S-Nitrosothiols signal hypoxia-mimetic vascular pathology (original) (raw)
NAC and SNOAC cause hypoxia-mimetic PAH. Three weeks of normobaric hypoxia (10%) increased RV systolic pressure, RV weight (relative to LV and septum [LV+S]) and muscularization of small (<80-μm) pulmonary arterioles in both C57BL/6 and C57BL/6/129SvEv mice (n = 4–14 each; Figure 1, A–E). The effects of 3 weeks’ exposure to either NAC (10 mg/ml [52 mM]) or SNOAC (1 mM) in the drinking water were similar to those of 3 weeks’ exposure to hypoxia in C57BL/6/129SvEv mice, increasing RV pressure (n = 3–4 each, P < 0.001 relative to normoxia), RV/LV+S weight ratio (n = 6–35 each; P < 0.02 relative to normoxia), and muscularization of the small pulmonary arterioles. (n = 3–8; P < 0.003 relative to normoxia) (Figure 1, A–E). Of note, PA pressures in the treatment groups were greater than 125% those in controls in the hypoxia, NAC, and SNOAC groups, despite anesthesia, which can blunt PAH (ref. 8; Figure 1C). Similar results were obtained in C57BL/6 mice (n = 4–12 each); however, the increase in RV/LV+S ratio was not significant in these mice in the case of SNOAC (Figure 1, A and B).
Systemic NAC and SNOAC cause hypoxia-mimetic PAH in mice. C57BL/6/129SvEv, C57BL/6, and eNOS–/– male mice were maintained in normoxia (N, red; 21% O2) or hypoxia (H, black; 10% O2); or were treated with NAC (blue) or SNOAC (green) in their drinking water for 3 weeks. (A) Relative RV weight was determined as the ratio of the weight of the RV to the LV+S weight. (B) RV systolic pressures were measured in the closed chest using a Millar 1.4 F catheter/transducer. (C) Representative RV pressure (RVP) tracings (each = 1 s). (D) Lung section images from C57BL/6/129SvEv mice immunostained for von Willebrand factor and α-SMA to illustrate changes in muscularization after 3 weeks of exposure to normoxia, hypoxia, NAC, or SNOAC. Scale bar: 100 μm (applies to all panels). (E) Changes in muscularization in C57BL/6/129SvEv mice in the small (<80-μm) vessels from histological sections (as in D) counted by an observer blinded to the protocol. FM, fully muscular; PM, partly muscular, NM, nonmuscular. Significant increases in muscularization in each treatment group were seen in comparison to normoxic controls. Data are mean ± SEM. ζ_P_ < 0.02, *P < 0.001, †P < 0.003, by 1-way ANOVA followed by pairwise comparison, all compared with normoxic control.
In dose-response experiments, higher-dose NAC (10 mg/ml) increased RV weight relative to normoxia at 3 weeks in C57BL/6/129SvEv mice (Figure 1A); however, at a lower dose (1 mg/ml), RV/LV+S was normal at 3 weeks (0.27 ± 0.005; n = 5; P = NS compared with no treatment). On the other hand, low-dose NAC increased RV systolic pressure at 3 weeks to 27.5 mmHg (n = 4; P < 0.04 compared with normoxia).
In time course experiments performed using C57BL/6/129SvEv mice, there was a significant increase in RV pressure after 1 week of NAC exposure (10 mg/ml; mean RV pressure with NAC, 22.8 ± 3.3 mmHg, versus without NAC, 13.7 ± 0.7 mmHg; n = 5–8 each; P < 0.01), persisting at 2 and 3 weeks. RV/LV+S increased more slowly: a significant difference between NAC-exposed and -unexposed animals was not observed until 3 weeks (n = 5–8 each at 1 and 2 weeks; P = NS). Concomitant 3-week exposure to both hypoxia and 10 mg/ml NAC did not result in a greater increase in mean RV/LV+S (0.35 ± 0.082 after NAC and hypoxia together versus 0.34 ± 0.020 after hypoxia alone; n = 4–6 each; P = NS) or RV pressure (21.4 ± 1.30 mmHg after NAC plus hypoxia versus 19.9 ± 1.03 mmHg after hypoxia alone; n = 4–6 each; P = NS), suggesting functional overlap between hypoxia- and NAC-stimulated pathways.
