Development of a red fluorescent light-up probe for highly selective and sensitive detection of vicinal dithiol-containing proteins in living cells

An environment-sensitive red fluorescent light-up probe for vicinal dithiol-containing proteins (VDPs) in living cells has been successfully developed.


Introduction
Vicinal dithiol-containing proteins (VDPs) are proteins that contain two space-closed cysteine (Cys)-sulydryl groups. In general, the vicinal thiols are derived from cysteines that are sequence proximal but may be formed by the juxtaposition of cysteinyl residuespresent in distant segments of the same or different polypeptides. 1 As the reductive end of this redox buffer network, VDPs can undergo reversible oxidative conversion to the intra-or interprotein disuldes and are highly effective in regulating the redox environment of internal cellular compartments under normal conditions. 2 In addition, VDPs hold a particularly prominent position in protein synthesis and function, and are responsible for many diseases such as cancer, 3 diabetes, 4 stroke, 5 and neurodegeneration. 6 Therefore, to explore the essential roles of VDPs in cellular redox homeostasis and protein function in living cells, there is a strong desire to develop uorescent probes for the sensitive and selective sensing of VDPs in living cells.
In recent years, two strategies have been adopted for the development of uorescent probes for VDPs. One method employs two maleimide groups as the recognition unit, which quenches the probe's uorescence until they both undergo thiol addition during the labeling reaction. These uorogenic probes have been used to label vicinal thiol-containing peptides/ proteins. 7 However, intracellular labeling still remains a challenge for these probes since intracellular glutathione (GSH) (1-10 mM) can undergo a similar addition reaction, thus leading to a nonspecic uorescent labeling reaction. 7b Alternatively, 1,3,2-dithiarsenolane was incorporated into a variety of uorophores to develop uorescent probes for VDPs which have been used to identify VDPs in living cells. 8 The method employs the fact that trivalent arsenicals can bind to vicinal thiol proteins with high affinities, 9 whereas proteins with thiols that are not vicinal (referred to as monothiols) interact weakly with arsenicals. 10 However, a potential drawback of these uorescent probes is the strong background uorescence from the unreacted probes inside cells which hinders the identication of labeled proteins. 8a-c To circumvent this problem, a tedious washing procedure (>15 min) is required to remove the unbound probes in order to reduce the background uorescence, which will inevitably delay the acquisition of microscopic data and thus makes the measurements prone to artifacts. To overcome this deciency, a ratiometric uorescent probe for VDPs was recently developed by Huang et al. based on the uorescence resonance energy transfer (FRET) mechanism. 8d Although ratiometric probes can overcome the inuence of a variety of factors such as instrumental efficiency, environmental conditions and the probe concentration, the proposed probe exhibits only moderate uorescence variations (ca. 6-fold) upon binding to VDPs. Furthermore, the probe shows a sluggish response to VDPs (>60 min), which makes it unsuitable for the real time monitoring of VDPs inside living cells. Therefore, it is highly desirable to develop uorescence turn-on probes for intracellular VDPs that have the combination of high selectivity, fast labeling rates, and high uorescent turn-on ratios.
Recently, a simple strategy for designing uorescence turnon probes for selective protein detection has been developed by taking advantage of environment-sensitive uorophores, 11 which generally involves the incorporation of an environmentsensitive uorophore into a ligand specic to the target protein.
Typically, these probes exhibit very weak uorescence in polar and protic environments, while the uorescence is enhanced when the environment becomes hydrophobic or less polar. They could provide a uorogenic response to their immediate environment, resulting in a variety of applications in bioanalytical chemistry. This light-up strategy has paved a new way for the detection of targeting proteins with high sensitivity and selectivity.
Although several environment-sensitive uorophores have been reported, 12 they show some crucial drawbacks. First, most of those used for light-up probe design are limited to those with blue or green emission, whereas far-red and near-infrared (NIR) dyes are advantageous for cellular studies due to lower photodamage, light scattering, and autouorescence in living systems. 13 Second, sensitivity to solvent polarity of these dyes is frequently not enough to detect subtle changes in the environment of the biomolecule of interest. Finally, polarity-sensitive dyes with red emission generally exhibit relatively lower sensitivity to polarity compared to blue dyes. 12b Therefore, currently intensive research is focusing on the design of environmentsensitive uorophores with red-shied emission.
