Introduction

There is an enormous personal and global socioeconomic burden of cardiovascular disease (CVD).^1,2 Oxidative stress is recognized to play a key role in mediating the pathophysiological effects of diabetes, hypertension, and aging on the arterial wall, all key risks factors for CVD. Yet, clinical trials of antioxidants have been disappointing.^3,4 This is partially attributable to the compartmentalization of redox signaling, and redox-sensitive proteins within the cell,⁵ which dietary antioxidants fail to effectively penetrate. Caveolae are important examples of such compartmentalization that may be impossible to influence with dietary antioxidants. These specialized invaginated plasma membrane domains are ≈50 to 100 nm in diameter and defined by positive caveolin expression. They can act as functional signalosomes where membrane receptors, signal transduction molecules, membrane channels, and transporters are housed in a concentrated space.^6,7 A large number of the ≈250 identified caveolar proteins are redox sensitive,⁸ with many of these redox modifications shown to be functionally important for the molecule and downstream signaling in the cell.^9–11

eNOS (endothelial nitric oxide [NO] synthase), a functional homodimer protein, is a well-known resident in the caveolae. In its healthy coupled state, eNOS catalyzes the generation of NO. This vasoprotective molecule regulates vascular tone and attenuates both platelet aggregation and neutrophil-endothelium interaction.¹² However, in diseases such as diabetes and hypertension, the elevated oxidative stress exacerbates eNOS dysfunction, resulting in a shift from NO production to superoxide generation.¹³ This eNOS uncoupling can occur via a number of mechanisms, including inhibition of phosphorylation (at serine 1177) and oxidization of the critical cofactor, tetrahydrobiopterin.¹⁴ More recently, the discovery that this uncoupling can be mediated by S-glutathionylation of cysteine residues in the eNOS reductase domain¹⁵ has led to a paradigm shift in our understanding of eNOS regulation. In this oxidative reaction, a reversible disulfide bond is formed between glutathione and the reactive cysteine residues of the enzyme. Protecting eNOS from S-glutathionylation and uncoupling during pathophysiological insults is predicted to halt the amplification process of reactive oxygen species (ROS)-induced eNOS uncoupling and ROS production in this critical signaling microdomain.

The FXYD protein family are type I membrane proteins and caveolae residents,⁸ known to colocalize with the Na⁺-K+ ATPase and play a role in kinase-dependent pump regulation.¹⁶ FXYD1 (FXYD domain containing ion transport regulator 1) (also known as phospholemman), expressed in both the heart and vasculature, is the only member of the family to have serine residues susceptible to phosphorylation. In contrast, 2 cytoplasmic cysteine residues are highly conserved. We have demonstrated a new role for FXYD proteins, dependent on one of these cysteine residues, in protecting the β1 subunit of Na⁺-K⁺ ATPase from oxidative inhibition.¹⁷ Given the protective effect of FXYD1 against oxidative S-glutathionylation and inhibition of the Na⁺-K⁺ ATPase, as well as the known colocalization of the Na⁺-K⁺ ATPase/FXYD complex¹⁷ with eNOS in caveolae, we postulated that FXYD1 may have a broader protective role beyond that of the Na⁺-K⁺ ATPase in the caveolae. Here, we identify a novel role for FXYD1 as an endogenous protector of eNOS from oxidative uncoupling both in vitro using human-derived cells and in vivo, in mouse models of redox stress. This has implications for all vascular diseases mediated by oxidative stress.

Methods

The authors declare that all supporting data are available within the article (and in the Data Supplement).

In Vivo Animal Studies

All animal studies performed were approved by the Northern Sydney Local Health District Animal Ethics Committee (approval numbers RESP/14/278, RESP/17/96, RESP/17/55) and conform to the National Health and Medical Research Council of Australia’s Code of Practice for the Care and Use of Animals for Scientific Purposes, and all procedures conform to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes. FXYD1 wildtype and knockout mice were derived as previously described.^17,18 FXYD1 hemizygous mice were interbred within the Kearns Facility, Kolling Institute of Medical Research, to produce FXYD1 wildtype and knockout littermates. For all studies, mice were randomized to groups, and blinded analysis of end point experiments was performed.

