Development of a Highly Selective and Sensitive Fluorescent Probe for Imaging RNA Dynamics in Live Cells

RNA imaging is of great importance for understanding its complex spatiotemporal dynamics and cellular functions. Considerable effort has been devoted to the development of small-molecule fluorescent probes for RNA imaging. However, most of the reported studies have mainly focused on improving the photostability, permeability, long emission wavelength, and compatibility with live-cell imaging of RNA probes. Less attention has been paid to the selectivity and detection limit of this class of probes. Highly selective and sensitive RNA probes are still rarely available. In this study, a new set of styryl probes were designed and synthesized, with the aim of upgrading the detection limit and maintaining the selectivity of a lead probe QUID−1 for RNA. Among these newly synthesized compounds, QUID−2 was the most promising candidate. The limit of detection (LOD) value of QUID−2 for the RNA was up to 1.8 ng/mL in solution. This property was significantly improved in comparison with that of QUID−1. Further spectroscopy and cell imaging studies demonstrated the advantages of QUID−2 over a commercially available RNA staining probe, SYTO RNASelect, for highly selective and sensitive RNA imaging. In addition, QUID−2 exhibited excellent photostability and low cytotoxicity. Using QUID−2, the global dynamics of RNA were revealed in live cells. More importantly, QUID−2 was found to be potentially applicable for detecting RNA granules in live cells. Collectively, our work provides an ideal probe for RNA imaging. We anticipate that this powerful tool may create new opportunities to investigate the underlying roles of RNA and RNA granules in live cells.


Introduction
RNAs exhibit complex dynamics in live cells, including temporal and spatial processing and transportation [1]. The complex spatiotemporal dynamics of RNA molecules affect diverse cellular function. RNAs dynamically interact with a large group of RNA-binding proteins that modulate RNA localization and function [2]. Such RNA-protein interactions govern RNA expression, processing, export from the nucleus, and assembly into translationally competent messages, as well as association into translationally inactive RNA granules, including stress granules (SGs) and processing bodies (P-bodies) [3,4]. Given the intricate connection between RNA dynamics and function, it is highly desired to develop tools for visualization of RNA molecules in live cells to study their functions.
Although specific RNA targets in live cells have been successfully visualized by using protein-based [5] and oligonucleotide-based probes [6], the global dynamics of RNA distribution and the spatiotemporal processing and transportation of RNA molecules remains to be analyzed in live cells and in vivo. To achieve this goal, much attention has been paid to RNA-selective small-molecule fluorescent probes [7]. As a powerful tool to visualize dynamic cellular processes in living organisms, small-molecule fluorescent probes have shown many advantages in the monitoring of endogenous RNA dynamics in a convenient, continuous, and real-time way [8][9][10]. Unfortunately, RNA-selective smallmolecule fluorescent probes suitable for RNA analysis are rarely available. To date, only one small-molecule fluorescent probe with RNA selectivity (named SYTO RNASelect) is commercially available. However, the practical applicability of SYTO RNASelect as an RNA-imaging probe in cells has been severely limited [11][12][13][14][15]. In addition, the molecular structure of SYTO RNASelect has not been published yet either, which makes it difficult to further improve the practical applicability based on SYTO RNASelect.
Recently, we discovered a novel RNA-selective fluorescent probe called QUID−1 by enzyme-digestion-based screening in HeLa cells [27]. QUID−1 is a quinoline-based indole-hemicyanine probe with a distinct turn-on response for RNA binding. Notably, QUID−1 exhibited significant selectivity for RNA versus DNA, which was significantly better than the commercial probe, SYTO RNASelect ( Figure S1). However, its detection limit for RNA is not satisfactory. The limit of detection (LOD) value of QUID−1 for baker's yeast RNA in solution was 26.5 ng/mL. Such data was much higher than data for commercial SYTO RNASelect with detection limit of 2.8 ng/mL for baker's yeast RNA ( Figure S2). These results impelled us to synthesize new probes based on QUID−1 scaffold aiming to improve its detection limit and maintain its selectivity for RNA.
To develop highly selective and sensitive RNA probes, we herein designed and synthesized a new set of styryl probes based on QUID−1 scaffold. Among these compounds, QUID−2 was the most promising candidate. The limit of detection (LOD) value of QUID−2 for the RNA was significantly improved in comparison with that of QUID−1. Further spectroscopy and cell imaging studies demonstrated the advantages of QUID−2 over a commercially available RNA staining probe, SYTO RNASelect, for RNA imaging. Owing to the high selectivity, high sensitivity, low cytotoxicity, and excellent photostability of QUID−2, we obtained long-term monitoring of global dynamics of RNA. Moreover, QUID−2 was found to have the potential to detect RNA granules in live cells.

