NADPH Oxidase 4 Is Expressed in Pulmonary Artery Adventitia and Contributes to Hypertensive Vascular Remodeling

Materials and Methods are available in the online-only Supplement.

Real-time reverse transcription polymerase chain reaction was used to determine relative expression levels of Nox enzymes and associated subunits in isolated PAs. In all 3 rat models of PH (fawn hooded rat [FHR], monocrotaline, and Sugen-hypoxia), Nox4 mRNA was significantly increased compared with the normotensive Sprague-Dawley rats (Figure 1A–1C). In addition, Western blot analysis revealed that Nox4 protein expression was significantly upregulated in PA from monocrotaline-treated rats (Figure 1D and 1E). Nox1 and Nox2 gene expressions were also increased in PA from monocrotaline-treated rats with PH (Figures IA and IIA in the online-only Data Supplement). However, Nox1 mRNA expression was not increased in PA from the FHR or the Sugen-hypoxia rat or in hypoxic mouse lung (Figure IB–ID in the online-only Data Supplement). Screening of p22phox, p47phox, and p67phox subunits revealed no significant differences in mRNA expression between the normotensive (Sprague-Dawley rats) and hypertensive (monocrotaline) PA (Figure IIC–IIE in the online-only Data Supplement). However, there was a significant downregulation of p47phox and p67phox and NOXA1 and NOXO1 in Sugen-hypoxia PA (Figure IID–IIG in the online-only Data Supplement). There was no change in the expression of p22phox, p47phox, p67phox, NOXO1, and NOXA1 between control rats and FHR (Figure IIH–IIL in the online-only Data Supplement).

To address the functional contribution of elevated Nox4 expression in the development of PH, we treated control and monocrotaline rats with 3 different Nox4 inhibitors. To confirm efficacy, we measured ROS in a cell type expressing Nox4. Inhibitor VCC588646 (A), VCC202273 (C), and GKT136901 (G) decreased Nox4 activity in a concentration-dependent manner (Figure 2A–2C). ROS production was completely inhibited by the flavoprotein inhibitor, diphenyleneiodonium, and by catalase, a scavenger of hydrogen peroxide (Figure 2C). Morphometric analysis of PA revealed significant medial remodeling by 4 weeks in monocrotaline-treated rats that was abrogated by both Nox4 inhibitor VCC588646 (A) and VCC202273 (C; Figure 2D and E). The relative activity of Nox4 inhibitors VCC588646 (A) and VCC202273 (C) against Nox1 activity was tested in a human embryonic kidney cell line expressing Nox1, NOXO1, and NOXA1 as previously described.²⁷ Both inhibitors reduced Nox1 activity but at concentrations higher than for Nox4 (Figure III in the online-only Data Supplement).

The effect of Nox4 inhibition on cardiac remodeling and indices of cardiopulmonary function is shown in Figure 3 and extended in Figure IV in the online-only Data Supplement. Monocrotaline significantly increased RV hypertrophy (RV/LV+S), which was inhibited with VCC202273 (C; Figure 3A) and also VCC588646 (A) and GKT136901 (G; Figure IVA and IVB in the online-only Data Supplement). Changes in RV function (RV systolic pressure, RV_max dp/dt), which were elevated in the monocrotaline-treated lungs, were also abrogated by Nox4 inhibition (Figure 3B and 3C; Figure IVC in the online-only Data Supplement). Noninvasive assessment of RV remodeling using digital ultrasound also revealed significant time-dependent increases in RV thickness in monocrotaline-treated rats, which were reduced by Nox4 inhibitors (Figure 3D and 3E; Figure IVD–IVE in the online-only Data Supplement). Cardiac output was decreased at 4 weeks post–monocrotaline treatment and improved with Nox4 inhibition (Figure 3F). There was evidence of PA remodeling as determined by analysis of PA hemodynamics, including a reduction in the velocity time integral, pulmonary ejection time, and PA acceleration time (Figure 3G–3I; Figure IVF–IVI in the online-only Data Supplement). Nox4 inhibition reduced RV thickness, increased cardiac output, and increased velocity time integral, pulmonary ejection time, and PA acceleration time (Figure 3E–3I; Figure IVD–IVI in the online-only Data Supplement), which collectively suggest a reduction in vessel stiffness (remodeling) and improved cardiac performance. Importantly, in all of the end points measured, all 3 inhibitors of Nox4 demonstrated significant efficacy in ameliorating the structural and functional changes with PH.

