Design of SP-C and the Transformation from SP-C to MC-C

According to the “sticker-and-spacer” model, small molecules consisting of two hydrophobic moieties (“stickers”) connected by a hydrophilic linker (“spacer”) tend to condense into liquid microdroplets via liquid-liquid phase separation (LLPS) under certain conditions in the aqueous solution.^{12, 13, 17} Here, to build photo-switchable phase-separated coacervates, we chose hydrophobic spiropyran (SP) as the sticker, as the non-fluorescent SP can undergo photo-responsive transformation to the fluorescent merocyanine (MC) form under UV light.⁵¹ Utilizing a PEG₈ molecule as the linker, SP-PEG₈-SP was synthesized (Figure 1a). First, we dissolve the compound SP-PEG₈-SP in DMSO to form a stock solution. Then, we diluted the SP-PEG₈-SP stock into water under the illumination of a 565 nm light; a white turbid solution was obtained (Figure 1b). Under a microscope, non-fluorescent microdroplets were observed (Figure 1c). Next, we stained the microdroplets with a fluorescent dye, Nile Red, and observed the fusion of small microdroplets into larger ones within 20 seconds under a fluorescent microscope (Figure S1). After we photobleached a portion of the droplet, the fluorescence intensity of the bleached area recovered within 300 seconds to 85 % of the initial fluorescence intensity before bleaching (Figures 1d, 1e). All these observations are consistent with the conclusion that the microdroplets formed by SP-PEG₈-SP in aqueous solution are phase-separated coacervates (termed SP-C here) with liquid-like properties. Imaging and turbidity analysis showed that pH and NaCl concentrations affected SP-C coacervates′ size and fusion rates (Figure S2). As a control, we synthesized a PEG-modified SP monomer called SP-PEG₆. We found SP-PEG₆ dissolved in the aqueous solvent, showing no sign of coacervate formation (Figure S3). Taken together, we have shown a successful design of a phase-separating compound based on SP.

Next, we examined the photo-responsive transformation of SP to MC within coacervates (Figures 1f, 1 g). Following illumination by a 365 nm light, SP-C solution turned purple, and the coacervates remained in the suspension (Figure 1h). Light illumination caused a gradual increase of a peak at 500–600 nm in the UV/Vis spectrum (Figure 1i), corresponding to the generation of MC-PEG₈-MC. Also, the microdroplets turned fluorescent (Figure 1j) and could still fuse (Figure S4). In a FRAP experiment, when a part of the droplet was photobleached, the fluorescence recovered within 100 seconds (Figures 1k, 1 l). Imaging and turbidity analysis showed that pH and NaCl concentrations affected MC-C coacervates′ size and fusion rates (Figure S5). All the data above indicate that microdroplets formed by MC-PEG₈-MC (termed MC-C here) are also phase-separated coacervates with liquid-like properties. Thereby, SP-C can transform into MC-C under UV light illumination.

Photochemical Reactions in Coacervates: ROS Generation and MC-C/SP-C Interconversion

Next, we tested whether photo-illumination of MC-C can induce reactive oxygen species (ROS) production under visible light. 1,3-Diphenylisobenzofuran (DPBF) was used as a probe for singlet oxygen detection; singlet oxygen will convert DPBF into 1,2-Dibenzoylbenzene (DBZB) (Figure 2a). By irradiating the mixture of DPBF and MC-C with a 565 nm light, we observed a decrease in DPBF and an increase in DBZB within 1 minute (Figure 2b). The structure of the product DBZB was confirmed using NMR (Figure S24) and mass spectrum. In contrast, irradiating the mixture of DPBF and SP-C with 565 nm light didn't cause an obvious decrease in DPBF and an increase in DBZB (Figure 2c). These results demonstrate that MC-C can induce ROS production under visible light irradiation, whereas SP-C cannot. Using DPBF as the ¹O₂ trapping agent and Rose Bengal (RB) as the standard compound, the ¹O₂ generation yield (Φ_Δ) of MC-C in DI water were determined to be about 0.40 (Figure S6).

At the same time, the 565 nm light also caused the conversion of MC-C to SP-C, evidenced by the gradual decrease of the peak at 500–600 nm in the UV/Vis spectrum (Figure 2d). The transformation of SP-C and MC-C is reversible: after four cycles of switch between the two states using an alternating 365 nm and 565 nm photo illumination, only a slight loss (20 %) of the fluorescence intensity was observed (Figures 2e, 2 f).⁴⁹ These data demonstrated multi-directional photochemical reactions of the coacervates: the 365 nm light turns the coacervates to MC-C, the fluorescent ROS “ON” state, whereas the 565 nm light induces ROS formation of the MC-C and at the same time shuts down the fluorescence and turns the coacervates into SP-C, the non-fluorescent ROS “OFF” state (Figure 2a).