Mouse Models

For baseline measures, male mice were used at 14 to 20 weeks of age. To investigate the role of FXYD1 under increased oxidative stress, several models were used. (1) Aged male mice were kept until 50 to 52 weeks of age; (2) a sub-pressor dose of Ang II (angiotensin II) was used to activate redox signaling in male mice aged 14 to 20 weeks; and (3) hyperglycemia was induced in male mice aged 10 to 12 weeks of age, with a model of insulin resistance that we have previously described.¹⁹ For the latter 2, mice were anesthetized using isoflurane (1.5%, in 100% O₂ at 0.4 L/min) and implanted with subcutaneous osmotic minipumps (Alzet; model 1004) containing Ang II (0.72 mg/kg per day, 4 weeks) or insulin-receptor antagonist, S961 (0.3 mg/kg per day in dimethyl sulfoxide, 4 weeks, KareBay Biochem Inc) to induce hyperglycemia.¹⁹ Of the total mice enrolled in the study, 2 mice (wildtype) died after telemetry pump implantation.

Blood Pressure Measurements by Radiotelemetry

Mice were implanted with radiotelemetric transmitters (TA11PA-C10, Data Sciences International) in the aortic arch via the left carotid artery while anesthetized (isoflurane, 1.5%–2%, in 100% O₂ at 0.4 L/min). Battery packs were tunneled under the skin to sit in the abdominal region. Mice were allowed to recover from surgery for 10 days before blood pressure (BP) measurement, as described previously.²⁰ BP was recorded continuously for 24 hours and average systolic, diastolic and mean arterial pressures, and heart rate and activity were determined. At the end of the recording period, all mice were euthanized by cervical dislocation while under isoflurane anesthesia (3%–4% in oxygen), following a blood sample and saline perfusion via the vena cava.

Acute BP Measurements

Mice were maintained under anesthesia (isoflurane, 1.5%, in 100% O₂ at 0.4 L/min) throughout the experiments, and body temperature was kept constant at 37 °C. Catheters filled with heparinized saline (20 U/mL, polyvinyl tubing, 0.61 mm outside diameter) were placed in the left carotid arteries for arterial BP measurements recorded with LabChart 6 using a bridge amplifier connected to a Powerlab system (AD Instruments, Australia).²¹ Similar catheters containing saline only were placed in jugular veins for intravenous injections of vasoactive substances and changes in pressor responses were recorded. Drugs used were bradykinin acetate (0.3–10 µg/kg), sodium nitroprusside (0.1–10 µg/kg), and Ang II (0.01–10 µg/kg) all sourced from Sigma Aldrich, Australia and were administered into the jugular vein in ≈50 µL bolus doses.²² Mice were euthanized as above at the end of the recording period, and tissue was collected and snap-frozen in liquid nitrogen and stored at −80 °C for analysis.

In Vitro Cell Culture

Primary human umbilical vein endothelial cells (HUVECs) or human coronary artery smooth muscle cells (commercially available, Lonza) were cultured in EGM+ culture media (Lonza, Australia) under standard tissue culture conditions, in 6 or 12-well plates (on glass coverslips as needed) and underwent the following experiments (detailed in full in the Data Supplement): (1) overexpression of FXYD1 by transfection of FXYD1 plasmid; (2) FXYD1 knockdown by siRNA transfection; (3) treatment with acetylcholine (1 mmol/L); (4) Ang II (500 nmol/L); (5) NO reactive fluorescent dye, 4-amino-5-methylamino-2′,7′-difluorofluoroscein diacetate (2.5 µmol/L); or (6) dihydroethidium (2.5 µmol/L) for O₂⁻; or (7) fixed for proximity ligation assay (Duolink). Vascularized cardiac spheroids were generated from isolated hearts of neonatal FXYD1 wildtype and knockout mice (bred from homozygous parents) as described previously,²³ using Perfecta 3D 96-well hanging drop plates (3D Biomatrix, Ann Arbor, MI).

