Controlling synthetic membraneless organelles by a red-light-dependent singlet oxygen-generating protein

Abstract

Membraneless organelles (MLOs) formed via protein phase separation have great implications for both physiological and pathological processes. However, the inability to precisely control the bioactivities of MLOs has hindered our understanding of their roles in biology, not to mention their translational applications. Here, by combining intrinsically disordered domains such as RGG and mussel-foot proteins, we create an in cellulo protein phase separation system, of which various biological activities can be introduced via metal-mediated protein immobilization and further controlled by the water-soluble chlorophyll protein (WSCP)-a remarkably stable, red-light-responsive singlet oxygen generator. The WSCP-laden protein condensates undergo a liquid-to-solid phase transition on light exposure, due to oxidative crosslinking, providing a means to control catalysis within synthetic MLOs. Moreover, these photoresponsive condensates, which retain the light-induced phase-transition behavior in living cells, exhibit marked membrane localization, reminiscent of the semi-membrane-bound compartments like postsynaptic densities in nervous systems. Together, this engineered system provides an approach toward controllable synthetic MLOs and, alongside its light-induced phase transition, may well serve to emulate and explore the aging process at the subcellular or even molecular level.

Fig. 1. Schematic illustration of a red-light-controlled protein condensate enabled by water-soluble chlorophyll-binding protein (WSCP).

The protein condensate formed via LLPS of RGG-Mfp-3-RGG (RMR) is decorated with WSCP (PDB: 2DRE), a tetrameric protein that generates singlet oxygen (¹O₂) under red light irradiation. Upon photooxidation, the protein condensate undergoes a liquid-to-solid phase transition, which restricts the diffusion of substrates into the enzyme-laden protein condensate and turns off the catalysis within.

Fig. 2. Concentration-dependent liquid-liquid phase separation of RMR.

a Schematic showing the phase separation of RMR. b Concentration-dependent phase separation of RMR at room temperature (23 °C). Scale bar: 5 μm. Images representative of n = 3. c Optical densities at 600 nm at different RMR concentrations. The threshold RMR concentration for phase separation at 23 °C is 4 μM. Data are presented as mean ± SD (n = 3).

Fig. 3. Temperature-dependent liquid-liquid phase separation of RMR in the presence and absence of WSCP.

a Schematic showing temperature-dependent phase separation of RMR. b Representative images showing temperature-dependent phase separation of RMR in the presence and absence of WSCP. Scale bars: 5 μm. Images representative of n = 3. c Schematic showing stabilization of RMR condensates by WSCP. Two RMR condensates are coalescing into a bigger one in the absence of WSCP, while remaining separate upon decoration with WSCP. Confocal imaging analysis revealed the presence of WSCP both on the surfaces and inside of the RMR condensates (Supplementary Fig. 4). The cartoons of WSCP inside the condensates are omitted for clarity. d Representative images showing RMR condensates in the absence and presence of WSCP at 23 °C with the passage of time. Scale bars: 1 μm.

Fig. 4. Red-light-dependent liquid-to-solid phase transition of RMR/WSCP condensates.

a Schematic illustration of red-light-induced liquid-to-solid phase transition process. b Appearance of an RMR/WSCP condensate at 23 °C under irradiation with the passage of time. Images representative of n > 10. c Percentages of solid particles throughout red-light irradiation with the passage of time. Solid particles were differentiated from the spherical liquid condensates based on their irregular shapes (Supplementary Fig. 3). Data are presented as mean ± SD (n = 3). d Schematic showing light-heat cycling experiments that comprise iterative red-light irradiation at 23 °C (5 min, 1 mW/cm²) and ensuing incubation at 52 °C in the dark (5 min). Images were taken at time points as indicated by red and black arrows. e Number of solid particles observed per image. Data are presented as mean ± SD (n = 4). f Percentage of heat-resistant solid particles observed after each light-heat cycle. Data are presented as mean ± SD (n = 4).