Molecular Design of Probes QUID−1-QUID−6
QUID−1 was identified as a highly selective fluorescent probe for RNA in our previous studies [27]. However, its detection limit for RNA is not satisfactory. To develop highly selective and sensitive RNA probes, we attempted to synthesize a new set of analogues based on QUID−1 scaffold to improve its detection limit and maintain its selectivity for RNA. The sensitivity of nucleic acid probes is caused by a high affinity for nucleic acids, in combination with a high quantum yield upon binding [28]. Thus, we envisaged the possibility of upgrading the detection limit of QUID−1 by improving its binding affinity and fluorescence quantum yield with RNA. Previous studies indicated that the binding affinities between ligands and their target RNA often correlate with the number of cationic side chains present in the ligands [25,29,30]. Considering the structure of QUID−1, which includes a 1-methylpiperazine side group (pKa was calculated to be 8.99), this terminal cationic group makes an important contribution to the electrostatic interaction of QUID−1 with RNA. However, this group alone is likely insufficient to achieve high-potency binding to RNA. On the basis of these results, we envisaged the improvement of probes' binding affinity for RNA via introducing two cationic groups on the QUID−1 scaffold. Therefore, another 1-methylpiperazine side group was introduced to the indole unit of QUID−1, yielding the resulting probe QUID−2. Three compounds named QUID−3, QUID−4, and QUID−5 were also designed by replacing the 1-methylpiperazine group on the indole unit of QUID−2 with OH, COOEt, and COOH, respectively, to determine the role of the cationic groups (1-methylpiperazine) in binding RNA (Figure 1). Because the carbon chain length of cationic group may sometimes influence the binding of probes to nucleic acid, QUID−6 was designed by replacing the 1-methylpiperazine group on the quinolinium unit with 1-(3-aminopropyl)-4-methylpiperazine to explore the effect of the distance between 1-methylpiperazine group and the quinolinium unit on probe performance ( Figure 1). interaction of QUID−1 with RNA. However, this group alone is likely insufficient to achieve high-potency binding to RNA. On the basis of these results, we envisaged the improvement of probes' binding affinity for RNA via introducing two cationic groups on the QUID−1 scaffold. Therefore, another 1-methylpiperazine side group was introduced to the indole unit of QUID−1, yielding the resulting probe QUID−2. Three compounds named QUID−3, QUID−4, and QUID−5 were also designed by replacing the 1-methylpiperazine group on the indole unit of QUID−2 with OH, COOEt, and COOH, respectively, to determine the role of the cationic groups (1-methylpiperazine) in binding RNA ( Figure  1). Because the carbon chain length of cationic group may sometimes influence the binding of probes to nucleic acid, QUID−6 was designed by replacing the 1-methylpiperazine group on the quinolinium unit with 1-(3-aminopropyl)-4-methylpiperazine to explore the effect of the distance between 1-methylpiperazine group and the quinolinium unit on probe performance ( Figure 1).

Screening QUID Fluorescent Probes
Upon synthesis, the spectroscopic properties of newly synthesized compounds were initially studied in vitro. These molecules structurally possess similar conjugation scaffold and thus they exhibited a slight difference in their absorption and emission maxima. The absorption maxima were found around 475 nm, and the emission maxima of the probes were found around 555 nm in TE buffer (Table 1 and Figure S3). Notably, these probes featured with the large Stokes shifts of ~80 nm, which was beneficial for RNA imaging in cells. To identify the most promising fluorescent probe for RNA imaging, newly synthesized compounds were screened for fluorescence performance with a focus on their detection limits and selectivity for RNA. Two representative nucleic acids, including baker's yeast RNA and salmon testes DNA, were employed in the assays, as these nucleic acids from biological species are often utilized in studies for RNA-targeting fluorescent probes [13,15,31]. As shown in Table 1, these candidate probes alone in buffer displayed extremely weak emission, even in the presence of DNA samples. By contrast, most of the candidates displayed strong fluorescence when adding the RNA samples. As expected, candidates bearing two cationic side chains (QUID−2 and QUID−6) exhibited a significantly better detection limit for RNA (Table 1 and Figure S4). However, the detection limits of QUID−3, QUID−4, and QUID−5 are not satisfactory. Considering their structural features, QUID−3, QUID−4, and QUID−5 contained terminal hydroxyl side chain, terminal ester side chain, and terminal carboxylic side chains, respectively. Neither of them was positively chargeable, thus demonstrating that the cationic amine side chain was indispensable. Among these candidates, QUID−2 was the best probe in view of its outstanding detection limit for RNA. In addition, QUID−2 displayed weaker fluorescence emission in the presence of DNA samples (fluorescence quantum yield ΦF = 8.64%) relative to QUID−6 (fluorescence quantum yield ΦF = 19.33%), thus showing its better selectivity towards RNA. Among the candidates, QUID−2 was chosen as the most promising RNA fluorescent probe for further detailed investigation.

