Thioredoxin Uses a GSH-independent Route to Deglutathionylate Endothelial Nitric-oxide Synthase and Protect against Myocardial Infarction*

Reversible glutathionylation plays a critical role in protecting protein function under conditions of oxidative stress generally and for endothelial nitric-oxide synthase (eNOS) specifically. Glutathione-dependent glutaredoxin-mediated deglutathionylation of eNOS has been shown to confer protection in a model of heart damage termed ischemia-reperfusion injury, motivating further study of eNOS deglutathionylation in general. In this report, we present evidence for an alternative mechanism of deglutathionylation. In this pathway thioredoxin (Trx), a small cellular redox protein, is shown to rescue eNOS from glutathionylation during ischemia-reperfusion in a GSH-independent manner. By comparing mice with global overexpression of Trx and mice with cardiomyocyte-specific overexpression of Trx, we demonstrate that vascular Trx-mediated deglutathionylation of eNOS protects against ischemia-reperfusion-mediated myocardial infarction. Trx deficiency in endothelial cells promoted eNOS glutathionylation and reduced its enzymatic activity, whereas increased levels of Trx led to deglutathionylated eNOS. Thioredoxin-mediated deglutathionylation of eNOS in the coronary artery in vivo protected against reperfusion injury, even in the presence of normal levels of GSH. We further show that Trx directly interacts with eNOS, and we confirmed that Cys-691 and Cys-910 are the glutathionylated sites, as mutation of these cysteines partially rescued the decrease in eNOS activity, whereas mutation of a distal site, Cys-384, did not. Collectively, this study shows for the first time that Trx is a potent deglutathionylating protein in vivo and in vitro that can deglutathionylate proteins in the presence of high levels of GSSG in conditions of oxidative stress.

Occlusion of coronary artery due to atherosclerotic plaque interrupts the flow of blood to the cardiac tissue, depriving it of nutrients and oxygen, resulting in the progressive death of affected tissue termed ischemia. When the occlusion is removed either by surgical or other interventions, ensuring the reflow of blood to the affected heart tissue, the process of reperfusion of ischemic tissue is established. Paradoxically, this reperfusion causes more damage to the affected heart tissue, which is known as reperfusion injury (1). The luminal surfaces of the coronary artery and their branches are lined with endothelial cells that are adversely impacted by ischemia-reperfusion resulting in loss or decreased production of vascular relaxing factors such as nitric oxide (NO). This dysfunction is known as endothelial dysfunction. Endothelial dysfunction is a key mechanism in the pathogenesis of myocardial infarction (MI) 2 due to ischemia/reperfusion (I/R) of the heart (2,3). Several studies have provided unequivocal evidence that increased production of vascular superoxide anion (O 2 . ) is a major cause of endothelial dysfunction (2)(3)(4). The sources of O 2 . in endothelial cells (ECs) include the mitochondrial electron transport chain, NADPH oxidases, activation of xanthine oxidase, and dysfunctional eNOS (1,5,6). Recent studies have shown that pretreatment with NO donors or drugs that increase NO synthesis prior to ischemia protects the myocardium against I/R injury (7-9) supporting a major role of eNOS dysfunction in the pathogenesis of myocardial injury. Although both eNOS and neuronal NOS are present in cardiomyocytes, endothelial cells within the heart express four times higher levels of eNOS compared with myocytes (10). Furthermore, 60% of the total non-myocyte cell population of heart is composed of endothelial cells (11), suggesting a crucial role of eNOS in the regulation of myocyte function. Consistent with this notion, NO in the heart affects the onset of ventricular relaxation that allows for a precise optimization of pump function (12), demonstrating that availability of endothelial NO is critical for heart function. Additionally, oxidative stress in I/R results in a dysfunctional eNOS that produces O 2 . by transferring electrons to molecular oxygen instead of L-arginine, resulting in uncoupling of the eNOS-L-arginine pathway of the production of NO. This uncoupling of eNOS further increases oxidative load and accentuates vascular dysfunction that sets the stage for various cardiovascular diseases such as atherosclerosis, hypertension, or myocardial infarction (13)(14)(15). It remains unclear whether maintaining the normal oxidation-reduction status (redox) of ECs during reperfusion of occluded coronary artery in vivo would prevent eNOS dysfunction and consequently protection against ischemia-reperfusion-mediated myocardial infarction in vivo. Besides uncoupling, eNOS is inactivated by conjugation of oxidized glutathione (GSSG) to cysteine moieties, referred to as "glutathionylation" under conditions of oxidative stress. Recently, it has been shown that increased concentration of GSSG due to the inhibition of glutathione reductase (GR) causes S-glutathionylation of eNOS at cysteine 689 and 908 residues of the reductase domain using cell culture studies (16,17). Furthermore, S-glutathionylated eNOS has been shown to uncouple eNOS leading to the production of O 2 . . eNOS S-glutathionylation is reversed by glutaredoxin (Grx1), a cytosolic oxidoreductase. Grx1 is an enzyme that catalyzes the transfer of electrons from glutathione (GSH) to glutathionylated proteins (PrS-SG). Utilization of GSH by Grx1 generates GSSG that is converted back to GSH via GR using NADPH (18). Grx1 efficiently reduces protein-mixed disulfide bonds by deglutathionylation in the presence of reduced GSH (18 -20). It has been shown that the efficiency of eNOS deglutathionylation by Grx1 is primarily influenced by the level of GSH (20). In addition to the GSH-Grx1 system, there is the ubiquitous presence of another powerful Trx redox system that is involved in the regulation of several important cellular functions. Trx is a 12-kDa small cytosolic redox protein that is primarily involved as an electron donor for ribonucleotide reductase, a rate-limiting enzyme in DNA replication (21). Trx also regenerates oxidatively inactivated proteins with reducing equivalents supplied by NADPH via thioredoxin reductase (TrxR) (22)(23)(24). Although potentially important, the role of Trx in deglutathionylation of mammalian proteins remains unknown. Because glutathionylation of eNOS in hypoxia reoxygenation is an underlying mechanism of endothelial dysfunction (25), we hypothesized that high levels of Trx would deglutathionylate eNOS in I/R due to its disulfide reductase activity. In this report, we show that I/R of the left anterior descending coronary artery (LAD) induced significant glutathionylation of eNOS concomitant with an increase in area-at-risk (AAR) in the heart. In contrast, coronary arteries from Trx-Tg mice with high levels of Trx failed to show significant glutathionylation of eNOS in I/R injury and were protected against coronary endothelial dysfunction in I/R. Furthermore, these mice showed significant reduction in AAR in I/R compared with non-transgenic littermate (NT) mice. However, mice with cardiac-specific overexpression of Trx (␣MHC-Trx-Tg) with normal levels of Trx in the coronary artery showed endothelial dysfunction in response to I/R, demonstrating a critical role of Trx in vascular endothelial function. Finally, we have shown that deglutathionylation of eNOS by Trx is a major mechanism of protection of eNOS in I/R in the presence of high levels of GSSG. Collectively, our data demonstrate a crucial role of Trx as an efficient deglutathionylating agent for eNOS in vivo in I/R that protects against endothelial dysfunction resulting in decreased MI.

