Probing Inhomogeneous Diffusion in the Microenvironments ofPhase-Separated Polymers under ConfinementMarjan Shayegan,† Radin Tahvildari,† Kimberly Metera,† Lydia Kisley,‡ Stephen W. Michnick,*,§

and Sabrina R. Leslie*,†

†Department of Physics, McGill University, Montreal, Quebec H3A 2T8, Canada‡Department of Physics, Case Western Reserve University, Cleveland, Ohio 44106, United States§Departement de Biochimie, Universite de Montreal, Montreal, Quebec H3C 3J7, Canada

*S Supporting Information

ABSTRACT: Biomolecular condensates formed by liquid−liquidphase separation of proteins and nucleic acids have been recentlydiscovered to be prevalent in biology. These dynamic condensatesbehave like biochemical reaction vessels, but little is known abouttheir structural organization and biophysical properties, which arelikely related to condensate size. Thus, it is critical that we studythem on scales found in vivo. However, previous in vitro studies ofcondensate assembly and physical properties have involvedcondensates up to 1000 times larger than those found in vivo.Here, we apply confinement microscopy to visualize condensatesand control their sizes by creating appropriate confinement lengthscales relevant to the cell environment. We observe anomalous diffusion of probe particles embedded within confinedcondensates, as well as heterogeneous dynamics in condensates formed from PEG/dextran and in ribonucleoprotein complexesof RNA and the RNA-binding protein Dhh1. We propose that the observed non-Gaussian dynamics indicate a hopping dif fusionmechanism inside condensates. We also observe that, for dextran-rich condensates, but not for ribonucleo condensates, probeparticle diffusion depends on condensate size.

■ INTRODUCTION

Liquid−liquid phase separation is found throughout natureand has important consequences in biology.1−4 For example,biomolecular condensates, or non-membranous organelles(NMOs), are dynamic structures within the cyto- andnucleoplasm of cells that form via reversible and highlycontrolled liquid−liquid phase separation of proteins andnucleic acids. Inside cells, different NMOs perform a variety offunctions:3,5−9 concentrating molecules to enhance biochem-ical reaction rates within the diverse and crowded cell interior;isolating catalytic processes; sequestering and storing materialsfor later release; and helping cells respond to stress. Some arefurther sub-compartmentalized and may generate gradientsand chemical potentials.6 Some appear to concentrate specificmolecules while simultaneously excluding others. Throughaberrant formation, regulation, or stability, NMOs may alsocontribute to neurodegenerative conditions including Alz-heimer’s and Huntington’s diseases. However, the principlesgoverning NMO assembly, disassembly, and functions remainunknown.10−13 Furthermore, the relationships between NMOsize, properties, and functions need to be clarified. NMOs scalewith the size of the cell,5,14 and cell size dysregulation isconnected to many diseases, including cancers.15

Detailed biophysical studies of phase-separated biomolecularcondensates will help clarify how these important structures

function. Important factors for the phase separation ofbiomolecular condensates include (1) the content of intrinsi-cally disordered regions in the component proteins and (2)interactions between folded protein or RNA domains withlinear regions.16 The mechanical properties of the phase-separated condensates (e.g., surface tension and viscoelasticity)and the interactions among their component proteins andnucleic acids are expected to control condensate coalescence,internal structuring, diffusion within the condensate, andexchange of molecules with the cyto- or nucleoplasm, and thusmust heavily influence condensate function.3 For example, ithas been proposed that the viscoelastic properties of NMOscontrol enzymatic processes by lowering the flux of specificmolecular species.17

In vitro studies of biomolecular condensates allow for controlof assembly and condensate properties and have probed arange of parameters controlling phase separation (refs 3,18−22 are recent reviews of in vitro studies). However, typicalin vitro techniques cannot approach simulating the naturallycrowded, complex, and confined conditions found in cells andreconstituted condensates can vary greatly in size and be up to10−1000 times larger than those observed in vivo (∼100 nm

Received: December 14, 2018Published: April 24, 2019

Article

pubs.acs.org/JACSCite This: J. Am. Chem. Soc. 2019, 141, 7751−7757

© 2019 American Chemical Society 7751 DOI: 10.1021/jacs.8b13349J. Am. Chem. Soc. 2019, 141, 7751−7757

Dow

nloa

ded

by M

CG

ILL

UN

IV a

t 10:

38:4

0:41

0 on

May

24,

201

9fr

om h

ttps:

//pub

s.ac

s.or

g/do

i/10.

1021

/jacs

.8b1

3349

.

for P-bodies8). Most theories on phase separation ofbiopolymers also assume infinite boundaries.23 To understandnatural NMOs, it is crucial to determine how these vastlydifferent size regimes might influence condensate biophysicalproperties.

