Membrane-less compartments, such as P bodies and stress granules, play major roles in cell biology and disease. The dynamics and physical properties of these compartments are very diverse, ranging from liquid-like to solid-like behavior. Importantly, changes in compartment dynamics have been linked to changes in compartment function or pathology. However, appropriate tools to investigate the physical properties of membrane-less compartments are still lacking.
The aliphatic alcohol hexanediol has been proposed as a tool to differentiate between liquid-like and solid-like assemblies in living cells. We show that in vitro reconstituted compartments formed by the RNA-binding protein FUS are rapidly dissolved by hexanediol. In contrast, solid-like fibers of FUS, which have been linked to the disease amyotrophic lateral sclerosis (ALS), are resistant to hexanediol treatment, supporting the idea that hexanediol can differentiate between liquid- and solid-like assemblies.
We further show that hexanediol can be used to examine the physical properties of membrane-less compartments in vivo. We find that hexanediol dissolves dynamic, liquid-like assemblies, such as P bodies, whereas solid-like assemblies, such as protein aggregates and cytoskeletal assemblies, are largely resistant to hexanediol. Finally, we report here that extended exposure of yeast and mammalian cells to hexanediol is cytotoxic and causes abnormal changes in cell morphology, which trigger the formation of aberrant assemblies.
We therefore urge care in the use of hexanediol, especially when cells are exposed to hexanediol for extended times. In summary, we propose that hexanediol is a powerful tool to probe the physical properties of membrane-less compartments, but only if adequate controls are included and precautions are taken.
(A) FUS droplets but not FUS fibers are hexanediol sensitive. Purified FUSG156E was phase separated at 25 µM and tested for sensitivity to 4% 1,6-hexanediol immediately, and after 24 h of shaking with subsequent fiber formation. An image of droplets treated with buffer only is shown as a control.
(B) Dissolution of yeast P bodies with 5% or 10% hexanediol. GFP-labelled Edc3 S.cerevisiae cells are shown in glucose-containing media, after 1 h stress in glucose-depleted media and after the addition of 5% or 10% hexanediol and 10 µg/ml digitonin.
(C) Hexanediol does not dissolve solid-like cellular assemblies. GFP-labelled Hsp104 and Pab1 S.cerevisiae cells are shown in glucose-containing media, after induction of protein aggregation in glucose-deprived media and after addition of 10% 1,6-hexanediol and 10 µg/ml digitonin.
(D) Cytoskeletal structures are not sensitive to hexanediol. GFP-labelled Tub1 (microtubules) and Sac6 (actin) S.cerevisiae cells in glucose-containing media and after the addition of 10% hexanediol and 10 µg/ml digitonin.
(E) Loss of membrane integrity causes the aggregation of P body proteins. Same yeast strain as in B. The cells were stressed in glucose-deprived medium for ~40 min. Afterwards, 10% hexanediol and 10 µg/ml digitonin were added together with trypanblue (1:50 diluted) as cell viability marker.
(F) Quantification of cells with P body foci and trypan blue signal shown in E. n = 90–150 cells from 3 independent experiments. Error bars are STDEV.
(G) Quantification of cell shrinkage shown in E after trypan blue signal increase. n = 22 cells from 3 independent experiments. The first value is set to 1. Error bars are STDEV.
(H) Hypertonic shock leads to reappearance of foci that are similar to RNP granules. GFP-labelled Edc3 and mCherry-labelled Pab1 cells are shown in glucose-containing media and after the addition of either 10% hexanediol with 10 µg/ml digitonin, hypertonic media with 10 µg/ml digitonin or hypotonic media including 10% hexanediol with 10 µg/ml digitonin. Please note that the reappearance of puncta is faster than in Panel E, which might be due to differences in the stress treatment protocol.
(I) Dissolution of mammalian stress granules with 2.5%, 3.5% or 5% hexanediol. G3BP2-GFP was expressed from a BAC construct in HeLa cells. Before the hexanediol treatment, the initially diffuse G3BP2 was aggregated through a 60 min pre-treatment with 1 mM sodium arsenate.