Results
Endothelial but Not Cardiomyocyte-specific Trx Overexpression Protects against MI-Because endothelial cells constitute about 60% of total non-myocyte cell population in the heart (11), we determined the relative contribution of endothelial Trx and cardiomyocyte Trx in protection against I/R-related MI by using ␣MHC-Trx mice and Trx-Tg mice. As shown in Fig. 1, A (lower panel), and B, ␣MHC-Trx-Tg mice failed to show protection against I/R. There was no difference in the AAR of NT or ␣MHC-Trx-Tg mice in I/R (Fig. 1B). However, Trx-Tg mice expressing Trx globally (in heart myocytes and endothelial cells) showed significant protection against I/R with decreased MI (Fig. 1, A, middle panel, B and C). To determine whether the level of protection is associated with the level of expression of Trx in cardiomyocytes, we isolated cardiomyocytes from adult Trx-Tg and ␣MHC Trx-Tg mice and determined Trx expression levels in these cells. We did not observe a significant difference in Trx expression levels in cardiomyocytes from Trx-Tg and ␣MHC-Trx-Tg mice (Fig. 1, D and E). Because endothelial cells constitute more than 60% of non-cardiac cells by number, we next determined the expression level of Trx in ECs versus cardiomyocytes in Trx-Tg mouse hearts. Indeed, as seen in Fig.  1, F and G, ECs did express a 20 -25% higher level of Trx compared with cardiomyocytes. The expression levels of Trx in cardiomyocytes isolated from Trx-Tg or ␣MHC-Trx-Tg are similar, yet ␣MHC-Trx-Tg mice showed significant MI due to I/R, demonstrating a significant role of endothelial Trx in protection against I/R injury, because other non-myocyte cells such as fibroblasts or inflammatory cells constitute a very low percentage of cells in the heart, and these cells do not express eNOS. We also found increased apoptosis in the heart sections in NT or ␣MHC-Trx littermates, but the TUNEL-positive nuclei were significantly lower in the Trx-Tg heart sections ( Fig. 1, H and I). These data show that high levels of functional hTrx expressed in ECs in the heart could be significant in protection against I/R-related MI. Taken together, our data show that high levels of endothelial Trx in the heart is critical for the protection of the heart against I/R-mediated MI.
High Levels of Trx Maintain Coronary Endothelial Function during I/R-Because endothelial dysfunction due to I/R is a major contributing factor to cardiomyocyte death and resultant MI (26), we analyzed the effect of high levels of Trx in coronary endothelial dysfunction in I/R. As shown in Fig. 2A, I/R significantly blunted endothelium-dependent acetylcholine (ACh)mediated relaxations in left coronary arteries (LCAs) derived from NT mice that underwent I/R compared with sham-operated LCAs. Endothelium-dependent ACh-induced relaxation was preserved in LCAs from Trx-Tg mice after I/R (Fig. 2B). Because NO is the major vasodilator, we determined the effect of Trx on eNOS function in I/R. As shown in Fig. 2C, basal NO inhibition by L-NAME did not result in an enhanced contraction in LCAs derived from NT mice that underwent I/R. This is in contrast to LCAs derived from Trx-Tg I/R mice, which showed a significant increase in basal tone after the addition of L-NAME, indicating increased basal NO release (Fig. 2D). To determine whether vascular Trx is specifically required for improved endothelial function in I/R, LCA from ␣MHC-Trx-Tg mice was subjected to I/R. As shown in Fig. 2E, LCA from these mice failed to show protection against endothelial dysfunction in I/R, demonstrating that vascular Trx is specifically required for endothelial function in I/R.
Hypoxia Reoxygenation (H/R) Causes S-Glutathionylation and Decreased Activity of eNOS-Recent studies have demonstrated that H/R impairs eNOS function resulting in decreased levels of NO release in bovine aortic endothelial cells (25). Because eNOS is inactivated via S-glutathionylation (16), we examined whether NO production in human coronary artery endothelial cells (HCAEC) via eNOS is impaired in H/R due to S-glutathionylation. Exposure of HCAEC to H/R caused S-glutathionylation of eNOS in a sustained manner as detected by immunoblotting following immunoprecipitation of total cellular lysate with eNOS antibody (Fig. 3A). Treatment with 2-mercaptoethanol released the GSH from eNOS-SG indicating that S-glutathionylation of eNOS was indeed induced due to H/R (Fig. 3B). We also confirmed the glutathionylation of eNOS in HCAEC using immunofluorescence that shows colocalization of eNOS and PrS-SG (Fig. 3C). To determine whether the enzymatic activity of eNOS is compromised following exposure of HCAEC to H/R, we measured the eNOS activity in HCAEC exposed to H/R. As shown in Fig. 3, D and E, eNOS enzymatic activity was decreased in a time-dependent manner, and after 8 h of H/R eNOS activity was reduced to less than 1%. Taken together, these data establish that HCAEC in H/R fail to maintain adequate NO production due to S-glutathionylation of eNOS that diminished its enzymatic activity.
