Dual‐Color Photoconvertible Fluorescent Probes Based on Directed Photooxidation Induced Conversion for Bioimaging

Abstract We herein present a new concept to produce dual‐color photoconvertible probes based on a mechanism called Directed Photooxidation Induced Conversion (DPIC). As a support of this mechanism, styryl‐coumarins (SCs) bearing Aromatic Singlet Oxygen Reactive Moieties (ASORMs) like furan and pyrrole have been synthesized. SCs are bright fluorophores, which undergo a hypsochromic conversion upon visible light irradiation due to directed photooxidation of the ASORM that leads to the disruption of conjugation. SC‐P, a yellow emitting probe bearing a pyrrole moiety, converts to a stable blue emitting coumarin with a 68 nm shift allowing the photoconversion and tracking of lipid droplet in live cells. This new approach might pave the way to a new generation of photoconvertible dyes for advanced bioimaging applications.

Photoconversion of fluorescent dyes in the visible range is a powerful tool in bioimaging to unambiguously track labeled biomolecules over large spatiotemporal scales. Although efficient photoconvertible fluorescent proteins have been developed, [1,2] molecular probes, characterized by their small size, improved optical properties, versatility and ease of use constitute a complementary and robust tool in the field of bioimaging, as they provide a homogeneous staining in cells and can be used in tissue and in vivo imaging. [3,4,5] In this context, Dual-Color Photoconvertible Fluorophores (DCPFs), able to convert from a bright emissive form to another, being advantageously detected prior to conversion, have recently drawn attention. Bright fluorophores like AlexaFluor 647 [6] and other cyanines [7] have been evaluated as DCPFs by studying the emissive form resulting from partial photobleaching. Conversely to this empirical approach, DCPFs with rational designs have been proposed through various mechanisms including: dealkylation of rhodamines, [8] phototruncation of cyanines, [9,10,11] photocyclodehydrogenation of diazaxanthilidene [12,13] and AIEgen, [14] spiropyranization of oxazine, [15] and photooxidative dehydrogenation. [16] Although these efforts led to efficient DCPFs, some limitations remain to be addressed including the undesired reversibility, low brightness of both initial and converted forms, irradiation with phototoxic UV, and low yields of conversion. Additionally, the kinetics of the photoconversion process should be adapted. Indeed, slow conversion could lead to phototoxicity of live samples, whereas fast conversion would make the probe unusable. Ideally, a more universal approach, applicable to various fluorophores, would enable to extend the palette of properties and available colors of DCPFs for various applications in bioimaging.
We thus aimed at establishing a rational design of DCPFs supported by a new mechanism. Herein, we introduce a strategy to obtain photoconvertible fluorophores based on Directed Photooxidation Induced Conversion (DPIC). We hypothesized that efficient DCPFs could be obtained from a directed partial photobleaching of the fluorophore leading to a hypsochromic shift of both excitation and emission spectra. Photobleaching occurs upon light irradiation where fluorophores generate singlet oxygen ( 1 O 2 ), which oxidizes the dye and mediate irreversible photochemical cleavage leading to non-emissive species. [17] We thus assumed that the conjugation of a fluorophore to an Aromatic Singlet Oxygen Reactive Moiety (ASORM) could 1) lead to a bathochromic shift by extension of the π system, and 2) upon generation of 1 O 2 , direct the oxidation towards the ASORM, thus provoking a disruption of the conjugation, leading to a hypsochromic shift in excitation and emission ( Figure 1).
For our study, coumarin was chosen as small and bright blue emissive fluorophore. Coumarins display high extinction coefficients and fluorescence quantum yields with low cytotoxicity making them a suitable platform to develop fluorescent probes for bioimaging. [18] Pyrrole [19] and furan [20] were chosen as ASORMs for their small size, and their ability to photooxidize in the presence of 1 O 2 resulting in dearomatization. The coumarin core fluorophore was thus conjugated to N-methylpyrrole and furan through a Wittig reaction to obtain the extended styryl-coumarins SC-P and SC-F respectively ( Figure 1). As comparison, styryl coumarins thiophene (SC-T) and anisole (SC-A) were synthesized to demonstrate the unique effect of Singlet Oxygen Reactive Moiety (ASORM) in the photoconversion process (Figure 1).
Additionally, a vinyl coumarin (SC-V) was synthesized as a model of product resulting from Directed Photooxidation Induced Conversion ( Figure 1). The styryl coumarins were obtained in pure form and were characterized by 1 H and 13 C NMR as well as high resolution mass spectrometry.
