Development of Dual-Reactivity Based Fluorescence Probes for Visualizing Intracellular Hydrogen Polysulfides Signaling

Abstract

Endogenous hydrogen polysulfides (H₂S_n, n > 1) have been recognized as important regulators in sulfur-related redox biology. H₂S_n can activate tumor suppressors, ion channels, and transcription factors with higher potency than H₂S. While H₂S_n are drawing increasing attention, their exact mechanisms of actions are still poorly understood. A major hurdle in this field is the lack of reliable and convenient methods for H₂S_n detection. In this work we report a H₂S_n-mediated benzodithiolone formation under mild conditions. This reaction takes advantage of the unique dual-reactivity of H₂S_n as both a nucleophile and an electrophile. It opens a door for better understanding H₂S_n's chemical biology. Based on this reaction, three fluorescent probes (PSP-1~3) were synthesized and evaluated. Among the probes prepared, PSP-3 showed a desirable off-on fluorescence response to H₂S_n and high specificity. It was successfully applied in visualizing intracellular H₂S_n.

Reactive sulfur species (RSS) are a group of sulfur-containing molecules that play regulatory roles in biological systems. Important RSS include biothiols and S-modified protein cysteine adducts. RSS also include hydrogen sulfide (H₂S) and sulfane sulfurs (such as cysteine persulfides (R-S-SH) and polysulfides (R-S-S_n-S-R', n > 0)). Understanding RSS's mechanisms of actions has become a very active research area in modern chemical biology, particularly from the methodological point of view.^[1] Among various RSS, hydrogen polysulfides (H₂S_n, n > 1) have attracted particular attention, mainly due to their involvement in H₂S-related redox biology.^[2] H₂S_n can be generated from endogenous H₂S upon reacting with reactive oxygen species (ROS).^[3] H₂S can also react with other sulfane sulfurs such as S₈ to form H₂S_n.^[2d,3c,4] Moreover, H₂S_n may have their own biosynthetic pathways. For example, while cystathionine γ-lyase (CSE)-mediated cystine metabolism can produce cysteine persulfide,^[5] it may also catalyze direct generation of H₂S_n from cystine. A series of recent works suggested that H₂S_n might be a new group of signaling molecules.^[2,3c,4] It was found that H₂S_n can activate tumor suppressors, ion channels, and transcription factors with higher potency than H₂S.^[4] Some physiological activities that were originally assigned to H₂S may actually be mediated by H₂S_n. One such example is S-sulfhydration.^[1a,5,6] This post-translational modification was previously thought to be the result of H₂S. However, recent results have demonstrated that H₂S_n are more effective than H₂S in S-sulfhydration.^[2,3b–c]

Although H₂S_n have now been recognized as potent physiological mediators, a number of issues remain to be clarified, such as the production and degradation pathways of H₂S_n, their regulatory mechanisms, and potential physiological stimuli that induce those regulatory mechanisms. To address these issues, it is critical to develop effective methods for H₂S_n detection. The traditional method is to measure UV absorption peaks at 290–300 nm and 370 nm, which has low sensitivity and is not applicable for biological detections. Fluorescence assays could be useful because of their high sensitivity and spatiotemporal resolution capability. While fluorescent probes for H₂S have been extensively studied,^[7] the probes for H₂S_n are still underdeveloped due to limited known chemistry of H₂S_n.

Our laboratory has initiated a program to study new chemistry/reactions of H₂S_n and develop reaction-based fluorescent probes for H₂S_n. In 2014, we reported the first H₂S_n-specific probes (DSP, Scheme 1) which employed a 2-fluoro-5-nitro-benzoic ester template to trap H₂S_n and promote an intramolecular cyclization to release a fluorophore.^[8] DSP probes, such as DSP-3, showed satisfactory sensitivity and selectivity for H₂S_n. Several other groups have adopted the same template to develop H₂S_n probes with interesting properties.^[9] However, a drawback of DSP is that 2-fluoro-5-nitro-benzoic esters can also react with biothiols. Although such reactions do not turn on fluorescence, the consumption of the probes is a problem. In theory, the reaction product, i.e. thioether 1, can further react with H₂S_n to switch on fluorescence. We found that the reaction was somewhat slow and non-productive when H₂S_n concentration was low (at μM levels). Therefore, high loading of DSP probes may be needed for biological detections. Very recently our lab discovered another H₂S_n-specific aziridine opening reaction and developed a probe (AP) based on this reaction.^[10] AP showed excellent selectivity for H₂S_n, it did not react with biothiols or H₂S in physiological concentrations. However, due to limited options for aziridine-based fluorophores, the sensitivity of AP for H₂S_n was not optimal and attempts to use AP for imaging H₂S_n in cells were not successful. Obviously further improvement of the probes is needed. Herein we report the design, synthesis, and evaluation of a series of new fluorescent probes (PSP1-3) that showed high sensitivity and selectivity for H₂S_n and can be used for endogenous H₂S_n imaging.

