Time-Resolved Observations of Liquid–Liquid Phase Separation at

Apr 15, 2019 - ... deposited onto a SiN window, covered, and closed with a second SiN window, .... It is noticeable that the initial growth kinetics o...
0 downloads 0 Views 3MB Size
Subscriber access provided by UNIV OF LOUISIANA

Article

Time-Resolved Observations of Liquid-Liquid Phase Separation at the Nanoscale using in situ Liquid Transmission Electron Microscopy Hortense Le Ferrand, Martial Duchamp, Bartosz Gabryelczyk, Hao Cai, and Ali Miserez J. Am. Chem. Soc., Just Accepted Manuscript • Publication Date (Web): 15 Apr 2019 Downloaded from http://pubs.acs.org on April 15, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Time-Resolved Observations of Liquid-Liquid Phase Separation at the Nanoscale using in situ Liquid Transmission Electron Microscopy Hortense Le Ferrand1, Martial Duchamp2, Bartosz Gabryelczyk1, Hao Cai1, and Ali Miserez1,3* 1

Biological and Biomimetic Material Laboratory and Center for Biomimetic Sensor Science,

School of Materials Science and Engineering, Nanyang Technological University (NTU), 50 Nanyang Avenue, Singapore 637553. 2

Laboratory for in situ & operando Electron Nanoscopy, School of Materials Science and

Engineering, Nanyang Technological University (NTU), 50 Nanyang Avenue, Singapore 637371. 3

School of Biological Sciences, Nanyang Technological University (NTU), 60 Nanyang Drive,

Singapore 637551. * To whom correspondence should be addressed: [email protected] Abstract Liquid-liquid phase separation (LLPS) of proteins into concentrated microdroplets (also called coacervation) is a phenomenon that is increasingly recognized to occur in many biological processes, both inside and outside the cell. While it has been established that LLPS can be described as a spinodal decomposition leading to demixing of an initially homogenous protein solution, little is known about the assembly pathways by which soluble proteins aggregate into dense microdroplets. Using a recently developed technique enabling the observation of matter suspended in liquid by transmission electron microscopy (TEM), we observed how a model intrinsically disordered protein (IDP) phase-separates in liquid environment. Our observations reveal for the first time dynamic mechanisms by which soluble proteins self-organize into condensed microdroplets with nano-scale and milli-second space and time resolution, respectively. With this method, the nucleation and initial growth steps of LLPS could be captured, opening the door for a deeper understanding of biomacromolecular complexes exhibiting LLPS ability.

Keywords Liquid-liquid phase separation, in situ liquid transmission electron microscopy, intrinsically disordered proteins, spinodal decomposition

ACS Paragon Plus Environment

1

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 24

Introduction Liquid-liquid phase separation (LLPS), or coacervation, of intrinsically disordered proteins (IDPs) is a ubiquitous phenomenon involved in many biological processes, both intra- and extracellularly, such as ion transport1, chemical reactions2,3, biomineralisation4, or disease development5,6. LLPS results in a metastable or stable equilibrium consisting of concentrated liquid microdroplets –also called coacervate microdroplets– separated from a dilute phase. Within the cells, these microdroplets form membraneless compartments that can exchange molecules with the adjacent milieu in a dynamic fashion7,8. If the strength of inter-molecular interactions within the microdroplets increases, the later can evolve into a protein gel, aggregates, or even crystals3,9. LLPS has also recently been proposed to be implicated in the development and processing of biomolecular composite tissues and biological adhesives10,11. Moreover, engineered IDPs with a stimuli-responsive LLPS properties have been used as carriers for therapeutics delivery12–14. Despite the consensus that dense coacervate microdroplets assemble from concentrated oligomers of collapsed proteins3,15–17, there remains questions as whether LLPS follows the classical nucleation theory, spinodal decomposition, non-classical theory, or another path18,19. Understanding

these

mechanisms

would

not

only

provide

insights

into

cellular

compartementalization20 or cells’ response to stress21, but may also yield bioinspired lessons to engineer novel responsive materials for biomedicine20,22. Towards this goal, there is a need to better understand the steps involved during LLPS at the nano-scale and in a time-resolved fashion. At the micro-scale, dynamic processes of LLPS have traditionally been visualized using techniques such as turbidity23, confocal microscopy6 and fluorescence recovery after photobleaching (FRAP)24,25. In one recent example, LLPS was induced using a relatively weak laser while concomitantly observing the dense microdroplets by optical microscopy26. LLPS has also been probed with spectroscopic methods such as nuclear magnetic resonance (NMR)27,28, light scattering29, circular dichroism (CD)23, and X-ray and neutron diffraction23,30. To image coacervate microdroplets with sub-micron resolution, cryo-transmission electron microscopy (cryoTEM)16 has been used. However, given the stringent conditions imposed by sample preparation, cryoTEM only provides information of the final structures arising from LLPS. Therefore, it remains challenging to dynamically image phase transitions of IDPs involved in LLPS with nanoscopic spatial resolution. So far, TEM has scarcely been used to image bare organic molecules such as proteins due to their degradation under the high intensity of the electron beam, damaging the sample and yielding imaging artefacts31. Contrasting agents that increase the density of biomolecular samples are also