The photophysical properties of the styryl coumarins (SCs) were evaluated in various solvents and showed high brightness ranging from 24 000 to 43 900 M À 1 cm À 1 with slight solvatochromic shifts ( Figure S1, Table S1). In methanol, SC-T, SC-F and SC-A displayed quite similar properties with λ Abs and λ Em around 425 and 512 nm respectively (Figure 2A), extinction coefficient (ɛ) around 30 000 M À 1 .cm À 1 and fluorescence quantum yields up to 0.77 (Table 1), whereas SC-P displayed a significant red-shifted emission spectrum ( Figure 2A, Table 1). As expected, SC-V is a typical blue-emissive coumarin (Figure 2A) though displaying a rather high brightness (Table 1: To evaluate their dual-color conversion properties, methanolic solutions of dyes were irradiated with a continuous wave 488 nm-laser and emission spectra were recorded over the time. The results showed that unlike SC-A (Figure 2B) and SC-T ( Figure S2) that slowly photobleached with no sign of conversion, SCs bearing ASORMs, namely SC-F ( Figure 2C) and SC-P ( Figure 2D), photoconverted rapidly towards a similar blue-shifted emissive form (λ Em =  486 and 483 nm respectively) with a hypsochromic shift of 24 nm and 68 nm, respectively (Table 1).
To further characterize the converted SC (cSC), HPLC/ UV/Visible/mass spectrometry analysis were performed before and after irradiation ( Figure 2E and S3-6). The results showed that both SC-P and SC-F converted towards isomers of oxidized forms (M + O 2 ), which displayed blueshifted absorption spectra due to dearomatization of the ASORMs. In addition, 1 H NMR of SC-P showed that after irradiation only the signals corresponding to the pyrrole moiety were changed ( Figure S7). According to these results a structure of cSC-P where the pyrrole got oxidized in its pyrrolidinone form is proposed in Figure 2E, which is in line with reported photooxidized form of pyrrole. [19] Interestingly, the photoproduct arising from SCs that did not photoconvert (cSC-A and cSC-T) also showed oxidized forms (M + nO 2 ), along with cleaved photoproducts (Figure S8), thus demonstrating that Singlet Oxygen Reactive Moieties (ASORMs) preferentially direct the photooxidation towards themselves.
The similarity of both emission and excitation spectra between the final converted forms, cSC-P and cSC-F and the model of conversion SC-V ( Figure 2F, Table 1), suggested that the conversion arose from the disruption of the conjugation between the coumarin and the ASORMs.
To prove that this process involved singlet oxygen, a set of experiments was conducted on SC-P. First, we showed that SC-P was stable over the time at ambient conditions in absence of light ( Figure S9). Then under irradiation in argon-degassed methanol the kinetics of photoconversion was significantly decreased ( Figure S10). The use of DPBF proved that SC-P generated 1 O 2 upon light irradiation ( Figure S11). Additionally, when 1 O 2 was independently generated through a 640 nm excited aluminum phthalocyanine, the conversion occurred within 10 s ( Figure S12), proving that the conversion arose from the chemical reaction between 1 O 2 and SC-P at its ground state. As SC-P could generate other ROS than 1 O 2 , the reaction of different ROS without light irradiation was also evaluated and showed that only hydroxyl radical (HO * ) could also trigger the conversion ( Figure S13). However, the use of HPF as HO * probe showed that SC-P does not generate HO * under irradiation ( Figure S14).
From those results, and knowing the photophysical properties of the SCs, their converted form (cSCs) similar to our model SC-V, and by monitoring the decrease of the fluorescence signal over the time (Figure S15), several photophysical constants involved in the phototransformation have been determined and are reported in Table 1. Although SC-F displayed an impressive estimated chemical conversion yield (η = 97 � 9%) compared to SC-P (η = 30 � 1%), the latter possesses more interesting photoconversion properties with a higher quantum yield of phototransformation (φ Pt ) and photoconversion (φ Pc ), depicting a higher photoreactivity (100-fold compared to SC-A), and a higher spectral shift upon conversion of 68 nm (24 nm for SC-F), which is preferable for bioimaging applications. Interestingly, the quantum yields of photobleaching (φ Bl ) of the converted forms (cSC-P and cSC-F) were shown to be slightly lower than for SC-V and for non-convertible SCs, thus depicting photostable photoproducts after conversion ( Table 1).