■ EXPERIMENTAL DETAILSConvex lens-induced confinement (CLiC)24−27 fluorescence micros-copy is applied to control condensate size and monitor how sizeaffects diffusion within them. In this technique, condensates areformed within tiny wells (450 nm deep and 3−50 μm in diameter) tolimit condensate growth. A fluid sample of biomolecules is loaded intoa custom-designed imaging chamber made from two glass coverslipsseparated by a spacer. As shown in Figure 1, the bottom coverslip (the

chamber floor) contains an array of embedded wells with highlycontrolled and well-defined nano- to microscale geometry. Aftersample injection, the deformable glass roof is pressed downward by aconvex lens, gently herding the biomolecules into the wells. The floor-ceiling distance can be controlled by posts attached to the chamberceiling to leave a small gap by which to introduce molecules/reagentsto the large, trapped biomolecules. The wells are deep enough tocontain freely diffusing and fluctuating biomolecules but shallowenough to keep them in the field of view for extended observationtimes. By squeezing out extra molecules, background fluorescence isreduced up to 20-fold, and samples can be tested at higherconcentrations compared to typical single-molecule techniques.Furthermore, the array of embedded wells allows data collectionwith significant statistics within a single experiment.Confinement microscopy enables a new approach to explore

protein phase separation under spatial confinement, relevant to thecellular interior. We visualized formation of condensates andcontrolled their growth by systematically varying confinementdimensions (Figures 2 and 3). We then explored the mesoscaleproperties of confined condensates by tracking the fluctuatingmotions of encapsulated fluorescent probe particles within thecondensates. We investigated two liquid condensate systems. First,phase-separated polyethylene glycol (PEG) and dextran condensateswere used to explore assembly, growth, and final size underconfinement. PEG-dextran condensates have been previously formedinside large vesicles, a conventional technique, as a simple model ofthe crowded cytoplasm.28−32 Second, we investigated biomolecularcondensates of ribonucleoprotein complexes formed of polyuridine(polyU) RNA and protein Dhh1 as a simple, few-component model

of natural NMOs. In both cases, we explored probe particle diffusionwith respect to condensate size.

■ RESULTS AND DISCUSSIONPEG-Dextran Mixtures Exhibit Anomalous Diffusion

and Size-Dependent Properties. In the first system,fluorescently labeled dextran was confined within wells, andcondensate formation and growth were monitored as PEG wasintroduced. The condensates formed under confinement(Supporting Information (SI), section S1) were demonstrably

Figure 1. Confinement microscopy: (a) Schematic of a Convex Lens-induced Confinement (CLiC) instrument used here; a convex lensdeforms the top coverslip of the flow cell. Posts fabricated into the topcoverslip maintain a space above the wells (450 nm deep and 3 μm indiameter) embedded in the flow cell floor, where condensates areformed. (b) Schematic of a microchannel fabricated into the bottomcoverslip that encircles the embedded pit array and facilitates reagentaddition and buffer exchange. (c) Schematic showing condensateformation on an open surface, i.e., no confinement.

Figure 2. Dextran-rich condensates condensed from a mixture ofPEG-dextran formed (a) on an open surface (i.e., withoutconfinement), and (b) inside a 3 μm well (i.e., under confinement).Using dual color imaging, fluorescently labeled beads (red) aredistinguished in dextran-rich condensates (green). Note: Red beadsappear yellow in the green channel when they overlap with green-labeled dextran-rich droplets. Dashed circle: well location. Inset: theparticle tracking of a red-labeled bead inside the condensate (greencircle). (c) Initiation of phase separation in an array of wellscontaining confined labeled dextran, as PEG is introduced from thebottom right corner. (d) Same array as shown in (c) 5 min later,condensates reach their final size.

Figure 3. Effect of confinement on the final size of the proteincondensates: condensates of (a) 2 μm, (b) 4 μm, and (c) 14 μmformed in wells with diameters of 3, 10, and 25 μm, respectively.Dashed circles correspond to the location of the wells. Condensatesare shown in gray, and bright dots inside condensates are probingparticles of 48 nm in diameter. (d) Schematic of ATP- and RNA-bound Dhh1 conformation, taken from a model proposed in ref 7, andhow these components may interact with each other inside a phase-separated condensate.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.8b13349J. Am. Chem. Soc. 2019, 141, 7751−7757