Whole-lung expression of hypoxia-inducible mitogenic factor (HIMF), fibronectin, and eNOS, proteins associated with the development of PAH in some models (16–20), was increased by hypoxia and by 3 weeks of NAC treatment (10 mg/ml; n = 3–6 each; P < 0.05 [HIMF and fibronectin] and P < 0.01 [eNOS]; Figure 2, A–D). Of note, VEGF-A and endothelin 1 protein expression did not increase significantly with hypoxia or NAC in our model (Figure 2, E and F). Though these proteins can be upregulated by low pO2 in vitro, results are variable in PAH models (21–26), suggesting that PAH-associated vascular remodeling is more complex than would be predicted based on the effects of low pO2 alone. Indeed, VEGF and eNOS expression can be associated with both protection against PAH and development of PAH, depending on the model and time course (18–29). NAC did not change expression of neuronal or inducible NOS isoforms at 3 weeks (data not shown). NAC exposure increased whole-lung HIMF mRNA expression and transiently increased eNOS mRNA (Figure 2, D and G). Interestingly, parallel increases in lung protein and mRNA were not observed for the expression of some genes known to be upregulated in hypoxia in vitro, either because of nonvascular expression measured in whole-lung homogenates and/or because of the complexity of pathways involved in the response to hypoxia and the development of PAH in vivo. For example, mRNA levels for VEGF-A increased after 3 weeks of NAC treatment (n = 3; P < 0.04; Figure 2G), though there was no significant change in VEGF-A protein levels (Figure 2E). On the other hand, the change in mRNA expression for fibronectin was not significant (n = 2–5 each; P = NS; Figure 2G), though expression of the corresponding protein increased (described above; Figure 2A). Further, mRNA for glucose transport protein–1 (Glut-1), conventionally upregulated by hypoxia, did not increase (n = 3; P = NS). Consistent with this complexity, our time course analysis revealed an NAC-induced increase in whole-lung eNOS mRNA that preceded the NAC-induced increase in eNOS protein expression but was transient (Figure 2D).
Three weeks of NAC treatment or hypoxia increases the whole-lung expression of certain genes associated with the development of PAH in mice. The expression of fibronectin (A), HIMF (B), eNOS (C and D), VEGF-A (E), and endothelin (F) in whole-lung homogenates from NAC-treated mice was examined by immunoblot. Fold increase in density relative to MAPK (equal loading control) was determined for each condition. The increases in fibronectin, HIMF, and eNOS (n = 3–5 each) were significant. (G) Three weeks of NAC treatment also increased whole-lung mRNA, assayed relative to 18S RNA by RT-PCR, for HIMF and VEGF-A but not fibronectin or Glut-1 (n = 3 each). Time course analysis of NAC-treated mice (D) revealed that the increase in whole-lung eNOS mRNA (filled squares, left axis) preceded the increase in eNOS protein expression (open circles, right axis) but decreased by 3 weeks. *P < 0.05; #P < 0.01.
Three weeks of NAC (10 mg/ml) did not affect systemic or portal vascular morphology (Supplemental Data; supplemental material available online with this article; doi:10.1172/JCI29444DS1) or hemoglobin (Hb) (12.5 ± 0.64 g/dl after NAC treatment versus 12.1 ± 0.65 g/dl in controls). Likewise, 3 weeks of oral SNOAC (1 mM) did not affect the systemic or portal vasculature (Supplemental Data).
NAC is converted to SNOAC in vivo and during erythrocytic deoxygenation in vitro and in vivo. The chemical mechanisms, including both inhibition of nitrosative/oxidative stress (30–32) and NO transfer chemistry (1, 33), by which NAC could cause PAH were investigated. NAC did not affect pulmonary vascular immunostaining for 3-nitrotyrosine (Supplemental Data), suggesting that its ability to cause PAH was not simply the result of tissue injury associated with altered nitrosative or oxidative stress.