Overall these sets of experiments confirmed our hypothesis that efficient dual-emissive photoconvertible probes can be obtained through DPIC mechanism and a detailed mechanism is proposed in Figure 3. Upon irradiation in the visible range SC-P reaches an excited state that can transfer its energy to triplet oxygen ( 3 O 2 ) to generate singlet oxygen ( 1 O 2 ) through intersystem crossing and deexcitation of the triplet state to the ground state. In the presence of the generated 1 O 2 , SC-P undergoes phototransformations (characterized by Φ Pt ). Indeed, at its ground state SC-P reacts with 1 O 2 either on the core coumarin leading to photobleaching (characterized by Φ Bl ) or in a directed manner towards the ASORM resulting in photoconversion (characterized by Φ Pc and the chemical yield η) and leading to a blue-shifted fluorescent dye (cSC-P) due to the disruption of the conjugation between the coumarin and the ASORM.
Other key parameters include the fluorescence brightness of both the initial probe (ɛ and Φ F ) and its converted form (ɛ c and Φ Fc ), as well as their quantum yield of 1 O 2 generation (Φ Δ ).  Proposed mechanism for the Directed Photooxidation Induced Conversion. 1 O 2 is generated upon irradiation and reacts at the ground state leading to a phototransformation (φ Pt ) which can be photoconversion (φ Pc ) or photobleaching (φ Bl ). The constants (in grey) were determined (Table 1) and support the presented mechanism (for definitions see Supporting Information). Φ Δ could not be determined due to the competitive reactivity of SCs and DPBF towards 1 O 2 (See Figure S11).
To demonstrate the application of DPIC mechanism in bioimaging, SC-P, which displayed the fastest conversion along with the highest shift in emission, was chosen for cellular imaging in Hela cells (Figure 4). Prior to cell experiments, cytotoxicity and phototoxicity assays were performed and showed that SC-P was not cytotoxic at 1 μM and not phototoxic after that the photoconversion was performed ( Figure S16). As shown in Figure 4A, conversion was successfully achieved after one frame as the intensity of SC-P and cSC-P were respectively decreased and increased upon high irradiation at 488 nm ( Figure 4B) with an average conversion yield of 91 � 8%. Advantageously, the conversion can be triggered at a desired timepoint ( Figure S17) as well as in various selected regions of interest without converting the surrounding regions ( Figure S18), showing a good spatiotemporal control of the conversion. Moreover, cSC-P was sufficiently stable to be tracked over time (at 1 Hz) over several frames ( Figure S17, S18). Using SMCy5.5 as a counterstain [21] (Figure 4C), we then showed that SC-P unexpectedly label lipids droplets (LDs) with a high association rate with SMCy5.5 of 83%, probably due to its non-charged and relatively lipophilic nature (cLogP = 2.97). Importantly, the converted form was shown to keep the ability to remain in LDs after conversion as cSC-P association rate with SC-P reached 94% ( Figure 4D). Additionally, taking advantage of MemBright-labeled endosomes, [22] we checked that SC-P was not endocytosed. Indeed, SC-P was very weakly associated with endosomes (1.38%, Figure S19). Finally, we proved that both SC-P and cSC-P could be tracked over 50 frames ( Figure 4E), and shared the same trajectories and average speed than SMCy5.5 upon tracking ( Figure 4F). Overall, these experiments demonstrated that SC-P and its readily obtained photoconverted form cSC-P, preferentially stains the LDs and can be both tracked over large spatiotemporal scales in bioimaging.
In conclusion, we presented a new concept based on the Directed Photooxidation Induced Conversion (DPIC) mechanism enabling to obtain bright fluorophores that readily photoconvert upon visible light towards photostable blueshifted photoproducts. Although furan was found to provide a cleaner conversion, pyrrole, due to its electron rich nature was found to be a more suitable Aromatic Singlet Oxygen Reactive Moiety (ASORM), as it provided an important emission shift once conjugated to the fluorophore and displayed faster conversion rate. This approach led to SC-P, a coumarin-based DCPF that was successfully used to photoconvert and track LDs in live cells. Preliminary results from our group suggest that substituted pyrroles provide conversion as well and that this method could be applied to other fluorophores, opening opportunities to develop new efficient DCPFs and paving the way to new photoconvertible fluorescent probes for advanced microscopy.