ACS Paragon Plus Environment

2

Page 3 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

commonly used for better imaging, but these additives may drive the proteins into a non-native state, which is not ideal for dynamic observations. In the past decade, new TEM techniques have been developed. These include liquid cells to observe matter in a liquid environment at ambient pressure and direct electron detectors to record TEM images with a higher signal-to-noise ratio and at a fast frame rate32. These developments have allowed the observation of crystal nucleation and growth in situ and directly in the liquid state32–35 and to follow gold nanoparticle growth36, polymerization-induced micelle formation37, and lysozyme aggregation38, for example. Since IDPs phase separate into dense liquid condensates, we hypothesized that they could be observed without the need for a contrasting agent, thus allowing their direct in situ observation in liquid TEM. It would then be possible to study the dynamic assembly of IDPs during LLPS. In this study, we therefore employed in situ liquid TEM to observe LLPS of a model selfcoacervating protein, namely Histidine-rich Beak Protein 2 (HBP-2). HBP-2 is an IDP found in the squid beak10 that phase separates into dense liquid microdroplets ca. 1 µm in diameter10,23 upon charge screening driven by hydrophobic interactions39. Following conformational change of HBP2 and the formation of microdroplets with salt concentration and time, we first determine the conditions at which the LLPS occurs with a kinetic slow enough to be probed by different techniques. Prior to the apparition of the first microdroplet, we observed indirectly by light scattering the formation of different populations of protein nanoclusters that evolved dynamically. These two characteristics, namely high dynamics and nanometer size, in addition to the high concentration of protein in the final microdroplets, called for direct in situ liquid TEM imaging. In these TEM experiments, we imaged in a time-resolved fashion and without contrasting agent the assembly and disassembly of HBP-2 proteins into nanoclusters of various sizes and contrast depending on the salt content, i.e. the level of charge screening. In situ liquid TEM appears to be an ideal method to observe the early stages of LLPS, in particular the formation of nanoclusters of collapsed proteins whose direct observations have so far been elusive and not well understood40. Applied to other biomolecular systems exhibiting LLPS, in situ liquid TEM of dynamically evolving proteins is expected to bring unprecedented insights into phase separation, aggregation, or crystallization events. Experimental section Protein and buffer preparation. HBP-2 protein was obtained by recombinant expression, purified by reverse-phase HPLC, and freeze-dried, following a procedure described elsewhere23. Prior to the experiments, protein powder was dissolved in acetic acid (grade AR, Schedelco) at 10

ACS Paragon Plus Environment

3

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 24

mM (pH 3-3.2) at a concentration of 1 mg/mL. The concentration in protein was measured using a spectrophotometer (Nanodrop 2000c, Thermo Fisher Scientific). Buffer solutions were prepared using TRIS buffer (tris(hydroxymethyl)aminomethane, MB grade, Affymetrix, USA) at 10 mM (pH 8) and the pH was adjusted using HCl to 5.5. NaCl (Emsure, Merck, Denmark) at different concentrations (noted [NaCl]) was added to these buffers. The phase separation was triggered by mixing the protein solution and the buffer at a ratio 1:1. Circular dichroism. Circular dichroism spectra were recorded in transmission using an Aviv 420 spectrometer (Aviv Biomedical Inc., New Jersey, USA) using a precision quartz cuvette (HellmaAnalytics, Germany) with a path length k = 0.2 mm. Each scan covered the wavelengths from 180 to 260 nm, with 1 nm steps and averaging time of 1s. The spectra were then superimposed using Matlab. Optical microscopy and microdroplet number. Increase in microdroplet number was measured under optical microscope (Zeiss Axio Scope, Germany) in a similar precision quartz closed cuvette as for the CD, to avoid drying and evaporation. Images were recorded every minute using a camera (AxioCam MRc-5) and the software (AxioVision). The areal density of microdroplet was obtained using Image J software by calculating the total area covered by the microdroplets and divided by the total area of the image. The measurements were repeated twice. Dynamic light scattering. The Zetasizer Nano SZ-100 (Horiba) was used to measure the dimensions of the protein nanoclusters during the phase separation. The diameters were recorded in terms of frequency. A total volume of 500 µL was placed into a disposable plastic cuvettes with 2 opening and measured under a light source at 532 nm at a detection angle of 173°. Each point was recorded over 30 s with 2 minutes between each point. The diameters were calculated using the standard mode. In situ liquid transmission electron microscopy. A static liquid cell with spacing 150 nm (Small E-chip EPB-55DS and Large E-chip EPT-55W, Protochips, USA) was prepared following the recommendations of the manufacturer. Before closing the chip and inserting it into the liquid cell holder (Poseidon, Protochips, USA), 1 µL drop of the mixture protein-buffer was deposited. After closing, the holder was directly inserted into the TEM (JEM-ARM200F) to pump the vacuum down. Observations were done in direct TEM mode under 200 kV and in diffraction mode for the conditions [NaCl] = 0.15 M and [NaCl] = 0.10 M. For [NaCl] = 0.125 M, we used TEM

ACS Paragon Plus Environment

4

Page 5 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

(GrandARM300) operating at 300 kV and confirmed our observations at other salt concentrations in these conditions. Videos were recorded using the Gatan one-view CMOS Camera (Gatan, Japan). Images were then analyzed using Image J with the Bioformat plugin and MATLAB. The cumulative electron doses recorded for the TEM images in Figure 2 are 45, 88, 126 and 1432 𝑒 ― / 𝐴2 for the conditions 0.05, 0.10, 0.125 and 0.15 M, respectively. The calculation of the doses was done using the recorded images and the following formula:

𝑒― 𝐴2

=

𝐶𝑜𝑢𝑛𝑡𝑠 ∙ 𝐸𝑥𝑝𝑜𝑠𝑢𝑟𝑒 𝑡𝑖𝑚𝑒 𝐴𝑟𝑒𝑎 ∙ 𝛾

, with 𝛾 the

conversion factor from counts into primary electron as calibrated by the manufacturer (Gatan). The images in Figure 3D,E were recorded after a cumulative exposure of ~63 and 22 𝑒 ― /𝐴2, respectively. To avoid large electron doses, the movies were recorded on areas that were not irradiated previously. The cumulative electron doses for the images Figure 3A range from 18 to 900 𝑒 ― /𝐴2, whereas the doses for the series Figure 3B range from 8 to 90 𝑒 ― /𝐴2. Overall, the doses used were lower than the typical does for which structural damage starts to occur in biological materials1. Results First, we established the phase diagram of the LLPS of HBP-2 by following with time the conformation change of HBP-2 using CD spectroscopy and the apparition of microdroplets by optical microscopy (Figure 1). HBP-2 contains a relatively high molar concentration of histidine residues (see Supplementary Figure 1 for the primary sequence), resulting in a net positive charge and electrostatic repulsion in the initial solution state in 10 mM acetic acid (pH ~ 3) (Figure 1A). In the vicinity of the isoelectric point (pH 6.5) and in the presence of a small amount of salt, the charges at the surface of the protein are screened, resulting in a decrease of Zeta potential (see Supplementary Figure 2A) and subsequently allowing hydrophobic as well as aromatic (mostly tyrosine-tyrosine) interactions to occur23. In the conditions explored in this study, we found that at pH 5.5, the addition of NaCl at concentrations above 0.1 M promoted these inter-molecular interactions, presumably through a decrease of the Debye length below 1 nm (details in Supplementary Figure 2B). In conditions of high charge screening, HBP-2 phase separated into microdroplets containing an extremely high protein concentration of ca. 885 mg/mL (see Supplementary Information and supplementary Figure 3 for details about these concentration measurements). In all experiments performed in the remainder of the paper, the pH of the proteinbuffer mixture was maintained constant at 5.5 and the initial protein concentration C0 at 0.5 mg/mL, while the charge screening effect was tuned via variation of the salt concentration [NaCl].

ACS Paragon Plus Environment

5

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 24

Figure 1. Time-dependence LLPS of HBP-2 at pH 5.5 in the presence of salt. With the exception of the plot in (G), all data were recorded using an initial protein concentration C0 = 0.5 mg/mL placed in a buffer with pH 5.5 and for different [NaCl] concentrations. (A) Schematic of the LLPS of HBP-2, initially soluble in acidic media and phase-separating into microdroplets ~1 µm in diameter after charge screening. (B) CD spectra representing the molar ellipticity vs. wavelength 𝜆 of the protein solution as a function of time, just after the addition of [NaCl] = 0.25 M and the rise in pH to 5.5. The minimum at 𝜆 = 204 nm is attributed to random coil conformation, whereas the minimum at 𝜆 = 215 nm indicates the presence of -sheet structures. (C) Plot of ellipticity [𝛩]215 (at 𝜆 = 215 𝑛𝑚) as a function of time and at various values of [NaCl]. The lines ―𝑡 𝜏

correspond to fits using 𝐴 ∙ 𝑒 +𝐵, where A and B are fitting parameters and 𝜏 a characteristic time. (D) Optical micrographs of the protein solution as a function of time (t = 0 when the buffer solution containing [NaCl] = 0.25 M was added). The solution was covered by a glass slide to avoid evaporation and the images were taken at identical locations in the center of the deposited

ACS Paragon Plus Environment

6

Page 7 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

liquid. The black arrow shows a microdroplet, whereas the white arrow indicates a similar microdroplet out of focus. (E) Optical areal density of protein microdroplets ddroplet as a function of time at different values of [NaCl], using optical micrographs as in (D). Insert is a close-up view. The lines correspond to linear fits and are offset with the apparition of the first microdroplets, from which a linear growth rate J is extracted. (F) Plots of time 𝜏 (extracted from (C)) and growth rate J (from (E)) as a function of [NaCl]. The lines are guides to the eye. (G) Schematic of the domains where LLPS of HBP-2 occurs, at pH 5.5 constructed using optical microscopy observations. The figure indicates the presence of microdroplets (black dots) for different initial protein concentrations C0 (x-axis) and salt concentrations [NaCl] (y-axis), 2 minutes after the buffer solution containing the salt was pipetted into the acidic protein solution. The darker grey area thus corresponds to the region where the phase separation occurs within 2 minutes. The light grey area indicates the absence of microdroplets at that time, but their appearance after a longer time. The white area represents conditions for which no phase separation was visible optically in the conditions of the experiments. The white dots point the compositions that were tested. The time-dependent CD spectrum measured at [NaCl] = 0.25 M shows the change of protein conformation from a random coil with a characteristic minimum around 200 nm, to a structure rich in 𝛽-sheet structures with a characteristic broad minimum around 215 nm (Figure 1B). These 𝛽 – sheet structures might be initially nucleated by transient hydrogen bonds between tyrosine and histidine side-chains followed by 𝜋 ― 𝜋 interactions between tyrosine side groups39. These effects could be responsible for the shift of the minimum from 200 nm to 204 nm (see Supplementary Figure 4). Repeating the measurement at various salt concentrations of NaCl, the kinetics of conformational change from random coil to -sheets could be obtained by plotting the ellipticity [𝛩]215 measured at 215 nm as a function of time (see Supplementary Figure 4 and Figure 1C). ―𝑡

The data were fitted with 𝐴 ∙ 𝑒 𝜏 +𝐵 to extract a characteristic time constant 𝜏 that was found to decrease with increasing salt concentration (Figure 1F and supplementary table 1 for fitting parameters). Below [NaCl] = 0.15 M, 𝜏 drastically increased as LLPS was inhibited, likely due to insufficient charge screening. These observations of a time-dependent folding of HBP-2 correlate well with optical observations where the density of microdroplets was found to depend on the salt concentration, and hence ionic strength, and time. The microdroplets that were suspended in the solution appeared as black or white dots by optical microscopy (Figure 1D) depending on their position vis-à-vis the focal plane. At salt concentrations of 0.25 M and 0.15 M, the formation of microdroplets occurred

