S-glutathionylation uncouples eNOS and regulates its cellular and vascular function

Abstract

Endothelial nitric oxide synthase (eNOS) is critical in the regulation of vascular function, and can generate both nitric oxide (NO) and superoxide (O(2)(•-)), which are key mediators of cellular signalling. In the presence of Ca(2+)/calmodulin, eNOS produces NO, endothelial-derived relaxing factor, from l-arginine (l-Arg) by means of electron transfer from NADPH through a flavin containing reductase domain to oxygen bound at the haem of an oxygenase domain, which also contains binding sites for tetrahydrobiopterin (BH(4)) and l-Arg. In the absence of BH(4), NO synthesis is abrogated and instead O(2)(•-) is generated. While NOS dysfunction occurs in diseases with redox stress, BH(4) repletion only partly restores NOS activity and NOS-dependent vasodilation. This suggests that there is an as yet unidentified redox-regulated mechanism controlling NOS function. Protein thiols can undergo S-glutathionylation, a reversible protein modification involved in cellular signalling and adaptation. Under oxidative stress, S-glutathionylation occurs through thiol-disulphide exchange with oxidized glutathione or reaction of oxidant-induced protein thiyl radicals with reduced glutathione. Cysteine residues are critical for the maintenance of eNOS function; we therefore speculated that oxidative stress could alter eNOS activity through S-glutathionylation. Here we show that S-glutathionylation of eNOS reversibly decreases NOS activity with an increase in O(2)(•-) generation primarily from the reductase, in which two highly conserved cysteine residues are identified as sites of S-glutathionylation and found to be critical for redox-regulation of eNOS function. We show that eNOS S-glutathionylation in endothelial cells, with loss of NO and gain of O(2)(•-) generation, is associated with impaired endothelium-dependent vasodilation. In hypertensive vessels, eNOS S-glutathionylation is increased with impaired endothelium-dependent vasodilation that is restored by thiol-specific reducing agents, which reverse this S-glutathionylation. Thus, S-glutathionylation of eNOS is a pivotal switch providing redox regulation of cellular signalling, endothelial function and vascular tone.

Figure 1. S-glutathionylation of heNOS occurs and inhibits NOS activity

a, Immunoblotting of heNOS S-glutathionylation. Top: immunoblotting of protein S-glutathionylation (PrS-SG) with anti-GSH antibody. Control non-S-glutathionylated heNOS (1 μg in 20 μl) or heNOS S-glutathionylated by 0.5, 1 or 2 mM GSSG at room temperature (23 °C) for 1 h. Treatment with 2-mercaptoethanol (ME) after S-glutathionylation with 2 mM GSSG reversed the S-glutathionylation. Bottom: immunoblotting with anti-eNOS antibody. b, Effect of S-glutathionylation and S-alkylation on heNOS activity. NOS activity was measured from control, S-glutathionylated (2 mM GSSG for 20 min) or alkylated (1 mM NEM for 20 min) heNOS. NOS activity of treated or untreated heNOS was fully inhibited by L-NAME (1 mM) or EGTA (1 mM). c, d, Effects of S-glutathionylation on O₂^•− generation from heNOS. O₂^•− generation was measured from control, S-glutathionylated (as in b) or alkylated (as in b) BH₄-bound heNOS by EPR spin trapping with 25 mM 5-diethoxyphosphoryl-5-methyl-1-pyrroline N-oxide (DEPMPO). c, Spin-trapping showed no signal in the absence of heNOS and only trace signal from control enzyme; S-glutathionylation triggered a marked increase in O₂^•− generation with a O₂^•−-adduct spectrum that was quenched by Cu, Zn superoxide dismutase (SOD) (200 U ml⁻¹). d, Effect of L-NAME (1 mM) and EGTA (1 mM) on O₂^•− generation from control, S-glutathionylated and alkylated BH₄-bound heNOS. Results in b and d are shown as means and s.e.m. (n = 3–5).

