RNA-mediated demixing transition of low-density condensates

Abstract

Biomolecular condensates play a key role in organizing cellular reactions by concentrating a specific set of biomolecules. However, whether condensate formation is accompanied by an increase in the total mass concentration within condensates or by the demixing of already highly crowded intracellular components remains elusive. Here, using refractive index imaging, we quantify the mass density of several condensates, including nucleoli, heterochromatin, nuclear speckles, and stress granules. Surprisingly, the latter two condensates exhibit low densities with a total mass concentration similar to the surrounding cyto- or nucleoplasm. Low-density condensates display higher permeability to cellular protein probes. We find that RNA tunes the biomolecular density of condensates. Moreover, intracellular structures such as mitochondria heavily influence the way phase separation proceeds, impacting the localization, morphology, and growth of condensates. These findings favor a model where segregative phase separation driven by non-associative or repulsive molecular interactions together with RNA-mediated selective association of specific components can give rise to low-density condensates in the crowded cellular environment.

Fig. 1. Quantitative refractive index imaging of intracellular condensates.

a (Top) Maximum intensity projection of the 3D RI tomogram of an intact live U2OS cell. (Bottom) Cross-sectional view of the same cell along a white dashed line. b–d (Top row) RI images of intact live cells. (Bottom row) Combined RI and immunofluorescence images of the same cell after fixation. The dashed boxes indicate areas for zoomed-in images on the right. b RI images of an intact live U2OS cell and the same cell after immunostaining with anti-NPM1. c RI images of an intact live U2OS cell and the same cell after double-immunostaining with anti-NPM1 and anti-HP1α. d RI images of an intact live NIH3T3 cell and the same cell after double-immunostaining with anti-NPM1 and anti-HP1α. e Quantification of refractive indices of intracellular regions in intact live cells. Data: center line = mean; box limits = [Q1, Q3]; whiskers = [Max, Min]; n = 73 (U2OS Nucleoli), 62 (U2OS Nucleoplasm), 79 (NIH3T3 Nucleoli), 98 (NIH3T3 Heterochromatin) and 42 (NIH3T3 Nucleoplasm). Distributions were statistically compared using the two-sided unpaired t-test. p = 1.83 · 10⁻³⁹; p = 9.63 · 10⁻⁸; p = 5.03 · 10⁻³²; ****p < 0.0001. Refractive index images are adjusted to the range of refractive index 1.34–1.37 for (a) and (d) and 1.337–1.37 for (b) and (c).

Fig. 2. Nuclear speckles and stress granules are low-density condensates.

a (Left) RI images of an intact live U2OS cell. (Right) Combined RI and immunofluorescence images of the same cell after fixation. Anti-SRSF2 is used to target nuclear speckles. b (Top) Combined RI and fluorescence images of a live U2OS cell expressing EGFP-SRSF2. (Bottom left) Zoomed-in images of a single nuclear speckle. (Bottom right) Intensity profile plot for normalized fluorescence intensities of EGFP-SRSF2 (green line) and RIs (gray line) along the white dashed line. c (Left) RI images of a live U2OS cell treated with 500 μM sodium arsenite. (Right) Combined RI and immunofluorescence images of the same cell after fixation. Anti-G3BP1 is used to target stress granules. d (Top) Combined RI and fluorescence images of a live U2OS cell expressing G3BP1-EGFP after the treatment of 500 μM sodium arsenite. (Bottom left) Zoomed-in images of a single stress granule. (Bottom right) Intensity profile plot for normalized fluorescence intensities of G3BP1-EGFP (green line) and RIs (gray line) along the white dashed line. e (Left) Time-lapse images of a fusion event between two stress granules. (Right) Intensity profile plots for normalized fluorescence intensities of G3BP1-EGFP (green line) and RIs (gray line) along white dashed lines. All refractive index images are adjusted to 1.337–1.37.

Fig. 3. Low-density condensates have highly porous and permeable internal organization.

a, b Representative images of U2OS cells stably expressing EGFP tagged NPM1, G3BP1, and SRSF2 and NIH3T3 cell stably expressing EGFP-HP1α. All cells are co-transduced with mCherry in (a) or tandem mCherry (mCh-mCh) in (b). G3BP1-EGFP expressing cells are treated with 500 μM sodium arsenite. For each condensate indicated with white dashed lines, intensity profile plots for normalized fluorescence of EGFP-tagged proteins and mCherry probes were shown in solid and dashed lines, respectively. c Partitioning coefficients of mCh and tandem mCh probes for each condensate type. Data: center line = mean; whiskers = [mean + std, mean - std]; n = 79, 86, 70, and 52 (mCh, respectively) 57, 57, 34, and 39 (tandem mCherry, respectively). Distributions were statistically compared using the two-sided unpaired t-test. p = 1.19 · 10⁻⁸; p = 1.67 · 10⁻⁸; p = 2.36 · 10⁻¹³; p = 1.56 · 10⁻⁴; ****p < 0.0001 and ***p < 0.001. d Schematic for the FLIP experiment. After acquiring 5 images for normalization, 100 cycles, each comprised of one bleaching event and 15 image acquisition, are followed. White dashed lines indicate the region where intensity profiles are measured for each image. e Time-lapse images of an NIH3T3 cell treated with 250 μM sodium arsenite during the FLIP experiment. The cell is expressing G3BP1-mCh, EGFP and miRFP. For every cycle, a bleaching event occurs at the white dashed circle for all three fluorescence channels. The white dashed line indicates the stress granule across which fluorescence changes are monitored over time. f Kymographs showing fluorescence changes in each channel, generated along the white dashed line in (e). g Temporal changes of fluorescence intensity of protein probes (left and middle) or G3BP1-mch (right) in and outside of the stress granule in (f). n = 3. The average lines were computed with a sliding window of 5 frames (15 s). Error bands denote standard deviations.