Delivery of FXYD1 Using Exosomes

To effectively deliver the protective FXYD1 protein into the vasculature without being cleared away or degraded, we packaged FXYD1 protein into exosomes by over-expressing myristoylation-palmitoylation tagged FXYD1 in HEK293T cells,²⁴ outlined in detail in the Data Supplement.

Tissue and Cell Analysis

Immunoblotting

Protein was extracted from cultured HUVECs and from mesenteric and aortic vessels isolated from mice. Ten to fifty micrograms of protein lysate was denatured and run under reducing conditions on SDS-PAGE as described previously.²⁵ Membranes were incubated in primary antibodies directed at determining expression of proteins as described in each figure. For co-immunoprecipitation of eNOS and FXYD1 protein, lysate was extracted from rabbit hearts and incubated with protein G dynabeads coated with anti-eNOS (1 mg) overnight at 4 °C.²⁵ The eNOS bound fraction was extracted using a magnet, and immunoblotting for FXYD1 and eNOS was performed as above.

Free Cysteine Assay

Cleared cell lysate was labeled with 1 µmol/L freshly made IRDye 800CW Maleimide (Licor, the United States in lysis buffer for 2 hours at room temperature in dark). Free labeling reagent was then removed by passing the lysate through the Zeb Spin Desalting Columns, 7K MWCO (Thermofisher) 3×. Lysate was then incubated with 2 μg anti-eNOS antibody overnight at 4 °C with rotation. Twenty-five microliters Protein G Dynabeads (Thermofisher) was used to immune-precipitate labeled eNOS which was eluted under reducing condition and resolved by Western blotting. The resulting membrane was probed with anti-eNOS and secondary antibody conjugated to IRDye 680RD.

Detection of Intracellular NO and Superoxide Generation

The intracellular NO level was measured 4-amino-5-methylamino-2′,7′-difluorofluoroscein diacetate, 2.5 µmol/L and intracellular levels of O₂⁻ were detected using dihydroethidium; 2.5 µmol/L according to manufacturer’s instructions (Thermofisher Scientific, Australia). Samples were prepared for confocal analysis (Leica SP5) as described in the Data Supplement.^23,25 For mouse vascular tissue, superoxide anion production was determined using lucigenin-enhanced chemiluminescence.²⁶

Statistical Analysis

Data are expressed as mean±SEM. Student t test was used for comparison between 2 groups. For multiple comparisons, 1- or 2-way ANOVA was used with Bonferroni post hoc analysis for multiple comparisons. A P value <0.05 was considered statistically significant. For all animal studies, power calculations were performed before the study to determine the appropriate group size.

Results

FXYD1 Physically Associates With eNOS

Although FXYD1 is known to be a caveolar protein, its physical association with eNOS has not been investigated. Given the known caveolar localization of both eNOS and FXYD1, and the functional effect of FXYD1 to protect against redox modification of the vicinyl β₁ subunit of the Na⁺-K⁺ ATPase in the caveolae, we examined specifically for their physical interaction. As shown in Figure 1A, FXYD1 co-immunoprecipitated with eNOS in freshly isolated rabbit hearts. A close physical interaction of the 2 proteins was supported by the proximity ligation assay results, with the Duolink staining of eNOS and FXYD1 in HUVECs showing a close proximity of <40 nm of the 2 proteins (Figure 1B).

