Optical characterization of molecular interaction strength in protein condensates

INTRODUCTION

Cells organize their complex inner structure to control the reactions of different components in space and time. This can be achieved by enclosing specific parts of the cellular interior within a membrane, as is the case for the nucleus, mitochondria, or transport vesicles. Another way, which has attracted increasing interest in recent years, is the formation of membraneless compartments via phase separation. Phase separation of macromolecules into a condensed phase occurs above threshold concentrations of the macromolecules, and the driving force for phase separation is modulated by the concentration of ions, molecular crowders, and ligands, and by physical parameters such as temperature (Brangwynne et al., 2015; Banani et al., 2017).

Due to their distinct molecular composition, biomolecular condensates may provide an environment with particular physical properties impacting the chemical and physical processes within the condensates (Woodruff et al., 2018; Dodson and Kennedy, 2020). Some condensates are highly dynamic accumulations of certain proteins and nucleic acids promoting specific biochemical reactions, such as the nucleolus (Lafontaine et al., 2021) or DNA damage repair sites (Altmeyer et al., 2015; Patel et al., 2015). However, there is also a continuum of more stable, less dynamic condensates (Woodruff et al., 2018), for example, Balbiani bodies during cell dormancy (Boke et al., 2016), or keratohyalin granules which are crucial for skin barrier formation (Quiroz et al., 2020). Moreover, the condensates’ material properties are variable and may change due to the partitioning of different molecules (Elbaum-Garfinkle et al., 2015; Chappidi et al., 2024), or in response to physical parameters like temperature (Alshareedah et al., 2024). Condensates that are viscoelastic fluids initially, may also transition into viscoelastic solids (Alshareedah et al., 2024) or solid cross-beta fibrils (Molliex et al., 2015; Patel et al., 2015) over time—a state often associated with pathology. For example, mutations in the fused in sarcoma (FUS) and hnRNPA1 proteins investigated in this study are implicated in a spectrum of neurodegenerative diseases, including amyotrophic lateral sclerosis (Kim et al., 2013; Li et al., 2013). The question of how material properties of biological condensates relate to their diverse functions has been referred to as a major question in the field of biological phase separation (Woodruff et al., 2018).

For the characterization of the complex viscoelastic properties that biological condensates may exhibit, or to uncover the underlying molecular interactions, the state of the art comprises the use of a set of techniques reporting on different characteristics. A molecular perspective, that is for example taken by nuclear magnetic resonance (Murthy and Fawzi, 2020) or Förster resonance energy transfer (Algar et al., 2019), helps to understand single residue dynamics and specific interaction patterns. For a more general characterization of bulk material properties, however, different approaches are needed. A common method used in the field is fluorescence recovery after photobleaching (FRAP). Assuming a purely viscous fluid, FRAP can be used for a qualitative assessment of relative viscosity (Alshareedah et al., 2021). Challenges of this technique mainly arise due to the required assumptions about the system probed and the consequential difficulty obtaining robust quantitative values (Taylor et al., 2019). Condensate coalescence times are typically utilized to determine inverse capillary velocities, which can be converted to surface tensions with the knowledge of viscosities—again assuming a purely viscous fluid (Ceballos et al., 2018). Optical traps (Jawerth et al., 2018; Alshareedah, Kaur, and Banerjee, 2021; Alshareedah, Moosa et al., 2021) or tracking of tracer particles (Fisher and Elbaum-Garfinkle, 2020) allow the extraction of complex elastic shear moduli, providing direct insight into the viscoelastic properties of condensates. The incorporation of beads or tracer particles into the condensates, however, can be experimentally cumbersome and potentially influence the results obtained.

Moreover, a simple measure of molecular interaction strengths applicable to all states of matter found in biological condensates is not yet available. With these challenges in mind, we chose a purely optical approach that does not rely on fluorescent labels or extensive assumptions and reports on macroscopic material properties and average molecular interaction strength. We employed Brillouin microscopy to investigate biological phase separation at intermediate length and timescales by quantifying the average molecular interaction strength that drives collective molecular motion. One form of collective molecular motion is a thermally excited pressure wave (Figure 1) propagating through a material at the speed of sound, v. Light that is scattered by such a propagating wave experiences a frequency shift , called Brillouin shift. This type of inelastic light scattering was described theoretically by Brillouin and Mandelstam (Brillouin, 1922; Mandelstam, 1926) in the 1920s. The absolute value of the Brillouin shift is proportional to the speed of sound (), which is determined by longitudinal modulus M and mass density ρ as . The longitudinal modulus can be interpreted as an average molecular coupling constant, giving rise to a resistance against uniaxial compression. Important to note is that this resistance is not specific to a particular type of molecule (e.g., proteins); any molecule within the probed volume contributes. A stronger coupling between molecules gives rise to a higher sound velocity and a greater Brillouin shift according to . The molecular motion probed takes place in the subnanosecond regime and averages molecular interactions over a few hundred nanometers.

While the Brillouin shift is associated with the elastic properties of the material (i.e., the longitudinal modulus), the linewidth is related to dissipation. In a liquid that is homogeneous on the length scale of the sound wave, sound attenuation is determined by the internal degrees of freedom of the material that dissipate energy and give rise to its viscosity (see also Supplemental Section S5) (Prevedel et al., 2019; Mattarelli et al., 2020).

While the application of Brillouin microscopy for biophysics research is a relatively recent development (Prevedel et al., 2019; Kabakova et al., 2024), the investigation of viscoelasticity with sound waves has a long history. Ultrasonic investigations were used broadly in material science and polymer physics, as well as to study phase separation (Singh and Singh, 1983; Jadhav et al., 2019). One specific advantage of using a sound-based method to study protein condensates is the possibility to quantitatively characterize material properties in any state of matter. There is no limitation to liquids or solids; in the case of Brillouin microscopy, optical accessibility is the only requirement. Furthermore, the determination of mechanical properties does not involve the application of a mechanical model which requires certain assumptions about the system probed. This is particularly advantageous for samples where the state of matter is not known or changes over time.

As the calculation of the longitudinal modulus based on the Brillouin shift requires knowledge about the refractive index (see Materials and Methods), a combination of Brillouin microscopy and quantitative phase microscopy such as quantitative phase imaging (QPI) or optical diffraction tomography (ODT) is advantageous (Schlüßler et al., 2022). Those techniques are able to capture the complex optical fields, including the phase information, while conventional light microscopes only probe the light intensity. The phase information can be converted into the sample's refractive index distribution, if the object geometry is known (e.g., spherical), or if the specimen is imaged from different angles in a tomographic manner (K Kim, KS Kim et al., 2013; Schürmann et al., 2016). Furthermore, the refractive index can be utilized to calculate protein concentration, volume fraction, as well as mass density, which is also required for longitudinal modulus computation (see Materials and Methods).

Brillouin microscopy has already been applied to membraneless organelles. While other groups have previously studied nucleoli (Antonacci and Braakman, 2016) and FUS condensates in fixed cells (Antonacci et al., 2018), our group has recently employed a combination of spatially resolved refractive index measurements and Brillouin microscopy to study nucleoli, FUS condensates, and poly-Q accumulations in living cells. A key finding was that nucleoli and poly-Q aggregates have a lower compressibility and higher refractive index than the surrounding material (Schlüßler et al., 2022). Following this proof of principle showing the applicability of Brillouin microscopy for the investigation of biomolecular condensates in general, a further establishment of this experimental method needed additional experimental work evaluating the sensitivity of this technique to changes in the physical state of the condensates in a systematic manner. Therefore, we tuned parameters known to affect phase separation and quantified their impact on the physical properties of different protein condensates. Here, we used a combination of Brillouin microscopy and QPI to investigate the impact of solution conditions, temperature, time, and amino acid sequence alterations on protein condensates in vitro, in a noninvasive, purely optical manner. We quantified changes in the molecular interaction strength in terms of Brillouin frequency shift, and dissipative properties in terms of Brillouin linewidth. Additionally, we compute protein concentration and volume fraction from refractive index measurements. We found that protein condensates exhibited higher values in the aforementioned optomechanical quantities in conditions with a stronger tendency for phase separation (i.e., low temperatures [Yoshizawa et al., 2018; Li et al., 2021], high, or low salt concentrations [Krainer et al., 2021]) revealing stronger molecular interactions. We also characterized the aging process of FUS condensates and found that their protein concentration and volume fraction, as well as their dissipative properties, remained unaffected. The aging process was also characterized by a gradual increase in Brillouin shift (i.e., molecular interaction strength); however, this effect was much weaker than the Brillouin shift elevation due to chemical fixation. Accordingly, the observed aging process may be characterized by a progressive establishment of a cross-linked polymer network across the condensates. Finally, we showed that the molecular grammar governing macromolecular condensation also has a strong impact on the physical properties of the condensates. Our experiments with variants of the low-complexity domain of the hnRNPA1 protein (A1-LCD) (Bremer et al., 2022) revealed distinct molecular interaction and dissipation strength, as well as different densities. Moreover, we found that the driving force for phase separation (quantified by the saturation concentration) showed a correlation with the molecular interaction strength in the condensates.