Scheme 1.

Previous designs of H₂S_n-specific fluorescent probes.

The key of developing optimum probes is to identify a highly specific H₂S_n recognition unit. Ideally, such a recognition unit should only react with H₂S_n and not react with other sulfur species, especially biothiols. From a chemistry perspective, H₂S_n are quite different from thiols or H₂S. The estimated pK_a values of H₂S_n are in the range of 3 to 5.^[11] For comparison, the pK_a values of H₂S and biothiols are in the range of 7 to 9.2. Under physiological pH, H₂S_n are expected to be weak acids, and stronger and more reactive nucleophiles than biothiols and H₂S due to alpha-effects. In addition, H₂S_n belong to the sulfane sulfur family. A character of sulfane sulfurs is that they can function as electrophiles and react with certain nucleophiles.^[12] Overall, H₂S_n have a unique dual-reactivity and can act as both nucleophiles and electrophiles. Taking advantage of this property, we envisioned template 2 might be specific for H₂S_n (Scheme 2). In 2, a thioester was employed to trap the nucleophilicity of H₂S_n. One might worry that the thioester group could also react with biothiols as the reactions between thioesters and thiols are known, for example in the well-known Native Chemical Ligation.^[13] However, those reactions are mostly for synthetic purposes with high concentrations of reactants and special solvent conditions. The relatively low concentrations of biothiols in biological systems and the mild, neutral, and aqueous environments may make the reactions slow and non-productive. In addition, we expected the manipulation of R groups (steric and electronic effects) in 2 could significantly differentiate its reaction rates toward thiols and H₂S_n. If H₂S_n could selectively react with an appropriate thioester group, the product, i.e. 3, should further react with H₂S_n (now serving as an electrophile) to form 4. The following spontaneous cyclization should release the fluorophore. Overall, this process would be specific for H₂S_n.

Scheme 2.

The design of dual-reactivity based probes.

Based on this idea, we synthesized three probes (PSP-1, PSP-2, and PSP-3, Scheme 3). Three different R groups (Me, tBu, and Ph) were used in order to explore the effects of acyl groups on the reactivity of thioesters toward biothiols. We first tested their fluorescence properties and responses to H₂S_n in PBS buffer. Freshly prepared Na₂S₂ solutions were used as the equivalents of H₂S_n. All three probes showed almost no fluorescence emission (Φ_PSP-1 = Φ_PSP-2 = 0.02; Φ_PSP-3 = 0.01) due to the protection of the two hydroxyl groups of fluorescein. Upon the treatment of Na₂S₂ (5 eq) for 30 min, the probes gave significant fluorescence enhancements (Figure 1A). PSP-1 and PSP-3 exhibited stronger fluorescence increases than PSP-2. Presumably the bulky t-butyl group of PSP-2 decreased the reactivity toward to H₂S_n. We also tested time-dependent fluorescence changes of the probes in the presence of Na₂S₂ (Figure 1B). The maximum emission intensities of PSP-1 and PSP-3 were reached within 2 min and 10 min respectively, indicating fluorescence turn-on was fast. The fluorescence turn-on of PSP-2 was slower (~20 min), again suggesting the steric effect played a role in its reactivity. For the purpose of reproducibility, a reaction time of 30 min was employed in all of the following experiments.

Scheme 3.

The structures of PSP probes.

Figure 1.

(A) Fluorescence intensity increases (ΔF) of PSP (10 μM) with Na₂S₂ (50 μM): (1) PSP-1; (2) PSP-2; (3) PSP-3. (B) Time-dependent fluorescence intensity changes of PSP (10 μM) in the presence of Na₂S₂ (50 μM).

Next we studied the stability of the probes in the presence of biothiols. As shown in Scheme 4, if the probes undergo thioester exchange with biothiols, the resultant product 5 should be sensitive to sulfane sulfurs like elemental sulfur (S₈) and produce strong fluorescence. Therefore, measuring fluorescence of the probes in a mixture of thiols and S₈ should be suitable for evaluating the stability of the probes to thiols.

Scheme 4.

Possible reactions of PSP probes with biothiols and S₈.

We then tested fluorescence responses of the probes (10 μM) to Cys (1 mM) and GSH (5 mM) in the absence or presence of S₈ (50 μM). As shown in Figure 2, Cys, GSH, or S₈ alone did not induce any responses. The mixtures of biothiols (GSH or Cys) and S₈ produced strong fluorescence for PSP-1. The same mixtures caused almost no fluorescence increase for PSP-2 and PSP-3. As a positive control, the thioester exchange product 5 gave strong fluorescence for S₈ under the same conditions. These results confirmed our hypothesis that the acyl group of thioester could affect its reactivity towards thiols. PSP-1 appeared to be labile to biothiols while PSP-2 and PSP-3 were stable. Therefore, the use of PSP-2 and PSP-3 should not be affected by cellular biothiols.