ACS Paragon Plus Environment

7

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 24

after a latency time of ca. 6 and 20 minutes, respectively. Furthermore, the density of microdroplets was found to increase linearly with time, recalling the linear nucleation rate in phase transformation systems36 (Figure 1E and supplementary table 2 for fitting parameters). We therefore extracted from these fits a growth rate parameter J that increased with the salt concentration (Figure 1F). We drew a map of the domains where the LLPS of HBP-2 at pH 5.5 occurs based on the observation of microdroplets 2 minutes after pipetting the acidic protein solution into the buffer containing the salt (Figure 1G). The diagram exhibits three regions: (i) a white region where no phase separation was observed over a 1 hour period; (ii) a lighter grey region where microdroplets formed after a latency time superior to 2 minutes, with a linear nucleation rate as described in Figure 1E; and (iii) a dark grey region where the formation of microdroplets occurred at fast linear nucleation rate, leading to a high density of microdroplets. Such a phase separation can occur either by spinodal decomposition, which does not require overcoming an activation energy barrier, or by nucleation and growth whereby nuclei with a large difference in protein content must initially be formed for phase separation to proceed41-43. But even for spinodal decomposition, nanoclusters with slightly enriched protein concentration should form in the initial stage of phase separation, with the protein content in the clusters rapidly increasing as they coarsen44. Because these clusters do not require an activation energy, their rate of formation is likely very high, which may have previously precluded their direct observation with conventional microscopy techniques due to their sub-micrometer length scale. Whether the phase transition occurs by spinodal decomposition or classical nucleation and growth, we hypothesized that initial clusters could be observed by liquid EM. To investigate further this hypothesis, we then used conditions in the light grey region to observe what happened during the lag time prior to the apparition of microdroplets.

ACS Paragon Plus Environment

8

Page 9 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Figure 2. Characteristic sizes of protein nanoclusters as a function of salt concentration [NaCl]. (A) Size of protein nanoclusters as a function of time t and salt concentration [NaCl] measured by DLS. t = 0 corresponds to the instant when the initial acidic protein solution was pipetted in the buffer solution. (B) Conditions where LLPS occurs for different initial protein concentration C0 and salt concentration [NaCl], similar to Figure 1G. Only the conditions used for liquid TEM experiments (C0 = 0.5 mg/mL) are presented. The two columns indicate the presence/absence of microdroplets after 2 minutes and 1 hour, respectively. (C) Schematics of the set-up used for in situ liquid TEM imaging conditions. The acidic protein solution was mixed with the relevant buffer and deposited onto the bottom window before the chip was sealed and imaged ca. t = 10 to 20 min later. (D) TEM micrographs taken ca. 1 hour after pipetting the protein into the buffer at different salt concentrations [NaCl]. The micrographs are representative of the observations recorded after that period of time.

ACS Paragon Plus Environment

9

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 24

To observe the early stage of LLPS, we selected conditions in the boundary region between the white and light grey areas featured in Figure 1G and followed the evolution of the size of protein nanoclusters via Dynamic Light Scattering (DLS), then via in situ liquid TEM (Figure 2). Figure 2A are plots representing the average diameter of protein nanoclusters measured by DLS as a function of time for decreasing salt concentrations, hence ionic strength (see Supplementary Figure 5 for the full distributions). In all conditions, HBP-2 exhibited a highly dynamic behavior. In strong coacervating conditions, i.e. with an initial composition located in the dark grey area of Figure 1G ([NaCl] = 0.5 M), the formation of ca. 1 µm microdroplets was instantaneous and was followed by growth and coalescence (Supplementary Figure 5). At conditions in the light grey area of Figure 1G, ([NaCl] = 0.25 M and 0.15 M), 500 nm nanoclusters were first measured that decreased in size before quickly growing into the 1 µm dense microdroplets (Figure 2A). This initial decrease may be attributed to the hydrophobic collapse of protein oligomers or nanoclusters, or to the disassembly and reassembly of proteins into nanoclusters of smaller size. In the vicinity of the light grey domain (0.10 ≤ [NaCl] ≤ 0.125 M), several populations of nanoclusters coexisted: 100-500 nm; 10-20 nm; and 2-3 nm, with high fluctuations for 0.125 M. Finally, in the conditions were no phase separation occurred ([NaCl] = 0.05 M), few nanoclusters formed with similar dimensions as for protein solubilized in acidic conditions (Supplementary Figure 5). We therefore postulated that at C0 = 0.5 mg/mL and at salt concentrations 0.10 ≤ [NaCl] ≤ 0.125 M (Figure 2B), it should be possible to observe the transition from the soluble state to the microdroplets, namely the early stage of the LLPS, within 1 h after pipetting the protein solution into the buffer. Selection of the appropriate imaging conditions is critical due to the requirements of the in situ liquid TEM set-up, where imaging starts 10 to 20 minutes after the protein has been placed in contact with the buffer solution (Figure 2C). Thus, one drop of proteins in buffer solution was deposited onto a SiN window, covered and closed with a second SiN window, leaving a gap of 150 nm. Electron micrographs taken ca. 1 hour after the protein had been placed in the buffer confirmed the DLS measurements (Figure 2D). For [NaCl] = 0.15 M, the system was in quasi steady state and evolved very slowly (see Supplementary Movie 1). We attribute the black area on the top left to a final and very dense microdroplet, with a porous interface containing a lower protein concentration, thus exhibiting a lower contrast. The spherical entities in this interface where found to slowly shrink over time (see Supplementary Figure 6). At [NaCl] = 0.125 M, dark nanoclusters initially appeared and grew, whereas at [NaCl] = 0.10 M similar dark nanoclusters were also present but then dissolved following their initial growth. After 1 hour, highly contrasted dark nanoclusters remained present at 0.125 M, whereas only a greyish grainy contrast was left at

ACS Paragon Plus Environment

10

Page 11 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

0.10 M. These two conditions are described in further detail in Figure 3. Finally, at [NaCl] = 0.05 M, very little contrast was observed indicating that most proteins remained soluble. Nevertheless, after 1 hour some darker spherical objects were present (Figure 2D, right-most micrograph).