Depletion of Trx in HCAEC Causes eNOS S-Glutathionylation and Reduces Its Activity-We have previously shown that Trx undergoes significant oxidation in hypoxia resulting in its inactivity in cancer cells (27). Therefore, we speculated that the FIGURE 1. Trx-Tg mice are protected from I/R-induced myocardial infarction. A, NT, Trx-Tg, and ␣MHC-Trx-Tg mice were subjected to 30 min of ischemia and 1 h of reperfusion, and TTC staining was performed as described under "Materials and Methods." TTC stains viable tissue brick red and necrotic tissue as white. B, infarct area in relation to AAR; *, p Ͻ 0.05 versus NT or ␣MHC-Trx-Tg. C, percent protection relative to NT; *, p Ͻ 0.05 versus Trx-Tg; D, isolated cardiomyocytes from Trx-Tg and ␣MHC-Trx-Tg were lysed and analyzed for Trx and ␣-actinin by Western blotting. E, quantitation of D. F, cardiomyocytes and endothelial cells were isolated from Trx-Tg mice and lysed, and an equal amount of protein from cell lysates was analyzed for Trx, ␣-actinin, and ␤-actin by Western blotting using their specific antibodies. G, quantitation of F. *, p Ͻ 0.05 versus cardiomyocytes. H, apoptosis was evaluated by TUNEL assay of heart sections from NT, Trx-Tg, and ␣MHC-Trx-Tg mice that were subjected to I/R surgery. Images were obtained from infarct zone from endocardial border to the center via a ϫ10 objective using Zeiss AxioObserver Z2 microscope. Green, apoptotic cells; red, cardiomyocytes (␣-actinin); blue, nucleus (DAPI). I, percent TUNEL positive nuclei; *, p Ͻ 0.05 versus NT or ␣MHC-Trx-Tg. unavailability of functional Trx might promote glutathionylation eNOS. To test this hypothesis, we depleted Trx in HCAEC and determined its effect on eNOS glutathionylation and activity. As shown in Fig. 4A, Trx was effectively down-regulated due to siRNA treatment of HCAEC . This depletion of endogenous Trx promoted significant increase in eNOS glutathionylation (Fig. 4, B and C). We further observed that down-regulation of Trx in HCAEC resulted in differential distribution of eNOS (Fig. 4C). Therefore, we sought to determine whether Trx depletion promoted the translocation of glutathionylated eNOS to a specific organelle within the cell. Because eNOS distribution pattern in Trx-depleted cells coincided with lysosome distribution, and lysosomal protein degradation is a major mechanism of elimination of dysfunctional proteins, we determined whether eNOS undergoes lysosomal degradation under Trx-deficient conditions. As shown in Fig. 4D, in Trxdepleted cells exposed to H/R eNOS was localized within lysosome-associated membrane protein 1 (LAMP1) vesicle. Because lysosomal degradation of eNOS could result in the loss of a specific epitope reactive to the antibody and hence decreased eNOS fluorescence, we treated cells with the lysosomal proton pump inhibitor lansoprazole and determined the localization of eNOS. As shown in Fig. 4E, there was not much difference in the eNOS intensity in lansoprazole-treated cells, suggesting that the eNOS may remain in the lysosome for a prolonged time without loss of reacting epitope. Thus, lysosomal localization of glutathionylated eNOS suggests that eNOS may be undergoing lysosomal degradation in response to H/R. Furthermore, eNOS glutathionylation due to Trx deple-tion decreased its enzymatic activity (Fig. 4F). Additionally, treatment with L-NNA, an eNOS inhibitor, abolished the production of [ 3 H]citrulline indicating the specificity of the reaction for eNOS activity (Fig. 4G). These data show for the first time that Trx is a critical regulator of glutathionylation even in the presence of an intact GSH-Grx system. Furthermore, these results indicate that Trx is required to prevent glutathionylation and maintain eNOS function.
Overexpression of Trx Protects against eNOS S-Glutathionylation in HCAEC-Because depletion of Trx induces glutathionylation of eNOS, we speculated that high levels of Trx would protect against H/R-induced glutathionylation. HCAEC were transduced with adenovirus to overexpress Trx (Fig. 5A) and exposed to H/R. As shown in Fig. 5B, H/R induced eNOS glutathionylation in LacZ-expressing cells but not in HCAEC overexpressing Trx. Additionally, immunofluorescence staining showed that the colocalization of eNOS and PrS-SG was increased after H/R in LacZ-expressing cells but not in Trxoverexpressed cells as detected by staining for PrS-SG (Fig. 5C). To further confirm the glutathionylation of eNOS, we implemented the proximity ligation assay (PLA) (28). As shown in Fig. 5D, exposure of HCAEC to H/R induced increased PLA signal for eNOS and PrS-SG staining suggesting a significant increase in eNOS glutathionylation. However, when PLA was performed in cells with increased expression of Trx (Ad-Trx infection), the PLA signal was significantly decreased, suggesting either protection against S-glutathionylation or deglutathionylation of eNOS by Trx in H/R (Fig. 5, D and E).
Overexpression of Trx Restores eNOS Activity and NO Release-We analyzed the effect of Trx on eNOS enzymatic activity and NO release. Trx overexpression protected against H/R-induced loss in eNOS enzymatic activity (Fig. 6A). In addition, overexpression of Trx restored the ability of eNOS to catalyze conversion of L-arginine to L-citrulline (Fig. 6B). Specificity of the reaction is shown by treatment of cells with NOS inhibitor, L-NNA, that completely prevented the conversion of L-arginine to citrulline. We further analyzed the effect of Trx on NO release using the NO-sensitive probe DAF-FM. As shown in Fig. 6C, H/R decreased ACh-induced NO release, which was restored by overexpression of Trx in response to 10 M ACh. Collectively, these data establish that high levels of Trx prevent eNOS glutathionylation and consequently maintain eNOS function in H/R. Because it is well established that the Grx1-GSH system is the predominant deglutathionylating system in cells (29), we next determined whether high levels of Trx could protect eNOS function in the presence of increased levels of GSSG.