• Download figure
• Download PowerPoint

FXYD1 Regulates eNOS Activity and Functional Coupling

Given FXYD1/eNOS were closely associated physically, we examined for any functional effect of FXYD1 on eNOS activity. NO bioavailability was initially assessed using the NO-sensitive fluorescent dye 4-amino-5-methylamino-2′,7′-difluorofluoroscein diacetate under basal conditions, and after stimulation with acetylcholine in HUVECs. FXYD1 knockdown resulted in lower NO bioavailability versus nonspecific control siRNA under basal conditions and when stimulated with acetylcholine (Figure 1C). S-glutathionylation of eNOS occurs at cysteine residues 689 and 908, in the reductase domain of the enzyme, and has been shown to mediate uncoupling.¹⁵ As an indirect measure of redox modification, we measured total free cysteines in HEK293 cells that were transiently transfected with either empty vector or recombinant FXYD1 protein. Conditions of high oxidative stress were simulated by exposing cells to hydrogen peroxide (H₂O₂). The co-immunoprecipitation of free cysteines and eNOS were detected in a series of individual experiments, and each immunoblot is shown in Figure 1D, with quantification performed by calculating the fold change of FXYD1 relative to the empty vector control in each experiment, after normalization to account for the considerable differences in intensity of bands detected on each occasion. Overexpression of FXYD1 increased the proportion of eNOS free cysteines compared with empty vector (Figure 1D) suggesting that FXYD1 protects against oxidative modification of these residues.

We then used tissue from mice deficient in FXYD1, with global knockout of FXYD1. To further validate our findings using monolayer cultures of HUVECs, we measured intracellular NO synthesis in 3-dimensional in vitro vascularized cardiac spheroids containing the diverse cell types found in vivo (endothelial, myocytes, and fibroblasts²³). These were generated from cells isolated from hearts of wildtype and FXYD1-deficient mice. We found a similar pattern in 3-dimensional cultures, where acetylcholine-stimulated NO production was lower in FXYD1-deficient mice compared with wildtype mice (Figure 2A). We next investigated eNOS enzyme expression in the vasculature of mice. Despite similar expression levels of the enzyme, eNOS phosphorylation at serine 1177, involved in activation of the enzyme, was lower in FXYD1 knockout mice than their wildtype controls (Figure 2B). This may contribute to lower NO bioavailability.

• Download figure
• Download PowerPoint

FXYD1 Protects Against Superoxide Generation and Nitrosative Stress

Given the emerging evidence that FXYD1 had a protective role in eNOS coupling suggested by enhanced NO bioavailability and proportion of eNOS free cysteines, we examined for any associated change in oxidative and nitrosative stress. In mesenteric tissue of mice, FXYD1 knockout was associated with an increase in nitrotyrosine levels (Figure 2C). We next investigated superoxide production, first in a real-time cell culture model, and then in freshly extracted tissue lysate from mice. HUVECs were incubated with dihydroethidium for 10 minutes before Ang II was added and incubated for an additional 50 minutes. Under basal conditions, FXYD1 knockdown had no significant effect on superoxide production (Figure 3A). However, under conditions of Ang II exposure, FXYD1 knockdown substantially augmented superoxide production by 67% compared with nonspecific siRNA (Figure 3A). We then measured superoxide generation in tissue from FXYD1 wildtype and knockout mice using a NADPH-dependent lucigenin assay. In the aorta from FXYD1 knockout mice, we saw higher generation of superoxide (Figure 3B). Yet, superoxide production in mesenteric resistance vessels was not different between the mice under baseline conditions (Figure 3C). To examine for additional enzymatic contributors to the superoxide levels, we studied NADPH oxidase isoform expression. Neither Nox 1 (Figure S1 in the Data Supplement) or Nox 2 (Figure 3D) expression were altered in association with FXYD1 knockout, and Nox4 isoform of NADPH oxidase was decreased ≈50% in the FXYD1 knockout compared with wildtype (Figure 3E).