7752

smaller and more monodisperse than those formed on afeatureless coverslip where condensate fusion may occur,frequently resulting in large variation in final condensate size(Figure 1c and Figure 2a). When the reconstituent elementswere confined in sealed wells, the maximum size of the phase-separated condensate was controlled and limited by theamount of material available inside the well, i.e., proportionalto the well size. For example, Figure 2 compares condensatesformed on a coverslip and within a large array of wells, inwhich the final size of the condensates averages below 2 μminside wells of 3 μm in diameter.Previous work indicated that smaller condensates may have

different network structure and physical behaviors compared tolarger ones such that some mechanical theories would fail toexplain their observed behavior.33,34 To test whether theproperties of confined condensates change with their size, weperformed passive microrheology experiments. The motions ofprobe particles (20 nm fluorescent latex microspherespassivated with dextran), randomly distributed within con-densates, were tracked using our in-house tracking algorithm35

(Figure 2b and SI, section S2). Note that condensates are notdrifting within the well. To characterize diffusion, wecalculated the probe particle’s time-averaged mean-squareddisplacements (MSDs)6 and performed correlation analyses onthe videos of particle diffusion. Correlation methods such asfluorescence correlation spectroscopy36,37 and image correla-tion spectroscopy38 are complementary to single-particletracking but are more objective in that individual point spreadfunctions do not need to be fit. As such, correlation techniquesare able to quantify faster diffusion coefficients and assess datawith lower signal-to-noise ratios.39 The self-similarity of thefluorescent intensity signal at the pixel with the highestintensity fluctuations within the condensate was analyzed bysecond-order correlation (SI, section S3).For solutions of pure dextran prior to PEG introduction, a

homogeneous ”confined” diffusion of beads was obtained inthe microscale (SI, Figure S1), in which at short time scalesBrownian diffusion is observed while at long time scales aparticle’s motion is confined by the wall of the well. Thesebehaviors result in a plateau in MSD values at approximatelythe well radius squared. Interestingly, after PEG was

introduced and phase separation occurred, dextran-richcondensates possessed heterogeneous properties (Figure 4a).The diffusion coefficient of the probe particle (proportional tothe y-intercept of the logarithmic MSD plot) varied by up to 1order of magnitude from one condensate to another and evenfrom one location to another within the same condensate.Note that since the probe particles did not escape thecondensates over the course of the measurement, at long timescales, we could expect the particle MSD curves to plateau as aresult of confinement within the “walls” of the condensate(rather than walls of the wells). However, over ourexperimental time-course, we do not observe that MSD plotsof beads within phase separated condensates reach meansquared radii plateau or deviate beyond of the size of thecondensates. Nonetheless, large variation in MSD values overall time scales indicates that the particles undergo heteroge-neous diffusion (Figure 4a, shaded area). Correlation analysis(SI, section S3) further supports the heterogeneous diffusivemotion of beads inside dextran condensates (Figure 4c). Theaveraged correlation curves were better fit with a two-component correlation function, i.e., two diffusion components(Figure 4c and SI, section S3), suggesting both a slow and afast mode of diffusion. The faster component had ananomalous factor, α > 1, indicative of a superdiffusive, hoppingmechanism.Particle motion is also affected by the confining surface (i.e.,

condensate−solvent interface) through solvent-mediatedhydrodynamic interactions.40 As a particle moves inside acondensate, it propagates perturbations in the velocity field ofthe medium traveling back and forth between the interface andthe particle. In a crowded environment, hydrodynamicinteractions become even more complex and can createheterogeneity in particle mobility within a condensate.PEG-dextran condensate properties are size-dependent.

Despite heterogeneity, ensemble-averaged MSD curves showenhanced mobility of probe particles inside smaller (1.5−3μm) condensates compared to inside larger (>3−15 μm) ones(Figure 4a). As fluorescence density inside condensates wasthe same for various condensate sizes, the observed size-dependent mobility of tracer particles inside condensate is not

Figure 4. Diffusion analysis for probe particles inside PEG-dextran condensates (N = 135). (a) Time-averaged mean squared displacement versuslag time of probe particles embedded in dextran-rich condensates. Red and black represent condensates that are smaller and larger than 3 μm,respectively. For each data set, the circles or squares show the averaged MSD (i.e., time and ensemble averaged), and the shaded area shows thedistribution of MSD curves within that set. Dashed line: a slope of 1 in log−log scale, indicative of normal Brownian diffusion. (b) Typicaldisplacement probability distribution, measured for a probe particle embedded inside a dextran-rich condensate, plotted logarithmically againstdisplacement. The dashed curve represents a fitted standard Gaussian distribution and highlights the poor fit to Brownian solutions with theexperimental data. Inset: probe particle trajectory over 1 min. (c) Correlation analysis performed on probe particle diffusion inside a dextran-richcondensates. Color schemes are the same as in (a). Circles or squares show the averaged correlation curve, G, for each data set, while the shadedareas show the distribution of correlation curves. Lines represent the fit which give two decay times (relevant to a slow and a fast diffusion mode)and anomalous diffusion exponents.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.8b13349J. Am. Chem. Soc. 2019, 141, 7751−7757

