Oxidized Forms of Ergothioneine Are Substrates for Mammalian Thioredoxin Reductase

Ergothioneine (EGT) is a sulfur-containing amino acid analog that is biosynthesized in fungi and bacteria, accumulated in plants, and ingested by humans where it is concentrated in tissues under oxidative stress. While the physiological function of EGT is not yet fully understood, EGT is a potent antioxidant in vitro. Here we report that oxidized forms of EGT, EGT-disulfide (ESSE) and 5-oxo-EGT, can be reduced by the selenoenzyme mammalian thioredoxin reductase (Sec-TrxR). ESSE and 5-oxo-EGT are formed upon reaction with biologically relevant reactive oxygen species. We found that glutathione reductase (GR) can reduce ESSE, but only with the aid of glutathione (GSH). The reduction of ESSE by TrxR was found to be selenium dependent, with non-selenium-containing TrxR enzymes having little or no ability to reduce ESSE. In comparing the reduction of ESSE by Sec-TrxR in the presence of thioredoxin to that of GR/GSH, we find that the glutathione system is 10-fold more efficient, but Sec-TrxR has the advantage of being able to reduce both ESSE and 5-oxo-EGT directly. This represents the first discovered direct enzymatic recycling system for oxidized forms of EGT. Based on our in vitro results, the thioredoxin system may be important for EGT redox biology and requires further in vivo investigation.

While the exact role of EGT in vivo has not been determined, it possesses remarkable antioxidant and cytoprotective properties in vitro. For example, EGT has been shown to be a powerful scavenger of singlet oxygen ( 1 O 2 ), hydroxyl radicals, and hypochlorite (HOCl) in vitro with the ability to protect other molecules from oxidative damage [2][3][4][5][6][7][8][9][15][16][17][18][19][20]. Experiments have shown that OCTN1 knockout cells are more prone to DNA damage, protein oxidation, and lipid peroxidation, suggesting EGT functions as an antioxidant and cytoprotective agent in vivo [5]. Another possible function for EGT is in the detoxification of xenobiotic electrophiles [21].  [26,27]. (Right) EGT has two tautomeric forms: thione (EGT) and thiol (ESH), with the thione being highly favored in neutral aqueous solution. This study investigated whether oxidized forms of EGT are also substrates for Sec-TrxR. The oxidized forms of EGT and Asc have clear structural similarities, as highlighted in blue and red. Sec-TrxR, a selenoenzyme, is also able to reduce the number of other small molecule substrates, including S-nitrosoglutathione, lipoic acid/lipoamide, lipid hydroperoxides, and ubiquinone [28][29][30][31].
In this report, we tested whether Sec-TrxR or glutathione reductase (GR) could reduce the oxidized forms of EGT shown in Figure 1. TrxR is part of the thioredoxin system comprised of TrxR, thioredoxin (Trx), and NADPH [32][33][34]. GR is part of the glutathione system, which is made up in part by GR, glutathione (GSH), and NADPH [35].
We found that both GR and Sec-TrxR could reduce the oxidized forms of EGT, but GR could not catalyze the reduction in the oxidized forms without the addition of GSH to the assay. In contrast, selenium-containing Sec-TrxR was able to efficiently reduce all of the oxidized forms tested without the addition of an exogenous thiol. Our in vitro results imply that the thioredoxin system may make an important contribution to the redox biology of EGT in mammalian systems.

Materials
Cys free-base was purchased from Calbiochem (Billerica, MA, USA). Histidine·HCl monohydrate and 3-mercaptopropionic acid were purchased from ACROS Organics (Pittsburgh, PA, USA). Rose bengal and bromine were purchased from Sigma-Aldrich  [26,27]. (Right) EGT has two tautomeric forms: thione (EGT) and thiol (ESH), with the thione being highly favored in neutral aqueous solution. This study investigated whether oxidized forms of EGT are also substrates for Sec-TrxR. The oxidized forms of EGT and Asc have clear structural similarities, as highlighted in blue and red. Sec-TrxR, a selenoenzyme, is also able to reduce the number of other small molecule substrates, including S-nitrosoglutathione, lipoic acid/lipoamide, lipid hydroperoxides, and ubiquinone [28][29][30][31].
In this report, we tested whether Sec-TrxR or glutathione reductase (GR) could reduce the oxidized forms of EGT shown in Figure 1. TrxR is part of the thioredoxin system comprised of TrxR, thioredoxin (Trx), and NADPH [32][33][34]. GR is part of the glutathione system, which is made up in part by GR, glutathione (GSH), and NADPH [35].
We found that both GR and Sec-TrxR could reduce the oxidized forms of EGT, but GR could not catalyze the reduction in the oxidized forms without the addition of GSH to the assay. In contrast, selenium-containing Sec-TrxR was able to efficiently reduce all of the oxidized forms tested without the addition of an exogenous thiol. Our in vitro results imply that the thioredoxin system may make an important contribution to the redox biology of EGT in mammalian systems.

Materials
Cys free-base was purchased from Calbiochem (Billerica, MA, USA). Histidine·HCl monohydrate and 3-mercaptopropionic acid were purchased from ACROS Organics (Pittsburgh, PA, USA). Rose bengal and bromine were purchased from Sigma-Aldrich (Milwaukee, WI, USA). Deuterium oxide (D, 99.9%) was purchased from Cambridge Isotope Laboratories (Andover, MA, USA). Hydrogen peroxide (30% in water) was purchased from Fisher Scientific (Waltham, MA, USA). All other chemicals were purchased from Sigma-Aldrich, Fisher Scientific, or ACROS Organics. 1 H-NMR and 13 C-NMR spectra were recorded with a Bruker Advance III HD 500 MHz NMR spectrometer (Bruker, Billerica, MA, USA). All enzyme kinetics assays were performed on a Cary 50 ultraviolet−visible (UV−Vis) spectrophotometer (Varian, Walnut Creek, CA, USA). Mass spectrometric (MS) analysis was either performed on a Thermo Scientific™ Q Exactive™ Hybrid Quadrupole-Orbitrap Mass Spectrometer (Waltham, MA, USA) or an Applied Biosystems QTrap 4000 hybrid triple-quadrupole/linear ion trap liquid chromatograph-mass spectrometer (LCMS) (SciEx, Framingham, MA, USA). The visible light source was a 500 W WorkForce™ lamp with a UV filter. Ergothioneine was a gift from Dr. E. Will Taylor of the University of North Carolina-Greensboro. Selenoneine was a gift from Dr. Florian Seebeck of the University of Basel.