Herein, we report the design and synthesis of a red uorescent light-up probe for the rapid detection of VDPs both in vitro and in vivo with excellent sensitivity and specicity. In the proposed sensing system, 2-(4-dimethylaminophenyl)-4-(2-carboxyphenyl)-7-diethylamino-1-benzopyrylium (F1) was selected as the environment-sensitive uorescence reporter and cyclic dithiaarsane as the specic ligand for VDPs. Our rationale is depicted in Scheme 1. We envisioned that the selective binding of protein vicinal dithiols to the trivalent arsenical of FAsH would bring the uorophore into the hydrophobic protein domain, and the hydrophobic environment would cause the uorophore to emit strong uorescence. In contrast, in the absence of the target protein, the probe would remain in aqueous solution and should emit only weak uorescence. Based on the above mechanism, we created a selective uorescence turn-on probe toward VDPs inside living cells with nowash procedures. Compared with the reported uorescent probes, FAsH is cell-permeable and shows a rapid response toward VDPs with high sensitivity. Moreover, the proposed probe can operate in the red region, which is favorable for biological applications in vitro and in vivo. The proposed probe has been used for rapid no-wash imaging of VDPs in living cells.

Probe design
In this work, we constructed a novel uorescent light-up probe for the selective detection of VDPs inside living cells by adopting a uorogenic mechanism based on an environment-sensitive uorophore. To obtain an optimal response, there is a strong need to select environment-sensitive dyes presenting both high environment-sensitivity and good uorescence properties. In this contribution, we focused our interest on a avylium uorescent dye F1, because it can be regarded as an electron acceptor system (benzopyrylium cation) connected to two different electron donor units (dimethylaminophenyl and diethylamino groups) ( Fig. 1), which is a typical characteristic of Scheme 1 (a) Proposed reaction between FAsH and rBSA; (b) schematic illustration of the fluorescence turn-on mechanism for rBSA detection with FAsH. representative solvatochromic uorescent molecules. 14 Although the biological applications of benzopyrylium dyes have been shown recently, 15 their environment-sensitive behavior remains unexplored.
We then studied the solvatochromic properties of F1 by measuring its absorption and emission spectra in different proportions of water and 1,4-dioxane with different polarities. As shown in Fig. S1 (ESI †), all the absorption spectra have maxima at about 598 nm, and there are no signicant changes observed in the solutions with different polarities. In contrast, solvent polarity had a dramatic effect on the emission spectra of F1. When the orientation polarizability (Df) of the solution decreased from 0.32 (99% water) to 0.292 (20% water), 16 the maximum emission wavelength of F1 shied from 646 to 635 nm, concomitant with a gradual increase in uorescence intensity (Fig. 2a). The uorescence intensity of F1 at 635 nm increased by a factor of 14.2. The above results reveal that F1 is a polar-sensitive (solvatochromic) uorescent dye.
Moreover, the multiple electron-donating groups in F1 are linked to the benzopyrylium unit via a single bond. Thus, it features high rotational exibility and possesses two different twisted intramolecular charge transfer (TICT) channels within the whole molecule (involving twisting of the dimethylaminophenyl and diethylamino groups, respectively, as shown in Fig. 1). 17 These intramolecular rotations lead to the nonradiative deactivation of the uorescent excited state, which might be another cause for the uorescence quenching of F1 in aqueous solution. To test this assumption, we checked the effect of solvent viscosity on F1 emission. The intramolecular rotation process is reported to be inuenced by the viscosity of the medium: the higher the viscosity of the medium, the slower the intramolecular rotation and hence the stronger the F1 emission. 18 We then evaluated the viscosity effect on the emission behavior of F1 in methanol/glycerol mixtures with different fractions of glycerol (f g ). As shown in Fig. 2b, the emission intensity of F1 is greatly enhanced as the solvent viscosity increases from 0.60 (methanol) to 950 cP (99% glycerol) at room temperature (25 C), 19 which is typically observed for molecular rotors. These experimental results support the fact that F1 is a viscosity-sensitive dye.