In contrast, the ratio of total erythrocytic _S_-nitrosothiol concentration to Hb (SNOrbc) (11) in NAC-treated mice (1.0 × 10–4 ± 0.7 × 10–4; n = 4) was lower than that in control animals (3.7 × 10–4 ± 3.2 × 10–4; n = 3; P < 0.05; Figure 3A). Decreased SNOrbc content in NAC-treated mice could be expected if the NO group on erythrocytic protein thiols were transferred (1, 33) to NAC according to the reaction: protein-S-NO + NAC → protein-SH + SNOAC. Therefore, the formation of plasma SNOAC was assayed by MS. SNOAC was identified in the RV plasma of the NAC-exposed animals (1.6 ± 0.9 μM versus 0 μM in unexposed animals; P = 0.04; Figure 3A), a finding confirmed both by coelution with 15N-labeled SNOAC and by NO displacement from the _S_-nitrosothiol bond using HgCl2 (11) (Figure 3B). SNOAC was not detected in the LV of NAC-treated animals (n = 7). Moreover, ex vivo human blood deoxygenation in the presence of NAC in a tonometer (in nitrogen with 5% CO2; pH maintained at 7.3; ref. 11) resulted in both loss of SNOrbc content and a nearly stoichiometric formation of SNOAC (n = 3; P < 0.05 by ANOVA followed by pairwise comparison with the maximal value; Figure 3, C and D). The concentration of SNOAC increased as the fraction of oxygenated Hb decreased in intact erythrocytes ex vivo (Figure 3D). There was no transfer from erythrocytes to NAC when blood was maintained at 100% saturation for 15 minutes (total SNOrbc was 2.1 × 10–4 ± 0.7 × 10–4 initially and 2.0 × 10–4 ± 0.6 × 10–4 at 15 minutes; SNOAC assayed by MS was undetectable). These findings are consistent with evidence that SNOrbc distribution is, in part, dependent on oxyhemoglobin saturation (3, 11, 12) and that thiols accelerate the desaturation-induced loss of SNOrbc content (11). SNOAC formation was pseudo-first order (in excess NAC) and relatively slow (k ~5.3 × 10–10 M/s).
SNOAC is formed from NAC in blood ex vivo and in vivo. (A) The SNOrbc in heparinized LV blood (black bars), measured by reductive chemiluminescence (11), was lower than normal following 3 weeks of treatment with 10 mg/ml NAC (n = 3–4 each). In the same mice, plasma SNOAC levels (gray bars; measured by MS) increased from undetectable to approximately 2 μM over the same time (*P < 0.05). (B) Serum SNOAC, measured by MS, formed in NAC-treated mice (3 weeks). Left: liquid chromatogram; right: MS spectrum. NAC-treated mice had a SNOAC peak (m/z 193; red) coeluting with 15N-labeled SNOAC standard (m/z 194; black) that was absent in untreated animals (green) and was not detected in NAC-treated mice after serum pretreatment with HgCl2 to displace NO+ from the thiolate (blue). (C) Oxygenated erythrocytes were deoxygenated ex vivo (argon; ref. 11) in the presence of 100 μM NAC; supernatant SNOAC was measured by MS (above). SNOAC concentration increased with oxyhemoglobin (Oxy Hb) desaturation (co-oximetry: inset), being maximal at 59.3% saturation (blue), less at 77.2% saturation (green), and undetectable at 98% saturation. (D) SNOAC (filled circles) accumulated as the concentration of _S_-nitrosothiol–modified Hb (SNOHb; open circles) and oxyhemoglobin saturation (Hb SO2; blue line) both decreased in heparinized whole blood using argon with 5% CO2 (pH 7.3) in a tonometer. Both the increase in SNOAC and the loss of SNOrbc between 0 and 20 minutes were significant (P < 0.01 by ANOVA followed by pairwise comparison to the maximum value; n = 3). #SNOAC levels were below the limit of detection when the oxyhemoglobin saturation was greater than 80%.
We also considered that NAC could be converted to circulating SNOAC by reacting with NO in endothelial cells. _S_-Nitrosothiol concentrations in whole-cell lysates (43 ± 7.3 nM) were not increased by 4 hours’ exposure to 50 μM NAC in the presence of 5 μM exogenous NO (45 ± 3.1 nM; P = NS; n = 3). However, cellular levels were increased by exposure to 5 μM SNOAC (2.0 ± 0.4 μM; n = 3; P < 0.05 when compared with NAC alone and control; Figure 4A).
