Genetically Anchored Fluorescent Probes for Subcellular Specific Imaging of Hydrogen Sulfide

Abstract

Imaging hydrogen sulfide (H₂S) at the subcellular resolution will greatly improve the understanding of functions of this signaling molecule. Taking advantage of the protein labeling technologies, we report a general strategy for the development of organelle specific H₂S probes, which enables sub-cellular H₂S imaging essentially in any organelles of interest.

Hydrogen sulfide (H₂S) is the most recently discovered member of gasotransmitters, along with nitric oxide (NO) and carbon monoxide.¹ H₂S is mainly produced by three enzymes, including cystathionine β-synthase (CBS), cystathionine γ-lyase (CSE), and 3-mercaptopyruvate sulfurtransferase (MST).² H₂S has been found to play many important roles in the regulation of blood pressure, cardiac response to ischemia injury, and inflammation.² It can also perform S-sulfhydration on proteins to modulate their activities.³ It is known that the subcellular distribution of H₂S producing enzymes changes under different physiological conditions. For example, liver ischemia can lead to the accumulation of CBS from cytosol into mitochondria.⁴ In addition, cytoplasmic CBS and CSE can be translocated into nucleus upon SUMO modification despite the roles of this translocation are unclear.⁵ Similarly, distribution and functions of H₂S in different organelles are not well understood. It was reported that H₂S can not only regulate mitochondrial bioenergetics functions,⁶ but also augment synaptic neurotransmission in the nucleus of the solitary tract.⁷ Because variations in the spatial and temporal production of H₂S can induce different downstream biological effects, it is important to track H₂S with subcellular resolution.¹ In this contribution, we present a new class of organelle-specific fluorescent probes for H₂S leveraging protein labeling techniques and demonstrate its applications in cells.

Several fluorescent H₂S probes for intracellular imaging were previously developed.^8-12 The groups of Chang and Wang took advantage of the reducing potential of H₂S to convert azides to amines to achieve a “turn-on” fluorescence response. Xian et al. reported a probe containing bis-electrophilic centers for H₂S detection.¹³ He et al. designed a Michael addition-based H₂S probe.¹⁴ These probes, however, lack organelle specificity. The groups of Kim and Guo applied positively charged molecules as targeting moieties to direct H₂S probes to mitochondria,^{15, 16} while Cui et al. developed a morpholine containing H₂S probe for lysosome targeting.¹⁷ However, these approaches cannot be generalized to other organelles and to H₂S producing enzymes, such as CBS and CSE. Ai et al. developed a genetically encoded H₂S probe based on an unnatural amino acid modified green fluorescent protein, which has the potential to fuse with any subcellular locating sequences to achieve organelle specific H₂S detection.¹⁸ The caveat for Ai's H₂S probe is currently limited to green fluorescence proteins.¹⁹ Therefore, a generalized alternative approach for organelle specific H₂S probe with broad color palette is needed for H₂S related research.

Fluorescent labeling of specific proteins in live cells has achieved significant advancement in recent years.²⁰ Among these techniques, SNAP/CLIP-tag^® technologies demonstrated to be reliable and applicable to wide range of target proteins. The SNAP/CLIP-tag^® technology, developed by Johnsson and co-workers, is based on the O⁶-alkylguanine-DNA alkyltransferase (AGT) fusion protein which, once it is fused with a protein of interest, reacts covalently with labeling molecules.²¹ These protein labeling techniques have been widely applied for specific labeling of proteins in live cells^{21, 22} and in animals,²³ and directing fluorescent probes to detect Ca²⁺, Zn²⁺ and H₂O₂ in specific organelles.^24-27

Our design of the organelle-specific fluorescent probes for H₂S is composed of an azide based H₂S fluorescent probe and a tagging substrate for SNAP/CLIP technologies (Figure 1). This strategy can be applied to any subcellular locations of interest. In this contribution, we demonstrate its applicability by developing H₂S fluorescent probes that specifically target mitochondria and for the first time – the nucleus.