Figure 2. Cysteine mutants (C689S, C908S and C689S/C908S) of heNOS resist S-glutathionylation and secondary uncoupling

WT heNOS and heNOS C689S, C908S and C689S/C908S mutants were treated with 2 mM GSSG. a, Percentage loss of NOS activity after treatment of heNOS with GSSG. b, Percentage increase in O₂^•− generation after treatment of heNOS with GSSG. c, Ratio of relative eNOS S-glutathionylation to eNOS protein. The relative intensity of eNOS S-glutathionylation/eNOS protein was normalized to the wild-type value. The Cys→Ser mutants maintained a NOS activity similar to that of the wild type (120 ± 12 n mol min⁻¹ mg⁻¹). Results are shown as means and s.e.m. (n =3).

Figure 3. Effect of redox stress on eNOS S-glutathionylation and function in endothelial cells

a, Immunostaining of eNOS (left column, green fluorescence) and S-glutathionylation (second column, red fluorescence) in control BAECs and cells preincubated with BCNU. The third column shows the merged S-glutathionylation/eNOS image along with 4′,6-diamidino-2-phenylindole (DAPI) staining of the nucleus (blue). eNOS staining and S-glutathionylation seem to co-localize. The right-hand column shows O₂^•− detection with dihydroethidine (DHE), which is oxidized by O₂^•− to a product with red fluorescence, and the cell nuclei were counterstained with DAPI (blue). Increased O₂^•− generation was seen in the BCNU-treated cells. Bottom: immunoprecipitation of eNOS from the BCNU-treated cells shows that eNOS S-glutathionylation occurs but is reversed by 1 mM DTT. The upper row shows immunoblotting with anti-GSH antibody; the lower row shows immunoblotting with anti-eNOS antibody. b, Effects of eNOS silencing from BAECs on BCNU-induced O₂^•− generation. Top: immunoblotting against eNOS to determine the efficiency of eNOS silencing. Middle: confocal microscopy of eNOS (upper row) and O₂^•− measurements with DHE (lower row). NOS3 short interfering RNA (siRNA) greatly decreases eNOS expression in BAECs (middle column). Bottom: graph of the effect of eNOS silencing on BCNU-induced O₂^•− generation. Results are shown as means and s.e.m. (n =5). Asterisk, P < 0.001.

Figure 4. Effect of redox stress on eNOS S-glutathionylation and function in vessels

a, Endothelium-dependent and endothelium-independent vasorelaxation in control and BCNU (80 μM)-treated rat aortic rings. BCNU markedly decreased endothelium-dependent relaxation to acetylcholine (left panel) but not endothelium-independent relaxation by the NO donor NONOate (right panel). DTT (1 mM for 20 min) reversed the BCNU-induced inhibition of relaxation (left panel). Aortic relaxation is plotted as the percentage decrease in phenylephrine (PHE)-induced contraction against agonist concentration on a logarithmic scale. Results are shown as means ± s.e.m.; P < 0.05, BCNU versus control or BCNU + DTT (n =4). b, Endothelium-dependent vasorelaxation in spontaneously hypertensive (SHR) and control (WKY) aortic rings. SHR rings showed a marked decrease in relaxation to acetylcholine; however, DTT (as above) re-established the acetylcholine response. Endothelium-independent relaxation (right) was similar for both SHR and WKY rings. Aortic relaxation is expressed as in a. P < 0.05, SHR versus WKY or SHR + DTT (n =4). See also Supplementary Fig. 10. c, eNOS S-glutathionylation of SHR and WKY aortae. Top: WKY and SHR aortae, either untreated or DTT-pretreated as in b, were homogenized. This was followed by immunoprecipitation with anti-eNOS antibody. The immunoprecipitation products were separated by SDS–PAGE followed by immunoblotting against anti-GSH and anti-eNOS antibodies. In SHR aortae, eNOS S-glutathionylation was markedly increased compared with WKY aortae and was abolished by pretreatment with DTT. Bottom: ratio of relative intensity of eNOS S-glutathionylation/eNOS, normalized to SHR aortae. There is only trace eNOS S-glutathionylation in WKY aortae, whereas high levels are seen in SHR aortae. There was no detectable (n.d.) NOS S-glutathionylation in DTT-pretreated WKY or SHR aortae. Asterisk, P < 0.001 versus SHR (n =5).