Screening QUID Fluorescent Probes
Upon synthesis, the spectroscopic properties of newly synthesized compounds were initially studied in vitro. These molecules structurally possess similar conjugation scaffold and thus they exhibited a slight difference in their absorption and emission maxima. The absorption maxima were found around 475 nm, and the emission maxima of the probes were found around 555 nm in TE buffer (Table 1 and Figure S3). Notably, these probes featured with the large Stokes shifts of~80 nm, which was beneficial for RNA imaging in cells. To identify the most promising fluorescent probe for RNA imaging, newly synthesized compounds were screened for fluorescence performance with a focus on their detection limits and selectivity for RNA. Two representative nucleic acids, including baker's yeast RNA and salmon testes DNA, were employed in the assays, as these nucleic acids from biological species are often utilized in studies for RNA-targeting fluorescent probes [13,15,31]. As shown in Table 1, these candidate probes alone in buffer displayed extremely weak emission, even in the presence of DNA samples. By contrast, most of the candidates displayed strong fluorescence when adding the RNA samples. As expected, candidates bearing two cationic side chains (QUID−2 and QUID−6) exhibited a significantly better detection limit for RNA (Table 1 and Figure S4). However, the detection limits of QUID−3, QUID−4, and QUID−5 are not satisfactory. Considering their structural features, QUID−3, QUID−4, and QUID−5 contained terminal hydroxyl side chain, terminal ester side chain, and terminal carboxylic side chains, respectively. Neither of them was positively chargeable, thus demonstrating that the cationic amine side chain was indispensable. Among these candidates, QUID−2 was the best probe in view of its outstanding detection limit for RNA. In addition, QUID−2 displayed weaker fluorescence emission in the presence of DNA samples (fluorescence quantum yield Φ F = 8.64%) relative to QUID−6 (fluorescence quantum yield Φ F = 19.33%), thus showing its better selectivity towards RNA. Among the candidates, QUID−2 was chosen as the most promising RNA fluorescent probe for further detailed investigation. Table 1. The optical data, limit of detection (LOD) value of probes (QUID−1-QUID−6) for baker's yeast RNA, and fluorescence quantum yields (Φ F ) of probes (QUID−1-QUID−6) in the presence of salmon testes DNA or baker's yeast RNA.