Synthesis of L-2-Thiohistidine
L-2-thiohistidine was synthesized according to the procedure from Erdelmeier et al., 2012 [36]. This reaction works best when performed on a 10 g or higher scale. Histidine·HCl monohydrate (14 g, 66.8 mmol, 1.0 eq.) was dissolved in 134 mL of deionized water. After the His was fully dissolved, the solution was cooled in an ice bath to 0 • C. Once the reaction was cooled, bromine (4.45 mL, 86.8 mmol, 1.3 eq.) was added rapidly dropwise, resulting in a bright orange solution. After 6 min, free-base Cys (24.3 g, 200.4 mmol, 3.0 eq.) was added to the reaction, resulting in a yellow solution. The solution was stirred at 0 • C for 1 h. An oil bath was preheated to 95 • C. After 1 h, 3-mercaptopropionic acid (34.9 mL, 400.7 mmol, 6.0 eq.) was added to the reaction, and the reaction was transferred to the oil bath at 95 • C. A condenser was attached to the reaction, and the reaction was stirred for 18 h at 95 • C, after which the reaction had turned dark brown. The reaction was removed from the oil bath and condenser and allowed to cool to room temperature. The aqueous solution was then extracted with ethyl acetate (4×, 100 mL). The aqueous layer remained dark brown after extraction. The aqueous layer was transferred to a clean flask and placed in an oil bath preheated to 40 • C. Nitrogen gas was bubbled through the solution at 40 • C to help remove excess HBr in solution. The pH of the solution was adjusted to 6.5 with 30% ammonia hydroxide while nitrogen was being bubbled through the solution to precipitate 2-thioHis. The reaction was cooled to room temperature and then chilled on ice to allow complete precipitation. The off-white precipitate was filtered out of the reaction and washed with cold deionized water and ethanol. The precipitate was dried under high vacuum to provide 5.04 g (26.9 mmol) of an off-white powder. The percent yield of this reaction varies between 40% and 50% in our hands, which is consistent with the findings of Erdelmeier and coworkers [36]. MS analysis revealed a peak at 188.1 m/z. 1

Sec-TrxR Activity Assay with ESSE
A 10 mM solution of EGT was prepared in 0.1 M potassium phosphate, pH 7.0. To this solution, 5 mM H 2 O 2 , freshly prepared in deionized water, was added to form the disulfide ESSE. This reaction was allowed to incubate at room temperature in the dark for 10 min. The activity of Sec-TrxR with ESSE was measured by monitoring the decrease in NADPH absorbance at 340 nm over 2 min using an extinction coefficient of 6220 M −1 cm −1 . A 20 mM stock solution of NADPH was prepared in deionized water and frozen in aliquots to be used for activity assays. The ability of Sec-TrxR to reduce ESSE was measured in 500 µL assays with 200 µM NADPH and oxidized ESSE concentrations ranging from 5 to 500 µM. All assays were initiated with the addition of 10 nM Sec-TrxR enzyme. Controls were performed with corresponding concentrations of H 2 O 2 and 2-thioHis or EGT alone using the same NADPH and enzyme concentrations. Parallel experiments were performed with 2-thioHis and selenoneine at concentrations ranging from 0.1 to 2 mM and 0.1 to 4 mM, respectively. The assays with 2-thioHis and selenoneine were initiated with 5 nM Sec-TrxR. Selenoneine exists as the diselenide in aqueous solution at neutral pH at room temperature, so it was not oxidized with H 2 O 2 [13]. All assays were performed in triplicate, and the average of these trials was reported.
Additional assays were performed with Sec-TrxR, Trx, and ESSE. These assays contained 10 µM Trx, 200 µM NADPH, 10 nM Sec-TrxR, and ESSE in concentrations ranging from 5 to 300 µM. The activity of Sec-TrxR with 10 µM Trx was subtracted from all data points to remove background activity.

GR/GSH Activity Assay with ESSE
These assays were performed using the same procedure as for Sec-TrxR with ESSE, except GR/GSH were used in place of Sec-TrxR. These assays contained 200 µM GSH, 0.60 nM GR, and 200 µM NADPH. The concentration of ESSE was varied from 10 µM to 1 mM. The decrease in NADPH absorbance at 340 nm was monitored for 2 min. The assay was performed in triplicate, and the results were averaged. Analogous experiments were performed with 2-thioHis disulfide in the concentration range of 50 µM to 4 mM and were initiated by the addition of 0.75 nM GR. The background activity of GR with 200 µM GSH was subtracted from all data points.

NMR Experiments with EGT and H 2 O 2
1 H-NMR was used to quantify desulfurization of EGT upon treatment with H 2 O 2 . One NMR sample was prepared with 10 mM EGT in D 2 O, and one was prepared with 10 mM EGT and 5 mM H 2 O 2 in D 2 O. Quantitative 1 H-NMR was obtained on a Bruker Advance III HD 500 MHz NMR spectrometer for both samples. To obtain quantitative spectra, a D1 of 69 s was used based on a T1 relaxation value of 14.42 s for the C2 proton of hercynine; 2 dummy scans and 8 scans were obtained over a time period of~12 min. The H 2 O 2 was added to the second sample exactly 10 min before recording NMR spectra.

p-Nitrosodimethylaniline (RNO) Bleaching
Singlet oxygen was produced using the photosensitizer rose bengal in a 90% D 2 O solution by shining light on the solution following the general methods described by Kochevar and Redmond and Herman and Neal [39,40]. Conditions for producing 1 O 2 and oxidizing EGT were optimized using RNO bleaching to monitor production/scavenging of 1 O 2 using the procedure from Herman and Neal and Kraljić and El Mohsni [40,41]. Samples with 10 µM rose bengal and 8 mM imidazole and 10 µM rose bengal alone were prepared in 20 mM potassium phosphate-buffered 90% D 2 O, pH 7.0. To both samples, 50 µM RNO was added. An initial absorbance scan of both samples was taken, noting the absorbance of RNO at 440 nm. Both samples were then incubated under light, and absorbance scans were obtained every 3-4 min for 10 min. In addition, samples of 1 mM 2-thioHis with 10 µM rose bengal in D 2 O were placed on ice and irradiated with light for 20 min while monitoring changes in absorbance of 2-thioHis at 255 nm. Mobile phase flow was maintained at 100 µL/min. Source temperature was maintained at 400 • C. Nitrogen was used for the sheath gas, auxiliary gas, and curtain gas. Sheath gas (GS1) flow was set at 40, auxiliary gas flow (GS2) at 50, curtain gas flow (CUR) at 30, and the declustering potential (DP) was set to 50. The mass spectrometer was operated in single quadrupole mode, scanning from m/z 100 to 1000.

Sec-TrxR Activity Assay with EGT Oxidized by 1 O 2
Singlet oxygen was produced using the procedure described in Section 2.8 with some alterations as described below. Stock solutions of 2 mM rose bengal in ethanol, and 10 mM EGT or 2-thioHis in deionized H 2 O were prepared and stored in the dark on ice and at room temperature, respectively. A 1 mL sample was prepared with 10 µM rose bengal and 1 mM EGT or 2-thioHis in D 2 O in a small glass test tube. This sample was incubated under a visible light source with a UV filter for 20 min. Following this incubation, the activity of Sec-TrxR with 1 O 2 oxidized 2-thioHis and EGT was measured by monitoring the decrease in NADPH absorbance at 340 nm over 2 min using an extinction coefficient of 6220 M −1 cm −1 . Aliquots of oxidized EGT or 2-thioHis were added to 500 µL assays in 0.1 M potassium phosphate buffer, pH 7.0 with 1 mM EDTA and 200 µM NADPH. Assays were initiated with 20 nM Sec-TrxR. To generate an activity curve for 1 O 2 oxidized EGT, 2 min assays were performed for concentrations ranging from 10 to 240 µM of the rose bengal/EGT solution. Analogous experiments were performed with 2-thioHis. The irradiated rose bengal/2-thioHis, or EGT solutions, were kept in amber tubes and either used fresh in assays or kept on ice and used within 30-45 min of irradiation by light as the oxidation products were not very stable at room temperature. Controls were performed with rose bengal and 2-thioHis alone added to Sec-TrxR and NADPH to ensure any observed activity was enzyme catalyzed.

