 9991-9999

Membraneless compartments resulting from liquid-liquid phase separation (LLPS) exhibit rapid internal mixing, fast exchange with the cytoplasm, and transform rapidly via a constant flux of molecules. In this context, kinetics, in addition to thermodynamics, define the system’s behavior. Despite the essential role of dynamics in living organisms, the mechanisms of assembly and disassembly have remained essentially unexplored, largely due to the lack of suitable tools to access the biologically relevant timescales and volumes of the dynamics of the phase transition. Here, we leverage microfluidics technology and develop new microfluidic mixers and reactors to reproduce liquid-liquid phase separation at the microscale. Specifically, we build on the ability of microfluidics to control multiphase flow and generate well-defined microcompartments, which allow us to rapidly mix reagents via chaotic advection and observe nucleation and growth events on a time scale of seconds in volumes comparable to cells (pL). In addition, we collect and store these microcompartments to monitor changes of the biomolecular condensates over longer incubation times of several minutes and hours. We demonstrate the power of this approach with the model protein DEAD-box ATPase Dhh1, which is associated to the formation of processing bodies and undergoes phase transition in the presence of ATP and RNA. The analysis indicates that increasing the volume of the compartment increases both the rate of formation and the size of the condensates. Importantly, we show that the kinetics of growth of the droplets is consistent with a mechanism based on droplet coalescence and not with Ostwald ripening. We further show that we can mimic the cytoskeleton by introducing a hydrogel into the droplets, which prevents the coalescence of the condensates into one single compartment.

Many membrane-less compartments in eukaryotic cells are protein-rich biomolecular condensates formed via phase separation from the cyto- or nucleoplasm. Condensate physicochemical properties, such as protein concentration, mesh size, and viscoelasticity, emerge from the interactions of the constituent molecules, and are thought to be tuned over evolutionary time to facilitate the specific biological functions of the compartments. However, a predictive understanding of how condensate properties are encoded by the amino acid sequences of scaffold proteins, which contribute the bulk of the non-aqueous condensate mass, is currently lacking. Building on the recent discovery that the saturation concentration of FUS/EWSR1/TAF15 protein family members is controlled by cation-pi interactions between Arg and Tyr residues, we examine the phase equilibria of constructs derived from the ancestral family member, TAF15. We use quantitative phase microscopy and optical diffraction tomography to measure the 3D refractive index distribution of protein-rich droplets following in vitro phase separation, from which we calculate the protein concentration of the condensed branch of the two-phase coexistence curve. We examine the role of electrostatic and cation-pi interactions in setting the condensate protein concentration with mutant constructs and by variations in ionic strength, and compare to the predictions for the Flory-Huggins mean-field theory. An improved understanding of residue-specific contributions to phase separation and the physical properties of the condensed phase should enable the modulation of condensate properties by rational mutagenesis, an invaluable tool for finally elucidating the degree to which the physical properties of biomolecular condensates are fine-tuned to facilitate specific biological functions in vivo.

To ensure correct chromosome segregation, cells must align their mitotic spindle perpendicular to the future division plane of the cell. In metaphase yeast, the Kar9-Bim1(EB1) complex connects astral microtubule tips to the actin motor Myo2. In turn, Myo2 pulls the spindle along polarized actin cables towards the bud (the future daughter cell). Interestingly, Kar9 localizes to only one astral microtubule emanating from the old spindle pole body (SPB, yeast centrosome). Consequently, only that spindle pole is pulled and oriented towards the bud, aligning the spindle perpendicular to the future cell division plane. It is unknown how the cell ensures this asymmetric recruitment of Kar9 to one microtubule, how Kar9 and Bim1 manage to stay attached to it through cycles of microtubule polymerization and depolymerization and how Kar9-Myo2 is maintained despite the forces generated by microtubule shrinkage. We propose liquid phase separation of Kar9, Bim1 and other microtubule plus-end tracking proteins (+TIPs) such as Bik1 (CLIP-170) into a ‘+TIP-body’ as a mechanism behind those puzzling observations. We show that the Kar9-Bim1 complex and Bik1 can phase separate together or individually in vitro. Mutations in oligomerization interfaces impair phase separation of the Kar9-Bim1 complex in vitro in absence of Bik1. Similarly, together with Bik1 deletion they cause Kar9 localization, nuclear segregation and viability defects in cells. These results indeed suggest a role of the +TIP body in spindle positioning.