• Download figure
• Download PowerPoint

FXYD1 Plays a Key Protective Role in Vascular Function In Vivo

Given the functional importance of protecting against ROS-induced eNOS uncoupling in cardiovascular health, we investigated the effect of FXYD1 absence on blood pressure in vivo. BP was monitored using radiotelemetry in freely moving, undisturbed mice, and normal circadian rhythms were observed. There were no differences in systolic or diastolic BP under basal conditions in FXYD1 knockout mice compared with wildtype controls (Figure 4A, Table S1). Heart rate was also similar in wildtype and knockout mice (Table S1). We next measured mean arterial BP in cannulated, anesthetized mice, and determined responses to the endothelial-dependent vasodilator bradykinin and the NO donor, sodium nitroprusside. Both of these drugs evoked substantial, concentration-dependent reductions in mean arterial BP in wildtype mice. FXYD1 knockout mice had substantially impaired BP responses to bradykinin (Figure 4B and 4C) but no effect on BP reductions using sodium nitroprusside (Figure 4D). These data suggest that FXYD1 plays an important role in endothelial-dependent relaxation, but signaling downstream of NO was normal.

• Download figure
• Download PowerPoint

Potential Therapeutic Opportunity—Delivery by Exosomes

To explore the therapeutic potential related to our findings, we developed a novel delivery method of FXYD1. FXYD1 was packaged into exosomes by over-expressing myristoylation-palmitoylation (MyrPalm) tagged FXYD1 in HEK293T cells and purifying exosomes from the HEK293T culture medium. The precipitant from harvested cell effluent was resuspended in PBS. FXYD1-exosomes, or control exosomes were administered intravenously to anesthetized FXYD1 knockout mice, following baseline bradykinin response measurements. Ninety minutes after FXYD1-exosome delivery, bradykinin responses were repeated. Delivery of FXYD1-exosomes to the knockout mice resulted in detectable FXYD1 expression in mesenteric tissue 90 minutes after delivery, although this was substantially less than the expression of that from tissues in wildtype animals (Figure 5A and Figure S2). Delivery of control exosomes (empty vector) did not significantly alter BP, or bradykinin-induced response (Figure 5B). However, despite the low expression, FXYD1-exosome delivery was associated with significantly augmented bradykinin-induced vasodilatation compared with knockout control (Figure 5B). The result was an almost complete rescue of the vascular dysfunction by FXYD1-exosomes in FXYD1 knockout mice. Our evidence of increased phosphorylated eNOS in the mesenteric vessels from these FXYD1 exosome-treated mice compared with empty vector controls, (Figure 5C) leads us to speculate that there may be increased eNOS activity after FXYD1 delivery.

• Download figure
• Download PowerPoint

FXYD1 Protects Against Redox-Mediated Vascular Dysfunction

Despite FXYD1 knockout resulting in reduced bradykinin-induced decreases in BP, reflecting endothelial dysfunction, we did not detect any major changes in superoxide generation in the resistance vasculature under baseline conditions. We next investigated the effects of FXYD1 knockout on vascular function under conditions of elevated oxidative stress. Ang II, well known to activate NADPH oxidase and shown in Figure 3 to increase superoxide generation in HUVECS, was infused in both an acute and chronic protocol in mice. FXYD1 knockout was associated with an exaggerated pressor response to acute bolus dosing of Ang II in both young and middle-aged mice compared with wildtype (Figure 6A). Baseline BPs from FXYD1 knockout mice measured by direct cannulation were not different between wildtype and knockout, similar to what was found using invasive radiotelemetry (Figure 4A and 6B). However, FXYD1 knockout mice had substantially increased BP in response to sub-pressor doses of Ang II infused over 4 weeks (Figure 6B). This does not appear to be due to increased Ang II receptor density, as expression of AT1R was not different between wildtype and knockout aged mice (Figure 6D). Stimulation of an endothelium-dependent vasodilator response using bradykinin was abolished in mice treated with Ang II, but this was not worsened in FXYD1 knockout mice (Figure 6C). We also determined whether aging or diabetes affected BP responses but found that anesthetized MABP responses were not different in FXYD1 knockout mice versus wildtype in either middle-aged or diabetic conditions, and in both cases, BP was comparable to young, normoglycemic control mice (Table S2). Despite the lack of effect of FXYD1 knockout on baseline BP under these conditions, acute responses to the endothelium-dependent vasodilator bradykinin were impaired in knockout diabetic (Figure 6E) and aged (Figure S3) mice compared with wildtype. Like the nondiabetic control mice in Figure 3C, mesenteric vessels from FXYD1 knockout normoglycemic mice showed no differences in mesenteric superoxide generation, but superoxide levels were elevated 5-fold by FXYD1 knockout in mice under diabetic conditions (Figure 6F).