Figure 2.

Fluorescence increase value (ΔF) of each probe (10 μM) in the presence of biothiols and S₈: (a) Probe + 5 mM GSH; (b) Probe + 1 mM Cys; (c) Probe + 50 μM S₈; (d) Probe + 5 mM GSH + 50 μM S₈; (e) Probe + 1 mM Cys + 50 μM S₈; (f) 10 μM Compound 5 + 50 μM S₈.

Having demonstrated the excellent reactivity and stability of PSP-3, this probe was selected for further studies. We next validated its selectivity for H₂S_n over other common RSS including Hcy, GSSG, H₂S, SO₃²⁻, S₂O₃²⁻, SO₄²⁻, and CH₃SSSCH₃. As shown in Figure 3A, no fluorescence increase was observed for these RSS (columns 2–8). Only Na₂S₂ and Na₂S₄ triggered strong fluorescence enhancements (columns 9 and 10). Additionally, It is known that H₂S_n could be generated from the reactions between H₂S and ROS. Therefore, we wondered if the probe could sense in situ H₂S_n formation from H₂S and ROS. As shown in Figure 3B, the responses of PSP-3 to a series of ROS including hydrogen peroxide (H₂O₂), hypochlorite (ClO⁻), superoxide (O₂⁻), hydroxyl radical (•OH), and singlet oxygen (¹O₂) were first tested. These ROS did not induce any fluorescence enhancements (columns 2–6). When H₂S was added to these ROS, a strong fluorescence signal was observed from ClO⁻ (column 9), but not from other ROS. These results agreed with previous discoveries that H₂S_n can be efficiently derived from H₂S and ClO⁻, but other ROS are not effective for H₂S_n generation under these concentrations.^[3,8–10] The responses of PSP-3 in the presence of other representative amino acids were also evaluated and no responses were observed (Figure S1).

Figure 3.

(A) Fluorescence intensity of 10 μM PSP-3 in the presence of various RSS: (1) probe alone; (2) 100 μM Hcy; (3) 100 μM GSSG; (4) 200 μM Na₂S; (5) 100 μM Na₂S₂O₃; (6) 100 μM Na₂SO₃; (7) 100 μM Na₂SO₄; (8) 100 μM CH₃SSSCH₃; (9) 50 μM Na₂S₂; (10) 50 μM Na₂S₄. (B) Fluorescence intensity of PSP-3 (10 μM) in the presence of ROS (with or without H₂S). (1) probe alone; (2) 200 μM H₂O₂; (3) 50 μM ClO⁻; (4) 50 μM O₂⁻; (5) 50 μM •OH; (6) 50 μM ¹O₂; (7) 200 μM H₂O₂ + 50 μM Na₂S; (8) 200 μM H₂O₂ + 100 μM Na₂S; (9) 50 μM ClO⁻ + 100 μM Na₂S; (10) 50 μM O₂⁻ + 100 μM Na₂S; (11) 50 μM •OH + 100 μM Na₂S; (12) 50 μM ¹O₂ + 100 μM Na₂S.

To further evaluate the sensitivity of PSP-3 for H₂S_n, a series of varied concentrations of Na₂S₂ (0 ~ 50 μM) were tested. The fluorescence intensities were linearly related to the concentrations of Na₂S₂ in the range of 0 ~ 20 μM (Figure S2). The detection limit was calculated to be 3 nM. This is the most sensitive probe for H₂S_n reported so far. To clarify fluorescence turn-on mechanism, we studied the reactions between the probes and Na₂S₂. All of the reactions gave fluorescein and benzodithiolone as the isolated products with good yields (80% ~ 90%) (Figure S3). As such, we believe the fluorescence turn-on is initiated by H₂S_n-mediated thioester cleavage. The resulting thiolate further reacts with H₂S_n and promotes cyclization to form benzodithiolone and release the fluorophore.

Next we wondered if PSP-3 could be used for imaging H₂S_n in cells. We first validated the capability of PSP-3 in visualizing exogenous H₂S_n. As shown in Figure 4, HeLa, RAW264.7, and Vero cells were respectively incubated with PSP-3 (5 μM) for 25 min. Extracellular probe was washed off, and no significant fluorescence was observed in cells. However, when cells were treated with Na₂S₂ or Na₂S₄ (30 μM) for 20 min, strong fluorescence was observed. These results demonstrated that PSP-3 has good cell permeability and could be used to monitor H₂S_n in cells. In addition, cell viability assay demonstrated that PSP-3 has no cytotoxicity to cells (Figure S4).

