Recent studies have demonstrated that hydrogen sulfide (H(2)S) is produced within the vessel wall from L-cysteine regulating several aspects of vascular homeostasis. H(2)S generated from cystathione γ-lyase (CSE) contributes to vascular tone; however, the molecular mechanisms underlying the vasorelaxing effects of H(2)S are still under investigation.
METHODS AND RESULTS:
Using isolated aortic rings, we observed that addition of L-cysteine led to a concentration-dependent relaxation that was prevented by the CSE inhibitors DL-propargylglyicine (PAG) and β-cyano-l-alanine (BCA). Moreover, incubation with PAG or BCA resulted in a rightward shift in sodium nitroprusside-and isoproterenol-induced relaxation. Aortic tissues exposed to PAG or BCA contained lower levels of cGMP, exposure of cells to exogenous H(2)S or overexpression of CSE raised cGMP concentration. RNA silencing of CSE expression reduced intracellular cGMP levels confirming a positive role for endogenous H(2)S on cGMP accumulation. The ability of H(2)S to enhance cGMP levels was greatly reduced by the nonselective phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine. Finally, addition of H(2)S to a cell-free system inhibited both cGMP and cAMP breakdown.
These findings provide direct evidence that H(2)S acts as an endogenous inhibitor of phosphodiesterase activity and reinforce the notion that this gasotransmitter could be therapeutically exploited.
Nitric oxide (NO) is believed to account for most of the endothelium-derived relaxing factor activity released within the vessel wall, at least in some vessels.1 On muscarinic stimulation, NO is produced following the conversion of l-arginine to NO by endothelial nitric oxide synthase (eNOS).2 NO diffuses from the endothelium to the underlying smooth muscle cell layer, where it stimulates soluble guanylate cyclase to produce cGMP. cGMP in turn activates protein kinase G (PKG), which initiates a cascade of events leading to relaxation.2,3 Hydrogen sulfide (H2S) is emerging as a new gaseous signaling molecule in the cardiovascular system.4,5 Vascular endothelial cells express cystathionine γ-lyase (CSE) and produce measurable amounts of this gasotransmitter.6 Recent evidence suggests that H2S exhibits endothelium-derived relaxing factor activity.6 In addition, it has been shown that muscarinic stimulation leads to CSE activation in the endothelium, triggering the conversion of l-cysteine to H2S and that CSE, like eNOS, is a calcium/calmodulin-dependent enzyme. Therefore, within the vascular wall, these 2 pathways coexist and serve a similar function. The relative amounts of NO versus H2S likely depend on the vascular bed studied7 or on the state of the tissue, eg, healthy versus diseased.5,8
Mice with targeted disruption of the CSE locus (CSE null mice) exhibit hypertension, similarly to what is observed in eNOS-null mice.6 Moreover, resistance arteries isolated from CSE knockout mice display a marked inhibition of methacoline-induced relaxation. These observations, along with data in the literature showing similar effects for the 2 gasotransmitters, strongly suggest that NO and H2S either cooperate dynamically to maintain vessel homeostasis or serve as each other’s backup system under pathological conditions.9,10
Although the mechanism of action of NO-induced vasorelaxation is well understood, that of H2S remains less clear. Glibenclamide attenuates the vasorelaxing action of H2S, but a residual dilatory response is observed,9,11 suggesting that although KATP channel activation is important for H2S-induced vasodilation, additional mediators contributing to tone reduction by H2S must exist. Therefore, the aim of the present study was to investigate the contribution of cyclic nucleotides in H2S-stimulated relaxation in vascular tissue.
Isolated Aortic Rings Experimental Protocol
Male Wistar rats 8 to 10 weeks of age were euthanized, and thoracic aortas were rapidly dissected and cleaned from fat and connective tissue. Rings 2 to 3 mm in length were cut and placed in organ baths (2.5 mL) filled with oxygenated (95% O2 to 5% CO2) Krebs solution at 37°C, mounted to isometric force transducers and connected to a Graphtec linearcorder. After ≈60 minutes of equilibration period, rings were challenged with phenylephrine (PE) (1 μmol/L) until the responses were reproducible. To verify the integrity of the endothelium, an acetylcholine (Ach) cumulative concentration-response curve (10 nmol/L to 30 μmol/L) was performed on PE-precontracted rings. Tissues that relaxed less than 80% were discarded. In a separate set of experiments, rings were precontracted with PE (1 μmol/L), once a plateau was reached, l-cysteine (1 μmol/L to 10 mmol/L), Ach (10 nmol/L to 30 μmol/L), and sodium nitroprusside (SNP) (1 nmol/L to 3 μmol/L) cumulative concentration-response curves were performed. To inhibit CSE, aortic rings were exposed to dl-propargylglyicine (PAG) and β-cyano-l-alanine (BCA) (1, 3, and 10 mmol/L) for 15 and 60 minutes, respectively, and then cumulative-concentration response curves to different vasorelaxing agents were performed.
