Using resonance synchronous spectroscopy to characterize the reactivity and electrophilicity of biologically relevant sulfane sulfur

Sulfane sulfur is common inside cells, playing both regulatory and antioxidant roles. However, there are unresolved issues about its chemistry and biochemistry. We report the discovery that reactive sulfane sulfur such as polysulfides and persulfides could be detected by using resonance synchronous spectroscopy (RS2). With RS2, we showed that inorganic polysulfides at low concentrations were unstable with a half-life about 1 min under physiological conditions due to reacting with glutathione. The protonated form of glutathione persulfide (GSSH) was electrophilic and had RS2 signal. GSS− was nucleophilic, prone to oxidation, but had no RS2 signal. Using this phenomenon, pKa of GSSH was determined as 6.9. GSSH/GSS− was 50-fold more reactive than H2S/HS− towards H2O2 at pH 7.4, supporting reactive sulfane sulfur species like GSSH/GSS− may act as antioxidants inside cells. Further, protein persulfides were shown to be in two forms: at pH 7.4 the deprotonated form (R-SS-) without RS2 signal was not reactive toward sulfite, and the protonated form (R-SSH) in the active site of a rhodanese had RS2 signal and readily reacted with sulfite to produce thiosulfate. These data suggest that RS2 of sulfane sulfur is likely associated with its electrophilicity. Sulfane sulfur showed species-specific RS2 spectra and intensities at physiological pH, which may reveal the relative abundance of a reactive sulfane sulfur species inside cells.


Introduction
Hydrogen sulfide (H 2 S) is a new gasotransmitter that serves many important regulatory roles in biological systems [1]. H 2 S is involved in vascular homeostasis, neurological function, cytoprotection, anti-inflammation, and revascularization [1][2][3]. However, accumulating evidences imply that H 2 S is converted to reactive sulfane sulfur, which plays the observed roles [4][5][6]. Reactive sulfane sulfur includes organic persulfides (R-SSH), organic polysulfides (R-SS n H or R-SS n R, n ≥ 2), and inorganic hydrogen polysulfides (H 2 S n , n ≥ 2) [7]. Reactive sulfane sulfur is different from thiols, as it often possesses both nucleophilic and electrophilic characteristics while thiols mainly function as nucleophiles [8]. The reactive sulfane sulfur can be produced from specific and nonspecific enzymatic oxidations of H 2 S [9,10] or from the metabolism of cysteine and N-Acetyl cysteine (NAC) [11][12][13]. GSSH is a key form of reactive sulfane sulfur in the sulfide oxidation pathway of heterotrophic bacteria and human mitochondria [14,15]. Reactive sulfane sulfur can modify cysteine residues in a large number of proteins by S-persulfidation (R-SSH), which can alter enzyme activity and influence biological processes via signaling [13,16]. For instance, rhodanese (thiosulfate:cyanide sulfurtransferase) that is present in almost all living organisms catalyzes the transfer of the sulfane sulfur from thiosulfate to cyanide via an intermediate (R-SSH) at its catalytic Cys residue [17,18]. Collectively, previous reports have revealed the significance of reactive sulfane sulfur in biological processes. Thus, a better understanding of the chemical and biochemical properties of biologically relevant reactive sulfane sulfur will help to advance the field [19,20].
Current methods used for the detection of reactive sulfane sulfur include sulfur chemiluminescence detection, ion chromatography, HPLC analysis of the monobromobimane derivative of H 2 S n , and the use of H 2 S n -sensitive fluorescent dyes in living cells or in vitro [5,7,21]. Gao et al. developed some fluorescent probes that serve as an effective imaging tool for tracing or monitoring concentration changes of endogenous sulfane sulfur [22,23]. All of these methods are reactionbased. A reaction-free method that can real-timely probe reactive sulfane sulfur has not been developed. Here, we report the discovery that reactive sulfane sulfur can be detected via resonance synchronous spectroscopy (RS 2 ) with a conventional spectrofluorometer by simultaneously scanning the excitation and emission (i.e. Δλ = λ em -λ ex ) [24]. This method is simple, fast, and nonintrusive for reactive sulfane sulfur analysis, allowing us to distinguish the protonated and

