Reaction-based �uorogenic probes for selective detection of cysteinyl oxidation in living cells

tools the products of unknown. Toward this goal, here we probes acid, a of protein cysteines inextricably linked to signaling and oxidative stress. The probes, called CysOx1 and CysOx2, are reaction-based, exhibit excellent cell permeability, rapid reactivity, and high selectivity with 14 minimal cytotoxicity. We applied CysOx2 in a cell-based 96-well plate assay to determine whether kinase 15 inhibitors modulate protein S -sulfenylation as well as O -phosphorylation. Analysis of these data revealed an unexpected positive association of S -sulfenylation and inhibition of select kinases within the TK, AGC, 17 and CMGC families including GSK3, a multitasking Ser/Thr kinase and emerging therapeutic target for neurodegenerative and mood disorders. Chemoproteomic mapping of sulfenic acid-modified cysteines in GSK3 inhibitor-treated cells shows that sites of S -oxidation localize to regulatory cysteines within key 20 components of antioxidant defense systems. Our studies with CysOx probes offer up new insights into kinase-inhibitor dependent modulation of sulfenylome dynamics and should accelerate future efforts in the modern era of translational redox medicine.

Measuring reactive oxygen, nitrogen and sulfur species in cells is established technology, but turn-on 10 fluorescence tools for detecting the products of their reaction with protein cysteines remain essentially 11 unknown. Toward this goal, here we describe fluorogenic probes for sulfenic acid, a redox modification 12 of protein cysteines inextricably linked to signaling and oxidative stress. The probes, called CysOx1 and 13 CysOx2, are reaction-based, exhibit excellent cell permeability, rapid reactivity, and high selectivity with 14 minimal cytotoxicity. We applied CysOx2 in a cell-based 96-well plate assay to determine whether kinase 15 inhibitors modulate protein S-sulfenylation as well as O-phosphorylation. Analysis of these data revealed Redox reactions and oxidative stress have been implied in the etiology of numerous diseases as well as 26 in the aging process 1 . The modern era in translational redox medicine seeks to identify types, sources, 27 metabolizers, and targets of oxidants in order to develop effective drugs and therapies for ROSopathies 2 . 28 In this context, turn-on fluorescence probes also referred to as fluorogenic probes that measure oxidative 29 stress in living cells have proven invaluable for redox-related biomedical research 3,4 . Analogous tools for 30 detecting the reaction products between biological oxidants and proteins are grossly underdeveloped, a 31 point exemplified by the lack of fluorogenic probes for detecting protein cysteine (i.e., cysteinyl) oxidation. 32 The latter issue is especially striking since protein cysteines are the major target of oxidants originating 33 from both endogenous and exogenous sources 5 . 34 35 Sulfenic acid (Cys-SOH) is a central redox modification of protein cysteines and is inextricably linked to 36 oxidant signaling and stress 6 . Sulfenic acid is generated by oxidation of a thiolate by reactive oxygen 37 species (ROS) such as hydrogen peroxide (H2O2) produced during cellular signaling and metabolism or 38 by hydrolysis of sulfenyl halides, and very polarized nitrosothiols and disulfides 7 . If stabilized by the 39 protein microenvironment, the thiol-sulfenic acid pair can operate as a switch that is triggered by redox 40 changes to regulate protein function, structure, and localization 8-11 . The electrophilic sulfur atom in 41 sulfenic acid can also react with a protein or low-molecular-weight thiol to form a disulfide 12,13 or, under 42 conditions of excess oxidative stress, can be oxidized further to sulfinic and sulfonic acids 14  Selective chemical detection of sulfenic acid is predicated on the chemical nature of this moiety in which 47 Fig. 1 | Design strategy for developing fluorogenic probes for detecting sulfenic acid. a, General structure of phenaline-1,3-dione scaffold and the tautomerization between enol and keto forms. Reaction between the nucleophilic C-2 and electrophilic sulfenic acid sulfur shifts the keto-enol equilibrium toward the keto form, which has fluorogenic potential, but is slow and poorly soluble. b, Colored circles denote the position for introducing electron-withdrawing groups to favor fluorogenic reaction with sulfenic acid (left). The resulting fluorogenic reactionbased probes are rapid, selective, and water soluble. group was replaced by the cyclic amine, azetidine (8) affording a compound with an observed reaction 96 rate of approximately 0.1 s -1 . Overall, the sulfonyl group dramatically increased nucleophilic reactivity 97 towards the sulfenic acid electrophile, which was tempered somewhat by the installation of amino and 98 fluorine groups, as expected. 99