MS Analysis of 1 O 2 -Oxidized EGT following Reduction with Sec-TrxR
To confirm that Sec-TrxR could reduce 1 O 2 -oxidized EGT, MS analysis was performed on a sample with 1 mM EGT and 10 µM rose bengal in 90% D 2 O, irradiated with light for 20 min, then diluted with pH 8.0 ammonium bicarbonate buffer and reacted with NADPH and Sec-TrxR for 10 min at 37 • C in a final volume of 1.262 mL. The final concentrations of compounds in the sample were as follows: 792 µM EGT, 7.92 µM rose bengal, 792 µM NADPH, and 190 nM Sec-TrxR in 80 mM ammonium bicarbonate buffer, pH 8.0. After incubation of the sample for 10 min, the sample was run through an Amicon ® Ultra 30K centrifugal filter unit (Millipore, Burlington, MA, USA), spinning the sample at 14,000× g for 5 min, to separate the enzyme from the rest of the sample. The flow-through from the filtration (no enzyme) was analyzed by direct-inject positive ESI MS. For these analyses, an Applied Biosystems QTrap 4000 hybrid triple-quadrupole/linear ion trap liquid chromatograph-mass spectrometer (SciEx, Framingham, MA, USA) was used. The same method and parameters described in Section 2.9 were used for this sample, except the mass range was altered to m/z 100 to 600. A control experiment was run using the same procedure outlined above, except 62 µL of deionized water was added in place of Sec-TrxR and NADPH to achieve the same final EGT and rose bengal concentrations. Following ultrafiltration, the flow-through from the control sample was analyzed using a Waters Xevo G2-S Q-TOF LCMS (Waters, Milford, MA, USA).

Selenium Dependency of ESSE Recycling by TrxR Experiments
The selenium dependency of reduction of ESSE by TrxR was determined by testing the ability of TrxR∆3 (truncated enzyme missing three C-terminal amino acids), TrxR-GCCG (Sec→Cys mutant), and DmTrxR to catalyze the reduction of ESSE. Assays were performed in 0.1 M potassium phosphate buffer containing 1 mM EDTA with 200 µM NADPH and 400 µM ESSE. The assays were initiated with 300 nM TrxR∆3, 114 (pH 7) or 68 (pH 8) nM TrxR-GCCG, and 80 nM DmTrxR. All enzymes were tested at both pH 7.0 and pH 8.0. The decrease in absorbance at 340 nm was monitored for 2 min for each sample.

Enzymatic Reduction of ESSE with TrxR
We generated ESSE by the reaction of EGT with H 2 O 2 (2:1) in 0.1 M potassium phosphate, pH 7.0. The disulfide of 2-thioHis was generated in an identical manner. We used 2-thioHis as a less expensive analog of EGT in order to experiment with the right conditions for generating various oxidized forms of EGT. Selenoneine does not exist in the reduced form and was provided as the diselenide form, and we, therefore, did not need to oxidize it further [13].
We analyzed the products of these reactions by electrospray MS. The results of these analyses are shown in Figure 2. Oxidation of EGT with H 2 O 2 shows more complete conversion to the disulfide form in comparison to 2-thioHis, with somewhat less desulfurization as shown by the presence of hercynine in the MS (Figure 2b). Hercynine is the name for the desulfurized form of EGT. Desulfurization results from overoxidation of ESSE, as previously discussed by Servillo and coworkers [15]. The overoxidation products of 2-thioHis that we detected are His, the sulfinic acid, and the disulfide-S-dioxide. One explanation for the differences in oxidation products between EGT and 2-thioHis is that oxidation of EGT results in a more hindered disulfide, which may be less susceptible to hydrolysis and overoxidation.
Early studies of the oxidation of EGT to ESSE reported that ESSE was only stable at acidic pH [42]. However, our data shows that ESSE is formed easily in water or neutral pH in the presence of EGT and H 2 O 2 in a 2:1 ratio and is stable for at least several hours. This agrees with more recent data from Servillo and coworkers, who showed that ESSE rapidly forms in the presence of various oxidants and then decomposes over a 36 h period [15]. They showed that oxidation of EGT by various oxidants proceeded to ESSE as an intermediate, which then underwent oxidative breakdown to the sulfinic acid and ESH when EGT and oxidant were combined in a 1:1 ratio [15]. As we combined EGT and oxidant in a 2:1 ratio, the lifetime of the disulfide should be considerably longer, as supported by our data.
After generating the disulfide in situ by addition of H 2 O 2 , we performed enzyme assays with Sec-TrxR by adding aliquots of the reaction mixture to an assay containing enzyme and NADPH. To ensure the activity that we were detecting was due to reduction by Sec-TrxR, we performed various controls without enzyme (or enzyme plus H 2 O 2 ), as shown in Figure 3. As is shown in the plot, activity is only present when the assay contains substrate (disulfide of 2-thioHis), NADPH, and enzyme. We subsequently repeated the same experiment by oxidizing EGT with H 2 O 2 . An important control in Figure 3 is the assay of Sec-TrxR in the presence of 1 mM H 2 O 2 (green trace), which shows no activity. Even though we have previously reported that H 2 O 2 is a "substrate" for Sec-TrxR, the K M for H 2 O 2 is~260 mM, with detectable activity only occurring at concentrations above 5 mM [38]. Under the conditions of our assay (1 mM), the presence of H 2 O 2 does not contribute to the observed activity. This assay was also performed either in the presence or absence of 10 µM E. coli Trx. We assayed selenoneine without the addition of H 2 O 2 because it autoxidizes to the diselenide form. Early studies of the oxidation of EGT to ESSE reported that ESSE was only stable at acidic pH [42]. However, our data shows that ESSE is formed easily in water or neutral pH in the presence of EGT and H2O2 in a 2:1 ratio and is stable for at least several hours. This agrees with more recent data from Servillo and coworkers, who showed that ESSE rapidly forms in the presence of various oxidants and then decomposes over a 36 h period [15]. They showed that oxidation of EGT by various oxidants proceeded to ESSE as an intermediate, which then underwent oxidative breakdown to the sulfinic acid and ESH when EGT and oxidant were combined in a 1:1 ratio [15]. As we combined EGT and oxidant in a 2:1 ratio, the lifetime of the disulfide should be considerably longer, as supported by our data.
After generating the disulfide in situ by addition of H2O2, we performed enzyme assays with Sec-TrxR by adding aliquots of the reaction mixture to an assay containing enzyme and NADPH. To ensure the activity that we were detecting was due to reduction by Sec-TrxR, we performed various controls without enzyme (or enzyme plus H2O2), as shown in Figure 3. As is shown in the plot, activity is only present when the assay contains substrate (disulfide of 2-thioHis), NADPH, and enzyme. We subsequently repeated the same experiment by oxidizing EGT with H2O2. An important control in Figure 3 is the assay of Sec-TrxR in the presence of 1 mM H2O2 (green trace), which shows no activity. Even though we have previously reported that H2O2 is a "substrate" for Sec-TrxR, the KM for H2O2 is ~260 mM, with detectable activity only occurring at concentrations above 5 mM [38]. Under the conditions of our assay (1 mM), the presence of H2O2 does not contribute to the observed activity. This assay was also performed either in the presence or The Michaelis-Menten plots of these enzyme assays are shown in Figure S1 of the Supporting Information. The addition of 10 μM E. coli Trx to the assay containing ESSE resulted in a 3-fold lower KM value as well as a lower kcat. This assay, in part, reflects in vivo conditions where Trx would be present. We chose a concentration of 10 μM Trx for the assay because higher values resulted in activity that was non-linear in the spectrophotometric assay even though 10 μM is less than the KM for E. coli Trx. While the data in  The Michaelis-Menten plots of these enzyme assays are shown in Figure S1 of the Supporting Information. The addition of 10 µM E. coli Trx to the assay containing ESSE resulted in a 3-fold lower K M value as well as a lower k cat . This assay, in part, reflects in vivo conditions where Trx would be present. We chose a concentration of 10 µM Trx for the assay because higher values resulted in activity that was non-linear in the spectrophotometric assay even though 10 µM is less than the K M for E. coli Trx. While the data in Figure 3 clearly shows that Sec-TrxR can directly reduce ESSE, Sec-TrxR and Trx can work together to reduce substrates. In such cases, Trx(SH) 2 can reduce ESSE to 2EGT, forming Trx(S-S). Sec-TrxR then reduces the Trx(S-S) back to Trx(SH) 2 to restart the cycle. This explains why there is a 3-fold lower K M value when Trx is added to the assay because Trx has a higher binding affinity for Sec-TrxR compared to ESSE.
One interesting phenomenon that we observed was that we could detect NADPH consumption by Sec-TrxR when either EGT or 2-thioHis was present in the assay, but in the absence of H 2 O 2 after a long lag phase, as shown in Figure 4. The standard method for assaying TrxR is to observe the consumption of NADPH in the presence of substrate for the first minute of the assay. We have typically observed 2 min of data in all of our studies on TrxR. When performing control experiments during this time interval, we observed no consumption of NADPH when either 2-thioHis or EGT was present in the assay in the absence of oxidants such as H 2 O 2 . Serendipitously, we made the observation that the addition of either EGT or 2-thioHis to the assay in the absence of H 2 O 2 resulted in consumption of NADPH after about 3.5 min. This activity is equivalent to~20% of the activity compared to when H 2 O 2 is added to the assay. One explanation is that EGT and 2-thioHis slowly autoxidize to the disulfide after about 3 min. However, it has been reported that EGT resists autoxidation due to it being present largely as the thione and not the thiol [6,8,15]. Alternatively, TrxR is known to have NADPH oxidase activity that produces H 2 O 2 in the absence of its cognate substrate, Trx [43]. The observed lag phase is consistent with this hypothesis as it would take a small amount of time to produce enough H 2 O 2 to oxidize the EGT or 2-thioHis in solution. While this is one possible explanation, we cannot definitively explain this phenomenon. However, our MS data makes it clear that the addition of H 2 O 2 to EGT or 2-thioHis results in the formation of the disulfide, which is a substrate for Sec-TrxR.