RS 2 analysis of reactive sulfane sulfur
RF-5301 PC Spectrofluoro Photometer (SHIMADZU) was used to measure the fluorescence. Sample was diluted into 2 ml argon-deoxygenated buffers (Tris-HCl 50 mM, pH 7.4) in a parafilm-sealed fluorometer cell (d = 1 cm). Cluster 5 chemicals were dissolved in acetone to make a 100 mM stock and then diluted into argon-deoxygenated buffer. RS 2 was acquired by simultaneously scanning the excitation (λ ex ) and emission (λ em ) on monochromators setting the offset (Δλ = λ em -λ ex ) to a constant [27]. All spectra were acquired with a scan rate of 60 nm/min. The measurement interval was 1.0 nm and slit width was 5 nm. For pH relevant RS 2 analysis, the concentrations of reactive sulfane sulfur were carefully selected to let the RS 2 intensities fell into the detection range of RF-5301. Known amounts of H 2 S n and GSSH were dissolved in 20 ml of 50 mM Tris-HCl solutions (pH 7.4) and 20 mM sodium phosphate solution (pH 6), respectively, and then were titrated with 500 mM NaOH via 10-μl additions. The solution mixture was vortexed, followed with pH measurement and RS 2 acquisition. The RS 2 intensities were used for determining pK a .

pK a determination method
The average signal intensities of GSSH (375 nm-384 nm) and DUF442-C34-SSH (444 nm-453 nm) were used for determining their pK a values, respectively. The pK a calculating equation is deduced as below: R 2 S 2 (I R2S2 obs ) was obtained by dividing the observed RS 2 intensity (I RS2 solu ) with the RS 2 intensity of the buffer (I RS2 solv ).
For RSSH/RSS − mixture, the R 2 S 2 is equal to that of fully protonated form (RS p ) times the fraction of its protonated form (f p ).
This equation can be rewritten as follow: According to the Henderson-Hasselbalch equation, p is hill slop; the fill status is succeeded (100).
Substituting the right-hand side of eq. (4) into eq. (3), we obtain: The detected R 2 S 2 intensity data of RSSH at different pH values were fitted with eq. (5) to obtain the pK a value.

1 H NMR and 13 C-1 H HMQC analysis
The 1 H NMR spectra were recorded on a Bruker spectrometer at 600 MHz with a 5-mm probe. 13 C-1 H HMQC spectra were recorded on the Bruker spectrometer at 600 MHz with a 5-mm-gradient salt-tolerant H/C probe. The pulse sequence was set according to a previous report [28]. Delay = 1.5s, Size of fid = 1024, Number of scans = 64. The NMR data were processed and analyzed with Mestrelab Mnova version 10.

Chemical reactions analysis
For RS 2 analysis of GSSH disproportionation, 50 μM of GSSH was transferred into 20 mM sodium phosphate buffer of different pH, and RS 2 was measured at selected time points as mentioned in the text. The reaction mixtures were also analyzed by HPLC-fluorescence and MS analysis.
For kinetics analysis, reactions were conducted in a fluorometer cell (d = 1 cm) sealed with parafilm. Reactions of H 2 S n with GSH were performed in deoxygenated HEPES buffer (100 mM, pH 7.4), started by adding 200 μM-5 mM of GSH to 10 μM of H 2 S n . RS 2 of 535 nm-545 nm was scanned immediately at 30-s intervals for 3 min. Reactions of RS n R with GSH were performed in deoxygenated HEPES buffer (100 mM, pH 7.4), started by adding 10 mM-20 mM of GSH to 500 μM of RS n R. RS 2 of 535 nm-545 nm was scanned at 1-min intervals for 8 min. The k obs value was calculated by plotting the ln [RS2] value against the reaction time. The apparent 2 nd -order reaction rate constant k was calculated with the formula: k obs = k × [GSH]. For H 2 S release detection, these reactions were performed in sealed tubes. Lead acetate papers were fixed in the gas phase of the tubes containing the reaction mixture. Reactions of DTT with H 2 S n and RS n R were similar to the GSH reaction above, and the calculations were also similar.
Reactions of antioxidants (H 2 S or GSSH) with H 2 O 2 were conducted in deoxygenated HEPES buffer (100 mM, pH 7.4), started by adding 50 μM-500 μM of the antioxidant to 50 μM of H 2 O 2 . H 2 O 2 reacted with GSSH to generate GSSSG [29], which had RS 2 . The RS 2 (535-545 nm) intensity of GSSSG was obtained and used to calculate the reaction rate. H 2 O 2 reacted with H 2 S to generate H 2 S 2 , which displayed RS 2 , and the RS 2 increase was used to obtain the reaction constant. The k obs value was calculated by plotting the L n [GSSSG] (or L n [H 2 S 2 ]) value against the reaction time. The apparent 2 nd -order reaction rate constant k was calculated using the formula: k obs = k × [antioxidant].

