A modular tool to query and inducibly disrupt biomolecular condensates

Abstract

Dynamic membraneless compartments formed by protein condensates have multifunctional roles in cellular biology. Tools that inducibly trigger condensate formation have been useful for exploring their cellular function, however, there are few tools that provide inducible control over condensate disruption. To address this need we developed DisCo (Disassembly of Condensates), which relies on the use of chemical dimerizers to inducibly recruit a ligand to the condensate-forming protein, triggering condensate dissociation. We demonstrate use of DisCo to disrupt condensates of FUS, associated with amyotrophic lateral sclerosis, and to prevent formation of polyglutamine-containing huntingtin condensates, associated with Huntington's disease. In addition, we combined DisCo with a tool to induce condensates with light, CRY2olig, achieving bidirectional control of condensate formation and disassembly using orthogonal inputs of light and rapamycin. Our results demonstrate a method to manipulate condensate states that will have broad utility, enabling better understanding of the biological role of condensates in health and disease.

Fig. 1. DisCo can be used to disrupt and prevent formation of FUS condensates.

a General overview of DisCo approach with liquid-like condensates. A ligand (C-BLOCK) is recruited by chemical-induced dimerization (CID) to bind to a “hook” domain on a scaffold that forms the condensate. Binding results in destabilization of the weak interactions that hold together the condensate and condensate dissolution. b Representative images of HEK293T cells transfected with EGFP-FRB-FUS-FRB, along with the C-BLOCK mCh-FKBP. Addition of 333 nM rapamycin induced fast disruption of FUS condensates. Graphs show the mean fluorescent intensities at the indicated regions in timelapse images for FUS in condensate (green, top right), FUS in dilute phase (blue, bottom right) and mCh-FKBP (red, bottom right). Scale bar, 10 µm. Arb. units, arbitrary units. Similar results were obtained in 5 independent repeats. c Recruitment of C-BLOCK to different orientations within FUS condensates does not affect the % disruption efficiency, defined as the % of initial condensate that is disrupted upon rapamycin addition. HEK293T cells expressing indicated constructs were monitored for ability of C-BLOCK recruitment to disrupt condensate states. Data shows average and error (s.d., n = 30 for EGFP-FRB-FUS-FRB, EGFP-FUS-FRB, and EGFP-FRB-FUS, and n = 12 for EGFP-DHFR-FUS). ns, not significant, Kolmogorov–Smirnov (two-tailed) test, p = 0.071(FRB-FUS-FRB vs. FRB-FUS) and p = 0.134 (FRB-FUS-FRB vs. FUS-FRB). d Quantification of change in FUS dilute phase signal after rapamycin addition. HEK293T cells expressing EGFP-FRB-FUS-FRB and mCh-FKBP were treated with 333 nM rapamycin as in (b). After addition, an increase in EGFP signal outside of the condensate is observed. Graphs show the normalized mean fluorescent intensity of EGFP in the dilute phase (cytosol) at three time points after rapamycin addition. Graph shows average and error (s.e.m) of three measurements for each cell, n = 20 cells from three separate experiments. e HEK293T cells expressing EGFP-FRB-FUS-FRB and indicated C-BLOCKs were treated with rapamycin as in Fig. 1b. Condensates were disrupted upon recruitment of FKBP, but not mCh-FKBP-FKBP. Representative images from a single experiment are shown; experiments were repeated twice with similar results. Scale bars, 10 µm. f For cells expressing EGFP-FRB-FUS-FRB and mCh-FKBP the % disruption efficiency was plotted with respect to a value representing the ratio of GFP signal (mean intensity) within condensates to signal in the dilute phase (Mean I_dense/Mean I_dilute). A correlation was observed between the degree of condensate disruption, and the Mean I_dense/Mean I_dilute value. g Disruption kinetics of a representative cell expressing EGFP-FRB-FUS-FRB and mCh-FKBP treated with rapamycin (333 nM). The amount of FUS protein within condensates as a function of time was quantified, then fit to a sigmoid function to obtain the characteristic time (tau). Arb. units, arbitrary units. h Characteristic time (tau) of DisCo-mediated FUS disruption for indicated constructs. Mutant FUS constructs contained the configuration EGFP-FRB-FUS-FRB. Data shows average and error (s.d., n = 10 cells examined over three independent experiments). ns, not significant (p = 0.02), non-parametric one-way ANOVA. i Fluorescence recovery after photobleaching (FRAP) analysis of EGFP-FRB-FUS-FRB condensates. Representative images of a photobleached condensate are shown at right. Graph shows average and error (s.e.m., n = 11 cells examined over three independent experiments). Scale bars, 10 µm. j Preemptive recruitment of C-BLOCK to FUS blocks FUS condensate formation. Representative images of HEK293T cells expressing EGFP-FRB-FUS-FRB and indicated C-BLOCK or mCherry control are shown. Cells were treated with 500 nM AP21967 (+Rapalog) for 18 h, added 4 h after transfection, and imaged at 22 h post-transfection. The experiment was repeated 3 times with similar results. Scale bars, 10 µm. k Quantification of FUS signal in condensates 22 h post-transfection, with or without prior C-BLOCK recruitment. HEK293T cells expressing EGFP-FRB-FUS-FRB and mCh-FKBP (or mCh as control) were incubated with 500 nM AP21967 (+rapalog) prior to protein expression (as in 1j). The condensate intensity ratio was determined by measuring the integrated intensity within all condensates in a maximum projection image normalized to the total integrated intensity for all the cells in the maximum projection image. Data shows average and error (s.e.m., n = 3 regions with cells examined from three independent experiments).