Figure 3. In situ liquid TEM observations of protein nanocluster formation and growth. (A) Time-lapse TEM micrographs of protein nanocluster growth in liquid at [NaCl] = 0.125 M. The nanoclusters did not re-dissolve over time and remained stable. The sequence was recorded at t1 ~ 20 min after the soluble protein solution was pipetted into the buffer. (B) Time-lapse TEM micrographs of protein nanoclusters growing in liquid at [NaCl] = 0.10 M. The dark areas first grew then re-dissolved over time during the course of observation. The sequence was recorded at

ACS Paragon Plus Environment

11

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 24

t2 ~ 20 min after the soluble protein solution was pipetted in the buffer solution. Note the higher contrast of the protein nanoclusters in (A) compared to (B), which indicates that the clusters formed at higher salt concentration are denser. (C) Plot of protein nanocluster diameter (assuming a spherical diameter) as a function of the observation time t (initial time t = t1 or t2) (square symbols: [NaCl] = 0.10 M; round symbols [NaCl] = 0.125 M). The imaging conditions were maintained constant during growth of the nanoclusters. Specific nanoclusters are represented by different shades of grey. The lines are curve fits following a diffusion-limited growth, where the nanocluster diameter scales as 𝑡𝑛 with n = 0.61. (D) High-resolution electron micrographs in liquid at [NaCl] = 0.10 M highlighting the aggregation of smaller protein oligomers into large nanoclusters. The image was taken after ca. 15 min of observation. The insets are closed-up views of the nanoclusters in the process of aggregating. (E) High resolution micrographs taken in similar conditions after a longer time (ca. 1 hour), highlighting the presence of dense protein nanocluster 5 to 20 nm in size. Since in situ liquid TEM allows the dynamic observation of different protein nanoclusters, we selected the two conditions where enriched protein nanoclusters formation and coarsening could be recorded in order to monitor the kinetics of phase-separation (Figure 3). The time-lapse micrographs presented in Figure 3A and B were recorded ca. 20 minutes after the soluble protein solution was pipetted in the buffer solution containing [NaCl] = 0.125 M and [NaCl] = 0.10 M, respectively (see Supplementary movies 2-4). In both conditions, dark nanoclusters were found to grow during the course of observation. The high contrast supports a high protein density in these nanoclusters, which are likely maintained in close proximity by the initial formation of hydrogen bonds followed by 𝜋 ― 𝜋 interactions39 (see the disappearance of the random coil minimum in the CD curves). In the case of [NaCl] = 0.10 M, the large nanoclusters re-dissolved with time suggesting that the salt concentration was not high enough to sufficiently screen the protein surface charges. However, the contrast of the liquid solution appeared different after re-dissolution than at the onset of observations (compared with Figure 2D for [NaCl] = 0.10 M). In the CD curve, we also note that the random coil minimum was never recovered. These observations were confirmed at several locations within the liquid cell (see Supplementary Figure 7) and suggest that at this salt concentration, hydrogen bonds between amino acid side-chains of HBP-2 could stabilize small oligomeric units but were only transient, resulting in re-dissolution of the nanoclusters. We also note that the nanoclusters were initially irregularly shaped but rapidly evolved towards a more rounded shape (Figure 3B). Therefore we can hypothesize that as the system evolves towards

ACS Paragon Plus Environment

12

Page 13 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

equilibrium (or meta-stable equilibrium), the nanoclusters initially formed by protein aggregates will adopt the spherical shapes of coacervate microdroplets as they coarsen. In addition, the initial rough shape indicates a diffusion-limited growth, which is confirmed by measuring the diameter of the nanoclusters, assuming an equivalent circular shape as they grow, as a function of time (Figure 3C). In both cases, the nanocluster growth followed 𝑑~𝐴 ∙ 𝑡𝑛, with an exponent 𝑛 = 0.61, which is close to a diffusion-limited growth19 (see Supplementary Table 3 for fitting parameters). The aggregation of nanoclusters into larger units is highlighted in Figure 3D for [NaCl] = 0.10 M. Dense 10-20 nm protein nanoclusters converged into larger 200-500 nm nanoclusters with a rough surface (see inset on the right). Furthermore, the white area around the nanoclusters suggests a strong depletion of proteins in their immediate vicinity (see Supplementary Figure 7 for micrographs as a function of time). It is noticeable that the initial growth kinetics of the smaller population of nanoclusters (up to ~ 400 nm) appeared to be faster at [NaCl] = 0.10 M than at 0.125 M. However, it is plausible that the same type of unstable nanoclusters observed at [NaCl] = 0.10 M were not captured at [NaCl] = 0.125 M due to a too fast kinetics of formation. One can suggests that at [NaCl] = 0.125 M, the nanoclusters were stabilized by hydrogen bonds inter oligomers. In addition, similar to previous studies describing nucleation events under electron irradiation45, the growth kinetics factor A was influenced by the illuminating electron dose and should thus be carefully considered (see Materials and Methods and Supplementary Figure 8). Nevertheless, at [NaCl] = 0.10 M re-dissolution of the nanoclusters always occurred after the initial growth and the system was very dynamic, whereas at [NaCl] = 0.125 M we never observed re-dissolution of nanoclusters. This is in agreement with the observation that no microdroplets formed at [NaCl] = 0.10 M, in turn indicating that the system was close to the conditions for phase separation to occur. A closer look at the protein nanoclusters at [NaCl] = 0.10 M revealed that proteins could collapse into nanoclusters of 3 to 20 nm, matching the values measured by DLS (Figure 3E). Since HBP-2 has a molecular weight of 18 kDa, its globular size is predicted to be around 1.4 nm46, suggesting that in the early stages of LLPS oligomers consist of 2 or more monomeric units. In addition, electron diffraction patterns indicating the presence of single salt crystals could also be recorded in some of these dark nanoclusters (see Supplementary Figure 9), whereas the final microdroplets exhibited no diffraction. At [NaCl] = 0.10 M, protein collapsing may have encapsulated ions until supersaturation conditions were reached, which is 6 M in the case of NaCl.