Since F1 features a dual dependency of emission intensity on both solvent polarity and viscosity, we thus expect that it would hold great promise in the development of environment-sensitive uorescent probes compared to the traditional solvatochromic uorescent dyes. Next, we checked the possible nonspecic interactions of F1 with serum proteins by introducing bovine serum albumin (BSA, 1.0 mg mL À1 ) to the aqueous solution of F1. As shown in Fig. S2 (ESI †), only negligible changes in the emission intensity (2.6-fold increase) of F1 were observed, suggesting there are few nonspecic interactions between F1 and BSA. This is crucial for the development of probes for in vivo imaging, as a high nonspe-cic background signal is the main reason for their failure. Finally, F1 contains a carboxylic acid group, which enables facile attachment of the recognition moiety. On the basis of the aforementioned results, F1 was selected as the uorophore to construct the probe.
Additionally, we selected 2-(4-aminophenyl)-1,3,2-dithiarsolane (PAO-EDT) as the recognition unit because its As(III) center can selectively discriminate vicinal dithiols from other forms of thiols through the interchange of 1,2-ethanedithiol (EDT) in cyclic dithiaarsanes with vicinal dithiols in proteins. 20 In addition, its 5-membered dithiarsolane ring is a more stable complex compared with the 6-and 7-membered ones. 21 In view of the above mentioned results, we rationally designed a red  uorescence turn-on probe FAsH for VDPs in living cells. The detailed synthetic procedures and characterization of FAsH are shown in Scheme 2. Meanwhile, F4 which lacks the 5membered dithiarsolane ring in its structure was also prepared for comparison purposes.

General spectral properties
With FAsH in hand, we studied its spectral properties. The absorption maximum of FAsH is located at 612 nm in aqueous solution (3 612 nm ¼ 2.68 Â 10 4 M À1 cm À1 ), which is red-shied by about 17 nm in comparison with that of F1 (Fig. S3a, ESI †). Furthermore, the emission spectra of FAsH were recorded in CH 2 Cl 2 and the results are shown in Fig. S3b (ESI †). In agreement with the red-shi in the absorption spectra, the uorescence spectrum of FAsH displays a 10 nm bathochromic shi when compared to F1. The optical spectra of F4 were almost the same as those of FAsH under identical conditions. Furthermore, it was observed that the uorescence intensity of FAsH shows no signicant difference to that of F4 (Fig. S3b, ESI †), indicating that no intramolecular uorescence quenching was induced by the arsenical moiety, which is apparently different from that seen in bisarsenical dyes. 20,22 Optical response The uorescence sensing behavior of FAsH toward VDPs was examined. Here, reduced bovine serum albumin (rBSA) was selected as the model protein because it has eight vicinal Cys pairs (its structure is shown in Fig. S4, ESI †) aer BSA is reduced with tris(2-carboxyethyl)phosphine (TCEP). 7b Initially, the kinetic behavior of FAsH toward rBSA was examined in phosphate buffer (20 mM, pH 7.4, containing 1% acetone as cosolvent). Upon introducing rBSA (1.0 equiv.) to the solution of FAsH, a dramatic increase in the uorescence intensity was observed within 2.5 min, which then reached a plateau as the reaction proceeded, whereas the uorescence background in the absence of rBSA remained unchanged under identical conditions (Fig. 3). The binding rate of FAsH to rBSA is dramatically accelerated in comparison with the previously reported probe 8d which is very important in monitoring the dynamic changes of VDPs in situ. Moreover, upon addition of EDT to the solution of the FAsH-rBSA complex, the uorescence intensity decreased by 42% within 1 min (Fig. S5, ESI †), which proves that the binding of FAsH with rBSA is reversible. 20 By contrast, BSA was added to the solution of FAsH and only a negligible uorescence increase was observed on the time scale of the experiments. These results indicate the high selectivity of FAsH toward vicinal dithiols in proteins.
