Structural Control of Dual Emission in Coumarin Fluorophores for Visualizing Protein Droplet Maturation Authors: Tomoya Yamamoto, Yusei Otsuka, Asuka Mori, Yuichiro Hori, Yuma Yamamoto, Motonari Uesugi, Kazuya Kikuchi Published: 2026 ABSTRACT ABSTRACTProteins droplets formed by liquid–liquid phase separation (LLPS) gradually lose fluidity and mature into aggregated and fibrillar states, initiating fibrillation of amyloidogenic proteins. Despite increasing interest in droplet maturation, high‐throughput methods for visualizing this process remain undeveloped. In this study, we report a fluorescence‐based method for real time visualization of droplet maturation using dual‐emission fluorophores. Coumarin‐based fluorophores conjugated with electron‐withdrawing rings exhibit dual fluorescence originating from multiple ground‐state conformers although predicting this behavior from molecular structure remains challenging. Here, we synthesized coumarin‐based fluorophores bearing directly conjugated electron‐withdrawing five‐membered ring to increase the abundance of twisted conformers through steric hindrance. The dual‐emission properties and fluorescence quantum yield were tuned by varying the heteroatoms within the five‐membered ring. Upon covalent conjugation of the probe to proteins, ratiometric confocal microscopy revealed that the emission ratio faithfully reflected the maturation process of liquid droplets, enabling quantitative evaluation of maturation timescale and heterogeneous internal environments. Furthermore, by combining this probe with the photoactive yellow protein (PYP) tag strategy, we visualized differences in the internal environments of protein droplets in living cells. This strategy allows continuous monitoring of droplet maturation and provides new insight into the mechanism of LLPS, amyloidogenic proteins fibrillations, and inhibitor screening against amyloid formation.Keywords: aggregation, amyloidogenic proteins, dual‐emission fluorophore, liquid–liquid phase separation, ratiometric fluorescence imaging Introduction Biomacromolecules including proteins, DNAs, and RNAs undergo liquid–liquid phase separation (LLPS) under crowded cellular environment to form fluid liquid droplets which regulates the activity, localization and compartmentation of the molecules [1, 2, 3]. In particular, the concentrated interior promotes protein aggregation, causing droplet maturation and loss of fluidity [4, 5]. Fibrillation of amyloidogenic proteins often originates in matured droplets [6, 7, 8, 9]. Visualization of the internal environment of droplets is crucial for elucidating the physicochemical mechanisms of protein droplet maturation [10, 11, 12]. Fluorescence recovery after photobleaching (FRAP) is currently the gold standard for monitoring the maturation process of liquid droplets. However, it requires bleaching individual droplets, preventing simultaneous observation of many droplets. In addition, variations in laser intensity and bleaching duration reduce reproducibility [2]. Hence, alternative, reproducible, and high‐throughput methods for monitoring droplet maturation are needed. Environment‐sensitive fluorophores are used to monitor protein aggregation and droplet maturation. Molecular rotor probes such as thioflavin T exhibit fluorescence when intramolecular rotation is restricted in aggregates [6, 7, 13, 14, 15, 16, 17, 18]. Although these dyes visualize droplet maturation, intensity‐based imaging is influenced by probe concentration and its partitioning into droplets. Polarity‐sensitive probes also visualize aggregation and droplet polarity through ratiometric fluorescence or λ‐scan wavelength shifts [11, 19, 20, 21]. However, the resulting intensity–ratio changes are small. Fluorescence lifetime imaging microscopy (FLIM) can evaluate the physicochemical states of droplets [11, 22]. However, it requires complex instrumentation, long acquisition times, and precise decay fitting, limiting high‐throughput analysis. To overcome the limitations of conventional fluorophores, we developed dual‐emission fluorophores to monitor droplet maturation sensitively. Coumarin‐conjugated dyes have recently gained attention for their dual‐emission fluorescence that arises from having multiple ground‐state conformations (Figure 1a) [23, 24, 25, 26, 27, 28]. The planar conformer shows extended conjugation with longer excitation (∼473 nm) and emission (∼530 nm) wavelengths, whereas the twisted conformer confines conjugation to the coumarin moiety, exhibiting shorter excitation (∼400 nm) and emission (∼450 nm) wavelengths [25, 26]. Thus, these conformers are excited independently at distinct wavelengths. The dual‐emission ratio is highly sensitive to environmental conditions such as polarity, viscosity, and steric hindrance, which influence