Overall, we demonstrated an all-optical approach that does not rely on fluorescent labels to quantify molecular interaction strength as well as protein concentration of protein condensates. We showed the sensitivity of our measurement techniques to known parameters affecting phase separation and physical properties of protein condensates. Such quantitative approaches are crucial to consolidate the current state of knowledge and to address open questions in the field of biological phase separation.

RESULTS

Experimental setup

Figure 2 illustrates the experimental setup as well as the analysis workflow. Brillouin microscopy was performed by doing point measurements in the condensed and dilute phases of phase-separated protein solutions. Surface wetting of the protein was minimized by polyethylene glycol coating, keeping the condensates in a spherical shape. This enabled the use of QPI to measure the refractive index and calculate protein concentration and volume fraction. Furthermore, the refractive index was utilized to calculate mass density and longitudinal modulus (see Materials and Methods).

Increasing temperature weakens molecular interaction strength in a polymer-rich phase

Temperature is an important thermodynamic control parameter for the formation of protein condensates. Previous reports have revealed that FUS condensation follows an upper critical solution temperature (UCST) phase diagram (Li et al., 2021). Therefore, we asked how temperature increase, which diminishes the driving force for phase separation (Yoshizawa et al., 2018; Li et al., 2021), affects the physical properties of FUS condensates.

To obtain a reference system illustrating the impact of temperature and solute concentration on the average molecular interaction and dissipation strength of a polymer solution, we first measured the Brillouin shift and linewidth of polyethylene glycol 3350 (PEG) solutions. We found that increasing the temperature of water (0% PEG) or a 10% PEG solution increased Brillouin shifts while decreasing linewidths (Figure 3A). Above a PEG concentration of 20%, the slope versus became positive, yielding both an increased Brillouin shift and linewidth with lower temperatures. Our results are in line with previous Brillouin scattering measurements of PEG 600 (Pochylski et al., 2006) and concentrated sucrose solutions (Hosea et al., 1990).

The increase of the Brillouin shift in a dilute polymer solution due to an elevation of temperature from 10°C to 50°C can be explained by the viscoelastic properties of water. Water exhibits a minimum in compressibility (resulting in a maximum of the bulk modulus K) at 45°C (Millero et al., 1969) and a vanishing shear modulus G = 0. Due to the relation , this implies a maximum of the longitudinal modulus (Santucci et al., 2006) and accordingly in the Brillouin shift (see also Supplemental Section S5). In summary, the viscoelastic properties of water dominate at low solute concentrations up to 10% PEG, while the solute dominates at concentrations above 20%.

After the establishment of the PEG-model system, we conducted in vitro experiments on FUS and A1-LCD condensates. To extend the accessible temperature range we added 5% dextran to the FUS buffer system preventing the dissolution of the condensates. In the dilute phase of the phase-separated protein solution (Figure 3B), we found an increase of the Brillouin shift and a reduction of the linewidth as a result of a temperature increase. The same trend was observed before for water and a 10% PEG solution (Figure 3A). The protein condensates on the other hand (Figure 3B), exhibited a reduction of Brillouin shift upon the same temperature change, similar to a 30% PEG solution (Figure 3A). Figure 3 also illustrates that the impact of polymer concentration on linewidth and Brillouin shift decreases for higher temperatures. This effect was observed for the phase-separated protein solution as well as the PEG model system. To compare our results with a widely used technique in the field, we performed FRAP to qualitatively inspect protein diffusion in the FUS condensates. We found a faster intensity recovery at higher temperatures (Figure 3B). This observation together with the measured trends in the Brillouin linewidth shows a reduction of dissipation upon temperature increase.

In addition, we investigated whether an increase in temperature would also result in a decrease in refractive index (Kolsk´a et al., 2019) and density (Gonzalez-Tello et al., 1994; Ebrahimi and Sadeghi, 2015) as previously reported for PEG solutions. Such behavior would be expected for a narrowing two-phase regime at increasing temperature in a UCST system. Indeed, we observed a subtle decrease in refractive index and protein concentration of FUS condensates upon a temperature increase from 25°C to 35°C (Figure 3C), similar to what was reported for mEGFP-TAF15 condensates (McCall et al., 2023). The same trend was found for A1-LCD, but the accessible temperature range was more restricted, as we did not use a crowding agent in this case. The onset of condensate dissolution above 30°C and the consequent lack of spherical condensates prevented QPI measurements at higher temperatures.

As the approximation is justified in our case (see Supplemental Section S6 for further details), it follows that and a measurement of the Brillouin shift is sufficient to detect changes in the longitudinal modulus. Accordingly, we report instead of M as a measure of molecular interaction strength in the following text. To facilitate the comparison of the different factors that were investigated in this study (temperature, ion concentration, and aging), we utilized the dilute phase for the normalization of Brillouin shift squared and linewidth of the dense phase. This approach helps to highlight the distinct changes of the physical properties of the dense phase compared with the dilute phase. Similar normalized quantities employing water as a reference were earlier introduced as “Brillouin elastic contrast” for the Brillouin shift and as “Brillouin viscous contrast” for Brillouin linewidth, respectively (Antonacci et al., 2020).

Plotting these normalized quantities for the phase-separated FUS system as a function of temperature shows a clear reduction of and upon temperature increase (Figure 3D). The local maximum of Brillouin linewidth for A1-LCD condensates at 15°C together with the monotonic trend in the Brillouin shift could denote a relaxation process (Bailey et al., 2020); that is, the timescale of molecular motion was slowed down into the regime of the experimentally accessed timescale of few hundred picoseconds.

In summary, we found a reduction in molecular interaction and dissipation strengths in the protein condensates upon temperature increase, accompanied by a reduction in protein concentration, where there was also evidence for a molecular relaxation process in the A1-LCD condensates indicated by a distinct signature in Brillouin shift and linewidth.

Variation of ion concentration reveals distinct interaction regimes

Another important parameter that can tune the interactions between proteins is the ion concentration in the surrounding medium. Ions have specific effects on protein–protein interactions as they directly affect the various possible interaction mechanisms of amino acids (Brangwynne et al., 2015) and the solubilities of proteins. A previous study demonstrated that FUS exhibits a reentrant phase behavior, resulting in two phase separating regimes, one at low and another one at high ion concentrations (Krainer et al., 2021). Also, the phase separation behavior of A1-LCD in response to different ion concentrations has already been investigated. In contrast to the FUS protein, a larger ion concentration leads to a lower saturation concentration of A1-LCD, whereas a reentrant phase separation was not observed (Martin et al., 2021). We hypothesized that these ion-dependent interaction regimes should be also reflected in the physical properties of the protein condensates. As above for exploring the effect of altering temperature, we added 5% dextran ensuring the presence of FUS condensates of suitable size across the full range of KCl concentrations.

Figure 4 illustrates the distinct impact of ion concentration on the physical properties of the protein condensates. While the buffer solutions showed a linear dependence of refractive index on ion concentration (see Supplemental Section S2 and Supplemental Figures S6, A and B), as well as the Brillouin shift of the dilute phases (Supplemental Figures S6, C and D), the dense phases exhibited more complex dependencies. The calculation of protein concentration and volume fraction (Figure 4A), which is based on the refractive index difference of buffer and sample (see Materials and Methods), showed that the maximum of protein concentration and volume fraction of FUS protein condensates is found at the lowest KCl concentrations. Toward higher KCl concentrations, in the region where FUS condensates would dissolve without the addition of a crowding agent (i.e., 0.5 M and 1 M KCl), protein concentration and volume fraction showed a minimum, before slightly increasing again at even higher KCl concentrations. A1-LCD condensates, on the other hand, were found to be generally denser than FUS condensates and showed a further increase of protein concentration and volume fraction upon higher NaCl concentrations.

Finally, as for the different temperatures in Figure 3, we computed normalized values for Brillouin shift squared and linewidth (Figure 4B). For FUS condensates, we identified a minimum in both quantities around a KCl concentration of 1M, while the largest values were located at the lowest ion concentrations, similar to the trends of protein concentration and volume fraction. In the same line, A1-LCD showed larger values of Brillouin shift and linewidth at higher NaCl concentrations.

In summary, our results show that the physical properties of the condensates change according to the previously identified interaction regimes. In addition, we found that low ion concentrations lead to denser FUS condensates with stronger molecular interactions compared with high ion concentrations.