SNOAC recapitulates in primary pulmonary arterial endothelial cells the hypoxia-mimetic whole-lung effect of chronic NAC administration on Sp3 expression in vivo. (A) One micromolar SNOAC, but not 50 μM NAC, treatment (4 hours each) increased intracellular _S_-nitrosothiol levels (assayed by Cu/cysteine chemiluminescence; ref. 11) in primary murine pulmonary endothelial cells (*P < 0.05 compared with SNOAC treatment). (B) Immunoblot showing increased Sp3 expression relative to MAPK in the whole-lung homogenates of mice treated for 3 weeks with 10 mg/ml NAC but not in those of control mice. By densitometry, this increase was significant (P < 0.01). (C) Paradoxically, however, NAC (50 μM; 4 hours) did not increase Sp3 expression relative to β-actin in primary murine pulmonary endothelial cells in vitro, while both SNOAC (1 μM; 4 hours) and hypoxia (10%; 4 hours) did.
eNOS-deficient mice are protected from the hypoxia-mimetic effects of NAC. eNOS is proposed to be important for maintaining SNOrbc content (34). We studied NAC-induced PAH in eNOS–/– mice. The SNOrbc content of these mice was 0.39 × 10–4 ± 0.15 × 10–4 (nearly a log order lower than in wild-type mice; P = 0.004). At baseline, eNOS–/– mice had slight increases in RV pressure and RV weight relative to those of the wild-type background (C57BL/6) mice, as reported previously (refs. 27–29; Figure 1, A and B). Strikingly, eNOS-deficient mice were completely protected from NAC-induced PAH — suggesting that the chronic effects of NAC to increase PAH are eNOS dependent — but they were not protected from SNOAC-induced PAH (Figure 1, A and B).
NAC could cause PAH in eNOS-replete mice by depleting endothelial NO or depleting eNOS-derived SNOrbc (ref. 34; Figure 3). However, (a) NAC was not converted to SNOAC in eNOS-replete pulmonary endothelial cells in culture, even in the presence of exogenous NO; (b) SNOrbc levels in eNOS–/– mice at baseline were lower than those in wild-type, NAC-treated mice; and (c) SNOAC caused, rather than ameliorating, PAH in eNOS–/– mice. Therefore, SNOAC excess, rather than endothelial NO or SNOrbc depletion, appears most likely to have caused PAH in our model.
Chronic GSNO exposure in normoxia does not cause PAH. GSNO is an endogenous _S_-nitrosothiol that is similar to SNOAC in the kinetics of its homolytic decomposition (35). Normoxic C57BL/6/129SvEv mice receiving 1 mM GSNO in their drinking water for 3 weeks did not develop PAH as measured by any parameter (n = 3 in the GSNO group and 20 in the untreated control group; P = NS).
The molecular effects of S-nitrosothiols in vitro are hypoxia mimetic and recapitulate whole-lung effects of NAC in vivo. _S_-Nitrosothiols can have hypoxia-mimetic gene-regulatory effects involving the expression of transcription factors relevant to pulmonary vascular disease, including the hypoxia-inducible factor (HIF) (5, 13) and specificity protein (Sp) families (14). HIF 1, HIF 2, Sp1, and Sp3 are involved in regulating the expression of the genes upregulated in the lungs of mice treated chronically with NAC (9, 16, 17, 24, 25, 36). Because the expression and activity of the HIF and Sp families are affected by _S_-nitrosothiols in vitro (5, 13, 14, 37, 38), we tested whether chronic NAC or SNOAC could affect their pulmonary expression in mice. Interestingly, NAC increased whole-lung Sp3 expression in vivo but not in primary pulmonary vascular endothelial cells in vitro (Figure 4, B and C). SNOAC, on the other hand, increased both intracellular _S_-nitrosothiol levels and nuclear Sp3 expression in primary murine pulmonary vascular endothelial cells (Figure 4, A and C). Thus, conversion of NAC to SNOAC (as shown in Figure 3) could be one explanation for the paradox that NAC can increase Sp3 expression in vivo but not in vitro.
Similarly, NAC increased HIF 1 activity in whole-lung extracts in vivo (Figure 5A), though NAC suppresses upregulation of HIF 1α, the oxygen-labile subunit of HIF 1, by both substance P (31) and by oxidized low-density lipoprotein (32) in vitro. As with Sp3, conversion of NAC to SNOAC may help to explain this paradox: SNOAC and hypoxia, but not NAC, increased HIF 1α expression, relative to β actin, in primary pulmonary vascular endothelial cells in vitro (Figure 5B). This is consistent with previous work showing HIF 1α stabilization by GSNO (5, 13); indeed, we found that GSNO reduced HIF 1α monoubiquitination in bovine pulmonary arterial cells (BPAECs) expressing HA-tagged HIF 1α and a histidine-tagged, dominant-negative ubiquitin that stops ubiquitin chain propagation (K48R DN-Ub) (Figure 5C).