100
In subsequent experiments, we evaluated the fluorescence properties of sulfonyl derivatives 4 through 101 8. For this purpose, the reaction product between each analog and CSA was prepared and isolated from 102 milligram-scale reactions. Fluorescence spectra of the resulting adducts were then recorded in organic 103 solvent or aqueous solution and compared to that of non-adducted compounds alone ( The increase in fluorescence intensity was 2.3-for CSA-7 and 11.4-fold for CSA-8 (quantum yields 114 increased by 2.1-and 2.6-fold, respectively) compared to 7 and 8 alone. UV-Vis spectroscopy analysis 115 of 7 and 8 with their respective CSA adducts showed an increase in extinction coefficient of 1.7-and 6.4-116 fold, respectively (Supplementary Table 1 and Supplementary Fig. 4-5). Finally, a bathochromic shift in 117 the absorption maxima of CSA-8 was also observed, in agreement with the shift observed in the 118 fluorescence spectra. Comparison of data obtained for compounds 7 and 8 to compound 6 demonstrates 7 121 Next, we measured the fluorescence intensity of 7 and 8 in real-time reactions with CSA. Reactions 122 containing 7 or 8 were excited at 443 nm and 447 nm, respectively. Fluorescence emission intensity 123 increased over time and was linear with probe concentration (Fig. 2e,f). Pseudo first-order rate constants 124 were then obtained at different probe concentrations to obtain second-order rate constants, which were 125 measured under pseudo-first-order conditions with small-molecule sulfenic acid model, CSA, in excess over phenaline-1,3-dione nucleophiles as described in Supplementary Methods. b, kobs (s -1 ) for compounds 1 through 8 were determined by curve fitting to a single exponential function. Second-order rate constant, k is extrapolated from kobs divided by the concentration of the limiting agent. c, Absorption and emission spectra of 7 and isolated CSA-7 (250 µM) in PBS:ACN pH 7.4 (9:1). d, Absorption and emission spectra of 8 and isolated CSA-8 (250 µM) in PBS:ACN pH 7.4 (9:1). e, Fluorescence intensity monitored over time for the reaction between 7 (5 to 100 µM) and CSA (1 mM) after excitation at 443 nm (left). Fluorescence intensity at 15 min plotted versus the concentration of 7 (right). n = 3. f, Fluorescence intensity monitored over time for the reaction between 8 (5 to 100 µM) and CSA (1 mM) after excitation at 447 nm (left). Fluorescence intensity at 15 min plotted versus the concentration of 8 (right); n = 3. g-h, Rate constants were measured under pseudo-first-order conditions with CSA in excess over phenaline-1,3-dione nucleophiles as described in Supplementary Methods. A representative trace from these collective reactions is shown. Linear fit of kobs (min -1 ) plotted versus CSA concentration gives 91 ± 3 and 105 ± 2 M -1 s -1 for 7 and 8, respectively (inset). n = 3. i-j, Box and whisker plot of normalized fluorescence intensity (F/F0) for reaction between 7 or 8 (50 µM) and various analytes (1 mM) after 1 h; n = 4. 91 ± 3 and 105 ± 2 M -1 s -1 for 7 and 8, respectively (Fig. 2g,h). To s c r e e n f o r p o t e n t i a l s i d e r e a c t i o n s , we 126 evaluated the reactivity of 7 and 8 with a panel of potentially reactive biomolecules, such as aldehyde 127 and disulfide electrophiles as well as related sulfur species including reduced glutathione, glutathione 128 nitrosothiol, glutathione sulfinic and sulfonic acid. No significant reaction took place between 7 or 8 and 129 any of the aforementioned compounds (Fig. i,j). Ta Supplementary Fig. 6). Fluorescence was distributed throughout 158 the cytoplasm with no apparent co-localization and minimal cytotoxicity ( Supplementary Fig. 7-8). Live-159 cell labeling by non-fluorogenic analog 6 was also investigated. In contrast to CysOx probes, compound 160 6 failed to promote oxidant-dependent "turn-on" fluorescence, instead giving weak signal throughout the 161 duration of the experiment ( Supplementary Fig. 9), consistent with our CSA studies. In-gel fluorescence 162 analysis of lysates derived from HeLa cells incubated with CysOx probes showed concentration-and 163 redox-dependent protein labeling by CysOx2 (Fig. 4c); however, signal from CysOx1 was much weaker 164 ( Supplementary Fig. 6c), reflecting the inherent difference in their fluorogenic intensities. Collectively, 165  Table 2). Protein S-sulfenylation was 178 modulated by inhibition of RTKs as well as select kinases from the AGC and CMGC families, including 179 the multitasking Ser/Thr kinase GSK3 (Fig. 4b). Upon closer inspection, three of the ten compounds 180 associated with the greatest increase in fluorescence intensity (SB-216763, BIM-IX, and BIO) were 181 identified as GSK3 inhibitors (Fig. 4a-c), suggesting a relationship between this idiosyncratic kinase and 182 protein S-sulfenylation. Underscoring this point, application of the most effective inhibitor identified in this 183 screen, SB-216763 promoted a fluorescence increase equivalent to that of exogenously applied peroxide 184 Data are representative of ten independent readings from five different frames. Error bars are ±SEM. Variance was analyzed by two-way ANOVA test. ns P > 0.05, * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001 when compared against cells treated with probe only. c, In-gel fluorescence analysis of lysates derived from HeLa cells incubated with CysOx2 and GOX (20 U/mL) or t-BOOH (500 µM). (Fig. 4c). The ability of SB-216763 to increase cellular protein S-sulfenylation was also confirmed through 185 fluorescence imaging microscopy (Fig. e,f) and in-gel fluorescence analysis ( Supplementary Fig. 11). 186 187