FUS condensate aging is ion concentration dependent

Multiple publications have investigated the time dependence of viscoelastic properties of FUS protein condensates. In general, the aging process of FUS condensates resulted in reduced protein mobility. The states of matter described for aged FUS condensates range from viscous fluids (Jawerth et al., 2020), and mixed liquid-gel states (Shen et al., 2023), to fibrillar solids (Patel et al., 2015; Shen et al., 2023). Given this broad range of results, we used Brillouin microscopy and QPI in various buffer conditions to provide an unbiased view of the aging of FUS condensates.

Previous publications reported morphological and mechanical changes in a time frame of about 12 to 24 h (Patel et al., 2015; Jawerth et al., 2020). Accordingly, we conducted QPI measurements shortly after sample preparation, as well as 12, 24, and 36 h later. In contrast to some previous publications, we could not find morphological changes even after 36 h, other than Ostwald ripening. Accordingly, bigger condensates were growing while smaller ones were shrinking (Figure 5A). In contrast to a previous study on PGL3 condensates reporting an increase in protein concentration over time (McCall et al., 2023), we could not detect a change in refractive index, protein concentration, or protein volume fraction (Figure 5B).

We also conducted Brillouin microscopy measurements over the same time course as the QPI measurements. We were aiming to investigate the same condensate population over the whole time course. In the case that a particular condensate became inaccessible we chose a new one that was then monitored over the remaining time. To minimize any possible effects of the laser light, we chose a laser power of about 11 mW, even though much higher laser powers (>40 mW at 660 nm) have already been shown to be safe for imaging live cells (Nikolić and Scarcelli, 2019). As data acquisition was easier for condensates with a diameter of at least 5 µm, due to reflections caused by the refractive index jump at the condensate interface, our measurements were performed mainly on condensates that were growing (Figure 5A).

We found that the aging process of the condensates was characterized by a shallow increase in of the condensate population in the case of low ion concentrations. At a KCl concentration of 2.5 M, however, this trend was suppressed (Figure 6A). Similar to the density of the condensates, the dissipative properties of the condensates were also mostly unaffected by the condensate aging (Figure 6B). An increase in viscosity that was reported previously (Jawerth et al., 2020) should have been reflected in an increase in Brillouin linewidth.

To create a reference system for a completely solidified protein condensate, we added 0.05% glutaraldehyde to a phase-separated FUS-protein system. To verify the solidification of the condensates we increased the temperature up to 40°C after the measurement. As the condensates did not dissolve, we concluded that the solidification was irreversible. The chemical cross-linking led to an increase in of 29% leading to the highest value of all conditions investigated in this study.

These findings suggest that the observed ion-dependent condensate aging process is not characterized by the formation of ordered crystalline structures, or an increase in viscosity, but rather by the development of an elastic network across the condensates.

Relationship between driving force for phase separation and the optomechanical properties of condensates

To interrogate the link between the driving force for phase separation and optomechanical properties, we conducted QPI and Brillouin microscopy experiments with designed variants of the A1-LCD. The amino acid sequences can be found in the supplement (Supplemental Table S14). The variants differ in their composition of aromatic residues, which were previously identified as the main drivers of pairwise interactions that generate three-dimensional networks and mediate homotypic phase separation in this system (Martin et al., 2020). By substituting the other aromatic residues with tryptophan, which has the largest aromatic ring system, we were able to increase the driving force for phase separation of the resulting protein variant. In the first mutant, only the seven tyrosine residues of the wild-type (WT) were substituted with tryptophan residues (YtoW). In a second mutant, all 12 phenylalanine residues were substituted with tryptophan residues (FtoW), further increasing the phase separation propensity. Finally, all 12 phenylalanine and seven tyrosine residues were mutated to tryptophan residues (allW) yielding a strongly phase-separating variant. Together, these four variants provided a broad range of phase separation propensities which is reflected in a range of saturation concentrations spanning 2 orders of magnitude (Alshareedah et al., 2024).

We observed an overall negative trend of the saturation concentration with the protein concentration, but no consistent correlation (Figure 7A). Figure 7B shows that follows a logarithmic relationship with the saturation concentration according to . The Brillouin linewidth, on the other hand, shows similar values for allW and FtoW, as well as for YtoW and WT, but does not follow a consistent trend with the saturation concentration.

In summary, we find that altering pairwise protein–protein interaction strengths by changing the molecular grammar of protein sequences does not only change the saturation concentration but also the physical properties of the respective condensates. However, the presented data highlights that even though measured quantities are related to intermolecular interactions, they are not necessarily connected by simple relationships.

DISCUSSION

In this work, we present Brillouin microscopy as a promising tool for the characterization of the physical properties of biomolecular condensates, as it introduces new ways to quantify the average molecular interaction and dissipation strength. The combination with QPI offers the possibility to obtain the condensate density and allows a precise calculation of the longitudinal modulus. In contrast to other commonly employed techniques, our experimental approach is noninvasive, does not rely on fluorescent labels, and can be applied in vitro and in vivo.

We showed that parameters with an impact on the phase diagram (i.e., temperature, ion concentration, and amino acid sequence) also affect the physical properties of protein condensates. Molecular interactions are therefore not only related to the saturation concentration but can be also accessed directly via the optomechanical properties of the dense phase. In general, the absolute values of longitudinal modulus and protein concentration are relatively high for condensed biological materials. We found protein concentrations between 0.26 g/ml for FUS condensates at 0.5 M KCl and 0.60 g/ml for A1-LCD allW and longitudinal moduli between 2.65 GPa for FUS condensates at 40°C and 4.26 GPa for A1-LCD WT at 0.5 M NaCl. Typical dry mass densities for HeLa cells for example cover a range from about 0.07 g/ml in the nucleoplasm to 0.15 g/ml in nucleoli (Kim and Guck, 2020), corresponding to a longitudinal modulus of about 2.4 GPa to 2.5 GPa. Intracellular FUS condensates in living cells fall in the same range (Schlüßler et al., 2022).

Similar values for protein concentration (0.26 g/ml) and longitudinal modulus (3.1 GPa) were previously found for intracellular polyQ aggregates in vivo (Schlüßler et al., 2022). Concentration measurements of phase-separated FUS-EGFP (Ahlers et al., 2021), as well as low-complexity domain molecules of FUS (Burke et al., 2015; Murthy et al., 2019) and A1-LCD (Martin et al., 2020; Bremer et al., 2022) found values of a few hundred mg/ml. Furthermore, a recent study employing QPI measured 0.34 g/ml for FUS-mEGFP (McCall et al., 2023) which is in line with our findings. Overall, these results show that the average molecular interaction strength and density of the investigated condensates in vitro are higher in living cells.

The differences between monocomponent condensates in vitro and complex multicomponent condensates in cells may underlie these substantial differences. Our study highlights the fact that the physical properties of protein condensates may be variable and context dependent, responding to external parameters such as thermodynamic variables. We also demonstrate how the physical properties can be tuned by altering the amino acid sequence. Therefore, the complex biochemical environment in living cells will modulate their properties, for example, due to the incorporation of additional molecules into the condensates or by posttranslational modifications (Mitrea et al., 2022).

Uncertainties for the calculation of the density, protein concentration, and related quantities (i.e., protein volume fraction, longitudinal modulus) arise from the application of dextran as a crowding agent in some of the conditions studied (see Supplemental Section S8 for further details). It was found that in the absence of specific interactions and with high molecular weight crowders, depletion effects dominate and the crowder molecules are excluded from the condensates (Chauhan et al., 2023). In light of these findings, the underlying assumption for our calculation of the density was that the crowder (dextran T500) is largely excluded from the protein condensates, and the values for α and θ were chosen accordingly. Nevertheless, we found that the addition of 5% dextran resulted in a small reduction in Brillouin shift and linewidth of the FUS condensates at low KCl concentrations (see Supplemental Section S9 for further details). Whether this difference arises from a partitioning of crowder molecules into the condensates still needs to be investigated. Furthermore, we assumed identical ion concentrations in the dilute and the dense phase, while we aware that ion partitioning in dilute and dense phase may differ (Posey et al., 2024). The impact of this uneven ion distribution on the refractive computation, however, should be negligible due to the weak dependency of ion concentration on refractive index (see Supplemental Section S2 for further details). It should be noted that refractive index measurements of nonspherical membraneless organelles in cells would require a tomographic imaging technique, such as ODT.

