Hydrogen Sulfide Metabolism and Pulmonary Hypertension

Pulmonary hypertension (PH) is a severe and multifactorial disease characterized by a progressive elevation of pulmonary arterial resistance and pressure due to remodeling, inflammation, oxidative stress, and vasoreactive alterations of pulmonary arteries (PAs). Currently, the etiology of these pathological features is not clearly understood and, therefore, no curative treatment is available. Since the 1990s, hydrogen sulfide (H2S) has been described as the third gasotransmitter with plethoric regulatory functions in cardiovascular tissues, especially in pulmonary circulation. Alteration in H2S biogenesis has been associated with the hallmarks of PH. H2S is also involved in pulmonary vascular cell homeostasis via the regulation of hypoxia response and mitochondrial bioenergetics, which are critical phenomena affected during the development of PH. In addition, H2S modulates ATP-sensitive K+ channel (KATP) activity, and is associated with PA relaxation. In vitro or in vivo H2S supplementation exerts antioxidative and anti-inflammatory properties, and reduces PA remodeling. Altogether, current findings suggest that H2S promotes protective effects against PH, and could be a relevant target for a new therapeutic strategy, using attractive H2S-releasing molecules. Thus, the present review discusses the involvement and dysregulation of H2S metabolism in pulmonary circulation pathophysiology.


Introduction
Pulmonary hypertension (PH) is a group of multifactorial and devastating cardiovascular conditions characterized by a progressive elevation of pulmonary arterial resistance (≥ 3 Wood) and mean pulmonary artery pressure (mPAP) above 20 mmHg [1], leading to right ventricular (RV) hypertrophy, failure, and ultimately to premature death [2,3]. PH is divided into five groups according to clinical, hemodynamic, and etiological characteristics, as well as treatment strategy: idiopathic and heritable pulmonary arterial hypertension (PAH) (group 1); PH due to left heart disease (group 2); PH due to lung diseases and/or hypoxia (group 3); PH associated with chronic thromboembolism (group 4); and finally, PH forms with unclear or multifaceted origins (group 5) [2]. PH development is associated with complex functional and structural modifications of the pulmonary arteries (PA), involving both pulmonary arterial endothelial cells (PAECs) and smooth muscle cells (PASMCs) [4]. Impairment of PA reactivity is a critical PH hallmark, and is linked to an imbalance of production and bioavailability of vasocontractile and vasorelaxant mediators [5,6]. This dysregulation is intimately linked to endothelial dysfunction [6,7], and is characterized by excessive constriction and decreased relaxation of PAs [5,8]. Concurrently, architectural changes of PA are mainly due to arterial wall remodeling, including media thickening of proximal arteries, intimal neo-formation, and fibrosis [4]. It is now well established that such changes are associated with PASMC and PAEC alterations during PH-namely, a pro-proliferative and apoptosis-resistant phenotype; an increased migration potential; extracellular matrix remodeling, mainly through excessive collagen deposition; and also endothelial-to-mesenchymal transdifferentiation [4,9]. Exacerbated inflammatory vascular responses also play a critical role in the development of PH. Actually, circulating concentrations of several proinflammatory cytokines, such as interleukin 6 (IL-6) and tumor necrosis factor alpha (TNF-α) [4,10], are abnormally increased in PAs regardless of the form of PH. This inflammatory context is associated with a worse clinical outcome in PH patients, and could contribute to exaggerated reactivity and remodeling of vascular cells. Furthermore, oxidative stress and associated enzymatic and/or mitochondrial reactive oxygen species (ROS) production are also considered to be important effectors in the development of PH, especially during PAs' responses to hypoxia [11][12][13]. Currently, therapeutic strategies developed against PH are based on the use of a treatment targeted to reduce exacerbated pulmonary vascular contractility and remodeling, along with a more general treatment to alleviate respiratory symptoms and RV failure [2]. However, these treatments fail to cure PH, and lung transplantation remains too often a requisite solution for eligible patients with severe forms of PH resistant to medical management. In the absence of suitable treatments, patients with idiopathic pulmonary arterial hypertension have a mean survival of 2.8 years [2]. As a consequence, the search for new therapeutic targets in the landscape of PH's cellular and molecular mechanisms remains essential.
Sulfur-based gases, such as hydrogen sulfide (H 2 S), have been known for a long time as environmentally poisonous and malodorous gases [14]. Aside from this harmful aspect, H 2 S is also involved in the regulation of several biological processes, and is currently considered to be a novel complementary gasotransmitter to the well-known nitric monoxide (NO) and carbon monoxide (CO). The growing importance of H 2 S in biological research is evidenced by the substantial increase in the number of original and review articles in the PubMed database dealing with this sulfur-based gas since 1996 ( Figure 1). Like NO and CO, H 2 S exhibits high solubility in aqueous solution and in lipids, suggesting an efficient ability to cross plasma membranes [15]. H 2 S is enzymatically produced under various physiological conditions, and acts as an endogenous signaling molecule that does not require a second messenger or any receptor activation [15][16][17][18]. Cystathionine-β synthase (CBS) and cystathionine-γ lyase (CSE)-two pyridoxal 5 -phosphate (PLP)-dependent enzymes-are considered to be the main sources of H 2 S in mammalian cells [19][20][21]. A third enzyme-3-mercaptopyruvate sulfur transferase (3-MST)-has also been shown to release H 2 S along with cysteine aminotransferase (CAT) activity [21][22][23][24]. H 2 S was initially described as a vasodilator, lowering blood pressure [25,26] through the ATP-sensitive potassium channel (K ATP ) openings in smooth muscle cells [27]. H 2 S also plays a critical role in numerous cardiovascular diseases, including atherosclerosis, diabetes-associated vascular disorders, and systemic hypertension [28][29][30][31][32], in which alterations of endogenous sulfur homeostasis can be observed [33]. In the context of the pulmonary circulation, several studies have already highlighted the essential role of endogenous H 2 S metabolism in both regulating physiological functions and contributing to the pathological hallmarks of PH [34][35][36][37]. CBS, CSE, and 3-MST are expressed in the intima and media of PAs [27,[38][39][40][41][42][43][44], where H 2 S is supposed to regulate oxygen sensing and associated hypoxic pulmonary vasoconstriction (HPV) [38,39,45,46]. H 2 S activates smooth muscle cell K ATP channels and, thus, contributes to PA relaxation [42], possibly in synergy with NO [47]. Interestingly, in vivo PH models are associated with dysregulated plasma H 2 S content and lung CSE expression [34,[48][49][50]. Moreover, oxidative and endoplasmic reticulum (ER) stress are attenuated by exogenous H 2 S supplementation in several PH models [51][52][53]. Such dysregulation of H 2 S metabolism in various cardiovascular diseases, including PH, leads to the consideration of new therapeutic strategies based on natural or synthetic H 2 S-releasing molecules such as GYY4137, the dithiolthione family, and garlic-derived donors [26,54,55], which have already exhibited interesting effects on pulmonary circulation pathophysiology [51,52,56,57]. Thus, the present review focuses on recent advances in the involvement of H 2 S metabolism and the dysregulation of its production in pulmonary circulation, and their consequences in the context of PH. Future prospects for the potential use of H 2 S donors as new therapeutic tools are also discussed.

H 2 S Metabolism in the Pulmonary Vasculature
H 2 S is endogenously produced under physiological conditions mainly through enzymatic pathways, but several findings suggest a role of non-enzymatic pathways in H 2 S mobilization [58][59][60][61]. In a physiological context, 28% of H 2 S is in the undissociated form, 72% is dissociated into hydrosulfide anions (HS -), and the rest is sulfide anions (S 2− ), although in negligible amounts [62]. The effects of the undissociated and dissociated forms cannot be distinguished, and are grouped in the nomenclature H 2 S. Initial reports suggested that H 2 S concentration was approximately 30 µM in lung tissue, and rose to more than 200 µM in the heart, brain, and plasma [63]. However, methodological considerations related to variability in pH, temperature, substrates concentration, etc., suggested that these concentrations were far from physiological relevance [64]. In fact, steady-state H 2 S concentration was reported to be in the nanomolar range in most tissues and in plasma [63,64]. This range is consistent with the fact that H 2 S undergoes a high turnover rate in physiological conditions, characterized by an important production associated with rapid oxidation [64]. At high concentrations ((H 2 S) > 20 µM), H 2 S triggers detrimental cellular consequences [65], characterized by mitochondrial poisoning due to inhibition of cytochrome c oxidase (complex IV) [33,66,67]. Thus, the duality of H 2 S's effects requires a fine regulation of the production/clearance balance.