Fig. 2. Use of DisCo with a photocleavable dimerizer allows reversible control of FUS condensate formation and disruption.

a Schematic showing use of the photocleavable heterodimerizer zapalog to reversibly control FUS condensate disruption. b Representative cell expressing EGFP-ecDHFR-FUS and mCh-FKBP incubated with 500 nM zapalog. Prior to illumination, FUS condensates are prevented from formation by C-BLOCK binding. Approximately 2–3 min after illumination with 405 nm light (50 ms pulse every 30 s for 5 min), FUS condensates can clearly be observed. After light treatment, incubation with zapalog in the dark for 34 min allows binding of uncleaved zapalog from the media and restores condensate blocking effect. A second round of 405 nm illumination induced further FUS condensate formation. The experiment was repeated an additional time with similar results. Scale bar, 10 µm. c Quantification of EGFP-ecDHFR-FUS condensate formation for experiment in (b), with data shown from a single experiment.

Fig. 3. HTT exon 1 condensates can be prevented from formation using DisCo but existing condensates are not disrupted.

a Representative images (left) and quantification of % disruption efficiency as a function of the initial EGFP mean signal in the dense phase normalized to the EGFP mean signal at the dilute phase ((Mean I_dense)/(Mean I_dilute)) (right) of HEK293T cells coexpressing HTTQ72-FRB-EGFP and mCh(K70N)-FKBP as C-BLOCK, exposed to 333 nM rapamycin. The experiment was repeated 3 times with similar results. Scale bars, 10 µm. b FRAP experiments showing Q72HTT-FRB-EGFP condensates do not recover from photobleaching within 600 s. Data shows average and error (s.e.m., n = 8). Representative images of a photobleached condensate are shown at right. Scale bars, 10 µm. c Preemptive recruitment of C-BLOCK to HTT prevents condensate formation. HEK293T cells coexpressing HTTQ72-FRB-EGFP and mCh(K70N)-FKBP were treated for 44 h with 500 nM AP21967 rapalog. Treated cells show a reduction in HTTex1 condensates compared with untreated cells or cells not expressing C-BLOCK. Representative images are shown at left, with quantification (as in Fig. 1k) at right. Data represents average and error (s.e.m, n = 3 regions with cells examined from 3 independent experiments). Scale bars, 10 µm.

Fig. 4. Light-activated CRY2olig condensates can be disrupted using DisCo.

a Schematic of DisCo approach with CRY2olig. Light triggers condensate formation, while rapamycin-mediated recruitment of C-BLOCK to a FRB ‘hook’ on CRY2olig scaffold disrupts condensates. b, c CRY2olig condensates are disrupted equally regardless of recruitment orientation. DisCo was performed on HEK293T cells expressing CRY2olig-FRB-mCh and mCh(K70N)-FKBP, or EGFP-FRB-CRY2olig and mCh-FKBP. Cells were illuminated with light (488 nm, 100 ms every 30 s) at 18–22 h post-transfection, then 333 nM rapamycin was added 5 min after light onset. Representative images shown in (b), with quantification of cytosolic disruption in (c). Data were quantified as in Fig. 1c and represent average and error (s.d., n = 10 cells examined from three independent experiments). Scale bars, 10 µm. d Kinetics of condensate disassembly. HEK293T cells expressing CRY2olig-FRB-mCh and mCh(K70N)-FKBP were illuminated 5 min (488 nm, 100 ms pulse every 20–30 s), 18–22 h after transfection. Black triangles, dark, no rapamycin. Blue squares, light (488 nm, 100 ms pulse every 30 s), 333 nM rapamycin. Red circles, control cells expressing CRY2olig-mCh (no FRB) and mCh(K70N)-FKB, with 333 nM rapamycin in dark. All experiments were performed at 33.5 °C. Data shows average and error (s.d., n = 5 cells examined from three independent experiments). e FRAP experiments with EGFP-FRB-CRY2olig condensates formed after 3 min light treatment. Data shows average and error (s.e.m, n = 6 cells examined from three independent experiments). Representative images of a photobleached condensate are shown at bottom. Scale bars, 10 µm. f Ca²⁺-dependent DisCo. HEK293T cells expressing EGFP-CaM-CRY2oligC9 and mCh-CBP formed condensates with light (488, 100 ms pulse every 30 s for 6 min). Addition of 2.5 mM CaCl₂ and 3 µM ionomycin resulted in condensate disruption. Image at far right shows control experiments subject to the same light and Ca²⁺ treatments, but without C-BLOCK. The experiment was repeated a second time with similar results. Scale bars, 10 µm. g Quantification of kinetics and extent of Ca²⁺-dependent disruption of CRY2olig condensates in experiments shown in (f).