Our investigation of PEG solutions and protein condensates at different temperatures revealed characteristic behavior for different solute concentrations. We found that water dominates the response in the dilute phase of the phase separated protein samples and in PEG solutions up to a concentration of about 10%. For PEG concentrations higher than 20%, and in the investigated protein condensates, the polymers have a greater impact on viscoelasticity. This fits with the fact that both polymers have now exceeded their overlap concentrations and have entered the semidilute regime, greatly increasing the likelihood of interactions and entanglements (Wang et al., 2021). Consequently, we find a decrease in molecular interaction and dissipation strength upon an increase in temperature. Such behavior is also known from glass-forming liquids (Klieber et al., 2013; Bailey et al., 2020). Consistent with the decrease in Brillouin linewidth, polymer solutions typically show lower viscosities at higher temperatures (Moore, 1961; Ebrahimi and Sadeghi, 2015). A reduction in viscosity with increasing temperature was also recently observed in A1-LCD condensates (Alshareedah et al., 2024). The lowering of molecular interaction strength due to a temperature increase was also reflected in smaller protein concentrations in the condensates. Our findings are in line with experimental phase diagrams of A1-LCD (Martin et al., 2020) as well as coarse-grained simulations of FUS condensates (Chou and Aksimentiev, 2020). We speculate that the local maximum of the Brillouin linewidth at 15°C could be the hallmark that the molecular relaxation time in the A1-LCD condensates moved into the regime of the probed timescale of few hundred nanoseconds (Prevedel et al., 2019). Such a transition was already observed for gelatin (Bailey et al., 2020) at protein weight fractions above 0.4 (corresponding to volume fractions of about 0.3) and would be consistent with a further increase of protein volume fraction in A1-LCD condensates for temperatures below 22°C. These results highlight that temperature strongly impacts the physical properties of protein condensates.

We also investigated the effect of varying ion concentrations on the physical properties of protein condensates, as ionic strength is a strong determinant of phase separation. In the case of FUS, the complex interplay of ion concentration and molecular interactions leads to reentrant phase separation (Krainer et al., 2021). At a low salt concentration below 125 mM, electrostatic interactions contribute to phase separation. These interactions are shielded by the presence of higher ion concentrations. At very high KCl concentrations above 1.5 M, however, hydrophobic and nonionic interactions drive the system again into a phase-separating regime.

Our results agree with these interaction regimes and further reveal the ion dependence of molecular interactions in FUS condensates. In the presence of a crowding agent, we found a drop of protein concentration in the dense phase at KCl concentrations above 100 mM associated with a reduction of Brillouin shift and linewidth. Furthermore, we discovered that molecular interactions were strengthened at KCl concentrations above 1 M but did not reach the levels of the low salt conditions, even above the previously reported reentrant point of 1.5 M KCl. Also, protein concentration and volume fraction reached their maximum at the lowest ion concentrations. We conclude that the overall molecular interactions are weaker in the high salt regime. Apparently, the reduction of electrostatic interactions is not completely compensated by the hydrophobic and nonionic interactions.

In the case of A1-LCD an increase of ionic strength leads to a reduction of the saturation concentration (Martin et al., 2021). Note that full-length hnRNPA1 shows the opposite behavior. As for the FUS condensates, we found that molecular interaction strength and density of the condensates correlated with the saturation concentration. Accordingly, a larger NaCl concentration led to higher protein concentration in the dense phase and a larger molecular interaction and dissipation strength.

Protein condensates are metastable fluids that can undergo changes in their morphologies and material properties. This may manifest in a slowing of molecular dynamics upon condensate aging or even in the formation of solid aggregates (Alberti and Hyman, 2021). Although the transition of the majority of FUS condensates into fibrous aggregates was previously reported to take place after 12 h (Patel et al., 2015), we only observed Ostwald ripening even 36 h after phase separation. Our observations are in agreement with the cryo-EM images of Jawerth et al. (Jawerth et al., 2020), where morphological changes were only occasionally visible on the condensate interface but not in the bulk material. Along the same line, recent publications suggest that the aging of FUS protein condensates is an inhomogeneous process resulting in different material properties within the same condensate (Garaizar et al., 2022), and solidification can be promoted by the condensate interface (Shen et al., 2023).

We found that the aging process of the FUS condensates is accompanied by an increase of the Brillouin shift at low ion concentrations, while a change in linewidth and protein volume fraction was not detected. Our results may be interpreted as increasing polymer crosslinking, similar to what was found in gelatin (Bailey et al., 2020) and gelatin methacryloyl (GelMA) (Rad et al., 2022). The change of Brillouin shift and longitudinal modulus induced by cross-linking of a polymer solution is typically small if the polymer volume fraction is kept constant (Jarry and Patterson, 1981; Bot et al., 1995; Bailey et al., 2020). An increase in shear viscosity as reported previously (Jawerth et al., 2020) should have been reflected in a larger Brillouin linewidth, although the relationship between shear viscosity and Brillouin linewidth is not straightforward (Illibauer et al., 2024) (see also Supplemental Section S5).

Interestingly, the increase of the Brillouin shift was suppressed at a KCl concentration of 2.5 M, suggesting that a high abundance of ions attenuates the molecular interactions driving the aging process. A similar effect (i.e., suppression of aging) of high salt concentrations was previously reported for PGL-3 condensates (Jawerth et al., 2020). These results suggest that the electrostatic interactions operating at low ion concentrations not only result in higher protein concentration and stronger average molecular interactions but are also important for the aging process of FUS condensates. The chemical cross-linking via the addition of glutaraldehyde led to a strong increase in the Brillouin shift and linewidth illustrating the effect of a complete condensate solidification.

We conclude that the ion-dependent aging that we observed was not the transition into semicrystalline fibers, as this would have been associated with morphological changes, a large increase in Brillouin shift, and probably changes in Brillouin linewidth and protein volume fraction.

Our investigations of A1-LCD variants with increasingly strong driving forces for phase separation revealed an overall negative trend of saturation concentration with protein concentration and Brillouin linewidth of condensates, but not an absolute correlation. Specifically, YtoW condensates have a higher density than FtoW condensates despite having a higher saturation concentration and, the FtoW variant has a higher Brillouin linewidth than may be expected compared with the YtoW and allW variants. These two observations may be related to the length scales of the underlying interactions or the modulation of hydrodynamic effects by the different types of aromatic interactions. We have recently observed a similar decoupling of viscosities from the driving force for phase separation in variants that are dominated by different types of aromatic residues (Alshareedah et al., 2024).

The negative trend with the saturation concentration is more apparent for the Brillouin shift; hence, the interaction strength between the proteins impacts both the saturation concentration and the average molecular interaction strength within the condensed phase. A comparison of A1-LCD and FUS shows that the FUS protein forms less dense condensates with weaker interactions compared with the A1-LCD variants, although the saturation concentration of FUS is relatively low (about 2 µM at 150 mM KCl) compared with most of the A1-LCD variants. One should note here that in the case of FUS, we studied the full-length protein with a GFP-tag, not the low-complexity domain in isolation. While we have not explicitly investigated the impact of a fluorescent label, the comparison of the properties of condensates assembled by labeled and unlabeled protein would be an interesting topic for future studies as many experimental techniques in the field of biomolecular condensates are based on the use of fluorescent labels. Our findings also further substantiate the sensitivity of the employed optical techniques to molecular interactions underlying phase separation. The sequence-dependent effects for protein concentration and Brillouin linewidth indicate an interesting decoupling of thermodynamics and hydrodynamics in variants that are dominated by different types of aromatic residues (Alshareedah et al., 2024).

Finally, we want to discuss our methods and results with regard to biological systems. Over the past years, fluorescence microscopy has been the workhorse of molecular biology. It casts a spotlight on a selected molecular species, allows to localize them, and thereby enabled the discovery of membraneless organelles with liquid-like properties (Brangwynne et al., 2009). The majority of the abundant molecules in the investigated organelle and around remains invisible, though. While intracellular condensates appear as dense accumulations under a fluorescence microscope, it often remains unclear to what extent such accumulations of molecules really provide a different physical environment than the surroundings. The material properties of the condensates, however, have a crucial impact on the behavior of their content.

Less dense condensates with similar material properties as the surrounding intracellular space may interact as reaction sites (Banani et al., 2017). Condensates exhibiting strong molecular interactions and high density can serve as a cellular mechanism to shut down biochemical reactions as a response to external stress or to enable cellular dormancy (Woodruff et al., 2018). Moreover, recent research suggests that some condensates might also exert forces on other cellular organelles (Quiroz et al., 2020). Therefore, cells need to regulate and sometimes actively change the physical properties of membraneless organelles, for example during germ cell development (Dodson and Kennedy, 2020) or in DNA repair sites (Chappidi et al., 2024). Our research also revealed that external stimuli, as well as intrinsic properties of the proteins, give rise to many subtle changes in the physical properties of the investigated condensates, which cannot be simply covered by qualitative terms such as “liquid-like” or “solid-like.”

Unraveling the implications of these diverse material properties and how cells control them will be a major task for the field of biomolecular condensates. This task requires the measurement of bulk physical properties in a quantitative, robust, and noninvasive way as reported here. The bulk quantities we measure, such as protein concentration, volume fraction, as well as sound velocity, and longitudinal modulus, may also be useful in refining coarse-grained simulations of condensates (Benayad et al., 2021; Tesei et al., 2021; Bremer et al., 2022; Farag et al., 2022) which can complement experimental results.