ACS Paragon Plus Environment

13

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 24

Discussion Based on our experiments, we found that early events of LLPS of HBP-2 featured the formation of nanoclusters enriched in protein that rapidly coarsen and concentrate (Figure 4). When sufficient charge screening is allowed, HBP-2 changes its conformation from random coils into 𝛽-sheets to form nanoscale nuclei that subsequently grow in a diffusion-limited fashion. Under these conditions, the nanoclusters do not re-dissolve as they can be stabilized by strong hydrophobic interactions and 𝛽–sheet structures. If the charge screening is not enough, the proteins collapse into small nanoclusters containing only a few oligomers. Based on this, we can assign each region in the map presented in Figure 1G as follow: (i) a soluble region where the proteins are in random coil configuration and remain in solution due to electrostatic repulsions; (ii) a metastable region where nuclei of proteins interacting via transient H-bonds and/or aromatic interactions can grow into irreversible -sheet structures; and finally (iii) the LLPS region that result within 2 minutes in microdroplets stabilized by -sheet structures.

Figure 4. LLPS HBP-2 at pH 5.5 showing the stabilization of protein oligomers via hydrogen bonds and aromatic interactions and dense protein coacervate microdroplets comprising 𝛽-sheet structures. A remarkable feature of these in situ liquid TEM measurements is our ability to observe directly, dynamically, and without staining or contrasting agents, the early stages of LLPS of IDPs. Since no chemical modification is required, namely neither staining nor labeling, the method can

ACS Paragon Plus Environment

14

Page 15 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

be directly applied to other biomacromolecular systems. This may be used to follow not only phase separation of protein coacervate microdroplets but also their transformation into irreversible aggregates such as amyloid fibers27, their crystallization47, or their capacity to encapsulate nanoparticles or molecules12,48. Liquid TEM technology can also be coupled with electron diffraction, which could reveal transitions from amorphous to crystal, or the presence and growth of -sheets-rich nanostructures47. The ability to observe multi-step stages of LLPS in a timeresolved manner with nano-scale resolution, in particular nanocluster formation in the early stages of phase-separation, opens the door to deepen our understanding of LLPS in other protein or protein/RNA complexes of biological relevance40. For example, we anticipate it may be applied to monitor the nucleation and early growth stages associated with the formation of membraneless organelles in reconstituted in vitro systems. Associated content Supplementary information and supplementary movies are provided. Author Contributions C.H. and B.G. synthesized and purified the proteins. H.L.F performed the experiments. B.G., M.D. and A.M. designed the experiments. H.L.F. performed the optical characterization, circular dichroism and dynamic light scattering. H.L.F and M.D. performed the electron microscopy experiments. H.L.F. and A.M. wrote the paper. All authors contributed to the experimental design, data analysis, and discussions. Acknowledgments This research was funded by the Singapore Ministry of Education (MOE) through an Academic Research Fund (AcRF) Tier 2 grant (# MOE2015-T2-1-062). The authors acknowledge the facilities for Analysis, Characterization, Testing and Simulations (FACTS) at Nanyang Technological University for access to TEM equipment. H.L.F. acknowledges support from the Swiss National Foundation for an individual post-doctoral fellowship (# P2EZP2_172169). M.D. acknowledges the financial support from Nanyang Technological University start-up grant M4081924. Notes The authors declare no competing financial interest.

ACS Paragon Plus Environment

15

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 24

References (1)

Chen, Q., Vekilov, P. G., Nagel, R. L. & Hirsch, R. E. Liquid-Liquid Phase Separation in Hemoglobins: Distinct Aggregation Mechanisms of the β6 Mutants. Biophys. J. 86, 1702– 1712 (2004).

(2)

Boeynaems, S., Alberti, S., Fawzi, N.L., Mittag, T., Polymenidou, M., Rousseau, F., Schymkowitz, J., Shorter, J., Wolozin, B., Van Den Bosch, L., Tompa, P. & Fuxreiter, M. Protein Phase Separation: A New Phase in Cell Biology. Trends Cell Biol. 28, 420–435 (2018).

(3)

Shin, Y. & Brangwynne, C. P. Liquid phase condensation in cell physiology and disease. Science 357, 1253-1266 (2017).

(4)

Bahn, S. Y., Jo, B. H., Choi, Y. S. & Cha, H. J. Control of nacre biomineralization by Pif80 in pearl oyster. Sci. Adv. 3, 1-9 (2017).

(5)

Hyman, A. A., Weber, C. A. & Julicher, K. Liquid-Liquid Phase Separation in Biology. Annu. Rev. Cell Dev. Biol. 30, 39-58 (2014).

(6)

Wegmann, S., Eftekharzadeh, B., Tepper, K., Zoltowska, K.M., Bennett, R.E, Dujardin, S., Laskowski, P.R., MacKenzie, D., Kamath, T. ,Commins, C., Vanderburg, C., Roe, A.D., Fan, Z., Molliex, A.M., Hernandez-Vega, A., Muller, D.. Hyman, A.A., Mandelkow, E., Taylor, J.P. & Hyman, B.T. Tau protein liquid–liquid phase separation can initiate tau aggregation. EMBO Journal 37, 1-21 (2018).