7753

likely due to different dextran densities within droplets ofdifferent sizes.Ribonucleo Poly(U)-Dhh1 Condensates Also Exhibit

Anomalous Diffusion. The next condensates we studiedwere the simple model NMOs formed from the yeast RNAbinding protein Dhh1 and the RNA analogue poly(U). Here,we pre-mixed the Dhh1 with the poly(U) and visualizedformation of Dhh1-poly(U) condensates with controlled sizesusing confinement microscopy (Figure 3).As in the PEG-dextran system, both particle tracking and

correlation analysis (Figure 5) indicated heterogeneousanomalous diffusion of embedded probe particles (48 nmfluorescent latex microspheres passivated with bovine serumalbumin, BSA). However, unlike the PEG-dextran condensates,the Dhh1-poly(U) condensates did not exhibit significant size-dependent internal properties, as MSD plots for small (1.5−3μm) condensates overlapped with those of larger ones (>3−15μm). It is worth noting that we tried alternative clusteringgroups of the MSD plots but still did not observe size-dependent properties.In addition, single-component Brownian curves failed to fit

the data (SI, Figure S2), while a two-component correlationmodel was better fit. This suggests the existence of non-Brownian and hopping mechanism of diffusion in Dhh1condensates. Similar to the single-particle tracking results,correlation analysis fitting results were statistically similarvalues for small and large Dhh1-poly(U) condensates (SI,Table S1).Anomalous diffusion and hopping mechanisms have recently

been reported in vivo, in NMOs called stress granules.41 In thatcase, biphasic partitioning of biopolymers resulted in theirhaving suppressed diffusion in some local microenvironmentsbut larger mobility at other locations.41,42 Recently, an in-vitrostudy also suggests a set of interactions that are likely to berelevant in formation of complex, multilayered structures anddiverse materials properties within condensates.43

In order to better understand the anomalous (non-Brownian) diffusion within condensates, we measured thedisplacement probability distribution. Compared to a Gaussianmodel, there are too many large displacement events (the tailsin Figure 5b at Δx values of greater than 0.3 or less than −0.3

μm) and a stronger confinement of motion (the sharp peakclose to zero displacement, Δx = 0). These observationsprovide further evidence for the “hopping diffusion” mecha-nism where a particle mostly resides in a caged structure madeby neighboring molecules but occasionally escapes and movesa significantly larger distance than a normal Brownian diffusionwould predict.42,44 The non-Gaussian distribution of particledisplacement was also observed in dextran condensates, butonly at large displacements (tails at displacements greater than4 μm in Figure 4b). Therefore, both systems showedindications of a heterogeneous environment, but for theprotein condensates, hopping diffusion was directly observedfrom the analysis of the particle trajectory (inset in Figure 4b).We hypothesize that the observed hopping diffusionmechanism is due to the particular biomolecular organizationinside the condensed phase and disappears when thecondensed phase disassembles, regardless of the confinement.To explain the diffusion mechanisms within the protein

condensates, poly(U) molecules may be unevenly distributedwithin the condensate to cause locally crowded environmentsthat inhibit particle motion. It is also possible that specificinteractions between the condensate components contribute tothe condensate interior properties and effectively suppressparticle movement.45 Binding of ATP and RNA to Dhh1 hasbeen previously shown to be necessary for condensateformation, presumably due to inducing specific conformationsof the Dhh1 protein.7 A schematic of this bound conformationof poly(U) and Dhh1 is shown in Figure 3d.To explore the effect of Dhh1-poly(U) bound conformation

in condensate assembly, we introduced the regulatory proteinNot1 to the confined condensates. Not1 enhances the ATPaseactivity of Dhh1, unbinds Dhh1 from the nucleic acid bychanging the Dhh1 conformation, and ultimately results indisassembly of the condensate.7 Interestingly, we observed thatwhen we added Not1, embedded beads started to diffuse morequickly after 1 min, although the condensate is not fullydisassembled (SI, Figure S3). After 5 min, disassembly wascomplete, and the beads diffused freely throughout the well inthe solution of the poly(U) and Dhh1 (SI, Figure S3). Particletracking measurements showed that, in an ATP- and RNA-bound Dhh1 conformation, i.e., assembled condensate, probe