The uorescence response of FAsH toward rBSA was examined by introducing increasing concentrations of rBSA (0-2.4 mM) to the solution of FAsH. As shown in Fig. 4, the free probe gives extremely weak uorescence in aqueous solution (4 f ¼ 0.006, using F1 in CH 2 Cl 2 as a reference). 23 However, the addition of an increasing amount of rBSA to the solution of FAsH elicits a gradual increase in the uorescence intensity and the nal enhancement factor is over 70-fold (4 f ¼ 0.21). This intensity increase was also accompanied by a hypsochromic shi in the emission spectra from 658 to 651 nm during the titration. The increase in the uorescence intensity and the hypsochromic shi of the uorescence emission maxima may be attributed to the binding of FAsH to the hydrophobic domain of rBSA. The uorescence intensity at 651 nm as a function of rBSA concentration was recorded, and a nearly linear relationship over the range of 0.06-0.9 mM was obtained (Fig. S6, ESI †). The detection limit (3d) for rBSA was calculated to be 0.015 mM. These results demonstrate that FAsH can detect rBSA with high sensitivity. Furthermore, to test the sensing behavior of FAsH toward different VDPs, we examined other reduced forms of proteins (human serum albumin, ovalbumin and lysozome) and found that reduced human serum albumin (rHSA) also induces a dramatic increment in emission intensity, while reduced ovalbumin affords a moderate uorescence enhancement. In the case of reduced lysozome, a very small increment in emission intensity is observed (Fig. S7, ESI †). This is apparently due to different VDPs having different reactivities with FAsH. Thus, FAsH is unsuitable for the quantitative determination of VDP content in complicated biological systems because different VDPs will afford different increments in emission intensity.

Selectivity studies
To further verify the selectivity of the probe for VDPs, FAsH was treated with a series of biologically relevant species in phosphate buffer (20 mM, pH 7.4). As shown in Fig. 5, only VDPs (rBSA, rHSA and reduced ovalbumin, 1 equiv. of each) induce a signicant uorescence increase, while other amino acids (Ile, Ser, Arg, Met, Thr, Tyr, Ala, Asp, His, Tpc, Pro, Gly, Glc, Trp, Glu, Lys, Leu, Val, Cys, Hcy, GSH, 100 equiv. of each), TCEP, ascorbic acid (100 equiv. of each) and proteins (cytochrome c, myoglobin, lysozyme and BSA, 1 equiv. of each) afford no obvious changes in emission intensity. Furthermore, FAsH can still retain its sensing behavior toward VDPs even in the presence of a large amount of biothiols (GSH, Cys and Hcy) or reductant (ascorbic acid) (Fig. S8, ESI †), which further conrms the high selectivity of FAsH towards VDPs over other cellular thiol-containing compounds. The specicity of FAsH toward VDPs was further veried by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). As shown in Fig. 6, a uorescence band was observed in the lane loaded with rBSA and FAsH, whereas no visible emission was observed in the case of BSA, which contains only one Cys residue (monothiol protein). Furthermore, no uorescence band was observed when control compound F4 was used for rBSA labeling, which provides strong evidence that the 5-membered dithiarsolane ring in FAsH is responsible for the specic binding of VDPs. By contrast, these protein bands were observed in the gel aer silver staining which proves that the uorescence band is related to the formation of the rBSA-FAsH complex. The excellent selectivity of FAsH toward VDPs can be explained by the fact that the cyclic dithioarsinite complexes (FAsH-rBSA) formed between trivalent arsenicals and vicinal thiols are markedly more stable than the noncyclic products formed from monothiols due to entropic considerations. 24

Mechanistic study
Some experiments were carried out to gain further insight into the uorogenic response of FAsH toward VDPs. The absorption spectra of FAsH with the introduction of rBSA were recorded, and it was observed that free FAsH exhibits two absorption bands at 448 and 612 nm. Upon addition of rBSA, a new blue-shied absorption band centered at 589 nm emerged, and all the absorption bands were increased with increasing rBSA concentration (Fig. S9, ESI †). The above spectra variation indicates that FAsH forms H-aggregates in the presence of rBSA. 25 Moreover, the aggregates' formation was evident from the trailing absorption trace of the probe solution in the presence of rBSA due to the light-scattering effects of the dye nanoparticles ( Fig. 7 and S9, ESI †). In the case of BSA, almost no spectral changes were observed, suggesting that there is almost no interaction between FAsH and BSA. These results indicate that rBSA can induce the aggregation of FAsH selectively. The aggregation was further conrmed using dynamic light scattering (DLS) measurements, and the solution of FAsH (2.0 mM) with rBSA (1.5 mM) shows an average particle size of 180.3 AE 7.3 nm (Fig. S10, ESI †). The above experimental results support the rBSA-induced aggregation of FAsH, which can then restrict the intramolecular rotation and thus lead to a great enhancement in the emission intensity. 26 Furthermore, to prove the uorescence enhancement of the sensing process is caused by the hydrophobic pocket of rBSA,   6 The selective binding of FAsH to VDPs was verified by SDS-PAGE. "+": the compound was present in the detection system; "À": the compound was absent in the detection system. guanidine hydrochloride (GdnHCl), a strong protein denaturant, was introduced into the solution of the FAsH-rBSA complex, and a signicant decrease in emission intensity was observed (Fig. S11, ESI †). This is apparently due to the unfolding of rBSA and the hydrophobic pocket in rBSA is destroyed. 27 As a result, the probe gets more exposed to the polar environment, which is undesirable for uorescence emission. The above results reveal the essential role of the hydrophobic cavities of the protein folding structure in the present sensing system. Collectively, we can conclude that the uorescence enhancement of the present sensing system is achieved by reducing the charge transfer between the uorophore and the polar media and restricting the intramolecular rotations via aggregation simultaneously (Scheme 1b).