Outlook

While the presented experimental approach has great potential to advance our understanding of the physical properties of biomolecular condensates in general, we identified some opportunities for further improvement. A useful modification of the Brillouin microscope could be the employment of a spectrometer with a larger free spectral range. While the free spectral range is sufficient for most biological samples, we found that in some of the experimental conditions, an unambiguous evaluation of the Brillouin shift was obstructed by an overlap of the Brillouin peaks.

A powerful addition to the current instrumentation would be the possibility to detect shear waves as demonstrated by Kim et al. (Kim et al., 2016). This would allow to directly measure the shear modulus, extract the bulk modulus from the longitudinal modulus, and derive an even more detailed physical description of the sample. Moreover, the detection of shear waves could be another modality to quantify a solidification process, because liquids are not permissive for shear wave propagation.

Furthermore, faster data acquisition will be important for future studies targeting condensates in living cells. While the refractive index distribution of a full field of view covering multiple cells can be measured in about a second (Kim and Guck, 2020), the employed Brillouin microscope with its confocal point scanning approach (∼0.1 to 1 s per pixel) poses a bottleneck as small and dynamic condensates might diffuse out of focus during the data acquisition. However, the acquisition speed of Brillouin microscopes is rapidly increasing and new technical innovations including full-field Brillouin microscopes are on their way (Bevilacqua and Prevedel, 2024).

Less dynamic and larger condensates, such as Balbiani and nuclear amyloid bodies, may therefore be promising next targets for Brillouin microscopy. There are a couple of burning questions regarding this type of condensates which urgently need fresh experimental concepts. What are the differences between physiological and pathological amyloid bodies? Why can cells dissolve the physiological ones? Could differences in the physical properties of physiological and pathological amyloid bodies explain this? We are convinced that the presented methods can be an important building block for future experiments that may answer these and other fundamental questions.

MATERIALS AND METHODS

Request a protocol through Bio-protocol.

Protein purification

FUS-GFP

FUS-GFP protein was expressed from insect cells using a baculovirus expression system (Lemaitre et al., 2019). Cells were lysed using an LM10 Microfluidizer (Microfluidics, USA) at 10,000 psi in lysis buffer containing 50 mM Tris-HCl pH 7.4, 1 M NaCl, 5% glycerol, 1 mM DTT, 33 U/ml benzonase, and complete, EDTA-free Protease Inhibitor Cocktail (Merck). Lysed cells were centrifuged at 60,000 RCF for 1 h to separate soluble from insoluble fractions. The soluble fraction was bound to amylose resin (NEB, USA) for 1 h and washed with 14 column volumes of wash buffer containing 50 mM Tris-HCl pH 7.4, 1 M NaCl, 5% glycerol, 1 mM DTT. The protein was cleaved from the MBP by loading one column volume of the same buffer supplemented with 6 µg/ml 3C protease (produced in house), and incubating for 3 h. Protein was eluted in two column volumes of the same wash buffer, and the elution fraction was concentrated to 5 ml using an Amicon Ultra Centrifugal Filter Unit (30 kDa MWCO, Merck). The concentrated protein was applied to size exclusion chromatography using a HiLoad 16/600 Superdex 200 pg (Cytiva) column on the Akta Pure (Cytiva) FPLC system in 50 mM Tris-HCl pH 7.4, 750 mM KCl, 5% glycerol, 250 mM urea, and 1 mM DTT storage buffer. The protein was concentrated to 60 to 100 µM using an Amicon Ultra Centrifugal Filter Unit (30 kDa MWCO, Merck) and stored at −80°C.

hnRNPA1-LCD

A1-LCD construct design and purification were carried out as described previously (Bremer et al., 2022; Alshareedah et al., 2024). In brief, the A1-LCD sequence was cloned into pDEST17, with a TEV cleavage site to allow for His-tag removal. Proteins were expressed in BL21(DE3) cells overnight at 37°C using autoinduction media. The protein was purified from the insoluble fraction using denaturing buffers and nickel-chelating resin. The tag was cleaved with His-tagged TEV protease, which was removed with a second nickel column. The proteins were further purified on a size exclusion column under denaturing conditions.

Brillouin microscopy

The Brillouin microscope was already described in detail elsewhere (Schlüßler et al., 2018). Briefly, a diode laser (TA pro 780, Toptica, Germany) was stabilized to the D2 transition of Rubidium 85 at 780.24 nm. To suppress amplified spontaneous emission, the light was transmitted through a Fabry–Pérot interferometer in a two-pass configuration and guided on a Bragg-grating. The light was coupled to a Zeiss Axiovert 200 M microscope stand (Carl Zeiss, Germany) and directed to the sample through a 40x/0.95 air objective (Carl Zeiss, Germany). The backscattered light was collected in a confocal manner and analyzed with a two-stage VIPA spectrometer. We performed ∼40- to 50-point measurements for several protein condensates for each condition. The number of evaluated condensates for each figure is given in the supplement (Supplemental Table S1). Dilute phase measurements consist of ∼ 30 spectra from one location.

Data analysis was conducted with a custom software written in Python (Raimund Schlüßler and others, 2023). Brillouin spectra were fitted with two individual Lorentzian peaks in the case of dilute phase and the sum of four Lorentzian peaks in the case of protein condensates. The latter approach was necessary as in most cases also the Brillouin peaks of the dilute phase were contributing to the spectrum while all peaks were overlapping. To improve the quality of the fit, a measurement of the dilute phase was used to constrain the Lorentzian peaks correspondingly. Note that reported linewidths were broadened by the spectrometer and also by the distribution of scattering angles collected by the objective (Antonacci et al., 2013). Temperature was controlled via a custom-built temperature stage (Mittasch et al., 2018).

QPI

QPI was performed with a previously described setup (Schlüßler et al., 2022) which is also suitable for ODT. It is based on Mach-Zehnder interferometry, employing a frequency-doubled Nd-YAG laser with a wavelength of 532 nm (Torus, Laser Quantum, UK). A 40x/1.0 water-dipping objective lens (Carl Zeiss, Germany) was used for sample illumination. The light diffracted by the sample was collected with a 63x/1.2 water immersion objective (Carl Zeiss, Germany). For all temperatures above room temperature (23°C), the sample temperature was set by two foil objective heaters (Thorlabs, USA), one for each objective, controlled by a TC200 temperature controller (Thorlabs, USA).

Phase images were computed from raw holograms that were acquired at a straight angle using qpimage (Müller, 2022). From the phase image covering the full field of view, individual FUS condensates were segmented by a custom Python software (Beck, 2023). The refractive index of the individual FUS-condensates was calculated from the segmented phase images by fitting a sphere using the Rytov approximation employing qpsphere (Müller et al., 2018; Müller, 2021). Only droplets with a diameter greater than 2 µm were considered for the evaluation.

Computation of density, protein concentration, and volume fraction

The mass density ρ was calculated with the following formula based on a binary mixture model:

employing the assumption of volume additivity and the Biot mixing rule of refractive indices (Schlüßler et al., 2018; Möckel et al., 2024). In Eq. 1, denotes the refractive index of the solvent, the density of the solvent, θ the proteins partial specific volume, and α its refractive index increment. The first term in Eq. 1 can be identified as the protein concentration c (Barer, 1952; Davies and Wilkins, 1952):

Using the information of the amino acid sequences of each protein, the values for θ and α were estimated based on the method proposed in (Zhao et al., 2011) (see Supplemental Table S13). Here, instead of employing the Wiener mixing rule for dilute solutions under use in (Zhao et al., 2011), the Biot mixing rule was used to maintain consistency with the above Eq. 1 (Möckel et al., 2024). The refractive index of the solvent was measured with an Abbe-refractometer (2WAJ, Arcarda, Germany) (see Supplemental Table S11B and Supplemental Figure S1).

The protein volume fraction ϕ was calculated based on the protein concentration with the following formula:

Effects of temperature on α, θ, and were neglected and the respective values at room temperature were used to calculate refractive index, dry mass, volume fraction, and longitudinal modulus.

Longitudinal modulus computation

All measurements were performed in the back-scattering configuration, where the scattering angle θ (as defined in Figure 2) is approximately zero. Accordingly, the longitudinal modulus can be calculated by

where λ denotes the incident wavelength of 780.24 nm. Equation 4 may be also written as

where denotes a setup-specific constant and . We found variations of b in the range of only about 2% (see Supplemental Section S6 for more details). Therefore, and are reasonable approximations in the context of the presented data. For specimens with a more complex composition, where not only proteins are present, this approximation needs a more cautious treatment (Möckel et al., 2024).

Data normalization

The values for the dilute phase, that were used to normalize longitudinal Modulus, Brillouin shift, and linewidth were either obtained by the respective median value of the dilute phase (Figure 3) or by taking the value corresponding value of a linear fit (Figure 4) in order to not transfer noise from the buffer measurement to the normalized values.