Trx Prevents Glutathionylation of eNOS in Presence of High Levels of GSSG-Inhibition of GR affects the redox balance of cells and causes glutathionylation of eNOS due to increased intracellular concentration of GSSG (16). We determined whether Trx overexpression would protect eNOS from S-glutathionylation in the presence of increased accumulation of GSSG due to inhibition of GR by 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU). As shown in Fig. 7A, treatment of cells with BCNU increased S-glutathionylation of eNOS, but overexpression of Trx protected eNOS from S-glutathionylation. We further tested whether depletion of GR by RNAi would promote  NOVEMBER 4, 2016 • VOLUME 291 • NUMBER 45 eNOS glutathionylation. As shown in Fig. 7B, GR was effectively down-regulated by GR siRNA, and this decrease in GR promoted eNOS glutathionylation. However, eNOS glutathionylation due to depletion of GR was abolished in the presence of high levels of Trx (Fig. 7C). Additionally, high levels of Trx restored eNOS enzymatic activity in the face of depletion of GR (Fig. 7D). Grx1 is the major deglutathionylation enzyme and uses reducing equivalents from GSH (20). To further delineate the role of Grx1, we depleted Grx1 (Fig. 7E) and determined eNOS glutathionylation. As shown in Fig. 7F, depletion of Grx1 induced eNOS glutathionylation. However, in the presence of a Trx-reducing system (Trx ϩ TrxR1 ϩ NADPH), glutathiony-lation of eNOS was rescued even in the absence of Grx1 (Fig.  7F). These data demonstrate that Trx is an efficient deglutathionylating agent and is functionally Grx1-independent. In addition, down-regulation of thioredoxin reductase 1(TrxR1) induced eNOS glutathionylation further demonstrating that the Trx redox cycle is an important regulator of eNOS function (Fig. 7, G and H). Taken together, our results demonstrate that Trx-dependent deglutathionylation of eNOS occurs in the absence of GSH, providing compelling evidence that Trx is a powerful deglutathionylating protein, and can function as an independent mechanism to protect eNOS against oxidant-mediated inactivation. A, HCAEC were exposed to H/R, lysed, and immunoprecipitated (IP) using anti-eNOS antibody, and the immunoprecipitates were analyzed by Western blotting using anti-PrS-SG or anti-eNOS antibodies. B, immunoprecipitates described in A were boiled with or without ␤-mercaptoethanol and analyzed by Western blotting using anti-PrS-SG or anti-eNOS antibodies. IB, immunoblot. C, HCAEC were exposed to H/R for indicated time periods, fixed, permeabilized, and immunostained with anti-eNOS and anti-PrS-SG antibodies. D, HCAEC were exposed to H/R for indicated time periods and lysed, and the cell lysates were analyzed for eNOS enzymatic activity as described under "Materials and Methods." *, p Ͻ 0.05 versus control. E, thin layer chromatography (TLC) of eNOS enzyme assay products from each group were spotted onto silica gel TLC plates and developed.

Thioredoxin Protects Coronary Endothelial Dysfunction
Accumulation of GSSG has been shown to cause defects in the endothelium-dependent vascular relaxation (16). To test the efficacy of Trx in the preservation of endothelial function in the presence of high levels of GSSG, we inhibited GR activity by BCNU (Fig. 7I) in myograph-mounted LCA, which significantly blunted the endothelium-dependent ACh-mediated relaxations (Fig. 7I). We next addressed whether elevated Trx levels would protect against the BCNU-induced impairment in ACh-mediated relaxation. Although BCNU resulted in a significant rightward shift in the concentration-response curves to ACh, maximal relaxation in response to ACh did not reach statistical significance for BCNU-treated LCA compared with FIGURE 4. Trx deficiency promotes eNOS S-glutathionylation and reduces its enzymatic activity. A, HCAEC were transfected with indicated concentrations of NT or Trx siRNA, and after 36 h the cell lysates were collected. The lysates were analyzed for levels of Trx and ␤-actin by Western blotting. B, cell lysates described in A were immunoprecipitated (IP) using anti-eNOS antibodies and analyzed by Western blotting using anti-PrS-SG and anti-eNOS antibodies. IB, immunoblot. C, NT or Trx siRNA (100 nM) transfected HCAEC were immunostained with anti-eNOS and anti-PrS-SG antibodies. D, HCAEC were transfected with NT/Trx siRNA (100 nM), exposed to H/R, and immunostained for eNOS and lysosome-associated membrane protein 1 (LAMP1). E, experiment described in D was performed in the presence of the proton pump inhibitor, lansoprazole (25 M). F, NT or Trx siRNA-transfected HCAEC lysates were analyzed for eNOS enzymatic activity; *, p Ͻ 0.05. G, specificity of production of [ 3 H]citrulline was analyzed in the presence or absence of L-NAA, an inhibitor of eNOS by thin layer chromatography. NOVEMBER 4, 2016 • VOLUME 291 • NUMBER 45

Thioredoxin Protects Coronary Endothelial Dysfunction
control LCAs from Trx-Tg mice (Fig. 7J). Addition of DTT fully restored ACh-mediated relaxations similar to control LCAs from Trx-Tg mice, suggesting that fully reduced Trx exerts maximal relaxing effect.