Anabolic Pathways of H 2 S
In mammalian cells, H 2 S is primarily produced through the action of specific enzymes involved in homocysteine metabolism and L-cysteine catabolism: CSE, CBS, and 3-MST/CAT coupling. CSE was initially predominantly found in cardiovascular tissues, with marked expression in the heart and vessels [15,28]. CBS is mainly present in brain cells and the central nervous system although anterior reports suggest that when homocysteine levels are high, H 2 S can be produced by CBS in endothelial cells [21,68]. A more recent study demonstrated that the loss of CBS in endothelial cells decreases H 2 S production by 50% and dysregulates endothelial signaling [68]. CSE and CBS are two PLP-dependent enzymes mainly localized in the cytosol. Interestingly, several reports showed a possible translocation of CSE and CBS into mitochondria under cellular stress conditions [69,70]. In vascular smooth muscle cells, translocation of CSE from the cytosol to the mitochondria is induced by cytosolic calcium increase and, thus, promotes H 2 S production within the mitochondria, where L-cysteine concentration is three times higher than in cytosol [70]. Likewise, CBS is transferred into the mitochondria in response to hypoxia [69]. CSE and CBS have been considered for a long time to be the main endogenous H 2 S enzymatic sources in cells [19][20][21] by acting on L-cysteine, their predominant substrate. L-cysteine is a sulfhydryl amino acid derived from nutrition or endogenously generated from homocysteine via the canonical transsulfuration pathway, involving CSE and CBS (Figure 2a). Briefly, CBS converts homocysteine and L-serine into cystathionine, which is then metabolized by CSE in L-cysteine and α-ketobutyrate [71]. To generate H 2 S from L-cysteine, CSE and CBS catalyze a desulfhydration reaction and generate pyruvate and L-serine, respectively. Under physiological conditions, intracellular L-cysteine amounts exceed those of homocysteine, suggesting that L-cysteine desulfhydration is the main pathway of CSEassociated H 2 S production. However, in hyperhomocysteinemia, CSE can catalyze H 2 S release from two molecules of homocysteine [21,71], suggesting that the latter constitutes an alternative substrate for H 2 S production. It should be noted that H 2 S production by CBS is more efficient when homocysteine acts as a co-substrate of L-cysteine [71]. CBS was also shown to be allosterically activated and stabilized by S-adenosyl-L-methionine-a methyl group donor [71]. On the other hand, CO and NO inhibit CBS activity, highlighting gasotransmitter crosstalk [71]. CSE expression is upregulated by multiple factors-namely, ROS [72], ER stress [73], and calcium increase [74,75]. CSE transcription is also upregulated by the action of TNF-α during inflammation [76]. CSE activity is enhanced by calmodulin in the presence of high calcium concentrations [28], although this finding remains controversial [71]. Contrary to CBS, CSE expression and activity are stimulated by NO in vascular tissues using NO donors [27].
PLP-independent 3-MST localization in tissue was initially mentioned in brain tissue [23], and later in liver, large intestine, and kidney tissues [77]. However, 3-MST is also expressed in the endothelium and media of the thoracic aorta [22]. Cytosolic CAT1 and mitochondrial CAT2 are widely expressed in liver, heart, and kidney tissues [78]. CAT1/2 were also detected in the thoracic aorta, but only in endothelial cells [22], leading to a proposed endothelial predominance of H 2 S production by CAT/3-MST coupling. 3-MST is mainly present within the mitochondrial matrix, but is also detectable in the cytosol [66,79]. To produce H 2 S, 3-MST requires the activity of CAT, which exists in two forms: CAT1 and CAT2 localized in the cytosol and the mitochondria, respectively [21]. Indeed, CAT activity catalyzes transamination between L-cysteine and α-ketoglutarate in order to produce 3-mercatopyruvate. The latter is then used as a substrate by 3-MST to release H 2 S in presence of reducing molecules, such as endogenous thioredoxin (Trx) or dihydrolipoic acid (DHLA) [21][22][23][24]. Trx is a class of ubiquitously expressed redox proteins, including cytosolic Trx1 and mitochondrial Trx2 [24]. DHLA is a reducing compound present in the mitochondria [24]. Although 3-MST-associated H 2 S release occurs in the cytosol and the mitochondria, the steady-state contribution of mitochondrial and cytosolic 3-MST to cellular H 2 S production remains to be determined. Due to its predominant mitochondrial localization, H 2 S production by 3-MST contributes to mitochondrial metabolism through oxidation by sulfide quinone oxidoreductase (SQR), which is part of the mitochondrial electron transport chain (ETC) [80] (Figure 2b). 3-MST expression is also enhanced in human umbilical vein endothelial cells in response to shear flow [81]. In contrast, 3-MST activity can be inhibited by oxidative stress through redox modifications of its molecular structure [82].
Recent advances in endogenous H 2 S metabolism study suggest that H 2 S can be generated independently of CSE, CBS, and 3-MST/CAT coupling. Although minor, nonenzymatic H 2 S release occurs during the chemical reduction of reactive sulfur groups in thiosulfates or polysulfides ( Figure 2a). Indeed, Benavides et al. demonstrated that dietary garlic (Allium sativum)-derived organic polysulfides (diallyl disulfide and diallyl trisulfide) can release H 2 S in presence of GSH and glucose metabolism [58]. In addition to being a substrate for H 2 S-generating enzymes, L-cysteine also generates H 2 S in physiological conditions in blood, in the presence of iron and vitamin B6 [59]. As a consequence, it has been suggested that basal circulating H 2 S content is due to this novel non-enzymatic process.

Figure 2.
Main H2S anabolic and catabolic pathways in mammalian cells. (a) Endogenous hydrogen sulfide (H2S) production is mainly due to enzymatic pathways through the activity of cystathionine γ-lyase (CSE), cystathionine β-synthase (CBS), and 3-mercaptopyruvate sulfur transferase/cysteine aminotransferase (3-MST/CAT) coupling. Cytosolic CBS and CSE are involved in the interconversion of homocysteine to L-cysteine (transsulfuration pathway). In the context of cysteine catabolism, CBS and CSE desulfhydrate L-cysteine (and homocysteine) to produce H2S. L-cysteine can also act as a substrate with α-ketoglutarate to produce 3-mercaptopyruvate through a transamination reaction via activity of the CAT1 and CAT2 enzymes (cytosolic and mitochondrial, respectively). CSE activity is selectively inhibited by DL-propargylglycine (PAG) and β-cyanoalanine (BCA). CBS and CSE activities-especially the latter-are both inhibited by AOA. In a reducing environment (presence of thioredoxin (Trx) and dihydrolipoic acid (DHLA)-two endogenous reducing molecules), 3-mercaptopyruvate is then used by 3-MST to release pyruvate and H2S, mainly in the mitochondria. To a lesser extent, non-enzymatic H2S production can occur in physiological conditions, caused by reactive sulfur groups of thiosulfates (S2O3 2− ) or polysulfides (RSnS) in the presence of glutathione (GSH) or reducing equivalents (nicotinamide adenine dinucleotide (NADH) and nicotinamide adenine dinucleotide phosphate (NADPH)). (b) Oxidation, the major catabolic pathway to maintaining H2S homeostasis, occurs in the mitochondria. H2S is rapidly oxidized by sulfide quinone oxidoreductase (SQR) to form persulfides (R-SSH), which undergo another oxidation step by persulfide dioxygenase (ETHE1) to produce sulfites (SO3 2− ). In this process, two electrons (e − ) are released to ubiquinone (Q) and transferred to complex III of the mitochondrial electron transfer chain (ETC) [33,83]. Sulfites are either converted to thiosulfates or directly to sulfates production is mainly due to enzymatic pathways through the activity of cystathionine γ-lyase (CSE), cystathionine β-synthase (CBS), and 3-mercaptopyruvate sulfur transferase/cysteine aminotransferase (3-MST/CAT) coupling. Cytosolic CBS and CSE are involved in the interconversion of homocysteine to L-cysteine (transsulfuration pathway). In the context of cysteine catabolism, CBS and CSE desulfhydrate L-cysteine (and homocysteine) to produce H 2 S. L-cysteine can also act as a substrate with α-ketoglutarate to produce 3-mercaptopyruvate through a transamination reaction via activity of the CAT1 and CAT2 enzymes (cytosolic and mitochondrial, respectively). CSE activity is selectively inhibited by DL-propargylglycine (PAG) and β-cyanoalanine (BCA). CBS and CSE activities-especially the latter-are both inhibited by AOA. In a reducing environment (presence of thioredoxin (Trx) and dihydrolipoic acid (DHLA)-two endogenous reducing molecules), 3-mercaptopyruvate is then used by 3-MST to release pyruvate and H 2 S, mainly in the mitochondria. To a lesser extent, non-enzymatic H 2 S production can occur in physiological conditions, caused by reactive sulfur groups of thiosulfates (S 2 O 3 2− ) or polysulfides (RS n S) in the presence of glutathione (GSH) or reducing equivalents (nicotinamide adenine dinucleotide (NADH) and nicotinamide adenine dinucleotide phosphate (NADPH)). (b) Oxidation, the major catabolic pathway to maintaining H 2 S homeostasis, occurs in the mitochondria. H 2 S is rapidly oxidized by sulfide quinone oxidoreductase (SQR) to form persulfides (R-SSH), which undergo another oxidation step by persulfide dioxygenase (ETHE1) to produce sulfites (SO 3 2− ).
In this process, two electrons (e − ) are released to ubiquinone (Q) and transferred to complex III of the mitochondrial electron transfer chain (ETC) [33,83]. Sulfites are either converted to thiosulfates or directly to sulfates (SO 4 2− ), thanks to rhodanese (Rhod) and sulfite oxidase (SO), respectively. The final catabolic products-sulfates-are finally excreted via urine. In an additional pathway, persulfides can be degraded to thiosulfates by sulfur transferase (SR). Thiosulfates can also be converted to sulfites and H 2 S by thiosulfate reductase (TR) in the presence of GSH, which, ultimately, leads to sulfate production by SO. H 2 S methylation is the other, although minor, clearance pathway. H 2 S is first converted to methanethiol (CH 4 S) and dimethyl sulfide ((CH 3 ) 2 S) via S-methyltransferase (TMT) activity. Then, Rhod breaks down dimethyl sulfide into sulfates, which are then excreted through urine. L-cysteine is degraded to cysteine sulfinate by cysteine dioxygenase (CDO) activity. Cysteine sulfinate is then converted to sulfites by CAT1/2 and, ultimately, to sulfate by SO, or to taurine by cysteine sulfinate decarboxylase (CSAD).