• Download figure
• Download PowerPoint

Discussion

Despite our longstanding knowledge regarding the role of dysregulated oxidative signaling in many forms of CVD, we have made little progress in therapeutically targeting this.²⁷ Furthermore, we have limited understanding of the cell’s own mechanism of redox protection in the important signaling domain of the caveolae. By harnessing naturally occuring protein interactions to inhibit oxidative dysregulation, or by fixing disturbed redox signaling within these sub-cellular redox hotspots, we may be able to overcome some of the challenges with translation of antioxidant therapy, that we have faced previously.¹⁰ Our findings from this study demonstrate that the small membrane protein FXYD1 has a novel functional partnership with eNOS, protecting it from uncoupling in the endothelium, and re-establishing NO/ROS balance in a number of different disease-relevant models. Harnessing this knowledge may have important therapeutic implications for redox-dependent vascular disease, including hypertension and atherosclerosis.

FXYD1 is associated with a reversal of S-glutathionylation of β₁ subunit the Na⁺-K⁺-ATPase and, therefore, protection against oxidative inhibition.¹⁷ Being colocated with the Na⁺-K⁺-ATPase in the caveolae, we speculated that FXYD1 might possess a broader role and protect eNOS in a similar manner. This is particularly important given the relative paucity of redox protective mechanisms identified in the caveolae, despite housing 2of the cell’s greatest producers of ROS.¹¹ Here, we have shown that eNOS and FXYD1 colocalize with each other in caveolae, using the proximity ligation assay, which supports a physical distance of <40 nmol/L. However, we need to take the size of the caveolae into consideration (≈60–80 nm as measured by electron microscopy),²⁸ and recognize that the strongly positive results of the proximity ligation assay may simply reflect their cohabitation in the caveolar microdomain. Complementary evidence for direct physical interaction of the proteins is provided by the co-immunoprecipitation experiments (Figure 1A), although this again may be indirect. Functional evidence was key to showing the relationship between eNOS and FXYD1.

The functional effect of FXYD1 to protect eNOS from uncoupling is supported by molecular data, superoxide measures in vitro and in vivo vascular function studies. S-glutathionylation of the key cysteine residues (689 and 908) in the reductase domain of the enzyme has been implicated as a major mechanism of eNOS uncoupling, associated biochemically with a decrease in NO production by the enzyme of 70% and an increase of superoxide production by 5-fold.¹⁵ We found an increase in eNOS free cysteines in cells over-expressing FXYD1 under simulated oxidative stress. Thus, it appears that a protection against S-glutathionylation with an associated increase in free cysteines could be contributing to the vascular protective effect observed in association with FXYD1 expression. Analysis of S-glutathionylation using immunoprecipitation in hearts from wildtype and knockout mice was attempted to further support this hypothesis; however, difficulties in distinguishing between glutathionylated eNOS and immunoglobulin G bands in the tissue lysate made the findings inconclusive.

The functional effect of this lack of NO bioavailability was clear, with a striking impairment of the vasodilation response when the endothelium-stimulating vasodilator, bradykinin was used. This was apparent even under nonstressed conditions but did not translate to a hypertensive response. However, once a physiologically relevant stress was induced (Ang II at a sub-pressor dose), FXYD1 knockout was associated with augmented pressure response in vivo. Our interpretation of this finding is that there may have been compensatory adaptation with either upregulation of other vasodilating pathways or downregulation of constrictor activity. The latter is less likely given that we demonstrated increased vasoconstrictor responses in FXYD1 knockout mice to both Ang II (Figure 6A) and phenylephrine (data not shown).