conformer population, twisted–planar conversion, and excited‐state stability of the dye. We speculated that this dual emission would be modulated by the microenvironmental changes on protein surfaces during droplet maturation (Figure 1b). Nevertheless, due to the aforementioned complexity of the dual‐fluorescence mechanism, designing fluorescent probes that exhibit dual emission remains challenging. We hypothesized that conformational flexibility controlled by steric hindrance would facilitate dual fluorescence. Therefore, we synthesized a series of coumarin‐based fluorophores in which a five‐membered ring was directly conjugated to the coumarin core using a conformationally less flexible olefin‐containing linker to determine whether such fluorophores would exhibit dual emission useful in studying protein droplet maturation. (a) Dual emission properties of coumarin‐conjugated fluorophores. (b) Application of coumarin‐conjugated fluorophores to visualization of the properties of liquid droplet. Results and Discussion Dual Emission Properties of Coumarin‐Based Fluorophores To develop dual‐emission fluorescent probes, we designed coumarin‐based fluorophores conjugated with various electron‐withdrawing five‐membered rings. Coumarin dyes conjugated with electron‐withdrawing rings have previously been reported to exhibit dual emission in donor–acceptor configurations [23, 24, 25, 26, 27, 28]. However, the mechanism underlying dual fluorescence is highly complex, involving the relative stability and interconversion dynamics of multiple conformers in both the ground and excited states, making precise structural prediction of dual fluorescence characteristics challenging. We designed coumarin‐based fluorophores 1a–1d that exhibited efficient dual emission by directly conjugating an electron withdrawing five‐membered ring to the coumarin scaffold. The five‐membered ring was connected to the coumarin scaffold using an olefinic linker to increase ground‐state steric hindrance and thus favor the twisted conformer (Figure 2). The fluorescence properties were further tuned by altering the type of heteroatom introduced to the five‐membered ring to enhance the quantum yield. In addition, the N‐ethyl group introduced to the five‐membered ring provided a site for installing protein‐reactive functionalities that would enable covalent protein labeling. Fluorescence spectra of 10 µM coumarin‐conjugated fluorophores excited at 390 nm for (a,b,d) or 370 nm for (c). Slit width: 2.5 nm for (a,c) and 5.0 nm for (b,d). Dual fluorescence properties were evaluated from the emission spectra of the twisted conformations in various solvents (Figure 2). Similar to unconjugated coumarin‐based fluorophores, excitation at 370 or 390 nm generates an emission peak at 400–460 nm if the twisted conformer is sufficiently populated and its conversion to the planar form in the excited state is suppressed. Notably, the fluorescence derived from the twisted conformer was exceptionally strong in toluene and could not be measured using the same slit width settings; therefore, the spectrum was presented as a normalized spectrum (Figure S1). These fluorophores independently exhibited fluorescence peaks corresponding to the planar conformer upon excitation at 430–480 nm (Figure S2). The fluorescence and absorption spectra of the planar conformer were analyzed using the Lippert–Mataga equation to evaluate the polarity‐dependent behavior of the planar conformer (Figures S2–S4) [29, 30]. All fluorophores exhibited polarity sensitivity, with fluorophores 1a and 1b showing particularly strong responses to solvent polarity. Furthermore, the viscosity responsiveness of these fluorescence spectra was investigated in glycerol/methanol mixed solvents [16, 31]. For fluorophores 1a, 1b, and 1d, the intensity of the peak attributed to the planar conformer showed pronounced viscosity sensitivity, whereas 1c exhibited only weak viscosity sensitivity (Figure S5 and S6). The fluorescence peak attributed to the twisted conformer also increased in intensity under high‐viscosity conditions. Notably, since the coumarin moieties of these probes are known to readily bind to bovine serum albumin (BSA), we also investigated how BSA binding affects the fluorescence intensity and dual‐emission behavior of the probes [25]. Upon addition of BSA, increases in dual‐emission signals were observed, presumably due to fluorophore‐BSA binding (Figure S7). In addition, the intensity ratios between the two emission peaks changed in many of the fluorophores, suggesting that binding to BSA altered their conformational distributions. The quantum yield was significantly increased in 1a and 1c, where the NH group is located next to the C═C bond connecting the five‐membered ring to the coumarin core (Table S1). The