Catabolic Pathways of H 2 S
Control of H 2 S metabolism requires an efficient clearance in order to avoid its detrimental effects on mitochondrial function [66]. Indeed, at high concentrations, H 2 S inhibits oxygen binding to mitochondrial complex IV, leading to ETC inhibition and the reduction of mitochondrial energetic production [33,66]. H 2 S catabolism mainly occurs in the mitochondria through oxidation pathways ( Figure 2b). From an evolutionary point of view, such mitochondrial location for H 2 S oxidation is unsurprising, owing to the primitive origin of this organelle (a sulfide-oxidizing bacterium). H 2 S clearance begins with its oxidation in the mitochondrial matrix by SQR, producing persulfide (R-SSH), and may occur in all cellular types ( Figure 2b) [71]. In physiological steady-state conditions, SQR actively participates in electron transfer to ubiquinone. However, in the oxidation process, H 2 S can donate electrons via SQR, which then follow the traditional transfer route to complexes III and IV and stimulate energetic metabolism [83]. Persulfides are then oxidized by persulfide dioxygenase (ETHE1) to generate sulfites (SO 3 2− ). Sulfites are further oxidized by sulfite oxidase (SO) to sulfates (SO 4 2− ), or by rhodanese (Rhod) to thiosulfates (S 2 O 3 2− ), which can also be reduced to sulfites and H 2 S by thiosulfate reductase (TR) in presence of GSH [84]. The terminal catabolic products of H 2 S-sulfates-are finally eliminated in urine. Additionally, persulfides generated by SQR can also be directly converted to thiosulfates by sulfur transferase (SR) [84]. In physiological conditions, H 2 S oxidation is very fast and efficient, maintaining a low level of H 2 S in tissues (i.e., half-lives of 2.0, 2.8, and 10.0 min in liver, kidney, and brain tissues, respectively) [64]. However, in hypoxic conditions, H 2 S oxidation decreases and H 2 S concentration logically increases, suggesting that H 2 S could be an efficient oxygen sensor [85].
Although considered to be minor, and much slower than oxidation, methylation of H 2 S also allows H 2 S elimination ( Figure 2b). In the cytoplasm, ubiquitous S-methyltransferase (TMT) converts H 2 S to methanethiol (CH 4 S) and dimethyl sulfide ((CH 3 ) 2 S). The latter is then oxidized by Rhod to sulfates, which are excreted via urine [15].
In addition to enzymatic routes, H 2 S can also be eliminated through expiration. Detection of exhaled H 2 S is possible when large amounts are present, such as during sodium sulfide administration [86]. H 2 S concentration in exhaled air is altered in various hypoxic diseases, such as chronic obstructive pulmonary disease (COPD) or asthma [87]. However, there is no standardized method to quantify this parameter, and very few data regarding H 2 S expiration in healthy population are available [88].
Likewise, intracellular L-cysteine concentration is tightly regulated, especially via oxidative pathways. Briefly, when it is excessive, the enzyme cysteine dioxygenase (CDO) oxidizes L-cysteine to cysteine sulfinate ( Figure 2b) [89]. The latter is further converted to sulfites thanks to the activity of cytosolic and mitochondrial CAT and SO, respectively, or to taurine by cysteine sulfinate decarboxylase (CSAD) [89].

H 2 S in the Pulmonary Circulation
Hosoki et al. first characterized CSE as the major source of H 2 S in vascular tissues [25]. CSE is abundantly expressed in bovine [38,39], broiler [40], and rat [27,[41][42][43][44] PAs. CSE expression was detected in pulmonary vessels early during post-natal lung development in mice [90]. Discrepancies in the localization of CSE expression were reported in the PA walls. Some studies mentioned that PA media and associated smooth muscle cells are the main or even exclusive sites of CSE expression in pulmonary circulation. Indeed, immunohistochemical staining showed that the presence of CSE proteins was limited to the media of rat peripheral lung vascular tissues [41], and was also found in cultured bovine PASMCs [39]. These findings are in accordance with other reports pointing out the expression of CSE in the smooth muscle cells of systemic vascular walls [22,25,27]. However, localization of CSE expression in the endothelium must be considered, because CSE was also detected in the endothelial layer of PAs and primary PAECs from rats [34,42,[91][92][93]. Those discrepancies in CSE localization within the PA walls could be explained by interspecies diversity of the studied models, as well as the pulmonary arterial segments considered (proximal versus distal) [18,45]. As a consequence, a dual role of CSE-associated H 2 S production in the media and intima of PAs should be considered.
Regarding CBS, interestingly, CBS mRNAs and proteins were also detected in rat [42,44], broiler [40], and adult mouse [91] PAs. Like CSE, CBS expression was observed in pulmonary vessels during post-natal lung development in mice [90]. CBS protein quantities in rat aortic and PA tissues were similar under physiological conditions [42]. However, unlike CSE, CBS proteins were clearly detected in the endothelial cells [94]-but not in the smooth muscle cells-of bovine PAs [39]. This endothelial predominance of CBS expression is in accordance with the first report of Zhao et al., showing the absence of CBS mRNA in endothelium-free rat PAs [27]. Although CBS mRNA and protein expression were lower than those of CSE in rat PAs [40], the CBS inhibitor aminooxyacetate (AOA) significantly altered H 2 S production [38]. In contrast, inhibition of CSE provided conflicting reports. shRNA-mediated CSE knockdown in rat PAECs significantly decreased H 2 S levels in supernatants [34], whereas DL-propargylglycine (PAG)-a presumed CSE inhibitor-did not inhibit H 2 S production of bovine PA homogenates in the presence of L-cysteine and pyridoxal-5 -phosphate. However, the selectivity of these commonly used pharmacological inhibitors to assess the functional involvement of CSE and CBS in a vascular context remains questionable [95]. Indeed, the selective CBS inhibitor AOA alters both CBS and CSE activity, and is even more efficient than PAG in reducing CSE activity. Altogether, CSE and CBS appear to be involved in endogenous H 2 S production in the pulmonary vasculature, even though the spatial distribution of H 2 S synthesis within arterial wall layers along the pulmonary vasculature remains to be clarified.
Regarding the final enzymatic source, 3-MST mRNA was observed in rat intrapulmonary arteries [43], and 3-MST proteins were detected in various human endothelial cell lines, including PAECs [18]. Immunohistochemical studies from another report indicate that 3-MST proteins are present in both endothelial and smooth muscle cells of small rat PAs [44] whereas less immunoreactivity was noticed in the media and adventitia compared to the endothelial layers of cow and sea lion PAs [39]. This broad distribution of 3-MST was also found in cultured bovine PAECs and PASMCs [39]. Interestingly, in comparison to aortic tissues, a higher 3-MST mRNA expression was quantified in PAs [42]. Cytosolic and mitochondrial CAT (CAT1 and CAT2, respectively) were notably detected in lung and systemic vascular tissues [22,44,[96][97][98]. Du et al. firstly indicated CAT1 and CAT2 mRNA presence in rat PAs [96]. Subsequently, CAT1 and CAT2 proteins were found to be expressed in cultured PAECs, PASMCs, and fibroblasts from rat PAs [34,99,100], suggesting a broad distribution within the pulmonary vascular wall. So far, few studies investigated the physiological relevance of CAT activity in H 2 S synthesis in the pulmonary vascula-ture. Madden et al. suggested an involvement of CAT activity in the generation of H 2 S through the CAT-3-MST pathway during acute hypoxic treatment of small rat PAs in the presence of cysteine and α-ketoglutarate [45]. However, CAT/3-MST coupling cannot be considered to be a direct H 2 S-producing pathway, because of the requirement of reducing molecules-such as endogenous Trx or DHLA-to produce free forms of H 2 S [22][23][24]. Cytosolic and mitochondrial Trx1 and 2 were detected in endothelial and smooth muscle cells from human PAs [101,102]. However, the link between Trx and CAT-3-MST-associated H 2 S production remains to be elucidated in pulmonary circulation.
Finally, H 2 S biogenesis indirectly depends on the level of its main precursor, L-cysteine, which, in turn, depends on the activity of CDO [78]. CDO was found to be greatly expressed and active in liver and kidney tissues, but also in lung tissues [84,97], and CDO proteins were also detected in the media of rat PAs [97].