Enzymatic Reduction of ESSE with GR/GSH
The other major antioxidant system in mammalian cells is the glutathione system. There are several enzymes of the glutathione system, including GR, glutaredoxin, and glutathione peroxidase [35]. The function of GR is to recycle oxidized glutathione (GSSG) to reduced glutathione (GSH) [35]. GSH is a general antioxidant in the cell and can function to reduce protein disulfide bonds and other low-molecular-weight disulfides such as ESSE [35,44]. Since mammalian GR is highly homologous to TrxR but lacks the Sec-containing C-terminal redox center, we decided to test whether GR alone or in combination with GSH could reduce ESSE [45,46].
We used identical conditions for generating the disulfide forms of EGT and 2-thioHis in our enzymatic assays with GR or GR/GSH as we used for our assays with Sec-TrxR. In

Enzymatic Reduction of ESSE with GR/GSH
The other major antioxidant system in mammalian cells is the glutathione system. There are several enzymes of the glutathione system, including GR, glutaredoxin, and glutathione peroxidase [35]. The function of GR is to recycle oxidized glutathione (GSSG) to reduced glutathione (GSH) [35]. GSH is a general antioxidant in the cell and can function to reduce protein disulfide bonds and other low-molecular-weight disulfides such as ESSE [35,44]. Since mammalian GR is highly homologous to TrxR but lacks the Seccontaining C-terminal redox center, we decided to test whether GR alone or in combination with GSH could reduce ESSE [45,46].
We used identical conditions for generating the disulfide forms of EGT and 2-thioHis in our enzymatic assays with GR or GR/GSH as we used for our assays with Sec-TrxR. In order to demonstrate that the consumption of NADPH that we observed in these experiments was due to enzymatic activity, we performed identical control experiments as was performed in our TrxR assays, starting with the disulfide of 2-thioHis. The result of this experiment is shown in Figure 5 and shows that the generated disulfide is a very poor substrate for GR itself. This is unlike Sec-TrxR. However, the addition of GSH to the assay greatly stimulates enzymatic activity. The reason for this is that GSH can attack the disulfide substrate (either ESSE or 2-thioHis disulfide) and form a mixed disulfide, as shown by Equation (1) in Figure 6. The mixed disulfide can be attacked by another equivalent of GSH to form GSSG (Equation (2)). The resulting GSSG can then be reduced by GR (Equation (3)). Another possibility is that GR efficiently reduces the mixed disulfide, as described by Equation (4). Comparing the kinetic parameters for the reduction of ESSE and 2-thioHis disulfide by GR provides insight into which mechanism is correct, as discussed below.  The Michaelis-Menten plots for the reduction of ESSE and 2-thioHis disulfide by the GR/GSH system are shown in Figure 7. One notable feature is that ESSE appeared to have an inhibitory effect on GR at high substrate concentrations, unlike 2-thioHis-disulfide, causing the activity to level off and then go back down ( Figure 7B). The different activity patterns observed between 2-thioHis-disulfide and ESSE were unexpected. The only difference between the molecules is that the amine of EGT is trimethylated, so it always car-  The Michaelis-Menten plots for the reduction of ESSE and 2-thioHis disulfide by the GR/GSH system are shown in Figure 7. One notable feature is that ESSE appeared to have an inhibitory effect on GR at high substrate concentrations, unlike 2-thioHis-disulfide, causing the activity to level off and then go back down ( Figure 7B). The different activity patterns observed between 2-thioHis-disulfide and ESSE were unexpected. The only difference between the molecules is that the amine of EGT is trimethylated, so it always carries a positive charge that is spread over the methyl groups. While this difference should The Michaelis-Menten plots for the reduction of ESSE and 2-thioHis disulfide by the GR/GSH system are shown in Figure 7. One notable feature is that ESSE appeared to have an inhibitory effect on GR at high substrate concentrations, unlike 2-thioHis-disulfide, causing the activity to level off and then go back down ( Figure 7B). The different activity patterns observed between 2-thioHis-disulfide and ESSE were unexpected. The only difference between the molecules is that the amine of EGT is trimethylated, so it always carries a positive charge that is spread over the methyl groups. While this difference should have little impact on the redox chemistry of the two compounds, it provides an important clue in differentiating whether GR is reducing GSSG in the assay as described by Equation (3) or whether GR is reducing the mixed disulfide (GSSE) as described by Equation (4). As listed in Table 1, the kcat values for the reduction of 2-thioHis disulfide and ESSE by GR/GSH are very similar, 16,425 min −1 and 20,035 min −1 , respectively. However, the KM values diverge by ~22-fold. If GSSG is being formed in the reaction of both disulfide substrates as described by Equation (3), one would expect very similar KM values. If the mixed disulfide is the substrate for GR, on the other hand, as described by Equation (4), different KM values are expected because one mixed disulfide has a trimethylated amine and the other one does not. The very different KM values argue strongly that the mixed disulfide is the substrate for GR. Another piece of evidence for this contention is that the concentration of GSH in the assay was kept constant at 200 μM. As shown by the data in Figure 7A, all of the assay points for 2-thioHis disulfide (except for two) were performed at a concentration of 500 μM or higher. At 500 μM 2-thioHis disulfide, 1 mM GSH is required to form GSSG, and this is 5-fold more than is actually present in the assay. The situation is amplified at the highest substrate concentration (4 mM), requiring 8 mM GSH in order to form GSSG. Other support for the mixed disulfide being the substrate comes from the literature. Eyer and Prodhradský investigated the mechanism of reduction of DTNB by GR/GSH and found that all of the GSH in the assay was consumed as described in Equation (1) of Figure  6 [47]. They concluded from their analyses that the mixed disulfide between GSH and 2nitro-5-mercapto-benzoic acid (TNB) was the substrate in their assay using GR/GSH. DTNB is a suitable model for ESSE as both disulfides are highly reactive due to the polar- As listed in Table 1, the k cat values for the reduction of 2-thioHis disulfide and ESSE by GR/GSH are very similar, 16,425 min −1 and 20,035 min −1 , respectively. However, the K M values diverge by~22-fold. If GSSG is being formed in the reaction of both disulfide substrates as described by Equation (3), one would expect very similar K M values. If the mixed disulfide is the substrate for GR, on the other hand, as described by Equation (4), different K M values are expected because one mixed disulfide has a trimethylated amine and the other one does not. The very different K M values argue strongly that the mixed disulfide is the substrate for GR. Another piece of evidence for this contention is that the concentration of GSH in the assay was kept constant at 200 µM. As shown by the data in Figure 7A, all of the assay points for 2-thioHis disulfide (except for two) were performed at a concentration of 500 µM or higher. At 500 µM 2-thioHis disulfide, 1 mM GSH is required to form GSSG, and this is 5-fold more than is actually present in the assay. The situation is amplified at the highest substrate concentration (4 mM), requiring 8 mM GSH in order to form GSSG. Other support for the mixed disulfide being the substrate comes from the literature. Eyer and Prodhradský investigated the mechanism of reduction of DTNB by GR/GSH and found that all of the GSH in the assay was consumed as described in Equation (1) of Figure 6 [47]. They concluded from their analyses that the mixed disulfide between GSH and 2-nitro-5-mercapto-benzoic acid (TNB) was the substrate in their assay using GR/GSH. DTNB is a suitable model for ESSE as both disulfides are highly reactive due to the polarization of the disulfide bond caused by electron-withdrawing groups on each sulfur atom. Another example from the literature is from Gruhlke and coworkers, who showed that the mixed disulfide between allicin and GSH was a very suitable substrate for GR [48]. Based on our analysis above and literature precedent, we conclude that the mixed disulfide (either GSSE or GSS2TH) is the substrate for GR in our assay.
We must strongly emphasize that our analysis above is predicated on the fact that the equilibrium constant for (1) is much greater than that of (2). If this is not the case and the equilibrium constants are similar such that the two equilibria are in competition, then our derived values of K M and k cat in Table 1 are meaningless.
Based on the structural difference between the two different types of mixed disulfides described above, we can offer a more specific reason for the inhibition observed in Figure 7B. The reason for the decrease in activity is that half of the disulfide that contains the positively charged trimethylamine of EGT could potentially mimic NADP + and disrupt the π-stacking and cation-π interactions between NADP + and FADH 2 at the NADPH binding site of GR [49][50][51][52]. This type of interaction is missing in the case of 2-thioHis. EGT is more likely to participate in stronger cation-π interactions than 2-thioHis due to the positively charged trimethylated amine group hosting a very stable cation [49][50][51]. Intracellular stores of EGT are typically kept in the low millimolar range [1,5,53]. A reduction in activity was not seen until a concentration of 1 mM ESSE (2 mM EGT) was reached, suggesting that the disruption of the π-stacking interaction by EGT occurs only at high physiological concentrations. Interestingly, we noticed that when we followed the reaction with ESSE over a long time period (10 min), the activity increased ( Figure S2 of the Supporting Information). A possible explanation for this lag phase is that once some of the disulfide is reduced, the inhibition is diminished.
A summary of the data for both TrxR and GR/GSH is given in Table 1. A comparison of ESSE as the substrate for both systems without Trx shows a~3-fold lower K M value and ã 7-fold higher k cat for GR/GSH. As a result, the catalytic efficiency of GR/GSH is~20-fold higher than TrxR alone. However, when comparing the GR/GSH system with the TrxR/Trx system, the difference in k cat values is~10-fold, and the K M values are similar. This results in a~10-fold higher catalytic efficiency for the GR/GSH system. However, the catalytic efficiency of both systems may be similar at a concentration of Trx that is close to the K M (physiological concentration), but we were unable to make this measurement in vitro. Our in vitro study cannot distinguish which antioxidant system is more important in vivo, and this remains a question for future investigation. However, TrxR has the advantage that it can directly reduce ESSE, and GR cannot.
Selenoneine is the selenium-analog of EGT and was discovered in 2010 in the blood of tuna fish [13,14]. Similar in function to EGT, selenoneine possesses powerful antioxidant properties in vitro and has also been shown to detoxify heavy metals such as mercury in vivo [14,54,55]. Humans obtain selenoneine by consuming fish and accumulating it in their cells via the cation/carnitine transporter OCTN1 [14,54,55]. It is interesting to note that selenoneine binds less tightly to Sec-TrxR, as reflected by a 5-fold higher K M value in comparison to ESSE but is only turned over 2-fold faster. As a result, the catalytic efficiency is~2-fold lower.
We must note that the activity of Sec-TrxR and GR/GSH toward ESSE is potentially higher than reported in Table 1 because MS analysis of the oxidation of EGT with H 2 O 2 ( Figure 2) revealed the presence of hercynine. This means that the calculated amount of disulfide substrate added to our assays is somewhat less than is reported in the plots of activity versus substrate concentration shown in Figure S1 of the Supporting Information.
Because the substrate was produced by the reaction of EGT with H 2 O 2 immediately before assaying, we have no way of correcting this small error. To determine if significant desulfurization occurred, 1 H-NMR experiments were performed with EGT in D 2 O and EGT in D 2 O incubated with H 2 O 2 for 10 min. The results of these experiments are shown in Figure S3 of the Supporting Information. The amount of desulfurization was calculated to be 3% in the oxidized sample compared to 2% in the unoxidized EGT sample, leading us to conclude that the amount of desulfurization that occurs as a result of oxidation of EGT with H 2 O 2 is not extensive in the time-frame of the enzyme kinetics assays.
While we used H 2 O 2 as the oxidant to generate ESSE in this study as a matter of convenience, Servillo and coworkers showed that several other biologically relevant oxidants oxidized EGT to ESSE at a much faster rate, especially HOCl [15]. HOCl is produced in neutrophils and monocytes as part of the innate immune system. We note that Servillo and coworkers not only studied the formation and decomposition of ESSE in vitro but demonstrated that ESSE forms in neutrophils and endothelial cells from various different oxidants, especially HOCl and superoxide [15,16]. Thus, there are multiple pathways for the biological formation of ESSE in animal cells, which could then be reduced back to EGT by Sec-TrxR or GR/GSH. We think it is logical that the cell would have a way to recycle oxidized forms of EGT, thus conserving it since the cell has a specific transporter for this dietary-derived nutrient.