Figure 1.

Illustration of genetically anchored fluorescent probes for subcellular specific imaging of hydrogen sulfide. In this specific example, cyan represents a nucleus localizing protein; orange represents CLIP-tag enzyme, which can specifically react with a CLIP-tag substrate CouN₃-BC. Upon reacting with H₂S, non-fluorescent CouN₃ is converted to fluorescent CouNH₂.

We chose 7-azido-coumarin (CouN₃) as the H₂S probe that fluoresces with 405 nm of excitation because i) it has been established that the azide group can quench the coumarin fluorescence, and upon reacting with H₂S, azide can be converted to an amine group, which restores the fluorescence of coumarin;²⁸ ii) coumarin can be applied in two-photon fluorescence imaging, which facilitates its in vivo applications;²⁹ and iii) our design can potentially complement with Ai's H₂S responsive GFP in a multi-color organelle specific H₂S detection.¹⁹ CouN₃ was conjugated with a O²-benzylcytosine (BC) moiety – the CLIP-tag substrate (Figure 2A). The synthesis of the CouN₃-BC conjugate is described in the Supporting Information (SI).

Figure 2.

Reaction specificities of H₂S fluorescent probe CouN₃-BC. (A) Chemical structure of organelle specific H₂S probe CouN₃-BC. (B) Normalized time dependent fluorescence emission spectra of CouN₃-BC (10 μM, λ_ex = 405 nm) upon reacting with 100 μM of H₂S in PBS (pH 7.4). The spectra were normalized to the emission intensity at 450 nm at time 0. (C) Normalized fluorescence changes of CouN₃-BC (10 μM, λ_ex = 405 nm, λ_em = 450 nm) upon reacting with a series of redox species in PBS (pH 7.4) for 1 h. The concentrations of GSH and cysteine are 5 mM and 500 μM, respectively. The concentrations of all the other redox species, including H₂S, are 100 μM. The fluorescence intensities at 450 nm were normalized to that of unreacted CouN₃-BC. Note: All the reactions were carried out under dark and anaerobic conditions to avoid potential photodegradation of the probe and oxidation of reductive species.

The spectroscopic changes and reaction kinetics of CouN₃-BC with H₂S were investigated. CouN₃-BC exhibits very minimal fluorescence with an excitation wavelength at 405 nm (Figures 2B and S1). Upon reacting CouN₃-BC with H₂S, the fluorescence at 450 nm increases 35 folds within 1 h (Figure 2B). It should be noted that Barrios et al. reported a 4-fold increase of fluorescence intensity for a similar H₂S probe upon reacting with 100 μM of H₂S for 1 h,²⁸ which is ~9 times lower than the fluorescence enhancement from our observation. Based on our experiments, we found that CouN₃-BC is photosensitive and aged CouN₃-BC usually has 4-5 folds higher baseline fluorescence than the newly prepared counterpart. Therefore, all our CouN₃-BC samples used in the fluorescence measurements were purified with flash chromatography in the dark and used immediately. In addition, because H₂S solution is prone to oxidation in air, we carried all the experiments in Figure 2 under anaerobic conditions. We suspect that the experimental details may account for the discrepancy between our and Barrios’ studies. In order to verify that CouNH₂-BC is the product for the reaction between CouN₃-BC and H₂S, we applied liquid chromatography-mass spectrometry (LC-MS) to monitor the reaction and found a new product with molecular weight that matches CouNH₂-BC (Figure S2).

