Cells recognize osmotic stress through liquid-liquid phase separation lubricated with poly(ADP-ribose)

Abstract

Cells are under threat of osmotic perturbation; cell volume maintenance is critical in cerebral edema, inflammation and aging, in which prominent changes in intracellular or extracellular osmolality emerge. After osmotic stress-enforced cell swelling or shrinkage, the cells regulate intracellular osmolality to recover their volume. However, the mechanisms recognizing osmotic stress remain obscured. We previously clarified that apoptosis signal-regulating kinase 3 (ASK3) bidirectionally responds to osmotic stress and regulates cell volume recovery. Here, we show that macromolecular crowding induces liquid-demixing condensates of ASK3 under hyperosmotic stress, which transduce osmosensing signal into ASK3 inactivation. A genome-wide small interfering RNA (siRNA) screen identifies an ASK3 inactivation regulator, nicotinamide phosphoribosyltransferase (NAMPT), related to poly(ADP-ribose) signaling. Furthermore, we clarify that poly(ADP-ribose) keeps ASK3 condensates in the liquid phase and enables ASK3 to become inactivated under hyperosmotic stress. Our findings demonstrate that cells rationally incorporate physicochemical phase separation into their osmosensing systems.

Fig. 1. ASK3 forms macromolecular crowding-driven condensates under hyperosmotic stress.

a, b Subcellular localization of ASK3 5 min after osmotic stress in Venus-ASK3-stably expressing HEK293A (Venus-ASK3-HEK293A) cells. Hypoosmotic stress: ultrapure water-diluted medium, hyperosmotic stress: mannitol-supplemented medium, DIC: differential interference contrast, white bar: 20 μm, red bar: 2.5 μm. Data: center line = median; box limits = [Q₁, Q₃]; whiskers = [max(minimum value, Q₁ − 1.5 × IQR), min(maximum value, Q₃ + 1.5 × IQR)], where Q₁, Q₃ and IQR are the first quartile, the third quartile and the interquartile range, respectively; n = 12 (700 mOsm), 14 (600 mOsm), 15 (400 mOsm), 16 (150 mOsm, 225 mOsm, 300 mOsm and 500 mOsm) cells pooled from four independent experiments. Note that the signal intensity of DIC cannot be compared among the images. c Schematic diagram of a computational model for protein diffusion and clustering in a two-dimensional grid space. Red squares: ASK3 units, black squares: obstacles, pale yellow arrows: potential movements, blue squares: surface positions of the clusters. See Supplementary Note for the full description of the model. d, e A computational simulation for the relationship between the grid space and the number/size of ASK3 clusters. Results after 5 × 10⁶ steps in the rejection kinetic Monte Carlo (rKMC) method at each grid space are presented. Red shading: the assumed range corresponding to hyperosmotic stress (details in Supplementary Discussion), dashed line: the minimum of ASK3 clusters definition. Data: mean ± SEM, n = 18 simulations.

Fig. 2. ASK3 condensates are liquid-demixing condensates induced by liquid–liquid phase separation.