(J) Hexanediol treatment induces aberrant assemblies labelled by G3BP2. The same cell line as in I was observed before and after addition of 3.5% hexanediol to unstressed cells.
(K) Aberrant stress granule-like assemblies form due to osmotic compression and cell shrinkage. The same unstressed cell line as in I is shown during treatment in hypotonic medium with 3.5% hexanediol and in hypertonic medium. Please note that the cells treated with hypertonic media show aberrant stress granule like assemblies, but not the cells treated with hypotonic media including hexanediol.
(L) Blebbing of HeLa cells under hexanediol treatment is reversible. The DIC images of the same cell are shown in the presence of hexanediol and after washout of hexanediol. The scale bar is 5 µm in all images.
(M) Schematic of the effects in cells upon hexanediol treatment. The green box describes the first direct effect of hexanediol treatment on liquid-like assemblies. Only here conclusions about the material state of examined assemblies can be made. The red box shows cellular changes due to an impairment of membrane integrity which happens after extended hexanediol treatment.
Intracellular compartments that lack membranes have gained a lot of interest in recent years. Two examples of such membrane-less compartments are cytoplasmic P bodies and stress granules, which increase in size and number during stress conditions. Interestingly, in budding yeast the material properties of these two assemblies are very different, and this is also reflected by their different functions. P bodies are active degradation sites of mRNA and have more liquid-like properties that allow for a fast entry and exit of components from the compartment.
In contrast, yeast stress granules are mRNA storage sites, which show more solid-like properties. Understanding the link between the physical properties of membrane-less compartments and their function is of utmost importance, also because transitions into more solid-like states have been linked to age-related diseases. The problem with determining the material properties of cellular compartments is that liquid-like assemblies and solid-like aggregates are morphologically very similar and cannot easily be distinguished by fluorescence microscopy.
Current approaches to determine the material properties of assemblies in living cells are measuring the sphericity, analyzing fusion events of compartments or recording fluorescent recovery after photobleaching. However, these methods have limitations, because the compartments under investigation are often very small and their mobility in cells is quite high, making such analyses difficult. Therefore, additional ways of characterizing the material properties of assemblies are required. In this paper, we provide evidence that the aliphatic alcohol 1,6-hexanediol can be a useful tool to determine the material properties of membrane-less compartments in vitro as well as in vivo.
The aim of this work is to show that hexanediol can be used as a tool to discriminate between liquid-like and solid-like assemblies in vitro as well as in yeast and mammalian cells.
Results & Discussion
Hexanediol can differentiate between liquid- and solid like assemblies in vitro
Previous studies have used the aliphatic alcohol 1,6-hexanediol to test the material properties of cellular assemblies. Hexanediol was shown to perturb FG repeat interactions between nucleoporins in nuclear pores and interactions between RNA-binding proteins in RNA-protein granules (RNP). These studies suggested that 1,6-hexanediol interferes with weak hydrophobic protein-protein or protein-RNA interactions that are required for these dynamic, liquid-like assemblies to form.
A well characterized protein that forms liquid-like compartments is FUS. FUS is primarily nuclear and involved in transcription, DNA repair and RNA processing. The protein also relocates to stress granules upon environmental perturbations. In both conditions, FUS forms compartments that have liquid-like properties. It was further reported that liquid-like FUS compartments undergo a liquid-to-solid transition in patients afflicted with ALS and related diseases. This makes FUS an ideal candidate protein to test how 1,6-hexanediol affects two different material states formed by the same protein.
To test the utility of hexanediol, we reconstituted FUS compartments as reported previously using dextran to mimic cellular crowding conditions (Panel A, top left). The formed droplets dissolved quickly upon the addition of 4% hexanediol, in contrast to the buffer-only control (Panel A, bottom). Next, we tested if hexanediol also affects the solid-like state of FUS. Solid FUS fibers form in an in vitro aging assay after 24 h (Panel A, top left). The fibers were incubated with 4% hexanediol (Panel A, bottom right). However, no dissolution was observed (Panel A). We therefore conclude that hexanediol can be used to differentiate liquid-like and solid-like states of protein assemblies.