We also observed different photochemical properties between the MC-C coacervates and the well-dispersed MC-(PEG)₆ in water. Firstly, compared to MC-(PEG)₆, the absorption spectrum results show a notable red shift of the MC-C absorption and a constant absorption shape lacking the appearance of a shoulder peak at around 400 nm during isomerization (Figures 2d, 2 g). Secondly, irradiating the mixture of DPBF and MC-(PEG)₆ within 60 s didn't cause an obvious oxidation of DPBF (Figures 2h, S7), which suggests that the efficiency of ROS production induced by dispersed MC-(PEG)₆ in water is much lower than that of the MC-C. All these results imply the J-type stacking of MC in MC-C.^{49, 54} However, using Rhodamine B (RhB) as the standard compound, the fluorescence quantum yield (Φ_f) of MC-C in DI water were determined to be 0.016 (Figure S8), which is lower than the reported values,^{49, 51} implying a fluorescence attenuation of the MC molecules in the coacervates. We suppose that the stacking of functional groups in coacervates can change the reaction properties in certain ways, and the coacervate state may provide a significantly different environment for the molecule compared to the soluble form.

Cell Entry and Intracellular Photochemical Reactions

We and others have shown that coacervates of small molecules or peptides can enter cells.^{17, 33, 39} Next, we explored whether SP-C and MC-C can enter mammalian cells and maintain their properties in cells. Fluorescence microscopy showed most of the coacervates were distributed in the cytoplasm or on the cell membrane (Figures 3a, 3b, S9), showing that both SP-C and MC-C can enter cells. It's worth noting that, after incubating SP-C or MC-C with cells in the dark for 1 day, intracellular coacervates of the two groups all showed very weak fluorescence intensity (Figures 3a, 3b), which suggests a spontaneous conversion from MC-C to SP-C inside cells at 37 °C, whereas the presence of some polar compounds in the cell may prevent the complete conversion from MC to SP.^{51, 55} We, therefore, call the equilibrated SP-C/MC-C coacervates spiropyran coacervates collectively. Extracellular results also proved the spontaneous transformation from MC-C to SP-C and the stability of SP-C at 37 °C (Figure S10a, S10b). However, due to the formation of the hydrogen bond, the thermal stability of MC in aqueous solution is usually higher than that of SP in aqueous solution.^{48, 56} For example, the UV/Vis absorption spectrum shows that water-soluble SP-(PEG)₆ transforms spontaneously to MC-(PEG)₆ at 37°C without illumination (Figure S10c). The opposite transformation direction of spiropyran in water solution and in coacervates is probably due to the much lower effective polarity in coacervates than in water.⁵⁷ These results suggest that coacervation stabilizes SP structure in the physiological enviroment.

Next, the photochemical properties of the spiropyran coacervates were detected. The fluorescence of MC-C in cells was switched on and off for at least three cycles under alternating 365 nm/565 nm light irradiation (Figure 3c), indicating that SP-C and MC-C can transform between each other reversibly under the condition of photo-illumination in cells. Then, we harnessed the power of the ROS probe, 2,7-dichlordihydro-fluorescein diacetate (DCFH-DA), to detect singlet oxygen in cells under photo illumination. The green fluorescence intensity of the oxidized product of DCFH-DA, dichlorofluorescine (DCF), is proportional to the level of ROS generated. After incubation with SP-C for 24 h, HeLa cells were incubated with DCFH-DA and then subjected to UV/Vis light illumination for different cycles (365 nm, 2 min/565 nm, 2 min) (Figure 4a). Under a fluorescent microscope, the green fluorescence from DCF was only observed with cells treated by both the coacervates and UV/Vis illumination cycles, while cells treated with only light, or only the coacervates, or coacervates and 565 nm light showed negligible DCF fluorescence (Figures 4b and S11). These results underscore the potential of MC-C to elevate intracellular ROS production under visible light, and the ability of this activation function to be turned “off” when MC-C transforms to SP-C, thus realizing an “on-and-off” switchable ROS production in cells, a significant advancement in the field. Moreover, since intracellular SP-C shows higher thermostability than MC-C (Figures 3a, 3b), undesired photosensitization of MC can be avoided.

Furthermore, we explored the spatiotemporal control of the photochemical reactions of coacervates in cells. Using the laser of a confocal fluorescence microscope as the light source, we irradiated a confined region of the cells pre-treated with SP-C first and then DCFH-DA. The fluorescence intensity of cells in the illuminated area was approximately 4 times higher than that in the unilluminated area (Figure 4c), a testament to the precision of our approach and the spatially resolved ROS production in cells loaded with the spiropyran coacervates. Besides, the fluorescence intensity of DCF increased with more irradiation cycles (Figure 4b), indicating an improvement of ROS content, which demonstrates the versatility of our method and the temporal resolution of ROS production.