Specificity Study of QUID−2 towards RNA In Vitro
Given the promising results obtained above, we further evaluated the fluorescence response of QUID−2 for RNA in detail. First, we investigated the fluorescence response of QUID−2 (1.0 µM) to baker's yeast RNA or salmon testes DNA (200 µg/mL) using fluorescence spectroscopy. As shown in Figure 2A, QUID−2 displayed negligible emission in the absence of nucleic acid. This can be explained by the nonradiative energy loss through free rotation around the methine bridge between the indole and N-methylated quinoline moieties, as typically observed for various monomethine cyanine probes [32]. The addition of baker's yeast RNA could significantly enhance the fluorescence signal of QUID−2 at approximately 555 nm. In addition, we observed a hyperchromicity and a red shift of the absorption band of QUID−2 when bound to RNA ( Figure S5), which suggests the intercalative binding mode of QUID−2 for RNA. Because QUID−2 has little tendency to aggregate in a buffer ( Figure S6), the observed turn-on response would arise from the suppressed rotation upon intercalation into the base pairs in the RNA. This assumption was further supported by the correlation between fluorescence intensity and solvent viscosity ( Figure S7) [33].
The fluorogenic properties (I RNA /I free , where I RNA and I free denote the fluorescence intensities in the presence and absence of RNA, respectively) of QUID−2 increased 120fold upon binding to RNA, whereas an increase of only 34-fold was observed upon binding to DNA (I DNA /I free , where I DNA and I free denote the fluorescence intensities in the presence and absence of DNA, respectively). Notably, the value of RNA/DNA specificity (I RNA /I DNA ) of QUID−2 was higher than the commercial probe SYTO RNASelect (Figure 2A-C). These results suggest that QUID−2 exhibited high selectivity toward RNA versus DNA. To further verify the selectivity of QUID−2, the fluorescent properties of QUID−2 interacting with RNA and DNA were investigated by fluorescence titration. As shown in Figure 2D-F, with the gradual addition of the baker's yeast RNA, an emission peak at approximately 555 nm was significantly enhanced. Nevertheless, negligible fluorescence enhancement was observed when titrating QUID−2 with salmon testes DNA. Collectively, these results demonstrated that QUID−2 specifically bind to RNA and showed much better RNA specificity than SYTO RNASelect. The fluorogenic properties (I RNA /I free , where I RNA and I free denote the fluorescence intensities in the presence and absence of RNA, respectively) of QUID−2 increased 120-fold upon binding to RNA, whereas an increase of only 34-fold was observed upon binding to DNA (I DNA /I free , where I DNA and I free denote the fluorescence intensities in the presence and absence of DNA, respectively). Notably, the value of RNA/DNA specificity (I RNA /I DNA ) of QUID−2 was higher than the commercial probe SYTO RNASelect (Figure 2A-C). These results suggest that QUID−2 exhibited high selectivity toward RNA versus DNA. To further verify the selectivity of QUID−2, the fluorescent properties of QUID−2 interacting with RNA and DNA were investigated by fluorescence titration. As shown in Figure 2D-F, with the gradual addition of the baker's yeast RNA, an emission peak at approximately 555 nm was significantly enhanced. Nevertheless, negligible fluorescence enhancement was observed when titrating QUID−2 with salmon testes DNA. Collectively, these results demonstrated that QUID−2 specifically bind to RNA and showed much better RNA specificity than SYTO RNASelect.

Specificity Study of QUID−2 towards RNA against Other Nucleic Acids in Cells
Encouraged by its desirable detection limit and specificity for RNA in vitro, we sought to evaluate the ability of QUID−2 to specifically detect RNA in cells. Accordingly, confocal fluorescence microscopy was employed to assess the cellular imaging performance of QUID−2. Fixed HeLa cells were used in this test, and SYTO RNASelect was also tested as a control under identical measurement conditions ([probe] = 0.5 µM, 15 min incubation). As shown in Figure 3A, fixed HeLa cells stained with 0.5 µM QUID−2 exhibited distinguishable bright nucleolar, cytoplasm staining, and faint nucleus. The staining patterns and distributions of QUID−2 were very similar to that of SYTO RNASelect ( Figure 3B) and other RNA fluorescent probes, suggesting that QUID−2 selectively detect RNA in the cell. Notably, in sharp contrast to SYTO RNASelect, QUID−2 displayed much brighter fluorescent emission in cells ( Figure 3C). This is probably due to its superior selectivity and sensitivity for RNA over SYTO RNASelect. cubation). As shown in Figure 3A, fixed HeLa cells stained with 0.5 μM QUID−2 exhibited distinguishable bright nucleolar, cytoplasm staining, and faint nucleus. The staining patterns and distributions of QUID−2 were very similar to that of SYTO RNASelect ( Figure  3B) and other RNA fluorescent probes, suggesting that QUID−2 selectively detect RNA in the cell. Notably, in sharp contrast to SYTO RNASelect, QUID−2 displayed much brighter fluorescent emission in cells ( Figure 3C). This is probably due to its superior selectivity and sensitivity for RNA over SYTO RNASelect. To further confirm the specificity of our probe towards RNA in cells, deoxyribonuclease (DNase) and ribonuclease (RNase) digestion experiments were carried out. In the DNase experiment, only the DNAs in the cell would be hydrolyzed. By contrast, in the RNase experiment, only RNAs would be hydrolyzed. Fixed-permeabilized HeLa cells were used in this test, and SYTO RNASelect was also selected as a control. As shown in Figure 4A, upon DNase I treatment, no fluorescence signal loss in either cytoplasm or nucleoli stained with QUID−2 was observed. In contrast, upon RNase A treatment, the originally fluorescence signal from cytoplasm and nucleoli region that stained with QUID−2 was dramatically disappeared. However, in the control group, the fluorescent signal of SYTO RNASelect after DNase I treatment was decreased because SYTO RNASelect also bound with DNA in cells ( Figure 4B) [15]. These results evidently indicate that QUID−2 is a more RNA-specific probe than SYTO RNASelect for imaging of RNA in cells. To further confirm the specificity of our probe towards RNA in cells, deoxyribonuclease (DNase) and ribonuclease (RNase) digestion experiments were carried out. In the DNase experiment, only the DNAs in the cell would be hydrolyzed. By contrast, in the RNase experiment, only RNAs would be hydrolyzed. Fixed-permeabilized HeLa cells were used in this test, and SYTO RNASelect was also selected as a control. As shown in Figure 4A, upon DNase I treatment, no fluorescence signal loss in either cytoplasm or nucleoli stained with QUID−2 was observed. In contrast, upon RNase A treatment, the originally fluorescence signal from cytoplasm and nucleoli region that stained with QUID−2 was dramatically disappeared. However, in the control group, the fluorescent signal of SYTO RNASelect after DNase I treatment was decreased because SYTO RNASelect also bound with DNA in cells ( Figure 4B) [15]. These results evidently indicate that QUID−2 is a more RNA-specific probe than SYTO RNASelect for imaging of RNA in cells.