Protein purification and modification with GSSH
The DUF442 domain of SQR (GenBank accession number: AAZ62946.1) was cloned from Cupriavidus pinatubonensis JMP134. Sitedirected mutagenesis was performed according to a revised method [30]. For protein expression, these genes were ligated into the pET30a vector with a His tag at the C-terminus and then expressed in Escherichia coli BL21 (DE3) ( Table S1). The recombinant E. coli was grown in LB at 30°C with shaking until OD 600nm reached about 0.6, and 0.3 mM IPTG was added; the cells were further cultivated at 20°C for 20 h. Cells were harvested and disrupted with crusher SPCH-18 (STANSTED); protein purification was carried out with nickel-nitrilotriacetic acid agarose resin (Invitrogen). Buffer exchange of the purified proteins was performed via PD-10 desalting column (GE Healthcare). The finally obtained protein was in HEPES buffer (25 mM, pH 8.0) containing 300 mM NaCl.
The purified protein (6.0 mg/ml) was mixed with 200 μM of GSSH in HEPES buffer (100 mM, pH 7.4). After incubated at 25°C for 20 min, the mixture was loaded onto a PD-10 desalting column to remove small molecules. The re-purified protein was then subjected to LC-MS/MS, RS 2 or sulfite reaction analysis. For RS 2 analysis, the protein was diluted to 0.1-0.5 mg/ml in the HEPES buffer so that the RS 2 intensities were within the detection range of our fluorometer (RF-5301). For protein-SSH pK a determination, we diluted DUF442-C34-SH (the C94S mutant) and GSSH reacted-DUF442-C34-SSH in HEPES buffers of different pH (3, 3.5, 4, 4.5 … …6.5, 7, 7.4), and then detected their RS 2 intensities. The pK a was determined using Eq. (5). Titrating HCl solution into a protein solution may cause protein denaturation.

HPLC-fluorescence and MS analysis of persulfides
LC-fluorescence and MS analysis of GSSH and protein-SSH was performed by following a previously reported protocol [10]. Briefly, samples were derivatized with monobromobimane (mBBr) and were injected onto a C18 reverse phase column (VP-ODS, 150 × 4 mm, Shimadzu). The column was maintained at 30°C and eluted with a gradient of solution A (0.25% acetic acid) and solution B (0.25% acetic acid and 75% methanol) in distilled water from 5% B to 70% B in 8 min, 70% B for 8 min, 100% B for 8 min at a flow rate of 0.8 ml/min. The fluorescence detector (LC-20A) was used for detection with excitation at 370 nm and emission at 485 nm. The ESI mass spectrometer (Ultimate 3000, Burker impact HD) was used with the source temperature at 200°C and the ion spray voltage at 4.5 kV. Nitrogen was used as the nebulizer and drying gas.

HPLC analysis of H 2 S n
H 2 S n (5 mM) was diluted in Tris-HCl buffer at different pH, derivatized with methyl trifluoromethanesulfonate (methyl triflate) and analyzed by reversed-phase liquid chromatography using a C18 reverse phase column (VP-ODS, 150 × 4 mm, Shimadzu) and eluted with pure methanol. HPLC analysis and peak position of dimethylpolysulfides from Me 2 S 2 to Me 2 S 8 and S 8 were found from calibration curves according to a published protocol [31].

Detection of sulfane sulfurs by using SSP4
Reactions of GSSH with SSP4 (Sulfane Sulfur Probe 4, Dojindo China Co., Ltd) were conducted by mixing 10 μM of SSP4 with 20 μM of GSSH in 100 μl of HEPES (0.1 M) buffer at different pH. The mixture was incubated at room temperature for 30 min, and then the fluorescence was detected by using Synergy H1 microplate reader. The excitation wavelength was set at 482 nm and the emission wavelength was set at 515 nm.

Whole cell analysis by RS 2
Wild-type E. coli BL21 and recombinant E. coli strain containing pBBR1-CpSQR were used for intracellular polysulfides analysis. The strain was incubated at 37°C until OD 600nm reached about 0.6 in LB medium. To induce CpSQR expression, 0.3 mM IPTG was added, and the cells were further cultivated at 30°C for 5 h. Cells were collected by centrifugation and washed twice with Tris buffer (50 mM, pH 7.4). Different concentrations of NaHS were added to cell suspension of 0.1 OD 600nm . H 2 S oxidation was performed at 30°C for 40 min. Then cells were collected and washed with Tris-HCl buffer (50 mM, pH 7.4). For RS 2 analysis, cell intensity was adjusted to 0.01 OD 600nm in Tris-HCl buffer. Wild-type E. coli BL21 was incubated in LB medium at 37°C. Cells were collected, washed twice with Tris-HCl buffer (50 mM, pH 7.4), and resuspended to 0.01 OD 600nm before RS 2 analysis.