CouN₃-BC is a highly specific probe for H₂S. In order to test the reaction specificity of CouN₃-BC towards H₂S, we incubated CouN₃-BC with various reductive and oxidative species for 1 h and monitored the changes of the fluorescence intensity at 450 nm. As shown in Figure 2C, the fluorescence intensity increased significantly when CouN₃-BC was incubated with PBS buffer containing 100 μM of H₂S. In contrast, there were minimal fluorescence changes observed when CouN₃-BC was incubated with other sulfur containing compounds, intracellular reductants or oxidants. Our result is consistent with previous studies using azides as a H₂S specific group^{18, 30} and demonstrate the H₂S specificity of CouN₃-BC. It should be noted that in order to simplify this model study, CouN₃-BC was used instead of the conjugate of CouN₃ and CLIP-tag protein, which is the functional unit inside cells. Because the BC moiety has little absorbance at 405 nm, we do not expect that it interferes with the fluorescence of CouNH₂. Furthermore, based on Lippard's study, AGT conjugated Zn²⁺ probe has very similar reactivity to the parent probe.²⁴ Therefore, we believe that CouN₃-BC is a reasonable model compound to study the spectroscopic changes and reactivity in the presence of H₂S and other redox species.

CouN₃-BC is cell-permeable and can cross plasma membrane in a concentration dependent manner. In order to apply CouN₃-BC for live cell imaging, it should be able to permeate cells. In addition, the excess CouN₃-BC that has not reacted with the CLIP-tag proteins should diffuse out of the cells in the washing steps. To test the cell permeability of CouN₃-BC, we incubated CouN₃-BC with HeLa cells for 30 min, followed by the addition of H₂S for another 30 min. We observed strong fluorescence inside the cells, indicating CouN₃-BC can enter cells (Figure S3). When incubation was followed by PBS washing, however, subsequent addition of H₂S solution did not induce any fluorescence increase, indicating that the excess CouN₃-BC that does not react with the CLIP-tag protein can diffuse out of cells (Figure S3).

CouN₃-BC can probe the changes of H₂S levels with subcellular specificity through conjugation with CLIP-tag fusion proteins. In this “proof-of-concept” study, we chose nucleus and mitochondria as the subcellular targets for CouN₃-BC because many of the known H₂S mediated biological processes are involved with these two organelles.^{3, 5, 7} Two plasmids pCLIP-H2B and pCLIP-COX8A were used to transiently express CLIP fusion proteins to localize CouN₃-BC to nucleus and mitochondria, respectively. The pCLIP-H2B construct targets the CLIP tag to the nuclear matrix with the help of the nucleus localization sequence on the fused histone H2B protein.³¹ The pCLIP-COX8A plasmid expresses cytochrome c oxidase subunit VIIIA (COX8A) with a fused CLIP tag that localizes at the inner membrane and the matrix of mitochondria.²⁵

HeLa cells were transiently transfected with pCLIP-H2B and pCLIP-COX8A for 72 h using lipofectamine^® 3000, and then incubated with CouN₃-BC to localize the H₂S probe to nucleus and mitochondria, respectively. Excess CouN₃-BC was removed by PBS washing. Subsequently, exogenous H₂S (250 μM) was added into the medium to induce the fluorescence responses from the H₂S probe. Secondary staining was also performed to confirm the organelle specificity of the H₂S probe. The negative control cells were transfected with control vectors. All the cells were treated with CouN₃-BC and H₂S, and their corresponding controls.

For the negative control cells, there was minimal fluorescence observed in the blue (405 nm) channel even in the presence of both CouN₃-BC and H₂S, indicating that CouN₃-BC can be efficiently removed if there are no CLIP-tag fusion proteins inside cells (Figure 3).

Figure 3.

Confocal microscopic study of the organelle specificity of CouN₃-BC (blue) and its responsiveness to cellular H₂S concentration changes. HeLa cells expressing pCLIP-H2B, pCLIP-COX8A, or a control plasmid were incubated with CouN₃-BC (10 μM), or NaHS (250 μM) or both. The organelle specificity was confirmed by a secondary staining with either NucRed^® Live 647 nuclear stain (red) or an Alexa 488 conjugated antibody against COX-IV (green). Quantitative colocalization analyses showed a significant correlation based on the Pearson's and Manders’ coefficients (Table S1). Scale bar is equal to 10 μm.