Imaging of RNA Dynamics in Live Cells
Mounting evidence has demonstrated that RNAs exhibit complex dynamics in live cells and the complex spatiotemporal dynamics of RNA molecules affect diverse cellular function [34,35]. However, the global dynamics of RNA in live cells have rarely been studied, leading to an incomplete understanding of the complex spatiotemporal dynamics of RNA molecules. Therefore, we sought to explore the capacity of QUID−2 in real-time monitoring the dynamics of RNA molecules in live cells.

Imaging of RNA Dynamics in Live Cells
Mounting evidence has demonstrated that RNAs exhibit complex dynamics in live cells and the complex spatiotemporal dynamics of RNA molecules affect diverse cellular function [34,35]. However, the global dynamics of RNA in live cells have rarely been studied, leading to an incomplete understanding of the complex spatiotemporal dynamics of RNA molecules. Therefore, we sought to explore the capacity of QUID−2 in real-time monitoring the dynamics of RNA molecules in live cells.
Upon QUID−2 treatment ([probe] = 0.5 μM, 30 min incubation), the live HeLa cells exhibited strong green fluorescent signal, with a distribution similar to that in fixed cells ( Figure 5A), indicating that QUID−2 selectively detect RNA in live cells. To ensure the practicability of QUID−2 during long-term cell imaging, the cytotoxicity and photostability of the probe were further investigated. The cytotoxicity of QUID−2 was evaluated in HeLa cells by standard MTT assay after 48 h of incubation. As shown in Figure 5B, QUID−2 in the concentration range of 0 to 40 μM exerted negligible toxicity to HeLa cells, suggesting that QUID−2 was safe at the working concentration of 0.5 μM for long term imaging in live cells. To study the photostability of QUID−2 in live cells, continuous laser irradiation was applied over an extended period. SYTO RNASelect was also tested as a control under identical measurement conditions. After 300 s of continuous irradiation, the fluorescence signals of SYTO RNASelect decreased by 70%. By contrast, QUID−2 did not show significant loss of the fluorescence signals ( Figure 5C). These results indicate that QUID−2 has satisfactory photostability and is suitable for long-term cell imaging.  Figure 5B, QUID−2 in the concentration range of 0 to 40 µM exerted negligible toxicity to HeLa cells, suggesting that QUID−2 was safe at the working concentration of 0.5 µM for long term imaging in live cells. To study the photostability of QUID−2 in live cells, continuous laser irradiation was applied over an extended period. SYTO RNASelect was also tested as a control under identical measurement conditions. After 300 s of continuous irradiation, the fluorescence signals of SYTO RNASelect decreased by 70%. By contrast, QUID−2 did not show significant loss of the fluorescence signals ( Figure 5C). These results indicate that QUID−2 has satisfactory photostability and is suitable for long-term cell imaging.
Considering the favorable specificity, sensitivity, and performance of QUID−2, we then employed QUID−2 to monitor the dynamics of RNA in live cells. Consistent with the dynamic nature of RNA [1,34,35], we observed complex spatiotemporal dynamics of QUID−2 fluorescence signals in live HeLa cells ( Figure 6A and Supplementary Materials Movie S1). In addition, we found some distinguishable bright QUID−2 foci in the cytoplasm. These bright QUID−2 foci were found to move in the cytoplasm, and several merging and splitting events were seen (Region 3, 4 in Figure 6A and Movie S1). Furthermore, we observed that during the measurement period some of the foci gradually disappeared, while new foci arose (Region 1, 2 in Figure 6A and Movie S1). This finding was similar to the results observed for RNA granules in live cells [36,37], suggesting that our probe might have the potential to detect RNA granules in live cells. To further test the practicability of using QUID−2 to track the dynamics of RNA granules, arsenite (NaAsO 2 ) was used to induce RNA granules (including SGs and P-bodies) formation [38]. As shown in Figure 7, upon exposure to arsenite (500 µM), QUID−2 foci increased in both size and number. In addition, we compared the localization of QUID−2 with BFP-G3BP1-labeled SGs (SGs are the most widely studied RNA granules and G3BP1 is a marker for SGs) in live HeLa cells treated with 500 µM sodium arsenite [36]. As expected, the fluorescent Considering the favorable specificity, sensitivity, and performance of QUID−2, we then employed QUID−2 to monitor the dynamics of RNA in live cells. Consistent with the dynamic nature of RNA [1,34,35], we observed complex spatiotemporal dynamics of QUID−2 fluorescence signals in live HeLa cells ( Figure 6A and Supplementary Materials Movie S1). In addition, we found some distinguishable bright QUID−2 foci in the cytoplasm. These bright QUID−2 foci were found to move in the cytoplasm, and several merging and splitting events were seen (Region 3, 4 in Figure 6A and Movie S1). Furthermore, we observed that during the measurement period some of the foci gradually disappeared, while new foci arose (Region 1, 2 in Figure 6A and Movie S1). This finding was similar to the results observed for RNA granules in live cells [36,37], suggesting that our probe might have the potential to detect RNA granules in live cells. To further test the practicability of using QUID−2 to track the dynamics of RNA granules, arsenite (NaAsO2) was used to induce RNA granules (including SGs and P-bodies) formation [38]. As shown in Figure 7, upon exposure to arsenite (500 μM), QUID−2 foci increased in both size and number. In addition, we compared the localization of QUID−2 with BFP-G3BP1-labeled SGs (SGs are the most widely studied RNA granules and G3BP1 is a marker for SGs) in live HeLa cells treated with 500 μM sodium arsenite [36]. As expected, the fluorescent probe QUID−2 accumulated in SGs and can be used to label SGs in live cells ( Figure S8). Collectively, these results confirmed that QUID−2 can detect RNA granules in live cells.