Whole cell analysis by SSP4
Wild-type E. coli BL21 cells were collected, washed twice with PBS buffer, and resuspended with PBS at 0.1 OD 600nm . SSP4 (10 μM) and CTAB (0.5 mM) were added to the cell suspension and incubated for 15 min at room temperature. After centrifugation (4000 rpm, 5 min), the supernatant was discarded and remaining cells were washed twice with PBS buffer. The cells were diluted to 0.1 OD 600nm in PBS buffer. The fluorescence was analyzed by using Synergy H1 microplate reader.

Discovery of strong RS 2 in H 2 S n and RS n R compounds
When analyzing H 2 S n using resonance synchronous spectroscopic (RS 2 ) [21], we found that it had strong fluorescence intensity. Then we set the offset (Δλ = λ em -λ ex ) to a constant between the excitation and detection wavelength, i.e. Δλ = 0,1,2 … 6 nm to scan the sample. The fluorescence intensity was the highest when Δλ = 1 nm and decreased along with Δλ increased. Thus, Δλ = 1 nm was used for all analyses. Distilled water and 50 mM Tris-HCl had low RS 2 , and we used the Tris-HCl or HEPES buffer (pH 7.4) for most analyses (Fig. 1A). To test whether it is a common property of sulfur-containing compounds, we totally analyzed 14 sulfur-containing chemicals that were sorted into 7 clusters (Table 1). Among them, clusters 2 and 3 are important cellular persulfides and polysulfides; cluster 5 contains diallyl polysulfides (RS n R); other clusters are not polysulfides but are all involved in polysulfide metabolism. All chemicals were tested at pH 7.4. In addition to H 2 S n , Bis [3-(triethoxysilyl)propyl], tetrasulfide (Tsp-SSSS-Tsp) and dimethyl trisulfide (Me-SSS-Me) also showed significant RS 2 (Fig. 1A).
We diluted different concentration of H 2 S n and cluster 5 compounds to analyze the RS 2 intensity. The RS 2 detection range was 0.2 μM-20 μM for H 2 S n at pH7.4 and 10 μM-2 mM for cluster 5 compounds (Me-SSS-Me, 100 μM-2 mM; Tsp-SSSS-Tsp, 10 μM-80 μM). To remove the interference of the buffer, the ratiometric resonance synchronous spectroscopy (R 2 S 2 ) value was obtained via dividing the sample RS 2 intensity (I RS2 solu ) with the solvent RS 2 intensity (I RS2 solv ) (Eq. (1) in methods) ( Fig. 1) [27]. The R 2 S 2 intensity of H 2 S n showed good responses to its concentrations and followed linear dependence at fixed pH values (Fig. 1B, Figs. S1A and S1B). Me-SSS-Me and Tsp-SSSS-Tsp also showed good responses to its concentrations and followed linear dependence ( Fig. 1C and D). For R 2 S 2 detection of GSSSG, we only showed the range 0.025 μM-0.5 μM, but the response is linear up to 5 μM or higher (Fig.  S1C). These results indicated RS 2 is not a common property of all sulfurcontaining chemicals, but a particular property of some chemicals that contain multiple sulfur atoms (n ≥ 2).