We went on to examine the consequence of FXYD1 knockout in a model of diabetic vascular oxidative stress,¹⁹ by infusing an insulin-receptor antagonist to simulate insulin resistance. While the diabetic mice remained normotensive, the impact of FXYD1 knockout led to impaired pressor responses in diabetics and this was accompanied by a 5-fold increase in superoxide generation in the resistance vessels. This would likely have impacted on BP if the model was taken out beyond the short period of 4 weeks. As far as we are aware, this is the first study to show a relationship between FXYD1 and vascular complications of diabetes.

A key observation from hemodynamic monitoring was that both young and middle-aged FXYD1 knockout mice had a heightened BP response to bolus doses of Ang II. We postulate that this is due to redox-NO imbalance that is exacerbated by Ang II-dependent NADPH activation. We showed that even under nonstressed conditions that NADPH oxidase-dependent superoxide generation was higher in the aortic vasculature of FXYD1 knockout mice. However, it may also be receptor-dependent, but we think that this is unlikely as AT1R expression was not altered. The effect may also be a result of modified calcium-dependent contractility via second messengers (ie, IP3-PKC mediated). Another plausible explanation is that reduced NO activity leads to increased sensitivity of smooth muscle contractile proteins to calcium, as we know that NO can decrease the calcium sensitivity and shift the myosin light chain kinase/phosphatase balance.²⁹ There are also hints that antioxidant activity may be altered in the presence of Ang II, and this could contribute to elevated BP seen in FXYD1 knockout mice in the setting of neurohormonal dysregulation that is a common feature driving human CVD.

FXYD1 expression was also associated with eNOS phosphorylation at serine 1177, known to be involved in enzyme activation, particularly in response to agonists such as VEGF.³⁰ The interdependence of post-translational modifications and their impact on the regulation of function is fascinating to consider, as has been reported for PKG1α.³¹ It is feasible that S-glutathionylation of cysteines 689 and 908 may have steric impact on the enzyme, decreasing phosphorylation at serine 1177, or vice versa- that phosphorylation at serine 1177 is protective against oxidative modification of the cysteines in the reductase domain. S-nitrosylation, as we observed in the setting of FXYD1 silencing, may also play into the equation. It is known to facilitate eNOS phosphorylation at serine 1177 in an autoregulatory fashion, with denitrosylation leading to decreased serine 1177-mediated eNOS activity by a mechanism that may involve depalmitoylation.³² The interaction of these post-translational modifications is of broad interest beyond this current study.

NO can induce serine-mediated phosphorylation of FXYD1 via a PKC-epsilon dependent mechanism, which has a regulatory role in cardiac Na⁺-K⁺ ATPase activity.²⁷ Here, we have shown a reciprocal relationship where FXYD1 regulates eNOS activity. The mechanism by which FXYD1 acts to protect eNOS from uncoupling and remains unclear. Although the co-immunoprecipitation and proximity ligation assays support a direct physical interaction with eNOS as a possible mechanism for FXYD1, it has not been shown to have direct deglutathionylating ability. There are 2 cysteines separated by a single amino acid in a C1-R-C2 motif that are highly conserved across the entire FXYD family, as well as across species.¹⁷ The resemblance to the CXXC motif seen in the redoxin family³³ is suggestive of a similar mechanism. However, in the case of the Na⁺-K⁺ ATPase’s β1 subunit, protection by FXYD1 against ROS-induced S-glutathionylation and inhibition of the Na⁺-K⁺ ATPase was shown to be dependent on just one of these—Cys40, that was flanked by basic amino acids, a classic feature of reactive cysteines.³⁴ Despite the clear role of Cys40 in the protective effect of FXYD1 on the β1 subunit, it is still likely that this occurs indirectly. In the crystal structure of the Na⁺-K⁺ ATPase,³⁵ it is challenging to consider how the reactive cysteine of the β₁ subunit that mediates redox inhibition of the Na⁺-K⁺ ATPase (Cys46)³⁶ could directly interact with the FXYD subunit of the complex. One possibility, given the known calcium sensitivity of eNOS, is that FXYD1 protection against redox inhibition of the Na⁺-K⁺ ATPase improves eNOS activity by altering sub-cellular Ca²⁺ concentrations. However, such a mechanism would be predicted to have the opposite effect than what is observed. A third possibility is that the FXYD1 effects that protect these 2 important caveolar proteins is mediated via an, as yet unknown, protein-protein interaction, for example somehow increasing the caveolar localization and activity of members of the redoxin family such as peroxiredoxin 6, or glutaredoxin 1 which are known to possess biochemical deglutathionylation capabilities.