cells.
Having identified an interesting connection between GSK inhibition and cysteinyl oxidation, we 189 next sought to site-specifically and quantitatively profile dynamic changes in the HeLa S-sulfenylome 190 induced by GSK-3 inhibitors using BTD-based chemoproteomics (Fig. 5a). BTD is a clickable chemical 191 probe for sulfenic acid detection 24 . In agreement with fluorescence screening data, SB-216763 treatment 192 resulted in the most significant perturbation to the S-sulfenylome, with 56.3% measured S-sulfenylation 193 events exhibiting at least a 1.5-fold change (Fig. 5b). Moreover, S-sulfenylation levels in nearly half of 194 quantified sites with Uniprot functional annotations were changed by SB-216763 (Fig. 5c)  Data are representative of ten independent readings from five different frames. Error bars are ±SEM. Variance was analyzed by two-way ANOVA test. ns P > 0.05, * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001 when compared against cells treated with probe only. catalytic cysteine is notoriously susceptible to overoxidation under oxidative stress, whereas exogenous 200 H2O2 downregulates this redox event in the same cell line 14 . GSK3 inhibition and H2O2 both decreased 201 sulfenylation at C152 of GAPDH, which is also well-known redox sensitive site ( Supplementary Fig. 12). 202 These findings suggest that redox changes induced by GSK3 inhibition might be more specific as 203 opposed to simple diffusion of H2O2. 204