Activity of Sec-TrxR toward 2-ThioHis and EGT Oxidized with 1 O 2
Singlet oxygen is a biologically significant oxidant produced in humans and other animal species [24,[56][57][58][59][60][61]. It is produced by cells of the immune system such as macrophages and neutrophils, where it is used to kill bacteria [62][63][64]. Singlet oxygen is generated by the enzymatic action of myeloperoxidase and superoxide dismutase, respectively, in these cell types [62][63][64]. It is notable that EGT, like vitamin C, is accumulated in these cells of the immune system [6,65,66]. We also note that EGT is a very good quencher of 1 O 2 , with a higher rate constant than GSH [24,67]. It has also been shown to be a superior quencher of 1 O 2 compared to Asc [20]. Singlet oxygen can also be produced in the skin and eye due to the presence of photosensitizers that catalyze the conversion of 3 O 2 (ground state) to 1 O 2 (excited state) [57,58]. In addition, the reaction of sunlight with protoporphyrin IX, the iron-free precursor of heme, produces 1 O 2 in erythropoietic cells [68].
We oxidized EGT using 1 O 2 that was generated by using the photosensitizer rose bengal in the presence of a visible light source in 90% D 2 O for 20 min. The decision to use D 2 O instead of water or buffer was made because the lifetime of 1 O 2 is significantly longer in D 2 O due to the shifted vibrational frequencies of the D-O bonds, allowing for more extensive oxidation of our compounds [69,70]. We then analyzed the products of this reaction using MS, which is shown in Figure 8. MS analysis of EGT oxidized with 1 O 2 showed the trace presence of the 5-oxo species at 244 m/z as well as the 5-hydroxy species at 246 m/z and ESSE at 229 m/z and 457 m/z ( Figure 7B). We note that the 5-hydroxy species is equivalent to the 5-oxo species, with the 5-hydroxy form existing at low pH (as occurs in the electrospray) and the 5-oxo species occurring at neutral pH. MS analysis of 2-thioHis oxidized with 1 O 2 is shown in Figure S4 of the Supporting Information.
The 5-oxo form of EGT is formed by oxidation of EGT with 1 O 2 by reaction of this excited form of oxygen with the C5 position to produce a hydroperoxide intermediate as shown in Figure 9, which then eliminates H 2 O to produce the 5-oxo form. This was shown recently by Gründemann and coworkers [20]. ESSE can then form from the 5oxo compound if another molecule of ESH attacks at the 4-position, forming a thioether intermediate. This step is followed by attack of a second molecule of ESH onto the sulfur of the thioether, resulting in the formation of ESSE. This is identical to the mechanism for EGT recycling by GSH proposed by Gründemann and coworkers resulting in the formation of GSSG [20]. The complete reduction of the 5-oxo form of EGT requires four equivalents of ESH to reduce 5-oxo-EGT to EGT, resulting in two equivalents of ESSE. The MS samples were run 30-45 min after completion of the reaction allowing for plenty of time for the formation of ESSE and other reduction products, which accounts for the significant amount of ESSE detected by MS, as shown in Figure 8B. We also note that the second reduction product has an m/z value of 230 for the doubly charged species, which is the same m/z value for singly charged EGT. Due to the resolution of our MS spectrum, we cannot determine if the 230 m/z peak is a mixture of these two species or not.
We oxidized EGT using 1 O2 that was generated by using the photosensitizer rose bengal in the presence of a visible light source in 90% D2O for 20 min. The decision to use D2O instead of water or buffer was made because the lifetime of 1 O2 is significantly longer in D2O due to the shifted vibrational frequencies of the D-O bonds, allowing for more extensive oxidation of our compounds [69,70]. We then analyzed the products of this reaction using MS, which is shown in Figure 8. MS analysis of EGT oxidized with 1 O2 showed the trace presence of the 5-oxo species at 244 m/z as well as the 5-hydroxy species at 246 m/z and ESSE at 229 m/z and 457 m/z ( Figure 7B). We note that the 5-hydroxy species is equivalent to the 5-oxo species, with the 5-hydroxy form existing at low pH (as occurs in the electrospray) and the 5-oxo species occurring at neutral pH. MS analysis of 2-thioHis oxidized with 1 O2 is shown in Figure S4 of the Supporting Information. The 5-oxo form of EGT is formed by oxidation of EGT with 1 O2 by reaction of this excited form of oxygen with the C5 position to produce a hydroperoxide intermediate as shown in Figure 9, which then eliminates H2O to produce the 5-oxo form. This was shown  recently by Gründemann and coworkers [20]. ESSE can then form from the 5-oxo compound if another molecule of ESH attacks at the 4-position, forming a thioether intermediate. This step is followed by attack of a second molecule of ESH onto the sulfur of the thioether, resulting in the formation of ESSE. This is identical to the mechanism for EGT recycling by GSH proposed by Gründemann and coworkers resulting in the formation of GSSG [20]. The complete reduction of the 5-oxo form of EGT requires four equivalents of ESH to reduce 5-oxo-EGT to EGT, resulting in two equivalents of ESSE. The MS samples were run 30-45 min after completion of the reaction allowing for plenty of time for the formation of ESSE and other reduction products, which accounts for the significant amount of ESSE detected by MS, as shown in Figure 8B. We also note that the second reduction product has an m/z value of 230 for the doubly charged species, which is the same m/z value for singly charged EGT. Due to the resolution of our MS spectrum, we cannot determine if the 230 m/z peak is a mixture of these two species or not. We confirmed that 1 O2 was generated in our reaction by measuring the change in absorbance of RNO after exposure to visible light in the presence of rose bengal for 10 min. Imidazole and its derivatives will react with 1 O2 and form a transannular peroxide, which then reacts with RNO resulting in "RNO bleaching", which can be detected by a change in absorbance at 440 nm [40,41]. In the absence of imidazole the UV-Vis spectrum showed little change at 440 nm, indicating very little oxidation of RNO, as expected. However, when we added imidazole to the same reaction, a large change at 440 nm was observed, indicating oxidation by 1 O2. An additional control was performed with rose bengal and 2-thioHis, monitoring the change in absorbance of 2-thioHis at 255 nm over a 20 min incubation. These control reactions are shown in Figures S5 and S6 of the Supporting Information.  [20]. Note that R = −CH2CH(COO − )N + (CH3)3.
We subsequently assayed the solutions of 2-thioHis and EGT that were oxidized by the 1 O2 that was generated from the photocatalytic reaction of rose bengal and 3 O2 in the  [20]. Note that R = −CH 2 CH(COO − )N + (CH 3 ) 3 .
We confirmed that 1 O 2 was generated in our reaction by measuring the change in absorbance of RNO after exposure to visible light in the presence of rose bengal for 10 min. Imidazole and its derivatives will react with 1 O 2 and form a transannular peroxide, which then reacts with RNO resulting in "RNO bleaching", which can be detected by a change in absorbance at 440 nm [40,41]. In the absence of imidazole the UV-Vis spectrum showed little change at 440 nm, indicating very little oxidation of RNO, as expected. However, when we added imidazole to the same reaction, a large change at 440 nm was observed, indicating oxidation by 1 O 2 . An additional control was performed with rose bengal and 2-thioHis, monitoring the change in absorbance of 2-thioHis at 255 nm over a 20 min incubation. These control reactions are shown in Figures S5 and S6 of the Supporting Information.
We subsequently assayed the solutions of 2-thioHis and EGT that were oxidized by the 1 O 2 that was generated from the photocatalytic reaction of rose bengal and 3 O 2 in the presence of visible light. The assay was performed by adding aliquots of the oxidized compounds immediately after exposure to 20 min of visible light to an assay mixture that contained 200 µM NADPH, potassium phosphate buffer, pH 7.0, and 20 nM Sec-TrxR. The result of this assay shows that NADPH was rapidly consumed in the presence of Sec-TrxR, but not when the enzyme was absent (compare red and blue traces in Figure 10A,B). When rose bengal is omitted from the reaction, there is very little consumption of NADPH for the first 3-4 min of the reaction. However, as the reaction progressed over a longer time period, we noticed consumption of NADPH that was less than when 1 O 2 was present due to the photocatalytic reaction of rose bengal and 3 O 2 , but more than when the enzyme was absent (compare black and blue traces in Figure 10A,B). This same phenomenon was observed previously in the oxidation of 2-thioHis/EGT by H 2 O 2, as shown in Figure 4. This apparent oxidation of 2-thioHis/EGT in the absence of added external oxidant is potentially explained by the NADPH oxidase activity of Sec-TrxR or by autoxidation of 2-thioHis/EGT in solution that is possibly catalyzed by trace metals in the solution. However, we cannot definitively explain this observation at this time. A further control was performed in which 2-thioHis/EGT was omitted from the reaction, but rose bengal, NADPH, and enzyme were all present. No consumption of NADPH was observed (green trace in Figure 10). This shows that rose bengal or an oxidized form of rose bengal is not a substrate for Sec-TrxR. The result of this assay shows that NADPH was rapidly consumed in the presence of Sec-TrxR, but not when the enzyme was absent (compare red and blue traces in Figure 10A,B). When rose bengal is omitted from the reaction, there is very little consumption of NADPH for the first 3-4 min of the reaction. However, as the reaction progressed over a longer time period, we noticed consumption of NADPH that was less than when 1 O2 was present due to the photocatalytic reaction of rose bengal and 3 O2, but more than when the enzyme was absent (compare black and blue traces in Figure 10A,B). This same phenomenon was observed previously in the oxidation of 2-thioHis/EGT by H2O2, as shown in Figure 4. This apparent oxidation of 2-thioHis/EGT in the absence of added external oxidant is potentially explained by the NADPH oxidase activity of Sec-TrxR or by autoxidation of 2-thio-His/EGT in solution that is possibly catalyzed by trace metals in the solution. However, we cannot definitively explain this observation at this time. A further control was performed in which 2-thioHis/EGT was omitted from the reaction, but rose bengal, NADPH, and enzyme were all present. No consumption of NADPH was observed (green trace in Figure 10). This shows that rose bengal or an oxidized form of rose bengal is not a substrate for Sec-TrxR. As evidenced by the Michaelis-Menten plots in Figure S7 of the Supporting Information and the kinetic data summarized in Table 2, Sec-TrxR displayed better saturation toward aliquots of the solution of EGT that was oxidized by 1 O2 compared to 2-thioHis resulting in a 2.5-fold lower KM value, but a ~4-fold lower kcat. The small differences in kinetic parameters may be attributed to greater steric hindrance in the oxidized form of EGT due to the trimethylated amine. We must note that our MS analysis shows that solutions of 2-thioHis or EGT that are oxidized with 1 O2 are a mixture of the 5-oxo and 5hydroxy forms as well as the disulfide form, which we have shown is also a substrate for Sec-TrxR. Thus, our observed values of kcat and KM do not reflect the true values for these parameters. Table 2. Kinetic parameters of Sec-TrxR with 2-thioHis and EGT oxidized with 1 O2 as substrates.