Figure 4.

Fluorescence images of exogenous H₂S_n in HeLa (a–c), RAW264.7 (d–f), and Vero (g–i). Cells were incubated with PSP-3 (5 μM) for 25 min, then washed and subjected to different treatments. (a, d, and g) controls (no Na₂S₂ or Na₂S₄); (b, e, and h) treated with 30 μM Na₂S₂; (c, f, and i) treated with 30 μM Na₂S₄.

H₂S_n are short-lived species and can readily decompose in buffers. In this regard, in-situ generation of H₂S_n from H₂S and ClO⁻ provides a more reliable and sustainable system for H₂S_n production. Herein PSP-3 was used to monitor in-situ generation of H₂S_n in cells. As shown in Figure S5, neither ClO⁻ nor H₂S gave noticeable fluorescence responses while the mixture of H₂S/ClO⁻ led to a strong fluorescence enhancement, which was even brighter than those obtained with exogenous H₂S_n. These results confirmed that in-situ generation of H₂S_n by H₂S and ClO- is a more effective system than simply using Na₂S₂ (or Na₂S₄) to maintain H₂S_n levels in cells.

Having demonstrated the capability of PSP-3 in detecting exogenous H₂S_n in cells, we then sought to use PSP-3 to monitor endogenous H₂S_n formation. Our recent studies found that CSE-overexpression causes significant elevation of persulfide and polysulfide levels in cells.^[5] Therefore, CSE over-expressed cells provide a suitable system for endogenous H₂S_n production. As shown in Figure 5a–d, normal and CSE-overexpressed^[5] COS-7 cells were separately treated with PSP-3 for 30 min. As expected, CSE-overexpressed cells induced much-enhanced fluorescent signals as compared to normal cells. In fact, appreciable amount of H₂S_n was directly generated from a recombinant CSE in a cell-free reaction mixture containing cystine as a substrate (Figure S6). These results demonstrate that PSP-3 is suitable for detecting endogenous H₂S_n. Additionally, Kimura et al recently described a new H₂S biosynthetic pathway from D-Cys in mammalian cells, which occurs predominantly in the kidney and the cerebellum.^[14] This process seems to be regulated by 3-mercaptopyruvate sulfurtransferase (3MST) and D-amino acid oxidase (DAO). It was suggested that H₂S produced from D-Cys might be stored as bound sulfane sulfur and the levels of sulfane sulfur can be significantly enhanced by D-Cys. In theory, H₂S_n could be produced from the reactions between H₂S and sulfane sulfurs. Therefore D-Cys stimulation in Vero cells may be a useful system of endogenous H₂S_n generation and this system was used to further validate the probe's sensitivity for endogenous H₂S_n. As shown in Figure 6e–h, the cells were incubated with PSP-3 (5 μM) for 25 min at 37 °C. The control (Figure 5e) showed almost no fluorescence. When cells were treated with 30 μM D-Cys, a modest but obvious fluorescence was observed (Figure 5f). When the concentration of D-Cys was increased to 1 mM, we observed strong fluorescence (Figure 5g). Moreover, when the cells were first treated with sodium benzoate, a DAO inhibitor, the stimulation by D-Cys (1 mM) led to significantly weakened fluorescence. Taken together, these results indicate that D-Cys and DAO may be responsible for cellular H₂S_n production.

Figure 5.

Fluorescence images of endogenous H₂S_n in COS-7 (a–d) and Vero (e–h). (a) Cells not treated with PSP-3; (b) Normal cells treated with PSP-3; (c) CSE-overexpressed cells treated with PSP-3; (d) Cells treated with 20 μM Na₂S₂. (e–g) Cells were incubated with PSP-3 (5 μM) for 25 min, washed, and subjected to different treatments. (e) cells were incubated with FBS-free media for 30 min; (f) cells were incubated with 30 μM D-Cys for 30 min; (g) cells were incubated with 1 mM D-Cys for 30 min; (h) cells were pre-treated with sodium benzoate (250 μM) for 30 min, then washed and incubated with PSP-3 (5 μM) for 25 min, and then washed and incubated with 1 mM D-Cys for 30 min.

In summary, we report here the development of three new fluorescent probes (PSP-1, PSP-2, PSP-3) based on the unique dual-reactivity of H₂S_n. Among the probes prepared, PSP-3 showed desirable off–on fluorescence response to H₂S_n and high specificity. PSP-3 was also successfully applied in visualizing both exogenous and endogenous H₂S_n in cells. This novel probe overcomes the major limitations of previously reported H₂S_n probes, and it is expected to serve as a useful tool in understanding the physiological functions of H₂S_n.