The Role of H 2 S in Lung and Pulmonary Circulation Development
In recent decades, the role of H 2 S in the modulation of respiratory rhythm and the function of epithelial and mucociliary clearance has been demonstrated [87]. Moreover, H 2 S is also a critical actor in lung development, including pulmonary circulation. The main H 2 S-producing enzymes-CBS and CSE-can be found in airway epithelial and pulmonary vessels from the early stages of post-natal development in mice [90]. Indeed, CBS and CSE are dynamically modulated during the first 10 days of mice's lives-a pivotal period for lung alveolarization. Firstly, genetic ablation of these two enzymes significantly increases small and medium PA muscularization in neonate mice. Secondly, CBS and CSE are involved in pulmonary vascular growth, since their absence decreases CD31 (endothelial cells marker) expression in newborn mice's lung homogenates. These results were confirmed by treatments with CBS siRNA or PAG, which induced a reduction in the length of tubes formed by human PAECs. GYY4137, a slow-releasing H 2 S donor, also showed a promoting influence on vascular growth. Altogether, H 2 S and both CBS and CSE activities seem to be important for the modulation of pulmonary vessels' architecture ( Figure 3).
3-MST is also primarily expressed in the smooth muscle cells and weakly in the endothelial cells of pulmonary vessels during the lung development of young mice [103]. Nevertheless, genetic deletion of 3-MST does not trigger any structural alteration, suggesting that, unlike CBS and CSE, its presence is not required for normal lung development [103]. Overall, these findings suggest that H 2 S could be a relevant target to treat pulmonary development diseases, such as bronchopulmonary dysplasia (BPD) and associated PH (BPD-PH), which will be discussed in Section 4.2.

The Role of H 2 S in Oxygen Sensing and Hypoxic Pulmonary Vasocontriction
H 2 S and oxygen have been linked for a long time in the history of the evolution of life on Earth. Primitive bacteria initially used sulfides from hydrothermal vents as a source of energy. The rise of the atmospheric oxygen fraction forced eukaryotic organisms to shift to oxidative metabolism, with oxygen as the final acceptor of mitochondrial ETC [85]. Interestingly, CBS's molecular structure exhibits a prosthetic heme group, which can interact with oxygen molecules according to oxygen partial pressure [69]. Teng et al. demonstrated the mitochondrial accumulation of CBS induced by ischemia/reperfusion or hypoxia challenges [69]. Under ischemic/hypoxic conditions, the decrease of CBS's oxygenation status modulates its interactions, resulting in alteration of its degradation by Lon protease, and subsequently in H 2 S production in the mitochondria [69]. In the vascular sphere, hypoxia (10% O 2 ) firstly increases 3-MST protein expression, and secondly its mRNA levels, suggesting a dual regulation of 3-MST under low oxygen levels [104]. More importantly, 3-MST-associated H 2 S production was involved in the hypoxia-induced migration of human umbilical endothelial cells [104]. These elements highlight the multifaceted inter-actions between H 2 S and oxygen with a transcriptional and a putative post-transcriptional regulation of H 2 S-producing enzymes.
In contrast to the hypoxic systemic vasodilation, pulmonary circulation responds to hypoxia by a contraction-a so-called "hypoxic pulmonary vasoconstriction" (HPV)-which is a physiological response to drive the distribution of pulmonary capillary blood flow to areas of the lungs with high oxygen availability, in order to maintain correct hematosis. HPV is characterized by biphasic vasoconstriction constituted firstly by a transient contraction, followed by transient relaxation and then a sustained contraction [105]. The identity of the oxygen sensor(s) involved in this critical physiological mechanism remains controversial. Endothelium-free precapillary vessels do constrict in response to hypoxia, demonstrating that sensor(s), transductor(s) and effector(s), are present in the PA walls [106]. Nevertheless, the role of the endothelium should not be neglected. Indeed, PAECs modulate HPV under conditions of sustained hypoxia [107]. ROS produced in the mitochondria by the oxygendependent ETC may be critical for HPV, although two opposing hypotheses (decrease or increase of ROS production in PAs during hypoxia, reviewed in [107]) are under discussion, and the question about the potential role of other cellular actors in HPV remains open.
Olson et al. initially hypothesized that hypoxia decreases H 2 S oxidation by SQR in the mitochondria via the attenuation of oxygen-dependent ETC function and, logically, increases H 2 S concentration [85]. Then, the coupling of H 2 S clearance and ETC function in the mitochondria could define H 2 S as a putative oxygen sensor. In this framework, the elevation of the partial pressure of oxygen promotes H 2 S consumption (i.e., H 2 S oxidation) in bovine lung homogenates and PASMCs, with half maximal consumption at a pO 2 of 3.2 and 6 mmHg, respectively [39]. Beyond reduced H 2 S oxidation by SQR, hypoxia may inhibit ETHE1 activity, which requires oxygen, to oxidize persulfides to sulfites [85], but this hypothesis has not yet been validated in pulmonary circulation. Interestingly, H 2 S production seems to be modulated by oxygen levels since, in rat lung homogenates and small PAs, H 2 S was produced, in the presence of L-cysteine and α-ketoglutarate, under marked hypoxic conditions, but decreased when the oxygen concentration raised [45]. Progressive increase of H 2 S levels was also observed in bovine PASMCs undergoing 24 h of low oxygen, suggesting that this relationship between H 2 S and oxygen could occur for sustained hypoxic challenges [46]. The elevation of H 2 S production under hypoxia could be explained by the mitochondrial reduction status, where the concentration of endogenous reducing molecules such as DHLA increases, triggering the release of H 2 S in the mitochondria from the catabolic intermediary, thiosulfates [60]. This production is quicker than enzymatic H 2 S production from L-cysteine in response to hypoxia, and could thus be an initial event in oxygen sensing [60]. Hypoxia-induced translocation of CBS and CSE to the mitochondria [69,70] may also participate in increased H 2 S production, but no evidence has been shown at this time in pulmonary circulation. Altogether, these reports confirm the elegant theory of H 2 S and oxygen coupling, and may involve a dual regulation of H 2 S production and clearance balance in the mitochondria according to the partial pressure of oxygen ( Figure 3). The link between H 2 S and oxygen is undeniable in pulmonary circulation [39,45,46], and was indicated in other systems, such as cardiomyocytes or carotid bodies [83]. Mechanisms underlying the regulation of H 2 S metabolism by oxygen partial pressure are still unclear, and require further experimentation, especially in pulmonary circulation.
To study the relationship between H 2 S and HPV, the comparison of the acute effects of H 2 S and hypoxia (pO 2 < 5 mmHg) on the vascular responses of pulmonary vessels from various species was assessed using H 2 S-releasing sulfide salts-sodium hydrosulfide (NaHS), or sodium sulfide (Na 2 S) (1 mM)-and indicated an intriguing similarity between contractile responses to H 2 S and hypoxia, especially in rat PAs [38]. Surprisingly, H 2 S or hypoxia triggered a biphasic response with a transient contraction (Phase 1-Ph1), followed by a transient relaxation and then a sustained contraction (phase 2-Ph2) in rat PAs. These results are consistent with recent work indicating that high concentrations (>100 µM) of Na 2 S induce a biphasic contraction of rat PAs, confirming a similar pattern to that observed in HPV [108]. To clarify the role of endogenous H 2 S production in HPV, β-cyanoalanine (BCA)-a potent CSE inhibitor-reduces Ph1 contraction and subsequent relaxation of rat PAs preconstricted with norepinephrine [38]. In addition, PAG, but not AOA, decreased the rise of PA pressure in response to hypoxia in rat perfused lung tissue [45]. Knowing the importance of a reducing environment in H 2 S release by mitochondrial thiosulfates, DTT and DHLA increased the amplitude of HPV in bovine PAs [60]. Interestingly, in contrast to the above results, a report by Prieto-Lloret et al. demonstrated that HPV is not inhibited by PAG when supplemented with physiological concentrations of H 2 S precursors [43]. This is consistent with the absence of the effect of CSE genetic deletion on hypoxia-induced elevation of pressure in murine perfused lung tissue [109]. Another discordant point is that the incubation of PAs with DTT does not potentiate, but rather reduces, the amplitude of HPV [43]. These results put initial reports on the role of H 2 S in HPV into perspective, and suggest that this process is not dependent on H 2 S production in PAs (depending on either the CSE pathway or H 2 S release from thiosulfates in a reducing environment). The observed similarity between PA contraction patterns triggered by either exogenous application of H 2 S or hypoxia does not necessarily entail a physiological relationship (direct or indirect), although it could be explained by shared common mechanisms. Mitochondrial ROS production is a key event in HPV, and activates calcium release from the sarcoplasmic reticulum, and subsequent contraction of PAs [107,110]. In a recent study, Prieto-Lloret et al. assessed the mechanistic issue of biphasic H 2 S-induced contraction [108]. Ph1 and Ph2 contraction and relaxation caused by H 2 S were not affected by L-NAME or endothelium denudation, suggesting that H 2 S (30-1000 µM) acts directly on PASMCs. H 2 S treatment of PASMCs increased ROS production, whereas SQR genetic deletion abolished this effect. The sustained Ph2 contraction was inhibited by myxothiazol (a complex III inhibitor), but not by rotenone (a complex I inhibitor). This report thus suggests that H 2 S stimulates the ETC through H 2 S oxidation by SQR, in turn stimulating ROS production from complex III and, thus, triggering the sustained Ph2 contraction. In contrast, this Ph2 sustained contraction induced by hypoxia during HPV is inhibited by rotenone [110], thus suggesting a different mechanism from that triggered by H 2 S. One hypothesis is that, when the intracellular H 2 S level increases, its poisoning effect on complex IV inhibits the ETC and decreases ROS production, thus inducing a transient relaxation. H 2 S evaporation decreases its level and again promotes the ETC via SQR oxidation, increasing ROS production by complex III (sustained Ph2 contraction). Overall, H 2 S's influence on PA contraction may be due to a balance between its promoting and blocking effects on ETC function ( Figure 3).
Initial reports indicated opposite dose-dependent effects of H 2 S on preconstricted bovine PAs [38]. Between 10 nM and 10 µM, H 2 S relaxes, whereas above 10 µM of H 2 S induces a constriction of bovine PAs. However, it is noteworthy that in most tissues and in plasma steady-state H 2 S levels do not exceed the nanomolar range [63,64]. In this range, H 2 S preferentially relaxes PAs. A putative increase in intracellular H 2 S levels from nanomolar range to 10 µM in response to hypoxia would induce detrimental consequences on the mitochondria. Numerous considerations about variability in the experimental approaches (H 2 S-releasing molecules, concentration, agonist-induced pre-tone of the vessel, oxygen levels, animal species, etc.) must be considered in order to explain opposing reports, and the development of reliable methods to control and measure intracellular H 2 S levels are required. As previously mentioned, the use of pharmacological inhibitors of H2S enzymatic production also represents a significant issue, due to their relative specificity and the detrimental absence of selective CBS inhibitors [95].