Enzyme
Substrate As evidenced by the Michaelis-Menten plots in Figure S7 of the Supporting Information and the kinetic data summarized in Table 2, Sec-TrxR displayed better saturation toward aliquots of the solution of EGT that was oxidized by 1 O 2 compared to 2-thioHis resulting in a 2.5-fold lower K M value, but a~4-fold lower k cat . The small differences in kinetic parameters may be attributed to greater steric hindrance in the oxidized form of EGT due to the trimethylated amine. We must note that our MS analysis shows that solutions of 2-thioHis or EGT that are oxidized with 1 O 2 are a mixture of the 5-oxo and 5-hydroxy forms as well as the disulfide form, which we have shown is also a substrate for Sec-TrxR. Thus, our observed values of k cat and K M do not reflect the true values for these parameters. In order to verify that the activity we measured in our assay was at least in part due to the ability of Sec-TrxR to reduce the 5-oxo form of 2-thioHis/EGT, we added 200 µL of 500 mM ammonium bicarbonate, pH 8.0, to 1 mL of the irradiated reaction mixture containing 10 µM rose bengal and 1 mM EGT in 90% D 2 O followed by addition of Sec-TrxR (190 nM) and NADPH (800 µM). This reaction was incubated for 10 min at 37 • C. The enzyme was removed by ultrafiltration, and the flow-through was submitted for MS analysis. The resulting MS data is shown in Figure 11. In order to verify that the activity we measured in our assay was at least in part due to the ability of Sec-TrxR to reduce the 5-oxo form of 2-thioHis/EGT, we added 200 μL of 500 mM ammonium bicarbonate, pH 8.0, to 1 mL of the irradiated reaction mixture containing 10 μM rose bengal and 1 mM EGT in 90% D2O followed by addition of Sec-TrxR (190 nM) and NADPH (800 μM). This reaction was incubated for 10 min at 37 °C. The enzyme was removed by ultrafiltration, and the flow-through was submitted for MS analysis. The resulting MS data is shown in Figure 11. Figure 11. Mass spectra of 1 O2-oxidized EGT followed by immediate addition of Sec-TrxR and NADPH. Comparison of the data above with the data in Figure 8B shows the disappearance of the peaks corresponding to 5-oxo-EGT and 5-hydroxy-EGT. Our interpretation is that Sec-TrxR can directly reduce the 5-oxo and 5-hydroxy forms back to EGT.
This MS experiment showed that Sec-TrxR completely reduced the 5-oxo and 5-hydroxy forms of EGT, as evidenced by the disappearance of the peaks at m/z = 244, m/z = 246, and m/z = 264. There was almost complete reduction of the disulfide ESSE (compare Figure 8B with Figure 11). In fact, the resulting MS spectrum contains almost all reduced EGT as the sodium or potassium adducts or two EGT molecules aggregated together with sodium or potassium. This MS experiment provides direct evidence of the reduction of ESSE, 5-hydroxy-EGT, and 5-oxo-EGT by Sec-TrxR, confirming that the recycling of these substrates by Sec-TrxR is responsible for the activity seen in our enzyme kinetics assays. The reduction of 5-oxo-EGT by Sec-TrxR, prior to MS analysis, was performed in ammonium bicarbonate buffer pH 8.0, which is different from our spectrophotometric assays, which were performed in potassium phosphate buffer, pH 7.0. As a control, we measured the rate of reduction of 1 O2-oxidized EGT by Sec-TrxR with spectrophotometry under the same conditions to ensure the enzyme had high activity in this buffer ( Figure S8 of the Supporting Information). A control MS experiment was run to verify that 5-oxo-EGT, 5hydroxy-EGT, and ESSE did not undergo decomposition due to ultrafiltration ( Figure S9 of the Supporting Information). All three of these oxidized species were detected in the control sample, confirming that they were stable to ultrafiltration.
As alluded to earlier, our computed kcat and KM values for the 5-oxo form are most assuredly greatly underestimated because the measured activity is a combination of the reduction of the 5-oxo form, 5-hydroxy form, and the disulfide form. However, one relevant point of comparison would be the reduction of dehydroascorbate by Sec-TrxR since there is a degree of structural similarity between 5-oxo EGT and dehydroascorbate, as Figure 11. Mass spectra of 1 O 2 -oxidized EGT followed by immediate addition of Sec-TrxR and NADPH. Comparison of the data above with the data in Figure 8B shows the disappearance of the peaks corresponding to 5-oxo-EGT and 5-hydroxy-EGT. Our interpretation is that Sec-TrxR can directly reduce the 5-oxo and 5-hydroxy forms back to EGT.
This MS experiment showed that Sec-TrxR completely reduced the 5-oxo and 5hydroxy forms of EGT, as evidenced by the disappearance of the peaks at m/z = 244, m/z = 246, and m/z = 264. There was almost complete reduction of the disulfide ESSE (compare Figure 8B with Figure 11). In fact, the resulting MS spectrum contains almost all reduced EGT as the sodium or potassium adducts or two EGT molecules aggregated together with sodium or potassium. This MS experiment provides direct evidence of the reduction of ESSE, 5-hydroxy-EGT, and 5-oxo-EGT by Sec-TrxR, confirming that the recycling of these substrates by Sec-TrxR is responsible for the activity seen in our enzyme kinetics assays. The reduction of 5-oxo-EGT by Sec-TrxR, prior to MS analysis, was performed in ammonium bicarbonate buffer pH 8.0, which is different from our spectrophotometric assays, which were performed in potassium phosphate buffer, pH 7.0. As a control, we measured the rate of reduction of 1 O 2 -oxidized EGT by Sec-TrxR with spectrophotometry under the same conditions to ensure the enzyme had high activity in this buffer ( Figure S8 of the Supporting Information). A control MS experiment was run to verify that 5-oxo-EGT, 5-hydroxy-EGT, and ESSE did not undergo decomposition due to ultrafiltration ( Figure S9 of the Supporting Information). All three of these oxidized species were detected in the control sample, confirming that they were stable to ultrafiltration.
As alluded to earlier, our computed k cat and K M values for the 5-oxo form are most assuredly greatly underestimated because the measured activity is a combination of the reduction of the 5-oxo form, 5-hydroxy form, and the disulfide form. However, one relevant point of comparison would be the reduction of dehydroascorbate by Sec-TrxR since there is a degree of structural similarity between 5-oxo EGT and dehydroascorbate, as shown in Figure 1. May and coworkers reported a k cat of 90 min −1 and a K M of 2.5 mM for the reduction of dehydroascorbate by Sec-TrxR [27]. Thus, 5-oxo EGT is a better substrate for Sec-TrxR by a factor of~100 as measured by k cat /K M .
As shown in Figure 1, the reduction of the oxidized forms of vitamin C by Sec-TrxR is analogous to the reduction of the oxidized forms of EGT by Sec-TrxR, with the exception of the disulfide form of EGT, which does not exist for ascorbate. The relationship in Figure 1 underscores the "vitamin-like" nature of EGT and the role of TrxR in recycling oxidized forms of ascorbate and EGT.