The pH effect on RS 2 of reactive sulfane sulfur and its applications
Since GSSH is a pivotal intermediate in cellular reactive sulfane sulfur metabolism [29], it was a surprise that RS 2 of GSSH was hardly detectable at pH 7.4 (Fig. 1A). When we analyzed GSSH at different pH, it showed clear R 2 S 2 at lower pH, especially at pH ≤ 6.0 ( Fig. 2A). The highest peak was around 300 nm, which is consistent with its absorbance peak at 300 nm [10].
We then used R 2 S 2 to determine pK a of GSSH/GSS − . GSSH was dissolved in 20 ml aliquots of 50 mM Tris-HCl solution (pH 6). The solution was titrated with NaOH, followed with pH measurement and RS 2 acquisition (375 nm-384 nm). The pK a value was determined as 6.9 via data fitting by using the Henderson-Hasselbach derived equation (Fig. 3B). GSSH can react with non-fluorescent SSP4 to release fluorescent fluorescein [7]. When GSSH and SSP4 were mixed at different pH, the reaction was rapid at pH 6 but not at pH 9.5 (Fig. S2). The results were logical from a chemical perspective, as GSSH is the electrophile and SSP4 is the nucleophile in the reaction. GSS − should be more nucleophilic, while GSSH should be more electrophilic. Therefore, the reaction of SSP4 with GSSH should be faster than that with GSS − . The reaction rates at different pH were determined. The data were fitted with the Henderson-Hasselbach derived equation to obtain the estimated pK a of GSSH. The value was 6.9, the same as that determined via R 2 S 2 (Fig. 3C). Interestingly, the SSP4 reaction rates and R 2 S 2 intensities at different pH were highly correlated (Fig. 3D), indicating that RS 2 correlates with the electrophilicity of GSSH.
The pH change did not show apparent effect to RS 2 of class 5 chemicals (RS n R) (data not shown). This is expected as RS n R compounds have no conditional protonation issues. When H 2 S n in Tris buffer at pH 7.4 was titrated with 500 mM NaOH, the RS 2 intensity decreased and reached the lowest level at pH around 8.5. However, RS 2 increased again when more NaOH was added into the H 2 S n solution. To confirm those results, we diluted H 2 S n in Tris buffers of different pH (7.4, 8.0, 8.5, 9.0, and 10). The R 2 S 2 intensity was high at pH 7.4 and low at pH 8.5 (Fig. 2B); R 2 S 2 increased again at pH 9.0 and pH 10, but the spectrum changed similarly to that of Tsp-S 4 -Tsp (Fig. 1D). When H 2 S n    was derivatized and analyzed by HPLC, the chain length distribution at various pH corresponded well to the calculated equilibrium distribution of polysulfide ions in aqueous solutions of different pH [31]. At pH 7.4, most H 2 S n was detected as S 8 , and a small peak of S 2 2− was also detectable (Fig. S3A). At low concentrations such as 10 μM, H 2 S 2 /HS 2 − has been detected as the dominant species [10,32]. , and S 7 2− were the dominant species (Fig. S3A). Large portions of S 8 were detected in most samples except at pH 10 ( Fig. S3), and we believe that this is likely due to the high concentration (5 mM) of H 2 S n used in the test for UV detection. Nonetheless, the variations in chain lengths associated with pH changes prevent us using R 2 S 2 to determine the pK a value of H 2 S n . The data also suggest that the R 2 S 2 spectra of H 2 S n depend on protonation as well as on the chain lengths (Fig. 2B). The chain length of H 2 S n detected here should reflect the length in the solutions, as the method is optimized to ensure the derivatization reaction was fast enough to minimize chain elongation reactions [31]. However, if the derivatization step is low and if the alkylating agent reacts with the sulfane sulfur in the middle of polysulfides, some conversion reactions could occur, which interferes with the chain length detection [32].

RS 2 of RS n R may correlate with the presence of thiosulfoxide
RS n R contains sulfane sulfur that may tautomerize to a thiosulfoxide bond (sulfur-sulfur double bond, e.g., R 2 S = S) [33]. We hypothesized that RS 2 of RS n R may correlate with the presence of thiosulfoxide. So we analyzed the structures of Pey-SSSS-Pey and Tsp-SSSS-Tsp by using 1 H NMR and 13 C-1 H heteronuclear multiple quantum correlation spectroscopy ( 13 C-1 H HMQC). In 13 C-1 H HMQC spectra, the two -CH 2groups connecting to sulfur atoms in Pey-SSSS-Pey (C a and C α ) showed two distinguishable peaks, while those of Tsp-SSSS-Tsp showed three (Fig. 4, Fig. S4 and Fig. S5), suggesting C a and C α are not symmetrical. In 1 H NMR spectra, protons linked to C a and C α had two or more groups of peaks, while those linked to other C's did not (Fig. S6 and Fig. S7). These results indicated the four sulfur atoms in these compounds are not linear and isomers containing the branched thiosulfoxide bond (> S]S) should exist, which causes the asymmetric configuration of C a and C α . The branched thiosulfoxide bond might lead to the generation of sulfane sulfur, and the RS 2 observed from cluster 5 chemicals might be caused by the presence of thiosulfoxide.