N‐ethylhydantoin‐conjugated fluorophore 1c showed clear dual fluorescence with peaks at 430–450 nm and 490–550 nm even in polar protic solvents (Figure 2c), while N‐ethylthiohydantoin conjugated fluorophore 1a mainly exhibited emission from planar conformation with peak at 500–550 nm (Figure 2a). This lack of dual emission is likely attributed to the facile twisted‐to‐planar transition in the ground state. These results suggest that the structure of fluorophore 1c is particularly suited for visualizing changes in the protein surface environment through dual fluorescence. Mechanistic Investigation of the Dual Emissions Coumarin fluorophores conjugated with electron‐withdrawing moieties have been reported to exhibit dual emission behavior similar to that observed for fluorophore 1c in this study [23, 24, 25, 26, 27, 28, 32]. Upon excitation at around 400 nm, these fluorophores typically show fluorescence at approximately 450 nm together with a more red‐shifted emission originating from a more extended conjugation system. In contrast, excitation at the major absorption wavelength predominantly produces only the longer‐wavelength fluorescence, suggesting the existence of different absorption species. The origin of this behavior has been discussed based on two distinct mechanisms: emission from different electronically excited states of the same molecular species, such as the S2 and S1 states [23, 24], and emission associated with conformational heterogeneity arising from rotation around the linker between the coumarin scaffold and the electron‐withdrawing moiety [25, 26]. However, the excitation‐wavelength‐dependent fluorescence behavior may not be fully explained by direct S2/S1 emission processes because rapid internal conversion is generally expected. The excitation‐wavelength‐dependent fluorescence peaks observed for these fluorophores are consistent with the presence of multiple absorbing species, potentially arising from conformational heterogeneity in the ground states. Therefore, in this study, we hypothesized that dual emission behavior originates from heterogeneous ground‐state conformers together with conformational relaxation in the excited state, and investigated the underlying mechanism. To investigate the mechanism underlying the dual emission of fluorophore 1c, quantum chemical calculations were performed to predict its conformations and HOMO–LUMO gaps. In the present fluorophore system, we focused on rotation around the single bond between the coumarin moiety and the olefin linker because this bond exhibits relatively high conformational flexibility in the ground state and is expected to affect the absorption and fluorescence properties of the fluorophore. Energy calculations were performed by varying the conformation around this bond, and the rotational energy profile as a function of the dihedral angle φ(C2–C3–C9–C10) was obtained. As a result of the dihedral angle scan, two stable conformations were obtained: one at a dihedral angle φ of approximately 0°, in which an intramolecular hydrogen bond is formed between the hydantoin and coumarin moieties, and another at a dihedral angle φ of approximately 180° (Figure 3a). The present conformational analysis employed the SMD solvent model, in which direct intermolecular interactions with solvent molecules are not explicitly considered. Therefore, the stabilization of the conformation is strongly influenced by intramolecular hydrogen bonding interactions. On the other hand, in actual systems where direct interactions with solvent molecules occur, this intramolecular hydrogen bond is expected to be weakened. In addition, the entropic contribution arising from conformational fluctuations should also be taken into account. Thus, the present calculations do not provide a direct indication of which conformation is more populated under experimental conditions; however, they suggest the existence of two stable conformations. (a) Potential energy profile as a function of the dihedral angle (φ: C2–C3–C9–C10), obtained by relaxed DFT scans every 10° at the B3LYP/6‐311++G(2d, 2p) level of theory using SMD solvent model for MeOH. (b,c) Excitation/emission energy diagrams and oscillator strength for the excitation and emission processes of the two representative conformers, together with the corresponding molecular orbitals calculated by TD‐DFT calculations at the CAM‐B3LYP/6‐311++G(2d, 2p) level of theory using the SMD solvent model for MeOH. For these conformations, TD‐DFT calculations were performed to evaluate their electronic structures in both the ground and excited states (Figure 3b,c). The calculated electronic energy gaps, which qualitatively correlate with the absorption and fluorescence wavelengths, differed depending on the molecular conformation. In the ground state, the conformation