The Role of H 2 S in Pulmonary Artery Relaxation
Numerous studies associate H 2 S with the regulation of vascular tone. The relaxing effects of exogenous H 2 S in portal veins and thoracic aortae of rats [25,27] were initially reported, and were associated with decreased arterial blood pressure [27,28,109]. For instance, CSE genetic deletion in mice reduces endothelium-dependent vasorelaxation, suggesting an important role of endogenous H 2 S production in this process [28].
Initial reports of the effect of exogenous H 2 S on PAs' vascular tone indicated that, at physiological levels, H 2 S relaxes bovine PAs in a dose-dependent manner, with a half maximal effective concentration of 550 ± 180 nM [38] (Figure 3). High concentrations of H 2 S (300 µM) also relax bovine PAs preconstricted by severe hypoxia [38]. NaHS induces a dose-dependent relaxation in rat PAs [42,47]. Exogenous H 2 S (20-500 µM)'s relaxing effects were also observed in isolated large-and medium-sized PAs from human donors, and were associated with a decrease in PA pressure in human perfused lung tissues [111]. However, it should be noted that the genetic deletion of CSE in mice does not impact steady-state mPAP in isolated perfused lung tissues [109], suggesting that endogenous H 2 S production by CSE is not involved in the regulation of basal mPAP, at least in mice. Derived organosulfur compounds from garlic, such as allicin, diallyl disulfide, and diallyl trisulfide, have for a long time been associated with H 2 S release under physiological conditions, triggering vascular relaxation in systemic vessels and decreasing blood pressure [58,112]. In pulmonary vascular beds, allicin-but not diallyl disulfide or trisulfide-dose-dependently relaxed rat PAs and reduced PA pressure in rat perfused lung tissues [113,114]. Interestingly, pre-treatment with garlic extracts significantly decreased endothelin-1 contraction of rat PAs [115]. However, the contribution of H 2 S release from these garlic-derived compounds to PA relaxation remains to be elucidated. Cellular mechanisms underlying the relaxing effect of H 2 S on PAs were primarily associated with K ATP channel activity. Indeed, Zhao et al. demonstrated that H 2 S promotes opening of K ATP channels, leading to smooth muscle cell hyperpolarization and, consequently, aortic relaxation [27]. This result was then reproduced in pulmonary circulation, where glibenclamide (a blocker of plasma membrane K ATP )-but not 5-hydroxydecanoate (blocker of mitochondrial K ATP )-partially inhibited H 2 S-induced PA relaxation in rats [42]. H 2 S has been shown to directly interact with the Kir6.1 regulating subunit of K ATP channels through S-sulfhydration (post-translational modification consisting in the formation of a persulfide group (R-SSH) on a cysteine residue) of cysteine residue (cys43) [116]. This modification of the K ATP regulating subunit alters ATP binding and, consequently, K ATP activity is enhanced, promoting hyperpolarization and vascular relaxation [116]. Although Kir6.1 was found in rat PAECs and PASMCs [42], no direct evidence links Kir6.1 S-sulfhydration and H 2 S-induced PA relaxation. In aortic tissue, H 2 S inhibits mitochondrial metabolism, thus resulting in decreased ATP production [117]. Alteration of ATP levels promoting K ATP activity could be another mechanism in H 2 S-induced PA relaxation. H 2 S has long been known to interact with other gasotransmitters, such as NO, through chemical or metabolic interplays [118]. For instance, H 2 S-induced relaxation of rat PAs is altered by Nω-nitro-L-arginine methyl ester (L-NAME), a potent NO synthase inhibitor [47]. PA relaxation triggered by sodium nitroprusside-an NO donor-is also partially reduced by PAG. Allicin-mediated PA relaxation is also dependent on the NO pathway [115]. Endogenous NO production thus seems to be involved in H 2 S relaxation, and vice versa, establishing a possible metabolic crosstalk for PA tone regulation (Figure 3). Although precise mechanism(s) remain to be determined in the pulmonary vascular bed, these results are in accordance with numerous reports showing H 2 S and NO pathway interactions in the modulation of systemic vascular tone [25,28,119,120]. Owing to the significant relaxing influence of H 2 S, many studies have shown a protective effect of H 2 S on increased mPAP during PH development, which will be discussed in the following section.  (1) promoting vascular growth and associated PAEC migration and (2) decreasing media muscularization. Variation in the partial pressure of oxygen is linked to regulation of H 2 S metabolism. Hypoxia exposure triggers a decrease in H 2 S clearance and an increase in H 2 S production, leading to a global elevation of intracellular H 2 S levels. Modulation of the H 2 S clearance/production balance may play a pivotal role in oxygen sensing in pulmonary circulation. PA treatment with H 2 S donors is associated with paradoxical influence on vascular tone. On the one hand, H 2 S induces biphasic contraction of PAs that could be explained by the balance of the promotion (oxidation by SQR) versus poisoning (blocking of complex IV) effects of H 2 S on mitochondrial ETC function. In other hand, H 2 S also dose-dependently relaxes PAs through the activation of K ATP channels, leading to vascular cells' hyperpolarization and, thus, to relaxation. Crosstalk between H 2 S and endothelial NO pathways was also observed, suggesting a potential role of the endothelium in the relaxing effects of H 2 S. 3-MST: 3-mercaptopyruvate sulfur transferase; CSE: cystathionine γ-lyase; GYY4137: morpholin-4-ium 4-methoxyphenyl (morpholino) phosphinodithioate; H 2 S: hydrogen sulfide; K ATP : ATP-sensitive K + channel; Na 2 S: sodium sulfide; NaHS: sodium hydrosulfide; NO: nitric oxide; PA: pulmonary artery; PAECs: pulmonary artery endothelial cells; PASMCs: pulmonary artery smooth muscle cells.