H2S determination was performed according to Stipanuk and Beck12 with some modifications. Aortic rings were homogenized in a lysis buffer (potassium phosphate buffer [100 mmol/L, pH=7.4] sodium orthovanadate [10 mmol/L], and protease inhibitor). The homogenates were added in a reaction mixture (total volume 500 μL) containing pyridoxal-5′-phosphate (2 mmol/L, 20 μL), l-cysteine (10 mmol/L, 20 μL), and saline (30 μL). After 30 minutes, ZnAc (1%, 250 μL) was added to trap H2S, followed by trichloroacetic acid (TCA) (10%, 250 μL). Subsequently, N,N-Dihethyl-p-phenyl-enediamine sulphate (DPD) (20 mmol/L, 133 μL) in 7.2 mol/L HCl and FeCl3 (30 mmol/L, 133 μL) in 1.2 mol/L HCl were added. Absorbance was measured at 650 nm after 20 minutes. The H2S concentration of each sample was calculated against a calibration curve of sodium hydrosulfide (NaHS) (3.12 to 250 μmol/L), and results were expressed as nmol of H2S/mg of protein.
Aortic rings were placed in Krebs solution at 37°C and incubated with vehicle, PAG, or BCA (10 mmol/L). Rings were stimulated with SNP (1 μmol/L); after 15 minutes, tissues were snap frozen and homogenized in 8 volumes of buffer containing 5% trichloroacetic acid per gram of tissue. cGMP was then extracted and measured using a commercially available enzyme immunoassay kit (Assay Designs, Ann Arbor, Mich) following the manufacturer’s instructions. For experiments in cultured cells, the cells were incubated for 5 minutes (unless otherwise indicated) with NaHS. In some experiments, cells were pretreated with 3-isobutyl-1-methylxanthine (IBMX). To overexpress CSE in smooth muscle cells, gene transfer was accomplished using an adenovirus at 10 multiciplities of infection (a green fluorescent protein–expressing adenovirus was used as a control). After 48 hours, cells were washed and incubated for 5 minutes with a solution containing l-cysteine (1 mmol/L) in the presence or absence of PAG pretreatment (10 mmol/L, 30 minutes). To extract cGMP, cells were lysed in 0.1 N HCl, and cGMP content was measured in the extracts using a commercial kit.
Cell Culture and Transfections
Human umbilical vein endothelial cells were grown in M199 medium supplemented with 15% fetal bovine serum, endothelial cell growth supplement, heparin, and antibiotics. Rat aortic smooth muscle cells were isolated from 12- to 14-week-old male Wistar rats and cultured in DMEM supplemented with 10% fetal bovine serum and antibiotics. For transfections, COS-7 cells were used. The cells were plated in 24-well plates at a density of 5×104 cells per well and allowed to grow overnight. Cells were then transfected with appropriate plasmids using ExGen 500 in vitro transfection reagent according to the manufacturer’s instructions. cGMP levels or expression of proteins was measured 48 hours after transfection. For transfections of human umbilical vein endothelial cells with the control or CSE short interfering RNA (siRNA), the following protocol was used. Cells were plated in a 24-well plate and allowed to reach 75% confluence. The medium was then replaced, and the transfection reagent (2 μL · 5 nmol−1 · L−1 siRNA) was prepared in serum-free medium and added to the cells. Plates were then vigorously agitated to disperse siRNA evenly. Twenty-four hours posttransfection, cells were used for cGMP measurements or were lysed to determine protein levels.