with a dihedral angle φ around 180° exhibited a smaller energy gap than that with φ around 0°. A similar trend was also observed for the excited‐state optimized structures, where the φ around 180° conformation retained a smaller energy gap. These results suggest that the conformer with a dihedral angle around 180° possesses a more extended effective π‐conjugation, leading to red‐shifted absorption and fluorescence properties compared with the 0° conformer. Similar conformational dependence of the electronic energy gaps was also observed for fluorophores 1a, 1b, and 1d (Figure S8). These trends are consistent with the experimental observations for fluorophore 1c, where excitation at 370 nm produced a fluorescence peak around 430 nm, whereas excitation at 430 nm produced a fluorescence peak around 500 nm. Although the calculated difference in the ground‐state HOMO–LUMO gaps was smaller than the experimentally observed spectral difference, this discrepancy may arise because explicit interactions with solvent molecules destabilize the 0° conformation. Such destabilization may favor conformations with reduced π‐conjugation and consequently larger HOMO–LUMO gaps. Collectively, the present calculations suggest that conformations with dihedral angles around 180° are likely associated with longer‐wavelength absorption and emission, whereas conformations represented by the 0° dihedral angle are likely associated with shorter‐wavelength absorption and emission. These results support the hypothesis that conformational heterogeneity in the ground state contributes to the excitation‐wavelength‐dependent dual emission behavior observed experimentally. The hypothesis that distinct conformers exhibit different excitation and emission wavelengths was investigated using spectroscopic analyses (Figure S9). First, the excitation spectra of fluorophore 1c were compared. The excitation spectrum monitored at 484 nm exhibited a broad excitation band spanning from 370 to 500 nm, whereas the excitation spectrum monitored at 420 nm showed a major excitation band in the range of 350–400 nm. These results suggest that the 484 nm emission and 420 nm emission originate from different absorbing species, consistent with the presence of multiple ground‐state conformers with distinct electronic structures. Next, fluorescence lifetime measurements of fluorophore 1c were performed for the 430 nm and 550 nm emissions (Table S2). Both emissions exhibited a short‐lived component below 1 ns together with a longer‐lived component of approximately 2–3 ns. The lifetime values differed between the two emissions, with the longer‐lived component of the 430 nm emission being approximately 0.6 ns longer than that of the 550 nm emission, suggesting that these emissions originate from distinct emissive states. In contrast, an additional long‐lived component was also observed for the 550 nm emission upon 370 nm excitation. Importantly, the sub‐nanosecond and 2–3 ns lifetime components observed upon 370 nm excitation were similar to those observed upon 454 nm excitation. These results suggest that excitation at 370 nm partially generates an emissive state similar to that directly populated upon 454 nm excitation, possibly through excited‐state conformational relaxation. These results suggest that the two fluorescence emissions are generated through different excitation pathways associated with distinct ground‐state conformers. Furthermore, upon 370 nm excitation, a portion of the excited fluorophores is suggested to undergo excited‐state conformational relaxation to generate an emissive state similar to that directly populated upon 454 nm excitation, possibly through conformational relaxation in the excited state. Collectively, these results support a model in which multiple ground‐state conformers possessing distinct absorption characteristics contribute to the excitation‐wavelength‐dependent emission behavior, while excited‐state conformational relaxation may additionally participate in generation of the longer‐wavelength emissive state. Visualization of Protein Droplets With Dual‐Emission Fluorophores Next, we attempted to visualize protein droplets using the developed fluorophores in protein solution systems. Many protein droplets are known to incorporate hydrophobic fluorescent molecules, and compounds structurally similar to our fluorophores have also been reported to partition into protein droplets. Therefore, we investigated whether the developed fluorophores could be incorporated into droplets formed by lactoferrin (LT) and γ‐globulin (GG). LT and GG droplets were prepared by mixing protein solutions with a PEG‐8000 solutions according to previous reports (Figure S10a,b) [5]. As a result, all fluorophores showed partitioning into LT droplets, whereas little fluorescence was