Fig. 5. Modulation of RMR/WSCP condensates with high spatial precision by optical tweezers.

a Schematic illustration of the use of a modified optical tweezer system consisting of trapping and triggering laser beams to modulate the protein condensates at the single-condensate level. Trapping laser: 1064 nm; 80 mW. Triggering laser: 532 nm; 2.5 mW. b Schematic illustration of the trapping of two protein condensates (Trap 1 and Trap 2) by optical tweezers, followed by selective irradiation of the chosen condensate (Trap 1; left) via the triggering laser beam. A live camera image of two trapped condensates is shown. Scale bar: 5 μm. c Liquid-to-solid phase transition of the chosen RMR/WSCP condensate (left) under the irradiation of the triggering laser beam within 15 s. The concentration of WSCP is 5 μM. Scale bar: 5 μm. Images representative of n = 9. d The RMR condensate in the absence of WSCP is inert to prolonged irradiation. Scale bar: 1 μm. Images representative of n = 2.

Fig. 6. Decoration of RMR condensates with His6-tagged GFP via metal coordination.

a Schematic illustration of Ni²⁺-induced recruitment of His6-tagged GFP into RMR condensates. b Fluorescence micrographs showing GFP-deprived and GFP-laden RMR condensates in the absence and presence of Ni²⁺ (50 μM), as well as those in the presence of Ni²⁺ (50 μM) and excess of EDTA (2 mM). Scale bars: 10 μm. Images representative of n = 3. c Relative distribution of GFP inside and outside RMR condensates. Enrichment efficiency was calculated using the equation, $Enrichment=GFP\left(in\right)/GFP\left(out\right)$. Data are presented as mean ± SD (n = 10); two-side t-test, p-value: **** < 0.0001 [p = 2.75 × 10⁻⁸ (+ Ni²⁺ vs - Ni²⁺); p = 3.19 × 10⁻⁵ (+ Ni²⁺ vs. + Ni²⁺ + EDTA)]. d Representative micrograph of a GFP-laden condensate and its normalized fluorescence intensity profile. Scale bar: 1 μm.

Fig. 7. Photo-controlled catalysis within RMR condensates.

a Schematic illustration of the mechanism for controlling catalysis within the caspase-3-laden condensate via red-light-induced phase transition. Substrates readily diffuse into a liquid condensate, but not into a solid particle. b Schematic illustration of the fluorogenic caspase-3 substrate, Z-DEVD-AFC. AFC, 7-amino-4-trifluoromethylcoumarin, with${\lambda }_{ex}=405\mathrm{nm}\mathrm{and}{\lambda }_{em}=500\mathrm{nm}$. c Influence of light and WSCP on bound caspase-3. Light intensity: 5 mW/cm²; irradiation duration: 120 min. Data are presented as mean ± SD (n = 3); two-side t-test, p-value: **** < 0.0001 (p = 3.41 × 10⁻⁷ (Light + WSCP vs Light ^-WSCP)); p = 7.92 × 10⁻⁷ (Light + WSCP vs. Dark + WSCP). d Influence of irradiation duration and intensity on catalysis. Data are presented as mean ± SD (n = 3).

Fig. 8. Phase separation in living cells.

a Fluorescent images of HEK293 cells transfected with the gene encoding RGG-Mfp-3-RGG-mCherry (RMR-mCherry), mCherry, or RGG-GFP-RGG (RGR). Scale bars: 5 μm. b Percentage of HEK293 cells that exhibit LLPS. Data are presented as mean ± SD (n = 8); two-side t-test, p-value: **** < 0.0001, p = 9.30 × 10⁻⁶ (RMR vs RGR); p = 8.89 × 10⁻⁶ (RMR vs mCherry). c 3D rendering and representative z-slice images of HEK293 cells producing RMR-mCherry. Scale bars: 10 μm. FRAP assays of RMR condensates in the cytoplasm (d) and in the outer membrane of HEK293 cells (e). The plots show the normalized fluorescence recovery after photobleaching. Scale bars: 10 μm. Data are presented as mean ± SD (n = 8 in d and n = 16 in e).

Fig. 9. Light-induced liquid-to-solid phase transition in HEK293 cells.

a Schematic illustration of light-induced phase transition in cells. b Fluorescent images of the cells harbouring chlorophyll (λ_ex, 405 nm; λ_em, 640 nm) while expressing RMR-mCherry and EGFP-WSCP. Scale bar: 10 μm. Images representative of n = 5. FRAP assays of RMR condensates in the cytoplasm (c) and in the outer membrane of HEK293 cells (d). The plots show the recovery of the fluorescence (normalized) after photobleaching. Scale bar: 10 μm. Data are presented as mean ± SD (n = 9 in c and n = 10 in d).