Liquid-like but not solid-like assemblies in living yeast cells are vulnerable to hexanediol
To evaluate whether hexanediol can differentiate between liquid- and solid-like assemblies in living yeast cells, we first determined the ideal concentration range for the use of hexanediol. We found that hexanediol concentrations of 10% quickly dissolved P bodies in budding yeast cells, whereas 5% hexanediol only dissolved P bodies after 20–30 min. These experiments were performed in the presence of 10 µg/ml digitonin, which makes yeast cells more permeable to hexanediol (Panel B).
Thus, we recommend using hexanediol at a concentration of 10% together with digitonin. Please note that in some cells assemblies reappear after 30–60 min. This will be addressed in the next section.
To determine whether hexanediol can be used to probe the material properties of cellular structures, we next tested several assemblies with more solid-like character for their sensitivity to hexanediol. We first looked at protein aggregates. Protein aggregates were induced through glucose starvation, and the chaperone Hsp104 was used as a marker. Prominent foci of Hsp104-positive aggregates were visible after 1 h. These Hsp104-positive protein aggregates were resistant to hexanediol (Panel C).
This is in agreement with the finding that yeast stress granules, which colocalize with Hsp104 and depend on Hsp104 for dissolution, are insensitive to hexanediol (Panel C). We next tested whether actin filaments and microtubules are affected by hexanediol treatment (Panel D). However, the hexanediol effects on the integrity of these cytoskeletal structures were mild and did not lead to a dissolution of actin filaments or microtubules. Thus, more solid-like protein assemblies such as protein aggregates and the cytoskeleton are not affected by hexanediol when used at a concentration of 10% in the presence of digitonin.
A previous study reported a partial dissolution of stress granules and cytoskeletal structures upon hexanediol treatment. However, under the conditions that we used, we did not find evidence for a hexanediol sensitivity of these structures. This may partially be due to the use of different growth and hexanediol treatment conditions. Another factor that can affect the outcome is the duration of the stress stimulus prior to the hexanediol treatment. This sensitivity to conditions (hexanediol concentration, stress period, growth conditions) should be taken into account when analyzing membrane-less compartments and assemblies with hexanediol.
We conclude that hexanediol can differentiate between liquid-like and more solid-like assemblies in yeast cells. However, the specific conditions under which the hexanediol experiment is performed are very important and will strongly impact the obtained results. We therefore recommend performing comparative experiments with assemblies for which the material properties are known.
Extended exposure of cells to hexanediol leads to loss of membrane integrity
When we performed extended time-lapse experiments in the presence of hexanediol, we have noticed that some yeast cells, which had initially lost P bodies, re-accumulated P body-like structures at a later time point (Panel E). This has been also observed by others. However, these structures also colocalized with SG markers (Panel H), something that is normally only observed in cells that are exposed to severe stress. To investigate the viability of these cells, we used the dye trypan blue.
This dye accumulates within cells that have lost their membrane integrity and can thus be used to differentiate between live and dead cells. Interestingly, the induction of late P body-like structures coincided with the influx of trypan blue into the cells (Panel E, bottom and Panel F). Loss of membrane integrity, as indicated by the accumulation of trypan blue within cells, also coincided with cell shrinkage (Panel G). This indicates that, upon prolonged exposure to hexanediol, yeast cells die. This results in the formation of aberrant structures that contain both P body and stress granule components.
We hypothesized that these assemblies form because the permeabilized cells experience altered osmotic conditions that lead to significant cell shrinkage and increased macromolecular crowding. To investigate this, we exposed yeast to a hypertonic shock. Indeed, under hypertonic conditions, we observed similar assemblies, which were positive for PB proteins (Panel H). Note that this hypertonic shock was applied in the absence of hexanediol. Importantly, hypotonic media delayed the formation of these aberrant assemblies in the presence of hexanediol (Panel H, compare top and bottom).