FRAP

FRAP experiments were carried out with an Andor spinning disk microscope (Nikon TiE inverted stand, Nikon Apo 100x, NA 1.49 Oil objective, Andor iXon+ camera) equipped with a FRAPPA unit (Andor, Northern Ireland). Temperature was controlled the same way as for the Brillouin microscopy experiments. Condensates were bleached with a 488 nm pulse with 20 µs dwell time. Fluorescence recovery was recorded in a single focal plane. For each recorded temperature recovery curves of five individual condensates were recorded and analyzed with a custom python software using trackpy to correct for stage drifting (Allan et al., 2021).

Sample preparation

FUS protein samples

If not stated differently, the protein stock solution (see Materials and Methods, Protein purification, FUS-GFP) was diluted in a buffer containing 1 mM DTT, 20 mM PIPES at a pH of 7.4 to achieve a concentration of 5 µM FUS. If indicated, a concentration of 5% (wt/wt) dextran T500 (Pharmacosmos A/S, Denmark) was used to ensure phase separation at intermediate salt concentrations and high temperatures. For imaging the sample was loaded into a chamber, built up by a PEG-coated cover slide at the bottom and a silicone spacer (Grace Bio-Labs, USA) with a height of 0.9 mm and a diameter of 4.5 mm, resulting in a sample volume of about 14 µl. PEG coating was performed according to a protocol published earlier (Alberti et al., 2018). For Brillouin microscopy and FRAP experiments, the glass surface of the temperature control stage served as the upper boundary of the sample chamber. For the QPI measurements, an untreated cover slide was used instead, as the temperature control stage was not compatible with the QPI setup.

Condensate fixation was performed by the addition of 0.05% glutaraldehyde after the initialization of phase separation by dilution of the stock protein solution with water.

hnRNPA1-LCD protein samples

Amicon Ultra centrifugal filters (Merck, Germany) were filled with 440 µl of 1 M MES pH 5.5 buffer. A total of 60 µl of ∼500 µM protein in 4 M GdmHCl was added. Filters were centrifuged for 15 min at 21,000 × g. Afterward, the sample was filled up to the initial volume of 500 µl with MES buffer. The centrifugation step was repeated until a 1000-fold dilution of the initial GdmHCl buffer was achieved. After the last centrifugation step, the centrifugal filter was refilled with 20 mM HEPES pH 7.0, and the previous procedure was repeated until a 10,000-fold dilution of the MES buffer was achieved. Phase separation was induced by adding NaCl to a final concentration of 150 mM. Samples were loaded on a PEG-coated cover slide and enclosed by an imaging spacer (Grace Bio-Labs, USA).

PEG 3350 solutions

Polyethylene Glycol (Sigma, USA) was dissolved in water to obtain a stock solution with a mass fraction of 50%. PEG solutions with a lower mass fraction were derived from the stock solution by dilution with water. Temperature control and sample mounting were conducted as for the FUS protein samples (see Materials and Methods, Sample preparation, FUS protein samples).