Conclusions
In summary, we have successfully developed a highly selective and sensitive fluorescent probe called QUID−2 for RNA imaging through the structural modification of precursor probe QUID−1. The detection limit of QUID−2 for the RNA was up to 1.8 ng/mL in solution. This value was significantly improved in comparison with its parent QUID−1, which had a detection limit of 26.5 ng/mL. We further demonstrated many advantages of QUID−2 over a commercially available RNA staining probe, SYTO RNASelect, for selective and sensitive RNA sensing by a systematic comparison of fluorescent properties for RNA. Moreover, QUID−2 possesses excellent photostability and low cytotoxicity in live cells, which is beneficial for long-term cell imaging. Thanks to many advantages of

Conclusions
In summary, we have successfully developed a highly selective and sensitive fluorescent probe called QUID−2 for RNA imaging through the structural modification of precursor probe QUID−1. The detection limit of QUID−2 for the RNA was up to 1.8 ng/mL in solution. This value was significantly improved in comparison with its parent QUID−1, which had a detection limit of 26.5 ng/mL. We further demonstrated many advantages of QUID−2 over a commercially available RNA staining probe, SYTO RNASelect, for selective and sensitive RNA sensing by a systematic comparison of fluorescent properties for RNA. Moreover, QUID−2 possesses excellent photostability and low cytotoxicity in live cells, which is beneficial for long-term cell imaging. Thanks to many advantages of QUID−2, the global dynamics of RNA were revealed in live cells. More importantly, QUID−2 was found to be potentially applicable for detecting RNA granules in live cells. Together, our work provides an ideal small-molecule probe for highly selective and sensitive RNA imaging. We anticipate that the further application of QUID−2 in combination with biological and imaging methods to explore RNA and RNA granules dynamics in live cells will uncover more information about the biological roles of RNA molecules.

Materials
SYTO RNASelect (Thermo Fisher Scientific, S32703) was purchased from Thermo Fisher (Waltham, MA, USA). Salmon testes DNA (Sigma, D1626) and baker's yeast RNA (Sigma, R6750) were purchased from Sigma (Saint Louis, MO, USA). RNase A and DNase I were purchased from Ambion (Austin, TX, USA). Hoechst 33,342 and DAPI were purchased from Sigma (Saint Louis, MO, USA). Sodium arsenite was purchased from Sigma (Saint Louis, MO, USA). TE buffer was purchased from Sangon Biotech (Shanghai, China). The HeLa cells used in this study were obtained from the cell bank of Sun Yat-Sen University Experimental Animal Center (Guangzhou, China).