The Role of H 2 S in Hallmarks of PH
Alterations in H 2 S production or clearance lead to a pathological drift contributing to vascular injury, notably through endothelial dysfunction, inflammation, oxidative stress, decreased vasorelaxation, vascular remodeling, and platelet aggregation (already reviewed in [29,32,121,122]). Depending on the considered pathological condition, circulating H 2 S levels are either decreased or increased in patients [123][124][125]. High levels of homocysteine that promote endothelial dysfunction are also observed in various clinical cardiovascular conditions [122]. Therefore, the global vasculoprotective role of exogenous H 2 S supplementation-especially through its anti-inflammatory, antioxidative, and vasorelaxant properties-argues in favor of H 2 S acting as a new therapeutic strategy for vascular diseases. In view of the implications of H 2 S in pulmonary circulation, we will discuss H 2 S metabolism alterations in the setup of PH, as well as the potential protective role of H 2 S in the development of associated pathological hallmarks.

H 2 S Metabolism in Human PH
According to PH classification (5 groups), PA pathological changes can be associated with a multitude of clinical conditions [2]. Congenital heart diseases (CHDs), such as Eisenmenger syndrome or systemic-to-pulmonary shunts, can promote PH development due to increased blood flow within pulmonary circulation (PH group 1) [2]. In children with CHD-associated PH, circulating H 2 S levels and CSE expression are lower than in children without PH [126]. In contrast, homocysteine levels are increased, but the low H 2 S levels may be due to the reduced CSE expression, thus inducing a decrease in homocysteine conversion into L-cysteine via the transsulfuration pathway [126,127]. In addition, low endogenous H 2 S was correlated with worse prognosis after surgical correction of CHD [128]. These results suggest that H 2 S and actors of its metabolism (namely homocysteine and CSE) could be potential biomarkers to determine the short-term prognosis and risk of CHD complicated with PH.
Chronic hypoxia exposure triggers a cascade of pathological manifestations inducing PH, characterized by a sustained PA contraction and decreased relaxation (endothelial dysfunction), PA remodeling, and inflammation, finally leading to right ventricle hypertrophy [4]. Indeed, lung hypoxic chronic diseases such as COPD, sleep apnea, or fibrosis/emphysema disorders are associated with PH development (PH group 3) [2,4]. PH is commonly observed in severe COPD, seen in more than 90% of patients with mPAP > 20 mmHg [129]. COPD is mainly characterized by a marked inflammation of the airway tract, lung parenchyma, and pulmonary vasculature, contributing to exacerbations and leading to possible pathological drift towards PH. An initial report mentioned that H 2 S serum levels in patients with acute COPD exacerbation are lower than those with stable COPD [130]. In addition, H 2 S levels are progressively reduced with the decrease of airway obstruction severity, and when systolic PAP is greater than or equal to 35 mmHg in COPD patients with acute exacerbation [130]. In accordance with these results, a complementary study indicated that CSE protein and CBS mRNA expression were significantly decreased in COPD patients [131]. A reduction in exhaled H 2 S levels was also found in COPD patients with significant quantities of eosinophils, suggesting a potential link with the modulation of inflammation [132].
BPD is a common lung developmental complication in premature newborns, characterized by an arrest of alveolarization and a decrease in angiogenesis due to the requirement of mechanical ventilation and hyperoxic treatment [133]. Those developmental alterations can lead to decreased pulmonary vascular density and promote PA muscularization and inflammation [134,135]. As a consequence, increased PA resistance and pressure can be observed, and 25-40% of premature infants with severe BPD will develop PH (BPD-PH). In case of BPD-PH, the death rate after 2 years is between 33 and 48% [136]. Although there are multiple pathophysiological mechanisms, dysregulation of H 2 S metabolism was demonstrated in the context of BPD. Indeed, cystathionine plasma levels are higher in premature than in full-term newborns. In contrast, L-cysteine levels are significantly lower in premature infants [137]. Moreover, hepatic CSE activity is lower in premature than in mature infants [138]. This result brings evidence of a possible alteration of the transsulfuration pathway in pre-term newborns. However, the link between dysfunction of H 2 S metabolism and the risk of developing BPD-associated PH is not yet elucidated.
Altogether, these results highlight an alteration of the endogenous H 2 S metabolism in CHD, COPD, and BPD, and could constitute an interesting biomarker of severity, prognosis, and risk to develop increased PAP.

H 2 S Metabolism in Experimental PH Models
To understand the complex pathophysiology of PH, various in vivo experimental models were developed to mimic clinical PH representations. Each experimental model has its own pathophysiology and etiology, although all exhibit, with different intensity, the primary markers of PH-namely, pronounced PA remodeling, increased mPAP, inflammation, and RV hypertrophy [139]. The monocrotaline (MCT) animal model consisted of one injection of MCT-a pyrrolizidine alkaloid extracted from the plant Crotalaria spectabilismetabolized into its active form (MCT pyrrole) in the liver by monooxygenase. MCT triggers pulmonary vascular injury, mainly via endothelial dysfunction and exacerbated inflammation, leading to increased mPAP and both PA and RV remodeling [140]. Interest-ingly, MCT-treated rats exhibit lower H 2 S levels in plasma and in lung tissue [34,141]. CSE protein expression and activity in lung tissue are also decreased in comparison to control rats [50,141]. These alterations are time-dependently downregulated in this PH model, with significant reductions in plasma H 2 S levels, lung CSE expression, and activity from 14 days after MCT injection [50]. A treatment of human PAECs with the MCT pyrrole decreases H 2 S production and CSE protein expression [141], suggesting a possible alteration of the endogenous H 2 S metabolism in PA endothelium.
In order to model secondary PH due to chronic hypoxic diseases such as COPD, animals are subjected to hypoxia housing in normobaric or hypobaric chambers. Similarly to the MCT PH model, daily exposure to normobaric or hypobaric hypoxia (10% O 2 ) for 3 weeks was associated with a significant decrease of H 2 S plasmatic concentration and lung H 2 S production [35,40,53,142,143]. Moreover, expression of CSE mRNA was altered in lung tissues [35,40]. Interestingly, López et al. compared the effects of altitude exposition (hypobaric hypoxia) on newborn sheep versus newborn llamas, which are adapted to chronic exposition to hypoxia at high altitudes [144]. First of all, altitude exposition increased homocysteine plasma levels in sheep, but not in llamas, which presented basal low levels of homocysteine regardless of altitude. In addition, increased PA pressure and resistance were observed in sheep, but not in llamas, in response to altitude. Thus, basal low levels of homocysteine in llamas could be explained by their catabolism by CSE and CBS via the transsulfuration pathway, and may prevent the effects of hypoxic exposure and associated PH. On the other hand, disruption of this route by an undefined hypoxia-dependent mechanism could trigger homocysteine accumulation and attenuate H 2 S bioavailability, promoting vascular injury and PH setup. PH secondary to COPD and emphysema has also been studied in mice subjected to tobacco smoke [94]. After 12 or 24 weeks of tobacco smoke exposure, H 2 S concentration was unchanged, whereas the capacity to produce H 2 S was decreased in murine lungs [94]. Such results could be explained by the significant reduction of CSE and CBS protein expression [94].
Beyond commonly used PH models, experimental PH can also be provoked by creating a shunt between the abdominal aorta and the inferior vena cava in order to increase pulmonary blood flow. After 11 weeks, rats with high pulmonary blood flow exhibited reduced plasma and lung H 2 S levels, as well as lung H 2 S production rates [92,145]. CSE, CBS, and 3-MST protein expression were also reduced in the PAs of these rats [44]. Surprisingly, four weeks after shunt, H 2 S concentration in the lung tissue was higher than in control rats. This elevation was suppressed with daily treatment with PAG (a CSE inhibitor), suggesting an increase in H 2 S production caused by CSE [93,145]. Indeed, we can hypothesize that the increase in H 2 S content in the lung tissue after four weeks could act as a compensatory mechanism to counteract the high PA pressures, perhaps via the relaxant properties of H 2 S on PAs. Prolongation of high PAP could induce a decompensation state, with a decrease in H 2 S production through the alteration of CSE, CBS, and 3-MST expression. Such hypotheses imply a cellular relationship between blood pressure detection and H 2 S metabolism. Already mentioned in recent reports on systemic vessels [81,146], this association needs to be demonstrated in pulmonary circulation.
Overall, experimental models of PH tend to reproduce clinical observations of dysregulated H 2 S metabolism in PH, whatever the PH group considered. Circulating and pulmonary H 2 S contents are decreased in most reports, and associated with altered expression and/or activity of H 2 S-generating enzymes. Dysregulation of CSE seems to be a pivotal common factor, since it is associated with various forms of PH in human and experimental models alike. Considering the major role of CSE in vascular tissues, and its involvement in endogenous H 2 S synthesis in PAs (see Section 2.2), reduction of its expression and/or activity in the lungs could be associated with high homocysteine levels and decrease of H 2 S vascular bioavailability. Finally, under pathological conditions, H 2 S metabolism alterations could, at least in part, result in PA injury through the development of endothelial dysfunction, remodeling, and decreased vasorelaxation. Altogether, current clinical and experimental research converges to a critical protective role of H 2 S metabolism in the pathophysiology of PH, which will be discussed in next section.