Photo-Induced Cytotoxicity to Cancer Cells

After showing the potential of the spiropyran coacervates in intracellular ROS generation, we explored the use of the spiropyran coacervates to kill cancer cells, taking advantage of the cytotoxicity of intracellular ROS.⁴⁶ Firstly, SP-C or MC-C at the concentration of 100 μM did not show any cytotoxicity in the dark (Figure S12). For photo-induced cytotoxicity, HeLa cells were incubated with SP-C for 24 h, illuminated with several UV/Vis illumination cycles, and then incubated for 24 h before being assayed for cell viability (Figure 5a). The results of CCK8 viability assay (Figure 5b), cell apoptosis assay (Figures 5c, S13), and LDH release assay (Figure 5d) all showed obvious cell death only in the group of treatment with both the coacervates and UV/Vis illumination cycles, while cells treated with irradiation only, coacervates only, or coacervates and 565 nm light (Figure S14) showed negligible death. These data correlate with MC-C-induced ROS production under visible light. For Hela cells, the IC₅₀ value of MC-C under 1 illumination cycle is around 140 μM (Figure S15). Also, the cell death rate increased with increasing irradiation cycles (Figure 5b). The cytotoxicity was further proved by the result of cell morphology change (Figure 5e). Spiropyran coacervates also entered other cell lines, such as SK-BR-3 and THP1 cells (Figures S16, S17), and elicited cytotoxicity under photo illumination (Figures 5f, 5 g). Besides, the coacervates also entered HeLa cell spheroids (Figure S18), showed photo-induced ROS production (Figure S19), and demonstrated a cell-killing effect (Figure 5h). All these results indicate that the spiropyran coacervates can be used for photo-controlled cell killing and has the potential to develop into new photodynamic therapeutics. In addition, water-soluble SP-(PEG)₆ exhibited significantly higher cytotoxicity than SP-C or MC-C at concentrations above 50 μM (Figure S12), suggesting that coacervation may affect the pharmacological properties of drug molecules.

Photo-Controlled Cancer Treatment in vivo

Lastly, we explored the in vivo efficacy of coacervates for cancer treatment. The coacervates showed similar hydrodynamic sizes in the serum-containing cell cultures (around 1 μm), according to DLS and turbidity measurements (Figure S20). Then the safety of the spiropyran coacervates was evaluated. According to the routine and biochemical blood analysis, mice that received SP-C did not show changes in the levels of white blood cells (WBC), red blood cells (RBC), platelets (PLT), and hemoglobin (HGB) at either day 1 or day 15. Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels remained normal (Figure S21b), indicating no toxicity to the liver. Histological examination data revealed normal cellular morphology and tissue structures (Figure S21c), suggesting that IV injection of spiropyran coacervates is well tolerated in mice.

Next, we examined the tissue distribution and tumor accumulation of the coacervates. BALB/c mice were subcutaneously implanted with 4T1 tumors. SP-C loaded with Nile Red were injected via the tail vein into tumor-bearing mice, and the mice were monitored using whole-body fluorescence imaging for 24 hours (Figure 6a). Nile Red signal appeared at the tumor site 3 hours after administration and peaked at the 6th hour before gradually diminishing (Figure 6b). Accumulation of the coacervates at the tumor site may be due to the prolonged retention of the spiropyran coacervates. Next, we dissected the tumor and performed OCT cryosectioning. Under a confocal microscope, fluorescent condensates were found in the tumor mass in the SP-C+Nile Red treatment group (Figure 6c), confirming the accumulation of the spiropyran coacervates in 4T1 tumors.

Next, we added a laser illumination step to induce in situ ROS generation from the coacervates at the tumor sites. Tumor-bearing mice received six different treatments: (I) PBS, (II) PBS+laser (404/565 nm), (III) SP-C, (IV) SP-C+laser (404 nm), (V) SP-C+laser (404/565 nm), and (VI) SP-C+laser (404/565*2) (Figure 6d). For group IV, we applied a laser illumination at 404 nm light to convert the spiropyran coacervates to the MC-C form. For group V, following the 404-nm light illumination, we applied a 565-nm light to elicit the ROS generation immediately. For group VI, in order to enhance the efficacy of ROS generation, we repeated 404/565 illumination at a laser power of 0.5 W/cm² for 4 minutes 6 hours and 8 hours after the drug administration, respectively. We observed that mice in all the groups behaved normally, and the body weights increased in 12 days (Figure 6e). Tumor growth in groups V and VI showed a remarkable decrease compared to groups I–IV, with the smallest tumor sizes observed in group VI (Figures 6f and 6 g). These data indicate that photo-induced in situ ROS generation is responsible for the tumor growth suppression effect. This conclusion was further supported by TUNEL and HE staining of tumor tissues: groups V and VI, massive cell necrosis and atrophy were observed, confirming the tumor damaging effect (Figure 6h). Lastly, we performed a DHE (dihydroethidium) assay to directly measure ROS levels in live cells. A comparison of groups IV and V showed a significantly higher signal in tumors that have received 565 nm light illumination (Figure 6i), consistent with our notion that the ROS induced by 565 nm light from the MC-C form was responsible for tumor cell killing and tumor growth suppression. Notably, although the cytotoxicity of MC-C was modest in vitro (IC₅₀ measured to be 140 μM in cell-based experiments), we observed a pronounced tumor-killing effect in vivo. We reason that the enrichment of MC-C at the tumor sites might explain such enhanced cytotoxicity in vivo. Taken together, these data prove that spiropyran coacervates have a strong tumor-site enrichment via IV injection, and a dual-wavelength switch of ROS production inhibits tumor growth and causes significant tumor cell killing.