Trx Inhibits H/R-or I/R-induced eNOS Dysfunction-To test whether Trx protects eNOS from oxidative modification and thereby retains it in a coupled state, we examined the effect of H/R on O 2 . production by DHE staining. As shown in Fig. 8 . production compared with either coronary arteries from Trx-Tg mice in I/R or coronary arteries from NT mice treated with L-NAME for I/R. In additional experiments, we also determined NO release in coronary arteries from I/R-sub- B, Ad-LacZ-or Ad-Trx-infected HCAEC were exposed to hypoxia (8 h) followed by reoxygenation (1 h) and lysed, and eNOS glutathionylation was analyzed by immunoprecipitation (IP) and Western blotting. C, Ad-LacZ-or Ad-Trx-infected H/R-treated HCAEC were analyzed for colocalization of eNOS and PrS-SG by immunofluorescence staining. D, Ad-LacZ-or Ad-Trx-infected HCAEC were exposed to H/R, and eNOS glutathionylation was examined by PLA as published previously (28). E, proximity signals (green foci) were counted and plotted as a bar graph. *, p Ͻ 0.05 versus Ad-LacZ; **, p Ͻ 0.05 versus Ad-LacZ ϩ H/R.
jected NT and Trx-Tg mice using EPR spin trapping. As shown in Fig. 8, E and F, coronary arteries from Trx-Tg mice generated higher levels of NO compared with NT mice in I/R. Cardiomyocytes constitute the major volume of myocardial tissue. Moreover, eNOS is also expressed in cardiomyocytes albeit to a lesser extent than the endothelium. Therefore, we determined the relative contribution of eNOS-mediated O 2 . in response to H/R in cardiomyocytes. As shown in Fig. 8  A, Ad-LacZ-or Ad-Trx-infected HCAEC were exposed to H/R, and eNOS enzymatic activity was assayed. *, p Ͻ 0.05 versus Ad-LacZ, and **, p Ͻ 0.05 versus Ad-LacZϩH/R. B, eNOS enzyme assay products were analyzed by TLC in the presence or absence of L-NAA, a specific inhibitor of eNOS. C, Ad-LacZ-or Ad-Trx-infected HCAEC were exposed to H/R, washed, loaded with DAF-FM, and stimulated with ACh (10 M) in the presence or absence of 100 M L-NAME, and images were acquired via ϫ20 objective.
Trx Directly Interacts with eNOS and Prevents S-Glutathionylation in Vivo-Next, we determined how Trx deglutathionylates eNOS in vivo. To differentiate the mechanism of Trxmediated protection of eNOS from S-glutathionylation, the HCAEC lysates were incubated with GSSG and/or recombinant hTrx, and its impact on glutathionylation was evaluated. As shown in Fig. 9A, addition of GSSG increased eNOS glutathionylation in the lysate of HCAEC. However, glutathionylation was restored to the control level when both GSSG and hTrx were added to lysates. Furthermore, hTrx interacted with eNOS and could be immunoblotted from eNOS immunoprecipitated samples, demonstrating that eNOS and hTrx interact in vitro (Fig. 9B), and this interaction was increased in H/R (Fig.  9C). We also studied the interaction of hTrx with eNOS using a PLA. As demonstrated in Fig. 9D (right panel), H/R induced a significant number of PLA signals compared with normoxic HCAEC (Fig. 9E). To determine whether we could rescue the loss of eNOS activity due to mutation of cysteine residues that undergo glutathionylation, we mutated the previously reported (16,19) glutathionylation-susceptible cysteines (384, 691, and 910) to serine in a bovine eNOS construct (Fig. 9, F and G). The mutant eNOS enzymatic analysis revealed that C691S and C910S mutations partially rescued the H/R-mediated decrease in eNOS activity, and Trx overexpression completely restored

. Overexpression of Trx prevents H/R-or I/R-mediated eNOS dysfunction and superoxide anion production.
A, Ad-LacZ-or Ad-Trx-treated HCAEC were exposed to H/R, washed, and treated with O 2 . indicator DHE in the presence or absence of L-NAME (100 M), and then images were acquired by a Zeiss microscope (ϫ20 objective). B, DHE fluorescence signals (mean gray values) were plotted as bar graph. *, p Ͻ 0.05 versus Ad-LacZ, and **, p Ͻ 0.05 versus Ad-LacZ ϩ H/R. C, O 2 . production in coronary arteries from sham or I/R-treated NT or Trx-Tg mice was measured by EPR spectrometry using spin trap BMPO as described under "Materials and Methods." D, height of peaks indicates the magnitude of O 2 . generation and was calculated and expressed as arbitrary units. *, p Ͻ 0.05 versus NT, and **, p Ͻ 0.05 versus NT I/R. E, LCAs from NT or Trx-Tg mice were isolated after sham or I/R surgery. NO formation by isolated LCA was detected by EPR spectroscopy using NO spin trap Fe-MGD as described under "Materials and Methods" in the presence or absence of L-NAME. F, height of peaks indicates the magnitude of NO generation and was calculated and expressed as arbitrary units. *, p Ͻ 0.05 versus NT, and **, p Ͻ 0.05 versus NT I/R; G, O 2 . production was measured via EPR spectrometry of isolated cardiomyocytes from adult (12-16 weeks) NT, Trx-Tg, and ␣MHC Trx-Tg mice after exposing them to 30 min of hypoxia followed by 60 min of reoxygenation. H, graphical presentation of EPR signal intensity shown in G. *, p Ͻ 0.05 versus respective normoxia control, **, p Ͻ 0.05 versus Trx-Tg or ␣MHC-Trx-Tg with L-NAME. NOVEMBER 4, 2016 • VOLUME 291 • NUMBER 45 the activity in either of the eNOS mutants. We believe that Trx deglutathionylated either cysteine 691 or 910 to restore eNOS activity during H/R. No rescue of activity in C384S-bovine eNOS mutant indicates that this cysteine is not susceptible to glutathionylation during H/R. To determine whether eNOS is glutathionylated in vivo, we implemented tissue PLA using anti-eNOS and anti-PrS-SG antibodies as described under "Materials and Methods." As shown in Fig. 9H (right panels), there was no PLA signal in the arteries within a heart section of NT mice that under went sham surgery, although isolectin staining show the presence of endothelial cells (Fig. 9H). However, a significant number of endothelial cells underwent eNOS glutathionylation in vivo in the coronary artery branches in response to I/R (Fig. 9, H, top lower panels, and I). In contrast, the PLA signals of eNOS-SG and isolectin were significantly lower in the endothelium of coronary arteries from Trx-Tg mice that underwent I/R (Fig. 9, H, middle panels, and I), but the PLA signal of eNOS-SG and isolectin did not change in coronary arteries of ␣MHC-Trx-Tg mice that underwent I/R (Fig. 9, H, bottom panels, and I). Collectively, these studies show that Trx deglutathionylate eNOS in vivo and in vitro in response to I/R or H/R.