a Transmission electron microscopy (TEM) analysis with immunogold labeling for ASK3. Venus-ASK3-HEK293A cells were sampled after hyperosmotic stress (800 mOsm, 3 h). White bar: 250 nm, red bar: 31.25 nm. A representative image from ten condensates across five micrographs is presented. b Dynamics and fusion of ASK3 condensates in Venus-ASK3-HEK293A cells. Hyperosmotic stress: 500 mOsm, white bar: 2 μm. A representative image set from four independent experiments is presented. c, d A computational prediction for the number/size of ASK3 clusters after grid space expansion. Simulation results during 35 × 10⁶ steps in rKMC method at the 120 × 120 grid space after the initial iteration with 5 × 10⁶ steps at the 55 × 55 grid space are presented as the unit of time t per 1 × 10⁵ steps. Red dashed line: the mean of initial values at the grid space expansion, black dashed line: the minimum of ASK3 clusters definition. Data: mean ± SEM, n = 12 simulations. e, f Reversibility of ASK3 condensates in EGFP-FLAG-ASK3-transfected HEK293A cells. After hyperosmotic stress (600 mOsm, 20 min), the extracellular osmolality was set back to the isoosmotic condition. White bar: 20 μm, red bar: 2.5 μm. Red dashed line: the mean of initial values after setting back to isoosmotic condition, black dashed line: timepoint at the osmotic stress treatment. Data: mean ± SEM, n = 8 cells pooled from three independent experiments. g, h Fluorescence recovery after photobleaching (FRAP) assay for ASK3 condensates in ASK3-tdTomato-transfected HEK293A cells. Prior to the assay, cells were exposed to hyperosmotic stress (600 mOsm, 30 min). White bar: 20 μm, red/blue bar: 2.5 μm. Data: mean ± SEM, n = 15 cells pooled from five independent experiments. i, j ASK3 condensation in vitro. Control: 150 mM NaCl, 20 mM Tris (pH 7.5), 1 mM dithiothreitol (DTT), 15-min incubation on ice. Ficoll and polyethylene glycol (PEG): a crowding reagent, white bar: 5 μm. Data: mean ± SEM, n = 3 independent experiments.

Fig. 3. ASK3 condensation is required for ASK3 inactivation under hyperosmotic stress.

a Schematic representation of ASK3 deletion mutants and fragments. The numbers indicate the amino acid (a.a.) positions in wild-type (WT). Black rectangle: kinase domain (652–908 a.a.), green rectangle: C-terminus coiled-coil domain (CCC: 1179–1225 a.a.), orange rectangle: C-terminus low-complexity region (CLCR: 1280–1293 a.a.). b, c Subcellular localization of ASK3 deletion mutants and fragments in HEK293A cells. Hyperosmotic stress: 600 mOsm, 10 min. DIC: differential interference contrast, white bar: 20 μm, red bar: 2.5 μm. A representative image set from four independent experiments is presented. Note that the signal intensity of DIC cannot be compared among the images. d Requirement of CCC and CLCR for ASK3 inactivation in HEK293A cells. Hyperosmolality (−): 300 mOsm; (+): 500 mOsm; 10 min. IB: immunoblotting, p-ASK_endo: phosphorylation bands of endogenously expressed ASK. ^†Nonspecific bands. A representative image set from four independent experiments is presented (quantification: Supplementary Fig. 5a). e Relationship between ANKRD52 and ASK3 condensates in HEK293A cells. Magenta: ASK3-tdTomato, green: ANKRD52-Venus, hyperosmotic stress: 500 mOsm, white bar: 20 μm, yellow bar: 2.5 μm. A representative image set from five independent experiments is presented.

Fig. 4. The NAD salvage pathway negatively regulates ASK3 activity under hyperosmotic stress.