Phosphodiesterase Activity Assay
The ability of NaHS to modulate phosphodiesterase (PDE) activity was evaluated by using a colorimetric cyclic nucleotide PDE assay kit. Briefly, the rationale of this assay relies on the cleavage of cGMP or cAMP by a cyclic nucleotide PDE. The 5′-GMP and 5′-AMP released were further cleaved into the nucleosides and phosphate. The phosphate generated is quantified using a colorimetric reaction using Biomol Green reagent in a modified Malachite Green assay. A nonspecific cyclic nucleotide PDE inhibitor, IBMX was included as a positive control for inhibitor screening. Different concentrations of NaHS (3, 10, 30 nmol/L) were tested and compared with IBMX (40 μmol/L); the procedure was performed according to manufacturer instructions.
All data are expressed as mean±SEM. Results were analyzed by 1- or 2-way ANOVA, followed by a post hoc test for multiple comparisons. A value of P<0.05 was considered significant.
Inhibition of CSE Attenuates Endothelium-Dependent and Endothelium-Independent Relaxations
Incubation of rat aortic rings with l-cysteine resulted in a concentration-dependent relaxation (Figure 1A and 1B). The maximal relaxation observed was 40%, occurring at 3 mmol/L l-cysteine. Treatment of aortic tissue with either one of the CSE inhibitors BCA or PAG inhibited the production of H2S in tissue homogenates (Figure 1C). The lack of complete inhibition of the colorimetric signal by the CSE inhibitors is likely due to the fact that this widely used assay also detects additional molecules carrying sulfhydryl groups or the presence of alternative pathways that generate H2S in vascular tissues. To determine whether endogenously produced H2S contributes to the Ach-induced vasodilatory response, rings were incubated with increasing amounts of BCA or PAG. In these experiments, we found that CSE inhibition reduced endothelium-dependent relaxations (Figure 2A and 2B).
To determine whether endogenous H2S alters endothelium-independent relaxation elicited by an NO donor (SNP), aortic rings were exposed to PAG or BCA before stimulation with SNP. Such pretreatment resulted in a rightward shift of the SNP relaxation curve and increased the EC50 of SNP (log EC50=−8.43±0.089, −7.41±0.062, and −7.71±0.07 for vehicle, PAG, and BCA, respectively; Figure 2C and 2D). It should be noted that the 2 CSE inhibitors had no effect on eNOS activity and did not quench the NO produced by directly reacting with it (supplemental Figure I, available online at http://atvb.ahajournals.org), suggesting that the observed effect was due to inhibition of H2S production rather than an interference of BCA or PAG with the eNOS/NO pathway.
H2S Regulates cGMP Levels
To study the mechanism through which H2S modulates the response to SNP, we determined the effects of H2S on cGMP levels. We found that blockade of CSE activity resulted in a significant reduction of SNP-induced cGMP accumulation in aortic tissue (Figure 3A). In addition, incubation of cultured rat aortic smooth muscle cells with NaHS, an H2S donor, led to a concentration-dependent increase in cGMP levels (Figure 3B) that reached a plateau at 50 μΜ NaHS. The NaHS-induced rise in cGMP was evident as early as 1 minute with cGMP levels, reached a maximum at 3 minutes, and remained elevated for at least 10 minutes (Figure 3C).
To confirm our observations that H2S has the ability to enhance cGMP levels, we infected smooth muscle cells with an adenovirus carrying the CSE cDNA. CSE overexpression elevated intracellular cGMP in a PAG-sensitive manner (Figure 4A). To confirm that endogenous H2S can raise intracellular cGMP levels, we used an siRNA approach to knock down CSE expression. Indeed, similarly to what had been observed with the pharmacological CSE inhibitors in tissue homogenates, reduction in CSE levels in cells attenuated cGMP levels (Figure 4B).
H2S Inhibits PDE Activity
To investigate whether the increase in cGMP observed after exposure to H2S was due to inhibition of its breakdown, we used the nonselective PDE inhibitor IBMX. We observed that H2S caused an increase in cGMP levels that was of similar magnitude to IBMX (Figure 5A). Moreover, in cells pretreated with IBMX, the ability of NaHS to enhance cGMP levels was greatly attenuated, suggesting that H2S and IBMX share a common target.