Synthesis
1 H and 13 C NMR spectra were recorded on a Bruker Ascend TM 400 or Ascend TM 500 spectrometer (Karlsruhe, Germany). Mass spectra (MS) were recorded on a Shimadzu LCMS-2010A instrument (Tokyo, Japan) with an ESI or ACPI mass selective detector and high-resolution mass spectra (HRMS) were recorded on a Shimadzu LCMS-IT-TOF (Tokyo, Japan). Flash column chromatography was performed with silica gel (200-300 mesh) purchased from Qingdao Haiyang Chemical Co. Ltd. (Qingdao, China). All chemicals were purchased from commercial sources unless otherwise specified. All the solvents were of analytical reagent grade and were used without further purification. QUID−1 was synthesized according to our previous report.
Synthesis of intermediate 4: To a solution of indole-3-carboxaldehyde (4 g, 27 mmol) and anhydrous K 2 CO 3 (3.8 g, 27 mmol) in acetonitrile (20.0 mL), 1,3-dibromopropane (28 g, 135 mmol) was added. The resulting mixture was heated to reflux for 12 h until the reaction was complete, and the solid was filtered away. After that, the remaining solution was concentrated under reduced pressure. The crude material was purified by column chromatography (DCM/PE 1:1) to give the product as colorless oil (4, 2.93 g, yield 40%). 1  To a solution of indole-3-carboxaldehyde (14.5 g, 100 mmol,) and anhydrous K 2 CO 3 (13.8 g, 100 mmol) in acetonitrile (40.0 mL), ethyl 4-bromobutyrate (19.5 g, 100 mmol) was added. The resulting mixture was heated to reflux for 12 h until the reaction was complete, and the solid was filtered away. After that, the remaining solution was concentrated under reduced pressure. The crude material was purified by column chromatography (DCM/PE 1:1) to give the product as colorless oil (5, 16.6 g, yield 64%). 1  To a solution of indole-3-carboxaldehyde (1.45 g, 10 mmol,) and anhydrous K 2 CO 3 (1.38 g, 10 mmol) in acetonitrile (5.0 mL), 3-bromo-1-propanol (1.39 g, 10 mmol) was added. The resulting mixture was heated to reflux for 12 h until the reaction was complete, and the solid was filtered away. After that, the remaining solution was concentrated under reduced pressure. The crude material was purified by column chromatography (DCM/PE 1.5:1) to give the product as colorless oil (6, 1.05 g, yield 52%). 1  Synthesis of intermediate 7: The compound 5 (2.6 g, 10 mmol) was dissolved in 20 mL 10% NaOH solution, and refluxed for 3 h (monitored by TLC). The mixture was allowed to cool to room temperature, then concentrated hydrochloric acid was added until the pH of the system was adjusted to about 3. The precipitate was filtered and dried under vacuum to afford white solid (1.39 g, 60%). 1  To a solution of 4 (1.33 g, 5 mmol,) and anhydrous K 2 CO 3 (0.69 g, 5 mmol) in acetonitrile (5.0 mL), 1-methylpiperazine (1 g, 10 mmol) was added. The resulting mixture was heated to reflux for 12 h until the reaction was complete, and the solid was filtered away. After that, the remaining solution was concentrated under reduced pressure. The crude material was purified by column chromatography (DCM/OH 80:1) to give the product as colorless oil (8, 0.63 g, yield 45%). 1

pKa Calculations
The pKa of 1-methylpiperazine side group was calculated using ChemBioDraw 14.0.

UV-Vis Absorption Studies
UV-Vis absorption studies were performed on a UV-2600 spectrophotometer (Shimadzu, Tokyo, Japan) using 1 cm path length quartz cuvette and the UV-Vis spectra in the range of 200-750 nm were recorded.

