Spatiotemporal Control of Intracellular Phase Transitions Using Light-Activated optoDroplets

Abstract

Phase transitions driven by intrinsically disordered protein regions (IDRs) have emerged as a ubiquitous mechanism for assembling liquid-like RNA/protein (RNP) bodies and other membrane-less organelles. However, a lack of tools to control intracellular phase transitions limits our ability to understand their role in cell physiology and disease. Here, we introduce an optogenetic platform that uses light to activate IDR-mediated phase transitions in living cells. We use this "optoDroplet" system to study condensed phases driven by the IDRs of various RNP body proteins, including FUS, DDX4, and HNRNPA1. Above a concentration threshold, these constructs undergo light-activated phase separation, forming spatiotemporally definable liquid optoDroplets. FUS optoDroplet assembly is fully reversible even after multiple activation cycles. However, cells driven deep within the phase boundary form solid-like gels that undergo aging into irreversible aggregates. This system can thus elucidate not only physiological phase transitions but also their link to pathological aggregates.

Keywords: RNA binding protein; aggregation; fluisomes; gelation; intracellular condensates; intrinsically disordered protein; liquid-liquid phase separation; membraneless organelles; optogenetics; ribonucleoprotein bodies.

Figure 1. Rapid light-dependent clustering of IDR fused Cry2

(A) Schematic diagram of the optogenetic platform. Three IDR-containing RNA binding proteins are used in this study: FUS, DDX4 and HNRNPA1. The “optoIDR” construct consists of the N-terminal IDR fused to mCherry fluorescent protein and the Cry2PHR domain. (B) Blue light activation of optoIDRs leads to rapid clustering in living cells. Representative fluorescence images of light-activated assembly of Cry2WT, optoFUS, optoDDX4 and optoHNRNPA1. All cells here are at similar expression levels and activated under identical conditions. Scale bar = 20 μm. (C) Temporal evolutions of integrated fluorescence intensity density of cytoplasmic (except optoHNRNPA1 which predominantly localized in nucleus) clusters for cells shown in (B).

Figure 2. The concentration of activated molecules is a key determinant for light-activated droplet assembly

(A) (Top) A sequence of increasing blue light intensity applied to optoFUS cells. (Bottom) Images of optoFUS cells taken during the activation sequence (time points indicated by arrows). Scale bar = 20 μm. See also Movie S1. (B) Images of optoFUS cells with varying expression levels (numeric values on the left, a.u.) exposed to identical blue light activation conditions. Scale bar = 10 μm. (C) Example images of optoFUS cells showing compensating effects of the expression level and blue light intensity. Cells with ~ 2-fold lower expression levels were activated with ~ 3.5-fold higher blue light intensity, yielding similar clustering kinetics. Scale bar = 10 μm. See also Figure S1.

Figure 3. OptoFUS clustering is a light-activated phase transition

All experiments in this figure were performed with the optoFUS construct. (A) Schematic diagram of light activation kinetics of Cry2. k_act and k₂ denote activation and inactivation rate constants, respectively. C_act and C_inact denote activated and inactivated molecule concentrations, respectively. Kinetic rate equations for light activation are given below the schematic. (B) Light activation protocols with varying lengths of the activation interval, T, are employed to activate cells and temporal evolution of background fluorescence intensity outside of clusters, C_bg (black solid line in the example plot), is measured during clustering until it reaches steady state (red dotted line). C_inact,st and F_inact,st denote steady-state concentration and fraction of inactivated molecules, respectively. C_tot and C_sat represent total and saturation concentration, respectively. (C) (Left) Steady-state background fluorescence intensities of individual cells (open circles) under the given activation interval increase linearly (solid lines) with total concentration of molecules. In our kinetic framework, the y-intercept and the slope of the linear fit correspond to the saturation concentration, C_sat, and inactivated molecule fraction at steady state, F_inact,st, respectively. (Right) The saturation concentration is independent of activation intervals used. A red dashed line represents average of saturation concentrations measured at 5 different activation intervals and error bars are 95 % confidence intervals of linear fits. (D) Steady-state fractions of inactivated molecules increase with either longer activation intervals or weaker blue light intensities. Blue light powers (488 nm, in μW) used in the measurements are specified. Solid lines denote a global fit to data using the kinetic model (Equation (8), See STAR Methods). Error bars are SDs. (E) Steady-state concentrations of activated molecules for all cells examined were calculated using measured kinetic parameters, which show a clear concentration threshold for light-mediated clustering. A dotted horizontal line indicates the saturation concentration measured in (C). (F) Phase diagram calculated using the mesoscale continuum model with X_AB = −3, X_AC = −3, X_BC = 3.75. The solid gold lines indicate two different expression levels of fixed ϕ̄_A + ϕ̄_B. The red circles and arrow highlight an example activation pathway through which phase separation is induced, and the solid gray lines are the line of steady-state concentration ratios preferred by the reaction terms for (k₁ = 0.01 and T = 50), (k₁ = 0.01 and T = 17.5) and (k₁ = 0.2 and T = 75) from top to bottom. The left-hand side of phase boundary represents the saturation concentration. (G) Snapshots of droplet assembly from the simulation for the phase transition pathway (red arrow) shown in (F). See also Figure S2 and S3.