Analysis of reaction kinetics by using RS 2
We used RS 2 as a real-time probe in assays of reactive sulfane sulfurinvolved reactions (Table 2). First, we tested the stability of RS n R chemicals (Me-SSS-Me and Tsp-SSSS-Tsp) in the presence of 100 mM GSH at pH 7.4 (100 mM HEPES). The RS 2 spectra of RS n R were unchanged, and there was no H 2 S released from the solution. Thus, RS n R is rather stable. Second, we tested the reaction of H 2 S n with GSH in deoxygenated HEPES buffer (100 mM, pH 7.4). After adding GSH (200 μM-5 mM) to 10 μM of H 2 S n , the RS 2 spectra of H 2 S n quickly decreased, and H 2 S was released. At low H 2 S n concentrations and at pH 7.4, H 2 S 2 is the dominant species [10]. By recording the RS 2 decreases, we determined the 2 nd -order rate constant of the reaction between H 2 S n and GSH as 0.89 M −1 s −1 ( Table 2). Because GSH is at least two-orders-ofmagnitude higher than H 2 S n [14,29,34], the reaction between H 2 S n and GSH should occur in the pseudo-first-order manner (e.g., t 1/2 = 78 s) at the physiological pH and GSH concentration. Third, because RS 2 has limitations, we could not use it to determine GSSH reduction at pH 7.4 due to its low RS 2 signal.   [14,16]. Using RS 2 intensity curves of H 2 S n and GSSSG concentration, we analyzed the kinetics of H 2 O 2 with H 2 S or GSSH at pH 7.4 and 25°C (Table 2). At pH 7.4 and 25°C, H 2 S reacted with H 2 O 2 slowly. The 2 nd -order rate constant was determined to be 0.46 M −1 s −1 , close to a previously reported value (0.73 M −1 s −1 ) determined at pH 7.4 and 37°C [14]. On the other hand, GSSH rapidly reacted with H 2 O 2 to produce GSSSG [29]; the 2 nd -order rate constant was 23.8 M −1 s −1 as determined with the RS 2 increase of GSSSG, 50-fold higher than that between H 2 S and H 2 O 2 . The rate constant is likely an underestimate, as the GSSH preparation contains equal molar GSH; the reaction between H 2 O 2 and GSSH/GSH primarily produces GSSG and GSSSG [29]. Considering GSSH is also more abundant than H 2 S inside cells, it has been proposed that GSSH is a major reactive oxygen species (ROS) scavenger other than H 2 S [29]; our finding proves the kinetic support for the hypothesis.
We also analyzed the reaction kinetics of GSSH with SSP4 at pH 7.4 and 25°C by recording the fluorescence increase of the released chromophore from SSP4; the rate constant of this reaction was 9.53 M −1 S −1 .