observed within GG droplets, indicating that the partition efficiency depended strongly on the type of droplet‐forming protein. In addition, fluorophore 1c showed little fluorescence upon 405 nm excitation, under conditions where blue fluorescence was expected, making observation of dual emission difficult. These results suggested that the non‐covalent introduction of fluorophores into protein droplets was not suitable for monitoring droplet maturation with ratiometric imaging based on dual emission. Extending from our fluorophore screening, we designed and synthesized probe 2 by introducing a butanoic acid moiety into the hydantoin‐conjugated fluorophore 1c (Figure 4a). The carboxylic acid group of probe 2 was converted into an N‐hydroxysulfosuccinimide ester, enabling covalent modification of lysine side‐chain amino groups and N‐terminal amino groups of proteins in aqueous solution [33, 34]. Probe 2 was covalently attached to four proteins—lactoferrin (LT), β‐lactoglobulin (β‐lac), β‐casein (β‐cas), and γ‐globulin (GG)—which are known to form protein droplets at relatively low concentrations upon mixing with PEG‐8000 solution [5]. The dual‐emission behavior of the resulting protein conjugate was then analyzed. The conjugated probes exhibited distinct dual fluorescence behaviors depending on the proteins (Figure 4b), consistent with earlier reports showing non‐covalent dye–protein interactions influencing dual emission from coumarin dyes [25]. (a) Scheme for covalent labelling of proteins with 2. (b,c) Fluorescence spectra of free 2 (2 only) and 2‐labelled γ‐globulin (2‐GG), β‐casein (2‐βcas), lactoferrin (2‐LT), and β‐lactoglobulin (2‐βlac) with (b) λ ex = 405 nm, slit width: 5.0 nm and (c) λ ex = 473 nm, slit width: 2.5 nm. While unconjugated probe 2 alone showed little twisted‐state emission (Figure 4b, black), this emission increased after protein conjugation with the highest twisted/planar intensity ratio observed with LT (Figure 4b, red). For βlac, the planar‐state emission exhibited a hypsochromic shift while no detectable twisted‐state emission appeared. Importantly, selective observation of each conformer is feasible using common fluorescence microscope channels: the blue channel (λ ex = 405 nm; λ em = 420–460 nm) for twisted‐state emission and the green channel (λ ex = 473 nm, λ em = 490–540 nm) for planar‐state emission (Figure 4c). Protein‐conjugated probe 2 showed quantum yields of 0.06–0.20, except when bound to β‐casein, which are adequate for fluorescence imaging (Table S3). These ratiometric differences likely arise from variations in the physicochemical properties of the protein surfaces, including steric hindrance and polarity. When probe 2 is conjugated in a sterically crowded environment, the ground‐state equilibrium between twisted and planar conformations may shift. This is consistent with the UV–Vis spectra, where proteins such as LT showing clear dual fluorescence also exhibited broadened absorption toward shorter wavelengths (Figure S11). Such a shift in absorption wavelength is likely derived from the ground‐state twisted conformer. The planar form undergoes intramolecular charge transfer in the excited state, producing polarity‐dependent fluorescence shifts [19], while high‐viscosity environments enhance planar emission by stabilizing this state [16, 35]. In contrast, polar solvents are reported to favor the twisted‐to‐planar conversion in the excited state [25]. These environmental factors are expected to change during protein droplet aggregation, thereby modulating the observed ratiometric behavior. To confirm on this idea, we measured dual‐emission intensity ratios as those proteins form liquid droplet under macromolecular crowding using confocal fluorescence microscopy [5]. Both twisted‐ and planar‐state emissions were detected from LT and GG droplets (Figure 5a,d), demonstrating that the droplet environment can be visualized through the fluorescence intensity ratio of blue (twisted) and green (planar) channels. (a) Confocal microscopy images of liquid droplets of lactoferrin (LT) in 20 mM sodium phosphate buffer (pH 7.4) containing 250 mM NaCl and 10 w/v% PEG‐8000; 2‐LT: LT at 2.5 µM. Non‐labelled LT: 37.5 µM. (b) Time‐course change of mean G/B ratio at different concentrations of total LT. (c) Fluorescence spectra of 2‐LT incubated in 20 mM sodium phosphate buffer (pH 7.4) containing 10% (w/v) PEG8000 for 1 h (black) and 72 h (red). (d) Confocal microscopy images of liquid droplets of γ‐globulin (GG) in 20 mM sodium phosphate buffer (pH 7.4) containing 10 w/v% PEG‐8000. λ ex: 405 nm for blue; 473 nm for green; 2‐GG: GG at 10 µM. (e) Time‐course change of mean G/B ratio of GG liquid droplets. (f) Enlarged confocal microscopy images of GG liquid droplets. White arrows indicate puncta where blue