To assess whether inhibition of Nox4 can reverse, attenuate, or halt the progression of experimental PH in monocrotaline-treated rats with established PH, we measured time-dependent changes in RV and PA remodeling via ultrasound. Reversal protocols were used because they are more relevant to the clinical predicament and also can determine whether the target of interest remains functionally important in the later stages of disease and whether pre-existing end points of PH can be improved. The Nox4 inhibitor VCC202273 (C) was administered to monocrotaline-treated rats on the first detectable increase in RV thickness, which occurred at 3 weeks post–monocrotaline treatment (Figure 3E). As shown in Figure 4, VCC202273 (C; 1 mg/kg per day) significantly attenuated the progression of compensatory RV hypertrophy (Figure 4A), velocity time integral (Figure 4B), and PA acceleration time (Figure 4C) in monocrotaline-treated rats by weeks 5 and 6 post–monocrotaline treatment. At the conclusion of the experiment (6 weeks), invasive and postmortem end points were recorded (RV systolic pressure and Fulton index). Nox4 inhibition attenuated the monocrotaline-induced increases in RV systolic pressure and Fulton index at 6 weeks postinitiation of monocrotaline treatment (Figure 4D and 4E). Although there was no indication that Nox4 inhibition promoted reversal of PH beyond pretreatment levels, there was significant evidence for the attenuation of progression of the disease.

To identify the regions of the blood vessel wall and cell types expressing Nox4 in hypertensive PA, we performed immunofluorescence staining of lung sections from monocrotaline-treated rats and in lungs from normal (control) and human PH. In control rat PA, Nox4 was detected in the intima (endothelial cells) and cells of the adventitia (Figure 5, top panel). In 4-week monocrotaline-treated rats, using 2 different antibodies that are selective for Nox4 (Epitomics and Abcam), there was a dramatic increase in Nox4-positive cells in the adventitia (monocrotaline; Figure 5, left panels). In PA from monocrotaline-treated rats, Nox4-positive cells were detected in the remodeled medial layer, but this was largely in areas devoid of α-actin expression (monocrotaline; Figure 5, right panels). Nox4 exhibited a staining pattern in sections of human lung from individuals with normal and elevated pulmonary blood pressure (Figure 5, lower panels) that was consistent with the monocrotaline rat model. In lung sections from animals treated with the Nox4 inhibitor VCC202273 (C), monocrotaline-stimulated vascular remodeling and Nox4 expression were significantly attenuated (Figure 5; monocrotaline plus C). Furthermore, a time course of Nox4 expression in monocrotaline-treated rats revealed increased Nox4 expression and progressive medial remodeling at 2 and 3 weeks post–monocrotaline treatment (Figure VI in the online-only Data Supplement). Negative controls for the fluorescent secondary antibodies were performed using nonimmune immunoglobulin G in human lung sections and were without significant staining as shown in Figure VII in the online-only Data Supplement.