Fig. 5. DisCo blocks CRY2olig self-interaction while maintaining other protein–protein interactions.

a Schematic of plasma membrane recruitment experiments, testing effect of DisCo on ability of CRY2 to interact with a plasma membrane anchored CIBN. b HEK293T cells expressing CIBN-CaaX, EGFP-FRB-CRY2olig, and mCh-FKBP as C-BLOCK were illuminated with light (488 nm, 200 ms pulse every 20 s) 18–22 h post-transfection. Light-induced CRY2olig condensate formation and recruitment of condensates to the plasma membrane through CIBN interaction. The addition of 333 nM rapamycin (3 min after light onset, ‘Light+Rap’) disrupted condensates but maintained CRY2olig-CIBN interaction. Magnified images in the second row correspond to the areas outlined in yellow. Graph below shows intensity profile of fluorescence at the plasma membrane from the magnified areas, indicating the loss of clustering with rapamycin addition. The experiment was repeated a second time with similar results. Scale bar, 10 µm. c C-BLOCK recruitment prior to light application prevents CRY2olig condensate formation, while allowing CRY2-CIBN association. Experiment was performed as in (b), but with rapamycin added 25 min prior to illuminating cells with light (488 nm, 200 ms pulse every 20 s for 8 min). The experiment was repeated a second time with similar results. Scale bar, 10 µm. d Schematic of mitochondrial recruitment experiments, testing effect of DisCo on interaction of CRY2 with CIBN anchored at the mitochondria. e COS-7 cells coexpressing TOM20-CIBN, CRY2olig-FRB-mCh and mCh(K70N)-FKBP as C-BLOCK were imaged in dark (‘Dark’), then treated with 333 nM rapamycin followed by light (488 nm, 100 ms every 30 s) (“Rap then light”). CRY2olig treated with rapamycin prior to light does not form condensates but is still recruited to CIBN. At far left are images of cells illuminated with light in the absence of rapamycin, which results in CRY2olig oligomerization at the mitochondrial membrane and mitochondrial compaction. The experiment was repeated a second time with similar results. All scale bars, 10 µm.

Fig. 6. Use of DisCo to rapidly reverse endocytosis blockage induced by CRY2olig and light.

a Schematic showing use of DisCo to reverse light-dependent blockage of clathrin-mediated endocytosis induced by CRY2olig clustering. b Timecourse (left) and results (right) of endocytosis assay. CRY2olig-clathrin light chain (CLC) constructs with or without a FRB hook were coexpressed in HEK293T cells along with mCh(K70N)-FKBP as C-BLOCK. 18–24 after transfection, cells were kept in dark (‘No Light’) or exposed to blue light (465 nm, 1 s pulse every 30 s for 10 min). Cells were treated with 333 nM rapamycin or vehicle, then incubated for 35 or 80 additional minutes in light (465 nm,1 s pulse every 30 s) (“Light”) or dark (“Dark recovery”) as indicated, then assayed for transferrin uptake. As a control for the reduction of transferrin uptake due to light treatment, cells were treated with light (465 nm, 1 s pulse every 30 s for 10 min) and assayed for transferrin uptake immediately after the light treatment (Light with no recovery). Transferrin uptake is expressed as a percent of results obtained from neighboring untransfected cells (set at 100%). Data represents average and error (s.e.m., n = 50 cells). Parametric t-test (two-tailed) between + /- rapamycin-treated cells showed a significant difference (***) with p < 0.0001 (Light 35 min, p = 1.4 ×10⁻¹⁴; Dark recovery 35 min, p = 1.4×10⁻¹⁴; Light 80 min, p = 1.7×10⁻¹⁰; Dark recovery 80 min, p = 5.5 ×10⁻¹¹).The experiment was performed an additional time with similar results. c Representative images of transferrin uptake in experiment shown in (b). Cells treated with DisCo (“+ Rap”) show recovery from CRY2olig-CLC clustering and normal uptake of transferrin even when incubated in light. The experiment was performed an additional time with similar results. Scale bars, 10 µm.