fluorescence is strong. Scale bar: 5 µm. λ ex: 405 nm for blue, 473 nm for green. λ em: 420–460 nm for blue, and 490–540 nm for green. Same laser power and sensitivity was used for each protein liquid droplets. Next, we evaluated the capability of the ratiometric fluorescence imaging method to monitor the maturation process of the droplets. Previous studies have reported that LT and GG lose internal fluidity within 72 h of droplet formation [5]. Therefore, time‐lapse fluorescence imaging was conducted to analyze temporal changes in the mean green/blue (G/B) ratio obtained from ratiometric images. Fluorescence acquisition was performed with identical laser power and detector sensitivity for each protein, enabling quantitative comparison of the two channels across different protein samples. In LT droplets, the G/B ratio decreased over time and with increasing protein concentration (Figure 5b). Consistently, fluorescence spectra revealed an increased contribution from twisted‐state emission at longer incubation times (Figure 5c). Given that LT droplet fluidity decreases over time, the changes in these ratios likely reflect protein aggregation within the droplets. Twisted‐state emission is enhanced by lower polarity and increased conformer population. Because aggregation is presumed to reduce local polarity and increased packing, it is thought to enhance twisted‐state fluorescence [25]. Similarly, higher protein concentrations intensified the twisted emission likely due to aggregation‐enhanced packing of LT molecules. GG droplets also exhibited time‐dependent decrease in fluidity accompanied by reduction in the G/B ratio values (Figure 5d,e). Notably, puncta with strong blue emission appeared within GG droplets (Figure 5f), suggesting local aggregation or environmental heterogeneity. In addition, these droplets exhibited lower fluorescence intensity ratio values at their periphery, suggesting preferential aggregation at the droplet interface. Thus, this ratiometric fluorescent probe effectively visualized distinct aggregation kinetics and spatial variations within protein droplets. Notably, no significant differences were observed in droplet formation, morphology, or time‐dependent morphological changes between protein droplets formed from covalently labeled proteins and those visualized through non‐covalent fluorophore partitioning (Figure S10). These results suggest that, at the labeling ratio used in this study, covalent modification has only a minor effect on the droplet formation. Analysis of Droplet Maturation for Amyloidogenic Protein FUS Next, we applied 2 to the amyloidogenic protein, fused in sarcoma (FUS), which is known to form liquid droplet and implicated in neurogenerative diseases such as amyotrophic lateral sclerosis and frontotemporal lobar degeneration [36, 37, 38]. For the FUS droplet‐forming system, we used a fusion protein in which maltose‐binding protein (MBP) and FUS are linked by a TEV protease‐cleavable sequence (MBP‐tev‐FUS) [39, 40]. Hydrophilic MBP served as a solubility tag, and removable of MBP with TEV protease triggered FUS droplet formation by reducing the FUS solubility. MBP‐tev‐FUS was labeled with 2 via covalent conjugation and applied to this system. After TEV protease addition, droplet formation was observed in both the blue and green channels (Figure 6a). During the first 6 h of incubation, the G/B ratio gradually increased (Figure 6b), opposite to the trend observed with LT and GG (Figure 5b,e). This difference may arise from the distinct droplet‐maturation mechanism among these proteins. The maturation of LT and GG is mainly attributed to protein aggregation without major structural rearrangements [5], whereas FUS maturation involves conformational transitions from intrinsically disordered states to cross‐β‐sheet structures typical of amyloidogenic proteins [41, 42, 43]. Planar‐state emission originates from an intramolecular charge‐transfer excited state, whose emission is enhanced when conformational flexibility is reduced particularly in viscous environments [44]. During FUS maturation, the formation of ordered fibrillar structures likely stabilizes the planar conformation in both ground and excited states, thereby enhancing green fluorescence. (a) Confocal microscopy images of liquid droplets of FUS labelled with 2 in dialysis buffer containing 75 mM NaCl. Scale bar: 5 µm. λ ex: 405 nm for blue, 473 nm for green. λ em: 420–460 nm for blue, and 490–540 nm for green. Same laser power and sensitivity was used for each protein liquid droplets. (b) Time‐course change of mean G/B ratio at different concentrations of NaCl. We monitored FUS droplet maturation at different salt concentrations, as the mechanism of FUS LLPS is known to depend on salt concentration [45]. Previous reports have shown that FUS