For pCLIP-H2B transfected cells, there was minimal fluorescence observed in the blue channel if either CouN₃-BC or H₂S was added to the cells, suggesting that there is low background from cells and CouN₃-BC (Figure 3). In contrast, significant fluorescence increase was observed in the nucleus only when cells were treated with both CouN₃-BC and H₂S, indicating an organelle specific H₂S response. It should be noted that these organelle-specific H₂S probes can only reflect the presence of H₂S in that organelle, but cannot determine whether H₂S is produced in the organelle due to the diffusive nature of H₂S inside cells.

In order to further confirm the organelle specificity of the H₂S probe, the cells were stained with NucRed^® Live 647 nuclear stain and an Alexa 488 conjugated antibody against COX-IV,³² cytochrome c oxidase complex as a marker of the mitochondrial inner membrane (off-target control). From the pseudo colored merged image of different channels, we can observe a substantial area in purple (overlay of the blue and red channels, Figure 3) indicating a significant accumulation of the H₂S probe in nucleus. Quantitative colocalization analyses further confirmed a significant correlation based on the Pearson's and Manders’ coefficients (Table S1). We could not detect any overlap between the blue and green channels, demonstrating minimal off-target effect of CouN₃-BC. To the best of our knowledge, this is the first example of a nuclear specific H₂S probe.

To demonstrate the general applicability of our genetically anchored H₂S probe, a similar experiment was performed using pCLIP-COX8A treated cells. We also observed substantial co-localization (cyan in Figure 3) of the fluorescence signals from the mitochondria targeted probe and the Alexa 488 conjugated antibody against COX-IV, a mitochondrial marker protein (Table S1). In contrast, little co-localization was observed between the blue channel (probe) and the red channel (nucleus), suggesting minimal off-target effect for the mitochondria targeted H₂S probe. It should be cautioned that the fluorescence of CouNH₂ can be quenched in acidic environment due to protonation of the aromatic amino group. Therefore, sensitivities of these probes may be affected if applied in endosomes or lysosomes.

The organelle specific CouN₃ probe is also responsive to endogenous H₂S. It is reported that pharmacological donors of NO can up-regulate cysteine uptake and CBS activity, resulting in increase of H₂S production.³³ To test whether CouN₃ probe is responsive to endogenous H₂S, HeLa cells were transiently transfected with pCLIP-H2B, and then incubated with CouN₃-BC to localize the H₂S probe to nucleus. Subsequently, cells were treated with sodium nitroprusside (SNP), an NO donor, to induce H₂S production. Consistent with previous study,³⁴ SNP treated cells showed significantly higher H₂S concentration in nucleus than the control cells (Figure 4). Considering that CBS is mainly distributed in mitochondria,³⁵ we speculate that H₂S produced in mitochondria by CBS can diffuse to the nucleus to trigger cellular signaling. Hydrogen peroxide (H₂O₂) and NO are known for their fast metabolism and short diffusion range. Based on our observation, different from NO and H₂O₂, H₂S may act as a long-range signaling molecule.

Figure 4.

Confocal microscopic study of the nuclear specific CouN₃-BC (blue) and its responsiveness to H₂S concentration changes induced by SNP, an NO donor. HeLa cells expressing pCLIP-H2B were incubated with CouN₃-BC (10 μM), and then treated with or without SNP (100 μM). The organelle specificity was confirmed by a secondary staining with either NucRed^® Live 647 nuclear stain (red) or an Alexa 488 conjugated antibody against COX-IV

In summary, the protein labeling technology provides an opportunity to direct the intracellular distribution of H₂S probes. We developed coumarin based H₂S probes, which can be specifically localized to nucleus and mitochondria depending on the tag fusion proteins used. This strategy can be extended to other fluorophores and cell labeling techniques, such as HaloTag^®, and applied to other organelles and H₂S producing enzymes, such as CBS and CSE, to elucidate H₂S related biology.