FOOTNOTES

REFERENCES

• Ahlers J, Adams EM, Bader V, Pezzotti S, Winklhofer KF, Tatzelt J, Havenith M (2021). The key role of solvent in condensation: Mapping water in liquid-liquid phase-separated FUS. Biophys J 120, 1266–1275. Crossref, Medline, Google Scholar
• Alberti S, Hyman AA (2021). Biomolecular condensates at the nexus of cellular stress, protein aggregation disease and ageing. Nat Rev Mol Cell Biol 22, 196–213. Crossref, Medline, Google Scholar
• Alberti S, Saha S, Woodruff JB, Franzmann TM, Wang J, Hyman AA (2018). A user's guide for phase separation assays with purified proteins. J Mol Biol 430, 4806–4820. Crossref, Medline, Google Scholar
• Algar WR, Hildebrandt N, Vogel SS, Medintz IL (2019). FRET as a biomolecular research tool — understanding its potential while avoiding pitfalls. Nat Methods 16, 815–829. Crossref, Medline, Google Scholar
• Allan DB, Caswell T, Keim NC, Wel CM, Verweij RW (2021). soft-matter/trackpy: Trackpy v0.5.0. Version v0.5.0. Google Scholar
• Alshareedah I, Borcherds WM, Cohen SR, Singh A, Posey AE, Farag M, Bremer A, Strout GW, Tomares DT, Pappu RV (2024). Sequence-specific interactions determine viscoelasticity and ageing dynamics of protein condensates. Nat Phys 20, 1482–1491. Crossref, Medline, Google Scholar
• Alshareedah I, Kaur T, Banerjee PR (2021). Chapter Six - Methods for characterizing the material properties of biomolecular condensates. In: Liquid-Liquid Phase Coexistence and Membraneless Organelles, Vol. 646, ed. CD Keating, 143–183. Pennsylvania State University, University Park, PA, United States: Academic Press. doi: https://doi.org/10.1016/bs.mie.2020.06.009 Crossref, Google Scholar
• Alshareedah I, Moosa MM, Pham M, Potoyan DA, Banerjee PR (2021). Programmable viscoelasticity in protein-RNA condensates with disordered sticker-spacer polypeptides. Nat Commun 12, 6620. Crossref, Medline, Google Scholar
• Altmeyer M, Neelsen KJ, Teloni F, Pozdnyakova I, Pellegrino S, Grøfte M, Rask MBD, Streicher W, Jungmichel S, Nielsen ML (2015). Liquid demixing of intrinsically disordered proteins is seeded by poly(ADP-ribose). Nat Commun 6, 8088. Crossref, Medline, Google Scholar
• Antonacci G, Beck T, Bilenca A, Czarske J, Elsayad K, Guck J, Kim K, Krug B, Palombo F, Prevedel R (2020). Recent progress and current opinions in Brillouin microscopy for life science applications. Biophys Rev 12, 615–624. Crossref, Medline, Google Scholar
• Antonacci G, Braakman S (2016). Biomechanics of subcellular structures by non-invasive Brillouin microscopy. Sci Rep 6, 37217. Crossref, Medline, Google Scholar
• Antonacci G, Foreman MR, Paterson C, Török P (2013). Spectral broadening in Brillouin imaging. Appl Phys Lett 103, 22. Crossref, Google Scholar
• Antonacci G, Turris V, Rosa A, Ruocco G (2018). Background-deflection Brillouin microscopy reveals altered biomechanics of intracellular stress granules by ALS protein FUS. Commun Biol 1, 139. Crossref, Medline, Google Scholar
• Bailey M, Alunni-Cardinali M, Correa N, Caponi S, Holsgrove T, Barr H, Stone N, Winlove CP, Fioretto D, Palombo F (2020). Viscoelastic properties of biopolymer hydrogels determined by Brillouin spectroscopy: A probe of tissue micromechanics. Sci Adv 6, eabc1937. Crossref, Medline, Google Scholar
• Banani SF, L HO, Hyman AA, Rosen MK (2017). Biomolecular condensates: Organizers of cellular biochemistry. Nat Rev Mol Cell Biol 18, 285–298. Crossref, Medline, Google Scholar
• Barer R (1952). Interference microscopy and mass determination. Nature 169, 366–367. Crossref, Medline, Google Scholar
• Beck T (2023). DropletQPI. Version 0.1.0. Available from: https://github.com/GuckLab/DropletQPI. Google Scholar
• Benayad Z, Von Bülow S, Stelzl LS, Hummer G (2021). Simulation of FUS protein condensates with an adapted coarse-grained model. J Chem Theory Comput 17, 525–537. Crossref, Medline, Google Scholar
• Bevilacqua C, Prevedel R (2024). Full-field Brillouin microscopy based on an imaging Fourier transform spectrometer. arXiv 2409.02092. Google Scholar
• Boke E, Ruer M, Wühr M, Coughlin M, Lemaitre R, Gygi SP, Alberti S, Drechsel D, Hyman AA, Mitchison TJ (2016). Amyloid-like self-assembly of a cellular compartment. Cell 166, 637–650. Crossref, Medline, Google Scholar
• Bot A, Schram RPC, Wegdam GH (1995). Brillouin light scattering from a biopolymer gel: Hypersonic sound waves in gelatin. Colloid Polym Sci 273, 252–256. Crossref, Google Scholar
• Brangwynne CP, Eckmann CR, Courson DS, Rybarska A, Hoege C, Gharakhani J, Jülicher F, Hyman AA (2009). Germline P granules are liquid droplets that localize by controlled dissolution/condensation. Science 324, 1729–1732. Crossref, Medline, Google Scholar
• Brangwynne CP, Tompa P, Pappu RV (2015). Polymer physics of intracellular phase transitions. Nat Phys 11, 899–904. Crossref, Google Scholar
• Bremer A, Farag M, Borcherds WM, Peran I, Martin EW, Pappu RV, Mittag T (2022). Deciphering how naturally occurring sequence features impact the phase behaviours of disordered prion-like domains. Nat Chem 14, 196–207. Crossref, Medline, Google Scholar
• Brillouin L (1922). Diffusion de la lumière et des rayons X par un corps transparent homogène - Influence de lágitation thermique. Ann Phys 9, 88–122. Crossref, Google Scholar
• Burke KA, Janke AM, Rhine CL, Fawzi NL (2015). Residue-by-residue view of in vitro FUS granules that bind the C-terminal domain of RNA polymerase II. Mol Cell 60, 231–241. Crossref, Medline, Google Scholar
• Ceballos AV, McDonald CJ, Elbaum-Garfinkle S (2018). Chapter Two - Methods and strategies to quantify phase separation of disordered proteins. In: Intrinsically Disordered Proteins, Vol. 611, ed. E Rhoades, 31–50. University of Pennsylvania, Philadelphia, PA, United States: Academic Press. doi: https://doi.org/10.1016/bs.mie.2018.09.037 Crossref, Google Scholar
• Chappidi N, Quail T, Doll S, Vogel LT, Aleksandrov R, Felekyan S, Kühnemuth R, Stoynov S, Seidel CA, Brugu´es J (2024). PARP1-DNA co-condensation drives DNA repair site assembly to prevent disjunction of broken DNA ends. Cell 187, 945–961.e18. Crossref, Medline, Google Scholar
• Chauhan G, Bremer A, Dar F, Mittag T, Pappu RV (2024). Crowder titrations enable the quantification of driving forces for macromolecular phase separation. Biophys J 123, 1376–1392. doi: 10.1016/j.bpj.2023.09.006. Crossref, Medline, Google Scholar
• Chou HY, Aksimentiev A (2020). Single-protein collapse determines phase equilibria of a biological condensate. J Phys Chem Lett 11, 4923–4929. Crossref, Medline, Google Scholar
• Davies HG, Wilkins MHF (1952). Interference microscopy and mass determination. Nature 169, 541. Crossref, Medline, Google Scholar
• Dodson AE, Kennedy S (2020). Phase separation in germ cells and development. Dev Cell 55, 4–17. Crossref, Medline, Google Scholar
• Ebrahimi S, Sadeghi R (2015). Density, speed of sound, and viscosity of some binary and ternary aqueous polymer solutions at different temperatures. J Chem Eng Data 60, 3132–3147. Crossref, Google Scholar
• Elbaum-Garfinkle S, Kim Y, Szczepaniak K, Chen CCH, Eckmann CR, Myong S, Brangwynne CP (2015). The disordered P granule protein LAF-1 drives phase separation into droplets with tunable viscosity and dynamics. Proc Natl Acad Sci U S A 112, 7189–7194. Crossref, Medline, Google Scholar
• Farag M, Cohen SR, Borcherds WM, Bremer A, Mittag T, Pappu RV (2022). Condensates formed by prion-like low complexity domains have small-world network structures and interfaces defined by expanded conformations. Nat Commun 13, 7722. Crossref, Medline, Google Scholar
• Fisher RS, Elbaum-Garfinkle S (2020). Tunable multiphase dynamics of arginine and lysine liquid condensates. Nat Commun 11, 4628. Crossref, Medline, Google Scholar
• Garaizar A, Espinosa JR, Joseph JA, Krainer G, Shen Y, Knowles TP, Collepardo-Guevara R (2022). Aging can transform single-component protein condensates into multiphase architectures. Proc Natl Acad Sci U S A 119, e2119800119. Crossref, Medline, Google Scholar
• Gonzalez-Tello P, Camacho F, Blazquez G (1994). Density and viscosity of concentrated aqueous solutions of polyethylene glycol. J Chem Eng Data 39, 611–614. Crossref, Google Scholar
• Hosea T, Ng S, Oates C (1990). A Brillouin scattering study of the glass transition in sucrose. Food Hydrocoll 4, 137–147. Crossref, Google Scholar
• Illibauer J, Clodi-Seitz T, Zoufaly A, Aberle JH, Weninger WJ, Foedinger M, Elsayad K (2024). Diagnostic potential of blood plasma longitudinal viscosity measured using Brillouin light scattering. Proc Natl Acad Sci U S A 121, 2017. Crossref, Google Scholar
• Jadhav V, Kumbhar R, Tamhankar B, Shinde S, Kolekar S, Sabale S (2019). Volumetric and compressibility studies and phase equilibria of aqueous biphasic systems of alcohols using phase diagram. SN Appl Sci 1. Crossref, Google Scholar
• Jarry JP, Patterson GD (1981). Brillouin scattering from polyepoxide solutions and gels. Macromolecules 14, 1281–1284. Crossref, Google Scholar
• Jawerth L, Fischer-Friedrich E, Saha S, Wang J, Franzmann T, Zhang X, Sachweh J, Ruer M, Ijavi M, Saha S (2020). Protein condensates as aging Maxwell fluids. Science 370, 1317–1323. Crossref, Medline, Google Scholar
• Jawerth LM, Ijavi M, Ruer M, Saha S, Jahnel M, Hyman AA, Jülicher F, Fischer-Friedrich E (2018). Salt-dependent rheology and surface tension of protein condensates using optical traps. Phys Rev Lett 121, 258101. Crossref, Medline, Google Scholar
• Kabakova I, Zhang J, Xiang Y, Caponi S, Bilenca A, Guck J, Scarcelli G (2024). Brillouin microscopy. Nat Rev Methods Primers 4, 8. Crossref, Medline, Google Scholar
• Kim HJ, Kim NC, Wang YD, Scarborough EA, Moore J, Diaz Z, MacLea KS, Freibaum B, Li S, Molliex A (2013). Mutations in prion-like domains in hnRNPA2B1 and hnRNPA1 cause multisystem proteinopathy and ALS. Nature 495, 467–473. Crossref, Medline, Google Scholar