(7)

Bergeron-Sandoval, L. P., Safaee, N. & Michnick, S. W. Mechanisms and Consequences of Macromolecular Phase Separation. Cell 165, 1067–1079 (2016).

(8)

Spitzer, J., Pielak, G. J. & Poolman, B. Emergence of life: Physical chemistry changes the paradigm. Biol. Direct 10, 1–15 (2015).

(9)

Stroo, E., Koopman, M., Nollen, E. A. A. & Mata-Cabana, A. Cellular Regulation of Amyloid Formation in Aging and Disease. Front. Neurosci. 11, 1–17 (2017).

(10)

Tan, Y., Hoon, S., Guerette, P.A., Wei, W., Ghadban, A., Hao, Cai, Miserez, A. & Waite, J.H. Infiltration of chitin by protein coacervates defines the squid beak mechanical gradient. Nat. Chem. Biol. 11, 488–495 (2015).

(11)

Wei, W., Tan, Y., Rodriguez, M., Yu, J., Israelachvili, J.N. & Waite, H.J. A musselderived one component adhesive coacervate. Acta Biomater. 10, 1663-1670 (2014).

(12)

Lim, Z. W., Ping, Y. & Miserez, A. Glucose-Responsive Peptide Coacervates with High Encapsulation Efficiency for Controlled Release of Insulin. Bioconjug. Chem. 29, 2176– 2180 (2018).

(13)

Blocher, W. C. & Perry, S. L. Complex coacervate-based materials for biomedicine. Wiley

ACS Paragon Plus Environment

16

Page 17 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Interdiscip. Rev. Nanomedicine Nanobiotechnology 9, 76–78 (2017). (14)

Amiram, M., Luginbuhl, K. M., Li, X., Feinglos, M. N. & Chilkoti, A. Injectable proteaseoperated depots of glucagon-like peptide-1 provide extended and tunable glucose control. Proc. Natl. Acad. Sci. 110, 2792–2797 (2013).

(15)

Mohammadi, P., Aranko, A.S., Lemetti, L., Cenev, Z., Zhou, Q., Virtanen, S., Landowski, C.P., Penttilä, M., Fischer, W.J., Wagermaier, W. & Linder, M.B. Phase transitions as intermediate steps in the formation of molecularly engineered protein fibers. Commun. Biol. 1, 86-98 (2018).

(16)

Kim, S., Huang, J., Lee, Y., Dutta, S., Yoo, H.Y., Jung, Y.M., Jho, Y.S., Zeng, H. & Hwang, D.S. Complexation and coacervation of like-charged polyelectrolytes inspired by mussels. Proc. Natl. Acad. Sci. 113, 847–853 (2016).

(17)

Menger, F. M., Peresypkin, A. V., Caran, K. L. & Apkarian, R. P. Sponge morphology in an elementary coacervate. Langmuir 16, 9113–9116 (2000).

(18)

Luiken, J. A. & Bolhuis, P. G. Prediction of a stable associated liquid of short amyloidogenic peptides. Phys. Chem. Chem. Phys. 17, 10556–10567 (2015).

(19)

Berry, J., Brangwynne, C. P. & Haataja, M. Physical principles of intracellular organization via active and passive phase transitions. Reports Prog. Phys. 81,1-43 (2018).

(20)

Banani, S. F., Lee, H. O., Hyman, A. A. & Rosen, M. K. Biomolecular condensates: organizers of cellular biochemistry. Nature Reviews 18, 285–298 (2017).

(21)

Saad, S., Cereghetti, G., Feng, Y., Piccotti, P., Peter, M. & Dechant, R. Reversible protein aggregation is a protective mechanism to ensure cell cycle restart after stress. Nat. Cell Biol. 19, 1202–1213 (2017).

(22)

Schuster, B. S., Reed, E.H., Parthasarathy, R., Jahnke, C.N., Caldwell, R.M., Bermudez, J.G., Ramage, H., Good, M.C. & Hammer, D.A. Controllable protein phase separation and modular recruitment to form responsive membraneless organelles. Nat. Commun. 9, 1-12 (2018).

(23)

Cai, H., Gabryelczyk, B., Manimekalai, M.S.S., Gruber, G., Salentinig, S. & Miserez, A. Self-coacervation of modular squid beak proteins-a comparative study. Soft Matter 13, 7740–7752 (2017).

(24)

Strom, A. R., Emelyanov, A.V., Mir, M., Fyodorov, D.V., Darzacq, X. & Karpen, G.H. Phase separation drives heterochromatin domain formation. Nature 547, 241–245 (2017).

(25)

Shin, Y., Berry, J., Pannucci, N., Haataja, M.P., Toettcher, J.E. & Brangwynne, C.P. Spatiotemporal Control of Intracellular Phase Transitions Using Light-Activated optoDroplets. Cell 168, 159–171 (2017).

ACS Paragon Plus Environment

17

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(26)

Page 18 of 24

Walton, F. & Wynne, K. Control over phase separation and nucleation using a lasertweezing potential. Nat. Chem. 10, 506–510 (2018).

(27)

Ambadipudi, S., Biernat, J., Riedel, D., Mandelkow, E. & Zweckstetter, M. Liquid-liquid phase separation of the microtubule-binding repeats of the Alzheimer-related protein Tau. Nat. Commun. 8, 1–13 (2017).

(28)

Reichheld, S. E., Muiznieks, L. D., Keeley, F. W. & Sharpe, S. Direct observation of structure and dynamics during phase separation of an elastomeric protein. Proc. Natl. Acad. Sci. 114, 4408–4415 (2017).