The location of ROS production in hypertensive PA was performed by immunofluorescence imaging for 8-hydroxydeoxyguanosine, a DNA nucleoside that is generated by ROS. 8-Hydroxydeoxyguanosine is used as an in vivo footprint of ROS production, and increased staining was detected in PA from 4-week monocrotaline-treated rats. The highest signal was observed in the adventitia, which overlapped significantly with the fibroblast marker, fibroblast-activating protein (monocrotaline; Figure 6A). A less pronounced increase in ROS levels was also observed in the media of hypertensive PA and was in proportion with the lower levels of Nox4 seen in the media compared with the adventitia. ROS levels in both the medial and adventitial layers were decreased in sections from rats treated with the Nox4 inhibitor VCC202273 (C; Figure 6A; monocrotaline plus C), suggesting that the elevated ROS production in hypertensive PA derives from increased Nox4 expression. In monocrotaline-treated PA, there was significant overlap between Nox4-positive cells in the adventitia and cells expressing fibroblast markers (cellular fibronectin [cellFN1], CD90) as well as the monocytic cell marker CD11B (Figure 6B). These data are supported by Western blot analysis of PA from control and 4-week monocrotaline-treated rats revealing increased protein expression of fibroblast markers such as CD90, cellFN1, periostin, tenascin N, and vimentin (Figure 7A and 7B; densitometry shown in Figure VA in the online-only Data Supplement). The effect of Nox4 inhibition on fibroblast and inflammatory markers was also determined in PA isolated from control, monocrotaline, and monocrotaline-treated rats treated with VCC202273 (C) using the reversal protocol (see Materials and Methods). Three-week exposure to a Nox4 inhibitor that was initiated after 3 weeks of monocrotaline resulted in prominent decreases in fibroblast markers (CD90, cellFN1, vimentin, periostin, and thrombospondin-4) as well as a marker of pericytes (chondroitin sulfate proteoglycan) compared with monocrotaline treatment alone (Figure 7B; densitometry shown in Figure VB in the online-only Data Supplement). A time course of Nox4 expression in isolated lung segments is shown in Figure VI in the online-only Data Supplement. These data, in conjunction with the functional indices of PA hypertension (Figure 3; Figure IV in the online-only Data Supplement), provide strong evidence for the pharmacological efficacy of Nox4 inhibitors and the importance of this pathway in the progression of PH.

To assess whether Nox4 modifies fibroblast function, we transduced human lung fibroblasts with an adenovirus expressing Nox4. Nox4-transduced fibroblasts exhibited a robust increase in cellular proliferation as demonstrated by real-time changes in electric impedance using electric cell–substrate impedance sensing arrays, as shown in Figure 8A. In addition, fibroblasts transduced with the Nox4 adenovirus displayed increased cell proliferation (total cell number; Figure 8B) and the number of viable cells as measured via the 3-(4,5-dimethylthiazol2-yl)-2,5-diphenyletrazolium bromide (MTT) assay (Figure 8C). Furthermore, using anti-CD90–immunomagnetic isolation, fibroblasts were isolated from control or 4-week monocrotaline-treated rat lungs. Western blot analysis revealed the expected enrichment of CD90-positive cells along with increased expression of Nox4 relative to the loading control, heat shock protein 90 (Figure 8D). Fibroblasts isolated from PA of monocrotaline-treated rats displayed a high level of proliferation that was decreased by the Nox4 inhibitor VCC202273 (C; Figure 8E). Adenoviral delivery of Nox4 also stimulated an increase in fibroblast motility (Figure 8F) compared with cells treated with a control virus. These data show that increased Nox4 expression alters fibroblast behavior, favoring a proproliferatory and promigratory phenotype. Lung fibroblasts exposed to the profibrotic growth factor transforming growth factor-β1 (TGF-β1) robustly increased Nox4 mRNA and protein levels as well as cellFN1 in a time-dependent manner (Figure 8G). To determine the relative ability of vascular cell types to express Nox4 in response to TGF-β1, we exposed human PA endothelial cells, human PA vascular smooth muscle cells (HPAVSMCs), and human lung fibroblasts to TGF-β1 for 24 hours. The phenotype of each cell type was confirmed using relevant cell-specific markers. We found the greatest induction of Nox4 in fibroblasts > human PA endothelial cells >> HPAVSMCs (Figure 8H). We next determined if Nox4 was important for the ability of TGF-β1 to induce profibrotic changes. In human fibroblasts, Nox4 small interfering RNA decreased Nox4 expression and subsequently prevented the expression of the TGF-β1–dependent genes, fibronectin, and ACTA (smooth muscle actin; Figure 8I).

A positive correlation between increased PA ROS production and elevated blood pressure of PH was observed >2 decades ago and has since been confirmed by numerous independent groups.^6,7,19,28,29 Although the consensus is for elevated ROS levels in PH, the source of ROS and its functional contribution to pathological remodeling of pulmonary blood vessels remain less well defined. The major goal of the current study was to determine the relative importance of Nox isoforms, a major source of cellular ROS, to the cause of PH.