H 2 S Exerts Protective Effects against PH
Research into the therapeutic potential of H 2 S against PH is based on H 2 S supplementation experiments using H 2 S-releasing molecules. The most commonly used are the sulfide salts NaHS and Na 2 S, which are inexpensive, water-soluble, and quickly release large amounts of H 2 S under physiological conditions [54,55]. Slow-releasing H 2 S donors, such as GYY4137 (morpholin-4-ium 4-methoxyphenyl (morpholino) phosphinodithioate), were further developed, and already exhibit interesting effects on cardiovascular diseases [26,54,55]. The beneficial effects of H 2 S-releasing molecules' administration on experimental PH development are summarized in Table 1.
In the MCT model, preventive daily injection of NaHS (56 µmol/kg, intraperitoneal) for 21 days significantly reversed the reduced H 2 S levels in plasma and in lung tissues, and decreased mPAP, RV hypertrophy, and PA remodeling via the reduction of media thickness [34,141]. This protective effect was also reported with a curative treatment with NaHS (1 mg/kg, intraperitoneal) 7 days after MCT injection [50]. More surprisingly, decreased CSE protein expression was also reversed in rat lung tissue, suggesting a feedback influence of H 2 S on its production routes [141]. In the same manner, low CSE protein expression observed in human PAECs treated with MCT pyrrole in vitro was reversed by the addition of H 2 S [141]. The protective influence of NaHS was associated with an inhibition of PH-associated inflammation, as shown by reduction of the plasmatic and pulmonary contents of the proinflammatory cytokines TNF-α, IL-6, and IL-8 [34,141]. H 2 S supplementation attenuated the activation of the nuclear factor kappa B (NF-κB) pathwaya pivotal signaling pathway in inflammation in PH [147,148]-in lung tissue from rats with MCT-associated PH [141] and in PAECs treated with MCT pyrrole [37,141]. Moreover, NaHS treatment reduces α-smooth muscle actin and increases VE-cadherin expression in the PAs of rats injected with MCT [50]. This effect was aggravated when rats were treated with PAG. NaHS also dose-dependently inhibited the in vitro phenotypic shift of human PAECs into mesenchymal cells induced by TGF-β1 treatment [50]. This process was mimicked by CSE overexpression. Altogether, these results suggest that H 2 S inhibits the endothelial-mesenchymal transition and associated remodeling observed in PH. In contrast to NaHS, dithiolthione (ADT-OH) was demonstrated to slowly release H 2 S in vivo [149]. Interestingly, 7 days after MCT injection, daily inhalation of ACS14 (a conjunction of ADT-OH and aspirin) encapsulated in a large porous microsphere decreased mPAP, PA remodeling, and RV hypertrophy, similarly to sildenafil, which is known to decrease such PH hallmarks [56]. Like NaHS, ACS14 treatment also reduced endothelial-mesenchymal transition in the PA walls of rats injected with MCT.
In the hypoxia model, daily administration of NaHS (14 µmol/kg) also demonstrated promising properties in PH. NaHS markedly inhibited mPAP in rats and broilers with hypoxic PH [35,40,53,142,143]. PA remodeling was also improved, with reduction of media thickness, the number of muscularized PAs, and the presence of collagen types I/III and elastin in PAs [35,143]. In addition, NaHS increased the total antioxidant capacity of lung homogenates, highlighting an improvement in cellular defenses against hypoxiainduced oxidative stress [53]. As in MCT models, NaHS treatment succeeded in reversing altered plasma and pulmonary H 2 S levels [35,40,53,142,143]. In the murine tobacco smokeinduced PH model, daily treatment with NaHS (50 µmol/kg) decreased TNF-α amounts in bronchial alveolar lavage-and 8-hydroxyguanine (a marker of DNA injury induced by ROS)-positive cells, suggesting an attenuation of lung inflammation and oxidative stress. Moreover, NaHS treatment restored murine lung expression of CSE and CBS proteins due to Akt protein activation [94]. Interestingly, these results were associated with a reduction in RV systolic pressure [94]. In another model of rats with COPD obtained via smoke exposure and lipopolysaccharide tracheal instillation, Ding et al. showed the beneficial effects of NaHS (56 µmol/kg) on PAEC apoptosis and associated endothelial injury [150]. Further-more, a recent work analyzed the therapeutic potential of a preventive treatment with a slow-releasing H 2 S donor-GYY4137-on hypoxic PH [51]. Partial reduction of mPAP and total PA resistance (mPAP/cardiac output ratio) was reported in rats subjected to 4 weeks of hypoxia and treated with GYY4137. PA media thickness was also decreased, which is consistent with the inhibition by GYY4137 (100 µM) of hypoxia-induced PASMC proliferation and migration, without inducing apoptosis [51]. ER stress was recently considered to be a pivotal manifestation during PH setup, and notably characterized by stimulation of the activating transcription factor 6 (ATF6) and disruption of ER-mitochondria interactions [151]. GYY4137 inhibited the expression of ER-stress-associated proteins binding the immunoglobulin protein (Grp78) and ATF6 in PAs [51]. However, as already demonstrated in HUVEC [152], the influence of GYY4137 on the ER-mitochondria unit remains to be defined in PA walls. It is important to note that, in addition to vascular benefits, NaHS and GYY4137 also demonstrated protective effects on cardiac function and remodeling, both in hypoxic PH and in PH secondary to COPD [35,51,53,94]. Beyond NaHS and GYY4137, garlic was studied for its short-term hemodynamic properties on pulmonary circulation. Indeed, daily treatment of rats with garlic by gavage (100 mg/kg) for 5 days significantly reduced the increase in mPAP induced by 90 min of hypoxia housing (10% O 2 ), without modifying systemic arterial pressure [57]. This effect was related to the relaxant influence of garlic on PAs (see Section 3.3).
An experimental model of BPD-PH was induced by exposing mouse or rat pups to hyperoxia during their first days of life. Such conditions promote lung inflammation, alveolar/vascular growth inhibition, increase in PA wall thickness and, consequently, PH [52,153]. Vadivel et al. assessed the benefits of daily administration with GYY4137 on hyperoxia (95% O 2 )-induced BPD-PH in rat pups [52]. This treatment reduced PASMC proliferation, PA media remodeling, and RV afterload, and partially attenuated RV hypertrophy, suggesting that GYY4137 could prevent vascular and cardiac adverse manifestations in BPD-PH. This beneficial effect was associated with an improvement of alveolar growth and pulmonary vessel density in lungs. Vascular growth alteration under hyperoxia is primarily due to a reduced networking of PAECs. Interestingly, GYY4137 treatment improved human PAEC network formation in normoxia or hyperoxia [52,90]. In relation to these results, GYY4137 also improved cell viability and reduced oxidative stress induced by hyperoxia in human PAECs, suggesting that such a compound could have a beneficial effect on PA remodeling [52]. The protective effects of H 2 S against BPD-PH are in total accordance with H 2 S's significant role in both lung and pulmonary circulation in post-natal development (see Section 3.1).
Finally, in the model of high pulmonary blood flow PH, H 2 S donors also revealed interesting properties. Indeed, daily NaHS (56 µmol/kg) administration during the 11 weeks of shunting partially attenuated increased mPAP and RV hypertrophy [48,143,145]. Improvement of PA structural alteration was observed with a reduction of PA remodeling and collagen type I/III staining in the PA walls [48]. In PA media from rats treated with NaHS, an increase in apoptosis markers-such as Fas and caspase-3-was observed [145]. It must be noted that NaHS also reversed the increased lung endothelin-1 levels-a potent vasoactive and pro-proliferative agent [48].
Beyond vascular manifestations, PH is also characterized by critical structural alterations (e.g., inflammation, fibrosis, remodeling, etc.) of the RV, resulting in RV failure and death [3]. H 2 S administration in PH experimental models showed beneficial effects on RV hypertrophy (see Section 4.2). Nevertheless, mechanisms underlying this influence are still undetermined; two options may merit further examination-namely, a direct effect on the myocardium, and/or an indirect process via attenuation of the RV afterload. Numerous studies reported the relationship between decreased plasma H 2 S levels and myocardial infarction or heart failure, suggesting an intrinsic role of H 2 S metabolism in cardiac homeostasis [154]. In both experimental and clinical cases, H 2 S administration using SG-1002-a slow-releasing H 2 S donor-exerted cardioprotective effects on heart failure through a proposed increase of both circulating NO bioavailability and the eNOS pathway [155,156].
In addition, liposomal ZYZ-802-another slow-releasing H 2 S molecule-reduced collagen fiber amounts and associated fibrosis in the myocardium in a heart failure model of rats [157].
At the mitochondrial level, PH is characterized by a cancer-like metabolic shift (Warburg effect) from oxidative phosphorylation to glycolysis in both PA and RV cells. Mitochondrial fission through increased dynamin-related protein 1 (DRP1) and decreased mitofusin 2 (MFN2) has also been demonstrated [158]. Interestingly, H 2 S oxidation by SQR stimulates oxidative phosphorylation, presenting the mitochondria as a privileged target of H 2 S [65,80]. H 2 S can also interact with the expression and function of proteins regulating mitochondrial dynamics. In fact, H 2 S decreases DRP1 and enhances MFN2 expression in the myocardium, and subsequently improves mitochondrial ultrastructure and function [159,160]. These beneficial effects were associated with an attenuation of myocardial hypertrophy in mice [159,160]. Moreover, CBS knockdown or inhibition using AOA was associated with a reduction of MFN2 expression in ovarian cancer cells [161]. In relation to its role in oxygen sensing and hypoxia responses in PAs (see Section 3.2), H 2 S could thus be of interest to reduce or reverse mitochondrial alterations associated with PH.
In summary, H 2 S-releasing molecules show significant protective effects on vascular and cardiac manifestations of various forms of PH in experimental PH models. H 2 S's beneficial effects on PH are multifaceted on various hallmarks, such as inflammation, PA remodeling (endothelial-mesenchymal transition; migration and proliferation of cells from the PA walls), and oxidative and ER stress ( Figure 4). H 2 S-based preventive and curative treatments improved endogenous H 2 S production, especially by CSE, which appeared to be critical in the regulation of PH hallmarks. To date, the lack of studies using GYY4137 or other slow-releasing H 2 S donors on PH does not allow for comparison of their effects with those of sulfide salts, such as NaHS. However, slow-releasing H 2 S donors seem to better control H 2 S rates, thus avoiding high H 2 S levels, which could be deleterious to mitochondria [54,55]. Although NaHS decreases lung ET-1 levels [48], the effect of preventive H 2 S donor supplementation on the decreased PA relaxation and hyper-reactivity observed in PH remains to be assessed. Table 1. Summary of reported effects of the use of H 2 S-releasing molecules on various experimental PH models. α-SMA: αsmooth muscle cell actin; ATF6: activating transcription factor 6; BAL: bronchial alveolar lavage; CO: carbon monoxide; ET-1: endothelin-1; ER: endoplasmic reticulum; GYY4137: morpholin-4-ium 4-methoxyphenyl (morpholino) phosphinodithioate; Grp78: binding immunoglobulin protein; HO-1: heme oxygenase-1; ICAM-1: intercellular adhesion molecule-1; mPAP: mean pulmonary artery pressure; NaHS,: sodium hydrosulfide; PA: pulmonary artery; PAECs: pulmonary artery endothelial cells; PASMCs: pulmonary artery smooth muscle cells; RV: right ventricle; TNF-α: tumor necrosis factor-α; IL-6/8: interleukin 6/8; VE-cadherin: vascular endothelial-cadherin.