undergoes LLPS through hydrogen bonding at low salt (∼75 mM) and hydrophobic interaction at high salt (∼2 M) concentrations, whereas droplet formation is less favorable at intermediate salt concentrations (∼150 mM). Based on this knowledge, maturation was monitored at 75 mM, 150 mM, and 2.5 M (Figure 6b). The increase in the G/B ratio was slightly faster at high salt (2.5 M) than low salt (75 mM) concentration, while little change was observed at 150 mM. These findings are in good agreement with the previously reported LLPS‐driven mechanism of FUS droplet maturation [45]. Thus, ratiometric fluorescence imaging based on dual‐emission probe enables screening of droplet maturation rates under various formation conditions. In addition to the time‐dependent changes in the overall ratio values of the droplets, spatially heterogeneous ratio distributions of within the droplets were also observed. While the overall ratio values of the droplets increased over time, aggregate‐like structures in the protein solution exhibited markedly lower ratio values (Figure S12a). This result suggests that the highly fluid droplet state and the aggregated state provide substantially different microenvironments for the fluorophore, resulting in distinct photophysical properties compared with those observed in the droplet state. Notably, under 1.0 M NaCl conditions, regions exhibiting low ratio values comparable to those of the aggregate‐like structures appeared at the droplet interface (Figure S12b). Thus, as observed for the GG droplet system, the early stages of aggregation from droplets could be visualized by ratiometric fluorescence imaging. To explore the potential application of this method to drug screening, we compared droplet maturation behavior in the presence of epigallocatechin gallate (EGCG), a compound widely studied as an inhibitor of amyloid formation (Figure S13) [46]. Under the conditions employed in this study, the addition of EGCG did not produce a significant change in the maturation rate. However, as demonstrated in this experiment, comparing droplet maturation in the presence of small molecules enables evaluation of whether these molecules inhibit droplet maturation. In the future, by automating imaging and data analysis, the developed fluorescent probe could be applied to the screening of compounds that inhibit droplet maturation. In this series of experiments using protein solutions, the probes were randomly conjugated to the proteins. For proteins such as LT and GG, whose molecular structures are not expected to undergo significant conformational changes during droplet maturation, the fluorescence response to maturation is likely to be relatively insensitive to the fluorophore labeling position. In contrast, for proteins such as FUS, in which droplet maturation is accompanied by structural transitions including amyloid‐like fibril formation, more pronounced changes in the fluorescence intensity ratio may be observed when the fluorophore is attached near regions involved in the formation of cross‐β‐sheet structures [41, 42, 43]. Therefore, to further improve the sensitivity to droplet maturation, it will be very important to site‐selectively introduce fluorophores at amino acid residues associated with structural transitions through cysteine mutagenesis and thiol‐selective labeling strategies. Visualization of the Internal Environment of Protein Droplets in Living Cells Finally, we applied the developed fluorophore to the imaging of droplets formed in living cells. We have previously established a method for covalent fluorescence labeling of droplet‐forming proteins in living cells using the Photoreactive Yellow Protein (PYP) Tag [47, 48]. In this system, droplet‐forming proteins were expressed in cells as fusion proteins containing PYPD71N E74Q D97N (PYPNQN), a highly reactive mutant of the PYP tag [49]. A ligand containing chloroacetyl and trifluoromethyl groups (PCAF ligand) covalently modifies the tag in a rapid and efficient manner [50]. Therefore, we synthesized probe 3, in which the PCAF ligand and the fluorophore 1c were connected by a linker, and investigated the application to this system (Figure 7a). The obtained probe 3 was confirmed to rapidly react with PYPNQN and covalently label the tag (Figure S14). (a) Chemical structure of probe 3. (b) Confocal fluorescence and green/blue ratio images of PYPNQN fusion proteins labeled with probe 3 in HEK293T cells. Scale bar: 20 µm. λ ex: 405 nm for blue, 473 nm for green. λ em: 420–460 nm for blue, and 490–540 nm for green. Same laser power and sensitivity were used for each protein. The contrast settings for blue and green fluorescence images were adjusted independently for each protein. Next, live‐cell imaging experiments were performed using PYPNQN fusion plasmids encoding five