Given the inability of any single animal model of PH to faithfully emulate the complexity of human PH, we first determined the expression of individual Nox isoforms in multiple rat models, including 2 inducible and 1 genetic model. Our studies were conducted primarily in rat models because they exhibit robust, irreversible pulmonary vascular remodeling, which is a defining characteristic of the human disease.^30,31 Changes in vascular gene expression were selectively measured in isolated PAs (down to the fourth branch). In PA, we found significantly increased expression of Nox4 mRNA in all 3 rat models. This correlated with increased protein expression in the monocrotaline model and suggests that Nox4 is a potential source of elevated ROS in hypertensive PAs. These data are consistent with prior studies in mice, rats, and in humans with PH.^{18–20,32,33} In the monocrotaline model, we observed increased expression of Nox1 and Nox2 mRNA, which is also consistent with previous reports.^34,35 However, Nox1 levels were not elevated in PA from the FHR, a genetic model of PH, or in the rat and mouse hypoxia models. Furthermore, in all 3 rat PH models, no increase in the expression of multiple accessory proteins that support the post-translational activation of Nox1 to Nox3 was observed. The significance of this is not known because the activity of Nox1 to Nox3 is dependent on stimulus-dependent assembly of a functional oxidase. In contrast, the elevated expression of Nox4 has additional significance because it is the only Nox isoform that is constitutively active. This means that increases in Nox4 gene expression translates to increased ROS.¹³ Collectively, these findings support elevated Nox4 expression in hypertensive PAs as a common variable in multiple rodent models of PH and in humans.

A major obstacle in the identification of functional roles for Nox isoforms in PH has been the lack of appropriate tools, particularly in the rat. To investigate the significance of Nox4 in PH, we used 3 structurally distinct Nox4 inhibitors.^36,37 This was done to ensure greater selectivity toward a common target (Nox4) and to minimize any possible nonspecific effects of individual inhibitors. The efficacy of these inhibitors against Nox4 activity was first confirmed in vitro and then in vivo in the monocrotaline rat model of PH. This model was selected based on the expeditious time course of PH (3–4 weeks versus 14 and 24 weeks for Sugen-hypoxia and FHR). We initially determined if inhibiting Nox4 would prevent the development of PH in monocrotaline-treated rats (day 1 treatment). Our objective in these studies was to determine if the increased Nox4 expression contributes to the induction of PH. We found that Nox4 inhibitors were effective at preventing PH as determined by indices of RV remodeling and noninvasive in vivo measurements of RV thickness, cardiac output, and remodeling of PA (velocity time integral, pulmonary ejection time, PA acceleration time). Our data are consistent with those reported recently by Green et al²⁰ in a mouse hypoxia model of PH. In that study, a related Nox4 inhibitor (GKT137831) improved indices of ventricular remodeling and PA wall thickening, but did not decrease RV systolic pressure. However, a major limitation of the mouse model is the absence of significant vascular remodeling. Our study revealed that Nox4 inhibitors were effective at preventing the robust hypertrophic remodeling of PA in monocrotaline-treated rats with PH. A further distinction of our study is that we also used Nox4 inhibitors in protocols to assess the ability to reverse established PH. This approach is more clinically relevant because the early stages of PH are generally silent, difficult to detect, and rarely treated. We found that Nox4 inhibitors were effective at halting the progression of PH. However, at the dose of inhibitor used and the time points studied, we did not observe an ability to fully halt the progression of the measured parameters of PH to levels below that of pretreatment in 3-week monocrotaline-treated rats. The reason for this is not yet known. In isolated PAs from rats with PH, we observed near-complete reversal of some indices such as the expression of CD90 and vimentin with Nox4 inhibition. However, incomplete reversal was seen with others including cellFN1, thrombospondin-4, and periostin. The reasons for diversity in the ability of Nox4 inhibitors to reverse changes in the expression of specific genes in PA and the inability to completely stop or reverse the progression of PH await further investigation. It remains possible that higher doses of Nox4 inhibitors or combination therapy with established therapeutics for PH (endothelin-1 antagonists, PDE5 inhibitors, etc) may enable more complete reversal of established PH. Germane to the overall thesis of this study, we found that inhibition of Nox4 was able to decrease the expression of genes in hypertensive PAs that are considered to be cellular markers of fibroblasts.