Figure 5. Diffusion analysis for probe particles inside protein-nucleic acid condensates (N = 140). (a) Time-averaged mean-squared displacementversus lag time of probe particles embedded in protein-rich condensates. Red and black represent condensates that are smaller and larger than 3μm, respectively. Dashed line: a slope of 1 in log−log scale, indicative of normal Brownian diffusion.(b) Typical displacement probabilitydistribution, measured for a probe particle embedded inside a protein-rich condensate plotted logarithmically against displacement. The dashed lineis a Gaussian fit that poorly fits the data, particularly where the experimental values deviate greatly from the Gaussian fit at Δx values of greater than0.3 or less than −0.3 μm. The non-Gaussianity observed can be attributed to the hopping diffusion. Inset: the trajectory of a probe particle over 1min. It suggests that the particle spends most of its time in a cage formed by its neighboring chain structures while during rare events moves asignificant distance due to a rearrangement of cage structure. (c) Correlation analysis performed on probe particles diffusing inside a protein-richcondensate. Color schemes are the same as in (a). Lines represent the fit to a two-component correlation function giving two decay times (relevantto a slow and a fast diffusion mode) and anomalous diffusion exponents.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.8b13349J. Am. Chem. Soc. 2019, 141, 7751−7757

7754

diffusion was highly suppressed and non-homogeneous acrossthe condensate, but as the Dhh1 conformation changed due tothe presence of Not1, probe particles diffused more freely anduniformly. MSD measurements of the embedded particlesbefore, during, and after disassembly showed that the size ofthe typical traversed region (calculated from the MSD)increased from 100 nm (t = 0) to 300 nm (t = 1 min) to1.5 μm (t = 5 min).Other results have indicated that permeability and diffusion

inside a protein condensate are affected by the length (MW) ofthe component RNA.46 These specific effects will be dictatedby the type of RNA- protein interactions. Short RNAs maymodulate protein−protein interactions to lead to a lowering ofthe viscosity, while larger molecules have the opposite effect,47

perhaps because larger molecules are involved in additionalphysical processes such as entanglement. In our protein study,the model RNA molecule, poly(U), is polydisperse in size (100kDa < MW < 1000 kDa), making it potentially morebiologically representative than a monodisperse RNA mole-cule. This polydispersity likely enhances the heterogeneity ofthe condensate interior structure, as indicated by our observedparticle diffusion. Future studies in which we form condensateswith monodispersed RNAs of a particular size would allow usto investigate the dependence of RNA length on diffusionwithin protein condensates.

■ SUMMARYThis study showcases a new application of confinementmicroscopy, which has enabled (1) control of condensate sizethrough spatial confinement (i.e., wells of controlled size), asshown in Figure 3a−c, and (2) the quantitative character-ization of microenvironments within condensates. Weobserved heterogeneous dynamics as well as anomalousdiffusion inside both dextran-rich and protein-rich conden-sates. In ribonucleo protein-rich condensates, we foundevidence of temporal heterogeneity such that embeddedparticles may diffuse within a microenvironment and occa-sionally “hop” to a neighboring microenvironment withdifferent physical properties. Furthermore, we observed a sizedependence of diffusion, and thus interior condensateproperties, for dextran condensates, although this was notobserved significantly for the Dhh1-poly(U) condensatestested.The presence of microenvironments inside NMO con-

densates has important biological implications. Due toheterogeneous distribution of nucleic acids within a con-densate, microenvironments likely facilitate biochemicalprocesses such as gene expression.48 Molecular crowding alsocontributes to the stochasticity of biochemical processes withinNMOs by decreasing diffusion of proteins and nucleic acidsand contributing to uneven distributions of biomolecules.48 Invitro studies at in vivo size scales should shed light oncondensate properties. Confinement microscopy is a promisingtechnique for a variety of NMO investigations, providinginsight into single molecule dynamics and reaction kinetics, bycreating “cell-like” spatial confinement and crowding. Thisplatform makes possible the means to directly and preciselystudy the effects of various parameters such as temperature andreagent (such as RNA, ATP, and salt) concentrations information of protein condensates (for instance, SI, Figure S4shows that protein concentration affects phase separation inthe absence of confinement) and in vitro study of enzymaticdisassembly of condensates at in vivo size scales.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/jacs.8b13349.