previously investigated phase‐separating proteins: FUS, FUS(P525L), TDP‐43, TDP‐43(A315T), and HNRNPA1 [47]. HEK293T cells were transfected with each plasmid, and labeling with probe 3 was carried out 24 h after transfection. Distinct punctate structures were observed for all proteins other than HNRNPA1 (Figure 7b). Fluorescence was weak in non‐transfected cells, suggesting that the probe 3 selectively bound to and labeled the PYP tag‐fused proteins (Figure S15). The G/B ratio was highest for HNRNPA1, whereas TDP‐43 and TDP‐43(A315T) exhibited the lowest values. FUS and FUS(P525L) showed intermediate ratio values (Figure S16a). These results correlate well with the propensity of these proteins to transition into solid aggregates reported in previous studies [47]. TDP‐43 and TDP‐43(A315T), which readily undergo solidification, likely exhibited lower ratio values due to aggregation‐induced decreases in local polarity and increased molecular packing. FUS(P525L) and TDP‐43(A315T), which contain mutations that promote transition into solid phases [51, 52, 53], showed little difference compared with their wild‐type counterparts. In addition, slight differences in the ratio values were observed among the punctate structures formed by FUS(P525L), suggesting that these differences may reflect droplet maturation (Figure S16b). Furthermore, as shown in Figure S16b, regions with different ratio values were observed even within the same droplet, suggesting that the internal heterogeneity within individual droplets can be visualized. Collectively, these results demonstrate that the dual‐emission fluorescent probe developed in this study can detect the different environments within intracellular droplets and has the potential to monitor the maturation of biomolecular condensates in live cells. Conclusion In this study, we developed a dual‐emission probe that enables real‐time monitoring of protein droplet maturation. By covalently conjugating this probe to droplet‐forming proteins, maturation can be visualized through quantifying the intensity ratio of dual emissions. This approach reveals differences in maturation kinetics and internal microenvironments of droplets formed under various conditions. Because the probe evaluates droplet properties through imaging, it allows analysis of environmental heterogeneity among and within individual droplets arising from differences in their size and morphology. Having demonstrated its applicability to amyloidogenic protein droplets, the probe can also be used to screen inhibitors of amyloid fibril formation. Furthermore, by combining the probe with a tag protein strategy, differences among intracellular droplets formed by different proteins could be visualized through the dual‐emission fluorescence intensity ratio. Overall, this work establishes a versatile fluorescence‐based platform for elucidating the molecular mechanism of protein droplet maturation. Author Contributions Tomoya Yamamoto: conceptualization, investigation, funding acquisition, methodology, writing – review and editing, writing – original draft, supervision, and visualization. Yusei Otsuka: methodology, investigation, writing – review and editing, and visualization. Asuka Mori: investigation and methodology. Yuichiro Hori: resources and investigation. Yuma Yamamoto: investigation and resources. Motonari Uesugi: investigation and resources. Kazuya Kikuchi: supervision, conceptualization, funding acquisition, writing – review and editing, and project administration. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Supporting File: The authors have cited additional references within the Supporting Information [54, 55, 56, 57, 58, 59, 60]. Acknowledgements This work was supported by a Grant‐in‐Aid for Transformative Research Area (A) “Latent Chemical Space” [23H04880 and 23H04881 (K. K.)], for Transformative Research Area (A) “Biological Cluster” [25H02370 to M. U.] for Transformative Research Area (A) “Autophagy expanded: decoding membrane interface biology” [26H01633 to T. Y.] from the Ministry of Education, Culture, Sports, Science and Technology, Japan. This research was also supported by JSPS KAKENHI (grant numbers: JP21H04706, JP23KK0106, and JP24H00494 to K. K.; 22K14787 to T. Y.); the JSPS CORE‐to‐CORE Program “Asian Chemical Biology Initiative”, the Grant Program of the Co‐Creation Organization, the University of Osaka. We thank the analytical instrument facility, Graduate School of Engineering, the University of Osaka, for HRMS analysis. Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request. References Associated Data Supplementary Materials Supporting File: The authors have cited additional references within the Supporting Information [54, 55, 56, 57, 58, 59, 60]. Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request.