a Distribution of gene candidates to regulate ASK3 inactivation in the previous primary screen. A sample with a higher B-score corresponds to a higher potential candidate. PPP6C: the catalytic subunit of PP6, an ASK3 phosphatase. b Diagram of the mammalian nicotinamide adenine dinucleotide (NAD) salvage pathway. Rectangle: NAD-related molecule, arrow: reaction, ellipse: enzyme, NAM: nicotinamide, NMN: nicotinamide mononucleotide, FK866: a NAMPT enzymatic inhibitor. c Effects of NAMPT depletion on ASK3 activity under hyperosmotic stress in FLAG-ASK3-stably expressing HEK293A (FLAG-ASK3-HEK293A) cells. d Effects of NAMPT depletion on endogenous ASK3 and SPAK/OSR1 activities under hyperosmotic stress in HEK293A cells. ^†Nonspecific bands. e Effects of NAMPT overexpression on ASK3 activity under hyperosmotic stress in HEK293A cells. WT: wild-type, S199D: homodimer-insufficient mutant, S200D: homodimer-null mutant. f Effects of FK866 and/or NMN pretreatment on ASK3 activity under hyperosmotic stress in FLAG-ASK3-HEK293A cells. FK866 (−): dimethyl sulfoxide (DMSO), solvent for FK866; FK866 (+): 10 nM FK866; NMN (−): ultrapure water, solvent for NMN; NMN (+): 1 mM NMN; 3 h pretreatment. g, h Effects of FK866 or NMN pretreatment on the interaction between ASK3 and PP6 under hyperosmotic stress in HEK293A cells. FK866 (−): DMSO; FK866 (+): 10 nM FK866; NMN (−): ultrapure water; NMN (+): 1 mM NMN; 24 h pretreatment. ^‡Remnant bands from prior detection of GFP. c–h Hyperosmolality (−): 300 mOsm; (+): 425 mOsm (with the exception in g and h, 500 mOsm); (++): 500 mOsm; 10 min. IB: immunoblotting, IP: immunoprecipitation. A representative image set from five (c, f) or four (d, e, g, h) independent experiments is presented (quantification: Supplementary Fig. 5b–g). Note that superfluous lanes were digitally eliminated from blot images in d and f as indicated by vertical black lines.

Fig. 5. Poly(ADP-ribose) keeps ASK3 condensates in the liquid phase for ASK3 inactivation.

a Diagram of the poly(ADP-ribosyl)ation (PARsylation) dynamics. Rectangle: NAD-related molecule, arrow: reaction, ellipse: enzyme, ADPr: ADP-ribose. b Effects of NAMPT overexpression or inhibition on the amount of PAR in HEK293A cells. FK866 (−): dimethyl sulfoxide (DMSO), solvent for FK866; (+): 10 nM FK866; 18–24 h pretreatment. NAMPT_exo: exogenously expressed NAMPT, NAMPT_endo: endogenously expressed NAMPT. c Effects of PARG overexpression on ASK3 activity under hyperosmotic stress in HEK293A cells. WT: wild-type, E673A/E674A: glycohydrolase-inactive mutant. Hyperosmolality (−): 300 mOsm; (+): 425 mOsm; 10 min. b, c IB: immunoblotting. A representative image set from three (b) or four (c) independent experiments is presented (quantification: Supplementary Fig. 5h, i). d, e Effects of PAR depletion on ASK3 condensates in ASK3-tdTomato-transfected HEK293A cells. Hyperosmotic stress: 600 mOsm, 10 min. White bar: 20 μm, red bar: 2.5 μm. A representative image set from four independent experiments is presented. f, g Effects of PAR depletion on the FRAP of ASK3 condensates in ASK3-tdTomato-transfected HEK293A cells. Prior to the assay, cells were exposed to hyperosmotic stress (600 mOsm, 30 min). Data (f): mean ± SEM; data (g): center line = median; box limits = [Q₁, Q₃]; whiskers = [max(minimum value, Q₁ − 1.5 × IQR), min(maximum value, Q₃ + 1.5 × IQR)], where Q₁, Q₃ and IQR are the first quartile, the third quartile and the interquartile range, respectively; n = 8 (FK866), 9 (DMSO), 14 (Control), 15 (+PARG WT and EA) cells pooled from three (top panels) or five (bottom panels) independent experiments. *P < 0.05, ***P < 0.001, n.s. (not significant) according to two-sided Welch’s t-tests (with the Bonferroni correction in the bottom panel). d–g DMSO: solvent for FK866; FK866: 10 nM FK866; 18–24 h pretreatment. Control: empty vector, PARG WT: wild-type PARG-Venus, PARG EA: E673A/E674A mutant PARG-Venus. h, i Effects of PAR on solid-like ASK3 condensation in vitro. Control: 150 mM NaCl, 20 mM Tris (pH 7.5), 1 mM DTT, 20% PEG, 15-min incubation on ice. White bar: 5 μm. Data: mean ± SEM, n = 4 independent experiments. *P < 0.05, **P < 0.01, n.s. (not significant) according to two-sided Dunnett’s test.