Fluorescence Studies
Fluorescence studies were performed on Fluoromax-4 Spectrofluorometer (HORIBA, Kyoto, Japan). A quartz cuvette with 2 mm × 10 mm path length was used for the spectra recorded at 5 nm excitation and emission slit widths unless otherwise specified. For the emission spectra of QUID−1-QUID−6 or SYTO RNASelect (1 µM) in TE buffer (pH 7.8-8.2), the fluorescence spectra in the range of 485-750 nm were recorded when excited at 470 nm. For titration experiments, small aliquots of a stock solution of baker's yeast RNA or salmon testes DNA were added into the solution containing probes (1 µM) in TE buffer (pH 7.8-8.2). After each addition of sample, the reaction was stirred and allowed to equilibrate for at least 2 min. The fluorescence spectra in the range of 485-750 nm was recorded when excited at 470 nm. The LOD values of probes for baker's yeast RNA in solution were calculated on the basis of the equation LOD = K × S b /m. The K value is generally taken to be 3 according to the IUPAC recommendation. The S b value represents the standard deviation for multiple measurements (n = 20) of blank solution. The m value is the slope of the calibration curve, which was derived from the linear range of a probe fluorescence titration curve with baker's yeast RNA and standards for the sensitivity of this method. The absolute fluorescence quantum yield (Φ F ) of QUID−1-QUID−6 was measured by Quanta-Phi module for Fluoromax-4 Spectrofluorometer.

Fixed Cell Staining Experiments
The HeLa cells were grown in MEM media containing 10% fetal bovine serum and cultured at 37 • C in a CO 2 /air (5%/95%) incubator. Cells were seeded in a glass-bottom 96-well plate (Thermo Fisher Scientific) and grew overnight. Cells were fixed with 4% paraformaldehyde in DEPC-PBS at room temperature for 15 min. After being washed with PBS for 5 min (3×), cells were stained with 0.5 µM probe (QUID−2 or SYTO RNASelect) and 0.5 µg mL −1 DAPI for 15 min at 37 • C. For deoxyribonuclease (DNase) and ribonuclease (RNase) digestion experiments, cells were permeabilized with 0.5% Triton X-100 in DEPC-PBS at 37 • C for 20 min, and then incubated with 200 units mL −1 enzymes (RNase A or DNase I) before staining. Samples were finally imaged on an FV3000 laser scanning confocal microscope (Olympus, Tokyo, Japan) with a 60× objective lens. The emission of probes (QUID−2 and SYTO RNASelect) was collected under excitation at 488 mm. The images were analyzed with Imaris software (Bitplane, Zurich, Switzerland).

Cell Viability Assay
The HeLa cells were grown in MEM media containing 10% fetal bovine serum and cultured at 37 • C in a CO 2 /air (5%/95%) incubator. HeLa cells were seeded in 96-well plates (5.0 × 10 3 cells per well) and exposed to various concentrations of QUID−2. After 48 h treatment, 20 µL of 2.5 mg/mL methylthiazolyl tetrazolium (MTT) solution was added to each well, and cells were further incubated for 4 h. The cells in each well were then treated with DMSO (100 µL per well), and the optical density (OD) was recorded at 490 nm. All experiments were performed in parallel and in triplicate, and the IC 50 values were derived from the curves of the mean OD values of the triplicate tests plotted against the drug concentration.

Plasmid Transfection
cDNAs encoding BFP-G3BP1 were obtained by gene synthesis from Convenience Biology (Changzhou, China). To generate BFP-tagged G3BP1 protein, cDNAs were inserted into the pcDNA3.1 vector and the construct was confirmed by sequencing. Plasmid transfection of cells was performed using Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Then, 100 ng of constructed plasmid were used per well of a glass bottom 96-well plate (Thermo Fisher Scientific) and incubated at 37 • C with CO 2 for 24 h.

Live Cell Staining Experiments
The HeLa cells were grown in MEM media containing 10% fetal bovine serum and cultured at 37 • C in a CO 2 /air (5%/95%) incubator. Cells were seeded in a glass-bottom 96-well plate (Thermo Fisher Scientific) and grew overnight. The cells were then stained with QUID−2 (0.5 µM) in FluoroBrite DMEM (Gibco, A1896701) in a 5% CO 2 atmosphere at 37 • C for 20 min, and then rinsed by FluoroBrite DMEM. For sodium arsenite treatment, cells were treated with (500 µM) sodium arsenite for 1 h before imaging. For co-staining with BFP-G3BP1, HeLa cells expressing BFP-G3BP1 were stained with 0.5 µM QUID−2 for 20 min and then stressed with 500 µM sodium arsenite for 1 h. Digital images were recorded using a FV3000 laser scanning confocal microscope (Olympus, Tokyo, Japan) with a 60× objective lens and analyzed with Imaris software (Bitplane, Zurich, Switzerland). The emission of QUID−2 was collected under excitation at 488 mm.

Conflicts of Interest:
The authors declare no competing interests.

Sample Availability:
The samples of the compounds used and/or analyzed during the current study are available from the corresponding author upon reasonable request.