Liquid-liquid phase separation of instrinsically disordered ERD14 chaperon protein induced by protein-RNA interaction MURVAI NIKOLETTA1, TANTOS ÁGNES1, SZABÓ BEÁTA1, SZEDER BÁLINT1, JOBBÁGY CSABA1, KOVÁCS DÉNES2 ÉS TOMPA PÉTER1,2 1 HAS Research Centre for Natural Sciences, Institute of Enzymology, Budapest, Hungary 2 VIB Department of Structural Biology, Vrije Universiteit Brussel, Brussels, Belgium Within the cell, various biochemical reactions often require specific conditions which demand the individual steps be separated in space. An alternative strategy can be the formation of membrane-free protein-based organelles. Among others, the nucleolus [1], PML bodies [2], Cajal bodies [3], P-bodies, and also stress and germinal granules are formed inside the cell. These cellular structures have been described as coacervates and can be characterized as optically small spherical micron-sized droplets [4]. The absence of the surrounding membrane makes it possible to efficiently track environmental changes, intracellular signals and respond quickly in order to keep cellular integrity and homeostasis [5]. It is a surprising feature of membrane-free organelles that the protein-based interior excludes most of the aqueous phase and thus they are able to form liquid-liquid phases within the cell. [6, 7]. ERD14 (Early Response to Dehydration 14) is a plant stress protein found in Arabidopsis thaliana, which expressed during abiotic stress (cold, drought and high salinity) in large quantities in all plant tissues and can fill in nearly 4% of the total protein content. The activity of ERD14 chaperone was also demonstrated in vitro and in vivo studies. However, the exact mechanism of protective effect is not well-known yet. The purpose of our research is to explore the mechanism and to map the connections between structure and function. Based on our observations, the protein is capable of interacting with nucleic acids and inducing phase separation under appropriate conditions in aqueous media. The resulting liquid droplets react dynamically to environmental changes, quickly dissolve or appear in large quantities. The organelles formation stabilizes through electrostatic interactions, which requires an exact charge pattern in the protein sequence. The droplets differentiate solubilizing nucleic acids, ERD14 protein is mainly capable of RNA binding for phase transition. Based on ERD14 in vitro and in-cell NMR measurements, a high degree of structural disorder is characterised, only a few conserved regions have transient secondary structures (K segments) likely to be involved in binding partner proteins and RNAs, thereby providing a protective effect. Irodalom [1] Montgomery, T.S. (1898). J. Morphol. 15: 265–582. [2] de The´, H., Chomienne, C., Lanotte, M., Degos, L., and Dejean, A. (1990) Nature 347: 558–561. [3] Cajal, S.R.y. (1903) Trab. Lab. Invest. Biol. Univ. Madrid 2: 129–221. [4] Hyman, A.A., and Brangwynne, C.P. (2011) Cell 21: 14–16. [5] Brangwynne, C.P., Eckmann, C.R., Hyman, A.A. et al. (2009) Science 324: 1729–1732. [6] Brangwynne, C.P., Mitchison, T.J., and Hyman, A.A. (2011) Proc. Natl. Acad. Sci. 108: 4334–4339. [7] Boeynaems, S., Tompa, P., Van Den Bosch, L. et al. (2017) Molecular Cell, 65,6: 1044 – 1055. [8] Szalaine Agoston, B., Kovacs, D., Tompa, P. & Perczel, A. (2011) Biomol NMR Assign 5: 189-193.

Enhancers are transferable DNA elements that regulate target genes within the confines of Topologically Associated Domains (TADs) through Cohesin-dependent looping. In this study we report that 17β-estradiol (E2) rapidly increases the spatial proximity of a cohort of particularly strong E2-responsive enhancers located in widely separated TADs. These estrogen receptor α (ERα) bound enhancers are characterized by high enhancer RNA (eRNA) induction and recruitment of mega-dalton sized multi-transcription factor complex (MegaTrans) which together function as a ribonucleoprotein complex. In order to test the hypothesis that spatial cooperation of distant enhancers is established through phase separation events on strong enhancers, we treated breast cancer cells with 1,6-Hexanediol (1,6-HD), an aliphatic alcohol that disrupt phase separated structures, followed by assessing transcriptional activity in response to E2 by Global Run-On Sequencing (GRO-Seq). Surprisingly, 1,6-HD selectively deactivated the MegaTrans bound strong ERα enhancers, not weak ERα enhancers or strong non- ERα enhancers. 1,6-HD specifically disrupts the assembly of MegaTrans on first tier ERα enhancers. Our data suggest that long distance enhancer association and co-operativity requires E2-induced assembly of ribonucleoprotein structures composed of eRNA, Condensins and the MegaTrans complex at strong ERα bound enhancers and their co-localization with phase-separated interchromatin granules (ICGs). These findings along with other evidences that support a new model for establishment and cooperative activation of inducible enhancers following principles of liquid liquid phase separation will be discussed.

Cellular functions are achieved and maintained by the spatiotemporal regulation of biological reactions within the crowded and heterogeneous milieu of the cellular cytoplasm. The bio-molecular components of the cytoplasm, particularly proteins and RNA, dynamically phase separate and form condensed aggregates (membrane-less compartments). While these membrane-less organelles have been hypothesized to form dynamic reaction containers, fusing with and splitting from each other, the interaction between membrane-less compartments and lipid membranes has remained unexplored. The physical phase (fluid-like, gel-like or solid/glassy) of these compartments is expected to affect not only the organization and physical properties (bending, fluidity etc.) of the membranes they are proximal to, but also the reactions between the components of the compartments and the lipid membranes. In this context, we consider the case of a membrane-less organelle, containing membrane binding and other interacting molecules, wetting a lipid membrane. We specifically study the effect of the local physical phase of the organelle on the binding/unbinding kinetics of membrane binding molecules with the membrane.

PtEPYC1 is a disordered repetitive protein predicted to be involved in pyrenoid assembly and phase separation in a model diatom Phaeodactylum tricornutum (Pt). Pyrenoids—chloroplastic suborganelles primarily composed of RuBisCO—are part of biophysical carbon concentrating mechanisms (CCMs) in many algal species. Confining RuBisCO—along with other mechanisms to concentrate CO2 around its active site—ensures carboxylation over oxygenation is favored by the enzyme. Evidence for liquid-like nature of Chlamydomonas reinhardtii pyrenoid was presented last year. The current model hypothesizes that multivalency of a disordered repetitive protein EPYC1 (Essential Pyrenoid Component 1) and its interactions with RuBisCO may drive the observed pyrenoid behavior. EPYC1 analog in Pt is a chloroplast-localized disordered protein with 7 repeats and low [CO2]-upregulated transcript. It contains a regularly spaced motif known to be recognized by a member of proprotein convertase family and our data indeed show the protein is cleaved suggesting different “repeat isoforms” are functional in vivo. What is the biological role of these isoforms? Do their ratios change under different CO2 and light regimes? Do they play a role in defining the phase-separated nature of the pyrenoid?