Selenium Dependence of the Reactions
After demonstrating that Sec-TrxR can reduce ESSE, we sought to determine if the reaction is selenium dependent. To test the selenium dependency of the reaction, EGT oxidized by H 2 O 2 was used as a substrate for TrxR∆3 (truncated enzyme missing three C-terminal amino acids), TrxR-GCCG (Sec→Cys mutant of Sec-TrxR), and DmTrxR (a Cys-ortholog from D. melanogaster). We note that this study uses the mitochondrial TrxR, which allows for better comparison to the Cys-ortholog TrxRs in this study as previously discussed by us [71].
The results of these assays are summarized in Table 3. The truncated enzyme TrxR∆3 had no ability to reduce ESSE even at 300 nM enzyme concentration, confirming that a C-terminus with Cys or Sec is necessary to reduce ESSE. The percent difference in activity between Sec-TrxR and TrxR∆3 was 99.8-99.9%, and activity showed little dependence upon pH. The mutant enzyme TrxR-GCCG had slightly higher activity than TrxR∆3, although it was 98.4-99.1% lower than Sec-TrxR. DmTrxR had more activity than either of the Sec-TrxR mutants and showed a greater dependence upon pH than either TrxR∆3 or TrxR-GCCG. The activity of DmTrxR toward ESSE was 94-97% lower than Sec-TrxR. Sec-TrxR showed the greatest pH dependence, with the activity toward ESSE almost doubling at pH 8 compared to pH 7. This is somewhat surprising because we had expected the activity of DmTrxR to increase more at higher pH because this would increase the proportion of Cys-thiolate relative to Cys-thiol. In contrast, the active site Sec residue should exist as the selenolate at both pH 7 and pH 8 since the pK a of Sec is~5.2 [72]. A possible explanation is that ESSE breaks down at pH 8 to ESOH (the sulfenic acid) and EGT, similar to the breakdown of DTNB at alkaline pH. The ESOH form could be a better substrate for Sec-TrxR than ESSE, with Sec-TrxR converting ESOH to EGT and water.

Conclusions
This report introduces a new and biologically relevant substrate for mammalian Sec-TrxR. We have demonstrated the ability of Sec-TrxR to directly reduce the 5-oxo and disulfide forms of EGT, produced by reaction with biologically relevant oxidants in vitro. The reduction of these oxidized forms of EGT is far more efficient with Sec-TrxR than it is with TrxR mutants without Sec or with the Cys-ortholog DmTrxR, showing that selenium has a major role in catalyzing the reaction. While this report demonstrates that Sec-TrxR can reduce oxidized forms of EGT in vitro, it remains to be seen how important the thioredoxin system is to recycling EGT in vivo since the glutathione system most likely makes an important contribution as well. Sec-TrxR has an advantage over GR as it can reduce ESSE directly without the aid of an exogenous thiol. This report brings us one step closer to understanding the biological chemistry and relevance of EGT.
While our results show that ESSE and 5-oxo-EGT are substrates for Sec-TrxR, we were unable to show that similar to Asc•, ES• was a substrate for Sec-TrxR because our initial experiments involved the use of various metals such as Fe 3+ to generate the radical and NADPH can directly reduce Fe 3+ in the absence of enzyme leading to very high background activity [73]. Since ES• and Asc• are obviously analogous structures (Figure 1), and May and coworkers have previously demonstrated that TrxR can reduce Asc• [26], we hypothesize that TrxR should be able to reduce ES• as well. EGT is a powerful radical scavenger and must form a radical intermediate during the reaction [74], but the half-life of ES• is most likely much shorter than that of Asc•, as ES• will quickly form the disulfide via a combination of two ES• radicals.