_S_-Nitrosothiols prevent normoxic ubiquitination and degradation of HIF 1α. (A) NAC treatment (10 mg/ml; 3 weeks) increased whole-lung HIF 1–DNA binding activity. Complexes were supershifted (ss) with anti–HIF 1β and eliminated with excess cold probe (P). (B) SNOAC (1 μM), like GSNO (5, 13, 38), increased normoxic HIF 1α expression in nuclear extracts isolated from primary murine pulmonary endothelial cells. NAC alone (50 μM) did not affect HIF 1α expression. β-Actin was used as a protein load control. (C) In BPAECs transfected with HA-tagged HIF 1α and dominant-negative His-6-Myc–tagged ubiquitin (DN-Ub), ubiquitinated proteins were isolated using a nickel column and immunoblotted for HIF 1α. Both hypoxia and GSNO (G; 10 μM) inhibited HIF 1α ubiquitination relative to normoxia. (D) In COS cells cotransfected with HA-tagged HIF 1α and FLAG-tagged pVHL, GSNO (10 μM) prevented the coimmunoprecipitation of HIF 1α with pVHL. (E) _S_-nitrosylation of pVHL by SNOAC (5 μM) in equal protein aliquots isolated from HeLa cells was identified by biotin substitution (49); in the absence of ascorbate, _S_-nitrosylated pVHL was not detected. (F) Similarly, SNOAC and GSNO (5 μM) increased pVHL _S_-nitrosylation in pVHL-overexpressing 786-O cells. (G) C162, but not C77, was identified by biotin substitution to be _S_-nitrosylated in BPAECs transfected with wild-type cysteine 77 to serine mutant (C77S), C162S, or combined C77S/C162S pVHL exposed to SNOAC (1 μM). Native pVHL and MAPK immunoblots represented the pVHL expression and protein load controls, respectively. All in vitro treatments were for 4 hours.
Mechanisms by which _S_-nitrosothiols inhibit HIF 1α ubiquitination and degradation are complex, involving both NO-radical reactions and transnitrosation reactions (1, 5, 13, 37–40). Low-micromolar SNOAC levels derived from NAC in vivo (Figure 3) would not generate concentrations of intravascular NO radical relevant to endothelial gene regulation (38) in the presence of millimolar concentrations of intravascular Hb (41); therefore, signaling through transnitrosation chemistry would seem more likely to be relevant to our in vivo model. NO transfer between thiols could stabilize HIF 1α through effects on prolyl hydroxylase, Akt signaling, and HIF 1α itself (13, 37–43). In addition, GSNO modifies E3 ubiquitin ligases through transnitrosation (44). Both SNOAC (Figure 5D) and GSNO (data not shown) inhibit the coimmunoprecipitation of HIF 1α with its E3 ligase, von Hippel–Lindau protein (pVHL) (45) in COS cells overexpressing both FLAG-tagged pVHL and HA-tagged HIF 1α. Strikingly, SNOAC _S_-nitrosylated native pVHL in HeLa cells (Figure 5E), and SNOAC and GSNO increased pVHL _S_-nitrosylation in 786-O cells stably overexpressing pVHL (45), in which baseline NOS activity and/or NO transfer reactions (1) also resulted in baseline pVHL _S_-nitrosylation (Figure 5F). Therefore, we studied the 2 potential pVHL _S_-nitrosylation targets (C77 and C162; refs. 45, 46). In BPAECs transiently transfected with wild-type pVHL or with cysteine-to-serine mutants C77S, C162S, or C77S/C162S, mutation of C162 eliminated SNOAC-induced pVHL _S_-nitrosylation (Figure 5G), consistent with evidence that C162 is required for pVHL to bind elongin C and ubiquitinate HIF 1α (46). Note that this effect could also reflect _S_-nitrosylation of a protein interacting with pVHL C162. Taken together, these data suggest that one element of the mechanism by which _S_-nitrosothiols may be hypoxia-mimetic involves prevention of HIF 1α ubiquitination, possibly through _S_-nitrosylation of pVHL C162. Additional studies are needed to determine the extent to which each mechanism is involved in the hypoxia-mimetic vascular effects of SNOAC in vivo.