H 2 S Significance and Perspectives in PH
Like NO and CO, the effect of H 2 S in the regulation of vascular homeostasis and fundamental functions is undeniable. H 2 S production takes part in pulmonary circulation development and oxygen sensing, and acts as a potent vasodilator. Alterations of endogenous H 2 S metabolism are observed in various clinical cardiovascular conditions [123][124][125], including PH and associated diseases such as CHD, COPD, and BPD (see Section 4.1.1). Although the cellular mechanisms of these dysregulations remain unclear, decreased H 2 S bioavailability is critical in the promotion of endothelial dysfunction, exacerbated inflammation, oxidative stress, and changes in the proliferative behavior of smooth muscle cells in the PA walls. In future clinical investigations, H 2 S and associated metabolism actors (namely, L-cysteine, homocysteine, CSE, CBS, and 3-MST) could thus be considered to be relevant biomarkers to characterize prognosis and risk of developing PH from CHD, COPD, or BPD. Enzymatic pathway production is considered to be the major route of H 2 S production in PAs. However, non-enzymatic H 2 S production by red blood cells from L-cysteine and iron has been recently evidenced as a new source of circulating H 2 S under physiological conditions [59]. Interestingly, patients with severe PH exhibit iron deficiency [162]. Hypothetically, reduced iron levels could lead to a reduction of nonenzymatic H 2 S production by red blood cells and, consequently, decreased circulating H 2 S bioavailability in vascular tissues. Although interesting, this hypothesis needs to be experimentally proven in the context of PH.

H 2 S-Releasing Molecules as a New Therapeutic Strategy for PH?
As discussed above, endogenous H 2 S alterations in PH patients and the beneficial impacts of H 2 S on the experimental pathophysiology of PH raise the issue of its clinical potential interest for PH. In the past few years, multiple pharmacological tools have been developed in order to manipulate H 2 S under physiological conditions. Sulfide salts were shown to release H 2 S in large amounts, allowing for quick and efficient distribution. In healthy humans, Toombs et al. showed that intravascular administration of Na 2 S induces an increase in the blood concentration of H 2 S and thiosulfates, as well as exhaled H 2 S levels, during the first minutes after injection [86]. However, this H 2 S bolus released by sulfide salts could activate the detrimental effects of H 2 S on mitochondrial function if high concentrations are reached, especially with susceptible individuals, as in patients with PH. Slow-releasing H 2 S donors, such as GYY4137 or dithiolthione compounds, thus represent an attractive alternative, since they exhibit interesting effects with prolonged liberation in vivo [55,149]. This characteristic avoids the bolus effect of sulfide salts, and allows a controlled release and a greater bioavailability of H 2 S to mimic endogenous H 2 S production rates and maintain a relevant concentration for a long period of time. Interestingly, numerous H 2 S donors are currently considered to be safe for a clinical application. Natural H 2 S donors, such as garlic extracts, have already demonstrated significant influence in lowering the blood pressure of hypertensive patients in multiple clinical trials [163]. Although efficient and highly safe, garlic supplementation should be considered as more of a co-treatment than a unique strategy, because of the lack of hindsight on long-term cardiovascular efficiency [163]. Anethole trithione (CAS number 532-11-6)-a dithiolthione compound-has been widely used and marketed for decades to treat salivary deficiency, cholecystitis, and hepatitis, with no known major side effects [164]. The structural versatility of dithiolthione compounds offers a multitude of plausible strategies to conjugate their H 2 S-releasing properties with other active molecules to improve PH hallmarks [54]. In this framework, dithiolthione was conjugated with sildenafil (ACS-6, http://www.ctgpharma.com, accessed on 10 June 2021) to associate the vasculoprotective features of H 2 S with an established treatment of PH (sildenafil) [165]. As sildenafil, ACS-6 inhibits phosphodiesterase type 5 (PDE5) activity and TNF-α-induced superoxide formation in PAECs in vitro [165]. H 2 S release by ACS-6 is more sustained than that by NaHS. Nevertheless, the preclinical potential of this molecule on experimental PH models has not yet been assessed. Other H 2 S-liberating molecules have been, or are being, studied in various cardiovascular clinical contexts, and subjected to patenting [166]. For instance, the aforementioned SG-1002 (clinicaltrials.gov; ID: NCT01989208) has been clinically tested to counteract H 2 S deficiency in heart failure patients. This compound is well tolerated at various doses, and no changes on hemodynamic parameters or clinical chemistry have been found in healthy or heart failure patients [155].
In summary, H 2 S-releasing molecules have already shown promising features to attenuate vascular and heart alterations in PH. Nevertheless, H 2 S donor research is emerging, and clinical data in the context of PH are still lacking. It will be pertinent to compare the efficiency of various H 2 S-releasing molecules on the development of PH hallmarks in experimental models. Moreover, it will be relevant to decipher the specific actions and mechanisms of such treatment on RV functional and structural alterations during PH development. Innovative synthetic as well as natural H 2 S-releasing molecules, with their broad spectrum of action on vascular and cardiac manifestations, thus constitute a relevant opportunity to develop new therapeutic strategies for PH, and further preclinical investigations on animal models are required prior to any clinical implementation.

Data Availability Statement:
No new data were created or analyzed in this study. Data sharing is not applicable to this article.