Technical details of the experimental methods andmaterials; details of the video tracking and particleposition analysis; analytical details of correlation analysisand fitting results; supplementary result for condensatedisassembly by addition of NOT1 protein; observationof the effect of protein concentration in Dhh1-Poly(U)system (PDF)

■ AUTHOR INFORMATIONCorresponding Authors*stephen.michnick@umontreal.ca*sabrina.leslie@mcgill.caORCIDMarjan Shayegan: 0000-0002-1892-4549NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSM.S., S.W.M, and S.R.L. acknowledge support from theCanadian Institutes of Health Research (CIHR) grantsMOP-GMX-152556 (S.W.M.), the Human Frontier ScienceProgram RGP0034/2017 (S.W.M.), the Natural Sciences andEngineering Research Council grant, Discovery and Accel-erator Grant Programs (S.R.L.), and CREATE Program on theCellular Dynamics of Macromolecular Complexes (M.S.). L.K.thanks the Arnold O. and Mabel M. Beckman Foundation forpartial support by the Beckman-Brown InterdisciplinaryPostdoctoral Fellowship. The authors thank Karsten Weisand Marie Hondele (ETH-Zurich) for gifts of Dhh1 and Not1proteins and also thank Christine Desroches for supportingprotein condensation experimental design.

■ REFERENCES(1) Banani, S. F. B.; Lee, H. O.; Hyman, A. A.; Rosen, M. K.Biomolecular condensates: organizers of cellular biochemistry. Nat.Rev. Mol. Cell Biol. 2017, 18, 285−298.(2) Shin, Y.; Brangwynne, C. P. Liquid phase condensation in cellphysiology and disease. Science 2017, 357, eaaf4382.(3) Bergeron-Sandoval, L.-P.; Safaee, N.; Michnick, S. W.Mechanisms and consequences of macromolecular phase separation.Cell 2016, 165, 1067−1079.(4) Gomes, E.; Shorter, J. The molecular language of membranelessorganelles. J. Biol. Chem. 2019, 294, 7115−7127.(5) Brangwynne, C. P. Phase transitions and size scaling ofmembrane-less organelles. J. Cell Biol. 2013, 203, 875−881.(6) Feric, M.; Vaidya, N.; Harmon, T. S.; Mitrea, D. M.; Zhu, L.;Richardson, T. M.; Kriwacki, R. W.; Pappu, R. V.; Brangwynne, C. P.Coexisting liquid phases underlie nucleolar subcompartments. Cell2016, 165, 1686−1697.(7) Mugler, C. F.; Hondele, M.; Heinrich, S.; Sachdev, R.; Vallotton,P.; Koek, A. Y.; Chan, L. Y.; Weis, K. ATPase activity of the DEAD-box protein Dhh1 controls processing body formation. eLife 2016, 5,No. e18746.(8) Stroberg, W.; Schnell, S. On the origin of non-membrane-boundorganelles, and their physiological function. J. Theor. Biol. 2017, 434,42−49.(9) Eulalio, A.; Behm-Ansmant, I.; Izaurralde, E. P bodies: at thecrossroads of post-transcriptional pathways. Nat. Rev. Mol. Cell Biol.2007, 8, 9.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.8b13349J. Am. Chem. Soc. 2019, 141, 7751−7757