We next used a heterologous expression system to determine whether H2S can inhibit PDE5, a cGMP-specific PDE that is widely distributed in the cardiovascular system. Stimulation of COS-7 with exogenous H2S (NaHS) increased cGMP levels, probably because of the inhibition of PDEs expressed endogenously by COS-7 cells. The effect of NaHS calculated as percentage increase in cGMP levels was significantly enhanced in COS-7 cells overexpressing PDE5A (236±26%) compared with that of β-galactosidase-expressing cells (168±23%) (Figure 5B and supplemental Figure II). This latter result suggests that H2S inhibits PDE5A activity, confirming our hypothesis that H2S acts as a PDE inhibitor.
To directly test whether H2S inhibits PDE activity, we used a cell-free assay. The assay uses a mixture of different semipurified PDE isoforms and allows the evaluation of PDE activity by measuring the breakdown products of cGMP and cAMP. Initially, we ruled out that NaHS interferes with any of the assay reagents; we then proceeded to demonstrate that nanomolar amounts of NaHS (3, 10, 30 nmol/L) inhibited cGMP-PDE activity in a concentration-dependent manner (Figure 5C and Supplemental Table I). In particular, 10 to 30 nmol/L NaHS significantly, and comparably to IBMX, inhibited PDE activity, causing a reduction in the breakdown product 5′-GMP (Figure 5C). Similar results were observed for cAMP catabolism PDEs (Figure 6A). To evaluate whether inhibition of endogenously produced H2S interferes with vasodilatory responses that are cAMP-mediated, rings were incubated with BCA or PAG before being exposed to isoproterenol (Figure 6B). In these experiments, we found that CSE inhibition shifted the isoproterenol-induced relaxation curves to the right.
The main findings of the present study are that (1) l-cysteine promotes vasodilation through activation of CSE and H2S production; (2) both endothelium-dependent and endothelium-independent relaxations in the rat aorta are attenuated on inhibition of endogenously produced H2S; (3) exposure of smooth muscle cells to H2S increases their intracellular cGMP content, whereas inhibition of cell-derived H2S reduces cGMP levels; and (4) H2S inhibits PDE activity and modulates cyclic nucleotide levels.
H2S is produced in mammalian cells both through enzymatic and nonenzymatic pathways.4,5 The pyridoxal-5′-phosphate-dependent enzymes cystathionine β-synthase and cystathionine-γ-lyase (CSE) use the amino acid l-cysteine as a substrate to generate H2S, with the latter being the major H2S-producing enzyme in the vasculature. Additional pathways (eg, 3-mercaptopyruvate sulfurtransferase) have been shown to contribute to H2S production by endothelial cells.13 In our experiments, incubation of rat aortic rings with l-cysteine elicited concentration-dependent relaxations that were inhibited by PAG and BCA, suggesting that l-cysteine is taken up by the tissue and converted to H2S, leading to relaxation. In line with our findings, Cheng et al observed that incubation of the mesenteric bed with l-cysteine causes dilation that is blocked by CSE inhibition.11 In our studies, inhibition of l-cysteine-induced relaxations were affected to a different extent by PAG and BCA, with the former exhibiting a greater inhibitory effect. Although both inhibit CSE, PAG is a noncompetitive/“suicide” inhibitor, whereas BCA is a competitive inhibitor14,15; thus combining PAG and BCA would not be expected to yield an additional effect. Similarly to what was observed with l-cysteine, endothelium-dependent relaxations triggered by Ach were attenuated by CSE inhibition. This finding is in agreement with that of Yang et al, who showed that cholinergic receptor stimulation in EC increases H2S release by 3-fold; other calcium-elevating agents, such as A23187 and vascular endothelial growth factor, also stimulate release of H2S from the endothelium.6,16 Most interestingly, BCA and PAG reduced the endothelium-independent relaxation elicited by exogenously applied NO. It has been shown that H2S under some conditions inhibits the vasorelaxant effects of NO donors, probably because of a direct reaction between NO and H2S and the formation of a nitrosothiol-like molecule.9,17,18 A similar rightward shift in the SNP effect was observed in mesenteric arteries harvested from CSE null mice, but it was reported not to reach statistical significance.6 The finding that inhibition of endogenously produced H2S altered the response to exogenously added NO indicates that H2S exerts its effects on vessel tone by modulating cGMP levels and cGMP signaling in smooth muscle cells and prompted us to investigate the effects of this agent on cGMP levels.