Figure 4. FUS_N-Cry2olig shows rapid clustering with lower saturation concentration than optoFUS

(A) Time-lapse images of cells with Cry2olig and FUS_N-Cry2olig upon blue light activation. FUS_N fusion leads to rapid cluster assembly. Scale bar = 20 μm. (B) Temporal evolutions of integrated fluorescence intensity of cytoplasmic clusters for cells shown in (A). (C) Steady-state background intensities of individual FUS_N-Cry2olig cells (open circles) under various activation intervals vs. total concentration of molecules. The cyclic activation protocol identical to one used for optoFUS (Fig. 3B–C) was applied for FUS_N-Cry2olig cells. Solid lines are linear fits to data, yielding 5-fold lower saturation concentration than optoFUS. See also Figure S4.

Figure 5. Localized phase transitions

(A) Time-lapse images of localized cluster formation for optoFUS. A circular area with a diameter of 1.9 μm (white dotted line) was periodically stimulated with blue light every 7.2 s. Scale bar, 10 μm. See also Movie S2. (B) Temporal evolution of integrated intensity density of clusters vs. distance away from the activation zone for clusters in (A). Solid lines are experimental data. Dashed lines are calculated droplet volume fraction profiles, θ_d(x,t), using coarse-grained phase transition model (All parameters used in the calculation are listed in Table S1). (C–D) Time-lapse images of optoFUS (C) and FUS_N-Cry2olig (D) showing clusters formed upon a single line activation (white dotted line). Scale bar, 10 μm. (E) Temporal evolution of standard deviations of integrated intensity density distributions for cells in (C and D). (Insets) Integrated intensity density distributions vs. distance away from the activation zone for time points in (C and D). Solid lines are experimental data and dashed lines are droplet volume fraction profiles, θ_d(x,t), calculated using the coarse-grained phase transition model (All parameters listed in Table S1). (F) Time-lapse images of cluster wave formation upon localized activation of FUS_N-Cry2olig under the identical activation cycle used in (A). Scale bar = 10 μm. See also Movie S3. (G) Temporal evolution of integrated intensity density of clusters vs. distance away from the activation zone for clusters in (F). Solid lines are experimental data. Dashed lines are calculated droplet volume fraction profiles, θ_d(x,t), using coarse-grained phase transition model (All parameters listed in Table S1). See also Figure S5 and S6.

Figure 6. Material state and reversibility of light induced clusters

(A) FRAP recovery curves for optoIDRs and Cry2olig variants. Cells were activated every 6 s for 10 min with weak blue light (~ 0.02 μW except for optoDDX4 and FUS_N-Cry2olig where 20 and 70% further reduced powers are used to account for their lower C_sat) to induce phase separation. Error bars represent SD (n = 7 – 14). (B) Normalized integrated intensities of light induced optoFUS clusters for three different supersaturation depths (See STAR Methods) after initiating blue light activation (time 0). Cells with similar expression levels are used. Integrated intensities were normalized with final values. (C) Distinct morphology of phase separated optoFUS clusters for shallow (top) and deep (bottom) supersaturation. Cell outlines are indicated with dashed lines. Scale bar = 5 μm. See also Movie S4. (D) Time-lapse images of optoFUS clusters forming upon deep supersaturation. Scale bar, 3 μm. See also Movie S4. (E) Example images of FRAP measurement for optoFUS gels. An optoFUS cell was activated every 6 s for 10 min with “deep” blue light condition. White dashed line indicates bleached area. Scale bar, 2 μm. (F) FRAP recovery curves of optoFUS clusters formed with varying supersaturation depths. The result for shallow activation is a replica of Fig. 6A. Error bars represent SD (n = 8 for both intermediate and deep). (G) Example images of Cry2WT and optoFUS cells under sequential cycles of assembly and disassembly. Cells are exposed to indicated blue light activation condition for 10 min to assemble clusters and then incubated in the absence of blue light for 25 min. Cell images before and after activation as well as at the end of each disassembly cycle are shown. Expression levels of these cells are similar. Scale bar = 10 μm. (H) Normalized number of clusters during disassembly cycles for cells in (G). The number of clusters for disassembly cycles was normalized with an initial cluster number for each cycle. FUS_N-Cry2olig data was added for comparison. Circles, squares and triangles indicate the first, second and third cycle, respectively. See also Figure S7.

Figure 7. Model for intracellular phase space

Conversion of molecular species from weak self-association state to high self-association one, for example through post-translational modification or exposure of RNA in RNP complexes, leads to liquid-liquid phase separation. When the depth is shallow, this process follows the green path to produces liquid droplets (i). Deep supersaturation along the red path results in the formation of solid-like gels, with arrested molecular dynamics (ii). Gels are initially reversible, but slow dynamics within promote the formation of irreversible aggregates over time (iii).