Discussion
We have shown that endothelial cells in coronary arteries undergo extensive S-glutathionylation of eNOS following I/R and consequently show impaired endothelial function. Using Trx-Tg and ␣MHC-Trx-Tg mice, we have shown that functional Trx in vascular endothelium is critically required for deglutathionylation of eNOS and maintenance of endothelial function following I/R. Furthermore, we demonstrated that I/R impairs endothelium-dependent relaxation of coronary arteries that is protected in Trx-Tg mice but not in ␣MHC-Trx-Tg mice. Thus, vascular Trx plays a critical role in the protection against I/R-mediated endothelial dysfunction. Because eNOS was glutathionylated following I/R and consequently its activity was decreased significantly in NT and ␣MHC-Trx-Tg mice but not in the coronary arteries of Trx-Tg mice, deglutathionylation by Trx constitutes a major protective mechanism in endothelial dysfunction in I/R by Trx. Further evidence is provided by the fact that Trx-Tg mice are protected from BCNU-induced impairment in vascular relaxation, demonstrating that Trx could independently deglutathionylate eNOS even when the GSSG/GSH ratio is quite high, as BCNU treatment increases GSSG due to inhibition of GR. Finally, we show that Trx directly interacts with eNOS to deglutathionylate it in I/R. This is the first report demonstrating that Trx is an efficient deglutathionylating protein even in the presence of normal levels of GSH and Grx1 in a cellular model and in vivo in an ischemia-reperfusion model of oxidative stress.
Trx has been shown to be up-regulated in oxidative stress conditions such as ischemia-reperfusion injury (30), hyperoxia (31), and also in hypoxia (27). However, the mechanisms associated with protection of myocardial ischemia-reperfusion injury due to high levels of Trx have remained unclear. We used both global Trx-overexpressing mice and mice that specifically overexpress Trx in the cardiomyocyte to determine the relative contribution of Trx in the protection against I/R. Although myocytes from ␣MHC-Trx-Tg and Trx-Tg showed almost equal expression of Trx, the ␣MHC-Trx-Tg mice underwent significant myocardial injury in I/R with a resultant increase in MI. The only major difference between these mice is that endothelial cells from Trx-Tg mouse hearts express significant levels of Trx. Because endothelial cells constitute 60% of total cell type of the heart (26), and endothelial cells have been shown to release survival factors in I/R that promote cardiomyocyte survival (32,33), we speculated a significant contribution of endothelial Trx for survival of myocytes in I/R. Myocyte eNOS constitutes about 25% or total eNOS of the heart (12). Therefore, endothelial eNOS is critically required for myocyte function in addition to myocyte eNOS. We may mention here that although non-myocyte cells such as fibroblasts and other cells might have higher levels of Trx in Trx-Tg compared with ␣MHC-Trx-Tg, we believe that higher Trx levels in the endothelial cells play a major role in the protection against MI due to deglutathionylation of eNOS, as fibroblasts or other inflammatory cells do not express eNOS. A dysfunctional eNOS due to glutathionylation is expected to cause endothelial dysfunction in I/R that impairs myocardial perfusion and resultant myocyte death. Our study shows that deglutathionylation of eNOS by high levels of vascular Trx is a critical endothelial mechanism by which myocytes survive during I/R with a consequent decrease in MI.
Cardiomyocytes isolated from Trx-Tg or ␣MHC-Trx-Tg mice had similar levels of O 2 . generation in H/R, which was lower than the NT mice but was significantly higher than myocytes in normoxia, and this O 2 . generation was not dependent on eNOS as L-NAME did not decrease their level. Thus, the source of O 2 . in cardiomyocytes could be other than eNOS, such as mitochondria or NADPH oxidase. Furthermore, the decrease in O 2 . in both myocytes was equal, yet the ␣MHC-Trx-Tg mice showed significant myocardial damage and MI. We believe that dysfunctional eNOS due to glutathionylation contributed significantly to the myocardial injury of ␣MHC-Trx-Tg mice, whereas Trx-Tg mice were protected due to deglutathionylation of eNOS resulting in a functional eNOS during I/R due to high levels of vascular Trx in addition to myocardial Trx. It is important to mention here that Trx is not a O 2 . scavenger but could scavenge hydroxyl radicals and singlet oxygen in a redox-independent manner (34). However, it rescues oxidatively inactivated proteins due to its disulfide reduc-  tase properties via NADPH and redox cycles using NADPH (35). Therefore, it is plausible that an impairment in the mitochondrial redox-sensitive enzymes such aconitase or NADH dehydrogenase (36) could occur due excessive O 2 . generation due to I/R. High levels of Trx in the myocytes could restore the activities of these enzymes and thus could bring down the levels of O 2 . in I/R in myocytes. We believe that the 50% decrease in the production of O 2 . in the myocytes from Trx-Tg or ␣MHC-Trx-Tg mice could account for the mitochondrial production of O 2 . . However, deglutathionylation of eNOS by Trx could provide much needed NO for the proper functioning of myocytes as endothelial NO is needed for a variety of myocyte function such as cardiomyocyte contractile function and survival in I/R (12,(37)(38)(39).