Fig. 6. Interaction between ASK3 and PAR is required for the liquidity of ASK3 condensates and ASK3 inactivation under hyperosmotic stress.

a Candidate PAR-binding motif (PBM) in ASK3. Schematic representation of ASK3 is the same as that in Fig. 3a. The numbers indicate the amino acid (a.a.) positions in wild-type (WT). b Subcellular localization of ASK3 PBM candidate mutants in HEK293A cells. Hyperosmotic stress: 500 mOsm, 10 min. White bar: 20 μm, red bar: 2.5 μm. A representative image set from three independent experiments is presented. c Inactivation of ASK3 PBM candidate mutants under hyperosmotic stress in HEK293A cells. p-ASK_endo: phosphorylation bands of endogenously expressed ASK. d Ability of the ASK3 PBM4 mutant to interact with PAR under hyperosmotic stress in HEK293A cells. FK866 (−): dimethyl sulfoxide (DMSO), solvent for FK866; (+): 10 nM FK866; 12 h pretreatment. c, d Hyperosmolality (−): 300 mOsm; (+): 500 mOsm; 10 min. IB: immunoblotting, IP: immunoprecipitation. A representative image set from six (c) or four (d) independent experiments is presented (quantification: Supplementary Fig. 5j, k). e, f FRAP assay for condensates of the ASK3 PBM4 mutant in HEK293A cells. Prior to the assay, cells transfected with each ASK3-tdTomato were exposed to hyperosmotic stress (600 mOsm, 30 min). Data (e): mean ± SEM; data (f): center line = median; box limits = [Q₁, Q₃]; whiskers = [max(minimum value, Q₁ − 1.5 × IQR), min(maximum value, Q₃ + 1.5 × IQR)], where Q₁, Q₃ and IQR are the first quartile, the third quartile and the interquartile range, respectively; n = 9 (CT), 11 (WT), 12 (PBM4) cells pooled from four independent experiments. ***P < 0.001 according to two-sided Welch’s t-tests with the Bonferroni correction. g, h Effects of PAR on the solid-like condensates of the ASK3 PBM4 mutant in vitro. Control: 150 mM NaCl, 20 mM Tris (pH 7.5), 1 mM DTT, 20% PEG, 15-min incubation on ice. PAR: 2.5 μM, white bar: 5 μm. Data: mean ± SEM, n = 4 independent experiments. *P < 0.05, n.s. (not significant) according to two-sided Student’s t-test with the Bonferroni correction.

Fig. 7. Schematic summary of the main findings and molecular mechanism models.

a Schematic representation of the main findings in this study. Under hyperosmotic stress, cells sense cell shrinkage immediately and induce appropriate responses to maintain homeostasis. By leveraging the “lens” of phase separation, we unveiled that cells rationally incorporate macromolecular crowding-driven phase separation into their osmosensing systems. In addition to the sensing machinery for osmotic stress, ASK3 condensates function as the specific accelerator of ASK3 inactivation under hyperosmotic stress (at least). Unlike the roles in condensates of most RNA-binding proteins, poly(ADP-ribose) (PAR) endows ASK3 condensates with the liquid property and enables the PP6-mediated inactivation of ASK3. b, c A current schematic model for the molecular mechanism of ASK3 condensates under hyperosmotic stress. Under the excluded volume effect, the heterogenous multivalent modular interactions via C-terminus coiled-coil (CCC) domain and low-complexity region (CLCR) would drive the condensate formation of ASK3 (indicated as red bars). At the same time, the flexible and bulky PAR would also multivalently interact with ASK3 via arginine residues in the PAR-binding motif (PBM4) of ASK3 and indirectly coordinate the multivalent modular interactions between ASK3 molecules, which would control the liquidity of ASK3 condensates (indicated as blue bars in b). When the multivalent interactions between ASK3 and PAR are absent, the modular interactions between ASK3 molecules would be strong enough to drive the maturation of ASK3 condensates into the solid phase (c).