Application of FAsH in biological systems
In order to show its application in biological samples, we examined the feasibility of FAsH for VDP assays in fetal bovine serum (FBS) samples. The introduction of FAsH to FBS solution induced a slight uorescence enhancement. However, upon adding dithiothreitol (DTT, 5 mM) to the above mixture, the uorescence intensity increased progressively (Fig. S12, ESI †). As a control, it was observed that DTT alone exhibits no uorescence increase in FAsH solution. The above experiments prove that FAsH can selectively respond to VDPs in biological samples.
Next, some experiments were performed to evaluate FAsH in live-cell imaging assays using human hepatoma cells (SMMC-7721) as a model cell line. Initially, the cytotoxicity of FAsH was evaluated using a standard MTT assay. Although PAO is quite toxic, the results showed that FAsH has minimal cytotoxicity at concentrations of 2-20 mM (Fig. S13, ESI †). This is apparently due to the EDT caging unit, which prevents FAsH from exerting acutely toxic effects. 20 Next, SMMC-7721 cells were incubated with FAsH for 20 min in PBS, and a strong uorescence signal was observed. Furthermore, in a control experiment, the cells were pretreated with 30 mM PAO (a selective VDP binding reagent) to reduce the amount of intracellular free VDPs prior to incubation with FAsH. A pronounced uorescence quenching was observed (Fig. S14, ESI †), which reveals that the above uorescence emission (Fig. 8) is indeed induced by VDPs. By contrast, F4 was used for cell staining and it affords negligible uorescence emission under the same conditions (Fig. 8). The semiquantitative calculation of the averaged uorescence intensity was further conducted. The emission intensity of FAsH-stained cells is about 19-fold higher than that of F4treated cells (Fig. S15, ESI †). The signicant difference in emission intensity indicates the selective binding of FAsH to endogenous VDPs inside live cells, and this selective binding is apparently due to the 5-membered dithiarsolane ring. These results demonstrate the capacity of FAsH for in situ imaging of VDPs in living cells.
To further investigate the subcellular localization of VDPs, a commercially available mitochondrial tracker (rhodamine 123) was used for a colocalization study with confocal microscopy. As displayed in Fig. 9, the observed uorescence signal from FAsH extensively overlaps with that of rhodamine 123, implying that FAsH-labeled VDPs are mainly localized to the mitochondria of these live cells. Furthermore, nuclear staining with DAPI indicates that the cells are viable throughout the imaging experiments. The above experiments prove that there is an abundance of VDPs distributed in the mitochondria of SMMC-7721 cells, which is consistent with previous studies. 8a,28

Conclusions
In summary, we have developed a uorescent light-up probe FAsH for selective detection of VDPs using a unique environment-sensitive avylium dye F1 as the uorescent reporter. The probe is almost non-uorescent in its free form, but exhibits strong uorescence emission upon specically binding to VDPs. Therefore, its application no longer requires tedious washing steps during protein labeling, which is favorable for the direct, noninvasive tracing of VDPs in a cellular redox environment. Compared with the widely used solvatochromic uorescent probes, FAsH possesses the characteristics of two types of environment-sensitive uorophores (molecule rotors and solvatochromic uorescent dyes) simultaneously. Thus, it shows a more sensitive light-up uorescence response toward VDPs. Preliminary imaging experiments reveal that FAsH can serve as a unique probe for the no-wash visualization of endogenous  VDPs in living cells. In addition, FAsH has the advantage of rapid binding kinetics, a high signal-to-noise ratio, red emission and will be an attractive tool for the in situ investigation of the essential role of VDPs in intracellular redox homeostasis and the exploration of its diverse pathophysiology.