(29)

Boire, A., Sanchez, C., Morel, M.H., Lettinga, M.P. & Menut, P. Dynamics of liquidliquid phase separation of wheat gliadins, Scientific Reports 8, 1-13 (2018)

(30)

Da Vela, S., Braun, M.K., Dorr, A., Greco, A., Moller, J., Fu, Z., Fajun, Z. & Schreiber, F. Kinetics of liquid-liquid phase separation in protein solutions exhibiting LCST phase behavior studied by time-resolved USAXS and VSANS. Soft Matter 12, 9334–9341 (2016).

(31)

Moser, T. H., Mehta, H., Park, C., Kelly, R.T., Shokuhfar, T. & Evans, J.E. The role of electron irradiation history in liquid cell transmission electron microscopy. Sci. Adv. 4, 1-8 (2018).

(32)

De Jonge, N. & Ross, F. M. Electron microscopy of specimens in liquid. Nat. Nanotechnol. 6, 695–704 (2011).

(33)

Yamazaki, T., Kimura, Y., Vekilov, P.G., Furukawa, E., Shirai, M., Matsumoto, H., Van Driessche, A.E.S. & Tsukamoto, K., Two types of amorphous protein particles facilitate crystal nucleation. Proc. Natl. Acad. Sci. 114, 2154–2159 (2017).

(34)

De Yoreo, J. J. & Sommerdijk, N. A. J. M. Investigating materials formation with liquidphase and cryogenic TEM. Nat. Rev. 1, 1-18 (2016).

(35)

Proetto, M.T., Ruch, A.M., Chien, M.P., Baeza, P.A., Patterson, J.P., Thompson, M.P., Olson, N.H., Moore, C.R., Rheingold, A.L., Andolina, C., Millstone, J., Howell, S.B., Browning, N.D., Evans, J.E. & Gianneschi, N.C., Dynamics of soft nanomaterials captured by transmission electron microscopy in liquid water, J. Am. Chem. Soc. 136, 1162-1165 (2014)

(36)

Filobelo, L. F., Galkin, O. & Vekilov, P. G. Spinodal for the solution-to-crystal phase transformation. J. Chem. Phys. 123, 1-8 (2005).

(37)

Touve, M. A., Figg, C.A., Wright, D.B., Park, C., Cantlon, J., Sumerlin, B.S. & Gianneschi, N.C. Polymerization-Induced Self-Assembly of Micelles Observed by Liquid Cell Transmission Electron Microscopy. ACS Cent. Sci. 4, 543–547 (2018).

ACS Paragon Plus Environment

18

Page 19 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(38)

Journal of the American Chemical Society

DiMemmo, L. M., Varano, A.C., Haulenbeek, J., Liang, Y., Patel, K., Dukes, M.J., Zheng, S., Hubert, M., Piccoli, S.P. & Kelly, D.F. Real-time observation of protein aggregates in pharmaceutical formulations using liquid cell electron microscopy. Lab Chip 17, 315–322 (2017).

(39)

Gabryelczyk, B., Cai, H., Shi, X., Swinkels, P.J.M., Salentinig, S., Pervushin, K. & Miserez, A. Aromatic Interactions within Repeat Motifs Trigger pH-Responsive LiquidLiquid Phase Separation of Intrinsically Disordered Peptides. (under revision in Nature Communications)

(40)

Brangwynne, C. P., Tompa, P. & Pappu, R. V. Polymer physics of intracellular phase transitions. Nat. Phys. 11, 899–904 (2015).

(41)

Porter, D. A., Easterling, K. E. & M.Y. Sherif. Phase transformations in Metals and Alloys. CRC Press, Tayler & Francis group, Boca Raton, USA. Third Edition (2009).

(42)

Sagui, C. & Grant, M. Theory of nucleation and growth during phase separation. Phys. Rev. E 59, 4175–4187 (1999).

(43)

Tanaka, H. Viscoelastic phase separation in soft matter and foods. Faraday Discuss. 158, 371–406 (2012).

(44)

Balsara, N.P. & Lin, C. Early stages of nucleation and growth in a polymer blend. Phys. Rev. Lett. 77, 3847–3850 (1996).

(45)

Zhang, Y., Keller, D., Rossell, M. D. & Erni, R. Formation of Au Nanoparticles in Liquid Cell Transmission Electron Microscopy: From a Systematic Study to Engineered Nanostructures. Chem. Mater. 29, 10518–10525 (2017).

(46)

Erickson, H. P. Size and shape of protein molecules at the nanometer level determined by sedimentation, gel filtration, and electron microscopy. Biol. Proced. Online 11, 32–51 (2009).

(47)

Zhang, F., Roosen-Runge, F., Sauter, A., Roth, R., Skoda, M.W.A., Jacobs, R.M.J., Sztucki, M. & Schreiber, F. The role of cluster formation and metastable liquid—liquid phase separation in protein crystallization. Faraday Discuss. 159, 313-325 (2012).

(48)

Black, K. A., Priftis, D., Perry, S.L., Yip., J., Byun, W.Y. & Tirrell, M. Protein encapsulation via polypeptide complex coacervation. ACS Macro Lett. 3, 1088–1091 (2014).

ACS Paragon Plus Environment

19

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 24

For Table of Contents only:

ACS Paragon Plus Environment

20

Page 21 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Figure 1 160x148mm (299 x 299 DPI)

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2 160x145mm (299 x 299 DPI)

ACS Paragon Plus Environment

Page 22 of 24

Page 23 of 24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Figure 3 160x162mm (299 x 299 DPI)

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4 160x90mm (299 x 299 DPI)

ACS Paragon Plus Environment

Page 24 of 24