Our data show that significant deglutathionylation of eNOS occurs with high levels of Trx in the absence of Grx1. Because Trx was able to decrease eNOS glutathionylation in the presence of increasing intracellular GSSG (by BCNU treatment), it is likely that high levels of Trx deglutathionylate eNOS in the presence of GSSG. Supporting this notion, we have shown that Trx directly interacts with eNOS and deglutathionylates it in an in vitro system. These studies point to the fact that deglutathionylation by Trx is an independent additional mechanism but not a compensatory mechanism in the absence of Grx1. This was further supported by the fact that Grx1 depletion-mediated glutathionylation of eNOS was rescued by increased levels of Trx. Further in vivo evidence is provided by Grx1 knock-out mice, which were not susceptible to I/R injury of the heart, suggesting that Grx1 is not essential for deglutathionylation in I/R (40). Thus, our studies establish for the first time that Trx is a potent deglutathionylating protein, which could rescue eNOS deglutathionylation during I/R. Grx1 is the major deglutathionylating enzyme with selective preference for PrS-SG (19). The catalytic efficiency of Grx1 is much higher for PrS-SG compared with Trx (41). For example, the k cat /K m is 5000-fold higher for PrS-SG compared with Trx (41). The Trx and Grx1 systems function in a very different manner for transfer of electrons from NADPH to a modified protein thiol. Whereas Grx1 uses a two-enzyme system (GSSG 3 GR-NADPH 3 GSH 3 Grx1 3 Pr-SH), Trx uses a oneenzyme system (NADPH 3 TrxR1 3 Trx-SH 3 R-SH). The role of Grx1 system has been intensely studied as a major deglutathionylating enzyme in several PrS-SGs (25,41,42), but the role of Trx in deglutathionylation remains poorly understood. The notion that Trx is inefficient as a deglutathionylating protein has been put forth based on the fact that Grx1 is more potent (Ͼ5000 times) than Trx as a deglutathionylating agent (41). However, one should note that Grx1 uses enzyme catalysis for reduction, whereas Trx uses stoichiometric transfer of electrons to oxidized proteins. The oxidized Trx is regenerated by thioredoxin reductase (TrxR1). Thus, the rate-limiting enzyme for Trx is the TrxR1. Therefore, a high level of Trx is expected to provide a much better reducing power for conversion of protein disulfides to thiols. It is important to mention here that for most protein cysteines with typical redox potential of K mix ϳ1, the intracellular GSH/GSSG ratio would have to decline from a level of 100:1 to 1:1 to achieve a 50% conversion of protein-SH to PrS-SG (29). Such low ratios were rare even in artificial con-ditions (29). Thus, glutathionylation of eNOS in conditions of I/R could occur in a much higher GSSG/GSH ratio resulting in efficient deglutathionylation by Trx.
In conclusion, our data establish that Trx prevents eNOS S-glutathionylation during I/R, thereby preventing uncoupling of eNOS resulting in improved NO release and decreased oxidative load, consequently maintaining coronary artery perfusion and endothelial function following I/R (Fig. 9J).

Materials and Methods
Animals and Cells-Wild-type C57BL6 strain (WT) was purchased from Charles River Laboratory. Transgenic mice with overexpression of hTrx (Trx-Tg) were bred and maintained in the animal facility of Texas Tech University Health Sciences Center and have been described previously (28,44). ␣MHC-Trx-Tg mice are a generous gift from Dr. Junichi Sadoshima, University of Medicine and Dentistry of New Jersey, and have been described previously (45). Both males and females were used in this study. All mice strains used in this study are from a C57BL/6 background and are 12-16 weeks of age. All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the Texas Tech University Health Sciences Center and were consistent with the Guide for the Care and Use of Laboratory Animals published by the National Institute of Health. HCAEC were purchased from Clonetics and propagated in endothelial basal medium supplemented with additives (Bullet kit, Clonetics). Mouse coronary artery endothelial cells and ECG media were purchased from Cell Biologics, Chicago, IL.
Isolation of Adult Mouse Cardiomyocytes-Adult cardiomyocytes were isolated from 12-to 16-week-old mice as described by O'Connell et al. (47) and Louch (46) with minor modifications. In brief, mice were anesthetized with 2% isoflurane and 100% O 2 (1.0 liter/min) and intraperitoneally injected with 50 IU of heparin. Thoracic cavity was opened, and hearts were dissected at aortic arch, immersed in ice-cold isolation buffer containing (in mM) NaCl 120.4, KCl 14.7, KH 2 PO 4 0.6, Na 2 HPO 4 0.6, MgSO 4 ⅐7H 2 O 1.2, Na-HEPES 10, NaHCO 3 4.6, taurine 30, butanedione monoxime 10, glucose 5.5. The aorta was cannulated just above the coronary artery branches under a dissection microscope. Hearts were retrograde-perfused through the aorta using a Langendorff perfusion apparatus with isolation buffer (3 ml/min) for 5 min and then switched to digestion buffer (Liberase TH (Roche Applied Science), 0.3 mg/ml trypsin, and 100 M CaCl 2 in isolation buffer) and further perfused for 15 min (4 ml/min). Hearts were placed in a 6-cm Petri dish containing 3 ml of digestion buffer; the atria were removed; the ventricles were cut into 4 -5 pieces, and cell suspension was made by gentle dispersion with transfer pipettes. The cells were filtered through a 200-m strainer, and stop solution (0.5% FBS in isolation buffer) was added. The cell suspension was allowed to settle for 10 min; the supernatant containing damaged myocytes and non-myocytes was removed, and the pellet was dissolved in stop buffer. The CaCl 2 was added sequentially, washed, resuspended in cardiomyocyte culture medium (minimum essential medium with insulin, transferrin, selenium, glutamine and antibiotics), and plated in laminin-coated dishes.
Endothelial Cell Isolation-Endothelial cells from mouse hearts were isolated as described by Jin et al. (48). In brief, mouse were anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg), and after opening the thoracic cavity, hearts were perfused, dissected, and placed in Hanks' balanced salt solution. Hearts were minced into small pieces and digested in 0.2% collagenase II from Worthington containing Hanks' balanced salt solution for 45 min. Then the mixture was triturated by passing through the 16-guage cannula and filtered with a 70-m strainer. After centrifugation, the pellet was dissolved in DMEM, and endothelial cells were selected by incubating with anti-platelet endothelial cell adhesion molecule (BD Biosciences)-coated sheep anti-rat Dynabeads (Invitrogen) for 30 min, washed, and trypsinized to detach cells from Dynabeads. Finally, the cells were suspended in EBM-2 medium and plated in gelatin-coated plates.