To determine a cellular mechanism by which Nox4 influences the progression of PH, we first performed immunofluorescence imaging in isolated lung sections. Nox4 expression was detected in both endothelium and adventitia of PA, and perivascular staining was prominently increased in animals and humans with PH. In the adventitia, Nox4 was detected in cell types expressing markers consistent with fibroblasts and monocytes. The perivascular expression of Nox4 also matched the location of ROS production and the markers of fibroblasts and monocytic cells. A surprising observation was the relative absence of Nox4 in PA vascular smooth muscle. Adventitial staining was confirmed using 2 distinct Nox4 antibodies, and in hypertensive PAs, the cell types expressing Nox4 in the medial layer did not overlap with those staining for smooth muscle actin. These results are supported by the higher levels of Nox4 detected in fibroblasts immunoisolated from monocrotaline-treated rat lungs and are consistent with previous reports showing elevated Nox4 in perivascular fibroblasts from individuals with idiopathic PA hypertension.¹⁷ Nox4 has been detected in most cell types with higher levels of expression seen in fibroblasts, particularly in the setting of pulmonary fibrosis.^17,26,38,39 Also consistent with previous studies, we observed that HPAVSMCs in culture have detectable levels of Nox4, which are increased in the presence of TGF-β1. However, these levels are less robust than those observed in fibroblasts.

The functional relevance of Nox4 in adventitial cells is not well described. The tunica externa or adventitia is a loosely defined collection of cells, including fibroblasts and immune cells, collagen, and elastic fibers that encircle the tunica media and intima layers of the blood vessel.⁴⁰ The adventitia orchestrates inflammation and vascular proliferation in response to injury, atherosclerosis, and pulmonary and systemic hypertension.⁴¹ The fibroblast is a primary cell type of the adventitia, responsible for the continual reorganization of the extracellular matrix via matrix deposition and secretion of growth factors, chemokines, and inflammatory cytokines.⁴² Aberrant vascular remodeling in PH occurs through increased inflammation, proliferation, and fibrosis, processes that collectively yield more muscular and less compliant PA.¹ Fibroblasts have key roles in these actions and actively secrete matrix, growth factors, and promote the inflammatory response by manipulating leukocyte recruitment and behavior.^43,44 We observed that increased expression of Nox4, in the absence of other stimuli, was sufficient to increase fibroblast migration and proliferation, and treatment of fibroblasts isolated from PAs from monocrotaline-treated rats with Nox4 inhibitors reduced cellular proliferation. Similarly, silencing Nox4 in fibroblasts decreased the ability of TGF-β to increase matrix and induce contractile gene expression, which is consistent with other reports.^26,39,45 A subset of bone marrow–derived cells, termed fibrocytes, can also be found in the adventitia, which have characteristics of both fibroblasts and macrophages.^23,46 Although we observed significant overlap of Nox4 staining with fibroblast and macrophage markers, it is not yet known if Nox4 is expressed in fibrocytes. In isolated PAs from monocrotaline-treated rats, we found increased expression of numerous markers that have been used to identify fibroblasts, including CD90, cellFN1, periostin, vimentin, and fibroblast activation protein. A role for Nox4 in regulating perivascular fibroblast behavior in PH is supported by results showing that Nox4 inhibition decreases the expression of fibroblast markers in isolated PAs from rats with established PH. These data suggest an ability of Nox4 to regulate the number and behavior of adventitial fibroblasts in animals with PH.

TGF-β1, an autocrine growth factor implicated in the pathophysiological vascular remodeling in PH,^47,48 robustly increased both Nox4 mRNA and protein levels in human lung fibroblasts. When comparing the ability of TGF-β1 to drive Nox4 expression in PA endothelial cells, PAVSMCs, and lung fibroblasts, we observed the greatest upregulation of Nox4 in fibroblasts, with less in VSMCs and none in PA endothelial cells. These data suggest that other mechanisms regulate Nox4 expression in endothelial cells. Indeed, numerous stimuli have been shown to modulate Nox4 expression, including hypoxia, angiotensin II, cyclic adenosine monophosphate, and protein kinase C,¹⁴ and it is possible that these signaling pathways contribute in varying degrees to the elevated Nox4 expression observed in PH. Our data are in agreement with previous studies,^32,49 and others have shown that TGF-β1 can upgregulate Nox4 expression in other cell types, including human cardiac fibroblasts, airway smooth muscle, and vascular smooth muscle.^32,38,50 The functional effects of Nox4 in vascular smooth muscle are similar to those reported in fibroblasts and include altered signaling, increased ability to proliferate and migrate,^32,33 and may reflect the altered phenotype of smooth muscle cells in culture. A role for fibroblasts in pathological remodeling in PH is supported by studies in transgenic mice with fibroblast-specific activation of TGF-β1 signaling. These mice develop mild PH with medial hypertrophy, inflammation, and fibrosis.⁵¹ Although this study strongly supports a role for fibroblast TGF-β1 signaling in aberrant pulmonary vascular remodeling, PH can be further exacerbated with additional stress on the endothelium and reflects the important contributions of multiple cell types in the development of PH.