While we have demonstrated a clear physical and functional relationship between eNOS and FXYD1 in human endothelial cells in vitro, our in vivo functional measures are limited by the lack of direct evidence of endothelium-dependency of the protective effect of FXYD1, given the systemic nature of the knockout model. While the differential protective effect of FXYD1 on bradykinin versus sodium nitroprusside is supportive of a predominant endothelial-dependent mechanism, it is feasible that FXYD1 may modulate vascular function, in part, via effects on the Na⁺-K⁺ ATPase of the vascular smooth muscle cell. We have previously shown that FXYD1 is expressed in human vascular smooth muscle cells, and that its protection against redox inhibition of the Na⁺-K⁺ ATPase has beneficial effects on vascular tone in the setting of Ang II.³⁷ However, this is unlikely to explain the impaired vasodilation response to bradykinin reflected in acute decrease in BP, a known endothelial-dependent response, that was not seen in response to a direct NO donor (Figure 4).

The caveolae is an intriguing signaling microdomain, which although facilitatory for receptor-coupled, physiological redox signaling in health, has the potential to be switched to a bonfire of dysregulated ROS,³⁸ with severe consequences for the function of redox-sensitive proteins such as receptors, channels, pumps, and signal-transduction molecules located in the immediate vicinity. The lack of clinical success of dietary antioxidants is not surprising in this context. However, our findings that FXYD1 is protective against eNOS uncoupling has therapeutic implications. To test the therapeutic potential of FXYD1 and its derivatives, we have packaged FXYD1 protein into exosomes which were able to be delivered through injection in a proof-of-principle study. The delivered FXYD1 protein rescued the bradykinin response in FXYD1 knockout mice in acute experiments. It is an ongoing challenge to optimize the delivery of FXYD1 acute and chronic therapeutic intent.

While this article has focussed on the vascular implications of FXYD1/eNOS interactions, the functional significance of FXYD1/eNOS in cardiac myocyte physiology where both proteins are independently known to be highly expressed and key to cardiac function and pathophysiology,^39,40 has not been explored. This is an important program of work beyond the scope of the current study.

In conclusion, we have identified a novel functional partnership of FXYD1 and eNOS. We concluded that FXYD1 protects the vasculature from redox-mediated dysfunction and hypertension, with beneficial effects in the setting of neurohormonal dysregulation and diabetes.

Perspectives

NO is a critical regulator of vascular health. Oxidative dysregulation of NO bioavailability underlies most vascular pathologies. Previous work has determined that S-glutathionylation of eNOS can render it inactive, leading to 70% reduction in NO and this is accompanied by out-of-control superoxide generation. In the current study, we showed for the first time that FXYD1 knockout is accompanied by vascular dysfunction that is associated with excessive superoxide generation. Our evidence suggests that FXYD1 may protect eNOS from S-glutathionylation-dependent oxidative inhibition, and this could be beneficial in CVDs including hypertension and diabetes. Future studies should determine the wider implications of this relationship in vascular diseases and investigate the most pragmatic approaches to using FXYD1 as a therapeutic target.

Acknowledgments

G.A. Figtree is supported by a National (Australia) Health and Medical Research Council (NHMRC) Practitioner Fellowship, as well as New South Wales Office of Health and Medical Research and Heart Research Australia. This study was funded by a project grant from the NHMRC (APP1080468).

Footnotes