Thus, we conclude that extended hexanediol treatment can lead to loss of membrane integrity and osmotic compression that causes the formation of aberrant assemblies in yeast cells. For this reason, we recommend performing live cell time-lapse imaging to separate early effects of hexanediol from later aberrant effects due to osmotically induced cell shrinkage. Time-resolved imaging will allow monitoring of individual cells and will reveal sudden changes in cellular morphology, and the appearance of aberrant aggregates with time. We also propose including vital dyes in the experiment such as trypan blue. These dyes can be added together with hexanediol and this will aid in identifying cells with compromised membrane integrity.
Similar to yeast cells, liquid-like assemblies in mammalian cells are dissolved through hexanediol
Similar to yeast P bodies, mammalian stress granules dissolve in a hexanediol-dependent manner over a time course of 10 to 20 min. This is due to the fact that stress granules in mammalian cells are much more liquid-like than stress granules in yeast cells. To further investigate this, we tested the concentration dependence of the hexanediol effect.
The dissolution of mammalian stress granules could be observed with hexanediol concentrations of 3.5% and higher (Panel I). Some cells also re-accumulated smaller stress granule-like structures labeled by G3BP2 under hexanediol treatment, which were especially prominent in unstressed cells, but these structures dissolved again (Panel J). This is reminiscent of the situation that we observed upon prolonged exposure of yeast cells to hexanediol, where the cells lost membrane integrity with time and formed aberrant structures.
To test whether the appearance of these late structures is due to increased membrane permeability and cell shrinkage and whether osmotic effects are involved, we repeated the experiment in hypotonic media. The cells were first imaged in regular media (time point “before”) and subsequently the media was replaced with hypotonic media containing hexanediol (Panel K, top). Indeed, in hypotonic media, aberrant stress granule-like structures did not form in the presence of hexanediol, while in hypertonic media stress granule-like structures formed even without hexanediol addition (Panel K, bottom).
This suggests that, similar to the situation in yeast cells, the formation of aberrant assemblies under extended hexanediol treatment is caused by a loss of membrane function and a potential loss of water from the cells that leads to an increase in macromolecular crowding. Thus, we conclude that hexanediol can be used to determine the material state of cellular assemblies in mammalian cells, but the extended exposure to hexanediol should be avoided.
1,6-hexanediol induces blebbing in HeLa cells, but it does not cause cell death
In contrast to yeast cells, HeLa cells are much more vulnerable to environmental changes. Exposure to adverse conditions often leads to the formation of membrane blebs, which can be the first signs of autophagy. Indeed, we noticed that the addition of hexanediol to HeLa cells caused blebbing.
However, cells that had been incubated with hexandiol for as long as 50 min could recover from blebbing and spread again when hexanediol was washed out and replaced with normal growth media (Panel L). Thus, incubation with hexanediol does not cause the immediate death of cultured mammalian cells such as HeLa cells, when the cells are incubated with hexanediol for a limited amount of time.
In this study, we report on the use of hexanediol as a probe to discriminate between liquid-like and solid-like membrane-less assemblies. Our findings shows that hexanediol can dissolve liquid-like but not solid-like assemblies in vitro and in vivo. If several precautions, e.g., optimizing the duration of hexanediol treatment, are taken, hexanediol can be a useful tool for the study of membrane-less compartments (Panel M). We recommend recording a time series with several independent replicates and inclusion of a positive and a negative control to ensure that hexanediol is working properly, as well a vital dye to monitor cell viability.
We do not yet understand, how hexanediol affects liquid-like assemblies. However, we hypothesize that hexanediol inhibits the formation of a liquid protein phase by interfering with the weak interactions between proteins or proteins and RNAs, as it was proposed for the nuclear pore complex. Evidence for this also comes from protein crystallization studies that often observe an intermediate liquid protein phase, and where hexanediol is used as an additive to favor the formation of the crystal.