• Kim K, Guck J (2020). The relative densities of cytoplasm and nuclear compartments are robust against strong perturbation. Biophys J 119, 1946–1957. Crossref, Medline, Google Scholar
• Kim K, Kim KS, Park H, Ye JC, Park Y (2013). Real-time visualization of 3-D dynamic microscopic objects using optical diffraction tomography. Opt Express 21, 32269–32278. Crossref, Medline, Google Scholar
• Kim M, Besner S, Ramier A, Kwok SJJ, An J, Scarcelli G, Yun SH (2016). Shear Brillouin light scattering microscope. Optics Express 24, 319. Crossref, Medline, Google Scholar
• Klieber C, Hecksher T, Pezeril T, Torchinsky DH, Dyre JC, Nelson KA (2013). Mechanical spectra of glass-forming liquids. II. Gigahertz-frequency longitudinal and shear acoustic dynamics in glycerol and DC704 studied by time-domain Brillouin scattering. J Chem Phys 138, 12A544. Crossref, Medline, Google Scholar
• Kolsk´a Z, Valha P, Slepiˇcka P, Svorˇc´ık V (2019). Refractometric study of systems water-poly(ethylene glycol) for preparation and characterization of Au nanoparticles dispersion. Arab J Chem 12, 5019–5027. Crossref, Google Scholar
• Krainer G, Welsh TJ, Joseph JA, Espinosa JR, Wittmann S, Csill´ery E, Sridhar A, Toprakcioglu Z, Gudiˇskyte˙ G, Czekalska MA (2021). Reentrant liquid condensate phase of proteins is stabilized by hydrophobic and non-ionic interactions. Nat Commun 12, 1085. Crossref, Medline, Google Scholar
• Lafontaine DL, Riback JA, Bascetin R, Brangwynne CP (2021). The nucleolus as a multiphase liquid condensate. Nat Rev Mol Cell Biol 22, 165–182. Crossref, Medline, Google Scholar
• Lemaitre RP, Bogdanova A, Borgonovo B, Woodruff JB, Drechsel DN (2019). FlexiBAC: A versatile, open-source baculovirus vector system for protein expression, secretion, and proteolytic processing. BMC Biotechnol 19, 20. Crossref, Medline, Google Scholar
• Li S, Yoshizawa T, Yamazaki R, Fujiwara A, Kameda T, Kitahara R (2021). Pressure and temperature phase diagram for liquid–liquid phase separation of the RNA-binding protein fused in sarcoma. J Phys Chem B 125, 6821–6829. Crossref, Medline, Google Scholar
• Li YR, King OD, Shorter J, Gitler AD (2013). Stress granules as crucibles of ALS pathogenesis. J Cell Biol 201, 361–372. Crossref, Medline, Google Scholar
• Mandelstam L (1926). Light scattering by inhomogeneous media. Zh Russ Fiz-Khim Ova 58, 146. Google Scholar
• Martin EW, Holehouse AS, Peran I, Farag M, Incicco JJ, Bremer A, Grace CR, Soranno A, Pappu RV, Mittag T (2020). Valence and patterning of aromatic residues determine the phase behavior of prion-like domains. Science 367, 694–699. Crossref, Medline, Google Scholar
• Martin EW, Thomasen FE, Milkovic NM, Cuneo MJ, Grace CR, Nourse A, Lindorff-Larsen K, Mittag T (2021). Interplay of folded domains and the disordered low-complexity domain in mediating hnRNPA1 phase separation. Nucleic Acids Res 49, 2931–2945. Crossref, Medline, Google Scholar
• Mattarelli M, Vassalli M, Caponi S (2020). Relevant length scales in Brillouin imaging of biomaterials: The interplay between phonons propagation and light focalization. ACS Photonics 7, 2319–2328. Crossref, Google Scholar
• McCall PM, Kim K, Ruer-Gruss M, Peychl J, Guck J, Hyman AA, Brugues J (2023). Label-free composition determination for biomolecular condensates with an arbitrarily large number of components. bioRxiv. 2020.10.25.352823. Google Scholar
• Millero FJ, Curry RW, Drost-Hansen W (1969). Isothermal compressibility of water at various temperatures. J Chem Eng Data 14, 422–425. Crossref, Google Scholar
• Mitrea DM, Mittasch M, Gomes BF, Klein IA, Murcko MA (2022). Modulating biomolecular condensates: A novel approach to drug discovery. Nat Rev Drug Discov 21, 841–862. Crossref, Medline, Google Scholar
• Mittasch M, Gross P, Nestler M, Fritsch AW, Iserman C, Kar M, Munder M, Voigt A, Alberti S, Grill SW (2018). Noninvasive perturbations of intracellular flow reveal physical principles of cell organization. Nat Cell Biol 20, 344–351. Crossref, Medline, Google Scholar
• Möckel C, Beck T, Kaliman S, Abuhattum S, Kim K, Kolb J, Wehner D, Zaburdaev V, Guck J (2024). Estimation of the mass density of biological matter from refractive index measurements. Biophys Rep 4, 100156. Google Scholar
• Molliex A, Temirov J, Lee J, Coughlin M, Kanagaraj AP, Kim HJ, Mittag T, Taylor JP (2015). Phase separation by low complexity domains promotes stress granule assembly and drives pathological fibrillization. Cell 163, 123–133. Crossref, Medline, Google Scholar
• Moore WR (1961). Viscosity-temperature relationships for dilute solutions of high polymers. Nature 191, 1292–1293. Crossref, Google Scholar
• Müller P (2021). qpsphere. Version 0.5.8. Available from: https://pypi.org/project/qpsphere/0.5.8/. Google Scholar
• Müller P (2022). qpimage. Version 0.7.5. Available from: https://pypi.org/project/qpimage/0.7.5/. Google Scholar
• Müller P, Schürmann M, Girardo S, Cojoc G, Guck J (2018). Accurate evaluation of size and refractive index for spherical objects in quantitative phase imaging. Optics Express 26, 10729. Crossref, Medline, Google Scholar
• Murthy AC, Dignon GL, Kan Y, Zerze GH, Parekh SH, Mittal J, Fawzi NL (2019). Molecular interactions underlying liquid-liquid phase separation of the FUS low-complexity domain. Nat Struct Mol Biol 26, 637–648. Crossref, Medline, Google Scholar
• Murthy AC, Fawzi NL (2020). The (un)structural biology of biomolecular liquid-liquid phase separation using NMR spectroscopy. J Biol Chem 295, 2375–2384. Crossref, Medline, Google Scholar
• Nikolić M, Scarcelli G (2019). Long-term Brillouin imaging of live cells with reduced absorption-mediated damage at 660 nm wavelength. Biomed Opt Express 10, 156. Crossref, Google Scholar
• Patel A, Lee HO, Jawerth L, Maharana S, Jahnel M, Hein MY, Stoynov S, Mahamid J, Saha S, Franzmann TM (2015). A liquid-to-solid phase transition of the ALS protein FUS accelerated by disease mutation. Cell 162, 1066–1077. Crossref, Medline, Google Scholar
• Pochylski M, Aliotta F, Blaszczak Z, Gapin´ski J (2006). Structuring effects and hydration phenomena in poly(ethylene glycol)/water mixtures investigated by Brillouin scattering. J Phys Chem B 110, 20533–20539. Crossref, Medline, Google Scholar
• Posey AE, Bremer A, Erkamp NA, Pant A, Knowles TPJ, Dai Y, Mittag T, Pappu RV, Authors C (2024). Biomolecular condensates are characterized by interphase electric potentials. bioRxiv. 2024.07.02.601783. Google Scholar
• Prevedel R, Diz-Mun˜oz A, Ruocco G, Antonacci G (2019). Brillouin microscopy: An emerging tool for mechanobiology. Nat Methods 16, 969–977. Crossref, Medline, Google Scholar
• Quiroz FG, Fiore VF, Levorse J, Polak L, Wong E, Pasolli HA, Fuchs E (2020). Liquid-liquid phase separation drives skin barrier formation. Science 367, eaax9554. Crossref, Medline, Google Scholar
• Rad MA, Mahmodi H, Filipe EC, Cox TR, Kabakova I, Tipper JL (2022). Micromechanical characterisation of 3D bioprinted neural cell models using Brillouin microspectroscopy. Bioprinting 25, e00179. Crossref, Google Scholar
• Raimund Schlüßler and others, (2023). BMicro. Version 0.8.1. Available from: https://github.com/BrillouinMicroscopy/BMicro Google Scholar
• Santucci SC, Fioretto D, Comez L, Gessini A, Masciovecchio C (2006). Is there any fast sound in water? Phys Rev Lett 97, 225701. Crossref, Medline, Google Scholar
• Schlüßler R, Möllmert S, Abuhattum S, Cojoc G, Müller P, Kim K, Möckel C, Zimmermann C, Czarske J, Guck J (2018). Mechanical mapping of spinal cord growth and repair in living zebrafish larvae by Brillouin imaging. Biophys J 115, 911–923. Crossref, Medline, Google Scholar
• Schürmann M, Scholze J, Müller P, Guck J, Chan CJ (2016). Cell nuclei have lower refractive index and mass density than cytoplasm. J Biophotonics 9, 1068–1076. Crossref, Medline, Google Scholar
• Schlüßler R, Kim K, Nötzel M, Taubenberger A, Abuhattum S, Beck T, Müller P, Maharana S, Cojoc G, Girardo S (2022). Correlative all-optical quantification of mass density and mechanics of subcellular compartments with fluorescence specificity. Elife 11, e68490. Crossref, Medline, Google Scholar
• Shen Y, Chen A, Wang W, Shen Y, Ruggeri FS, Aime S, Wang Z, Qamar S, Espinosa JR, Garaizar A (2023). The liquid-to-solid transition of FUS is promoted by the condensate surface. Proc Natl Acad Sci U S A 120, e2301366120. Crossref, Medline, Google Scholar
• Singh Y, Singh R (1983). Compatibility studies on solutions of polymer blends by viscometric and ultrasonic techniques. Eur Polym J 19, 535–541. Crossref, Google Scholar
• Taylor NO, Wei MT, Stone HA, Brangwynne CP (2019). Quantifying dynamics in phase-separated condensates using fluorescence recovery after photobleaching. Biophys J 117, 1285-1300. Crossref, Medline, Google Scholar
• Tesei G, Schulze TK, Crehuet R, Lindorff-Larsen K (2021). Accurate model of liquid–liquid phase behavior of intrinsically disordered proteins from optimization of single-chain properties. Proc Natl Acad Sci U S A 118, e2111696118. Crossref, Medline, Google Scholar
• Wang X, Bowman J, Tu S, Nykypanchuk D, Kuksenok O, Minko S (2021). Polyethylene glycol Crowder's effect on enzyme aggregation, thermal stability, and residual catalytic activity. Langmuir 37, 8474–8485 Crossref, Medline, Google Scholar
• Woodruff JB, Hyman AA, Boke E (2018). Organization and function of non-dynamic biomolecular condensates. Trends Biochem Sci 43, 81–94. Crossref, Medline, Google Scholar
• Yoshizawa T, Ali R, Jiou J, Fung HYJ, Burke KA, Kim SJ, Lin Y, Peeples WB, Saltzberg D, Soniat M (2018). Nuclear import receptor inhibits phase separation of FUS through binding to multiple sites. Cell 173, 693–705.e22. Crossref, Medline, Google Scholar
• Zhao H, Brown PH, Schuck P (2011). On the distribution of protein refractive index increments. Biophys J 100, 2309–2317. Crossref, Medline, Google Scholar