7755

(10) Patel, A.; Lee, H. O.; Jawerth, L.; Maharana, S.; Jahnel, M.;Hein, M. Y.; Stoynov, S.; Mahamid, J.; Saha, S.; Franzmann, T. M.;Pozniakovski, A.; Poser, I.; Maghelli, N.; Royer, L. A.; Weigert, M.;Myers, E. W.; Grill, S.; Drechsel, D.; Hyman, A. A.; Alberti, S. Aliquid-to-solid phase transition of the ALS protein FUS accelerated bydisease mutation. Cell 2015, 162, 1066−1077.(11) Buchan, J. R. mRNP granules: assembly, function, andconnections with disease. RNA Biol. 2014, 11, 1019−1030.(12) Anderson, P.; Kedersha, N.; Ivanov, P. Stress granules, P-bodiesand cancer. Biochim. Biophys. Acta, Gene Regul. Mech. 2015, 1849,861−870.(13) Toretsky, J. A.; Wright, P. E. Assemblages: functional unitsformed by cellular phase separation. J. Cell Biol. 2014, 206, 579−588.(14) Weber, S. C.; Brangwynne, C. P. Inverse size scaling of thenucleolus by a concentration-dependent phase transition. Curr. Biol.2015, 25, 641−646.(15) Yang, X.; Xu, T. Molecular mechanism of size control indevelopment and human diseases. Cell Res. 2011, 21, 715.(16) Schutz, S.; Noldeke, E. R.; Sprangers, R. A synergistic networkof interactions promotes the formation of in vitro processing bodiesand protects mRNA against decapping. Nucleic Acids Res. 2017, 45,6911−6922.(17) Mitrea, D. M.; Kriwacki, R. W. Phase separation in biology;functional organization of a higher order. Cell Commun. Signaling2016, 14, 1−20.(18) Aumiller, W. M., Jr.; Keating, C. D. Experimental models fordynamic compartmentalization of biomolecules in liquid organelles:Reversible formation and partitioning in aqueous biphasic systems.Adv. Colloid Interface Sci. 2017, 239, 75−87.(19) Uversky, V. N. Intrinsically disordered proteins in overcrowdedmilieu: membrane-less organelles, phase separation, and intrinsicdisorder. Curr. Opin. Struct. Biol. 2017, 44, 18−30.(20) Zhu, L.; Brangwynne, C. P. Nuclear bodies: the emergingbiophysics of nucleoplasmic phases. Curr. Opin. Cell Biol. 2015, 34,23−30.(21) Luo, Y.; Na, Z.; Slavoff, S. A. P-Bodies: Composition,Properties, and Functions. Biochemistry 2018, 57, 2424−2431.(22) Mason, T. O.; Michaels, T. C. T.; Levin, A.; Dobson, C. M.;Gazit, E.; Knowles, T. P. J.; Buell, A. K. Thermodynamics ofPolypeptide Supramolecular Assembly in the Short-Chain Limit. J.Am. Chem. Soc. 2017, 139, 16134−16142.(23) Lin, Y.-H.; Forman-Kay, J. D.; Chan, H. S. Theories forSequence-Dependent Phase Behaviors of Biomolecular Condensates.Biochemistry 2018, 57, 2499−2508.(24) Berard, D. J.; Michaud, F.; Mahshid, S.; Ahamed, M. J.; McFaul,C. M.; Leith, J. S.; Berube, P.; Sladek, R.; Reisner, W.; Leslie, S. R.Convex lens-induced nanoscale templating. Proc. Natl. Acad. Sci. U. S.A. 2014, 111, 13295−13300.(25) Ahamed, M. J.; Mahshid, S.; Berard, D. J.; Michaud, F.; Sladek,R.; Reisner, W. W.; Leslie, S. R. Continuous Confinement Fluidics:Getting Lots of Molecules into Small Spaces with High Fidelity.Macromolecules 2016, 49, 2853−2859.(26) Henkin, G.; Berard, D.; Stabile, F.; Shayegan, M.; Leith, J. S.;Leslie, S. R. Manipulating and visualizing molecular interactions incustomized nanoscale spaces. Anal. Chem. 2016, 88, 11100−11107.(27) Berard, D. J.; Shayegan, M.; Michaud, F.; Henkin, G.; Scott, S.;Leslie, S. Formatting and ligating biopolymers using adjustablenanoconfinement. Appl. Phys. Lett. 2016, 109, 033702.(28) Helfrich, M. R.; Mangeney-Slavin, L. K.; Long, M. S.; Djoko, K.Y.; Keating, C. D. Aqueous phase separation in giant vesicles. J. Am.Chem. Soc. 2002, 124, 13374−13375.(29) Li, Y.; Lipowsky, R.; Dimova, R. Membrane nanotubes inducedby aqueous phase separation and stabilized by spontaneous curvature.Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 4731−4736.(30) Li, Y.; Lipowsky, R.; Dimova, R. Transition from Complete toPartial Wetting within Membrane Compartments. J. Am. Chem. Soc.2008, 130, 12252−12253.