Adenovirus Production-AdenoX system was obtained from Stratagene Corp. (La Jolla, CA), and LacZ or Trx cDNA was cloned into pAdenoX vector as described previously (49). Recombinant virus was allowed to infect HEK293 cells for generation of viral particles.
Cell Culture and H/R-HCAEC and mouse coronary artery endothelial cells in complete medium and isolated cardiomyocytes were flushed with a 95% N 2 , 5% CO 2 gas mixture while in a Billups-Rothenberg modular chamber to create an hypoxic environment. The oxygen level was kept below 1% by measuring with an oxygen electrode. Chambers were kept inside the incubator at 37°C for indicated periods of time and followed by 1 h of reoxygenation in normoxic condition.
Myocardial Ischemia and Reperfusion-Trx-Tg, ␣MHC-Trx-Tg, or NT littermates were anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg). After an equilibration period of 10 min, the left thoracotomy was performed in the fourth intercostal space, and the pericardium was opened to expose the heart. An 8-0 silk suture was passed around the LAD at a point two-thirds of the way between its origin near the pulmonary conus and the cardiac apex. Coronary artery occlusion was achieved by ligating the left descending coronary artery using a slipknot. Following ischemia, the slipknot was released, and the myocardium was reperfused. Sham mice underwent the same procedure without the slipknot tied.
Determination of Infarct Size-Myocardial infarct size was determined as described previously (43). Briefly, after reperfusion, animals were sacrificed, and the aorta was cannulated and perfused with saline to remove blood. 0.25 ml of 1.5% Evans blue was perfused after religating the coronary artery to demarcate remote myocardium (blue) and AAR. 1.0-mm heart sections were taken and stained with 1.0% triphenyltetrazolium chloride (TTC) for 15 min at 37°C. After TTC staining, paraformaldehyde-fixed heart sections were imaged with a Nikon camera. TTC stained and unstained areas (infarct) at AAR were quantified.
eNOS Activity Assay-eNOS activity was assayed using Cayman's NOS activity assay kit (catalog no. 781001) following the manufacturer's instruction.
Detection of Nitric Oxide and Superoxide Anion-HCAEC were cultured on glass coverslips, which were placed in the wells of a four-well plate. For detection of NO after H/R, cells were loaded with DAF-FM (10 M) and then treated with ACh (10 M) in the presence or absence of eNOS inhibitor L-NAME. For O 2 . detection, H/R-exposed cells were treated with 5.0 M DHE in the presence or absence of L-NAME, fixed, and counter-stained with Hoechst 33342. Immunoprecipitation and Immunoblotting-HCAEC were lysed in 25 mM Tris-HCl (pH 7.4) buffer containing 1% Nonidet P-40, 150 mM NaCl, and 20 mM NEM and protease inhibitors. The cell lysate was then incubated with the bead-conjugated eNOS antibody overnight at 4°C under constant rotation. eNOS was then eluted from the bead-antibody-eNOS complex using the loading buffer without reducing agent, and the supernatant was separated by SDS-PAGE.
Isolation and Mounting of Left Coronary Artery Segments-The heart was removed and placed in cold Krebs-Ringer buffer (KRB), and the left coronary artery below the ligation point was carefully dissected and mounted in a wire myograph (model 620 M; Danish Myotechnology, Aarhus, Denmark) for the recording of isometric force development.
Relaxation Responses-During contraction with a single concentration of 5-HT (30 mol/liter), relaxing responses to ACh (0.001-10 mol/liter) were recorded with or without inhibitors.
Immunofluoresence and PLA-PLA and immunofluorescence staining were performed as described in our recently published report (28).
TUNEL Assay-TUNEL assay was performed by In Situ Cell Death Detection kit from Roche Applied Science (catalog no. 11684795910) following the supplier's protocol.
In Situ PLA-Deparaffinized heart sections were permeabilized with 0.1% Triton X-100 for 10 min at room temperature, blocked with 5% donkey serum and 3% BSA in PBS for 1 h at room temperature, and incubated with primary antibodies in 50% Da Vinci Green antibody diluent (Abcam, Cambridge, MA). PLA was performed following the supplier's instructions using Duolink anti-rabbit PLUS and anti-mouse MINUS PLA probes and Duolink green detection reagent (Duolink, Sigma).
Electron Paramagnetic Spectrometry (EPR) for Detection of O 2 . and NO-Superoxide production by coronary artery or isolated cardiomyocytes was measured by EPR spectrometry using spin trap BMPO. Isolated coronary arteries (1.5-3 mm) were incubated in 50 l of Krebs-HEPES buffer containing 25 mM BMPO. Superoxide generated by coronary arterial sections was detected as BMPO-OOH adduct using Bruker EMX X-band spectrometer at room temperature. NO formation by isolated LCA was detected by EPR spectroscopy using NO spin trap Fe 2ϩ -(N-methyl-D-glucamine dithiocarbamate) 2 (Fe-MGD). LCA was longitudinally opened and washed with minimum Eagle's medium. LCA was incubated in 50 l of minimum Eagle's medium containing 10 M ACh, 0.1 mM sodium ascorbate, and 2 mM Fe-MGD. NO generated by LCA was detected as paramagnetic NO-Fe 2ϩ -MGD 2 adduct using Bruker EMX X-band spectroscope at room temperature. Statistical Analysis-All mouse numbers are mentioned in the figure legends. A minimum of n ϭ 3 was used in this study. The experiments were performed in triplicate and repeated for minimum of 2 times. All cell culture studies were performed in triplicate and repeated at least twice. Data were statistically analyzed by analysis of variance for multiple means with Tukey's post hoc analysis. Student's t test was used to compare two means. Prism software (Version 6.0) was used for all statistical analyses.