Fig. 4. RNA tunes the biomolecular density of intracellular condensates.

a Schematics of RNA depletion experiments. After activating NLS-RNase L by poly(I:C) treatment or treating cells with either DRB or ActD, a series of quantitative characterizations were conducted. b (Left) Representative images of U2OS cells stably expressing EGFP-SRSF2 and NLS-RNase L-P2A-BFP after poly(I:C) treatment. After live-cell images were collected, poly-dT RNA FISH images were acquired for the same cells. (Right) Distribution of poly-dT intensities within individual nuclear speckles. Only cells expressing both constructs were included. The normalized counts were averaged with a sliding window of 0.2 (a.u.). n = 146 (mock) and 119 (poly(I:C)) c (Left) Time-lapse images of a live U2OS cell stably expressing EGFP-SRSF2, mCherry, and NLS-RNase L after poly(I:C) treatment. Time 0 is defined when the nuclear speckle morphology begins to change. (Right) Normalized intensity profiles for mCherry across either a nuclear speckle or a nucleolus labeled with arrowheads in (c; Left). d Temporal changes of the partition coefficients of mCherry for nuclear speckles or nucleoli in U2OS cells after NLS-RNase L activation. Time is defined as in (c). The bold lines denote average values. n = 14 (nuclear speckles) and 6 (nucleoli). e Representative images of U2OS cells expressing EGFP-SRSF2, mCherry, and NLS-RNase L-P2A-BFP under mock treatment (Left) or poly(I:C) treatment (Right). After imaging live-cells, poly-dT RNA FISH images were acquired for the same cells. Normalized intensity profiles of EGFP-SRSF2 (green), RI (gray), mCh (red), and poly-dT (pink) along white dashed lines are shown. RI images are adjusted to the range of 1.34–1.37. f Averaged RI and fluorescence images of nuclear speckles for each RNA depletion condition. Individual images of nuclear speckles (same datasets as in g; Left) were center-aligned using fluorescence signals before averaging. ΔRI images were obtained by subtracting the minimum RI pixel value of each averaged RI image and adjusted to the range of 0–0.0025. g For individual nuclear speckles, either refractive indices (Left) or partition coefficients of mCherry (Right) were plotted against poly-dT intensities. Each experimental condition is color-coded: gray (DMSO or Mock), orange (NLS-RNase L), green (ActD), and blue (DRB). Data: center dot = mean; whiskers = [mean-std, mean+std].

Fig. 5. Intracellular structures influence the distribution, morphology and growth of condensates.

a Time-lapse images of stress granule formation within the G3BP1-EGFP expressing U2OS cell after the treatment of 500 μM sodium arsenite. b Zoomed-in images of the same cell in (a). c Normalized 2D autocorrelation of G3BP1-EGFP images and cross-correlation between G3BP1 and RI images. Plotted are the average (filled circles) and standard deviation (errorbars) of the correlation, at each binned radial shift (n = 11). d For zero radial shift, normalized cross-correlation values between G3BP1 and RI images are plotted in the presence and absence of stress granules. n = 55 (with stress granule) and 11 (no stress granule). Error bar was defined as standard deviation. e Combined RI and fluorescence images of a live G3BP1-EGFP expressing U2OS cell one hour after treatment of 500 μM sodium arsenite. Blue and red boxes represent two ROIs in (h). f Temporal changes of the size of individual stress granules. Granules with color codes are those found in the corresponding ROIs in I. Data from 12 stress granules were shown. g The size of individual stress granules at the last time point is plotted against the mean RI values of areas surrounding each granule. RIs within 5 pixels, 385 nm, from each granule are used to compute the mean and the standard deviation (errorbar) of the surroundings. Data from 12 stress granules were shown. h Time-lapse images for stress granule growth in two different ROIs. (Top) the blue box ROI in (e). (Bottom) the red box ROI in (e). All refractive index images are adjusted to 1.337–1.37.

Fig. 6. Demixing transition of low-density condensates in living cells.

a Schematics of phase diagrams for associative- (left) and segregative phase separation (right). In the simple ternary systems, different intermolecular interactions can give rise to immiscible phases of distinct compositions, which can be represented in terms of phase diagrams. The ternary mixtures with initial compositions labeled as black dots undergo phase separation along the tie line to form two immiscible phases. b (Left) Representative fluorescence and ΔRI images of the mixture of DNA and poly-l-lysine (PLL). The mixture contains 89.5 μM DNA phosphates (10% FAM labeled) and 0.068 w/v% PLL amines in 100 mM NaCl buffer. (Right) Representative fluorescence and ΔRI images of PEG-dextran aqueous two-phase system (ATPS). 5 wt% TMR-labeled dextran and 4 wt% PEG are used in the ATPS system. The inset image is adjusted to the range of ΔRI 0–0.075 for comparison with the DNA-PLL image. ΔRI images are obtained by subtracting the minimum RI pixel value of each raw RI image. c (Top) Combined RI and fluorescence images of stress granules in U2OS cells expressing G3BP1-EGFP. (Bottom) Combined ΔRI and fluorescence images of nuclear speckles in U2OS cells expressing EGFP-SRSF2. For comparison, the scale of RIs is adjusted to be identical to that in (b, right). d Schematic illustration of the demixing transition model of low-density condensates. RNA-mediated associative phase separation can lead to the assembly of condensates with reduced biomolecular density, compared to protein-only ones. An increase in the available internal space facilitated by RNA promotes the access of cellular proteins into condensates. Within living cells, diverse macromolecular complexes as well as organelles may segregatively work to further enhance phase separation of condensate components. Together, the content of living cells undergoes the demixing transition to form low-density condensates where a minimal density difference is observed across condensate boundary.