In addition to Nox4, other Nox isoforms have been implicated in the development of PH. A recent study by Veit et al³⁵ reported increased Nox1 expression in isolated pulmonary microvessels and cultured PAVSMCs isolated from monocrotaline-treated rats. They did not observe increased expression of Nox4 (or Nox2) in cultured PAVSMC from monocrotaline-treated rats. The low expression of Nox4 in PAVSMC is in agreement with our study showing that in intact PA, Nox4 expression is low in the medial (smooth muscle) layer relative to the prominent expression of Nox4 observed in adventitial and endothelial cells. The importance of cell type to the expression of individual Nox isoforms is further emphasized in a study from Li et al,¹⁷ who showed increased Nox4, but not Nox1, expression in adventitial fibroblasts isolated from humans with idiopathic PA hypertension. However, a potential role for Nox1 is important to address because the inhibitors we have used to determine a role for Nox4 in PA hypertension also inhibit Nox1.³⁷ In our study, we also found that Nox1 expression was increased in the monocrotaline model of PA hypertension, but expression levels were not increased in rat or mouse hypoxia models or in the FHR. The functional importance of Nox1 was recently assessed in a study by Iwata et al⁵² using Nox1-knockout mice. Contrary to expectations, the loss of Nox1 did not protect against PH but instead promoted PH. At least in the mouse hypoxia model, Nox1 is proposed to repress PAVSMC proliferation through actions on potassium channels. Further studies are needed to define a functional role for Nox1 in rodent models and establish its importance in human PH. In contrast, Nox2-knockout mice are protected from PH.^7,18 How Nox2 contributes to PH is not fully understood. It is highly expressed in immune cells such as macrophages and emits superoxide rather than the preferential generation of hydrogen peroxide from Nox4. It may directly influence the contractile responses of PA,⁵³ but it has also been shown that Nox2-knockout mice fail to upregulate Nox4 in response to hypoxia.¹⁸ In contrast to Nox4, Nox2 requires cytosolic subunits for activation. We did not observe increased expression of these subunits in hypertensive PA, but it is not known whether this is important for assembly of a functional oxidase in PH. In our study, a functional role for Nox4 can be deduced from the use of Nox4/1 inhibitor GKT136901, which is not selective for Nox2,⁵⁴ and also from other studies using GKT137831, which has been shown to be effective in a mouse model of PH.²⁰

Nox4 has gained considerable attention as a primary source of ROS and cellular proliferation in the pathogenesis of both idiopathic pulmonary fibrosis and PH. The results of the current study are in agreement with both concepts and connect Nox4 as a common variable in fibroblasts (and other perivascular cells) that contributes to the remodeling of hypertensive pulmonary arterioles. The remodeling of blood vessels requires the participation of all 3 layers, and although numerous studies have proposed a central role for endothelial cells (inside-out remodeling), it has also been shown that vascular remodeling can be driven by changes in the adventitia (ie, outside-in). In both human and animal models of PH, prominent inflammation, activation, and restructuring of the adventitia are observed.⁵⁵ The adventitial location of Nox4 is, therefore, highly suited to orchestrate the changes in vascular inflammation and matrix deposition that are widely observed in PH. Current therapies for PH are ineffective in the long term, and new therapeutic strategies are needed. Our study supports the effectiveness of Nox4 inhibitors in halting the progression of experimental PH, although (as previously addressed) this strategy was unable to completely reverse the cardiopulmonary functional and morphometric changes that have already occurred. It remains to be determined whether this is a limitation of modalities that target Nox4 or whether a combination of therapeutic approaches will have superior efficacy.

We acknowledge the technical assistance of Louise Meadows, Yevgeniy Kovalenkov, and Stephen Haigh.