Discussion 205
In contrast to cysteinyl oxidation products, turn-on fluorescent probes for the detection of low-molecular-206 weight thiols like cysteine and glutathione have enjoyed a lengthy history of development and research 25 . 207 Such reagents have utility in cells and often show a degree of discrimination among different biothiols. 208 Despite such advances, the degree to which these tools report on cysteine residues in proteins is not 209 clear. Furthermore, this general class of reagents only detects biothiol oxidation as a loss of fluorescence 210 signal. To a d d r e s s t h i s s i g n i f i c a n t g a p , t his study presents the first instance of fluorogenic tools for direct 211 detection of protein cysteine or cysteinyl oxidation, specifically monitoring the formation of sulfenic acid. 212

213
The fluorogenic character of CysOx probes stems, at least in part, from the fluorescence quenching effect 214 of the phenaline-1,3-dione keto tautomer, in which nearly parallel carbonyl groups gives the largest large 215 dipole moment 26 . Reaction of the phenaline-1,3-dione scaffold at nucleophilic C-2 with the electrophilic 216 sulfur atom in sulfenic acid serves to stabilize the fluorogenic enol tautomer. A second key contributor to 217 the fluorescence of these probes is the electron-withdrawing fluorine atom attached to the naphthalene 218 core, which serves to stabilize the excited state 27 . A final feature of CysOx probes is the push-pull system 219 wherein the electron-donating amino group is in conjugation with fluorine via the intervening p-system 28 . 220 221 Another aspect of this study was to apply CysOx technology to identify kinase inhibitors that modulate S-222 sulfenylation in cells. The curated screening library used in this study includes inhibitors of lipid, receptor 223 and non-receptor tyrosine, serine/threonine, and dual specificity kinases 29 . Several interesting patterns 224 emerged from this screen. For instance, among inhibitors associated with a significant increase in cellular 225 S-sulfenylation, about 40% targeted kinases in the RTK family. This is consistent with reports from our 226 lab and others indicating that growth factor signaling is intimately linked to ROS metabolism in cells 5,8 . 227 The balance of "hits" within this cohort target members of the AGC and CMGC families and, of special 228 interest, three of these inhibit GSK3 function 30 . GSK3 in an unusual kinase in that it is constitutively 229 active and its substrates (over 100 are known) need to be pre-phosphorylated by another kinase, and it 230 is inhibited as opposed to activated by the two pathways known to converge on GSK3, the insulin and identifying agents or conditions that modulate protein S-sulfenylation, an area of central importance to 233 many fields including covalent targeting of semi-conserved cysteines which is a growing strategy in drug 234 design 32,33 . 235

236
To independently verify the relationship between GSK3 inhibition and cysteinyl oxidation we performed 237 chemoproteomic analysis using the sulfenic acid-selective probe, BTD. We found that GSK3 inhibitors 238 did indeed increase cellular S-sulfenylation with SB-216763 exhibiting the greatest effect, followed by 239 BIM-IX and BIO, consistent with the findings from our kinase inhibitor screen. In this regard, we note 240 well were directly analyzed in an Olympus FluoView IX81 confocal microscope using the FV10-ASW 3.0 software. For tBOOH treatment: Oxidant was added at the indicated concentrations in EMEM for 10 min 308 at 37 o C followed by the addition of CysOx probe at the indicated concentration for additional 15 min at 309 37 °C. The media was then aspirated, cells washed twice with DPBS, and fresh DPBS added. The wells 310 were then analyzed in an Olympus FluoView IX81 confocal microscope using FV10-ASW 3.0 software. 311 The 458 nm laser channel was used to excite incorporated CysOx probes. Fluorescence signal was 312 filtered using a SDM560 dichroic mirror, followed by BA505-605 band pass filter to observe the CysOx 313 fluorescence. Five frames were recorded per condition. Images were analyzed using ImageJ software, 314 where the pixel intensity of the cellular cytoplasmic regions was measured with at least 10 measurements 315 per condition.