Intracellular cGMP levels reflect the difference between the rate of cGMP synthesis and breakdown, the former being regulated by guanylyl cyclases and the latter by phosphodiestearases.19,20 It has been reported that although H2S binds with high affinity to heme,21 it does not appear to activate soluble guanylate cyclase22 and that H2S-induced vasorelaxation is not inhibited by a soluble guanylate cyclase inhibitor.23 Initially, we observed that addition of exogenous H2S caused a concentration- and time-dependent increase in cGMP levels, whereas pharmacological inhibitors or knockdown of CSE reduced cGMP levels in cells and tissues. To determine whether the changes that we observed in smooth muscle cell cGMP levels were due to activation of soluble guanylate cyclase or inhibition of PDE activity, we incubated cells with an nonselective PDE inhibitor. Such pretreatment reduced the H2S-stimulated increase in cGMP from 70-fold to less than 55% and suggested that H2S and IBMX share a common target, supporting the hypothesis that H2S inhibits PDE activity. Our hypothesis was further substantiated by the finding that H2S ameliorated the reduction in cGMP levels brought about by overexpression of PDE5A. Finally, to obtain direct evidence for the inhibition of PDE activity by H2S, we used a mixture of semipurified PDE isoforms in a commercially available cell-free assay. Results from these experiments confirmed that H2S is a nonselective inhibitor of PDE activity. Although our results do not unravel the mechanism through which H2S blocks PDE activity, at least 2 possibilities exist. PDEs are Zn-containing enzymes24; Zn is coordinated with histidine and aspartic acid residues in the substrate binding pocket and its removal abolishes PDE activity.24 At the same time, H2S is known to bind Zn5 and alter the activity of Zn-dependent enzymes. A second mechanism through which H2S might regulate PDE activity is through sulfhydration. Recently, mass spectrometric analysis revealed the attachment of an additional sulfur to the thiol (-SH) groups of cysteines, yielding a hydropersulfide (-SSH) moiety25 This posttranslational modification affects 10% to 25% of some proteins that include, but are not limited to GAPDH, β-tubulin, and actin. Most importantly, sulfhydration alters the proteins’ function, as it increases GAPDH activity by 7-fold and it enhances actin polymerization. It is therefore possible that H2S modifies some of the critical cysteine residues that regulate PDE activity.
On the basis of the above, the mechanism through which H2S promotes vasorelaxation would likely depend on the relative expression of KATP channels, the PDE isoform expressed, and the amount of H2S present in the microenvironment. For example, tissues expressing KATP channels and no or low PDE levels would be expected to dilate in a glibenclamide-inhibitable manner. The molecular basis for KATP channel activation by H2S was recently proposed to require the cysteine residues C6 and C26 in the extracellular loop of the SUR1 subunit,26 which most likely become sulfhydrated. On the other hand, tissues expressing high PDE5 levels and abundant amounts of PKG, but no KATP channels, would dilate in a cGMP-dependent manner. Because H2S was found to also inhibit cAMP breakdown and to modulate the relaxation to the cAMP-elevating agent isoproterenol, H2S responses in cells containing high amounts of cAMP-specific PDEs would be expected to be mediated mainly by PKA. In agreement with this hypothesis, in porcine pulmonary artery endothelial cells, inhibition of PKA but not PKG blocked the effects of NaHS on gp91(phox) expression.27 Another important observation that needs to be kept in mind is that PKG activates K+ channels, leading to hyperpolarization and relaxation.28,29 This latter finding allows us to speculate that in cells and tissues in which H2S responses are mediated by K+ channels, there might be both a cGMP-independent and a cGMP/PKG-dependent component.
In conclusion, we have demonstrated that H2S causes vasorelaxation by acting as a nonselective endogenous PDE inhibitor that boosts cyclic nucleotide levels in tissues. Interactions between the H2S and NO has long been suspected to occur. The 2 gasotransmitters have been proposed to interact in multiple ways, ranging from regulation of each other’s expression or activity to direct chemical reaction.4,9 In many cases, conflicting results have been obtained. For example, NO has been shown to enhance or reduce the effects of H2S on vascular tone.4 Our findings establish the existence of cross-talk between NO and H2S. Moreover, our data will aid in reconciling conflicting or difficult-to-explain observations regarding H2S-induced vasorelaxation.