Detection of GSSH disproportionation reactions
Trace amounts of GSSSH and GSSSG have been found in cancer cells, whether they are from GSSH disproportionation reactions (Fig. 5A) are still inconclusive [29,35,36]. We studied these reactions using the RS 2 method. When GSSH was incubated at pH 9.5, no appearance of RS 2 was detected. At pH 6.9, RS 2 spectra of protonated GSSH was initially observed, then it gradually changed to a spectrum overlapping those of GSSH and GSSSG (Fig. 5B). At pH 6.0, the RS 2 spectral change was also observed with a slower increase of the GSSSG peak. In consistent, LC-ESI-MS analysis (Fig. S8) indicated the amount of unreacted GSSH (remaining in solution) was the highest at pH 9.5 and the lowest at pH 6.9 (Fig. 5C). GSSSG was produced the most at pH 6.9 with less at pH 6.0 and the lowest at pH 9.5 (Fig. 5D). GSSSH was also produced, but at about one order of magnitude lower than that of GSSSG, following the same trend at various pH values ( Fig. 5E; pH 6.9 > pH 6.0 > pH 9.5). At pH 9.5, a small amount of GSSSSG was also detected, which should be produced from GSS − oxidation (Fig. 5F). These results indicated that GSSH disproportionation occurred most efficiently at its pK a . Considering GSSH can be as high as 100 μM in cancer cell and its pK a is close to the intracellular pH [29], it is highly possible that the intracellular GSSSH and GSSSG are produced from these reactions.
3.6. RS 2 separates protein S-persulfidation into active and inactive forms DUF442, a domain of Cupriavidus pinatubonensis JMP134 sulfide:quionone oxidoreductase (GeneBank: AAZ62946.1), has rhodanese activity and catalyzes the reaction of GSSH with sulfite to produce thiosulfate [10]. The DUF442 domain consists of 128 amino acid residues with two cysteine residues, C34 and C94, and only C94 is conserved and functionally essential [10]. We used GSSH to react with DUF442 and LC-MS/MS to analyze the modification. Both C34-SSH and C94-SSH modifications were detected (Fig. S9 and Fig. S10). The modified DUF442 displayed significant RS 2 , which was not observed from unmodified protein at pH 7.4 (Fig. 6A). In addition, the C34S/C94S double-mutant DUF442 showed no RS 2 after reacting with GSSH (Fig. 6B). Next, we reacted GSSH with the two single-mutants of DUF442 (C34S and C94S) at pH 7.4. C34S mutant showed significant RS 2 , while C94S mutant did not, although the individual Cys residues were modified by GSSH treatment to form persulfides (confirmed via LC-MS/MS analysis). These results indicated that C94-SSH is likely in the protonated form (C94-SSH) and C34-SSH is in the deprotonated form (C34-SS -) at pH 7.4. When reacted sulfite, the RS 2 intensity of C94-SSH (C34S mutant) significantly decreased with the production of thiosulfate (C94-SSH + SO 3 2− →C94-SH + S 2 O 3 2− ) [10]; whereas, C34-SSH (C94S mutant) without RS 2 did not produce thiosulfate when reacted with sulfite. When C34-SSH was titrated with HCl, the RS 2 intensity was increased at low pH. In the control containing C34-SH (GSSH unreacted), RS 2 was not detectable at all the tested pH. Considering HCl titration may cause aggregation of protein, which disturbs RS 2 measuring, we used different pH buffer for the titration. We diluted C34-SH or C34-SSH protein in HEPES buffers of different pH. The pK a was determined by using R 2 S 2 method to be 6.29 (Fig. 6C). Thus, only C94-SSH in the DUF442 wild type or the DUF442C34S mutant is protonated at pH 7.4 and the sulfane sulfur can be transferred to sulfite to produce thiosulfate.
To inspect what makes DUF442-C94-SSH in the protonated form (C94-SSH) at pH 7.4, we modeled 3D structures of DUF442 with a putative rhodanese from Neisseria meningitides z2491 (PDB ID: 2F46) as the template (39% sequence similarity). The C94 sulfur was located at the bottom of a cradle-like pocket surrounded by basic side chains, generating a positively electrostatic field (Fig. 6D). The distances between the sulfur atom and the circumjacent nitrogen atoms of the peptide backbone are in the range of 3.1 Å-4.0 Å, and the NH 3 + group of R99 is also nearby (Fig. 6E). Therefore, either C94-SSH is not dissociated in the pocket or C94-SS-forms a hydrogen bond with one of these groups as revealed by RS 2 . In contrast, sulfur atom of C34 is not located in a positively electrostatic field (Fig. 6F), and it should exist in the deprotonated form C34-SS-at pH 7.4, which showed no RS 2 and electrophilicity.

RS 2 method application in whole cells
We also used the RS 2 method to analyze intracellular changes of reactive sulfane sulfur in wild-type E. coli. E. coli contained more reactive sulfane sulfur at the stationary phase of growth (12 h) than at the log phase (6 h), as revealed by RS 2 intensity (Fig. 7A) and SSP4 analysis (Fig. 7B). The sulfane sulfur species have different RS 2 spectra at pH 7.4 (Figs. 1A, 5B and 6A). The RS 2 peak of whole cells around 400 nm suggest the presence of R-SS n H and R-SS n R (n ≥ 2), as well as persulfides (R-SSH); however, most protein persulfides and GSSH do not contribute much to the RS 2 signal at the physiological pH. The low intensity of persulfides at physiological pH is likely responsible for a smaller increase in the R 2 S 2 signal than the increase of sulfane sulfur detected by SSP4 for cells at the stationary phase (Fig. 7B). Further, the RS 2 peak of whole cells at 450 nm suggest the possible presence of H 2 S n .
Previously, we reported that recombinant E. coli strain expressing a sulfide:quinone oxidoreductase of Cupriavidus pinatubonensis JMP134 (CpSQR) can oxidize H 2 S to H 2 S n , and the produced H 2 S n is associated with the cell [10]. Herein, we used the recombinant E. coli to oxidize H 2 S. After the cells oxidized H 2 S, the cells were harvested, washed, and diluted for RS 2 measurement. The RS 2 peak at 450 nm increased, suggesting the production of H 2 S n ; however, the increase slowed down with increased H 2 S oxidation (Fig. 7C). On the basis of the RS 2 spectra, other reactive sulfane sulfur species inside E. coli also increased after H 2 S oxidation, possibly including protein-SSH.