In order to fully discriminate the effect of hexanediol on liquid-like assemblies, we need a better molecular understanding of the mode of action of hexanediol. The most appropriate system to test this will be structural studies with in vitro reconstituted liquid and solid assemblies.
Yeast genetic techniques, strains, and media
The media used were standard synthetic media or rich media containing 2% D-glucose. The yeast strain backgrounds were W303 ADE+ (leu2-3,112; his3-11,-15; trp1-1; ura3-1; can1-100; [psi-]; [PIN+]) or BY4741 (his3Δ1; leu2Δ0; met15Δ0; ura3Δ0; [psi-]; [PIN+]) with a GFP or mCherry tag fused to the indicated endogenous protein. Cells were cultured at 25˚C and stressed for 45–60 min in glucose-deficient medium were indicated.
Cell culture of HeLa BAC cell lines
HeLa cells were cultured in DMEM supplemented with 10% FBS and penicillin-streptomycin (Gibco Life Technologies, USA). Cells were maintained at 37°C in a 5% CO2 incubator. HeLa Kyoto cells containing G3BP2-GFP BAC constructs were used to visualize stress granules. If stated, the cells were stressed for 1 h with 1 mM sodium arsenate.
Protein Purification & In vitro reconstitution
The protein FUSG156E was purified as described under in the protein purification facility at the MPI-CBG, Dresden. The reconstitution of FUSG156E assemblies was done in 50 mM Tris/HCl pH 7.2 buffer with 500 mM KCl, 1 mM DTT, 2.5% glycerol and 10% dextran in low-binding tubes.
Wide-field fluorescence microscopy
Yeast cells were grown in cultures of 50–100 ml at 25°C to an OD600 not higher than 0.5. The yeast cells were then immobilized on concanavalin A-coated four chamber dishes (MatTek, USA), allowing us to image four different conditions in the same experiment. The HeLa cells were plated on 35 mm coverslip bottomed dishes (MatTek, USA). Microscopy was performed using a Deltavision microscope system with softWorx 4.1.2 software (Applied Precision).
The system was based on an Olympus IX71 microscope, which was used with a UPlanSApo 100×1.4 numerical (NA) oil objective for yeast and a Plan Apo 60×1.42 NA oil immersion objective for mammalian cells. The images were collected with a Cool SnapHQ camera (Photometrics) and a pixel size of 0.13 µm. When indicated 5%–10% of 1,6-hexanediol (Merck, Germany) solution for yeast and 2.5%–5% 1,6-hexanediol solution for mammalian cells was added to perturb RNP granule integrity.
If mentioned, yeast cells were treated with 10 µg/ml digitonin. Trypan blue (Sigma) was added to cells together with hexanediol. For yeast cells, the media was diluted 5 times and for mammalian cells 4 times with sterile water to reduce the osmolarity of culture media (hypotonic media). To obtain hypertonic media, the culture media was supplemented with 0.8 M D-sorbitol for yeast and 0.4 M NaCl for mammalian cells.
All images were deconvolved using standard softWorx deconvolution algorithms (enhanced ratio, high noise filtering). Shown images are maximum intensity projections of 8–15 individual images. Figures show representative cells. To count Edc3 foci as well as trypan blue-positive cells, the Fiji Cell Counter Plugin was used. For the quantification of yeast cell size, individual cells were thresholded and converted to binary before using the Fiji Particle Analyzer.
Sonja Kroschwald was supported by a springboard-to-postdoc fellowship by the Dresden International PhD program. Shovamayee Maharana was supported by a postdoctoral fellowship by the Alexander von Humboldt Foundation. We also acknowledge funding by the Max Planck Society.
We thank the MPI-CBG protein purification facility for supplying us with purified FUSG156E. We thank Titus Franzmann, Christiane Iserman and Avinash Patel for commenting on the manuscript.
Conflict of interest
The authors declare no conflicts of interest.
Ethics committee approval was not required because only cell lines were used.
No fraudulence is committed in performing these experiments or during processing of the data. We understand that in the case of fraudulence, the study can be retracted by ScienceMatters.