(31) Helfrich, M. R.; Mangeney-Slavin, L. K.; Long, M. S.; Djoko, K.Y.; Keating, C. D. Aqueous Phase Separation in Giant Vesicles. J. Am.Chem. Soc. 2002, 124, 13374−13375.(32) Ban, T.; Fukuyama, T.; Makino, S.; Nawa, E.; Nagatsu, Y. Self-Propelled Vesicles Induced by the Mixing of Two Polymeric AqueousSolutions through a Vesicle Membrane Far from Equilibrium.Langmuir 2016, 32, 2574−2581.(33) Style, R. W.; Hyland, C.; Boltyanskiy, R.; Wettlaufer, J. S.;Dufresne, E. R. Surface tension and contact with soft elastic solids.Nat. Commun. 2013, 4, 1−6.(34) Claessens, M. M. A. E.; Tharmann, R.; Kroy, K.; Bausch, A. R.Microstructure and viscoelasticity of confined semiflexible polymernetworks. Nat. Phys. 2006, 2, 186−189.(35) Scott, S.; Xu, Z.; Kouzine, F.; Berard, D. J.; Shaheen, C.; Gravel,B.; Saunders, L.; Hofkirchner, A.; Leroux, C.; Laurin, J.; Levens, D.;Benham, C. J.; Leslie, S. R. Visualizing structure-mediated interactionsin supercoiled DNA molecules. Nucleic Acids Res. 2018, 46, 4622−4631.(36) Elson, E. L.; Magde, D. Fluorescence correlation spectroscopy.I. Conceptual basis and theory. Biopolymers 1974, 13, 1−27.(37) Schwille, P.; Ries, J. Biophotonics: Spectroscopy, Imaging, Sensing,and Manipulation; Springer: Dordrecht, 2011; pp 63−85.(38) Cooper, J. T.; Harris, J. M. Imaging Fluorescence-CorrelationSpectroscopy for Measuring Fast Surface Diffusion at Liquid/SolidInterfaces. Anal. Chem. 2014, 86, 7618−7626.(39) Kisley, L.; Brunetti, R.; Tauzin, L. J.; Shuang, B.; Yi, X.;Kirkeminde, A. W.; Higgins, D. A.; Weiss, S.; Landes, C. F.Characterization of Porous Materials by Fluorescence CorrelationSpectroscopy Super-resolution Optical Fluctuation Imaging. ACSNano 2015, 9, 9158−9166.(40) Cervantes-Martínez, A. E.; Ramírez-Saito, A.; Armenta-Calderon, R.; Ojeda-Lopez, M. A.; Arauz-Lara, J. L. Colloidaldiffusion inside a spherical cell. Phys. Rev. E 2011, 83, 030402.(41) Niewidok, B.; Igaev, M.; Pereira da Graca, A.; Strassner, A.;Lenzen, C.; Richter, C. P.; Piehler, J.; Kurre, R.; Brandt, R. Single-molecule imaging reveals dynamic biphasic partition of RNA-bindingproteins in stress granules. J. Cell Biol. 2018, 217, 1303−1318.(42) Skaug, M. J.; Mabry, J.; Schwartz, D. K. Intermittent MolecularHopping at the Solid-Liquid Interface. Phys. Rev. Lett. 2013, 110,256101.(43) Boeynaems, S.; Holehouse, A. S.; Weinhardt, V.; Kovacs, D.;Van Lindt, J.; Larabell, C.; Van Den Bosch, L.; Das, R.; Tompa, P. S.;Pappu, R. V.; Gitler, A. D. Spontaneous driving forces give rise toprotein−RNA condensates with coexisting phases and complexmaterial properties. Proc. Natl. Acad. Sci. U. S. A. 2019, 116, 7889−7898.(44) Valentine, M. T.; Kaplan, P. D.; Thota, D.; Crocker, J. C.;Gisler, T.; Prud’homme, R. K.; Beck, M.; Weitz, D. A. Investigatingthe microenvironments of inhomogeneous soft materials withmultiple particle tracking. Phys. Rev. E: Stat. Phys., Plasmas, Fluids,Relat. Interdiscip. Top. 2001, 64, 061506.(45) Brady, J. P.; Farber, P. J.; Sekhar, A.; Lin, Y.-H.; Huang, R.; Bah,A.; Nott, T. J.; Chan, H. S.; Baldwin, A. J.; Forman-Kay, J. D.; Kay, L.E. Structural and hydrodynamic properties of an intrinsicallydisordered region of a germ cell-specific protein on phase separation.Proc. Natl. Acad. Sci. U. S. A. 2017, 114, E8194−E8203.(46) Wei, M.-T.; Elbaum-Garfinkle, S.; Holehouse, A. S.; Chen, C.C.-H.; Feric, M.; Arnold, C. B.; Priestley, R. D.; Pappu, R. V.;Brangwynne, C. P. Phase behaviour of disordered proteins underlyinglow density and high permeability of liquid organelles. Nat. Chem.2017, 9, 1118.(47) Elbaum-Garfinkle, S.; Kim, Y.; Szczepaniak, K.; Chen, C. C.-H.;Eckmann, C. R.; Myong, S.; Brangwynne, C. P. The disordered Pgranule protein LAF-1 drives phase separation into droplets withtunable viscosity and dynamics. Proc. Natl. Acad. Sci. U. S. A. 2015,112, 7189−7194.(48) Hansen, M. M. K.; Meijer, L. H. H.; Spruijt, E.; Maas, R. J. M.;Ventosa Rosquelles, M.; Groen, J.; Heus, H. A.; Huck, W. T. S.Macromolecular crowding creates heterogeneous environments of

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.8b13349J. Am. Chem. Soc. 2019, 141, 7751−7757

7756

gene expression in picolitre droplets. Nat. Nanotechnol. 2016, 11,191−197.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.8b13349J. Am. Chem. Soc. 2019, 141, 7751−7757

7757