Discussion
Individual fluorophores are often considered as simultaneous photon absorbers and emitters with no significant light scattering due to small sizes [37]. Consequently, RS 2 is often observed from the aggregated fluorophores, which usually are simultaneous photon absorbers, scatters, and fluorescence emitters [21]. However, RS 2 of reactive sulfane sulfur is most likely from soluble, individual molecules, since GSSH at pH 6 and DUF442-C94S-SSH at pH 7.4 are soluble and have RS 2 (Figs. 2 and 6). At pH 8 to 8.5, H 2 S n is mainly present as S 2 2− (Fig.   S3) and the solution has low RS 2 (Fig. 2B). At pH > 9, long chain S n 2− species are dominant and RS 2 was increased ( Fig. 2B and Fig. S3). Our data showed that GSSH in the protonated form has RS 2 and is electrophilic, and the deprotonated form does not have RS 2 and is not electrophilic (Fig. 3A&B). The ratio of GSSH/GSS − at various pH determines the electrophilicity and RS 2 intensity (Fig. 3D). Apparently, the electrophilicity of the sulfane sulfur in GSSH is strongly affected by deprotonation because the negatively charged terminal sulfur (S − ) affects its adjacent sulfur through the α-effect, making both sulfur atoms with negative charge. The same logic may also apply to long chain S n 2− (n > 4) with the terminal sulfur (S − ) having minimal effects on distant S in the middle (Fig. S3). The electrophilicity of sulfane sulfur can be explained in the form of thiosulfoxide (RR'S=S) [33,38]. Although our NMR analysis showed evidence to support the presence of thiosulfoxide in R-SSSS-R (Fig. 4), for GSSH and H 2 S n , whether the sulfane sulfur is in present as thiosulfoxide or in a linear form is still unsettled [39,40]. Thus, our results associate RS 2 with the electrophilicity of sulfane sulfur; the deprotonated persulfides (R-SS -) are nucleophilic and prone to oxidation but does not react with SSP4 [41].
The pK a values of thiols are critical to their reactivity at physiological pH. The pK a of Cys thiol at active center of enzyme may be lowered so that the thiol is deprotonated at neutral pH, which are strongly nucleophilic and are prone to oxidation by ROS. The pK a values of R-SSH are also likely important and have previously been reported within the range of 4.3-6.23 [16,42], implying that the persulfides should be mostly in the deprotonated form (RSS -) at pH 7 and displaying nucleophilic properties. The pK a value of cumyl-SSH has recently being determined as 7.0 [43], close to the value of GSSH (6.9) that we determined with two different approaches (Fig. 3). According to this value, the ratio of deprotonated form and protonated form of GSSH is within the range of 2-9 at physiological pH range (7.2-7.8). Disproportionation of GSSH requires both the deprotonated and protonated forms with one playing an electrophile and another acting as a nucleophile, which is consistent with our observation that the disproportionation was the most efficient at pH closed to its pK a (Fig. 3). These chemical reactions might be the origin of intracellular GSSSH and GSSSG. Sulfane sulfur prefers to move from a high reactive polysulfide to form a lower one [10,44]. Thus, in the cell the flow of sulfane sulfur is likely from H 2 S n to GSSH and then to GSSSG.
Protein S-persulfidation is common inside cells [39]. Here we showed that like cystinyl thiols at the active site, the pK a values of protein persulfides can also be affected by its location. Most protein persulfides are likely deprotonated at physiological pH because they have no apparent RS 2 and cannot react with sulfite (Fig. 6), but the sulfane sulfur at the active site of rhodanese is not deprotonated, due to its location in a positive electrostatic field. Rhodanese can then transfer the sulfane sulfur to small nucleophiles, such as cyanide and sulfite, which act as sulfane sulfur acceptors. Our finding implies that the catalysis of rhodanese is likely to generate an electrophilic sulfane sulfur that is easily transferred between two nucleophilic substrates, such as from GSS − to SO 3 2− , producing S 2 O 3 2− .

Conclusions
We discovered reactive sulfane sulfur species have RS 2 properties only when the molecules contain an electrophilic sulfane sulfur. It can be applied to reactive sulfane sulfur analyses, such as pK a determination, reaction kinetics, pH-dependent sulfane reactivity of small and protein persulfides, etc. For whole cell analysis, it may reveal the relative abundance of a reactive sulfane sulfur species. The RS 2 method is rapid, sensitive and convenient, allowing us to reveal several new chemical and biochemical properties of biologically relevant reactive sulfane sulfur. The results that were reported here, such as the pK a of GSSH, the reaction parameters, the distribution of H 2 S n species at different pH, may fill some gaps in the field.